Search is not available for this dataset
article
stringlengths 4.36k
149k
| summary
stringlengths 32
3.35k
| section_headings
listlengths 1
91
| keywords
listlengths 0
141
| year
stringclasses 13
values | title
stringlengths 20
281
|
|---|---|---|---|---|---|
Functional redundancy limits detailed analysis of genes in many organisms . Here , we report a method to efficiently overcome this obstacle by combining gene expression data with analysis of gene-indexed mutants . Using a rice NSF45K oligo-microarray to compare 2-week-old light- and dark-grown rice leaf tissue , we identified 365 genes that showed significant 8-fold or greater induction in the light relative to dark conditions . We then screened collections of rice T-DNA insertional mutants to identify rice lines with mutations in the strongly light-induced genes . From this analysis , we identified 74 different lines comprising two independent mutant lines for each of 37 light-induced genes . This list was further refined by mining gene expression data to exclude genes that had potential functional redundancy due to co-expressed family members ( 12 genes ) and genes that had inconsistent light responses across other publicly available microarray datasets ( five genes ) . We next characterized the phenotypes of rice lines carrying mutations in ten of the remaining candidate genes and then carried out co-expression analysis associated with these genes . This analysis effectively provided candidate functions for two genes of previously unknown function and for one gene not directly linked to the tested biochemical pathways . These data demonstrate the efficiency of combining gene family-based expression profiles with analyses of insertional mutants to identify novel genes and their functions , even among members of multi-gene families .
Gene inactivation by the insertion of T-DNA is a common tool used for functional studies of genes in model plants such as Arabidopsis or rice . T-DNA insertional mutants have been generated for virtually all of the annotated genes in Arabidopsis thaliana . Recently , researchers working with Arabidopsis have combined the use of microarray technology with screens of T-DNA insertional mutant libraries to identify and characterize genes of unknown function . This strategy has successfully been used to characterize genes related to the formation of the secondary cell wall , those differentially regulated in phytochrome-mediated light signals , and those involved in host responses to pathogens [1] , [2] , [3] . These studies did not address the issue of multi-gene families , which limits functional analysis of genes in many plant species . Rice is a staple crop for more than half of the world's population and a model for other cereal crops . Therefore , studies of the rice genome are expected to help elucidate the function of genes in other major cereal crops , all of which have much larger genomes than rice . So far the functions of only a handful of rice genes have been characterized . The rice research community has recently generated a large collection of rice insertional mutant lines and thus far 27 , 551 gene loci with knockout mutations have been collected and 172 , 500 flanking sequences tagging those mutations have been sequenced ( http://signal . salk . edu/RiceGE/RiceGE_Data_Source . html ) [4] . Four microarray platforms covering nearly the entire rice transcriptome have also been developed [4] , [5] , [6] , [7] . Accordingly , studies similar to those carried out in Arabidopsis , combining microarray-derived expression data with reverse genetics to address gene function , can now be carried out in rice . Plants are continuously exposed to biotic and abiotic factors , light being one of the most important . In addition to providing the source of energy , light is also involved in regulating growth and development throughout the plant life cycle [8] . Genome-wide gene expression profiles of various light responses can help reveal the complicated physiological networks with which plants adapt to environmental changes . Toward that end , plant researchers have carried out microarray experiments with various plant species in efforts to identify light-regulated genes [9] , [10] , [11] . One of the major challenges facing scientists in the field of functional genomics , even those working with relatively simple model organisms like Arabidopsis , is the prevalence of gene families [12] , [13] . Such family members often encode redundant functions [14] , [15] . Because of the presence of gene families with functional redundancy , it is often the case that a gene , when mutated , will display no detectable phenotype . For example , an Arabidopsis line carrying a T-DNA insertion in a gene encoding a protein with a putative R3H domain , At5g05100 , hypothesized to be of importance in seed organ development , was reported to have no observable mutant phenotype [13] . Because Arabidopsis also has three genes , At2g40960 , At3g10770 , and At3g56680 , the lack of a mutant phenotype in this line is probably due to one or more of these genes encoding a function redundant of At5g05100 [13] . Efforts to overcome such functional redundancy have mostly focused on generating mutants that disrupt more than one gene family member simultaneously [16] , [17] . For example , RNA interference ( RNAi ) technology has also been used to simultaneously silence multiple members of a gene family [18] . Rice researchers encounter even larger gene families , more difficulty in scaling up experiments and greater time constraints associated with rice's longer life cycle than do researchers working with Arabidopsis . For example , whereas Arabidopsis has more than 600 genes in its receptor-like kinase ( RLK ) family , 80 in the bric-a-brac/tramtrack/broad ( BTB ) protein family , and 37 genes in the GARS family , rice has more than 1131 , 149 , and 52 members in those families , respectively [19] , [20] , [21] . Accordingly , a strategy for overcoming these difficulties and accelerating functional genomics analysis is especially helpful in crops like rice . Interestingly , the presence of multiple gene family members does not always mask phenotypic changes in gene “knockout” mutants . For example , 43 . 9% of the Arabidopsis seedling-lethal mutants and 45 . 2% of the Arabidopsis embryo-defective mutants phenotypically identified in two Arabidopsis studies carried defects in genes that were members of gene families [22] , [23] . These results suggest that all members of gene families are not necessarily functionally equivalent and individuals among them may play predominant roles [24] . Here we report the use of a microarray that covers nearly the entire rice transcriptome , the National Science Foundation-supported 45K microarray ( NSF45K , http://www . ricearray . org/ ) , to identify light-responsive genes in rice and subsequent functional analysis of a subset of those genes through screening gene-indexed mutant lines of rice . We describe the advantages of utilizing candidate genes derived from rice gene expression profiles to screen rice insertional mutant collections and discuss the biological significance of our findings .
We performed expression-profiling experiments with a newly-developed NSF45K microarray ( http://www . ricearray . org/ ) using two weeks-old rice leaf tissues harvested from plants grown under light and dark conditions ( See Materials and Methods ) . Because genetic backgrounds can affect expression profiles [25] and we wanted to select light-responsive genes that behaved similarly in different genetic backgrounds , identical experiments were carried out with four different rice varieties: Kitaake , Nipponbare , Tapei309 and IR24 . We used the LMGene Package developed by Rocke ( 2004 ) to identify 10361 , 4962 , 1933 , and 453 genes differentially expressed in response to light versus dark treatment at FDR p-values of 0 . 01 , 10−4 , 10−6 , and 10−8 , respectively ( Table S1 ) . To assess the difference in gene expression profiles among the varieties , we calculated correlation coefficients . Correlation coefficient values were 0 . 89–0 . 92 between the three japonica varieties ( i . e . Kitaake , Nipponbare , and Tapei309 ) whereas correlation coefficients values between subspecies ( i . e . japonica and indica ) were 0 . 80–0 . 82 . This result indicates that differences in genetic background clearly affect the expression patterns . Accordingly , by focusing on genes whose expression pattern is similar between varieties , we can avoid genes that show cultivar-specific light responses . To assess the usefulness of our microarray data we surveyed the expression patterns we obtained through microarray analysis for seventeen genes , including gene family members , which encode proteins for the seven steps in the well-characterized chlorophyll biosynthetic pathway ( Figure 1 ) . The expression patterns of all of the candidate genes in the seven steps of this pathway were validated using reverse transcriptase- ( RT- ) PCR ( Figure 1 ) . Based on this analysis we identified four unique genes ( Figure 1B , 1a , 1b , 1c , and 3 ) , and two predominantly expressed light-responsive candidate genes ( Figure 1B , 2-1 and 7-1 ) as good targets for studying the biological functions of genes involved in rice chlorophyll biosynthesis . A predominantly expressed light-responsive candidate gene is the one that is most significantly induced by light ( compared to other members of the multi-gene family ) . Previous studies in rice showed that the knockout mutants of three of the unique genes ( encoding the subunits comprising magnesium chelatase complexes in rice: magnesium chelatase subunit H ( CHLH , 1a ) , magnesium-chelatase subunit I family protein ( CHLI , 1b ) , and magnesium chelatase ATPase subunit D protein ( CHLD , 1c ) exhibited chlorina phenotypes [26] , [27] ) ( Figure S1 ) . The remaining unique gene ( magnesium-protoporphyrin IX monomethyl ester cyclase , MPE ) at step 3 is predicted to have a function similar to that of its Arabidopsis ortholog , CHL27 [28] . Knockout lines of the two predominantly expressed light-responsive candidate genes ( rice magnesium-protoporphyrin O-methyltransferase ( ChlM;Os06g04150; and gene 2-1 ) and rice chlorophyll a oxygenase ( Cao1; Os10g41780;and gene 7-1 ) ) had been previously identified and shown to display with light response-related phenotypes [29] , [30] ( Figure S1 ) . These results indicate that identification of unique genes or the predominantly expressed member of a gene family in the light is an effective method to target genes for further functional analysis . More descriptions on this strategy are shown in Figure S2 and Text S1 . ( RiceGE , http://signal . salk . edu/RiceGE/RiceGE_Data_Source . html ) [4] ( Table 1 ) . From our NSF45K light vs . dark experiment , we selected 365 candidate genes showing at least an 8-fold induction at the 10−4 FDR p-value . We then utilized collections of rice insertional mutants to identify lines carrying mutations in the light-responsive genes identified through expression analysis . To date , some 172 , 500 sequences have been generated from regions flanking insertional mutants in rice and they are publicly available at the Rice Functional Genomic Express Database ( http://signal . salk . edu/cgi-bin/RiceGE ) . Because we wanted to include 2 independently derived mutant alleles in our analysis of each candidate gene so as to help discriminate between phenotypic changes generated by somaclonal variation versus those resulting from the insertional mutations themselves , we limited our phenotypic analysis to 74 mutant lines with T-DNA insertions in a total of 37 candidate genes . The overall scheme we used for functional analysis based on our microarray experiment is presented in Figure 2A and Figure S2 ) . We classified the 37 candidate genes for which we had corresponding mutants into two groups according to whether the candidate gene belonged to a gene family or not . There were 12 unique genes ( those without gene family members ) ( Figure 2B and Figure S3A ) and 25 belonging to gene families ( see Materials and Methods ) . The latter class was further divided into two subgroups by considering the predominance of each gene's expression in the light based on the NSF45K light vs . dark array dataset . As a result there were 13 predominantly expressed-light-induced gene family members ( referred to as “P” in Figure 2B and as “Predominant” , marked with asterisks in Figure S3B ) and 12 gene family members that were not the predominantly expressed in the light ( referred to as “NP” marked in Figure 2B and as “Non predominant” , marked with sharps in Figure S3C ) . Because other more predominantly or equally expressed gene family members might compensate for a defective gene family member in the light , the non-predominantly expressed gene family members were not considered good candidates for functional analysis and were initially excluded from the functional analysis ( Figure S3C ) . Next , we identified 5 genes , Os03g48030 , Os11g05050 , Os02g58790 , Os09g37620 , and Os09g16950 , for which light responses between the NSF45K and BGI/Yale light vs . dark array datasets were significantly inconsistent and deleted them from our primary list of candidate genes for the initial round of functional analysis ( Figure 2 ) . Of these , Os03g48030 was a unique gene and Os11g05050 , Os02g58790 , Os09g37620 , and Os09g16950 were the members of their respective gene families the most predominantly expressed in the light in the NSF45K array data set . We also included one unique gene ( Os07g46460 ) and one predominantly light-induced gene family member ( Os03g37830 ) in our candidate gene list based only on our own data because information on their light responses was not available among the BGI/Yale data ( Figure 2 ) . We then screened the remaining mutants , those associated with 11 unique genes and 9 predominantly light-induced gene family members , to determine their phenotypes ( Figure 3 ) . We also assayed the phenotypes of the knockout lines associated with the 17 genes we had eliminated from our primary list of candidate genes to check the efficiency of identifying mutant phenotypes for the not predominantly light-induced genes in a gene family and/or genes with expression patterns that weren't consistent between the NSF45K and BGI/Yale light vs . dark array datasets . ( Detailed data regarding all 37 of the genes that were functionally analyzed are presented in Table S2 . ) Our criteria for selecting candidate genes that respond to light might have inadvertently eliminated from consideration genes among the BGI/Yale data that exhibit condition-dependent light responsiveness . We initially carried out functional analysis for 20 selected candidate genes , 11 unique genes and 9 predominantly light-induced gene family members ( Figure 3A and 3B ) . First , we identified defective phenotypes associated with six of the 11 unique genes that we analyzed for function from our list of top candidate candidates . Of these , the phenotypes of knockout lines associated with Os01g01710 ( 1-deoxy-D-xylulose 5-phosphate reductoisomerase , Dxr ) and Os03g04470 ( Expressed protein ) were albino and displayed chlorotic leaves , respectively , 2 weeks post-sowing ( Figure 3A ) . Knockouts of Os01g71190 ( Photosystem II subunit 28 , Psb28 ) , Os02g57030 ( Expressed protein ) , and Os07g46460 ( ferredoxin-dependent glutamine:2-oxoglutarate aminotransferase , Fd-GOGAT ) displayed pale green phenotypes ( Figure 3A ) . Of these , a mutation in the Fd-GOGAT gene displayed photo-bleached leaves two weeks later after revealing pale green leaves ( Data not shown ) . Knockouts of Os04g37619 ( Zeaxanthin epoxiydase , Aba1 ) produced dwarf mutants ( Figure 3A ) . We noted that the mutant phenotypes associated with three of these 6 genes , those encoding DXR , Fd-GOGAT , and ABA1 , were similar to those associated with their Arabidopsis orthologs [31] , [32] , [33] . The Arabidopsis ortholog of Os01g71190 encodes photosystem II ( PSII ) reaction center Psb28 protein ( Psb28 ) that was first identified from PSII of Synechocystis 6803 with Psb27 [34] . The function of this PSB28 has not yet been well- characterized , although it is predicted to serve a role as a regulatory protein based on its substoichiometric amount [34] , [35] . Similarly , Arabidopsis lines carrying a mutation in Psb27 did not display a severe phenotype . Recovery of PSII activity after photoinhibition was delayed in the Arabidopsis psb27 mutant supporting a role in PSII for this gene [36] . The mild phenotype displayed by the Psb28 T-DNA insertional rice plants in Psb28 gene suggests that this gene product might serve as a regulatory protein to stabilize PSII activity . The other two unique genes for which we identified corresponding mutant phenotypes as a result of our analysis encode as yet unidentified proteins . The endogenous retrotransposon Tos17 has been shown to be an efficient insertional mutagen in rice and phenotypes of some 50 , 000 M2 generation insertion lines carrying Tos17 insertions have been reported [21] . We used the available Tos17 phenotypic data to assess the functions of some of our candidate genes . Phenotypes similar to those identified in the T-DNA insertional mutants for Os01g71190 ( Psb28 ) , Os03g47610 ( Dxr ) , and Os04g37619 ( Aba1 ) have also been observed for the corresponding Tos17 insertional mutant lines ( Table 2; [21] , [37] ) . We could not identify mutant phenotypes for the other unique genes containing T-DNA insertional mutations that we analyzed . We assume that phenotypes associated with two genes , Os01g40710 and Os08g40160 , could not be determined due to confounding somaclonal variations ( Figure S4 ) in the one mutant line available for each of these genes for analysis . We could not analyze the second mutant line corresponding to each of these genes due to poor seed set . Lines with T-DNA insertions in Os03g47610 , lines 1A-08723 and 1A-17021 , did not produce any homozygous progenies and as a result the phenotypes associated with these mutations were also not determined ( Table 2; Table S2 ) . On the other hand , while we did identify homozygous progenies and their siblings among mutants with insertions in Os03g06230 and Os03g17960 , we did not observe phenotypic differences between them ( Table 2 ) . Targeting functional analysis to unique genes is an effective way to significantly increase the efficiency of identifying genes corresponding to defective phenotypes . Utilizing microarray-derived expression profiles of unique genes can increase the efficiency of functional analyses [1] , [22] , [38] . The efficiency ( 6/11 ) with which we were able to identify phenotypes associated with mutations in unique genes demonstrates the power of combining knowledge of gene copy number and gene expression patterns . Of the nine predominantly light-induced gene family members that were consistently induced in the light , we found four for which mutant phenotypes co-segregated with a T-DNA insertion in the gene ( Figure 3B ) . These four genes encode carbonic anhydrase 1 ( CA1 ) , serine hydroxymethyltransferase 1 ( SHMT1 ) , nitrate reductase 1 ( NR1 ) , and Vitamin C defective 2 ( VTC2 ) , respectively . The rice T-DNA insertional line carrying a mutation in the Shmt1 gene ( Os03g52840 ) , line 1D-03944 , displayed variegated chlorina leaves ( Figure 3B ) . The line 2B-60065 with a T-DNA insertion in Ca1 ( Os01g45274 ) showed an oxidative stress-related phenotype , necrosis in the middle of the leaf , and a little growth retardation ( Figure 3B ) . Line 4A-50280 , with a T-DNA insertion in Nr1 ( Os08g36480 ) , displayed dwarfism , and line 3A-14221 , with a T-DNA insertion in the Vtc2 gene ( Os12g08810 ) , exhibited a pale green and later photo-bleached leaves phenotype ( Figure 2B ) . The Tos17 line with an insertion in the rice Shmt1 gene exhibited the same phenotype as the corresponding T-DNA mutant , line 1D-03944 ( Table 2 ) . The phenotypes associated with mutations in Shmt1 ( Os03g52840 ) , Nr1 ( Os08g36480 ) , and Vtc2 ( Os12g08810 ) were also reminiscent of the phenotypes associated with mutations in the orthologous genes in Arabidopsis [39] , [40] , [41] , [42] . Mutant phenotypes associated with T-DNA insertions in Os01g08460 , and Os05g47540 were not determined due to confounding somaclonal variations ( Figure S4 ) . The homozygous progenies of mutant lines with insertions in two other genes , Os03g37830 and Os06g04510 , did not show visible phenotypic changes ( Table 2 ) . Our success in conducting functional analyses of gene family members that are the predominantly expressed in the light , and consistently so from one microarray experiment to the next , suggests that the functions of gene family members can be successfully analyzed by utilizing data obtained using microarrays representing nearly complete plant transcriptomes . Analyses that consider both sequence similarity and predominance of gene expression has also been reported to be quite effective in functional analysis of yeast genes [14] . We also carried out functional analysis of 5 genes that showed inconsistent gene expression patterns when we compared our NSF45K light-response experiments and the BGI/Yale light vs . dark experiments . One of them , Os03g48030 ( designated U9 in Figure 2 and Figure 3 ) , is a unique gene and the other four genes , encoding lectin protein kinase ( Os09g16950 ) , stem-specific protein TSJT1 ( Os11g05050 ) , flavin-containing monooxygenase family protein ( Os09g37620 ) , and an expressed protein ( Os02g58790 ) , were the predominantly light-induced members of their respective families among the NSF45K array-derived data but not in the BGI/Yale light vs . dark dataset . No defective phenotypes were observed among the mutant lines with T-DNA insertions in any of these genes ( Table S3 ) . One possible reason for this result is the presence of redundant metabolic networks or the absence of appropriate screening conditions [15] , [43] , [44] . Mutants carrying insertions in 12 genes that were not the predominantly expressed light-induced member of their respective gene family were also examined in this study . A visible phenotype was observed in the homozygous mutant progenies associated with only one of these genes , Os07g05000 ( R8-2 ) ( Figure S3C and Figure S5 ) , which belongs to the family of genes encoding aldo/keto reductases . Phenotypic changes were not observed in the homozygous segregants of lines with mutations in any of the other genes ( Table S3 ) . The absence of detectable abnormal phenotypes associated with these other genes is generally believed to be due to one ( or more ) family members compensating for the function of the mutated gene [14] , [15] . Line 3A-03008 , which carries a T-DNA insertion in Os07g05000 , showed a weakly pale green phenotype and slight growth retardation ( Figure S5 ) . Identification of phenotypes associated with the other mutations may require specific conditions under which there will be no compensatory gene expression from other family members . In cases of gene families without a predominantly expressed member under specific experimental conditions , microarray data can still be used to identify the multiple significantly expressed genes in a family so that they can be subjected to RNA-silencing techniques as has been carried out by Miki et al . [18] , [45] for the rice genes encoding homologs of mammalian Rac GTPase , OsRac1 and OsRac5 . Of our list of non-predominantly light-induced genes , there were two genes from same family expressed under light conditions ( Os12g03070 and Os11g03390 ) . Both encode an FHA domain , which is a putative nuclear signaling domain found in protein kinases and transcription factors . However , we did not observe a phenotype in knockout lines of Os11g03390 ( R12-1 ) [46] ( Table S3 ) . Similarly , we did not observe phenotypic changes associated with T-DNA insertions in members of the gene families encoding ABC1 proteins ( Os02g57160 , R3-1 and Os04g54790 , R5-1 ) , S1-RNA binding domain proteins ( Os04g54790 , R4-1 ) , and glycine dehydrogenases ( Os06g40940 , R7-1 ) . Further experiments to generate double mutants for these family members and their light-induced relatives will be required to elucidate their functions . In Arabidopsis , the light-responsive functions of genes in gene families such as POR and PHOT have been clarified by using double mutants of two family members , porbporc and phot1phot2 , respectively [47] , [48] . When we consider severity of phenotypes , two of six lines carrying defects in unique genes and one of four lines carrying defects in the gene family member predominantly expressed in the light died at early seedling stage ( Figure 3 ) . Therefore , these three genes are essential for survival . Overall , phenotypes of mutants with defects in unique genes were more frequently observed than those of mutants carrying defects in predominantly light-induced gene family members ( compare Figure 3A with Figure 3B ) . We suspect , therefore , that other members in a gene family may carry out a somewhat compensating role [14] , [15] . Similarly , we did not find phenotypic changes associated with mutations in genes that were not the predominantly light-induced members of their respective families , with the exception of mutations in a gene encoding an aldo/keto reductase protein . This result also supports our hypothesis that other gene family members can compensate for the mutation . Despites these observations , we can not rule out the possibility that the absence of phenotype is due to non-optimal environmental conditions [44] . In summary , we identified phenotypic changes in rice lines carrying mutations in 10 out of 20 unique genes or genes that were the most predominantly light-induced members of their respective families . In contrast , we discovered only one phenotypic change among the lines carrying mutations in the 17 other genes that either showed inconsistently light-induced expression among different microarray data sets or were not the predominantly light-induced members of their respective gene families ( Table S3 ) . Microarray data were very useful as criteria for prioritizing candidate genes for functional analyses . Consideration of the expression patterns of all the genes within a gene family is an effective way to approach study of the functions of gene family members [14] . Functional profiling of genes related to the phytochrome-mediated signaling pathway in Arabidopsis was recently carried out [3] . We used this data set to further test the usefulness of our method ( see Materials and Methods ) . Thirty two genes were selected for this functional profiling analysis . Of these , mutants in seven genes displayed statistically significant photomorphogenic phenotypes . Except for one gene ( At2g46970 ) whose gene expression profiling data was not available , we found that six genes were either unique sequences or the predominantly expressed gene family member in the red light . In contrast , mutations in the remaining 25 genes showed less significant photomorphogenic phenotypes; thirteen of them displayed mild or severe defects in photoresponsiveness and 12 did not showed distinguishable phenotypes ( Figure S6 ) [3] . Of these , 10 genes were not the predominantly expressed gene family member in the red light whereas 13 genes ( except two genes; At3g21550 and At3g21330 whose gene expression profiling data was not available ) were either unique sequences or the predominantly expressed gene family member in the light ( Figure S6 ) . Thus , this functional profiling analysis in Arabidopsis also indicates that predominantly expressed gene family members as well as unique genes are good targets for functional validation . Mutant phenotypes clearly suggest functions for targeted genes and also for the pathways those genes are associated with . Therefore , understanding relationships among multiple pathways containing mutants defective in the plant's response to light will help us elucidate the light response . To do this , we identified genes among various pathways that were co-expressed with the 10 genes for which we had identified mutant phenotypes in this study . We found that ten pathways ( http://www . gramene . org/pathway/ ) involved 7 of the genes for which we had identified mutants ( Figure 4; Table S4 ) . Sixty nine genes in these 10 pathways were selected as described in Materials and Methods . Additionally , three single-step reactions unlinked to any of these pathways but involving the other three genes for which we identified mutant phenotypes in this study , U4 ( Os02g57030 ) , U5 ( Os03g04470 ) , and Ca1 ( P2-1 , Os1g45274 ) were also included in this analysis ( see Table 2 ) . All together , 72 genes involved in the 10 pathways and three reactions were selected for hierarchical clustering analysis ( Table S4 ) . Then , we selected 10 datasets with which to carry out the analysis ( Table S4 ) : log2 fold change values of NSF45K light vs . dark and four different types of light vs . dark datasets generated by BGI/Yale array [49] , and log2 fold change values of five different tissues vs . cultured cells [50] . We selected candidate genes in each pathway that are unique or are the predominantly light-induced gene family member ( except several steps not represented by predominantly light-induced gene family members ) . We found that gene expression patterns of 67 out of the 72 genes are light-inducible in at least two of the light treatments ( Figure 4 ) . Fifty-five genes have GO terms in the cellular component category and 46 of them have a chloroplast GO term in the cellular component category . Seven genes are predicted to have role in the mitochondrion ( Figure 4 ) . Most of the genes used for the clustering analysis are predicted to perform their light response-related function in chloroplasts or mitochondria ( Figure 4 ) . As a result of the hierarchical clustering analysis we identified 10 gene clusters ( Figure 4 ) . As has been previously reported [50] , [51] , [52] , co-expression analysis is useful for revealing functionally coherent groups of genes . We next looked for relationships among different pathways by utilizing Cytoscape software to analyze the results we obtained from our co-expression analysis ( see Materials and Methods ) . Cytoscape is an open source software for integrating biomolecular interaction networks with high-throughput expression data [51] . The results of this analysis are shown in Figure S7 . First , cluster III ( purple lines in Figure 4 and Figure S7 ) contained 3 components of PSI , one of PS II , three from the photorespiratory pathway , one from the ammonia assimilation pathway , and one from the chlorophyll biosynthetic pathway . Of these , ferredoxin-dependent glutamine:2-oxoglutarate aminotransferase ( Fd-GOGAT , U11 ) couples with glutamine synthetase 2 ( GS2 ) for assimilating ammonium produced by photorespiration [33] , [52] . In Figure 4 and Figure 5 , the Fd-GOGAT gene ( U11 ) , which generates glutamate at step 2 of the ammonia assimilation pathway is co-expressed with the PS I and PS II components Os08g44680 , Os12g23200 , Os07g25430 , and Os01g64960 . GS2 ( OS04g56400 ) at step 1 of this pathway is co-expressed with other PS I and PS II components ( Os09g30340 and Os08g10020 ) . PS I and PS II supplies ATP for the reaction of GS2 and reduced ferredoxin ( Fdrd ) for Fd-GOGAT [53] . This dependency of the ammonia assimilation pathway on photosynthesis is supported by the co-regulation of several PS I and PS II components with two genes in the ammonia assimilation pathway ( Figure 4 , Figure 5 , and Figure S7 ) . The photorespiratory pathway is known to be tightly correlated with nitrogen cycles [54] . Co-expression patterns of genes encoding three early steps of the photorespiratory pathway with the Fd-GOGAT gene suggest a close relationship of photorespiration with the ammonia assimilation pathway ( purple lines in Figure 4 , Figure 5 and Figure S7 ) . This makes sense because the ammonia assimilation pathway plays a role in preventing loss of ammonia , which is generated from step 6 of the photorespiratory pathway by refixing it to glutamate [53] , [55] , [56] . Furthermore , the resulting amino acid ( i . e . glutamate ) reacts with glycine to synthesize glutathione in the peroxisome , suggesting that this pathway is important for protecting photosynthetic apparatus ( e . g . PS II ) from toxins such as free radicals [55] . Also , co-regulation of genes in the three early steps of the photorespiratory pathway with several PS I and II components suggests that oxygen generated by photosynthesis triggers the photorespiration pathway [57] , [58] . Co-regulation of genes involved in photosynthesis , ammonia assimilation and photorespiration can explain their coordinated functions [56] . Of the genes in cluster III , the phenotypes of a mutant line with a T-DNA insertion in Fd-GOGAT ( U11 ) and a mutant lines under-expressing the Rca1 gene ( Os11g47970 ) were characterized ( Figure 4 , Figure 5 and Figure S7 ) [59] . The Rca1 under-expressed mutant displays chlorotic leaves [59] . The mutant ( 3A-01082 , this study ) with a T-DNA insertion in Fd-GOGAT gene displayed pale green leaves shortly after germination ( Figure 3 ) . The same mutant displayed chlorotic leaves four weeks after germination and is similar to the phenotype of one of the Rca1 under-expressed mutants ( data not shown ) [59] . These results indicate that co-expressed groups of genes carry out closely related functions as reported in other species [1] , [33] , [60] . Therefore we can predict that mutations in other genes in this cluster will display similar phenotypes to those observed in the Fd-GOGAT and Rca1 mutant lines . Those predicted to display such phenotypes include phosphoglycolate phosphatase ( Pgp , Os04g41340 ) [61] at step 2 of the photorespiration pathway . In support of this hypothesis , a mutant with defects in Arabidopsis ( PGLP1 ) displayed chlorotic leaves at the early seedling stage , the phenotype was similar to that observed for Rca1 and U11 mutant lines [61] . The MEP pathway is a unique and essential process for plants , algae and bacteria [67] , [68] , [69] . The final metabolites of the pathway are isopentenyl pyrophosphate ( IPP ) and its isomer dimethylallyl pyrophosphate ( DMAPP ) , which are used for the synthesis of isoprenoids ( such as isoprene ) , carotenoids , plastoquinones , phytol conjugates ( such as chlorophylls and tocopherols ) , and hormones ( such as gibberellins and abscisic acid ) [60] , [70] , [71] , [72] , [73] , [74] . Co-expressed Cluster I consists of genes ( Os05g33840 , Os07g36190 and Os05g33840 ) controlling steps 1 , 2 , and 8 of the MEP pathway , genes ( Os04g37619 and Os07g10490 ) controlling steps step 3 and 6 of the carotenoid biosynthetic pathway , genes ( Os03g36540 and Os03g59640 ) encoding two components ( CHLI and CHLD ) controlling step 1 of the chlorophyll biosynthetic pathway , gene ( AK059143 ) encoding Psba of photosystem II , and gene ( OS09g29070 ) controlling step 1 of the vitamin C biosynthetic pathway . Co-expression of genes in the MEP pathways with those in the chlorophyll and carotenoid biosynthesis pathways supports the hypothesis that the syntheses of pigments mediated by metabolite ( s ) resulted from the MEP pathway are dependent on photosynthesis ( Figure 4 and Figure S7 ) . In Arabidopsis , isopentenyl-PP or dimethylallyl-PP cause feedback regulation of step 1 ( 1-deoxy-D-xylulose-5-phosphatesynthase , DXS ) of the MEP pathway ( Figure 6 ) [72] . The co-regulation of genes controlling step 1 and step 8 suggest the existence of feedback regulation between these two steps in rice as reported in Arabidopsis [72] ( Figure 4 , Figure 6 , and Figure S7 ) . In an effort to deduce the hierarchical sequence of the enzymes involved in this pathway , we assessed the expression patterns of 12 genes controlling all steps in the MEP pathway in the homozygous mutant dxr ( step2; progenies of Line 1A-14224 , Ho-1 and Ho-2 in Figure 6 ) and its wild-type segregants ( WT1 and WT2 in Figure 6 ) . This analysis revealed that a defect at step 2 inhibits the previous step ( step 1 ) in this pathway . These results indicate that the gene products controlling step2 and step8 of the MEP pathway cause feedback regulation of step 1 and that these three steps might be controlled by the same regulatory molecule ( Figure 4 , Figure 6 , and Figure S7 ) . In addition to the predicted feedback regulation of step 2 to step 1 , transcription analysis of 12 candidate genes in MEP pathway in the dxr mutant revealed probable compensating rotes that could be taken to make up for the defect in the family member , Dxs1 , that is predominantly expressed in the light . There are three gene family members ( Dxs1 , Dxs2 , and Dxs3 ) associated with step 1 ( DXS ) . The induction in the dxr mutant of the two other gene family members of step1 , Dxs2 and Dxs3 , supports a compensating roles for gene family members . The observed expression patterns of genes in other steps in this mutant facilitated predictions as to how these compensating routes would affect hierarchical sequence of the enzymes involved in this pathway . For example , as predicted , the Dxr ( Os01g01710 ) gene at step 2 was not expressed in the mutant . The C-methyl-D-erythritol4-phosphate cytidylyl transferase ( Cmt , Os01g66360 ) gene at step 3 was also repressed and the effect of a mutation in the Dxr gene might be extended to step 3 ( Figure 6 ) . However , the Cmk ( Os01g58790 ) gene at step 4 ( 2-C-methyl-D-erythritol4-phosphate cytidylyl transferase 1 , Cmk ) was up-regulated in the rice dxr mutants suggesting that a compensating route might be associated with this step as well . The predominantly light-induced gene family members in the four steps following the Cmt1 ( Step 4 ) , 2-C-methyl-D-erythritol 2 , 4-cyclodiphosphate synthase ( Mcs , Os02g45660 ) , 4-hydroxy-3-methylbut-2-en-1-yl diphosphate synthase ( Hds , Os02g39160 ) , 4-hydroxy-3-methylbut-2-enyl diphosphate reductase 1 ( Hdr1 , Os03g52170 ) , and Isopentenyl-diphosphate delta-isomerase 1 ( Idi1 , Os07g36190 ) , exhibited expression levels similar to those observed for the wild-type segregants . Interestingly , Hdr2 and Idi2 ( in step 7 and step 8 of Figure 6 , respectively ) , which are not the predominantly expressed family members in light-grown wild-type rice , were up-regulated in the dxr mutants ( Figure 6 ) . This result suggested that there might be another compensation mechanism occurring to cope with the upstream blockage of the MEP pathway . As a consequence of up-regulation in step 7 ( Hdr2 , Os03g52180 ) , the gene expression of step 8 ( Idi2 , Os05g34180 ) would be increased . However , these compensating routes must not to be predominant because the original route has evolved to be the most effective . Finally , plants homozygous for this mutant locus were lethal despite the functioning of these two putative compensating pathways . The specifics of these proposed compensating routes await validation through further analyses . Most of genes involved in the carotenoid , abscisic acid , chlorophyll , and the tocopherol ( Vitamin E ) biosynthesis pathways , predicted to be downstream of the MEP pathway [60] , [75] , are light-responsive in various light vs . dark experiments ( Figure 4 and Figure S7 ) . These results indicate a probable metabolic connections between the MEP pathway and these four downstream pathways . The connections are likely made through intermediates synthesized via the MEP pathway in the light . In support of the importance of this pathway in metabolism , we observed an albino leaf phenotype in homozygous progenies of lines 1A-14224 and 1C-07431 , carrying T-DNA insertions in the Dxr gene at step 2 and 2-C-methyl-D-erythritol 2 , 4-cyclodiphosphate synthase ( Mcs , Os02g45660 ) at step 5 , respectively ( Figure 3 , Figure 6 , and Figure S8 ) . A mutation in Arabidopsis Mcs1 revealed a similar albino phenotype . These plants died at the early seedling stage [76] . In Arabidopsis or tobacco , lines carrying mutations in other genes of this pathway displayed albino- or chlorotic-leaf phenotypes [72] , [77] , [78] , [79] , [80] , [81] . Thus , our results indicate that the MEP pathway in rice performs key roles in generating pigments such as chlorophyll or carotenoid as it does in other plant species [31] , [67] , [82] . This study provides a method for identifying sets of candidate genes involved in specific biochemical pathways . Our results reveal that light regulated gene expression controls diverse metabolic networks [83] , [84] and that co-expression analysis is an effective strategy for elucidating the plant response to light .
Nipponbare , Kitaake , TP309 , and IR24 rice seeds were germinated and grown in the greenhouse . Nipponbare , Kitaake , and TP309 are japonica cultivars and IR24 is indica . For light treatments seedlings remained in the greenhouse for two weeks . For dark treatments , seedlings were moved after 7 days to a dark incubator ( Percival Scientific , Inc . , Perry , IA ) and maintained at 28°C for another 7 days . Of 365 candidate genes showing at least an 8-fold light induction during our NSF45K light/dark experiment , 161 had T-DNA insertions in them corresponding to mutant lines in the plant functional genomics lab ( PFG ) and 45 of those had at least two insertionally mutated alleles ( and corresponding mutant lines ) present in the database ( RiceGE , http://signal . salk . edu/cgi-bin/RiceGE ) . However , 8 of the 45 had the same or nearly the same insertion sites in the two different mutant lines and so those lines were excluded . Finally , we ordered seventy-four T-DNA insertional lines containing mutants in 37 genes . Sixty-eight of these knockout lines ( japonica cv . Dongjin or Hwayoung ) , after excluding six lines which had insufficient seeds , were grown in the greenhouse . We observed the phenotypes of these knockout lines for 4 weeks after they germinated . The progenies showing phenotypic changes and their wild-type siblings from individual lines were harvested to extract genomic DNAs ( described below ) for co-segregation analyses as indicated in Figure S9 . Usually , 15–20 rice plants are used for testing co-segregations and repeat two or three times this experiment . We selected lines having at least two co-segregating mutants and also repeated it at least twice . We visually picked out progenies displaying expected phenotypes such as color defects ( albino or pale green ) , growth retardation or oxidative stress-related symptoms . Next , genomic DNA was extracted from mutants and from their phenotypically normal siblings ( Figure S9 ) . The genotypes of the siblings in each mutant family were determined by carrying out PCRs using two sets of primers: one designed to identify the rice gene that had been knocked out by using primers containing target gene sequences in front of and behind the T-DNA insertion site , the other designed to verify the insertion of T-DNA in the gene by using primers that amplify the hygromycin phosphotransferase ( hph ) gene contained within the T-DNA insert ( Figure S9 ) . The primers located upstream and downstream of the T-DNA insertional sites in each line were designed based on sequence information available from the Rice Functional Genomic Express Database ( Rice GE , http://signal . salk . edu/cgi-bin/RiceGE ) . The PCR amplifications were carried out in 20 µl volumes of a mixture that contained 20 ng of plant DNA , 10× Taq buffer , 0 . 2 mM dNTP , 0 . 5 unit Taq polymerase ( Invitrogen ) , and 0 . 2 µM of the primers for 35 cycles at 94°C for 60 s , 60°C for 60 s , and 72°C for 150 s . All primers ( Sigma ) for genotyping are described in Table S5 . Leaves from rice plants grown for 2 weeks in the greenhouse or in a dark incubator were collected and total RNA was isolated using TRIZOL reagent according to the manufacturer's instructions ( Invitrogen , Carlsbad , CA ) . The total RNA was DNaseI-treated for 15 minutes then purified using the RNeasy Midi Kit ( Qiagen , Germantown , MD ) . The total RNA was then enriched for poly-A RNA by using the Oligotex mRNA Kit ( Qiagen ) . All steps were performed according to the manufacturer's instructions . The quantity of total RNA and mRNA were determined by measuring absorbance at 260 nm and 280 nm by using a Nanodrop ND-1000 ( Nanodrop , Wilmington , DE ) . In addition , the level of protein contamination in the RNA was determined based on the A260/A280 ratio . Only RNA samples with ratios of 2 . 0–2 . 2 were used for these experiments . Reverse transcriptase- ( RT- ) PCR were carried out as used in previous study [85] . All hybridizations were done at the Arraycore Microarray Facility at the University of California , Davis Arraycoreucdavis . edu . Probe labeling , hybridizations with the NSF45K microarray , slide scanning and identification of spot intensity were as described ( Jung et al . submitted ) . To minimize variations caused by experimental procedures , replicated data was normalized using the Lowess normalization method in the LMGene Package [86] , [87] . To identify differentially expressed genes , we used the publicly available R program LMGene developed by Rocke [86] . FDR ( false discovery rate , adjusted p-value ) and log2 fold changes of light over dark were generated for all genes . The expression data from these experiments are available through Gene Expression Ominibus ( GEO ) ( Accession # GSE8261 ) . To identify genes consistently expressed in response to light among different array platforms , we selected genes that were induced in our NSF45K array experiments and also showed at least 0 . 5 log2 values ( 1 . 4-fold induction ) in more than two light intensity conditions of the BGI/Yale light vs . dark array data . The Rice Multi-platform Search page ( http://www . ricearray . org/matrix . search . shtml ) is a tool that allows users the ability to search across four different rice oligo microarray platform types ( Affymetrix , Agilent , BGI/Yale , and NSF45K ) to determine which oligos from each platform represent to a common gene target . More detailed information on Rice Multi-Platform Search Tool is available at http://www . ricearray . org/matrix . search . shtml . The data from the BGI/Yale light vs . dark array dataset and the Affymetrix array data for seedling leaves and shoots for the seventeen genes in the chlorophyll biosynthesis pathway , twelve genes in the MEP pathway , and the thirty-seven genes on which functional analyses were conducted were extracted by using the rice multi-platform microarray search tool and publicly available Affymetrix and BGI/Yale array data . The detailed information on the multiplatform array data used in this study is presented in Table S6 and is also available at the NCBI GEO ( http://www . ncbi . nlm . nih . gov/geo/ ) . To make images using the multi-platform microarray data , we used TIGR MultiExperiment Viewer software ( MeV , http://www . tm4 . org/mev . html ) . We generated two tab-delimited multiple-samples files ( tdms files ) consisting of log2 ratios of light over dark treatment from the NSF45K and the BGI/Yale light vs . dark array datasets and of log2-transformed spot intensities of 23 Affymetrix array datasets related to development ( Table S6 ) . The tdms files were loaded onto the MeV and the resulting image data were used for creating Figure 2 , Figure 4 , Figure S6 , and Figure S10 . Gene families within the predicted rice proteome were identified using Pfam and BLASTP and more detail on these identifications can be found at http://www . tigr . org/tdb/e2k1/osa1/para . family/para . method . shtml . A total of 3842 paralogous protein families containing a total of 20729 proteins were identified ( http://rice . tigr . org/tdb/e2k1/osa1/para . family/index . shtml ) . Table S7 shows the gene families we identified in rice . To identify Arabidopsis gene family members for Figure S6 , we used GenomeNet browser ( http://www . genome . jp/ ) and the protein sequences which had more than 200 score bits or 40% similarity were considered members of the same gene family . As a result , 9 genes were identified as being unique and 23 other genes had a total of 87 additional gene family members . RiceCyc ( http://pathway . gramene . org/RICE/class-instances ? object=Pathways ) is a web-based tool curated by Gramene ( http://www . gramene . org/ ) and provides biochemical/metabolic pathways . Seven pathways in RiceCyc are associated with 7 mutants identified in this study . U11 is a part of the ammonia assimilation pathway and two genes in this pathway are displayed . U1 is a part of the MEP pathway and eight genes in this pathway are displayed . U10 is a part of the carotenoid biosynthesis pathway and eleven genes in this pathway are displayed . P9-1 is a part of the nitrate assimilation pathway and one gene in this pathway is displayed . P5-1 is a part of the photorespiration pathway and eight genes in this pathway are displayed . U3 is a part of the photosynthetic light reaction pathway by PSI and PSII and fifteen genes in this pathway are displayed . P13-1 is a part of the Vitamin C biosynthesis pathway and eight genes in this pathway are displayed ( Figure 4; Table S4 ) . The abscisic acid ( ABA; there are 3 genes ) , chlorophyll ( 9 genes ) , and vitamin E biosynthetic ( 5 genes ) pathways mediated by precursor or final product of carotenoid biosynthesis are added for co-expression analysis [31] , [82] ( Table S4 ) . Unique sequences or predominantly expressed gene family members were selected for this analysis . Single-step reactions unlinked to these or other pathways were established for the other genes for which mutant phenotypes were identified during this study , i . e . U4 ( Os02g57030 ) , U5 ( Os03g04470 ) , and Ca1 ( P2-1 , Os1g45274 ) ( see , for example , Table 2 ) . The pathways used in this study were developed using Gramene RiceCyc ( http://pathway . gramene . org/RICE/class-instances ? object=Pathways ) . Candidate genes in the pathways having evidence of expression , such as expressed sequence tags ( ESTs ) or full length cDNAs , were available at the above website . Additional candidate genes for which no evidence of gene expression was available were selected based on Arabidopsis best-hit genes homologous to TIGR version 5 gene models . Probable gene family members of all candidate genes were checked against Table S7 which lists all rice gene family members . For the co-expression analysis , we selected 72 genes in 13 biochemical/metabolic pathways or reactions associated with the 10 genes for which we had identified mutant phenotypes in this study ( Figure 4 ) . As a result , 10 gene clusters were identified . By using Cytoscape software , we deduced relationships among the different biochemical pathways or reactions ( Figure S7 ) . To do this , we considered the 10 gene clusters classified by co-expression analysis in Figure 4 as different functional groups . ( The twelve boxes with different colors indicate the different biochemical pathways or reactions in Figure S7 . Of them , the gray-colored box indicates reactions of unknown genes U4 and U5 ) . In addition , co-expressed gene-clusters in Figure 4 were marked with 10 differently colored lines in Figure S7 . Each line indicates an enzyme reaction carried out by the product of one of 72 genes in Figure 4 . These gene products are numbered in the 10 biochemical pathways with different colored rectangular boxes ( Figure S7 ) . Each rectangular box ( node ) indicates chemical compounds in the pathway .
|
Rice , a model monocot , is the first crop plant to have its entire genome sequenced . Although genome-wide transcriptome analysis tools and genome-wide , gene-indexed mutant collections have been generated for rice , the functions of only a handful of rice genes have been revealed thus far . Functional genomics approaches to studying crop plants like rice are much more labor-intensive and difficult in terms of maintaining the plants than when studying Arabidopsis , a model dicot . Here , we describe an efficient method for dissecting gene function in rice and other crop plants . We identified light response-related phenotypes for ten genes , the functions for which were previously unknown in rice . We also carried out co-expression analysis of 72 genes involved in specific biochemical pathways connected in lines carrying mutations in these ten genes . This analysis led to the identification of a novel set of genes likely involved in these pathways . The rapid progress of functional genomics in crops will significantly contribute to overcoming a food crisis in the near future .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"and",
"genomics/genomics",
"genetics",
"and",
"genomics/gene",
"discovery",
"genetics",
"and",
"genomics/gene",
"expression",
"genetics",
"and",
"genomics/functional",
"genomics",
"plant",
"biology",
"plant",
"biology/plant-environment",
"interactions",
"genetics",
"and",
"genomics/gene",
"function",
"plant",
"biology/plant",
"growth",
"and",
"development",
"plant",
"biology/agricultural",
"biotechnology",
"plant",
"biology/plant",
"genetics",
"and",
"gene",
"expression",
"genetics",
"and",
"genomics/plant",
"genetics",
"and",
"gene",
"expression"
] |
2008
|
Identification and Functional Analysis of Light-Responsive Unique Genes and Gene Family Members in Rice
|
Reward-related dopaminergic influences on learning and overt behaviour are well established , but any influence on sensory decision-making is largely unknown . We used functional magnetic resonance imaging ( fMRI ) while participants judged electric somatosensory stimuli on one hand or other , before being rewarded for correct performance at trial end via a visual signal , at one of four anticipated financial levels . Prior to the procedure , participants received either placebo ( saline ) , a dopamine agonist ( levodopa ) , or an antagonist ( haloperidol ) . Principal findings: higher anticipated reward improved tactile decisions . Visually signalled reward reactivated primary somatosensory cortex for the judged hand , more strongly for higher reward . After receiving a higher reward on one trial , somatosensory activations and decisions were enhanced on the next trial . These behavioural and neural effects were all enhanced by levodopa and attenuated by haloperidol , indicating dopaminergic dependency . Dopaminergic reward-related influences extend even to early somatosensory cortex and sensory decision-making .
A role for dopamine in Pavlovian and instrumental learning , as well as in consolidating plastic changes in corticostriatal pathways , is well established [1] , [2] . Although research on reward has focused on learning , there is growing interest in a possible reward-mediated modulation of perception and sensory decision-making [3]–[5] . However , it remains unclear whether effects of reward on human sensory processing are influenced by dopamine . Here , we examined possible dopaminergic modulatory influences on neural activity in human primary somatosensory cortex ( PSC ) and on sensory decisions . We exploited a new somatosensory paradigm for which we recently showed that increased financial rewards not only improve sensory performance , but also modulate PSC at the point of reward delivery , even when the financial reward is presented only visually [6] . To examine any contribution of dopamine to reward modulation of somatosensation , we now combine the sensory decision-making paradigm with concurrent functional magnetic resonance imaging ( fMRI ) ( see Materials and Methods , and Figure 1 ) in the context of both agonist and antagonist dopaminergic pharmacological manipulations . In a placebo-controlled , double-blind , fully randomized design , participants received pills comprising either 100-mg levodopa , 2-mg haloperidol , or placebo ( see Materials and Methods , and Figure S2 ) . Levodopa is well established for increasing brain dopamine levels , as commonly used as therapy for Parkinson disease [7] . Haloperidol is an antidopaminergic drug ( selective D2 receptor antagonist ) , frequently used to treat psychosis [8] .
We found a clear impact of dopaminergic modulation on somatosensory decisions . During scanning , across all reward levels , percentage correct somatosensory judgments comprised 70 . 4% of trials for the placebo group , increased to 76 . 3% for the levodopa group , and reduced to 66 . 4% for the haloperidol group . In terms of the specifics of reward effects , within the present placebo group ( and in accord with our recent nonpharmacological study [6] ) , increased potential reward led to enhanced accuracy of sensory decisions ( Figure 2 , top row; linear parametric effect of reward level F ( 1 , 9 ) = 9 . 99 , p = 0 . 012 ) for judgments about the left or right hand ( no significant main effects or interactions with factor of side , all p>0 . 99 ) . This effect of reward on somatosensory decisions was affected by our pharmacological manipulation ( see Figure 2 , comparing different rows ) , leading to a significant interaction between drug group and reward level ( F ( 2 , 27 ) = 3 . 81 , p = 0 . 035 ) . Again , this outcome did not depend on the hand judged ( no significant main effects or interactions involving side , all p>0 . 5 ) . Planned comparisons for the impact of the different drugs showed that under levodopa ( middle row in Figure 2 ) , overall accuracy was significantly higher than for placebo ( F ( 1 , 18 ) = 5 . 68 , p = 0 . 028 ) , whereas higher reward levels still systematically increased discrimination accuracy ( F ( 1 , 9 ) = 19 . 16 , p = 0 . 002 ) . By contrast , haloperidol ( bottom row in Figure 2 ) not only attenuated the effect of reward level relative to placebo ( F ( 1 , 18 ) = 5 . 32 , p = 0 . 03 ) , but actually eliminated the impact of reward level ( F ( 1 , 9 ) = 0 . 03 , p = 0 . 85 , n . s . ) . Thus , these data show that for sensory decisions involving the left or right hand , agonist and antagonist dopamine manipulations enhance accuracy or reduce reward-related effects on somatosensory discrimination performance , respectively . We next examined the fMRI data acquired concurrently with task performance for all three pharmacological groups , analyzing these with standard approaches ( SPM5 software , see Materials and Methods for details ) . To anticipate , we observed effects of the dopaminergic manipulations on brain activity related to reward and to somatosensory processing that corresponded with the effects on somatosensory decisions reported behaviourally above , and that shed light on the neural mechanisms involved . During the somatosensory discrimination phase of each trial , we found activation of a task-related network of brain areas including PSC and secondary somatosensory cortices/parietal ventral cortex , as well as prefrontal cortex ( PFC ) , supplementary motor area ( SMA ) , premotor cortex ( PMC ) , posterior parietal cortex ( PPC ) , insula , caudate nucleus , and striatum in both hemispheres ( see Text S1 and Table S1 ) . This accords with the involvement of similar areas for related somatosensory tasks in other work [6] , [9] . To identify brain regions where dopamine level specifically influenced reward-related activation , we next focused on blood oxygen level–dependent ( BOLD ) signals during the visual reward outcome presented at trial end ( see Figure 1 ) , in the absence of somatosensory stimulation . When testing for the interaction of group and outcome ( rewarded versus nonrewarded ) at trial end , we observed greater differential BOLD signal in ventral striatum and orbitofrontal cortex ( OFC ) . Both these regions showed a reliable group-by-reward interaction , attributable to an enhancement of the reward-related signal for the levodopa ( increased central dopamine ) group and an attenuation of reward effects for the haloperidol ( D2 receptor antagonist ) group , relative to intermediate reward-related responses seen under placebo ( see Figure 3 , plus Table S2a ) . Thus , BOLD signals in two key regions implicated in reward , ( i . e . , ventral striatum [10] and OFC [11] ) , showed a response profile at reward delivery that clearly depended on dopamine level . The same comparison ( interaction of drug group with reward versus no reward ) also revealed dopamine-related influences on activity within PSC itself ( see Table S2a ) . Note that this reward-dependent somatosensory activation was expressed at a time point corresponding to the delivery of visual rewards at trial end . We confirmed that these activations originated from PSC , by restricting our examination of BOLD signals to primary somatosensory areas BA1 , BA2 , and BA3b [12] , as defined by a computerized atlas based on cytoarchitectonic data [13] ( see Materials and Methods for further details ) . These analyses confirmed a reward effect ( relative to nonrewarded trials ) in PSC , at the time point corresponding to visual reward delivery . Moreover , this somatosensory effect also depended on dopamine level , as manipulated here pharmacologically ( see Figure 4 , plus Table S2a and S2b ) . The involvement of PSC in this impact of reward , at the reward delivery point during a sensory decision task , accords with our recent nonpharmacological findings [6] . That previous study also showed that visually signalled financial rewards can “reactivate” PSC in the context of a somatosensory-discrimination task . This suggests that reward outcome provides a form of teaching signal that may be fed back to task-relevant sensory cortex . The present data now show that the effectiveness of reward in influencing PSC in this way depends on dopamine , as evident in our new demonstration that the impact on somatosensory cortex itself is enhanced under levodopa and attenuated under haloperidol ( see Figure 4 , Table S2a and S2b ) , analogously to the effects we found also for more classic reward-related regions ( see Figure 3 ) . Importantly , these dopaminergic influences on reward effects in somatosensory cortex were expressed specifically in the PSC that was required for the preceding decision that led to the reward . Separate analyses of trials in which the left or right index finger had been judged revealed that only somatosensory cortex contralateral to the currently judged hand was affected by reward delivery and by drug group in this way ( Figure 4 and Table S2b; peak at xyz = −36 , −36 , 60 for left PSC when judging the right hand; and at 36 , −30 , 48 for right PSC when judging the left hand ) . This underlines that the dopaminergic reward influences were indeed specifically expressed only in the portions of PSC that were relevant for correct performance of the preceding task . Figure 5 plots the percent signal changes at reward delivery for reward minus nonreward trials , extracted from independently defined regions of interest ( ROIs , see [14] , Materials and Methods , and Discussion ) contralateral to the rewarded index finger , for each drug group ( separate rows in Figure 5 ) . This reveals that our pharmacological manipulation of dopamine level influenced BOLD responses in PSC ROIs specifically as a function of the different financial levels achieved on reward trials ( significant interaction between drug group and parametric reward-level; F ( 2 , 27 ) = 10 . 27 , p<0 . 001 ) . Under placebo ( top row in Figure 5 ) , reactivation of contralateral PSC by visual reward feedback increased systematically with financial magnitude ( F ( 1 , 9 ) = 14 . 34 , p = 0 . 004 ) . Such an increase was also found under levodopa ( F ( 1 , 9 ) = 15 . 94 , p = 0 . 003 ) , with a trend towards a steeper slope than under placebo ( F ( 1 , 18 ) = 9 . 53 , p = 0 . 07 ) . Haloperidol , by contrast ( see bottom row in Figure 5 ) , completely eliminated the impact of a parametric reward level on PSC at reward delivery ( F ( 1 , 9 ) = 0 . 88 , p = 0 . 37 ) , with this flat function differing significantly from the linear increase under placebo in a direct comparison ( F ( 1 , 18 ) = 14 . 68 , p = 0 . 001 ) . All of these influences of reward level during visual reward delivery upon PSC were specific to the positive trials in which financial reward was delivered , with no effect of financial reward level on somatosensory cortex being found for nonrewarded trials instead , for all three groups here ( all p>0 . 2 ) . Thus , these effects are indeed due to the actual receipt of reward , rather than just general feedback on task performance . Our somatosensory decision task allows assessment of whether ( higher ) reward delivery on a given trial can enhance behavioural decisions and related PSC activity on the next trial [6] . Such trial-to-trial effects of reward delivery could explain why receiving higher rewards leads to better performance overall . Accordingly , we examined how dopamine may affect such trial-to-trial reward effects on somatosensory discrimination . We recently reported that the conditional probability of the next trial being correct after receiving a reward on the preceding trial is enhanced for higher rewards [6] . We now show that this behavioural effect is strongly modulated by dopamine , as manipulated here pharmacologically ( interaction of reward level and the three drug groups , F ( 2 , 27 ) = 7 . 6 , p = 0 . 002 ) . Under placebo ( see green line in Figure 6A ) , the findings confirm our recent nonpharmacological study [6] . The beneficial impact of receiving reward on a given trial ( n−1 ) for accurate performance on the subsequent trial ( n ) is stronger for higher reward levels ( F ( 1 , 9 ) = 14 . 62 , p = 0 . 004 ) . This reward level–dependent trial-to-trial effect was even more pronounced ( F ( 1 , 9 ) = 35 . 02 , p<0 . 001 ) under levodopa ( see blue line in Figure 6A ) , with a significantly steeper slope against the reward-level factor than for placebo ( F ( 1 , 18 ) = 7 . 49 , p = 0 . 014 ) . Levodopa also enhanced the overall trial-to-trial effect ( pooled over reward level ) relative to placebo ( F ( 1 , 18 ) = 4 . 68 , p = 0 . 044 ) . For haloperidol , by contrast ( see red line in Figure 6A ) , the parametric increase in trial-to-trial performance ( as a function of reward level obtained on the preceding trial ) was completely eliminated ( F ( 1 , 9 ) = 0 . 102 , p = 0 . 757 ) and hence reduced relative to the placebo group ( F ( 1 , 18 ) = 3 . 16 , p = 0 . 09 ) . Haloperidol likewise reduced the trial-to-trial effects relative to placebo when pooling across reward levels ( F ( 1 , 18 ) = 4 . 109 , p = 0 . 05 ) . This aspect of our behavioural findings thus establishes dopamine-dependence for the enhancing effect of receiving a ( higher ) reward on the previous trial upon sensory decisions for the next trial , with this enhancement being even more pronounced under levodopa , but eliminated by haloperidol . Our final results confirm that such a dopamine-related trial-to-trial effect of reward level was not only present for sensory performance , but also impacted on the BOLD response of PSC during somatosensory discrimination for the next trial ( interaction between drug group and parametric reward level , F ( 2 , 27 ) = 4 . 48 , p = 0 . 021 , see Figure 6B ) . Using the same independently defined ROIs for PSC as previously ( see Figure 4 , and Materials and Methods for discussion on ROI selection ) , we found BOLD signal increases in PSC contralateral to the judged hand , now during the somatosensory stimulation/discrimination phase of a given trial , if a higher reward had actually been received on the previous trial . In line with our predictions [6] , these trial-to-trial enhancements of PSC BOLD response by the level of reward actually received on the previous trial were present under placebo ( F ( 1 , 9 ) = 11 . 79 , p = 0 . 007 ) . Our new pharmacological manipulation revealed that these reward-dependent trial-to-trial enhancements of PSC were even more pronounced under levodopa ( F ( 1 , 9 ) = 23 . 58 , p = 0 . 001; F ( 1 , 18 ) = 3 . 16 , p = 0 . 09 in direct comparison with placebo ) , but were completely abolished under haloperidol ( F ( 1 , 9 ) = 0 . 003 , n . s . ; F ( 1 , 18 ) = 5 . 53 , p = 0 . 03 in comparison with placebo ) . This pattern of results shows that pharmacologically manipulated dopamine level modulates the impact of reward for a given trial upon sensory performance ( Figure 6A ) and the response of PSC ( Figure 6B ) for the subsequent trial . Figure 7 plots the BOLD responses in ventral striatum for each financial reward level ( OFC showed comparable signal changes for these comparisons ) . At the point of reward delivery ( see Figure 7A ) , the ventral striatum showed a significant reward-by-drug interaction ( F ( 2 , 27 ) = 5 . 59 , p = 0 . 009 , for ventral striatum , see Figure 7A; and F ( 2 , 27 ) = 5 . 26 , p = 0 . 012 , for OFC ) . At this time point , BOLD responses under levodopa showed an impact of reward versus nonreward for ventral striatum ( F ( 1 , 9 ) = 5 . 99 , p = 0 . 03 ) that was enhanced relative to placebo ( F ( 1 , 18 ) = 5 . 57 , p = 0 . 03 ) and likewise for OFC ( F ( 1 , 9 ) = 7 . 49 , p = 0 . 02; levodopa vs . placebo: F ( 1 , 18 ) = 3 . 51 , p = 0 . 07 ) . However , the effects of reward delivery on ventral striatum and OFC differed from those observed in PSC ( cf . Figure 5 ) . In the placebo group , neither striatum ( F ( 1 , 9 ) = 0 . 01 , p = 0 . 91 ) nor OFC ( F ( 1 , 9 ) = 3 . 6 , p = 0 . 09 ) showed a significant increase in BOLD response with rising reward level at reward delivery point ( consistent with [6] ) . Instead , the impact of reward level on striatum and OFC at reward delivery was only significant under levodopa ( striatum: F ( 1 , 9 ) = 5 . 99 , p = 0 . 03; OFC: F ( 1 , 9 ) = 7 . 49 , p = 0 . 02 ) . We found no parametric reward-level effect under haloperidol , neither in the striatum ( F ( 1 , 9 ) = 0 . 59 , p = 0 . 46; haloperidol vs . placebo: F ( 1 , 18 ) = 0 . 002 , p = 0 . 96 ) nor in OFC ( F ( 1 , 9 ) = 0 . 01 , p = 0 . 9; haloperidol vs . placebo: F ( 1 , 18 ) = 3 . 6 , p = 0 . 07 ) at reward delivery . During the earlier stimulation/discrimination phase , both these reward-related regions showed an interaction of drug and reward level ( striatum: F ( 2 , 27 ) = 3 . 98 , p = 0 . 03; OFC: F ( 2 , 27 ) = 3 . 42 , p = 0 . 04 , for correct minus incorrect trials , see Figure 7B ) . For this relatively early point in the trial , BOLD responses under placebo replicated our recent nonpharmacological study [6] in showing a monotonic effect of increased anticipated reward level for ventral striatum ( F ( 1 , 9 ) = 5 . 12 , p = 0 . 05 ) and OFC ( F ( 1 , 9 ) = 7 . 38 , p = 0 . 02 ) , in advance of actual reward delivery . This pattern was also found under levodopa ( striatum: F ( 1 , 9 ) = 7 . 03 , p = 0 . 02; OFC: F ( 1 , 9 ) = 6 . 03 , p = 0 . 03 ) , but it was attenuated , and indeed eliminated , under haloperidol ( striatum: F ( 1 , 9 ) = 0 . 1 , p = 0 . 75; placebo vs . haloperidol: F ( 1 , 18 ) = 4 . 75 , p = 0 . 04; OFC: F ( 1 , 9 ) = 0 . 02 , p = 0 . 87; placebo vs . haloperidol: F ( 1 , 18 ) = 6 . 55 , p = 0 . 02 ) . Our data indicate that ventral striatum and OFC , both classic reward-related regions , also show a pattern of dopaminergic reward-related modulation , but this pattern differed from that seen in PSC . During the reward delivery phase , an influence of reward level was observed in ventral striatum and OFC only under levodopa ( not during placebo , as in PSC ) . However , reward level affected BOLD signal in ventral striatum and OFC during the earlier stimulation/discrimination phase ( where no effects was seen in PSC ) . Thus , the classic reward-related areas in ventral striatum and OFC mostly showed anticipatory effects of reward level at the earlier stimulation/discrimination point , whereas PSC was only influenced significantly by reward level at the later reward delivery phase ( see above and Figures 4 and 5 ) . Nonetheless , all of these effects were typically enhanced by levodopa but eliminated by haloperidol .
How the brain harnesses reward-related information to control a wide range of overt behaviours [15] , [16] is a central topic in decision neuroscience . Much recent discussion concerns the likely dopaminergic mediation of such effects [1] . An emerging new question is whether reward may influence early sensory processing [6] , [17] , [18] , and if so , whether these influences are dopaminergically mediated . Here , we establish a dopamine dependence for reward-based influences on human PSC in a sensory decision-making task . We found that participants pretreated with levodopa showed increased reward-related effects for both tactile decisions and for hemodynamic responses ( or “reactivation” ) in PSC at the point of reward delivery ( see Figures 4 and 5 ) . Haloperidol , in contrast , eliminated these influences of reward on somatosensory performance and on processing in PSC . This demonstrates that dopaminergic neural processes , enhanced by levodopa but attenuated by haloperidol , are involved in modulating the impact of reward upon activity and function for primary somatosensation . There was no parametric effect of reward level on PSC at the time point corresponding to the earlier discrimination phase for correct minus incorrect trials ( see Figure S1 , and Text S1 ) . Instead , the effect of financial reward level was expressed only at the end of the trial , timelocked to positive reward delivery via visual feedback ( see Materials and Methods for how these different phases of the trial were separated ) . This confirms that the effects on PSC ( as seen in Figures 4 and 5 ) must reflect feedback due to reward receipt , rather than modulation of sensory processing during the stimulation , as observed during attention [19] . Dopamine is a key neurotransmitter implicated in incentive motivation [20] , memory formation [21] , [22] , and reinforcement learning [23] , [24] . Furthermore , dopamine release optimises response selection in skilled nonautomatic tasks [25] and improves cognitive function by enhancing information processing [26] , [27] and attentional accuracy [28]–[30] , possibly via suppression of background noise and enhancement of task-related signals [31] . However , there seems to be a trade-off between dopamine levels and performance since individuals with pathologically increased dopamine activity ( i . e . , schizophrenia ) have reduced function of attentional and sensorimotor systems , and administration of strong dopamine antagonists in these patients ameliorates their deficits [32] . In humans , dopamine-mediated reward effects are well established in midbrain , ventral striatum , and OFC; these structures represent key components of the human reward system [10] , [11] and were activated here . However , reward-sensitive areas are tightly interconnected with other cortical regions , including via thalamocortical loops [18] , suggesting that a complex network of dopaminergic projections [33] can also affect processing in other brain areas , such as PSC [34] , [35] . It is tempting to speculate that this interconnected architecture provides the basis for a pervasive influence of reward on a wide range of cognitive processes , and our present results appear consistent with this perspective , extending the range of influences to include PSC and sensory decisions . When inspecting our data for effects of reward level on ventral striatum and OFC , we found dopamine-related anticipatory effects of monetary incentive arising early in the trial , during sensory processing and prior to reward delivery ( Figure 7B ) . Note that these effects arose much earlier than a later effect on PSC , expressed solely when a reward was actually received ( see Figure 5 ) . The finding that ventral striatum and OFC were affected by reward level more in an anticipatory than outcome-related fashion may appear at odds with a number of studies showing reward outcome effects in these structures ( e . g . , [36] , [37] ) . On the other hand , our findings are compatible with several other studies showing anticipatory reward effects in ventral striatum and OFC [38]–[41] . One possible explanation for our specific results may be that we used relatively low levels of reward on each trial ( i . e . , 0 , 20 , 50 , or 80 pennies per correct trial ) , since other studies showing outcome-related effects in OFC and ventral striatum often used much larger amounts of monetary reward ( e . g . , 0 to $10 as in [42] ) . Furthermore , in our design , reward level was always explicitly signalled by a visual cue indicating the potential reward for the next blocked series of trials . The ventral striatum is known to encode not only reward prediction , like in our study during stimulation/discrimination period , see also [43] , but also reward prediction errors , which reflect a difference between predicted and received reward level during feedback/outcome ( see also [1] , [2] , [44] , [45] ) . Thus , the explicit predictability of reward value in our task ( signalled blockwise via a visual cue ) may explain an absent reward effect at the outcome/feedback point in ventral striatum . Taken together , our findings suggest that when reward outcome depends on a veridical sensory decision , reward signals that arise in putative reward regions ( such as ventral striatum during stimulation/discrimination period here ) can be propagated to early sensory systems that are critical for sensory judgements ( in this case , the PSC ) . These reward-related modulations reflect the magnitude of reward actually received , and may thus provide a possible dopaminergic “teaching signal” based upon reward delivery . This suggests that dopamine-related interplay between striatum , OFC , and sensory cortex may allow incentive motivation and feedback to shape cortical responses [23] , in line with the recent finding [46] that corticostriatal interactions during processing of incentive stimuli covary with the COMT val158met polymorphism , which is linked to higher synaptic dopamine levels . Our present fMRI findings clearly establish that dopamine levels can affect reward-related influences on PSC . Future invasive neurophysiological studies in animals may shed further light on the fine-grained neural mechanisms and circuits involved in reward-related dopaminergic modulation of PSC function . Some aspects of our results already provide an initial step towards a mechanistic account for how reward can impact on somatosensory discrimination performance . Notably , we found that the “reactivation” in PSC by reward delivery at trial end influenced both performance and evoked somatosensory responses for the next trial ( Figure 6; see also [6] ) . An important new finding here is that this trial-to-trial effect of reward outcome was also mediated by dopaminergic transmission , being enhanced by levodopa and abolished by haloperidol . These modulatory trial-to-trial effects of dopamine on somatosensory performance and cortical processing specifically depended on the financial level of reward received , and thus did not simply indicate some form of general “resetting” for the next trial [47] . Instead , our results suggest that these effects reflect a dopamine-mediated learning signal [48] , fed back to task-specific primary sensory cortex [6] , [17] , that enhances the response of somatosensory cortex and somatosensory performance for the next trial , thereby leading to enhanced outcomes , and to the improvement in sensory decisions under higher rewards . Our findings show that dopamine mediates a reward influence on early human sensory cortex in a sensory decision-making task . Recent invasive studies in rats [17] , [18] , and monkeys [4] , [49] , [50] had begun to incorporate reward considerations into mechanistic accounts for motor choice , and increasingly for perceptual decisions [5] . The present human study indicates that even basic sensory discriminations and the function of early sensory structures ( here , PSC ) are influenced by dopaminergic transmission [7] . Thus , dopamine-dependent reward signals arising in classic reward-related structures appear to be propagated back to early somatosensory cortex so as to shape basic sensory discrimination , leading to enhanced reward outcome . This raises the tantalising possibility that specific pharmacological manipulations ( e . g . , those affecting dopaminergic systems ) might modulate reward-related brain processes for possible neuro-rehabilitation of sensory processing .
Participants first practiced the somatosensory frequency-discrimination task in an initial session inside the scanner , but without functional images being collected . This practice session had the same length as the subsequent experiment , but we presented only 0-pence trials to avoid habituation to reward magnitudes . Participants were then removed from the scanner , and drugs were administrated in a placebo-controlled , double-blind , fully randomized design . Since levodopa reaches peak plasma concentration within 1 h after intake , whereas haloperidol peaks 3 h later , we followed a recently described method [16] to ensure that peak plasma concentration of both drugs coincided with fMRI ( see Figure S2 ) . Participant always received two pills; the first immediately after the practise session , and the second 3 h later . The main experiment involving scanning started 1 h after the participant received the second pill . If a participant was assigned to the placebo group , both pills contained placebo . In the levodopa group , the first pill contained placebo , the second 100 mg of levodopa . Accordingly , latency between levodopa administration and main experiment was 1 h , which is the time a single dose of 100 mg needs to reach peak plasma concentration [16] . In the haloperidol group , the first pill contained 2-mg haloperidol , the second pill placebo . Thus , latency between haloperidol intake and main experiment was 4 h , in which time haloperidol is known to reach peak plasma concentration [16] . This drug administration schedule thus ensured that the peak plasma concentration of both drugs was matched across participants , without the necessity for further pharmacokinetic characterisation . Thirty right-handed healthy participants gave written informed consent in accord with local ethics . Ten participants ( seven male ) were included in each group in a fully randomized , double-blind fashion ( placebo: aged between 21 and 35 y , mean 27±5 . 3 y; dopamine: aged between 19 and 31 y , mean 26±3 . 5 y; haloperidol: aged between 20 and 33 y , mean 27±4 . 5 y ) . All participants were European students . All females took contraceptives and were not scanned during menses . All participants were first interviewed and examined by an experienced physician ( B . P . ) to exclude any psychiatric/neurological symptoms and history of significant drug use . We used a 3T head-scanner ( Magnetom Allegra; Siemens ) to acquire functional and structural brain scans . For functional brain scans , we used a BOLD-sensitive gradient echo T2* weighted echo-planar imaging ( EPI ) sequence ( TE = 30 ms , TR = 2 . 21 s , flip angle = 90° , in-plane resolution = 3×3 mm2 , slice-thickness = 2 mm , interslice distance = 1 mm ) optimized for fMRI studies of the orbitofrontal cortex ( for further information , see [51] ) . One MRI scan ( or volume ) consisted of 34 oblique slices ( transversal-coronal tilt: −10° ) covering the whole cerebrum . During each fMRI session we acquired 875 volumes continuously . After drug administration , volunteers underwent an fMRI experiment in which they repeatedly discriminated the frequency of two electrical stimuli , applied sequentially to the index finger ( both index fingers were in fact stimulated twice in succession on each trial , but only one hand or the other was judged ) . Participants experienced the stimulation as a prickling and tingling sensation , and reported that their decision was based on comparing the speed or rhythm of the two stimuli . With each trial , participants first perceive a stimulus , hold it in working memory , and finally make a decision by comparing it with a second stimulus ( see also [9] , [52] ) . For a detailed description of the fMRI design and the stimuli used; see [6] . Participants signalled their judgment via a foot response , and received visual feedback indicating positive or negative reward outcome ( for correct or incorrect trials , respectively ) after a variable temporal delay ( see Figure 1 ) . This temporal separation , and other standard aspects of event-related fMRI ( e . g . , [53] ) , allowed separation of BOLD signals attributable to somatosensory encoding from those due to the subsequent visual reward outcome ( see Materials and Methods , and also [6] for in-depth discussion ) . Thus , any somatosensory reactivations due to the visual reward delivery must reflect reward-related signals , not the initial processing or level of attention during somatosensory input . We examined the influence of dopaminergic manipulations on reward-related processes at four different monetary reward levels ( 0 , 20 , 50 , or 80 pennies per correct trial ) . These reward levels were organised into miniblocks of four successive trials ( see Figure 1 ) . The onset of each miniblock was signalled by a distinct visual cue indicating the potential reward for each of the next four trials , and also whether the right or left index finger should be judged for all those trials . Thus , the participant knew both the financial stake and which hand to judge in advance of each miniblock . Apart from this miniblock structure , levels of rewards were randomly intermingled , as was judged side . Our design enabled us to examine dopaminergic dependence of reward-related influences on somatosensory judgments and on related brain activity , both for overall effects of reward ( regardless of financial level ) , as well as ( orthogonally ) for the parametric impact of increased potential rewards ( i . e . , 0 , 20 , 50 , or 80 pence per correct judgment ) . For a high-resolution structural brain scan , which was acquired after the functional MRI session , we used an isotropic 3D spoiled gradient-recalled ( SPGR ) sequence with 107 sagittal-orientated slices covering the whole brain . The anatomical images across participants were used to calculate a mean group image . For initial spatial assignment of functional changes , parametric maps showing the group statistics were superimposed onto this mean structural image . We used SPM5 software ( http://www . fil . ion . ucl . ac . uk/spm/ ) to assess event-related BOLD responses [53] . During the first six volumes per session , BOLD signal reached steady state . These volumes were discarded from further analysis . The remaining 869 volumes entered realignment and unwarping to remove movement artefacts [54] . Volumes were then spatially normalized to the standard template of the Montreal Neurological Institute [55] . As for our recent nonpharmacological study [6] , we smoothed volumes using a 10-mm ( full-width half-maximum ) isotropic , three-dimensional Gaussian filter , in accord with the standard SPM approach . To assess reliability of effects across participants , we used random-effects SPM analysis . We report all brain regions that survived family-wise error ( FWE ) -corrected thresholds . We further assessed whether particular hemodynamic changes could specifically be attributed to PSC , by restricting the analysis to PSC in both brain hemispheres . For this , we used a cytoarchitechtonic computerized anatomical atlas [13] ( see http://www . fz-juelich . de/inb/inb-3//spm_anatomy_toolbox ) to create masks according to the broad definition of PSC as encompassing BA1 , BA2 , and BA3b , based on separate postmortem data [12] , [13] , [56] . This ROI definition by means of anatomy prevented any selection bias and hence potential artificial inflation of our ROI statistics [14] . We identified effects attributable to distinct events using distinct stick functions ( convolved with the default HRF in SPM ) . These stick functions encoded the timing of tactile stimulation , or of later reward feedback for each trial . We also used a stick function timelocked to the actual pedal response on each trial , and a further stick function timelocked to the visual cue at the start of each miniblock . This meant that all event types were coded as distinct events , except that the successive pair of somatosensory stimuli on each trial was coded as a single composite event since they were not jittered relative to each other ( see Figure 1 ) . We further distinguished between event types depending on the rewarded side ( right or left ) , reward magnitude , and whether the judgment was correct or not ( rewarded vs . nonreward trials ) . Null trials provided an implicit baseline . To consider general half-life issues of the drugs ( i . e . , haloperidol and levodopa ) , we followed a recently described procedure ( see above , and [16] ) that allowed us to use a statistical model for our fMRI data without taking pharmacokinetic characterization of the different drugs into account . It was important for our experimental design to distinguish brain activity attributable to somatosensory stimulation/discrimination , from that due to later visual feedback signalling reward presentation . Unlike previous event-related fMRI studies on reward effects , which were not tailored to distinguish reward anticipation from sensation effects ( see , e . g . , [41] , [57] ) , trial phases in our experiment could be separated due to a combination of a jittered timing ( 3–5 s intervening; 6–8 s from first somatosensory input ) plus the fact that not all trials were rewarded ( only correct ) and reward could reflect four different monetary levels . The critical regressors for our analysis were demonstrably uncorrelated and therefore independent/orthogonal ( see also [6] ) . The actual correlations between critical regressors across all three groups ( levodopa , haloperidol , and placebo ) were as follows: for correct discrimination versus reward feedback: r = −0 . 02; and for incorrect discriminations versus no-reward feedback: r = −0 . 08 ( see also [6] ) . These vanishingly small ( insignificant , null ) correlations allowed us to separate discrimination-related versus feedback-related activity changes with a standard event-related SPM analysis ( see also [58]–[60] for similar use of standard methods for decorrelating regressors in fMRI analyses ) . Given other recent results from pharmacological fMRI involving levodopa and haloperidol [16] , and in accord with their established impact on central dopamine action , we expected reduced reward-related effects ( if dopaminergic ) in the haloperidol group , and enhanced reward effects in the levodopa group , relative to placebo . Accordingly , on the between-group level , we coded the groups as three successive steps ( haloperidol , then placebo , then levodopa ) for parametric contrasts in the general linear model ( i . e . , weighting them as “1 , ” “2 , ” “3” ) . Nonreward trials were equally coded for all three groups ( weighted as “−2” ) . For further analysis of trial-to-trial effects , in terms of whether performance was rewarded or not on the preceding trial at a particular monetary level , we had to eliminate the last trial from each miniblock from consideration of possible effects on the next trial , as a different miniblock instruction intervened . Hypothesis-driven ROI analyses [6] , [61] were implemented using 5-mm spheres centred the peak coordinates for the categorical reward versus nonreward feedback effect in PSC , contralateral to the judged side ( i . e . , at −36 , −36 , 60 for the right and at 36 , −30 , 48 for the left index finger , as shown in Figure 4 and Table S2b ) . Note that these ROIs were fully unbiased , being derived from a categorical contrast orthogonal to ( and hence independent of ) any parametric effect related to monetary level of reward . In the context of our fully balanced factorial design , this regressor orthogonality ensures unbiased ROI statistics [14] . We averaged the signal within these spheres and submitted the values to conventional tests for significance across participants . Note that tests for any one particular factor were applied to ROIs whose location had been defined orthogonally by an independent contrast , to avoid circularity or bias in ROI selection .
|
The rewards one receives during decision-making has a profound impact on learning . Much recent interest has focused on the role of the neurotransmitter dopamine in the basal ganglia for influencing learning and behaviour . Here , we ask whether reward can influence low-level sensory processing , for instance in primary sensory cortex , and how dopamine mediates this process . We show in humans that dopamine level , as manipulated with a dopamine agonist and antagonist in a double-blind placebo-controlled design , is involved in reward modulation of primary somatosensory cortex . Higher anticipated reward improved tactile decisions , and receipt of visual reward signals reactivated primary somatosensory cortex for the judged hand as measured using functional neuroimaging . After receiving a higher reward on one trial , somatosensory activations and decisions were enhanced on the next trial , suggesting that reward outcome provides a form of teaching signal that may be fed back to task-relevant sensory cortex . All these behavioural and neural effects of reward on somatosensory decision-making were strongly modulated by the availability of dopamine as the mediating neurotransmitter . These findings raise the tantalising new possibility that reward manipulations in conjunction with dopaminergic drugs might be used to enhance pathologically deficient or lapsed sensory processes , analogous to how rewards can be used to shape or correct behaviour .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"neuroscience/behavioral",
"neuroscience",
"neuroscience/cognitive",
"neuroscience",
"neuroscience/sensory",
"systems"
] |
2009
|
Influence of Dopaminergically Mediated Reward on Somatosensory Decision-Making
|
Handheld light microscopy using compact optics and mobile phones may improve the quality of health care in resource-constrained settings by enabling access to prompt and accurate diagnosis . Laboratory technicians were trained to operate two handheld diagnostic devices ( Newton Nm1 microscope and a clip-on version of the mobile phone-based CellScope ) . The accuracy of these devices was compared to conventional light microscopy for the diagnosis of Schistosoma haematobium , S . mansoni , and intestinal protozoa infection in a community-based survey in rural Côte d’Ivoire . One slide of 10 ml filtered urine and a single Kato-Katz thick smear from 226 individuals were subjected to the Newton Nm1 microscope and CellScope for detection of Schistosoma eggs and compared to conventional microscopy . Additionally , 121 sodium acetate-acetic acid-formalin ( SAF ) -fixed stool samples were examined by the Newton Nm1 microscope and compared to conventional microscopy for the diagnosis of intestinal protozoa . The prevalence of S . haematobium , S . mansoni , Giardia intestinalis , and Entamoeba histolytica/E . dispar , as determined by conventional microscopy , was 39 . 8% , 5 . 3% , 20 . 7% , and 4 . 9% , respectively . The Newton Nm1 microscope had diagnostic sensitivities for S . mansoni and S . haematobium infection of 91 . 7% ( 95% confidence interval ( CI ) 59 . 8–99 . 6% ) and 81 . 1% ( 95% CI 71 . 2–88 . 3% ) , respectively , and specificities of 99 . 5% ( 95% CI 97 . 0–100% ) and 97 . 1% ( 95% CI 92 . 2–99 . 1% ) , respectively . The CellScope demonstrated sensitivities for S . mansoni and S . haematobium of 50 . 0% ( 95% CI 25 . 4–74 . 6% ) and 35 . 6% ( 95% CI 25 . 9–46 . 4% ) , respectively , and specificities of 99 . 5% ( 95% CI 97 . 0–100% ) and 100% ( 95% CI 86 . 7–100% ) , respectively . For G . intestinalis and E . histolytica/E . dispar , the Newton Nm1 microscope had sensitivity of 84 . 0% ( 95% CI 63 . 1–94 . 7% ) and 83 . 3% ( 95% CI 36 . 5–99 . 1% ) , respectively , and 100% specificity . Handheld diagnostic devices can be employed in community-based surveys in resource-constrained settings after minimal training of laboratory technicians to diagnose intestinal parasites .
Neglected tropical diseases have considerable detrimental impacts in resource-constrained settings as they can result in chronic disability and stigmatization , and have profound negative economic consequences [1 , 2] . Microscopy is an essential tool in the diagnosis and surveillance of many neglected tropical diseases and is a vital component in virtually every clinical and public health laboratory worldwide . Unfortunately even basic microscopy facilities are lacking in many resource-constrained settings where the greatest needs exist , and where neglected tropical diseases are rife [3] . Schistosomiasis is a neglected tropical disease and an important public health threat , with countries in sub-Saharan Africa affected most [4] . Chronic infection with Schistosoma mansoni may result in disability and death due to complications of portal hypertension , while chronic infection with S . haematobium frequently results in genitourinary morbidity and mortality , with bladder cancer as a well-known complication [5 , 6] . Intestinal protozoa , such as Entamoeba histolytica and Giardia intestinalis , are common pathogens accounting for widespread morbidity and mortality in resource-constrained settings . For example , E . histolytica is responsible for an estimated 40 , 000 to 100 , 000 deaths annually . G . intestinalis , a common cause of diarrheal illness , has an estimated prevalence of 20–30% in low-income countries [7 , 8] . These infections are diagnosed primarily by stool microscopy ( or urine microscopy in the case of S . haematobium ) . Recently , portable handheld microscopes [9–11] and mobile phone-based microscopes [12–14] have been evaluated for the diagnosis of neglected tropical diseases ( e . g . , schistosomiasis , opisthorchiasis , and soil-transmitted helminthiasis ) and malaria . Most of the prior studies evaluating handheld and mobile phone-based microscopy have utilized expert microscopists in field settings , or were implemented under laboratory conditions . Prior to wide-scale utilization , these devices must be validated in real-world clinical and public health settings , and operated by individuals who will use them in routine daily practice . Here , we integrate a handheld light microscope ( i . e . , Newton Nm1 microscope; Newton Microscopes; Cambridge , United Kingdom ) [15] and a handheld mobile phone-based microscope ( i . e . , clip-on version of the reversed-lens CellScope ) [16] for the diagnosis of S . mansoni , S . haematobium , and intestinal protozoa into a community-based survey in rural Côte d’Ivoire . These devices were chosen because of their compact design , ease of use , and sufficient resolution to detect intestinal and urogenital parasites . We assessed the accuracy of these devices by comparing them to routine microscopy .
This study was embedded into a larger , cross-sectional , community-based survey in Côte d’Ivoire . Ethical approval was granted by the Ministry of Health and Public Hygiene of Côte d’Ivoire ( reference no . , 32/MSLS/CNER-dkn ) . Written informed consent was obtained from adults aged 18 years or older , and parents or legal guardians on behalf of children . Children , in addition , assented orally . Anthelmintic treatment was offered to all participants at the end of the study ( i . e . , praziquantel , 40 mg/kg of body weight for schistosomiasis and albendazole , 400 mg for soil-transmitted helminthiasis ) . This cross-sectional study was conducted in the village of Grand Moutcho in southern Côte d’Ivoire ( geographic coordinates: 4 . 181 N latitude and 5 . 961 E longitude ) . The village belongs to a region that is highly endemic for schistosomiasis [17] . The study was carried out between April and June 2014 . Study participants were between 6 and 19 years of age . Early morning stool and urine samples , collected between 10:00 and 12:00 hours [18] , were processed and evaluated on the spot in a community clinic . Fresh stool samples were processed with the Kato-Katz technique [19] . In brief , standard 41 . 7 mg thick smears were placed on microscope slides for evaluation of S . mansoni and soil-transmitted helminth eggs . In addition , approximately 2 g of unprocessed stool from each individual was fixed in a standard solution of sodium acetate-acetic acid-formalin ( SAF ) for subsequent laboratory processing and diagnosis of intestinal protozoa infections . Urine samples were first shaken , then 10 ml was extracted and pressed through a 13 mm diameter meshed filter with 20 μm pores ( Sefar AG; Heiden , Switzerland ) . One drop of Lugol’s iodine solution was placed over the filter prior to examination . We selected one Kato-Katz thick smear slide and one filtered urine slide from each individual on their first day of participation in the study . Each slide was subjected to three microscope techniques shortly after collection , and eggs were identified and quantified at the field site . All microscopists were blinded to prior diagnoses on each slide . Slides were first evaluated by ‘gold’ standard microscopy , with an Olympus CX21 microscope under 10x and 40x lenses ( Olympus; Volketswil , Switzerland ) . Laboratory technicians read slides , and 10% of all slides were re-examined by a senior expert microscopist ( JTC , IIB ) blinded to prior results for quality control and validation . Each slide was subsequently examined by two experimental microscopes; the Newton Nm1 Portable Field Microscope and the mobile phone-mounted reversed-lens CellScope ( Fig 1 ) . The Newton Nm1 microscope is a handheld , commercially available device , weighing 480 g with modular objective lenses ( 10x , 40x , and 100x ) , and has been described in field use elsewhere [11 , 13] . The reversed-lens CellScope fits a 3D printed plastic attachment weighing 5 . 2 g over an iPhone 5s ( Apple; Cupertino , California , United States of America ) , with an embedded lens superimposed over the iPhone lens . This device harnesses the mobile phone’s light source to illuminate a specimen [14 , 16] . Laboratory technicians were provided with a half-day of training with direction on the operation of each microscope prior to initiating the study . This training consisted of didactic teaching sessions followed by supervised , hands-on training with multiple test slides . Briefly , the Newton Nm1 is operated by placing a slide in an XY translation stage mounted above the objective , focusing the objective on the sample , and then scanning the sample as it is viewed through the eyepiece of the microscope . The reversed-lens CellScope is operated by holding the mobile phone microscope above the sample and manually moving the device above the sample at the same time as maintaining focus and viewing the images on the screen . Since the slide evaluation was performed live using the screen , which displays images with a lower resolution than still photographs that capture the full resolution of the microscope , the effective resolution of the device used in this study was 14 μm . One month after completion of the field study , SAF-fixed stool samples were subjected to an ether-concentration method , performed in the laboratory of the Centre Suisse de Recherches Scientifiques en Côte d’Ivoire near Abidjan , using a standard protocol [20] . In brief , the SAF-fixed stool samples were re-suspended and placed into a centrifuge tube and centrifuged for 1 min at 500 g . The supernatant was discarded and 7 ml of 0 . 85% NaCl plus 3 ml of ether were added to the remaining pellet . After shaking for 30 sec , the tube and its content were centrifuged for 5 min at 500 g . Finally , from the four layers formed , the three top layers were discarded . The bottom layer ( including sediment ) was placed on a microscope slide . One slide from each participant was created . Slides were examined by microscopy with an Olympus CX21 microscope ( Olympus; Volketswil , Switzerland ) by the same laboratory technicians for the presence or absence of intestinal protozoa , with 10% of the slides re-examined by an expert microscopist ( JTC ) for quality control . Identification of the presence or absence of intestinal protozoa was recorded . Laboratory technicians blinded to earlier results re-examined each slide with the Newton Nm1 microscope and recorded the presence or absence of intestinal protozoa . The clip-on version of the reversed-lens CellScope was not used for intestinal protozoa evaluation in the current study given its effective resolution of 14 μm , as described above . Data were double entered into an Excel spreadsheet , transferred into EpiInfo version 3 . 2 ( Centers for Disease Control and Prevention; Atlanta , Georgia , United States of America ) and cross-checked . All analyses were conducted using R ( R Foundation for Statistical Computing; Vienna , Austria ) . Prevalences were expressed as proportion and we calculated sensitivity , specificity , positive predictive value ( PPV ) , and negative predictive value ( NPV ) of the experimental microscopes for each parasite , using conventional light microscopy as ‘gold’ standard . Using logistic regression models , we determined the sensitivity of the experimental microscopes for detecting any eggs , as a function of the egg count determined by conventional microscopy . Linear association of egg count estimates was assessed by Pearson’s correlation coefficient .
One slide of filtered urine and a single Kato-Katz thick smear were examined from 226 individuals by conventional microscopy and the two experimental microscopes . Conventional microscopy identified 12 positive Kato-Katz thick smears for S . mansoni ( 5 . 3% prevalence ) , and 90 positive slides for S . haematobium ( 39 . 8% prevalence ) . No infections with Ascaris lumbricoides or Trichuris trichiura were noted . Hookworm eggs were detected in 22 of the Kato-Katz thick smears ( 9 . 7% prevalence ) . However , these were not further subjected to experimental microscopes given the concerns over rapid egg degradation after collection of stool samples and laboratory work-up , pending microscopy analysis , thus affecting the validity of our results [21] . Fig 2 demonstrates S . mansoni eggs visualized by conventional microscopy , the Newton Nm1 microscope , and a clip-on version of the reversed lens CellScope . Table 1 outlines the operating characteristics of the Newton Nm1 microscope and the reversed-lens CellScope for S . mansoni and S . haematobium diagnosis . Sensitivities for S . mansoni and S . haematobium with the Newton Nm1 microscope were 91 . 7% ( 95% confidence interval ( CI ) 59 . 8–99 . 6% ) and 81 . 1% ( 95% CI 71 . 2–88 . 3% ) respectively , and specificities were 99 . 5% ( 95% CI 97 . 0–100% ) and 97 . 1% ( 95% CI 92 . 2–99 . 1% ) . S . mansoni and S . haematobium diagnosis with the reversed-lens CellScope demonstrated sensitivities of 50 . 0% ( 95% CI 25 . 4–74 . 6% ) and 35 . 6% ( 95% CI 25 . 9–46 . 4% ) , respectively , and specificities of 99 . 5% ( 95% CI 97 . 0–100% ) and 100% ( 95% CI 86 . 7–100% ) , respectively . The diagnostic sensitivity for the Newton Nm1 microscope for S . haematobium was 100% for egg counts ≥10 eggs/10 ml of urine . The clip-on version of the reversed-lens CellScope had limited sensitivity at low egg counts , but sensitivity improved as infection intensity increased , culminating in a sensitivity of >90% at 40 eggs/10 ml of urine or higher ( Fig 3 ) . Compared with conventional microscopy , estimates of egg counts for S . haematobium had a Pearson’s correlation coefficient of 0 . 98 using Newton Nm1 and 0 . 92 using the CellScope . Overall , 121 slides were examined for evidence of intestinal protozoa infection by both conventional microscopy and the Newton Nm1 microscope , with results outlined in Table 2 . Based on conventional microscopy , the prevalence of E . histolytica/E . dispar and G . intestinalis of SAF-fixed stool samples subjected to an ether-concentration method were 4 . 9% and 20 . 7% , respectively . The Newton Nm1 microscope demonstrated a sensitivity for E . histolytica/E . dispar and G . intestinalis of 83 . 3% ( 95% CI 36 . 5–99 . 1% ) and 84 . 0% ( 95% CI 63 . 1–94 . 7% ) , respectively , while specificity was 100% . For other intestinal protozoa , sensitivity of Newton Nm1 microscopy was variable ( 39–88% ) , while specificity was excellent ( 98–100% ) , as shown in Table 2 .
Our study shows that handheld microscopes such as the Newton Nm1 portable field microscope and the mobile phone-based CellScope can be successfully implemented into public health settings , after minimal training of laboratory technicians , for the diagnosis of gastrointestinal parasitic infections in rural African settings . Novel diagnostic approaches for common parasitic infections could have a positive impact on the quality of care delivered in resource-constrained settings [3] . Handheld microscopes may be useful tools in such settings as they are lightweight and easily transportable , enabling the delivery of quality diagnostics to individuals in rural , remote , or under-serviced locations rather than transporting people or specimens to distant laboratories . In addition , these devices are battery powered and are helpful in settings where there is no or only intermittent electricity . Indeed , mobile phone-based microscopes have several attributes that make them attractive for use in epidemiologic and public health settings . For example , mobile phone microscopes have the capacity to digitize images such that they can be saved and easily catalogued , or rapidly sent to other practitioners [22] . Digitization of images also allows for the attachment of geographic coordinates that may aid in mapping of infectious diseases and risk profiling of neglected tropical diseases [23] . Lastly , valuable clinical information associated with each image can be stored and catalogued , enabling healthcare providers for patient management . Additionally , the digitization of samples via mobile phone microscopy allows for computer vision and machine learning technology to aid in automated diagnoses and quantification of infectious diseases , such as malaria [24] , schistosomiasis [22] , giardiasis [25] , and filariasis [26] . One potential barrier to widespread implementation is that handheld and mobile phone microscopy only addresses the issue of enabling microscopy in underserviced settings . Developing simple , reliable , and low-cost approaches to standardized sample and slide preparation are required as well , and have received comparably little attention . To date , virtually all studies have evaluated handheld and mobile phone microscopes either in controlled laboratory settings or as used by expert microscopists . Prior to broader implementation , such devices must be rigorously validated in real-world settings and operated by front-line healthcare professionals . Our data confirm and add to findings from previous studies [9] demonstrating that laboratory technicians can reliably use handheld microscopes after minimal training . The Meade Readview handheld microscope was used by laboratory technicians in a Ugandan field study for Schistosoma diagnosis and demonstrated a sensitivity and specificity of 85% and 96% , respectively , compared to conventional microscopy . However , this device was limited by a smaller field of view , limited movement of the stage , and has not demonstrated widespread scale-up since its introduction [9] . Similarly , Ugandan laboratory technicians were trained to operate the Newton Nm1 microscope ( as used in this study ) , to evaluate a set of pre-selected malaria slides , and demonstrated a sensitivity and specificity of 93 . 5% and 100% , respectively [11] , although this study was not conducted in a true field setting . Our study adds to this prior work by implementing and evaluating the handheld microscope devices in a real-world field setting , demonstrating the utility of this device in day-to-day community-based diagnostic testing . In the current study , the clip-on version of the reversed-lens CellScope demonstrated excellent diagnostic specificity for S . mansoni and S . haematobium infection , but only modest sensitivity for these trematode eggs . This observation is consistent with a prior study evaluating the reversed-lens CellScope for S . haematobium diagnosis [14] . We suspect sensitivities were low because users must manually hold the device and guide the lens over the entire surface area of a slide . The CellScopes used in this study were not anchored to a solid structure nor did they utilize a microscope stage such as with conventional microscopy or the Newton Nm1 microscope ( Fig 1 ) , however future iterations of this device will have the ability to reliably read an entire slide . Hence , the operators are likely undercounting schistosome eggs due to the challenges of manually maneuvering the device over a microscope slide . Interestingly , the CellScope rapidly gains sensitivity at higher egg counts . For example the sensitivity of CellScope reaches that of the Newton Nm1 microscope ( >95% ) for S . haematobium diagnosis when used by laboratory technicians at infection intensities of 40 eggs per 10 ml of urine , which is still considered to be a low-intensity infection ( Fig 3 ) [27] . Diagnosing moderate- and high-intensity infections may be useful in clinical settings where worm burdens closely correlate with symptoms [28 , 29] . However it is still crucial to have the most sensitive tests available to diagnose even very low intensity infections for disease mapping , epidemiologic surveys , particularly after drug interventions , and rigorous surveillance . Newer versions of the CellScope are currently in development that will enable automated sample scanning and image interpretation . Limitations of our study include only commenting on the presence or absence of intestinal protozoa rather than quantifying these organisms . Future work should evaluate the diagnosis of intestinal protozoa in field settings with experimental microscopes rather than under laboratory conditions . Also , there were no infections with A . lumbricoides or T . trichiura in this setting , and hence , it would be useful to validate the diagnostic performance of these devices for these nematode eggs in other epidemiologic settings given the considerable global health importance of soil-transmitted helminthiasis [2] , in addition to other endemic infections . Lastly , our study was restricted to a community-based setting , and future studies should validate the diagnostic capabilities of these devices in clinical environments . In conclusion , handheld light microscopes have considerable potential for use in clinical and public health settings in resource-constrained environments . The clip-on reversed-lens CellScope , while convenient and low-cost to produce , was only modestly sensitive as currently used , however improvements are under development that could make it more appropriate for field deployment in the future . The Newton Nm1 handheld microscope , on the other hand , had good sensitivity and excellent specificity , and hence , could be readily integrated into real-world public health settings to diagnose intestinal parasitic infections .
|
Handheld light microscopes are new technologies that may be helpful in enabling better access to diagnostic testing for people living in resource-constrained settings in tropical and subtropical countries . Recent studies evaluating the accuracy of such devices have focused on their use by expert microscopists and were mainly conducted in laboratories . We evaluated the operating performance of two handheld microscopes ( Newton Nm1 microscope and clip-on version of the reversed-lens CellScope ) in comparison to conventional microscopy for the diagnosis of urogenital and intestinal schistosomiasis , when integrated into routine use in a community-based survey carried out in Côte d’Ivoire . Additionally , we evaluated the same microscopist’s diagnostic performance with the Newton Nm1 microscope for intestinal protozoa in a laboratory set-up . The Newton Nm1 microscope demonstrated excellent diagnostic sensitivity and specificity for schistosomiasis and intestinal protozoa . The CellScope had high specificity but only modest sensitivity for schistosomiasis diagnosis . Taken together , handheld diagnostic tools show promise to improve the quality of clinical and public health care delivered in resource-constrained settings .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"schistosoma",
"invertebrates",
"schistosoma",
"mansoni",
"medicine",
"and",
"health",
"sciences",
"engineering",
"and",
"technology",
"helminths",
"light",
"microscopy",
"animals",
"cell",
"phones",
"protozoans",
"microscopy",
"digestive",
"system",
"research",
"and",
"analysis",
"methods",
"schistosoma",
"haematobium",
"communication",
"equipment",
"gastrointestinal",
"tract",
"people",
"and",
"places",
"technicians",
"professions",
"diagnostic",
"medicine",
"anatomy",
"equipment",
"biology",
"and",
"life",
"sciences",
"population",
"groupings",
"organisms"
] |
2016
|
Accuracy of Mobile Phone and Handheld Light Microscopy for the Diagnosis of Schistosomiasis and Intestinal Protozoa Infections in Côte d’Ivoire
|
Genetic screens are powerful methods for the discovery of gene–phenotype associations . However , a systems biology approach to genetics must leverage the massive amount of “omics” data to enhance the power and speed of functional gene discovery in vivo . Thus far , few computational methods for gene function prediction have been rigorously tested for their performance on a genome-wide scale in vivo . In this work , we demonstrate that integrating genome-wide computational gene prioritization with large-scale genetic screening is a powerful tool for functional gene discovery . To discover genes involved in neural development in Drosophila , we extend our strategy for the prioritization of human candidate disease genes to functional prioritization in Drosophila . We then integrate this prioritization strategy with a large-scale genetic screen for interactors of the proneural transcription factor Atonal using genomic deficiencies and mutant and RNAi collections . Using the prioritized genes validated in our genetic screen , we describe a novel genetic interaction network for Atonal . Lastly , we prioritize the whole Drosophila genome and identify candidate gene associations for ten receptor-signaling pathways . This novel database of prioritized pathway candidates , as well as a web application for functional prioritization in Drosophila , called Endeavour-HighFly , and the Atonal network , are publicly available resources . A systems genetics approach that combines the power of computational predictions with in vivo genetic screens strongly enhances the process of gene function and gene–gene association discovery .
The demand by systems biology for bona fide , in vivo validated , biochemical interaction data and high quality functional annotations is much higher than the supply that geneticists are able to provide , principally because genetic approaches mainly focus on generating data on a gene-by-gene basis . On the other hand , computational predictions of gene function alone remain far from being accurate enough to be considered high-quality biological data . Integrated solutions , that combine the advantages of several approaches , should in theory provide both fast and physiologically relevant genetic data , while simultaneously increasing our understanding of biological processes . Genetic interactions in model organisms constitute a potentially invaluable source of in vivo interaction data for systems biology provided that throughput and speed can be increased . Currently , the number of known genetic interactions remains much smaller than the number of annotated physical interactions . For example , the BioGRID [1] database currently contains approximately 53 , 000 genetic interactions compared to almost 100 , 000 physical interactions . Clearly , the power of genetic approaches is that they produce - by definition - data that is directly relevant in a living system . Genetic screens , either for specific phenotypes or for modifiers of gene function , are thus a valuable source of large-scale interaction data . However , the main disadvantage of large-scale genetic screens is that they are costly , labor intensive , and time consuming . Turning in vivo genetic screens into a staple of systems biology by making them easier and faster without compromising their accuracy would therefore represent a major advance . In the bioinformatics community , process- or disease-related genes are , as of recently , being computationally predicted by taking advantage of the large amount of available sequence , function , annotation , and interaction data [2]–[13] . However to our knowledge , none of these methods have been used in combination with large-scale genetic experiments . Therefore , it remains unclear to what extent genome-wide , or even large-scale , computational predictions of gene-gene or gene-pathway associations , are biologically meaningful . Carrying out such screens on a large scale is difficult in human or mouse genetics , but the availability of genetic tools in Drosophila melanogaster together with collections of deficiency lines , mutants , and insertion lines , makes it an ideal model organism to investigate the concept of integrating genetic screens with gene prioritization methods . Here , we integrate genetics and computational biology to identify genetic interactions underlying neural development in the Drosophila Peripheral Nervous System ( PNS ) , a well-established model for neurogenesis . Proneural genes encoding proteins of the basic-helix-loop-helix ( bHLH ) super-family of transcription factors are essential for the initiation of neuronal lineage development in all species [14]–[18] . They act by forming heterodimers with the widely expressed bHLH E-proteins to bind a DNA motif called the E-box [19] and regulate the transcription of target genes . The highly conserved members of the Atonal ( Ato ) family are one example of proneural genes whose activity is required for the development of multiple lineages in vertebrates and invertebrates [14] , [20]–[22] . Despite a solid understanding of when and where ato-like genes are required in the Drosophila PNS and how they interact with Notch signaling to select neural precursor cells ( NPCs ) , the mechanisms that mediate their activity within NPCs and their specificity in inducing neuronal differentiation remain largely obscure . To identify genes involved in ato mediated neural development we propose a strategy for functional gene prioritization in Drosophila called Endeavour-HighFly that uses the same data fusion method and user interface as the human gene prioritization method Endeavour [3] , [23] . We identify 18 genes that interact with ato in two different contexts , including 2 previously uncharacterized genes , and use them to predict a core Ato interaction network . Furthermore , to broaden our strategy to other developmental processes , we prioritize the entire Drosophila genome for each of ten canonical biological pathways and generate a freely available database of candidate members or interactors for each pathway .
Three amino-acids within the basic domain of the first helix have been shown to mediate the specificity of ato function [24] , and the same motif enables specific transcriptional activation of the nicotinic acetylcholine receptor beta-3 subunit by the ato orthologue Ath5 [25] . Substituting the same amino acids in the Ato-related mouse proneural protein Neurogenin 1 ( Ngn1 ) for Ato group-specific residues ( NgnbAto ) allows Ngn1 to induce neurogenesis in Drosophila . This induction mimics that caused by Ato itself and depends on the fly E-protein Daughterless ( Da ) and the proneural co-factor Senseless ( Sens ) . Also , like endogenous proneural activity , it is antagonized by the Notch signaling pathway . Expression of the “Atonalized” form of mouse Ngn1 , NgnbATO ( Figure 1A ) under the control of dpp-Gal4 induces an average of ∼30 ectopic sensory bristles on the adult wing vein ( n = 30; Figure 1B , C ) . This is in contrast to an average of only ∼7 bristles induced by Ngn1 itself ( n = 26; p<0 . 001 ) , but is similar to the number induced by Ato ( n = 26 , n . s . ; Figure 1C ) . However , unlike Ato , NgnbATO induces significantly less lethality and many fewer wing deformities making it much easier to use in a large scale , quantitative genetic screen . In addition , just like for Ato , removal of one copy of sens reduces the number of NgnbATO-induced bristles by 55 . 6% ( Figure 1C ) . In order to bring the screen to a dosage critical value , a heterozygous sens mutant was introduced into the background of UAS::NgnbATO; dpp-Gal4 . The number of ectopic bristles with this system provides a sensitized and quantitative read out in which to screen for modifiers of Ato function . To test the feasibility of isolating dominant modifiers of the number of ectopic bristles , we crossed UAS::Ngnbato/Cyo;sens , dpp-Gal4/TM6c , flies to da or Notch mutant flies . We find that removal of a single copy of da almost completely suppressed NgnbATO induced bristle formation ( average of 0 . 7±0 . 9 bristles; n = 27 , p<0 . 001 ) , while removal of one copy of Notch strongly enhanced the phenotype ( average of 43 . 5±4 . 1 bristles; n = 23 , p = 0 . 002; Figure 1C ) . All together , these data suggest that the assay is both robust and sensitive and should enable the identification of specific quantitative modifiers involved in ato-dependent neurogenesis in the Drosophila PNS . Following this strategy , a deficiency screen of the second and the third chromosomes for modifiers of Ngnbato misexpression was performed . The deficiency kit is a collection of fly stocks that each carries a deficiency , or deletion , chromosome uncovering multiple genes . The different deficiencies encompass most of the chromosome and deficiency screening is an established and rapid assay to identify chromosomal regions with enhancer and suppressor loci for a given phenotype or pathway [26] . To identify chromosomal loci that influence ato-induced neural development , 180 deficiency fly lines were crossed to UAS::Ngnbato/Cyo;sens , dpp-Gal4/TM6c , flies . Loci were considered positive if they altered the number of ectopic bristles on the adult wing vein by more than 30% compared to the number of bristles induced in sibling control flies , as well as in wild type Canton S flies , and if the change in bristle number was strongly statistically significant ( p<0 . 01 ) . Following these stringent criteria , 17 positive regions on chromosome 2 and 14 positive regions on chromosome 3 were identified . Since induction of ectopic bristles is a common property of all proneural genes , the identified loci might be involved in both achaete-scute and ato dependent neurogenesis . In order to identify Ato-specific loci , the individual candidate deletion stocks were tested with flies expressing UAS::ato , UAS::ngn1 , and UAS::sc , respectively , under the control of dpp-Gal4 . The loci which modified Ato misexpression , but not that of Sc or Ngn1 were considered to be Ato-specific loci . Of the 31 loci identified in the primary screen , only one failed to interact with any of the genes in the secondary screen . We find that 15 of the 31 loci interact with both ato and at least one other proneural gene , while 2 loci interact only with ngn1 and 1 locus interacts only with sens ( data not shown ) . The remaining 12 loci ( 6 on chromosome 2 and 6 on chromosome 3 ) interact specifically with ato . Examining the breakpoints of the overlapping deletions uncovering these 12 loci shows that they harbor 1056 annotated genes ( Figure 1D and Table S1 ) . Each of these loci is expected to harbor one or more ato-interacting genes . The identification of the individual modifier genes from these regions is similar to the problem in human genetics where for a given human phenotype and its underlying chromosomal locus , identified by cytogenetic studies or linkage mapping for example , the individual disease-causing gene ( s ) need ( s ) to be identified . Besides directly providing interaction candidates , the twelve positive regions resulting from the deficiency screen provide an excellent opportunity to test the principle of gene prioritization on a large scale and in an unbiased setup . First we present a redesign of an existing gene prioritization approach that is specifically tuned towards the Drosophila genome , and then we use it to select the most promising candidates from the 1056 genes within the twelve positive regions . To prioritize Drosophila genes we upgraded the existing Endeavour tool for gene prioritization [3] , [23] by including Drosophila data sources ( Table 1 and Materials and Methods ) and we name this version Endeavour-HighFly , or HighFly for short . To test the performance of each individual Drosophila data source we carried out leave-one-out cross-validations ( LOOCV; see Experimental Procedures ) on several gene sets . Each set contains genes that are “similar” to each other for different reasons , for example genes with similar expression patterns or genes from the same pathway . We tested whether HighFly could identify the correct members of each set by leaving out one gene at a time and calculating the similarity between the left-out gene and the rest of the set . We found that HighFly ranks highly the left-out genes when at least one data source holds the information that this gene is related to the remainder of the gene set ( for example , the expression data source is informative for the expression-related gene set ) ( Figure 2 ) . Importantly , regardless of which particular data sources show the strongest performances , the performance of the combined or fused ranking ( last column in Figure 2 ) is highly robust for all sets , it is not influenced by non-informative data sources , and it is almost always greater than 90% compared to a performance of ∼50% for randomly assembled sets of genes ( Figure 2 ) . These results validate the technical aspects of the implementation and suggest that HighFly performs robust prioritizations on Drosophila data sources . Next , we investigated whether HighFly would be capable of finding genes that interact in vivo with ato . A training set , called TRAIN_Ato1 was assembled with the following genes: ato , Brd , rho , Takr86C , pnt , dpp , Egfr , da , wg , sens , chn , and sca . Because different sizes and compositions of training sets are possible , we tested the suitability of this training set for ato-related gene prioritization , by performing two tests . First , we assessed the content of some of the trained submodels . The trained GO submodel for this set contains “peripheral nervous system development” , “cell fate specification” , “eye morphogenesis” , “sensory organ development” , etc . as highly over-represented terms ( p value<10−09 ) . The Text submodel contains stemmed terms like “cell fate” , “notch” , “egfr” , “disc” . The InterPro submodel has no highly over-represented domains , but “Basic helix-loop-helix dimerization region bHLH” is marginally over-represented ( corrected p-value = 0 . 07 ) . Secondly , we tested the homogeneity of TRAIN_Ato1 , by subjecting it to LOOCV and obtained an AUC performance of 98 . 5% , suggesting that TRAIN_Ato1 is a coherent and internally consistent training set . To test the possibility of obtaining biologically meaningful prioritizations , we performed a pilot test by prioritizing the right arm of chromosome 3 ( chr3R ) using TRAIN_Ato1 and then divided all the genes on the list into three groups: the top 1/3 , the middle 1/3 , and the bottom 1/3 . From each group the top 30 genes for which stocks with mutant alleles are available from the public stock centers were examined for their modification of ato's proneural activity , using the same bristle induction assay described above . Four positive genes were found in the top group ( rn , Antp , gro , and pros ) , none in the middle group , and none in the bottom group ( Table 2 and Table S2 ) . Although the power of this preliminary test is greatly limited due to the relatively small number of genes tested ( 90 ) and the variability of available alleles , we found these results sufficiently encouraging to proceed with HighFly prioritizations of all twelve modifier loci found in the deficiency screen . However , to further evaluate HighFly , we intentionally chose a less stringent threshold of further validating the top 30% of ranked genes so as to compare the rankings of positive and negative genes with a sufficiently large sample size at the end of the screen . To identify candidate genes within the positive regions , all genes in each of the twelve positive regions were prioritized separately using TRAIN_ATO1 as training set and all 12 HighFly data sources ( Table S3 ) . For all genes that were ranked within the top 30% , a mutant stock , when available , was ordered from the public stock centers . Each mutant was then crossed to the sensitized tester fly stock ( uas::ngnbato/Cyo;sens , dpp-Gal4/TM6c ) and the bristles at the anterior-posterior margin ( where dpp-Gal4 is expressed ) were counted and compared to the number of bristles observed in the control flies as described above . For twelve genes , namely toc , lilli , Sbb , fj , mus209 , zip , shg , Egfr , dom , smg , cas and ppan , the number of bristles was significantly lower or higher ( p<0 . 01 ) than in the control flies ( Table 2 , bottom panel ) . Each of these mutants were then tested against uas::sc; uas::ngn1 , and uas::ato under the control of dpp-Gal4 to check for the specificity of ato interaction . All of the genes modified only the ato gain of function phenotype ( data not shown ) . We note that although mutants of genes that ranked in the top 30% of each locus were tested , 11 of the 12 ranked in the top 6% of their locus ( Table 2 and Table S3 ) , suggesting that HighFly prioritizations enrich strongly for positive interactions . Similar prioritizations were obtained by using a different high-quality training set ( LOOCV AUC = 99 . 5% ) , assembled by selecting all 18 known interactors of ato from BioGRID ( data not shown ) . In contrast when the same 12 genes were prioritized using 100 randomly assembled training sets , the median rank ratio was 0 . 247 compared to 0 . 02 for the ato training set ( Figure 3 ) . An alternative analysis , instead of prioritizing each deficiency region separately , is to pool all candidate genes from the positive deficiency regions and prioritize this set in one analysis . We performed such a prioritization as post-analysis and found all 12 positives ranked in the top 10% ( Table S1 and Figure S1 ) . An examination of the contribution of individual data sources to the high rankings of the positive genes shows that for all positives , their high ranking is caused by high rankings for several data sources , rather than a single high ranking for one of the data sources ( Figure S1 ) , which supports our initial assumption of the added value of data integration for gene prioritization . In a second post-analysis , by comparing HighFly with existing online tools such as FlyBase [27] , UCSC Gene Sorter [28] , and STRING [29] , we found that the use of a training set of genes related to ato is more favorable than a single gene query; and also that a gene ranking is more favorable for gene identification than a gene filtering ( e . g . , using a selection of Gene Ontology terms or a selection of FlyBase expression terms ) ( Text S1 ) . Functional inspection of the 16 positive genes ( 12 from the deficiency screen +4 from the pilot screen of 90 genes on chromosome 3R ) by Gene Ontology statistics [30] revealed that this gene set is significantly enriched for developmental processes that require ato such as eye development and regulation of transcription ( Table 3 ) . Finally , we compared the phenotypic distribution of the effects of the modifier genes identified in our screen with the distribution documented for saturating forward genetic screens and cellular siRNA screens [31] . We find that despite the relatively small number of genes that need to be tested in a HighFly screen , the distribution of phenotypes mirrors that obtained in genome wide forward and reverse genetics screens ( Figure 4 ) . These data further support the power and accuracy of the integration of computational biology and genetics . ato acts as a proneural gene for two different types of founder cells . The first is a subset of sense organ precursor ( SOP ) of the body wall and appendages and the second is the R8 founder cell of the retina . The major difference between the SOP and the R8 is that the SOP undergoes cell division to generate the sensory organ , whereas the R8 cell terminally differentiates . However , both cells share the property of recruiting neighboring cells into the ato-dependent fate; a property unique to ato , not shared by other proneural genes . We assessed whether genes identified in one context , also operate in the other . To this end , we tested the relationship between ato and its putative interactors in the developing fly retina , where ato function is well described [32] . In the retina , ato specifies the first photoreceptor , or R cell , the R8 ( Figure S2A , B ) . The R8 then releases an ato-dependent EGF signal that organizes the rest of the retinal field and specifies the R1–R7 . Loss of ato function in the retina results in the complete failure of retinal specification [33] . Expression of an ato-RNAi construct ( A kind gift of A . P . Jarman ) in the eye in ato heterozygous flies ( uas::ato-RNAi;h-Gal4 , ato1; see Materials and Methods ) reduces R8 specification and consequently the recruitment of other R cells in a dose dependent fashion ( Figure S2C , D ) . One copy of ato-RNAi produces a smaller eye with approximately half the normal number of ommatidia ( Figure 5A , B ) . Mutants for the 16 genes identified in the screen were crossed to the ato-RNAi flies and scored for their ability to dominantly modify the ato RNAi phenotype . Ten of the 16 tested genes , namely gro , rn , EGFR , cas , ppan , toc , sbb , fj , shg and dom dominantly enhanced the ato-RNAi phenotype , with nine showing further reductions in eye size to approximately 250 ommatidia ( Figure 5C , D ) . An 11th gene , pros , was semi lethal . The remaining five genes , namely Antp , smg , lilli , mus209 , and zip , did not appear to alter the ato-RNAi induced small eye size . The data thus far suggest that at least 10 of the 16 genes we identified in the sensory bristle screen also interact with ato during retina development . Some of these genes such as pros are known for their role in neurogenesis [34] , while the EGF receptor is well known for its close interactions with ato [35] . However , most of the genes we identified as genetic interactors of ato have not , to our knowledge , been previously shown to play a role in ato-dependent neurogenesis . Next we asked if these genes might be co-expressed with ato in the various PNS anlagen that derive from ato expressing precursors . We were able to obtain LacZ enhancer trap lines from stock centers for 10 of the 16 interacting genes ( dom , fj , lilli , mus209 , pros , rn , sbb , shg , toc and zip ) to examine their expression patterns in the third instar larval ( L3 ) imaginal discs . In the eye , antennal , leg and wing L3 discs , Ato marks the progenitor pools and the very early precursor cells of specific neuronal lineages . Senseless then marks the precursor cells during and after Ato expression . One enhancer trap , rn , did not show any obvious expression relationship to ato . Two of the 10 genes , mus209 ( fly PCNA ) and sbb are generally expressed . An additional two lines , toc and zip showed expression in the posterior part of the eye disc ( Figure S3A ) , suggesting a later function than that of ato . Finally , other five of the 10 tested enhancer traps showed a clear expression relationship with Ato ( Figure S3 ) . We observed strong lacZ expression in the L3 discs in Ato-expressing and Ato-dependent cells in the eye disc ( fj , lilli , shg , pros ) , in the antennal Johnston organ precursor cells ( dom , shg , pros ) , in the chordotonal organ precursor cells of the wing and leg imaginal disc ( dom , shg , pros ) ( Figure S3 and data not shown ) . It should be noted that enhancer trap lines might reflect only part of the total expression pattern of the trapped gene . The data above support the feasibility of rapidly and accurately identifying gene function through the fusion of in silico gene prioritization and in vivo genetic screens . One issue that faces all gene prioritization approaches is an expected bias towards genes with at a large amount of pre-existing information in several databases . Although this is still valuable in assigning novel functions to known genes , we reasoned that it would be interesting to test the performance of HighFly in the prioritization of genes about which there is little explicit information . Genes with limited annotations can potentially be ranked high due to data sources that are independent of existing knowledge , such as sequence similarity , protein domains , gene expression data , or protein-protein interaction data from high-throughput experiments . Indeed , 30 out of 96 genes , known only by their CG numbers , ranked in the top 10% of the ato-specific deletion loci identified in the initial bristle screen ( Table S4 ) . The recent availability of a genome-wide in vivo Drosophila RNAi library [36] allowed us to test these genes for their interaction with ato . When no off-target effects were predicted , available RNAi lines were ordered and crossed to the ato-RNAi flies driven by the h-Gal4 driver in an ato heterozygous background ( uas::ato-RNAi;h-Gal4 , ato1 ) , as well as two different control lines; h-Gal4 , ato1 and h-Gal4 alone . To avoid potential artifacts resulting from the RNAi approach , we set relatively stringent criteria: we searched for genes that show synthetic lethality specifically and only in combination with ato-RNAi , but show no phenotype under the two control conditions . We were able to obtain a total of 36 RNAi lines for 24 uncharacterized genes ranking in the top 10% of positive deficiency regions . Eleven RNAi lines were lethal under all conditions and could not be evaluated further . The 25 remaining RNAi lines allowed us to perform knockdown of 17 genes . Of these , 2 genes ( CG1024 , CG1218 , ) caused lethality only in combination with atoRNAi , but not under control conditions ( Table 4 ) . As a further confirmation for the specificity of these interactions , we tested 51 RNAi lines for the bottom 10% ranking genes in each deficiency . None of these lines showed specific synthetic lethality in combination with atoRNAi ( data not shown ) . Thus , the combination of HighFly prioritization , the RNAi library and genetic screening allows the rapid functional identification of previously uncharacterized genes . The combination of forward and reverse genetics tools and computational biology allowed the identification of 18 , mostly novel , genetic interactions with the proneural gene ato . We sought to determine if the identified positive genes are functionally associated with each other , with ato , and with any of the other training genes that were used originally to identify these genes . To this end we used the STRING [29] protein-protein association predictions at 0 . 8 confidence level and determined the optimally connected sub-network that can be formed among the 18 positive genes , via maximally two other proteins ( see Materials and Methods ) . We find that a network can be constructed that includes 12 of the 16 known genes ( data not shown ) . As expected , the 2 unknown genes play no role in this analysis because of the lack of STRING data at this high confidence level . This analysis discovers Ato itself as member of the best network that connects the positive genes . We found that the maximal confidence level at which Ato is still part of the network is 0 . 842 , and therefore used this stringency for further analyses . The network formed by the 16 known genes at this confidence level ( Figure 6A ) contains 84 nodes and 250 edges , and now includes 12 of the 16 positive genes and 6 training genes , including ato . Egfr is directly connected to ato; fj , Antp and gro are connected to ato via one other protein; pros , rn , shg , lilli are connected to ato via two other proteins and cas , smg , zip , mus209 via three other proteins . To determine the significance of finding a large interconnected network , which includes ato , starting from the 16 positive known genes , we generated 1000 random sets of 16 known genes . Specifically , we used only genes with a name in FlyBase and at least one GO biological process annotation . Only 29 of the 1000 networks contain ato and , on average , they contain 0 . 70 ( S . D . = 1 . 13 ) training genes , 7 . 83 nodes ( S . D . = 9 . 09 ) , and 13 . 07 edges ( S . D . = 19 . 28 ) . An example of such a network is shown in Figure 6B . With a p-value of 0 . 029 to find Ato in the real network , p<0 . 001 to obtain 84 nodes , p<0 . 001 to obtain 250 edges , and p = 0 . 001 to recover 6 of the 11 training genes , we conclude that the positive genes we identified are strongly associated with each other and with Ato and its known interactors . A particular feature of the HighFly tool is the speed of prioritization . We wondered whether this computational efficiency makes it possible to prioritize whole chromosomes or even the entire genome . To this end we asked if it is possible to rank the 16 known genes identified in our screen on their respective chromosomes , and if so , whether these rankings would be high . Table 2 shows the chromosomal rankings of these genes . All except one of the known genes rank within the top 10% of their respective chromosome ( Table 2 ) . These data suggest that it is possible to obtain strongly meaningful gene prioritizations across large data sets . We sought to illustrate the general applicability of fly gene prioritization and simultaneously generate a second community-wide resource by prioritizing the entire genome to identify genes that are related to , or potentially involved in , either of ten signaling pathways , namely Transforming Growth Factor beta ( TGFβ ) receptor signaling pathway ( GO:0007179 ) , Epidermal Growth Factor Receptor ( EGFR ) signaling pathway ( GO:0007173 ) , Fibroblast Growth Factor Receptor ( FGFR ) signaling pathway ( GO:0008543 ) , Notch ( N ) signaling pathway ( GO:0007219 ) , Sevenless ( Sev ) signaling pathway ( GO:0045500 ) , Smoothened/Hedgehog ( Smo/H ) signaling pathway ( GO:0007224 ) , Toll signaling pathway ( GO:0008063 ) , Extracellular signal-Regulated Kinase ( ERK; GO:0007259 ) , JAK-STAT ( GO:00016055 ) and Wnt signaling pathway ( GO:0016055 ) . To investigate the rankings in terms of biological processes we calculated GO over-representations for each top 100 ranked genes , excluding the training genes . We also excluded genes that were ranked in the top 100 for more than two pathways and GO-terms that were over-represented in more than four pathways . We find that typical overrepresented functions are cell adhesion and photoreceptor fate commitment for EGFR-related genes; cell migration for FGFR; neuroblast fate determination and equator specification for Notch; defense response for Toll; and ectoderm development for Wnt , suggesting that the prioritizations are biologically meaningful . Finally , we compared prioritizations for 4 of the 10 pathways- namely ERK , Wnt , Hh and JAK-STAT- for overlap with published genome-wide siRNA screens . We find significant overlap between the top 10% of the genome as prioritized by HighFly and the genes scored as positives in these screens for 3 of these pathways ( Figure 7 ) . Only the Hh pathway screen shows poor overlap with the prioritizations . Prioritizations and functional analyses , as well as the HighFly software , are available at http://med . kuleuven . be/cme-mg/lng/HighFly .
The molecular unraveling of biological processes in the post-genome era is characterized by the use of high-throughput experiments and the integration of prior knowledge ( e . g . , the use of GO-statistics to select microarray generated gene clusters ) , and is therefore supported and guided by bioinformatics . Genetic screens in model organisms such as Drosophila melanogaster are also high-throughput experiments , but they are yet to be aided by computational techniques , as an integral part of the screen itself . We sought to demonstrate the power of an integrated approach that combines high-throughput in silico and in vivo genetic approaches . This integration allowed us to quickly identify novel genetic interactions during neural development in the fly PNS , while significantly reducing the workload of the genetic screen . First , a classical deficiency modifier screen is performed . Then , instead of assaying all the genes located within the positive deficiency regions , the best candidates are selected computationally . This is done by integrating multiple heterogeneous genome-scale data sources , both representing published knowledge ( e . g . , functional gene annotations or protein-protein interactions ) , genome sequences , and experimental data ( e . g . , gene expression data or phenotypes ) . As such , we were able to assign novel functions for known genes whose involvement in ato-dependent neural development was unknown , as well as describe functions for uncharacterized genes . A major advantage of genetic screens is that they are unbiased: they can reveal a function for a previously unknown gene . Although gene prioritization based on available data would have been expected to affect this property of screens , our data indicate that this is not necessarily the case . Even genes with very little explicit information , and no known function could be identified both as high ranking and as bone fide interactors in vivo in our HighFly supported screen . In addition , our data suggest that the combination of HighFly prioritizations and transgenic RNAi lines can result in very rapid functional gene discovery . The use of an integrative screening strategy combining computational biology with medium or high-throughput screening assays is likely to be applicable to a broad range of screening assays ( from in vitro to in vivo assays ) beyond Drosophila genetics . Essentially any assay designed around evaluating a given gene , and for which whole-genome screening is outside the reach of the typical lab , could benefit from strategies similar to ours . Even with more extensive resources , it may be more productive ( at equal time and cost ) to evaluate several prioritized screens than a single whole-genome screen . Obviously , the strategy we propose is not applicable in the case where extremely little is known about the molecular basis of a phenotype ( because of lack of a training set ) while a genetic screen would still be feasible . It is a clear research challenge for computational biology to develop methods applicable to such a situation . A further advantage of our integrated systems genetics approach is the combination of speed and accuracy of gene function discovery . In this work we tested a total of 180 deletion lines , 220 mutants and 36 RNAi lines to identify 18 ato interacting genes , representing a discovery rate of ∼5% . It should be noted that the 220 mutants tested include 90 mutants examined only for the purposes of testing the prioritizations as well as 78 mutants ranking between 10% and 30% of their deletion regions . Our data clearly indicate that testing genes ranked in the top 10% only will suffice to discover the vast majority of sought after genes: 17 of the 18 genes identified ( ∼94% ) rank in the top 10% of their tested regions . Thus , assuming all genes have available RNAi lines or mutant alleles , testing only 96 genes , after the initial deficiency screen , would have identified at least 17 ato interacting genes , a discovery rate of almost 18% . In this regard we note that Endeavour-based prioritizations appear to outperform existing tools . We believe this to be due to three main properties namely the use of a multi-gene training set , the integration of multiple data sources , and the production of gene rankings . The genes we find to interact with ato reveal an interaction network underlying early neural differentiation . Network analysis reveals two important aspects of the screen . Although neither Ato nor its known interactors were included in the query , the best network found includes Ato and almost all of its known interactors . In addition network analysis yields a number of interesting insights . First , most of the 89 genes in this network are signaling molecules and transcription factors belonging to the Notch , Wnt , EGFR , Dpp and Hh pathways . These pathways are known to interact with ato and our data suggest that the newly identified ato interacting genes may be members of these pathways or may implement the interactions between ato and these pathways . Second , most of the genes tested for both bristle formation and retinal development interact with ato in both assays . This suggests that ato may work with a core group of genes to implement context-specific neural fate decisions . One exception to this appears to be genes acting in cell division ( mus209 , lilli , zip ) that , not surprisingly , interact in the bristle assay , but not the R8 assay . Third , we note that HighFly was able to predict the interaction of uncharacterized genes with ato , which network analysis alone , would have not been able to predict . In summary , a systems genetics [37] approach not only identifies novel functions for individual genes with great speed and accuracy , but , as would be desirable in a systems biology context , also uncovers the structure and functional attributes of the network formed by these genes . Yet , the main advantage of systems genetics over other systems biology approaches is that the results are physiologically relevant by definition , because they are discovered directly in vivo . The HighFly tool can perform prioritizations on the entire fly genome . We have done this for ten major signaling pathways , but many other prioritizations are possible , depending on the interest of the user . HighFly and its prioritizations are public resources that we hope will contribute to enhancing the speed and accuracy of functional gene discovery in vivo and establishing classical genetics as a fundamental tool of systems biology .
All crosses were performed at 25°C , except for the atoRNAi eye screen crosses which were performed at 28°C , on standard fly food . Deficiency kits , LacZ enhancer trap flies and all mutant lines were obtained from the Bloomington and Szeged stock centre . The atoRNAi lines were kindly provided by Andrew Jarman , and the RNAi lines for uncharacterized genes were obtained from the Vienna Drosophila RNAi Center ( VDRC ) . Third instar larval imaginal discs were dissected in 1× PBS . Discs were fixed with 4% formaldehyde in 1× PBT for 15 minutes . Then , washed five times ( 15 min/T ) in 1× PBT . Blocking and antibody incubation were performed as described [38] . The antibodies used were: sheep anti-ATO ( 1∶250 ) , rabbit anti-GFP ( 1∶1000 ) , rat anti-Elav ( 1∶100 ) , guinea pig anti-SENS ( 1∶1000 ) mouse anti-βgal ( 1∶1000 ) , rabbit anti–βgal ( 1∶1000 ) . Secondary antibodies were always used 1 in 500 . Samples were mounted in Vectashield mounting medium and detected using confocal microscopy ( BioRad 1024 , Hercules , California , United States and Leica DM-RXA , Wetzlar , Germany ) . The fly strain w; UAS::ngnbato/CyO; sens , dpp-GAl4/TM6 was used to set up crosses with deficiency lines . The number of the ectopic bristles was used as a parameter to reflect the strength of the proneural function of Ato in this context [24] . When a deficiency region caused a significant change in the number of ectopic bristles , the corresponding deficiency line was further crossed to three fly lines UAS::ato; dpp-Gal4 , UAS::ngn; dpp-Gal4 and UAS::sc; dpp-Gal4 and the number of ectopic bristles was counted . Deficiencies were considered ato specific when they altered the amount of bristles generated by UAS::ato/cyo; dpp-Gal4/TM6 , and not by UAS::ngn , dpp-Gal4/TM6 or UAS::sc , dpp-Gal4/TM6 . Within these deficiency regions , high-ranking mutant lines available in the stock centre were ordered and crossed to w; uas::Ngnbato/CyO; sens , dpp-GAl4/TM6 . If a mutant still caused a significant change in bristle number , the corresponding gene interacts with Ato . The positive genes were tested with flies expressing UAS::ato , UAS::ngn1 and UAS::sc respectively under dpp-Gal4 control to check for specificity . All ectopic bristles were counted under stereomicroscope . For all statistic analysis , the sample number is n = 10 , and a significant difference between two average values is defined as p≤0 . 01 . The eye phenotype screen was performed by crossing w; UAS::atoRNAi/CyO; h-Gal4 , ato1/TM6C , which reduced the eye size in 50% of the flies , with the mutant strains identified in the bristle screen . Positive genes for retinal modifiers of ato were mutants that enhanced or suppressed the atoRNAi phenotype . RNAi strains were crossed to h-Gal4; ato1/TM6 and h-Gal4 as controls . Only the one showing synthetic lethality specifically with w; UAS::atoRNAi/CyO; h-Gal4 , ato1/TM6C , but not with two controls was considered as positive . The gene prioritization method [3] , [23] works as follows . First , a set of training genes is defined to describe the particular process under study . For each data source , the following data for the training genes are assembled: ( 1 ) a gene's function derived from FlyBase GO annotation , textual information extracted from PubMed abstracts , SwissProt keywords and KEGG pathway membership; ( 2 ) a gene's expression pattern derived from two general Drosophila microarray data sets [39] , [40] and embryonic in situ expression patterns from the Berkely Drosphila Genome Project ( BDGP ) ; ( 3 ) a gene's protein sequence from Ensembl and its protein domains from InterPro; ( 4 ) described mutant phenotypes from FlyBase; and ( 5 ) described genetic interactions or predicted protein-protein associations from BioGRID and STRING . The applied training and scoring strategies for each data source are described in Table 1 . For each gene in a “test set” the similarity with a submodel is calculated and the ranks according to individual submodel scores are integrated using order statistics , yielding a q-value . The q-value is transformed into a p-value according to fitted distributions , depending on the number missing values . Finally , the test genes are ranked according to this p-value . We assembled sets of genes involved in the same signaling pathway , tested on eight pathways defined by GO; genes with similar expression patterns using an expression cluster from Arbeitman et al . [39] and a second cluster of all genes expressed in Bolwig's organ from FlyBase; genes with the same protein domain , namely the bHLH domain; all genes that interact with the same gene , tested on all interactors with Atonal from BioGRID; and genes that are co-cited with a specific gene in PubMed abstracts , namely genes cited with ato , extracted using iHOP [41] . In LOOCV , every gene from every validation set is , in turn left out , and the ranking of the left-out gene within a set of 99 randomly selected genes is recorded . From all these rankings , Receiver Operating Characteristic ( ROC ) curves are generated and the area under this ROC curve is used as a measure of the performance of each individual data source and of the integrated prioritization . The aim of the network extraction is to obtain a subgraph that connects the genes of interest ( the seed genes ) . Network connections were extracted from the STRING protein-protein associations , using a minimum edge confidence ( above 0 . 8 ) . We define the connecting nodes ( the non-seed genes ) in the subgraph as the nodes that are on the shortest path ( s ) between two or more seed genes . To identify those connecting nodes , a multiple sources breadth-first search is performed , which is initialized with the seed genes . During the search , the minimum distance to the seed genes is recorded until seed genes are reachable from one another . Upon completion , the final network is obtained by exploring the shortest paths , starting from the seed genes , that have a maximum length of 4 and that connect at least two seed genes . Hence , the extracted network is made of one or more connected components and may not include all the seed genes . The obtained networks were visualized using Cytoscape [42] .
|
Genome sequencing and annotation , combined with large-scale molecular experiments to query gene expression and molecular interactions , collectively known as Systems Biology , have resulted in an enormous wealth in biological databases . Yet , it remains a daunting task to use these data to decipher the rules that govern biological systems . One of the most trusted approaches in biology is genetic analysis because of its emphasis on gene function in living organisms . Genetics , however , proceeds slowly and unravels small-scale interactions . Turning genetics into an effective tool of Systems Biology requires harnessing the large-scale molecular data for the design and execution of genetic screens . In this work , we test the idea of exploiting a computational approach known as gene prioritization to pre-rank genes for the likelihood of their involvement in a process of interest . By carrying out a gene prioritization–supported genetic screen , we greatly enhance the speed and output of in vivo genetic screens without compromising their sensitivity . These results mean that future genetic screens can be custom-catered for any process of interest and carried out with a speed and efficiency that is comparable to other large-scale molecular experiments . We refer to this combined approach as Systems Genetics .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"and",
"genomics/animal",
"genetics",
"genetics",
"and",
"genomics/bioinformatics",
"neuroscience/neurodevelopment"
] |
2009
|
Integrating Computational Biology and Forward Genetics in Drosophila
|
For Chagas disease , the most serious infectious disease in the Americas , effective disease control depends on elimination of vectors through spraying with insecticides . Molecular genetic research can help vector control programs by identifying and characterizing vector populations and then developing effective intervention strategies . The population genetic structure of Triatoma infestans ( Hemiptera: Reduviidae ) , the main vector of Chagas disease in Bolivia , was investigated using a hierarchical sampling strategy . A total of 230 adults and nymphs from 23 localities throughout the department of Chuquisaca in Southern Bolivia were analyzed at ten microsatellite loci . Population structure , estimated using analysis of molecular variance ( AMOVA ) to estimate FST ( infinite alleles model ) and RST ( stepwise mutation model ) , was significant between western and eastern regions within Chuquisaca and between insects collected in domestic and peri-domestic habitats . Genetic differentiation at three different hierarchical geographic levels was significant , even in the case of adjacent households within a single locality ( RST = 0 . 14 , FST = 0 . 07 ) . On the largest geographic scale , among five communities up to 100 km apart , RST = 0 . 12 and FST = 0 . 06 . Cluster analysis combined with assignment tests identified five clusters within the five communities . Some houses are colonized by insects from several genetic clusters after spraying , whereas other households are colonized predominately by insects from a single cluster . Significant population structure , measured by both RST and FST , supports the hypothesis of poor dispersal ability and/or reduced migration of T . infestans . The high degree of genetic structure at small geographic scales , inferences from cluster analysis and assignment tests , and demographic data suggest reinfesting vectors are coming from nearby and from recrudescence ( hatching of eggs that were laid before insecticide spraying ) . Suggestions for using these results in vector control strategies are made .
Chagas disease is a parasitic disease in which the pathogenic agent , Trypanosoma cruzi is transmitted by hematophagous insects of the sub-family Triatominae . Triatoma infestans is the major vector in the Andean highlands where the disease is endemic and has infected humans for over 9000 years [1] . Chagas disease is the most important parasitic disease in the Americas in terms of mortality and economic impact [2] . In Bolivia the endemic area covers 55% of the country and , in 1985 , more than one million people were infected [3] . In 1991 a public health program , the Southern Cone Initiative was launched by the World Health Organization to eliminate vector populations [4] , through spraying of houses and surrounding areas with pyrethroid insecticides [5] . In Argentina , Brazil , Chile , and Uruguay , T . infestans is exclusively domestic or peri-domestic , thus eradication of the vector in these regions , followed by vigilance against re-infestation , has proven largely successful in reducing transmission of T . cruzi and thus the prevalence of Chagas disease [6] . In contrast , in Bolivia the vectors occur in domestic , peri-domestic , and sylvatic environments [7]; thus , control of T . infestans in towns and homesteads is confounded by the possible re-infestation from surrounding sylvatic areas . Molecular genetic research can help vector control programs by identifying and characterizing genetically distinct vector populations and then developing effective intervention strategies [8] . Several genetic markers including isozymes and the mitochondrial cytochrome b gene have proved useful in studying the genetic diversity of T . infestans [9] , [10]; however , markers with more resolution would aid vector control efforts . DNA based microsatellite markers have been widely used in population studies because of their large polymorphism information content , widespread distribution in the eukaryotic genome and robust methodology . To reduce transmission of Chagas disease , estimates of population differentiation are crucial to understand vector dispersal , sources of reinfestation , and gene flow; this genetic information is an important tool for effective management of insect control programs . Here we aimed to investigate the population genetic structure and inferred the source of colonization of vectors in the department of Chuquisaca , Bolivia using ten highly polymorphic microsatellite markers . The geographic region has high levels of human infection and house infestation and is located in a region thought to be the evolutionary origin of T . infestans .
Insects were collected from 23 localities including both peri-urban ( inhabited areas in the immediate vicinity of a city ) and rural sites ( less than 2000 inhabitants ) in the provinces of Oropeza , Zudañez , Azurduy , Yamparaez , Tomina , Belisario Boeto and Hernando Siles within the Department of Chuquisaca , in the Bolivian highlands ranging from 1079 to 3020 meters above the sea level ( Table 1 , Figure 1 ) . This area presents a broken topography with numerous valleys and small plateaus characterized by very diverse climates . In the Andean highlands , wheat is grown predominantly in small-scale , subsistence farming systems . In higher precipitation areas , potato is the preferred crop . Rainfall in these areas ranges from approximately 300 to 600 mm per year . In the Andean Plateau the average temperature is less than 10°C and there is less than 500 mm of annual precipitation . The Andean valleys present moderate climates , with average temperatures of 18°C and approximately 500 and 600 mm of rain every year . The relative humidity varies throughout the year , showing a similar pattern to the other climatic parameters . The majority of the vegetation in the plateau is grassy plain with a rich variety of grasses and dichotomous herbs , but also shrubs and some trees . The valleys contain fertile soils where vegetables , cereals and fruits are grown . Specimens of T . infestans included in the present study were a mixture of nymphs and adults , collected from inside as well as the immediate vicinity of homes . Collections were made in the months of the Southern hemisphere summer 2002 , spring 2005 and fall 2005 . Forty-four insects came from a single corral in the community of Jackota in the province of Zudañez , 78 insects were collected in the community of Zurima in the province of Oropeza , and 37 were collected in Sucre the capital and main city of Chuquisaca located in the province of Oropeza . The remaining 71 insects came from collections in 20 localities throughout Chuquisaca . All insects included in the study were identified as T . infestans using taxonomic keys [11] . Insects from the first collection were frozen live . Those from subsequent collections were placed in 95% ethanol while alive . Specimens then were sent to Vermont , USA for molecular analysis . DNA was extracted from three legs or 25 mg of tissue obtained from the posterior part of the abdomen of a given specimen using the Qiagen DNeasy DNA extraction kit ( Qiagen , Inc . , Valencia , CA ) . Care was taken to avoid sampling from the mid-abdomen as the stomach may inhibit the PCR reaction [12] . We investigated population genetic structure at both ecological and geographic levels ( Table 1 a–e ) . Ecological grouping included: Eastern , low altitude ( 97 individuals ) vs . Western , high altitude ( 133 individuals ) regions ( Table 1 a ) and domestic ( 36 individuals ) vs . peri-domestic habitats ( 42 individuals ) within Zurima ( Table 1 e ) . The geographic groupings included: among 5 communities within a 100 Km diameter with a total of 193 individuals ( Table 1 b ) , among 7 households within a 750 m diameter ( defined as a house and the associated peri-domestic buildings and corrals , with 4 , 7 , 14 , 7 , 6 , 11 and 3 insects respectively ) within Zurima ( Table 1 c ) , and 36 nymphs from a single corral in Jackota ( Table 1 d ) . Four insects from a household in Zurima were collected in 2002 before spraying , all other specimens were sampled in 2005 , up to 6 months after spraying and were re-infesting insects .
There was significant genetic differentiation among populations based on RST and FST estimates for all hierarchical levels analyzed ( Table 2 ) . Between low altitude East and high altitude West , RST and FST are statistically significant ( RST = 0 . 08 , FST = 0 . 02 ) ; both measures are also significant among the five communities <100 Km apart ( RST = 0 . 12 , FST = 0 . 06 ) and among houses in Zurima ( RST = 0 . 14 , FST = 0 . 07 ) . We also observed significant differentiation between domestic and peri-domestic populations within the community of Zurima ( RST = 0 . 05 , FST = 0 . 03 ) . Although East and West were genetically differentiated , we did not observe a trend towards higher diversity at higher altitude when we compared the Western populations with a mean altitude of 2600 m , which comprises the provinces of Oropeza and Yamparaez , with the Eastern populations having a mean altitude of 2300 m which includes the provinces of Zudañez , Belisario Boeto , Azurduy , Tomina and Hernando Siles . The mean number of alleles per locus was 15 . 3±2 . 23 and 13 . 6±2 . 31 at the high and low altitudes respectively ( t-test , P>0 . 05 ) . The dendogram based on Nei's genetic distances showed a cluster comprising populations from Zurima , El Chaco and Sucre differentiated from a sister cluster with the Jackota population ( Figure 2 ) . These two clusters were well differentiated from a cluster containing populations from the more distant Serrano ( Table 3 ) . Pairwise estimates of RST and FST among communities ( Table 4 ) support the conclusion that El Chaco , Zurima and Sucre are genetically similar to each other and that these communities differ from Jackota and Serrano . Within the town of Zurima , the estimates of RST and FST among the 7 households are shown in Table 5 . With respect to RST , households 4 and 5 are the most different from other households . These households represent peri-domestic samples and their difference from the other households is also shown by the significant difference among habitats ( Table 2 e ) . Five clusters were identified among the 5 communities ( Table 3 ) . When assigning individuals to genetic populations based on these communities , 78–86% of the individuals were assigned . The clusters represent insects with similar genotypes . Assignment tests can be viewed in terms of the number and evenness of communities in a single cluster and with respect to the number and evenness of clusters represented in a single community . Cluster 1 was a mixture of insects from the three close localities , Sucre , El Chaco and Zurima . The other four clusters contained insects from primarily one locality: clusters 2 and 3 were primarily from Zurima ( 24/29 = 83% and 24/28 = 86% respectively ) ; cluster 4 from Jackota ( 32/33 = 97% ) and cluster 5 from Serrano ( 18/28 = 64% ) . About 15–20% of the insects from each community were not assigned . From the community perspective , most of the insects from four of the communities are from a single genetic group: Jackota ( 73% from cluster 4 ) , Sucre ( 67% from cluster 1 ) , El Chaco ( 56% from cluster 1 ) and Serrano ( 72% from cluster 5 ) . Zurima contains a mixture of groups , 13% group 1 , 31% from group 2 and another 31% from group 3 . At the household level , five genetic clusters were identified from the seven households ( Table 6 ) . Insects from households 1 , 2 , 5 and half of those from household 7 were collected in peri-domestic settings , all the others came from domestic structures . The assignment test was quite successful for some households ( 100% assigned ) , yet for other households none of the insects were assigned . There does not seem to be any tendency for insects collected from domestic vs . peri-domestic sites to be assigned . With respect to the life stage and household of origin for the insects in each cluster , clusters 3 and 5 were mostly from a single household ( 86% and 100% respectively ) with cluster 5 being composed only of the most geographically isolated insects and cluster 3 containing 5 nymphs and one adult from household 3 along with one adult male from household 6 . Cluster 2 contains insects from 5 of the 7 households and cluster 1 contains insects coming from 4 households . Cluster 4 contains only nymphs , five from household 3 and four from household 2 . The fifth cluster was a mix of adults and nymphs coming exclusively from Z-6 . All four insects from the pre-spraying collection were not assigned to any cluster ( Z-1 ) ( Table 6 ) . Relatedness of insects in nine out of seventeen houses was not significantly different from 0 ( Table 7 ) . From these nine households , in six cases at least one adult was collected and in three cases only nymphs were collected . For one household ( S-1 ) , r<0 ( P<0 . 05 ) indicating significant outcrossing . For seven houses r>0 ( P<0 . 05 ) . A value of r≈0 . 25 ( half sibs ) was obtained for four households , and although the relatedness was similar in these households , the composition of the insect collection varied ( 1 site only adults 1 site only nymphs and 2 sites a mix of adults and nymphs ) . For the sites with the highest relatedness values ( r≈0 . 33 , 0 . 44 and 0 . 48 ) , in 2 houses a single adult and 2–4 nymphs were collected and for one household only nymphs were collected . The estimates of the effective number of migrants per generation , Nm , among towns <40 Km apart was higher ( 2 . 03 ) compared with those among more distant communities ( 1 . 42 ) and among houses within the town of Zurima ( 0 . 99 ) . The Mantel test of isolation by distance revealed a non-significant correlation between Slatkin's linearized FST and Nm vs . the natural log of geographic distance ( R2 = 0 . 001 , P = 0 . 294; R2 = −0 . 184 , P = 0 . 725 respectively ) . Non-significant results were also observed when applying the Mantel test for a correlation between Nei's genetic distances and geographic distances among populations ( R2 = 0 . 00056 , P = 0 . 135 ) , and altitude ( R2 = −0 . 000012 , P = 0 . 548 ) . The Mantel tests had low power because of the small samples within many of the communities .
Previous studies on population genetics and morphometry of T . infestans from Bolivia have found geographical variation in patterns of population structure in this vector; therefore we examined distinct ecological and geographic hierarchical groups ranging from a single goat corral to comparing western and eastern regions of Chuquisaca . Genetic analysis over twenty-three localities throughout the department of Chuquisaca have revealed moderate but highly significant levels of genetic variation among populations . Both FST and RST showed differentiation even within a community . Previous study in the same area using a mitochondrial cyt b gene [10] failed to verify significant genetic diversity comparing distant rural and peri-urban settings . However , significant differentiation was revealed when populations from Chuquisaca ( Andean ) were compared with non-Andean populations from Brazil , Argentina and the Bolivian Chaco . Cytogenetic [26] and allozyme [9] studies have also confirmed genetic differences between T . infestans from highlands ( >1800 m ) and lowlands ( <500 m ) . We examined insects from eastern and western Chuquisaca that significantly differ in altitude , both groups are >2000 m , and we detected significant differentiation at this ecological level . In our study , RST values were larger than FST , suggesting polymorphism is high and rates of migration are low [27] . The IAM-based estimates ( FST ) indicate lower differentiation because they do not distinguish among shared alleles in different populations that are not identical by descent . Similar values of RST and FST are only to be expected when mutation rates are negligible in comparison to migration and drift . When the SMM contributes to population differentiation , RST values should be larger than FST values [28] . When comparing the 5 communities ( Table 4 ) , in general , pairwise RST>FST suggesting that mutation contributes to differences at this geographic level . However , there is no such pattern for pairwaise RST and FST among households suggesting that mutation does not contributes much to differentiation at this level . As suggested by RST>FST , T . infestans has a low capacity for active dispersal [29] but can passively disperse over long distances when associated with human migration . It seems that this has been the structuring pattern of T . infestans in Chuquisaca . In our study , the results of the assignment of individuals to genetic clusters ( Table 3 ) shows the assignment of insects to genetic populations located >100 Km apart . Several studies using isozymes have examined population structure in T . infestans and report variation among regions in the spatial scale of population differentiation . Variation in population structure among regions was established using twelve isozymes [9] , [30] . There was significant differentiation of T . infestans populations between villages located 50 Km apart in Vallegrande , Santa Cruz yet in the Yungas of La Paz , populations only a few Km apart showed significant differences . Using 19 isozyme loci , significant differences in allele frequencies between populations separated by 20 Km were found in central Bolivia [31] , but this study failed to detect differentiation between sylvatic and domestic populations of T . infestans . By contrast , incipient differentiation between sylvatic and domestic populations was revealed using morphometry of the head capsule [32] . Other studies [33] have indicated that the panmictic unit may be no larger than a single household , based on the finding of significant differentiation within households in Yungas , Bolivia . Differences have also been detected between geographically close populations based on wing geometric morphometry [34] . The results of our study show significant population structure among communities . These results are supported by cluster analysis , which identified the geographically isolated communities as separate clusters ( Jackota and Serrano , Table 3 ) ; however the closer communities are not as genetically distinct ( Sucre , El Chaco and Zurima , Table 3 ) . If migration depends on habitat quality , when insects find favorable conditions at the microhabitat level it can reduce their dispersal tendency and consequently reduce gene flow . Within the community of Zurima we sampled 7 houses and statistical analysis estimated 5 clusters within an area of 750 m diameter . These results suggest the single household is not the panmictic unit in this area of Chuquisaca and is in accordance with a study on dispersal capacity in the towns of Trinidad and Mercedes , Argentina , that clustered the source of re-infestation at ∼500 meters [35] . The isolation-by-distance tests based on allozyme markers in populations from several areas in Bolivia and Peru found a positive correlation between genetic and geographic distances [9] . We found no evidence of isolation by distance within this area of Chuquisaca . Differences between the two studies may result because our study had low statistical power due to sampling a relatively small number of communities , few samples per community and microsatellite data , because of the high number of alleles , require large sample sizes . However , the non-significant results may also be because our study covers a small geographic area of Chuquisaca characterized by a high human migration rate in the last 40 years [36] . Previous studies [37] identified unique local characteristics in landscape and vegetation , distances between houses , the abundance of bugs and hosts , and presence of many peri-domiciliary structures in conjunction with the existence of sylvatic populations as contributing to spatial patterns of re-infestation . Identification of the source of re-colonizers can direct control programs in the surveillance phase . We have found significant differentiation at the household level in populations from Chuquisaca , Bolivia . Cluster analysis , relatedness estimates and life stage data can be combined to understand pre-spraying population dynamics and infer patterns of re-colonization . Within Zurima , individuals collected in the most geographically isolated household ( Z-6 ) were assigned to one cluster . The relatedness of insects in Z-6 was significantly greater than 0 ( Z-6 , r>0 . 17 , c . i . = 0 . 15 , Table 7 ) . Eight of the nine adults and the two nymphs in Z-6 were assigned to a single cluster , but this house also had insects from two other clusters . The reinfestation patterns for individual houses are quite variable including repeated colonization from several sources ( Z-2 , seven peri-domestic adults , r≈0 . 10 , c . i . 0 . 13 , Table 7 ) , a single multiply mated female ( S-3 , 1 adult 5 nymphs , r≈0 . 26 , c . i . = 0 . 21 , Table 7 ) , multiple colonization from a single source ( Z-5 , 3 males and 3 females , r≈0 . 23 , c . i . = 0 . 21 , Table 7 ) , recrudescence of full sibs ( Z-10 , 3 nymphs , r≈0 . 48 , c . i . = 0 . 45 , Table 7 ) and recrudescence of unrelated eggs ( Z-3 , 14 insects mostly nymphs , r≈0 . 05 , c . i . = 0 . 07 , Table 7 ) . Of course there are multiple possibilities for each household and these inferences are to show the range of possibilities , not to infer a given scenario for a specific household . The presence of adults in many households less than 6 months after spraying suggests that for many cases , structures around human habitations may be playing a key role as the source of insects invading houses . The presence of nymphs in houses where no adults were found suggests recrudescence . Hence , recrudescence from a residual population and colonists from peri-domicile structures , rather than reinvasion from surrounding localities , seems to be a probable explanation of the source of re-colonists found during surveillance activities in this area . The variety of results suggest that continuous surveillance consisting of analyzing relatedness among reinfesting insects at the household level is critical to maintain insect free houses and optimize insecticide spraying in this region .
|
Chagas disease is a protozoan infection caused by the parasite Trypanosoma cruzi . Chagas is prevalent throughout Central and South America , and it remains a chief concern in Bolivia . A movement that began in 1991 called the Southern Cone Initiative has been successful in reducing the incidence of Chagas disease in the Southern Cone countries of Argentina , Brazil , Chile , and Uruguay; but due to socio-economic and other factors , incidence remains high in Bolivia . The most important mode of transmission of T . cruzi to humans and other mammals is through feces of triatomine bugs . Thus , disease control and transmission prevention focus on elimination of triatomine vectors , and more specifically in Bolivia , it focuses on the elimination of Triatoma infestans . This study focuses on T . infestans in the Department of Chuquisaca , Bolivia . Ten highly variable microsatellite markers were used to analyze the population structure of insects collected in different towns . Statistical analyses show that T . infestans are highly structured , which means that they colonize on a small geographic scale . The results also suggest little active dispersal . These findings should be implemented during control efforts so that insecticide spraying focuses on geographic areas of colonization and re-colonization .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"infectious",
"diseases/neglected",
"tropical",
"diseases",
"public",
"health",
"and",
"epidemiology/infectious",
"diseases",
"genetics",
"and",
"genomics/population",
"genetics"
] |
2008
|
Microsatellites Reveal a High Population Structure in Triatoma infestans from Chuquisaca, Bolivia
|
Head lice , Pediculus humanus capitis , occur in four divergent mitochondrial clades ( A , B , C and D ) , each having particular geographical distributions . Recent studies suggest that head lice , as is the case of body lice , can act as a vector for louse-borne diseases . Therefore , understanding the genetic diversity of lice worldwide is of critical importance to our understanding of the risk of louse-borne diseases . Here , we report the results of the first molecular screening of pygmies’ head lice in the Republic of Congo for seven pathogens and an analysis of lice mitochondrial clades . We developed two duplex clade-specific real-time PCRs and identified three major mitochondrial clades: A , C , and D indicating high diversity among the head lice studied . We identified the presence of a dangerous human pathogen , Borrelia recurrentis , the causative agent of relapsing fever , in ten clade A head lice , which was not reported in the Republic of Congo , and B . theileri in one head louse . The results also show widespread infection among head lice with several species of Acinetobacter . A . junii was the most prevalent , followed by A . ursingii , A . baumannii , A . johnsonii , A . schindleri , A . lwoffii , A . nosocomialis and A . towneri . Our study is the first to show the presence of B . recurrentis in African pygmies’ head lice in the Republic of Congo . This study is also the first to report the presence of DNAs of B . theileri and several species of Acinetobacter in human head lice . Further studies are needed to determine whether the head lice can transmit these pathogenic bacteria from person to another .
The head louse , Pediculus humanus capitis , and the body louse , P . h . humanus , are obligatory hematophagous parasite that thrived exclusively on human blood for thousands of years [1 , 2] . The two lice are now usually considered members of a single species as opposed to separate species [3 , 4] , each louse lives and multiplies in a specific ecological niche: hair for head lice and clothing for body lice [5 , 6] . Molecular analysis of mitochondrial genes has permitted the classification of Pediculus humanus into three several clades or haplogroups , referred to as A , B , and C [1 , 2 , 7 , 8 , 9] . Haplogroup A is the most common , and possesses a global distribution , including both head and body lice [1 , 2 , 6 , 8 , 9] . Clade B comprises only head lice , is confined to the New Word , Europe , Australia and was recently reported in North and South Africa [2 , 6 , 10 , 11] . Clade C includes only head lice and is mainly found in Africa and Asia [2 , 5 , 9 , 10] . Most recently , a novel clade D , comprising both head and body lice , was described in Democratic Republic of the Congo [6] . Prior research suggested that the known lice clades evolved on different lineages of Homo , similarly to those which are known to have existed 2 . 3 to 0 . 03 million years ago ( MYA ) [1 , 11] , and accordingly their geographic distribution may provide information regarding the evolutionary history of the lice as well as their human hosts [1 , 2 , 22] . Clade A lice are most likely to have emerged in Africa and to have evolved on the host linage that led to anatomically modern humans ( Homo sapiens ) , showing the signs of a recent demographic expansion out of Africa about 100 , 000 years ago , first to Eurasia and subsequently to Europe , Asia , and the New World [1 , 5 , 12] . Haplogroup B diverged from haplogroup A between 0 . 7 and 1 . 2 MYA and may have evolved on archaic hominids , such as the Homo sapiens neanderthalensis , who spread across Europe and Asia , only becoming associated with modern humans during the period of overlap as the result of a recent host switch [1 , 5 , 12] . Head lice are one of the most prevalent parasitic infestations in contemporary populations , particularly in children . They often cause intense itching and , in some cases , insomnia . As a result , they represent a major economic and social concern worldwide [6 , 13 , 14] . Body lice , unlike head lice , are nowadays less prevalent and tend to appear mainly in indigent individuals living in poor sanitary conditions [6 , 9 , 13] . They do , however , present a far more serious threat to public health because they transmit at least three deadly bacterial pathogens that have killed millions of peoples , namely: Rickettsia prowazekii , Bartonella quintana , and Borrelia recurrentis , responsible for epidemic typhus , trench fever , and relapsing fever , respectively [5 , 9 , 13] . Body lice are also suspected of transmitting the agent of plague , Yersinia pestis and the nosocomial pathogen , Acinetobacter baumannii [6 , 15 , 16] . Until recently , it was believed that head lice cannot transmit louse-borne diseases [17] . Recently , however , its status as a vector of pathogens has been brought into question , since , they have been found to carry the DNA of B . quintana , B . recurrentis , A . baumannii , and Y . pestis in natural settings [6 , 18 , 19 , 20 , 21 , 22 , 23] . Furthermore , experimental infections have shown that head lice may also act as a vector of louse-borne diseases [24 , 25] , justifying a detailed understanding of their genetic diversity and distribution worldwide . In Central Africa , studies on head lice , particularly those involving indigenous individuals , have received little prior attention . Of these indigenous populations , the African Pygmies are hunter-gatherers who live scattered in the equatorial forest . They are characterized by having a very short stature [26] . The Eastern and Western Pygmies represent the two principal groups of African Pygmies [26] . The Western group is estimated to include 55 , 000 individuals living in the Western Congo basin , across the countries of Cameroon , Republic of Congo , Gabon and Central African Republic , and its subgroups are identified by different names , including the Binga , Baka , Biaka and Aka or Atsua [26] . Furthermore , the detection of B . recurrentis in African lice remains limited to only a small number of countries . Currently , this bacterium is endemic in Eastern Africa ( Ethiopia , Eritrea , Somalia , and Sudan ) with the highest number of cases observed in Ethiopia , where it is the seventh most common cause of hospital admission and the fifth most common cause of death [27 , 28] . Nevertheless , this borreliae has not been reported in any of the Central African countries cited above . In this work , we aimed to study the genetic diversity of head lice collected from African Pygmies in the Republic of Congo and to look for louse-borne pathogens in these lice .
This study was approved by the Health Ministry of the Republic of Congo ( 000208/MSP/CAB . 15 du Ministère de la Santé et de la Population , 20 August 2015 ) . All necessary permits were obtained from the individuals involved or their legal representatives in the case of children . All permissions were granted orally , because the participants are illiterate . The representatives of a local Health Center and the village elders accompanied the researchers to ensure that information was correctly translated into local languages and that the villagers were willing to take part in the study . A total of 630 head lice samples were collected from 126 apparently healthy authochthonal individuals ( pygmies ) in the Republic of Congo ( Congo-Brazzaville ) in August 2015 . The collections were conducted in three different villages: i ) Thanry-Ipendja , where 137 lice were isolated from 18 people , ii ) Pokola , where 163 lice were isolated from 36 people , and iii ) Béné-Gamboma , where 330 lice were isolated from 72 people ( Fig 1 ) . All the sampled individuals were thoroughly examined for the presence of both head and body lice . All visible head lice were removed from hair using a fine-tooth comb . Lice were then collected from the clean white tissue with forceps . No body lice were found during the examination . All the lice were preserved in 70% ethanol and transported to our laboratory in Marseille ( France ) . The head lice specimens were removed from the 70% ethanol , washed three times in distilled water , and cut in half . The genomic DNA of each half louse was extracted using a DNA extraction kit , QIAamp Tissue Kit ( Qiagen SAS , Courtaboeuf , France ) with the EZ1 apparatus following the manufacturer’s protocols . The extracted head lice DNA was assessed for quantity and quality using a Nano Drop spectrophotometer ( Thermo Scientific , Wilmington , United Kingdom ) . The genomic DNA was stored at -20°C under sterile conditions until the next stage of the investigation . The qPCR was performed to screen all lice samples using previously reported primers and probes for Borrelia spp . , Bartonella spp . , Acinetobacter spp . , Rickettsia spp . , Rickettsia prowazekii , Y . pestis , and Anaplasma spp . ( Table 1 ) . All qPCRs were performed using a CFX96 Real-Time system ( Bio-Rad Laboratories ) and the Eurogentec Master Mix Probe PCR kit ( Eurogentec ) . We included the DNA of the target bacteria as positive controls and master mixtures as a negative control for each test . We considered samples to be positive when the cycle’s threshold ( Ct ) was lower than 35 Ct [38] . To identify the species of bacteria , all positive samples from qPCRs for Acinetobacter spp . and Borrelia spp . were further subjected to standard PCR , targeting a portion of the rpoB gene ( zone1 ) and a portion of the flab gene , respectively , using the primers and all conditions as described previously [33 , 36] . Successful amplification was confirmed via gel electrophoresis and amplicons were prepared and sequenced using similar methods as described for cytb gene for lice above . For comparison , the head lice DNA sequences obtained in this study were combined with the 30 cytb haplotypes reported by Drali et al . [39] . We then complemented this dataset with newly available sequences in GenBank , then assigned them to haplotypes using DnaSP v5 . 10 [40] . Finally , we created a dataset that consisted of 51 haplotypes . These haplotypes span 41 geographic locations ( countries ) in five continents ( S1 Table ) . In order to investigate the possible relationships between the haplotypes , the median-joining ( MJ ) network using the method of Bandelt was constructed with the program NETWORK4 . 6 ( www . fluxus-engineering . com/sharenet . htm ) [41] . Phylogenetic analyses and tree reconstruction were performed using MEGA software version 6 . 06 [42] with 500 bootstrap replications .
A total of 160 head lice cytb sequences were analyzed in this work yielding 83 variable positions defining 15 different haplotypes , including 11 new ones: five from haplogroup A ( 35 . 7% ) , four from haplogroup D ( 28 . 5% ) , and six from haplogroup C ( 42 . 8% ) ( Table 3 ) . These haplotypes , together with references from all the body and head lice haplogroups were used to construct a maximum-likelihood ( ML ) tree and a median-joining ( MJ ) network ( Figs 2 and 3 ) . ML and MJ analyses had similar results: all the cytb sequences were divided across the four major supported clades , represented by four connected subnetworks distinct groups as shown in the MJ network ( Fig 2 ) corresponding to the known clades: A , D , B , and C . The 15 haplotypes in our study fell into all of the three haplogroups , A , D , and C . The haplogroup A subnetwork was star-like in structure , with the most prevalent and widespread haplotype being A5 ( 78% of locations and 45 . 4% of the 1 , 005 analyzed human lice ) in the center . 24 ( 15% ) of our cytb sequences have this A5 haplotype and are all from the village of Béné-Gamboma , while a total of 64 ( 40% ) cytb sequences ( 34 sequences from Thanry-Ipendja and 34 sequences from Pokola villages ) have the A17 haplotype , which is the second most common A-haplotype and derived from the A5-haplotype by one mutation step . The remaining five clade A sequences , four from Thanry-Ipendja and five from Pokola , defined three novel haplotypes , named here A57 , A58 , and A59 . These three novel haplotypes derived from A17-haplotype by one mutation step . Haplogroup D , which is genetically close to A , only consists of haplotypes from Ethiopia and the Republic Democratic of Congo ( RDC ) . The 45 ( 45/160 ) pygmy head lice sequences within clade D defined four haplotypes , of which three are novel ( named here: D71 , D72 , D73 ) , while the fourth haplotype possessed D65 haplotype from RDC . The clade C , representing the most divergent lineage in which two sub-clades can be defined , here referred to as sub-clade C1 , which consists of head lice from Ethiopia , France and the Asian continent , and sub-clade C2 , which consists of head lice from Senegal and Mali . These two subclades are separated by 12 mutations steps . Interestingly , all 45 ( 45/160 ) pygmy head lice sequences within clade C yielded six novel haplotypes , named here as C74-C79 and are parts of sub-clade C1 . In this study , the qPCR investigation of all 630 lice samples for Bartonella spp . , Rickettsia spp . , R . prowazekii , Y . pestis , and Anaplasma spp . produced no positive results . However , we obtained positive results when testing for the presence of Borrelia spp . and Acinetobacter spp . The DNA of Borrelia spp . was detected in 11/630 ( 1 . 74% ) head lice collected from 7/126 ( 5 . 55% ) individuals . All Borrelia-positive lice were clade A and found only in Pokola . The DNA of Acinetobacter spp . was detected in 235/630 ( 37 . 3% ) head lice collected from 93/126 ( 73 . 8% ) people . Of the 235 positive lice , 176 ( 26% ) were clade A , 24 ( 3 . 8% ) clade D , and 47 ( 7 . 5% ) clade C . Sixty-one of these infected lice were from Pokola , forty-one from Thanry-Ipendja , and one hundred and thirty-three from Béné-Gamboma .
Here , we report the first molecular data on human head lice , P . h . capitis , infesting the pygmy population in the Republic of Congo in Western Africa . In this study , we established and evaluated for the first time , qPCR assay based on two duplex designed from the cytb gene , which is very well established in the study of lice , in order to identify all known clades of P . humanus . The assay adopted herein proved itself to be fast , specific , sensitive and fully compatible when routinely analyzing large collections of lice specimens . The mtDNA analysis of 630 head lice , collected from 126 pygmies , showed the presence of three major mitochondrial haplogroups: A , C and D , indicating high mtDNA diversity among the head lice studied . Haplogroup A was the most prevalent ( 56% ) followed by haplogroup C ( 5% ) . The data confirm that clade A has worldwide distribution , as reported by others [6 , 8 , 9 , 10] . Previous studies reported that clade C is limited to Nepal and Thailand [1 , 5 , 23] , Ethiopia , Senegal and Mali [5 , 9 , 18 , 22]; this is the first report of clade C which has been found in the Republic of Congo . The remaining samples ( 10 . 3% ) were from new haplogroup D , which is known only to exist in Democratic Republic of the Congo and Ethiopia [2 , 6] . In addition to inter-haplogroup diversity , P . humanus also presents intra-haplogroup diversity , illustrated by many distinct A , B and C haplotypes [2 , 12 , 39] . These results are supported by our finding , that , of the 160 head lice cytb sequences analysed , 15 different haplotypes were identified , of which 11 were novel . B . recurrentis is the known causative agent of relapsing fever which , if untreated , can be fatal in up to 40% of patients [13 , 43 , 44] . It has long been established that body lice are the main vector for this bacterial pathogen [13 , 27] . In the present study , the DNA of B . recurrentis was detected in 10/630 ( 1 . 58% ) head lice belonging to clade A collected from 6/126 ( 4 . 76% ) individuals . Specifically , all positives cases were only found in Pokola , suggesting that a small , unnoticed outbreak may have occurred in the population in this area . This is the second report of the presence of B . recurrentis DNA in human head lice . Recently , this bacterium was also detected in 23% of head lice clade C from patients with louse-borne relapsing fever in Ethiopia and , because these patients were also infested with body lice , the authors hypothesize that head lice might be contaminated by blood that is infected with B . recurrentis [21] . In this study , the discovery of B . recurrentis in the clade A head lice , the same clade that includes body lice , and the absence of body lice may support the hypothesis that B . recurrentis may be transmitted by clade A head lice . Nevertheless , evidence for the presence of the DNA of this bacterium in head lice by PCR cannot distinguish between transient infections , accidentally acquire the pathogen from the blood of infected individuals , and those established in a competent vector , maintain and transmit the pathogen . Further studies are needed to determine whether the head louse can act as a vector of B . recurrentis . Interestingly , one of the Borrelia-positive lice was identified as B . theileri . This is the first report of the presence of the DNA of this species in human head lice . B . theileri is a spirochete that causes borreliosis in cattle , a relapsing fever-like illness , transmitted by hard ticks , such as Rhipicephalus ( Boophilus ) [45] . This infection can be considered as being rediscovered , appears to exist in regions where diagnostic ability is limited and its impact on livestock is largely unexplored [45] . In this study , two hypotheses can arise from the detection of B . theileri in human head lice . The first one is that the presence of this bacterium results from environmental and/or laboratory contaminations . This hypothesis is hardly possible , because , our work was carried out in a laboratory where B . theileri had never been worked on , nor had B . theileri DNA been extracted . Indeed , each PCR assay was systematically validated by the presence of positive and negative controls . Moreover , our collection contains lice only and didn’t contain another specimens like ticks that could be an important source of environmental contamination . The second hypothesis is that , as head lice feed only on human blood [5] , the acquired infection would be from the blood of patients with ongoing bacteremia . Although , humans infected with this spirochete have not been described in the literature , the transmission of this pathogen to humans may not be ruled out . Moreover , the sequence generated in this study was more similar by flaB sequence comparison to those reported from Ornithodoros sp . soft tick ( GenBank KP191621 ) collected from cave in Israel , than , those reported from Rhipicephalus hard tick ( GenBank KF569936 ) from Mali , as shown in the phylogenetic tree ( Fig 4 ) . Ornithodoros ticks can feed from multiple warm-blooded vertebrates , including humans , and are known to transmit several species of Borrelia to humans [27 , 43] , thus taking in consideration that the epidemiology of B . theileri is not yet completely discovered , hypothetically it may be transmitted to humans . Finally , if our hypothesis of B . theileri bacteremia in persons harboring head lice is true , this may merely reflect ‘accidental spill-over’ from animal hosts infection , such phenomena has already been described in the literature , with the finding of the DNA of B . duttonii , the species that is only know to infect ticks and humans , in chickens and swine living close to their human owners [43] . Findings from this study also show widespread infection of head lice with several species of Acinetobacter . In total , eight Acinetobacter species were detected in 144 samples; A . junii was the most prevalent , followed by A . ursingii , A . baumannii , A . johnsonii , A . schindleri , A . lwoffii , A . nosocomialis and A . towneri . The DNA of A . towneri was only found in clade C head lice , the DNA of A . lwoffii was only found in clades A and D , while the DNA of the remaining species was found in all three clades A , D and C . Previous studies demonstrated that A . baumannii is the most commonly found species in body and head lice [23] , as shown by its detection in 21% of body lice collected worldwide [15] , in 33% of head lice collected from Parisian elementary school children , belonging to the clade A [19] and in 71% body and 47% head lice collected from healthy individuals from Ethiopia [20] . Another study , performed in head lice samples collected from elementary school children in Thailand , showed the presence of the DNA of three Acinetobacter species in 3 . 62% head lice belonging to both clade A and C . The Acinetobacter species identified were A . baumannii , A . schindleri and A . radioresistens [23] . When comparing the panel of Acinetobacter species found in all these studies with our findings , A . radioresistens was the only species that we did not identify in our head lice specimens . Conversely , our sampling showed , for the first time , the presence of the DNA of A . junii , A . ursingii , A . johnsonii , A . lwoffii , A . nosocomialis and A . towneri in human head lice , but further study is needed to determine the significance of this finding . Furthermore , it is still unknown how these lice acquire their Acinetobacter infections . Some authors have argued that the infection could occur after the ingestion of infected blood meal from individuals with ongoing bacteremia , or may possibly be derived from superficial contamination through human skin while feeding [15] . An experimental study showed that the human body louse , feeding on bacteremic rabbits , is able to acquire and maintain a persistent life-long infection with A . baumannii and A . lwoffii [46] . Furthermore , another study performed a comparison between two sequenced genomes of A . baumannii and showed that the A . baumannii SDF strain , isolated from a human body louse , had several hundred insertion sequence elements which have played a crucial role in its genome reduction ( gene disruptions and simple DNA loss ) compared to the human multidrug-resistant A . baumannii AYE strain , and also been shown to have low catabolic capacities , suggesting the specific adaptation of this strain to the louse environment [47] . However , Acinetobacter species are widespread in nature ( water , soil , living organisms , and the skin of patients and healthy subjects ) [47] , and because the frequency of with which these species associate with the skin of pygmy population is unknown , it is not possible to rule out the infection of lice by external contamination . Clinically , A . baumannii is known to be a major cause of nosocomial infections in humans and it is an increasing public health concern due to the increasing resistance to antibiotic treatment which has been identified worldwide [47] . Other Acinetobacter species include A . lwoffii and A . junii are also often identified as the cause of infection in humans [48] . However , it still not clear whether these Acinetobacter strains present in lice are the same as those that are responsible for human infections [20] .
In conclusion , the qPCR adopted in this study proved to be a fast , sensitive and specific tool that is fully compatible when routinely analyzing a large collections of lice specimens . Our results showed the presence of three major mitochondrial haplogroups: A , C and D , indicating high mtDNA diversity among the pygmy head lice studied . We identified the presence of a dangerous human pathogen , B . recurrentis , the causative agent of relapsing fever , in ten clade A head lice , which had not previously been reported in the Republic of Congo . Findings from this study also show the widespread infection of head lice with several species of Acinetobacter . Despite several investigations into the transmissibility of numerous infectious agents , no conclusive evidence has demonstrated the transmission of disease by head lice . That said , we believe that pathogens detected in head lice may be an indirect tool for evaluating the risk of louse-borne diseases in humans .
|
Head lice , Pediculus capitis humanus , and body lice , Pediculus h . humanus , are obligatory ectoparasites that feed exclusively on human blood . Currently , the body louse is the only recognized vector of at least three deadly bacterial pathogens that have killed millions of peoples , namely: Rickettsia prowazekii , Bartonella quintana and Borrelia recurrentis , responsible for epidemic typhus , trench fever and relapsing fever , respectively . In this work , we aimed to study the genetic diversity of head lice collected from African Pygmies in the Republic of Congo and to look for louse-borne pathogens in these lice . We detected B . recurrentis in head lice belonged to clade A that is prevalent in the Republic of Congo . Our study also show , for the first time , the presence of DNAs of B . theileri and several species of Acinetobacter in human head lice . Despite several investigations into the transmissibility of numerous infectious agents , no conclusive evidence has demonstrated the transmission of disease by head lice . That said , we believe that pathogens detected in head lice may be an indirect tool for evaluating the risk of louse-borne diseases in humans .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion",
"Conclusions"
] |
[
"invertebrates",
"borrelia",
"infection",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"evolutionary",
"biology",
"acinetobacter",
"infections",
"pathogens",
"population",
"genetics",
"microbiology",
"animals",
"bacterial",
"diseases",
"body",
"lice",
"population",
"biology",
"insect",
"vectors",
"bacteria",
"haplogroups",
"bacterial",
"pathogens",
"infectious",
"diseases",
"medical",
"microbiology",
"epidemiology",
"microbial",
"pathogens",
"borrelia",
"insects",
"disease",
"vectors",
"head",
"lice",
"arthropoda",
"acinetobacter",
"haplotypes",
"genetics",
"biology",
"and",
"life",
"sciences",
"lice",
"organisms"
] |
2016
|
Head Lice of Pygmies Reveal the Presence of Relapsing Fever Borreliae in the Republic of Congo
|
Retrotransposons are mobile genetic elements abundant in plant and animal genomes . While efficiently silenced by the epigenetic machinery , they can be reactivated upon stress or during development . Their level of transcription not reflecting their transposition ability , it is thus difficult to evaluate their contribution to the active mobilome . Here we applied a simple methodology based on the high throughput sequencing of extrachromosomal circular DNA ( eccDNA ) forms of active retrotransposons to characterize the repertoire of mobile retrotransposons in plants . This method successfully identified known active retrotransposons in both Arabidopsis and rice material where the epigenome is destabilized . When applying mobilome-seq to developmental stages in wild type rice , we identified PopRice as a highly active retrotransposon producing eccDNA forms in the wild type endosperm . The mobilome-seq strategy opens new routes for the characterization of a yet unexplored fraction of plant genomes .
Transposable elements ( TEs ) are major players in the evolution of animal and plant genomes [1–3] . The observation of both a complex epigenetic repression of TE expression and a large compartment occupied by TE copies in most sequenced eukaryotic genomes reflects a fine-tuned interaction between TEs and their host genomes [1–4] . TE proliferation in genomes leads to increased genomic diversity through mutations , genomic rearrangements like translocations or inversions [2] , and epigenetic modifications [5] . This proliferation can also have a regulatory effect on gene expression that has been proposed to potentially result in adaptive traits [1 , 6 , 7] . According to their mode of transposition , TEs are organized into two main classes: retrotransposons ( RTs ) and DNA transposons ( DNA-TEs ) . RTs multiply using a « copy and paste » strategy mediated by an RNA-intermediate , whereas DNA-TEs use a « cut and paste » mechanism [8] . During their life cycle TEs thus can exist as integrated DNA , mRNA and extrachromosomal linear DNA ( S1 Fig ) . The extrachromosomal linear form , typical of actively proliferating TEs , can be detected by the host and may be circularized by DNA repair processes . The non-homologous end-joining mechanism and/or homologous recombination between flanking repeat sequences have been proposed to promote the circularization of extrachromosomal DNA into extrachromosomal circular DNA ( eccDNA ) [9–12] . There is no evidence that these eccDNAs can be re-integrated into the plant genome . Thus the formation of eccDNAs by the host could be a mechanism to limit the number of new insertions of active TEs in the genome ( S1 Fig ) . Different types of active TEs have been detected as eccDNAs in plants such as Tto1 [13] , Mu [14] and Ac/Ds [15] , however no genome-wide analysis of these forms has been performed yet . The mobilome consists of all mobile genetic elements in a cell that can be plasmids in prokaryotes or TEs in eukaryotes [16] . We will hereafter refer to the extrachromosomal forms of TEs as the reverse-transcribed mobilome . Multiple approaches have been used to identify actively proliferating TEs at different steps of their life-cycle: ( 1 ) positional cloning of genes altered by a TE insertion ( for example in rice the hAT DNA-TE [17] or the Long Terminal Repeat RT ( LTR-RT ) Houba [18] , ( 2 ) search for TE-insertion polymorphisms using transposon display on candidate TEs ( for example rice mPing and Pong [19] , ( 3 ) transcription studies on candidates TEs using primers targeting conserved domains , for example rice LTR-RT Tos17 [20] or through genome-wide transcriptomic analyses , for example the LTR-RT Lullaby in rice calli [21] . Today the most advanced technique to identify actively proliferating TEs in species where the genome sequence is available consists of whole-genome resequencing and detection of TE-associated polymorphisms using paired-end mapping [22–24] . The techniques listed above have important limitations . The analysis of transcripts by RNA-seq allows the description of transcriptionally active retrotransposons but does not take into account their capacity to produce proteins . As transcription is the first step in a retrotransposon life cycle , most copies do not go further this point , either because of post-transcriptional gene silencing activities or because they have accumulated mutations that prevent the translation of mature proteins , although some TEs with non functional proteins might parasite other TEs [25 , 26] . The analysis of neo-insertions through genome resequencing is very powerful to reduce the complexity of transcriptionally active TEs to the ones that effectively produce new insertions . This approach detects breakpoints between neo-insertions and a reference genome and thus requires a high sequencing coverage more adapted to small genomes . Furthermore only fixed , transgenerational neo-insertions can be detected with high accuracy . Finally , despite the numerous pipelines developed to characterize these neo-insertions [27 , 28] , only insertions into non repetitive regions of the genome can be accurately detected , leaving a large part of the structural variations caused by TEs undetectable . Alternative approaches initially developed in mammals , such as retrotransposon-capture sequencing , consist in the enrichment and identification of the flanking sequences of a particular retrotransposon [28–30] , but these techniques require prior knowledge of the active TE families in the species of interest . We therefore endeavor to develop a genome-wide strategy that could efficiently track potentially active TEs without full genome resequencing . We sought to take advantage of the presence of circular forms of active TEs in the eccDNA compartment to identify active TEs in plants . Extrachromosomal DNA circles were identified decades ago in Drosophila [9 , 31] and observed by electron microscopy in Vigna radiata [32] and by two-dimensional gel-electrophoresis in carcinogen-treated cells [33] and in plants [34] . These eccDNAs can be formed by homologous recombination between adjacent repeats such as amplified genes [35] , tandem repeats ( satellite , telomeric , centromeric and ribosomal repeats ) [34 , 36] or they can result from the linear extrachromosomal forms of active TEs [37] . These eccDNAs are ubiquitous elements and heterogeneous populations of eccDNAs seem to be present in all eukaryotic organisms [38] . Recently , sequencing of eccDNAs was experimented in mouse cells where microDNAs originating from chromosomal micro-deletions at specific gene loci [39 , 40] were identified . Numerous eccDNAs were detected in yeast cells [41 , 42] , although no new active TE could be identified . Therefore , the abundance and identity of eccDNAs specifically resulting from the circularization of extrachromosomal TE DNA in multicellular organisms is not well documented . Here , we used the identification of TE eccDNA as a tool to investigate TE activation in plants and developed a dedicated computational pipeline to address this question . We analyzed the active mobilomes from the two plant species Arabidopsis thaliana and Oryza sativa . As a proof of concept , we selected plant material where active TEs had previously been identified: a partially hypomethylated line for A . thaliana [43] and a callus tissue for O . sativa . Our mobilome-seq analyses clearly identified the two known active LTR-RTs EVD [44] and Tos17 [45] , in A . thaliana and O . sativa samples respectively , in their eccDNA forms . To investigate novel TE activity we applied mobilome-seq to wild type rice seeds and identified PopRice LTR-RTs as producing large amounts of eccDNAs specifically in the endosperm tissue . We propose that the mobilome-seq strategy could help identifying mobile TEs in different species to better understand the impact of the active mobilome on the host genome .
In order to isolate and to sequence eccDNAs , total DNA was first extracted from plant tissues ( Fig 1 ) . Linear genomic DNA was digested with an exonuclease and the remaining eccDNA molecules were then amplified by rolling circle amplification ( RCA ) using random primers . This method does therefore not require any a priori knowledge on TEs for a given sample . We first performed this experiment on samples from A . thaliana Columbia wild type plants as a negative control ( Col WT ) and on an epigenetic recombinant inbred line ( epiRIL12 hereafter called epi12 ) where an hypomethylated retrotransposon ( EVD/ATCOPIA93 ) was detected as actively proliferating [44] , as a positive control . Southern blot validation assays using an EVD specific probe were performed to analyze the enrichment of eccDNAs before and after the RCA step ( Fig 2A ) . A signal corresponding to digested EVD eccDNAs was detected in samples from siliques and flowers from epi12 plants after RCA , but not in samples from WT plants . No signal could be detected in the absence of RCA indicating that most genomic DNA had been degraded after the exonuclease treatment . We used this material for high throughput sequencing . We performed mobilome-seq on Col WT and epi12 siliques samples as shown in Fig 1 . After mapping the reads on the reference genome of A . thaliana we detected peaks of high coverage in both WT and epi12 mobilome-seq libraries ( S2 Fig ) corresponding to ribosomal DNA ( rDNA ) loci that are known to produce eccDNAs [34] . All peaks of high coverage corresponding to TEs in both WT and epi12 are listed in S5 Table . In particular , peaks corresponding to EVD were specifically detected in epi12 ( Fig 2B , S3A Fig and S5 Table ) . EVD is a 5 , 3 kilobases ( kb ) -long LTR-RT present in two full-length copies in the genome of A . thaliana ecotype Columbia . EVD is transcribed and mobilized in met1-derived epiRILs [44] and produces eccDNA copies [46] . Due to the repetitive nature of TEs , reads corresponding to EVD eccDNA can map against full-length and also truncated copies present in the genome explaining why all regions corresponding to EVD are more or less covered . Nevertheless , the two full-length copies ( on chromosomes 1 and 5 ) are the most significantly covered with a p-value < 10−8 ( S3B and S3C Fig ) . The EVD locus on chromosome 5 is highly covered in the epi12 mobilome-seq library compared to the WT library , with a depth of coverage ( DOC ) of 3500X versus 1X , respectively ( Fig 2C ) . To further identify the presence of reads corresponding to eccDNA junctions , we specifically detected split reads ( SRs ) as paired-reads that are not correctly mapped onto the reference genome ( see Methods ) . We could detect a high number of SRs at both 5’ and 3’ ends of EVD in the epi12 mobilome-seq data compared to WT ( Fig 2C and S4 Fig ) suggesting the presence of reads spanning the circular junction . A closer examination of some of these reads revealed that they indeed correspond to 2LTR junctions ( S5 Fig ) . While 142 TEs were detected as overexpressed in epi12 at the transcriptional level [47] , the mobilome-seq data suggest that only EVD produce circular copies ( S6 Fig ) . We then analyzed mobilome-seq libraries from O . sativa ssp japonica cv Nipponbare , a species with a larger genome ( 400Mb ) than A . thaliana ( 135Mb ) and a three times bigger proportion of TEs ( 45% in O . sativa against 15% in A . thaliana ) , using both leaf material and callus tissue . TEs with high coverage in O . sativa mobilome-seq libraries are listed in S5 Table . More specifically , peaks corresponding to the Tos17 family were detected in the mobilome-seq libraries of callus tissue but not in leaves ( Fig 3A and 3B , S7 Fig ) . Tos17 is a 4 , 1 kb-long LTR-RT present in two copies in the O . sativa genome ( on chromosomes 7 and 10 ) , the copy on chromosome 7 being active in calli [13] . The DOC analysis indicated a clear enrichment ( DOC = 200X ) at the Tos17 locus on chromosome 7 in the callus mobilome-seq library compared to the leaf mobilome-seq library ( <1X ) ( Fig 3B and S7B Fig ) . SRs were detected on both ends of Tos17 suggesting the presence of reads spanning the junction . The presence of Tos17 eccDNAs was confirmed by an inverse PCR assay ( Fig 3C ) and a closer inspection of SRs identified reads spanning the 2LTR-circle junction ( S8A Fig ) . Moreover we have also analyzed the coverage of Lullaby , a LTR-RT active in calli [21] . A low coverage from 10X to 12X was detected in the callus mobilome-seq library and the presence of reads spanning junction of Lullaby eccDNAs was confirmed ( S8B Fig ) . Altogether , these results show that known active LTR-RTs could be detected using the mobilome-seq approach , suggesting that this technique can be used to identify new active TEs in plants . Epigenomic studies have revealed a release of TE transcriptional silencing during plant development [48–51] . In a first attempt to understand the possible role of TEs reactivation during plant development , we performed mobilome-seq analyses on DNA extracted from whole rice seeds . Some TE regions were significantly highly covered in this mobilome-seq library ( Fig 4A and 4B , S9 Fig ) , most of these regions corresponding to TEs belonging to a single subfamily of Osr4 . Osr4 [52] is a large family of 5 . 7 kb-long LTR-RTs comprising 47 members in the O . sativa ssp japonica cv Nipponbare genome ( Fig 4C and S3 Table ) . To differentiate Osr4 active and non-active members we hereafter refer to the subfamily enriched in the seed mobilome-seq library as the PopRice family . The PopRice family is composed of 17 full-length copies in the reference genome . Some of these loci are highly covered in the seed mobilome-seq library with a DOC reaching 300X ( Fig 4B ) , showing that some members of this subfamily are actively producing eccDNAs in wild type rice seeds . We detected SRs located on both 5’ and 3’ ends of some PopRice loci ( Fig 4B and S9 Fig ) . A closer examination of reads spanning junctions has also confirmed the presence of PopRice eccDNAs ( S10 Fig ) . Further sequence analyses of PopRice family revealed that the most active PopRice copies form a subgroup of 5 members ( Fig 4C ) . We performed de novo assembly of mobilome-seq libraries to determine whether EVD , Tos17 and PopRice could be detected without mapping on a reference genome . We did not detect scaffolds corresponding to EVD when performing de novo assembly on the WT mobilome-seq library . In the Arabidopsis epi12 mobilome-seq library , de novo assembly resulted in three main scaffolds corresponding to EVD ( S11 Fig ) . These three scaffolds all result from the assembly of a high number of reads ( 59 , 943; 49 , 098 and 19 , 424 reads per million ( rpm ) , respectively , p-value < 0 . 05 , negative binomial distribution ) . In the rice callus mobilome-seq library the most highly covered scaffold ( 3 , 906 rpm ) with homology to TEs corresponded to Tos17 ( 100% identity over 4 , 532 base pairs ( bp ) , Fig 3D ) . This suggests that de novo assembly can be used to identify active RTs . In the seed mobilome-seq library , the most significantly covered scaffold ( 3 , 473 rpm ) showed 99% of sequence identity with a PopRice consensus sequence over 4 , 960 bp ( Fig 4D ) . Only the ends of PopRice were not assembled in this scaffold , likely due to the repetitive nature of LTR sequences . To further validate the presence of extrachromosomal DNA fragments originating from PopRice in WT rice seeds , we performed a Southern blot experiment using non-amplified and non-digested genomic DNA ( Fig 5A ) . Using a PopRice specific probe , a signal corresponding to a 5 kb fragment was identified in genomic DNA samples extracted from seeds but not from leaves , revealing a massive accumulation of PopRice extrachromosomal copies in wild type seeds . A Southern blot performed on genomic DNA obtained from dissected seed tissues further revealed that PopRice extrachromosomal DNA could only be detected in the endosperm tissue but not in the embryo or seed coat ( Fig 5B ) . This result was confirmed by inverse PCR assays ( S12 Fig ) . To characterize the kinetics of PopRice activation during plant development we used inverse PCR to detect the presence of PopRice eccDNAs at different developmental stages . PopRice eccDNAs seemed to be specific of seed tissues from the embryo developmental stage ( corresponding to immature seeds , from 3 to 5 days after pollination ) to the germination , however eccDNAs were not detected in roots and cotyledons after germination ( Fig 5C ) . To rule out the possibility that PopRice circles could originate from homologous recombination between its endogenous LTR sequences , we identified mobilome-seq reads corresponding to 2LTR junctions in the seed libraries ( S10 Fig ) . The presence of these reads confirmed that non-homologous end-joining of reverse transcription products , and not homologous recombination at the endogenous genomic location , is responsible for the formation of PopRice eccDNA molecules . Additionally we analyzed PopRice transcription , the mRNAs being the precursors of the eccDNAs . RT-qPCR assays showed that PopRice and Osr4 members are highly transcribed in seeds compared to leaves and flowers ( Fig 5D ) . The level of expression seems higher when all elements of the Osr4 family are considered suggesting that the whole Osr4 family is transcriptionally active , although only PopRice eccDNAs could be detected ( Fig 4C ) .
In all eukaryotic organisms , eccDNA molecules are ubiquitous elements and constitute an heterogenous population of circular molecules that can originate from repeats such as rDNA clusters through homologous recombination [34 , 53] or from active TEs ( through circularization of linear extrachromosomal forms ) . We took advantage of the detection of eccDNAs by next generation sequencing ( NGS ) to explore the extrachromosomal circular mobilome in plants . As a proof of concept we analyzed samples from A . thaliana and O . sativa material for which actively proliferating TEs had previously been characterized [44 , 45] . Identification of two well-characterized active LTR-RTs , EVD and Tos17 in an A . thaliana hypomethylated line and in rice callus tissue , respectively , confirmed that this method is efficient to capture actively proliferating retrotransposons in plants . The detection of rDNA circles validates the enrichment of eccDNAs in our libraries and thus constitutes another positive control . Moreover our observations suggest that some TEs might form different circles where SRs spread into internal regions of TEs reflecting a possible heterogeneity of these extrachromosomal circles . The mobilome-seq strategy exploits the advantages of NGS and requires a low sequencing coverage for each library . Indeed only a minor fraction of the genome is sequenced , opening the future possibility of applying the technique to very large genomes , for which resequencing techniques are not affordable and technically challenging . Furthermore , the de novo assembly analyses might represent a precious and powerful method to study the active mobilome of species for which a reference genome is lacking . Developmental relaxation of TE control has been documented in plant tissues accompanying the gametes: vegetative nucleus for the pollen and endosperm for the ovary [51] . In rice , DNA methylome analyses revealed a global hypomethylation in the endosperm [50 , 54 , 55] confirming previous results in Arabidopsis [49 , 56 , 57] . This suggests that TE activity could be increased in these tissues; however , to our knowledge , only the proliferation of a DNA-TE Mule element has been documented so far in the A . thaliana pollen [48] . Here , our seed mobilome-seq analyses reproducibly revealed that the PopRice family of autonomous LTR-RTs produces extrachromosomal copies . These copies can be detected on a Southern blot analysis of untreated genomic DNA showing that PopRice extrachromosomal copies accumulate in wild type rice endosperm . Further studies will help evaluating the proliferation of PopRice in the endosperm genome . PopRice transcripts could be detected in seeds suggesting that eccDNAs indeed originate from reverse transcription of these transcripts . Genomic imprinting could explain the transcriptional activity of some TEs in the endosperm . According to a study by Luo et al . [58] , only two relatively ancient copies ( PopRice_16 and Osr4_28 ) are localized in introns of paternally imprinted genes ( LOC_Os11g09329 and LOC_Os08g24420 , respectively ) suggesting that imprinting might not be the only trigger for PopRice/Osr4 transcriptional activation in the endosperm . Recently , Cheng et al . have shown that members of the Osr4 LTR-RT family could retrotranspose in oscmt3 mutants , affected in a chromomethylase involved in DNA methylation , through genome resequencing [59] . Interestingly all neo-insertions are due to the PopRice subfamily suggesting that this subfamily contains all potentially active members and that they are under the epigenetic control of OsCMT3 . This transgenerational control is reminiscent of the regulation of Onsen , an A . thaliana LTR-RT that produces eccDNA molecules after heat stress . However in the case of Onsen transgenerational neo-insertions are only detected in mutants affected in the RNA-directed DNA methylation ( RdDM ) pathway , but not in the CMT3 pathway [60] . The precise role of the RdDM pathway in the transgenerational control of these LTR-RTs neo-insertions is not yet elucidated [61] . Using a newly developed method to sequence and identify eccDNAs originating from TEs , we have characterized a yet unexplored fraction of plant DNA . This study revealed the reactivation of an endosperm-specific LTR-RT in rice . This LTR-RT family seems to be under the control of both epigenetic and post-transcriptional regulation . Furthermore the identification of this only LTR-RT family active in the endosperm suggests that the global hypomethylation occurring in this tissue is not sufficient to trigger a massive reactivation of TEs . By giving an insight into actively proliferating retrotransposons in plants , the mobilome-seq approach is likely to expand our understanding of TE activity in plants and of their putative contribution in response to stress and during plant development .
Arabidopsis thaliana WT ecotype Columbia-0 and epiRIL12 plants from the eighth generation [43] were grown in soil under a 16h/8h ( light/dark ) cycle after 2 days at 4°C for stratification . Florets and 1–2 cm green siliques were harvested 3 days to 2 weeks after pollination , respectively . Oryza sativa ssp . japonica cv . Nipponbare rice plants were cultivated in a growth chamber ( Percival , USA ) under a 12h light-dark cycle ( 12h-28°C/12h-26°C ) and with a relative humidity of 80% during the day and 70% during the night . The light intensity varied gradually in 40 min at the beginning and end of the day . Grain material was harvested 3 to 5 to 15 days after pollination for the immature and mature stage , respectively . Seeds were germinated in the dark on a humid Whatman paper for 5 days before harvest . Dissection of seeds was performed under the binocular on mature seeds . Callus material was previously described [21] . For each plant sample , total DNA was extracted using the plant DNeasy mini kit ( Qiagen ) according to the manufacturer’s instructions . A DNA pre-extraction was performed for rice grains to optimize DNA quantity and quality . Grains were grinded in an extraction buffer ( Tris-HCl pH8 , NaCl 250mM , EDTA 50mM , 0 . 2% SDS ) and were incubated 30 min at 65°C . DNA samples were precipitated with 0 . 7 volume of isopropanol and the DNA pellet was directly resuspended in the plant DNeasy mini buffer ( Qiagen ) . To remove large genomic linear fragments 5μg of genomic DNA for each sample were purified using a Geneclean kit ( MPBio ) according to the manufacturer’s instructions . eccDNA was isolated from 28μl of the GeneClean product using the PlasmidSafe DNase ( Epicentre ) according to the manufacturer’s instructions , except that the 37°C incubation was performed for 17h . The PlasmidSafe exonuclease digests double-stranded linear DNA to deoxynucleotides while leaving circular DNA intact . DNA samples were precipitated by adding 0 . 1 volume of 3M sodium acetate ( pH 5 . 2 ) , 2 . 5 volumes of ethanol and 1 μl of glycogen ( Fisher ) and incubating overnight at -20°C . The precipitated circular DNA was amplified by random RCA using the Illustra TempliPhi kit ( GE Healthcare ) . For this , the DNA pellet was directly resuspended in the Illustra TempliPhi Sample Buffer , and the reaction was performed according to the manufacturer’s instructions except that the incubation was performed for 65h at 28°C . One tenth of each amplified DNA sample was digested with restriction enzymes and loaded on an agarose gel electrophoresis to control the DNA quality and amplification . Then , the DNA concentration was determined using the DNA PicoGreen kit ( Invitrogen ) following the manufacturer’s instructions , the fluorescence being read using a LightCycler480 ( Roche ) . The samples were diluted to a final concentration of 0 . 2 ng/μl in order to prepare the libraries for sequencing . One nanogram of DNA from each sample was used to prepare the libraries using the Nextera XT library kit ( Illumina ) according to the manufacturer’s user guide . Each mobilome-seq library was amplified by 12 cycles of PCR using index primers . DNA quality and concentration were determined using a high sensitivity DNA Bioanalyzer chip ( Agilent Technologies ) . Samples were pooled and loaded onto a flow cell and 2x250 nucleotides paired-end sequencing was performed using the MiSeq platform ( Illumina ) . Up to twelve mobilome-seq libraries were pooled into one run and an average of 1 million reads per library were obtained ( S1 and S2 Tables ) . Illumina reads were collected for the analysis as FASTQ files . To analyze the sequencing reads we anticipated that the eccDNAs of interest originating from mobile TEs should represent a very small fraction of the genome and consequently that the loci from where these eccDNAs were produced should be highly covered . Furthermore , as these molecules are circular , reads spanning the junction of the circles should not map properly on the reference genome because such junctions do not exist in the chromosomes . However , these reads might map on two different locations ( start and end of the element ) . Thus the eccDNAs of interest should fit to the two following criteria: ( 1 ) high DOC and ( 2 ) presence of SRs when mapped to a reference genome . Finally , due to the repetitive nature of TEs , we reasoned that the read-mapping coverage could be less sensitive for large TE families as reads could be dispersed amongst related TE copies . Therefore we should be able to identify the most abundant eccDNAs by analyzing highly covered scaffolds after de novo assembly . Quality control of FASTQ files was evaluated using the FastQC tool ( version 0 . 10 . 1 www . bioinformatics . babraham . ac . uk/projects/fastqc ) . To remove any read originating from organelle circular genomes , reads were mapped against the mitochondria ( NCBI GenBank Y08501 . 2 for A . thaliana; GenBank NC_011033 for O . sativa ) and chloroplast genomes ( GenBank AP000423 . 1 for A . thaliana; GenBank X15901 for O . sativa ) using the program BOWTIE2 version 2 . 2 . 2 [62] with—sensitive local mapping . Unmapped reads were considered for the next analysis and were mapped against a genome of reference , TAIR10 ( The Arabidopsis Information Resource , http://www . arabidopsis . org ) for A . thaliana , IRGSP1 . 0 ( International Rice Genome Sequencing Project version 5 http://rgp . dna . affrc . go . jp/E/IRGSP/Build5 . html ) for O . sativa . The parameters used for the mapping were as follows:—sensitive local mapping , no multiple-mappings ( -k 1 ) so that only the best hit is kept per read-pair . DNA from both mitochondria and chloroplast genomes are integrated in nuclear genomes . To completely eliminate these regions from our data , sequencing reads were simulated from organelle genome using the dwgsim program ( version 0 . 1 . 10 ) and the . fasta files were mapped against the corresponding reference genome . A total of 816 , 300 bp and 1 , 697 , 400 bp were masked in A . thaliana and O . sativa , respectively , and TE containing regions cover 24 , 786 , 000 bp and 194 , 224 , 800 bp in A . thaliana and O . sativa , respectively . A . bam file with all genome regions corresponding to organelle integrated sequences was obtained for each species and was used to filter our alignment files using the intersect module of BEDTools version 2 . 21 . 0 ( option -v ) . Finally , for each library , a . bam alignment file corresponding to enriched genomic regions was considered for statistical analysis and visualized with the Integrative Genomics Viewer ( IGV ) software ( https://www . broadinstitute . org/igv/home ) and Circos [63] . For each species , the reference genome was split into consecutive windows of 100bp for each library and the coverageBED module of BEDTools was used to determine the read coverage depth of these non-overlapping windows . The coverage data was normalized by the total number of reads which mapped on the genome expressed in rpm and statistical analysis was performed on these files . First we determined covered regions using the Poisson distribution that best fits our data with a p-value <10−5 for each library . All uncovered regions were removed from our coverage files . On the covered regions we applied a negative binomial distribution to identify peaks of higher coverage with a p-value <10−3 . Finally , regions corresponding to the peaks were selected and annotated using . gff files ( S5 Table ) . All statistical analysis and graphics were performed using R ( Rstudio package version 0 . 98 . 1091 , www . r-project . org/ ) . Mobilome-seq reads were assembled de novo using the A5-miseq pipeline [64] . For each library , . fasta and . bam files were obtained and the idxstats module of SAMtools was used to determine the read number corresponding to each assembled scaffold . The coverage data was normalized by the total number of reads used for the de novo assembly expressed in million reads ( rpm ) . We applied a negative binomial distribution to identify significantly covered scaffolds ( p-value < 0 . 05 ) . Filtered scaffolds were annotated using a BLAST analysis ( -p -m 8 ) against organelle genomes and a TE database allowing for one hit per scaffold ( -b 1 -v 1 options ) and for an e-value < 10−2 . For A . thaliana we used the TE database based on TAIR10 annotation and established by H . Quesneville ( www . arabidopsis . org ) ; for O . sativa we used an in house curated database ( www . panaudlab . org ) . Resulting hits were filtered to keep only scaffolds with a HSP ≥ 100bp . Reads spanning 2 LTR junctions constitute an evidence of a circular TE and were detected using a SR mapping strategy . Reads were aligned against the reference genomes using the segemehl software [65] with the following parameters: -S ( SR mapping ) -A 95 ( accuracy of 95% ) -U 24 ( minimum score of 24 ) -Z 25 ( minimum length of 25 ) -W 95 ( alignment covered on 95% of the read ) . Split reads were collected from . bam files based on the FLAG field of each read , using the view module of SAMtools . Therefore , only reads which were not mapped in a proper pair ( -f 14 ) and which have multiple primary alignments ( -F 256 ) were considered as SR . The coverageBED module of BEDTools was used to determine the read coverage depth of these SR . bam files were visualized with IGV . To determine TE loci of interest in each library , we first filtered the . bam files obtained from bowtie2 mapping ( for the DOC ) for TEs covered for 90% of their length and with a DOC >10 rpm . Using the . bam files obtained from segemehl mapping we selected the TE loci with a SR coverage >10 rpm . We selected the TE loci fitting both criteria and which length is > 100bp ( S6 Table ) . The TE loci were visualized with Circos [63] . The in-house developed code for DOC and SR detection and for the establishment of the candidate TE list can be accessed upon request . Total genomic DNA was extracted using the CTAB method [66] and samples were loaded on a 0 . 8% agarose gel and transferred onto Hybond-N+ nylon membrane ( GE Healthcare ) . The Southern blot on A . thaliana material was performed as previously described [44] using a radioactive probe . The Southern blot on O . sativa material was performed using a non-radioactive probe and the hybridization signal was detected with the DIG system ( Roche ) following the manufacturer’s instructions . Stringency washes were performed at 65°C in 0 . 5X SSC . Probes were amplified from genomic DNA by PCR using primers listed in the S4 Table . The Southern blots on Fig 5A and 5B were repeated twice using biological replicates . Total RNAs were isolated from leaves , flowers and seeds using the Tri-reagent ( MRC ) according to the manufacturer's instructions . RNAs were treated with DNAse from RQ1 kit ( Promega ) and 1 . 25 μg were reverse-transcribed into cDNAs using the GoScript kit ( Promega ) . Analyses by quantitative real-time PCR ( qRT-PCR ) were established using 7 to 35 ng of cDNA . qRT-PCRs were run on a LightCycler 480 ( Roche ) using Takyon No Rox SYBR MasterMix dTTP Blue Kit ( Eurogentec ) according to the manufacturer’s instructions . The qRT-PCR conditions were the following: a first denaturation step at 95°C for 5 min followed by 40 cycles at 95°C for 15s , an annealing and elongation step at 60°C for 60s , and a melting curve analysis at 95°C for 10s , 60°C for 10s , an increase of 0 . 04°C per second until 95°C and a final step of cooling at 40°C for 30s . Two biological replicates were analyzed for each tissue . PopRice and Osr4 expression levels relative to eIF-5a [67] were calculated using the formula: 2- ( mean ( CT PopRice—CT internal references ) ) . Primers were used with a concentration of 2μM and primers details are given in the S4 Table and S13 Fig . PCR reactions were performed using 2 μl of DNA ( before or after the RCA amplification ) in a final volume of 15 μl , using the GoTaq polymerase ( Promega ) . All primer pairs were designed using Primer3 ( www . primer3 . ut . ee ) and quality-checked using OligoCalc ( www . basic . northwestern . edu/biotools/oligocalc . html ) and BLAST ( www . ncbi . nlm . nih . gov/BLAST/ ) . Target sequences and primers used are shown in S4 Table . The PCR conditions were the following: a first denaturation step at 95°C for 5 min followed by 30 cycles at 95°C for 30s , an annealing step ( temperature details in S4 Table ) for 30s , an elongation step at 72°C for 20 seconds , and a final extension step at 72°C for 5 min . 8 μl of PCR products were deposited on a 1 , 5% agarose gel and run at 135mV for 30 min . DNA was stained using a GelRed dye ( Biotium ) . Gel pictures were obtained using an UGenius gel imaging system ( Syngene ) . All PCR assays were repeated at least twice using biological replicates . To determine the evolutionary story of PopRice elements , a consensus sequence of PopRice was used to detect by BLAST all LTR-RTs from the IRGSP-1 . 0 reference genome belonging to the same family ( HSP>4000bp , minimum of 70% of identity , e-value < e-50 ) . All selected sequences were aligned with MAFFT ( http://mafft . cbrc . jp/alignment/server/ ) . Alignments were analyzed using SEAVIEW ( http://doua . prabi . fr/software/seaview ) and all incomplete elements were removed . 47 elements were selected for the Osr4 family ( comprising PopRice sequences ) and a phylogenetic tree was built with PhyML and visualized with FigTree ( http://tree . bio . ed . ac . uk/software/figtree/ ) . Sequencing data generated in this study have been deposited at the European Nucleotide Archive ( ENA , www . ebi . ac . uk/ena ) under the accession number PRJEB13537 .
|
Long time considered as « junk DNA » , the evolutive force of transposable elements ( TEs ) is now well established and TEs contribute strongly to eukaryote genome plasticity . However , it is difficult to fully characterize the mobile part of a genome , or active mobilome , and tracking TE activity remains challenging . We therefore propose to use the detection of extrachromosomal circular DNA as a diagnostic for plant TE activity . Our mobilome-seq technique allowed to identify a new active retrotransposon in wild type rice seeds , and will represent a powerful strategy in characterizing the somatic activity of TEs to evaluate their impact on genome stability and to better understand their adaptive capacity in multicellular eukaryotes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"biotechnology",
"retrotransposons",
"brassica",
"plant",
"science",
"model",
"organisms",
"rice",
"genome",
"analysis",
"experimental",
"organism",
"systems",
"genetic",
"elements",
"plant",
"genomics",
"molecular",
"biology",
"techniques",
"plants",
"arabidopsis",
"thaliana",
"research",
"and",
"analysis",
"methods",
"genomic",
"libraries",
"grasses",
"artificial",
"gene",
"amplification",
"and",
"extension",
"plant",
"genetics",
"molecular",
"biology",
"genetic",
"loci",
"plant",
"and",
"algal",
"models",
"polymerase",
"chain",
"reaction",
"genetics",
"transposable",
"elements",
"biology",
"and",
"life",
"sciences",
"genomics",
"mobile",
"genetic",
"elements",
"plant",
"biotechnology",
"computational",
"biology",
"organisms"
] |
2017
|
Sequencing the extrachromosomal circular mobilome reveals retrotransposon activity in plants
|
Single nucleotide polymorphisms ( SNPs ) have been increasingly utilized to investigate somatic genetic abnormalities in premalignancy and cancer . LOH is a common alteration observed during cancer development , and SNP assays have been used to identify LOH at specific chromosomal regions . The design of such studies requires consideration of the resolution for detecting LOH throughout the genome and identification of the number and location of SNPs required to detect genetic alterations in specific genomic regions . Our study evaluated SNP distribution patterns and used probability models , Monte Carlo simulation , and real human subject genotype data to investigate the relationships between the number of SNPs , SNP HET rates , and the sensitivity ( resolution ) for detecting LOH . We report that variances of SNP heterozygosity rate in dbSNP are high for a large proportion of SNPs . Two statistical methods proposed for directly inferring SNP heterozygosity rates require much smaller sample sizes ( intermediate sizes ) and are feasible for practical use in SNP selection or verification . Using HapMap data , we showed that a region of LOH greater than 200 kb can be reliably detected , with losses smaller than 50 kb having a substantially lower detection probability when using all SNPs currently in the HapMap database . Higher densities of SNPs may exist in certain local chromosomal regions that provide some opportunities for reliably detecting LOH of segment sizes smaller than 50 kb . These results suggest that the interpretation of the results from genome-wide scans for LOH using commercial arrays need to consider the relationships among inter-SNP distance , detection probability , and sample size for a specific study . New experimental designs for LOH studies would also benefit from considering the power of detection and sample sizes required to accomplish the proposed aims .
Single nucleotide polymorphisms ( SNPs ) are common DNA sequence variations and have been widely investigated for their roles in disease causation [1] or association [2 , 3] , heterogeneous responses to drug therapies [4–6] , genetic linkage analysis [7 , 8] , and evolutionary biology [9 , 10] . This has led to the characterization of whole-genome patterns of a large number of common SNPs in a few ethnic groups [11] . Distinct from constitutive genome studies , SNPs have also been used extensively to study the somatic development of cancer [12] ( also see review by Engle , et al . [13] ) . Alterations of the copy number of DNA sequences ( DNA amplification or deletion ) and those that result in a loss of genetic information ( loss of heterozygosity; LOH ) occur frequently in neoplastic tissues and tumors , and changes in the copy number or heterozygosity of SNPs allow these alterations to be detected and mapped in the genome . Low density , whole genome analyses have previously been sufficient to allow gross characterization of critical genetic alterations that occur during neoplastic progression . However , much finer-scale mapping of these alterations is frequently required both for furthering basic understanding of the genetic events that occur during progression to cancer and for developing diagnostic tests with sufficient sensitivity and specificity for translation into clinical practice . In addition , commercial high-density SNP platforms tend to be both expensive and biospecimen-intensive , making them impractical for high-throughput , fine-scale mapping of specific chromosomal regions . The alternative is to develop a custom panel of SNPs that can characterize the genomic region of interest . Detection of LOH requires SNPs to be heterozygous ( i . e . , informative ) . In the largest public SNP database , dbSNP ( http://www . ncbi . nlm . nih . gov/SNP ) , the heterozygosity ( HET ) rates estimated for a substantial number of SNPs have large estimated variances , likely due to small sample sizes , among other reasons [14] . Using SNPs with large HET rates variances may lead to ambiguous experimental results ( e . g . , an under-powered study ) . A better understanding of how the distribution of SNPs in the genome and the variance of SNP HET rates affect the ability of a panel of SNPs to detect LOH would allow improved design of SNP-based assays for somatic genetic alteration studies . Statistical models for classifying subjects by LOH profile that take into account noninformative markers have been developed [15] . We used real genotype data to investigate the relationship between detection probabilities ( resolution ) and LOH sizes using all currently characterized SNPs in Hapmap , a relationship that is closely related to sample size and power calculations in LOH detection experimental design . As well , the study evaluates the key factors governing the selection of a group of SNPs for designing custom assays for particular chromosomal regions . Specifically , we first evaluated the variances in SNP HET rate estimation currently reported in the dbSNP database , and then addressed sample size issues related to directly inferring SNP HET rates for the purpose of selecting SNPs for LOH detection . We propose two statistical approaches to determine the minimum number of individuals in a population that would need to be examined to determine if a SNP HET rate was above or below a specified threshold . Finally , we evaluate the relationships between the number of SNPs , SNP HET rates , and sensitivity ( resolution ) for detecting LOH using real whole-genome genotype data .
The frequency distribution of the average SNP HET rates for each SNP reported in the dbSNP database is shown in Figure 1 . The genome-wide mean HET rate is 0 . 263 ( SD = 0 . 171 ) . The observed pattern was similar to a beta distribution , although the data do not exactly fit a formal beta distribution . Figure 2 shows the distribution of the estimated coefficient of variation ( CV = SD/Mean% ) of SNP HET rates in the dbSNP database . These results indicate that a significant number of the SNPs in dbSNP have large estimated variances , which would not provide enough precise information for designing studies requiring the accurate estimation of SNP HET rates ( i . e . , those using SNPs for LOH detection for molecular diagnoses ) . Traditionally , for diallelic alleles with p1 and p2 allele frequencies , the HET rate could be estimated as hr = 2p1p2 , although this formula is appropriate only for alleles in HWE . Another approach , which is robust to HWE assumptions , is to estimate the HET rate ( and its variance ) directly by population allele frequencies [16] . This method requires large sample sizes in order to achieve accurate estimation of HET rates . Here we consider the case where the HET rate is measured directly using techniques like DNA sequencing , microarray analysis , Pyrosequencing , or MALDI-TOF . Using hypothetical parameters for true HET rates and sample sizes , we first show the relationships among true HET rates , estimated HET rates , their estimated variances , and sample sizes using the score method with continuity correction ( exact binomial method may result in larger CI ) ( Table 1 ) . The variance of HET rates could be estimated by ( hr ( 1 − hr ) ) / ( N − 1 ) , where N is the sample size . Table 1 shows that even with moderately large sample size ( e . g . , N = 100 ) , the confidence interval or CVs are quite large for all values of HET rates listed , particularly for lower HET rates ( some upper bounds of the CI even exceeded the theoretical maximum value of hr = 0 . 5 ) . With 500 subjects tested , the estimated CI and variance are small , but such a sample size is prohibitively large for many studies . We introduce two different approaches to deal with the unrealistically large sample size requirement . In using SNPs to evaluate LOH in a specific chromosomal region , it is desirable that the HET rates of selected SNPs used in the region be higher than a specific value to increase the probability that at least one SNP will be informative for each patient . Therefore , the question is to test the statistical hypothesis for the HET rate of a specific SNP hrs versus a prespecified HET rate value hr0 ( i . e . , H0: hr ≥ hr0 versus H1: hr < hr0 ) . With a given power and sample size n , we have: To get sample size , we have: = Zβ , where Zα and Zβ are the 100 ( 1−α ) th and 100 ( 1−β ) th percentile of the standard normal distribution . Using Equation 1 , Table 2 shows the sample sizes needed for testing whether the HET rate of a given SNP is significantly higher than a desired threshold . The sample size required to reject H0 is reasonably small in most cases; e . g . , when hr0 = 0 . 2 , and hrs = 0 . 35 , only 50–72 subjects need to be tested . However , when a SNP HET rate hrs is near the desired threshold value hr0 , the required sample size becomes much larger . Table 2 utilizes a fixed sample size method that is easy to use , but may not be optimal when considering the number of subjects needed . A sequential sampling technique based on the sequential probability ratio test ( SPRT ) [17] can be used for directly inferring SNP heterozygous rates . With this method , samples are tested one by one and a decision will be made to determine whether or not the HET rate of a SNP has reached a prespecified value after each sample is tested . This method generally requires less time and reagents and conserves biospecimens that are frequently unique and difficult to obtain . Specifically , SPRT tests the SNP HET rate h using the hypothesis H0: h = h0 versus H1: h = h1 , ( h1 < h0 ) . The likelihood ratio is For type I error ( false positive ) level α , and type II error ( false negative ) level β , ( power = 1 − β ) , it has been shown that sample testing should continue if ln < ln ( λ ( x1 , … , xn , h0 , h1 ) ) < ln ; if ln ( λ ) reaches or passes beyond the two bounds , then sample testing should stop . The hypothesis H0 will be accepted when ln ≤ ln ( λ ) , or H0 will be rejected and H1 accepted when ln ( λ ) ≥ ln . In this process , the total number of samples tested is a random variable based on the distribution specified by parameters h0 , h1 , α , β , and the underlying HET rate h of a specific SNP . In the SPRT approach , for fixed h , α , and β , the ASN ( average sample number ) depends on h0 and h1 . Table 3 shows simulation results for testing h0 = 0 . 3 , and h1 = 0 . 2 against various true ( sample ) SNP HET rates h . For example , if the true SNP HET rates under testing are h = 0 . 4 or above , approximately 15 to 40 subjects need to be tested , on average , to make a decision on whether h = h0 , and , under the most optimistic situations , only four subjects are necessary to determine the HET rate regarding hypothesis H0 . Depending on the goals of a study , the SPRT method could be used to significantly reduce the testing sample size required for SNP HET rate inference ( e . g . , compare to values in Table 1 ) . We also examined the number of SNPs needed for reliable detection of LOH for random chromosomal regions of a specific length assuming the SNP HET rate distribution shown in Figure 1 . If SNPs are being used to effectively detect the loss of a chromosomal segment , the segment should contain at least one or more heterozygous SNPs . If all SNPs have an identical HET rate ht ( 0 < ht ≤ 0 . 5 ) , then k SNPs are needed such that 1 − ( 1 − ht ) k ≥ threshold ( i . e . , threshold = 0 . 95 or 0 . 99 ) to guarantee at least one or more heterozygous SNP will be in the lost segment . However , ht is not constant across all SNPs ( Figure 1 ) . Therefore , k SNPs are needed to have: where the threshold ( i . e . , threshold = 0 . 95 or 0 . 99 ) is the probability of having at least one or more heterozygous SNP in the chromosome segment . Based on the distribution pattern of HET SNP rates ( Figure 1 ) , Monte Carlo simulation was used to estimate the number of SNPs needed ( k ) to satisfy Equation 2 at the α level ( i . e . , 0 . 05 or 0 . 01 ) which guarantees that the left-hand-side of Equation 2 will lie beyond the threshold ( 1 − α ) 100% of the time . The probability density distribution of k is shown in Figure 3 based on the results of these simulations . Similarly , the simulation indicates that if SNPs with HET rates ≥0 . 3 are randomly selected for use , then the required number of SNPs ( k ) is 10 , and for a SNP HET rate ≥0 . 4 , the required number of SNPs ( k ) is 9 ( both calculated at α = 0 . 01 level using the cumulative density function and threshold = 0 . 95 , unpublished data ) . Given the non-random distribution pattern of SNP HET rates in the genome , the next obvious question is how long ( in base pairs ) must a random chromosomal segment be to contain one or more heterozygous SNPs so that LOH is detected with a high probability ( e . g . , 0 . 95 or 0 . 99 ) . Based on HapMap data , we used three approaches to ascertain this relationship , including simulation using the fitted dbSNP HET rate distribution pattern in Figure 1 , modeling of the SNP HET rate distribution within various chromosome deletion sizes using a negative binomial distribution ( model not shown ) , and random sampling along a chromosome based on real genotyping data . The results from the three approaches are shown in Figure 4 using Chromosomes 1 , 3 , 9 , and 17 , which frequently undergo alterations in many cancers , as examples . Many publications [11 , 18–20] have reported the mean/median distance between SNPs ( inter-SNP distance ) on specific arrays used in various studies . Therefore , we explored the relationships between inter-SNP distances , SNP HET rate , and detection probability of LOH to determine the chromosome segment size in base pairs required to have a reasonable chance of containing an informative SNP . Let s be the size ( in nucleotide base pairs ) of the DNA being lost on a chromosome , d the distance ( in nucleotide base pairs ) between two SNPs ( inter-SNP distance ) , and hhet the SNP HET rate , assuming the SNPs to be evenly distributed . If s ≤ d , the probability of the lost DNA segment containing a SNP can be estimated as p = , and the probability of detecting of LOH with HET SNPs is pd = phhet ( Figure 5A ) . When s > d , the number of SNPs within the lost region is k = ( ⌊ ⌋ representing the largest integer equal to or smaller than s/d ) . The probability that at least one SNP is heterozygous can be estimated as pd = 1 − ( 1 − hhet ) k . The relationships are shown in Figure 5 . Finally , we used a bootstrap method to randomly sample the heterozygous SNPs on Chromosomes 1 , 3 , 9 , and 17 genotype data within a 500 kb window in two human subjects from the HapMap database . Figure 6 shows the spatial distribution of LOH detection probabilities with heterozygous SNPs on Chromosomes 1 , 3 , 9 and 17 for various loss sizes ( 5 kb , 10 kb , 30 kb , and 100 kb ) , assuming all known SNPs in that region were used . The mean of the detection probabilities of each loss size are very similar to the results shown in Figure 4 .
Using SNPs for LOH detection is of great value for chromosomal instability studies and cancer risk prediction , but a better understanding of the resolution of the technique and how to select an informative panel of SNPs for a given application is needed . The variances of SNP HET rates are large for a large number of SNPs . In most cases , this is likely to be due to the small sample sizes used for estimation of allele frequencies in most cases . Differences in ethnic groups might also contribute to the variance of averaged HET rates . Relatively large sample sizes are needed to accurately estimate SNP HET rates using traditional methods . In order to reduce sample size for practical use , we presented two statistical methods that could be used to determine the number of individuals in the population that would need to be examined to determine if a SNP HET rate was above or below a specified threshold . The Monte Carlo simulation was performed on SNPs in dbSNP with HET rate estimation values ≤0 . 5 as well as all SNPs , with essentially no change in the conclusion of the study ( Figure 3 ) . Only 0 . 2% of the SNPs in dbSNP have a HET rate estimation higher than 0 . 5 , some of which may be truly higher due to violations of Hardy-Weinberg equilibrium and some due to other factors such as estimation from a small sample size . Based on specific study goals or technologies , more study specific methods such as truncated SPRT schemes [17] could potentially be used to minimize the sample size when the HET rate is close to the testing rate . In addition , since different human populations ( e . g . , Asian versus African versus European ) may have different SNP distribution patterns [21–23] , the sample size calculation methods may only be applicable within specific populations instead of across mixed populations . Finally , although the SNP HET rates could be inferred using linkage disequilibrium information ( i . e . , pair-wise linkage disequilibrium r2 ) , the estimation of r2 and variance of r2 themselves are subject to the effects of sample size and evolutionary history of specific SNPs [24] . Therefore , the sample size and variance of r2 should be considered when r2 are used for inferring SNP HET rates if a study has stringent requirements ( i . e . , development of clinical diagnostic markers ) . We did not distinguish coding and non-coding regions of the genome in this simulation since Cargill et al . [5] reported that there is no significant difference in SNP density between coding and non-coding regions , and since the breakpoints of chromosome loss are poorly understood . We also examined the detection probability of LOH due to various sizes of chromosome loss assuming all known SNPs were used . This question is closely related to sample size and statistical power calculation in the experimental design for a neoplastic progression study; e . g . , a small segment of chromosome loss has a lower detection probability for LOH , and in order to detect it , large sample sizes are needed . We also verified the simulation results directly using the SNP genotype data ( all SNPs were used ) from 90 individual subjects from the HapMap database ( Figure 4 , red line ) . The detection probabilities based on simulation methods are reasonably close to the observed data , but may be overly optimistic to a certain degree . Such differences may be due to bias in the SNP HET rate estimation distribution [25] toward common SNPs ( Figure 1 ) or to a non-random distribution of SNPs . Our study showed that a region of LOH greater than 200 kb could be detected with high probability ( >90% ) , with losses smaller than 50 kb having a substantially lower detection probability when using all SNPs currently in the HapMap database ( Figure 4 ) . Higher densities of SNPs exist in certain chromosomal regions that provide the opportunity for reliably ( p > 0 . 95 or 0 . 99 ) detecting LOH of segment sizes smaller than 50 kb ( Figure 6 ) . Finally , we evaluated the LOH detection probability for the given inter-SNP distances as reported for many commercial products ( e . g . , SNP-based genotyping arrays ) or in published studies . For inter-SNP distances of 120 kb to 200 kb , the probability of detecting LOH for LOH of 300 kb or smaller ranges from 20% to 60% depending on SNP HET rates . The detection probability appears close to 1 if the region of loss is 900 kb or larger . The detection probabilities with inter-SNP distances 120 or 200 kb indicated in Figure 5 are substantially lower than the results shown in Figure 4 for a similar size of LOH . This is because the results in Figure 4 assume all SNPs currently reported in HapMap were used , whereas for Figure 5 , SNPs with fixed inter-SNP distances ( fewer SNPs ) were used to calculate the detection probabilities . To increase detection probability , more SNPs should be used , or inter-SNP distance should be minimized ( red line in Figure 5 ) ; however , this might be limited by the actual number of HET SNPs in a given chromosome segment . An alternate solution would be to increase sample size ( statistical power ) to detect small size of loss in an experiment . To a certain degree , improvements in LOH detection algorithms will increase the LOH detection probability . Improvements might include increasing the sensitivity of LOH detection in mixed cell populations ( i . e . , the neoplastic changes in somatic tissue ) . Combining copy number measurements and allele ratio measurements will increase detection of deletions but not copy neutral LOH . However , the resolution of LOH will still be constrained by the informative SNP distribution pattern itself . Sequencing or screening more human subjects to find more new SNPs could improve the theoretical detection probability as shown in Figures 4 and 6 only if the future-discovered SNPs are of great abundance and have high HET rates . For example , a SNP chip with 1 million SNPs to cover the 3 billion bp human genome would have a 3 kb mean inter-SNP distance . If the SNPs were evenly distributed throughout the genome to maximize coverage , the regions of LOH would need to be 32kb or larger in order to be detected with 0 . 95 probability assuming a SNP HET rate of 0 . 25 ( and 26kb or larger for a HET rate of 0 . 3 ) . Due to uneven distribution of SNPs in actual sequences , the detection probability will fluctuate with similar patterns shown in Figure 6 . Using dbSNP and HapMap data , this study evaluated the distribution of SNP HET rates and resolution of LOH genome wide . The results of this study have two important implications that might improve design and interpretation of future genome wide LOH screens of cancers and premalignant tissues . First , retrospective review of previous genome-wide LOH screens indicate that technology limitations ( i . e . , SNP density of arrays ) used in the experiments could have missed significant numbers of LOH events that were below the resolution of the SNP array [26–31] . By using the analysis methods reported in this paper , reports of genome wide LOH could discuss the limitations of the resolution of the study in terms of what might have been missed in addition to the important loci that were discovered . A well-designed study using carefully selected SNP sets for evaluating specific regions on several chromosomes still had more than 280 kb distance on average between two informative SNPs [32] . However , in general , 280 kb is still relatively large considering an average gene size is 3 to 20 kb in the human genome , and smaller regions of LOH ( i . e . , <50 kb ) might still be important , especially for early stages of neoplastic progression . The characteristics of LOH resolution mentioned above still apply to higher-density SNP arrays . LOH has been frequently proposed as a candidate biomarker for cancer risk prediction . The ability to detect an LOH event will depend on informativity , SNP density , and the size of the LOH event . Our results could improve sample size calculations for design of future LOH studies . If one would like to detect the effect of an LOH event on the risk of progression to cancer , then the sample size depends on the LOH detection probability . For example , in a study with a 1:5 ratio of cases and controls , a minimum detectable relative risk of the LOH of 5 , a statistical detection power 0 . 9 , and an LOH prevalence rate of 30% among informative subjects , at least 23 cases and 117 controls will be needed if the LOH detection probability is 100% ( large region loss or high density of informative SNPs ) . However , if the LOH detection probability is 0 . 7 or 0 . 3 , for example , ( e . g . , a smaller loss event , or fewer informative SNPs ) , then at least 44 cases and 190 controls or 116 cases and 468 controls will be needed , respectively . All the results obtained in this analysis are based on the assumption that heterozygous SNPs are required for detection of LOH . New technologies are emerging that could be used to detect chromosome copy number changes ( including deletion ) using homozygous SNPs with a reasonably high accuracy [33 , 34] . However , since LOH can result from mechanisms that do not change copy number [35 , 36] , using copy number approaches can only yield a partial picture of the LOH status of a region of interest . Combining the analyses presented in this study and copy number could lead to a high level of reliability and a higher resolution in LOH detection for neoplastic progression research and biomarker development .
The data for SNPs HET rates were downloaded from dbSNP ( build 126 ) ( ftp://ftp . ncbi . nih . gov/snp/organisms/human_9606/database/organism_data/ ) . HapMap SNP data for the human genome were downloaded from the HapMap Web site ( July 2006 release ) ( http://www . hapmap . org/genotypes/ ) . We only used the CEU population ( Utah residents with Northern and Western European ancestry ) data from HapMap . Our methods can easily be extended to other ethnic group data . The estimated SNP HET rates >0 . 5 were dropped from the analysis of HET rate distribution . The estimated variances for SNP HET rates were directly obtain from dbSNP . Data from dbSNP were used to summarize the HET rate distribution pattern of SNPs ( Figure 1 ) and evaluate the estimated variances of HET rates in dbSNP ( Figure 2 ) . To estimate the number of SNPs needed for LOH detection in any given chromosomal region , a Monte Carlo simulation method was used . In this process , a SNP was selected and the determination of its heterozygosity was based upon the HET SNP distribution shown in Figure 1 . This process was repeated until the cumulative probability of HET SNP reached the threshold at a predetermined α level ( i . e . , α = 0 . 05 or 0 . 01 ) which guarantees that the left-hand-side of Equation 2 will lie beyond the threshold ( 1 − α ) 100% of the time ( Figure 3 ) . The simulation for chromosome segment deletion ( Figure 4 ) was done using the genotype data from the HapMap CEU population data . In the simulation process , for each of the Chromosomes 1 , 3 , 9 , 13 , 17 , and 18 ( results of Chromosome 13 and 18 are unpublished data ) , a random segment was removed from the chromosome ( mimicking the region of LOH on a chromosome ) , and the number of SNPs in the region was examined based on the genotype data of the individuals . The process was repeated 20 , 000 times for each segment size on a chromosome . The segment sizes of loss used in the simulation are: 5 , 10 , 20 , 30 , 50 , 100 , 200 , 300 , 500 , 1 , 000 , 2 , 000 , 3 , 000 , 4 , 000 , and 5 , 000 kb . Based on these data , three methods ( negative binomial model fitting , Monte Carlo simulation , and bootstrap ) were used to investigate the relationship between the size of chromosome loss and probability of LOH detection . For negative binomial model fitting , which was found to fit the data best among the various theoretical distributions we evaluated , the discrete frequency distribution patterns of HET SNPs for each segment size listed above were fitted to a negative binomial model . Specifically , for the data of each segment size of loss , the HET SNP counts in each sample along a chromosome were used to estimate the parameters of negative binomial distribution with maximum likelihood method . The random numbers of HET SNPs were then generated based on the fitted negative binomial distribution parameters for each size of segment loss . This was repeated 10 , 000 times for each segment size and the detection probabilities were calculated based on the process for each segment ( Figure 4 , magenta lines ) . For the Monte Carlo simulation ( Figure 4 , blue and black lines ) , for each size of deletion listed above , the number of SNPs for each segment was counted , and the number of HET SNPs and detection probabilities were determined based on the empirical distribution pattern shown in Figure 1 . For the bootstrap method , the observed detection probability ( Figure 4 red line ) was obtained by directly counting the HET SNPs in each segment based on the real genotyping data in the bootstrap sampling process . The results in Figure 5 were obtained by the probability model described in the text . To examine the spatial pattern of LOH detection probability along a chromosome ( Figure 6 ) , we chose the 500 kb window size along Chromosomes 1 , 3 , 9 , and 17 , and within each window samples were randomly taken with various loss sizes to calculate the probabilities of LOH detection within each window along the chromosome . Similar patterns were found on other chromosomes ( unpublished data ) . All analyses and simulations were carried out with Matlab ( version 7 . 1 , The MathWorks ) .
|
More than 99% of each person's genome is identical to everyone else's . Many of the differences involve single base pairs , termed single nucleotide polymorphisms ( SNPs ) . SNPs are used as genetic markers to facilitate identification of disease-causing genes , as well as in cancer studies by aiding in determining which regions of the genome may be lost ( LOH ) or amplified during neoplastic progression . One drawback to SNPs is their low informativity: a SNP is only informative if it is polymorphic on the two different alleles found on each chromosome of a pair; and if there is not an informative SNP in the region of genome of interest , it is impossible to detect alterations occurring there through LOH . A common solution to this problem is to use arrays containing hundreds of thousands of SNPs to ensure adequate coverage , but for many studies this is prohibitive on a cost and sample amount basis . In addition , SNP distribution itself can constrain the size of loss that can be reliably detected at the population level . We examined the relationship between chromosome loss sizes and detection probability of LOH genome-wide . The study provides useful information for researchers designing LOH-related studies and evaluating results obtained from such studies .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"homo",
"(human)",
"genetics",
"and",
"genomics"
] |
2007
|
Direct Inference of SNP Heterozygosity Rates and Resolution of LOH Detection
|
PROSPERO CRD42012002292
Dengue is a growing health concern in the Philippines . Outbreaks were reported in1926 [1] , [2] , and the first recorded epidemic in Southeast Asia occurred in Manila in 1954 [3] , [4] . Further epidemics occurred in 1966 , 1983 , and 1998 , with increasing reported cases of dengue disease [5]–[8] . The 1998 epidemic had the highest recorded incidence rate ( 60 . 9 cases per 100 , 000 population ) and case fatality rate ( CFR; 2 . 6% ) [5] . The rising incidence of dengue disease can be explained by several factors . Dengue is caused by one of four dengue viruses ( DENV-1 , -2 , -3 , or -4 ) transmitted primarily by the Aedes aegypti ( Linnaeus ) mosquito , which breeds in open water containers , and can survive year round in tropical and subtropical climates . During World War II , the movement of people and equipment expanded the geographic distribution of Ae . Aegypti and dengue disease in Southeast Asia [3] . Since then , virus propagation in the region has been facilitated by rapid urbanization , environmental degradation , the lack of a reliable water supply , and improper management and disposal of solid waste [3] , [9] . In the Philippines , the percentage of the population living in urban areas increased from 27 . 1% in 1950 to 58 . 5% in 2000 [10] . Dengue has been a notifiable disease in the Philippines since 1958 [11] . During the review period the Philippines employed both passive ( outpatient and inpatient ) and sentinel surveillance across all ages [12] . Prior to 2006 , the National Epidemic Sentinel Surveillance System , managed by the National Epidemiology Center ( NEC ) of the Department of Health ( DoH ) , maintained surveillance of notifiable diseases , including dengue disease . The National Epidemic Sentinel Surveillance System monitored the total number of hospital cases and deaths that were admitted to 250–400 selected sentinel hospitals throughout the Philippines and , up until 2005 , did not differentiate between dengue fever ( DF ) , dengue haemorrhagic fever ( DHF ) , or dengue shock syndrome ( DSS ) . To improve surveillance , in 2005 , the system changed to separate reporting of DF , DHF , and DSS . In 2007 , the Sentinel Surveillance System was expanded to include up to 1662 disease reporting units ( including sentinel hospitals , private hospitals , and rural health facilities ) to develop an all-case ( suspected and probable ) reporting system ( Philippines Integrated Disease Surveillance and Response System ) . In addition , virological surveillance of dengue disease was implemented in 2008 [13] . The Field Health Surveillance Information System , also managed by the NEC of the DoH , is a passive reporting system that consolidates public health statistics due to notifiable diseases , including dengue disease , from all levels of government health facilities in the Philippines . Most reported dengue cases are suspected or probable cases according to standard definitions and are not laboratory confirmed . In a recent review of research needs for dengue surveillance and emergency response [14] ( an update of a review of surveillance systems in dengue-endemic countries by Gubler [15] ) laboratory capability for DENV serology was rated as ‘good’ in the Philippines but laboratory capability for dengue disease virology was rated as ‘exists’ ( rather than ‘good’ ) [14] . In addition , at the time of the review there was a lack of laboratory capacity to confirm cases [16] . The 2009 World Health Organization ( WHO ) classification of dengue by levels of severity is currently used in the Philippines: non-severe dengue with or without warning signs , and severe dengue ( severe plasma leakage , severe bleeding , or severe organ involvement ) [17] . An objective of the Philippines DoH is to ensure that this system is consistently applied in the country . However , at the end of 2012 , the 1997 WHO classification [18] continued to be used by some reporting physicians . This older system grouped symptomatic dengue into three categories: undifferentiated fever , DF , and DHF; DHF was further classified into four severity grades , with grades III and IV being defined as DSS [18] . The Republic of the Philippines comprises 7107 islands in Southeast Asia; the three island groups of Luzon ( Regions I–V , Cordillera Administrative Region [CAR] , and National Capital Region [NCR] ) , Visayas ( Regions VI–VIII ) , and Mindanao ( Regions IX–XIII and Autonomous Region in Muslim Mindanao ) are split into 17 regions ( Figure 1 ) . The population is 92 , 337 , 852 ( 2010 census ) , with an average annual population growth rate of 1 . 90% for the period 2000–2010 [19] . A large proportion of the population ( 37 . 3% ) lives in three regions: Calabarzon ( Region IV-A; 11 . 74 million people ) , the capital , Metro ( metropolitan ) Manila , also known as NCR ( 11 . 55 million people ) , and Central Luzon ( Region III; 9 . 72 million people ) [20] . The Philippines has a tropical marine climate , with an average annual temperature of 27°C , annual dry seasons from December to May , and annual wet seasons from June to December [21] . A systematic literature review was conducted to describe the available epidemiology of dengue reported in the Philippines between 1 January 2000 and 23 February 2012 . Our objectives were to describe the recent epidemiology of dengue ( national and regional incidence [by age and sex] , seroprevalence and serotype distribution and other relevant epidemiological data ) and to identify gaps in epidemiological knowledge requiring further research , but not to provide an exhaustive picture of the history of dengue in the country . Given the 3–5-year periodicity of dengue outbreaks [3] we estimated that at least one decade of data would be necessary to provide an accurate image of recent evolution of epidemiology . Furthermore , a 10 year period was determined to observe serotype distribution over time and through several epidemics . For convenience , we chose to start our review period on 1 January 2000 and set the cutoff as 23 February 2012 , the date when we initiated this review . An additional rationale for selecting 1 January 2000 as the start date for this review , as opposed to an earlier date , was that we hypothesized that this would limit the bias that any differences in surveillance practices over time would have on the results .
Search strings for each database were designed with reference to the expanded Medical Subject Headings thesaurus , encompassing the terms ‘dengue’ , ‘epidemiology’ , and ‘Philippines’ . Different search string combinations were used for each electronic database with the aim of increasing the query's sensitivity and specificity . Only studies published in English between 1 January 2000 and 23 February 2012 were included . For databases that did not allow language and/or date limitations , references not meeting these criteria were deleted manually at the first review stage . No limits by sex , age and ethnicity of study participants or by study type were imposed , although single-case reports were excluded , as were studies that only reported data for the period before 1 January 2000 . As duplicate publication of data ( e . g . , in meta-analyses and other reviews ) could lead to oversampling and overestimates of the incidence of dengue disease , literature reviews and editorials involving previously published peer-reviewed data were also excluded . In March 2012 we searched the following databases ( 1 ) PubMed ( http://www . ncbi . nlm . nih . gov/pubmed/ ) ; ( 2 ) WHO Library database ( WHOLIS: http://www . who . int/publications/en/ ) ; ( 3 ) WHO Western Pacific Region ( WPRO: http://www . wpro . who . int/en ) ; ( 4 ) Index Medicus for South-East Asia Region ( IMSEAR: http://imsear . hellis . org/ ) ; ( 5 ) WHO Regional Office for Southeast Asia ( WHOSEAR: http://www . searo . who . int/en/ ) ; ( 6 ) Philippines Ministry of Health official bulletins ( http://www . doh . gov . ph/ ) ; ( 7 ) Philippines National Institute of Health ( http://nih . upm . edu . ph/ ) ; ( 8 ) Philippine Council for Health Research and Development ( http://www . pchrd . dost . gov . ph/ ) ; and ( 9 ) National Epidemiology Center , DoH , Philippines . We also included data from several other sources to complement articles identified by the primary literature review: two national journals ( the Pediatric Infectious Disease Society of the Philippines journal; http://pidsphil . org , and the Philippine Society for Microbiology and Infectious Disease journal; http://www . psmid . org . ph/ ) and The Western Pacific Surveillance and Response ( WPSAR; http://www . wpro . who . int/wpsar/en/ ) open-access journal dedicated to the surveillance of and response to public health events were searched; other reports and guidelines published on-line by relevant organizations; conference papers and posters from infectious disease , tropical medicine , and paediatric conferences , and grey literature ( e . g . , lay publications ) were sought through general internet searches ( e . g . , Google and Yahoo; limited to the first 50 search results ) . Additional publications and unpublished data sources meeting the search inclusion criteria were included if recommended by a consensus of the Literature Review Group . After removing duplicate citations , the Literature Review Group reviewed the titles and abstracts and identified those for which the full text was retrieved . A second review was performed on the full text to make the final selection of relevant articles to include . Studies were reviewed by the Literature Review Group to ensure they complied with the search inclusion and exclusion criteria . In particular , publications of duplicate data sets were excluded , unless the articles were reporting different outcome measures . We chose not to exclude articles and other data sources nor formally rank them on the basis of the quality of evidence . Indeed while it is recognized that assessing study quality can potentially add value to a literature review , the consensus of the Literature Review Group was that given the expected high proportion of surveillance data among the available data sources and the nature of surveillance data ( passive reporting of clinically-suspected dengue ) , such quality assessment would not add value in this case . As our primary objective was to describe the recent evolution of dengue , rather than to quantify disease in absolute terms , we therefore retained all available data sources . The selected data sources were collated and summarized using a data extraction instrument developed as a series of Excel ( Microsoft Corp . , Redmond , WA ) spreadsheets . Data were extracted into the spreadsheets according to the following categories for analysis: incidence , age , sex and serotype distribution , serotype data , seroepidemiology or seasonality and environmental factors , by national or regional groups . Data from literature reviews of previously published peer-reviewed studies and pre-2000 data published within the search period were not extracted . The original data sources and the extraction tables were made available to all members of the Literature Review Group for review and analysis . Additional data on dengue were provided by the Philippines DoH NEC on 28 May 2012 [23] . The NEC Library , formerly the Field Epidemiology Training Program ( FETP ) Library ( http://nec . doh . gov . ph/index . php ? option=com_content&view=article&id=43&Itemid=58 ) , established in 1989 under the FETP project of the DoH , contains reports of local infectious disease outbreaks submitted to them by FETP Fellows and by Regional Surveillance Units . The reports of dengue disease outbreaks that occurred between 1987 and 2011 were manually searched , and data were collated and summarized using the data extraction instrument . The NEC also provided data from the urban area of Quezon City , NCR , and the rural area of Rizal , Region IV-A . Data included the population numbers and the number of dengue disease cases by year ( 1999–2011 ) and by age ( <1 year , 1–10 years , 11–20 years , 21–30 years , 31–40 years , >40 years ) . These data were integrated with the other data sources in the data extraction tool .
Among the included sources , no complete and comparable data were found for the entire review period . The most complete datasets for the number of cases of dengue disease in the Philippines and the number of dengue-related deaths were reported by the DoH [5] , [25]–[29] and the WHO [6] , [16] , [20] , [21] , [30]–[32] ( Figures 3A and 2B , Table S2 and Table S3 ) , although a number of sources did report similar , but isolated , data during the survey period [13] , [33] , [34] . An overall summary plot of these data would be of little value in identifying trends over time . Despite possible bias therefore it is useful to view the data made available during the review period from the WHO and DoH ( Figures 3A and 3B ) . These data show that the reported number of dengue disease cases fluctuated throughout the review period , with an overall increase in cases observed over time ( Figure 3A ) . There was a sharp rise in the number of cases in 2001 ( 23 , 235 cases ) compared with the previous and following year , and in a similar fashion high numbers of cases were also reported in 2003 ( 22 , 789 cases ) and 2007 ( 23 , 773 cases ) as shown by DoH data . The incidence per 100 , 000 population was 30 cases in 2001 , 28 . 1 cases in 2003 , and 28 . 2 cases in 2007 [5] , [25] , [29] . Possibly as a result of data extrapolation from incomplete submissions from some regions , the WHO data showed consistently higher numbers of cases than the DoH , but the same general pattern . A large increase in the number of cases was recorded in 2010 , with 131 , 976 and 173 , 033 cases reported by the DoH and the WHO , respectively , compared with 56 , 545 and 57 , 819 cases , respectively , in 2009 [30]–[33] . There were also a large number of cases reported by the DoH in 2011 ( 118 , 868 ) [31] . Overall , the CFR ranged from 0 . 5% to 1 . 7% [5] , [6] , [16] , [29] , [30] , [32] . There were 548 fatal cases in 2009 ( CFR 0 . 95% ) , increasing to 788–793 in 2010 ( CFR 0 . 60–0 . 94% ) [30] , [32] , [33] . Data on the severity of dengue disease cases were inconsistently reported over the review period . However , the available data from the DoH showed an increase in the number of DHF/DSS cases reported and the incidence of DHF/DSS per 100 , 000 population in the middle of the decade ( 2006–2008: 11 , 915–14 , 310 DHF/DSS cases , 14 . 1–17 . 7 per 100 , 000 population ) [29] . Data from the DoH also showed that peaks in the number of dengue disease-related deaths were observed at the beginning of the decade ( 2001: 641 deaths ) and from 2003 to 2006 ( 2003: 831 deaths; 2004: 761 deaths; 2005: 887 deaths; 2006: 1017 deaths ) . Overall , the CFR was in the range 0 . 5–1 . 7% ( DoH ) or 0 . 5–1 . 2% ( WHO ) and decreased after 2005 ( Figure 3B ) . The numbers of dengue disease cases being reported were highest in the most populated urban areas , such as NCR [25] , [28] , [29] , [35]–[40] . However , the incidence of dengue disease per 100 , 000 population varied by year and by region . The dengue disease incidence rates per 100 , 000 population were highest in the NCR in 2000 , CAR in 2001 , Region VI in 2002 , Region VII in 2007 , Region XI in 2003 , 2004 , 2008 and 2009 , and Region XII in 2005 and 2006 ( Table S4 ) [25] , [29] . Incidence rates for 2010 and 2011 were not available . However , in 2010 , the highest number of cases by region was in Western Visayas ( Region VI; 17 , 593 cases; 84 deaths; CFR 0 . 48% ) [36] . In 2011 , the highest number of cases by region was in NCR ( 15 , 427 cases; 93 deaths; CFR 0 . 60% ) , and the NCR area with the highest number of cases was Quezon City ( 4611 cases; 32 deaths; CFR 0 . 69% ) [36] . The highest numbers of fatal dengue disease cases were in the NCR in 2001 , 2003 , 2004 , 2005 , and 2006 ( 121 , 148 , 131 , 185 , and 345 cases , respectively ) [28] . By contrast , in 2002 , the region with the highest number of fatal dengue disease cases was Central Visayas ( Region VII; 107 cases ) [28] . The numbers of reported dengue disease cases were substantially higher in Quezon City than in Rizal , even though the populations are similar: 2 . 7 million and 2 . 5 million in Quezon City and Rizal , respectively , in 2011 [23] . Nevertheless similar patterns of reported dengue were seen in both regions with an increase in cases over the study period and higher numbers of cases in 2006 , 2008 , 2010 and 2011 compared with other years . The DoH has also recorded local outbreaks of dengue disease from 2000 from FETP Fellows reports and from Regional Surveillance Units reports [23] . At least one local outbreak was reported in each year and in all regions except Regions VIII and XIII . The regions with the highest numbers of local outbreaks reported were CAR ( 10 outbreaks ) , Region III ( five outbreaks ) , Region IV ( four outbreaks ) , and Region X ( four outbreaks ) . The outbreak with the highest number of dengue disease cases was reported in Zamboanga City , Region IX , in 2010 ( 2122 cases; 22 fatal cases; CFR 1 . 04% ) . Where data were available over the review period , children aged 5–14 years old represent the age group with the highest proportion of dengue disease in the Philippines ( Table S5 ) [6] , [20] , [25] , [29] , [36] , [41] , [42] . Dengue disease cases were reported by age group to the DoH in 2000–2003 and in 2005–2009 . In 2000–2003 and 2005–2009 , the highest proportions of cases were reported in individuals who were 5–14 years old ( 28 . 6–50 . 6% of cases ) , followed by those who were 15–49 years ( 21 . 2–37 . 3% % of cases ) and 1–4 years ( 15 . 4–31 . 1% of cases ) [29] . In 2010–2011 , the largest proportion of dengue disease cases reported to the DoH was in individuals aged 1–10 years ( around 25 , 800 of 70 , 204 [36 . 8%] cases ( value estimated from Figure 2 in Disease Surveillance Report , DoH , 2011 [36] ) . Where incidence data were available , the highest rates in 2000 , 2003 , and 2005 were reported in individuals who were 5–14 years old , followed by those who were <5 years old , and then 15–49 years old . In 2006 , the highest incidence rates were reported in individuals who were 0–4 years old , followed by those who were 5–14 years old , and then in those who were 15–49 years old ( Table S5 ) . In both Quezon City ( NCR ) and Rizal ( Region IV-A ) , there was a general increase in the numbers of dengue disease cases reported over time in each age group . The highest numbers of cases were reported in individuals aged 1–9 years , followed by 10–19 years , with these two age groups representing over 75% of reported cases in each year [23] . Furthermore , only 1% of patients with dengue admitted to San Lazaro Hospital , in Manila , Luzon were over the age of 35 years [43] . The WHO reported similar findings with respect to dengue disease incidence by age: in 2008 , of 7880 patients with dengue disease admitted to different sentinel hospitals nationwide from 1 January to 29 March , the median age was 12 years ( range <1 month to 87 years ) [20] . A prospective community-based study of dengue disease in infants 2–15 months of age was conducted from January 2007 to May 2009 in the semi-urban community of San Pablo , Laguna , Calabarzon ( Region IV-A ) . Between January 2007 and January 2008 , the modal age for symptomatic dengue disease in these infants was 8 months ( median 7 . 2 months ) [44] . The age-specific incidence of infant DHF was 0 . 5 per 1000 persons aged 3–8 months and zero among those aged ≥9 months [44] . The DoH has reported the numbers of dengue disease-related deaths by age group in 2003–2005 [25] . Over 80% of the fatal cases in each year occurred among individuals aged <20 years [25] . Among those aged <10 years , there were 477–562 fatal cases in 2003–2005 ( 62 . 7–66 . 1% of all dengue disease-related deaths ) . Data from the WHO showed that the majority of dengue disease-related deaths occurred among children aged <9 years [6] . Few data were available for CFR by age . Available data for 2003 and 2005 showed that the CFR decreased from age 0–4 years ( CFR range 0 . 29–0 . 37% ) to 5–14 years ( 0 . 18–0 . 23% ) and to 15–49 years ( 0 . 09–0 . 13% ) . By contrast , CFR increased for the age groups 50–64 years ( 0 . 13–0 . 17% ) and ≥65 years ( 0 . 42–0 . 95% ) . Data regarding the sex distribution of dengue disease in the Philippines are scarce and thus it is difficult to discern any distribution pattern . In one report from the WHO ( covering the period 1 January to 29 March , 2008 ) , the majority of dengue cases ( 53% ) were male [20] . Similar proportions were found in the analysis of urban Quezon City versus rural Rizal in 2000 to 2011 [23] . Another WHO report suggested that dengue disease cases and related deaths occur in approximately equal proportions among males and females [16] . The only other data found were from a 10-month , prospective cohort study of 42 dengue disease patients admitted to a tertiary referral hospital in Cardinal Santos Medical Center , San Juan ( NCR ) , from November 2006 to August 2007 , in which dengue disease occurred in more females ( 57 . 1% ) than males [45] . Few studies show any analysis of the seroepidemiology of dengue . Nevertheless , a longitudinal prospective cohort study of fever in 4441 infants in San Pablo Laguna conducted from January 2007 to May 2009 , showed that 11% of all presenting undifferentiated febrile illness was dengue ( 40 cases of dengue diagnosed out of 353 cases of fever ) . All cases of dengue but one were primary dengue ( as determined using IgG/IgM and paired sera ) and DHF was seen only in infants under 8 months of age . DENV-3 predominated in this cohort ( as noted below ) and infants with high levels of anti-dengue 3 antibodies at birth developed dengue infections later than infants with low levels of antibody at birth . The overall infection rate ( as determined by seroconversion in a subset of this cohort ) was nearly 12% , between January 2007 and January 2008 , and 87% of these dengue infections were asymptomatic or only mildly symptomatic [44] , [46] ( Table S1 ) . In a prospective study of children admitted with fever without a clear focus to St Lukes Medical Centre in Quezon City , Metro Manila , from January 1999 to December 2001 , 71 . 4% had dengue ( confirmed by IgM and/or RT-PCR ) and 1/3 had DHF [47] , [48] . Furthermore , in another prospective fever surveillance study of patients with a mean age of 18 years admitted to San Lazaro Hospital , in Manila , Luzon , 87% of those with fever without a clear focus of infection had dengue , 7% of the cases were primary dengue infections ( determined by IgM/IgG ELISA ) [43] . Studies published during the review period that observed DENV serotypes were hospital or community based and involved low numbers of cases; no published studies examined national or regional serotype distribution . All four DENV serotypes were reportedly present in the Philippines at some time during the review period ( Figure 4 ) [43]–[50] . DENV-1 and -2 appeared to be more prevalent ( 2000–2001 ) in a prospective study of hospitalized paediatric patients in NCR ( January 1999 to December 2001 ) [47] , [48] and in isolates from dengue disease outbreaks in the Philippines ( 1995–2002 ) [50] . In studies towards the end of the review period DENV-3 became more predominant [43]–[46] , [49] , DENV-4 was either not present [44] , [46]–[48] , [50] , or was present in up to 7% of the dengue disease cases [45] , [49] , [50] in the studies included in this review . The molecular epidemiology of DENV-2 isolated from patients with DF , DHF and DSS in the Philippines between 1995 and 2002 was examined by Salda et al . who have shown evolution from the Asian 2 genotype to the Cosmopolitan genotype , first identified in their sample of virus isolates in 1998 [50] . The genetic sequence of DENV-3 circulating in 2008–2010 outbreaks was characterized by Destura et al . to help assess the relationship between genotype mutations and the potential to cause outbreaks of severe or attenuated disease . The isolates did not fall into any of the groups of the reported genotypes suggesting the identification of a new genotype [51] . The LRG were aware of data that suggests the number of cases increasing 1–2 months after the onset of the rainy season , resulting in a peak of dengue cases in July to November , especially August . However , these are unpublished statistics and thus there is a gap in the available data that can address the influence of seasonal factors on the incidence of dengue disease . One study that assessed climactic factors associated with dengue disease incidence showed that high rainfall ( but not temperature ) was significantly associated with increased dengue disease incidence in Metro Manila in 1996–2005 [40] . Local outbreak data from FETP Fellows reports and from Regional Surveillance Units reports showed that dengue disease incidence was linked to flooding and/or to the unsuitable storage of water , e . g . , in open containers , which are potential breeding areas for mosquitoes [23] .
Seasonal and environmental factors affecting dengue in the Philippines are under-studied . Only one study attempts to correlate climatic factors with annual peaks in dengue cases , and this study was based in metro Manila . The Philippines has four distinct climate types across its extensive and diverse geography and further study is needed to better understand the seasonal patterns affecting the whole country . The high proportion of dengue disease and related deaths reported in children versus other age groups may reflect the age profile of the population; approximately one-third of the population is aged <15 years . Additionally , because dengue is highly endemic in the Philippines , most adults are immune . Understanding age distribution of dengue disease can aid the identification of groups with a high risk of dengue disease , provide information on age-related severity . From the data used in this review , no conclusions can be drawn regarding the sex distribution of dengue disease in the Philippines . Comprehensive national and regional data that describe the proportion of severe dengue disease cases , including hospitalizations and mortality , are lacking . The incidence of the severe forms of the disease , DHF/DSS , appeared to increase in the middle of the decade . The number of dengue disease-related deaths varied throughout the Philippines , with several peaks observed . Interestingly , the CFR also fluctuated . There are several potential reasons for this observation , including greater public awareness and early case detection , the possibility that certain dengue serotypes/genotypes cause less severe disease , and variations in clinical case management of dengue disease during 2000–2011 . Another important reason why severe cases may have spiked midway through the decade is that prior to 2005 the reporting forms did not facilitate the reporting of DHF and DSS separately from DF . The introduction of the 2009 WHO classification of dengue disease [17] may improve the completeness of severity data in the Philippines although it may take , several years before the majority of both public and private institutions use this classification and report their data accordingly . All four serotypes were present during 2000–2011 , with the co-circulating types varying temporally and spatially . However , few data were available , and study findings do not represent the national or regional distribution . The data suggest a shift towards a prevalence of the DENV-3 serotype towards the end of the review period . However , studies assessing DENV serotype distribution were mainly hospital or community based and involved low numbers of cases; no published studies examined national or regional serotype distribution . Robust surveillance of serotype distribution is essential to monitor changes in the relative prevalence of DENV serotypes ( or their variants ) and any potential effects this may have on dengue disease incidence or severity and to help predict epidemics . The presenting signs and symptoms in individuals with dengue disease are similar to those with other non-dengue acute undifferentiated febrile illnesses . Thus , in dengue-endemic regions clinicians should maintain a high index of suspicion for dengue disease . Available data on the laboratory confirmation of dengue disease cases were scarce , but showed that DENV antibody tests were used for most confirmed cases . Additionally , the few prospective fever surveillance studies that were reported during the review period showed that a variable proportion of patients with fever presenting to , or admitted to , hospital had dengue . This variation was most likely related to differences in study design , and a high proportion of children and adults admitted to hospital with fever , without a clear focus of infection had dengue . Serological analysis showed that most children and adults admitted to hospital were experiencing a secondary dengue infection ( determined by IgM/IgG ELISA ) . The only study to determine an incidence of infection ( in infants ) , showed an incidence of 12% , which is comparable to that seen in other prospective studies in highly endemic countries [52] . In the study by Capeding et al . , all cases of DHF were observed in younger infants ( aged <8 months ) [44] , consistent with previous observations that younger age groups are particularly vulnerable to severe disease [53] . Understanding the spectrum of dengue disease is essential to combating the disease . However , gaps in the epidemiological information available during 2000–2011 have been highlighted in this review and provide indications for avenues of future research . Comprehensive and continuous data are lacking for the review period , in particular national and regional age-stratified incidence rates and sex distribution data . This limits the possibility of making comparisons and drawing firm conclusions over the years , across regions , and among different ages . Although data are available on the number of dengue disease cases nationally and regionally , there are relatively few reported data on the incidence of dengue disease per 100 , 000 population . There were relatively few published studies of regional data on dengue disease epidemiology for the period 2007–2009 . Availability of comprehensive data sets would allow a more systematic evaluation of the trends and informed assessments about their impact on surveillance procedures and control measures . Further studies exploring DENV serotype distribution and seroprevalence data as well as the associations between DENV serotype and disease severity are necessary . For example , studies in some countries have demonstrated that DENV-3 is associated with a significant proportion of severe complications , and the displacement in the predominant serotype has been related to local outbreaks of disease [54] . This knowledge can help guide the introduction of additional public health measures , including vector control intervention , educational communication , and the adequate provision of medical supplies . Estimates of the extent of dengue disease under-reporting would also be valuable . Whilst the epidemiology of dengue disease varied between the three island groups both spatially and temporally , the data do not reveal any geographical patterns at an island level in incidence or other epidemiologic parameters . Although the regions with the highest incidence , morbidity , and mortality were generally urban centres , regions with the highest incidence rates and peaks in the number of dengue disease-related deaths and the intensity and magnitude of dengue cases changed each year . The Philippines is severely affected by extreme weather events and is vulnerable to climate change . Vector-borne diseases , such as dengue disease , may be particularly sensitive to both periodic fluctuations and sustained changes in global and local climates . A programme of regional dengue disease burden surveillance studies will provide scientific data on which to base decisions regarding priorities , resource allocation , and geographical areas for targeting vector control . In addition , studies on the effects of continuing urbanization ( including the effects of human density and population movement ) as well as information relating to the effects of housing conditions , water supplies , and waste management on the incidence of the disease would be useful . Strengths of this systematic review include the complementary information provided from national surveillance data and local studies . However , there are several limitations to note . There is a general scarcity of published information on the epidemiology of dengue disease in the Philippines . Additionally , some of the studies identified may have weaknesses , such as inadequately described case selection and a lack of sound statistical methods , which were not accounted for as no assessment of quality of evidence was conducted . Another limitation of the data generated by this review is the discrepancies in the reported dengue disease rates between the WHO and the DoH . As already noted , this may be due to data extrapolation from incomplete submissions from some regions of the Philippines . There are also inherent limitations associated with the surveillance data due to changes in reporting behaviour , the systems used , misclassifications , and under-reporting [3] , [13] . Importantly , a proportion of the increase in the number of dengue disease cases towards the end of the review period may be an artefact of the changes in the surveillance system , including the separate reporting of DF , DHF , and DSS since 2005 , and the transition to the all-case reporting surveillance system with the increase from 250–400 sentinel hospitals to a network of up to 1662 disease reporting units since 2006 [33] . Another limitation of this review is that the number of cases of dengue disease may be under-reported by the surveillance system in the Philippines . Although some hospitals may over-diagnose dengue disease cases , there may be an overall under-reporting of cases due to variability in defining dengue disease , the passive surveillance system used in the Philippines [3] , [13] , the sentinel system used prior to 2006 [33] , and the exclusion of data from privately treated patients . Different applications or interpretations of case definitions over the review period limit the ability to make valid temporal comparisons . This long-term review highlights an increase in the reported incidence of dengue disease in the Philippines . All regions reported cases of dengue disease , although more cases were reported from the most populated , urbanized areas . The reported number of cases of dengue disease fluctuated throughout 2000–2011 , with an overall increase in cases over time . The highest incidence of dengue disease cases per 100 , 000 population was reported in children 5–14 years of age , followed by children 0–4 years old , and 80% of all dengue disease-related deaths were reported in individuals aged <20 years . In the regions of the Philippines , the incidence of dengue disease per 100 , 000 population varied , with particularly high incidences observed in the regions of the island of Mindanao . The increasing incidence of dengue disease may be related to a growing population , increasing urbanization , improvements in surveillance , and the limited success of vector control measures . All four DENV serotypes were present; however , there was a shift to DENV-3 towards the end of the literature review period . Recent improvements to the surveillance system and more consistent use of the 2009 WHO classification of dengue disease [17] may help standardize the approach to data collection and reporting of dengue disease in the Philippines .
|
Dengue disease is a tropical and subtropical mosquito-borne viral illness and is a major health concern in the Philippines . To determine the dengue disease burden in the Philippines and identify gaps and future research needs , we conducted a literature analysis and review to describe the epidemiology of dengue disease . We used well-defined methods to search and identify relevant research conducted between 2000 and 2011 . This long-term review highlights an increase in the reported incidence of dengue disease in the Philippines . The rising incidence of dengue disease may be related to a growing population , increasing urbanization , improvements in surveillance , and the limited success of vector control measures . Gaps in the epidemiological information available in the Philippines during the period 2000–2011 include comprehensive national and regional data that describe the proportion of severe dengue disease , including hospitalizations and mortality , and incidence data per 100 , 000 population . More comprehensive data are also needed for age , serotype , and seroprevalence on both national and regional levels . The data presented enable the observation of epidemiological characteristics , both within and across years . Such assessments are essential at national and regional levels to improve both preparedness and response activities relating to dengue disease outbreaks .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"epidemiology"
] |
2014
|
Epidemiology of Dengue Disease in the Philippines (2000–2011): A Systematic Literature Review
|
Post-translational modifications ( PTMs ) add a further layer of complexity to the proteome and regulate a wide range of cellular protein functions . With the increasing number of known PTM sites , it becomes imperative to understand their functional interplays . In this study , we proposed a novel analytical strategy to explore functional relationships between PTM sites by testing their tendency to be modified together ( co-occurrence ) under the same condition , and applied it to proteome-wide human phosphorylation data collected under 88 different laboratory or physiological conditions . Co-occurring phosphorylation occurs significantly more frequently than randomly expected and include many known examples of cross-talk or functional connections . Such pairs , either within the same phosphoprotein or between interacting partners , are more likely to be in sequence or structural proximity , be phosphorylated by the same kinases , participate in similar biological processes , and show residue co-evolution across vertebrates . In addition , we also found that their co-occurrence states tend to be conserved in orthologous phosphosites in the mouse proteome . Together , our results support that the co-occurring phosphorylation are functionally associated . Comparison with existing methods further suggests that co-occurrence analysis can be a useful complement to uncover novel functional associations between PTM sites .
In addition to gene expression and translation , post-translational modification ( PTM ) represent another level of regulation that expands the functional capacity of proteins . It play a crucial role in a plethora of biological processes including regulation of gene expression [1] , modulation of enzymatic activity [2 , 3] , and control of protein-protein interaction ( PPI ) [4] . More than 400 different types of PTMs have been discovered , including phosphorylation , acetylation , methylation , ubiquitination and SUMOylation , with phosphorylation at serine/ threonine/ tyrosine ( S/T/Y ) residues being the most abundant and well characterized type [5] . Different types of PTMs usually cooperate with each other to carry out specific functions . PTM at different sites of the same protein can physically interact with each other or jointly carry out a specific biological function , referred to as PTM cross-talk [6] . For example , in the human p53 protein , phosphorylation of S37 promotes phosphorylation of S33 which together activate p53’s transcriptional activity [7] . In the human CDC25C ( cell division cyclin 25 homolog c ) protein , phosphorylation at S214 prevents phosphorylation at its nearby site S216 and promotes cells to enter mitosis under cancerous conditions [8] . PTM cross-talks are not limited to within the same protein . For instance , ubiquitination of histone H2B forms the basis for the methylation of K79 of histone H3 [9–11] . Residue-specific cross-talk has also been shown between phosphorylation of S21 in EZH2 ( enhancer of zeste homolog 2 ) and methylation of L27 in histone H3 [12] . Thanks to the recent advances of mass spectrometry ( MS ) technology , the number of known PTM sites has increased rapidly [13] . It motivated several computational studies to systematically characterize their functional relationships . Functional associations between PTM types could be revealed by statistical enrichment of different PTM type combinations observed within proteins [14] , although it did not delineate relationships between individual PTM sites . At individual site level , target sites modified by more than one types of PTM is the simplest case of cross-talk [15] . Apart from that , PTMs occurring in proximity were presumed to interact and used to identify motifs [16] . Indeed , phosphor-acceptor residue nearby a modified lysine ( L ) was found significantly more likely to be phosphorylated [17] . And the sequence and spatial distances between phosphosites are closer than expected [18] . Some disordered protein regions with very dense PTM aggregation were recently highlighted as important for combinatorial PTM regulation [15] . There was also statistical evidence supporting co-operation among locally clustered phosphorylations [19] . In addition to sequence distance , residue co-evolution between PTM sites represent another type of surrogate for their functional associations , and was used to establish a proteome-wide PTM type association network [20] . Previously , we developed a naive Bayesian classifier for PTM cross-talk prediction , which integrated sequence and structural distances , co-localization within same disordered region , as well as residue and modification co-evolution [21] . Based on a manually curated cross-talk data set , we demonstrated that integration of different features achieved better performance for cross-talk prediction than relying on individual features . In addition , cross-talk between PTM sites can also be revealed from specialized MS experiments . For example , “top-down” or “middle-down” MS strategy can be used to directly identify PTMs co-existing within the same peptide segments [22 , 23] . Quantitative MS data can also be exploited to infer co-existing PTMs based on their co-varying modification levels that change interdependently across different experimental conditions [24 , 25] . However , both strategies cannot scale to the entire proteome , and were only applied to study the interactions among various modifications on histones . Another recent study investigated the co-modification of phosphorylation and ubiquitination in the entire proteome by developing a novel MS experimental strategy that enrich both modification types at the same time [26] . But it is unclear if similar methods can also be developed to study other PTM combinations . Furthermore , this strategy cannot verify if two distant co-modification exists on the same peptide sequence , and it falls short of identifying PTM cross-talks between different proteins . Publicly available proteome-wide PTM data are mostly generated by “bottom-up” MS strategies that allow high throughput protein identification but lose connections between modifications [13 , 27] . Such data are not quantitative only providing binary modification on-off status for each PTM site . Motivated by those early studies , we hypothesized that the correlation of binary modification status between two PTM sites can also suggest functional association . Following [20] , functional association here is a broad concept that not only stands for cross-talk but also describes general association like involvement in the same signaling pathway or biological function . We benchmarked this idea on human phosphorylation data , because it was the only PTM type with proteome-wide coverage over large number of conditions by the time of this study . We showed that co-occurrence of phosphorylation can be used to distinguish known functionally connected phosphosite pairs from negative ones . Then we applied method to all pairwise combinations of phosphosite sites , either within proteins or between interacting partners , to identify pairs that tend to be modified under the same condition ( co-occurrence ) . We systematically compared the observed co-occurrence with randomized data set . Site pairs showing significant co-occurring phosphorylation status were then characterized by their location preference , sharing of functional annotations and catalytic kinases , and residue co-evolution . We also compared predictions with other existing methods to highlight differences .
To make use of the phosphorylation status at different conditions , we assembled human phosphorylation identified from high-throughput MS analyses across 88 different conditions from PhosphoSitePlus [28] . They include 16 human tissues as well as 28 cultured cell lines , 44 of which are disease cells ( S1 Table ) . The conditions were selected with a minimal 1000 modification sites proteome-wide to ensure coverage . In total , the collected data contains 165 , 201 potential phosphosites ( 58 . 7% S , 23 . 6% T and 17 . 8% Y ) of 17 , 819 proteins , along with their modification status across all conditions ( S1 Dataset ) . They include 55 , 145 sites of 10 , 868 proteins ( 56 . 2% S , 18 . 5% T and 25 . 3%Y ) that are phosphorylated in at least 3 different conditions ( high-frequency sites ) . Compared with the sites phosphorylated in less than 3 conditions ( low frequency sites ) , high-frequency sites are preferentially located in disordered protein regions ( 79 . 8% vs . 75 . 0% , p-value<1E-5 by permutation test ) , evolutionary more conserved ( median Residue Conservation Score ( RCS ) : 0 . 87 vs . 0 . 82 , p-value<1E-5 by permutation test ) , and showed 3-fold increase of annotated functional terms ( p-value<1E-5 by permutation test ) . The results suggest high-frequency phosphosites may be functionally more important , consistent with the previous study that suggested higher proportion of functional phosphorylations among high abundance sites [29] . To measure the tendency of two sites being phosphorylated at the same condition , we cross-tabulated the times that a pair of sites are phosphorylated across different conditions by a contingency table and calculated p-value from one-sided Fisher’s exact test ( FET ) . Since low-frequency sites are unlikely to show statistically significant co-occurrence with other sites , we only included high-frequency sites in the co-occurrence analysis . As a known example , transcription factor c-Jun is phosphorylated at four high-frequency sites in our data ( Fig 1A ) , three of which ( T239 , S243 and S249 ) are clustered in a short segment upstream of its DNA binding domain . They are usually phosphorylated together by GSK-3 in epithelial or fibroblast cells to inhibit c-Jun activity in resting cell states [30] . Consistent with their functional cross-talk , phosphorylation status at T239 , S243 and S249 are more likely to occur together ( pairwise FET p-values: 1 . 8E-4 to 6 . 4E-6 ) than their combinations with S58 ( FET p-values: >1E-3 ) . As a proof of principle , we first examined the phosphosites which showed experimental validated evidence of cross-talk which were collected from literature as part of our previous study [21] . Twenty-two of them are composed of high-frequency sites in our collected data . Compared with all phosphosite pairs within proteins , their FET p-values are predominantly distributed at the lower end ( median p-value: 5 . 0E-5; Fig 1B ) . We also made use of curated functional annotations for phosphosites and defined homo-functional ( hetero-functional ) pair as two sites that execute the same ( different ) biological function ( s ) within the same protein ( Materials and Methods ) . The homo-functional pairs defined in this way are expected to enrich with functional associations; whereas the hetero-functional pairs are more likely functionally unrelated . Consistent with the expectation , the FET p-values of 380 homo-functional pairs are similar as known cross-talk pairs and significantly lower than those of 35 hetero-functional pairs ( median: 9 . 74E-5 vs . 1 . 30E-2; Fig 1C ) . Since homo-functional pairs have only five overlaps with the known cross-talk set , we then combined known cross-talk and homo-functional pairs to form the positive set , and used hetero-functional pairs as the negative set ( S2 Table ) . The co-occurrence test can be used to distinguish the two classes ( Table 1 ) , achieving an area under the ROC curve of 0 . 713 ( S1 Fig ) . The above analyses suggest that we can identify functionally associated phosphosite pairs by using the co-occurrence of their modification status across conditions . We applied co-occurrence test to all pairwise combinations of high-frequency phosphosites within proteins , resulting in a total of 521 , 321 site pairs in 10 , 868 proteins . At different p-value thresholds to define co-occurring pairs , we consistently observed higher proportions among all pairs than hetero-functional pairs ( Table 1 ) , suggesting that the identified co-occurring pairs enrich true functional associations . We also noted higher than expected proportions of co-occurrence in hetero-functional pairs . It can be explained by our incomplete knowledge of functional associations between phosphosite and/or due to confounding factors . For example , different protein abundances across conditions due to biological reasons can induce artificial co-occurrence of phosphorylation states . When a peptide segment is at high abundance level in certain conditions , its phosphorylation sites are likely to be all detected . In some other conditions , when it is at low abundance , none of the phosphorylation sites could be detected . The major effect is to inflate the significance level of association test by erroneously taken missing data as non-modification status . Although one can address this issue by incorporating protein abundance levels , this information is only available for less than 20% conditions used in the co-occurrence analysis ( based on PaxDB ) . Furthermore , the data set we collected were generated by heterogeneous MS experiments , which employed different enrichment strategies and could introduce further variations to the digested peptide segments . The technological issue may be the major confounding factor , because we did not observe higher proportions of co-occurring pairs within housekeeping proteins [31] which are highly expressed in all tissues ( S3 Table ) . To evaluate the effect of confounding factors , we compared the observed proportions of co-occurring pairs in the original data under different p-value cutoffs to those of randomized data in which the number of phosphorylations at the protein level or within short peptide segments were kept the same as observed . We first shuffled phosphorylation states among all potential phosphosites of each protein under each condition . The procedure was repeated 100 times , resulting in 135 , 954 , 681 high-frequency within-protein phosphosite pairs . We found consistent enrichments of co-occurring pairs in the original data across different p-value thresholds ( Table 2 ) . For example , at the most stringent threshold of 1E-6 , 6 . 68% ( 34 , 835 ) pairs in 3 , 722 proteins of the original data show co-occurring phosphorylation as compared with only 0 . 12% ( 169 , 617 ) pairs in randomized data . The fold increase of co-occurring pairs in the original data decrease with relaxing p-value thresholds ( Table 2 ) . We next examined protein sub-sequences that range from 10 to 100 amino acids long and contain at least 5 potential phosphosites across conditions . We treated those short segments as individual proteins and generated randomized data sets as above . Again , consistent enrichment of co-occurrence can be seen in the original data . The enrichment level generally decrease with shorter fragment length ( S4 Table ) . For example , at the fragment length of 10 and p-value cutoff 1E-5 , we identified 2 , 357 ( 37 . 6% ) co-occurring pairs compared to 147 , 290 ( 15 . 9% ) in the randomized data . By contrast , at the length of 100 with the same p-value cutoff , the corresponding proportions in the original and randomized data are 21 . 3% and 2 . 1% . However , phosphorylation sites at proximity are known a priori to have functional association , so randomized data in this case are not devoid of functional association . Nevertheless , we still observed higher fold increase of co-occurrence especially at more stringent p-value thresholds ( S4 Table ) . Our permutation test procedures above fixed the observed number of phosphorylation sites at the protein level or within peptide segments . It can give a rough estimate of false discoveries if confounding factors mainly influences the observed number of phosphorylations , which may not be true in the process of generating real data . Despite of the caveat , we believe the observed co-occurring phosphorylation pairs can capture the functional association given the results on the known positive and negative sets . And we opted to use a stringent p-value threshold of 1E-5 to define co-occurring pairs for further functional characterization . This threshold recovers 42 . 3% of known cross-talk and homo-functional pairs and 11 . 4% of hetero-functional pairs; achieving a proper tradeoff between sensitivity and specificity . At the p-value threshold 1E-5 , we identified 63 , 760 ( 12 . 23% ) co-occurring pairs in 5 , 109 human proteins ( S2 Dataset ) . For comparison , we also defined 94 , 391 within-protein phosphorylation pairs as controls whose FET p-values are no less than 0 . 5 . Compared with control pairs , sites in co-occurring pairs tend to be located closer to each other in primary protein sequences ( median: 166 vs . 415 , p-value<1E-5 by permutation test; Fig 2A ) . Using the protein structure data from the PDB database [32] , we were able to calculate the 3D structural distances for 1 , 251 co-occurring pairs and 2 , 022 control pairs . Phosphosites in co-occurring pairs also situated significantly closer in 3D space than control pairs ( median: 13 . 58 Å vs . 28 . 21 Å , p-value<1E-5 by permutation test; Fig 2B ) . The results are consistent with the fact that modification sites that are closer to each other are more likely to have physical interactions and functionally associated . We then retrieved expert curated annotations from literatures for selected phosphosites . Those include 4 , 971 terms of biological process for 3 , 133 sites , 8 , 434 terns of molecular function for 5 , 186 sites , and 364 different kinases for 7 , 119 sites . Compared with controls , co-occurring pairs are more likely to contain both sites with annotations ( 1083 vs . 256 , FET p-value <1E-5 ) . Among pairs with both sites annotated , co-occurring pairs tend to share at least one annotation ( biological process: 93 . 80% vs . 64 . 62% , FET p-value = 1 . 2E-6; molecular function: 92 . 59% vs . 79 . 63% , FET p-value = 1 . 2E-4 ) , and are more likely catalyzed by the same protein kinase ( 74 . 72% vs . 28 . 43% , FET p-value <1E-5 ) . The top enriched functions among co-occurring pairs include protein degradation , induced cell growth , and cell mobility . We devised a score to measure the sharing of annotations ( kinases ) that accounts for the number of terms ( kinases ) annotated to each site and information content of each term ( kinase ) ( Materials and Methods ) , and demonstrated the increased levels of function sharing among co-occurring pairs ( p-values<1E-5; Fig 2C–2E ) . To account for the sequence distance when comparing annotation sharing , we selected a subset of control pairs with similar distribution of sequence distances as co-occurring pairs ( S2 Fig ) , and found the difference between co-occurring and distance matched controls remained significant ( p-values <1E-5; Fig 2C–2E ) . Together , the results suggest co-occurring pairs contains more sites of known functions , tend to be involved in the same biological pathways , and catalyzed by similar kinases , which cannot be fully explained by their physical proximity . Because only a small proportion of phosphorylation sites had literature annotations , the functional sharing can only be analyzed for a limited number of phosphosites . To extend the functional analysis , the co-evolution of a pair of modified residues can be used as a proxy for their functional association [20] . We mapped all high-frequency phosphorylation sites to the orthologous positions across vertebrates using sequence alignments from eggNOG database [33] . Residue co-evolving scores as measured by normalized mutual information ( nMI; Materials and Methods ) could be calculated for 51 , 370 co-occurring pairs and 63 , 400 control pairs . Co-evolving scores of the co-occurring pairs were significantly higher than those of the control pairs ( median nMI: 0 . 422 vs . 0 . 312 , permutation test p-value < 1E-5; Fig 2F ) . Similarly , the increasing level of co-evolution in co-occurring pairs remained significant after accounting for their sequence distances ( median nMI: 0 . 422 vs . 0 . 346 , permutation test p-value < 1E-5; Fig 2F ) . We also compared evolutionary conservation of the co-occurring and the control pairs . As expected , modification sites within co-occurring pairs are slightly more conserved than control pairs ( median RCS: 0 . 573 vs . 0 . 557 , p-value = 2 . 4E-3 by permutation test; S3A Fig ) . The differences in conservation levels increases after excluding sites shared by the co-occurring and the control pairs ( median RCS: 0 . 623 vs . 0 . 571; p-value<1E-5 by permutation test; S3B Fig ) . In addition to residue co-evolution , we also tested if mouse orthologous phosphosites of human co-occurring pairs show tendency of co-occurrence of phosphorylation status . To this end , we collected 67 , 555 mouse phosphosites in 10 , 237 proteins across 34 different conditions . After removing low frequency sites ( phosphorylated in less than 3 conditions ) , 32 , 887 phosphosites remained for co-occurrence analysis . We could map 23 , 918 ( 37 . 5% ) co-occurring and 5 , 147 ( 5 . 45% ) controls pairs from human phosphoproteins to the mouse orthologous phosphosites . Co-occurrence analysis in the mouse data shows FET p-values for mouse orthologs of the co-occurring pairs are significantly lower than that of control pairs ( p-value < 1E-5 by permutation test; S4 Fig ) . It suggests that functional association between phosphosites captured by the phosphorylation states co-occurrence tends to be conserved from human to mouse provided their phosphorylation status are conserved . Together , the results above lend further support to the notion that co-occurrence of phosphorylation status can be used to infer functional association between phosphosites . We note in passing that all the above results are quantitatively similar when the p-value threshold used to define co-occurring pairs was changed to 1E-4 ( S5 Fig ) or 1E-6 ( S6 Fig ) . Functional associations between phosphorylation sites not only exist within same protein , but also between different proteins . Phosphorylation and other types PTMs are known to mediate the binding between PPI partners [34] . And recent studies have shown that phosphoproteins had more PPI partners than non-phosphorylated ones and both interacting proteins tend to be phosphorylated [35 , 36] . It is also well characterized in signal transduction pathways that phosphorylation activates kinases to phosphorylate their substrates ( kinase cascade ) . Some stable kinase-substrate relationships can also be captured by PPI . So we extend the co-occurrence analysis to the phosphosites between interacting proteins . We used 13 , 944 experimental validated high quality PPI pairs from CCSB Human Binary Interactome database [37] . A total of 55 , 145 high-frequency sites could be mapped to the CCSB PPI set , resulting in 132 , 360 pairs between 2 , 959 interacting partners . One-sided FET was performed to test the tendency of co-occurrence of modification status between each phosphosite pairs . As within-protein analyses , we first contrast phosphosite pairs that have higher chances of functional associations with those most likely unrelated . One recent computational study found that phosphorylations are selectively accumulated in protein complexes , and some complexes tend to integrate phosphorylation signals on distinct subunits [15] . So we examined the co-occurrence between phosphorylations from different sub-units of phosphorylation enriched complexes reported by that study . A total of 3 , 654 phosphosite pairs can be mapped to 30 protein pairs within same complexes ( positive set ) ; and 124 , 594 phosphosite pairs mapped to 4 , 719 protein pairs that are not part of the complexes ( negative set ) . Phosphosite pairs in the positive set are enriched with small FET p-values as compared with the negative set ( Fig 3A , Table 3 ) . It is consistent with the functional associations between phosphosites within some of these complexes . Phosphorylation states co-occurrences between interacting proteins may also be confounded by the co-varying abundance level of interacting proteins or their digested peptides across conditions . Using the same permutation approach in which the total number of phosphosites of each protein was kept fixed , we also observed a significant excess of observed co-occurring pairs compared with randomized data , with decreasing level of fold increase at relaxing p-value thresholds ( Table 4 ) . We also examined short peptide fragments ( 10–100 amino acids ) containing at least 5 potential phosphosites . By controlling the observed number of phosphorylations within each short segment , we can observe a higher proportion of co-occurring pairs from short segments between interacting proteins than random expectations ( S5 Table ) . The same caveat as in the within-protein analysis applies to interpret the result of randomization . Compared to within-protein analysis at the same p-value threshold , the observed proportion of co-occurring pairs between interacting proteins is much lower . For example , at the p-value cutoff 1E-5 , 2 . 73% of all high-frequency phosphosite pairs ( and 7 . 17% of those within same phosphorylation enriched complexes ) were identified as co-occurring pairs , whereas 12 . 23% were found within proteins . Notably , the proportions of identified co-occurring pairs among all pairs are consistently higher than those of the negative set ( Table 3 ) , although phosphosite pairs in the negative set defined above are not fully unrelated in functions , suggesting that co-occurrence analysis can still uncover some functional associations . For further functional characterization , we defined co-occurring phosphorylation pairs between interacting proteins by FET p-value less than 1E-5 and control pairs by p-value no less than 0 . 5 . A total of 3 , 610 co-occurring pairs ( S3 Dataset ) and 53 , 016 control pairs are identified and used in the following comparisons . To evaluate if two phosphosites of co-occurring pairs tend to be closer in space , we test whether two sites of co-occurring pairs are more likely to co-localize at the interaction interfaces of interacting proteins . Structurally resolved interactions for 6 , 585 human protein pairs were obtained from the INstruct database [38] . A total of 427 co-occurring pairs are found between protein pairs with interaction domain information , and 56 . 44% of them are in the interacting interfaces . By contrast , among 1872 control pairs mapped to Instruct , only 32 . 32% of them are in the interfaces ( Fig 3B ) . To test if co-occurring pairs tend to be catalyzed by similar kinases , we used a computational approach ( Materials and Methods ) to link phosphosites to 39 different kinases . In total , kinase information could be obtained for both sites of 882 co-occurring pairs and 6 , 446 control pairs . Co-occurring pairs are more likely to share at least one kinase ( 34 . 92% vs . 14 . 71% , p-value<1E-5 by FET ) and have higher kinase sharing scores ( Fig 3C ) . Together , the above results can be interpreted that physically binding proteins tend to be phosphorylated by the same enzyme , and their structural proximity would facilitate this process . We also analyzed residue co-evolution and conservation of co-occurrence in mouse proteome . For phosphosite pairs whose co-evolving score can be calculated ( 54 , 46% co-occurring pairs and 57 . 88% control pairs ) , the co-occurring pairs have slightly higher nMI scores than controls ( nMI: 0 . 2 vs . 0 . 19 , p-value = 0 . 019 ) . For phosphosites whose phosphorylation status are conserved in the mouse orthologous positions , they contain 625 co-occurring and 1 , 442 control pairs . FET p-values of the co-occurrence test in mouse proteome were significantly lower for the orthologs of co-occurring pairs than control pairs ( median: 0 . 00193 vs . 0 . 409 , p-value<1e-5 by permutation test; Fig 3D ) . The results resemble those of within-protein pairs showing co-occurring phosphosite pairs are more likely to co-evolve and their co-occurrence states are conserved in mouse . The PTMcode database ( http://ptmcode . embl . de ) contains PTM associations of different PTM types collected based on multiple evidence channels . Among phosphosites , their functional relationships were computationally predicted based on residue co-evolution and space proximity in 3D structure [39] , with the overwhelming majority ( >99% ) based on co-evolution . To compare with PTMcode , we focus on the common set of 4 , 617 proteins on which the functional relationship between phosphosites can be analyzed by both PTMcode and our study . On these genes , 504 , 849 phosphosite pairs were annotated by PTMcode as functionally associated , and 30 , 779 phosphosite pairs were identified as co-occurring pairs ( defined by FET p-value <1E-5 ) . The much higher number of pairs identified by PTMcode mainly because co-occurrence analysis is limited to high-frequency phosphosites . Indeed , only 62 , 732 ( 12 . 43% ) of pairs identified by PTMcode are composed of both high-frequency sites . Intersecting PTMcode and co-occurring pairs results in an overlap of only 9 , 809 pairs . We further compared 495 , 040 pairs specific to PTMcode with 20 , 970 pairs specific to co-occurrence analysis . Co-occurrence specific predictions contain higher proportion of pairs with both functional annotated sites than PTMcode specific predictions ( 313 ( 1 . 49% ) vs . 2 , 378 ( 0 . 48% ) , p-value<1E-5 by FET ) , and higher proportion of pairs with kinase information ( 334 ( 1 . 59% ) vs . 4 , 478 ( 0 . 90% ) , p-value<1E-5 by FET ) . Among pairs with functional annotations , co-occurrence specific pairs more likely to share functional annotations ( 286 ( 91 . 37% ) vs . 1 , 736 ( 73 . 00% ) , p-value<1E-5 ) and at least one catalytic kinase ( 263 ( 78 . 74% ) vs . 2 , 761 ( 61 . 66% ) , p-value<1E-5 ) , and have slightly higher functional sharing scores ( S7A–S7C Fig ) . PTMcode v2 also predicted functional associations for PTM sites between high-confidence interacting proteins annotated by the STRING database [40] , which comprehensively catalogs known and predicted interacting proteins . Given the difference of PPI sets and data requirement for different methods , comparison between them is difficult . Here we focus on the 102 protein pairs that were analyzed by both PTMcode and by our study . On these protein pairs , 9 , 177 were annotated as functionally associated by PTMcode , 219 were identified as co-occurring pairs , with only 68 pairs in common . Bioinformatics analysis predicted kinases for 31 ( 14 . 16% ) of co-occurrence specific pairs and 214 ( 2 . 33% ) PTMcode specific pairs . Co-occurrence specific pairs are more likely to be catalyzed by the same predicted kinase ( 11 ( 35 . 48% ) vs . 46 ( 21 . 50% ) ) , and have higher kinase sharing scores ( S7D Fig ) . The results indicated that although the application of co-occurrence is limited by the availability of the data which will continue to expand in the future , it can be complementary to the existing method like residue co-evolution in predicting functionally associated phosphorylation . And co-occurrence analysis is more likely to identify site pairs with share functional annotations and catalyzed by the same kinase .
More than half of all human proteins can be phosphorylated . And phosphorylation dynamically regulates enzymatic activity , protein stability , subcellular localization , and transmit signals to downstream pathways , etc . [41] . Its function can be fine-tuned by multiple phosphorylation sites within protein or protein complex [42] . To better understand function relationships between different phosphosites , in this study , we exploited co-occurrence of phosphorylation status across conditions from public available high-throughput MS data . To mitigate the influence of confounding factors , we opted to use a stringent p-value threshold in statistical test . And a series of simulations were performed to systematically compared the number of identified co-occurring phosphosite pairs with random expectation controlling for the number of phosphorylations within protein or short peptide segments . Although we consistently observed higher proportion of co-occurring pairs than random permutation , we did not derive false discovery rates from this comparison because randomization may not fully capture the effect of confounding factors and would most likely under-estimate false positives . In the future , incorporating protein abundance information and applying uniform MS protocols across conditions can better address this issue . We also benchmarked the discriminative performance of the co-occurrence p-value in classifying the positive set that are enriched for known functional associations and the negative set that are more likely to be functionally unrelated . The performance based on this type of analysis should be interpreted as a lower bound , given the lack of golden standard for the positive and negative sets . Taken together the method’s ability to distinguish the positive and negative sets , higher discovery rates in the original than randomized data , and high a priori functional association between phosphosites within proteins , we believe the identified co-occurring pairs within proteins are enriched for functional associations . We then sought additional evidence to support the functional relevance of the identified co-occurring pairs by their sequence/structural proximity and residue co-evolution , which are also two commonly used measures of functional association [16 , 20] . Phosphorylations closer to each other have higher chance of physical interaction . Indeed , for more than two thirds of known cross-talk pairs , two phosphosites are within 20 amino acids . Common mechanisms of functional association include that phosphorylation at one site facilitate the phosphorylation the other site ( e . g . , S33 and S37 of p53 [7] ) , or several phosphosites need to be simultaneously phosphorylated to fulfill a molecular function ( e . g . , T239 , S243 , S249 of c-Jun [30] , Y342 and Y346 of Sky [43] ) . Functional associations are certainly not limited to nearby phosphosites . Among 63 , 760 significant co-occurring pairs within proteins , the primary sequence distance is less than 20 amino acids for 26 . 77% of all pairs , and more than 57 . 63% of them are separated by at least 100 amino acids . A similar proportion ( 60 . 27% ) were also observed for functionally associated phosphosite pairs in PTMcode [39] identified by the co-evolution method . Our comparison with PTMcode showed limited overlap and suggested co-occurrence analysis can be used as a complement . Identified co-occurring pairs include several well characterized long-range functional cross-talks , including Y707 and Y806 of CDCP1 that are phosphorylated by Src family kinases ( SFKs ) and activate CDCP1 to promote cell growth and SFK activities [44] , S22 and S390 of lamin-A that are phosphorylated by BGLF4 and promote the reorganization of the nuclear lamina [45] , and S612 and T365 of Rb protein which function together to prevent its association with E2F transcription factor [46] . Notably , the first two examples were not uncovered by PTMcode . While it is plausible that those long range functional cross-talks play roles in allosteric and orthosteric regulation of proteins [47] , another possibility could be that a group of phosphosites in a protein region contributes to the modification collectively through an aggregate property irrespective of precise locations [48 , 49] , e . g . through bulk electrostatics [50] . Under this model , natural selection would act to maintain the total number of phosphosites but individual phosphosites may not be conserved [51] . Co-evolution may fail to identify the functional association in such cases . A well-studied example is pre-replication complex: several sub-units in human ( CDC6 , CDT1 , MCM2 , MCM4 , and ORC1 ) contain cluster of phosphosites that are phosphorylated by CDKs . Those phosphosites showed rapid evolutionary turnover even when the local cluster of site is preserved . Co-occurrence analysis identified functional associated pairs in all five proteins , notably no evidence of co-evolution for phosphosites of ORC1 and CDT1 . Other examples of clustered co-occurring phosphorylations include DNA repair protein ERCC5 , DNA polymerase subunit RFC1 , RNA polymerase subunit POLR2A , etc . Throughout this manuscript , we focus on co-occurrence or positive correlation of modification status , because we found few negative correlations in the original and randomized data ( S6 Table ) . Although most known cross-talks between phosphorylations are positive ( activating or co-operative ) , negative ( inhibit or steric hindrance ) example does exist , for example the mutual inhibitory between phosphorylation of S214 and S216 of CDC25C [8] . In other cases , phosphorylations at different sites in the same protein can have opposite effects on protein activity causing activation or inhibition of downstream function [52] . We examined the modification status for the CDC25C example and 35 hetero-functional pairs in which one site has activating function and the other has inhibiting function ( S7 Table ) , but found none of them show negative correlation trend . There are two possible reasons . First , the steric hindrance case cannot be revealed in the data because we observe both sites are phosphorylated under multiple conditions . In the case of CDC25C , we indeed observed moderate evidence of co-occurrence between S214 and S216 ( p-value = 1 . 1E-4 ) . It is because the phosphorylation status detected by the “bottom-up” MS technology reflects the stoichiometry of phosphorylation in a pool of multiple isoforms , even if two sites are mutually exclusive on one isoform , they can still be detected under the same condition because both isoforms exist . This issue can be address by “top-down” or quantitative MS approaches . Second , sites with opposite functional categories cannot also be identified as negative correlation because in many conditions neither site is phosphorylated due to biological ( low protein abundance ) or technical reasons ( low enrichment of digested peptides ) . Indeed , over 80% of phosphosites analyzed in this study are phosphorylated in less than 10 conditions and over 60% are phosphorylated in less than 5 conditions ( S8 Fig ) . After removing conditions when neither sites are phosphorylated , we found that hetero-functional pairs tend to be phosphorylated in only one site , consistent with their opposite functions . The two reasons above can be unified under the same statistical principle that the presumed negative correlation is indeed conditional association . For mutual inhibitory pairs , negative correlation is conditional on being on the same protein isoform; for hetero-functional pairs , negative correlation is conditional on at least one site is phosphorylated . The co-occurrence analysis of binary phosphorylation status can only identify marginal association which is not necessarily the same as conditional association . We also applied co-occurrence analysis to interacting protein pairs . Given lower a priori of functional association for sites between proteins and issues of confounding , we expect to observe higher rate of false positives . Despite of this , the observed global trends support the co-phosphorylation of interacting proteins by the same kinase [35 , 53] , and the role of signal integration in phosphorylation enriched protein complex [15] . We found several between co-occurring pairs with suggestive functional evidence . For example , GRB2 and one of its interaction partner GAB2 contain more than 20 inter-protein co-occurring phosphosite pairs . GRB2 is an adaptor protein involved in signal transduction and cell communication [54 , 55] , and GAB2 is a multi-site docking protein and serve as the gateway into GRB2 activation [56] . The co-occurring phosphosite pairs between these two proteins may play roles in mediating interaction between different signaling pathway and signal integration . As another example , Y705 of STAT3 and Y323/Y352 of SYK show co-occurrence under more than half of biological conditions consistent with the established genetic and biomedical evidence that STAT3 is a substrate of SYK [57] , so the co-occurrence in this case may represent part of signal cascading . In this study , interacting proteins are defined by a set of experimentally validated high-quality PPI . One limitation is that PPI may not include many kinase-substrate relationships [58] . We collected 8 , 666 known kinase-substrate pairs from PhosphoSitePlus and only found 10 overlap with the CCSB PPI , possibility because kinase usually interact with phosphosites in a transient manner . When applying the co-occurrence analysis to kinase-substrate pairs , the identified proportion of co-occurring pairs are similar as the PPI set ( S8 Table ) . The enrichment of co-occurring pairs in phosphorylation enriched complexes and in interaction interfaces are also observed ( S9 Fig ) . For the co-occurrence analysis between interacting proteins , we suggest it be applied to the cases in which there is strong evidence for functional association between phosphorylations , and interpretation should be made by considering the function of interacting proteins . Together our study shows that co-occurring phosphorylation are functionally associated , and suggests the utility of mining co-occurrence of modification status to reveal functional association between PTM sites . With the increasing coverage of other PTM types , the co-occurrence can potentially be integrated with other methods to identify novel functional associations between different PTMs . We also found that phosphosites of the co-occurring pairs are more likely to contain functional annotations and evolutionary conserved , suggesting they are more likely to be functional . While previous studies to prioritization functional phosphorylations focus on individual sites [17 , 29 , 59] , the functional associations identified by the co-occurrence analysis in our study can provide further lines of evidence for this purpose .
We downloaded all experimentally observed human phosphorylation sites from the PhosphoSitePlus database ( http://www . phosphosite . org , last access: 2016–02 ) [28] . The observed modification sites were further stratified into different laboratory ( cell line vs . tissue ) and physiological ( disease vs . non-disease ) conditions , resulting in a total 656 data files . To ensure proteome-wide coverage , we only retained 88 different conditions with at least 1000 modification sites . The final data set used in the analysis contains 55 , 145 sites of 10 , 868 proteins and their binary phosphorylation status under the 88 conditions ( S1 Table ) . To investigate if the observed functional association between phosphorylation sites in the human proteome were conserved in mouse , we also downloaded all mouse phosphorylation sites and processed with the same criteria as human . The final data set includes 67 , 555 sites of 10 , 237 mouse proteins along with their phosphorylation status under 34 different conditions . Orthologous proteins of human and mouse were downloaded from InParanoid database v8 [60] . Only the 1-to-1 orthologs with confidence scores greater than 0 . 9 were kept . To map human phosphorylation sites to the orthologous positions in mouse , human and mouse protein sequences from the UniProt database were aligned by MUSCLE 3 . 8 . 31 [61] . We mapped phosphosites to the disordered protein regions predicted by DisEMBL [62] . The 3D structural positions were obtained from PDB database [32] . And structural distance between phosphosites was defined as the distance between the two α-carbon atoms adjacent to the carboxyl group of amino acids . For each pair of phosphorylation sites , we cross-tabulated the times that two residues are phosphorylated under different conditions into a 2-by-2 contingency table . The p-value of one-sided FET was used to evaluate the tendency of phosphorylation to co-occur under the same conditions . The procedure applied to phosphosites within proteins or between pre-specified protein pairs . In the main text , we chose the threshold of 1E-5 to define co-occurring pairs . Residue sites that are phosphorylated in less than 3 conditions were removed prior to the calculation because pairs with one rarely phosphorylated site cannot not reach the desired significance level . To explore how many co-occurring phosphorylation pairs can be identified in the randomized data sets , we performed a series of permutation tests in the following way . For each protein , we first identified its potential phosphosites which were modified at least once across conditions . Then we randomly introduced the same amount of phosphorylation status as observed at each condition among those potential sites . The process was repeated 100 times; and one-sided FET was performed as above to identify co-occurring pairs within proteins or between protein pairs . The above procedure generated randomized data sets by fixing the total number of modification sites within protein , we also considered controlling the number modification sites within a short peptide segment . To do this , we identified all non-overlapping windows of 10 , 20 , 50 , 70 or 100 amino acids long which contain no less than 5 potential phosphorylation sites , treated them as individual proteins and then applied the same permutation procedure as above to generate randomized data . We downloaded multiple sequence alignment ( MSA ) and the species tree of vertebrates non-supervised orthologous groups ( veNOG ) from the eggNOG database ( v4 . 5 ) [33] . Human phosphorylation sites were mapped to the orthologous positions across vertebrates . The normalized mutual information ( nMI ) [63–65] was used to measure the co-evolution of residues at two modification sites: nMI ( X;Y ) =MI ( X;Y ) ∑x∈AXp ( x ) log ( p ( x ) ) ∑y∈AYp ( y ) log ( p ( y ) ) ( 1 ) MI ( X;Y ) =∑y∈AY∑x∈AXp ( x , y ) log ( p ( x , y ) p ( x ) p ( y ) ) ( 2 ) where x and y represent amino acids or alignment gaps at the orthologous positions of human phosphosites X and Y across species; p ( x ) and p ( y ) are the marginal frequencies , and p ( x , y ) is the joint frequency of x and y across MSA . Only sites with at least three non-conserved residues across species were included in the nMI calculation . Residue conservation score ( RCS ) was used to measure the conservation of phosphorylation sites , and was calculated using the method of [20] . Briefly , for each site we first determine the maximum branch length ( MBL ) in the species based on residues the species that have the same amino acid as human . MBL is calculated as the ratio relative to the two most distant species . Then we built a sub-tree containing the most common ancestor of species with the same amino acid as human , and calculated the ratio of conserved residues ( RCR ) among species in the sub-tree . Finally , RCS was obtained by the product of MBL and RCR . Given both MBL and RCR are defined as ratios , RCS will take values from 0 to 1 . We retrieved experts curated annotations about biological process , molecular function , and catalyzing kinases for phosphosites from the PhosphoSitePlus database . For sites with annotations , they were classified into three board functional categories ( activate/ inhibit/ dual ) . We defined homo-functional pair as two phosphosites that belong to the same category and share at least one annotation term , and hetero-functional pair as those that belong to different categories and do not share any annotation term . To measure the sharing of functional annotations between two phosphorylation site , we devised a score that account for the information content ( specificity ) of annotation terms which is defined as: 1NA×NB∑Ai∈A , Bj∈B[I{Ai==Bj}×1F ( Ai ) ] ( 3 ) where A and B denote the sets of annotation terms for two modification sites , NA and NB are the cardinality of the sets , and Ai and Bj are individual terms . F ( Ai ) is the frequency of the term Ai that appear in the entire database; I{Ai == Bj} is an indicator function taking value of 1 only if Ai and Bj are the same . Effectively , this formula accounts for the number of annotations of each site and puts higher weights to the sharing of specific ( low frequency ) terms than the general ( high frequency ) terms . We used position weight matrix ( PWM ) to represent kinase-specific substrate motifs defined as the 15 amino acids sequence context centering on the modification site: M=[MA , 1 , MA , 2 , … , MA , 14 , MA , 15MC , 1 , MC , 2 , … , MC , 14 , MC , 15MD , 1 , MD , 2 , … , MD , 14 , MD , 15…MY , 1 , MY , 2 , … , MY , 14 , MY , 15] ( 4 ) Mk , j=log2 ( qk , jbk , j ) ( 5 ) where k ∈ {A , C , D , E , F , G , H , I , K , L , M , N , P , Q , R , S , T , V , W , Y} i , j ∈ {1 , 2 , … , 15} . qk , j and bk , j denotes the frequency of amino acid k at motif position j in the foreground and background set respectively . To derived the foreground set , we downloaded all known kinases-substrate relationships from the PhosphoSitePlus database , and selected 39 kinases with at least 50 different site-specific substrates . Then for each kinase , the sequences of 15 residues centering on the substrate phosphorylated sites were extracted and positional frequencies of amino acids were calculated . For the background set , we retrieved 15 amino acid context centering on S , T , or Y sites of all human proteins . To predict whether one phosphosite was catalyzed by a kinase , we made use of this kinase’s PWM and score the phosphosite given its sequence context: Score=∑j=115Msj , j where sj is the amino acid at motif position j for this phosphosite . For each kinase , we first scored all its known substrates , and took the median score as the cutoff for prediction . Then for phosphorylation sites with unknown kinase , if its score by the kinase’s PWM exceeded the cutoff , the site was predicted to be catalyzed by this kinase . Note , in this way one phosphosite can be predicted to be catalyzed by multiple kinases .
|
In addition to gene expression and translation control , post-translational modifications ( PTMs ) represent another level to regulate proteins functions . Different PTM sites within a protein usually co-operate to fulfill their functional roles . Recent advances in high-throughput mass spectrometry ( MS ) technologies have facilitated the proteome-wide identification of PTM sites , giving rise to both challenge and opportunity to understand their functional relationships . Previously , several data mining approaches have been developed to explore the global PTM interplays . In this study , we proposed to infer functional associations between PTM sites from the correlation of their modification status across many biological conditions , which was not exploited before . In practice , we tested if a pair of sites are modified together under the same condition significantly more often than expected ( co-occurrence ) . As a proof of principle , we applied this analytical strategy to human phosphorylation because we could collect data sets of proteome-wide coverage under 88 different conditions . We demonstrated that sites with co-occurring phosphorylation status are functionally associated from several lines of evidence . The co-occurrence analysis can also uncover functionally connected phosphosites with clear biological evidence which are missed by other approaches . With increasing proteome-wide data for other types of PTMs under different conditions , the co-occurrence analysis can be integrated with other methods to identify novel PTM associations .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"phosphorylation",
"mathematics",
"signaling",
"cascades",
"protein",
"structure",
"sequence",
"motif",
"analysis",
"protein",
"structure",
"databases",
"discrete",
"mathematics",
"combinatorics",
"research",
"and",
"analysis",
"methods",
"protein",
"kinase",
"signaling",
"cascade",
"sequence",
"analysis",
"bioinformatics",
"proteins",
"biological",
"databases",
"molecular",
"biology",
"biochemistry",
"signal",
"transduction",
"sequence",
"databases",
"permutation",
"peptides",
"post-translational",
"modification",
"proteomes",
"cell",
"biology",
"database",
"and",
"informatics",
"methods",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"cell",
"signaling",
"macromolecular",
"structure",
"analysis"
] |
2017
|
Co-occurring protein phosphorylation are functionally associated
|
The analysis of introgression of genomic regions between divergent populations provides an excellent opportunity to determine the genetic basis of reproductive isolation during the early stages of speciation . However , hybridization and subsequent gene flow must be relatively common in order to localize individual loci that resist introgression . In this study , we used next-generation sequencing to study genome-wide patterns of genetic differentiation between two hybridizing subspecies of rabbits ( Oryctolagus cuniculus algirus and O . c . cuniculus ) that are known to undergo high rates of gene exchange . Our primary objective was to identify specific genes or genomic regions that have resisted introgression and are likely to confer reproductive barriers in natural conditions . On the basis of 326 , 000 polymorphisms , we found low to moderate overall levels of differentiation between subspecies , and fewer than 200 genomic regions dispersed throughout the genome showing high differentiation consistent with a signature of reduced gene flow . Most differentiated regions were smaller than 200 Kb and contained very few genes . Remarkably , 30 regions were each found to contain a single gene , facilitating the identification of candidate genes underlying reproductive isolation . This gene-level resolution yielded several insights into the genetic basis and architecture of reproductive isolation in rabbits . Regions of high differentiation were enriched on the X-chromosome and near centromeres . Genes lying within differentiated regions were often associated with transcription and epigenetic activities , including chromatin organization , regulation of transcription , and DNA binding . Overall , our results from a naturally hybridizing system share important commonalities with hybrid incompatibility genes identified using laboratory crosses in mice and flies , highlighting general mechanisms underlying the maintenance of reproductive barriers .
Most research into the genetic underpinnings of speciation is based on laboratory studies of specific reproductively isolating phenotypes [1] , [2] . A small number of these studies have identified individual genes that confer lower fitness when in a foreign or admixed genetic background [3]–[6] . However , it is not clear whether these genes and corresponding phenotypes - often identified in crosses between highly divergent species pairs - were ever relevant to restricting gene flow in nature , or whether they simply represent an inevitable byproduct of functional divergence between species [1] , [7] . Moreover , it is not clear that hybrid incompatibilities accumulated long after the cessation of gene flow are strongly representative of the early stages of speciation . The direct analysis of naturally hybridizing populations provides an alternative and arguably more direct way to study the genetic basis of reproductive isolation [8] . This approach assures relevance to the early stages of speciation and allows the fitness of hybrid genotypes to be evaluated under natural conditions [9] , [10] . Natural hybridization has been extensively studied in the last few years revealing the mosaic nature of gene flow across species boundaries [11] . It is now well established that genomic regions harboring hybrid incompatibilities are expected to display low levels of introgression , whereas regions not harboring incompatibilities should introgress more freely [12]–[14] . In spite of these general expectations , the identification of individual genes causing reproductive isolation in nature has remained elusive for at least three reasons . First , many hybrid zone studies have lacked mapping resolution due to limited sampling of the genome . Second , numerous historical and biological factors can inherently limit the fine-scale resolution of individual incompatibilities within many hybrid zones . For example , time since secondary contact , rates of dispersal and hybridization , and genomic architecture will all influence the effective rate of recombination and thus mapping resolution within a hybrid zone . Third , in many species , widespread differentiation across much of the genome limits the ability to detect loci contributing to isolation . In principle , some of these limitations can be overcome by applying high-throughput sequencing to the study of introgression between partially-isolated populations that are characterized by a long history of extensive gene flow . Genome-wide data from such populations promises to provide insight into the genetic basis of reproductive isolation in nature [15]–[19] and thereby provide a powerful counterpart to laboratory studies of speciation . Two naturally hybridizing subspecies of the European rabbit ( Oryctolagus cuniculus algirus and O . c . cuniculus ) are a good model for the early stages of reproductive isolation . These two subspecies currently occur in parapatry , hybridize over a large area in the central part of the Iberian Peninsula ( Figure S1 ) , and have diverged approximately 1 . 8 million years ago [20] , [21] . The current secondary contact is thought to result from a range expansion associated with the amelioration of climatic conditions after the last glacial maximum ( ∼18000 ya ) [20] , although previous glacial/interglacial cycles during the Pleistocene may have provided multiple opportunities for vicariant and past hybridization events . Both O . c . algirus and O . c . cuniculus span the three major climatic zones in the Iberian Peninsula ( Mediterranean , Oceanic , and semi-arid ) with little evidence for morphological differentiation along these ecological transitions [22] , . The two subspecies can hybridize in the lab and produce viable hybrids , but hybrid males suffer from reduced fertility ( Ferrand , unpublished results ) . Genetic studies using samples along the distribution range of both subspecies and based on a few dozen nuclear markers have revealed loci of relatively high divergence ( 0 . 3–1 . 2% ) between subspecies embedded in a genome otherwise characterized by low levels of differentiation and high levels of bidirectional gene flow [21] , [24]–[26] , likely facilitated by high effective population sizes , high dispersal , and a relatively short generation time [21] , [24]–[26] . Differentiation appears most pronounced on the sex chromosomes and in centromeric regions , albeit based on limited and biased genomic sampling . A more recent study using individuals sampled across the hybrid zone and 22 markers showed that some loci are characterized by concordant and stepped clines that do not co-localize with any obvious physical or ecological barriers [27] , suggesting an overall barrier to gene flow likely to be primarily maintained by selection against intrinsically unfit hybrids . This same study revealed large inter-locus variation in cline width and cline center among some loci , consistent with substantial variation in patterns and levels of introgression as indicated by previous work [21] , [24]–[26] . Long tails of introgression were often detected such that populations from the extreme ends of the Iberian Peninsula , which are well removed from the primary hybrid zone , were not genetically pure [27] ( Figure S2 ) . Thus , introgression seems not only to affect a large portion of the genome but frequently occurs through the entire range of parental populations . These previous studies provided important information regarding the dynamics of contact and hybridization . However , they were not designed to delineate regions of the genome resisting introgression nor investigate the frequency and types of genes involved in genetic incompatibilities , because they were based on a relatively small number of loci ( less than 50 ) and the sampling of the genome was non-random ( e . g . overrepresentation of the X-chromosome or regions near centromeres ) . The combination of extensive introgression between rabbit subspecies together with the availability of a rabbit reference genome sequence provides ideal conditions to resolve the genetic basis and architecture of persistent reproductive barriers between hybridizing taxa in the early stages of divergence . In this study , we used Illumina sequencing to provide a high-resolution map of genetic differentiation between rabbit subspecies . Our main goal was to identify specific genes or genomic regions that have resisted introgression and are likely to confer reproductive barriers in natural conditions . We further investigated the proportion of the genome that is resisting introgression and the number and size of such regions . We also tested whether previously hypothesized positional effects on genetic differentiation between rabbit subspecies ( i . e . centromeres and X-chromosome ) are robust to a more dense and unbiased sampling of the genome . We used custom microarrays to capture and Illumina sequence 6 Mb of targeted intronic DNA , and combined this dataset with published protein coding sequences derived from transcriptome sequencing [28]–[29] . The combination of genome-wide data and population genetic inference allowed the fine-scale identification of genes likely to underlie reproductive barriers in nature .
We analyzed a large dataset consisting of ∼326 , 000 SNPs ( ∼177 , 000 from the targeted DNA capture of intronic sequence and ∼149 , 000 from the transcriptome resequencing ) distributed on all rabbit chromosomes , providing an unbiased picture of genetic differentiation between rabbit subspecies at the genome-wide level . This dataset was obtained in rabbits sampled at some distance from the primary contact zone ( Figure S1 ) , but in geographic regions that nonetheless show high levels of introgression [21] , [24]–[27] ( see Figure S2 ) . We found that overall levels of genetic differentiation between the rabbit subspecies were low to moderate ( Figure 1 ) . Fixed differences in our dataset accounted for 0 . 3% of the total number of mutations , and shared polymorphisms were ∼100 fold more numerous ( 31 . 2% ) . The percentage of fixed differences in nature may be even lower given the relatively small number of individuals used in our study . We estimated an average fixation index ( FST ) value of 0 . 084 across all polymorphisms and of 0 . 169 across polymorphisms with a minor allele frequency ( MAF ) threshold of 10% ( 132 , 199 SNPs ) . Thus , the overall baseline of differentiation is moderate even after excluding low frequency variants , which are more likely than high-frequency variants to represent sequencing and genotyping errors [30] . The mean levels of differentiation reported here ( FST autosomes = 0 . 082; FST X-chromosome = 0 . 185 ) were lower than previously reported values using Sanger sequencing of small intronic fragments ( FST autosomes = 0 . 146; FST X-chromosome = 0 . 447 ) [26] and remained lower for the X-chromosome after the removal of low frequency variants ( FST autosomes MAF>10% = 0 . 164; FST X-chromosome MAF>10% = 0 . 368 ) . This difference likely reflects the biased sampling of the genome in previous studies towards regions of higher differentiation ( e . g . centromeric regions ) . For comparison , our genome-wide estimates in rabbits are similar to that between human population groups , namely Africans , Caucasians , and Asians ( FST≈0 . 120 ) [31] . There are very few examples in the literature describing such low levels of genetic differentiation in the context of hybridizing taxa showing partial reproductive barriers . Anopheles mosquitoes provide perhaps one of the most extreme examples . Recent results describe numerous regions of differentiation between forms but large portions of the genome remain lowly differentiated [16] , [32] , [33] . However , they are thought to have diverged very recently - within the past 10 , 000 years - associated with the development of agriculture . In contrast , genetic divergence at multiple loci suggests that rabbits started diverging approximately 1 . 8 million years ago [20] , [21] . A more comparable example is found in sunflowers . A recent study found that genetic differentiation was much lower throughout a large fraction of the genome between the now sympatric species pair Helianthus annuus and H . petiolaris , which have diverged for ∼1 . 0 million years , than between more closely related species with non-overlapping geographical ranges [34] , [35] . The shared variation between rabbit subspecies is likely due to both gene flow and retention of ancestral variation . Evidence for high levels of gene flow comes from several sources . First , previous estimates of gene flow between rabbit subspecies using an Isolation-with-Migration model [36] , [37] and a similar sampling scheme to the one used here suggest moderate to high levels of gene exchange ( 2Nm≈1 . 2 for the autosomes averaged in both directions ) [26] . These estimates are at the higher end of the range of values reported in a recent review of similar studies of hybridizing taxa [11] . Second , the spatial distribution of allelic frequencies across the hybrid zone revealed a subset of coincident , narrow and stepped clines consistent with reproductive barriers acting to prevent introgression in a fraction of the genome , but also that cline width varied by a factor of 50 with introgressed alleles often reaching the distribution ends of both subspecies [27] ( Figure S2 ) . Evidence for unsorted ancestral variation comes from a consideration of the divergence time and the effective population size ( Ne ) . Effective population sizes ( Ne ) in rabbits are on the order of 106 [21] . Since the average coalescence time for alleles within a population is 4Ne generations for the autosomes and 3Ne for the X-chromosome , ancestral variation is still expected to be segregating between subspecies , even in the absence of gene flow ( Ne≈106 , divergence time≈1 . 8 Mya , and a generation time of one year ) [21] , [38] . It is exceptionally difficult to resolve the relative contribution of these two forces in structuring genetic variation between the subspecies . Nonetheless , given high levels of gene flow ( 2Nm>1 ) and persistence of ancestral variation , fixed differences between the subspecies are expected to be very rare across the genome . To quantify this expectation given the number of individuals sampled , we conducted simulations to estimate the expected number of fixed differences between the subspecies assuming previous estimates of the degree of gene flow [21] , . We found that both the autosomes and X-chromosome were significantly enriched for fixed differences relative to the simulated neutral expectation ( P<0 . 0001; Figure S3 ) . Thus , genome-wide patterns of variation are characterized by extremely low levels of differentiation punctuated by an unusually high number of fixed differences . Given this pattern , genomic regions that are enriched for fixed differences are likely to be maintained by selection against introgression and thus contain genes involved in hybrid incompatibilities ( or loci experiencing selection at linked sites within subspecies; see below ) . To investigate patterns of differentiation across the genome we first used a 100 kb sliding window approach ( Figure 2; values are detailed in Dataset S1 ) . Levels of differentiation varied greatly across the genome . Similar to our results for individual SNPs , the great majority of windows showed low to moderate differentiation ( median FST MAF>10% = 0 . 139 ) , and extended portions of the genome were completely devoid of fixed differences . Several windows , however , were strongly differentiated relative to the genome-wide average . For the following analyses , we used SNPs with MAF>10% , although similar results are obtained when all SNPs are used ( see Materials and Methods ) . For each window , we calculated the Z-score or the number of standard deviations departing from the median FST over all windows , the proportion of fixed differences versus shared polymorphisms , and also the absolute number of fixed differences . 3 . 2% of windows showed a Z-score greater than three , which corresponded to an FST value of 0 . 590 or higher . Intervals defined by a high Z-score also tended to show an enrichment of the absolute number of fixed differences and the proportion of fixed differences relative to shared polymorphisms ( Figure 2 ) . We found a 100% overlap between windows falling in the top 1% of the distribution of the absolute number of fixed differences and windows showing a significantly elevated proportion of fixed differences to shared polymorphisms when compared to the genome-wide average . We then used a randomized permutation test across SNPs within the genome ( 1000 replicates ) to test if the clustering in levels of differentiation was unusual , given the genome-wide distribution of FST MAF>10% values for individual SNPs and fixed differences . None of the random permutations exceeded the observed differentiation ( P<0 . 001 ) , indicating that these regions of differentiation are unlikely to reflect simple stochastic variation in differentiation across the genome due to the reduced number of individuals sampled . The spatial distribution of differentiation across the genome is potentially informative about the genetic architecture of reproductive isolation . To further demarcate contiguous genomic regions of high differentiation ( i . e . , genomic islands of differentiation ) , we identified all single or consecutive windows with significantly elevated ratios of fixed differences to shared polymorphisms . Using these criteria , we identified 140 independent regions of differentiation that together encompass ∼46 . 8 Mb of genomic sequence ( ∼1 . 8% of the genome; Table S1 ) . Alternatively , we identified 173 regions ( 60 . 6 Mb and ∼2 . 3% of the genome; Table S2 ) when we used a Z-score of three or higher to define highly differentiated regions . Importantly , we found a high degree of overlap between these two metrics of differentiation ( 102 overlapping regions ) ( Table S3 ) and most differentiated regions were well demarcated ( i . e . appearing as single peaks surrounded by background levels of much lower differentiation; Figure 2 ) . Our approach to identifying islands of divergence relies , in part , in levels of variation within populations and therefore is potentially sensitive to any process that reduces diversity within either subspecies , and thus high divergence may not be directly associated with signatures of reduced gene flow due to incompatibilities . For example , regions of high differentiation may be identified in our study due to selection at linked sites acting independently in each subspecies reducing within population variation relative to divergence between populations [39] , [40] . We used two separate approaches to address this issue . First , we used a measure of differentiation , relative node depth ( RND ) [41] , which does not depend on a within-population component of variation and therefore should be useful for distinguishing between both scenarios [42] , [43] . RND takes into account possible mutation rate differences among genomic regions by incorporating divergence to a third more distantly related species , and is expected to be inversely proportional to the amount of gene flow . To generate appropriate outgroup data , we used a custom microarray to capture and sequence the same 5 , 000 introns used for rabbits in a single Hare ( Lepus timidus ) and then calculated RND for each region . We found much higher variance in RND values for intronic fragments consisting of a reduced number of sites , likely due to stochastic variation in the mutation process . To reduce sampling noise , we restricted our analyses to the 1 , 665 fragments with more than 1 . 0 kb of aligned Lepus data . Consistent with variable levels of introgression playing a substantial role in determining patterns of genomic differentiation , we found a significant increase in mean RND values for fragments residing within regions of differentiation defined both using the Z-score or the proportion of fixed differences versus shared polymorphisms ( P<0 . 05 for all comparisons; Figure 3 ) . Second , we tested for evidence of more frequent selection within islands of differentiation . If regions of high differentiation reflect the independent action of positive selection within either subspecies ( i . e . , subspecific selective sweeps ) , then we would expect an excess of high frequency derived alleles within regions of differentiation [44] . We did not find significant differences or deviations in a consistent direction in the frequency of derived alleles in islands of divergence within O . c . cuniculus or in either subspecies when defining differentiation based on the proportion of fixed differences versus shared polymorphisms ( Figure 3 ) ; however , intervals of high FST in O . c . algirus did show a significant skew towards high frequency derived alleles . Our results do not rule out that a simple hitchhiking model or a combination of both factors ( i . e . hitchhiking plus reduced introgression ) may explain some of the regions of differentiation , particular within O . c . algirus . Indeed , several admixture scenarios [44] , [45] and selection at linked sites are expected to generate an excess of high frequency derived alleles , but the observation that most islands of differentiation are not enriched for high frequency derived alleles suggests that selection at linked sites is likely to have a relatively small contribution overall to the occurrence of areas of high differentiation . Moreover , the fact that the inferred skew towards high frequency derived alleles was only detected in one subspecies is mostly consistent with asymmetric migration . Taken together , our results indicate that both FST ( i . e . Z-score ) and the ratio of shared polymorphisms to fixed differences are capturing signatures of reduced gene flow and that regions of high differentiation indentified in our study are likely to be enriched for regions involved in reproductive barriers . Given that combining different summary statistics is likely to reduce the number of false positives , for the remainder of the analysis we defined highly differentiated regions as the 102 genomic intervals that show significant differentiation for both metrics . Our results further show that signatures of reduced introgression are spread across the genome , suggesting that reproductive barriers in rabbits appear to have a complex genetic basis controlled by many unlinked genes . Genome-wide patterns of isolation do not provide direct information regarding the mechanisms or phenotypes that contribute to reproductive barriers , underscoring both the power of this population genetic approach because it is not constrained by choice of a specific phenotype , but also its limitations . Information regarding hybrid phenotypes in rabbits is scarce because wild animals are exceptionally difficult to maintain in captivity . Nonetheless preliminary data from crosses between subspecies indicate that hybrid males are less fertile than hybrid females or parental individuals . Hybrid male sterility often evolves very quickly and may show a complex genetic basis even at the earliest stages of divergence [e . g . 46] , which could account for reduced introgression at multiple regions of the genome as inferred here . Yet , we cannot discard that other hybrid phenotypes , both resulting from intrinsic or extrinsic causes , also act as reproductive barriers between rabbit subspecies in nature . The physical size of differentiated regions was highly heterogeneous ( Figure 4 ) . The largest region of differentiation spanned more than ∼1 . 3 Mb but the majority of differentiation occurred on a physical scale smaller than 200 kb . The large number of small and independent regions of elevated differentiation in a background of low differentiation indicates that the time since secondary contact and subsequent hybridization has been sufficient for recombination to decouple most of the genome from adjacent isolation factors . Since the average gene density in a mammalian genome is ∼1 gene per 100 Kb , this scale of differentiation approaches gene-level resolution ( see below ) . Differentiated regions and their size were not randomly distributed across the genome . First , the X-chromosome exhibited particularly strong genetic differentiation ( Figure 2 ) . Mean FST values were significantly higher on the X-chromosome ( FST MAF>10% = 0 . 368 ) when compared to the autosomes ( FST MAF>10% = 0 . 164; Mann-Whitney U test , P<0 . 001 ) , as was the ratio of fixed to shared polymorphisms ( 0 . 283 vs 0 . 007; Fisher's Exact Test , P<0 . 001 ) . This elevated differentiation was reflected in a significantly larger size of differentiated regions on the X than that on the autosomes ( MeanAutosomes = 202 Kb; MeanX-chromosome = 379 Kb; Mann-Whitney U test P<0 . 001; Figure 4 ) . Moreover , while the X-chromosome represents ∼5% of the rabbit genome , it contained a significant enrichment of differentiated regions ( 27 . 5% , P<0 . 001 ) . In contrast , the mean RND value was significantly lower for the X-chromosome ( RND = 0 . 176 ) when compared to the autosomes ( RND = 0 . 209; P<0 . 001 ) . A similar pattern between chimpanzees and humans has been interpreted as evidence for complex speciation scenarios involving differential gene flow [47] , but several authors have pointed out that several other neutral and simpler explanations were not considered [e . g . 48] . RND values are corrected for mutation rate differences among loci using divergence to an outgroup; however , the X-chromosomes spends only one third of its evolutionary history in males . Thus , variation in the strength of male-biased mutation across time due to changes in reproductive system or generation time in the rabbits versus Lepus comparison could contribute in part to the autosome/X-chromosome difference . Additionally , the ancestral Ne is expected to be smaller for the X chromosome than for the autosomes and this is expected to lead to lower RND for the former than for the latter . Second , we observed strong differentiation close to several centromeric regions , which are known to have reduced rates of recombination in several species [e . g . 49] . Mean FST and RND were significantly higher within 5 Mb of centromeres ( FST MAF>10% = 0 . 246; RND = 0 . 228 ) when compared to the remainder of the genome ( FST MAF>10% = 0 . 164; RND = 0 . 213; Mann-Whitney U test , P<0 . 05 for both tests ) . Two of the most notable such regions are found on chromosomes 4 and X ( Figure 2 ) . Regions of differentiation found within 5 Mb of centromeres were larger when compared to the rest of the genome ( Meancentromeres = 339 kb; Meangenome = 225 kb; Mann-Whitney U test P<0 . 07; Figure 5 ) , and significantly more numerous since they represent ∼9% of the genome and contain 22 . 5% of the total number of differentiated regions ( P<0 . 01 ) . We note that differentiation was not universally elevated near centromeres and many differentiated regions were found along chromosomal arms ( Figure 2 ) , including some larger than 500 Kb ( Figure 5 ) . These genome-wide data confirm and extend previously described positional effects on genetic differentiation in rabbits [21] , [24]–[26] . These findings underscore the central role that the X chromosome plays in the evolution of reproductive isolation [2] , [50] , and support the theoretical prediction that low recombination might facilitate species divergence in the face of gene flow [reviewed in 51] . According to these hypotheses , these regions should be enriched for genes affecting reproductive isolation . However , we cannot exclude the possibility that some of the regions of higher differentiation on the X or in centromeric regions are driven by selection at linked sites or reduced Ne [39] , [40] . The availability of an annotated rabbit genome provides a good opportunity to access candidate genes within differentiated regions . Overall , these regions contained 410 known or predicted genes ( Table S3 ) , and of these genes , 337 were annotated as protein-coding genes , 2 as retrotransposed genes , 57 as noncoding RNAs , and 14 as pseudogenes . Consistent with the variation in size we also found a heterogeneous number of genes contained within each region ( ranging from 1–31 ) . Nine regions ( 8 . 8% ) contained more than 10 genes each . In such regions it will be difficult to identify candidate variants underlying reproductive isolation . However , most differentiated regions contained very few annotated genes: 59 segments ( 57 . 8% ) contained three or fewer genes and 30 ( 29 . 4% ) localized to a single gene . Thus , while many regions throughout the genome appear to contribute to reproductive isolation , the substrate of selection within most regions is likely to be one or a few genes . These findings further underscore the utility of the rabbit hybrid zone for fine-scale mapping of candidate genes likely to be involved in the early stages of reproductive isolation . Due to the reduced number of genes within most islands of isolation , our results allow us to explore the functional underpinnings of putative reproductive barriers at a higher resolution than most comparable genomic studies . We found that genes within islands of differentiation were enriched for several functional classes ( Table S4 ) , and most terms were related to transcription and epigenetic pathways ( e . g . chromatin organization and modification , regulation of transcription and translation , DNA binding ) . These results are consistent with the emerging paradigm that epigenetic and transcription regulation frequently underlies the evolution of postzygotic hybrid dysfunctions [4] , [5] , [52] , [53] . Contrary to several case studies on hybrid incompatibility genes [54]–[57] , we found no statistical evidence for more rapid protein evolution within islands of differentiation ( Table 1 ) . Single-gene islands provide the strongest candidates for hybrid incompatibilities . Given this , we performed additional examination of their molecular functions ( Table 2 ) . We found a wide variety of molecular functions among these candidates , and interactions between distinct elements ( e . g . DNA binding , RNA binding , and protein binding ) , and several transcription activities ( e . g . transcription corepressor activity , transcription factor binding , and histone binding ) , were particularly common . Moreover , several candidate genes were associated with male sterility , which is the only hybrid phenotype detected so far in crosses between rabbit subspecies . For example , EIF4G3 encodes a eukaryotic translation initiation factor , and mutations in this gene have been shown to cause male limited infertility in mice by inducing meiosis and spermatogenesis arrest in meiotic prophase [58] . CDKN2C is a member of the INK4 family of cyclin-dependent kinase inhibitors and its expression is largely confined to spermatocytes undergoing meiosis in seminiferous tubules . Male mice bearing a null CDKN2C mutant suffer from increased germ cell apoptosis , resulting in abnormal spermatogenesis and oligozoospermia [59] . Finally , when we consider our findings in the context of hybrid male sterility genes previously uncovered in laboratory crosses , some intriguing commonalities are apparent . Prdm9 was the first hybrid male sterility gene discovered in vertebrates and was identified from crosses between mouse subspecies [5] . It contains a KRAB motif , a histone H3 Lysine-4-methyltransferase domain , and several zinc fingers that likely mediate sequence-specific binding to DNA . OdsH , a hybrid male sterility gene identified from crosses between Drosophila mauritiana and D . simulans , is thought to encode a heterochromatin-binding protein , and differential DNA binding properties between alleles result in altered heterochromatic localization and chromosome decondensation [4] . Ovd , another hybrid male sterility gene in Drosophila , is also thought to function as a DNA-binding protein [6] . It is noteworthy that several genes in our candidate gene list have nucleic-acid binding properties , and at least five genes contain zinc-fingers and two interact with histones similarly to Prdm9 ( Table 2 ) . Crossing experiments and follow-up functional studies will ultimately be necessary to confirm the involvement of these genes in reproductive isolation and to determine the specific phenotypes which they are controlling . The findings presented here provide a manageable list of candidate genes with explicit phenotypic predictions .
We have used genome-wide data and a population genetic approach to dissect the genetic basis of reproductive isolation in the European rabbit by taking advantage of natural hybridization between subspecies in the Iberian Peninsula . While most of the genome is lowly differentiated between subspecies , we identified numerous regions of strong differentiation , suggesting that the genetic basis of reproductive isolation may be highly polygenic . In addition , the architecture of differentiation is such that these regions were of small size and contained very few genes . This observation and the availability of a reference rabbit genome allowed us to identify several candidate genes for reproductive isolation in this system that suggest an important role for epigenetic and transcription regulation in the maintenance of reproductive barriers . Most interestingly , the molecular functions of some genes identified here using patterns of genetic differentiation in nature strongly parallel those of the hybrid male sterility genes identified using laboratory crosses . Genome-wide analyses of differentiation , such as has been pursued here , are now feasible in a large number of organisms , and are poised to provide new genetic insights into the early stages of speciation .
|
A number of laboratory studies of speciation have uncovered individual genes that confer lower fitness when in a divergent genetic background . It is , however , unclear whether these genes were ever relevant to restricting gene flow in nature , or whether they are indirect consequences of functional divergence between species that in most cases can no longer hybridize in natural conditions . Analysis of introgression across the genome of divergent populations provides an alternative approach to determine the genetic basis of reproductive isolation . Using two subspecies of rabbits as a model for the early stages of speciation , we provide a genome-wide map of genetic differentiation . Our study revealed important aspects of the genetic architecture of differentiation in the early stages of divergence and allowed the identification of genomic regions that resist introgression and are likely to confer reproductive barriers in natural conditions . The gene content of these regions , which in several cases reached gene-level resolution , suggests an important role for epigenetic and transcription regulation in the maintenance of reproductive barriers . Some of the genes identified here in a natural system are similar in function to the hybrid male sterility genes identified in laboratory studies of speciation .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Conclusions"
] |
[
"genome",
"sequencing",
"genome",
"analysis",
"tools",
"gene",
"ontologies",
"genomics",
"genome",
"scans",
"speciation",
"introgression",
"population",
"genetics",
"hybridization",
"biology",
"evolutionary",
"biology",
"gene",
"flow",
"evolutionary",
"processes",
"transcriptomes"
] |
2014
|
The Genomic Architecture of Population Divergence between Subspecies of the European Rabbit
|
Light and microRNAs ( miRNAs ) are key external and internal signals for plant development , respectively . However , the relationship between the light signaling and miRNA biogenesis pathways remains unknown . Here we found that miRNA processer proteins DCL1 and HYL1 interact with a basic helix-loop-helix ( bHLH ) transcription factor , phytochrome-interacting factor 4 ( PIF4 ) , which mediates the destabilization of DCL1 during dark-to-red-light transition . PIF4 acts as a transcription factor for some miRNA genes and is necessary for the proper accumulation of miRNAs . DCL1 , HYL1 , and mature miRNAs play roles in the regulation of plant hypocotyl growth . These results uncovered a previously unknown crosstalk between miRNA biogenesis and red light signaling through the PIF4-dependent regulation of miRNA transcription and processing to affect red-light-directed plant photomorphogenesis .
Light is one of the most important environmental factors to regulate multiple growth and developmental processes of plants , including germination , de-etiolation , phototropism , flowering , leaf and stem growth , circadian clock adjustment , stomatal opening , chloroplast relocation , and anthocyanin synthesis [1 , 2] . Plants utilize at least four distinct families of photoreceptors , including phytochromes , cryptochromes , phototropins , and the ultraviolet B photoreceptor , to perceive light signals [3 , 4] . Phytochromes ( phys ) are primarily responsible for detecting red and far-red light . The Arabidopsis genome encodes five phys , phyA to E [3] . Of these , phyA and phyB have the most prominent functions: phyA is responsible for perceiving far-red ( FRc ) light , and phyB for continuous monochromatic-red ( Rc ) light [5] . Members of the basic helix-loop-helix ( bHLH ) family of transcription factors play a central role in phytochrome-mediated signal transduction . Among bHLH factors , phytochrome-interacting factor 4 ( PIF4 ) acts as a negative regulator in the phyB signaling pathway by selectively binding to the biologically active Pfr form of phyB and regulating a subset of downstream genes [6] . It also acts as a mediator in the auxin-signaling pathway at high temperature , playing a key role in modulating developmental responses to both light and temperature [7] . In addition , PIF4 integrates the brassinosteroid ( BR ) and light signals by interacting with BZR1 and binding to nearly two thousand common target genes , and synergistically regulating many of these target genes [8] . Recently , it has been reported that PIF4 interacts with cryptochrome 1 ( CRY1 ) to regulate high temperature-mediated hypocotyl elongation under blue light [9] . The 20–22 nt-long miRNAs are essential regulators for many biological processes in almost all eukaryotes [10] . MiRNAs are processed from long stem-loop primary transcripts ( pri-miRNAs ) , which are transcribed by DNA-dependent RNA polymeraseII [11] . In animals , the pri-miRNAs are first cropped in the nucleus by the RNAse-III-like endonuclease Drosha and its partner DGCR8 , a double-stranded RNA ( dsRNA ) binding ( dsRBD ) protein , to release the foldback precursor miRNAs ( pre-miRNAs ) . After exportin-5-mediated export to the cytoplasm , the pre-miRNAs are cut into the miRNA/miRNA* duplexes by the Drosha homolog Dicer with the assistance of TAR RNA-binding protein 2 ( TRBP ) [12 , 13] . In plants , however , the two processing steps are completed in the nucleus by a single RNase-III enzyme , DICER-LIKE1 ( DCL1 ) [14 , 15] . Other proteins involved include the dsRBD protein , HYPONASTIC LEAVES1 ( HYL1 ) [16] , and the zinc finger domain protein , serrate ( SE ) [17 , 18] . In addition , several transcription factors including CDF2 , CDC5 , MeCP2 , and NOT2 facilitate miRNA processing [19–22] . Recently , the regulatory mechanism of miRNA processor stability has been partially revealed . In plants , HYL1 is modulated by the light signaling factor COP1 , and destabilized by an unidentified protease [23] . In animals , Dicer is degraded through autophagy [24] . In this study , we show that DCL1 interacts with PIF4 which integrates miRNA biogenesis and red light signaling by regulating the transcription of a group of miRNA genes and the stability of DCL1 during dark-to-red-light or red-light-to-dark transitions . Our results also revealed a previously unknown role for the miRNA processing enzyme DCL1 in red light signaling .
To study the function of DCL1 , we performed yeast two-hybrid screens to identify proteins that interact with the two C-terminal DsRBDs of DCL1 ( DCL1-RBD ) which are important for protein–protein interaction [25] . In addition to the transcription factor CDF2 [21] , we obtained another transcription factor , PIF4 . We then examined the interactions between PIF4 and full-length DCL1 or HYL1 by yeast two-hybrid assays . Our results show that PIF4 can interact with DCL1 and HYL1 , the essential components in miRNA processing ( Fig 1A ) . We confirmed the interaction between DCL1-RBD and PIF4 by MBP pull-down assays . The fusion proteins MBP-DCL1-RBD and GST-PIF4 were expressed in E . coli and purified with amylose resin and glutathione sepharose beads , respectively . We incubated GST-PIF4 with the MBP-DCL1-RBD captured by amylose resin beads ( Fig 1B ) . Parallel assays were performed using GST and MBP proteins as negative controls ( Fig 1B ) . The results indicated that MBP-DCL1-RBD can interact with GST-PIF4 , whereas no interaction was observed in the negative controls ( Fig 1B ) . To verify the interaction between DCL1 and PIF4 in vivo , we performed the bimolecular fluorescence complementation ( BiFC ) assay . We fused DCL1 to the N-terminal fragment of yellow fluorescent protein , YFP ( YFPN ) , and PIF4 to the C-terminal fragment of YFP ( YFPC ) . The fusion pairs were transiently co-transformed into tobacco leaf epidermal cells by Agrobacterium-mediated infiltrations . Strong BiFC signals were detected in nuclear bodies for DCL1 and PIF4 , HYL1 and PIF4 , and DCL1 and the C-terminal fragment of PIF4 ( PIF4-C ) ( Fig 1C ) . In contrast , we did not observe any BiFC signals between the N-terminal fragment of PIF4 ( PIF4-N ) and DCL1 , PIF4 and YFP , or PIF4 and DCL1-9 , a C-terminal DsRBD-truncated form of DCL1 from 1733aa-1911aa [26] . We performed coimmunoprecipitation ( Co-IP ) experiments to further confirm the interaction in vivo . We extracted total proteins from the 4-day-old seedlings of Arabidopsis plants transformed with pPIF4::PIF4-HA and pDCL1::DCL1-YFP [21] ( S1 Fig ) , and we extracted the proteins from plants co-expressing pPIF4::PIF4-HA and p35S::YFP as a negative control . We incubated the extracted proteins with HA-conjugated agarose beads to immunoprecipitate DCL1 . We separated DCL1-containing complexes using SDS-PAGE and immunoblotted them with anti-GFP and anti-HA antibodies . Fig 1D shows the physical interaction between PIF4 and DCL1 . PIF4 has an N-terminal active phytochrome binding ( APB ) motif [27] and a C-terminal bHLH domain [28] . We mapped the region of PIF4 that interacts with DCL1 by yeast two-hybrid assays . We found that the C-terminal region of PIF4 interacts with DCL1 as strongly as the full-length PIF4 ( Fig 1E ) , and this was consistent with the BiFC results ( Fig 1C ) . The transcriptional activation activity of the C-terminal region of PIF4 is shown in S2 Fig . PIF4 is regulated by red light [17 , 18] , and the accumulation level of PIF4 protein decreases during the transition from darkness to red light ( Fig 2A , S3A Fig , S4A Fig ) . We examined whether the accumulation level of DCL1 changes during red-light-to-dark and dark-to-red-light transitions . Four-day-old wild-type ( WT ) seedlings were grown in continuous red light or darkness and then transferred to the opposite light condition for 3 and 5 h . Western blots of the protein extracts from these seedlings revealed that the DCL1 protein level decreases during dark-to-red-light transition ( Fig 2B , S3B Fig , S4B Fig ) , and increases during red-light-to-dark transition at 3 and 5 h ( Fig 2B , S3B Fig , S4B Fig ) . To further address the stability of DCL1 protein in vivo , we generated pDCL1::DCL1-YFP/Col lines and the protein accumulation level of DCL1-YFP was monitored by Western blots using an anti-GFP antibody . We found that red light promotes the destabilization of DCL1 ( S5 Fig ) . We then compared DCL1 protein levels in WT seedlings and pif4-2 mutant seedlings in response to red light for 3 and 5 h , and found that the DCL1 protein level gradually decreases in the pif4-2 mutant during red-light-to-dark transition , but increases during dark-to-red-light transition at 3 and 5 h ( Fig 2B , S3B Fig , S4B Fig ) . These results suggested that PIF4 promotes the destabilization of DCL1 during dark-to-red-light transition at 3 and 5 h . As the C-terminal region of PIF4 interacts with DCL1 and has transcriptional activation activity , we examined the DCL1 levels in two transgenic lines expressing N-terminal ( PIF4-N ) or C-terminal ( PIF4-C ) protein fragment of PIF4 in the pif4-2 mutant background ( S1 Fig ) . The results show that DCL1 protein levels in 35S::PIF4-N/pif4-2 transgenic lines were similar to those in the pif4-2 mutant ( Fig 2B and 2C , S3B and S3C Fig , S4B and S4C Fig ) . In contrast , the DCL1 levels in 35S::PIF4-C/pif4-2 transgenic lines were similar to those in the WT seedlings ( Fig 2B and 2C , S3B and S3C Fig , S4B and S4C Fig ) . These results indicated that the C-terminal fragment of PIF4 can rescue the DCL1 destabilization in the pif4-2 mutant during red-light-to-dark transition . We then examined the accumulation level of HYL1 and tested the effect of PIF4 on HYL1 stability by comparing HYL1 levels in Col and pif4 mutant backgrounds during red-light-to-dark or dark-to-red-light transitions . The results showed that the HYL1 protein level decreases during dark-to-red-light transition and increases during red-light-to-dark transition at 3 and 5 h in the WT seedlings ( S6 Fig ) . In contrast , the HYL1 protein level increases during dark-to-red-light transition and decreases during red-light-to-dark transition at 3 and 5 h in the pif4 mutant ( S6 Fig ) , indicating that PIF4 promotes the destabilization of HYL1 during dark-to-red-light transition . To elucidate whether the expressions of DCL1 and HYL1 were modulated at the mRNA level under red light , we tested the DCL1 and HYL1 transcript levels in WT and pif4-2 seedlings by real time quantitative PCR ( qRT-PCR ) . We found that the DCL1 and HYL1 transcript levels in pif4-2 are similar to those of corresponding genes in WT under red light ( S7A and S7B Fig ) , indicating that the accumulation levels of DCL1 and HYL1 proteins under red light is mainly regulated at the post-transcriptional level . Since the ubiquitin-proteasome ( UPS ) and autophagy are two major proteolytic pathways for protein degradation in plants [29 , 30] , we decided to test whether the instability of DCL1 protein is regulated by these two pathways . First , we applied several proteolysis inhibitors , including MG132 ( a reversible proteasome inhibitor ) , MG115 ( a proteasome-specific inhibitor ) , CLL ( clasto-lactacystin b-lactone , a specific irreversible proteasome inhibitor in UPS ) , proteolysis inhibitors ( PIs ) ( cocktail , a protease inhibitor mixture ) , PMSF ( phenylmethylsulfonyl fluoride , a serine protease inhibitor ) to examine whether proteolysis is responsible for DCL1 stability , using the solvent dimethyl sulfoxide ( DMSO ) as a negative control . We found that only the broad-spectrum PIs stabilize the DCL1 protein , whereas the proteasome-specific inhibitors MG132 , MG115 , and CLL have no stabilizing effect on the protein ( Fig 3A ) . These data indicated that the destabilization of DCL1 is not driven by the UPS pathway . To investigate whether the autophagic pathway regulates the proteolysis of DCL1 , we treated 4-day-old WT seedlings for 18h with several protease inhibitors , including PIs and E64D ( an irreversible cysteine protease inhibitor which can strongly inhibit autophagic degradation ) . As shown in Fig 3B and 3C , the prolonged treatments of these inhibitors at high concentrations could increase the DCL1 protein level . Given that the autophagic pathway is composed of three steps: autophagopore formation; substrate loading into the autophagosome; and transition of autophagosome into vacuoles for substrate degradation , we treated WT seedlings for 18 h with 3-MA , an inhibitor of type III phosphatidylinositol 3-kinases that blocks the formation of autophagosomes , or BTH , a salicylic acid analog that stimulates the formation of autophagosomes , to determine if these autophagic processes are involved in the control of DCL1 levels . We found only a mild change in the DCL1 protein levels under either of the 3-MA and BTH treatments ( Fig 3D and 3E ) , suggesting that further studies are necessary to uncover if and how the DCL1 protein undergoes autophagy during dark to red light transition . As red light modulates DCL1 stability , we applied high-throughput sequencing using WT seedlings grown in dark for 4 days and then exposed to continuous red light for 2 h and 8 h to determine the global effects of red light on miRNA accumulation . The sequencing data for all known miRNAs were subjected to hierarchical clustering in an unsupervised manner [31] to analyze the expression of 2-fold differential miRNAs among plants held under red light for 2 h and 8 h compared with their miRNA levels at 0 h ( S8 Fig ) . The results show that the expression levels of many miRNAs changed under 2 h and 8 h red light treatments . As PIF4 interacts with DCL1 and affects its stability , we analyzed the global miRNAs in 4-day-old WT seedlings and pif4-2 mutant seedlings under continuous red light to determine the effect of PIF4 on miRNA accumulation , using hyl1 mutant seedlings as a control . The data quality is summarized in S1 and S2 Tables . The sequencing data for all known miRNAs were subjected to hierarchical clustering in an unsupervised manner to analyze the expression of 2-fold differential miRNAs [31] ( Fig 4A and S9 Fig ) . The small RNA-seq data chosen from the changed miRNAs were validated by Northern blots of miR319 , miR165 , miR171 , and miR160 ( Fig 4B ) . Among 189 miRNAs detected in both WT and pif4-2 mutant seedlings , we found that 22 miRNAs show at least 1 . 5-fold changes in the pif4-2 mutant compared with those in the WT . Among these 22 miRNAs , 21 ( 95 . 5% ) were down-regulated , and one ( 4 . 5% ) was up-regulated . Taking these observations together , we concluded that the accumulation of a group of miRNAs is affected by PIF4 . In the sequencing data , we noticed that the levels of some miRNAs decrease in the pif4-2 mutant in which the DCL1 accumulation level increases , so we speculated that PIF4 may play a role in miRNA transcription . As PIF4 is a transcription factor in the red light signaling pathway , we analyzed the expression levels of 14 pri-miRNAs , for which their mature miRNAs show differences in the pif4 mutant compared with those in the WT ( S1 Table ) , by qRT-PCR in seedlings of pif4-2 mutant and PIF4 overexpression ( p35S::PIF4-YFP ) lines . As shown in Fig 5A , the relative expression levels of pri-miRNAs between pif4-2 and p35S::PIF4-YFP were predominantly opposite for all 14 pri-miRNAs , although some of pri-miRNAs were up-regulated , whereas others were down-regulated in the pif4-2 mutant line ( Fig 5A ) , indicating that PIF4 might act as a transcription factor for these miRNA genes . To address whether PIF4 binds to the promoters of miRNA genes , we performed chromatin immunoprecipitation-PCR ( ChIP-PCR ) using GFP antibody-precipitated chromatin from pPIF4::PIF4-YFP/Col plants . We focused on the promoters of miRNA genes for which the expression levels of pri-miRNAs were found to have changed in the pif4-2 mutant ( Fig 5A ) . The promoter fragments of miRNA genes were amplified from GFP antibody-immunoprecipitated , but not from HA antibody-immunoprecipitated pPIF4::PIF4-YFP/Col samples . In addition , no apparent enrichment of fragments in the WT seedlings was observed ( Fig 5B and 5C ) . We performed DNA competitive electrophoretic mobility shift assays to verify the direct interaction between PIF4 and a fragment of miR160b promoter containing a G-box DNA-sequence motif ( CACGTG ) , which is similar to the E-box motif ( CANNTG ) known to be the binding site of PIFs [32] . The reactions were performed using decreasing amounts of PIF4 , and the results showed that PIF4 can directly bind to the promoter of miR160b gene ( S10 Fig ) . We concluded that PIF4 is a transcription factor for a group of miRNA genes . To further test the effect of PIF4 on miRNA expression , we used a β-glucuronidase ( GUS ) reporter gene driven by the promoter of miR172a whose expression is repressed by PIF4 ( Fig 5D ) . This system was previously used to determine the function of CDF2 , DDL , CDC5 , and NOT2 in the regulation of miRNA gene transcription [19–21 , 33] . We crossed pif4-2 mutant with transgenic plants containing pmiR172a::GUS . In the second ( F2 ) generation , we obtained PIF4/PIF4 , PIF4/pif4 , and pif4/pif4 genotypes containing pMIR172a::GUS . The expression level of GUS increases in pif4/pif4 compared with that in PIF4+ plants ( Fig 5D ) . Quantitative RT-PCR analysis indicated that the GUS level in the pif4 mutant increases compared with that in WT plants ( Fig 5E ) . We concluded that PIF4 regulates the transcription of miR172a gene . As PIF4 interacts with red light receptor phyB [6] and miRNA processing enzyme DCL1 ( Fig 1 ) , we investigated the role of DCL1 in photomorphogenesis by examining the light inhibition of hypocotyl elongation , the most widely used phenotype to study photomorphogenesis [34] . We compared the hypocotyl growth of 4-day-old WT seedlings and pif4-2 , hyl1-2 , and dcl1-9 mutant seedlings grown under continuous red light at four different intensities ( 0 . 1 , 0 . 5 , 1 , and 10 μmol s−1m−2 ) , or continuous darkness . The hypocotyl lengths of the mutants were indistinguishable with those of the WT grown under dark condition; only slightly shorter hypocotyls were observed in hyl1-2 and dcl1-9 mutants grown under dark condition ( S11A and S11B Fig ) . The relative hypocotyl lengths under red light were normalized to the hypocotyl lengths of dark-grown seedlings [35] , and analyses indicated that dcl1-9 and hyl1-2 had shorter hypocotyl lengths than WT seedlings ( Fig 6A and 6B ) . To diminish the developmental effects of dcl1-9 and hyl1-2 mutants on the analysis of photomorphogenesis based on hypocotyl lengths , we examined the hypocotyl growth of 4-day-old dcl1-9 and hyl1-2 mutants under continuous white light ( 30 μmol s−1 m−2 ) , and we found that the hypocotyl lengths of dcl1-9 and hyl1-2 mutants are similar to those of WT seedlings grown under white light ( S12A and S12B Fig ) . To further examine the function of DCL1 and HYL1 under red light , we generated DCL1 and HYL1 overexpressing lines in WT background under the control of the 35S promoter ( p35S::PIF4-YFP/Col , p35S::DCL1-YFP/Col and p35S::HYL1-YFP/Col ) . High expression levels of these genes were confirmed in transgenic lines ( S13 Fig ) . The seedlings of p35S::PIF4-YFP , p35S::DCL1-YFP , and p35S::HYL1-YFP had longer hypocotyls than those of WT seedlings grown under any red light intensity ( 0 . 1 , 0 . 5 , 1 , and 10μmol s−1 m−2 ) ( Fig 6A and 6B ) . Together , these results indicated that DCL1and HYL1 act as negative regulators for plant photomorphogenesis in the red light signaling pathway . To determine the genetic relationship between PIF4 and miRNA processor DCL1 or HYL1 , we generated a pif4-2/hyl-2 double mutant , as the homozygous dcl1-9 line is sterile [26] . We found the pif4-2/hyl-2 double mutant has a shorter hypocotyl length than those of hyl1-2 and pif4-2 single mutants grown under red light ( Fig 6A and 6B ) , indicating that HYL1 mutation can enhance the photomorphogenesis phenotype of the pif4-2 mutant grown under red light; therefore , HYL1 and PIF4 have a synergistic function in the regulation of hypocotyl elongation . To investigate the genetic interaction of phyB and HYL1 , we generated a phyB/hyl1-2 double mutant . The hypocotyl length of phyB/hyl1-2 mutant is shorter than that of phyB mutant and longer than that of hyl1-2 single mutant ( Fig 6A and 6B ) grown under red light , demonstrating that HYL1 acts genetically downstream of PHYB in the regulation of plant photomorphogenesis . Because the microprocessors DCL1 and HYL1 regulate plant photomorphogenesis ( Fig 6A and 6B ) , we speculated that mature miRNAs might also serve as regulators of hypocotyl growth . We examined the phenotypes of several mutants of miRNAs which are regulated by PIF4 ( Fig 4; S1 and S2 Tables ) and found that miRNA319b and miRNA160b mutants displayed longer hypocotyl phenotypes , while miRNA167b and miRNA848 exhibited shorter hypocotyl phenotypes compared with WT seedlings grown under red light ( Fig 7A and 7B ) . The hypocotyl lengths of these miRNA mutants were similar to those of WT seedlings grown under dark conditions ( S11A and S11B Fig ) . Therefore , we concluded that mature miRNAs , including miRNA319b , miRNA160b , miRNA167b , and miRNA848 , regulate plant photomorphogenesis either positively or negatively .
In this study , we uncovered a crosstalk between the light signaling pathway and the miRNA biogenesis pathway in Arabidopsis . It was known that PIF4 plays a negative role in the phytochrome B signaling pathway under red light , [6] and a positive role in cell elongation [36] . We found that PIF4 integrates red light signaling and miRNA biogenesis through the regulation of miRNA transcription and processing by affecting DCL1 stability during dark/red-light transitions . In addition , we found that the miRNA processing enzyme DCL1 and a group of miRNAs play an important role in plant photomorphogenesis . For the biogenesis of miRNAs , at least three steps are necessary: the transcription of primary miRNAs ( pri-miRNAs ) from miRNA genes; the processing of pri-miRNAs to precursor miRNAs ( pre-miRNAs ) ; and the processing of pre-miRNAs to mature miRNAs . It was known that the interaction of PIF4 with the photo-activated Pfr form of phyB modulates a subset of downstream factors through binding to the promoters of these genes [6] . In this study , we found that the transcription factor PIF4 functions at both the transcriptional and post-transcriptional levels to regulate miRNA biogenesis . At the transcriptional level , PIF4 can bind directly to the promoters of a group of miRNA genes and control their transcription ( Fig 8 ) . In addition , we noticed that the expressions of some miRNAs increase , while others decrease ( Fig 5A ) , implying that PIF4 may serve as either a positive or negative transcription regulator for a miRNA gene . At post-transcriptional level , PIF4 interacts with DCL1 and HYL1 to promote the destabilization of these essential microprocessor proteins and regulate the levels of mature miRNAs during dark-to-red-light transition ( Fig 8 ) . The PIF4-dependent DCL1 degradation during dark-to-red-light transition might also provide a logical explanation for small RNA-seq data which show that the accumulation levels of most miRNAs decrease , whereas only a few miRNAs increase in plants grown under red light conditions ( S6 Fig ) . The light-dependent stabilization of HYL1 is maintained by the light signal negative factor COP1 [23] . Because the miRNA processing activity is impaired when DCL1 or HYL1 is degraded , plants might be able to adjust their levels of precisely processed miRNAs by modulating their DCL1 and HYL1 levels in response to different light conditions . A recent study reported that the levels of miR167 , miR168 , miR171 , and miR398 increase by day and decrease by night , but the oscillatory pattern is not regulated by the circadian clock [37] . Therefore , the biological significance of DCL1 fluctuation can be inferred by the diurnal oscillation of the short-lived miRNAs . We suspected that DCL1 is degraded by a protease or several proteases to adjust miRNA processing , similar to the report that HYL1 is destabilized by COP1 in response to light/dark transition . In animals where miRNA homeostasis including DICER1 and AGO2 is regulated by autophagy [23] . Future studies will focus on revealing if and how autophagy is involved in the regulation of DCL1 or HYL1 stability and identifying the protease ( s ) responsible for plant microprocessor degradation . We identified the miR167b mutant which exhibited a shorter hypocotyl phenotype . One of the mi167b targets is ARF8 , which inhibits hypocotyl elongation under red and blue light [38]; this supports that miR167b plays a role in plant photomorphogenesis . We found that miR167b and miR848 play negative roles , whereas miR319b and miR160b play positive roles in plant photomorphogenesis . Conversely , DCL1 or HYL1 is a negative regulator of photomorphogenesis . We speculated that the overall phenotypes of dcl1 and hyl1 mutants in photomorphogenesis result from the coordinated effects of the diverse functions of miRNAs in the regulation of hypocotyl growth . Multiple environmental and hormonal signals for plant growth regulation in Arabidopsis are integrated by PIF4 [17 , 18] . In this study , we placed the microprocessor DCL1 and HYL1 as a critical node in the red light signaling pathways by binding them to PIF4 , and further studies will reveal whether DCL1 also acts as a node in other signal pathways . Our previous results indicated that blue light signaling factor CDF2 is involved in miRNA biogenesis [23] , and further studies will direct toward testing whether the destabilization of DCL1 is dependent on blue light and the potential crosstalk between blue light signaling and miRNA biogenesis .
Arabidopsis thaliana ( ecotype Col-0 ) , phyB ( SALK_022035C ) , dcl1-9 [25 , 39] , hyl1-2 ( Salk_064863 ) , pif4-2 ( CS66043 ) , miR160b ( SALK_152649 ) , miR319b ( SALK_037093C ) , miR167b ( CS872594 ) and miR848 ( CS812734 ) mutants were used . All plants were grown in soil or Murashige and Skoog ( MS ) medium at 16 hr light/8 hr dark photoperiod unless specifically indicated otherwise . The PIF4 , DCL1 and HYL1 coding sequences were cloned into pCambia1301 with the CaMV35S promoter and a YFP tag [25] and confirmed by sequencing . Primers used to amplify these genes are listed in S3 Table . The upstream regulatory sequence of PIF4 and coding region were cloned into pCambia1301 with a YFP tag to generate pPIF4::PIF4-YFP vector and confirmed by sequencing . Primers are listed in S3 Table . All binary vectors were introduced into Agrobacterium tumefaciens ( strain GV3101 ) by electroporation . Plants were transformed by floral dip [40] , and the transformats were selected on MS medium with hygromycin ( 50 mg/l ) . Lines containing a single T-DNA insertion were selected on the basis of the segregation ratio of the resistant and susceptible plants to hygromycin in the progeny of these primary transformats . Homozygous stocks were selected from these lines and at least 15 T2 independent lines were analyzed for each construct . Yeast transformation and library screening were performed according to the Pro-Quest Two-Hybrid System Manual ( Matchmaker user’s manual , Invitrogen ) . The experiments was performed as described in Sun et al . [21] . The size of F1 fragment of PIF4 was 309 aa ( from aa 123 to aa 431 ) , F2 fragment was 221 aa ( from aa 211 to aa 431 ) , F3 fragment was 177 aa ( from aa 255 to aa 431 ) and F4 fragment was 119 aa ( from aa 256 to aa 374 ) . The cDNAs of PIF4 and DCL1 were amplified and subcloned to the pMAL-c2x and pGEX4T-1 plasmids , and confirmed by sequencing . Primers used to amplify the genes are listed in S3 Table . MBP-DCL1-RBD and MBP were expressed in E . coli BL21 ( DE3 ) and purified according to the manufacturer’s protocol ( New England Biolabs ) . GST-PIF4 and GST were purified using glutathione–agarose 4B ( Peptron ) beads . For protein pull-down assays , GST-PIF4 was incubated with the MBP-DCL1-RBD bound to amylose resin , mixed with total protein extracts in 1ml protein pull-down buffer ( 40 mM HEPES-KOH , pH 7 . 5 , 10 mM KCl , 3 mM MgCl2 , 0 . 4 M sucrose , 1 mM EDTA , 1 mM DTT , and 0 . 2% Triton X-100 ) , and then incubated at 4°C for 1 h with agitation . GST and MBP proteins were used as negative controls in parallel assays . The beads were washed four times with the binding buffer . Proteins were eluted and further analyzed by immunoblotting using appropriate antibodies . The lines of coexpressing hemagglutinin ( HA ) -labeled PIF4 ( pPIF4::PIF4-HA ) and YFP-labeled DCL1 ( pDCL1::DCL1-YFP ) were generated by crossing pDCL1::DCL1-YFP/dcl1-9 [21] with pPIF4::PIF4-HA/pif4-2 obtained by crossing pPIF4::PIF4-HA/Col with pif4-2 mutant . For control lines , plants co-expressing pPIF4::PIF4-HA/pif4-2 and p35S::YFP/Col were generated by transforming the pPIF4::PIF4-HA/pif4-2 line with the p35::YFP construct . The homozygous lines were grown in MS medium for 4 days , then seedlings were ground in liquid nitrogen and homogenized in three volumes of extraction buffer ( 50 mM Tris-HCl at pH 8 . 0 , 150 mM NaCl , 0 . 5% TritonX-100 , 0 . 2% 2-mercaptoethanol , 5% glycerol ) containing complete proteinase inhibitor cocktail ( Roche ) using a mortar and pestle set . Cell debris was pellet by centrifugation for 10 min at 13 , 000 g . The Co-IP experiments were performed using HA Tag IP/Co-IP Kit according to the manufacturer’s protocol ( Thermo Pierce ) and the previous report [21] . The coding sequence of PIF4 [41] , DCL1 , HYL1 , DCL1-9 , the N-terminal fragment of PIF4 ( PIF4-N , from aa 1 to 254 ) , and the C-terminal fragment of PIF4 ( PIF4-C , from aa 255 to 431 ) were subcloned into pCambia1301 Vectors with YFPN or YFPC tag [25] and confirmed by sequencing . Primers used to amplify these genes are listed in S3 Table . Plasmid pairs were expressed in tobacco leaves , the relevant negative control was performed at the same time . 48 hours after co-inoculation , BiFC signals were visualized with a DeltaVision Personal DV system ( Applied Precision ) using an Olympus UPLANAPO water immersion objective lens ( 60×/1 . 20 numerical aperture ) [42] . Degradation assay was performed as previous reported with some modifications [43] , 4-day-old seedlings of wild-type ( 18h light/6-h dark photoperiod ) were ground in liquid nitrogen and resuspended in a buffer ( 25 mM Tris pH 7 . 5 , 10 mM MgCl2 , 5 mM DTT , 10 mM NaCl and 10 mM ATP ) . Cell debris was pellet by centrifugation and equal amounts of extract were transferred to individual tubes , which were incubated at room temperature for 2 h , and reactions were stopped by adding an equal volume of 2×protein gel-loading buffer . Equal amounts of sample were then analyzed by western blots with an anti-DCL1 antibody and an anti-ACTIN antibody . For chemical treatments , 4-day-old seedlings of WT ( 18h light/6-h dark photoperiod ) were treated with protease inhibitor cocktail tablets ( PIs , Roche ) , E64D ( Sigma ) , 3-MA ( Sigma ) or BTH ( Sigma ) at the indicated concentrations . Blots were detected with an anti-DCL1 antibody and an anti-ACTIN antibody . Real-time quantitative PCR assays were performed as previously described [21] . Briefly , the first strand cDNA was synthesized from the total RNA ( 1 μg ) with M-MLV reverse transcriptase ( Promega ) and used as the template for subsequent PCR amplification . The real-time quantitative PCR ( RT-qPCR ) for examination of pri-miRNA expression was carried out with a BIO-RAD CFXTM Real-Time System . The ACTIN gene was used as an internal control for normalization of the cDNA template . Each PCR was repeated at least three times . The data were analyzed with a Bio-Rad iCycler iQ Real-Time Detection System . The expressions of genes were calculated using the relative 2–ΔΔCt method [44] . The ChIP assay was performed as described [45] using 4-d-old seedlings . Transgenic seedlings harboring pPIF4::PIF4-YFP were harvested in cross-linking buffer ( 0 . 4M sucrose , 10mM Tris-HCl ( pH8 . 0 ) , 1mM PMSF , ImM EDTA , 1% formaldehyde ) for 10 min using vacuum infiltration . The cross-linking was stopped in 2M glycine . After chromatin shearing , about 5 μg anti-GFP monoclonal antibody ( Sigma , G6495 ) was added to the samples and incubated at 4°C overnight . Beads were then washed and eluted with the lysis buffer ( 0 . 1M NaHCO3 , 1%SDS ) . DNA was precipitated using ethanol , and resuspended in 50 μl water after reversing cross-linking . The immunoprecipitated genomic DNA were used for PCR and quantified by real-time PCR . ACTIN gene was used as an internal control . Primers used to amplify the promoters of some miRNA genes are listed in S3 Table . The EMSA was performed as described previously [46] with modification . The sense and antisence sequences of miR160 promoter ( S3 Table ) were synthesized and the sense sequence was 3’end-labeled with biotin . The promoter fragment synthesized has a G-box DNA-sequence motif ( CACGTG ) , a variant of the canonical E-box motif ( CANNTG ) which was known to be the binding site of PIFs [32] . The single-strand DNAs were mixed , denatured at 95°C for 5 min , and slowly cooled to room temperature to form the double-strand DNAs . The unlabeled double-strand DNA with the same sequence was used as a competitor . Using Light Shift Chemiluminescent EMSA Kit ( Thermo Pierce ) , EMSA assays were performed in 20 μl reaction buffer according to manufactory’s protocol . The mixtures were incubated on ice for 30 min and then fractioned on polyacrylamide gels in 1×TBE buffer for about 60 min . The gels were transferred to a nylon membrane ( GE Healthcare ) and then the biotin-labeled oligonucletides were detected by Chemiluminescence ( Thermo Pierce ) . GUS staining was performed according to the standard procedure [47] with a modified buffer ( 1mg/ml 5-bromo-4-chloro-3-indolyl-b-D-glucuronic acid cyclohexylammonium salt , 50 mM sodium phosphate , pH 7 . 0 , 0 . 1% Triton X-100 , 2 mM potassium ferrocyanide , 2 mM potassium ferricyanide , and 10 mM EDTA ) . Plant tissues were incubated in the buffer at 37°C in the dark overnight , and then cleared with 75% ethanol , followed by observation [21] . Light treatment of seedlings was performed as described [48] with some modifications . Seedlings of each line were grown on the same 100 mm Petri dish for each repeat . Seeds were sterilized by soaking for 5 min in 70% ethanol , followed by 5 min in 95% ethanol . The seeds were vernalizated for 3 days at 4°C in the dark . Plates were placed into white light for 3 h to induce germination and placed in a growth chamber at 21°C in darkness for 21 h before being transferred to the experimental light conditions . Seedlings were moved to red light ( 0 . 1 , 0 . 5 , 1 and 10 μmol s-1 m-2 ) and white light ( 10 μmol s-1 m-2 ) growth chambers ( Percival Scientific Inc . , Perry , Iowa , USA ) at 21°C for 4 days , respectively . Dark control seedlings were kept in darkness . Hypocotyl lengths of at least 30 seedlings were measured after the light treatments . Small RNAs were isolated from seedlings of 4-day-old plants using mirVana™miRNA Isolation Kit ( Ambion , AM1561 ) . Small RNA about three micrograms was fractionated on a 15% polyacrylamide gel containing 8M urea , and then transferred to a nylon transfer membrane ( GE Healthcare ) . The antisense oligonucletides ( S3 Table ) were synthesized and 3’end-labeled as probes with biotin . Hybridization was performed overnight at 42°C in hybridization buffer ( Ambion , AM8663 ) . A probe complementary to U6 ( 5’CATCCTTGCGCAGGGG CCA 3’ ) was used as a loading control . RT-qPCR assays were performed as previously described [21] . Small RNAs were isolated from seedlings of 4-day-old plants using mirVana™miRNA Isolation Kit ( Ambion , AM1561 ) and sequenced by Illumina Solexa high-throughput sequencing . The sequencing data for all known miRNAs were subjected to hierarchical clustering in an unsupervised manner to analyze the extent of differential miRNA expression [31] .
|
External light and internal miRNAs are important for plant development . This study revealed that the miRNA-processing enzyme DCL1 interacts with the red-light-regulated transcription factor PIF4 , which modulates the stability of DCL1 during dark-to-red-light or red-light-to-dark transitions and acts as a transcription factor for some miRNA genes . This study revealed that DCL1 and mature miRNAs play roles in the red light signaling pathway to regulate plant photomorphogenesis . These results shed light on the crosstalk between miRNA and red light signaling pathways .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"plant",
"anatomy",
"plant",
"growth",
"and",
"development",
"gene",
"regulation",
"plant",
"embryo",
"anatomy",
"regulatory",
"proteins",
"dna-binding",
"proteins",
"plant",
"physiology",
"developmental",
"biology",
"micrornas",
"plant",
"science",
"transcription",
"factors",
"seedlings",
"molecular",
"biology",
"techniques",
"plants",
"research",
"and",
"analysis",
"methods",
"transcriptional",
"control",
"plant",
"embryogenesis",
"plant",
"development",
"proteins",
"artificial",
"gene",
"amplification",
"and",
"extension",
"gene",
"expression",
"molecular",
"biology",
"biochemistry",
"rna",
"eukaryota",
"photomorphogenesis",
"nucleic",
"acids",
"polymerase",
"chain",
"reaction",
"embryogenesis",
"genetics",
"biology",
"and",
"life",
"sciences",
"biosynthesis",
"non-coding",
"rna",
"hypocotyl",
"organisms",
"fruit",
"and",
"seed",
"anatomy"
] |
2018
|
Coordinated regulation of Arabidopsis microRNA biogenesis and red light signaling through Dicer-like 1 and phytochrome-interacting factor 4
|
Cysteine peptidases play a central role in the biology of Leishmania . In this work , we sought to further elucidate the mechanism ( s ) by which the cysteine peptidase CPB contributes to L . mexicana virulence and whether CPB participates in the formation of large communal parasitophorous vacuoles induced by these parasites . We initially examined the impact of L . mexicana infection on the trafficking of VAMP3 and VAMP8 , two endocytic SNARE proteins associated with phagolysosome biogenesis and function . Using a CPB-deficient mutant , we found that both VAMP3 and VAMP8 were down-modulated in a CPB-dependent manner . We also discovered that expression of the virulence-associated GPI-anchored metalloprotease GP63 was inhibited in the absence of CPB . Expression of GP63 in the CPB-deficient mutant was sufficient to down-modulate VAMP3 and VAMP8 . Similarly , episomal expression of GP63 enabled the CPB-deficient mutant to establish infection in macrophages , induce the formation of large communal parasitophorous vacuoles , and cause lesions in mice . These findings implicate CPB in the regulation of GP63 expression and provide evidence that both GP63 and CPB are key virulence factors in L . mexicana .
The protozoan Leishmania parasitizes macrophages and causes a spectrum of human diseases ranging from self-healing cutaneous lesions to a progressive visceral infection that can be fatal if left untreated . Infection is initiated when promastigote forms of the parasite are inoculated into the mammalian host by infected sand flies and are internalized by phagocytes . Inside macrophages , promastigotes differentiate into amastigotes to replicate within phagolysosomal compartments also known as parasitophorous vacuoles ( PVs ) . Upon their internalization , L . donovani and L . major promastigotes arrest phagolysosomal biogenesis and create an intracellular niche favorable to the establishment of infection and to the evasion of the immune system [1 , 2] . Disruption of the macrophage membrane fusion machinery through the action of virulence factors plays an critical role in this PV remodeling . Hence , insertion of the promastigote surface glycolipid lipophosphoglycan ( LPG ) into the PV membrane destabilizes lipid microdomains and causes exclusion of the membrane fusion regulator synaptotagmin V from the PV [2–4] . Similarly , the parasite GPI-anchored metalloprotease GP63 [5 , 6] redistributes within the infected cells and cleaves key Soluble NSF Attachment Protein Receptors ( SNAREs ) and synaptotagmins to impair phagosome functions [1 , 7] . Whereas L . major and L . donovani multiply in tight individual PVs , parasites of the L . mexicana complex ( L . mexicana , L . amazonensis ) replicate within large communal PVs . Relatively little is known about the host and parasite factors involved in the biogenesis and expansion of those communal PVs . Studies with L . amazonensis revealed that phagosomes containing promastigotes fuse extensively with late endosomes/lysosomes within 30 minutes post-infection [8] . At that stage , parasites are located within small individual compartments and by 18 to 24 hours large PVs containing several parasites are observed . The rapid increase in the size of those PVs requires extensive fusion with secondary lysosomes and correlates with the depletion of those organelles in infected cells [9–11] . Homotypic fusion between L . amazonensis-containing PVs also occurs , but its contribution to PV enlargement remains to be further investigated [12] . These studies highlighted the contribution of the host cell membrane fusion machinery in the biogenesis and expansion of large communal PVs and are consistent with a role for endocytic SNAREs in this process [13] . Interestingly , communal PVs interact with the host cell’s endoplasmic reticulum ( ER ) and disruption of the fusion machinery associated with the ER and Golgi inhibits parasite replication and PV enlargement [14–16] . The Leishmania-derived molecules involved in the expansion of the communal PVs remains to be identified . LPG and other phosphoglycans do not play a significant role in L . mexicana promastigote virulence and PV formation [17] , in contrast to L . major and L . donovani [2] . Cysteine peptidases ( CP ) are a large family of papain-like enzymes that play important roles in the biology of Leishmania [18] . Three members of these papain-like proteases are expressed by L . mexicana and the generation of CP-deficient mutants revealed that CPB contributes to the ability to infect macrophages and to induce lesions in BALB/c mice [19–21] . The precise mechanism ( s ) by which CPB participates in the virulence of L . mexicana is poorly understood . Previous studies revealed that CPB traffics within and outside infected macrophages [18] . In infected macrophages , CPB alters signal transduction and gene expression through the activation of the protein tyrosine phosphatase PTP-1B and the cleavage of transcription factors responsible for the expression of genes involved in host defense and immunity [20 , 22] . The observation that CPs interfere with the host immune response through the degradation of MHC class II molecules and invariant chains present in PVs housing L . amazonensis [23] , raises the possibility that CPB participates in the modulation of PV composition and function . In this study , we sought to gain insight into the mechanism by which CPB contributes to L . mexicana virulence , with a focus on the PV . We provide evidence that CPB participates in PV biogenesis and virulence through the regulation of GP63 expression .
Formation and expansion of communal PVs hosting L . mexicana involve fusion between PVs and endocytic organelles , as well as homotypic fusion among PVs [10–12] . To identify the host and parasite factors involved in this process , we embarked on a study to elucidate the fate of endosomal SNAREs during infection of macrophages with L . mexicana . Given the requirement of CPB for L . mexicana to replicate normally inside macrophages [19] , we included a L . mexicana CPB-deficient mutant ( Δcpb ) in our investigation . We infected BMM with either WT or Δcpb L . mexicana promastigotes for 2 h and we assessed the distribution of the endosomal SNAREs VAMP3 and VAMP8 by confocal immunofluorescence microscopy . As previously observed during infection with L . major promastigotes [1] , we found a notable reduction in the staining intensity for both VAMP3 ( Fig 1A ) and VAMP8 ( Fig 1B ) in BMM infected with WT L . mexicana , but this was not observed with Δcpb . This reduction in staining intensity correlated with a down-modulation of VAMP3 and VAMP8 proteins in BMM infected with WT L . mexicana , compared to cells infected with Δcpb ( Fig 1C ) . These results suggested that L . mexicana causes the reduction of VAMP3 and VAMP8 levels in infected BMM through the action of CPB . However , we considered the possibility that CPB acted indirectly on VAMP3 and VAMP8 because we previously found that GP63 targets those SNAREs in L . major-infected BMM [1] . We therefore ensured that similar levels of GP63 were present in lysates of BMM infected with WT and Δcpb L . mexicana promastigotes . As shown in Fig 2 , GP63 was detected in lysates of BMM infected with WT L . mexicana up to 72 h post-infection , when the parasites replicate as amastigotes . The important reduction in GP63 levels at this time point is consistent with previously published data showing a 90% reduction in the amount of GP63 detected in amastigotes with respect to promastigotes [24 , 25] . Surprisingly , we found that GP63 was barely detectable in BMM infected with Δcpb at all time points tested . This observation raised the possibility that the lack of VAMP3 and VAMP8 down-regulation in Δcpb-infected BMM was due to defective expression of GP63 . To address the issue of GP63 down-regulation in L . mexicana Δcpb , we first determined whether complementation of Δcpb with the CPB gene array ( Δcpb+CPB ) restores wild type GP63 levels . As shown in Fig 3A , GP63 levels and activity are down-modulated in the Δcpb mutant , and complementation with the CPB array restored GP63 levels and activity similar to those observed in WT parasites . It was previously reported that expression of the cell surface glycolipid LPG and of GP63 may share common biosynthetic steps [26–29] . We therefore evaluated the levels of LPG in lysates of WT , Δcpb , Δcpb+CPB , and Δcpb+GP63 parasites by Western blot analysis . Strikingly , similar to GP63 , LPG levels were also down-modulated in the Δcpb mutant and complementation with either the CPB array or GP63 restored wild type LPG levels . To further investigate the possible role of CPB in the regulation of GP63 expression , we determined the levels of GP63 mRNA in WT , Δcpb , Δcpb+CPB , and Δcpb+GP63 parasites by RT-PCR . As shown in Fig 3B , GP63 mRNA levels were highly down-regulated in Δcpb and complementation with the CPB array restored wild type levels of GP63 mRNA . Interestingly , complementation of Δcpb with L . major GP63 did not increase endogenous GP63 mRNAs . RT-PCR using L . major GP63-specific primers showed that this gene is expressed only in Δcpb+GP63 . Together , these results suggest that CPB controls GP63 mRNA levels at the post-transcriptional level . Clearly , additional studies will be required to elucidate the underlying mechanism ( s ) . Our results also raised the possibility that down-modulation of GP63 in the Δcpb mutant may have accounted for the inability of Δcpb to down-regulate VAMP3 and VAMP8 . The finding that expression of GP63 in Δcpb restored LPG levels was unexpected and suggested a role for GP63 in the expression of LPG in L . mexicana . As it is estimated that at least 25 genes are required for the synthesis , assembly , and transport of the various components of LPG [30] , it may be difficult to determine whether GP63 acts on the expression of a LPG biosynthetic gene or on a biosynthetic step . Assessment of LPG2 gene expression revealed that it was equally expressed WT , Δcpb , Δcpb+CPB , and Δcpb+GP63 parasites . Further studies will be necessary to understand how GP63 expression restores LPG synthesis in Δcpb . Since LPG does not play a major role in the virulence of L . mexicana [17] , the Δcpb mutant expressing exogenous GP63 provides a unique opportunity to address the impact of GP63 on SNARE cleavage , as well as on the in vitro and in vivo virulence of L . mexicana . We next assessed the impact of GP63 on VAMP3 and VAMP8 during L . mexicana infection . To this end , we infected BMM with either WT , Δcpb , Δcpb+CPB , or Δcpb+GP63 L . mexicana promastigotes for various time points , and we assessed VAMP3 and VAMP8 levels and intracellular distribution . Fig 4A shows that GP63 is present at high levels in lysates of BMM infected for 2 h with WT , Δcpb+CPB , and Δcpb+GP63 promastigotes ( compared to lysates of BMM infected with Δcpb ) . At 72 h post-infection , GP63 levels are strongly reduced in BMM infected with WT and Δcpb+CPB , whereas they remain elevated in BMM infected with the Δcpb+GP63 ( Fig 4A ) [25] . The high levels of GP63 present in BMM infected with Δcpb+GP63 for 72 h may be related to the fact that expression of the L . major GP63 gene from the pLEXNeo episomal vector [31] is not under the control of endogenous GP63 3' untranslated regions . Western blot analyses revealed that down-regulation of VAMP3 and VAMP8 correlated with GP63 levels expressed by the parasites . Consistently , the staining intensity of VAMP3 and VAMP8 was reduced in BMM infected with GP63-expressing parasites , as assessed by confocal immunofluorescence microscopy ( Fig 4D and 4E ) . These results suggest that GP63 is responsible for the down-modulation of the endosomal SNAREs VAMP3 and VAMP8 in L . mexicana-infected BMM . Interestingly , we observed recruitment of VAMP3 to PVs containing L . mexicana parasites at later time points , when promastigotes have differentiated into amastigotes , with the exception of Δcpb+GP63 L . mexicana promastigotes ( Fig 4B ) . In contrast , we found that VAMP8 is excluded from L . mexicana-containing PVs both at early and later time points post-infection , independently of GP63 levels , suggesting that additional mechanisms regulate VAMP8 recruitment to L . mexicana PVs . Since GP63 was shown to contribute to L . major virulence [32] , we next sought to determine whether expression of GP63 is sufficient to restore the ability of Δcpb to replicate inside macrophages and to cause lesions in mice [19] . To this end , we first infected BMM with either WT , Δcpb , Δcpb+CPB , or Δcpb+GP63 stationary phase promastigotes and we assessed parasite burden and PV surface area at various time points post-infection . We found that Δcpb had an impaired capacity to replicate inside macrophages and to induce the formation of large communal PVs compared to WT and Δcpb+CPB parasites ( Fig 5A , 5B and 5C ) . Strikingly , expression of GP63 in Δcpb restored its ability to replicate in macrophages and to induce large communal PVs up to 72 h post-infection . These results underline the role of GP63 in the ability of L . mexicana to infect and replicate in macrophages , even in the absence of CPB . Following inoculation inside the mammalian host , promastigotes are exposed to complement and both GP63 and LPG confer resistance to complement-mediated lysis [32 , 33] . L . mexicana promastigotes were therefore analyzed for their sensitivity to complement-mediated lysis in the presence of fresh human serum . As shown in Fig 6A , over 40% of Δcpb was killed after 30 min in the presence of 20% serum , whereas Δcpb+CPB , and Δcpb+GP63 were more resistant to serum lysis at 14% and 10% , respectively . Absence of both GP63 and LPG may be responsible for the serum sensitivity of Δcpb . Finally , to assess the impact of GP63 on the ability of Δcpb to cause lesions , we used a mouse model of cutaneous leishmaniasis . Susceptible BALB/c mice were infected in the hind footpad with either WT , Δcpb , Δcpb+CPB , or Δcpb+GP63 promastigotes and disease progression was monitored for 9 weeks . Consistent with its reduced capacity to replicate inside macrophages , Δcpb failed to cause significant lesions compared to WT parasites [19] and Δcpb complemented with CPB ( Fig 6B ) . Remarkably , expression of GP63 in Δcpb restored its capacity to cause lesions , albeit to a lower level than Δcpb complemented with CPB . Lesion size correlated with parasite burden , as measured at 9 weeks post-infection ( Fig 6C ) . Collectively , these results indicate that expression of GP63 is sufficient to restore virulence of Δcpb .
This study aimed at investigating the mechanism ( s ) by which CBP contributes to L . mexicana virulence . To this end , we initially examined PV biogenesis by assessing the impact of L . mexicana infection on the trafficking of VAMP3 and VAMP8 , two endocytic SNAREs associated with phagosome biogenesis and function [1 , 34] . We found that both SNAREs were down-modulated in a CPB-dependent manner , which hampered VAMP3 recruitment to PVs . We also discovered that expression of GP63 , which we previously showed to be responsible for cleaving SNAREs in L . major-infected macrophages [1] , was down-modulated in the L . mexicana Δcpb . Strikingly , restoration of GP63 expression in Δcpb bypassed the need for CPB for SNARE cleavage . Similarly , episomal expression of GP63 enabled the Δcpb mutant to establish infection in macrophages , induce larger PVs and cause lesions in mice . These findings imply that CPB contributes to L . mexicana virulence in part through the regulation of GP63 expression , and provide evidence that GP63 is a key virulence factor for L . mexicana . The observation that CPB regulates GP63 expression at the mRNA levels was both unexpected and intriguing . Insight into the possible mechanism ( s ) may be deduced from a recent study on the role of cathepsin B in L . donovani , which is homologous to the L . mexicana CPC [35] . Similar to L . mexicana Δcpb , L . donovani ΔcatB displays reduced virulence in macrophages . To investigate the role of cathepsin B in virulence , the authors performed quantitative proteome profiling of WT and ΔcatB parasites and identified 83 proteins whose expression is altered in the absence of cathepsin B , with the majority being down-modulated [35] . Among those were a group of proteins involved in post-transcriptional regulation of gene expression ( RNA stability , processing , translation ) [35] . Whether this is the case in Δcpb deserves further investigation . Clearly , a detailed analysis of wild-type and Δcpb parasites may provide the information required to understand the extent of the impact of CBP on the expression and synthesis of virulence factors and the exact role of CPB in L . mexicana virulence . The observation that episomal expression of GP63 in Δcpb restored LPG synthesis is an intriguing issue , as it suggests that GP63 acts on a LPG biosynthetic step . This role for GP63 is likely redundant , since L . major Δgp63 promastigotes express LPG levels similar to that of wild type parasites ( S1 Fig ) . It has been proposed that expansion of the PVs hosting parasites of the L . mexicana complex leads to the dilution of the microbicidal effectors to which the parasites are exposed , thereby contributing to parasite survival [36] . Both host and parasite factors may be involved in the control of PV enlargement . On the host side , it has been previously reported that L . amazonensis cannot survive in cells overexpressing LYST , a host gene that restricts Leishmania growth by counteracting PV expansion [37] . Similarly , disrupting the fusion between PVs housing L . amazonensis and the endoplasmic reticulum resulted in limited PV expansion and inhibition of parasite replication [15 , 16] . On the parasite side , virulence of L . amazonensis isolates was shown to correlate with the ability to induce larger PVs [38] . Our results indicate that the inability of Δcpb to multiply inside macrophages is associated with smaller PV size , and that expression of GP63 is sufficient to restore the capacity of Δcpb to survive within macrophages and to induce PV expansion . How does GP63 modulate L . mexicana virulence and PV expansion ? In addition to the numerous macrophage proteins known to be targeted by GP63 , it is possible that SNARE cleavage is one of the factors associated with L . mexicana virulence and PV expansion . For instance , we previously reported that VAMP8 is required for phagosomal oxidative activity [1] . One may envision that its degradation by GP63 is part of a strategy used by L . mexicana to establish infection in an environment devoid of oxidants , thereby contributing to parasite survival . The LYST protein is a regulator of lysosome size and its absence leads to further PV expansion and enhanced L . amazonensis replication [37] . It is interesting to note that LYST was proposed to function as an adaptor protein that juxtaposes proteins such as SNAREs that mediate intracellular membrane fusion reactions [39] . In this context , cleavage of SNAREs that interact with LYST may interfere with its function and promote PV expansion . Further studies will be necessary to clarify these issues , including the potential role of VAMP3 and VAMP8 in PV biogenesis and expansion . Previous studies using Δcpb parasites led to the conclusion that CPB enables L . mexicana to alter host cell signaling and gene expression through the cleavage of various host proteins [20 , 22] . Hence , CPB-dependent cleavage of PTP-1B , NF-κB , STAT1 , and AP1 in L . mexicana-infected macrophages was associated with the inhibition of IL-12 expression and generation of nitric oxide , both of which are important for initiation of a host immune response and parasite killing , respectively . Our finding that GP63 expression is down-modulated in the Δcpb mutant raises the possibility that cleavage of those transcription factors may actually be mediated by GP63 . Indeed , GP63 cleaves numerous host macrophage effectors , including PTP-1B , NF-κB , STAT1 , and AP1 [40] . Revisiting the role of CPB in the context of GP63 expression will be necessary to elucidate whether , and to which extent , CPB is acting directly on the host cell proteome . In sum , we discovered that CPB contributes to L . mexicana virulence in part through the regulation of GP63 expression . Complementation of Δcpb revealed the importance of GP63 for the virulence of L . mexicana , as it participates in the survival of intracellular parasites , PV expansion , and the formation of cutaneous lesions .
Experiments involving mice were done as prescribed by protocol 1406–02 , which was approved by the Comité Institutionnel de Protection des Animaux of the INRS-Institut Armand-Frappier . In vivo infections were performed as per Animal Use Protocol #4859 , which was approved by the Institutional Animal Care and Use Committees at McGill University . These protocols respect procedures on good animal practice provided by the Canadian Council on Animal Care ( CCAC ) . The mouse anti-GP63 monoclonal antibody was previously described [41] . The mouse anti-phosphoglycans CA7AE monoclonal antibody [42] was from Cedarlane and the rabbit polyclonal anti-aldolase was a gift from Dr . A . Jardim ( McGill University ) . Rabbit polyclonal antibodies for VAMP3 and VAMP8 were obtained from Synaptic Systems . Bone marrow-derived macrophages ( BMM ) were differentiated from the bone marrow of 6- to 8-week-old female 129XB6 mice ( Charles River Laboratories ) as previously described [43] . Cells were cultured for 7 days in complete medium ( DMEM [Life Technologies] supplemented with L-glutamine [Life Technologies] , 10% heat-inactivated FBS [PAA Laboratories] , 10 mM HEPES at pH 7 . 4 , and antibiotics ) containing 15% v/v L929 cell–conditioned medium as a source of M-CSF . Macrophages were kept at 37°C in a humidified incubator with 5% CO2 . To render BMM quiescent prior to experiments , cells were transferred to 6- or 24-well tissue culture microplates ( TrueLine ) and kept for 16 h in complete DMEM without L929 cell–conditioned medium . Promastigotes of L . mexicana wild-type strain ( MNYC/BZ/62/M379 ) and of L . major NIH S ( MHOM/SN/74/Seidman ) clone A2 were grown at 26°C in Leishmania medium ( Medium 199 supplemented with 10% heat-inactivated FBS , 40 mM HEPES pH 7 . 4 , 100 μM hypoxanthine , 5 μM hemin , 3 μM biopterin , 1 μM biotin , and antibiotics ) . The isogenic L . mexicana CPB-deficient mutant Δcpbpac ( thereafter referred to as Δcpb ) and its complemented counterpart Δcpbpac[pGL263] ( thereafter referred to as Δcpb+CPB ) were described previously [21] . L . mexicana Δcpb promastigotes were electroporated as described [44] with the pLEXNeoGP63 . 1 plasmid [32] to generate Δcpb+GP63 parasites . L . mexicana Δcpb+CPB and Δcpb+GP63 promastigotes were grown in the presence of 50 μg/ml hygromycin or 50 μg/ml G418 , respectively . The L . major NIH clone A2 isogenic Δgp63 mutant and its complemented counterpart Δgp63+gp63 have been previously described [32] . Cultures of Δgp63+gp63 promastigotes were supplemented with 50 μg/ml G418 . Prior to macrophage infections , promastigotes in late stationary phase were opsonized with DBA/2 mouse serum . For synchronized phagocytosis using parasites , macrophages and parasites were incubated at 4°C for 10 min and spun at 167 g for 1 min , and internalization was triggered by transferring cells to 34°C . Macrophages were washed twice after 2h with complete DMEM to remove the non-internalized parasites and were further incubated at 34°C for the required times . Cells were then washed with PBS and stained using the Hema 3 Fixative and Solutions kit ( Fisher HealthCare ) , or prepared for confocal immunofluorescence microscopy . Macrophages on coverslips were fixed with 2% paraformaldehyde ( Canemco and Mirvac ) for 40 min and blocked/permeabilized for 17 min with a solution of 0 . 05% saponin , 1% BSA , 6% skim milk , 2% goat serum , and 50% FBS . This was followed by a 2 h incubation with primary antibodies diluted in PBS . Macrophages were then incubated with a suitable combination of secondary antibodies ( anti-rabbit Alexa Fluor 488 and anti-rat 568; Molecular Probes ) and DAPI ( Life technologies ) . Coverslips were washed three times with PBS after every step . After the final washes , Fluoromount-G ( Southern Biotechnology Associates ) was used to mount coverslips on glass slides , and coverslips were sealed with nail polish ( Sally Hansen ) . Macrophages were imaged with the 63X objective of an LSM780 microscope ( Carl Zeiss Microimaging ) , and images were taken in sequential scanning mode . Image analysis and vacuole size measurements were performed with the ZEN 2012 software . Prior to lysis , macrophages were placed on ice and washed with PBS containing 1 mM sodium orthovanadate and 5 mM 1 , 10-phenanthroline ( Roche ) . Cells were scraped in the presence of lysis buffer containing 1% Nonidet P-40 ( Caledon ) , 50 mM Tris-HCl ( pH 7 . 5 ) ( Bioshop ) , 150 mM NaCl , 1 mM EDTA ( pH 8 ) , 5 mM 1 , 10-phenanthroline , and phosphatase and protease inhibitors ( Roche ) . Parasites were washed twice with PBS and lysed in the presence of lysis buffer containing 0 . 5% Nonidet P-40 ( Caledon ) , 100mM Tris-HCl ( Bioshop ) and 150 mM NaCl at -70°C . Lysates were thawed on ice and centrifuged for 10 min to remove insoluble matter . After protein quantification , protein samples were boiled ( 100°C ) for 6 min in SDS sample buffer and migrated in 12% or 15% SDS-PAGE gels . Three micrograms and 15 μg of protein were loaded for parasite and infected macrophage lysates , respectively . Proteins were transferred onto Hybond-ECL membranes ( Amersham Biosciences ) , blocked for 1 h in TBS 1X-0 . 1% Tween containing 5% skim milk , incubated with primary antibodies ( diluted in TBS 1X-0 . 1% Tween containing 5% BSA ) overnight at 4°C , and thence with appropriate HRP-conjugated secondary antibodies for 1 h at room temperature . Then , membranes were incubated in ECL ( GE Healthcare ) , and immunodetection was achieved via chemiluminescence . Membranes were washed 3 times between each step . For zymography , 2 μg of lysate were incubated at RT for 6 min in sample buffer without DTT and then migrated in 12% SDS-PAGE gels with 0 . 2% gelatin ( Sigma ) . Gels were incubated for 1 h in the presence of 50 mM Tris pH 7 . 4 , 2 , 5% Triton X-100 , 5 mM CaCl2 and 1 μM ZnCl2 and incubated overnight in the presence of 50 mM Tris pH 7 . 4 , 5 mM CaCl2 , 1μM ZnCl2 and 0 , 01% NaN3 at 37°C [45] . Late stationary phase promastigotes were incubated for 30 min in complete DMEM medium with 20% human serum from healthy donors . Parasites were then incubated in LIVE/DEAD Fixable Violet Dead Cell Stain Kit ( Life technologies ) and fixed in 2% paraformaldehyde . Flow cytometric analysis was carried out using the LSRFortessa cytometer ( Special Order Research Product; BD Biosciences ) , and the BD FACSDiva Software ( version 6 . 2 ) was used for data acquisition and analysis . Male BALB/c mice ( 6 to 8 weeks old ) were purchased from Charles River Laboratories and infected in the right hind footpad with 5x106 stationary phase L . mexicana promastigotes as described [46] . Disease progression was monitored by measuring footpad swelling weekly with a metric caliper , for up to 9 weeks post-infection . Footpads were then processed to calculate parasite burden using the limiting dilution assay . After 9 weeks of infection , mice were euthanized under CO2 asphyxiation and subsequently by cervical dislocation . The infected footpads were surface-sterilized with a chlorine dioxide based disinfectant followed by ethanol 70% for 5 min . Footpads were washed in PBS , lightly sliced , transferred to a glass tissue homogenizer containing 6 ml of PBS , and manually homogenized . The last step was repeated two to three times , until complete tissue disruption was achieved . Final homogenate was then centrifuged at 3 , 000 x g for 5 min and resuspended in the appropriate volume of PBS . 100 μl of homogenate were added in duplicates to 96-well plates containing 100 μl of complete Schneider’s medium in each well ( twenty-four 2-fold dilutions for each duplicate ) . Plates were incubated at 28°C . After 7–10 days , the number of viable parasites was determined from the highest dilutions at which promastigotes were observed using an inverted microscope [47] . Total RNA was extracted from promastigotes using the TRIzol reagent ( Invitrogen Life Technology , Carlsbad , CA ) and reverse transcribed . One-tenth of the resulting cDNA was amplified by PCR on a DNA thermal cycler , version 2 . 3 ( Perkin-Elmer Corporation , Norwalk , CT ) , with the following primer pairs: for the L . mexicana GP63 C-1 5'-ACCGTCTGAGAGTCGGAACT-3' ( forward ) , 5'-GTAGTCCAGGAATGGCGAGT-3' ( reverse ) ; the L . major GP63-1 5'-TCTGAGGCACATGCTTCGTT-3' ( forward ) , 5'-GTCAGTTGCCTTCGGTCTGA-3' ( reverse ) , the L . mexicana LPG2 5'CATTTGGTATCCTGGTGCTG-3' ( forward ) , 5'-GAGGAAGCCACTGTTAGCC-3' ( reverse ) , and the L . mexicana α-tubulin 5'-CTATCTGCATCCACATTGGC-3' ( forward ) , 5'-ACTTGTCAGAGGGCATGGA-3' ( reverse ) . The PCR products were analyzed by electrophoresis on a 3% ( wt/vol ) agarose gel , and the pictures were taken using AlphaImager 3400 imaging software ( Alpha Innotech Corporation , San Leandro , CA ) . Statistically significant differences were analyzed by ANOVA followed by the Tukey post-hoc test using the Graphpad Prism program ( version 5 . 0 ) . For the limiting dilution assay , the non-parametric Mann-Whitney or Kruskal-Wallis test was used . Values starting at P<0 . 05 were considered statistically significant . All data are presented as mean ± SEM .
|
The parasite Leishmania mexicana expresses several cysteine peptidases of the papain family that are involved in processes such as virulence and evasion of host immune responses . The cysteine peptidase CPB is required for survival within macrophages and for lesion formation in susceptible mice . Upon their internalization by macrophages , parasites of the L . mexicana complex induce the formation of large communal parasitophorous vacuoles in which they replicate , and expansion of those large vacuoles correlates with the ability of the parasites to survive inside macrophages . Here , we found that CPB contributes to L . mexicana virulence ( macrophage survival , formation and expansion of communal parasitophorous vacuoles , lesion formation in mice ) through the regulation of the virulence factor GP63 , a Leishmania zinc-metalloprotease that acts by cleaving key host cell proteins . This work thus elucidates a novel Leishmania virulence regulatory mechanism whereby CPB controls the expression of GP63 .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"blood",
"cells",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"parasite",
"replication",
"pathogens",
"immunology",
"microbiology",
"parasitic",
"diseases",
"protozoan",
"life",
"cycles",
"parasitic",
"protozoans",
"parasitology",
"developmental",
"biology",
"protozoans",
"leishmania",
"promastigotes",
"white",
"blood",
"cells",
"animal",
"cells",
"life",
"cycles",
"pathogenesis",
"cell",
"biology",
"leishmania",
"major",
"virulence",
"factors",
"host-pathogen",
"interactions",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"protozoology",
"macrophages",
"organisms"
] |
2016
|
Cysteine Peptidase B Regulates Leishmania mexicana Virulence through the Modulation of GP63 Expression
|
The nature of the “toxic gain of function” that results from amyotrophic lateral sclerosis ( ALS ) - , Parkinson- , and Alzheimer-related mutations is a matter of debate . As a result no adequate model of any neurodegenerative disease etiology exists . We demonstrate that two synergistic properties , namely , increased protein aggregation propensity ( increased likelihood that an unfolded protein will aggregate ) and decreased protein stability ( increased likelihood that a protein will unfold ) , are central to ALS etiology . Taken together these properties account for 69% of the variability in mutant Cu/Zn-superoxide-dismutase-linked familial ALS patient survival times . Aggregation is a concentration-dependent process , and spinal cord motor neurons have higher concentrations of Cu/Zn-superoxide dismutase than the surrounding cells . Protein aggregation therefore is expected to contribute to the selective vulnerability of motor neurons in familial ALS .
Amyotrophic lateral sclerosis ( ALS ) is an adult-onset neurodegenerative disease with roughly 10% of the cases being inherited or familial [1] . The cause of sporadic ALS ( sALS ) is unknown while familial ALS ( fALS ) is known to be caused by mutations in six different genes and six different chromosomal loci [2–4] . One of these genes encoding Cu/Zn-superoxide dismutase ( SOD1 ) was found to associate with 20% of fALS , and at least 119 fALS-associated SOD1 mutations have been characterized in humans [1 , 5] . Fifteen years after the discovery that SOD1 mutations can cause ALS [1] , the mechanisms of toxicity are still not well understood . The dominant inheritance of most SOD1 mutations and the literature as a whole indicate that SOD1 mutations result in a toxic gain of function rather than a loss of function [6–9] . Numerous hypotheses have been proposed , reviewed in [10 , 11] , and can be broken down conceptually into the ( nonexclusive ) toxic mechanisms that converge to SOD1 protein structure–function and those that converge elsewhere ( downstream effects ) . Popular hypotheses for SOD1 variant structure and function changes include decreased stability of apo or metallated SOD1 [12–15] , increased hydrophobicity [16] and aggregation propensity [17 , 18] , susceptibility to posttranslational modification [19–25] , loss of metals [22 , 26–32] , and aberrant chemistry [33–37] . Popular hypotheses for downstream effects [38–40] include impairment of axonal transport [41–43] , impairment of proteasome [39 , 44 , 45] or chaperone activity [46 , 47] , and mitochondrial [9 , 48–53] or endoplasmic reticulum–Golgi dysfunction [54 , 55] . Notably , the only potentially toxic property thus far shared by all fALS SOD1 variants is an increased propensity to form proteinaceous aggregates [56–62] . Due to the clinical similarities between fALS and sALS , research into SOD1 mutation-related fALS may provide insight into sporadic cases . Here we demonstrate that two properties , namely , increased aggregation propensity and instability ( loss of stability ) , are major contributors to SOD1 toxicity in ALS patients . On the basis of these results we rationalize the determinants of aggregation , the selective vulnerability of neurons , and patient survival times .
The goal of this study is to discover the mechanisms of toxicity of fALS SOD1 mutations . Neurologists often publish the age at onset and the time from disease onset to death ( also termed survival or disease duration ) for their ALS patients , thereby enabling epidemiological studies that assess the risk of a given variable [63–66] , which for this study include given mutations' relative toxicity and physical characteristics ( physicochemical parameters ) . Previous studies revealed that different SOD1 mutations have inherently different toxicities ( encode different mean disease durations ) [67] . We expanded upon these studies with a larger set of fALS-causing SOD1 mutations as well as larger patient cohorts . Hazard ratios ( relative risk of dying at a given time ) of fALS SOD1 mutations and non-SOD1-related fALS compared to that of sALS were obtained from the Cox proportional hazard model ( Table 1 ) . From this result , fALS SOD1 mutations with data from at least five individual patients ( full criteria for inclusion are defined in the Materials and Methods section ) were significantly related to different hazards . Kaplan–Meier survival curves from patients with fALS-causing SOD1 mutations , non-SOD1-related fALS , and sALS were generated . Figure 1 illustrates that fALS SOD1 mutations encode different prognoses , ranging from considerably better ( e . g . , H46R , hazard rate = 0 . 075 × sALS ) to considerably worse ( e . g . , A4V , hazard rate = 5 . 7 × sALS ) than sALS . Moreover , the Log rank , Breslow , and Tarone–Ware tests , which also compare patient survival rates ( i . e . , each mutation versus every other fALS-causing SOD1 mutation , non-SOD1-related fALS , and sALS ) using different mathematical functions , confirm that different SOD1 mutations have inherently different prognoses ( Table 2 ) . To test the hypothesis that changes to the physicochemical properties of SOD1 variants are toxic , specifically those properties known to influence protein aggregation , physicochemical properties for each protein variant ( hydrophobicity , propensity to lose α-helices , form β-sheets , protein net charge , etc . ) were evaluated in a Cox proportional hazard model ( Table 3 ) . The hazard ratios were significantly higher than 1 . 0 for mutations that either increase hydrophobicity , lose α-helices , or form β-sheets . In contrast , mutations that decrease the magnitude of the protein net charge correlate with hazard ratios significantly smaller than 1 . 0 . These results indicate that changes in the SOD1 variants' properties , specifically increases in hydrophobicity and propensity to lose α-helices and to form β-sheets , correlate with decreased fALS patient survival , while decreases of the magnitude of net charge correlate with increased fALS patient survival , in contradiction with previous reports [15 , 68] . Dobson and co-workers [69] introduced an equation ( termed the Chiti–Dobson equation herein ) to predict the changes of aggregation rates of unfolded peptides or proteins upon point mutations by their physicochemical properties . This equation was derived empirically by modeling how three physicochemical properties , hydrophobicity , secondary structure ( including loss of α-helix and gain of β-sheet ) , and protein net charge , change upon mutations ( the hazard analyses for each of these properties were reported in the previous paragraph ) . These physicochemical property changes then were related to changes in protein aggregation rate , yielding an equation that predicts how any mutation will change the rate of protein aggregation ( the predicted change of aggregation rate is referred to as the aggregation propensity ) . Although this equation is empirical , it is based upon first physical/chemical principles and approximates how a given mutation will change the energy ( and thus the equilibrium ) between a solvated and an aggregated protein . The Chiti–Dobson equation is ln ( νmut/νwt ) = 0 . 633ΔHydr + 0 . 198 ( ΔΔGcoil-α + ΔΔGβ-coil ) – 0 . 491Δcharge , in which ln ( νmut/νwt ) represents change of aggregation rate upon mutation , and ΔHydr , ΔΔGcoil-α , ΔΔGβ-coil , and Δcharge represent the changes of hydrophobicity , free energy change for the process from α-helix to random coil , free energy change for the process from random coil to β-sheet , and protein net charge from the mutation , respectively . In their landmark study , it was demonstrated that increases in hydrophobicity , losses of α-helices , gains of β-sheets , and decreases in the magnitude of protein net charge increase the rate of protein aggregation . The Chiti–Dobson equation and the many equations it inspired are robust and versatile , having successfully predicted aggregation rates of diverse disease-associated proteins [70] , including amyloid β-peptide [69] , tau [69] , α-synuclein [69] , amylin [69] , lysozyme [71] , etc . Moreover , increases in the predicted rates of aggregation of various mutations in amyloid β-peptide were shown to relate to increased neuronal dysfunction and degeneration in a Drosophila model of Alzheimer's disease [72] . To test the hypothesis that protein aggregation propensity is related to fALS patient survival , the Chiti–Dobson equation was used to predict the aggregation propensities of fALS-causing SOD1 mutations . We started this study by validating the Chiti–Dobson equation , taking all experimental protein aggregation rate data available at the inception of our study ( data reported as of 2005 , listed in Table 4 ) and recalibrating the equation . The detailed results of the validation are reported in Figure 2 . In summary , the Chiti–Dobson equation was verified for use in fALS , and the statistical correlation between the physicochemical parameters ( hydrophobicity , net charge , and secondary structure ) and the aggregation propensity remained and changed only marginally . Since the time we validated the Chiti–Dobson equation , a number of papers also validated their general approach [73–75] . Even so , we have included our own analysis since it provides exposure to the physical basis of aggregation propensity . Furthermore , inclusion of this data makes this study self-contained so that all of the data necessary to support or disprove our model are contained herein . Notably , this paper's conclusions were the same using both the original and the recalibrated Chiti–Dobson equation . The average patient survival times for different SOD1 variants with measured thermodynamic stabilities were plotted against corresponding predicted aggregation propensities , and linear regression analysis weighted by the number of patients for each mutation yielded R ( multiple correlation coefficient , with a larger value indicating a stronger relationship ) and P ( value less than 0 . 05 implies a significant result ) values of 0 . 58 and <0 . 001 , respectively ( Figure 3A ) . The severity of fALS thus is related to mutation-induced increases in SOD1 aggregation propensity . The same plot was performed with the linear regression analysis not weighted by the number of patients ( Figure 4A ) , yielding R and P values of 0 . 23 and 0 . 2 , respectively . Unfortunately , the published epidemiology data do not provide the information necessary to stratify for known ALS covariates , including lifestyle ( diet and smoking ) [76–79] , palliative care [80] , bulbar onset , etc . , and weighted data are more likely to account for differences in these factors . The Chiti–Dobson equation results for all fALS-causing SOD1 mutations with patients' survival data also were evaluated in a univariate Cox proportional hazard model ( Table 3 ) . The hazard ratio for the Chiti–Dobson equation result was significantly higher than 1 . 0 , which also indicates that aggregation propensity is a risk factor for fALS . Previous studies of Huntington's disease revealed an inverse relationship between the length of glutamine repeat of huntingtin and age of disease onset . The authors of this previous study concluded that disease onset correlates with rate of nucleation of aggregation [81] . We demonstrate here an inverse relationship between the rate of aggregation elongation after nucleation and the disease duration after onset . On the basis of our observation that predicted increased protein aggregation correlates with increased disease severity and previous data indicating that protein unfolding or misfolding promote aggregation [82–85] , we tested the hypothesis that a loss of protein stability also could be a risk factor for ALS . For the sake of simplicity , we use the term instability throughout this article , with instability defined as the inverse of either the normalized ΔΔG ( unfolding free energy change difference between mutant and wild-type SOD1 ) or normalized ΔTm ( melting point difference between mutant and wild-type SOD1 ) . Instability was considered for two reasons: ( 1 ) the Chiti–Dobson equation predicts the aggregation rates of unfolded proteins ( it was derived from the aggregation rates of proteins in high trifluoroethanol concentrations that contained secondary but no tertiary structure ) , and therefore , formally , unfolding must occur prior to aggregation , and ( 2 ) unfolding is known to speed protein aggregation in vitro to the extent that without chemically induced unfolding induction periods extend from months to years , as demonstrated for SOD1 [32] . Aggregation in vivo therefore may require protein unfolding . Before using stability data published by different laboratories using different methods ( melting point , which yields ΔTm , or chaotroph-induced unfolding , which yields ΔΔG ) , we sought to determine the reliability of the data . If different laboratories reported similar values of stability for the same mutants , then the data could be deemed reliable . Therefore , all published measurements of apo SOD1 stability ( metallated SOD1 calorimetry data often bear the characteristics of irreversible denaturation , probably via Cu-catalyzed disulfide bond formation , and is therefore less reliable ) [15 , 31 , 60 , 86–88] were compiled , and the experimental values of ΔΔG and ΔTm were normalized to the range from 0 to 1 ( described in the Materials and Methods section ) , with 0 representing the least stable , and 1 representing the most stable ( highest stability ) variant . Through the use of all of the data from mutants where ΔΔG and ΔTm were measured by different laboratories , a plot of normalized ΔΔG versus normalized ΔTm was created . Good interlaboratory correlation of measured stability values was observed ( slope = 0 . 94 , R = 0 . 90 , P = 0 . 002; Figure 5 ) , and we therefore deemed the stability data reliable for use . Next , patient survival data for fALS-causing SOD1 variants were plotted against corresponding instability values , and linear regression analysis weighted by the number of patients for each mutation yielded R and P values of 0 . 71 and <0 . 001 , respectively ( Figure 3B ) . The same plot was performed with the linear regression analysis not weighted by the number of patients ( Figure 4B ) , yielding R = 0 . 34 and P = 0 . 07 . A gain of SOD1 instability ( loss of stability ) upon mutation therefore is related to decreased fALS patient survival . Increased in vitro instability is consistent with previous findings that the in vivo half-lives of SOD1 variants are decreased [89] . Previous results from computer simulations indicate a multistep process for aggregation via destabilization [90] , encouraging us to understand the combined effect of aggregation propensity and protein instability upon ALS patient survival . On the basis of their respective multiple correlation coefficients and slopes , aggregation propensity and instability are equal contributors to fALS patients' survival . Moreover , no obvious correlation between protein instability and aggregation propensity was observed for the SOD1 variants used in Figure 3 ( Figure S1 ) , indicating that increased instability is not responsible for the increased predicted protein aggregation propensity . The combination of instability and aggregation propensity represents the relative energy in proceeding from folded to unfolded apo SOD1 and then from unfolded to aggregated states . Patient survival was plotted against corresponding summed instability and aggregation propensity values . A linear regression analysis weighted by the number of patients for each mutation yielded R and P values of 0 . 83 and <0 . 001 , respectively ( Figure 3C ) . The same plot was performed with the linear regression analysis not weighted by the number of patients ( Figure 4C ) , yielding R = 0 . 47 and P = 0 . 01 . The improved statistical result of predicting patient survival after combining instability and aggregation propensity indicates that aggregation occurs from unfolded or partially unfolded SOD1 . The stability data used herein were for apo SOD1 , and therefore the absence of metals is implicit . The R2 value was 0 . 69 from the weighted data , indicating that 69% of the intrinsic variability these fALS patients' survival resulted from the combination of increased aggregation propensity and instability . Additionally , aggregation propensity and instability were evaluated in a Cox proportional hazard model ( Table 5 ) . The hazard ratios were significantly higher than 1 . 0 for both factors . The sum of aggregation propensity and instability also was evaluated in a univariate Cox proportional hazard model ( Table 5 ) . The hazard ratio for this sum was also significantly higher than 1 . 0 , further indicating that aggregation propensity and instability are synergistic risk factors for fALS . Note that the aggregation propensity and instability tested in Table 5 were normalized to the range from 0 to 1 ( as in Figures 3 and 4 ) , while the values tested in Table 3 were not normalized . As a result of normalization , which decreased the value range of tested factors , the hazard ratios of Table 5 are much larger than Table 3 , and therefore the large values of hazard ratios reported in Table 5 should not be overinterpreted . Significantly , a fALS patient with an SOD1 mutation of relatively low aggregation propensity and high stability is expected to survive longer after disease onset . It has not escaped our attention that the rate of protein aggregation has implications in both sporadic diseases and aging; for example , the toxicity of a given posttranslational modification is a function of its effect on protein stability and aggregation propensity .
We describe here synergistic gains of toxic functions of SOD1 in ALS . These are the first results in any neurodegenerative disease demonstrating that protein instability and aggregation propensity are synergistic risk factors . The fact that there are two synergistic risk factors rather than a single toxic gain of function probably has delayed the discovery of the mechanisms of fALS mutant SOD1 toxicity . The SOD1 stability data used in this paper were measured from apo SOD1 , and the aggregation rate data used to create the Chiti–Dobson model were from in vitro unfolded proteins . Therefore , formally , the combination of instability and aggregation propensity represents the relative energy in proceeding from apo folded to unfolded SOD1 and then from unfolded to aggregated states . It has been demonstrated experimentally that apo SOD1 has a faster rate of aggregation than that of holo forms [32] . Partial unfolding/misfolding also can lead to aggregation [28 , 91–96] , and our results cannot rule out a role for the aggregation of partially folded , including metallated , SOD1 . Previous studies revealed a correlation [15] and conversely a lack of correlation [86] between SOD1 variant stability and patient disease duration . Correlation between SOD1 variant stability and patient disease duration , however , required that the authors omit stability data of 4 of the 15 variants from their regression analysis ( on the basis that these variants change the net charge of SOD1 ) . As presented in Table 3 , SOD1 variants' loss of net charge correlates with increased patient survival , while gain of hydrophobicity , loss of α-helix , and gain of β-sheet propensity are ALS risk factors . On the basis of Dobson and co-workers' related work [69 , 73 , 97] , a loss of net charge is predicted to increase the aggregation propensity of unfolded proteins . If aggregation is toxic , then one would expect loss of net charge to be toxic . In contrast to the synergistic effects for aggregation propensity and instability presented in Table 5 , the correlation of loss of net charge with increased survival has an effect of decreasing the hazard ratio presented in the univariate model presented in Table 3 . We demonstrate that mutations causing the entire protein to approach neutrality are protective in the context of fALS ( Table 3 ) rather than deleterious as proposed by Oliveberg and co-workers [15 , 68] . These results should be cautiously interpreted since in contrast to our Cox proportional hazard model result that loss of net charge is protective , the mean patient survival for loss of net charge and gain of net charge mutations , unweighted by the number of patients , are 7 . 1 and 6 . 9 years , respectively . Further study clearly is required to understand the role of charge in ALS etiology . In contrast with the strong familiality shown for disease duration after onset ( Table 1 ) , SOD1-mediated ALS showed modest familiality with respect to onset , accounting for only 42% of the variability in A4V and D90A fALS patients [98] , and with only G37R and L38V mutations of SOD1 being significant covariates of age of onset [67] . The same analysis shown in Figures 3 and 4 was performed using age at disease onset rather than disease duration as the dependant variable ( Figure S2 ) , and little or no relationship between disease onset and aggregation propensity or instability was observed . The Chiti–Dobson equation predicts the rate of aggregation after nucleation ( rate of elongation ) . It is tempting therefore to speculate that the rate of nucleation is a determinant of age at onset . Testing this hypothesis would require the development of a model that can predict nucleation rates based upon physicochemical parameters , a task that is hampered by the stochastic nature of in vitro nucleation times [99 , 100] but that should now be possible given our recent development of methods for modeling in vitro nucleation kinetics [101] . Although our model accounts for 69% of the variability in fALS patient survival after onset , there are clearly genetic components of fALS that our model cannot account for . For example , while D90A is normally a dominantly inherited mutation in North America , 2 . 5% of people in Sweden and Finland are heterozygous asymptomatic carriers of the D90A SOD1 mutation [102 , 103] and require two mutant alleles before presenting ALS symptoms . Notably , our results and conclusions were unaffected by including or excluding D90A survival times during data analysis . It is postulated that diseases for which protein aggregation contributes to patient death will ( 1 ) develop in cells with the highest concentration of the aggregation-prone protein in accordance with the concentration dependence of aggregation rates [101 , 104] and ( 2 ) have a prognosis influenced by the aggregation propensity of the aggregating protein , in accordance with the results reported herein . Motor neurons are the cells in the ventral horn of the spinal cord with the highest SOD1 concentration [39 , 105] , perhaps explaining an aspect of the selective vulnerability of these cells . Protein aggregation is a hallmark of many neurodegenerative diseases , including ALS . The toxicity of aggregation is fiercely debated ( reviewed in [106 , 107] ) , fueled by reproducible evidence that aggregates can be toxic [108 , 109] , have no effect [110] , or be protective [111] . We propose that aggregates on either extreme of size , i . e . , small protofibrillar aggregates [108 , 112] or aggregates large enough to clog axons , are more toxic , while midsize microscopically visible aggregates are less toxic [106] . Our data indicate that the increased aggregation propensity of SOD1 is related to decreased survival of ALS patients . Notably aggregation of SOD1 has been demonstrated in fALS [113 , 114] and a subset of sALS patients [114 , 115] , 18 fALS rodent models of 13 different SOD1 mutations [7–9 , 23 , 59 , 116–128] , and at least 13 SOD1 mutants in cell models [56 , 58 , 59 , 116 , 118 , 119 , 129–131] .
Familial ALS patients' data were taken from all of the available literature . Disease duration was initiated with onset of the first symptoms until the patient's death or when respiratory assistance was required for patients' survival . The average duration and onset for each mutation were calculated as the weighted average based on the number of patients ( Table 6 ) . If the patients were still reported to be alive without respiratory assistance at the end of a study , then their disease durations were not used to calculate the average unless the known duration value was larger than the average calculated with only durations from patients deceased or with respiratory assistance . For studies reporting average disease duration and Kaplan–Meier curves , the reported average durations were used to calculate the weighted averages . The current unavailability of http://www . alsod . org/ made it impossible to review the references provided by the website ( from which we had taken survival times before it became unavailable ) , which created the risk of counting a patient's disease onset or survival twice , and made reproducing our study impossible for other groups . We therefore opted not to report data from this website in this study , thereby eliminating no more than 67 ( there were 67 http://www . alsod . org/ patients' data without accompanying literature references that may , or may not , have been represented by our literature search ) of 1319 patients' data . However , we did perform a complete , alternative set of analyses that did include http://www . alsod . org/ data ( unpublished data ) , and the statistical correlations in the figures and tables shown herein persisted . Mean values of disease durations also were obtained from Kaplan–Meier curves and tested on SOD1 mutations with known experimental thermodynamic stabilities , and the results were comparable to those in Figure 3 . Since the weighted average method can provide disease duration regardless of the number of patients , we opted for its use . Kaplan–Meier curves of survival for different fALS-causing SOD1 mutations , non-SOD1-related fALS , and sALS were generated . The hazard ratios of different fALS-causing SOD1 mutations and non-SOD1-related fALS compared to sALS were tested as a category variable by Cox proportional hazard model analysis . For studies reporting Kaplan–Meier curves but without individual patients' data , the Engauge Digitizer 4 . 1 software was used to obtain coordinates for cumulative survival at each time point . This information was used to calculate the number of patients not surviving at each time point under the assumption that there is no censored patient ( with unknown exact survival time because of being alive at the end of study , lost to follow-up , or withdrawal from the study ) within the course of survival curves . For cumulative survival not reaching 0 at the end of study , those fractions of patients were treated as censored . The error of the estimated number of patients is less than 5% of the number reported . To eliminate the chance that one or two patients' survival data bias the analysis result , a rule of thumb [132] requiring that each tested fALS-related SOD1 mutation includes at least five noncensored patients was applied . Since patients' survival was reported only as an average from a group of patients and individual patient's survival information was not described in some publications , one or two publications' survival data might bias the analysis result . To eliminate this chance of bias , the rule of thumb was modified as requiring at least five independent descriptions of noncensored patients' survival data ( a reported average without individual patient's survival information was treated as one description ) . The statistical analysis was performed with the software SPSS 15 . 0 ( SPSS , Inc . ) . The hydrophobicity , β-sheet propensity , and charge values for the amino acid residues were obtained from the Supplementary Information of [69] . While applying the AGADIR algorithm at http://www . embl-heidelberg . de/Services/serrano/agadir/agadir-start . html to obtain α-helical propensities for wild-type ( wt ) and mutant ( mut ) Pαwt and Pαmut values for ΔΔGcoil-α calculation for human SOD1 , the parameters of pH 7 , 310 K , and ionic strength of 0 . 100 were used . For the protein human SOD1 , the N terminus is acetylated , and the C terminus is free in vivo . After the prediction at the residue level was output , the value in the column “Hel” at a specific residue was taken as Pα . If a value of 0 for Pα was obtained , then 0 . 1 was added to both Pαwt and Pαmut values for the correct mathematical meaning of ln ( Pαwt/Pαmut ) ( F . Chiti , personal communication ) . The Chiti–Dobson equation terms , Δhydrophobicity , Δcharge , ΔΔGcoil-α , and ΔΔGβ-coil , were calculated based on equations illustrated in the legend of Table 1 of [69] . The ln ( νmut/νwt ) values were calculated based on Equation 1 from [69] and normalized from 0 to 1 using the equation normalized aggregation propensity = ( aggregation propensity before normalization – MINap ) / ( MAXap – MINap ) , with MINap and MAXap as the minimum and maximum aggregation propensities of fALS-causing mutations with known thermodynamic stabilities , respectively , so that the larger normalized values correlate to larger aggregation propensities . The free energy change difference ( ΔΔG ) and melting temperature difference ( ΔTm ) of unfolding a pathogenic variant and wild-type protein are parameters used to characterize the thermodynamic stability of a protein . ΔΔG values were taken from Table 2 of [15] . To graph with other protein stability data , the ΔΔG values were normalized by applying the equation normalized ΔΔG = ( ΔΔG values before normalization – MINΔΔG ) / ( MAXΔΔG – MINΔΔG ) , with MINΔΔG and MAXΔΔG as the minimum and maximum values of ΔΔG in this dataset , respectively . ΔTm values were taken from Table 1 of [86] , Table 1 of [60] , Table 2 of [87] , Table 3 of [88] , and Table II of [31] . ΔTm values from [60 , 87 , 88] were averaged for each mutation . Those results then were averaged with the ΔTm values from [31 , 86] to determine the ΔTm values for given mutations . The normalized ΔTm values were obtained by applying the equation normalized ΔTm = ( average ΔTm values before normalization – MINΔTm ) / ( MAXΔTm – MINΔTm ) , with MINΔTm and MAXΔTm as the minimum and maximum values of averaged ΔTm values in this dataset , respectively . The instability values for SOD1 variants were obtained from the equation normalized instability = 1 – average of normalized ΔΔG and normalized ΔTm for each mutation , so instability values are simply ( 1 – normalized ΔΔG or ΔTm ) , and larger values correlate to less stable variants . The normalized aggregation propensity and instability for each variant were summed and normalized to the range from 0 to 1 to consider the two factors together .
|
Amyotrophic lateral sclerosis ( ALS ) , also known in America as Lou Gehrig's disease , is a fatal neurodegenerative disease with no effective treatment . Paralysis occurs as the result of the death of cells that connect the brain to various muscles , namely , the motor neurons of the brain and spinal cord . Ninety percent of ALS is sporadic and of unknown cause . A landmark discovery in ALS research was that mutations in the gene coding for Cu/Zn-superoxide dismutase cause at least 2% of ALS , and researchers have since discovered at least 119 such mutations . Neurologists also discovered that different mutations have remarkably different prognoses . For example , patients with the A4V mutation survive an average of 1 year after diagnosis , whereas patients with the H46R mutation survive an average of 18 years . Biochemists discovered that different mutations result in remarkably different physical properties , for example , stability of Cu/Zn-superoxide dismutase . In this article we apply an algorithm that predicts how fast a given Cu/Zn-superoxide dismutase will aggregate ( stick to other proteins ) and demonstrate that faster aggregation relates to faster death of ALS patients . We also demonstrate that loss of Cu/Zn-superoxide dismutase stability relates to faster ALS patient death . Our findings imply that aggregation of unfolded SOD1 is toxic for ALS patients , and in fact accounts for 69% of the variability in mutant Cu/Zn-superoxide-dismutase-linked familial ALS patient survival times .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"neurological",
"disorders"
] |
2008
|
Protein Aggregation and Protein Instability Govern Familial Amyotrophic Lateral Sclerosis Patient Survival
|
As antimicrobial resistance increases , it is crucial to develop new treatment strategies to counter the emerging threat . In this paper , we consider combination therapies involving conventional antibiotics and debridement , coupled with a novel anti-adhesion therapy , and their use in the treatment of antimicrobial resistant burn wound infections . Our models predict that anti-adhesion–antibiotic–debridement combination therapies can eliminate a bacterial infection in cases where each treatment in isolation would fail . Antibiotics are assumed to have a bactericidal mode of action , killing bacteria , while debridement involves physically cleaning a wound ( e . g . with a cloth ) ; removing free bacteria . Anti-adhesion therapy can take a number of forms . Here we consider adhesion inhibitors consisting of polystyrene microbeads chemically coupled to a protein known as multivalent adhesion molecule 7 , an adhesin which mediates the initial stages of attachment of many bacterial species to host cells . Adhesion inhibitors competitively inhibit bacteria from binding to host cells , thus rendering them susceptible to removal through debridement . An ordinary differential equation model is developed and the antibiotic-related parameters are fitted against new in vitro data gathered for the present study . The model is used to predict treatment outcomes and to suggest optimal treatment strategies . Our model predicts that anti-adhesion and antibiotic therapies will combine synergistically , producing a combined effect which is often greater than the sum of their individual effects , and that anti-adhesion–antibiotic–debridement combination therapy will be more effective than any of the treatment strategies used in isolation . Further , the use of inhibitors significantly reduces the minimum dose of antibiotics required to eliminate an infection , reducing the chances that bacteria will develop increased resistance . Lastly , we use our model to suggest treatment regimens capable of eliminating bacterial infections within clinically relevant timescales .
Antimicrobial resistance ( AMR ) is on the rise [1–3] and with it the need to develop and apply novel treatment strategies [4 , 5] . In this paper , we formulate and analyse mathematical models of combination therapies , bringing together traditional antibiotics and debridement with a new anti-adhesion treatment , seeking to determine if a combination therapy could succeed in eliminating an AMR infection in cases where antibiotics alone would fail . It has been predicted that unless steps are taken to combat its rise , AMR could lead to as many as 10 million deaths per annum by the year 2050 [1] . Antibiotics are the standard treatment option for microbial infections . They may be classified into two broad categories: bactericidal and bacteriostatic [6] . Bactericidal antibiotics act by killing bacteria , while bacteriostatic antibiotics inhibit their growth ( we note that some antibiotics may exhibit both modes of action ) . While effective in general , antibiotic use has the unfortunate consequence of selecting for those members of a bacterial population which are resistant to the antibiotic being applied . Resistance then spreads through the bacterial population via vertical ( parent to daughter ) and/or horizontal ( cell to cell ) gene transfer , until the resistant phenotype comes to dominate [7–9] . One solution to this problem is to use multiple antibiotics; however , this runs the risk of selecting for multi-drug resistant bacteria , or ‘super bugs’ [10] . An alternative approach is to use a class of treatments known as anti-virulence therapies , either in place of , or in addition to , antibiotics . Anti-virulence therapies are diverse [11–13]; however , they have the common aim of preventing or limiting disease in the host [6] . By using these therapies in combination with more traditional treatments , such as antibiotics and debridement ( physical clearance of a wound e . g . with a cloth ) , it is hoped that bacteria can be cleared from a host more rapidly , while reducing the risk of resistant phenotypes emerging [14 , 15] . In this paper we shall consider a particular form of anti-virulence therapy , known as anti-adhesion therapy , which operates by preventing bacteria from binding to the cells of an infected host , thus rendering them more susceptible to physical clearance [13] . Krachler et al . [12] have developed an anti-adhesion treatment based upon a protein , discovered earlier by the same group , which they named multivalent adhesion molecule ( MAM ) 7 [16] . MAM7 is anchored in the extracellular side of the outer membrane of many Gram-negative bacteria , where it is responsible for mediating the initial stages of attachment to host cells [16 , 17] . By chemically coupling polystyrene microbeads to MAM7 , adhesion inhibitors ( henceforth inhibitors ) can be constructed which , when applied to an infection site , competitively inhibit the binding of bacteria to host cells [18] . Burn wound infections provide a promising application of this treatment [18–20] . Nosocomial ( hospital-acquired ) infections pose a major challenge in the treatment of burn wound patients , as these wounds create a significant opportunity for bacteria to penetrate host defences [21–25] . Here we consider the potential of an anti-adhesion–antibiotic–debridement combination therapy to clear an infection , preventing further tissue damage and sepsis . The mathematical model developed in the present study extends our earlier model in Roberts et al . [26] , which considered the response of a purely susceptible bacterial infection to treatment with inhibitors and debridement . Our models predicted that , when combined with debridement , the bacterial burden could be significantly reduced and , in some cases , eliminated . The present study extends this model by considering mixed susceptible and resistant infections and an augmented treatment strategy , combining inhibitors and debridement with antibiotics . This is the first mathematical modelling study: ( i ) to consider the effects of antibiotic in a situation where bacteria can exist in either bound or unbound states in the absence of a biofilm ( [27] and [28] , noted below , do not include antibiotic treatment ) ; ( ii ) to consider a treatment combining antibiotics with anti-adhesion therapy , or ( iii ) to predict optimal antibiotic-inhibitor-debridement treatment regimens . As in [26] , the present mathematical model is based upon the in vivo rat burn wound model described in Huebinger et al . [18] . In each experiment , a burn wound was administered to the back of a rat and a portion of the resulting necrotic tissue later excised . An inoculum of the Gram-negative Pseudomonas aeruginosa ( P . aeruginosa ) bacteria was then applied to the wound , together with an active or inactive form of the inhibitor . The bacterial burden was monitored for six days , after which each rat was euthanised ( see [18] for further details ) . The treatment was found to effect a marked reduction in the total bacterial burden compared to controls . A mathematical model of a generalised anti-virulence treatment combined with antibiotics was proposed by Ternant et al . [29] . This ODE model conceived of anti-virulence treatment as providing a boost to the immune system , though it did not consider an anti-adhesion therapy specifically . The model predicted that antibiotics and anti-virulence treatments could be effective when used in combination , in cases where neither is effective in isolation , provided the therapies are administered in staggered doses . A number of modelling studies have considered bacterial infections of burn wounds [30–35] , the binding of bacteria to surfaces [27 , 28] and anti-virulence treatments which interfere with quorum sensing [33 , 36–42] . There is also a large literature on the mathematical modelling of antibiotic therapy ( see [43–48] for reviews ) . In this paper we develop an ODE model to describe and predict the bacterial population dynamics in an infected burn wound , under treatment regimens combining antibiotic , inhibitor and debridement therapies . We fit our antibiotic-associated parameters to new in vitro data collected for this study . We use our models to gain insight into how these combination therapies operate , to predict treatment outcomes and to suggest ways in which therapy could be optimised in a clinical setting . Crucially , it is found that anti-adhesion–antibiotic-debridement combination therapies can eliminate bacterial infections in situations where each treatment would fail when used in isolation .
We construct a mathematical model of an infected burn wound , focussing upon the bacterial population and the treatment strategies employed to clear an infection . For our purposes a burn wound consists of a layer of host cells , over which lies a fluid layer , exuded by the host cells , called the exudate . The exudate is partially covered by a layer of necrotic tissue , except in the region of a surgical excision , where it is exposed to the air and from which fluid may leak . If left undisturbed , a scab forms across the excision after 24 hr , preventing further fluid loss ( see Fig 1 ( a ) ) . The environment-dependent parameters used in this paper were fitted to an in vivo rat model with the bacterial species P . aeruginosa in [26]; however , this model is also of relevance to burn wounds in humans and for any bacterial species for which host cell attachment is partly mediated by MAM7 . See the ‘Experimental set-up’ and ‘Model formulation’ sections of [26] for more details . Our model considers three types of species: bacteria , inhibitors and antibiotics . Both bacteria and inhibitors may exist in one of two physical states , either swimming/floating freely in the exudate or bound to the host cells , while antibiotics remain in solution in the exudate at all times . Further , bacteria come in two varieties: those which are more vulnerable to antibiotic ( susceptible bacteria ) and those which have developed resistance to the antibiotic ( resistant bacteria ) . In this study we model a bactericidal antibiotic , employing parameter values fitted to newly measured in vitro kill curves for P . aeruginosa using the antibiotic meropenem ( see Parameter fitting and justification and S1 Text for more details ) . Meropenem is commonly used to treat P . aeruginosa burn wound infections [49] and can be administered intravenously; thus , it is a natural choice for this study . Inhibitors are applied directly to the exudate , whereas antibiotics are applied systemically , entering the wound through the host cell layer , having reached the wound via the bloodstream . Assuming , as in [26] , that the system is well-mixed , we define an ordinary differential equation ( ODE ) model for free susceptible bacteria density , B F S ( t ) ( cells cm−3 ) , free resistant bacteria density , B F R ( t ) ( cells cm−3 ) , bound susceptible bacteria density , B B S ( t ) ( cells cm−2 ) , bound resistant bacteria density , B B R ( t ) ( cells cm−2 ) , free inhibitor concentration , IF ( t ) ( inhib . cm−3 ) , bound inhibitor concentration , IB ( t ) ( inhib . cm−2 ) , and antibiotic concentration A ( t ) ( μg cm−3 ) , over time , t ( hr ) ( the dependent and independent variables are summarised in Table 1 ) . It is assumed that the total binding site density on the host cells , consisting of both free and occupied sites , is conserved , such that the free binding site density E ( t ) = E t o t a l - ϕ B a c ( B B S ( t ) + B B R ( t ) ) - ϕ I I B ( t ) ( sites cm−2 ) , where Etotal ( sites cm−2 ) is the total density of binding sites ( both free and bound ) , and ϕBac ( sites cell−1 ) and ϕI ( sites inhib . −1 ) are the number of binding sites occupied by a bacterium or an inhibitor respectively . The model , summarised in Fig 1 ( b ) and 1 ( c ) , is described by the following governing equations d B F S d t = [ r F S B F S H ( V B F S - 1 ) ︸ logistic growth + ρ B F R H ( V B F R - 1 ) H ( K F - B F S - B F R ) ︸ segregation ] ( 1 - B F S + B F R K F ) + ( 1 - η ( E ) ) H ( K B - B B S - B B R ) ︸ daughter cells freed from host cells upon division × 1 h [ r B S B B S H ( A r B B S - 1 ) ︸ logistic growth + ρ B B R H ( A r B B R - 1 ) ︸ segregation ] ( 1 - B B S + B B R K B ) - α B a c A r B F S E ︸ binding to host cells + β B a c h B B S ︸ unbindingfrom host cells - E m a x S A A 50 S + A B F S ︸ killing byantibiotic - λ B F S B F R ︸ conjugation - ψ B a c ( t ) B F S ︸ natural clearance , ( 1 ) d B F R d t = [ ( 1 - c H ( K F - B F S - B F R ) ) r F S B F R ︸ logistic growth - ρ B F R H ( K F - B F S - B F R ) ︸ segregation ] × ( 1 - B F S + B F R K F ) H ( V B F R - 1 ) + ( 1 - η ( E ) ) H ( K B - B B S - B B R ) ︸ daughter cells freed from host cells upon division × 1 h [ ( 1 - c ) r B S B B R ︸ logistic growth - ρ B B R ︸ segregation ] ( 1 - B B S + B B R K B ) H ( A r B B R - 1 ) - α B a c A r B F R E ︸ binding to host cells + β B a c h B B R ︸ unbindingfrom host cells - E m a x R A A 50 R + A B F R ︸ killing byantibiotic + λ B F S B F R ︸ conjugation - ψ B a c ( t ) B F R ︸ natural clearance , ( 2 ) d B B S d t = [ 1 + ( η ( E ) - 1 ) H ( K B - B B S - B B R ) ] ︸ a proportion , η , remain attached × [ r B S B B S H ( A r B B S - 1 ) ︸ logistic growth + ρ B B R H ( A r B B R - 1 ) H ( K B - B B S - B B R ) ︸ segregation ] × ( 1 - B B S + B B R K B ) + α B a c V B F S E ︸ binding to host cells - β B a c B B S ︸ unbindingfrom host cells - δ B B B S ︸ phagocytosis - ω E m a x S A A 50 S + A B B S ︸ killing byantibiotic , ( 3 ) d B B R d t = [ 1 + ( η ( E ) - 1 ) H ( K B - B B S - B B R ) ] ︸ a proportion , η , remain attached × [ ( 1 - c H ( K B - B B S - B B R ) ) r B S B B R ︸ logistic growth - ρ B B R H ( K B - B B S - B B R ) ︸ segregation ] × ( 1 - B B S + B B R K B ) H ( A r B B R - 1 ) + α B a c V B F R E ︸ binding to host cells - β B a c B B R ︸ unbindingfrom host cells - δ B B B R ︸ phagocytosis - ω E m a x R A A 50 R + A B B R ︸ killing byantibiotic , ( 4 ) d I F d t = - α I A r I F E ︸ binding tohost cells + β I h I B ︸ unbindingfrom host cells - ψ I ( t ) I F ︸ naturalclearance , ( 5 ) d I B d t = α I V I F E ︸ binding tohost cells - β I I B ︸ unbindingfrom host cells , ( 6 ) d A d t = { - δ A A ︸ elimination discrete dosing , 0 constant concentration , ( 7 ) where parameter definitions and values are given in Tables 2–4 . See Parameter fitting and justification for details on how the parameter values were obtained . We note that this model differs from that presented in [26] in the following respects: These key extensions to our model in [26] facilitate investigation into how the combination therapies presented here can best be employed to tackle an otherwise untreatable antibiotic resistant infection . Both free and bound bacteria are assumed to grow logistically with carrying capacities KF ( cells cm−3 ) and KB ( cells cm−2 ) respectively . In our model , the carrying capacities represent the maximum number of bacteria that can be sustained by available nutrients and are such that bacterial division is negligible when B F S ( t ) + B F R ( t ) = K F or B B S ( t ) + B B R ( t ) = K B ( see [62 , 63] ) . It is important to note that the number of bacteria that can be supported by nutrients near the host cells is not in general equal to the number of available binding sites on the host cells ( KB ≠ Etotal/ϕBac ) , indeed KB < Etotal/ϕBac for all parameter sets considered here ( see Tables 2 and 4 ) . Susceptible bacteria have intrinsic growth rates r F S ( hr−1 ) ( free ) and r B S ( hr−1 ) ( bound ) , while resistant bacteria incur a fitness cost , 0 < c < 1 ( dimensionless ) , such that their intrinsic growth rates are ( 1 - c ) r F S ( hr−1 ) ( free ) and ( 1 - c ) r B S ( hr−1 ) ( bound ) . This fitness cost only operates when the logistic terms represent bacterial growth . If the density of free cells , B F S ( t ) + B F R ( t ) , exceeds the free carrying capacity , KF , then the free logistic growth term becomes a death term , and likewise for bound bacteria . In this case the intrinsic growth rates of resistant bacteria revert to those of susceptible bacteria , since resistant bacteria are assumed to die at the same rate as susceptible bacteria . This is achieved through the use of Heaviside step functions , H ( K F - B F S ( t ) - B F R ( t ) ) and H ( K B - B B S ( t ) - B B R ( t ) ) , in Eqs 2 and 4 , where H ( x ) ≔ { 0 if x < 0 , 1 if x ≥ 0 . ( 8 ) Further , the growth of any bacterial subtype ( free-susceptible/free-resistant/bound-susceptible/bound-resistant ) ceases once the number of bacteria in that subtype falls beneath one , since at least one cell is required in order for division to be possible . This is achieved using the Heaviside step functions H ( V B F S ( t ) - 1 ) , H ( V B F R ( t ) - 1 ) , H ( A r B B S ( t ) - 1 ) and H ( A r B B R ( t ) - 1 ) , in Eqs 1–4 , where H is defined in Eq 8 . Daughter cells derived from bound bacteria may enter either the bound compartment ( in the proportion 0 ≤ η ( E ( t ) ) ≤ 1 ( dimensionless ) ) or the free compartment ( 1 − η ( E ( t ) ) ) , the proportion entering the bound compartment increasing as the density of free binding sites , E ( t ) , increases . We model this dependence using a Hill function as follows η ( E ) = η m a x E n γ n + E n , ( 9 ) where ηmax ( dimensionless ) is the maximum proportion of daughter cells which may remain bound to the surface , γ ( sites cm−2 ) is the binding site density at which η ( E ) = ηmax/2 and n ( dimensionless ) is the Hill coefficient . We use a Heaviside step function , H ( K B - B B S ( t ) - B B R ( t ) ) , in Eqs 1–4 to restrict cell death due to the bound logistic growth term to the bound compartment when B B S ( t ) + B B R ( t ) > K B , where H is defined in Eq 8 . The resistant strain of P . aeruginosa used in our in vitro experiments , PA1004 Evo10 , transfers resistance genes vertically , but not horizontally . Therefore , throughout most of this study we neglect horizontal gene transfer and segregation . We include conjugation and segregation terms in Eqs 1–11 so as to make our model relevant to a wider class of infections , performing a sensitivity analysis on these parameters in Sensitivity analysis . In those cases where horizontal gene transfer does occur , resistant bacteria transfer plasmids conferring resistance to susceptible bacteria via conjugation at a rate λ ( cm3cell−1hr−1 ) . It is assumed that this process occurs within the free compartment , but not within the bound compartment or between the two compartments , since bound bacteria are typically physically separated from each other and free bacteria are unlikely to interact with bound bacteria . Horizontal gene transfer can also occur via transformation and transduction; however , we consider only conjugation here since it is the most common of the three mechanisms [10] . When a bacterium divides , its plasmids are segregated ( divided ) between the resulting daughter cells . A portion of the daughter cells of resistant bacteria produced upon division fail to inherit the resistance plasmid , leading to the production of susceptible offspring ( by resistant bacteria ) at a rate ρ ( hr−1 ) ( see [53] for an example ) . Similarly to the processes described above , segregation only occurs where the number of free or bound bacteria are below carrying capacity and where the number of free or bound resistant bacteria is greater than one . This is achieved through the use of Heaviside step functions , H ( K F - B F S ( t ) - B F R ( t ) ) , H ( K B - B B S ( t ) - B B R ( t ) ) , H ( V B F R ( t ) - 1 ) and H ( A r B B R ( t ) - 1 ) , in Eqs 1–4 , where H is defined in Eq 8 . Bacteria and inhibitors bind to and unbind from the host cells with respective binding rates αBac ( hr−1 sites−1 ) and αI ( hr−1 sites−1 ) , and unbinding rates βBac ( hr−1 ) and βI ( hr−1 ) , in accordance with the law of mass action . Neutrophils are present only on the surface of the host cells and are fully upregulated throughout an infection , such that bound bacteria can be assumed to decay exponentially at rate δB ( hr−1 ) , where δB accounts for neutrophil density . We use Michaelis-Menten terms for the killing rates of susceptible and resistant bacteria by antibiotics to capture the saturating effects of increased antibiotic concentration . The maximum killing rates are given by E m a x S ( hr−1 ) ( susceptible ) and E m a x R ( hr−1 ) ( resistant ) , where E m a x S > E m a x R , while the Michaelis constants A 50 S ( μg cm−3 ) ( susceptible ) and A 50 R ( μg cm−3 ) ( resistant ) give the antibiotic concentrations at which the killing rate is half-maximal , where A 50 R > A 50 S ( see Table 3 ) . We multiply the bound bacteria antibiotic killing terms by a factor ω ( dimensionless ) , to account for the potential difference in the antibiotic potency against bound bacteria as compared with free bacteria . Bound bacteria may be less vulnerable to antibiotic than free bacteria , in which case ω < 1; however , they may also be exposed to higher concentrations of antibiotic , which enters the wound through the host cell layer , in which case they may be more vulnerable , such that ω > 1 . If bound bacteria are equally as vulnerable to antibiotic as free bacteria then ω = 1 . The clearance of bacteria and inhibitors ( ψBac ( t ) ( hr−1 ) and ψI ( t ) ( hr−1 ) ) is assumed to occur at a constant rate for the first 24 hours , after which it ceases when a scab forms over the excision . Therefore , clearance occurs at rates ψ B a c ( t ) = ψ ˜ B a c H ( 24 - t ) and ψ I ( t ) = ψ ˜ I H ( 24 - t ) , ( 10 ) where ψ ˜ B a c ( hr−1 ) and ψ ˜ I ( hr−1 ) are the constant clearance rates which apply in the first 24 hours , and H is a Heaviside step function ( see Eq 8 ) . Antibiotics may either be administered in discrete doses or applied continuously , such that the antibiotic concentration remains fixed . In the former case , antibiotic is assumed to be eliminated from the system ( e . g . through degradation and clearance into the bloodstream and surrounding tissues ) at a rate δA ( hr−1 ) following a dosing event . It is assumed that the loss of antibiotic through its interaction with bacteria is negligible in comparison to its elimination rate , and hence we do not include it in the model . Further , we do not include an antibiotic clearance term , similar to those given in Eq 10 for bacteria and inhibitors , since , while some antibiotic will leave the wound within the leaking exudate , this will not affect the antibiotic concentration in the remaining exudate ( which is replenished via passage cross the host cell layer ) . It is assumed that inhibitor degradation , if it occurs , is sufficiently gradual that it can be neglected . Several of the terms in Eqs 1–7 contain the exudate height , h , or volume , V , or the wound area , Ar , as a factor in order to ensure dimensional consistency . We retain them in explicit form in the interests of clarity , though we note they could have been combined with their multipliers to create new parameters . Bacteria are applied to the burn wound , following the excision , at time t = 0 ( hr ) . This is also the first occasion upon which inhibitor or antibiotic treatment may be applied . Therefore , initially BFS ( 0 ) =BFSinit , BFR ( 0 ) =BFRinit , BBS ( 0 ) =0 , BBR ( 0 ) =0 , IF ( 0 ) =IFinit , IB ( 0 ) =0 , A ( 0 ) =Ainit , ( 11 ) where B F S i n i t , B F R i n i t , I F i n i t and Ainit are constants . The bound compartments are empty initially , since bacteria and inhibitors have not had an opportunity to bind to the host cells . See Tables 2–4 for parameter values . We note that we retain equations in dimensional form to ease biological interpretation . Previously we considered a susceptible only bacterial population , treated using inhibitors and debridement [26] . There the focus was upon optimising inhibitor properties to improve treatment . Here we consider how to optimally combine antibiotic , inhibitor and debridement therapies so as to eliminate a mixed susceptible-resistant population of bacteria . Antibiotics are applied systemically and may be administered either in discrete doses ( e . g . administered orally as tablets ) or continuously ( e . g . administered intravenously via a drip ) . In the continuous case the antibiotic concentration is held at a constant value such that A ≡ Ainit . Hellinger et al . [59] have shown that meropenem dosages as high as 6 g day−1 can be used in humans without increasing the frequency of adverse effects , a result which has been confirmed by other groups [60 , 61] . Furthermore , Roberts et al . [60] found that continuous dosing of meropenem at 3 g day−1 in humans resulted in subcutaneous tissue concentrations of 4 μg cm−3 . Therefore , since the daily dosage could be up to twice this value , we can infer ( assuming a linear scaling ) that subcutaneous tissue ( and hence burn wound ) concentrations up to 8 μg cm−3 are achievable . In the discrete dosing case , antibiotic degrades and is cleared from the body following each dosing event . We assume that discrete doses may not exceed tissue concentrations of 8 μg cm−3 , consistent with the continuous case . We take the dosing frequency to be once a day , at the same times at which inhibitors are applied ( see below ) , thus ensuring that our treatment regimens are feasible to implement clinically . Inhibitors are applied topically to the wound . There is no hard limit on how frequently inhibitors may be applied; however , twice daily is a reasonable upper limit ( that is , at 0 , 12 , 24 , 36 , … hr ) , fixing the frequency at daily dosing in the present study for simplicity . We take the dose used by Huebinger et al . [18] in their experiments , that is 3×108 inhibitors ( which , when added to the exudate , corresponds to a concentration of 6 . 12×107 inhib . cm−3 ) , as standard . The total number of inhibitors in the system ( free and bound ) is conserved in the absence of debridement , except during the first 24 hr ( after the necrotic tissue is first excised ) , when free inhibitors are lost through leakage of the exudate . Debridement involves the mechanical cleansing of a wound , for example with a cloth . In our model this corresponds to the instantaneous removal of the exudate and with it all of the free bacteria and inhibitors . The exudate is quickly replenished ( on the timescale of a few minutes ) such that its volume fluctuation can be neglected . Debridement can be administered at most once daily , starting from the first day after the excision is made ( that is at 24 , 48 , 72 , … hr ) . In those cases where debridement and dosing with inhibitors coincides , debridement is performed first , to avoid immediately removing the newly administered inhibitors . Since debridement involves the removal of the scab that forms over the wound , clearance of bacteria and inhibitors is re-established in the first 24 hr after each debridement event . The parameters in Table 3 were fitted to newly gathered in vitro data . Susceptible , PA1004 WT , and resistant , PA1004 Evo10 , strains of P . aeruginosa were grown both in the absence of antibiotic and in the presence of a range of concentrations of meropenem . Simplified equations , containing only logistic growth and antibiotic killing terms were then fitted to the data using the Matlab routine fminsearch , providing fits for c , E m a x S , E m a x R , A 50 S and A 50 R . See S1 Text for further details . The parameters in Table 2 come from [26] where they were fitted to in vivo data from the rat burn wound model described in [18] . Twelve valid parameter sets were identified , which were grouped into four qualitatively distinct cases ( Case A–Case D ) . Treatment with inhibitors is effective in Cases A and B , worsens an infection in Case C and has little effect in Case D . Each parameter set gave an equally good fit to the data , while insufficient experimental data is currently available to distinguish between them . In the present work we use a single parameter set from each case , Set 2 from Case A , Set 6 from Case B , Set 10 from Case C and Set 12 from Case D . Set 2 was chosen as it is the most biologically realistic parameter set in Case A , Sets 6 and 10 were chosen since they are the most resistant to treatment , allowing us to consider the worst-case-scenario , and Set 12 was chosen since it is the only parameter set in Case D . We note that we used Sets 3 and 8 , rather than Sets 6 and 10 , in the main text of [26] in Cases B and C respectively . The combination of the parameters fitted to in vitro data in Table 3 with the parameters fitted to in vivo data in Table 2 is valid , both since the effects and processes with which each set of parameters are associated are independent from each other , and because the in vitro data were gathered using the same bacterial species ( P . aeruginosa ) as the in vivo data and using a growth medium which replicates the nutrient levels in a burn wound exudate ( see S1 Text ) . Each of the parameters in Table 4 were either measured , calculated , derived from the literature or estimated , as indicated in the fourth column . The parameters ϕBac , ϕI , V , Ar , h , n and Etotal are justified in [26] ( where Etotal is written as Einit ) , while I F i n i t and Ainit are justified above in Treatment types . We set the conjugation and segregation rates , λ and ρ , to zero unless otherwise stated . This is because the resistance genes to meropenem in the PA1004 Evo10 strain of P . aeruginosa under consideration are chromosomal and hence cannot be transferred by conjugation or lost through segregation , which requires the resistance gene to be carried on a plasmid . While we have the PA1004 Evo10 strain in mind throughout this study , we have included terms for conjugation and segregation in order to make our model sufficiently general to account for other bacterial strains . In S3 Text we perform a sensitivity analysis to investigate the effect of these parameters on the bacterial dynamics , using values informed by the literature as described below . Hall et al . [50] measured the intraspecific conjugation rates of P . fluorescens and P . putida to be 10−11±0 . 2 cell−1hr−1 and 10−14±0 . 4 cell−1hr−1 respectively ( these values must be multiplied by V = 4 . 9 cm3 to make them dimensionally consistent with our model ) , which fall within the range of values measured by [51–53] , while Simonsen et al . [51] have measured conjugation rates as high as 10−9 cm3 cell−1hr−1 in E . coli . Smets et al . [52] measured a plasmid loss rate of 2 . 52×10−4hr−1 which informed the value of 1×10−4hr−1 used in [50 , 53] and falls within the range of values measured by [54] . The antibiotic elimination rate , δA , has been measured to lie in the range 0 . 62–1 . 72 hr−1 for meropenem in both humans and pigs [55–57] , thus we choose δA = 1 hr−1 as a typical value . The factor difference in antibiotic potency against bound bacteria ω is assumed to be one ( i . e . no difference ) by default and in the absence of further information . We perform a sensitivity analysis on ω in Sensitivity analysis , varying it within the range ω ∈ [0 . 5 , 2] . The initial bacterial burden is taken to be 5×106 CFU ( colony-forming units ) , corresponding to an initial free density ( all bacteria are free initially ) of 1 . 02×106 cells cm−3 , in accordance with the in vivo model in [18] . The initial ratio of susceptible to resistant bacteria may vary; however , susceptible bacteria will be in the majority prior to treatment with antibiotic due to the resistance-associated fitness cost . Therefore , we assume that only 2% ( 2 . 04×104 cells cm−3 ) of the initial bacterial population exhibits the resistant phenotype , the remaining 98% ( 1 . 00×106 cells cm−3 ) being susceptible .
We begin by considering a steady-state analysis of Eqs 1–10 , performed using Maple , to determine the number of steady-states exhibited by the system under various conditions , together with their stability properties . While the system may take a long while to approach steady-state in practice , depending upon the choice of parameters and initial conditions , this analysis is instructive for at least two reasons . Firstly , it allows us to ensure that we are not overlooking any potential stable steady-state solutions in the time-dependent simulations presented below . Secondly , it allows us to make more clear-cut comparisons between different scenarios , looking beyond the transient dynamics resulting from the choice of initial conditions . We consider four scenarios: untreated , antibiotic treatment only , inhibitor treatment only , and treatment with both antibiotics and inhibitors , comparing Cases A–D in each scenario . We use the maximum continuous concentration ( 8 μg cm−3 ) for antibiotic treatment and a single standard dose ( 6 . 12×107 inhib . cm−3 ) for inhibitor treatment ( see Treatment types ) . We set the clearance terms to zero ( ψ ˜ B a c = 0 hr−1 and ψ ˜ I = 0 hr−1 ) since fluid only leaks from the wound in the first 24 hr . Further , we neglect conjugation and segregation ( λ = 0 cm3cell−1hr−1 and ρ = 0 hr−1 ) , and assume that there is no difference in the potency of antibiotics against bound bacteria as compared with free bacteria ( ω = 1 ) ( see Parameter fitting and justification ) . We also remove the Heaviside step functions preventing the logistic growth of bacteria when their population size goes beneath one ( H ( V B F S - 1 ) , H ( V B F R - 1 ) , H ( A r B B S - 1 ) and H ( A r B B R - 1 ) ) since these are only required in dynamic simulations to prevent biologically unrealistic regrowth when bacteria have been eliminated . Lastly , where antibiotics are applied , we assume a constant dose , since with a discrete dose the antibiotic concentration is zero at steady-state , being identical to the equivalent scenario without antibiotic treatment . Following these simplifications , the governing equations reduce to Eqs A–E in S2 Text . All remaining parameter values are as given in Tables 2–4 . The results of the steady-state analysis are described in detail in S2 Text and summarised here , in Table 5 and in Fig 2 . In all cases except Case A under the inhibitor only treatment the system is monostable , the number of steady-states ( stable plus unstable ) varying between one and three depending upon the treatment scenario and the parameter set . In the absence of antibiotics , resistant bacteria go extinct at the stable steady-state , while free and bound susceptible bacteria survive . The situation is reversed in the presence of antibiotics , with susceptible bacteria going extinct at the stable steady-state , while free and bound resistant bacteria survive . There are three exceptions to this rule . The first two are for the scenario in which treatment with both antibiotics and inhibitors is applied , in Cases B and C , for which all bacteria go extinct at the stable steady-state . The third is Case A under the inhibitor only treatment ( noted above ) , for which there exist no isolated stable steady-states . Instead , there exists a region of non-isolated steady-states [64] , in which susceptible and resistant bacteria may coexist , including the extremes ( unstable steady-states ) at which only one of these subtypes survives . As such , the state to which the system settles depends upon the initial conditions . For simplicity of exposition , we plot the unstable steady-state solution in which only susceptible bacteria survive in Fig 2 , this being the typical state in the absence of antibiotics under most parameter sets . Treatment with antibiotics alone reduces the total number of bacteria , BT , in Cases A , C and D; however , it slightly increases the bacterial burden in Case B ( this counter-intuitive result is discussed below in Sensitivity analysis ) . As was the case in [26] , treatment with inhibitors reduces the total number of bacteria in Cases A and B , increases the bacterial burden in Case C and has little effect in Case D ( see Sensitivity analysis and [26] for discussion ) . Treatment with antibiotics and inhibitors in combination is more effective than treatment with either therapy in isolation , eliminating the bacterial burden in Cases B and C and greatly reducing it in Cases A and D . It is evident from these results that antibiotics and inhibitors are predicted to work together in a synergistic manner , as opposed to an additive one , their combined effect reducing the total bacterial burden by a greater quantity in Cases B–D than the sum of the reductions when applied in isolation , and by a smaller quantity in Case A . Having examined the behaviour of the system at steady-state , we consider the bacterial population dynamics over time in response to treatment . We use the Matlab routine ode15s , a variable-step , variable-order solver based upon numerical differentiation formulas , to solve the time-dependent problem ( Eqs 1–11 ) both here and throughout the paper . The untreated scenario is compared with four treatment scenarios: regular antibiotic and inhibitor dosing with and without regular debridement , and constant antibiotic concentration with regular inhibitor dosing , with and without regular debridement ( see Fig 3 ) , for Cases A–D . We note that while only the total number of bacteria , BT , is plotted for clarity , the simulations include susceptible/resistant and free/bound bacteria . Regular antibiotic/inhibitor/debridement treatments are performed every 24 hr , antibiotic/inhibitor dosing occurring for the first time at t = 0 hr and debridement being performed for the first time at t = 24 hr . Antibiotic doses of 8 μg cm−3 and standard inhibitor doses of 6 . 12×107 inhib . cm−3 are used in all cases , while each debridement event results in the removal of all free bacteria and inhibitors . The antibiotic concentration is held fixed at A = 8 μg cm−3 in the constant antibiotic scenarios ( see Treatment types for more details ) . Constant antibiotic concentration with regular inhibitor dosing and debridement is the most effective treatment , eliminating the bacterial burden in all cases and doing so more rapidly than the other strategies . Constant antibiotic concentration with regular inhibitor dosing and no debridement eliminates the bacterial population in Cases B and C , but has a more modest effect in Cases A and D . Regular antibiotic and inhibitor dosing with debridement eliminates all bacteria in Cases A and B , but is ineffective in Cases C and D . Lastly , regular antibiotic and inhibitor dosing without debridement is the least effective strategy , having little effect in Cases A–D . We note that while constant antibiotic concentration with regular inhibitor dosing and debridement reduces the number of antibiotic resistant bacteria , B R = V B F R + A r B B R , in all cases ( see inset graphs , Fig 3 ) , the remaining strategies increase BR above untreated levels in some cases; indeed , regular antibiotic and inhibitor dosing without debridement does so in all cases . A comparison between regular inhibitor and debridement treatment , which is the most effective therapy in the absence of antibiotics ( see [26] ) , and treatment which combines regular inhibitor and debridement therapy with a constant antibiotic dose , shows that the combined therapy is significantly more effective ( see Fig . A in S3 Text ) . Inhibitors and debridement alone eliminate the bacterial burden in Case A only , whereas , when combined with antibiotics , the bacterial burden is eliminated in all four cases ( A–D ) . In the results that follow we consider the effect of varying the antibiotic and inhibitor doses , and other key parameters , upon the bacterial population dynamics and their steady-state values . In the cases where time-dependent simulations are employed , the solutions are shown at 4 weeks ( 672 hr ) , with solutions at 1 week ( 168 hr ) and 1 year ( 365 days = 8760 hr ) provided in S3 Text . Results are given at 1 week since ideally we would like to clear an infection within this time , while results are shown at 4 weeks and 1 year to demonstrate the dynamics of more persistent infections and since the sensitivity of the system to changes in parameter values varies over time . Informed by the preceding sensitivity analyses , we used our mathematical model to predict the optimum treatment regimens in Cases A–D under certain constraints . Two sets of initial conditions were considered , each consisting of a mixture of susceptible and antibiotic resistant bacteria . The first set is the standard initial conditions given in Table 4 , which corresponds to a new infection in which bacteria have not yet had time to bind to host cells . The second set corresponds to an established infection . Here we chose the initial conditions to be the untreated steady-states for each parameter set ( Cases A–D ) , in which all surviving bacteria are susceptible , modified so that 2% of the free and bound bacteria are resistant . We chose to optimise the treatment over the period of a week—this being a standard period over which to treat a bacterial infection and also reducing the number of regimens over which to search compared with longer periods—exploring combination therapies including antibiotics , inhibitors and debridement . We assume continuous dosing with antibiotics , fixing the concentration at its maximum value of A = 8 μg ml−1 since this was found to have the greatest effect against bound bacteria ( see Fig 4 ) , while free bacteria can be removed using debridement . Debridement may be applied at the beginning of days 2–7 , but not at the start of the first day ( see Treatment types ) , giving 26 = 64 possible treatment regimens . Lastly , we assume that inhibitors may be applied in multiples of the standard dose ( 6 . 12×107 inhib . cm−3 ) , using exactly seven standard doses worth of inhibitors over the week ( 4 . 28×108 inhib . cm−3 ) , and that inhibitors may only be applied at t = 0 hr and immediately following a debridement event . This brings the total number of possible treatment regimens to 14 , 407 . We note that in preliminary work we used a genetic algorithm approach to investigate optimum solutions; however , there is no guarantee of identifying the global optimum via this method . Rather , by accounting for the clinical constraints on the treatment regimen ( as described above ) , we sample the complete space of possible treatment regimens , enabling us to identify the global optimum , subject to these constraints . While the clinical constraints imposed on our optimisation limit the options to a discrete set of points in decision space , we note that the theoretical range of treatment options lies on a continuum ( e . g . the timing and concentration of inhibitor doses ) . Four separate optimisations were performed for each of Cases A–D and for each set of initial conditions , each using a different objective function which we sought to minimise . The first objective function gives the final number of bound bacteria , BB ( 168 ) , the second gives the final total number of bacteria , BT ( 168 ) , the third gives the mean number of bound bacteria over the week , 〈BB〉 , while the fourth gives the mean total number of bacteria over the week , 〈BT〉 . We seek to optimise for each objective function individually , rather than performing a multi-objective optimisation , since we wish to find the regimens which fully-optimise each criteria and to compare between these . A unique optimal regimen can always be found under the first two optimality criteria . In those cases where multiple treatment regimens are equally optimal under the third and fourth criteria , we designate that regimen which gives the lowest final value of BB ( for the 〈BB〉 criterion ) or BT ( for the 〈BT〉 criterion ) as being optimal . We performed separate optimisations for the bound bacterial burden since it is bound bacteria , rather than free bacteria , that actively damage host tissue . Therefore , it may be more important to remove bound bacteria than free bacteria . Further , we performed separate optimisations for the final and mean number of bacteria since we aim both to eliminate the bacterial burden as rapidly as possible ( final ) , while also keeping the bacterial burden low during treatment ( mean ) . In each case we search through the full set of 14 , 407 possible treatment regimens . We note that , unlike in the steady-state and sensitivity analyses above ( with the exception of Fig 7 ) , clearance of free bacteria and free inhibitors is included in these simulations , occurring both in the first 24 hr and in the first 24 hr after each debridement event . Fig 9 shows the optimum treatment regimens for the BB ( 168 ) and BT ( 168 ) objective functions ( columns ) and for each parameter set ( rows ) in the new infection scenario . It is predicted to be optimal to apply all of the inhibitors at the start of the first day under both optimality conditions in Cases A–C and to distribute inhibitors more evenly across the week in Case D . Further , it is predicted to be optimal to debride every day ( days 2–7 ) under both optimality conditions in Cases A and D , and to debride only on some of the later days in Cases B and C . The results under the 〈BB〉 and 〈BT〉 criteria are similar ( see Fig 10 ) . Fig 11 shows the dynamics of BB and BT under the optimal BB ( 168 ) and BT ( 168 ) treatment regimens in the new infection scenario . The total bacterial burden is eliminated by the end of the week in Case A , is reduced to O ( 10 ) in Case B , to O ( 102 ) in Case C and to just below 103 in Case D ( where BT ( 168 ) = O ( 107 ) to O ( 108 ) in the untreated scenario in Cases A–D ) . The results under the 〈BB〉 and 〈BT〉 criteria are presented in Fig 12 . The difference in the bacterial dynamics between the different optimisation regimens is minor . Both here and in Figs 15 and 16 we plot just the total number of bacteria and the number of bound bacteria for clarity . In both cases the majority of bacteria are susceptible for approximately the first 2 days , after which antibiotic resistant bacteria dominate . Fig 13 shows the optimum treatment regimens for the BB ( 168 ) and BT ( 168 ) objective functions ( columns ) and for each parameter set ( rows ) in the case of an established infection . It is predicted to be optimal to apply all or most of the inhibitors at the start of the first day under both optimality conditions in Cases B–D and to distribute inhibitors more evenly across the week in Case A . In this respect the optimal debridement is similar to the new infection scenario for Cases B and C , differing in Case A , where all inhibitors were used on the first day , and Case D , where inhibitor doses were distributed throughout the week . The predicted optimal debridement patterns differ markedly from the new infection scenario , with debridement being less frequent in all cases and entirely absent in Cases C and D under the BB ( 168 ) criterion . The results under the 〈BB〉 and 〈BT〉 criteria differ from all of those discussed above ( see Fig 14 ) . Here , it is predicted to be optimal to use all inhibitors at the start of day 1 for Cases A–D under the 〈BB〉 criterion and in Case B under the 〈BT〉 criterion , while it is better to distribute inhibitors across multiple days in Cases A , C and D under the 〈BT〉 criterion . Further it is predicted to be optimal to debride every day in Cases A–D under the 〈BT〉 criterion , not at all in Cases A and D under the 〈BB〉 criterion and only on some later days in Cases B and C under the 〈BB〉 criterion . Fig 15 shows the dynamics of BB and BT under the optimal BB ( 168 ) and BT ( 168 ) treatment regimens in the case of an established infection . Here the efficacy is more modest in comparison with the new infection scenario , as would be expected . The total bacterial burden is eliminated by the end of the week in Case A , is reduced to O ( 102 ) in Case B , to O ( 104 ) –O ( 105 ) in Case C and to O ( 105 ) –O ( 106 ) in Case D ( where BT ( 168 ) = O ( 107 ) to O ( 108 ) in the untreated scenario in Cases A–D ) , lower values corresponding to the BT ( 168 ) optimality condition and higher values to the BB ( 168 ) optimality condition where ranges are given . The results under the 〈BB〉 and 〈BT〉 criteria are presented in Fig 16 . The difference in the bacterial dynamics between the different optimisation regimens is minor , except under the 〈BB〉 optimum regimen in Case A , for which the total bacterial burden is not eliminated since debridement is not employed .
The rise in antimicrobial resistance ( AMR ) poses a real and increasing challenge in treating microbial infections . Anti-adhesion therapy provides one way of meeting this challenge , preventing bacteria from binding to the cells of an infected host , thus rendering them more susceptible to physical clearance e . g . through debridement , and less harmful to the host . In this paper we have used mathematical modelling to elucidate and predict the effects of therapies combining traditional treatments , namely antibiotics and debridement , with anti-adhesion therapy to treat antimicrobial resistant infections . We consider the particular context of a burn wound , infected by a mixture of antibiotic-resistant and antibiotic-susceptible strains of P . aeruginosa , using the bactericidal antibiotic meropenem; fitting the antibiotic-associated parameters in our ODE model to in vitro data , collected as part of this study . While the parameters used in the model are specific to P . aeruginosa and meropenem in a rat burn wound , the model structure can also be used to model burn wound infections in other species ( e . g . in humans ) , with other bactericidal antibiotics and with any bacterial species that uses MAM7 to enable it to bind to host cells ( e . g . Vibrio parahaemolyticus , Yersinia pseudotuberculosis and Vibrio cholerae [16] ) . We begin by providing a brief summary of our key results , before discussing them in more detail below: In each of the results presented we considered four parameter sets , denoted as Cases A–D , each of which provides a good fit to the experimental data , but qualitatively different behaviour beyond the time frame of the experimental results ( see Parameter fitting and justification and [26] for details ) . Steady-state analysis demonstrated that the system is monostable for all parameter sets considered , with one exception ( see Steady-state analysis ) . In the absence of antibiotic , susceptible bacteria survive while antibiotic resistant bacteria go extinct due to the fitness cost associated with resistance . However , in the presence of ( sufficient quantities of ) antibiotic , resistant bacteria survive and susceptible bacteria go extinct , since the asymmetric killing rate of susceptible and resistant bacteria by antibiotic outweighs the fitness cost experienced by resistant bacteria . Treatment with antibiotics and inhibitors in combination is more effective than treatment with either therapy in isolation , eliminating the bacterial population in Cases B and C , and significantly reducing it in Cases A and D . Interestingly , the combined effect is synergistic , as opposed to additive , effecting a greater reduction in the bacterial burden than the sum of the reductions achieved through either therapy in isolation in Cases B–D and a lesser reduction in Case A . Indeed , the elimination of bacteria in Cases B and C is surprising given that antibiotics alone increase the total bacterial burden , BT , in Case B , while inhibitors alone significantly increase BT in Case C . While these results are encouraging , it is important to note that it can take on the order of days to months for the system to approach steady-state . Therefore , it is important to consider the dynamic behaviour of the system . Simulations of the full time-dependent problem revealed that a constant antibiotic concentration is more effective , often significantly so , than regular dosing at the same concentration . This is to be expected , in part , since the antibiotic killing rate is maintained at a high level in the constant concentration scenario , whereas it drops off as antibiotic is eliminated from the body in the regular dosing scenario . However , the difference in efficacy is more significant that might be expected . Combination therapy , combining a constant antibiotic concentration with regular inhibitor dosing and debridement , was the most effective treatment strategy considered , eliminating the bacterial population in Cases A–D in times ranging between 1–30 days . While all bacteria , including the resistant subpopulation , were eliminated in this latter therapy , other strategies were found to increase the number of resistant bacteria , compared with the untreated scenario , in some cases . This highlights the fact that the choice of treatment regimen can have a significant effect on the spread of AMR within a host . Steady-state sensitivity analyses for antibiotic and inhibitor doses ( A and I F i n i t respectively ) applied in isolation show that these treatments can both decrease and ( surprisingly ) increase BT , depending upon the dosage used and upon the parameter set under consideration . The increase in BT is caused either by an increase in the logistic growth rate of bound bacteria ( Fig 4 Cases B–D and Fig 5 Case B ) or by a decrease in the per-bacteria binding rate of free bacteria to host cells ( Fig 5 Case C , see Sensitivity analysis for a detailed discussion ) . Each treatment is effective in reducing the total bacterial burden when used in isolation , provided the dosage is sufficiently large; however , our model predicts that the antibiotic dose would have to be made infeasibly large in Case B to be effective in isolation ( A > 10 μg cm−3 ) and similarly for the inhibitor dose in Cases C and D ( hundreds to thousands of times the standard dose ) . Further experimental studies are required to test these predictions to determine under what circumstances they hold . Sensitivity analysis for antibiotic and inhibitor combination therapy without debridement predicts that the bacterial burden can be eliminated within four weeks in Case C using realistic doses and significantly reduced in Case B , whereas BT can be reduced by at most an order of magnitude for realistic doses in Cases A and D , which are less sensitive to treatment . We further predicted that treatment efficacy can be enhanced by including debridement , eliminating bacteria in Cases A–C using realistic levels of antibiotic and inhibitor , and clearing an infection more rapidly . Importantly , our model predicts that the use of inhibitors significantly reduces the antibiotic dose required to clear an infection , both in terms of the maximum antibiotic concentration required and also in terms of the total quantity of antibiotic administered over the course of an infection , given that combination therapy may clear an infection more quickly . We speculate that this could also reduce the chances of bacteria developing resistance to antibiotic therapies . The model is insensitive to the rates of conjugation and segregation ( λ and ρ respectively ) within realistic ranges; hence , it is reasonable to neglect these processes from the model . By contrast , the system is sensitive to the factor difference in antibiotic potency against bound bacteria compared with free bacteria , ω , an increase in this parameter effecting a decrease in BT . We have assumed that ω = 1 in the present work; however , it would be valuable to measure this parameter experientially for different bacterial species , antibiotics and infection sites to determine its true value in a variety of contexts , and thus to incorporate this into future models . Optimal treatment regimens , combining antibiotics and inhibitors with debridement over the period of a week , were predicted for Cases A–D . For each case , two scenarios were considered: the first , corresponding to a new infection , in which bacteria have not yet had an opportunity to bind to host cells and the second , corresponding to an established infection , including both bound and free bacteria . Both scenarios consisted of mixed populations of susceptible and resistant bacteria . The inhibitor dosing and debridement regimens were allowed to vary , while the antibiotic concentration was assumed to take its maximum value based upon the preceding sensitivity analyses . Four separate optimisations were performed for each parameter set and scenario using different optimality criteria: minimising the final bound bacterial burden , BB ( 168 ) , minimising the final total bacterial burden , BT ( 168 ) , minimising the mean bound bacterial burden , 〈BB〉 , and minimising the mean total bacterial burden , 〈BT〉 . We focused upon bound bacteria in particular , since it is bound bacteria , rather than free bacteria , that damage host cells . In the new infection scenario it is almost always optimal to use the full weekly quota of inhibitors at the beginning of the first day in Cases A–C and to distribute inhibitor dosing more evenly across the week in Case D , while it is best to debride every day to minimise 〈BB〉 and 〈BT〉 in most cases , the optimal debridement regimen varying between parameter sets under the BB ( 168 ) and BT ( 168 ) criteria . Given that the bound and total bacterial burdens evolve similarly in Cases A–D under each of the optimal regimens , we suggest that , in the new infection scenario , it would be best to use the full inhibitor quota at the beginning of the first day and to debride every day in a clinical setting . The optimal treatment regimens are predicted to eliminate the bacterial burden within a week in Case A and to significantly reduce the bacterial burden in Cases B–D . Further experimental studies are required to test these predictions . In the established infection scenario optimal treatment regimens vary greatly between parameter sets and optimality criteria . Since it is most important that we eliminate the bound bacterial burden , we suggest that a regimen which minimises BB ( 168 ) or 〈BB〉 would be best . Under these criteria it is almost always optimal to use the full inhibitor quota at the beginning of the first day of treatment and to delay the first debridement event for as long as possible to allow inhibitors time to outcompete bacteria for binding sites before debridement removes their free contingent . If daily debridement is required then the optimal strategy will depend upon the parameter set . In Cases A and B , the ratio of bacterial binding to unbinding rates , αBac/βbac , is lower than the ratio of inhibitor binding to unbinding rates , αI/βI . Therefore , inhibitors quickly outcompete bacteria for binding sites , such that using the full inhibitor dose at the start of the first day would be a good strategy . In Cases C and D , αBac/βbac > αI/βI . Therefore , it takes inhibitors longer to displace bacteria , such that distributing inhibitor doses evenly across the week would be a good strategy . In future work we will develop our mathematical modelling in a number of new directions . This will include the development of partial differential equation models to account for the spatial distribution of bacteria , antibiotics , inhibitors and binding sites ( ODE models being incapable of adequately accounting for non-uniform distributions or diffusive/migratory processes ) , allowing us to investigate issues such as how a localised application of inhibitors would affect treatment efficacy; the development of stochastic and cellular automata models to account for the random behaviour of the system at a more finely-resolved spatial scale; and a more detailed stability analysis of ODE systems involving treatment with inhibitors and antibiotics . Future models could also consider the use of bacteriostatic antibiotics , quorum sensing and biofilm formation . Possible future experimental studies are noted in the discussion above . In conclusion , our model predicts that antibiotics and inhibitors have a synergistic effect when used together , that combination therapy is more effective than either treatment in isolation and that treatment may be further enhanced through the use of debridement . Further , our model predicts that , in general , when treating over the period of a week , the optimal strategy is to maintain a constant antibiotic dose at the maximum allowable concentration , to use the full quota of inhibitors at the beginning of the first day of treatment and to debride daily , though this could be further enhanced if a patient-specific parameter set is identified . Lastly , our models predict that using inhibitors lowers the minimum antibiotic dose required in order to eliminate a bacterial infection , reducing the selection pressure and , potentially , the probability that bacteria will develop resistance to the antibiotic .
|
Since the development of the first antibiotics , bacteria have utilised and developed resistance mechanisms , helping them to avoid being eliminated and to survive within a host . Traditionally , the solution to this problem has been to treat with multiple antibiotics , switching to a new type when the one currently in use proves ineffective . However , the development of antibiotics has slowed significantly in the past two decades , while multi-drug resistant strains , otherwise known as ‘super bugs’ , are on the rise . In answer to this challenge , alternative approaches , such as anti-adhesion therapy , are being developed as a complement or alternative to traditional antimicrobials . In this paper we formulate and analyse a mathematical model of a combination therapy , applied in the context of an infected burn wound , bringing together antibiotics , anti-adhesion therapy and debridement ( the physical cleaning of a wound ) . We use our models to make sense of how these treatments interact to combat a bacterial infection , to predict treatment outcomes for a range of strategies and to suggest optimal treatment regimens . It is hoped that this study will guide future experimental and clinical research , helping biomedical researchers to identify the most promising approaches to treatment .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"cell",
"binding",
"antimicrobials",
"cell",
"physiology",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"dose",
"prediction",
"methods",
"pathogens",
"drugs",
"microbiology",
"pseudomonas",
"aeruginosa",
"antibiotic",
"resistance",
"pharmaceutics",
"antibiotics",
"pharmacology",
"bacteria",
"adjustment",
"of",
"dosage",
"at",
"steady",
"state",
"bacterial",
"pathogens",
"pseudomonas",
"antimicrobial",
"resistance",
"medical",
"microbiology",
"microbial",
"pathogens",
"host",
"cells",
"cell",
"biology",
"virology",
"microbial",
"control",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2019
|
Mathematical model predicts anti-adhesion–antibiotic–debridement combination therapies can clear an antibiotic resistant infection
|
During viral infection , a massive demand for viral glycoproteins can overwhelm the capacity of the protein folding and quality control machinery , leading to an accumulation of unfolded proteins in the endoplasmic reticulum ( ER ) . To restore ER homeostasis , cells initiate the unfolded protein response ( UPR ) by activating three ER-to-nucleus signaling pathways , of which the inositol-requiring enzyme 1 ( IRE1 ) -dependent pathway is the most conserved . To reduce ER stress , the UPR decreases protein synthesis , increases degradation of unfolded proteins , and upregulates chaperone expression to enhance protein folding . Cytomegaloviruses , as other viral pathogens , modulate the UPR to their own advantage . However , the molecular mechanisms and the viral proteins responsible for UPR modulation remained to be identified . In this study , we investigated the modulation of IRE1 signaling by murine cytomegalovirus ( MCMV ) and found that IRE1-mediated mRNA splicing and expression of the X-box binding protein 1 ( XBP1 ) is repressed in infected cells . By affinity purification , we identified the viral M50 protein as an IRE1-interacting protein . M50 expression in transfected or MCMV-infected cells induced a substantial downregulation of IRE1 protein levels . The N-terminal conserved region of M50 was found to be required for interaction with and downregulation of IRE1 . Moreover , UL50 , the human cytomegalovirus ( HCMV ) homolog of M50 , affected IRE1 in the same way . Thus we concluded that IRE1 downregulation represents a previously undescribed viral strategy to curb the UPR .
During viral replication large amounts of viral proteins must be synthesized , folded , and posttranslationally modified . Folding , maturation and multi-subunit assembly of secreted and transmembrane proteins take place in the endoplasmic reticulum ( ER ) and require an elaborate system of chaperones , lectins , and carbohydrate-processing enzymes . Whereas correctly folded proteins are transported to the Golgi , misfolded or unfolded proteins are arrested in the ER and diverted for degradation via the ER-associated protein degradation ( ERAD ) pathway [1] . However , the high levels of viral envelope glycoproteins that are being synthesized particularly during the late phase of the viral life cycle can overwhelm the folding and processing capacity of the ER and cause accumulation of unfolded and misfolded proteins in the ER [2] . In addition , large quantities of secreted and immunomodulatory viral proteins can contribute to ER stress [3] . To reduce protein load and restore ER homeostasis , eukaryotic cells activate various ER-to-nucleus signaling pathways , which are collectively referred to as Unfolded Protein Response ( UPR ) [4] , [5] . The UPR is initiated by three sensor proteins that recognize ER stress: protein kinase R-like ER kinase ( PERK ) , activating transcription factor 6 ( ATF6 ) , and inositol-requiring enzyme 1 ( IRE1 ) . The ER chaperone BiP ( immunoglobulin heavy chain binding protein ) , also known as glucose-regulated protein 78 , is thought to bind these sensors and keep them inactive under normal conditions . However , when unfolded and misfolded proteins accumulate in the ER , BiP dissociates from these sensors to perform its chaperone function . As a consequence , the sensors are activated and initiate UPR signaling . Activation of PERK leads to phosphorylation of the α subunit of eukaryotic translation initiation factor 2 ( eIF2α ) , resulting in global attenuation of protein translation [6] , [7] . However , if ER stress persists eIF2α initiates expression of activating transcription factor 4 ( ATF4 ) , which induces expression of the proapoptotic transcription factor C/EBP-homologous protein ( CHOP , also known as growth arrest and DNA damage-inducible protein 153 ) . CHOP expression promotes apoptosis by downregulating the antiapoptotic protein Bcl-2 [8] , [9] . Activated ATF6 translocates to the Golgi where it is cleaved by site 1 and site 2 proteases [10] . The active transcription factor is imported into the nucleus where it induces transcription of chaperone genes [11] . The IRE1 pathway is the most conserved branch of the UPR [12] . Mammalian cells encode two IRE1 isoforms , IRE1α and IRE1β . IRE1α , the most abundant isoform , is expressed in most cells and tissues and is hereafter referred to as IRE1 . By contrast , IRE1β ( also known as IRE2 ) is expressed to significant levels only in intestinal epithelial cells [13] . Upon activation , IRE1 dimerizes and transphosphorylates itself . This leads to activation of a site-specific endoribonuclease activity in the cytosolic tail of IRE1 , which mediates an unconventional splicing of the X-box binding protein 1 ( XBP1 ) mRNA in the cytosol [14] , [15] . The transcription factor XBP1s , which is translated from the spliced Xbp1 transcript , translocates to the nucleus and induces expression of ERAD enzymes [1] , [12] . If ER stress is too severe to overcome and ER homeostasis cannot be restored , IRE1 can also activate c-Jun N-terminal kinase ( JNK ) to commit damaged cells to apoptosis [16] . Increasing evidence indicates that viruses selectively modulate the UPR to take advantage of the beneficial effects and inhibit those detrimental to viral replication [2] . For instance , hepatitis C virus and other members of the Flaviviridae activate beneficial components of the UPR such as BiP in certain cell types to facilitate their replication but trigger ER stress-induced apoptosis in other cells [17]–[20] . Members of the Herpesviridae also modulate the UPR to their own advantage . The molecular mechanisms , however , appear to differ from one virus to another [21] . For example the viral glycoprotein gB of herpes simplex virus type 1 ( HSV-1 ) inhibits PERK activation [22] . By contrast , varicella-zoster virus , another alphaherpesvirus , activates the PERK and IRE1 pathways [23] . UPR modulation also takes place in gammaherpesvirus-infected cells . Epstein-Barr virus ( EBV ) latent membrane protein 1 activates PERK to enhance its own expression [24] . In addition , reactivation of EBV from latent infection is induced by extrinsic ER stress while XBP1 induces EBV lytic gene expression [25] . From these and other examples it has been concluded that UPR regulation plays an important role in viral infection and pathogenesis [2] . Several studies have investigated the ability of human cytomegalovirus ( HCMV ) , a betaherpesvirus , to cope with ER stress and manipulate the UPR to its own benefit . HCMV is a major hazard for immunocompromised individuals such as transplant recipients and the leading infectious cause of birth defects [26] . To enhance viral replication HCMV has adopted several strategies to modulate the UPR . For example , HCMV induces PERK activation , but limits eIF2α phosphorylation . By doing this the virus prevents a global protein synthesis shutoff but allows eIF2α phosphorylation-dependent activation of transcription factor ATF4 [27] . HCMV also uses PERK to induce lipogenesis by activating the cleavage of sterol regulatory element binding protein 1 [28] . In addition , HCMV increases expression of the ER chaperone BiP to facilitate protein folding and virion assembly [29] , [30] . Moreover , the viral UL38 protein was shown to prevent ER stress-induced JNK activation and apoptosis [31] . A recent study has revealed that murine cytomegalovirus ( MCMV ) , a related betaherpesvirus , influences the UPR in a similar manner [32] . Particularly , MCMV was shown to activate the PERK–ATF4 pathway and upregulate expression of the ER chaperone BiP . However , in most cases the exact mechanisms by which human and murine cytomegaloviruses modulate the UPR remain undefined . In the present study , we investigated the influence of MCMV infection on the IRE1 pathway . This pathway has been characterized in yeast as well as in mammalian cells and represents the most evolutionary conserved branch of the UPR [5] . IRE1 mediates an unconventional splicing of Xbp1 , which in turn triggers expression of ERAD proteins [33] . We discovered an interaction between IRE1 and the viral protein M50 . The viral M50 was previously characterized as a type II transmembrane ( TM ) protein that associates with the viral M53 protein . M50 and M53 are essential components of a complex that dissolves the nuclear lamina [34] . Proteins homologous to M50 are found in all herpesviruses studied thus far , and these proteins are involved in nuclear egress of viral capsids . Moreover , M50 and its homologs are essential for lytic replication of beta- and gammaherpesviruses [35] , [36] . We show that M50 expression induces a robust downmodulation of IRE1 levels in transfection and infection experiments suggesting that M50 induces IRE1 degradation . The N-terminal conserved region of M50 proved to be required for IRE1 binding and degradation . We further showed that UL50 , the HCMV homolog of M50 , has a similar function . We propose that inhibition of IRE1 signaling by removal of the sensor IRE1 represents a previously unrecognized viral strategy to curb the UPR .
As it has been shown that cytomegaloviruses inhibit the IRE1-dependent UPR pathway by an unknown mechanism [27] , [32] , we wanted to investigate how MCMV modulates this pathway . First , we measured Xbp1 mRNA splicing during MCMV infection by semiquantitative RT-PCR and real-time RT-PCR . Since the 26 nt intron , which is spliced out by IRE1 , contains a PstI restriction site [14] , [15] , only the unspliced RT-PCR product is cleaved by PstI . MCMV infection of NIH-3T3 fibroblasts induced a slight and transient increase in Xbp1 splicing similar to the one induced by treatment with a very low dose of tunicamycin ( Tun ) , an established ER stress inducer ( Fig . 1A and B ) . The ratio of spliced to unspliced transcripts returned almost to baseline levels around 8 hours postinfection ( hpi ) and remained constant until 48 hpi . To test whether MCMV actively suppresses Xbp1 splicing , we treated MCMV-infected fibroblasts with Tun and measured Xbp1 splicing . As shown in Figure 1C and D , Tun-induced Xbp1 splicing was strongly reduced at 24 hpi and almost completely blocked at 48 hpi . A similar inhibition of Xbp1 splicing was observed when infected cells were treated with the ER stress inducer thapsigargin ( Fig . 1E ) . We also determined the protein levels of transcription factor XBP1s by immunoblot analysis . Consistent with the RT-PCR results , Tun-induced XBP1s protein expression was inhibited at 24 and almost completely blocked at 48 hpi ( Fig . 1F ) . Moreover , ER stress-induced transcription of the XBP1s target gene ERdj4 , which encodes an ERAD protein [37] , was also inhibited ( Fig . 1G ) , further confirming the conclusion that MCMV actively inhibits the IRE1 pathway at late times postinfection . As activated IRE1 is the only enzyme mediating Xbp1 mRNA splicing , we hypothesized that MCMV might express a protein that interacts with IRE1 . To identify IRE1 interaction partners , we stably transfected NIH-3T3 cells with a plasmid encoding IRE1 with a C-terminal tobacco etch virus ( TEV ) protease cleavage site and an HA epitope tag . IRE1-TEV-HA-expressing cells were infected with MCMV , and protein lysates were loaded onto an anti-HA affinity matrix . After washing , IRE1 was released from the matrix by TEV protease digestion . Eluted proteins were separated by gel electrophoresis and silver stained ( Fig . 2A ) . Bands not present in the control lane ( uninfected cells ) were excised and analyzed by protein mass spectrometry . In an approx . 32 kDa band two MCMV proteins were identified: M50 and M85 . M50 is a type II transmembrane ( TM ) protein with a C-terminal TM anchor . It is found in the ER membrane and the nuclear envelope and is known to play a crucial role in nuclear egress of viral capsids [34] , [38]–[40] . M85 is the MCMV minor capsid protein [41] and is not known to be associated with ER membranes . To confirm or dismiss the two MCMV proteins as specific interaction partners of IRE1 , HEK 293 cells were transfected with expression plasmids encoding HA-tagged IRE1 and Flag-tagged MCMV proteins . Flag-tagged M50 coprecipitated with IRE1-HA , and IRE1-HA coprecipitated with M50-Flag ( Fig . 2B and C ) . IRE1 did not interact with Flag-tagged m144 or Calnexin ( CNX ) , two ER-localized control proteins . IRE1 also did not interact with Flag-tagged M85 in co-immunoprecipitation experiments ( data not shown ) . Therefore , M85 was not further investigated as a modulator of the IRE1 signaling pathway . Next we tested whether IRE1 interacts with M50 during MCMV infection . As endogenous IRE1 is expressed at low levels and is difficult to analyze , we used cells expressing epitope-tagged IRE1 from a retroviral vector – a procedure used in several previous studies [42] , [43] . 10 . 1 fibroblasts stably expressing myc-tagged IRE1 were infected with MCMV mutants expressing HA-tagged M50 ( MCMV-M50HA ) or HA-tagged m41 ( MCMV-HAm41 ) . Cell lysates were harvested 17 and 31 hpi and subjected to immunoprecipitation and immunoblot analyses . Fig . 2D shows that M50 coprecipitated with IRE1 , consistent with the affinity purification experiment ( Fig . 2A ) , but m41 , an unrelated MCMV type 2 TM protein [44] , did not . Likewise , HA-tagged M50 , but not m41 , interacted with endogenous IRE1 in MCMV-infected NIH-3T3 cells ( Fig . 2E ) . Next we analyzed the subcellular localization of IRE1 and M50 by immunofluorescence ( IF ) . To do this , we cotransfected cells with expression plasmids for IRE1 and M50 or UL56 , an unrelated type 2 TM protein of Herpes Simplex Virus type 1 [45] . As shown in Figure 3A , IRE1 and M50 colocalized in transfected NIH-3T3 cells , but IRE1 and UL56 did not . We also tested whether IRE1 and M50 colocalize in MCMV-infected fibroblasts . As M50 is a late protein , infected cells had to be fixed and stained at late time points , but not too late in order to avoid cell rounding and detachment as a result of the MCMV-induced cytopathic effect . Moreover , M50 is known to change its localization during MCMV infection: it first localizes to the ER , but is subsequently redistributed to the nuclear rim as a consequence of its interaction with the nuclear MCMV protein M53 [34] , [40] . When we infected 10 . 1 fibroblasts expressing myc-tagged IRE1 with MCMV-M50HA , we saw that a substantial portion of M50 retained a cytoplasmic distribution despite an obvious accumulation at the nuclear rim , and this portion of M50 colocalized with IRE1 ( Fig . 3B ) . We also noticed that IRE1 levels appeared to be reduced at late times in MCMV-infected cells compared to neighboring uninfected cells ( Fig . 3B , 20 hpi ) . To rule out the possibility that the detection of HA-tagged M50 in the cytoplasm resulted from an unspecific binding of the anti-HA antibody to MCMV-infected cells , fibroblasts were infected with wt MCMV or MCMV-M50HA and subjected to IF analysis . Using the same anti-HA antibody and the same staining conditions , a cytoplasmic staining was detected only in MCMV-M50HA- , but not in wt MCMV-infected cells ( Fig . 3C ) . Next we investigated whether M50 inhibits IRE1 phosphorylation , which is required for activation of its endoribonuclease activity . To do this , we transfected NIH-3T3 cells with an IRE1 expression plasmid and cotransfected increasing amounts of an M50 or an m144 expression plasmid . Overexpression of IRE1 is known to cause its activation by autotransphosphorylation [46] . As shown in figure 4A , the levels of phosphorylated IRE1 decreased with increasing M50 expression but not with increasing expression of the m144 control protein . Moreover , total IRE1 levels were also decreased , indicating that M50 reduces IRE1 levels rather than just inhibiting its activation . Nevertheless , IRE1 phosphorylation might be required for its downregulation . To test this hypothesis , NIH-3T3 cells were cotransfected with expression plasmids for M50 and either wildtype ( wt ) IRE1 or a kinase-inactive mutant ( K599A ) . As expected , overexpressed IRE1 K599A was not phosphorylated . However , it was downregulated by M50 just like wt IRE1 ( Fig . 4B ) . Thus we concluded that IRE1 downregulation is independent of its phosphorylation . As M50 also interacts with the viral M53 protein at the nuclear envelope , we tested whether M53 expression affects the M50-induced IRE1 downregulation . NIH-3T3 cells were cotransfected with plasmids encoding IRE1 , M50 , and M53 . As shown in figure S1 , M50 expression induced IRE1 downregulation also in the presence of M53 . However , downregulation was reduced when larger amounts of M53 expression plasmid were cotransfected , suggesting that M53 and IRE1 compete for binding to M50 . To check whether IRE1 was also downregulated during MCMV infection , IRE1 levels were determined in an infection time course experiment . Figure 4C shows that IRE1 levels decreased during the course of infection as M50 levels increased . The observed IRE1 downregulation is consistent with the inhibited Xbp1 splicing in normal fibroblasts , which express only endogenous IRE1 ( Fig . 1 ) Next we investigated whether IRE1 downregulation occurred at the transcriptional level . RNA was isolated from MCMV-infected cells and Ire1 transcripts were quantified by real-time RT-PCR . The results showed that Ire1 transcripts did not decrease but rather increased slightly during the course of MCMV infection ( Fig . 4D ) , indicating that IRE1 downregulation occurred at the posttranscriptional level . We then tested whether M50 induces IRE1 degradation . To do this , HEK 293 cells were cotransfected with M50 and IRE1 expression plasmids , and IRE1 stability was determined by pulse-chase analysis . Indeed , IRE1 stability was reduced significantly when M50 was coexpressed ( Fig . 4E and F ) , strongly suggesting that M50 induces IRE1 degradation . To test whether ubiquitylation was necessary for IRE1 degradation , IRE1 and M50 were coexpressed in ts20 cells , which have a temperature sensitive E1 ubiquitin-activating enzyme [47] . In these cells , M50 expression reduced IRE1 levels at both the permissive and restrictive temperatures ( Fig . S2A ) , suggesting that ubiquitin conjugation was not required . The IRE1 downregulation seen in immunoblot experiments was also not inhibited by proteasome inhibitors MG132 or lactacystin ( Fig . S2B ) . We also investigated whether IRE1 degradation could be inhibited by lysosomal protease inhibitors ( PI ) or NH4Cl , an inhibitor of lysosome acidification . Neither NH4Cl nor a PI cocktail inhibited IRE1 downregulation by M50 ( Fig . S2C ) . Collectively these data suggested that IRE1 is degraded neither by the proteasome nor in lysosomes but rather cleaved by another cellular protease . It is also conceivable that lysosomal proteases that are not inhibited by these drugs are responsible for IRE1 degradation . M50 consists of an N-terminal conserved region , a variable region , a TM domain , and a short C-terminal tail [39] . The N-terminal region is conserved among the herpesviruses , particularly those of the same subfamily [48] . To determine which parts of M50 are required for IRE1 downregulation , a number of N- and C-terminal truncation mutants and mutants with internal deletions were constructed ( Fig . 5A ) . These mutants were tested for their ability to downregulate IRE1 levels and interact with IRE1 . In cotransfection experiments , M50 mutants lacking the entire conserved region were unable to downregulate IRE1 , whereas mutants lacking only a part of the conserved or the variable region downregulated IRE1 ( Fig . 5B ) . The 141–317 mutant repeatedly displayed an intermediate phenotype , i . e . a moderate downregulation of IRE1 . Truncated M50 proteins lacking up to 140 aa from the N-terminus coprecipitated with IRE1 , but mutants lacking the entire conserved region did not ( Fig . 5C ) . The M50 1–276 mutant , which lacks the TM domain , was also incapable of downregulating IRE1 ( Fig . 5B ) but coprecipitated with IRE1 ( Fig . S3A ) and colocalized , at least partially , with IRE1 in transfected cells ( Fig . S3B ) . However , when the M50 TM domain was substituted by the TM domain of an unrelated type 2 TM protein , HSV-1 UL56 , IRE1 downregulation was restored , suggesting that the M50 protein needs a TM anchor for IRE1 downregulation but not for interaction with IRE1 . We wanted to test whether M50 is responsible for the IRE1 downregulation observed in MCMV-infected cells ( Fig . 4C ) . This could be done with an MCMV M50 deletion mutant or a virus mutant expressing an M50 protein lacking the conserved region . Unfortunately , M50 is essential for MCMV replication as it mediates , together with M53 , nuclear egress of viral capsids [34] . The conserved region of M50 , which mediates interaction with IRE1 ( Fig . 5 ) , is also required for its interaction with M53 [38] , [39] . Until recently , all attempts to generate M50 trans-complementing cell lines for the propagation of an M50-deficient MCMV had failed because stable M50 expression was not tolerated by cells [34] . This obstacle was recently overcome with an MCMV-inducible expression system based on an episomal replicating plasmid containing the MCMV origin of lytic replication and the M50 gene [49] . In NIH-3T3 cells stably carrying this plasmid M50 expression was silenced . Upon MCMV infection , however , the vector was replicated and M50 expression was strongly induced . An MCMV mutant lacking M50 ( MCMVΔM50 ) could be propagated on these trans-complementing cells [49] and used for further experiments . When we infected 10 . 1 fibroblasts stably expressing myc-tagged IRE1 with MCMVΔM50 or the parental control virus , we observed a strong downregulation of IRE1 levels by the parental MCMV , but only a slight reduction by the MCMVΔM50 virus ( Fig . 6A ) . MCMV infection caused a modest reduction of Ire1 transcripts in these cells ( Fig . S4 ) . This reduction was seen for both viruses , indicating that M50 is not responsible for this effect . However , it is possible that the slightly reduced IRE1 protein levels observed in MCMVΔM50-infected cells ( Fig . 6A ) were caused by reduced Ire1 transcription . In NIH-3T3 fibroblasts ( expressing only endogenous IRE1 ) , splicing of Xbp1 transcripts and transcription of ERdj4 was strongly inhibited upon infection with the MCMV control virus , but only moderately diminished upon infection with MCMVΔM50 ( Fig . 6B and C ) . These results showed that M50 is primarily responsible for inhibition of the IRE1-XBP1 pathway during MCMV infection . However , the moderate reduction in Xbp1 mRNA splicing and ERdj4 transcription in MCMVΔM50-infected cells suggest that additional mechanisms contribute to the inhibition of the IRE1-XBP1 signaling pathway . M50 is a protein conserved among the Herpesviridae family , and the functional conservation was reported to be particularly strong among members of the same subfamily [48] . Hence we tested whether UL50 , the HCMV homolog of M50 , has a similar function . Indeed , UL50 coimmunoprecipitated with IRE1 like M50 did ( Fig . 7A ) , and UL50 expression downregulated IRE1 levels in transfected cells ( Fig . 7B ) . Moreover , IRE1 levels in HCMV-infected fibroblasts decreased over time ( Fig . 7C ) , and this decrease correlated with a suppression of Xbp1 splicing following Tun treatment ( Fig . 7D ) . Therefore we concluded that the novel function of M50 described in this report is not unique for MCMV but conserved in the related human pathogen , HCMV .
In this study we showed that MCMV and HCMV repress IRE1-mediated ER-to-nucleus signaling , the most conserved branch of the UPR ( Fig . 8A ) . The viral proteins M50 and UL50 , respectively , interact with IRE1 and downregulate IRE1 levels in transfected or infected cells ( Fig . 8B ) . Thereby , IRE1-mediated Xbp1 mRNA splicing , synthesis of transcription factor XBP1s , and expression of XBP1s target genes are inhibited . These results are consistent with two previous studies , which have reported an inhibition of EDEM ( an XBP1s target gene ) expression by HCMV [27] and a block to Xbp1 mRNA splicing by MCMV [32] , respectively . In these studies , the underlying mechanism of these effects and the viral proteins involved were not investigated . However , other previous studies have shown that HCMV upregulates the ER chaperone BiP through increased transcription and activation of translation by using the BiP internal ribosome entry site [29] , [30] . BiP was shown to be important for HCMV virion assembly [29] . Moreover , since BiP binds to the ER stress sensors PERK , ATF6 , and IRE1 and keeps them inactive , it has been suggested that BiP upregulation might also dampen the UPR [29] . We and others have also observed BiP upregulation in MCMV-infected cells ( Fig . S5 and [32] ) , and this effect might in fact be responsible for the moderate inhibition of Xbp1 splicing and ERdj4 transcription observed in MCMVΔM50-infected cells ( Fig . 6B and C ) . It should also be noted that an interaction between UL50 and BiP has been described in a previous study [50] . It remains to be investigated whether UL50 interacts with BiP directly or rather indirectly via IRE1 . Collectively , the data of the present and previous studies suggest that M50/UL50 and increased BiP levels have a synergistic inhibitory effect on the IRE1-dependent signaling pathways . Apart from the strong inhibition of the IRE1-XBP1 axis at late times postinfection , MCMV infection causes a modest induction of Xbp1 mRNA splicing at very early times after infection ( Fig . 1B , 2 hpi ) , which decreases within the following hours . The cause of this modest effect was not investigated in this study and remains unknown . It is unlikely that viral glycoprotein expression is responsible for this very early induction of Xbp1 mRNA splicing as viral glycoproteins are not expressed in large quantities so early after infection . However , it is possible that the high-MOI infection itself causes ER stress , for instance , by inducing a rapid and transient Ca2+ release from the ER as it has been described for HSV-1 infection [51] . It also remains unknown whether the initial ER stress induction occurs only transiently , or whether it is actively inhibited by a virally induced mechanism . M50 is expressed only at late times and becomes detectable around 16 hpi ( Fig . 3B ) . By contrast , BiP upregulation starts already 8 to 12 hpi ( Fig . S5 ) and might contribute to inhibition of the very early Xbp1 splicing . By downregulating IRE1 the CMVs can avoid cellular responses that are likely detrimental for viral replication . Many XBP1s target genes encode ERAD proteins , which reduce the protein load in the ER by enhancing ER-associated protein degradation [1] . Particularly in the late phase of the viral replication cycle , when large quantities of viral glycoproteins are needed for progeny production , this counter-regulatory mechanism should have a negative impact on viral replication . XBP1s has also been reported to enhance interferon β production in dendritic cells [52] , providing another good reason for the virus to block the IRE1-XBP1 pathway . Moreover , IRE1 has a role in several other pathways: Besides Xbp1 mRNA splicing , its endoribonuclease activity also mediates cleavage and inactivation of glycoprotein-encoding mRNAs [53] as well as certain microRNAs [54] ( Fig . 8C ) . In addition , IRE1 can initiate ER stress-induced programmed cell death by recruiting the adaptor protein TRAF2 and activating caspase-12 or JNK [16] , [55] , [56] ( Fig . 8D ) . Activated JNK phosphorylates and inhibits the antiapoptotic Bcl-2 and activates proapoptotic BH3 proteins [57] , [58] . One can assume that the viral mediated downregulation of IRE1 , which we described in this study , inhibits all IRE1-dependent pathways . However , further in-depth studies will be necessary to fully characterize all consequences of IRE1 downregulation by M50/UL50 and a potential synergism with UL38 , an HCMV protein that inhibits ER stress-induced JNK activation and apoptosis [31] . While it is clear that M50 interacts with IRE1 and downregulates IRE1 levels by reducing its half-life , the exact mechanism of IRE1 removal remains to be determined . As viruses often abuse host mechanisms for their own benefit , it is possible that the CMVs activate a cellular IRE1-inhibiting mechanism . For instance , the cellular BAX inhibtor-1 ( BI-1 ) protein interacts with IRE1 and inhibits the IRE1-XBP1 signaling pathway [59] . However , BI-1 has not been reported to downregulate IRE1 protein levels , indicating that M50 and UL50 operate in a different manner . By contrast , the cellular protein synoviolin interacts with IRE1 and induces its ubiquitylation and degradation by the proteasome [60] . It remains to be determined whether or not M50 and UL50 operate in a similar fashion . However , the viral mediated IRE1 downregulation appears to be stronger than the one reported for synoviolin , and the preserved IRE1 downregulation by M50 in the presence of proteasome inhibitors and in ubiquitylation-deficient cells ( Fig . S2 ) argue for a proteasome-independent mechanism . Besides its effect on IRE1 , M50 has an essential role in the export of viral capsids through the nuclear envelope . It interacts with the nuclear-localized M53 protein and facilitates primary envelopment at the inner nuclear membrane [34] . It should be worthwhile to separate the functions of M50 in capsid export and IRE1 inhibition in order to study them separately during viral infection . This is probably a very challenging task as both functions require the conserved N-terminal domain . However , with a suitable mutant virus one could investigate the importance of IRE1 inhibition for CMV replication in cell culture as well as in the mouse model . The essential function of M50 in nuclear egress is highly conserved not only among the CMVs , but among all herpesviruses analyzed thus far [35] , [36] . Hence it would be interesting to investigate whether the IRE1-downregulating function of M50 and UL50 is also conserved beyond the betaherpesviruses . Clearly , increasing evidence argues for additional , nuclear egress-unrelated functions of the M50 homologs in both alpha- and betaherpesviruses [40] , [61] .
NIH-3T3 ( ATCC CRL-1658 ) , 10 . 1 [62] , 293T ( ATCC CRL-11268 ) ; 293A ( Invitrogen ) , telomerase-immortalized human foreskin fibroblasts ( HFF ) [63] , ts20 cells [47] , and MRC-5 ( ATCC CCL-171 ) cells were grown under standard conditions in Dulbecco's modified Eagle's medium supplemented with 5% neonatal or 10% fetal calf serum , 100 units/ml penicillin , and 100 µg/ml streptomycin . Wildtype MCMV , MCMV-GFP [64] , MCMV-M50HA [40] , and MCMV-HAm41 [65] were grown and titrated on 10 . 1 fibroblasts . HCMV AD169-GFP [66] was grown and titrated on HFF . MCMVΔM50 and the corresponding control virus were propagated and titrated on M50-complementing cells as described [49] . Viral titers were determined using the median tissue culture infective dose ( TCID50 ) method [67] . Plasmids pcDNA-hIRE1α and pCMVTAG-NEMO were purchased from Addgene , pCR3-IgM53 [34] was provided by Walter Muranyi . For pcDNA-IRE1-TEV-HA , the murine IRE1α cDNA was PCR-amplified ( introducing the TEV-HA sequence with the reverse primer ) and inserted between the EcoRI and XbaI sites of pcDNA3 ( Invitrogen ) . The IRE1-TEV-HA sequence was also cloned in pBRep , an episomal replicating plasmid vector [68] . Plasmids pcDNA-hIRE1-HA , pcDNA-M50 , pcDNA-M50-Flag , pcDNA-m144-Flag , and pcDNA-UL56-Flag were generated by PCR amplification and insertion of the coding sequence between the HindIII and XhoI sites of pcDNA3 . Plasmids encoding N- and C-terminal truncations of M50 were generated in the same way . Deletions within the M50 variable region were made as described [39] . Substitutions of the M50 TM domain were made using a three-step PCR-based procedure essentially as described elsewhere [69] . pcDNA-UL50-Flag and pcDNA-CNX-Flag were also generated by PCR cloning using the EcoRI and XhoI sites of pcDNA3 . The K599A mutation was introduced by QuikChange site-directed mutagenesis ( Stratagene ) into pcDNA-IRE1-HA . Transient transfections were done using ployethyleneimine ( Sigma ) or PolyFect transfection reagent ( Qiagen ) according to the manufacturer's protocol . Within each experiment , the total amount of transfected DNA was kept constant by addition of empty vector plasmid . Tunicamycin , thapsigargin , puromycin , and protease inhibitor cocktail ( 104 mM AEBSF , 80 µM Aprotinin , 4 mM Bestatin , 1 . 4 mM E-64 , 2 mM Leupeptin , 1 . 5 mM Pepstatin A ) were purchased from Sigma , MG132 from Merck , and lactacystin from Biomol . HA-tagged murine IRE1α was PCR amplified , digested with BglII and XbaI , and inserted into pMSCVpuro ( Clontech ) . Murine IRE1α with a 3xmyc tag was PCR amplified , digested with BglII and HpaI , and inserted into pMSCVhyg ( Clontech ) . HA-tagged human IRE1α was excised from pcDNA-hIRE1-HA and inserted between the PmlI and XhoI sites of pRetroEBNA [70] . Retrovirus production using the Phoenix packaging cell line and transduction of target cells was done as described [71] . Cells transduced with MSCVpuro vectors were selected with 6 µg/ml puromycin ( Sigma ) and cells transduced with MSCVhyg vectors were selected with 200 µg/ml hygromycin B ( PAA Laboratories ) . NIH-3T3 cells were transfected with pBRep-IRE1-TEV-HA and selected as bulk culture for 14 days with 200 µg/ml hygromycin B . IRE1-TEV-HA expression was verified by immunoblot . 8×107 cells were mock treated or MCMV infected at an MOI of 1 . After 48 h , cells were lysed with RIPA buffer ( 50 mM Tris pH 7 . 2 , 150 mM NaCl , 1% TritonX100 , 0 . 1% SDS , 1% sodium deoxycholate , and Complete protease inhibitor cocktail [Roche] ) and centrifuged for 10 min at 16000 g . Supernatants were loaded onto anti-HA 3F10 affinity columns ( Roche ) and washed with 20 mM Tris-HCl pH 7 . 5 , 0 . 1 M NaCl , 0 . 1 M EDTA , 0 . 05% Tween-20 . IRE1 and associated proteins were eluted by digestion with 100 units of AcTEV protease ( Invitrogen ) for 1 h at room temperature . Eluted proteins were concentrated with StrataClean resin beads , separated by SDS-PAGE , and silver-stained [52] . In-gel digestion of excised gel bands was done as described [72] . Peptide extracts were analyzed on an Orbitrap XL mass spectrometer ( Thermo Scientific ) , online coupled to a bioinert Ultimate 3000 nano HPLCs ( Thermo Scientific ) . Peptides were pre-concentrated on a self-packed Synergi HydroRP trapping column ( 100 µm ID , 4 µm particle size , 100 Å pore size , 2 cm length ) and separated on a self-packed Synergi HydroRP main column ( 75 µm ID , 2 . 5 µm particle size , 100 Å pore size , 30 cm length ) at 60°C and a flow rate of 270 nL/min using a binary gradient ( A: 0 . 1% formic acid , B: 0 . 1% formic acid , 84% acetonitrile ) ranging from 5% to 50% B in 40 min . After each sample a dedicated wash blank was applied to clean the columns . MS survey scans were acquired from 350–2000 m/z in the Orbitrap with a resolution of 60 , 000 using the polysiloxane m/z 445 . 120030 as lock [73] . The five most intense signals were subjected to MS/MS in the LTQ with a normalized collision energy of 35 and a dynamic exclusion of 30 s . Automatic gain control target values were set to 106 for MS and 104 for MS/MS scans . Raw data were searched with the Proteome Discoverer Software 1 . 2 ( Thermo Scientific ) and Mascot 2 . 2 ( Matrix Science ) against Uniprot mouse and murid herpesvirus 1 databases . Search settings were as follows: ( i ) Trypsin as enzyme with a maximum of two missed cleavage sites , ( ii ) carbamidomethylation of Cys as fixed modification , ( iii ) phosphorylation of Ser/Thr/Tyr , and oxidation of Met as variable modifications , ( iv ) MS and MS/MS tolerances of 10 ppm and 0 . 5 Da , respectively . Only proteins with at least two peptides having ( i ) a Mascot score above 35 and ( ii ) a mass deviation ≤4 ppm and ( iii ) between 6 and 22 amino acids , were considered for data evaluation For immunoprecipitation 293A cells were transfected in 10 cm dishes and lysed after 24 h with RIPA buffer . Insoluble material was removed by centrifugation . Proteins were precipitated using antibodies against HA , Flag , or myc epitopes and protein A or protein G Sepharose ( GE Healthcare ) , respectively , washed 6 times , eluted by boiling in sample buffer , and subjected to SDS-PAGE and immunoblotting . For immunoblot analysis whole cell lysates were analyzed using antibodies against Flag epitope ( M2 or F7425 , Sigma ) , HA epitope ( 16B12 , Covance Inc . , or 3F10 , Roche ) , myc epitope ( 4A6 , Millipore ) , β actin ( AC-74 , Sigma ) , MCMV IE1 ( CROMA101; provided by Stipan Jonjic , University of Rijeka , Croatia ) , HCMV IE1/2 ( 3H4; provided by Thomas Shenk , Princeton University , USA ) , M50 [34] , M55/gB ( SN1 . 07 , provided by Stipan Jonjic ) , BiP ( E-4 , Santa Cruz ) ; IRE1α ( 14C10 , Cell Signaling ) , IRE1α[pSer724] ( Novus Biologicals ) , XBP1s ( M-186 , Santa Cruz ) , HP1α ( Cell Signaling ) , p53 ( FL-393 , Santa Cruz ) . Secondary antibodies coupled to horseradish peroxidase were purchased from Dako . NIH-3T3 or 10 . 1 cells were transfected or infected on coverslips , washed with PBS , and fixed for 20 min in 4% paraformaldehyde in PBS . Cells were incubated with 50 mM ammonium chloride , permeabilized with 0 . 3% TritonX-100 , and blocked with 0 . 2% cold-water fish skin gelatin ( Sigma ) and 2% horse serum ( when the anti-M50 antiserum was used ) . Cells were then incubated with primary antibodies for 1 h at room temperature ( RT ) , washed three times with PBS , and incubated for 1 h with secondary antibodies coupled to AlexaFluor555 or AlexaFluor488 ( Invitrogen ) . Nuclei were stained using Draq5 ( BioStatus ) . Samples were washed , mounted on slides with Aqua-Poly/Mount ( Polysciences ) , and analyzed by confocal laser scanning microscopy using a Zeiss LSM510 Meta microscope . The Pearson correlation coefficient was calculated using JACoP for ImageJ [74] . 293T cells were transfected with pcDNA-IRE1-HA and pcDNA-M50 at a 1∶4 ratio using polyethyleneimine . 48 h after transfection cells were incubated with methionine-deficient DMEM for 45 min and pulse-labeled with 35S-methionine ( IsoLabel L-[35S] , Izotop , Hungary ) for 30 min . Cell were chased for up to 4 h in DMEM containing 50-fold excess of cold methionine . Cells were then harvested and lysed as described previously [75] . HA-tagged IRE1 was immunoprecipitated with 12CA5 anti-HA monoclonal antibody and protein G-conjugated sepharose ( Santa Cruz ) . Immunoprecipitates were washed extensively with NET buffer containing 0 . 1% SDS , boiled in Laemmli sample buffer and separated by SDS-PAGE . The gel was processed for autoradiography as previously described [75] . Total RNA was isolated from murine fibroblasts using innuPREP RNA Mini Kit ( analytik-jena ) . After DNase treatment ( Turbo DNA-free Kit , Ambion ) cDNA was synthesized from 1 µg RNA using 200 U RevertAid H Minus Reverse Transcriptase , 100 pmol Oligo[dT]18 , and 20 U RNase inhibitor ( Thermo Scientific ) . For semiquantitative analysis , murine Xbp1 was amplified by using primers 5′-AAACAGAGTAGCAGCGCAGACTGC-3′ and 5′-AAACAGAGTAGCAGCGCAGACTG C-3′ . Primers 5′-GCCAGAGGAGGAACGAGCT-3′ and 5′-GGGCCTTTTCATTGTT TTCCA-3′ were used to amplify c-myc . PCR reaction was performed under the following conditions: 40 cycles of 30 s at 95°C , 30 s at 48°C , and 30 s at 72°C . PCR products were digested with PstI and analyzed on an ethidium bromide-stained agarose gel as described [76] . Quantitative RT-PCR reactions employing SYBR Green fluorescent reagent ( Applied Biosystems ) were run in an Applied Biosystems 7900HT Fast Real-Time PCR System . The following primers were used: 5′-GAGTCCGCAGCAGGTG-3′ and 5′-GTGTCAGAGTCCATGGGA-3′ murine Xbp1s , 5′-GTGTCAGAGTCCATGGGA-3′ and 5′-GTGTCAGAGTCCATGGGA-3′ for murine Xbp1u , 5′-GAGTCCGCAGCAGGTG-3′ and 5′-CAATACCGCCAGAATCCA-3′ for human XBP1s , 5′-CACTCAGACTATGTGCACCTC-3′ and 5′-CAATACCGCCAGAATCCA-3′ for human XBP1u , 5′-ATAAAAGCCCTGATGCTGAAGC-3′ and 5′-GCCATTGGTAAAAGCACTGTGT-3′ for murine ERdj4 , 5′-CGGCCTTTGCTGATAGTCTC-3′ and 5′-AGTTACCACCAGTCCATCGC-3′ for murine Ire1 and 5′-CCCACTCTTCCACCTTCGATG-3′ and 5′-GTCCACCACCCTGTTGCTGTAG-3′ for human and murine GAPDH . Reactions were performed under the following conditions: 45 cycles of 3 s at 95°C and 30 s at 60°C . Three or four replicates were analyzed for each condition , and the relative amounts of mRNAs were calculated from the comparative threshold cycle ( Ct ) values by using GAPDH as reference . GenBank accession numbers of proteins and genes mentioned in this study: murine ATF4 ( NP_033846 ) , ATF6 ( NP_001074773 ) , BiP ( P20029 ) , BI-1 ( NP_001164507 ) , CNX ( P35564 ) , ERdj4 ( NM_013760 ) , GAPDH ( NM_008084 ) , IRE1 ( AF071777 ) , IRE2 ( Q9Z2E3 ) , PERK ( NP_034251 ) , SYVN1 ( NP_001158181 ) , XBP1s ( NM_001271730 ) , XBP1u ( NM_013842 ) ; human EDEM ( NP_055489 ) , ERdj4 ( NM_012328 ) , GAPDH ( NM_002046 ) , IRE1 ( NM_001433 ) , XBP1s ( NM_001079539 ) , XBP1u ( NM_005080 ) ; MCMV IE1 ( P11210 ) , m41 ( ADD10423 ) , M50 ( ADD10432 ) , M53 ( ADD10435 ) , M55 ( ADD10436 ) , M85 ( ADD10456 ) , m144 ( ADD10510 ) ; HCMV IE1 ( P13202 ) , IE2 ( P19893 ) , UL50 ( P16791 ) ; HSV-1 UL56 ( AEQ77088 ) .
|
Viruses abuse the cell's protein synthesis and folding machinery to produce large amounts of viral proteins . This enforced synthesis overloads the cell's capacity and leads to an accumulation of unfolded proteins in the endoplasmic reticulum ( ER ) resulting in ER stress , which can compromise cell viability . To restore ER homeostasis , cells initiate the unfolded protein response ( UPR ) to reduce protein synthesis , increase degradation of unfolded proteins , and upregulate chaperone expression for enhanced protein folding . The most conserved branch of the UPR is the signaling pathway activated by the ER stress sensor IRE1 . It upregulates ER-associated degradation ( ERAD ) , thereby antagonizing ER stress . Some of the counter-regulatory mechanisms of the UPR are detrimental for viral replication and are , therefore , moderated by viruses . In this study we identified the first viral IRE1 inhibitor: The murine cytomegalovirus M50 protein , which interacts with IRE1 and induces its degradation . By this means , M50 inhibits IRE1 signaling and prevents ERAD upregulation . Interestingly , the M50 homolog in human cytomegalovirus , UL50 , also downregulated IRE1 revealing a previously unknown mechanism of viral host cell manipulation .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"cellular",
"stress",
"responses",
"molecular",
"cell",
"biology",
"viral",
"immune",
"evasion",
"virology",
"biology",
"microbiology",
"pathogenesis"
] |
2013
|
Cytomegalovirus Downregulates IRE1 to Repress the Unfolded Protein Response
|
Double strand breaks ( DSBs ) and interstrand crosslinks ( ICLs ) are toxic DNA lesions that can be repaired through multiple pathways , some of which involve shared proteins . One of these proteins , DNA Polymerase θ ( Pol θ ) , coordinates a mutagenic DSB repair pathway named microhomology-mediated end joining ( MMEJ ) and is also a critical component for bypass or repair of ICLs in several organisms . Pol θ contains both polymerase and helicase-like domains that are tethered by an unstructured central region . While the role of the polymerase domain in promoting MMEJ has been studied extensively both in vitro and in vivo , a function for the helicase-like domain , which possesses DNA-dependent ATPase activity , remains unclear . Here , we utilize genetic and biochemical analyses to examine the roles of the helicase-like and polymerase domains of Drosophila Pol θ . We demonstrate an absolute requirement for both polymerase and ATPase activities during ICL repair in vivo . However , similar to mammalian systems , polymerase activity , but not ATPase activity , is required for ionizing radiation-induced DSB repair . Using a site-specific break repair assay , we show that overall end-joining efficiency is not affected in ATPase-dead mutants , but there is a significant decrease in templated insertion events . In vitro , Pol θ can efficiently bypass a model unhooked nitrogen mustard crosslink and promote DNA synthesis following microhomology annealing , although ATPase activity is not required for these functions . Together , our data illustrate the functional importance of the helicase-like domain of Pol θ and suggest that its tethering to the polymerase domain is important for its multiple functions in DNA repair and damage tolerance .
DNA double strand breaks ( DSBs ) and interstrand crosslinks ( ICLs ) compromise both strands of the DNA duplex and must be repaired to ensure cellular survival . While DSBs disrupt the physical integrity of DNA , ICLs tether DNA strands together through a covalent bond . Thus , both types of lesions can impede critical processes such as replication and transcription . ICLs require a complex network of proteins for their removal and involve multiple DNA repair pathways ( reviewed in [1 , 2] ) . Because repair of ICLs can generate one- or two-ended DSB intermediates , the identification of proteins that are involved in both ICL and DSB repair might provide insight into their common mechanisms [3 , 4] . In Drosophila melanogaster , DNA Polymerase θ ( Pol θ ) has emerged as one of these dual-function proteins . It was first identified in a mutagen sensitivity screen where mutations in mus308 , the gene encoding Pol θ , caused hypersensitivity to ICL-inducing agents but not to other DNA alkylating agents [5 , 6] . Pol θ is an A-family polymerase with homology to E . coli Pol I and contains an N-terminal helicase-like domain connected to the polymerase domain through a long , unstructured central domain [7–9] . Both the polymerase and helicase domains are conserved among metazoans . In vertebrates , the polymerase domain also contains three insertion loops that are not present in Pol I and which are poorly conserved in invertebrates ( reviewed in [10] ) . Early work on Drosophila Pol θ showed that it likely acts downstream of ICL “unhooking” [5] , where one strand of the DNA helix is incised on both sides of the crosslink . Subsequent studies in C . elegans revealed a conserved role for Pol θ in ICL repair [11] . However , the mechanisms by which Pol θ promotes ICL repair remain uncharacterized . Numerous studies have shown that Pol θ is also critical for alternative end-joining repair of DSBs via a pathway named microhomology-mediated end joining ( MMEJ ) [12–17] . MMEJ occurs when short , complementary DNA sequences located at broken DNA ends anneal and serve as primers for fill-in synthesis . An early study of Pol θ-mediated MMEJ in Drosophila , using a transposon-induced gap repair system , demonstrated that repair involving annealing of 5–10 nucleotide pre-existing microhomologies was dependent on Pol θ [18] . Pol θ has also been shown to promote MMEJ in C . elegans [19 , 20] , zebrafish [21] , and mice [13] . While MMEJ acts independently of classical NHEJ proteins , 5’→3’ resection stimulates MMEJ and there is some overlap between proteins involved in the intial steps of both MMEJ and homologous recombination [22] . Following resection , microhomologies located internal to the break site can also be used for annealing , although this requires trimming of non-homologous flaps prior to synthesis [23] . In vitro , the polymerase domain of Pol θ is sufficient to join DNA ends with limited microhomologies , though the optimal length of DNA overhangs varies by system [12 , 13] . Additionally , the polymerase domain can mediate the alignment of both internal and terminal microhomologies and displace annealed DNA during template extension in vitro [12] . In addition to the use of microhomologies , another signature of many alternative end-joining events in metazoans is the presence of small insertions that appear to be derived from flanking DNA sequences [24] . It is now clear that Pol θ is often utilized during the generation of these templated insertions . In Pol θ-null mouse B cells , programmed DSBs formed during class-switch recombination are repaired at a similar rate to wildtype but lack insertions of >1 nt , indicating that Pol θ is required to generate these insertions [13] . Studies with I-SceI induced chromosomal DSBs in Drosophila documented repair events with >4 nt templated insertions that were largely dependent on Pol θ [25] . In addition , Pol θ plays a critical role in the generation of insertions during alternative end joining repair of collapsed replication forks in C . elegans [19 , 20 , 26] and during T-DNA insertion in Arabidopsis [27] . Though the function of the polymerase domain of Pol θ has been studied extensively in vitro , thus far no defined role exists for the helicase-like domain . The helicase-like domain displays DNA-dependent ATPase activity in vitro , but DNA unwinding activity has not been demonstrated in vitro or in vivo [8 , 28] . Pol θ null mouse embryonic fibroblasts are extremely sensitive to the DSB-inducing agent bleomycin and the addition of the polymerase domain alone is sufficient to restore resistance to these cells [13] . The helicase-like domain , therefore , does not appear to be required for efficient DSB repair in mammalian cells and its vital role ( s ) , if any , in DNA metabolism have remained largely uncharacterized . Here , we report our efforts to further elucidate the roles of the helicase-like and polymerase domains of Drosophila Pol θ in ICL and DSB repair . We show that purified Pol θ can bypass model unhooked ICL substrates and promote the initial steps of MMEJ in vitro , with minimal requirement for ATPase activity . In contrast , our in vivo experiments establish that both the ATPase activity of the helicase-like domain and DNA polymerase activity are critical for tolerance or repair of ICLs and for the generation of templated insertions during alternative end joining . Tethering of these two enzymatic domains may therefore promote complex DNA transactions necessary for both types of repair .
Drosophila with mutations in the mus308 gene , encoding Pol θ , were originally identified by their hypersensitivity to the bi-functional alkylating agent nitrogen mustard , which induces ICLs [5] . To determine which enzymatic functions of Pol θ are required during ICL repair , we generated a series of transgenic Drosophila lacking the endogenous mus308 gene and possessing Pol θ transgenes with a mutation in the Walker A box of the helicase-like domain that prevents ATP binding ( K262A; ATPase-dead ) or the catalytic function of the polymerase domain ( D1826A , F1827A; pol-dead ) ( Fig 1A ) . In addition , we created a transgenic rescue stock with a full-length wildtype copy of Pol θ . All of the transgenes included the full-length mus308 promoter to ensure endogenous expression . We exposed larvae from these transgenic stocks to increasing concentrations of nitrogen mustard and quantified survival to adulthood . Similar to previous studies , we found that homozygous mutants lacking Pol θ are hypersensitive to nitrogen mustard , as compared to heterozygous controls ( Fig 1B ) . Addition of the full-length transgene rescued this sensitivity to wild-type levels . Pol-dead mutants were hypersensitive to low concentrations of nitrogen mustard , indicating a requirement for DNA synthesis in Pol θ-mediated ICL repair . Intriguingly , we found that ATPase-dead mutants were equally as sensitive to nitrogen mustard as pol-dead mutants or flies lacking Pol θ . In mammals , the polymerase activity of Pol θ is required for double-strand break repair , specifically for MMEJ [12 , 13 , 17] . Interestingly , the polymerase domain of purified Pol θ protein alone is sufficient to join DNA ends in vitro , indicating that ATPase activity is dispensable [12] . To test whether this is also true in Drosophila , we exposed flies expressing different Pol θ transgenes to ionizing radiation ( IR ) in the absence of endogenous Pol θ . Our previous study showed that flies lacking Pol θ are only sensitive to IR in the absence of homologous recombination [18] . Therefore , we conducted these experiments in flies lacking RAD51 ( encoded by the spn-A gene ) . As expected , mus308Δ , spn-A mutants were highly sensitive to IR and addition of the full-length transgene rescued this sensitivity ( Fig 1C ) . Indeed , a dose of 500 rads was lethal to mutants lacking Pol θ and RAD51 , while 50 percent of spn-A mutants were able to survive this exposure . Pol-dead and Pol θ null mutants were equally hypersensitive to IR . However , ATPase-dead mutants behaved identically to the wild-type control . Thus , the ATPase activity of Pol θ is needed for survival following exposure to DNA interstrand crosslinking agents , but not for damage induced by ionizing radiation . To further delineate the functions of the ATPase and polymerase domains in ICL repair , we turned to an in vitro system . We purified full-length wild-type , ATPase-dead , and pol-dead proteins to near homogeneity from insect cells . SDS-PAGE and Western blotting confirmed the identity of full-length Pol θ , although some minor degradation products were visible ( S1 Fig ) . Current models suggest that during replication-coupled ICL repair , two incisions are made on either side of the ICL to release it from one strand , generating an “unhooked” ICL in which the excised fragment remains attached to other strand through the ICL itself . However , the exact location of the incisions and the length of the remaining olionucleotide attached to the duplex are currently unknown [29] . Thus , in vitro studies of ICL bypass typically use model unhooked ICLs of different lengths [30] . We used our purified Pol θ proteins in primer extension assays with substrates mimicking 6 base pair ( bp ) or 20 bp unhooked nitrogen mustard ICLs and corresponding control substrates ( Fig 2A and 2B ) [31 , 32] . Both wild-type and ATPase-dead Pol θ were able to effectively extend a fluorescently labeled primer on a single-strand control template , although their activity on a partial double-strand control template was somewhat reduced compared to the Klenow fragment of E . coli DNA polymerase I ( Fig 2C and 2D ) . While both Klenow fragment and wild-type Pol θ could insert a nucleotide opposite an ICL , Klenow was unable to extend past the ICL with both the 6 bp and 20 bp unhooked crosslinks ( Fig 2C and 2D ) . By contrast , Pol θ was able to efficiently carry out extension to full product for the 6 bp ICL . The efficiency of extension was reduced with the 20 bp ICL , similar to previous reports with other polymerases [31 , 32] . We observed only moderately reduced extension activity with ATPase-dead Pol θ compared to wild-type , suggesting that ATP hydrolysis is not required for bypassing an unhooked ICL in vitro . Previous work from our lab showed that hypomorphic and point mutations in Pol θ lead to a significant decrease in DNA ligase 4-independent alternative end joining [18] . During Pol θ-mediated end joining of I-SceI or P element-induced double-strand breaks , microhomologies are often used to align DNA ends and insertions are present in approximately 25% of repair junctions [18 , 25] . To determine which domains of Pol θ might be involved in the generation of these insertions , we utilized a well-characterized site-specific gap repair assay [33] . In this assay , a 14 kilobase transposon ( P{wa} ) is inserted into an intron of the scalloped gene on the X chromosome . Excision of P{wa} is catalyzed by P transposase , resulting in a two-ended double-strand break with 17-nucleotide non-complementary 3’ overhangs ( Fig 3A ) . While this break can be repaired proficiently through homologous recombination , in the absence of RAD51 the break is repaired solely through end joining . Repair events that occur in the pre-meiotic male germline can be recovered from the female progeny that inherit the repaired chromosome and sequenced . As expected , expression of wild-type Pol θ in a mus308Δ mutant background resulted in a similar frequency of alternative end-joining events to flies expressing endogenous Pol θ ( Fig 3B ) . Ablation of ATPase activity did not affect the frequency of end joining repair ( Fig 3B ) , in agreement with previously published results that ATPase function is not required for MMEJ [12 , 13] . In contrast , we saw a significant decrease in end joining with the pol-dead allele , comparable to mus308Δ mutants . In the P{wa} system , repair events accompanied by large deletions ( >1 . 5 kilobases ) that remove part of the coding region of the scalloped gene can be scored by the presence of female progeny possessing scalloped wings [34] . These deletions are thought to occur through a resection-based mechanism and were previously observed in flies with Pol θ-inactivating point mutations [18] . Notably , we observed a substantial increase in deletion-associated repair events in the polymerase-dead mutant compared to the full-length control and ATPase-dead mutant ( S1 Table ) . Thus , our data suggest that Pol θ polymerase , but not ATPase activity , is required to promote normal levels of alternative end joining and to suppress alternative , deletion-prone repair pathways . Analysis of the repair junctions recovered from the P{wa} assay showed that Pol θ-proficient flies often utilized long microhomologies during break repair . We saw a particularly high usage of an 8 bp imperfect internal microhomology in both the wildtype and the ATPase-dead backgrounds ( Tables 1 and 2 ) . In addition , we recovered many repair products from both backgrounds that appeared to be generated through annealing at short microhomologies , with a preference for microhomologies closer to the 3’ ends of the single-stranded overhangs . Thus , it appears that ATPase activity is not required to generate products through annealing at pre-existing microhomologies . We also observed frequent insertions in wild-type repair products that appeared to be templated from sequences near the break site . This insertion class accounted for just 3% of repair junctions recovered from the ATPase-dead mutants , significantly less than the 24% of wild-type repair junctions with insertions ( p = 0 . 03 , Fisher’s exact test; Tables 1 and 2 and Fig 3C ) . The sole insertion event generated in the ATPase-dead mutant was a four-nucleotide insertion apparently templated from sequence immediately adjacent to the break . In contrast , repair events in flies with endogenous mus308 expression or the control transgene contained insertions templated from sequences immediately adjacent to the break as well as internal sequences ( Table 1 ) . Flies expressing wild-type Pol θ also contained more complex events that can be explained by an iterative process of multiple rounds of synthesis , dissociation , and reannealing . These events were absent in ATPase-dead mutants . To further test whether the ATPase activity of Pol θ might be important for the complex insertions observed during alternative end joining , we utilized purified proteins with substrates that can only support templated DNA synthesis following annealing of 2 nt or longer terminal microhomologies . We began with a partial single-stranded DNA substrate that was previously used with the human Pol θ polymerase domain to simulate MMEJ-like synthesis reactions [12] ( Fig 4A ) . Similar to the human protein , wild-type Drosophila Pol θ can carry out DNA synthesis with this substrate ( Fig 4A ) . The size of the product is consistent with annealing of two molecules of the substrate at CCGG microhomologies , followed by extension and strand displacement by Pol θ . Interestingly , the amount of product is reduced with the ATPase-dead Pol θ ( Fig 4A ) . Next , we tested the ability of Pol θ to utilize a partial single-stranded DNA substrate mimicking the intermediates formed in the P{wa} assay ( Fig 4B ) . Surprisingly , wild-type Pol θ was able to catalyze more extension with this substrate , even though the terminal microhomology was only a 2-nt TA . While the products formed in this reaction were more variable in size , the length of the largest products was consistent with intermolecular annealing of two substrate molecules at the TA terminal microhomologies and strand-displacement synthesis to the end of the template . Repeated attempts to clone and sequence the reaction products were unsuccessful . Therefore , while the smaller products likely result from annealing at other microhomologous sequences in the substrate , it is unknown whether these are intermolecular or intramolecular events . Similar to the results with the first substrate , ATPase-dead Pol θ produced fewer extension products , particularly the full-length products ( Fig 4B ) . The polymerase domain of human Pol θ has more MMEJ activity on partial single-stranded substrates compared to single-stranded oligonucleotides with similar terminal microhomologies [12] . However , when we tested the ability of both full-length and ATPase-dead Drosophila Pol θ to utilize a single-stranded oligonucleotide corresponding to one strand of the P{wa} substrate , we observed an enhanced extension activity compared to that observed with the partial single-stranded substrate ( Fig 4B ) . The sizes of the products with the single-stranded substrate were smaller than what would be expected with intermolecular annealing at the TA microhomology and synthesis to the end of the template . Taken together , these data suggest that Drosophila Pol θ can utilize short microhomologies present in partially and fully single-stranded DNA as primers for DNA synthesis and that ATPase activity from the helicase-like domain promotes the efficiency of these reactions .
One possible role for Pol θ in ICL tolerance is the bypass of unhooked crosslinked duplex DNA following single-stranded nicking by nucleases [30] . The polymerase active site of Pol θ contains several features that make it an ideal enzyme to bypass bulky lesions such as crosslinks , including unique insertion loops that provide a stable interaction between a poorly matched primer and template strand [43] . Here , we have shown that Drosophila Pol θ is able to bypass model unhooked ICL substrates in vitro . Specifically , it can insert a nucleotide opposite a crosslinked base and subsequently extend from the insertion point , in the context of duplexes of variable lengths ( 6–20 bp ) duplexes surrounding the ICL . To date , only polymerase eta has been shown to carry out extension of these substrates with similar efficiency [30 , 31] . Interestingly , while the ATPase activity of Drosophila Pol θ is required for ICL tolerance in vivo , it plays only a minor role during bypass in vitro . Perhaps ATP hydrolysis is only required for efficient bypass of an ICL in a cellular context , where the helicase-like domain could be required to displace other proteins from the site of the lesion . In support of this , protein displacement is a major function of several other SF1 and SF2 helicases [44–48] . Alternatively , Pol θ ATPase activity might be necessary for efficient repair of one or two-ended DSBs that form during intermediate stages of ICL repair [1] . Our mutagen sensitivity assays were performed by treating larvae with nitrogen mustard during a time in larval development when imaginal disc cells , which are necessary for metamorphosis and adult survival , are undergoing rapid divisions . Mutations that result in an inability to repair DSBs during this developmental window will cause organismal death . A recent study using zebrafish found that Pol θ-mediated end joining is most critical at the earliest stages of blastulation , when cells divide every 15 minutes [21] . Thus , Pol θ-mediated end joining may be particularly crucial for the repair of crosslink-related DSBs that occur in rapidly dividing cells [49] . Although Pol θ is essential for ICL tolerance in Drosophila , Arabidopsis , and C . elegans , it is not required in mouse or human cell lines [5 , 11 , 50 , 51] . A possible explanation for this difference is that another TLS polymerase with redundant activity might substitute for Pol θ in mammals . One potential candidate is Pol ν , which has been suggested to participate in ICL repair [52]; human cell lines lacking Pol ν are sensitive to ICL-inducing agents [53 , 54] . Intriguingly , the organisms which utilize Pol θ for ICL tolerance ( Drosophila , Arabidopsis , and C . elegans ) lack a Pol ν homolog [55] . However , recent reports also demonstrate that recombinant human Pol ν cannot bypass cisplatin-induced crosslinks in vitro [56] , inconsistent with a substitution model . We found that mutations that abolish the ATPase activity of Pol θ do not increase sensitivity to IR-induced DNA damage . Additionally , loss of ATPase activity does not change the frequency of Pol θ-mediated end-joining repair following creation of a site-specific break . This reflects a similarity between Drosophila and mammalian cells , which also require polymerase activity , but not the helicase-like domain , for Pol θ-mediated end joining [12 , 13] . Pol θ has also been shown to generate templated insertions during alternative end joining [13 , 17 , 19 , 57] . These insertions might result from a cell’s attempt to generate microhomologies suitable for alternative end joining with initially incompatible DNA ends . Strikingly , when analyzing our repair junction sequences , we found that the ATPase-dead mutants had a significant , 8-fold reduction in insertion events compared to the controls and a corresponding increase in the number of repair events using 2–3 nt microhomologies . Similarly , purified Pol θ lacking ATPase activity was less efficient at annealing and extending single-stranded DNA substrates . Previous biochemical studies have established that the polymerase domain of human Pol θ aligns substrates with long GC-rich microhomologies more efficiently than AT-rich microhomologies and prefers using terminal microhomologies over internal ones [12] . We observed a similar predisposition with our partial single-stranded substrates . However , in the in vivo P{wa} repair system , end joining occurs using a 17 nucleotide AT-rich 3’ overhang . We see frequent usage of a long , 10-nt microhomology internal to the break site during junction formation . Utilization of internal microhomologies has also been observed in human cells [23] . Though our sequence is quite AT-rich , the most frequently used internal microhomologies have several GCs , supporting the idea that Pol θ more readily aligns GC-rich sequences . Additionally , in our system complex insertion events are typically templated from sequences directly adjacent to the break site . One potential explanation for this is that the AT-rich nature of the overhang results in less stable annealing between DNA ends , promoting multiple rounds of synthesis . In vitro , the human Pol θ polymerase domain efficiently uses partially single-stranded DNA with a 3’ overhang as a substrate for MMEJ , while fully ssDNA is used much less efficiently [12] . While we do observe annealing and extension products with partially single-stranded DNA , we found that full length Drosophila Pol θ robustly uses ssDNA as a template for synthesis in vitro . Because we failed to observe products consistent with intermolecular annealing at a TA terminal microhomology ( Fig 4C ) , we postulate that Pol θ may also use intramolecular hairpin-like structures as a template for synthesis , in a process that we have called synthesis-dependent microhomology-mediated end joining [25] . Similar observations have been made by others studying human Pol θ [57] .
While we have learned much about the roles of Pol θ in alternative end joining and ICL tolerance from biochemical studies , our findings illustrate that its functions in vivo are likely affected by additional factors . These might include variations in DNA sequence context or chromatin structure near DNA lesions and by interactions with other proteins that remain to be identified . Furthermore , the higher-order structure of Pol θ in vivo is likely important for its function . Crystal structures of the polymerase and helicase-like regions of Pol θ suggest that these subdomains can exist as dimers and tetramers , respectively [28 , 43] , but the oligomeric state of the full-length protein necessary for its in vivo functions is currently unknown . In summary , our studies with full-length Drosophila Pol θ suggest that the helicase-like and polymerase domains of Pol θ play important roles during both ICL repair and alternative end joining . Given the frequent overexpression of Pol θ in human cancers , our findings highlight the potential utility of therapeutically targeting one or both domains to modulate its activity [14] .
Fly stocks were maintained on standard cornmeal agar medium at 25°C . The mus308D2 , spn-A057 , and spn-A093 alleles were obtained from Bloomington Stock Center . The mus308Δ allele was generated through an imprecise excision screen of the P-element P{GSV7}GS23034 ( Kyoto Drosophila Genomics and Genetics Resources stock center ) . The mus308Δ allele deletes 14 , 250 bp spanning mus308 , its endogenous promoter , and part of the neighboring gene Men . To generate the transgenic alleles , genomic DNA encoding mus308 and 1 . 25 kb of upstream sequence , including its endogenous promoter , was isolated and amplified with primers containing BglII and Acc65I restriction enzyme sites . The DNA was cloned into the pMTL vector and mutagenized through amplification-based site-specific mutagenesis . Mutagenized alleles were then cloned into the pattB expression vector . The plasmids were injected into embryos containing attP integration sites by BestGene injection services and genome integration was achieved through the Fly-C31 lambda phage recombination system [58 , 59] . Transformants were identified using a white+ marker present on pattB . Control , polymerase-dead , and ATPase-dead transgenes were inserted at the ZH-attp-51C site ( Bloomington stock number 24483 ) on chromosome 2 . For nitrogen mustard assays , 5–8 mus308 transgene; mus308Δ/TM6B female flies were mated with 3–4 mus308 transgene; mus308Δ/TM6B males flies in a standard vial containing 5 mL cornmeal agar . Flies were allowed to mate and lay eggs for 3 days before being transferred to fresh vials for an additional 2 days . The first set of vials was treated with 250 μL of nitrogen mustard solution while the second control set was treated with 250 μL water . For irradiation assays , 40–50 mus308 transgene; mus308Δ , spnA057/TM3 females were mated to 10–15 mus308Δ , spnA093/TM6B males and eggs were collected on grape juice agar plates for 12 hours . Eggs were allowed to hatch and mature to third instar larvae , then irradiated using a Gammator 1000 irradiator . Percent relative survival was calculated using the ratio of mus308 homozygous to heterozygous adult survival in mutagen treated vials compared to control vials . Each experiment consisted of at least 5 independent vials or 1 grape agar plate and experiments were repeated in triplicate . End joining repair of a double strand break was monitored after the excision of a P{wa} element as described previously [60] . A second chromosome transposase source ( CyO , H{w+ , Δ2–3} ) was used to excise P{wa} . Single males of the genotype P{wa}; mus308 transgene/CyO , H{w+ , Δ2–3}; mus308Δ , spnA057/mus308D2 , spnA093 were mated to P{wa} females and individual repair events were recovered in the female progeny . Progeny containing end-joining events lose a functional copy of the white gene , thus end-joining repair events can be quantified and recovered in female progeny with yellow eyes . For each genotype , individual male crosses were scored for eye color of female progeny . The percentage of progeny from each repair class was calculated on a per vial basis , with each vial representing an independent experiment . Statistical comparisons were done with a non-parametric ANOVA followed by Tukey’s test using InStat3 ( GraphPad ) . Female progeny containing the repaired P{wa} construct were collected from the single male crosses . To isolate genomic DNA , whole flies were manually disrupted in 50 uL squishing buffer ( 10 mM Tris-HCl pH8 , 1 mM EDTA , 25 mM NaCl , 200 g/ml Proteinase K ) and incubated at 37°C for 30 minutes then 95°C for 3 minutes . Repair junctions were amplified by PCR using primers near the junction Sd5320 ( ACCATTGCAAGCTACATAGCTGAC ) and Sd5941R ( GCCTTGCTTCTTCCACACAGCGTG ) . PCR products were sequenced using the Sd5320 primer . Full-length mus308 cDNA ( Drosophila Genomics Resource Center , clone LP14642 ) was amplified with primers containing SalI and Acc65I restriction sites and cloned into pFastbac1 ( Invitrogen , gift from Timur Yusufzai ) in frame with the 6X His and FLAG tags to make pFBFL308 . ATPase-dead and pol-dead mutant clones were created by site-directed mutagenesis using Q5 polymerase ( NEB , Ipswich , MA ) . These constructs were used with the Bac to Bac Baculovirus Expression System ( Life Technologies ) for expression in Sf9 cells ( Orbigen , San Diego , CA . ) Sf9 cells ( 1 L , 6 . 0 × 106 cells/ml ) were infected with each construct for 72 hr at 25°C and harvested by centrifugation . Cells were sonicated in lysis buffer containing 20 mM Hepes , pH 7 . 6 , 500 mM NaCl , 1 . 5 mM MgCl2 , 10% glycerol , 0 . 1% Triton X-100 , 1 mM phenylmethylsulfonyl fluoride ( PMSF ) , and complete EDTA-free protease inhibitor ( Roche ) , with three 10 second bursts at 20% duty cycle with 10 seconds rest between bursts . Debris was removed by centrifugation and the supernatant was incubated for 4 hr at 4°C with 200 μl of anti-DYKDDDDK G1 affinity resin ( Genscript ) . The resin was washed with 10 volumes of the lysis buffer and eluted with 250 μg/ml 3X FLAG peptide ( gift of S . Fuchs ) . The eluate was spun through a Zeba 40K desalting column ( ThermoFisher ) following the manufacturer’s directions . Protein amounts were quantified by comparison with a BSA standard curve following transfer from an SDS-PAGE gel using Fastblot stain ( G Biosciences ) . 5-atom ICL-containing substrates were synthesized according to ( Roy et al . , 2016[61] ) . These were annealed with P15 ( 5’-CACTGACTCTATGATG-3’ ) labeled at the 5’ end with 6-FAM . ICL substrates ( 150 nM ) and 5’ labeled primer P15 ( 50 nM ) were annealed in 10 mM Tris-HCl pH 8 . 0 , 50 mM NaCl , overnight at room temperature to ensure the stability of the ICLs . The ICL substrates/primers ( 5 nM , with respect to the primer ) were incubated with Klenow ( exo- ) fragment or Pol θ in a reaction volume of 10 μL . For assays with Klenow ( exo- ) , 1 nM enzyme was used in reaction buffer NEB2 ( 50 mM NaCl , 10 mM Tris-HCl , 10 mM MgCl2 , 1 mM DTT ) . 0 . 2 nM Pol θ was used in a reaction buffer containing 25 mM Tris-Cl pH 7 . 5 , 10 mM MgCl2 , 200 μM ATP , 100 μg/mL BSA and 4 . 8% glycerol . Reactions with Klenow ( exo- ) were incubated for 5 minutes , and with Pol θ for 10 minutes at 37°C . Reactions were stopped by addition of 10 μL of formamide buffer ( 80% formamide , 1 mM EDTA , 1 mg/mL Orange G ) , denatured at 95°C for 2 minutes and chilled on ice . The products of the reaction were resolved on a 10% 7M Urea PAGE and FAM labeled DNA was visualized using a Typhoon 9400 scanner ( GE Healthcare ) . Images were analyzed and quantified using ImageQuant software ( Molecular Dynamics ) The polymerase assay was adapted from [8] . To create the 26nt pssDNA substrate with CCGG microhomology , PAGE-purified oligo 5’-CTAAGCTCACAGTG-3’ ( IDT ) was 5’ end-labeled with polynucleotide kinase ( NEB ) and ATP , [γ-32P]- 3000Ci/mmol 10mCi/ml ( Perkin-Elmer ) . The labeled oligo was annealed to an oligo of sequence 5’-CACTGTGAGCTTAGGGTTAGAGCCGG-3’ in STE buffer ( 100 mM NaCl , 10mM Tris-HCl , pH 8 . 0 , 1 mM EDTA ) by heating to 85°C and slowly cooling to room temperature . To create the 33nt pssDNA substrate with TA microhomology , PAGE-purified oligos 5’-GTCTGGGTCAGCAGGG-3’ and 5’-CCCTGCTGACCCAGACCATGATGAAATAACATA-3’ were annealed under identical conditions . Reaction mixtures ( 20 μl ) contained 20 mM Tris–HCl pH 7 . 5 , 4% glycerol , 80 μg/ml bovine serum albumin ( BSA ) , 8 mM MgCl2 , 16 fmol of substrate , 100 μM dNTPs , and 0 . 25–0 . 5 ng of Pol θ . After incubation for 10 min at 37°C , reactions were terminated by adding gel loading buffer ( formamide , 0 . 1% xylene cyanol , 0 . 1% bromophenol blue , 20 mM EDTA ) and boiling . Products were separated by 20% denaturing SDS-PAGE and band intensities were analyzed with a Biorad phosphorimager .
|
Error-prone DNA Polymerase θ ( Pol θ ) plays a conserved role in a mutagenic DNA double-strand break repair mechanism called microhomology-mediated end joining ( MMEJ ) . In many organisms , it also participates in a process crucial to the removal/repair of DNA interstrand crosslinks . The exact mechanism by which Pol θ promotes these processes is unclear , but a clue may lie in its dual-domain structure . While the role of its polymerase domain has been well-studied , the function of its helicase-like domain remains an open question . Here we report an absolute requirement for ATPase activity of the helicase-like domain during interstrand crosslink repair in Drosophila melanogaster . We also find that although end joining frequency does not decrease in ATPase-dead mutants , ATPase activity is critical for generating templated insertions . Using purified Pol θ protein , we show that it can bypass synthetic substrates mimicking interstrand crosslink intermediates and can promote MMEJ-like reactions with partial double-stranded and single-stranded DNA . Together , these data demonstrate a novel function for the helicase-like domain of Pol θ in both interstrand crosslink repair and MMEJ and provide insight into why the dual-domain structure has been conserved throughout evolution .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Conclusion",
"Materials",
"and",
"methods"
] |
[
"invertebrates",
"nucleic",
"acid",
"synthesis",
"enzymes",
"annealing",
"(genetics)",
"dna-binding",
"proteins",
"enzymology",
"animals",
"microhomology-mediated",
"end",
"joining",
"phosphatases",
"animal",
"models",
"drosophila",
"melanogaster",
"model",
"organisms",
"polymerases",
"experimental",
"organism",
"systems",
"dna",
"drosophila",
"dna",
"synthesis",
"chemical",
"synthesis",
"research",
"and",
"analysis",
"methods",
"dna",
"annealing",
"proteins",
"biophysics",
"insects",
"adenosine",
"triphosphatase",
"biosynthetic",
"techniques",
"arthropoda",
"physics",
"biochemistry",
"dna",
"polymerase",
"nucleic",
"acids",
"nucleic",
"acid",
"thermodynamics",
"genetics",
"biology",
"and",
"life",
"sciences",
"dna",
"repair",
"physical",
"sciences",
"organisms"
] |
2017
|
Drosophila DNA polymerase theta utilizes both helicase-like and polymerase domains during microhomology-mediated end joining and interstrand crosslink repair
|
Visual stimuli evoke activity in visual cortical neuronal populations . Neuronal activity can be selectively modulated by particular visual stimulus parameters , such as the direction of a moving bar of light , resulting in well-defined trial averaged tuning properties . However , given any single stimulus parameter , a large number of neurons in visual cortex remain unmodulated , and the role of this untuned population is not well understood . Here , we use two-photon calcium imaging to record , in an unbiased manner , from large populations of layer 2/3 excitatory neurons in mouse primary visual cortex to describe co-varying activity on single trials in neuronal populations consisting of both tuned and untuned neurons . Specifically , we summarize pairwise covariability with an asymmetric partial correlation coefficient , allowing us to analyze the resultant population correlation structure , or functional network , with graph theory . Using the graph neighbors of a neuron , we find that the local population , including both tuned and untuned neurons , are able to predict individual neuron activity on a moment to moment basis , while also recapitulating tuning properties of tuned neurons . Variance explained in total population activity scales with the number of neurons imaged , demonstrating larger sample sizes are required to fully capture local network interactions . We also find that a specific functional triplet motif in the graph results in the best predictions , suggesting a signature of informative correlations in these populations . In summary , we show that unbiased sampling of the local population can explain single trial response variability as well as trial-averaged tuning properties in V1 , and the ability to predict responses is tied to the occurrence of a functional triplet motif .
In the visual system , decades of research have probed stimulus parameters that evoke responses in single neurons and these responses have been generally trial-averaged [1] . These response properties have revealed principles of functional organization in primary visual cortex ( V1 ) , such as orientation columns [2] , and canonical computations , such as divisive normalization [3] . However , responses are variable across trials [4] , making the relationship between perceptual stability and neuronal single-trial stimulus representations unclear [5] . The fluctuations of response strength are not independent across neurons , and this shared variability impacts population level representations of visual stimuli [6 , 7] . Neurons are highly interconnected and connection likelihood is biased toward spatially proximal neurons [8] , suggesting that trial-to-trial response variability may be in part the manifestation of the state of the surrounding neuronal population [9 , 10] . Pairwise interactions within a population can shape information representation [11 , 12 , 13] and can be regulated by top-down influences [14] . Therefore , comprehensive descriptions of stimulus representations in primary sensory cortex require a network perspective . Here , we used two-photon imaging to record from large populations of L2/3 excitatory neurons in mouse V1 to study effects of local population activity on trial-to-trial variability . Understanding the sources and consequences of response variability is necessary to extend theories of sensory computation from the average case to single trials . Perception and behavior take place in real time , after all , so variable responses must be taken into account to understand stimulus representations in cortex . Shared variability in neural responses is commonly quantified by the set of pairwise correlations between neurons , and the structure of these correlations can have constructive or destructive effects on stimulus encoding in populations of neurons [15 , 16] , highlighting the importance of its characterization . Moreover , complex patterns of population activity in retina can be captured by taking into account only neuron firing rates and pairwise correlations [17] . Covariability can also be shaped by cognitive properties such as attention in order to improve perceptual acuity [14] . Whether trial-to-trial variability is harnessed to improve the fidelity of sensory representations , or accounted for when decoding from noisy signals , the properties of response variability have a large impact on neural function . Research on the correlation structure of population activity is still incomplete , however , and can be meaningfully expanded by incorporating a more comprehensive sampling of the network [18] . V1 populations consist of neurons whose activity is not modulated by , or is untuned to , a given stimulus . It is still an open question how this subpopulation contributes to neuron correlations . Two-photon imaging results in a relatively unbiased sampling of spatially proximal neurons including both tuned and untuned subpopulations . Neurons unrelated to a behavioral task can help predict activity in neighboring neurons in hippocampal CA1 [9] . In V1 , it has been shown that untuned neurons can help to decode the orientation of drifting gratings [19] . We investigate how co-fluctuations in the activity within tuned and untuned neurons interacts with responses to drifting grating stimuli . We characterize population activity and correlations between tuned and untuned subpopulations in order to understand the relationship between single-cell response properties and recurrent network dynamics . Traditional noise correlation analyses study covariability independent from stimulus-driven activity [20] . However , untuned neurons have no stimulus modulation , so we use an analogous partial-correlation based method that additionally accounts for population-wide covariability . Fluctuations common across a local population are an important determinant of single-trial responses in mouse V1 [21 , 22] . Capturing this additional variable allows us to study the correlations in the entire unbiased sampling of the population . Additionally , the partial correlation matrices are asymmetric and relatively sparse , and can thus be represented as a weighted , directed graph . Graph theory analysis is used to summarize structure in complex networks and can resolve emergent properties resulting from pairwise relationships . Connectivity patterns , or motifs , in graphs can be characterized within small groups of neurons [23] or across an entire population [24 , 25] . Motifs have proved to be impactful for understanding complex biological processes including transcription networks [26] and spike propagation [23] . More generally , motifs patterns impact information representation in complex systems [27 , 28] and have increasingly been a subject of interest among neuroscientific disciplines . From the graph neighbors of a given neuron , we can accurately predict activity on single trials using a simple , linear model . The local population contains information sufficient to predict trial-to-trial variability and recapitulates average tuning properties . Furthermore , neurons that are well-modeled by the activity of their neighbors have specific signatures of functional connection motifs . Across the entire graph , the most predictive motifs are also the most prevalent , suggesting that this structure is responsible for the overall quality of reconstruction observed . The triplet motif that facilitates predictions of neural activity may have a broad impact on information representation in graphs . Notably , total variance explained in a field of view scales with the number of neurons imaged , suggesting larger sample sizes are required to fully capture local network interactions .
To study interactions and response variability in local , cortical populations ( <800μm diameter imaging plane ) , we imaged L2/3 excitatory neurons ( 72–347 neurons; 25–33 Hz; n = 8 animals; 23 distinct fields of view; Fig 1A ) in mouse V1 during presentation of drifting gratings ( Fig 1B and 1C ) . Square-wave gratings at 12 directions were presented in pseudo-random order for 5 seconds each , interleaved with 3 seconds of mean-luminance matched grey screen . We designed grating stimuli with slightly longer durations than many studies [16 , 29 , 30] for two reasons: first , to allow for the slow decay of the calcium indicator to fall to baseline in order to remove any confounds from the previous grating , and second , to study the sustained response in the population , rather than a transient response to stimulus onset [31] . Mice were awake and allowed to freely run on a linear treadmill . The majority of neurons showed significantly increased activity to one or more gratings over grey screen ( 3023/4535 ) . Of the responsive subpopulation , most neurons were significantly tuned to orientation or direction ( 2073/3023; 540/3023 respectively ) . Neuron tuning was measured by fitting an asymmetric circular Gaussian tuning curve to the trial-averaged mean fluorescence in each grating direction ( Fig 1D ) . These numbers of tuned and untuned neurons are in line with other population studies of awake , mouse V1 [16 , 30] . In subsequent analyses , we pooled all visually responsive neurons without significant direction or orientation tuning into a class of ‘untuned’ neurons , differentiating two distinct subpopulations in V1 by their responsivity to drifting gratings . Single trial responses to gratings showed a high degree of variability , even in strongly tuned neurons , manifesting as occasional strong responses to null-directions and weak or absent responses to preferred directions ( Figs 1C and 2A ) . Tuning curves described average response strength of tuned neurons well ( 0 . 70+/-0 . 20 R2 ) . Responses across single trials in tuned neurons , however , were not well-described by their tuning curves . The mean fluorescence for each direction only explained a small fraction of the total trial-to-trial variance which we calculated by subtracting mean fluorescence in each direction from fluorescence in each trial ( Fig 2B ) . This finding matches earlier results in awake , mouse V1 [30] . We computed the distribution of response strength ( mean fluorescence across a grey or grating presentation ) within single trials , z-scoring to account for neurons with different activity levels . Single trial response distributions were skewed , with most responses weaker than the mean ( Fig 2C ) . Tuned neurons showed slightly stronger responses during gratings , as compared to untuned neurons which had nearly identical response distributions in grey and grating trials . Tuned response distributions were strongly overlapping , however , consistent with the hypothesis that individual neuron activity is not solely driven by tuning properties . To further describe population activity , we computed the time-varying activity during the presentation of a grating and its preceding and following grey presentations . For each trial , we removed neurons that were silent ( defined as no fluorescence change 2*S . D . above baseline ) and computed the time-varying z-scored fluorescence across neurons ( Fig 3A ) . Despite the long duration of stimulus presentations , adaptation effects were minimal in these L2/3 neurons , as tuned neurons showed sustained activity throughout the 5 second stimulus presentation . Untuned neurons have weak modulation to the onset and offset of the stimulus and are equally active during the grey period . However , these effects are small in comparison to variability across trials , as indicated by strong overlap between the activity of the two subpopulations . In the awake animal , running speed is known to strongly influence spike rates [32] , and we similarly observe that periods of high population activity are very likely to occur during periods of running ( Fig 3B , anecdotally in 1B ) . However , mice did not preferentially run during grating or grey presentations ( probability of running 9 . 7+/-7 . 7% during gratings; 10 . 0+/-7 . 4% during greys; p = 0 . 278 paired t-test ) . While we used changes in fluorescence for all other analysis , we expanded our comparison of both subpopulations by estimating spike rates using a spike-inference from calcium fluorescence algorithm [33] . We found that untuned neurons exhibited an identical firing rate distribution to tuned neurons ( Fig 3C ) . Therefore , differences between subpopulation dynamics cannot be explained by differences in firing-rates . Untuned neurons are a large proportion of total neurons , exhibit similar spike rates to tuned neurons , and are likely to contribute to correlation structure in the population . To begin describing how dynamics are affected by stimuli in populations containing tuned and untuned neurons , we first analyzed pairwise correlations during grating and grey presentations . For each pair of neurons , we computed the correlation coefficient between the mean fluorescence ( averaged over time ) in either grating or grey trials . We did not remove signal-dependent responses , nor did we shuffle responses to eliminate simultaneous cofluctuations , therefore , these correlations are a combination of signal and noise correlations . Overall , within-subpopulation correlations are weak ( 0 . 014+/-0 . 027 tuned; 0 . 033+/-0 . 047 untuned ) , and beween-subpopulation activity is slightly anti-correlated ( -0 . 019+/-0 . 017 ) . Comparing mean pairwise correlations across all pairs according to their subpopulation , only within-subpopulation correlations are affected by grating stimuli , while between-subpopulation correlations are unchanged between stimulus and grey conditions ( Fig 4A ) . Tuned neurons show a strong decrease in mean correlations during gratings , as seen in macaques [34] . Conversely , untuned neurons are more strongly correlated during gratings , and yet correlations between tuned and untuned neurons do not change in magnitude between stimulus and grey conditions . Untuned neuron activity is not directly modulated by the stimulus , so changes within this subpopulation most likely reflect changes in activity from the tuned subpopulation propagating through local synaptic connectivity . However , this occurs without a change in mean correlation between tuned and untuned neurons , bringing to question the mechanism involved . The pairwise correlations in tuned neurons during the stimulus are a function of their preferred grating directions , as expected for signal correlations ( Fig 4B ) . Similarly tuned neurons show strong correlations , while orthogonally tuned neurons show negative correlations . This structure is not present during activity in grey periods , however . This is surprising , because if local connectivity underlies correlations in the grey condition , one should expect the structure seen during gratings to remain in part because similarly tuned neurons are more likely to be connected [35] . Overall correlations in populations including untuned neurons begins to reveal properties of local population activity , but in order to study the sources and structure of trial-to-trial shared variability , researchers attempt to remove the stimulus-dependent portion of responses leaving only variability , or ‘noise’ [20] . Correlated fluctuations between the remaining responses are therefore often called ‘noise correlations . ’ However traditional noise correlation analysis was not appropriate in this case for the following three reasons: 1 ) the untuned neuronal subpopulation has no stimulus-dependent response; 2 ) in tuned neurons stimulus-driven response explains only a small portion of overall variability in tuned neurons ( Fig 2B ) , making traditional noise-correlation analysis unsatisfactory for these neurons as well; 3 ) finally , V1 populations in mice , cat , and macaque are characterized by global cofluctuations common to every neuron [22 , 42 , 43] such as covariance driven by running ( Figs 1B and 3B ) . Therefore in order to study pairwise noise correlations within and between subpopulations we used partial-correlation analysis that allowed us to account for stimulus-driven responses and population-wide co-fluctuations in both tuned and untuned neurons . Visual stimuli were presented in 5-minute blocks . Each block of visual stimuli contained three repetitions of each direction in pseudo-random order and corresponding luminance matched grey periods . While the order of grating stimuli were pseudo random in each block the order was maintained between blocks . For each pair of neurons , the average activity across all remaining blocks , other than the block considered at that time , represented the stimulus-dependent responses capturing tuning properties , when present . Additionally , we accounted for the mean within-block population-wide activity of all remaining neurons . Consequently , we were able to compute a partial correlation coefficient in each block between the activity of every pair of neurons , controlling for stimulus responses and population co-activity ( Fig 5A ) . The mean partial-correlation across blocks is taken as the final correlation strength and entered as an edge weight into the functional connectivity matrix . We also added directionality to the partial-correlation by examining the mean cross-correlogram across blocks for the neuron pair . If the peak value occurs at lag 0 ( i . e . within the same imaging frame ) , the edge was bidirectional , otherwise the edge was in the direction of positive lag . Lags greater than 500ms were thrown out and correlations set to zero . This resulted in a functional network described by a directed weight matrix . Though we interpret these partial-correlations as equivalent to noise correlations , the correlation matrices are different from traditional noise correlations in two important ways . First , many pairs of neuron correlations are exactly zero ( 51 . 7+/-7 . 2% ) , and second , non-zero correlations are asymmetric ( Fig 5B ) . This allows us to analyze these matrices from a graph-theoretic perspective representing the functional partial correlations as a weighted , directed graph . Graph representations of pairwise edges allow us to analyze population-wide statistical features of the correlation structure . Overall , partial-correlation strengths , synonymous with edge weights , were long-tailed , centered slightly above zero ( Fig 5C ) , similar to noise-correlations observed elsewhere [36] . As expected , tuned edge weights were similar to signal correlations [36 , 37] , with similarly tuned neurons having larger edge weights on average ( Fig 5D ) . We next analyzed the partial-correlations within and between tuned and untuned subpopulations . The graphs exhibit dense correlations with varying strengths among subpopulations ( Fig 6A ) . To analyze biases in edge strengths , we thresholded the matrices at increasing values , setting all edges below each threshold to zero . Among all edges , within-tuned connections are more likely , while within-untuned and between-subpopulations are less common ( Fig 6B; within-tuned 54 . 2+/-8 . 6% , within-untuned 44 . 1+/-6 . 0% between 43 . 9+/-6 . 2% ) . Between-subpopulation connections remain the least likely at higher thresholds , but among the strongest edges , within-untuned connections are the most likely . We then recomputed partial-correlation matrices , exclusively using frames during the grey condition or during grating condition to see how correlation strengths were affected , despite controlling for mean stimulus-dependent activity . The two resulting matrices did not have significantly different edge strength ( stimulus 0 . 12+/- . 02; grey 0 . 12+/- . 03; p = 0 . 75 one-way ANOVA ) . However , probability of connection was higher within grey frames ( grating 45 . 3+/-6 . 5%; grey 58 . 0+/-8 . 2%; p = 5 . 8*10^-7 one-way ANOVA ) , indicating a higher degree of interconnectivity in the population in the absence of stimuli . When analyzing magnitude of edge strength differences between grey and grating matrices , we found that all connection types changed similarly with a mean near zero ( Fig 6C; within-tuned -0 . 011+/-0 . 097; within-untuned -0 . 005+/-0 . 103; between -0 . 008+/-0 . 093 ) . Since population dynamics are partly constrained by synaptic connectivity [38] , we evaluated whether there was a spatial component to the weight matrices . Edge probability fell monotonically with distance between neurons , similar to traditional noise correlations [36] and synaptic connections [8 , 39] . Notably , the decay in connection probability is slower within tuned neurons compared to within untuned neurons ( Fig 6D ) . Between-population decay lies in the middle . If the spatial structure of untuned correlations is exclusively driven by local connectivity , then bottom-up sensory drive is the most likely source for the longer range of functional correlations among tuned neurons . Furthermore , the mean lag ( delay of the cross-correlogram peak used to determine edge directionality ) is greater over longer distances and accumulated evenly across subpopulation connection type ( Fig 6E ) . Assuming a linear change in lag over space , these data suggest the speed of functional correlations in this preparation is roughly 25 mm/s . Despite allowing for directional edges , nearly half of all edges were bidirectional ( 42 . 5+/-16 . 8% ) . As a function of edge strength , bidirectional edges are more prominent within-subpopulations , and are strongly biased toward the strongest weights ( Fig 6F ) . Thresholding at increasing edge strengths sparsifies the matrices , so we normalized the bidirectional edge counts by the probability of bidirectional edges assuming connections are placed randomly . Among all edges , bidirectional edges occur less often than random . The strongest edges , however , are roughly 5 times more likely to be bidirectional than random . Overall , zero-lag connections are less frequent between tuned and untuned neurons , suggesting a transmission or propagation of information , rather than simultaneous representation of the same information between subpopulations . Because trial-averaged tuning poorly captures trial-to-trial responses we asked whether information in the local population could better explain V1 trial-to-trial response variability . Pairwise correlations have been shown to capture a significant portion of the complexity in population activity [17] . We tested whether the activity of a neuron could be modeled from the activity of its neighbors that had a non-zero correlation . Since correlation coefficients capture the linear relationship between neuron coactivity , we used a simple linear combination of the input neuron activity with partial-correlation coefficients ( edge strengths ) as weights . This model gave a time-varying prediction of the fluorescence of a given neuron , which was then rescaled by an offset and a gain to account for different numbers of input neurons ( Fig 7A ) . In many cases , this model resulted in highly accurate predictions of activity . Mean squared error of the reconstruction was small , and often near optimal compared to weights estimated by regression ( Fig 7B ) . Tuned neurons were slightly better-modeled on average than untuned neurons ( tuned MSE 0 . 014 median , 0 . 037 inter-quartile range; untuned MSE 0 . 017 median , 0 . 046 inter-quartile range ) , possibly because of the additional stimulus-dependent information captured by other tuned inputs . The ratio of MSE to mean-squared fluorescence from the true trace , subtracted from one , gives the percent variance explained of the model for each neuron . The optimal reconstructions from regression explained 77 . 2+/-15 . 2% of the variance in activity across all neurons ( n = 4531 ) . Using partial-correlation coefficients as weights performed near this upper-bound , at 65 . 8+/-17 . 1% . We tested the robustness of the partial-correlation based predictions modeling 5 minutes of fluorescence change , corresponding to one block , using a functional network built using 45 min , corresponding to the nine other blocks , of non-overlapping recordings to recompute the weights . We note that one block , i . e . five minutes of stimulus epochs , corresponded to three stimuli at each of the 12 directions and interleaved grey periods . We then tested our predictions on the left-out 5 minute dataset ( repeated to leave out each dataset once ) . On this cross-validation procedure , the average performance on the left out epoch reached 55 . 4+/-18 . 9% variance explained . We selectively removed either tuned or untuned input neurons to evaluate the relative contributions to model predictions . The accuracy of prediction decreased more when removing within-subpopulation inputs as compared to removing between-subpopulation inputs , but the decrease is small and distributions across neurons strongly overlap ( Fig 7C ) . Within-subpopulation inputs are more frequent , however , so removing a larger fraction of inputs should be expected to have a larger impact on our ability to model neuronal fluorescence . On a single neuron basis , correlations between tuned and untuned neurons both contribute to predicting time-varying activity . We next asked whether our model based on local population activity could also predict trial-averaged tuning properties . For tuned neurons , we used the modeled fluorescence traces to recompute the mean fluorescence in each grating direction . The average responses in direction space were added together to obtain a mean tuning vector . The modeled fluorescence had very similar tuning properties to the data , as measured by the cosine similarity between model and data tuning vectors ( Fig 7D ) . The modeled activity was constructed from a neuron’s input edges; we also asked if a neuron’s outgoing edges were related to tuning properties , which may indicate neurons that are good at decoding the stimulus ( strong tuning ) were also good at decoding activity in its neighboring neurons ( strong outgoing edges ) . However , there was nearly zero correlation between the tuning strength of tuned neurons and its mean outgoing connection strength ( r = 0 . 04 , p = 0 . 02 ) . Local population activity therefore contains information to capture trial-to-trial variability as well as trial-average stimulus response properties , but a neuron’s dependence on the stimulus does not affect its local population correlations . To investigate how individual neurons inputs contribute to reconstruction of activity , we selectively removed input neurons based on their edge strength and measured the increase in reconstruction error ( MSE ) , normalized by total mean squared fluorescence in the neuron . As expected , the strongest edges contribute the most to activity reconstruction with the strongest 25% of weights containing over half of the reconstruction capability , whereas removing half of the weakest edges had no discernible effect ( Fig 8A ) . Interestingly , randomly removing edges does not linearly reduce prediction performance . This suggests a level of redundancy in the predictive information of input neuron activity . Accounting for the cumulative weight removed , we still found worse performance when removing the strongest edges first , and removing half of the total weight using only the weakest edges has minimal effect on reconstruction error ( Fig 8B ) . Normalizing for the total weight removed reveals a nearly-linear increase in reconstruction error when removing the strongest weights , suggesting that these neurons may hold independent information from the remaining input pool . To address the possibility of synergistic or redundant information between input neurons , we analyzed connections between triplets of neurons termed ‘motifs’ in graph theory literature . Triplet connection motifs are built up from pairwise connectivity and collectively represent higher-order connectivity patterns that cannot be captured by individual edges , and can have strong implications for computation and information propagation within graphs [27] . We looked at the clustering of triplet motifs for each type of triangle that can be formed with directed edges ( Fig 8C ) [40] . Clustering is a measure of how many motifs are present among a neuron’s neighbors given its input and output connections . In comparison to Erdos-Reyni graphs , which have uniform , low levels of clustering across motifs , cycles of edges in the data are less clustered on average , with all other motifs showing elevated clustering . The middle-man motif shows the strongest clustering . Similar results have been found in activity generated by simulated and ex vivo neural networks , although fan-in clustering was higher than middle-man [23] . To address the possibility that triplet motifs are responsible for explaining more of the neuronal response than pairwise edges alone , we analyzed the relationship between motif clustering in single neurons and the variance explained by the predicted activity . Because the magnitude of clustering is different across motifs , we first z-scored clustering coefficients , then computed the mean across neurons with different levels of variance explained . Neurons with the best reconstruction showed higher clustering of middle-man motifs and lower cycle clustering , relative to other neurons in the population ( Fig 8D ) . Total clustering , as well as fan-in and fan-out , had weak , positive correlation with variance explained . Together , these relationships map directly onto the overall prevalence of the graph motifs , suggesting that the graph structure has an important function in representing population information . Because the motif with the highest mean clustering was also most indicative of model performance , we hypothesized that the partial-correlation structure might underlie our ability to predict neuron activity from its local population . Interestingly , across fields of view , the total variance explained in the population increased with number of neurons imaged ( Fig 8E; r = 0 . 58 ) . To compute population variance explained , we took the sum of all squared prediction errors ( residuals ) across neurons in a field of view , divided by the sum of the squared population fluorescence activity , and subtracted from one . This improvement of variance explained across the population with more recorded neurons suggests that , in addition to motifs , total neurons sampled in the population determines our ability to measure a neurons’ single trial dependence on its local population . Moreover , the linear trend had no discernible plateau , so representing the local population may require recording from more than 300 neurons simultaneously .
We sought to describe the interrelationships within local populations of V1 neurons , including tuned and untuned neurons , as they relate to single-trial responses to grating stimuli . We used two-photon imaging to record from L2/3 excitatory populations constitutively expressing calcium indicator GCaMP6s . Neurons with similar response properties showed stronger co-variability on average , but across the entire population there was a broad distribution of correlations driven , presumably , by a confluence of sensory drive and activity in the local population . The functional correlations in the recorded populations were sufficient to predict activity in individual neurons , far surpassing predictions from tuning characteristics alone . The dependence on local population activity reinforces theories of layer 2/3 acting under strong modulation with sparse activity and weaker dependence on sensory drive than layer 4 [41] . We summarized the structure of correlations as directed , sparse matrices in order to analyze population dynamics from a graph theoretic perspective . We demonstrate that a simple population model capable of predicting single-trial neural responses is also able to accurately predict trial-averaged tuning responses , a key feature of V1 function . We found that the prevalence a specific triplet connectivity motif , built up from pairwise correlations , corresponded with our ability to predict activity on single-trials . This result could not have been observed from only studying pairwise correlations and motivates the continued use of graph theory to study neural population dynamics . The single-neuron response properties in our data replicate imaging and electrophysiological results in awake recordings of V1 including proportion of tuned neurons [29 , 30 , 36] , and variance explained by the mean tuning curve [30] . Both tuned and untuned neurons have low firing rates on average , though estimation of firing rates from calcium imaging is not always straightforward . We note that single neuron properties , including firing rate and trial-to-trial variance , can change substantially between animal models and in different states of anesthesia [42 , 43] . We found that the magnitude of signal correlations between tuned and untuned subpopulations do not change between grey and grating stimulus conditions , while within subpopulation correlations do change . Perhaps ubiquitous changes would be more expected , or exclusive changes among neurons modulated by the stimulus . Within subpopulation correlations change in opposite directions , however , and could serve as a mechanism to balance changes between the subpopulations . We found that the spatial organization of the network is a strong determinant of correlation structure with correlations decaying over distance , consistent with paired patch clamp recordings [8] and the correlation structure of activity in isolated preparations [44] . Correlation matrices were computed as an asymmetric partial correlation coefficient , accounting for stimulus and population effects , to allow the incorporation of untuned neurons into traditional noise correlation analyses . While this approach differs from standard noise correlation estimates , the magnitude of correlations and dependence on tuning similarity replicate previous results measuring noise correlations [36] . For this reason , we interpret the partial correlations as measuring trial-to-trial covariability , and as being broadly equivalent to noise correlations . Tuned neurons , which combine feedforward sensory inputs with recurrent inputs , show a slower spatial decay of correlational values than untuned neurons , the latter of which presumably are driven more exclusively by recurrent , local inputs . These data show than subdividing the population by their response to grating stimuli can differentiate rates of spatial decay within the network . Noise correlations are hypothesized to be driven in part by shared synaptic input [45] . Spatial dependencies of feedforward and recurrent connectivity can qualitatively change the spiking activity in network models [46] . The relatively small diameter of the fields of view imaged here ( <1mm vs 10mm ) does not allow us to differentiate between the two modes of activity predicted by these models . These data suggest , however , that functional recurrent correlations have less spatial extent than feed-forward functional correlations . To our knowledge , analogous synaptic connectivity estimates do not exist for mouse V1 . In addition to spatial structure of correlations , we found an increase in mean temporal delay of correlations over distance . This delay spans roughly 20–50 ms in our field of view and is therefore likely to reflect timescales of functional correlations rather than monosynaptic transmission delays . We have previously found that functional correlations are indicative of synaptic connections , in some cases , when considering synaptic integration rather than synaptic delays [23] . The implied speed of this delay accumulation is much faster than propagating LFP waves observed in macaque M1 [47] after normalizing for total cortical surface area [48] . The propagation of beta-oscillations and temporal delays of functional correlations likely have different underlying mechanisms , which may explain the different speeds . We were able to extract a substantial amount of predictive information from the local population using a simple , linear model . Other approaches have successfully predicted single-trial responses from ongoing population activity at a larger spatial scale , averaging over many neurons in a population [49] . In contrast , we use a large , unbiased sampling of the local population with single cell resolution to predict variability on single trials . Alternative models incorporating known neuronal nonlinearities [50] , or more sophisticated or biologically-relevant predictive models [51] , may have improved performance . Consequently , the increase in total variance explained with population size may show different trends with alternative models . Nevertheless , we chose this modeling approach to maintain ease of interpretation and utilize the linear correlation coefficients estimated from the data in a straightforward manner . The linear models had good performance , and allowed us to remove input neurons without laboriously re-training the models . This straightforward modeling approach may not capture all of the information present in the local population , but its performance sets a lower bound on predictive information in the population . A small subset of neurons have high mean-squared error of predictions , and we speculate that we have not sampled enough of the population to predict these neurons given that total variance explained scales with population size . Neuron activity was only predicted from the in-degree neurons , rather than the entire population , reducing the number of parameters by 52+/-22% across neurons , though in either case , parameters scale by order N . These results substantiate previous models showing that the collection of pairwise relationships in large populations can explain complex activity patterns [17] . The predictions of single-trial activity that are obtained from local activity are , on one hand , a description of inter-neuronal dependencies in recurrent networks , and on the other hand , are capable of recapitulating trial-averaged tuning properties , extending these dependencies to stimulus encoding . We further explored this idea by identifying a specific motif of pairwise correlations underlying accurate predictions of neuron activity on single-trials . The middle-man motif underlying the most accurate predictions is a specific pattern of pairwise correlations and represents a higher order feature of covariability . Local populations have been shown to contain such high-order correlations , but were not seen at larger spatial scales [52 , 53]; further experimentation is necessary to test whether functional triplet motifs occur across thousands of microns . The finding that middle-man motifs underlie the best predictions of activity may be initially surprising from the connection pattern of the motif . Compared to the fan-in motif , which has two input connections , the middle-man has one input and one output connection , and acts to route connections from its input to its target . However , the motifs were quantified using the clustering coefficient , which normalizes motif count by the total number of possible motifs ( i . e . high fan-in clustering doesn’t correspond to high in-degree ) . The functional significance of any given triangle motif clustering is unknown and is likely dependent on the underlying system represented by the graph . In a neural network , middle-man clustering may indicate a hub-like property common in the brain [54] , having a combination of convergent and divergent correlations . The cycle motif similarly has a combination of convergence and divergence , yet its clustering has a negative effect on prediction accuracy . The difference in the two motifs is the direction of connections between neighbors . In these networks , it may be that cycle clustering reflects recurrent , redundant correlations reducing our ability to predict activity from the population . Conversely , the middle-man motif is isometric to fan-in and fan-out motifs , and could allow for transfer of information between motifs , and in turn increase predictive power . Finally , this result demonstrates that in addition to providing insights into synaptic mechanisms underlying dynamics [23] , network science can also provide insights into predictions of single trial neural responses as we have demonstrated here . V1 is the first stage in which visual information is encoded in densely recurrent cortical networks . Thus , in order to understand activity patterns in V1 , one must take into account visual drive as well as local network activity . We have provided a quantitative comparison of the relative influences of these two factors in awake , ambulating mice . Local network effects dominate on single trials , highlighting the importance of investigating cortical computation from a population perspective in order to understand how information is encoded in single trials . Populations of neurons exhibit emergent properties beyond the sum of their individual neurons and connections , and we use the analytic framework of graph theory to begin unraveling this emergent structure .
All procedures were performed in accordance and approved by the Institutional Animal Care and Use Committee at the University of Chicago . Data was collected from C57BL/6J mice of either sex ( n = 4 female; 4 male ) expressing transgene Tg ( Thy1-GCaMP6s ) GP4 . 12Dkim ( Jackson Laboratory ) between ages P84 –P191 . After induction of anesthesia with isoflurane ( induction at 4% , maintenance at 1–1 . 5% ) , a 3mm diameter cranial window was implanted above V1 by stereotaxic coordinates and cemented in place alongside a custom titanium headbar . Mice recovered for at least 8 days before intrinsic signal imaging to identify V1 followed by two-photon data collection . Boundaries of V1 were identified by intrinsic signal imaging post-surgery [55] ( Fig 1A , left] . Mice were anesthetized with isoflurane and head-clamped under a CCD camera ( Qimaging Retiga-SRV ) . A vertical white-bar stimulus ( 100% contrast , 0 . 125Hz ) was repeatedly presented on an LED monitor ( AOC G2460 ) approximately 20cm from the contralateral eye while capturing cortical reflectance under 625nm illumination . The retinotopic mapping of V1 was then estimated at each pixel from the phase of peak reflectance driven by increases in activity-dependent blood flow . Two-photon imaging was collected from awake , head-fixed mice on a linear treadmill . Running speed was measured with a rotary encoder attached to the treadmill axle . A L2/3 field of view ( roughly 800μm diameter ) in V1 was identified with galvanometer-mirror raster scanning ( Cambridge Technologies; 6215H ) . Once a suitable field of view was found , raster scans ( 1Hz ) were continuously acquired for roughly 10 minutes alongside visual stimulation . Neurons ( n = 72–347 per field of view ) were then automatically identified using custom image processing software for imaging during visual stimulation using Heuristically Optimal Path Scanning [56] at 25–33 Hz ) . All imaging was performed at 910nm ( Coherent; Chameleon Ultra ) with a 20X 1 . 1NA Olympus objective and GaAsP PMT ( Hamamatsu; H10770A-40 ) . Field of view size was estimated by fitting circles to a single raster scan of 15μm fluorescent microbeads and used for each dataset , though true field of view size may vary up to 8% across datasets from realignment of laser beam path . Drifting grating stimuli were presented on an ASUS VG248QE , 20cm from the contralateral eye at 60Hz; 60cd/m2 . The mean luminance was measured and gamma correction was performed and confirmed using a luminance meter . Square-wave gratings were shown at 80% contrast , 2Hz , 0 . 04 cyc/deg at 12 evenly spaced directions . Gratings were presented for 5s , interleaved with 3s mean-luminance grey screen . Three repetitions of each orientation were presented in a pseudo-random order , resulting in a roughly 5min stimulus movie . The grating order was preserved between movie presentations , and mice were shown 8–11 repetitions of the movie ( 24–33 repetitions of each direction ) . Stimulus presentation was monitored with a photodiode ( Thorlabs ) and synchronized with running speed and imaging frames at 2kHz . For each neuron , baseline fluorescence was estimated from raw fluorescence by thresholding to eliminate spike-induced fluorescence transients and smoothed with a 4th-order , 81-point Savitzky-Golay filter . Fluorescence time-series were then normalized to percent change from baseline ( dF/F0 ) using this time-varying baseline . Spike inference from fluorescence traces was performed using the OASIS algorithm [33] implemented using software made freely available ( github . com/j-friedrich/OASIS ) . Inference outputs probability of spiking at each time point . As commonly done [57] , probabilities were thresholded to obtain a binary spike train . Neurons were classified as visually responsive if the mean response to any grating was significantly greater than the mean response across all grey periods by Dunnett-corrected one-way ANOVA ( alpha = 0 . 01 ) . In these analyses , each trial response is the mean fluorescence across the entire grating presentation ( 5000ms ) , or the last half of the grey presentation ( 1500ms ) to allow for fluorescence from grating responses to decay . Responsive neurons were then tested for statistically significant orientation- or direction-tuned responses according to the trial vectors in orientation or direction spaces ( [58] for more detailed methods] . For significantly tuned responses , tuning curves were then fit with an asymmetric-circular Gaussian to significantly tuned neurons . Tuning curve parameters ( baseline , tuning width , peak amplitudes , and preferred direction ) were fit repeatedly using randomized initial conditions . The parameter set that minimized mean-squared error was maintained . For each neuron , the mean responses in each trial ( using the same time windows as for tuning classification ) for either stimulus or grey trials were used as response vectors . Typically , these response vectors had 360 elements ( 12 directions , 30 trials each ) . The correlation coefficient between each pair of these vectors was used to compute a pairwise correlation matrix for the grating and grey conditions . We did not shuffle responses , therefore these values measure the combination of signal and noise correlations . For each pair of neurons , pairwise-correlation was computed as the mean partial correlation between their fluorescence across movies while accounting for three variables . This was computed with a built-in MATLAB function ( partialcorr . m ) . We computed these correlations on time-varying traces of activity , rather than time-averaged trial activity as done in previous correlation analysis . Controlling for a single variable can be computed with the following equation , where rx , y is the correlation coefficient between time-varying fluorescence traces x ( t ) , y ( t ) , and rx , y|z denotes the partial correlation between x ( t ) and y ( t ) , controlling for z ( t ) : rx , y∨z=rx , y−rx , zry , z1−rx , z21−ry , z2 Here , x ( t ) is the fluorescence trace of the ‘input’ neuron , y ( t ) is the fluorescence trace of the ‘output’ neuron , and z ( t ) represents the three control fluorescence traces . Controlling for more than one variable can be achieved by successive iterations of this procedure . This was computed for each movie , or 5-minute block of grating presentations that was repeated in each experiment . The first two control variables are the mean response of the two neurons in all other movies , accounting for the stimulus-driven responses . This is similar to normalizing by the mean response as in traditional noise correlation estimates . The third control variable is the within-movie mean fluorescence of all other neurons and was included to control for population-wide covariability , for example running speed effects . Furthermore , the cross-correlogram between the two neurons’ fluorescence traces , averaged across movies , was used to compute directionality of the correlation . The time-lag of the cross-correlogram global maximum determined the direction and lag of the edge . If the lag was zero , the correlation was bidirectional . If the lag was greater than 500ms ( roughly 14 imaging frames ) , no edge was included . The partial-correlation matrix could equivalently be analyzed as a directed , weighted graph . Open source software ( Gephi ) was used for visualization , with node layout determined by the Yifan-Hu algorithm and tuned by hand . Edge weights less than 0 . 05 were set to zero for visualization clarity . Erdos-Reyni ( ER ) null graphs were generated for each graph to match the mean connection probability . The mean directed clustering coefficient across nodes was calculated across 50 ER graphs and averaged for comparison with data . Clustering coefficients were computed with binary matrices ( nonzero weights set to one ) . To model neuron responses , we used a linear weighting of the fluorescence of every in-degree using the weights in the partial-correlation matrix . At each time point , a weighted sum was calculated , resulting in a time-varying predicted fluorescence trace . Because different numbers of input neurons and varying calcium transient amplitudes , the modeled trace was then fit with a linear offset and a gain to minimize mean-squared error with the true fluorescence trace . These two parameters were not changed when input neurons were removed ( 7C , 8A , 8B ) . For tuned neurons , we also recomputed the trial-averaged tuning response of modeled activity . The fluorescence over a grating presentation was averaged , then trials were averaged over directions to obtain a mean direction-response . The sum of these vectors in direction space gave the model-estimated tuning vector , and the cosine similarity with the data-derived tuning vector was used to quantify the reconstruction of the tuning properties . Cosine similarity was computed in direction or orientation space according to each neuron’s tuning properties . To compute total population variance explained , modeled traces were subtracted from the data traces to obtain residuals , and the ratio of total sum of squares across neurons were subtracted from one as 1−∑i∑t ( Xi ( t ) −X˜i ( t ) ) 2∑i∑t ( Xi ( t ) ) 2 where Xi ( t ) is the time-varying fluorescence trace of neuron i , and Xi ( t ) tilde is the predicted trace . Optimal weights for all incoming edges were computed for each neuron by LASSO regression as minβ , β0 ( 12N∑i=1N ( yi−β0−xiTβ ) 2+λ∑j=1p|βj| ) For weights β , and offset β0 , with modeled neuron fluorescence as y and input neuron activity as x . Weight estimation and 5-fold cross validation to estimate MSE standard error was performed with MATLAB R2016a implementation . The maximum regularization parameter ( λ ) whose mean-squared error did not exceed the standard error of the minimum MSE was used to find the set of optimal weights .
|
V1 populations have historically been characterized by single cell response properties and pairwise co-variability . Many cells , however , do not show obvious dependencies to a given stimulus or behavioral task , and have consequently gone unanalyzed . We densely record from large V1 populations to measure how trial-to-trial response variability relates to these previously understudied neurons . We find that individual neurons , regardless of response properties , are inextricably dependent on the population in which they are embedded . Specifically , patterns of correlations between groups of neurons , allow us to predict moment to moment activity in individual neurons . Only by studying large , local , populations simultaneously were we able to find an emergent property of this information . These results imply that understanding how the visual system operates with substantial trial-to-trial variability will necessitate a network perspective that accounts for both visual stimuli and activity in the local population .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"social",
"sciences",
"vertebrates",
"neuroscience",
"animals",
"mammals",
"primates",
"mathematics",
"network",
"analysis",
"computational",
"neuroscience",
"vision",
"neuroimaging",
"neuronal",
"tuning",
"old",
"world",
"monkeys",
"research",
"and",
"analysis",
"methods",
"computer",
"and",
"information",
"sciences",
"imaging",
"techniques",
"network",
"motifs",
"monkeys",
"animal",
"cells",
"graph",
"theory",
"macaque",
"cellular",
"neuroscience",
"psychology",
"eukaryota",
"cell",
"biology",
"single",
"neuron",
"function",
"neurons",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"sensory",
"perception",
"computational",
"biology",
"physical",
"sciences",
"amniotes",
"organisms"
] |
2018
|
Functional triplet motifs underlie accurate predictions of single-trial responses in populations of tuned and untuned V1 neurons
|
Kaposi’s sarcoma ( KS ) is a highly prevalent cancer in AIDS patients , especially in sub-Saharan Africa . Kaposi’s sarcoma-associated herpesvirus ( KSHV ) is the etiological agent of KS and other cancers like Primary Effusion Lymphoma ( PEL ) . In KS and PEL , all tumors harbor latent KSHV episomes and express latency-associated viral proteins and microRNAs ( miRNAs ) . The exact molecular mechanisms by which latent KSHV drives tumorigenesis are not completely understood . Recent developments have highlighted the importance of aberrant long non-coding RNA ( lncRNA ) expression in cancer . Deregulation of lncRNAs by miRNAs is a newly described phenomenon . We hypothesized that KSHV-encoded miRNAs deregulate human lncRNAs to drive tumorigenesis . We performed lncRNA expression profiling of endothelial cells infected with wt and miRNA-deleted KSHV and identified 126 lncRNAs as putative viral miRNA targets . Here we show that KSHV deregulates host lncRNAs in both a miRNA-dependent fashion by direct interaction and in a miRNA-independent fashion through latency-associated proteins . Several lncRNAs that were previously implicated in cancer , including MEG3 , ANRIL and UCA1 , are deregulated by KSHV . Our results also demonstrate that KSHV-mediated UCA1 deregulation contributes to increased proliferation and migration of endothelial cells .
Kaposi’s sarcoma-associated herpesvirus ( KSHV ) is an opportunistic human oncovirus , which causes Kaposi’s sarcoma ( KS ) , Primary Effusion lymphoma ( PEL ) and Multicentric Castleman’s disease ( MCD ) in immunocompromised individuals , primarily AIDS patients and organ-transplant recipients [1] . KSHV uses the lytic mode of replication for spread of infection , and latency for persistence in the host . All tumor cells isolated from KS patients test positive for latent viral episomes [1] . Latent KSHV expresses only 10% of its 140-kb dsDNA genome , encoding primarily four latency proteins ( Kaposin , vFLIP , vCyclin and LANA ) and 25 mature miRNAs [1] . miRNAs are 21–23 nt long non-coding RNAs that recognize target mRNAs using 7 bp ‘seed sequences’ and silence them ( see [2] for review ) . To identify the means by which KSHV causes tumors , KSHV latency proteins and miRNAs have been studied extensively [1] . Ribonomics approaches to identify targets of KSHV miRNAs have focused exclusively on mRNAs [3 , 4] . Recently , lncRNAs have emerged as important regulatory molecules in cancer [5] . LncRNAs play a variety of regulatory roles in both the cytoplasm and nucleus [6 , 7] . This group includes all RNA molecules longer than 200 nt with no apparent coding potential , and they have diverse functions ranging from acting as a scaffold , sponge/decoy or guide aiding in cell-signaling [6 , 8] . Owing to their diversity , over 95% of the lncRNAs remain uncharacterized . Disease association is a starting point for identifying and characterizing lncRNAs with important regulatory roles . Using this approach with different cancer types , oncogenic lncRNAs such as MALAT-1 , ANRIL , UCA1 , and tumor suppressor lncRNAs like Gas-5 and MEG3 have been functionally characterized [5] . Another important group of disease-relevant lncRNAs includes those involved in the innate immune response following viral or bacterial infections [9] . A few studies have addressed the roles of host lncRNAs during viral infections , for example HULC ( Hepatitis-B ) and NRON ( HIV ) [10] . However , the question of whether viruses manipulate specific host lncRNAs to their advantage remains largely unexplored . Understanding deregulation of specific host lncRNAs , especially cancer-related lncRNAs by persistent oncoviruses , such as the γ-herpesviruses , would shed light on how these viruses drive oncogenesis . Regulatory cross-talk is known to occur between miRNAs and lncRNAs , at multiple levels . LncRNAs like BIC1 and H19 act as precursors for miRNAs [11 , 12] and lncRNAs such as HULC and CDR1-AS act as sponges for miRNAs [13 , 14] . Conversely , human miRNA miR-9 represses the expression levels of the lncRNA MALAT1 [15] . Work from the Steitz laboratory demonstrated that the viral lncRNAs HSUR1 and HSUR2 , encoded by Herpesvirus Saimiri , act as sponges for cellular miR-16 , miR-142-3p and miR-27 and thereby silence some of these miRNAs in T-lymphocytes , suggesting that γ-herpesviruses can utilize virus lncRNAs to target host miRNAs[16] . Conversely , whether herpesvirus miRNAs can target and downregulate host lncRNAs remains an open question . In this study , we demonstrate that latent KSHV infection of endothelial cells alters the host lncRNA profile . We provide evidence that KSHV deregulates hundreds of host lncRNAs including many cancer-associated lncRNAs such as UCA1 , ANRIL and MEG3 in both a miRNA dependent and independent manner . Furthermore , KSHV appears to manipulate the host lncRNAs to favor proliferation and migration of latently infected endothelial cells .
Previously , we identified the mRNA targetome of viral miRNAs in PEL cells by High Throughput Sequencing-Crosslinking Immuno Precipitation ( HITS-CLIP ) analysis of the Ago protein [3] . The PEL cell lines we studied were BC-3 and BCBL-1 , which are KSHV positive B-cell lines . We reanalyzed the HITS-CLIP data for enriched lncRNAs and compared our results with a similar reinvestigation of Ago PAR-CLIP data from lymphoblastoid cell lines infected with Epstein-Barr Virus ( EBV ) [17] , a related γ-herpesvirus that causes cancer . We found that approximately 357 and 750 lncRNAs were a part of the KSHV and EBV miRNA targetome , respectively , and 64 lncRNAs were potentially targeted by miRNAs from both viruses ( S1 Table ) . We aimed to determine the effect of latent KSHV infection on the lncRNA expression profile of endothelial cells and specifically question whether KSHV encoded miRNAs targeted endothelial lncRNAs . To address these questions , we used Telomerase Immortalized Vein Endothelial ( TIVE ) cells , an in vitro model system to study KS [18] . We performed lncRNA expression profiling on latently infected TIVE cells harboring either the wt-KSHV or Δcluster-KSHV [19 , 20] , in which a region containing 10 of the 12 miRNA genes is deleted , and used the lncRNA profile of mock-infected TIVE cells as reference . The KSHV latency-associated region of the wt and mutant bacmid backbones used for this experiment is shown in Fig 1A . The profiling analysis revealed that wt-KSHV and Δcluster-KSHV infections deregulate 858 and 2372 host lncRNAs , respectively ( Table 1 ) , indicating that latent KSHV infection globally affects lncRNA expression . The higher count of deregulated lncRNAs in Δcluster-KSHV infection is likely due to increased spontaneous reactivation rate in the absence of viral miRNAs [20 , 21] . The differentially expressed lncRNAs are listed in S2 Table . We grouped the deregulated lncRNAs into three categories based on a cut-off of fold change ≥ 2 . 0: upregulated , downregulated and rescued . We defined rescued genes as those that were downregulated in wt-KSHV-infected cells compared to mock , and were upregulated in Δcluster-KSHV-infected cells compared to wt-infected cells . We validated using qRT-PCR two downregulated lncRNAs , two upregulated lncRNAs , and three rescued lncRNAs that were identified from the microarray analysis ( S1 Fig ) . We identified 126 candidates in the rescued category , which are putative direct targets of viral miRNAs ( Fig 1B ) . Based on lncRNA localization data from HUVEC cells [22] , at least 9 of the 126 putative lncRNA targets of viral miRNAs we identified are exclusively nuclear localized , and 32 of them are partially nuclear localized . It is important to note that the localization information was available for only 72 out of the 126 rescued lncRNAs . Similarly , the 357 lncRNAs identified from Ago HITS-CLIP of PEL cells include nuclear resident lncRNAs such as ANRIL ( CDKN2B-AS1 ) and MALAT-1 . Moreover , several of the uncharacterized candidates of these 357 lncRNAs may be nuclear localized . miRNA-mediated regulation of nuclear localized lncRNAs seemed paradoxical at the outset , as mature miRNAs and RISCs including the Ago family proteins are believed to reside and function in the cytoplasm . Recently , several groups showed that Ago-2 complexes can be present in the nuclei of different cell types [23 , 24] . Moreover , studies in Hodgkin’s lymphoma lines identified that several lncRNAs co-isolate with Ago protein [25] . To determine whether KSHV miRNAs could regulate nuclear lncRNAs , we investigated the nuclear/cytoplasmic distribution of viral miRNAs and Ago-2 in PEL cells . We fractionated BCBL-1 cells into nucleus and cytoplasm and analyzed the distribution of KSHV miRNAs using stem-loop RT-qPCR , which amplifies mature miRNAs but not their precursors . Mature KSHV miRNAs were found in both the cytoplasmic and nuclear fraction ( Fig 2A ) . It is important to note that a cellular miRNA hsa-miR-16 is also distributed between the nucleus and the cytoplasm ( Fig 2A ) , and such partial nuclear localization of mature miRNAs has been previously reported in other cell lines [26 , 27] . We probed the fractions for Ago-2 using western blotting ( Fig 2B ) . Calnexin , an ER resident , was used as a control to ensure that the nuclear preparations were free of endoplasmic reticulum . A significant fraction of Ago-2 was localized in the nucleus of BCBL-1 cells . This observation is consistent with a study by Gagnon et al . , which reported comparable amounts of Ago2 in the nucleus and cytoplasm of multiple cell lines including HeLa , T47D , A549 and fibroblasts [23] . These results were confirmed using immunofluorescence analysis ( IFA ) of Ago-2 in isolated BCBL-1 nuclei by confocal microscopy and 3D-reconstruction . The images in Fig 2C show Ago-2 in all planes of view ( XY , YZ and ZX ) with and without DAPI , and it is evident that Ago-2 is present inside the BCBL-1 nuclei . We observed similar results with IFA performed on KSHV-infected TIVE cells ( Movie S1 ) . Thus , we concluded that Ago2 and viral miRNAs are present in the nuclei of infected cells , and miRNAs could potentially interact via Ago2 with nuclear lncRNAs . Of the 126 rescued lncRNAs identified based on transcriptional profiling , 98 contained seed sequence matches for at least one KSHV miRNA . Repeated sampling of 126 sequences from randomly generated DNA sequences , controlling for lncRNA length , revealed that the presence of KSHV miRNA seed matches in 98 out of 126 lncRNAs is statistically significant ( p-value = 5 . 79 x 10−8 , one-sided t-test ) . These data provide genetic evidence for miRNA-dependent deregulation of host lncRNAs during KSHV latency . In order to validate that KSHV miRNAs can target host lncRNAs in the absence of KSHV infection , we chose four lncRNAs from the 98 containing seed sequences , and transfected pools of corresponding miRNA mimics into uninfected TIVE cells . The pools of mimics transfected were specific to the seed matches that those lncRNAs contained . Their respective mimic pools when compared to control mimic significantly knocked down all four lncRNAs tested , demonstrating that the viral miRNAs target lncRNAs in the absence of KSHV infection ( Fig 3A ) . The miRNA-dependent downregulation of lncRNAs could result from direct targeting of lncRNAs by miRNAs , or from an indirect secondary effect ( e . g . , through miRNA-mediated downregulation of transcription factors ) . To investigate direct interaction between KSHV miRNAs and lncRNAs , we performed miRNA pull-down experiments in TIVE-Ex-LTC cells . TIVE-Ex-LTC cells were derived from TIVE cells ( see Materials and Methods ) , but grow much faster compared to TIVE cells . KSHV negative TIVE-Ex-LTC cells were transfected with biotinylated miRNA mimics for either miR-K12-6-5p , miR-K12-11* or siGLO ( lacks biotin ) and pull-down experiments were performed 24 h post-transfection . It is important to note that the mimics are dsRNAs that require loading into the RISC in order to bind their targets . Loc541472 has one binding site for miR-K12-6-5p but none for miR-K12-11* , and CD27-AS1 has one binding site for miR-K12-11* but none for miR-K12-6-5p . Biotinylated miR-K12-6-5p mimic pulled down 43 . 7% of Loc541472 and none of CD27-AS1 , and miR-K12-11* mimic pulled down 12 . 9% of CD27-AS1 , but no Loc541472 , thus confirming direct miRNA-lncRNA interaction ( Fig 3B ) . The fact that we identified putative lncRNA targets of viral miRNAs in PEL and endothelial cells by Ago HITS-CLIP and viral genetics , together with biochemical evidence for direct miRNA-lncRNA interaction , demonstrated that KSHV deregulates a subset of host lncRNAs in a miRNA-dependent fashion . To date a very small percentage of all lncRNAs are functionally annotated , making interpretation of lncRNA expression data challenging . As a starting point , we analyzed lncRNAs that were deregulated ( upregulated , downregulated and rescued ) in response to latent KSHV infection for known or proposed functions in disease processes . Comparison of our dataset to two public databases [28 , 29] identified 54 lncRNAs that were previously shown to be aberrantly expressed in various human cancers ( S3 Table ) . These include HOTTIP , DLEU2 , HOTAIRM1 , ANRIL , MEG3 and UCA1 . Ten of the 54 lncRNAs are listed in Table 2 , and include oncogenic and tumor suppressor lncRNAs . HOTTIP is upregulated in hepatocellular carcinoma , osteosarcoma , lung , prostate and other cancers [30]; DLEU2 is deleted in lymphocytic leukemia and epigenetically silenced in myeloid leukemia [31 , 32] . Knockdown of HOTARM1 has been shown to promote proliferation in promyelocytic leukemia cells [33] . ANRIL is an oncogenic lncRNA that promotes proliferation in numerous cancers including basal cell carcinoma ( BCC ) , glioma , prostate and ovarian cancers [34] . UCA1 is upregulated in multiple cancers , including bladder , endometrial and pancreatic cancer and acts as an oncogenic lncRNA [35] . Loss of MEG3 expression has been reported in a wide spectrum of malignancies ranging from gliomas to colon and liver cancers [36] . To understand the mechanisms by which cancer-related lncRNAs are deregulated by KSHV , and their contribution to pathogenesis , we chose to initially study UCA1 , ANRIL and MEG3 . MEG3 is a tumor suppressor lncRNA which is proposed to act by enhancing transcription from p53-dependent promoters [36] . Studies in HCT116 and U2OS cell lines have identified that MEG3 is a nuclear localized lncRNA [41] , which was also confirmed in GM12878 cells by the GENCODE project [42] . According to the microarray data ( S2 Table ) , MEG3 was slightly upregulated during latent KSHV infection . However , when validating MEG3 expression by qRT-PCR , it behaved in a rescued pattern , being suppressed in wt-KSHV infection and restored in Δcluster-KSHV-infected cells , suggesting regulation by KSHV miRNAs ( Fig 4A ) . MEG3 contained seed sequence matches for miR-K12-3 , K12-5 , K12-6-5p , K12-8* and K12-9* . Uninfected TIVE cells were transfected with a pool of three KSHV miRNA mimics ( miR-K12-5 , K12-6-5p and K12-8* ) . MEG3 expression was reduced by almost 80% ( Fig 4B ) . Furthermore , miRNA pull-down assays using biotinylated miR-K12-6-5p mimic pulled-down 24 . 5% of MEG3 ( Fig 4C ) . miR-K12-11* mimic did not pull down MEG3 lncRNA . These data are consistent with viral miRNAs directly binding to and downregulating MEG3 . ANRIL is a nuclear localized oncogenic lncRNA that drives proliferation by silencing the INK4 tumor suppressor gene by recruiting PRC2 complexes [34] . The fact that ANRIL was downregulated in KSHV-infected cells from the microarray data suggested that ANRIL does not have a direct role in proliferation; however , ANRIL has recently also been implicated in innate immune responses , albeit in the context of bacterial infection [43] . Analysis of ANRIL expression by qRT-PCR showed a very strong 100-fold downregulation in wt-KSHV-infected cells , and a slightly reduced inhibition in the Δcluster-KSHV-infected TIVE cells ( Fig 5A ) . Such strong repression is not typical of miRNAs , however , the cDNA of ANRIL had a total of 17 6-mer seed matches for 12 of 25 mature KSHV miRNAs . To investigate whether the large number of KSHV miRNA seed sequence matches in ANRIL are targeted by KSHV miRNAs , we ectopically overexpressed the shortest isoform ( transcript variant 12 ) of ANRIL from a CMV promoter-driven vector in wt-KSHV-infected and uninfected TIVE-Ex-LTC cells . Since TIVE cells are highly resistant to plasmid transfection , we used TIVE-Ex-LTC cells for this experiment . As shown in Fig 5B , the ANRIL expression levels achieved in wt-KSHV-infected cells were 80% less compared to uninfected cells . We note that this expression difference was not due to differences in transfection efficiencies , since a control gene ( LSD-1 ) , expressed from the same vector , was expressed at similar levels in both cell lines ( Fig 5B ) . Hence , the reduced ANRIL expression levels in infected cells compared to control cells strongly suggested post-transcriptional miRNA-dependent regulation of ANRIL . To test this , we transfected a pool of four miRNA mimics ( miR-K12-1* , K12-6-5p , K12-2* and K12-11* ) which led to a strong knock-down of ANRIL expression in uninfected TIVE cells compared to the control mimic ( Fig 5C ) . Additionally , pull-down experiments in TIVE cells using biotinylated miR-K12-6-5p and miR-K12-11* mimics , for which ANRIL contains two seed matches each , significantly pulled-down 12 . 7% and 22 . 7% of ANRIL transcripts , respectively ( Fig 5D ) . Together these data show that ANRIL is targeted by multiple viral miRNAs . Since ANRIL also contained miRNA seed sequence matches for miR-K12-10 and K12-12 , which are still present in Δcluster-KSHV ( Fig 1A ) , we wanted to test ANRIL expression in the absence of all viral miRNAs . To this end we analyzed ANRIL expression in TIVE cells by infecting with a virus lacking all 12 miRNA genes ( Δall-KSHV ) . Surprisingly we did not observe significantly altered ANRIL expression compared to wt-KSHV-infected cells ( Fig 5E ) . These data suggested that ANRIL may also be negatively regulated by latency associated proteins . To directly address this question we ectopically expressed the major latency associated proteins of KSHV ( LANA , vCyclin , vFLIP and Kaposin ) and monitored ANRIL expression by qRT-PCR . Since TIVE-Ex-LTC cells do not express detectable levels of ANRIL , this experiment was performed in HeLa cells , which are known to robustly express ANRIL [44] . vFLIP and vCyclin downregulated ANRIL expression by almost 75% and 53% , respectively ( Fig 5F ) . LANA and Kaposin did not have significant effects . The observation that ANRIL is negatively regulated by both miRNAs and latency associated proteins is in congruence with other host genes that are targeted by multiple viral mechanisms [45] . Urothelial Cancer Associated 1 ( UCA1 ) is a lncRNA which was identified as highly upregulated in bladder cancer and has since been implicated in other cancers like colorectal , ovarian and renal carcinomas [35] . UCA1 is partially localized in both the nucleus and the cytoplasm and plays distinct roles in different sub-cellular compartments [46 , 47] . Recently , it was shown that UCA1 transcription is induced by HIF-1α , to enhance hypoxic proliferation , migration and invasion of bladder cancer cells [35] . UCA1 was upregulated by approximately 90-fold during wtKSHV infection and approx . 300-fold during Δcluster-KSHV infection ( Fig 6A ) . Since UCA1 was upregulated under both infection conditions and its cDNA sequence contained no seed matches for any KSHV miRNAs , UCA1 is presumably not regulated by a miRNA-dependent mechanism . To determine which of the four major latency-associated proteins ( LANA , vCyclin , vFLIP and Kaposin ) upregulates UCA1 , we transfected TIVE-Ex-LTC cells with expression vectors either alone or in combination . Ectopic expression of vCyclin and Kaposin led to a 3 . 9 and 5 . 7-fold upregulation of UCA1 as monitored by qRT-PCR , respectively . Furthermore , co-transfection of vCyclin and Kaposin increased UCA1 to almost 15-fold compared to empty vector suggesting synergy ( Fig 6B ) . LANA and vFLIP had no effect . The fact that the upregulation observed in transfected cells is much less than in the context of infection could be a consequence of either an altered stoichiometry or absolute expression levels of latency proteins , or mean that other viral genes might contribute to UCA1 upregulation . To address whether UCA1 directly contributes to KS-associated phenotypes , we knocked-down UCA1 expression using siRNAs in KSHV-infected TIVE cells . At 24 , 48 , 72 and 96 h post-transfection we observed 60–85% knockdown of UCA1 expression ( Fig 6C ) . First , we assayed for proliferation using the MTS assay . We measured proliferation at 24 , 48 , 72 and 96 h post-transfection and observed a statistically significant and dose-dependent decrease in proliferation of cells treated with siUCA1 as compared to scrambled control ( Scr ) . Upon treatment with 10 nM siUCA1 , the proliferation rate dropped to 72% by day 1 and then progressively to 52% by day 4 ( Fig 6D ) . We observed a similar decrease in proliferation of uninfected TIVE cells transfected with siUCA1 ( S2A Fig ) suggesting that UCA1 contributes to endothelial cell proliferation in general . To test whether latent KSHV upregulates UCA1 in all infected cells , we measured UCA1 expression levels in uninfected and KSHV-infected iSLK cells . UCA1 was upregulated by almost 5-fold in KSHV-infected iSLK cells ( S3A Fig ) . Knockdown of UCA1 in uninfected and KSHV-infected iSLK cells led to a mild reduction in proliferation of these cells ( S3B and S3C Fig ) . The magnitude of effect observed in iSLK cells was much lower than that in TIVE cells , presumably because iSLK cells are transformed and unlike TIVE cells form tumors in nude mice [18] . Next , we assayed the effect of UCA1 knockdown on migration of KSHV-infected TIVE cells . The migration assay ( wound healing ) involves introduction of a scratch in a monolayer of cells and measuring the percentage of the clear area that gets covered by migration at 12 hours post introduction of the scratch under serum-free conditions ( Fig 6E ) . siUCA1-treated cells were consistently slower in migration from day 1 through day 4 , as they recovered only between 12–15% of the scratch area , while Scr-treated cells recovered between 26–35% of the area ( Fig 6F ) . A similar reduction in migration was observed on days 1 and 2 when UCA1 was knocked down in uninfected TIVE cells , however , no difference was evident after day 3 ( S2B Fig ) . This suggests that high UCA1 levels in KSHV-infected endothelial cells contribute to increased migration of these cells . These data demonstrate that the induction of UCA1 by the KSHV latency-associated proteins Kaposin and vCyclin promotes proliferation and migration , and likely contributes to KSHV pathogenesis and tumorigenesis .
Here we show that latent KSHV infection significantly alters the lncRNA expression profile of endothelial cells . Deregulation of lncRNAs has implications in diseases such as diabetes , neurological disorders , viral infections and cancer [48 , 49] . Our study establishes that KSHV employs its latency proteins and miRNAs , either alone or in combination , to target specific lncRNAs and potentially contribute to sarcomagenesis . Post-transcriptional regulation of lncRNA expression by miRNAs is a newly described phenomenon . Yoon et al showed let-7 loaded RISCs targeted lincRNA-p21 in a HuR-dependent manner in cervical carcinoma cells , eventually destabilizing and degrading lincRNA-p21 [50] . In bladder cancer , UCA1 and miR-1 expressions were inversely correlated , and overexpression of miR-1 phenocopied the knockdown of UCA1 [51] . Further , MALAT-1 , a nuclear lncRNA , was reported to be targeted by miR-9 in an Ago-2-dependent manner in the nuclei of Hodgkin’s lymphoma and glioblastoma cell lines [15] . We identified 126 lncRNAs as potential targets of viral miRNAs in endothelial cells , and we verified direct miRNA/lncRNA interactions by pulldown experiments with biotinylated KSHV miRNA mimics targeting Loc541472 , CD27-AS1 , ANRIL and MEG3 . Results from the Ago HITS-CLIP experiment further suggest that this regulation proceeds in an Ago and hence RISC-dependent manner . As per our current understanding , RISC-mediated silencing of mRNAs proceeds via translation repression and induction of mRNA turnover [52 , 53] . RNA destabilization followed by degradation is perhaps the mechanism relevant to silencing of lncRNAs . However , the details of the mechanism , especially for lncRNAs lacking a cap and/or a poly-A tail , remain to be uncovered . An alternative and not mutually exclusive mechanism that involves direct engagement of miRNAs and lncRNAs is miRNA sponging by lncRNAs [54] . LincRNA-RoR sponges miR-145-5p thereby increasing the expression of pluripotent stem cell factors Oct4 , Nanog and Sox2 [55] . The Steitz lab showed that lncRNAs encoded by Herpesvirus Saimiri , called HSURs , sequester host miRNAs in infected T-lymphocytes [16] . It is plausible that some host lncRNAs could sponge KSHV miRNAs , thereby derepressing downstream targets instead of being targeted by miRNAs themselves . We demonstrated that viral latency proteins vCyclin and Kaposin synergistically upregulate UCA1 while vFLIP and vCyclin downregulate ANRIL . Thus , aside from miRNAs , the latency proteins play a pronounced role in perturbing lncRNA expression . This is not surprising given we identified 858 differentially expressed lncRNAs during wt-KSHV infection and only 126 were potential miRNA targets . vCyclin , an ortholog of cellular Cyclin D , upregulates expression of cell cycle regulatory genes [56] . Moreover , Kaposin stabilizes cytokine mRNAs thereby increases their turnover time [57] . vCyclin and Kaposin may act cooperatively by augmenting transcription and simultaneously preventing turnover of UCA1 . We also showed that ectopically expressed vFLIP strongly downregulates ANRIL . STAT1-mediated activation of the ANRIL locus in vascular endothelial cells has been reported based on GWAS studies [58] . Studies using a mutant virus that lacks vFLIP in HUVEC cells showed activation of STAT1 in a NF-κB-dependent manner , suggesting that vFLIP probably inhibits STAT1 to downregulate ANRIL expression [59] . A recent study in endothelial cells demonstrated that ANRIL expression is induced by pro-inflammatory molecules , especially NF-κB and TNF-α , and silencing of ANRIL expression led to a reduction in IL6/IL8 response [60] . This further underlines the role of ANRIL in immunity and supports the notion that KSHV may downregulate ANRIL to evade innate immune responses . KSHV drives latently infected cells towards proliferation by a variety of mechanisms such as encoding orthologs for cell cycle proteins like vCyclin , or interfering with the p53 pathway through LANA [61] , encoding miR-K12-11 , an ortholog of oncomir-155 [62] , and the induction of the oncogenic host miRNA cluster miR-17/92 [45] . Here we demonstrate that KSHV also upregulates UCA1 to drive proliferation and migration in endothelial cells . UCA1 has also been shown to promote the Warburg effect [63] , an effect that has been shown to be required for maintenance of latent KSHV in endothelial cells [64] . We found that 53 additional lncRNAs previously shown to be aberrantly expressed in various malignancies are deregulated by KSHV , suggesting that UCA1 exemplifies how KSHV could similarly exploit lncRNAs that contribute to phenotypes such as proliferation and migration in the context of tumorigenesis . Given that the majority of lncRNAs we catalogued in this study remain uncharacterized , the repertoire of cancer-relevant lncRNAs regulated by KSHV may be much larger . Although cancer is the pathological consequence of KSHV infection , KSHV could target lncRNAs of biological significance in other cellular processes , for example , lncRNAs involved in inflammation and innate immunity [9] . KSHV continually evades the innate immune response using several approaches , like suppressing TGF-β signaling [45] , activation of NF-κB response genes [65] and encoding trace amounts of v-IL6 , a truncated version of human IL-6 , during latent infection [66] . Loc541472 , which we show here is targeted directly by KSHV miRNAs , is antisense to the hIL-6 promoter , suggesting that targeting of this lncRNA contributes to regulation of IL-6 expression . Indeed , preliminary experiments suggest a correlation between Loc541472 and hIL-6 expression and mechanistic studies are currently ongoing . We identified a novel paradigm by which KSHV , an oncogenic herpesvirus , regulates cellular gene expression by targeting host lncRNAs with viral miRNAs and latency proteins . Studying lncRNAs deregulated by KSHV may yield novel mechanisms by which viruses evade the host immune response and in the case of EBV and KSHV contribute to tumorigenesis , as exemplified by our data on UCA1 which modulates migration and proliferation . Finally , studies on aberrantly expressed lncRNAs in KSHV-infected cancer cells may aid the functional characterization of cellular lncRNAs and at the same time identify novel virus-specific therapeutic targets for KS .
The viruses used in this study , wt-KSHV , Δcluster-KSHV and Δall-KSHV , have the viral genome cloned into a Bac-16 backbone , as described in Brulois et al . [19] and Jain et al . [20] . Transcript variant 12 ( RefSeq ID: NR_047542 . 1 ) of ANRIL was expressed from a pcDNA3 . 1 vector [67] . LANA , vCyclin , vFLIP and Kaposin were expressed from pcDNA3 . 2 vectors [68] . Telomerase immortalized vein endothelial cells ( TIVE ) and long-term cultured KSHV infected cells ( TIVE-LTC ) were generated by immortalizing passage 2 HUVEC cells ( kindly provided by Dr . Keith McCrae , Case Western Reserve University ) in our laboratory as described [18] . All uninfected and infected TIVE cells were grown in complete Medium-199 ( 1% Pen-Strep , 20% FBS ) , supplemented with Endothelial cell growth supplement ( Sigma ) . TIVE-Ex-LTC cells were obtained by culturing TIVE-LTC cells as single cell dilutions without antibiotic selection , and have lost all copies of viral episomes . TIVE-Ex-LTC cells grow faster and are more transfectable compared to TIVE cells . All uninfected and infected TIVE-Ex-LTC cells were grown in complete DMEM ( 1% Pen-Strep , 10% FBS ) . Latently infected TIVE and TIVE-Ex-LTC cells were maintained under hygromycin ( 10 μg/mL ) to prevent episome loss . Body-cavity-based lymphoma ( BCBL-1 ) cell line was derived from KSHV positive primary effusion lymphoma ( PEL ) and was kindly provided by Dr . Don Ganem at UCSF [69] . BCBL-1 cells were grown in complete RPMI ( 2% Pen-Strep , 10% FBS ) . HeLa cells and iSLK cells were grown in complete DMEM ( 1% Pen-Strep , 10% FBS ) . The method for isolating nuclear and cytoplasmic fractions was adapted from [71] . Briefly , 1 x 107 BCBL-1 cells were pelleted and washed twice with ice cold PBS . Cells were resuspended smoothly by gentle pipetting in Sucrose buffer I ( SB-I: 0 . 32 M Sucrose , 3 mM CaCl2 , 2 mM Mg ( Ac ) 2 , 0 . 1 mM EDTA , 10 mM Tris-HCl ( pH 8 ) , 1 mM DTT , 0 . 5 mM PMSF and 0 . 5% NP-40 ) using 100 μL buffer per 1 x 107 cells . Lysis was at room temperature for 60–90 s . The nuclei were pelleted at 800 x g , 4 ˚C for 5 min and the supernatant ( cytoplasmic fraction ) was frozen immediately and stored at -80C . The pellet was resuspended smoothly by gentle pipetting in 50 μL of SB-I and allowed to sit for 30 s at RT . The nuclei were pelleted again at 800 x g , 4 ˚C for 5 minutes . The supernatant was discarded and the pellet ( now whiter ) was washed twice in 1 mL ice cold PBS . The resuspension was smooth and easy indicating no nuclear rupture . 10 μL of the 1 mL suspension from the second wash was trypan blue stained and checked by microscopy to verify the purity and integrity of the isolated nuclei . The nuclear fraction was frozen immediately and stored at -80 ˚C . TIVE cells were grown overnight on coverslips at a dilution of 1 x 104 cells per well in a 6-well plate . Nuclei isolated from PEL cells were prepared as described [72] , and fixed with a 1:1 ratio of methanol and acetone for 10 min in a humid chamber at 4 ˚C . The samples were blocked in PBS with 3% BSA for 1 h at room temperature , and then incubated overnight at 4 ˚C with either primary anti-Ago2 antibody or blocking solution ( control ) . After washing , the samples were incubated with Alexa-468 anti-rat secondary antibody for 1 hour at room temperature . The slides were then stored at -20 ˚C and imaged using a LEICA TCS SP2 AOBS Spectral Confocal microscope . The images were analyzed and figures were generated using the freeware Vaa3D [73] . The movie was generated using Volocity® 6 . 3 . SDS-PAGE and Western blotting were performed using whole cell lysates , or cytoplasmic or nuclear fractions prepared from 100 , 000 cells/well . The following antibodies were used to probe the membrane: Ago2 ( 11A9 , [74] ) , β-Tubulin ( Millipore , CP06-100UG ) , Sm antigen ( Dr . Joan Steitz’s lab , Yale University ) , Lamin A/C ( Active Motif , 39288 ) , Calnexin ( ENZO Lifesciences , ADI-SPA-865-D ) . Total RNA was isolated with RNA-Bee ( Tel-Test Inc . ) using the protocol provided by the manufacturer . Total RNA ( 5–10 μg ) was treated with DNase I ( NEB ) according to the manufacturer’s instructions and ethanol precipitated overnight . Genome-wide lncRNA microarray analysis was performed with ArrayStar using Human LncRNA Array v3 . 0 ( 8 x 60K , Arraystar ) . A fold change cut-off of 2 . 0 was applied to filter lncRNAs into different categories ( upregulated , downregulated and rescued ) for further analyses . Three technical replicates for each of the three samples were analyzed . Total RNA preparations from PEL cell fractions were reverse transcribed using the TaqMan MicroRNA Reverse Transcription Kit ( ThermoScientific ) . Stem-loop qPCR was performed using the TaqMan Gene Expression Master Mix and appropriate miRNA assays from Applied Biosystems . Total RNA ( 2 μg ) was reverse transcribed using SuperScript III ( Life Technologies ) using random hexamers according to the manufacturer’s instructions . cDNA corresponding to 50–100 ng RNA was used per 10 μL of qPCR reaction . Instruments used for real-time PCR included ABI StepOne Plus ( Applied Biosystems ) and LightCycler96 ( Roche ) . qPCR primer sequences are listed in S4 Table . TIVE cells were seeded in 48-well plates ( 50 , 000 cells/well ) and transfected with pools of miRNA mimics ( in equimolar ratios and a final concentration of 5 nM ) purchased from Qiagen . At 48 h post transfection , the lncRNA expression levels were measured using the Power SYBR Green Cells-to-CT Kit ( ThermoFisher ) . In the cases of ANRIL and MEG3 , 10 cm plates were seeded to 70% confluency and qRT-PCR analysis was performed using the conventional approach described above . Biotinylated miRNA mimics ( miR-K12-6-5p and miR-K12-11* ) were purchased from Exiqon . Pulldown was performed from TIVE and TIVE-Ex-LTC cells according to the previously published protocol [75] with minor changes . Each replicate started with 6x106 cells for TIVE-Ex-LTCs ( instead of 4x106 ) and 8x106 cells for TIVE cells . Input RNAs saved for analysis were 5% and 20% for TIVE-Ex-LTC and TIVE cells , respectively . TIVE-Ex-LTC cells were reverse transfected in 6-well plates ( 300 , 000 cells/ well ) with 2 μg of plasmid DNA using FuGENE HD according to the manufacturer’s protocol . HeLa cells were seeded in 6-well plates ( 150 , 000 cells/ well ) and were transfected 24 h later with 2 μg plasmid DNA using Lipofectamine 3000 according to the manufacturer’s protocol . DMEM ( 10% FBS ) was used for transfection of both cell types . Comparable transfection efficiencies were ensured by co-transfection of pmaxGFP . Total RNA was harvested from transfected cells at 72 h post-transfection . wtKSHV-infected or uninfected TIVE cells were plated in 96-well plates ( 20 , 000 cells/well for MTS assay ) and 48-well plates ( 250 , 000 cells/well for wound healing assay ) . Uninfected and wtKSHV-infected iSLK cells were plated in 96-well plates ( 8000 cells/well for MTS assay ) . siRNAs ( 5nM or 10 nM ) against UCA1 ( Qiagen ) were transfected using Lipofectamine RNAiMAX reagent ( ThermoFisher ) according to the manufacturer’s protocol . ON-TARGETplus Non-targeting Control siRNA ( Dharmacon ) was used as the scrambled negative control . At 4 h post-transfection , the serum free medium was replaced by complete Medium-199 ( TIVE ) or DMEM ( iSLK ) . Comparable transfection efficiencies were ensured by co-transfection of siGLO ( Dharmacon ) . Statistical analyses on experimental measurements were done using two-tailed student’s t-test assuming unequal variances for all experiments reported . Raw data files from the microarray experiment were deposited to the Gene Expression Omnibus under the accession number GSE89114 .
|
KS is the most prevalent cancer associated with AIDS in sub-Saharan Africa , and is also common in males not affected by AIDS . KSHV manipulates human cells by targeting protein-coding genes and cell signaling . Here we show that KSHV alters the expression of hundreds of human lncRNAs , a broad class of regulatory molecules involved in a variety of cellular pathways including cell cycle and apoptosis . KSHV uses both latency proteins and miRNAs to target lncRNAs . miRNA-mediated targeting of lncRNAs is a novel regulatory mechanism of gene expression . Given that most herpesviruses encode miRNAs , this mechanism might be a common theme during herpesvirus infections . Understanding lncRNA deregulation by KSHV will help decipher the important molecular mechanisms underlying viral pathogenesis and tumorigenesis .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Material",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"gene",
"regulation",
"pathogens",
"endothelial",
"cells",
"microbiology",
"long",
"non-coding",
"rnas",
"epithelial",
"cells",
"viruses",
"oncology",
"micrornas",
"dna",
"viruses",
"bioassays",
"and",
"physiological",
"analysis",
"herpesviruses",
"research",
"and",
"analysis",
"methods",
"animal",
"cells",
"medical",
"microbiology",
"gene",
"expression",
"microbial",
"pathogens",
"biological",
"tissue",
"kaposi's",
"sarcoma-associated",
"herpesvirus",
"microarrays",
"biochemistry",
"rna",
"carcinogenesis",
"cell",
"biology",
"nucleic",
"acids",
"anatomy",
"viral",
"persistence",
"and",
"latency",
"virology",
"viral",
"pathogens",
"genetics",
"epithelium",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"non-coding",
"rna",
"organisms"
] |
2017
|
microRNA dependent and independent deregulation of long non-coding RNAs by an oncogenic herpesvirus
|
Genome-wide association studies have identified more than 200 genetic variants to be associated with an increased risk of developing multiple sclerosis ( MS ) . Still , little is known about the causal molecular mechanisms that underlie the genetic contribution to disease susceptibility . In this study , we investigated the role of the single-nucleotide polymorphism ( SNP ) rs1414273 , which is located within the microRNA-548ac stem-loop sequence in the first intron of the CD58 gene . We conducted an expression quantitative trait locus ( eQTL ) analysis based on public RNA-sequencing and microarray data of blood-derived cells of more than 1000 subjects . Additionally , CD58 transcripts and mature hsa-miR-548ac molecules were measured using real-time PCR in peripheral blood samples of 32 MS patients . Cell culture experiments were performed to evaluate the efficiency of Drosha-mediated stem-loop processing dependent on genotype and to determine the target genes of this underexplored microRNA . Across different global populations and data sets , carriers of the MS risk allele showed reduced CD58 mRNA levels but increased hsa-miR-548ac levels . We provide evidence that the SNP rs1414273 might alter Drosha cleavage activity , thereby provoking partial uncoupling of CD58 gene expression and microRNA-548ac production from the shared primary transcript in immune cells . Moreover , the microRNA was found to regulate genes , which participate in inflammatory processes and in controlling the balance of protein folding and degradation . We thus uncovered new regulatory implications of the MS-associated haplotype of the CD58 gene locus , and we remind that paradoxical findings can be encountered in the analysis of eQTLs upon data aggregation . Our study illustrates that a better understanding of RNA processing events might help to establish the functional nature of genetic variants , which predispose to inflammatory and neurological diseases .
In the past 10 years , genome-wide association studies ( GWAS ) assaying hundreds of thousands of single-nucleotide polymorphisms ( SNPs ) rapidly expanded our knowledge of genetic loci contributing to complex multifactorial diseases . The current GWAS catalog contains already more than 89000 unique SNP-trait associations [1] , and shared risk variants between diseases are increasingly recognized . However , despite these important insights , timely diagnosis and appropriate treatment of autoimmune diseases remains challenging , because our understanding of pathogenesis is still limited . We have to bear in mind that association does not imply causation . Moreover , onset and course of autoimmune diseases are influenced by a combination of different risk factors , including multiple genetic , epigenetic , immunological , and environmental factors . As in chaotic systems related to the n-body problem , all these factors may constantly interact with each other , which provokes unpredictable outcomes [2] . Multiple sclerosis ( MS ) is a chronic inflammatory disorder of the central nervous system ( CNS ) and the most common cause of neurological disability in young adults . The disease is characterized by immune cell infiltrates in the CNS , demyelination , and axonal loss [3 , 4] . Most patients are initially diagnosed with relapsing-remitting MS ( RRMS ) , which lasts for around 2 decades before transition to secondary progressive MS ( SPMS ) , whereas 10–15% of the patients show a gradual worsening of neurological functions from disease onset without early relapses ( primary progressive MS , PPMS ) [5] . The individual course of the disease is very heterogeneous [6] . Severity and diversity of clinical symptoms largely depend on frequency and distribution of lesions in brain and spinal cord . However , the precise triggers of neuroinflammation and neurodegeneration in MS have not been resolved yet . T-cells have been in the focus of intense research , but evidence is accumulating that B-cells play a crucial role as well . In fact , basically all approved disease-modifying therapies for MS affect the B-cell population [7] . The biological underpinnings of MS are highly complex , involving multifaceted interactions between immune cells as well as between risk genes and environmental factors ( e . g . , exposure to tobacco smoke and infection with Epstein-Barr virus , EBV ) [8 , 9] . The hereditary component of MS has been first demonstrated in studies of twins . In comparison to the general population , a monozygotic twin of an MS patient has a more than 100 times higher risk to develop the disease [10] . It is well-established that the human leukocyte antigen ( HLA ) class II region exerts the strongest genetic effect in MS [11] . Beyond that , international GWAS so far identified about 200 independent MS-associated non-HLA variants , each exerting a small effect on disease risk [12–14] . However , still little is known about causal genetic variants and the biological mechanisms that underlie MS susceptibility . Any of the genetic variants within a risk-associated locus that are in strong linkage disequilibrium ( LD ) with a GWAS-reported SNP could account for the association . Moreover , the vast majority ( >90% ) of MS-associated tag SNPs are located in intronic or intergenic regions [14] . There is currently a lack of studies aimed at deciphering the functional implications of these loci in MS . It is also unclear to which extent genetic variants may influence the course of MS ( relapsing and progressive ) and the degree of disease activity . One of the top non-HLA loci conferring risk of MS is found within the CD58 gene locus [15 , 16] . In particular , SNPs in the first intron of CD58 ( e . g . , dbSNP identifiers rs12044852 , rs10801908 , rs1335532 , and rs2300747 ) are associated with MS , with odds ratios ranging from 1 . 30 in the latest multi-national GWAS [13 , 14] to 2 . 13 in a regional cohort from Germany [16] to 2 . 63 in familial MS cases [17] as compared to controls . Interestingly , a microRNA ( miRNA ) , hsa-miR-548ac , is encoded in this intron . Thus , both CD58 mRNA and hsa-miR-548ac are processed from the same primary transcript ( Fig 1A ) . The miRNA hsa-miR-548ac belongs to a large family of miRNAs that evolved from a primate-specific transposable element ( Made1 ) [18] , and it has been first identified in 2010 by Jima et al . based on high-throughput sequencing ( HTS ) of small RNAs from human B-cells [19] . By post-transcriptional gene silencing , miRNAs are able to modulate the expression of hundreds of genes either directly or indirectly [20 , 21] . However , the specific target genes of hsa-miR-548ac have so far not been explicitly determined experimentally . It is also important to note that there is a SNP , rs1414273 , located in the miRNA stem-loop sequence , 11 nt downstream of the 3' Drosha cleavage site ( Fig 1B ) . This SNP is very commonly inherited together with the MS-associated SNPs from GWAS , with r2 and D' as measures of LD being close to 1 ( Fig 1C ) [16] . Furthermore , the MS risk alleles of these SNPs are the major alleles in individuals of European ancestry but the minor alleles in East Asian populations ( Fig 1D ) . In 2009 , De Jager et al . could show by an expression quantitative trait loci ( eQTL ) analysis based on 149 individuals from the HapMap collection that carriers of the MS-associated haplotype express lower levels of CD58 mRNA [15] ( as later confirmed at the protein level [22] ) . The authors also further established the function of CD58 protein in the activation of T-cells through engagement of the CD2 receptor , and they demonstrated that decreased CD58 expression , as particularly seen during a relapse , leads to dysfunction of regulatory T-cells [15] . However , the widespread expression of CD58 in many cell types necessitates caution in the interpretation of the role of CD58 in the pathogenesis of MS . Moreover , the study by De Jager et al . did not consider hsa-miR-548ac , which was not known at that time . We speculated that the MS-associated SNPs within the CD58 gene locus affect the expression of mature hsa-miR-548ac and that , more specifically , SNP rs1414273 is the causal genetic variant that acts as cis-mRNA-eQTL and cis-miR-eQTL . Here , we provide evidence that the genotype alters the efficiency of the Drosha-DGCR8 complex in processing the miRNA stem-loop . This appears to modulate the expression levels of both hsa-miR-548ac and CD58 mRNA in an inverse and allele-specific manner . We further compared the levels of hsa-miR-548ac in different immune cell subsets circulating in the blood , and we obtained first insights into the function of this miRNA by screening for target genes . In exploring the data , we stumbled on two paradoxes of a three-variable relationship: ( 1 ) the non-transitivity paradox of positive correlation and ( 2 ) the Simpson's paradox . These are well-known statistical phenomena , especially in the social and biomedical sciences [23 , 24] . We discovered similar eQTL paradoxes for other human miRNAs . Our study illustrates how disease susceptibility variants may regulate uncoupling of miRNA production from host gene expression . This underscores the contribution of miRNAs in disease pathogeneses and their potential use as molecular biomarkers , which has not been fully exploited yet .
The eQTL analysis employed microarray data ( HapMap cohort ) , HTS data ( Geuvadis cohort ) , and real-time PCR data ( MS cohort ) . Firstly , we analyzed CD58 mRNA expression in lymphoblastoid cell lines ( LCLs ) derived from a total of 726 individuals from 8 global populations from the HapMap project [25–27] . The normalized gene expression data correlated significantly with the genotypes of the intronic SNP rs1335532 ( used as proxy ) for 4 populations: CHB ( F-test p = 0 . 019 ) , GIH ( p = 0 . 00008 ) , JPT ( p = 0 . 0004 ) , and MEX ( p = 0 . 030 ) . In all these populations , homozygous carriers of the MS risk allele showed , on average , the lowest CD58 transcript levels ( Fig 2A ) . This clearly confirms the eQTL and the protein QTL previously described in LCLs by De Jager et al . [15] and Wu et al . [22] , respectively . The analysis of the other 4 populations did not reach statistical significance , which may be explained by the limited numbers of individuals ( n≤135 ) and distinct risk allele frequencies across the populations ( from 0 . 34 in JPT up to 0 . 86 in CEU ) . Moreover , an analysis of variance ( ANOVA ) revealed huge differences in CD58 gene expression between the 8 populations ( p = 1 . 0×10−68 ) , impairing the association analysis . In fact , when considering the data of all 726 individuals in a simple linear regression ( SLR ) model , the eQTL effect could not be seen ( p = 0 . 472 ) due to this confounding . This is reminiscent of Simpson's paradox [23] , as elaborated later in this article . The issue of combining different groups of data can be more adequately addressed using an analysis of covariance ( ANCOVA ) , which blends ANOVA and regression . This analysis demonstrated a significant main effect for the rs1335532 genotype ( p = 0 . 027 ) and an interaction between genotype and population ( p = 0 . 0007 ) ( Fig 2D ) . Next , we evaluated RNA sequencing ( RNA-seq ) data of the Geuvadis consortium , which provided the levels of CD58 mRNA and hsa-miR-548ac in LCLs derived from 465 individuals from 5 populations [28] . In this cohort , the risk allele frequency of SNP rs1414273 was 0 . 86 in individuals of European ancestry ( CEU , FIN , GBR , and TSI ) and 0 . 40 in YRI . As for the microarray data of the HapMap cohort , significant differences in CD58 expression could be observed in the HTS data between the populations ( ANOVA p = 3 . 3×10−10 ) . This population effect was modest for hsa-miR-548ac ( p = 0 . 062 ) , which was actually detected in only 59 . 7% of the samples due to limited sequencing depth , with an overall average of 1 . 2 million miRNA reads per sample after quality control [28] . The eQTL analysis again reflected a Simpson-like paradox: When combining all data , the association of CD58 mRNA expression with the genotype of SNP rs1414273 was not significant in the SLR ( p = 0 . 447 ) but in the ANCOVA ( p = 0 . 004 ) , which included the population as independent variable ( Fig 2D ) . The data thus confirm the result of the HapMap cohort analysis , with individuals homozygous for the allele conferring risk of MS having a moderately lower level of CD58 gene transcripts than individuals homozygous for the alternative allele and heterozygous carriers showing an intermediate level of expression . On the other hand , the intronic SNP was also significantly associated with hsa-miR-548ac sequencing counts ( p = 0 . 022 and p = 0 . 014 for SLR and ANCOVA , respectively ) , however , in the opposite direction: The genetic risk variant correlated with higher levels of this miRNA . The pattern of increased miRNA expression and decreased CD58 mRNA expression in carriers of the MS-associated allele was noticed in all 5 populations , but it did not reach statistical significance per population given the limited number of LCLs analyzed ( n≤96 ) . In Fig 2B , we visualized the HTS data for non-CEU Europeans ( FIN , GBR , and TSI ) , because they are independent from the LCLs included in the HapMap cohort . In this geographically more proximate subset , the apparent inverse regulatory effect of the rs1414273 polymorphism on levels of CD58 ( F-test p = 0 . 017 ) and hsa-miR-548ac ( p = 0 . 017 likewise ) can be seen . To verify the findings obtained from the LCL data , we studied peripheral blood mononuclear cells ( PBMC ) from 32 MS patients from north-east Germany . Using quantitative real-time PCR , we were able to detect mature hsa-miR-548ac molecules in each of the triplicate reactions ( threshold cycle Ct<45 ) . This demonstrates that the measurement sensitivity of the MS cohort analysis is much better than of the HTS-based Geuvadis cohort analysis . Regarding SNP rs1414273 , only two MS patients had the TT genotype ( with respect to the forward strand of the reference genome ) . Thus , most patients carried the disease risk variant C at least once ( n = 14 CT heterozygotes and n = 16 CC homozygotes ) . The inverse correlation of the risk allele with lower CD58 mRNA expression and higher hsa-miR-548ac levels , which was seen in the Geuvadis data , could , in principle , be seen in the real-time PCR data as well ( Fig 2C ) . Because of limited sample size and small effect sizes , the genotypic association was not significant for CD58 ( F-test p = 0 . 172 ) and hsa-miR-548ac ( p = 0 . 158 ) alone but for the ratio of the levels of miRNA and host gene ( p = 0 . 040 ) . Moreover , Welch t-tests indicated significantly elevated miRNA expression in patients with the susceptibility variant for MS ( Fig 2C ) . There was no evidence that the data could be otherwise explained by differences in age , gender , therapy , or disease course ( F-test p>0 . 18 ) . To conclude , the MS-associated haplotype is implicated with significantly decreased CD58 mRNA levels ( HapMap cohort and Geuvadis cohort data ) . On the other hand , significantly increased levels of hsa-miR-548ac can be seen in risk allele carriers ( Geuvadis cohort and MS cohort data ) . The cis-mRNA-/miR-eQTL analysis thus suggests that SNP rs1414273 affects the processing of both mRNA and miRNA from the same primary transcript of the CD58 gene locus . The eQTL analysis revealed two observations , which are not intuitive . Firstly , different global populations showed similar eQTL results , but the association disappeared when combining the data . Secondly , although transcriptional control of CD58 and hsa-miR-548ac is coupled , there appears to be an inverse regulatory relationship to SNP rs1414273 . The latter is presumably driven by a genotype-dependent Drosha-mediated processing of the intronic miRNA stem-loop . As this cleavage occurs cotranscriptionally before splicing catalysis , the mRNA levels of the host gene may be affected as well . Two data sets were simulated for investigating the two paradoxes further . The Simpson-like paradox is shown in Fig 3A . For both populations examined ( n = 96 and n = 84 ) , the eQTL is highly statistically significant ( F-test p = 2 . 1×10−31 and p = 1 . 5×10−24 ) . However , when ignoring the subpopulations and looking at the entire cohort ( n = 180 ) , the average expression value is exactly 18 for each genotype group . Thus , the genotype effect completely disappeared ( p = 1 . 0 ) , because the expression is more strongly influenced by population than by genotype . In a more extreme situation , even a reversal of the genetic association can happen , with the same data leading to opposite conclusions . When incorporating the populations in an ANCOVA , the clear link between SNP and gene expression becomes evident again ( p = 2 . 5×10−54 ) . This demonstrates that biological or technical factors may mask important effects and that we must be careful when aggregating data . The second data set refers to the non-transitivity property of correlation ( Fig 3B ) . Here , we investigated a three-way relationship between two RNAs and a SNP ( n = 20 individuals per genotype ) . In these simulated data , individuals with a higher allele count tend to show higher levels of the first RNA ( Pearson coefficient r = 0 . 472 and p = 0 . 0001 ) , and the levels of the first RNA and the second RNA are significantly positively correlated as well ( r = 0 . 508 and p = 0 . 00003 ) . One might thus assume that a higher allele count is also associated with higher levels of the second RNA . However , this reasoning fails . In fact , the correlation between the SNP allele and the second RNA is negative ( r = -0 . 471 and p = 0 . 0001 ) . This seems to be a paradox , but it actually reflects a common misconception concerning transitivity of correlation [24] . The simulations thus provided examples of two statistical phenomena that may occur in eQTL studies . In both cases , there are plausible explanations that can help resolving the paradoxes . Therefore , causal considerations are necessary to avoid contradicting conclusions . We employed the HTS data from the Geuvadis consortium [28] to expose similar cis-miR-eQTLs as for hsa-miR-548ac . The data contained the levels of mature miRNAs from a total of 741 miRNA stem-loops and their respective host genes . For 63 of those , we found that the expression of the miRNAs depends on the genotypes of SNPs mapping to the stem-loop sequence or the flanking bases up to positions -25/+25 . Nine SNPs were each associated with the production of two mature miRNAs from the same locus—one from the 5' strand and one from the 3' strand of the precursor—either in the same direction or in the opposite direction ( e . g . , SNP rs2910164 in hsa-mir-146a ) . For 7 stem-loop regions , 2–3 SNPs were filtered ( not just one ) . Altogether , 80 SNP-miRNA associations were detected by ANCOVA with p-value<0 . 05 for the genotype variable ( S1 Table ) . A subset of 74 associations also showed statistical significance in a linear mixed-effects model ( LMM ) analysis ( S1 Table ) . In 11 cases , the eQTL effect was not significant when not controlling for the population structure of the data ( Simpson-like paradox ) . In 36 cases , asynchronous correlations between SNP , miRNA , and primary transcript were observed ( non-transitivity paradox ) , and in 27 cases , the miRNA expression was considerably uncoupled from host gene expression ( S1 Table ) . Such uncoupling suggests that local sequence features modulate the activity of RNA processing factors , although the causal events underlying these associations cannot be unequivocally resolved based on this analysis . Different steps in miRNA biogenesis might be independently affected by genetic variants , in particular transcriptional regulation ( dependent on LD block length ) and RNA cleavage by Drosha ( cropping ) and Dicer ( dicing ) . Drosha/DGCR8-independent mechanisms also play a role as 7 of the stem-loops are mirtrons , which are processed to precursor miRNAs through splicing , lariat debranching , and trimming [29] . It remains speculative whether the allele-specific expression of some of the identified miRNAs is linked to host gene mRNA destabilization . Several of the gene loci have been reported in GWAS to be affiliated with diseases such as schizophrenia ( n = 8 ) and systemic lupus erythematosus ( n = 5 ) ( S1 Table ) . Intriguingly , apart from hsa-mir-548ac , two more of the determined miR-eQTLs ( in hsa-mir-941-4 and hsa-mir-4664 ) are located within a distance of <200 kb from MS susceptibility tag SNPs [14] . Insights into causal relationships might come into reach by further delineating the genetic effects on the levels of the mature miRNAs . As the eQTL analyses showed that the MS risk locus in the CD58 gene is associated with CD58 mRNA and hsa-miR-548ac expression , we speculated that SNP rs1414273 is the causal variant at this locus . Dependent on the genotype of this SNP , there is either a Watson-Crick base pair ( A-U ) or a wobble base pair ( G-U ) at the base of the miRNA stem-loop ( Fig 1B ) [16] . Cell culture experiments and real-time PCR analyses were conducted to examine whether this difference affects the enzymatic processing of the miRNA . After HeLa cells were transiently transfected with precursor miRNA expression vectors carrying either the G allele ( mir-548ac-G ) or the A allele ( mir-548ac-A ) ( Fig 4A ) , precursor transcripts as well as mature hsa-miR-548ac molecules could be detected in all PCR wells ( n = 24 each ) , with an average raw Ct value of 35 . 1 and 34 . 8 , respectively . In contrast , both the precursor RNA and the mature miRNA were not detected ( Ct>45 for 23 out of 24 wells ) after transfection with a negative control vector with scrambled sequence , which implies that they are not endogenously expressed in HeLa cells . For the mir-548ac-G and the mir-548ac-A plasmids , we observed similar changes in expression over the course of time , with levels of the mature miRNA being increased by 96% and 137% and levels of the primary miRNA transcript being decreased by 96% and 93% at 48 h versus 24 h post-transfection , respectively ( Fig 4B and 4C ) . However , when normalizing the mature hsa-miR-548ac levels to the levels of the stem-loop-containing precursor RNA , a 1 . 5-fold ( at 24 h ) and 3 . 4-fold ( at 48 h ) higher relative expression ratio was measured in mir-548ac-G-transfected cells compared with mir-548ac-A-transfected cells ( Fig 4D ) . Although these differences were not statistically significant ( Welch t-test p>0 . 06 ) , the data do support a role of SNP rs1414273 in the recognition and/or processing of the primary miRNA by the Drosha-DGCR8 complex . The real-time PCR data thus corroborated what we have seen in the eQTL analysis: The MS-associated G allele mediates a more efficient biogenesis of hsa-miR-548ac . While the effect of the genetic variant on miRNA expression is moderate , it may influence the regulation of many target genes and , in consequence , constitute an important link to the pathogenesis of MS . Gene expression is post-transcriptionally regulated by miRNAs via destabilization of target gene transcripts and repression of mRNA translation , although indirect mechanisms can also lead to transcriptional activation of specific genes [20 , 21] . We set out to identify direct target genes of hsa-miR-548ac by microarray-based screening , bioinformatic assessment , and experimental validation . For the target screening , HeLa cells were transiently transfected with the mir-548ac-G plasmid or the control plasmid . After 24 h and 48 h , processed mature hsa-miR-548ac molecules could be measured with real-time PCR in cells transfected with the precursor miRNA expression plasmid ( Ct<34 . 2 for all 54 wells ) but not in cells transfected with the control vector ( Ct>45 for 27 out of 36 wells ) . As in the former experiment ( Fig 4B ) , the mir-548ac plasmid evoked higher levels of the miRNA at the second time point ( Fig 5A ) . The gene expression of the cells was then profiled using high density Affymetrix Human Transcriptome Arrays ( HTA ) 2 . 0 . These microarrays contain >6 million oligonucleotide probes matching to 67528 protein-coding and non-coding genes . The analysis of the data ( probe set signal intensities ) revealed overall 333 transcripts , which were statistically significantly downregulated by >33 . 3% ( raw p-value<0 . 05 and fold-change≤-1 . 5 ) in mir-548ac-transfected cells as compared to negative controls ( S2 Table ) . The expression differences were more pronounced 24 h ( n = 257 transcripts ) than 48 h ( n = 99 transcripts ) post-transfection ( Fig 5B ) . Many of the genes are known to be expressed in blood and brain cells , and a functional analysis showed that they are significantly overrepresented in biological processes and pathways related to , e . g . , "cytokine signaling in immune system" , "MAPK family signaling cascades" , "de novo posttranslational protein folding" and "response to unfolded protein" ( S2 Table ) . Interestingly , a relatively huge fraction of transcripts ( n = 50 ) was found to be paralogous to the mir-548 family . Their downregulation suggests that there are mutual interactions between the miRNA and various other small RNAs , which presumably evolved from the same palindromic transposable element [30] . Of the 333 potential target transcripts of hsa-miR-548ac , 6 were already listed as such in the miRTarBase catalog [31] with weak evidence from HTS data , and 35 were consistently predicted to contain a binding site for the miRNA in the 3' untranslated region ( UTR ) using the miRWalk 2 . 0 platform [32] ( S2 Table ) . These small overlaps are not surprising due to the lack of studies on this primate-specific miRNA and the common assumption of prediction tools that binding sites should be evolutionarily conserved . We prioritized the 333 transcripts in order to select probable direct protein-coding targets for a validation experiment . Nine genes showed consistency of downregulation , strong expression in HeLa cells , and a stable interaction of the 3' UTR with hsa-miR-548ac as calculated by RNAhybrid [33] ( Fig 5C , S2 Table ) . Luciferase reporter assays were then performed with 3' UTR constructs for 3 of these genes: SDC4 , SEL1L , and TNFAIP3 . HeLa cells transfected with these constructs always produced lower relative luminescence intensities than HeLa cells transfected with a control vector containing a minimal 3' UTR . This suggests that the 3' UTRs of these genes contain binding sites for other miRNAs or other sequence features that affect mRNA stability [34] . When comparing the effect of the mir-548ac-G plasmid versus the plasmid with scrambled sequence , a significant decrease in luciferase activity was seen for SDC4 ( -19 . 8% ) , SEL1L ( -23 . 7% ) as well as TNFAIP3 ( -31 . 5% ) 24 h after cotransfection ( Fig 5D ) . As in the microarray-based screening , the regulatory effect was still present , albeit less pronounced , at the 48 h time point ( Welch t-test p-value<0 . 05 only for SEL1L ) . We could thus verify direct interactions between hsa-miR-548ac and selected candidate target genes . This establishes a functional link , where the MS-associated locus within the CD58 gene may exert its effect on disease risk by post-transcriptional modulation of gene expression via hsa-miR-548ac . The data also indicated numerous interactions between the miRNA and related non-coding transcripts , which may cause secondary effects that remain to be delineated . Because we could detect hsa-miR-548ac molecules in all PBMC samples in the miR-eQTL analysis , we were interested in finding out which specific immune cells express this miRNA and to which extent it is coexpressed with CD58 . For this purpose , we analyzed publicly available microarray and RNA-seq data for a variety of blood cell subpopulations , including T-cells , B-cells , monocytes , and natural killer ( NK ) cells . On the one hand , we used transcriptome profiles obtained using Affymetrix microarrays for leukocyte populations , which were separated from blood samples via cell surface markers [35] . The microarrays provided expression values for CD58 but not for hsa-miR-548ac . In these data , CD58 was expressed at high mRNA levels in NK cells , monocytes , and neutrophils , followed by effector memory T-cells and myeloid dendritic cells ( DCs ) ( Fig 6A ) . On the other hand , we used the human miRNA expression atlas , which has been built based on short RNA-seq data from the Functional Annotation of Mammalian Genome 5 project ( FANTOM5 ) [36] . This expression atlas allowed us to compare the levels of hsa-miR-548ac across 118 different cell types and tissues . The abundance of this miRNA was relatively low in 113 of the analyzed samples ( z-score of expression <0 . 15 ) . However , enriched expression ( z-scores >0 . 15 ) was seen in CD8+ T-cells , NK cells , monocytes , and neutrophils ( Fig 6B ) as well as in endothelial progenitor cells , which are extremely rare in normal peripheral blood [37] . Therefore , the expression signatures of hsa-miR-548ac and its host gene CD58 were overall similar , with elevated levels in diverse immune cells , but notable differences could be ascertained in particular for CD8+ T-cells and myeloid DCs . The observation that the miRNA and the mRNA are not necessarily proportionally produced across cell subsets suggests that cell type-specific mechanisms may modulate the processing of both molecules from the same primary transcript and/or affect their turnover . Such partial uncoupling of expression could yield a somewhat more diversified miRNA output .
In this study , we investigated the regulatory effects of the MS-associated haplotype within the CD58 gene locus . The MS risk allele was associated with significantly lower CD58 mRNA levels in the HapMap microarray data and in the Geuvadis RNA-seq data . On the other hand , significantly increased levels of hsa-miR-548ac were seen in risk allele carriers in the PCR data and in the Geuvadis data . These findings suggested a genotype-dependent processing of the intronic miRNA stem-loop by Drosha-DGCR8 , a mechanism which likely explains decoupling of eQTL effects for other miRNAs as well . The data exhibited two paradoxes , which may serve as examples illustrating the pitfalls of aggregating data , an issue of particular relevance in meta-analyses and multi-omics studies . Our study is the first that explicitly measured mature hsa-miR-548ac molecules in the blood of MS patients , and it is also the first that specifically aimed to identify the target genes of this miRNA . These analyses shed light on the role of this particular miRNA in MS , but also pointed out general challenges in elucidating the genetic basis of autoimmune diseases . A relatively high proportion of MS-associated SNPs ( 25–50% ) has been recently shown to be in strong LD with cis-eQTL SNPs established in blood [38–41] . In line with this , we here described an eQTL shared across multiple populations and with opposing effects on the levels of CD58 mRNA and hsa-miR-548ac in blood-derived cells , while notably the block of LD for this association does not include the gene's promoter region [16 , 41] . Previous studies already indicated that the haplotype , which is overrepresented in subjects with MS , is linked to moderately reduced CD58 mRNA and protein levels in LCLs [15 , 22 , 42] . We here confirmed this colocalization signal in data from more than 1000 individuals . We also showed that the eQTL effect can be obscured by differences in gene expression and genotype distribution across global populations . Moreover , we discovered a non-intuitive inverse relationship for the production of mature miRNA molecules from the same primary transcript . The further verification of this phenomenon is currently complicated by the lack of eligible miR-eQTL mapping data . In fact , the microarray- and PCR-based measurement platforms used in the majority of studies do not contain an assay for the mature form of hsa-miR-548ac , which can be explained by the fact that this miRNA is recorded in the miRBase database only since release 17 ( accession MIMAT0018938 ) [43] . However , the rapidly increasing breadth and depth of HTS data should deliver deeper insights into genetic regulatory elements of non-coding and short RNAs ( sRNA ) in the near future . We performed cell culture experiments to gain further evidence towards our hypothesis that the causal SNP of the CD58 gene locus is located in the miRNA stem-loop sequence . In these in vitro analyses , the ratio of mature hsa-miR-548ac miRNAs relative to precursor transcripts was up to 3 . 4 times higher for the MS risk allele G than for the allele A of SNP rs1414273 . This difference most likely reflects an influence of this SNP at position +11 of the basal stem on the efficiency of miRNA cropping by the Drosha-DGCR8 complex . On the other hand , multiple steps affect biogenesis and turnover of miRNAs , e . g . , nuclear export , Dicer-mediated processing , RNA-induced silencing complex ( RISC ) formation , and miRNA decay [29] . Therefore , kinetic assays may be employed to quantify the genotype-dependent velocity of Drosha cleavage more accurately [44] . Others have shown that specific short sequence motifs and structural features are enriched in efficiently cleaved human primary miRNAs [20 , 45–47] . These findings pointed toward the base of the stem region ( around position -13/+11 ) as a regulatory hub for miRNA processing . At the same time , a growing number of cofactors were found to enhance cleavage activity [29 , 48] . For instance , the prevalent CNNC motif that is also present downstream of hsa-mir-548ac ( position +18 to +21 ) is bound by SRSF3 , which is a splicing factor [45] . It is important to note that all major RNA processing activities occur cotranscriptionally at RNA polymerase II transcription sites . This includes the initial processing of miRNA stem-loops , which usually precedes intron splicing [46 , 49] . Introns are known to contain regulatory elements that enhance or silence exon recognition . Exonucleolytic degradation of cleaved intronic sequences thus can affect the splicing regulation of nascent transcripts [50] . As a consequence , the maturation of numerous mRNAs is expected to be influenced by Drosha activities in the nucleus [51] , even if sometimes no authentic precursor miRNAs are generated for subsequent cytoplasmic processing by Dicer . While these events may provide a reason why hsa-miR-548ac miRNA levels and CD58 mRNA levels are inversely affected by the genotype , it still remains to be elucidated which precise auxiliary factors for splicing and Drosha-mediated miRNA processing may contribute to this MS-associated eQTL and whether other SNPs play a role as well . The cooperative control of the cleavage by various RNA-binding proteins allows a partial uncoupling from host gene expression and diversification of mature miRNA production from the shared primary transcript [51] . Accordingly , a transcriptome-wide analysis of chromatin-associated RNAs by Conrad et al . demonstrated that miRNA processing and basal transcription are independently regulated processes [46] . This is of relevance as human genetic variants can lead to structural differences in the stem-loop and in the flanking regions , thereby interfering with the recognition of primary miRNAs . In case of SNP rs1414273 , there is either a Watson-Crick base pair ( A-U ) or a wobble base pair ( G-U ) at the base of the helix . Similar base substitutions have already been investigated for hsa-mir-142 [52] and hsa-mir-510 [53] . The cleavage of hsa-mir-142 precursor transcripts was also shown to be modulated in response to A-to-I editing by adenosine deaminases [52] . We identified several other SNPs with potential impact on miRNA biogenesis in the Geuvadis HTS data . This cis-miR-eQTL analysis revealed miRNAs encoded in the vicinity of MS-associated loci ( miR-941 and miR-4664-3p ) [14] , miRNAs previously found to be dysregulated in MS ( miR-146a-5p and miR-629-5p ) [54] as well as other mir-548 family members ( miR-548j-3p/-5p and miR-548l ) . Substantial uncoupling from host gene expression was seen for 27 mature miRNAs . However , the analysis was limited to autosomal SNPs and to miRNAs being expressed in LCLs from annotated primary transcripts . Further research is needed to obtain a better understanding of the sequence determinants and the factors that causally influence the level of miRNAs , which may provide deeper insights on the misregulation of miRNAs in disease . In the eQTL analysis , we encountered a Simpson-like paradox , as the genotype effect on CD58 gene expression disappeared in the LCL data when not controlling for population substructure . Simpson's paradox refers to a sign reversal of an association between two variables , with a trend appearing in one direction for separate groups but in the opposite direction when the groups are combined [23] . We did not observe an actual reversal of the relationship between genotype and gene expression but an unexpected masking of the eQTL effect when aggregating the data of the human subpopulations . This phenomenon is of general relevance in data science and it is a reminder against simplistic reasoning , demonstrating that caution is needed when describing relations and their implications . The paradoxical element arises from incorrect assumptions and incomplete causal information . Biological and technical reasons might explain the differences in CD58 mRNA levels in LCLs between different global populations . However , it is usually not known which factors have to be included in the analysis to check whether a relationship in the data is consistent . The data can be partitioned in many ways , e . g . , by gender and age of the subjects , by lab and technician involved in preparing the samples , or by means of cluster analysis . Thus , simply pooling the data of individual study cohorts can yield spurious results [55] , and various elaborated statistical models have been developed to account for confounding sources of expression variation [56 , 57] . The literature provides examples of eQTLs that are cell type-specific , context-specific , and population-specific [38 , 58–63]–attributes that are not yet well represented in eQTL databases . For instance , SNP rs738289 constitutes an eQTL to two different genes from the same locus , depending on cell type: In B-cells , this SNP is associated with MGAT3 expression , whereas in monocytes , this SNP is associated with SYNGR1 expression [58] . As another example , the strength of the eQTL SNP rs285205 on MYBL2 expression is modulated by the expression of the transcription factor EBF1 [38] . The significance of a SNP might be also masked by the effects of other nearby genetic variants [64 , 65] , which requires a conditional analysis to delineate the independent associations . Moreover , the different alleles of a SNP can be related to the expression of different RNA isoforms of the same gene ( exon-level eQTLs ) [38] . If such effects are not taken into consideration , a Simpson-like paradox may occur and conceal the regulatory role of a genetic variant . We also noticed the non-transitivity of correlation [24] , as hsa-miR-548ac and CD58 mRNA , which showed elevated and reduced levels in risk allele carriers , respectively , originate from the same transcript . The simulated data illustrated the paradox that discrepant directions of effect are possible in pairwise correlations between three quantitative variables . With regard to the eQTL SNP rs1414273 , the relative contributions of miRNA cleavage efficiency and primary transcript expression to mature miRNA biogenesis determine the extent of uncoupling from the expression of the host gene CD58 . Similar relationships can result from SNPs in bidirectional promoters or promoter-interacting regions of accessible chromatin . For instance , the alleles of SNP rs71636780 were shown to have opposite effects on the expression of ARID1A and ZDHHC18 [60] . Conversely , it is possible that two SNPs in weak LD have an independent and antagonistic regulatory influence on the level of the same transcript . Another remarkable example is the inverse miR-eQTL association that we noted for SNP rs2910164 and the two mature miRNAs from the 5' arm and the 3' arm of hsa-mir-146a . One should also keep in mind that correlations are not in general transitive when analyzing multiple types of molecular data ( e . g . , genetic variants , DNA methylation , chromatin state , and RNA and protein expression ) . Hence , proving causal links between different molecular changes is often challenging . Paradoxical conclusions might arise if the data is not supplemented with extra information of the context . In our study , we aimed at further uncovering the biological implications of the MS-associated genetic locus captured by the SNP rs1414273 . Whereas the host gene CD58 is known to encode a costimulatory membrane-bound protein [15 , 16] , the functions of the intronic miRNA have not been established so far . The characterization of this particular miRNA is impeded by the fact that it belongs to a huge primate-specific family , which evolved from a transposable element [18] . As a consequence , mir-548 family members overlap in their function and probably interact with each other [30] . Their sequence similarities have complicated their discovery and their accurate annotation in the genome , and as several of these miRNAs seem to be expressed at low levels , methods with high sensitivity and specificity are needed . For the target gene analysis of hsa-miR-548ac , we employed a precursor miRNA expression plasmid , which triggers the production of mature miRNA molecules via the endogenous miRNA biogenesis pathway . However , in comparison to the natural hsa-mir-548ac transcript , this expression construct may not adequately represent the influence of processing cofactors such as spliceosomal proteins . It is also important to note that alternative maturation might lead to different miRNA isoforms , possibly even from the 5' arm of the duplex , which may recognize different target sites . In addition , the reporter gene might carry over unprocessed miRNA stem-loops to the cytoplasm . Nevertheless , we assumed that the main regulatory effect is mediated by the canonical mature form of hsa-miR-548ac as annotated in miRBase [43] and as measured by real-time PCR . While our study benefited from a transcriptome-wide approach , we had to accept diverse limitations: ( 1 ) hsa-miR-548ac may suppress protein translation of some targets with little effect on mRNA stability , ( 2 ) fragments of degraded target RNAs may still hybridize on the microarrays , ( 3 ) feedback mechanisms may counteract the regulatory effect of the miRNA , ( 4 ) experimental conditions change over time ( e . g . , transient expression , miRNA turnover , and cell viability ) , and ( 5 ) we missed target genes not expressed in HeLa cells and non-human targets ( e . g . , viral genes ) . The screening identified 333 potential primary and secondary target transcripts , of which 23 transcripts were filtered at both time points ( 24 h and 48 h ) . Drosha , DGCR8 , and CD58 itself were not among the significantly downregulated transcripts ( p>0 . 20 ) , but there were several small non-coding RNAs with mir-548-related palindromic sequence . For 3 prioritized protein-coding genes , we could validate the binding of hsa-miR-548ac at the 3' UTR with luciferase-based assays . SDC4 is a transmembrane receptor that regulates focal adhesions [66] as well as B-cell migration and germinal center formation [67] . SEL1L guides misfolded proteins to endoplasmic reticulum-associated degradation [68] and manages a checkpoint in early B-cell development [69] . TNFAIP3 is an ubiquitin-editing enzyme involved in cytokine-mediated immune responses . Its role in various innate and adaptive immune cells has been well established [70] , and multiple SNPs in the vicinity of the TNFAIP3 gene have been associated with MS ( e . g . , rs17780048 [14] ) as well as other autoimmune diseases [70] . The gene set enrichment analysis revealed that also other putative target genes of hsa-miR-548ac are known to promote inflammatory responses and/or the folding and degradation of proteins ( e . g . , DNAJC3 , HERPUD1 , and members of the Hsp70 chaperone family ) . The filtering of several interferon-induced genes ( e . g . , IFI6 and OAS1 ) may correspond to previous studies indicating that mir-548 members downregulate , at least indirectly , antiviral mechanisms [71 , 72] . Further efforts are needed to elucidate the complex molecular network regulated by these miRNAs in order to achieve a better understanding of their role in health and disease . To obtain further insights on hsa-miR-548ac , we had a closer look at its expression profile across cell types and tissues . The expression levels were taken from the new FANTOM5 miRNA expression atlas [36] . Compared to other data sets , this atlas provides more accurate information and includes also novel miRNAs expressed at low levels . These improvements have been achieved by advances in RNA-seq , which led to an increase in sequencing depth from nearly 1300 reads per sRNA library in the first miRNA atlas [73] to nearly 4 . 4 million reads per library in FANTOM5 [36] . The FANTOM5 data clearly indicated that hsa-miR-548ac is preferentially expressed in immune cells from the blood . We also found that CD58 mRNA is similarly expressed in a range of immune cells such as CD8+ T-cells , NK cells , monocytes , and neutrophils . This is a consequence of the fact that the same primary transcript is subjected to both Drosha-DGCR8 and splicing activities . However , we did expect higher levels of hsa-miR-548ac in B-cells , where it has been identified for the first time by Jima et al . According to their data , hsa-miR-548ac is mainly produced by lymphoma cell lines , EBV-transformed B-cells , and memory B-cells [19] . It is therefore possible that this miRNA is specifically expressed in memory B-cells but not in naive B-cells , which is the major B-cell subset in blood , especially in young people [74] . As naive and memory cells were not distinguished in FANTOM5 , this remains to be investigated in further studies . Interestingly , compared to host gene expression , levels of hsa-miR-548ac were relatively high in CD8+ T-cells and relatively low in myeloid DCs . Such partial uncoupling of cell type-specific expression of CD58 mRNA and hsa-miR-548ac , if validated , might be explained by cofactors of Drosha-mediated cleavage [29 , 46] . Auxiliary factors for cotranscriptional splicing , RNA editing , and RNA processing may also modify the eQTL effect of SNP rs1414273 in different cell populations and at distinct stages of development . The regulation of transcription and mRNA/miRNA maturation and turnover thus deserves subsequent investigation , if possible , at single-cell resolution . In the past years , thousands of associations between genetic loci and human diseases have been discovered [1] , with a large number of risk factors being shared across different neurological and immunological conditions [58 , 60 , 75] . However , their functional interpretation remains a challenge . Ongoing research on the causal mechanisms will deliver further insights into the pathogenesis and maintenance of such diseases and their treatment , but it has to be considered that the underlying dynamic processes constantly influence each other and thus exhibit a complex chaotic behavior [2] . One approach for addressing this issue is to employ mathematical methods that better estimate the trajectories of the biological system by integrating multiple levels of data . Fortunately , a huge diversity of large-scale molecular data sets is generated by multinational initiatives [76 , 77] for mapping genetic effects on gene expression , splicing , DNA methylation , histone modifications , etc . Some of these efforts also included the analysis of miRNA expression profiles from hundreds of individuals [28 , 78] . Still , because MS susceptibility loci may alter different processes in blood and brain , a more detailed view on specific cell types is needed to expedite the identification of causal regulatory variants . Moreover , it is important to examine combinations of genetic factors and the molecular interplay with environmental and lifestyle factors ( e . g . , nutrition ) . Studies have shown that interactions of HLA risk alleles with smoking , EBV infection , and adolescent obesity result in odds ratios for MS of >14 , which is higher than the sum of the individual effects [9] . The local environment can affect disease risk phenotypes by modulating eQTL effects [63] . It is therefore difficult to disentangle beneficial and adverse gene-environment relationships . The finding that most human mRNAs are regulated by miRNAs suggests that they influence the majority of developmental processes and pathologies [20] . Intriguingly , a number of miRNAs are encoded near MS-associated tag SNPs ( e . g . , hsa-mir-26a-2 and hsa-mir-934 ) [14] , and the MS risk haplotype from hsa-mir-548ac predisposes for neuromyelitis optica [79] and primary biliary cholangitis [80] as well . A better understanding of how processing , modification , and target binding of miRNAs depend on genetic variation and external stimuli is essential and may help defining biomarker constellations with large effect sizes concerning disease risk , severity , prognosis , and therapy response , which would aid in counseling patients and their relatives . To conclude , our analysis suggests that SNP rs1414273 is possibly implicated in the development of MS . The MS-associated allele was confirmed to be linked with moderately reduced CD58 mRNA expression in blood-derived cells from more than 1000 individuals . On the contrary , HTS and real-time PCR data revealed increased levels of mature hsa-miR-548ac molecules in carriers of the risk haplotype , which we attribute to a genotype-dependent processing of the miRNA stem-loop from the first intron of the CD58 gene . We showed that hsa-miR-548ac is preferentially expressed in immune cell populations , and we found that some of its determined target genes participate in inflammation and proteostasis . Moreover , we discussed that eQTLs can be cell type-specific , context-specific , and population-specific , giving rise to paradoxical phenomena . Other disease susceptibility loci similarly contain SNPs that potentially impact the efficiency of primary miRNA cleavage and cotranscriptional splicing . As single miRNAs are able to regulate hundreds of transcripts , their aberrant expression is expected to play a critical role in various pathological conditions . The deeper investigation of disease mechanisms thus requires further studies on how genetics and environment affect cooperative miRNA-target interaction networks .
We investigated the effect of the MS-associated allele on CD58 mRNA expression using microarray data generated by the Wellcome Trust Sanger Institute for human cell lines derived from individuals that have been genetically well-characterized in the international HapMap project [25] . This analysis is very similar to the one previously performed on HapMap Phase I data [81 , 82] by De Jager et al . [15] . The main difference is that , since then , much more microarray data have been produced . Meanwhile , gene expression levels in EBV-transformed LCLs of 730 individuals from HapMap Phase II + III [25] have been measured using Illumina Human-6 v2 Expression BeadChips [26 , 27] . The individuals represent 8 human populations and the numbers per population are n = 112 CEU ( Caucasians of northern and western European ancestry living in Utah , USA ) , n = 80 CHB ( Han Chinese from Beijing , China ) , n = 82 GIH ( Gujarati Indians in Houston , TX , USA ) , n = 82 JPT ( Japanese in Tokyo , Japan ) , n = 83 LWK ( Luhya in Webuye , Kenya ) , n = 45 MEX ( Mexican ancestry in Los Angeles , CA , USA ) , n = 138 MKK ( Maasai in Kinyawa , Kenya ) , and n = 108 YRI ( Yoruba in Ibadan , Nigeria ) . The data in raw and normalized form are publicly available and have been downloaded from the ArrayExpress database ( series accession numbers E-MTAB-198 and E-MTAB-264 ) . With Illumina BeadChip technology , levels of RNA ( but not miRNA ) are quantified by hybridization to gene-specific 50mer probes . The probe identifier for CD58 is ILMN_1785268 , with the corresponding oligonucleotide sequence spanning exon 2 and exon 3 . Normalized signal intensity values in log2 scale were used as input for the eQTL analysis . The genotype data were obtained from HapMap ( Phase II + III ) [25] and the 1000 Genomes project ( Phase 3 ) resource [83] . SNP rs1335532 was used for distinguishing the MS risk allele ( SNP rs1414273 was not covered in HapMap ) . Genotypes and expression data were both available for LCLs of 726 individuals . SNP information was missing for four samples ( CEU: NA12274; MKK: NA21306 , NA21443 , and NA21649 ) . The eQTL analysis was performed in the statistical environment R using linear models with genotypes ( numerical ) and populations ( categorical ) as independent variables ( either alone or in combination ) and gene expression as dependent variable . F-tests were calculated for all main and interaction effects based on type II sums of squares with the Anova ( ) function of the car package for R [84] . This analysis is equivalent to SLR when using the genotype variable only . A one-way ANOVA was conducted when using the population variable only . An ANCOVA was performed for the model including both variables and a genotype-by-population interaction term . Moreover , two-sample two-tailed Welch t-tests were used for pairwise comparisons of genotype groups to further inspect the alleles' effect on CD58 expression . Statistical significance was defined as p-value<0 . 05 . For further investigating the eQTL , we used data obtained by high-throughput RNA sequencing . A huge mRNA and small RNA sequencing analysis has been conducted by the Geuvadis consortium [28] on 465 LCL samples from 5 populations of the 1000 Genomes project [83] . The numbers per populations are n = 92 CEU , n = 95 FIN ( Finnish in Finland ) , n = 96 GBR ( British in England and Scotland ) , n = 93 TSI ( Toscani in Italia ) , and n = 89 YRI . A subset of n = 86 CEU ( 93 . 5% ) and n = 81 YRI ( 91 . 0% ) individuals of the Geuvadis cohort also belongs to the HapMap cohort mentioned in the previous section . The HTS data were generated using the Illumina HiSeq 2000 platform , with paired-end 75 bp mRNA-seq and single-end 36 bp small RNA-seq . After applying quality control filters , mRNA and miRNA data were processed for 462 and 452 individuals , respectively . The Geuvadis RNA-seq data and quantification files as well as the complete genotype data are freely available in ArrayExpress ( series accession numbers E-GEUV-1 and E-GEUV-2 ) [28] . For CD58 gene expression levels ( Ensembl identifier ENSG00000116815 ) , we downloaded the reads per kb per million mapped reads ( RPKM ) normalized for sequencing depth and transcript length . For hsa-miR-548ac levels , we downloaded the read counts normalized by the total number of miRNA reads . The values of replicates were averaged . Similar to the HapMap cohort analysis , we then tested whether the regulation of CD58 mRNA and hsa-miR-548ac is affected by genetic variation . The mRNA-/miR-eQTL evaluation was performed with SNP rs1414273 genotypes using F-tests of linear models as well as Welch t-tests as described for the HapMap cohort analysis . The Geuvadis data [28] were also used for a genome-wide cis-miR-eQTL analysis . For all miRNAs ( as of miRBase release 21 ) that were measured in these data together with the respective host gene ( Ensembl release 90 ) , we considered all biallelic SNPs ( dbSNP build 150 ) located in the stem-loop or within 25 bases of flanking sequence up- and downstream of the cleavage site . SNPs in this region can lead to alterations of structural properties that are important for miRNA biogenesis , including the recognition by Drosha-DGCR8 , the export to the cytoplasm , and the processing by Dicer [29 , 45–47] . ANCOVA were calculated for both the miRNA and the annotated primary transcript , and all SNPs with a significant genotype effect ( p-value<0 . 05 ) on the levels of the mature miRNA were filtered . As an alternative approach , we fitted LMM that include the population variable as random-effects term ( random intercept model ) by minimizing the restricted maximum likelihood criterion with the lme4 R package [85] and assessed the genotype effect by type II Wald chi-square tests . Additionally , population-adjusted Pearson correlation coefficients r were computed for all pairwise relationships between miRNA , host gene , and SNP . These coefficients , which were based on data from 449 LCLs , were used to check for opposite directions of correlation ( due to non-transitivity ) and uncoupling of miRNA production from basal transcription . A difference between rmiRNA vs . SNP and rhost gene vs . SNP of >0 . 2 was considered as substantial uncoupling . In this case , transcription initiation and miRNA processing are suspected to be controlled in an independent and additive manner . Otherwise , the miR-eQTL association might be largely driven by other SNPs at the promoter region in the same block of LD . For all resulting gene loci , we checked for associated diseases/traits for which there are at least 10 studies in the GWAS catalog ( as of 09/27/2018 ) [1] . We used sensitive real-time polymerase chain reactions ( PCR ) for investigating the putative causal nature of the genetic risk locus within the CD58 gene in blood samples of MS patients . With written informed consent , 20 ml of peripheral venous blood was obtained from 32 MS patients ( n = 27 RRMS patients in remission and n = 5 SPMS patients , n = 21 females and n = 11 males , average age of 48 . 3 years ) . The patients were diagnosed with MS according to the revised McDonald criteria [86] . Routine medical care was provided to all patients . They were treated and monitored according to the European Medicines Agency labels , following the consensus treatment guidelines and recommendations of the German Society of Neurology . The study was approved by the University of Rostock's ethics committee ( permit number A 2014–0112 ) and carried out in line with the Declaration of Helsinki . DNA was purified from the blood samples using the PAXgene Blood DNA Kit ( Qiagen ) . PBMC were separated using a Ficoll gradient ( Histopaque-1077 , Sigma-Aldrich ) and stored in mirVana lysis buffer at -20 °C . Total RNA enriched with small RNAs was isolated using the mirVana miRNA Isolation Kit ( Thermo Fisher Scientific ) . Genotyping was done by PCR with allele-specific TaqMan probes for SNP rs1414273 , which is located at the base of the hsa-mir-548ac stem-loop sequence [16] . The Custom TaqMan Assay Design Tool ( Thermo Fisher Scientific ) was used to construct the probes and primers ( forward primer: 5'-ACCTGGTATTAAAAAGTGGAACATAAAATCTCT-3' , reverse primer: 5'-ATGGCAAAAACCGGCAATTACTTT-3' , VIC dye-labeled TaqMan minor groove binder ( MGB ) probe: 5'-TGCACTAACCTAATAGTTAC-3' , FAM dye-labeled TaqMan MGB probe: 5'-TGCACTAACCTAGTAGTTAC-3' ) . PCR amplification was carried out following manufacturer's instructions in a 7900HT Fast Real-Time PCR System ( Applied Biosystems ) . Endpoint analysis was used to determine the genotypes of the individual patients by allelic discrimination . Quantitation of transcript levels and miRNA expression was performed with TaqMan single-tube assays from Thermo Fisher Scientific . We measured CD58 ( assay identifier Hs01560660_m1 , amplicon spanning exon 1 and exon 2 ) and hsa-miR-548ac ( 464325_mat ) as well as GAPDH ( Hs99999905_m1 ) and hsa-miR-191-5p ( 002299 ) [87] as reference gene and miRNA , respectively . From each sample , 400 ng of total RNA was reverse transcribed with random primers for mRNA species ( High-Capacity cDNA Reverse Transcription Kit ) , and 10 ng of total RNA was reverse transcribed with specific primers provided with each TaqMan miRNA assay to convert mature miRNAs to cDNA ( TaqMan MicroRNA Reverse Transcription Kit , Thermo Fisher Scientific ) . The quantitative real-time PCR was then run in triplicates with the predesigned primers and TaqMan probes according to the manufacturer's protocols with 45 cycles in the 7900HT instrument ( Applied Biosystems ) . Threshold cycle ( Ct ) values were computed using SDS 2 . 3 and RQ Manager 1 . 2 software ( Applied Biosystems ) ( S3 Table ) . For mRNA and miRNA data preprocessing , we calculated the mean Ct value of each triplicate , normalized the means to those of reference gene and miRNA ( ΔCt method ) , converted the values to the linear scale using the equation 2-ΔCt , and scaled the result by a factor of 1000 for convenience . Afterwards , SLR was used for the eQTL analysis based on CD58 gene expression , hsa-miR-548ac levels , and the number of SNP rs1414273 risk alleles per patient . Welch t-tests at the 0 . 05 level of significance were calculated for pairwise comparisons of genotype groups . We simulated genotype distributions and gene expression data to illustrate that counter-intuitive results may be obtained in eQTL analyses . The first scenario is posed by Simpson's paradox and the second by the non-transitivity paradox of correlation . Simpson's paradox ( also called amalgamation paradox ) refers to phenomena , in which a trend appears in different groups of data but reverses when these groups are combined [23] . Formally , a relationship between two variables A and B is seen in one direction when partitioning the data by variable C , but the opposite direction is seen when variable C is ignored . According to the classical definition , the variables A , B , and C are categorical , but the paradox applies to quantitative data as well . Moreover , while Simpson's paradox denotes an actual reversal in the relationship when not controlling for the confounding variable C [23] , we here consider associations that disappear rather than reverse signs upon data aggregation . This more general statistical problem has been referred to as Simpson-like paradox [88] . In eQTL studies , gene expression levels ( variable A ) are compared between genotypes ( variable B ) in data obtained from individuals that can be grouped by demographic information , such as population structure ( variable C ) . Simpson-like paradox then happens if two events occur together: ( 1 ) the number of individuals per genotype × subpopulation group are very different and ( 2 ) the population has a large effect on the measured gene expression . An exemplary data set was simulated in R with 60 individuals for each genotype , with the major allele in one population being the minor allele in the other population , and with a continuous genotype-dependent trend in the gene expression values ( means of 15 , 10 , and 5 for the first population and means of 30 , 25 , and 20 for the second population ) . The data were then analyzed using linear models to demonstrate that the eQTL association disappears when examining the data without accounting for subpopulations . A second exemplary data set was used to show that correlation is not transitive [24] . Due to this , when considering a chain of correlation where A correlates with B and B correlates with C , it does not necessarily follow that A correlates with C . This peculiarity is often regarded as paradoxical , but it is basically just a reminder that correlation is not causation , because the three bivariate relationships may have different underlying reasons . We simulated expression data of two RNAs ( variables A and B ) for 60 individuals equally distributed among the three possible genotypes of a biallelic SNP ( variable C ) . The variables for the two RNAs are quantitative and we also treated the variable for the SNP as quantitative ( representing the number of a certain allele ) . We supposed the RNAs to be positively correlated in expression with some random variation in the data . Additionally , we introduced an inverse dependency on the genotype for the levels of both RNAs . The eQTL effect was then assessed by SLR , which is equivalent to calculating the p-value for Pearson's product-moment correlation . We performed cell culture experiments to analyze whether the genotype of SNP rs1414273 is linked to mature miRNA levels . For this purpose , a miExpress precursor miRNA expression plasmid was purchased from GeneCopoeia ( HmiR1085-MR04 ) and amplified in E . coli . This plasmid contains a 315 bp long DNA fragment from the first intron of the CD58 gene that comprises the miRNA stem-loop sequence as well as flanking regions . This fragment is coexpressed with a reporter gene coding for enhanced green fluorescent protein as a combined transcript . The expression of this construct is driven by a cytomegalovirus promoter ( vector pEZX-MR04 ) . A second plasmid was derived that only differs from the ordered plasmid at the position of SNP rs1414273 . The ordered plasmid encodes the G allele of this SNP at the base of the hsa-mir-548ac stem-loop ( mir-548ac-G ) . We generated a G to A base substitution using the QuikChange Lightning Site-Directed Mutagenesis Kit ( Agilent Technologies ) and two oligonucleotide primers that contain the desired mutation and anneal to the same sequence on opposite strands of the plasmid ( Life Technologies , forward primer: 5'-TTTTGCACTAACCTAATAGTTACTACAAAAA-3' , reverse primer: 5'-TTTTTGTAGTAACTATTAGGTTAGTGCAAAA-3' ) . The mutagenesis was verified by bidirectional sequencing using a GenomeLab GeXP Genetic Analysis System ( Beckman Coulter ) . The sequencing primers were designed as suggested in the vector data sheet from GeneCopoeia and synthesized by Life Technologies ( forward primer: 5'-CCGACAACCACTACCTGA-3' , reverse primer: 5'-ATTGTGGATGAATACTGCC-3' ) . Visual inspection of the obtained chromatograms was done with the Chromas 2 . 6 . 4 software ( Technelysium ) . For the comparative analysis , HeLa cells were transiently transfected with either precursor miRNA expression plasmid ( mir-548ac-G and mir-548ac-A ) using the FuGENE HD transfection reagent ( Promega ) . A vector with a scrambled sequence ( CmiR0001-MR04 , GeneCopoeia ) served as negative control . Total RNA was isolated 24 h and 48 h after transfection with the miRNeasy Mini Kit ( Qiagen ) and an on-column DNase digestion step . All these experimental procedures were done in duplicates , which resulted in 12 different RNA samples . Reverse transcription and quantitative real-time PCR were then conducted basically as described in section "Real-time PCR-based eQTL analysis ( MS cohort ) " . In brief , 400 ng and 10 ng of total RNA from each sample were reverse transcribed for the analysis of mRNA and miRNA , respectively . TaqMan single-tube assays from Thermo Fisher Scientific were employed for the detection of mature hsa-miR-548ac molecules as well as their precursor transcripts . The custom assay for the primary RNA was designed with the Primer Express 3 . 0 software ( Applied Biosystems ) so that the PCR amplicon contains the entire miRNA stem-loop sequence ( forward primer: 5'-GGCTAAAGAAGTGCTCTCAAATAGAAG-3' , reverse primer: 5'-GACCTGGTATTAAAAAGTGGAACATAAA-3' , FAM dye-labeled TaqMan MGB probe: 5'-TTTCATTTCGACATGTATTAGG-3' ) . Additionally , GAPDH and the reference miRNA hsa-miR-191-5p were measured for normalization [87] . The real-time PCR was performed in triplicates with 45 cycles in a 7900HT Fast Real-Time PCR System ( Applied Biosystems ) . The obtained Ct values are provided in S3 Table . The relative abundance of hsa-mir-548ac precursor RNAs and mature miRNAs was assessed based on the ΔCt method . Accordingly , median Ct values per triplicate were normalized to those of reference gene and miRNA , respectively ( ΔCt ) , and then transformed to linear scale ( 2-ΔCt ) . To compare the two SNP rs1414273 alleles with regard to hsa-miR-548ac expression , the mature miRNA levels were normalized to the precursor levels using the 2-ΔΔCt method , with ΔΔCt= ( CtmiR−548ac−CtmiR−191−5p ) − ( Ctprecursormir−548ac−CtGAPDH ) . Finally , the values in linear scale were scaled so that the data for the A allele and the 24 h time point are equal to 1 , respectively . Bar plots were created for displaying the resulting expression ratios . For the initial screening , we conducted a transcriptome profiling analysis of HeLa cell cultures . The cells were transiently transfected with a non-viral precursor miRNA expression plasmid ( mir-548ac-G ) or the negative control with scrambled sequence , and total RNA from these cells was prepared for real-time PCR analysis to confirm in vitro expression of mature hsa-miR-548ac molecules essentially as described in the previous section . The RNA samples were isolated 24 h and 48 h after transfection of the mir-548ac plasmid ( n = 9 replicates ) or the control plasmid ( n = 6 replicates ) . The quantitative real-time PCR was then run with 45 cycles , and if the signal did not pass the threshold , the data were set to Ct = 45 . The resulting Ct values ( S3 Table ) were averaged over triplicates , normalized to the reference miRNA hsa-miR-191-5p [87] , transformed to linear scale , and scaled . A subset of 10 RNA samples was used for the microarray gene expression analysis , with triplicates and duplicates from both time points for mir-548ac-transfected cells and negative controls , respectively . To this end , 210 ng of total RNA from each sample was used as starting material for the Affymetrix GeneChip Whole Transcript Sense Target Labeling Assay protocol , which allows to generate amplified , fragmented , and biotinylated single-stranded sense strand DNA from the entire expressed genome . The hybridization on high-resolution HTA 2 . 0 microarrays ( Affymetrix ) was then carried out for 16 h at 45 °C in a GeneChip Hybridization Oven 645 ( Affymetrix ) following the manufacturer's protocol . After washing and staining in the Affymetrix Fluidics Station 450 , the microarrays were scanned using a GeneChip Scanner 3000 7G ( Affymetrix ) . The scanned images were imported to the Affymetrix GeneChip Command Console version 4 . 0 to extract the signal intensities for the >6 million 25mer oligonucleotide probes per array . The data were further processed using the Transcriptome Analysis Console ( TAC ) version 4 . 0 ( Affymetrix ) with the signal space transformation robust multi-array average ( SST-RMA ) algorithm for background adjustment , quantile normalization , gene-level probe set summarization , and log2 transformation . This resulted in intensity values for 44699 probe sets for protein-coding genes and 22829 probe sets for non-coding genes , including many miRNA precursors . The raw and processed microarray data are publicly available in the GEO database ( accession number GSE120769 ) . Binding of RISC-miRNA complexes to complementary sites primarily mediates RNA decay [20] . We thus examined the microarray data for significantly downregulated transcripts in mir-548ac-G-transfected HeLa cells . The analysis of differential expression was performed according to the default workflow in TAC 4 . 0 , which is based on the limma method with eBayes correction [89] . Fold-changes were calculated in linear scale from robust means per sample group , while in case of downregulation , the ratio has been inverted and multiplied by -1 . For both the 24 h and the 48 h time point , we filtered probe sets with fold-change≤-1 . 5 at the nominal significance level of α = 5% . A series of database-driven analyses was done with the obtained combined list of potential target genes of hsa-miR-548ac . First , we used the web tool Enrichr [90] to determine Gene Ontology biological process categories [91] and Reactome cell signaling pathways [92] with a combined gene set enrichment score >20 . Second , we studied the genes' expression in cell types of the blood and brain based on the microarray data from Novershtern et al . [35] ( GEO series GSE24759 ) and the RNA-seq data from Darmanis et al . [93] ( downloaded via http://celltypes . org/brain/ [94] ) and by demanding an RNA level of >100 . Third , we checked whether the filtered probe sets interrogate transcripts , which are sequence-related to the mir-548 family . For this purpose , the corresponding genetic sequences were fetched from the reference genome ( GRCh37 assembly ) and tested for similarity with human mir-548 stem-loop sequences using the search tool in miRBase release 21 [43] with E-value<10 and score>100 as cutoffs . Fourth , we evaluated whether the genes were also predicted to be regulated by hsa-miR-548ac in the miRWalk 2 . 0 interaction information retrieval system [32] . We requested that at least 6 out of the 12 different algorithms provided by the web-interface consistently predict the binding of the miRNA at the respective 3' UTRs . Fifth , we used miRTarBase release 6 . 1 [31] to determine the overlap with experimentally determined miRNA targets collected from the literature . Because some gene expression changes observed in the screening experiment may result from secondary effects , we prioritized the genes in order to select the most likely direct target transcripts of hsa-miR-548ac for validation . On the one hand , we determined the subsets of genes , which were nominally downregulated at both time points , that is 24 h and 48 h post-transfection with the mir-548ac plasmid , and which showed robust expression in HeLa cells , with an average log2 signal >10 for either group of samples transfected with the negative control plasmid . On the other hand , we used the RNAhybrid program ( version 2 . 2 ) [33] for assessing evidence of direct miRNA-target interactions in silico . RNAhybrid determines the most favorable hybridization of two RNAs . As input , we used the 3' UTR sequences of the downregulated protein-coding transcripts , which were gathered via the BioMart portal of the Ensembl database ( release 89 ) [95 , 96] . Moreover , we opted for a maximum free energy of -15 and a helix constraint for the seed region of the miRNA ( position 2 to 7 ) . The result of this analysis revealed a subset of genes with at least one predicted 3' UTR binding site for hsa-miR-548ac . Luciferase reporter assays were conducted for 3 of the prioritized target genes . For this purpose , miTarget miRNA 3' UTR target plasmids ( vector pEZX-MT05 ) were purchased from GeneCopoeia for SDC4 ( HmiT016661-MT05 ) , SEL1L ( HmiT016743-MT05 ) , TNFAIP3 ( HmiT061984-MT05 ) , and a negative control ( CmiT000001-MT05 ) . These constructs enable the in vitro expression of the respective 3' UTR sequences as insert downstream of a reporter gene ( Gaussia luciferase , GLuc ) , and they contain a second constitutively expressed reporter ( secreted alkaline phosphatase , SEAP ) as an internal control . They were cotransfected with the mir-548ac-G plasmid or the scrambled sequence expression plasmid in HeLa cells in 3 biological replicates . Following transfection for 24 h and 48 h , supernatants were collected , and GLuc and SEAP activities were measured twice using the Secrete-Pair Dual Luminescence Assay Kit ( GeneCopoeia ) and a GloMax-Multi Detection System ( Promega ) according to the manufacturers' recommendations ( S3 Table ) . Afterwards , ratios of the average luminescence catalyzed by GLuc and SEAP were calculated , normalized to those resulting for cotransfections with the 3' UTR control vector , and compared using Welch t-tests . We analyzed microarray and RNA-seq data to compare the levels of CD58 mRNA and hsa-miR-548ac in different immune cell populations circulating in the blood . The microarray data set was generated by Novershtern et al . using Affymetrix Human Genome U133 GeneChips [35] . These data provided gene expression profiles for purified subpopulations of B-cells , T-cells , NK cells , monocytes , granulocytes , and DCs from the blood of healthy volunteers . We downloaded the RMA-normalized data from the GEO repository ( accession number GSE24759 ) and extracted the signal intensities for CD58 as summarized on the basis of 11 different 25mer oligonucleotide probes ( probe set identifier 205173_x_at ) . The cell type-specific abundance of mature hsa-miR-548ac molecules was evaluated using the expression atlas of human miRNAs , which has been recently created as part of the FANTOM5 project [36] . In FANTOM5 , deep sequencing of sRNA libraries from a large collection of human primary cells , cell lines , and tissues has been performed using an Illumina HiSeq 2000 sequencer . The number of sRNA tags to each miRNA locus were counted , normalized to counts per million , averaged across sample donors , and converted to z-scores . The z-scores for hsa-miR-548ac were taken from the mature miRNA expression table provided at the FANTOM5 web interface . Bar plots were used for displaying the levels of hsa-miR-548ac and CD58 mRNA in leukocyte populations .
|
More than 200 genetic loci have been associated with an increased risk of developing multiple sclerosis ( MS ) . Here , we investigated the role of a single-nucleotide polymorphism ( SNP ) , which is located within the microRNA-548ac stem-loop sequence in the first intron of the CD58 gene . We analyzed expression data of blood-derived cells of about 1000 subjects and observed that MS risk allele carriers have reduced CD58 mRNA levels but increased hsa-miR-548ac levels . Our findings suggest that Drosha cleavage activity is affected , perhaps attributable to the specific SNP . This may contribute to partial uncoupling of CD58 gene expression and hsa-miR-548ac production from the shared primary transcript in immune cells . We discovered that the mature microRNA downregulates genes involved in inflammatory processes and in controlling the balance of protein folding and degradation . Our study exemplifies that paradoxical findings can be encountered in the analysis of genetic variants regulating transcription and/or RNA processing .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"multiple",
"sclerosis",
"natural",
"antisense",
"transcripts",
"gene",
"regulation",
"neurodegenerative",
"diseases",
"immunology",
"plasmid",
"construction",
"micrornas",
"clinical",
"medicine",
"demyelinating",
"disorders",
"dna",
"construction",
"molecular",
"genetics",
"molecular",
"biology",
"techniques",
"bioassays",
"and",
"physiological",
"analysis",
"research",
"and",
"analysis",
"methods",
"artificial",
"gene",
"amplification",
"and",
"extension",
"gene",
"expression",
"molecular",
"biology",
"genetic",
"loci",
"microarrays",
"biochemistry",
"rna",
"nucleic",
"acids",
"clinical",
"immunology",
"neurology",
"genetics",
"autoimmune",
"diseases",
"biology",
"and",
"life",
"sciences",
"non-coding",
"rna",
"polymerase",
"chain",
"reaction"
] |
2019
|
A genetic variant associated with multiple sclerosis inversely affects the expression of CD58 and microRNA-548ac from the same gene
|
Pneumococcal conjugate vaccination has proved highly effective in eliminating vaccine-type pneumococcal carriage and disease . However , the potential adverse effects of serotype replacement remain a major concern when implementing routine childhood pneumococcal conjugate vaccination programmes . Applying a concise predictive model , we present a ready-to-use quantitative tool to investigate the implications of serotype replacement on the net effectiveness of vaccination against invasive pneumococcal disease ( IPD ) and to guide in the selection of optimal vaccine serotype compositions . We utilise pre-vaccination data on pneumococcal carriage and IPD and assume partial or complete elimination of vaccine-type carriage , its replacement by non-vaccine-type carriage , and stable case-to-carrier ratios ( probability of IPD per carriage episode ) . The model predicts that the post-vaccination IPD incidences in Finland for currently available vaccine serotype compositions can eventually decrease among the target age group of children <5 years of age by 75% . However , due to replacement through herd effects , the decrease among the older population is predicted to be much less ( 20–40% ) . We introduce a sequential algorithm for the search of optimal serotype compositions and assess the robustness of inferences to uncertainties in data and assumptions about carriage and IPD . The optimal serotype composition depends on the age group of interest and some serotypes may be highly beneficial vaccine types in one age category ( e . g . 6B in children ) , while being disadvantageous in another . The net effectiveness will be improved only if the added serotype has a higher case-to-carrier ratio than the average case-to-carrier ratio of the current non-vaccine types and the degree of improvement in effectiveness depends on the carriage incidence of the serotype . The serotype compositions of currently available pneumococcal vaccines are not optimal and the effectiveness of vaccination in the population at large could be improved by including new serotypes in the vaccine ( e . g . 22 and 9N ) .
The bacterial pathogen Streptococcus pneumoniae ( the pneumococcus ) is a major cause of morbidity and mortality worldwide . Pneumococcal conjugate vaccines ( PCV ) were introduced over a decade ago and have proved highly effective in eliminating vaccine-type ( VT ) pneumococcal carriage and invasive disease ( IPD ) in countries where PCV has been included in the infant immunisation programme [1]–[3] . So far , three different PCVs have been in use: PCV7 ( with vaccine serotypes 4 , 14 , 6B , 9V , 18C , 19F and 23F ) , PCV10 ( additional serotypes 1 , 5 and 7F ) and PCV13 ( additional serotypes 3 , 6A and 19A ) . Despite the successes of PCVs against their respective vaccine types , the overall public health impact of pneumococcal conjugate vaccination remains less clear . In particular , the lost VT carriage has almost invariably been replaced by non-vaccine-type ( NVT ) carriage [4] , [5] . Depending on the invasive potential of the serotypes involved , replacement in carriage leads to a varying degree of replacement in disease and may have undesirable implications on the overall pneumococcal disease burden in the population at large . As population-wide changes in serotype-specific carriage and disease will not fully emerge until several years after the onset of a vaccination programme [6] , mathematical models are indispensable in predicting the implications of serotype replacement on the use of pneumococcal vaccines . To date , dynamic models have been used in describing pneumococcal transmission , serotype competition as the mechanism of replacement and the net effectiveness of vaccination [7]–[9] . In addition , statistical models utilising serotype-specific carriage data and estimates of invasiveness ( probability of disease per carriage episode ) have been applied to predict post-vaccination disease patterns [10]–[12] . These and other authors have pointed out the importance of the invasiveness of replacing non-vaccine types when assessing the net effectiveness of PCV vaccination [3] , [4] , [13] . In this paper , we elaborate the above ideas to develop a concise model for serotype replacement and present a ready-to-use tool for the prediction of patterns in post-vaccination pneumococcal incidence of carriage and disease , based solely on pre-vaccination data on carriage and disease . For a given vaccine composition , corresponding either to a current or a prospective vaccine , we show in detail how the net effectiveness of vaccination under serotype replacement depends on the invasiveness of the vaccine types relative to that of the non-vaccine types . We demonstrate how differences in the invasiveness across serotypes imply that the disease incidence may either decrease or increase after vaccination and introduce a sequential algorithm for the identification of the most optimal additional serotypes to current vaccine formulations .
The data on the prevalence of pneumococcal carriage in Finland originated from Syrjänen et al . [14] ( <2 year olds ) , Palmu et al . [15] ( 4–5 year olds ) and Leino et al . [16] ( 5+ year olds ) . The serotype distribution in carriage for under 5 years olds was based on data from <2 year old children in Finland [14] and for the rest of the population ( 5+ years ) on data from England and Wales [4] . The UK data were available separately for the adolescent ( 5–19 year olds ) and the adult ( 20+ year olds ) populations . We combined these into a single set of serotype proportions corresponding to the 5+ age class by calculating the weighted averages of the proportions in the two age groups with weights 25% and 75% , respectively . The age- and serotype-specific annual average incidence of invasive pneumococcal disease ( IPD ) in 2000–2009 was retrieved from the National Infectious Disease Registry data ( Finland ) . Both carriage and IPD samples were utilised on the serogroup level , except for PCV7 serotypes , for which carriage data on the serotype level were available . IPD samples with serotype 6A were re-analysed to distinguish between serotypes 6A and 6C [17] . This information was not available from the carriage samples . The proportion of 6C carriage among 6A/C carriers was assumed to be one third in both age classes [4] , [18] , [19] . The sensitivity of the results on this assumption was explored by assuming alternative proportions ( 0% and 50% ) . Our analysis of serotype replacement is based solely on age-specific serotype distributions in carriage and disease in the pre-vaccination era . Here , serotype distribution refers to the stationary ( steady-state ) distribution , assumed to be applicable in the pre-vaccination era or under a PCV programme with the current vaccine composition . We first consider a single age stratum within which the proportion of vaccine-type ( VT ) carriage does not have notable trends with age . The pre-vaccination incidence of carriage and disease with serotype i are denoted by ci and di , respectively . The case-to-carrier ratio ri is defined as the probability of disease per carriage episode , i . e . ri = di/ci . At the aggregate level of all vaccine types , the corresponding quantities are For the set of non-vaccine serotypes ( NVT ) , the quantities CNVT , DNVT and RNVT are defined similarly . We note that the aggregate case-to-carrier ratios are weighted averages of the respective serotype-specific case-to-carrier ratios , the weights being the carriage incidences . In the applications of this paper , disease is equivalent to invasive pneumococcal disease ( IPD ) , but the proposed model is applicable to any disease outcome . Our model of serotype replacement is built on two assumptions regarding the new steady-state after vaccination: ( A1 ) the relative serotype proportions among the non-vaccine types are not affected by vaccination ( proportionality assumption ) ; ( A2 ) the case-to-carrier ratios remain at their pre-vaccination levels . It follows from assumptions ( 1 ) and ( 2 ) that the case-to-carrier ratios remain at their pre-vaccination values also for the aggregate VT and NVT sets . Let q be the proportion of VT carriage eliminated by vaccination and p the proportion of eliminated VT carriage that is replaced by NVT carriage in the new steady state . Under assumptions ( A1 ) and ( A2 ) , the disease incidence after vaccination is ( for clarification , see Figure 1 ) The reduction in the disease incidence is thus ( 1 ) As RVT = DVT/CVT and RNVT = DNVT/CNVT , the following expression is equivalent to ( 1 ) : ( 2 ) According to these , the disease incidence can either decrease or increase after vaccination . In particular , whether or not vaccination will be beneficial depends on the magnitude of the case-to-carrier ratio of the vaccine types compared to that of the non-vaccine types . According to equation ( 1 ) , a reduction in the disease incidence requires the ( average ) VT case-to-carrier ratio ( RVT ) be larger than an ( adjusted ) NVT case-to-carrier ratio ( p*RNVT ) . An alternative characterisation follows from ( 2 ) , according to which a disease reduction requires the vaccine types possess a larger share of pre-vaccination disease than their ( adjusted ) share of pre-vaccination carriage . These characterisations are independent of the degree of elimination q . Of note , if the mean duration of carriage is the same for all serotypes , the above expressions are equally valid if carriage prevalences are used instead of the incidences CVT and CNVT .
Rewriting equation ( 2 ) for a single serotype i with q = 1 , we obtain a criterion for the serotype whose addition to the vaccine would result in the largest decrease in the disease incidence . In particular , by substituting DVT = di ( disease incidence of serotype i ) and DNVT = D – di , where D is the total disease incidence , and likewise CVT = ci and CNVT = C - ci , equation ( 2 ) can be written as ( 3 ) The optimal serotype is the one that maximises expression ( 3 ) . Clearly , the best single vaccine type is neither necessarily the one with the highest pre-vaccination incidence of IPD nor the one with the highest case-to-carrier ratio . In fact , the optimal single serotype may not correspond to the highest value of either of these two quantities . Importantly , the optimal type may be different for different age groups . Furthermore , if ci is equal ( or close ) to 0 , the decrease in IPD incidence ( 3 ) is equal ( or close ) to di , i . e . the decrease is determined solely by the IPD of the serotype . Equation ( 3 ) can be applied sequentially to find the optimal vaccine with a given number of serotypes . At each step the current vaccine composition is supplemented by the best single serotype among those not already included in the vaccine . This is repeated until a desired number of vaccine serotypes is reached . In most cases this procedure can be expected to lead either to the maximum possible reduction in IPD or at least to a reduction very close to the maximum . Note that the optimal serotype composition is independent of the assumed level of VT elimination . If the elimination of VT carriage is expected to be incomplete ( q<1 ) and proportion ( 1-q ) *100% of VT carriage remains after vaccination , the projected incidences of carriage and IPD are obtained as weighted averages of the model projections assuming complete elimination and of the original incidences with weights q and 1-q , respectively . A program code implementing the tools proposed above , including instructions on how to use the code , is provided in File S1 . The program is written in the R programming language , which is freely available .
As small changes in the VT/NVT carriage proportions may result in notable shifts in projected IPD incidences , uncertainties in carriage data should be accounted for using sensitivity analysis . When assessing the net effectiveness of vaccination with a given vaccine composition , the effect of a change in the VT/NVT carriage proportions is obtained directly from ( 2 ) . To investigate the robustness of the optimal serotype composition , we calculated the order of inclusion of individual serotypes in the optimal vaccine composition for a large number of sets of carriage proportions , which were generated from an uncertainty distribution . For more details , see File S2 .
We investigated the similarity of pre- and post-vaccination serotype proportions for the non-vaccine types , based on published data from three different locations [4] , [10] , [19] . This analysis shows that the proportionality assumption ( A1 ) is solid as a general rule ( see Figure S1 ) .
We applied the replacement model to the pre-vaccination IPD incidence and carriage prevalence data to predict the post-vaccination IPD incidence in Finland .
We derived a simple expression for the expected net effectiveness of childhood vaccination against invasive pneumococcal disease ( IPD ) under serotype replacement . This expression depends only on the pre-vaccination incidences of vaccine-type ( VT ) and non-vaccine-type ( NVT ) carriage and disease . Our analysis explicates that vaccination will result in a notable reduction in the IPD incidence only if the average case-to-carrier ratio of the vaccine types clearly exceeds that of the non-vaccine types . In Finland , this would be expected to occur among children with any of the currently available PCV formulations . However , the same might not hold to the same extent in the general population not targeted by the vaccination , and our analysis indicates there are vaccine compositions with higher expected net effectiveness . These compositions are projected to be no worse than the current ones among children while clearly outperforming them in older age categories . We formulated the expected net effectiveness of vaccination in terms of serotype-specific incidences of carriage and disease . Equivalently , one could use either of the two quantities together with the case-to-carrier ratios ( i . e . disease incidences divided by carriage incidences ) as any two of the three quantities determine the third one . Importantly , while beneficial vaccine serotypes can be identified using carriage and disease data , they are not necessarily those with the highest carriage incidence , disease incidence or case-to-carrier ratios . The net effectiveness will be improved only if the average VT case-to-carrier ratio is larger than the average NVT case-to-carrier ratio . Furthermore , the above rule is not transparent , unless some serotype had the largest incidence of either carriage or disease and at the same the largest case-to-carrier ratio . In addition , a trivial rule applies in two special situations . First , if all serotypes have identical case-to-carrier ratios , as may be a good first approximation in case of pneumococcal otitis media , there is essentially no change in disease incidence . Second , if there is no replacement in carriage , the expected change in disease only depends on the pre-vaccination disease incidence . Stratification of carriage and disease data by age is essential in the analysis of replacement . For example , an individual serotype included in a vaccine may decrease IPD in one age category while increasing it in another ( cf . serotype 6B among <5 and 5+ year olds in Finland; Figure 2 ) . If an optimal protection for the whole population is of interest , such opposite effects across different age categories may be partly compensated by the inclusion of a sufficient number of serotypes . In particular , while the analysis shows that among the <5 year olds in Finland PCV13 cannot be much improved , the adverse effect among adults of including 6B in the vaccine could be compensated by including additional serotypes ( e . g . 9N , 12 , 22 ) . In practice , evaluation of optimal PCV vaccination for the whole population should be based on a cost-effectiveness analysis that takes into account health benefits and costs in the vaccine target population as well as in the older cohorts . Of note , all of the serotype compositions we consider refer to an infant vaccination programme assuming an adequate level of coverage of vaccination to induce a substantial herd effect in the whole population . The algebraic simplicity of our model is a direct consequence of the two key assumptions that neither the serotype proportions in carriage nor the invasiveness ( case-to-carrier ratios ) of the non-vaccine types are altered by vaccination . Examples of post-vaccination scenarios not covered by our model are a disproportionally large increase in carriage of a previously rare invasive type or an increase in invasiveness of a commonly carried type . Either of these scenarios would increase the IPD in excess of our model predictions . However , there is some empirical evidence in support of the key assumptions . In particular , our analysis confirmed the similarity in pre- and post-vaccination serotype proportions using four different datasets ( Figure S1 ) . The unchanged invasiveness of individual non-vaccine serotypes under vaccination is suggested by their relatively stable case-to carrier ratios in different populations pre-vaccination [3] . The pre- and post-vaccination incidences of carriage and disease involved in our method correspond to the respective stationary ( steady-state ) serotype distributions . These distributions can be characterised by the average annual serotype-specific incidences over a period where they do not manifest any systematic trends . The post-vaccination stationary distribution is typically achieved some years after the onset of a new infant vaccination programme [1] , [2] , [20] . Importantly , if post-vaccination data on IPD are available , the reliability of the model predictions can be monitored at an intermediate stage after the onset of a vaccination programme even before the post-vaccination stationary distribution has been reached . This is achieved by comparing the observed NVT disease reduction to the corresponding predicted reduction calculated under the assumption that the proportion of eliminated VT carriage q in the model equals the observed proportion of eliminated VT disease . There are further assumptions that may pose limitations on the applicability of the proposed method . An identical degree of elimination and replacement in all age classes , i . e . the same values of p and q in equation ( 1 ) irrespective of age was assumed . In reality some differences across age classes may exist , but at least VT elimination has been observed to be nearly complete regardless of age class due to a strong herd effect . Of note , while full elimination in VT carriage is assumed , assumptions regarding vaccine efficacy against disease are irrelevant . We assumed the long-term impact of vaccination on VT carriage is the same for all serotypes , i . e . complete or partial elimination . In addition , NVT carriage was assumed to be affected by vaccination only through replacement and vaccine-induced cross-protection was not included ( for 6A , however , see Table S1 ) . If differences between serotypes with respect to degree of elimination are assumed to exist , they could be taken into account in our model . Furthermore , as carriage data are typically available in terms of prevalences rather than incidences , also uncertainties regarding serotype specific differences in the duration of carriage affect the reliability of model predictions . Some serotypes are identified in carriage samples very rarely ( e . g . serotypes 1 and 5 ) . The current analysis suggests that the evaluation of rarely carried serotypes can rely solely on the disease incidence data . Poor availability or reliability of serotype-specific carriage data across all age classes may limit the applicability of the model . In particular , the predictions are typically sensitive to assumptions regarding the VT carriage proportion . However , carriage data on the adult population are often sparse . Overestimating the VT proportion among adults may lead to underestimation of the effectiveness of vaccination . We demonstrated this by using two alternative VT carriage proportions ( 53% and 62% for PCV10 ) in the non-target population ( individuals 5+ years of age ) in the context of existing PCVs ( Figure 3 ) . In practice , the similarity of the VT proportion among children ( 68% UK vs . 62% in Finland for PCV10 ) suggests that the VT proportion among adults in the UK ( 53% ) should be a good approximation to the corresponding proportion in Finland . Several authors have previously discussed the importance of the invasiveness and the carriage incidences of the vaccine types relative to the non-vaccine types in assessing the net effectiveness of a vaccine [3] , [4] , [13] . Shea et al . [10] applied a replacement model similar to ours to predict the amount of pneumococcal acute otitis media following PCV13 . Another line of related work is based on regression models utilising pre- and post-vaccination IPD incidences [11] or pre- and post-vaccination distributions of carriage [12] . Our results , however , appear to be the first to express the relationship between the invasiveness , the carriage incidences and the predicted IPD incidences in an operational manner that allows for the identification of optimal vaccine types in addition to calculation of their predicted effects if added to the vaccine composition . Moreover , our formulation enables an easy and explicit assessment of the role of age stratification in evaluating different PCV programmes . Based on whole-genomic sequencing data , Croucher et al . [21] suggest that serotype replacement in carriage manifests itself within sequence clusters so that VT strains are replaced by related NVT strains while the cluster specific prevalences are largely unaffected by vaccination . Of note , taking into account within sequence cluster replacement of this type in our model is analogous to the way we have above handled replacement within age categories . We have proposed tools for the quantification of the relative importance of individual pneumococcal serotypes in conjugate vaccine compositions under serotype replacement . Our examples used IPD data from Finland and carriage data from Finland and the UK . Contingent on the availability of data , our tools are easily applicable in other settings as well . However , in contrast to the relative succinctness of the underlying model , the data requirements for a successful application of the proposed tools are not straightforward to satisfy and underline the importance of the availability of age- and serotype-specific data on both pneumococcal carriage and disease .
|
The bacterial pathogen Streptococcus pneumoniae ( pneumococcus ) is a major contributor to child mortality worldwide . Hence , effective pneumococcal vaccination programmes are globally among the most cost-effective public health interventions . Three different conjugate vaccine compositions , targeting 7 , 10 or 13 pneumococcal serotypes , have been used in infant vaccination programmes . The use of these vaccines has both decreased the disease burden and changed the patterns of pneumococcal carriage in locations where they have been in use . However , due to serotype replacement , where the lost vaccine serotype carriage is replaced by carriage of the non-vaccine serotypes , the net effect of vaccination on the disease burden has generally been milder than expected . Here , we apply a concise model for serotype replacement and present a ready-to-use tool for the prediction of patterns in post-vaccination pneumococcal incidence of carriage and invasive disease . We introduce a sequential algorithm for the identification of the most optimal additional serotypes to current vaccine formulations and demonstrate how differences in the invasiveness across serotypes imply that the disease incidence may either decrease or increase after vaccination . The methods we outline have direct relevance in decision making while reviewing the performance of the current pneumococcal vaccination programmes .
|
[
"Abstract",
"Introduction",
"Methods",
"Search",
"for",
"optimal",
"vaccine",
"serotype",
"compositions",
"Uncertainty",
"in",
"carriage",
"proportions",
"Proportionality",
"assumption",
"(A1)",
"Results",
"Discussion"
] |
[
"algorithms",
"medicine",
"bacterial",
"diseases",
"infectious",
"diseases",
"computer",
"science",
"pediatric",
"epidemiology",
"mathematics",
"pneumococcus",
"epidemiology",
"statistics",
"global",
"health",
"applied",
"mathematics",
"infectious",
"disease",
"control"
] |
2014
|
Optimal Serotype Compositions for Pneumococcal Conjugate Vaccination under Serotype Replacement
|
Given that dengue disease is growing and may progress to dengue hemorrhagic fever ( DHF ) , data on economic cost and disease burden are important . However , data for Mexico are limited . Burden of dengue fever ( DF ) and DHF in Mexico was assessed using official databases for epidemiological information , disabilities weights from Shepard et al , the reported number of cases and deaths , and costs . Overall costs of dengue were summed from direct medical costs to the health system , cost of dengue to the patient ( out-of-pocket expenses [medical and non-medical] , indirect costs [loss of earnings , patient and/or caregiver] ) , and other government expenditures on prevention/surveillance . The first three components , calculated as costs per case by a micro-costing approach ( PAATI; program , actions , activities , tasks , inputs ) , were scaled up to overall cost using epidemiology data from official databases . PAATI was used to calculate cost of vector control and prevention , education , and epidemiological surveillance , based on an expert consensus and normative construction of an ideal scenario . Disability-adjusted life years ( DALYs ) for Mexico in 2016 were calculated to be 2283 . 46 ( 1 . 87 per 100 , 000 inhabitants ) . Overall economic impact of dengue in Mexico for 2012 was US$144 million , of which US$44 million corresponded to direct medical costs and US$5 million to the costs from the patient’s perspective . The estimated cost of prevention/surveillance was calculated with information provided by federal government to be US$95 million . The overall economic impact of DF and DHF showed an increase in 2013 to US$161 million and a decrease to US$133 , US$131 and US$130 million in 2014 , 2015 and 2016 , respectively . The medical and economic impact of dengue were in agreement with other international studies , and highlight the need to include governmental expenditure for prevention/surveillance in overall cost analyses given the high economic impact of these , increasing the necessity to evaluate its effectiveness .
Dengue fever ( DF ) is a vector-borne viral infection , the incidence of which has increased and expanded geographically over the past 50 years . The World Health Organization estimates reported a total of 50–100 million infections per year for the period 2010 to 2013 [1] . According to the Global Burden of Disease 2016 Study [2] , there was a significant increase in mortality from dengue between 2006 and 2016; from 20 , 800 deaths ( 95% uncertainty interval [UI] 6000–26 , 500 ) in 2006 to 37 , 800 ( 95% UI 10 , 900–52 , 700 ) deaths in 2016 , while age-standardized rates increased from 0 . 3 deaths per 100 , 000 ( 95% UI 0 . 01–0 . 4 ) in 2006 to 0 . 5 ( 95% UI 0 . 2–0 . 7 ) deaths per 100 , 000 in 2016 . Modeling of the incidence of dengue , accounting for under-reporting of cases , has also showed an increase between 1990 ( 8 . 3 million cases [95% uncertainty estimate 3 . 3 million–17 . 2 million] ) and 2013 ( 58 . 4 million cases [95% uncertainty estimate 23 . 6 million–121 . 9 million] ) [3] . Clinical symptoms of DF can lead to a wide range of clinical manifestations; these are usually mild but some patients may be hospitalized , and progress to a more severe and life-threatening form of the disease that requires admission to an intensive care unit ( ICU ) [4] . The potentially severe consequences of infection , allied with the high prevalence , especially during epidemic years , make for a high burden of disease and high economic cost [5] . Nevertheless , other health issues compete for limited overall resources , so it is important to have reliable figures to quantify as accurately as possible the burden and costs of dengue to enable rational budget allocation . The economic and disease burden of dengue in Mexico has previously been estimated for the period 2010–2011 by Undurraga et al [6] , and was recently updated by Tiga et al [7] who also considered the persistent symptoms of dengue . In addition , Shepard et al [8] estimated the direct and indirect costs of hospitalized and ambulatory dengue episodes from several countries , including Mexico . Mexico has complex arrangements for healthcare in which two main systems , the Secretariat of Health ( SS ) and the Mexican Social Security Institute ( IMSS ) , coexist [9] . We have previously reported the estimated direct costs per case for both the health system and for patients , along with the indirect costs per case , using a micro-costing approach known as the program , actions , activities , tasks , inputs ( PAATI ) method [10] . These costs per case then need to be scaled up to the overall population , and the additional cost of prevention/education programs and other governmental expenses ( such as the cost of surveillance ) need to be considered to obtain a more complete picture of the overall medical and economic impact of dengue in Mexico . This article reports epidemiological data extracted from databases provided by the National Center for Disease Control and Prevention ( Centro Nacional de Programas Preventivos y Control de Enfermedades , CENAPRECE ) , the National System for Epidemiological Surveillance ( Sistema Nacional de Vigilancia Epidemiológica , SINAVE ) and the General Health Directorate . With this information , data from other federal sources on additional costs ( such as prevention/education programs and surveillance ) , and the results of surveys , we report here the estimates for the overall direct medical costs to the health system , costs from the patient perspective , governmental costs of the system , and economic impact of dengue in Mexico for 2012 to 2016 .
Given that only limited analyses of epidemiological data for dengue have been published for Mexico at the time of analysis , epidemiological data were derived from databases provided by CENAPRECE , SINAVE and the General Health Directorate for 2012–2016 . For each of these years , the overall number of cases of DF and DHF and the number of deaths attributed to dengue infection were extracted . The disease burden was estimated for 2016 by calculating three measures of Years of Life Lost ( YLL ) , Years Lost due to Disability ( YLD ) , and Disability-Adjusted Life-Years ( DALYs ) . The YLL were estimated as the difference between the age of death and the life expectancy corresponding to those who survive at that age at the time of death; life expectancy was based on the average life expectancy reported by INEGI ( National Institute of Statistics and Geography ) in Mexico [11] . The YLD were calculated as the product of the number of cases that have a certain health status , the duration of that health status and the weight of the disability for that state of health , adjusted to annual values . To estimate the YLD we considered a systematic review by Shepard et al , whom estimated specifically dengue , not general infections [12] . The disability weight used was 0 . 032 ( 0 . 018–0 . 044 ) for DF and 0 . 036 ( 0 . 022–0 . 050 ) for DHF , and the average duration days were 11 . 5 days for DF and 14 . 2 days for DHF . The expectancy of life ( 75 years ) was considered to be distributed homogeneously for the cases of DF and DHF [11] . The incidence ( number of cases ) by age was determined by consulting SINAVE [13] . We present a second calculation with all the previous estimations but considering all symptomatic dengue infections with an expansion factor of 5 . 6 for ambulatory and 2 . 0 for hospitalized , as described in Undurraga et al [6] . The expansion factor was applied to DF , DHF , and deaths . A multi-method approach was used to calculate the economic impact of dengue . The different elements of the estimation are presented in the following formula: Economicimpactofdengue=[costpercase]×[totalepisodes]+[costofdenguepreventionandsurveillanceprogram] Each element of the formula is explained below . All costs were calculated in local currency ( Mexican pesos ) , converted to US Dollars using the exchange rate on September 12 , 2012 ( 1 US$ = 13 . 03 Mexican pesos ) and then adjusted for inflation for each year . Previous studies in Mexico have reported a high level of under-reported and under-notification of cases [6 , 21 , 22] . Initially not all symptomatic patients were considered , only cases reported to the system . Nevertheless , in order to allow for comparability with other studies , a sensitivity analysis was conducted , where the base case is with no factor expansion , followed by two different scenarios with two expansion factors taken from the study by Undurraga et al [6]; the dengue patients who visited a health facility ( 3 . 7 ambulatory and 1 . 4 hospitalized ) , and considering all symptomatic dengue infections ( 5 . 6 ambulatory and 2 . 0 hospitalized ) . Finally the calculation follows the same assumptions as the “Economic Impact of dengue” section . It is important to clarify that the ambulatory expansion factor by Undurraga et al is the only one applicable to this cost data . In that sense Martinez-Vega , et al reported a general 33 . 3% under-reporting and 68 . 2% under-notified [22]; while Sarti et al reported an 8 . 4-fold local expansion factor [21] . In both studies we do not know the differences by healthcare setting , and so these calculations were not included in our study . Although comparisons with other countries may be illustrative , firm conclusions cannot be drawn given the differences in economic development , population size , and healthcare systems , as well as the methodology used for the estimates . Estimation of the economic cost of dengue should also include the cost of vector control and prevention , education , and epidemiological surveillance . As for cost per case , costs for dengue prevention and surveillance activities were also calculated using the PAATI approach , following the same scheme previously mentioned ( Fig 1 ) . An ideal protocol was prepared using a literature review and guidelines from the Mexican program [16 , 17] , and then the support and consensus of federal technical advisor users of the program who also participated in the expert group discussion sponsored by the Ministry of Health was sought ( see Betancourt-Cravioto et al . [20] ) . Six major activities were identified ( Fig 2 ) ( epidemiological surveillance , virological surveillance , environmental surveillance , insect surveillance , vector control and education ) . The calculation of the economic impact of dengue was based on three assumptions: ( 1 ) the direct costs were considered by the IMSS clinical real PAATI; although , we have data from a previous study for IMSS and SS , we consider costing for the system with the best quality in following the protocol , in order to have the most accurate estimation , and so IMSS alone was used; ( 2 ) as not all patients have out-of-pocket expenses , the costs from a patient perspective were estimated to be a proportion of the population , taken directly from previous data: 38 . 5% , 4% and 5% ( proportion of patients with expenses ) of the total cost per case for hospital cases , outpatients and patients in the ICU , respectively; and ( 3 ) the estimated cost of the dengue prevention and control program is the ideal epidemiological PAATI for the population of 25 endemic states . Despite attempting to estimate cost of dengue on the health system , patients and government , it was not possible to determine all types of costing from these sources . For example , persistent cases or sequelae following infection were not considered [7 , 23] . Data limitations prevented the quantification of the impact of tourism , or the impact of dengue on tourism revenues . Although , dengue cost might include other hidden items [24] . The study was delimitated to the points made in the methodology section .
Consultation of SINAVE , run by the General Directorate of Epidemiology , revealed three peaks in DF during the period studied ( from 2012 through 2016 , the most recent update to the database ) . The peak is presented in 2013 with 105 , 973 DF and 19 , 822 DHF cases ( Table 1 ) . The highest number of deaths was reported in the same year and the case fatality rate for DHF have increased in this period , with the highest rate of 1 . 70% reported in 2015 . Using the above epidemiological data , the DALYs for Mexico in 2016 were calculated to be 2283 . 46 which correspond to 1 . 87 DALYs per 100 , 000 inhabitants ( Table 2 ) . With the application of expansion factors for all symptomatic cases ( Table 3 ) , we observed that DALYs were 7156 . 22 which correspond to 5 . 85 DALYs per 100 , 000 inhabitants . As shown in Figs 3 and 4 , this burden fell largely on the younger age groups ( 1–4 years ) . The main driver corresponded to the ‘early death’ component ( YLL ) as the ‘living with disability’ component ( YLD ) contributed less than the overall DALYs ( Tables 2 and 3 ) . On application of the expansion factor , the change is proportional in each age group . Table 4 provides the total cost by year and number of cases for direct medical costs and costs from the patient’s perspective , for the years 2012–2016 , by care setting and according to the assumptions on the direct medical and patient costs ( see Methods section ) and the cost per case estimated previously [10] . This expense ( shown in Table 4 ) represents the cost to the system and the cost to the patient and caregiver . Governmental costs related to dengue were prevention , surveillance and control programs , and total cost of the ideal epidemiological PAATI were estimated to be US$11 , 766 . 52 per 10 , 000 inhabitants . As shown in Table 5 , vector control was the biggest cost driver , accounting for approximately 40% of the overall nonmedical governmental costs , followed by epidemiological surveillance , which accounted for approximately 25% . Finally , Table 6 shows the overall economic impact of dengue infection for 2012–2016 , according to the assumptions mentioned previously , and shows the differences between our results with and without the expansion factors between 2012 and 2016 . The overall economic impact of dengue infection ( DF and DHF ) for 2012 to 2016 shows how the cost is proportional to the number of cases obviously , but the real impact on cost is the prevention and control program . It is important to mention that the cost of the prevention program was not calculated for each year because the method essentially involves identifying the tasks and inputs from an ideal protocol and then assigning a unit cost to each . Therefore , there was no estimated average for the program . It is also important to emphasize that this estimation is not a projection . It was made using the reported cases to the system per year in Mexico and the cost calculation previously explained . The economic impact of dengue was presented by year from 2012 ( reference year ) to 2016 , the most recent year for which epidemiological information was available , with adjustment for inflation by year . As shown in Tables 1 and 6 , 2012 had 65 , 892 DF cases and an economic impact of US$144 million . The increment in the cost of the program is not proportional to the cases because execution of the program does not depend on the number of cases . Although 2016 had fewer cases than the others , the overall cost is only slightly affected . Additionally , expansion factors for patients with dengue who visited a health facility and all symptomatic dengue infections were used in order to compare these results with the previous studies [6] . Although comparisons with other countries may be illustrative , firm conclusions cannot be drawn given the differences in economic development , population size , and healthcare systems , as well as the methodology used for the estimates .
In the present analysis , considering only the reported cases , the cost of dengue infection in Mexico in 2012 was estimated to be US$144 million . In other studies of the disease in Mexico , Undurranga et al [6] estimated a cost of US$149 million ( 95% CI: $136–$231 ) in 2011 and an average of US$170 million ( 95% CI: $151–$292 ) for 2010 and 2011 . Tiga et al [7] reported an estimated average annual cost of US$192 million ( 95% CI: $171– $325 ) in 2012; however , it has to be considered that these estimates include persistent symptoms , and therefore must be compared to our outcomes with caution . The number of cases reported in 2011 was lower than in this study ( 2012–2016 ) . To investigate the impact this lower incidence might have , estimated costs for 2011 using the same assumptions , with adjustment for inflation were also calculated . The cost obtained for 2011 was US$103 , showing that these estimates are lower than other published estimates for each year . First , this must be considered as a consequence of using the PAATI methodology which , as a micro-costing methodology , starts with a detailed inventory and measurement of inputs that detects small differences in cost . It is more accurate than a macro-costing or a top-down approach , which are the typical analysis methods for aggregate data [25 , 26] . It is important to mention that only the clinical PAATI had a top-down approach , the epidemiology data used a bottom-up approach , although both are micro-costing . Second , PAATI was calculated using the average use reported per input ( ie resource or cost type ) . Each separate type of cost incurred was examined , which allows control of the variability for each component of the healthcare process , rather than using other methodologies where the average overall cost per patient is calculated and then assigned to each activity . Third , the expansion factor also has an impact; when these finding are compared with the adjustments in Table 6 it is clear that the estimation is getting closer to the other published studies . Although the expansion factor was added to enable comparison with other studies , the raw number of cases reported to the surveillance system in the 25 endemic states were used initially; it is important to clarify that it was not the aim to pronounce this as the best estimation . As mentioned previously , the cost of the government programs account for the majority of the overall cost and the actual number of cases has a limited impact . Nevertheless , the direct costs , derived from the number of cases , may be an underestimate of the actual cost given that DF is thought to be subject to considerable under-reporting . During the latest years , the number of cases of dengue have increased considerably , and several studies have shown number of cases to be underestimated . For example , a recent review of the global distribution and burden of dengue concluded that total infection was more than three times the dengue burden estimate of the World Health Organization [27] . In the study by Shepard and colleagues [28] , an expansion factor of 15 for DF was recommended for Mexico the recommended expansion factor for DHF was 2 . 3 . The results of a prospective cohort study , whose main aim was to investigate peridomestic infection as a determinant of dengue transmission , supported the expansion factors used by Shepard and colleagues on crosschecking the infections detected in their study with official databases [29] . In a recent study by Sarti et al , expansion factors indicated significant underreporting , and the authors reported that their use should be interpreted with caution [21] . Another study in Puerto Rico estimated the degree of underreporting of dengue cases by building a model using data from two different surveillance systems; the estimated rates of between 2 . 1 ( for inpatients ) to 7 . 8 ( for outpatients ) per 1 , 000 population compare with reported rates of 0 . 4 for outpatients and 0 . 1 for inpatients per 1 , 000 population [30] . In recent years , other countries have evaluated the economic impact of dengue . Colombia estimated a financial cost of US$167 . 8 million for 2010 , US$129 . 9 million for 2011 , and US$131 . 7 million for 2012 [31] . In Brazil , the annual total cost for dengue was estimated to be US$164 million ( 90% CI: $123–$205 ) from the public payer perspective and increased to US$ 447 million ( 90% CI: $335–$559 ) with adjustment for underreporting [32] . For the societal perspective which includes weight by public and private sector , the estimated cost was US$ 468 million ( 90% CL: 349–590 ) and US$ 1 , 212 million ( 90% CL: 904–1 , 526 ) , adjusting for under-reporting [32] . Finally , the overall cost of dengue presented in this study does not include preventive measures taken by households to limit transmission ( for example , use of insecticide and bed nets [33] ) . Clearly , this component may be non-negligible , but it would be extremely difficult to provide a reliable estimate for a number of reasons . First , if a family buys insecticide , for example , it would be difficult to determine what proportion of this expense was specific to the mosquitos that transmit dengue . Second , any survey would be subject to considerable recall bias , not to mention the problems with obtaining a representative sample . The DALYs calculated for dengue in Mexico were 2283 . 46 , which corresponds to 1 . 87 per 100 , 000 inhabitants or 18 . 67 DALYs per millon . Comparing with the global burden of dengue [3] , with fatal and non-fatal outcomes together , dengue was responsible for 1 . 14 million ( 0 . 73 million–1 . 98 million ) DALYs in 2013 , which appear to be higher in Mexico . We should consider between these estimations that ongoing factors , such as increased urbanization [34] and climate change [35] , have been associated with increased incidence . It seems likely that the number of cases in peak years will continue to increase if preventive measures are not taken . In addition , the DALYs are calculated using only mortality and disability; disruption of healthcare services , productivity losses and broader economic impacts also characterize the burden of dengue illness and are not taken into account in DALYs . It is also illustrative to compare the burden of dengue obtained in this study with figures reported for other studies , which tend to report a markedly higher burden . For example , a study from before 2000 in Puerto Rico reported a burden of dengue of 658 DALYs per million inhabitants [36] . Another study in Thailand reported a value of 427 DALYs per million inhabitants [37] , while a study of several southeast Asian countries reported 372 DALYs per million inhabitants [38] . In Mexico , Undurraga [6] reported the total disease burden for the adjusted average of dengue episodes was 83 . 5 and 46 . 7 DALYs per million inhabitants in 2010 and 2011 respectively . All those studies used expansion factors in their calculations . In principle , this could explain the difference between the figure found in this study and the other studies . It is of note , however , that a major component of the overall DALYs was premature death , accounting for more than three-quarters of the overall figure . As most deaths due to dengue are likely to occur in a hospital setting , with attention by specialist professionals , and given the more stringent reporting requirements for deaths , it is hard to imagine that the underreporting rate is high , although this point has been contended [39] . Nevertheless , given the discussion we decided to calculated DALYs with expansion factors and the value was incremented to 58 . 5 per million . Although we used the expansion factor even in the deaths , the DALYs are still higher in the studies mentioned above . The study is subject to a number of limitations . IMSS has better data as well as more complete and expensive health care delivery services , so we accept that we could overestimate the cost of the entire system , although; we used costs and not charges , given that the system is public and IMSS do not add a mark up to its costs . In the real PAATI we used current costs as reported by IMSS . At the same time since the system costs are over represented the magnitude of DF is not diminished . The impact of underreporting on both DALYs and economic cost is likely to mean that the costs and burden of dengue presented here are underestimates . This study used the raw number of cases to provide the base estimate of the real costs when using a PAATI methodology . Nevertheless , when these results are compared using the expansion factor considering the cases of DF not detected by the national surveillance system ( 3 . 7 ambulatory and 1 . 4 hospitalized ) a 20% increase is seen , still within the ranges presented in other studies . In addition , as has been discussed previously with the detailed data on costs per case [10] , the survey used to derive the real costs is subject to possible sampling bias and incomplete data . However , to compensate , this study has a large sample size which provides more confidence in the stability of the data . It is therefore reassuring that the costs estimated using this micro-costing approach are generally in line with previously published estimates using different methodology . Finally , given the low responses by those responsible for the dengue program in each of the states , it was not possible to present a “real epidemiological PAATI” , and realistic estimates of surveillance and prevention costs would have been a useful contribution . The ideal dengue program will have to be evaluated in future research , with estimates for real costs of the program . In conclusion , the present study of dengue infection in Mexico shows that the economic impact of dengue is considerable , in broad agreement with other international studies . In particular , the results highlight the need to include not only direct medical and non-medical costs but also the costs linked to surveillance , vector control and prevention in any overall analysis of dengue infection . Given the high economic impact of the prevention , surveillance and control program , their effectiveness needs to be assessed; it cannot be assumed that the lower response in our study is a lack in the program but should be trigger point for research , in the same way for other alternatives , such as vaccination , should also be considered in the next evaluation . However , it is important to remember that many aspects of this program also impact other diseases transmitted by the same vector . Another factor that could be considered in future studies is the geographic distribution of dengue in Mexico . Some regions are highly endemic while others have limited dengue transmission; a review of an endemic sub-region would further inform a more tailored approach in these regions . Although the burden of dengue compared with other infectious diseases may appear limited , the epidemiological trends of mortality , particularly during epidemics , the main driver in the economic impact of this disease , do not leave room for complacency .
|
Dengue fever is caused by a flavivirus transmitted predominantly by the mosquito Aedes aegypti . Infection causes a broad spectrum of clinical signs and symptoms , from mild disease , such as dengue fever to a life threatening form known as dengue hemorrhagic fever . The disease is widespread in tropical regions . Measures such as vector control can slow the spread of infection , and most countries where the disease is endemic , Mexico included , have programs in place to this end . However , faced with other health issues that also require attention , it is important to quantify the suffering caused by dengue and also its economic costs . In this study , we aimed to produce detailed figures for Mexico to complement and refine those available from international studies . Such information will help guide how the money budgeted for health in dengue is spent .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"disabilities",
"economic",
"analysis",
"tropical",
"diseases",
"geographical",
"locations",
"social",
"sciences",
"health",
"care",
"north",
"america",
"neglected",
"tropical",
"diseases",
"infectious",
"disease",
"control",
"public",
"and",
"occupational",
"health",
"infectious",
"diseases",
"economic",
"impact",
"analysis",
"health",
"economics",
"dengue",
"fever",
"epidemiology",
"economics",
"people",
"and",
"places",
"infectious",
"disease",
"surveillance",
"disease",
"surveillance",
"viral",
"diseases",
"mexico"
] |
2018
|
Economic impact of dengue in Mexico considering reported cases for 2012 to 2016
|
In the fungal pathogen Cryptococcus neoformans , the switch from yeast to hypha is an important morphological process preceding the meiotic events during sexual development . Morphotype is also known to be associated with cryptococcal virulence potential . Previous studies identified the regulator Znf2 as a key decision maker for hypha formation and as an anti-virulence factor . By a forward genetic screen , we discovered that a long non-coding RNA ( lncRNA ) RZE1 functions upstream of ZNF2 in regulating yeast-to-hypha transition . We demonstrate that RZE1 functions primarily in cis and less effectively in trans . Interestingly , RZE1’s function is restricted to its native nucleus . Accordingly , RZE1 does not appear to directly affect Znf2 translation or the subcellular localization of Znf2 protein . Transcriptome analysis indicates that the loss of RZE1 reduces the transcript level of ZNF2 and Znf2’s prominent downstream targets . In addition , microscopic examination using single molecule fluorescent in situ hybridization ( smFISH ) indicates that the loss of RZE1 increases the ratio of ZNF2 transcripts in the nucleus versus those in the cytoplasm . Taken together , this lncRNA controls Cryptococcus yeast-to-hypha transition through regulating the key morphogenesis regulator Znf2 . This is the first functional characterization of a lncRNA in a human fungal pathogen . Given the potential large number of lncRNAs in the genomes of Cryptococcus and other fungal pathogens , the findings implicate lncRNAs as an additional layer of genetic regulation during fungal development that may well contribute to the complexity in these “simple” eukaryotes .
In many human fungal pathogens , the morphological transition from yeast to hypha plays a central role in pathogenesis [1 , 2] , as demonstrated in the ascomycetes Candida albicans , Penicillium marneffei , Histoplasma capsulatum , Coccidioides immitis , and Paracoccidiodides brasiliensis [3–6] . Different morphotypes also display different levels of pathogenicity in the basidiomycetous fungus Cryptococcus neoformans [1 , 7] , the causative agent of the deadly cryptococcal meningitis [8] . Although primarily considered as yeasts , Cryptococcus undergoes yeast-to-hypha transition during unisexual mating ( self-fruiting ) or bisexual a-α mating [9–11] . The zinc finger transcription factor Znf2 ultimately controls this morphotype transition . During mating , Znf2 is activated by the pheromone MAPK pathway controlled by the HMG domain transcription factor Mat2 [12–15] ( Fig 1A ) . Mat2 is essential for pheromone sensing and response , which leads to the cell fusion event . Hyphal growth commences after cell fusion and eventually gives rise to fruiting structures and meiotic spores [9 , 16] . However , Mat2 does not control hyphal morphogenesis per se [12] . By contrast , Znf2 governs hypha generation and it is dispensable for the early mating events like cell fusion [12 , 17] ( Fig 1A ) . Under non-mating inducing conditions , Znf2 could be activated by the matri-cellular signal protein Cfl1 through a positive feedback regulation [18 , 19] . It is unknown whether other host or environmental factors can also regulate Znf2 activity . Znf2 is an anti-virulence factor in the mouse model of cryptococcosis [12 , 20] . The deletion of the ZNF2 gene locks the fungal cells in the yeast form , making them more virulent [12] . Conversely , the activation of ZNF2 drives filamentation and attenuates Cryptococcus virulence [21 , 22] . The ZNF2 overexpression cells , either in the live or heat-killed form , can protect the hosts from a subsequent challenge with otherwise lethal wild-type cells [22] . Thus manipulation of ZNF2 activity could be a potential means to alleviate cryptococcosis . Besides its anti-virulence effect during cryptococcal infection in a mammalian host , Znf2 also shapes cryptococcal interaction with other heterologous hosts , such as the soil amoeba Acanthamoeba castellanii and the insect Galleria mellonella [23] . The essential role of Znf2 in Cryptococcus sexual cycle and its pivotal role in regulating cryptococcal interaction with various host species make this transcription factor a potential target for multi-layered regulation in response to various stimuli . To identify the upstream regulators of ZNF2 , we conducted a forward genetic screen to find mutations that cause similar phenotypes as those caused by the disruption of ZNF2 . The screen led to the discovery of RZE1 that functions upstream of ZNF2 . Further investigation revealed that RZE1 functions primarily as a cis-acting , and less efficiently a trans-acting , lncRNA . Furthermore , this lncRNA is functionally restricted to its native nuclei based on heterokaryon assay . We found that RZE1 exerts its impact on cryptococcal morphogenesis by regulating ZNF2 transcription and by influencing the nuclear versus cytoplasmic distribution of ZNF2 transcripts , which consequently affects ZNF2’s ability to get translated into protein . Although genomic and transcriptomic data suggest the presence of lncRNAs showing infection- or tissue-specific expression in fungal species pathogenic to plants and animals [24] , it is unclear if lncRNAs have any biological relevance in the life cycle or the development of human fungal pathogens . RZE1 is the first lncRNA that is functionally characterized in a human fungal pathogen . The importance of RZE1 in cryptococcal morphogenesis raises the possibility that lncRNAs may be important regulators that contribute to the complexity in genetic regulation in these eukaryotic pathogens .
Cryptococcus undergoes filamentation in response to the mating signal and other environmental cues . Znf2 is the essential regulator of this morphological switch and it bridges morphogenesis and virulence in this fungal pathogen [12 , 17–19] . To identify the regulatory network of the Znf2-controlled filamentation pathway , we conducted a random insertional mutagenesis screen for znf2Δ-like phenotype in the self-filamentous strain XL280 [25] . XL280 has been commonly used in morphogenesis studies [12 , 20 , 25] . It has a publically released genome sequence ( ~19 Mb , ~7000 genes ) [26] and a well characterized congenic pair [21] . We generated 63 , 000 insertional mutants via Agrobacterium-mediated transformation in XL280 and screened these mutants on filamentation-inducing V8 medium for non-filamentous phenotypes . Among a set of selected non-filamentous mutants , 15 had their insertion site identified by inverse PCR and sequencing ( Table 1 ) . Of the 15 insertion sites identified , one T-DNA insertion ( Tn ) in mutant X261 was found to be in the genetic locus that we named RZE1 . The RZE1 gene encodes a 1 , 268 nt long transcript in XL280 based on our primer walking and RACE PCR results ( S1 Fig ) . Only one transcription start site and one transcription stop site were identified for RZE1 under the tested condition . The rze1Tn mutant , like the znf2Δ mutant , is non-filamentous under mating-inducing conditions such as on V8 media ( Fig 1B ) . To confirm the role of RZE1 in hyphal growth , we deleted the RZE1 gene in the XL280 background . The targeted deletion of RZE1 also abolished self-filamentation ( Fig 1B ) . To ensure that the non-filamentous phenotype of the rze1Tn and the rze1Δ mutants was attributable to the disruption of RZE1 and not due to other cryptic mutations , a wild-type copy of the RZE1 gene was re-introduced ectopically into the rze1Tn and the rze1Δ mutants . The ectopically integrated RZE1 partially restored the filamentation defects in these rze1 mutants ( S2 Fig ) . We also introduced the wild-type RZE1 back to its native locus in the rze1Tn and the rze1Δ mutants . The introduced RZE1 gene at its native locus effectively restored the ability of both mutants to filament ( Fig 1B ) , indicating that RZE1 indeed is required for filamentation . The pheromone sensing pathway is critical for filamentation under mating-inducing conditions . Disruption of the pheromone sensing pathway ( e . g . the deletion of HMG domain transcription factor Mat2 or the MAPKK Ste7 ) , even in one mating partner , will abolish filamentation produced by bisexual mating between a and α partners [12 , 14] . The lack of the bisexual mating hyphae is due to lack of cell fusion . It is previously established that the disruption of ZNF2 does not impair pheromone pathway or abolish cell fusion [12] . Thus a unilateral cross involving the znf2Δ mutant and a wild-type partner still produces mating filaments , in contrast to the unilateral cross involving one mat2Δ mating partner [12] . We then decided to assess if this also holds true for RZE1 . To avoid the complication due to self-filamentation in the XL280 background , we chose to delete RZE1 in the non-self-filamentous strain H99 . In the H99 background , hyphae can only result from bisexual mating following cell fusion between a and α cells activated by the pheromone pathway . We found that the unilateral crosses between the rze1Δ ( or the rze1Tn ) mutant with a wild-type mating partner produced mating filaments ( Fig 1C ) , as in the znf2Δ mutant and unlike the mat2Δ mutant ( Fig 1C ) . Consistently , the pheromone production in the rze1Δ mutant was at par with the wild type , indicating that RZE1 , like ZNF2 , is not required for pheromone sensing pathway . Under non-mating inducing conditions like on the YPD medium , ZNF2 and thereby filamentation can be activated by the matricellular signaling protein Cfl1 through a positive feedback loop [18] . Znf2 is absolutely necessary for Cryptococcus to respond to exogenous Cfl1 signal to promote filamentation [20] . Similarly , we found that RZE1 is also required for the recipient strain to filament in response to the CFL1 protein released from nearby donor cells under mating-suppressing conditions ( Fig 1D ) . Morphological switch from the yeast state to the filamentous state is inversely correlated with virulence in Cryptococcus . Consequently , the loss of ZNF2 modestly increases fungal virulence [20] . Thus we hypothesize that the upstream regulator of ZNF2 might also influence cryptococcal virulence potential . Since mice are highly susceptible to XL280 ( C . neoformans var . neoformans , serotype D ) [21 , 27] and H99 ( C . neoformans var . grubii , serotype A ) [28 , 29] , we decided to compare the virulence between the wild-type strain and the rze1Δ mutant made in both XL280 and H99 backgrounds using the fungal burden assays . We inoculated the mice with the rze1Δ mutants and their corresponding wild-type strains through inhalation and measured the fungal burdens in the lungs at day 10 post inoculation . The fungal burdens in the lungs infected by the rze1Δ mutants were close to 2 fold higher than those infected by the corresponding wild-type controls ( Fig 1E ) , indicating enhanced virulence of the rze1Δ mutants . Given that an increase in virulence with gene disruption is rare , these results are consistent with the observed enhanced virulence of the znf2Δ mutant , which showed about 2 . 4 fold increase in lung fungal burden compared to the wild type at the same time point during infection [12 , 20] . In line with our previous observations for the znf2Δ mutant [12 , 20] , the rze1Δ mutant did not show any apparent difference from the wild type in classic virulence traits such as melanization , capsule production , or thermo-tolerance ( S3 Fig ) . We also did not observe any apparent alteration in the susceptibility of the rze1Δ mutant to stressors like SDS , caspofungin ( inhibitor of β -1 , 3-glucan synthase ) , calcoflour white ( inhibitor of chitin ) , iron chelator , UV radiation , or oxidative stress ( H2O2 ) when compared to the wild-type control ( S3 Fig ) . Taken together , these observations indicate that the disruption of RZE1 recapitulates in vitro and in vivo phenotypes caused by the deletion of ZNF2 . Thus RZE1 might be a highly selective regulator of ZNF2 or a major target of ZNF2 . RZE1 is located ~2 . 5 kb upstream of ZNF2 . Although the average intergenic space is less than 800 bp in Cryptococcus [30] , the physical location of RZE1 raises a concern that the RZE1 transcript could be part of the ZNF2 transcript and the disruption of RZE1 would cause the disruption of ZNF2 itself . To address this concern , we amplified and compared four regions ( I—IV in Fig 2A ) that cover RZE1 , the beginning of the ZNF2 coding region , and the region between RZE1 and ZNF2 , using templates of total cDNA or genomic DNA derived from the wild-type XL280 strain . Region II that lies between RZE1 and the 5’ region upstream of the ZNF2 ORF failed to yield any detectable amplicon when cDNA was used as the template ( Fig 2A ) . This suggests that RZE1 is likely produced as a separate transcript independent from ZNF2 . Furthermore , the ability of the ectopically introduced RZE1 to partially complement the rze1Tn and the rze1Δ mutants ( S2A Fig ) reinforces the idea that RZE1 is a functionally independent transcript from that of ZNF2 . In addition , since the phenotype of the znf2Δ mutant can be complemented effectively using an ectopic copy of ZNF2 with 1kb sequence upstream of the ZNF2 ORF that does not include RZE1 ( S4 Fig ) , it is reasonable to conclude that RZE1 and ZNF2 encode separate transcripts that can function independently . To investigate the relationship between RZE1 and ZNF2 , we examined the RZE1 transcript level in the wild type , the znf2Δ mutant , and the ZNF2oe strain under three different growth conditions ( YPD/rich medium , serum/host relevant , and V8/mating-inducing ) . The transcript level of RZE1 in the wild-type H99 strain was relatively stable , and there was no dramatic difference when cells were cultured in YPD medium , on V8 medium , or in serum ( Fig 2C ) . Under the same culture conditions , the RZE1 transcript level remained constant among the different strains ( black bars in Fig 2D–2F ) , even though the ZNF2 transcript level was drastically different among these strains ( grey bars in Fig 2D–2F ) . This indicates that RZE1 is not responsive to changes in the ZNF2 transcript level . This result supports the earlier conclusion that RZE1 is a transcript independent of ZNF2 and likely functions in the filamentation pathway upstream of ZNF2 . Consistently , the constitutive expression of ZNF2 driven by the GPD1 promoter led to the partial restoration of filamentation to these rze1 mutants ( Fig 2B ) , indicating that ZNF2 indeed functions downstream of RZE1 . These observations are in accordance with the idea that RZE1 functions in the filamentation pathway upstream of ZNF2 . Based on the Cryptococcus EST databases , the RZE1 transcript is present in all subspecies of the Cryptococcus neoformans species complex , namely C . neoformans var . grubii ( serotype A ) , C . neoformans var . neoformans ( serotype D ) , and the subspecies C . gatii ( serotype B and C ) . An analysis of the RZE1 transcript sequence suggests that the RZE1 gene is not a typical protein-coding gene . Rather than encoding one open reading frame ( ORF ) , as expected for most mRNAs in fungi , RZE1 contains 5 short potential ORFs ( Fig 3A ) , with three potential translation start codons ( ATG ) in a poor translation context ( S1 Fig ) . BLAST searches using these potential ORFs against fungal genome databases or GenBank did not yield any significant hits . Thus , these potential ORFs do not encode conserved protein products . To test if products from any of the small ORFs confer RZE1 function in filamentation , we inserted the constitutive active GPD1 promoter [31] ( Fig 3B ) or the CTR4 inducible promoter [32] ( S5 Fig ) upstream of each of the five ORFs contained within the RZE1 transcript . We introduced these constructs into the rze1Δ mutant and then tested the transformants for their ability to filament . None of the PGPD1-ORF transformants were able to produce filaments ( Fig 3B ) . Similarly , none of the PCTR4-2-ORF transformants were able to filament under inducing conditions ( S5 Fig ) . The results suggest that the potential ORFs carried within the RZE1 transcript are unlikely to produce protein products that function in the filamentation pathway . We postulate that RZE1 may function as a lncRNA instead . To further determine whether RZE1 functions as a protein or a transcript , we did site-directed single nucleotide mutagenesis to alter the translation start codon of each of the five potential ORFs in RZE1 ( Fig 3C ) . We mutated ATG to ATA or AAG as such changes are known to almost abolish translation initiation in other fungi [33] . These codon changes are expected to prevent or considerably reduce the translation of these potential ORFs . We then introduced each of these mutated RZE1 alleles into the rze1Δ mutant and selected transformants with the mutated RZE1 allele integrated into the RZE1’s native locus . We then examined the ability of these transformants to produce hyphae . To our amazement , all the mutated RZE1 alleles were able to restore the filamentation defect of the rze1Δ mutant ( Fig 3C ) , although to a lesser degree compared to the wild-type RZE1 allele integrated at the native locus ( Fig 3C ) . This finding indicates that these RZE1 alleles are at least partly functional in promoting filamentation , even though the specific nucleotides that are potential start-codons are mutated . Collectively , our results strongly suggest that RZE1 is a long non-coding RNA regulating morphogenesis in Cryptococcus . As a lncRNA , it is possible to be functionally restricted to its target within its native nucleus , as XIST or MALAT-1 [34–37] , or to be functional in the cytoplasm [38] . This is in contrast to protein-coding mRNAs , which need to be exported from the nucleus to the cytoplasm for translation to make the functional products . Proteins produced by mRNAs would be able to function with its targets produced by any nucleus that shares the same cytoplasm . To narrow down the potential mode of action of RZE1 , we decided to first examine if RZE1 is functionally restricted to its native nucleus . For this purpose , we designed a heterokaryon assay where two cells conjugate and thus share the cytoplasm , but their nuclei do not fuse . Heterokaryon assay was previously used to show that the lncRNA MALAT1 is functionally restricted to the nucleus [39] . Forming heterokaryons is a natural process during the a-α bisexual mating in Cryptococcus [9 , 40 , 41] ( Fig 4A ) . The formation of α-a dikaryon after mating conjugation brings the homeodomain proteins Sxi1α and Sxi2a together [42] , which triggers the expression of ZNF2 and the production of dikaryotic mating hyphae [9 , 12 , 43] . It is important to note that RZE1 controls filamentation but not conjugation ( cell fusion ) , as established previously for Znf2 ( Fig 1C ) [12 , 17] . Furthermore , the dikaryon will be blocked from filamentation only when ZNF2 is disrupted in both mating partners [12 , 43] . For this assay , we used cryptococcal strains in H99 background because H99 does not self-filament . As a result , filaments produced from the α-a bisexual mating are all derived from conjugated dikaryons . This allows us to examine if RZE1 produced by one nucleus could compensate the loss of RZE1 in the other nucleus that shares the same cytoplasm . The idea is that if the RZE1 from one nucleus was able to function in the cytoplasm or in another nucleus in the same conjugated cell , then it would have been able to regulate ZNF2 activity even if ZNF2 is produced by the other nucleus ( Fig 4A ) . As expected , the unilateral crosses rze1Δ α x a , znf2Δ α x a , and rze1Δznf2Δ α x a all filamented at a reduced level compared to the WT cross α x a ( Fig 4A ) . However , the dikaryons generated from the cross rze1Δ α x znf2Δ a or the cross rze1Δ a x znf2Δ α failed to filament ( Fig 4A ) . These dikaryons contain one nucleus with RZE1 but no ZNF2 , and the other nucleus with ZNF2 but no RZE1 . The inability of these dikaryons to filament indicates that RZE1 produced by one nucleus failed to support the activity of ZNF2 transcribed by the other nucleus even though the two nuclei shared the same cytoplasm . Because the diploid α/a wild-type cells in H99 background did not filament under this condition , it precludes us from further assessing whether the diploid cells derived from nuclear fusion of the rze1Δa x znf2Δα heterokaryon could filament . The decreased ability of the heterozygous diploid cells ( α/a ) compared to the corresponding dikaryons ( α-a ) to continue through the mating program appears to be a common phenomenon in basidiomycetes . Nonetheless , these findings indicate that RZE1 is functionally restricted to its native nucleus and as such it is highly unlikely for RZE1 to directly affect the ZNF2 translation or other post-translational processes that occur in the cytosol . To further test the hypothesis that RZE1 does not play a direct role in translation or Znf2 protein localization to the nucleus , we used a system where the expression of an ectopically introduced ZNF2-mCherry is controlled by the CTR4 promoter . We compared the fluorescent intensity ( indicative of protein level ) and the subcellular localization of Znf2 in the presence or absence of RZE1 . We found that the loss of RZE1 does not affect either the intensity or the localization of Znf2-mCherry ( Fig 4B and 4C ) . This is consistent with the idea that RZE1 does not directly regulate ZNF2 at the level of translation or protein translocation . Taken together , these results strongly suggest that RZE1 is functionally restricted in its native nucleus , and is not directly involved in the Znf2 protein processing or translocation . In the nucleus , however , it is possible that RZE1 could directly or indirectly regulate ZNF2 at different levels ( Fig 4D ) : ( i ) promote ZNF2 transcription or transcript stability , ( ii ) support the production of the functional ZNF2 transcript isoform through its influence on alternative start , alternative termination , or alternative splicing [28] , or ( iii ) assist the export of the ZNF2 transcript from nucleus to cytosol . Regulatory activity of RZE1 in any of the above processes could result in an effect on the Znf2 protein level or activity in the wild type strain . The evidence presented so far suggests that RZE1 functions in its native nucleus and it functions in a regulatory capacity upstream of ZNF2 ( Fig 4D ) . We decided to examine the first hypothesis of transcriptional control . To obtain a holistic view of the effect of the RZE1 deletion on cryptococcal transcriptome , we first conducted a pilot comparative transcriptome analysis between the wild-type XL280 and the rze1Δ mutant undergoing self-filamentation by RNA sequencing ( RNA-seq ) . A total of 265 genes with the 1 . 5 fold change as the cut-off threshold ( P<0 . 05 ) were differentially expressed in the rze1Δ mutant relative to the wild type at 24h post inoculation on filamentation-inducing V8 media ( S1 Table ) . These genes are classified into eight functional categories ( Fig 5A ) . Among the enriched GO terms , genes involved in the septin complex , cell cycle , and DNA replication are known to affect morphogenesis [44–46] , which is consistent with the regulatory function of RZE1 in morphogenesis . Based on the RNA-seq data , the transcript level of ZNF2 in the rze1Δ mutant was lower compared to the wild type at 24 hours post inoculation ( Fig 5B ) . The reduction of the transcript level of ZNF2 in the rze1Δ mutant was less obvious at 72 hours post inoculation compared to the 24h and 28h time points ( S6B Fig ) . The lower ZNF2 transcript level in the rze1Δ mutant was further confirmed by quantitative real-time PCR ( Fig 5C ) . These results suggest that RZE1 , directly or indirectly , up-regulates ZNF2 at the transcriptional level . Since filamentation is obvious at 24h in XL280 under inducing condition , we suspect that the ZNF2 gene expression might be induced earlier . We therefore expanded our examination of the ZNF2 transcript level in the rze1Δ mutant and the wild type to include more time points ( 12h , 18h , 24h , 48h , and 72h post inoculation on V8 medium ) by qPCR . As expected [12] , the ZNF2 transcript level in the wild-type strain of all the later time points examined was increased compared to the reference level at 12h ( Fig 5C ) . To our surprise , the ZNF2 transcript level in the rze1Δ mutant is also increased , suggesting that the signal transduction to promote filamentation still occurs in the mutant . However , the level of ZNF2 transcripts in the rze1Δ mutant is lower than that in the wild type , with the biggest difference observed at 18h ( ~4 fold ) and smaller differences at 48h and 72h ( Fig 5C ) . This might be consistent with the observation that the number of differentially expressed genes in the rze1Δ mutant is much higher at the 24h time point than the later 48h and 72h time points based on RNA-seq ( S1 Table ) . These observations suggest an intriguing possibility that RZE1 might play an important regulatory role in driving filamentation at the early stage of morphogenesis . Given the modest effect on the ZNF2 transcript level by the disruption of RZE1 , we were surprised by the blocked filamentation in the rze1Δ mutant . Other mutants with reduced levels of ZNF2 typically show reduced but not abolished filamentation . Thus it is enigmatic that the disruption of RZE1 could completely abolish self-filamentation . The paradox between the observed reduction in the transcript level of ZNF2 and the abolished filamentation in the rze1Δ mutant thus prompted us to examine the effect of RZE1 deletion on the previously characterized downstream targets of Znf2 that are known to be important for filamentation: CFL1 , FAD1 , and FAS1 [17 , 18] . Both qPCR and RNA-seq results indicated that the transcript levels of these filamentation markers were considerably reduced in the rze1Δ mutant ( Fig 5D and S7 Fig ) , often more comparable to those observed in the znf2Δ mutant ( Fig 5D ) . The drastic reduction in the transcript level of CFL1 , an auto-inducer working in positive feedback with ZNF2 [18] , is also in accordance with the results of the confrontation assay ( Fig 1D ) . Thus it appears that although the deletion of RZE1 only exerts a modest impact on the transcript level of ZNF2 , it exerts a drastic impact on factors downstream of ZNF2 and may inactivate at least part of the ZNF2 regulon that controls filamentation . This raises the possibility that RZE1 may affect additional processes between the ZNF2 transcription and translation . As transcript isoforms different from those in the wild type could be non-functional , we decided to examine the impact of RZE1’s disruption on ZNF2 isoforms ( Fig 4D ) . In C . neoformans 99 . 5% of all expressed genes have introns and an average gene contains multiple introns [30] . Thus alternative splicing could be one major way of generating different transcript isoforms . Indeed , multiple introns were found in ZNF2 wild type allele in the serotype D strain XL280 ( Fig 5B and S6A Fig ) as well as in the serotype A strain H99 [28] . The RNA-seq data showed that introns of the ZNF2 transcripts were spliced in both the wild-type strain XL280 and in the rze1Δ mutant ( Fig 5B ) . Consistently , amplicons of the transcript that covered the whole ZNF2 ORF region in the wild type and in the rze1Δ mutant by reverse transcription PCR showed no apparent difference in size , again supporting the idea that introns of the ZNF2 transcripts could be spliced in the absence of RZE1 . To examine if there was any quantitative difference in splicing of the ZNF2 transcript at different regions , we analyzed the transcript level of five different regions of the ZNF2 transcripts in the wild type and in the rze1Δ mutant during self-filamentation . These five regions included two regions in the 5’UTR , the junction between exon 1 and 2 , the junction between exon 3 and 4 , and the exon 5 region ( Fig 6A ) . The levels of all five tested regions of the ZNF2 transcripts were lower in the rze1Δ mutant relative to those in the wild type , again with the biggest difference observed at 18h and smaller differences observed at 48h and 72h ( Fig 6A ) . The pattern was similar to the earlier observation based on the conserved exon 5 region using the P5 primer set ( Fig 5C ) . Collectively , the results suggest that RZE1 does not affect intron splicing of the ZNF2 transcripts . To test if RZE1 plays a role in other aspects of ZNF2 transcript processing , we performed northern blot and probed with the ZNF2 ORF . As ZNF2 basal expression is low and is induced during bisexual mating [12] , we first examined purified poly ( A ) -RNAs extracted from wild type , the znf2Δ mutant , and the rze1Δ mutant at the 12h and 24h time points during bisexual mating . One band that was larger than 3 , 000 nt and close to 4 , 000 nt was detected in both wild type and the rze1Δ mutant , but not in the znf2Δ mutant ( Fig 6B ) . The size is consistent with the predicted approximately 3 , 600 nt-long mature transcript for ZNF2 . The reduction in ZNF2 transcript level in the rze1Δ mutant during bisexual mating was apparent at the 12h time point , but not at the 24h time point ( Fig 5C and S6B Fig ) . This might be because the impact of RZE1 deletion on the ZNF2 transcript level becomes more modest at later time points , as we observed in the unisexual mating cells ( S6B Fig ) . We also compared the ZNF2 transcript size by northern hybridization using purified poly ( A ) RNAs extracted from wild type , the znf2Δ mutant , and the rze1Δ mutant at 48h during self-filamentation . We found that the size of the ZNF2 transcript remained ~3 . 6kb in wild type and the rze1Δ mutant ( Fig 6B ) . Taken together , these results indicate that ZNF2 only has one transcript isoform under the conditions we tested . Furthermore , the loss of RZE1 does not appear to affect cryptococcal ability to process introns of the ZNF2 transcripts . We decided to test the third hypothesis: the effect of RZE1 on the export of the ZNF2 transcripts from the nucleus to the cytosol ( Fig 4D ) . For this purpose , we used the technique called single molecule Fluorescent In Situ Hybridization ( smFISH ) , which helps visualize and quantify transcripts in the cell [47] . Unlike qPCR or RNA-seq that measure the total transcript level from the whole population , smFISH can measure the level of a particular transcript in a single cell and can be used to examine the heterogeneity in gene expression in that population . This microscopy based technique can also reveal the subcellular localization of the examined transcript . Although this technique has been used in research in other organisms ( e . g . mammalian cells and model yeasts ) [47–49] , it has not been applied to Cryptococcus because of the unique challenge of making spheroplast/protoplast in this organism due to the presence of cell wall and polysaccharide capsule . Thus , we did pilot studies to optimize the experimental conditions for Cryptococcus using the U2 spliceosomal RNA as the positive control . We chose the U2 snRNA as the control because of its abundance in intron-rich eukaryotic cells and its exclusive nuclear localization [50] . Consistently , we found abundant U2 snRNA in Cryptococcus nuclei ( Fig 7A ) . To further establish a control of messenger RNAs that will be localized to the nucleus ( for transcription ) and cytosol ( for translation ) , we chose to examine the actin transcripts ( ACT1 ) in cells plated on mating media ( V8 medium , 18h ) ( Fig 7B ) . We found that 62 . 7% of cells in wild type ( n = 884 ) and 59 . 3% of cells in the rze1Δ mutant ( n = 899 ) were positive for the actin probe , likely due to incomplete cell wall digestion and the consequent occlusion of the probe from the remaining cells ( S2 Table ) . However , digestion for longer period led to visible damage in some cells and was thus undesirable for the purpose of assessment of the transcripts’ subcellular localization . Thus we decided not to extend enzymatic digestion to prevent cellular damage . We therefore assumed ~50–60% efficiency in probe penetration of cells prepared under such conditions . Among the actin transcripts examined , 70–80% of them were found in the cytosol and 20–30% in the nucleus ( S2 Table ) . Such predominant distribution in the cytosol is expected for messenger RNAs as they primarily serve as templates for translation . Interestingly but not surprisingly , the deletion of RZE1 did not affect the subcellular distribution of ACT1 , with still 71% found in the cytosol in the rze1Δ mutant ( Fig 7C and S2 Table ) . We then proceeded to visualize the ZNF2 transcripts in cryptococcal cells in the absence or the presence of RZE1 under the filamentation-inducing condition that induces the expression of ZNF2 ( Fig 7E and 7F ) . The znf2Δ mutant cells processed under the same conditions were used as the negative control . We found that 49 . 18% of cells in wild type ( n = 612 ) and 43 . 9% of cells in the rze1Δ mutant ( n = 664 ) expressed ZNF2 ( S2 and S3 Tables . ) . We believe that the percentage of wild-type cells expressing the ZNF2 gene was an under-estimation . This is because some of the wild type cells expressing higher levels of ZNF2 were turning into hyphae under this condition . These hyphal cells were more likely to be damaged during cell collection from solid agar plates and during the fixation and digestion process . Thus , wild-type cells that likely expressed higher levels of ZNF2 were also more likely to be excluded from the analysis . Nonetheless , as expected for an mRNA , we observed ZNF2 transcripts in both the nucleus and the cytosol ( Fig 7E and 7F ) . Interestingly , we observed a higher percentile of ZNF2 transcripts localized in the nuclei in the rze1Δ mutant ( 44 . 7% cytoplasmic: 55 . 2% nuclear distribution compared to that in the wild type ( 69 . 6% cytoplasmic: 30 . 4% nuclear distribution ) ( Fig 7G and S2 and S3 Tables ) . The increased ratio of nucleus versus cytosol distribution of ZNF2 transcripts in the rze1Δ mutant could be caused by a defect in exporting ZNF2 transcripts to the cytoplasm , or alternatively increased cytoplasmic degradation of ZNF2 transcripts . Improperly processed transcripts are known to be subject to degradation in the cytoplasm by RNA surveillance mechanisms [51] . However , we did not detect any apparent difference in intron splicing of the ZNF2 transcripts generated in the rze1Δ mutant ( Fig 5B ) . If ZNF2 transcripts undergo increased degradation in the cytoplasm , the involvement of RZE1 in that process must be , if any , indirect . This is because RZE1 is functionally restricted to its native nucleus ( Fig 4A ) . Thus , we favor the other possibility of increased nuclear retention of ZNF2 transcripts ( or reduced ZNF2 transport ) in the absence of RZE1 . That said , although the direct effect of RZE1 may be in the nucleus , we cannot exclude any indirect effect happening in the cytoplasm . Future investigation is warranted to further distinguish these hypotheses . It was previously shown that functionally nuclear restricted lncRNAs could be physically confined to nuclei ( as exemplified by lncRNAs MALATI and XIST [36 , 37] ) . LncRNAs with nuclear function could also be distributed both in the cytoplasm and nuclei , although the cytoplasmic transcripts are considered non-functional as deductible from their mode of action ( e . g . FLO11 lncRNAs in S . cerevisiae [52] ) . We therefore decided to examine the subcellular localization of RZE1 transcripts in the wild-type cells under a mating-inducing condition . We found that RZE1 was predominantly localized in the nucleus ( 29 . 3% cytoplasmic: 70 . 6% nuclear distribution; n = 514 ) ( Fig 7H and S2 Table ) . The predominant nuclear localization of RZE1 again is consistent with the evidence presented earlier supporting RZE1 acting as a transcript rather than as an mRNA .
Non-coding RNAs constitute a large part of genomes in higher eukaryotes [53 , 54] . Among ncRNAs , lncRNAs are typically regulatory ncRNAs arbitrarily defined as transcripts of over 200 nt with minimum protein-coding potential [55] . The first discovered lncRNA H19 [56] was classified as a lncRNA in 1990 after the analysis of cloned gene sequence revealed its poor coding potential and its lack of observed sedimentation with the ribosomes [57] . The famous Xist RNA was identified by virtue of its exclusive expression by the inactivated X chromosome and its association with the X-Inactivation Center in 1991 [58–61] . Xist was classified as a lncRNA a year later due to lack of conserved open reading frames in its sequence and its exclusive nuclear localization [62] . Among the functionally characterized lncRNAs in higher eukaryotes , many are involved in growth , development , and differentiation [63–68] . Consequently , mis-regulation of these lncRNAs could result in tumerogenesis [69 , 70] , cardiovascular disorders [71 , 72] , or neurological diseases like Alzheimer’s and Parkinson’s [73] . Some of these lncRNAs are being exploited for the detection and treatment of cancers and lung diseases [74–76] . The mode of action by Xist in silencing the whole duplicated chromosome is being explored to treat diseases caused by trisomies [77] . Thus the investigation into the function of lncRNAs contributes significantly to our understanding of eukaryotic biology and diseases . In lower eukaryotes such as fungi , however , few lncRNAs involved in the development and stress response have been characterized , mostly in the model systems such as Saccharomyces cerevisiae and Schizosaccharomyces pombe [52 , 78 , 79] . In Neurospora crassa and Aspergillus flavus , natural antisense transcripts ( NATs ) and lncRNAs are found to be differentially expressed in response to specific stimuli that induce developmental changes [80 , 81] . One N . crassa lncRNA named qrf is experimentally shown to be important in maintaining circadian clock rhythmicity and clock resetting [82] . Although lncRNAs have been identified in phytopathogenic oomycete Phytophthora , and NATs in Ustilago are implicated in its virulence [83] , none have been functionally characterized [84 , 85] . Unfortunately , the prevalence or the potential functions of lncRNAs is unknown for human fungal pathogens . Here we discovered and functionally characterized the lncRNA RZE1 , which controls cryptococcal morphogenesis through its regulation of ZNF2 . An ortholog of RZE1 could not be identified outside of the Cryptococcus species complex by sequence similarity searches . Even within the Cryptococcus species complex , the level of sequence conservation of the RZE1 gene is lower than protein-coding genes such as ZNF2 , GPD1 , or ACT1 . This is not unexpected as lncRNAs are known to evolve at a faster rate and are sometimes species-specific . This , however , does raise the possibilities that RZE1 either is confined to the Cryptococcus species , or it retains evolutionary conservation across different fungal species but does so without sequence conservation . In higher eukaryotes , instances of functional conservation of lncRNAs without sequence conservation have been observed [85–87] . Thus , a functional equivalent of RZE1 in related and distant relatives of Cryptococcus is a possibility worth exploring . Another intriguing possibility is that in Cryptococcus , RZE1 might have evolved to be a regulator dedicated to ZNF2 and/or some of Znf2’s downstream targets . Consistently , the loss of RZE1 does not affect the transcript level of genes CNAG_03365 and CNAG_03367 that are adjacent to ZNF2 , reflective of RZE1’s selectivity towards ZNF2 and/or its regulon . This makes RZE1 different from some other lncRNAs that can act globally ( e . g . Xist for the entire X chromosome [86–88] ) . RZE1 appears to act on ZNF2 primarily through transcription regulation . There are several instances of lncRNAs in the model yeast Saccharomyces that regulate by turning on or off their target transcription directly or indirectly [48 , 89] . In this case , the amplitude of ZNF2 transcription induction was reduced about 2–3 folds in general in the rze1Δ mutant . Thus the role of RZE1 in ZNF2 transcriptional regulation appears to be more of modulating role than that of a switch . Given that other mutants with reduced ZNF2 expression level only shows decreased but not blocked filamentation , it stands to reason that the regulatory function of RZE1 goes beyond its effect on ZNF2 transcription . Intriguingly , RZE1 directly or indirectly regulates the distribution of ZNF2 transcripts , although the exact mechanism of RZE1 in regulating Znf2 activity is yet to be established . Another enigmatic aspect of RZE1 is its position effect . The genetic position of RZE1 seems to have more drastic effect on Znf2 activity than RZE1’s expression level per se . For instance , the introduction of RZE1 in the ‘cis’ position seems to restore filamentation to the rze1Δ mutant much more effectively than an ectopic copy of RZE1 or the multi-copy RZE1 in episomally maintained vector ( Fig 1B and S2 and S5 Figs ) . When comparing the level of RZE1 transcripts in the ‘cis’ complemented strain to the strains complemented with the site-directed mutated RZE1 alleles in the cis position , there were wide variations in the transcript level ( S8 Fig ) . Most of these strains expressed RZE1 at a level higher than that in the wild type , yet their transcript level does not correlate with their robustness in filamentation ( Fig 3C and S8 Fig ) . Similarly , ectopically expressing RZE1 using the strong CTR4 or the GPD1 promoter in the rze1Δ mutant yielded poorer filamentation than the RZE1 integrated at its native locus ( S5 Fig ) . Collectively , these observations indicate that the physical distance between RZE1and ZNF2 , and consequently their newly made transcripts , matters greatly for their functional efficiency . While we speculate that RZE1 specifically regulates ZNF2 , we did not find any specific repeats or regions in RZE1 that share sequence similarity with ZNF2 . This suggests that their direct interaction , if it exists , is unlikely to be based on simple sequence pairing . The identification of molecules directly interacting with this lncRNA might help solve that mystery . Nonetheless , RZE1 is the first of its kind identified in Cryptococcus . The RZE1-ZNF2 can serve as a paradigm and a stimulus for the investigation of other regulatory lncRNAs in fungal genomes . Our preliminary transcriptional analysis of XL280 and H99 strains suggests that there are more than one thousand conserved lncRNAs in serotype D and serotype A , an equivalent of 1 lncRNA gene for every 6 protein-coding genes in Cryptococcus . Thus , exploring the function of these lncRNAs may give novel insights into the regulatory networks that control the morphogenesis and virulence of this organism . Such an additional layer of genetic regulation might explain the complexity of life cycle and pathogenic strategies of these “simple” eukaryotes .
All the animal experiments were performed according to the guidelines of NIH and Texas A&M University Institutional Animal Care and Use Committee ( protocol numbers: 2011–22 and 2014–0049 ) . The strains and plasmids used in this study are listed in S4 Table . Yeast cells were grown routinely on YPD media unless otherwise specified . Mating assays were conducted on V8 agar medium in the dark at 22°C as previously described [12] . Cells collected for qPCR were grown on V8 media at 22°C in ambient air or in serum at 37°C in presence of 5% CO2 . Transformed strains were obtained by electroporation [90] or biolistic methods [91] and they were selected on YPD with 100 μg/ml of nourseothricin ( NAT ) or neomycin ( NEO ) . Strains carrying genes driven by the inducible promoter of the CTR4 gene [32] were grown in media supplemented with 25 μM of CuSO4 for suppression or 200 μM of bathocuproine disulphonate ( BCS ) for induction [32] . Agrobacterium tumefaciencs strain EHA105 containing the Ti plasmid pPZP-NATcc was used for the insertional mutagenesis as described previously [12 , 92] . Briefly , the bacterium was grown overnight at 22°C , washed twice with sterile water , and transferred to an induction medium containing 100μM acetosyringone and incubated for 6h . Overnight culture of Cryptococcus grown in liquid YPD was washed in induction medium and re-suspended to get 107 cells/ml . Fungal and bacterial aliquots ( 200μl each ) were mixed and plated on induction medium and co-cultured for 3 days at 22°C . The co-cultured cells were scraped and plated onto selective medium of YPD containing the antibiotic cefotaxime to remove Agrobacterium and the selective drug NAT 100μg/ml for fungal transformants . A total of 63 , 000 transformants were generated and screened for filamentation defect on V8 juice agar media using an Olympus SZX16 stereoscope . The colonies showing filamentation defect were further tested for intact pheromone sensing pathway by using unilateral crosses with the mating type a reference strain JEC20 . The insertion site in selected 15 transformants with znf2Δ phenotype was identified using inverse PCR and sequencing as described below . Genomic DNA from the selected insertion mutants having znf2Δ like phenotype was extracted , digested with restriction enzymes , and then self-ligated as described previously [12 , 92] . Primers AI076 and AI077 were used for inverse PCR . The PCR amplicons were sequenced and the sequence was BLAST searched against the C . neoformans ( JEC21 ) serotype D genome database at GenBank to identify the insertion site . Phenotypic assays were performed as described previously [93] . The strains to be tested were grown overnight in YPD . The cells were washed , adjusted to the same cell density ( OD600 = 1 . 0 ) , and serially diluted . The dilutions were plated onto appropriate media for various tests . To analyze the capacity of the strain to melanize , the original culture and the dilutions were plated onto media containing L-dihydroxyphenylalanine ( L-DOPA ) and incubated at 37°C in the dark . To observe capsule production , cells were plated on Dulbecco’s Modified Eagle’s medium ( DMEM ) ( Invitrogen ) and grown at 37°C under 5% CO2 . Capsule was visualized by India ink exclusion and examined under a light microscope . For testing the ability of the fungus to grow at high temperatures and their sensitivity to UV radiation , equal number of cells were plated onto YNB agar media and grown at 37°C or exposed to 300 J/m2 of UV for 1 , 5 , and 10s . Cells were then cultured at 22°C for additional 2 days . To test the tolerance of fungal cells to osmotic and other stressors , cells were grown at 22°C for 2 days on YNB medium that was supplemented with H2O2 ( 21mM ) , calcoflour white ( 200 μg/ml ) , iron chelator-bathophenanthroline disulfonic acid ( 300 μg/ml ) , SDS ( 0 . 1% , 0 . 01% and 0 . 001%/shown in SF3 ) , or caspofungin ( 16 μg/ml ) . Confrontation assays were performed as described earlier [18] . Donor cells ( OD600 = 3 ) were dropped onto YPD plates 2 . 5 days before dropping the recipient . The recipient cells ( 3μl ) of OD600 0 . 8 were dropped in close proximity without touching . The colonies were observed 60h after dropping of the recipient and photographed with Olympus SZX16 stereoscope . Gene Racer kit ( Invitrogen , CA , USA ) was used to amplify 5’ and 3’ ends of RZE1 and ZNF2 RNA . Full length transcript of RZE1 was amplified using the primers indicated in S5 Table from RNA extracted from cells grown on V8 for 24h following the manufacturer’s instructions . All the 3 sets of 5’ and 3’ RACE primers yielded the same start and stop termini . The amplified bands were separated on 0 . 8% agarose gel and extracted using the gel purification kit ( Invitrogen ) , cloned in TOPO 2 . 1 ( Invitrogen ) and sequenced . The targeted deletion of RZE1 ( excluding the first 100 bp after the transcription start ) was conducted by introducing the deletion construct with the 1 kb flanks of the gene and split part of the dominant marker NAT , by electroporation or biolistic transformation as described earlier . Mutants generated were confirmed by PCR and by the genetic linkage assay using meiotic progeny dissected from genetic crosses [94] . The deletion mutant in the MATa background was obtained by crossing the MATα rze1Δ strains with the corresponding congenic wild type a strains . For complementation , RZE1 transcript with 1kb of promoter region was amplified by PCR , digested , and inserted into the plasmid pXL1 [20] . Overexpression construct was created by amplifying the entire ORF by PCR and similarly inserting into pXL1 behind GPD1 or CTR4 promoters as described before [31 , 32] . The RZE1 multi-copy expression strains were generated by cloning the RZE1 with 1kb of its own promoter into the pPM8 vector [95] . The linearized vector was introduced into the auxotrophic rze1Δ ura5 strain by electroporation . The primers used for the generation of the mutants and for their analysis are listed in S5 Table . RNA extraction and real-time qPCR were carried out as we previously described [20] . Briefly , total RNA was extracted using PureLink RNA mini kit ( Life Technologies ) . DNase ( Ambion ) treated RNA samples were analyzed on a denaturing formaldehyde agarose gel for assessing quality and concentration . Superscript III cDNA synthesis kit ( Life Technology ) was used for the first strand cDNA synthesis following the manufacturer’s instructions . Constitutively expressing housekeeping gene TEF1 was used as endogenous control for the normalization of expression of the other genes studied . For Northern blot , total RNA was extracted from cultures on V8 medium ( pH = 7 . 0 ) from the indicated strains under the conditions as described in the texts . Poly ( A ) tailed RNAs were purified using the PolyATtract mRNA Isolation System IV ( Promega ) according to the manufacture’s instruction . PolyA tailed RNAs were then separated on denaturing formaldehyde agarose gels and then transferred to nylon membrane . The Random Primers DNA Labeling System ( Life technologies ) was used to generate probes . Primer sequences used to make gene-specific templates are listed in S5 Table . Wild-type strain XL280 and the rze1Δ mutant were cultured on YPD and V8 ( pH = 7 ) media . Cells were collected at 24 , 48 and 72h from mating media and used for isolation of total RNA . Strand-specific RNA-seq was performed at the TAMU genomic facility according to the standard protocol for Illumina Genome Analyzer IIx . Sequenced reads were aligned to the XL280 reference sequence [26] using Tophat [96] . Reads that aligned uniquely to the reference sequence were considered for gene expression quantification with Cufflinks [97] . Gene expression was normalized using the DESeq package [98] in R [99] . Differential expression analysis comparing mutant to wild type was performed with using a 5% false discovery rate . RNA-Seq data is deposited at NCBI ( BioProject ID PRGNA278291 , accession 278291 ) . To introduce single nucleotide mutations in the potential translation start sites of RZE1 , Quick Change II XL site directed mutagenesis kit from ( Agilent Technologies ) was used following manufacturer’s instructions . Primers for introducing the mutations ( S5 Table ) were designed using the QuickChange primer design program from Agilent Technologies . RZE1 along with 1kb of its 5’ flank were amplified by PCR and cloned into pGEMT easy vector ( Promega ) . The entire plasmid with insert was then amplified using specific primers to change each of the ATGs to AAGs or ATAs . The modified inserts were then amplified from the plasmid using PCR with high fidelity Phusion enzyme ( New England Biolabs ) and used as template for overlap PCR with part of NEO marker gene to obtain a merged construct of RZE1 with 5’ flank and part of the marker . Similar overlap PCR with 3’ flank of RZE1 and part of NEO was used to obtain the other half of insertion construct . The two parts of insertion construct were then mixed and introduced biolistically into Cryptococcus rze1Δ mutant strain to replace the rze1Δ deletion construct . The stable transformants that showed resistance to NEO ( indicative of the integration of the mutant RZE1 allele ) and sensitivity to NAT ( indicative of the loss of the rze1Δ construct ) were selected and further analyzed by PCR to confirm the replacement . Female A/J mice of 6–8 week were intranasally inoculated with 5x104 cells/mouse of serotype A strains or 5x105 cells/mouse of serotype D strains with 5 mice each group as we previously described [100–102] . The mice were sacrificed 10 days post inoculation . The organs from sacrificed animals were dissected and homogenized in phosphate buffered saline , serially diluted , and plated on YNB plates . After 2 days of incubation at 30°C , the CFUs were counted . One way ANOVA was used to analyze the fungal burden . All statistical analyses were performed using the Graphpad Prism 5 program and and P values lower than 0 . 05 was considered statistically significant . All strains in H99 background were grown overnight in YPD liquid medium . Cells were washed and suspended in PBS with the same cell density ( OD600 = 1 . 0 ) . Equal number of cells of each pair of strains as indicated in the figure and the text were mixed and spotted onto V8 agar pH 5 . 0 and incubated at 22°C in the dark for 2 . 5 weeks . The colony morphology was observed and photographed using an Olympus SZX16 stereo microscope . The fluorescent intensity of Znf2 tagged with mCherry was measured using the Zen Lite software from Carl Zeiss . The microscopic images taken on a Zeiss Axioplan 2 microscope were analyzed using the interactive fluorescence measurement feature of Zen . Fluorescence intensity of Znf2-mCherry from forty cells in either the wild-type background or the rze1Δ mutant background was measured using the software and statistically analyzed using Prism V from Graphpad . The smFISH experiments were performed as described before [103–105] with modifications for Cryptococcus . Strains either grown to exponential growth phase in YPD or the co-culture of the congenic mating pairs on V8 media were collected and fixed in 4% paraformaldehyde for 30 minutes at 22°C . The cells were digested in sphaeroplasting buffer ( 1M sorbitol , 10 mM EDTA , and 100 mM sodium citrate ) using 100mg/ml of lyzing enzyme ( Sigma ) for 45 min . Sphaeroplasted cells were permeabilized after being incubated in 70% ethanol at 4°C overnight . Cells were then hybridized with TAMRA labeled ZNF2 specific probes or ALEXA FLOUR labelled RZE1 probes ( Stellaris , Biosearch Technologies ) in buffer containing 10% formamide for 16h at 37°C . Counter staining to visualize nuclei was done using 10 mg/ml of DAPI . The cells were visualized on a Zeiss Imager M2 microscope and the z-stack images were then subjected to 3D deconvolution by using AutoQuant software ( Media Cybernetics ) . After deconvolution , the images were imported in Zen Blue software and the channels were merged and analyzed .
|
The involvement of non-protein regulators in developmental processes in higher eukaryotes is an area that has come to light more recently . Earlier known as the dark matter of the genome , the non-protein coding genes are now recognized for their important regulatory roles in the life of eukaryotes . Using forward genetic screen , we identified RZE1 as a lncRNA with key function in regulating morphogenesis in Cryptococcus . We further discovered that RZE1 regulates the transcription and transcript export of ZNF2 , which encodes the key morphogenesis transcription factor . RZE1 is the first functionally characterized lncRNA in a human fungal pathogen . Given the potential large number of lncRNAs in Cryptococcus and other fungal pathogens , the RZE1-ZNF2 regulatory system could serve as a paradigm for the investigation of lncRNAs in development and virulence in eukaryotic pathogens .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
The lncRNA RZE1 Controls Cryptococcal Morphological Transition
|
The Drosophila sex determination hierarchy controls all aspects of somatic sexual differentiation , including sex-specific differences in adult morphology and behavior . To gain insight into the molecular-genetic specification of reproductive behaviors and physiology , we identified genes expressed in the adult head and central nervous system that are regulated downstream of sex-specific transcription factors encoded by doublesex ( dsx ) and fruitless ( fru ) . We used a microarray approach and identified 54 genes regulated downstream of dsx . Furthermore , based on these expression studies we identified new modes of DSX-regulated gene expression . We also identified 90 and 26 genes regulated in the adult head and central nervous system tissues , respectively , downstream of the sex-specific transcription factors encoded by fru . In addition , we present molecular-genetic analyses of two genes identified in our studies , calphotin ( cpn ) and defective proboscis extension response ( dpr ) , and begin to describe their functional roles in male behaviors . We show that dpr and dpr-expressing cells are required for the proper timing of male courtship behaviors .
Genomic microarray approaches have allowed scientists to address questions that were previously intractable using molecular-genetic approaches . For example , in studies of Drosophila , microarrays can facilitate the identification of genes that underlie complex behaviors in the adult . These genes are difficult to identify using standard genetic approaches , as most genes are utilized during early development [1] . Given the likelihood of a developmental phenotype , adult-specific function is difficult to ascertain . The terminal genes in the Drosophila sex hierarchy encode sex-specific transcription factors that have been shown to play key roles in specifying sex-specific behaviors . Animals with mutations in the genes that encode these transcription factors , and that display adult-specific behavioral phenotypes , are particularly well suited for the identification of genes that underlie complex behaviors . In this study , we use genomic approaches to identify genes that are regulated downstream of the sex hierarchy transcription factors , thereby providing insight into the molecular basis of male courtship behavior . The Drosophila sex determination hierarchy consists of an alternative pre-mRNA splicing cascade that culminates in the production of sex-specific transcription factors encoded by doublesex ( dsx ) and fruitless ( fru ) ( Figure 1 ) ( reviewed in [2] ) . dsx produces both male- and female-specific transcription factors ( DSXM and DSXF , respectively ) [3] . fru is a complex locus with at least four promoters; it is the product of the P1 promoter ( fru P1 ) that is sex specifically spliced and produces male-specific FRU isoforms ( FRUM ) [4] . In Drosophila , male courtship behavior is an innate , genetically programmed behavior that consists of a series of steps performed by the male to attract a mate ( reviewed in [5] ) . The male orients towards the female , follows her , taps her with his foreleg , produces a species-specific courtship song via wing vibration , contacts the female genitalia with his proboscis , and then attempts copulation . If the female has not recently mated , she typically will allow copulation to proceed . fru P1 is necessary for specifying the potential for all of these male courtship steps [4 , 6–9] . In addition , when FRUM is produced in females , in the homologously positioned neurons in which it is normally produced in males , the early steps of the male courtship ritual are observed [10 , 11] . dsx is required for specifying all adult morphological differences between the sexes [12] , most of which are required for courtship performance [13 , 14] . dsx also functions in the central nervous system ( CNS ) to specify the potential for courtship wing song and wild-type levels of courtship performance [13–17] . The genes regulated downstream of dsx and fru P1 remain largely unknown in most tissues of the adult fly . Several genes have been identified that are regulated downstream of DSX activity in the adult internal genitalia [18] , and the Yolk protein genes have been shown to be direct targets of DSX in fat-body tissues [19–21] . Genes shown to be regulated downstream of FRUM activity , include yellow [22] , takeout [23] , and neuropeptide F [24] . To gain insight into how sex-specific behaviors are specified , sex-differential gene expression was examined in adult head tissue and dissected CNS tissue . Using a microarray approach , we have identified genes regulated by the sex hierarchy that are either direct or indirect targets of DSX and/or FRUM , in either adult head tissues or in the CNS . By extending the gene expression analyses , we have also determined new modes of DSX-regulated gene expression . We present additional molecular-genetic analyses of two genes , calphotin ( cpn ) and defective proboscis extension response ( dpr ) . We demonstrate that cpn is more highly expressed in the retina in males , as compared to females , and is downstream of DSX activity . In addition , we show that dpr is regulated downstream of fru P1 and is expressed in fru P1-expressing cells in the CNS . We demonstrate a role for dpr and for fru P1 in dpr-expressing cells in male courtship performance .
A major goal of this work is to understand , at a molecular-genetic level , how sex-differential gene expression is established in adult head tissues , to gain insight into how the potential for reproductive behaviors are established . Here , we analyzed gene expression using a glass-slide microarray approach ( see Materials and Methods ) . For each sex-differentially expressed gene identified , we sought to determine at which level or branch of the somatic sex hierarchy , sex-differential expression is established . This will provide an understanding of how a multitiered and branched genetic regulatory hierarchy deploys the genome to bring about developmental and physiological differences . The logic of our approach is as follows ( see Figure 1B ) . We first compared gene expression between wild-type male and female adult heads to identify the genes that are differentially expressed between the sexes . We next determined if regulation of the sex-differentially expressed genes is downstream of transformer ( tra ) ( see Figure 1 ) . A gene that is sex differentially expressed and not regulated downstream of tra would be inferred to be regulated at the level of sxl , or because of differences in sex chromosome composition ( Figure 1A ) . tra and sxl both encode pre-mRNA splicing factors . sxl is at the top of both the tra branch and the branch of the sex hierarchy that controls dosage compensation , the process that equalizes the amount of transcript produced from the single male X chromosome to that of the two female X chromosomes ( Figure 1A; reviewed in [2 , 25] ) . For those genes that are regulated downstream of tra , we distinguished between possible regulation by sex-specific transcription factors encoded by dsx and fru P1 by performing additional gene expression analyses . Finally , we distinguished between sex-differential expression in the CNS versus other tissues in the adult head . We first identified genes that are sex differentially expressed in wild-type adult head tissues . We compared gene expression in 0–24-h adult male and female heads , from two different wild-type strains , Canton Special ( CS ) and Berlin . We used two strains to ensure that we focused on key genes underlying the differences between the sexes . To identify genes with significant differences in gene expression , a False Discovery Rate ( FDR ) method was employed [26] . FDR is the proportion of false positives among all the genes initially declared as being differentially expressed . FDR has become a standard for multiple testing paradigms such as whole-genome microarray analyses [27 , 28] . Throughout this study , we used an FDR cutoff of 0 . 15 ( q < 0 . 15 ) , unless otherwise noted . In our analyses , candidate genes passed the FDR cutoff for multiple independent array experiments . Thus , we reasoned that allowing up to 15% of all declared differentially expressed genes within each test as false positives would be a conservative cutoff to obtain high confidence in our analyses . We combined the data from the two wild-type strains for statistical analyses , with the expectation that sex-differentially expressed genes would show the same direction of change in the two strains . We identified 754 genes that displayed significant , sex-differential expression ( see Materials and Methods ) . For these 754 genes , the range of expression values was between ∼130-fold higher in females to ∼45-fold higher in males , with 94% of genes displaying differences less than 2-fold . Three hundred thirty and 424 genes were more highly expressed in males and females , respectively . Of these 754 genes , 46 genes displayed both significant and substantial sex-differential expression ( q < 0 . 15 and fold change [FC] > 2; Table S1 ) . We confirmed our array experiments accurately identified sex-differentially expressed genes . First , we identified several genes previously shown to display robust sex-differential expression . In particular , Yolk protein 1 ( Yp1 ) , Yolk protein 2 ( Yp2 ) , and Yolk protein 3 ( Yp3 ) are highly expressed in the female , but not male , adult fat body [29 , 30] . Consistent with this , we observed over 100 times more Yp1 and Yp2 transcripts , and 40 times more Yp3 transcript in adult head tissues from females , than males . Second , we analyzed data from array features that are specific for the transcripts CG11094-RA and CG11094-RB that produce DSXM and DSXF isoforms , respectively . CG11094-RA and CG11094-RB showed significantly higher expression in wild-type males and females , respectively . Finally , several genes in this list have previously been shown to display sex-differential expression , such as female-specific independent of transformer ( fit ) and female-enriched sex-specific enzyme 2 ( sxe2 ) [31] . To distinguish between sex differences in gene expression in the CNS from those contributed by other tissues of the head , such as the fat body , we compared gene expression between male and female dissected brain and ventral nerve cord tissues . We identified four genes , including fit and CG8007 , a male-biased gene inferred to be involved in serotonin receptor signaling [32] , with significant , sex-differential expression . The other two genes are CR32998 and rna on the x2 ( rox2 , CR32665 ) , both of which encode small RNAs and are more highly expressed in males . CR32998 encodes a transcript with sequence similarity to small nuclear RNAs ( snRNA , U425F ) , and rox2 encodes a RNA component of the dosage compensation complex . In Drosophila , the dosage compensation complex mediates up-regulation of gene expression on the X chromosome in males . rox2 RNA is part of the dosage compensation complex in males [33–35] , further validating our microarray data . The identification of only four genes with sex-differential expression in the CNS is not due to poor quality data , as we had the power to detect 94 . 7% of gene expression differences ( see Materials and Methods ) . The Pearson correlation statistic among all hybridizations for these experiments was r2 = 0 . 96 ( see Materials and Methods ) , further demonstrating low experimental variation among our replicates . Thus , most sex-differential expression in adult head tissue we detected was because of expression outside of the CNS . Our findings are consistent with previous studies that have suggested that there are hundreds of genes expressed in the adult head fat body that are not expressed in the CNS [36] . It is possible that many fat body–specific mRNAs are sex differentially expressed . Indeed , previous studies have identified sex-biased fat body–specific mRNAs , such as Cyp4d21 [23 , 36] , turn on sex-specificity ( tsx ) , sxe2 [31] , drosomycin [37] , and mRNAs from the gene family takeout [23 , 36 , 38] . In our microarray data analyses of dissected CNS tissue , fit showed significantly higher expression in females than males ( see above ) , similar to a previous study that detected female-enriched expression of fit , largely in head fat cells [31] . Even if fit were solely expressed in the fat body , this is the only gene previously identified as expressed in the fat body that was identified in our analyses of gene expression in the CNS , suggesting that our dissections minimized contamination from peripheral perineuronal fat-body tissues . Our data indicate that many genes are expressed at similar levels in males and females , when gene expression in the entire CNS is assayed . However , these experiments would not allow for the identification of genes that display sex-specific spatial expression differences , but that overall are expressed at similar levels in males and females , or genes that are sex differentially expressed , but below the limit of our microarray detection , such as those genes that are expressed in very few cells of the CNS . Indeed , we did not detect dsx , yellow , or npf , in these CNS microarray comparisons , though we know that they are sex differentially expressed in the CNS [15 , 24 , 39 , 40] . We next sought to identify genes that are regulated downstream of tra in adult head tissues ( Figure 1A ) . We compared gene expression in tra mutant animals ( tra pseudomales ) to wild-type females . The tra pseudomales are chromosomally XX , but are phenotypically almost identical to wild-type males . tra is required for splicing dsx and fru P1 pre-mRNAs in females . Thus , tra pseudomales produce both DSXM and FRUM , which direct male somatic development ( Figure 1A ) . We identified 117 genes that are sex differentially expressed between wild-type males and females , and between tra pseudomales and wild-type females ( one-tailed t-test ) ( Table S2 ) ; 32 and 85 genes are more highly expressed in males and females , respectively . The remaining 637 genes out of the 754 sex-differentially expressed genes either ( 1 ) showed no significant differential expression ( 613 genes ) , or ( 2 ) showed significant differential expression in the opposite direction than in wild-type males versus females ( 24 genes , see Figure 2A ) . We examined the data for the 613 genes that showed no significant difference in expression in the tra comparison , to further determine if the sex-differential expression we observed in wild-type animals is independent of tra . One hundred sixty out of 613 genes were missing data in one or more of the four experimental replicates . For these 160 genes we had much less statistical power , compared to what we had in our experiments analyzing wild-type animals ( eight replicates ) and so were not further considered . We next examined the data for the remaining 453/613 genes , for which we had data in all four of the tra comparisons . If we reduced the stringency of our statistical cut-off ( q < 0 . 25 ) , 35/453 showed significant differential expression downstream of tra and so might be regulated downstream of TRA activity ( see Figure 2A ) . This suggests that the remaining 418/453 genes are sex differentially expressed , but are not downstream of tra . Indeed , as expected for a gene regulated independent of tra , nearly all of the 418 genes are also not downstream of dsx and/or fru P1 . Only 19/418 genes were significantly differentially expressed in our dsx and/or fru P1 microarray comparisons and thus are possible false negatives in the tra data analyses ( see below and Figure 1A ) . For the 418 genes , the median and average fold difference in expression between males and females is 1 . 14 and 1 . 07 , respectively . One possibility is that for these 418 genes , the small fold differences in gene expression between the sexes are due to having different numbers of sex chromosomes between wild-type males and females , and/or the efficiency of dosage compensation , which is regulated at the level of sxl . In the sex hierarchy , sxl is above the level of tra ( see Figure 1A ) . In the tra pseudomale to female comparison , tra pseudomales and females are both chromosomally XX and produce SXL , so gene expression differences due to differences in sex chromosomes and/or dosage compensation would be eliminated . To determine if the 418 genes we identified as sex differentially expressed , but not dependent on tra , are due to differences in sex chromosome composition , we determined the chromosomal distribution of the 418 genes . We observed a significant over-representation of genes on the X chromosome ( p = 2 . 25E-07 , hypergeometric test ) , but no significant over-representation of genes on any other chromosome . Taken together , these observations suggest that these 418 genes that display small fold differences in sex-differential gene expression and are not regulated downstream of tra . It appears they are regulated at the level of sxl and dosage compensation and/or because of differences in the number of sex chromosomes between males and females . Of the 24 genes for which expression was in the opposite direction from wild-type males versus females , six were more highly expressed in females compared to tra pseudomales , and 18 were more highly expressed in tra pseudomales than wild-type females ( Table S3 ) . The expression profiles of these 24 genes may reflect complex modes of sex-hierarchy regulation , as previously described in pupal-distal-leg segments [41] . However , given the fact that for these 24 genes there is significant over-representation ( p < 6 . 6E-3 ) of genes involved in stress-induced humeral defense against bacterial infection or high temperature , including turandot A , turandot C [42] , and immune induced molecule [43] , this suggests that many of these 24 genes are not sex-hierarchy regulated , but are induced by infection or stress . Next , we identified genes that are regulated by DSX activity in adult heads . Here we compared gene expression in chromosomally XX , dsxD pseudomales to wild-type females . The dsxD pseudomales are transheterozygous for one dsx null allele and one dsx allele that can only produce DSXM . These animals look almost identical to wild-type males , as DSXM directs male somatic development and physiology . We employed one-tailed statistical tests , reasoning that we could predict the direction of gene expression in dsxD pseudomales comparisons , based on the wild-type and tra pseudomale array comparisons ( Table S2 ) . We identified 54 genes that displayed significant sex-differential expression between wild-type males and females , as well as significant sex-differential expression between tra pseudomales and females , and dsxD pseudomales and females ( one-tailed t-tests; Table S4 ) . Forty-seven and seven genes were more highly expressed in females and males , respectively . These 54 genes will hereafter be referred to as the “DSX set . ” We identified functional annotation categories that are enriched in the DSX set , to infer processes regulated by DSX . The functional annotation analysis tool Database for Annotation , Visualization , and Integrated Discovery ( DAVID ) can be used to determine significant enrichment of functional annotations of genes in an input gene list , against a background gene list ( see Material and Methods ) [44] . Using DAVID , we identified metal ion binding ( involved in apoptosis ) and calcium ion binding ( involved in cell signaling ) as enriched functional categories in the DSX set ( Table S4; p < 0 . 05 ) . DSX might establish sex-specific differences at the morphological level by controlling cell proliferation , perhaps by regulating apoptosis or cell division , as shown previously [45] . DSX effectors may also be involved in controlling cell differentiation and physiology through calcium-mediated cell signaling . In addition , several genes identified in the DSX set have been shown to be involved in light and olfactory sensory responses . Differences in these sensory responses may underlie differences in behavioral outputs of adult males and females . DSX regulation of these genes suggests that DSX has a broader role than just specifying morphological differences between the sexes , but may also play a role in establishing the potential for behavior , in collaboration with FRUM . We identified 117 genes that are sex differentially expressed and regulated downstream of tra , but only 54/117 genes are also regulated downstream of DSX activity , leaving open the possibility that the other 63 genes are regulated by fru P1 or by another branch of the somatic sex hierarchy ( see Figures 1A and 2A ) . Of these 63 genes , we removed 25 genes from consideration , as we either did not have data in all four dsxD comparisons , or the gene displayed significant differential expression when we employed a less stringent statistical criteria ( q < 0 . 25 ) , leaving open the possibility of regulation downstream of dsx . Of the 38 remaining genes , 14 genes displayed significant differential expression in the fru P1 comparisons , suggesting regulation by FRUM ( q < 0 . 25 , see Materials and Methods and below ) . Thus , there were 24 genes that displayed significant sex-differential expression downstream of tra and were not regulated downstream of dsx or fru P1 ( Table S5 ) . This raises the possibility that there is another branch of the sex hierarchy that is responsible for differences in transcript abundances that is at the same level as dsx and fru . Another transcription factor , dissatisfaction ( dsf ) has been postulated to function in the sex hierarchy at the same level as dsx and fru [46] . Alternatively , it is possible that sex-specific differences in transcript abundance for this set of genes are regulated directly by TRA . In the latter case , a pre-mRNA transcript that is differentially spliced by TRA might have different stability , as compared to one spliced by the default-splicing pathway . Yp1 and Yp2 are direct targets of DSX that are highly expressed in female fat body tissues and show little or no expression in males [19–21 , 30] . On the basis of analyses of the Yp1 and Yp2 regulatory region , a model for how DSX regulates gene expression has been put forth [21 , 30] . In this model , the sex-specific DSX isoforms , given their identical DNA binding domains but different trans-activation and protein-protein interaction regions , have opposite regulatory functions for a target gene in the two sexes [47 , 48] . The model is based on the observation that DSXF activates Yp1 expression in females , and DSXM represses Yp1 in males . An extension of the model postulated that a male-specific gene would be transcriptionally activated by DSXM activity in males and repressed by DSXF activity in females [21 , 48–50] . Activation of gene expression by DSXM activity in males , and repression by DSXF activity in females , has been reported for several genes including takeout [18 , 23] . To examine if the model based on Yp transcriptional regulation is a general mechanism for how DSX regulates target gene expression , we performed microarray experiments that compared gene expression in adult head tissues of chromosomally XX and XY dsx null intersexual animals ( dsxd+r3/dsxm+r15 ) to wild-type females and males , respectively . Thus , for a gene that showed sex-differential expression , we could determine if DSX activated , repressed , or had no effect on the expression of the gene , in both males and females . Of the 54 genes in our DSX set , there were 19 genes that showed significant differential expression downstream of dsx , in both dsx null comparisons ( Figure 1B ) . The direction of gene expression changes suggests there are at least three different modes of DSX regulation for these 19 genes . Surprisingly , only four of these 19 genes ( Yp1 , Yp2 , Yp3 , and CG7607 ) showed the previously postulated Yp-like DSX regulation . Whereas , seven genes showed lower expression in both sexes when there was no DSX . This suggests that these genes are usually activated downstream of dsx in both sexes . In contrast , there were eight genes that showed higher expression in both dsx null genotypes compared to wild type , suggesting these genes are usually repressed downstream of dsx in both sexes . This was similar to what was first observed in our analyses of DSX-regulated gene expression at pupal stages ( L . Sanders , M . Lebo , and M . N . Arbeitman , unpublished data ) . Given this unexpected observation , we wanted to have a larger number of genes to examine the modes of DSX regulation . We defined an additional DSX-regulated set using less stringent FDR tests than we previously employed . We identified 106 genes that displayed significant sex-differential expression between both wild-type males and females ( q < 0 . 25 ) , and between females versus both tra pseudomales and dsxD pseudomales ( one-tailed t-tests , q < 0 . 25 ) ( see Figure 2B ) . Of these 106 genes , 40 genes ( Figure 3 ) also showed significant differential expression in both dsx null comparisons , suggesting DSX regulates their expression in both sexes . We first examined these 40 genes to ascertain if any appeared to be regulated in a manner similar to Yp1 and identified only six genes ( Figure 3 , gray region ) . Five of the six have data consistent with being activated by DSXF in females and repressed by DSXM in males ( Figure 3 , section IV ) . We identified one gene , CG4979 ( encodes a predicted sex-specific enzyme involved in lipid metabolism ) , which appears to be activated by DSXM activity in males and repressed by DSXF activity in females , suggesting that DSXM can be an activator in males , as predicted by the early model and in other studies [18 , 21 , 48–52] ( Figure 3 , section V ) . Thus , for a small set of genes , DSX regulation of gene expression can be described by the early model that was based on Yp1 regulation . However , for the genes in this study , regulation by DSX might be direct or indirect . The majority of genes ( 34/40 genes ) that displayed sex-differential expression downstream of dsx in both sexes appear to be activated or repressed downstream of DSX activity in both sexes , but the extent of activation or repression is sex-specific ( Figure 3 , white region ) . Of these 34 genes , 14 genes showed significantly lower expression in both chromosomally XX and XY dsx null animals , compared to wild-type female and male animals , respectively ( Figure 3 , section I ) . 20/34 genes showed significantly higher expression in both chromosomally XX and XY dsx null animals , compared to wild-type female and male animals , respectively ( Figure 3 , sections II and III ) . The majority of male-biased genes were repressed by DSX activity in both sexes , but DSXF activity repressed to a greater extent in females than DSXM activity did in males , resulting in higher expression in males . In contrast , the majority of female-biased genes were activated by DSX activity in both sexes , but DSXF activity activated these genes in females to a greater extent than DSXM did in males , resulting in higher expression in females . If any of these genes are direct targets of DSX , this observation is consistent with the fact that the sex-specific region of DSXF interacts with another protein encoded by intersex , and this interaction is required for DSXF transcriptional activity , potentially making DSXF a more potent transcriptional activator and repressor than DSXM [53] . We have performed additional microarray experiments in which we individually overexpressed each DSX isoform in adult head tissues . This expression data validated that many genes that appear to be activated or repressed by both DSX isoforms , on the basis of the dsx null comparisons , show the predicted expression changes in the ectopic expression experiments ( T . D . Goldman and M . N . Arbeitman , unpublished data ) . Thus far , we have described the modes of DSX regulation for genes that are regulated by dsx in both sexes . However , it is possible that DSX activity is only required in one sex for sex-differential expression . This has previously been suggested for four genes expressed in the male accessory gland and one gene expressed in the female spermathecae [18] , as well as by others [51 , 52] . We determined if the remaining 66/106 genes of DSX-regulated genes , identified above , are regulated by DSX in only one sex ( see Figure 2B ) . For 34/66 genes we either did not have sufficient array data , or the data were highly variable and at the borderline of our significance cut-offs above , and so we did not consider these genes further . Twelve of the remaining 32 genes appear to be regulated by DSXF and not DSXM . Of these 12 genes regulated only by DSXF , nine appear to be activated and three repressed ( Table S6 ) . The remaining 20 genes showed significant expression differences in the male dsx null comparisons only , suggesting that these genes are regulated only by DSXM and not DSXF; here 17 and three genes appeared to be repressed or activated by DSXM , respectively . Again , this suggests that the primary mode of DSXM regulation is repression . Taken together , we have shown that for 106 genes regulated downstream of DSX activity , there are four main modes of regulation: ( 1 ) DSX is either an activator or repressor in both sexes , ( 2 ) DSX acts as an activator in one sex and a repressor in the other , ( 3 ) genes are only regulated by DSXF activity , and ( 4 ) genes are only regulated by DSXM activity . We chose one gene , calphotin ( cpn ) , regulated by DSX to analyze more extensively . CPN is present in photoreceptor cells of the developing eye imaginal disc [54] and functions in rhabdomere and photoreceptor development , as cpn mutants display photoreceptor cell death and have misoriented and disrupted rhabdomere structures [55] . In the adult , cpn flies lack pigment in some parts of the eye and show a rough eye phenotype [55] . Additionally , adult cpn flies show a slight reduction in phototaxis towards visible light [55] . Although rhabdomere structure is abnormal , phototaxis functionality appears to be largely intact , leaving the role of cpn in adult physiology an unanswered question . CPN contains a leucine zipper and has been suggested to have calcium-binding properties and play a role in signal transduction by regulating free calcium levels in photoreceptor cells [54 , 56] . Several genes that are thought to be involved in phototransduction or in light-induced release of internally sequestered calcium ions , including cpn , were identified as regulated downstream of DSX activity . This includes no receptor potential A ( CG3620 ) , lightoid ( CG8024 ) , neither inactivation nor afterpotential A ( CG3966 ) , neither inactivation nor afterpotential C ( CG5125 ) , G protein 49B ( CG17759 ) , and bride of sevenless ( CG8285 ) ( see Figure 3 ) . Given that several genes involved in phototransduction and genes encoding rhodopsins have previously been identified with soma-biased , sex-differential expression in adult flies [37] , we wanted to determine if cpn might be involved in sex-specific photoreceptor electrophysiological responses in the adult . These sex-specific differences in visual transduction could account for differences in male and female behaviors that require vision , such as differences in circadian activity [57–59] and reproductive behaviors [60–62] . Consistent with the idea that cpn may play a role in behaviors , a previous whole-genome array study to identify clock-controlled genes identified cpn as a gene under circadian regulation in the adult head [59] . We first performed real-time ( RT ) PCR assays to confirm our microarray results and observed significantly higher cpn expression in dsxD pseudomales than wild-type females ( ∼3 . 5-fold , p < 0 . 015; Figure 4A ) . Our RT-PCR results are consistent with our microarray results where ∼2 . 2-fold higher cpn expression was observed in dsxD pseudomales than females . On the basis of our dsx null microarray comparisons , cpn appears to be repressed by DSX in both males and females , but to a greater extent by DSXF than DSXM , which ultimately leads to higher expression in males than females . We wanted to determine if sex-differential expression within head tissues was due to cpn being expressed in the same regions , but more highly in males than females , or if cpn is expressed in broader spatial domains in males than females . Accordingly , we performed frozen section in situ analysis and observed higher expression in dsxD pseudomales and wild-type males as compared to wild-type females ( Figure 5 ) . The observed cpn expression is consistent with previous reports of expression restricted to the photoreceptor cells throughout the retina [54 , 56] . The spatial expression regions are similar in males , females , and pseudomales . This suggests that the sex-specific differences in cpn expression levels are not due to broader spatial expression in males , but are due to higher expression in the same tissue in which cpn is expressed in females . These results , together with the observation that several genes that underlie visual transduction are sex differentially expressed , suggest that quantitative differences in gene expression in photoreceptor cells between males and females may account for differences in sex-specific visual physiological and behavioral responses . Given the observation that most genes do not display sex-differential gene expression in CNS tissues ( see above ) and that fru P1 ( transcripts from P1 promoter produce FRUM ) acts predominately in the nervous system [4 , 10 , 63 , 64] , we propose that fru P1 functions to specify male-specific behaviors by modulating expression of genes that are expressed in both male and female head tissue . Given the modulatory nature of cis-regulatory promoter regions , these genes are likely to be regulated in males and females by different transcription factor sets , in different cells , and thus may have the same overall expression levels in males and females , but have different spatial patterns . Thus , we do not expect genes regulated by FRUM activity to be male-specific or enriched , but we do expect them to show differential expression when comparing expression in fru P1 mutant males to wild-type males . Accordingly , we identified a set of genes regulated by FRUM , by comparing gene expression in adult heads , between males from two different wild-type strains ( CS and Berlin ) versus males from two different fru P1 allele combinations ( fru440/p14 and fruw12/cham5; see Figure 1 ) . The fru P1 allele combinations are null/strong loss-of-function allele combinations for fru P1 transcript classes [6 , 65 , 9] . We note that one of these fru allele combinations ( fru440/p14 ) also removes fru P2 transcripts and some expression differences we report may be due to the loss of transcripts produced from the P2 promoter . Additionally , the fru P1 mutant allele combinations are transheterozygotes for deficiency chromosomes and so may be hemizygous for loci adjacent to fru . By requiring that the genes show expression differences between two wild-type and two fru P1 allele combinations , we can reduce the likelihood of identifying genes that are differentially expressed because of differences in strain background that affect transcription , which can be substantial [66 , 67] , or because of the other genes located near the fru locus that are removed by each fru deficiency combination used here [6 , 65] . We combined the data from the two wild-type male versus fru P1 male comparisons for statistical analyses , with the rationale that differentially expressed genes should show the same direction of change in the two mutant combinations as compared to wild type . In our identification of the DSX set , we used three genotype comparisons and thus used stringent statistical criteria , as a gene had to pass three different FDR tests . For the fru P1 comparisons , we only used one comparison to define genes regulated by FRUM ( wild type versus fru P1 males ) . Thus , we used a more stringent FDR cutoff ( q < 0 . 05 ) , as well as FC > 2 , as criteria . This resulted in a list of 90 genes , of which 54 and 36 showed higher expression in fru P1 males and wild-type males , respectively ( Table S7 ) . This set of fru P1-regulated genes will hereafter be referred to as the “FRUM head set” . As confirmation of the microarray expression data , fruitless was identified in the FRUM head set , showing higher expression in wild-type males than in fru P1 males . Using DAVID , we identified functional categories that are enriched in the FRUM set; 88/90 genes were used in this analysis , as two of the 90 FRUM head set genes did not have GenBank accession numbers ( Table S7; p < 0 . 05 ) . The FRUM head set was significantly enriched for functional categories involved in sensory perception , including the categories “response to light” ( p < 0 . 019 ) and “response to both physical and chemical stimuli” ( p < 3 . 9E-3 ) . These over-represented functional categories are consistent with recent reports that fru P1 is localized not only to the CNS [64] , but also the peripheral nervous system ( PNS ) [10 , 63] . We also report significant enrichment ( p < 2 . 9E-3 ) for cytochrome p450 genes , which are involved in steroid metabolism and xenobiotic detoxification . Cytochrome p450 genes are known to have sex-differential expression in the fat body [23 , 31 , 36] . Male courtship behavior has been shown to be disrupted via feminization of the fat body [38] . Perhaps one way that FRUM regulates behavior is by modulating steroid metabolism and affecting circulation levels of steroid hormones [38] . This may occur by influencing gene expression in non-neuronal tissues indirectly , as fru P1 has not been detected in fat body tissues using various techniques [64 , 65 , 68] . Additional possibilities are that subsets of the cytochrome p450 genes are expressed in the nervous system , or alternatively , fru P1 may be expressed at low levels in fat body tissues . Other genes that that are enriched in our FRUM head set are those involved in circadian rhythm processes ( p < 2 . 4E-5 ) . Other studies have shown that the establishment and maintenance of neurons involved in circadian processing impacts the number of fru P1-expressing cells . For example , the number of fru-expressing neurons decreases when toxic proteins are expressed in timeless ( tim ) cells , suggesting the possibility that tim-expressing cells might synapse on fru-expressing cells and that this is required for maintenance and modulation of a subset of fru-expressing cells [36] . Additionally , neuropeptide F ( npf ) -expressing cells are a subset of tim-expressing neurons , and fru P1 brains show reduced npf expression [24] . CG9377 is among several genes encoding serine-type peptidases identified as having differential expression as a consequence of fru P1 expression , and shows circadian-dependent expression levels [57 , 58] . The enrichment of genes that are under circadian regulation that are also regulated by fru P1 , supports the idea that there is a direct molecular tie between regulating circadian rhythms and courtship behaviors , as would be expected for a behavior that displays periodicity based on the circadian clock . To identify genes that underlie behavior , and to distinguish between genes that show differential expression as a consequence of fru P1 expression in the CNS from those expressed in other tissues of the adult head , we performed microarray experiments using RNA extracted from dissected CNS tissues . Microarray experiments were performed using RNA extracted from both dissected brains and ventral nerve cords from wild-type males and fru P1 males ( see Materials and Methods ) . We identified 26 genes ( hereafter called the “FRUM CNS set” ) with significant differential expression , of which 17 and 9 showed higher expression in fru P1 male and wild-type male dissected CNS tissue , respectively ( Table S8 ) . While the feature for the fru transcript showed borderline FDR significance , a modified Student's t-test showed a significant difference in expression ( p < 0 . 004 ) , and in all six experiments , the data from this fru array element showed higher expression in males with an average FC > 1 . 5 . We identified capability ( capa ) in our FRUM CNS set with higher expression in dissected CNS tissues of wild-type males than fru P1 males , thus capa may be induced by FRUM . capa is a gene predicted to be involved in neuropeptide hormone signaling and ion transport [69–71] . We validated the microarray expression results using RT-PCR and report significantly higher expression in the dissected CNS tissue of males than fru P1 males ( ∼1 . 4 fold , p < 9 . 2E-4; Figure 4C ) . Using DAVID , we identified “ion transport” ( p < 0 . 02 ) and “establishment of localization” ( p < 0 . 03 ) as functional categories over-represented in the FRUM CNS set . The FRUM CNS set includes CG8713 and CG11710 , genes whose products are inferred to play a role in potassium ion transport and transmission of nerve impulses , and to have transcription cofactor activity , respectively [72 , 73] . Additionally , the FRUM CNS set includes capa ( see above ) , and resistant to dieldrin , a gene shown to have gamma-aminobutyric acid ( GABA-A ) receptor and neurotransmitter activity [74] . In light of recent studies that showed that the fru P1-expressing circuit is present at a morphologically indistinct level in both males and females , [10 , 63 , 75] with the exception of small differences in cell number [75] , the functional categories that we identified are consistent with FRUM playing a role in neurophysiology or fine-scale connectivity , as has been suggested [10 , 63] . Our data showed substantially more genes with significant differential expression between wild-type males and fru P1 males when we assayed RNA extracted from adult heads , versus RNA extracted from dissected brains and ventral nerve cords . This was also the case when we analyzed sex-differential expression between RNA from males and females from head tissues , compared to RNA derived from CNS tissues . We have the power to detect 99 . 5% of true positives and report a high Pearson correlation ( r2 = 0 . 96 ) among all FRUM CNS set experiments ( see Materials and Methods ) , demonstrating that the reason we identified fewer genes in the FRUM CNS set is not because of a technical problem . This suggests that the majority of genes that were identified as regulated by FRUM in the adult head are expressed outside of the brain and ventral nerve cord . Given this gene list , these genes are most likely expressed in the fat body and other PNS tissues or perhaps glial cells . There have been several studies that implicate the fat body as playing an important role in behaviors , by potentially producing secreted circulating proteins . One such gene , takeout ( to ) , was shown to encode a secreted signaling molecule that can be found in the hemolymph and is regulated by the sex hierarchy [23] . TO has much higher levels in males as compared to females , and to mutants show a reduced courtship index [23 , 38] . On the basis of these studies it was proposed that TO may be a fat body–diffusible factor that serves hemolymph-brain communication to help regulate courtship activity [23 , 38] . As noted above , other studies have identified many sex-differentially expressed genes in adult head tissue [23 , 31 , 36] , however previous studies did not distinguish between the numbers of genes that were differentially expressed in the CNS versus other tissues of the head . Furthermore , those studies did not determine the role of fru P1 in establishing gene expression levels in these other head tissues , as it is thought that the primary role of FRUM is in the nervous system . Our expression results identified hundreds of genes ( 754 ) with significant sex-biased expression in the adult head that are not significantly sex differentially expressed in wild-type CNS tissues . Although our stringently defined FRUM head set contains 90 genes ( q < 0 . 05 , FC > 2 ) , by relaxing the statistical stringency , we have identified over 1 , 000 additional genes with significant differential expression in fru P1 versus wild-type adult male heads that are not significantly differentially expressed in fru P1 versus wild-type male CNS tissue comparisons ( q < 0 . 15 , 1 , 554 genes , unpublished data ) . Furthermore , none of the six genes identified as having significant differential expression in both the FRUM adult head and FRUM CNS datasets were identified as having significant sex-biased expression in either the wild-type adult head , or the wild-type dissected CNS datasets . Taken together , our results suggest the possibility of independent sex-specific and fru P1-specific regulation of fat body genes . Furthermore , these studies bolster the previously suggested idea that fru P1 likely influences adult-male fat body–gene expression [23 , 76] . Given the complexity of male courtship behavior and the essential role of FRUM in specifying this behavior , it is surprising that a greater number of genes that are regulated downstream of FRUM activity were not identified in the CNS . FRUM is expressed in ∼1 , 200–1 , 300 adult CNS cells [10 , 64] . Although we detect differences in fru P1 levels in our experiments , perhaps FRUM targets are expressed at lower levels . It is also possible that FRUM targets are not regulated by fru P1 throughout the circuit , or are under the regulation of other transcription factors in other cells of the CNS , thus making it difficult to detect expression level differences in the comparisons we have made . Future studies analyzing gene expression in subsets of FRUM-expressing cells will address this concern . Given the observation that both males and females have nearly the same number of homologously positioned fru P1 neurons and these neurons have similar axonal projection patterns [10 , 63 , 75] , it has been proposed that FRUM in males may either modulate neuronal activity or fine-scale connectivity to bring about the potential for male courtship behaviors [10 , 63] . Thus , we searched our dataset to find genes whose functions are consistent with these roles to further analyze and chose defective proboscis extension response ( dpr ) . dpr was identified in a genetic screen for genes that underlie the behavioral response of proboscis removal from a high salt solution [77] . dpr is the founding member of a family of genes encoding predicted cell adhesion molecules that contain two Ig domains . Drosophila Ig-containing proteins have been classified as either secreted or membrane-bound ligands , cell adhesion molecules , or transmembrane receptors . Since the cytoplasmic domain of DPR is only 75 amino acids , DPR is inferred to be a cell adhesion molecule , or a membrane-bound ligand . Previous studies identified a family of 20 dpr-related genes ( dpr1–dpr20 ) that encode predicted proteins with 30%–52% amino acid similarity within both Ig domains of DPR [77] . Nineteen of the 20 dpr-family genes were represented as features on our microarrays . Several members of the dpr family appear to be regulated by FRUM ( Figure 4B ) . Statistical analyses of our array data using FDR identified five dpr-family genes with significantly higher expression in wild-type males than fru P1 males ( q < 0 . 1 ) . The observation that five out of 19 dpr-family members are differentially expressed is significantly greater than what is expected , on the basis of the number of dpr genes in the genome ( p < 0 . 01 , hypergeometric test ) . We furthered these analyses by determining if there is an enrichment of dpr-family members in genes that are regulated by FRUM , by examining the fru440/p14 and fruw12/ cham5 versus wild-type male comparisons independently . Here we reasoned that differences in the fru allele combinations might affect dpr-family gene expression differently . We found eight and two additional dpr-family genes with significantly higher expression in wild-type males than fru4–40/p14 males and fruw12/cham5 males , respectively . Overall , 15 of the 19 dpr-family genes showed significantly higher expression in males than either fru440/p14 or fruw12/cham5 , or both . This is significantly greater than what is expected , on the basis of the number of dpr-family genes in the genome ( p < 1 . 0E-4 , hypergeometric test ) , suggesting this family of genes may be regulated in a similar manner by FRUM . Here we focus on dpr , which showed higher expression in wild type than fru P1 head tissues ( ∼1 . 3 fold , q < 0 . 036 ) . We confirmed our microarray results ( Figure 4B ) by RT-PCR and found that dpr is significantly higher ( ∼4 . 4 fold , p < 0 . 05 ) in wild-type males as compared to fru P1 mutants . We next determined if FRUM is expressed in subsets of dpr-expressing cells , to ascertain if it is possible that direct regulation of dpr by FRUM could account for reduced dpr expression levels in the fru P1 mutant . We visualized FRUM and dpr-expressing cells using immunohistochemistry . dpr-expressing cells were detected using an upstream activating sequence ( UAS ) directing expression of nuclear green fluorescent protein ( GFP ) reporter , driven by GAL4 that is inserted in the dpr locus ( hereafter called dpr-GAL4 ) . The dpr-GAL4 expression pattern has been previously characterized and shown to recapitulate most endogenous dpr expression [77] . We observed colocalization of FRUM and dpr-expression in the nuclei of ten to 15 neurons that are below the median bundle axon tracts ( Figure 6J ) . When we examined the dpr-expressing cells projection patterns ( Figure 6G , arrow ) , it appears that they are part of the ascending median bundle neurons . Subsets of neurons in the ascending median bundle were previously described as playing a role in courtship gating , the process that controls the timing and the sequence of progression through the courtship ritual [78] . The nuclei of these ascending median bundle neurons are in the subesophegial ganglion ( SOG ) region; the SOG contains neurons that are innervated by primary gustatory sensory neurons that extend from the proboscis to the brain [79] . The SOG is also the termination site of labellar and tarsal gustatory neurons [80–83] . We also observed colocalization of FRUM and dpr-expressing cells in three to six cells of the first thoracic segment in the ventral nerve cord ( Figure 6H and 6K ) . This region of the CNS has been implicated in wing song formation during courtship performance [84 , 85] , though further experimentation is required to ascertain if altered dpr expression in fru P1 mutants affects courtship song . We observed close positioning of many of the dpr-expressing cells with FRUM-expressing cells in the SOG and the pars intercerebralis ( PI ) region ( Figure 6G , labeled “PI” ) , raising the possibility that fru P1 influences dpr expression via synaptic connections and neuronal activity of FRUM-expressing cells , rather than directly regulating transcription . The PI region is known to contain large neurosecretory cells . It has been shown that ablation or perturbation of four neurons in the pars intercerebralis region , leads to increased courtship activities [86] . Perhaps , dpr-expressing cells in the PI are involved in modulating courtship activity rate and latency , which is consistent with the behavioral phenotypes we observed for dpr ( see below ) . Overall , given that some dpr-expressing cells overlap with fru P1-expressing cells , it is possible that FRUM directly regulates levels of dpr expression in subsets of dpr-expressing cells . Alternatively , wild-type activity of fru P1-expressing neurons may be important for the maintenance of dpr expression levels , especially in dpr-expressing cells that are in close proximity to fru P1-expressing cells , like those in the SOG and PI regions of the brain . Given the observation that fru P1 is expressed in the PNS [10 , 63] , we wanted to determine if FRUM plays a role in regulating dpr expression in the PNS . However , because FRUM and dpr-expressing cells in the PNS are both only detectable using the GAL4 system , we were limited in our ability to determine colocalization [10 , 63 , 75] and so here describe our observation of dpr expression in the PNS . We observed dpr-expressing cells in the forelegs and proboscis ( Figure S1 ) , consistent with previous descriptions [77] , and near previously reported fru P1-expressing cells [10 , 63] . Thus , an intriguing possibility is that dpr may play a role in both primary sensory cells , as well as in cells involved in higher order processing of gustatory information , such as those we detect in the SOG region . Given that dpr was identified as being regulated by FRUM , we sought to determine if the number and pattern of dpr-expressing cells is sexually dimorphic , in the adult or pupal CNS . In the adult and pupal male brain and ventral nerve cord ( Figure 6A–6C ) we observed a broad distribution of dpr-expressing cells throughout the midbrain , optic lobes , and ventral nerve cord , which is consistent with previous DPR expression studies in adult male head sections [77] . We did not observe any major differences in dpr-expression patterns at the gross morphological level between adult ( Figure 6D–6F ) or 48-h pupal male and female brains and ventral nerve cords ( compare Figure 6A and 6C to Figure 6I and 6L ) , though quantitative differences are difficult to detect using the GFP reporter . Possibly , other non–sex-specific transcription factors establish the dpr spatial expression patterns , and in males FRUM is required to maintain the levels of this expression . Alternatively , normal transcription levels of dpr in males may require fru P1-expressing cells to have wild-type neuronal activity , and the effects on dpr expression may be indirect . In females , other transcription programs/neuronal activity would be needed to maintain the dpr expression pattern To determine if dpr and fru P1 expression in dpr-expressing cells have a role in courtship behavior , we performed courtship analyses on dpr mutants and mutants in which the fru P1 transcript is reduced in dpr-expressing cells . The fru P1 transcript was reduced using a UAS-RNAi transgene called fru P1-RNAi [78] , with expression driven by a dpr-GAL4 transgene . To account for any possible background genetic effects , all transgenic strains described here were backcrossed to the same strain of w , CS flies ( see Materials and Methods ) . The dpr-GAL4 P-element insertion causes a loss-of-function mutation at the dpr locus [77] . We quantified the amount of time it takes for a male to initiate wing extension , one of the steps in the male courtship ritual that can be reliably assayed . We observed a significantly reduced time to first wing extension in homozygous dpr-GAL4 mutant males , as compared to wild-type males , and UAS-fru P1-RNAi transgene males ( Figure 7A ) . Interestingly , we also saw a significant reduction in time to first wing extension in males that are transheterozygous for the dpr-GAL4 allele and a UAS-GFP RNAi transgene , suggesting that the dpr-GAL4 allele is a dominant allele with respect to initiation of wing extension or there is a nonspecific effect on wing extension due to the expression of the GFP-RNAi transgene . We observed an even greater reduction in time to first wing extension in dpr-GAL4; fru P1-RNAi males , compared to all the other male genotypes , suggesting that both dpr and fru P1 expression in dpr-expressing cells are important for the wild-type timing of wing extension initiation . We next quantified the time to first attempted copulation and observed a significant reduction in the amount of time it takes dpr-GAL4; fru P1-RNAi transgene males ( 95 . 3 +/− 10 . 8 s , Figure 7B ) , as compared to all the other male genotypes we assayed . No significant differences in time to attempted first copulation were observed in males homozygous or heterozygous for dpr-GAL4 , as was observed for wing extension . This suggests that the dpr-GAL4 mutation had an effect on courtship timing up through wing song , but that abrogating fru P1 function had an additional effect on the gating of courtship progression through to attempted copulation . Furthermore , a higher percentage of dpr-GAL4; fru P1-RNAi male flies performed wing extension ( WE ) and attempted copulation ( AC ) ( WE: n = 66 , 100% and AC: n = 63 , 95% , respectively ) within the allotted time , as compared to wild-type males ( WE: n = 30 , 91% and AC: n = 27 , 82% , respectively ) , suggesting that the reduction of fru P1 expression in dpr-expressing cells manifests in quicker initiation of courtship activity throughout several stages of the courtship ritual . Given that dpr-GAL4; fru P1-RNAi males are heterozygous for the dpr-GAL4 allele , it is possible that some of these phenotypes are an additive effect of removing dpr function and removing fru P1 function . The dpr-GAL4; fru P1-RNAi courtship phenotype is consistent with what was observed when fru P1 transcript was reduced in the ascending median bundle neurons using a different GAL4 driver called p52a-GAL4 [78] . p52a-GAL4; fru P1-RNAi mutant males courted females much more rapidly; <10 s for latency for courtship initiation , versus ∼70 s in wild-type CS flies , and ∼80 s for other control flies [78] . In addition , under wild-type conditions , initiation of courtship by one male towards a female will delay the initiation of courtship of a second male towards the same female [78] . In contrast , multiple p52a-GAL4; fru P1-RNAi mutant males were shown to immediately court and attempt copulation with a female , suggesting that inhibitory cues from other males or male–female pairs act through the ascending median bundle neurons [78] . We also observed this effect , as multiple dpr-GAL4; fru P1-RNAi males courted females without delay ( unpublished data ) . We observe this phenotype only when the dpr-GAL4; fru P1-RNAi flies were raised at 29 °C , but not at 25 °C , suggesting the fru P1-RNAi transgene is more effective at 29 °C than at 25 °C . Taken together , the courtship phenotypes associated with reducing fru P1 in the ascending median bundle suggest that the ascending median bundle neurons are important for timing of courtship progression . Besides reduced latency in courtship initiation and attempted copulation , dpr-GAL4 and dpr-GAL4; fru P1-RNAi males appear normal morphologically and are not sterile . Additionally , these flies do not appear to have any mate-recognition problems , as they do not show any male–male chaining behavior . To determine if the reduction in time to WE and AC in dpr-GAL4; fru P1-RNAi , and dpr-GAL4 strains was due to a specific effect on courtship behaviors versus a nonspecific effect of increased locomotor activity , we assayed activity of these flies compared to controls ( Figure S2 ) . Here we observed a reduction in locomotor activity in the dpr-GAL4; fru P1-RNAi , and dpr-GAL4 strains , compared to controls . During the time we performed our courtship assays the differences were not as robust as other times of the day ( Zeitberger Time [ZT] 5–9 ) . This suggests that the increased courtship activity we observed in dpr strains was not due to nonspecific increases in general locomotor activity . We cannot rule out that decreased locomotor activity results in increased courtship . However , it has been suggested that decreased locomotor activity is associated with increased latency to copulation [87] , and that peak mating frequency occurs at times when flies are most active [62] , suggesting that the courtship phenotypes we observed in the dpr mutants were not due to decreased locomotor activity . In this study we used microarray approaches to identify and describe genes that are sex differentially expressed and/or regulated by DSX ( 54 genes ) and FRUM in adult head ( 90 genes ) and adult CNS tissues ( 26 genes ) . On the basis of these large datasets , we have described new models for DSX regulation of gene expression . One of the most striking observations from these studies is that the majority of genes that display sex-differential expression in the adult head are likely expressed in tissues outside of nervous tissue , such as those of the adult fat body . Furthermore , a large number of genes that are regulated by FRUM are also expressed outside the CNS and appear to also be expressed in adult fat body . These results , together with recent work from other groups , suggest that modulation of adult behaviors including reproductive [38] , feeding [88] , aggressive [89] , and circadian behaviors [36] is achieved through the activity of the fat body , perhaps through secreted molecules that affect neuronal function . In the future , functional analyses of genes expressed in the fat body will provide further insight into how this tissue influences sex-specific behavioral activities . We focused our functional analyses on male-biased genes that are expressed in the CNS , but many of the genes we identified may underlie the potential for other sexually dimorphic behavioral activities , including female receptivity , egg laying , feeding behaviors , circadian behavior , and aggression . Here we provide molecular inroads for the future study of these behaviors .
All stocks were grown using standard conditions at 25 °C unless otherwise noted . Wild-type stocks were CS , Berlin , and w; Berlin . Chromosomally XX , tra pseudomales are the genotype y , w , P [w+cM , ubi-gfp]/w; tra1/ Df ( 3L ) st-j7 . For the following strains , chromosomal sex appears in parentheses . Chromosomally XX , dsxD pseudomales are the genotype y , w , P [w+cM , ubi-gfp]/w; dsxD , Sb/dsxm+r15 ( XX ) . dsx intersexual animals are the genotypes y , w , P [w+cM , ubi-gfp]; dsxd+r3/dsxm+r15 ( XX ) and dsxd+r3/dsxm+r15 ( XY ) . The genotypes of fru P1 males are fru4–40/frup14 ( XY ) and w; fruw12/fru cham5 ( XY ) . The dpr-GAL4 [77] line was kindly provided by C . Montell ( The Johns Hopkins University School of Medicine , Baltimore , MD , United States ) . The UAS-fruRNAi line ( w; P [w+cM , UAS-fruRNAi]/CyO; P[w+cM , UAS-fruRNAi]/ P[w+cM , UAS-fruRNAi] ) was kindly provided by D . Manoli in the laboratory of B . Baker , ( Stanford University , Stanford , CA , United States ) . All strains used in courtship assays were outcrossed for six generations to w , CS kindly provided by D . Guarnieri , in the laboratory of U . Heberlein ( University of California at San Francisco , San Francisco , CA , United States ) . We employed a two-color DNA microarray approach [90] , using glass-slide arrays spotted with 15 , 158 oligonucleotide probes representing all known and predicted open-reading frames , based on release 4 . 1 of the D . melanogaster genome . The long oligonucleotide sequences were designed by the International Drosophila Array Consortium ( INDAC ) using a custom implementation of OligoArray2 [91] . The oligonucleotides were designed with size ranges between 65–69 nucleotides , a minimal Tm window , bias towards the 3′-ends of transcripts , and minimal sequence similarity to other genes [92] . The oligonucleotides were synthesized by Illumina . The oligonucleotide sequences can be downloaded from Flymine: http://www . flymine . org/release-5 . 0/aspect . do ? name=INDAC . Microarrays were printed in the laboratory of Eric Johnson ( University of Oregon , Eugene , OR , United States ) using slides coated with aldehyde chemistry and were postprocessed using the Nunc SuperChip Aldehyde protocol ( Thermo Fisher Scientific ) . For experiments in which RNA was extracted from adult head tissues , flies were collected under CO2 anesthesia ( ZT 2 ) and allowed 8 h ( ZT 10 ) to recover to minimize identifying genes induced by exposure to CO2 , before being snap frozen in liquid nitrogen . All collections were timed such that all adult flies were between 8–24 h posteclosion when snap frozen . For experiments performed using RNA from dissected CNS tissues , animals were lightly anesthetized at the time of dissection and dissections were performed rapidly . Total RNA was extracted from fly heads or dissected CNS tissues by homogenization and extraction using TRIzol reagent ( Invitrogen ) . All experiments included dye-flip hybridizations , to minimize identifying genes because of differences in dye incorporation , with one half of the comparisons labeled in one dye orientation and the other half the other dye orientation . For experiments using RNA derived from head tissues , total RNA ( 20 μg ) was used as template to make cDNA in the presence of Cy-labeled nucleotides ( direct labeling ) . Direct labeling was achieved through a 2-h reverse transcription reaction at 42 °C . Final concentrations are indicated in parentheses: oligo dT primer ( Operon , 3 . 75 μM ) , dithiothreitol ( Invitrogen , 10 mM ) , First Strand Buffer ( Invitrogen , 1× ) , dNTPs minus dTTP ( Invitrogen , 0 . 5 μM ) , dTTP ( Invitrogen , 50 nM ) , Cy-labeled dUTP ( Perkin-Elmer , 0 . 625 nM ) , and Superscript II reverse transcriptase ( Invitrogen , 10 U/μL ) . The reaction was quenched using NaOH ( 167 mM ) and EDTA ( 83 mM ) and purified using the Qiaquick PCR purification kit ( Qiagen ) . Targets were dried and resuspended in formamide ( 25 μM ) , 3 M NaCl , and 0 . 3 M sodium citrate buffer ( SSC , 3 . 3× ) , SDS ( 1 . 1% ) , Denhardts ( 5 . 56× ) , and polyA solution ( 8 . 88 μM ) , and boiled for 2 min . The resuspended labeled cDNA was applied to the microarrays and allowed to hybridize 16–18 h at 42 °C . Slides were washed in 300 ml of a 1 . 5% SDS , 1× SSC solution for 5 min , followed by a 5-min wash in 0 . 2 × SSC , and twice for 10 min in dH2O , spun dry , and scanned with 635-nm and 532-nm lasers using a Genepix 4100A scanner ( Axon Instruments ) . cRNA probes were made from RNA derived from CNS tissue using the Amino Allyl MessageAmp II aRNA Amplification Kit ( Ambion ) . All experiments were performed using at least four independent biological samples of ∼100 fly heads each for head experiments and ∼20 dissected brains and VNCs for CNS experiments; for the experiments comparing wild-type male versus female and fru P1 male versus wild-type male , eight and six independent biological samples were used , respectively . Poor quality features , identified by visualization of the scanned microarray , were removed from the raw data before analysis using Genepix software . Resulting Genepix data were analyzed using the R Bioconductor software package [93] . Filtering identified and kept features containing >75% of total pixels above 1 standard deviation above background for further study . Data were normalized using the Bioconductor lowess method [94] . All analyses were performed on log-transformed ratio values . To adjust for multiple testing , the positive FDR [26] was calculated for each gene using the Bioconductor FDR package to determine the significance of differences . All significance tests used a q < 0 . 15 for significance cutoff unless otherwise noted . Antilogarithm ( base 2 ) was applied to the data to obtain FC values . Post hoc power calculations were based on a t-test using <0 . 05 to measure the difference between the means of two independent groups and were computed using the GPower3 . 0 . 3 program [95] . The effective size for power was calculated using the mean intensity and standard deviation of all the genes within the 50th percentile of genes with the least variance . Pearson correlation ( r2 ) was determined across all experiments using dye-flip intensity values for all genes with a calculated variance . The expression data are publicly available at http://www . ncbi . nlm . nih . gov/geo/ . An analysis of the combined data for CS and Berlin demonstrates that we have the power to detect 87 . 1% of gene expression differences ( see “Microarray data analyses” in Materials and Methods ) . We calculated the Pearson correlation ( r2 ) between all hybridizations and report r2 = 0 . 88 for these experiments , suggesting that we do not have a high amount of experimental variation between our replicates . Sex-biased , tra-independent genes were identified by considering only significant sex-biased genes that did not pass a q < 0 . 25 statistical cutoff and had data in all four tra versus wild-type female array replicates . Similarly , tra-regulated , but DSX- and FRUM- independent genes , were identified by considering only genes that were significant in the tra comparison; did not pass q < 0 . 25 for the dsxD comparison; had data for all four dsxD versus wild-type female array comparisons; did not pass q < 0 . 25 for the fru P1 comparisons; and had data in at least six of the eight fru P1 versus wild-type male array comparisons . Total RNA was extracted using standard TRIzol protocols and treated with DNaseI ( Ambion ) and MgCl2 . cDNA was synthesized following the SuperScript II Reverse Transcriptase protocol using Random Primers ( Invitrogen ) . Real-time PCR was performed using a DNA Engine Opticon 2 detection system ( BioRad ) , utilizing AmpliTaq Gold PCR Master Mix ( Applied Biosystems ) and SYBR Green ( Applied Biosystems ) fluorescence . The following primer pairs were used: cpn ( 5′-TGGAGCGACAGCCACTTCTG and 5′-GCAGACGTTGCTCCACCTGA ) , dpr ( 5′-CGCCAATTGGACACTGCAAA and 5′-GCTCGTGGGGTCCTTGCATA ) , capa ( 5′- CCACTGGCTTTCTTTTGGAA and 5′-AGTCTGCGCGACGGATTAG ) , and rp49 ( 5′- GCCAAACTGATGCTAGGC and 5′-CCACCTCCACTTCAGGATAC ) . Cycling was for 10 min at 94 °C , followed by 45 cycles of 94 °C for 20 s , 60 °C for 30 s , and 72 °C for 30 s . Each of four independent samples for each genotype , consisting of ∼1 . 2 μg RNA , was assayed in three technical replicates . Expression for each gene relative to rp49 expression was calculated using Data Analysis for Real Time ( DART ) PCR [96] . t-Tests were used to determine the significance of differences between the genotypes assayed . Detection of cpn mRNA in frozen sections prepared from adults heads was performed as described by Goodwin et al . [65] , using digoxigenin-labeled anti-sense and sense cpn cRNA probes , made from cDNA clone GH08002 , available from the Berkeley Drosophila Genome Project ( BDGP ) . The bioinformatics tool DAVID [44] was used to determine significant enrichment of known functional annotations within our gene lists . DAVID calculates the statistical probability for representation of genes within a given functional category for an input list relative to the total number of genes , within the same category for a background list . The background list consisted of 13 , 614 GenBank accession numbers representing each unique transcript on our arrays . Search parameters included all available categories from two GO ontologies , biological process ( BP ) , and molecular function ( MF ) . All significantly ( p < 0 . 05 ) enriched functional groups within all levels of BP and MF were reported in the Supporting Information tables . We reported level four BP and MF for all genes without an enriched functional category . If level four BP and MF were not available , we reported level three . An additional literature search parameter was added to the DAVID analysis to achieve significant enrichment for the circadian and oxidative-stress functional categories . dpr-GAL4;UAS-nlsGFP 0–24-h adult flies were dissected in 1× phosphate buffered saline ( PBS ) , fixed in 4% paraformaldehyde , 1× PBS for 20 min and stained as described in Lee et al [64] . Here , brains and ventral nerve cords were stained using primary rat polyclonal FRUM antisera ( 1:100 ) and secondary Cy5-conjugated goat anti-rat IgG and Cy3-conjugated rabbit anti-GFP IgG ( 1:500; Molecular Probes ) . We generated the FRUM antisera ( Josman Laboratories ) against the 101 amino acid male-specific region , as described previously [64] . Adult peripheral tissues ( forelegs , proboscis , antennae , and external genitalia ) in 1× PBS were fixed , washed twice in 10 mM Tris-Cl , 150 mM NaCl , 0 . 05% Tween-20 buffer ( TNT ) , and mounted in Vectashield mounting media ( Vector Labs ) . Optical section stacks were obtained using a Zeiss LSM 5 Pascal confocal microscope and processed using Adobe Photoshop . Adult male flies were reared individually in vials and aged 4–6 d in a 12-h light: 12-h dark cycle . For courtship assays , a single male fly of each genotype was paired with a 4–6-d-old CS virgin female in a 10-mm diameter chamber and video recorded for 10 min . To account for any temperature or circadian effects , all courtship assays were performed at 22 °C , ∼60% humidity , and at ZT 5–9 . Assays for control and experimental flies were performed on the same day . Courtship latency was considered the time until first wing extension . All courtship recordings were analyzed using Noldus software ( Wageningen , Netherlands ) . Data were averaged , and standard error ( SE ) was determined for each genotype . Whitney-Mann ( nonparametric ) rank-sum tests were calculated to determine the significance of differences . Three individual vials with food and containing 20 male flies of each genotype were assayed . Flies were entrained in a 12-h light: 12-h dark cycle for at least 4 d . Line crossings were counted in 30-min bins over a 48-h period using the Drosophila Activity Monitoring System ( Trikinetics ) in the same light:dark cycle as entrainment . Data were averaged , and SE was determined for each genotype .
|
The fruit fly Drosophila is an excellent model system to use to understand the molecular-genetic basis of male courtship behavior , as the potential for this behavior is specified by a well-understood genetic regulatory hierarchy , called the sex determination hierarchy . The sex hierarchy consists of a pre-mRNA splicing cascade that culminates in the production of sex-specific transcription factors , encoded by doublesex ( dsx ) and fruitless ( fru ) . dsx specifies all the anatomical differences between the sexes , and fru is required for all aspects of male courtship behavior . In this study , we measure gene expression differences between males and females , and between sex hierarchy mutants and wild-type animals , to identify genes that underlie the differences between males and females . We have performed these studies on adult head and nervous system tissues , as these tissues are important for establishing the potential for behaviors . We have identified several genes regulated downstream of dsx and fru and more extensively characterized two genes that are more highly expressed in males . One gene regulated downstream of dsx is expressed in the retina and is known to have a function in visual transduction . The other gene , regulated downstream of fru , plays a role in the timing of male courtship behavior .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] |
[
"neuroscience",
"drosophila",
"genetics",
"and",
"genomics"
] |
2007
|
Genomic and Functional Studies of Drosophila Sex Hierarchy Regulated Gene Expression in Adult Head and Nervous System Tissues
|
Salmonella enterica serovar Typhi is transmitted by fecally contaminated food and water and causes approximately 22 million typhoid fever infections worldwide each year . Most cases occur in developing countries , where approximately 4% of patients develop intestinal perforation ( IP ) . In Kasese District , Uganda , a typhoid fever outbreak notable for a high IP rate began in 2008 . We report that this outbreak continued through 2011 , when it spread to the neighboring district of Bundibugyo . A suspected typhoid fever case was defined as IP or symptoms of fever , abdominal pain , and ≥1 of the following: gastrointestinal disruptions , body weakness , joint pain , headache , clinically suspected IP , or non-responsiveness to antimalarial medications . Cases were identified retrospectively via medical record reviews and prospectively through laboratory-enhanced case finding . Among Kasese residents , 709 cases were identified from August 1 , 2009–December 31 , 2011; of these , 149 were identified during the prospective period beginning November 1 , 2011 . Among Bundibugyo residents , 333 cases were identified from January 1–December 31 , 2011 , including 128 cases identified during the prospective period beginning October 28 , 2011 . IP was reported for 507 ( 82% ) and 59 ( 20% ) of Kasese and Bundibugyo cases , respectively . Blood and stool cultures performed for 154 patients during the prospective period yielded isolates from 24 ( 16% ) patients . Three pulsed-field gel electrophoresis pattern combinations , including one observed in a Kasese isolate in 2009 , were shared among Kasese and Bundibugyo isolates . Antimicrobial susceptibility was assessed for 18 isolates; among these 15 ( 83% ) were multidrug-resistant ( MDR ) , compared to 5% of 2009 isolates . Molecular and epidemiological evidence suggest that during a prolonged outbreak , typhoid spread from Kasese to Bundibugyo . MDR strains became prevalent . Lasting interventions , such as typhoid vaccination and improvements in drinking water infrastructure , should be considered to minimize the risk of prolonged outbreaks in the future .
Salmonella enterica serovar Typhi ( Salmonella Typhi ) is the Gram-negative bacillus that causes typhoid fever , a systemic infection transmitted through food and water contaminated with human feces . Typhoid fever is characterized by numerous non-specific symptoms , including high fever , headache , malaise , joint pain , abdominal pain , and gastrointestinal symptoms such as nausea , vomiting , constipation , and diarrhea . The case fatality rate is less than 1% with prompt and effective antimicrobial treatment , but may reach 41% in developing countries where access to care is limited [1] . The most serious complication , intestinal perforation , occurs in approximately 3 . 8% of patients in the developing world; in these areas , reported rates of intestinal perforation range from 0 . 1–39% [2] . Intestinal perforation has been associated with male gender , older age , delayed or inappropriate antimicrobial therapy , and short duration of symptoms [1] , [3] , [4] . Typhoid is endemic in many countries with poor sanitation and hygiene and limited access to safe water . Although well-studied in South and Southeast Asia , where it is widely endemic , typhoid fever has only recently been recognized as a significant contributor to the burden of febrile illness in sub-Saharan Africa . Since 2008 , severe typhoid fever outbreaks have been reported in rural Malawi [5] and Uganda [6] , and in the capital cities of Zimbabwe [7] and Zambia [8] . Typhoid was recently shown to be highly endemic in an urban population in Kenya; the typhoid incidence of 247 cases per 100 , 000 persons in this area was similar to that found in highly endemic areas in Southeast Asia [9] . Typhoid diagnosis in malaria endemic areas of sub-Saharan Africa is challenged by the similar clinical presentations of typhoid and malaria and the limited availability of laboratory resources in many countries . Blood culture is time and resource intensive , and rapid diagnostic tests , such as TUBEX-TF , are useful for preliminary assessment of potential outbreaks but do not have adequate sensitivity and specificity for diagnosis of individual patients [10]–[12] . Limited data from Africa suggests that antimicrobial resistance is increasing among Salmonella Typhi isolates , including a rise in the prevalence of multidrug resistance ( MDR ) , defined as resistance to the three traditional first-line antimicrobials , ampicillin , chloramphenicol , and trimethoprim-sulfamethoxazole [13] , and emergence of decreased susceptibility to ciprofloxacin [5] , [9] , [14] . Kasese and Bundibugyo are neighboring rural agricultural districts in western Uganda that border the Democratic Republic of the Congo ( Figure 1 ) . Epidemics of waterborne and foodborne diarrheal diseases , such as cholera [15] and typhoid fever [6] , have historically plagued both districts . Malaria is endemic . Water treatment is largely an individual responsibility , as coverage with improved water sources is low , and neither district has municipal water systems that deliver chlorinated water . An outbreak of typhoid fever with a high rate of intestinal perforation began in 2008 in Kasese district [6] . Among 21 Salmonella Typhi isolates obtained through surveillance conducted from March 4 to July 31 , 2009 , only 1 ( 5% ) was multidrug-resistant and no resistance to nalidixic acid or ciprofloxacin was detected [6] . We report the continuation of the outbreak in Kasese district , present evidence for its extension into Bundibugyo , a neighboring district , and describe a dramatic increase in the prevalence and extent of antimicrobial resistance since 2009 . In August 2011 , an outbreak of undiagnosed febrile illness , which we later confirmed as typhoid fever , began in the neighboring district of Bundibugyo , and a high number of patients with intestinal perforation were noted in Kasese . In response to a request to investigate from the Uganda Ministry of Health and in collaboration with the Uganda Ministry of Health and Kasese and Bundibugyo District Health Offices , we implemented laboratory-enhanced prospective case finding , conducted retrospective case finding for typhoid fever via medical record reviews , and tested drinking water sources in both districts to determine the scope and likely vehicles of the outbreaks . Outbreak strains were characterized by molecular subtyping and antimicrobial susceptibility testing .
A suspected case of typhoid fever was defined as surgically diagnosed intestinal perforation consistent with Salmonella Typhi infection or illness characterized by fever and abdominal pain for ≥1 day and at least one of the following signs or symptoms — gastrointestinal disruptions , such as vomiting , diarrhea , or constipation , general body weakness , joint pain , headache , clinical suspicion for intestinal perforation , or failure to respond to antimalarial medications — with onset from January 1 to December 31 , 2011 in Bundibugyo residents and from August 1 , 2009 to December 31 , 2011 in residents of other districts , including Kasese . The source populations for identifying cases of typhoid fever in Kasese and Bundibugyo districts were persons seeking care at government-affiliated and private not-for-profit health facilities in these districts and at Fort Portal Regional Referral Hospital in Kabarole district , which borders Kasese and Bundibugyo districts . Cases of surgically-diagnosed intestinal perforation were identified retrospectively from operating room logbooks and chart abstractions of intestinal perforation patients at Kilembe Mines Hospital , Kagando Hospital , Bwera Hospital , St . Paul's Health Center IV , Bundibugyo Hospital , and Fort Portal Regional Referral Hospital . Additional cases were identified retrospectively through linelists of patients with surgically-diagnosed intestinal perforation maintained by Kasese hospitals , and linelists of suspected typhoid fever cases maintained since August 2011 by the District Health Office in Bundibugyo . Beginning in October 2011 , typhoid cases were identified prospectively through patient and caregiver interviews in health care facilities and highly affected communities , and through laboratory-enhanced case finding . Systematic , laboratory-enhanced prospective case finding was conducted in Kasese from November 1 to December 31 , 2011 and in Bundibugyo from October 28 to December 31 , 2011 . Eighty-two health facilities in Kasese and 21 in Bundibugyo were provided with case report forms eliciting information about clinical history , potential risk factors , and socioeconomic status . Health facilities with the capacity to collect specimens in Kasese ( 15 ) and Bundibugyo ( 3 ) were asked to collect blood and stool from all patients with suspected intestinal perforation and from the first two ( Kasese ) or three ( Bundibugyo ) patients who presented at the facility and met the case definition each day . Blood , serum , and stool were collected according to the above criteria at each facility daily from October 28 to December 31 , 2011; prior to this , from October 18 to October 28 , 2011 , specimens were collected from a sample of clinically-suspected typhoid patients . Supplies for specimen collection and testing were provided by the US Centers for Disease Control and Prevention ( CDC ) . CDC , the Kenya Medical Research Institute ( KEMRI ) , and the Uganda Central Public Health Laboratory ( CPHL ) trained local laboratory technicians in microbiologic and serologic techniques for typhoid fever diagnosis at one hospital and one upper level health facility in Kasese and at one hospital in Bundibugyo . Blood ( persons 10 years of age and older: 10 ml; children <10 years: 1 ml of blood per year of age ) was collected using standard methods . Eight ml of blood ( patients ≥10 years ) or one-half the sample volume ( children <10 years ) were inoculated into an Oxoid signal blood culture bottle; the remaining blood was placed in a serum separator tube . Stool was collected according to standard guidelines and inoculated into Cary-Blair transport medium [16] . All samples were held at ambient temperature and transported to district referral laboratories within 72 hours of collection . Blood and stool cultures were performed per standard protocols for isolation of Salmonella Typhi [17] , [18] . Isolates that were biochemically or serologically typical of Salmonella Typhi were forwarded to CPHL for confirmation and to CDC-Atlanta for further characterization , including serotyping , pulsed-field gel electrophoresis ( PFGE ) , and antimicrobial susceptibility testing ( AST ) . PFGE , which is the current gold standard subtyping technique for Salmonella , was conducted per standard protocols [19] and PFGE patterns were analyzed using BioNumerics software version 5 . 01 ( Applied Maths , Inc . , Austin , TX , USA ) . AST was performed using disk diffusion [20] and broth microdilution ( Sensititre; Trek Diagnostics ) according to the manufacturer's instructions . Where applicable , Clinical and Laboratory Standards Institute ( CLSI ) 2012 interpretive criteria were used to categorize antimicrobial susceptibility results [20] . For drugs that lack CLSI interpretive criteria , results were classified using interpretive criteria from the CDC National Antimicrobial Resistance Monitoring System [21] . Serological testing was performed with TUBEX- TF ( IDL Biotech ) per product insert . Water samples collected from drinking water sources in Kasese and Bundibugyo were tested for the presence of total coliforms and Escherichia coli in 100 ml , 1 ml , or 100 µl of water using the Colilert-18 test kit ( IDEXX , Westbrook , ME ) . One-ml samples and 100-µl samples were diluted in 100 ml of sterile , distilled water before addition of the Colilert reagent . In Kasese , two surface water sources and six drinking water taps were sampled . In Bundibugyo , river water upstream of a gravity flow scheme ( GFS ) intake , the GFS outflow tank , two GFS taps , and one tap from a town municipal water supply were sampled . Data were entered into electronic databases and analyzed using SAS 9 . 3 ( SAS Institute , Cary , NC ) . Statistical testing was done using the Fisher's exact test for categorical data and the Wilcoxon rank-sum test for continuous data . For analysis of prospectively-identified cases , non-yes responses for symptoms , animal ownership , and assets were structurally missing on most forms and were imputed as negative responses . P values<0 . 05 were considered significant . The primary purpose of this activity was to identify , characterize , and control disease in response to an immediate public health threat . As such , the human subjects research designee in the Division of Foodborne , Waterborne , and Environmental Diseases at CDC determined that the activities constituted public health response rather than research . Patients with suspected typhoid fever were offered diagnostic testing through routine culture of stool and blood specimens and through serologic testing as part of standard clinical care , and informed consent specifically for this testing was not obtained .
During the period August 1 , 2009 to January 6 , 2012 , 1 , 341 suspected typhoid fever cases were identified , including 1 , 049 ( 78% ) identified retrospectively and 292 ( 22% ) identified prospectively . Among 1 , 165 patients for whom district of residence was reported , 709 ( 61% ) resided in Kasese , 333 ( 29% ) resided in Bundibugyo , and 83 ( 7% ) resided in Kabarole district . Thirty-seven were from other districts and three were residents of the Democratic Republic of Congo . Among Kasese patients , more cases with intestinal perforation were recorded in 2011 compared to previous years ( Figure 2A ) ; during the period of laboratory-enhanced case-finding ( November to December 2011 ) , cases with intestinal perforation represented a small fraction of all cases identified . In Bundibugyo , a sharp increase in cases of typhoid fever with and without intestinal perforation was observed beginning in August 2011 ( Figure 2B ) ; the outbreak was reported to the Uganda Ministry of Health later that month . Intestinal perforation status was recorded or imputed for 697 ( 98% ) of 709 Kasese patients and 293 ( 88% ) of 333 Bundibugyo patients; the frequency of intestinal perforation was 82% and 20% , respectively ( Table 1 ) . In Kasese , but not Bundibugyo , all sources of retrospective case finding recorded only patients with intestinal perforation instead of all patients with suspected typhoid . Males were disproportionately affected by intestinal perforation in both districts , accounting for 59% and 66% of Kasese and Bundibugyo patients with intestinal perforation , and only 46% and 40% of patients without intestinal perforation , respectively . The median age of Bundibugyo patients was 13 years ( range: <1–68 years ) . Among Bundibugyo patients with intestinal perforation , the median age of males , 15 years , was older than that of females , who had a median age of 10 years ( P = 0 . 03 ) . Female patients with intestinal perforation were younger than females without intestinal perforation , who had a median age of 18 years ( P = 0 . 008 ) . Among Bundibugyo patients , the proportion with intestinal perforation was significantly higher for males than females for patients aged 20–29 years ( 40% vs . 0% , P = 0 . 0006 ) and 30–39 years ( 42% vs . 0% , P = 0 . 03 ) ( Figure 3 ) . Similar associations between age , gender and intestinal perforation status were also identified in Kasese patients . Kasese patients resided in all 21 sub-counties in the district . During the 29-month case finding period , there were 98 cases of clinically-diagnosed typhoid fever per 100 , 000 persons district-wide; in 2011 alone , there were 58 cases per 100 , 000 persons . The Bukonzo West health subdistrict , comprising the western sub-counties of Bwera , Ihandiro , Isango , Karambi , and Mpondwe-Lhubiriha Town Council , had the highest typhoid incidence from August 2009 to December 2011 , 115 cases per 100 , 000 persons . The incidence of typhoid fever was 139 per 100 , 000 in Bundibugyo District for the period January 1 , 2011 to December 31 , 2011 . During this period , four neighboring Bundibugyo sub-counties had an incidence of typhoid fever greater than 100 cases per 100 , 000 persons: Kirumya ( 990 per 100 , 000 ) , Bubukwanga ( 234 per 100 , 000 ) , Bukonzo ( 199 per 100 , 000 ) , and Bundibugyo Town Council ( 155 per 100 , 000 ) . Case report forms for 149 Kasese residents , 128 Bundibugyo residents , and 13 residents of other districts or for whom district of residence could not be determined were completed at health facilities from October 11 , 2011 to January 6 , 2012 . The most common symptoms of the 277 case-patients from Kasese and Bundibugyo , other than fever and abdominal pain ( both required by the case definition ) , were weakness ( 84% ) , headache ( 82% ) , and joint pain ( 71% ) ( Table 2 ) . Six percent of patients had intestinal perforation . Before the visit where the case report form was completed , patients reported feeling ill for a median of 7 days ( range 1–240 days , n = 267 ) , including having fever for a median of 6 days ( range 1–240 days , n = 268 ) , and abdominal pain for a median of 4 days ( range 1–730 days , n = 262 ) . Seventy percent of respondents reported seeking care for their illness before the health care visit during which the case report form was completed ( Table 3 ) . Patients most frequently sought care at a drug shop or pharmacy ( 49% ) or a health center or hospital ( 48% ) ; a small fraction consulted an herbalist ( 7% ) or traditional healer ( 1% ) . Antibiotics were taken by 45% of patients . Metronidazole , which is not effective against Salmonella Typhi , was the antimicrobial most frequently reported and was used by 36% of patients who reported taking antimicrobials . Ciprofloxacin , co-trimoxazole , and amoxicillin were taken by 27% , 26% , and 24% of patients who reported taking antimicrobials , respectively; only 5% reported taking chloramphenicol . Nearly one-quarter ( 23% ) of patients who reported taking antimicrobials had taken ≥2 agents to treat the same illness episode . In Kasese , tap water was the most commonly reported primary source of drinking water during the month before illness onset and was reported by 84 ( 62% ) of 136 patients . Other primary drinking water sources were stream or river water , spring water , and wells , used by 22 ( 16% ) , 14 ( 10% ) , and 2 ( 1% ) patients , respectively . In Bundibugyo , tap water was also the most common primary source of drinking water and was reported by 56 ( 46% ) of 122 patients . Forty-three ( 35% ) patients used stream or river water , 11 ( 9% ) used spring water , 5 ( 4% ) used well water , and 2 ( 2% ) used bottled water . Among the 50 Bundibugyo patients who reported tap water as their primary drinking water source and also reported their sub-county of residence , 24 ( 48% ) resided in subcounties where tap water was provided by a single GFS , the Kirumya-Bubukwanga GFS , and 8 ( 16% ) resided in Bundibugyo Town Council , the district's largest town . Among the 35 Bundibugyo patients who used river water as their primary drinking water source and reported the name of the river they used , 15 ( 43% ) used the Kirumya River , which is the source of the Kirumya-Bubukwanga GFS . Clustering by drinking water source was not observed among Kasese patients . Among 208 patients , 30 ( 14% ) reported treating their water in the month before they became ill . Among those who treated their water , 20 ( 67% ) boiled water , 7 ( 23% ) used a chlorine product , 1 ( 3% ) used PuR , a chlorination-flocculation product , and 2 ( 0 . 7% ) used a treatment method not listed . Significant differences were observed in the symptoms , clinical histories , and socioeconomic status of the 18 patients with confirmed or suspected intestinal perforation and the 250 patients without intestinal perforation who were identified through prospective laboratory-enhanced case-finding ( Table 4 ) . Patients with intestinal perforation were more likely than those without intestinal perforation to have sought health care for the same illness episode before the visit when the enrollment form was completed ( 100% vs . 69% , P = 0 . 004 ) and to report that care was sought at a health center or hospital ( 88% vs . 46% , P = 0 . 0008 ) . Patients with intestinal perforation reported not responding to antimalarials more frequently than patients without intestinal perforation ( 71% vs . 31% , respectively; P = 0 . 003 ) , although reported antimalarial use did not differ by intestinal perforation status . Patients with intestinal perforation were more likely than those without intestinal perforation to report taking antibiotics ( 75% vs . 44% , respectively , P = 0 . 02 ) , and among patients who took antibiotics , were more likely to report taking chloramphenicol ( 25% vs . 1% , respectively; P = 0 . 003 ) . Compared to patients without intestinal perforation , more than twice the proportion of patients with intestinal perforation reported taking ≥2 antibiotics to treat the same illness episode ( 58% vs . 24%; P = 0 . 04 ) . Patients with intestinal perforation were less likely to own two or more listed household items ( radio , mobile telephone , foam mattress , bicycle , and motorcycle ) than patients without intestinal perforation ( 56% vs . 80% , respectively; P = 0 . 03 ) ; employment and number of animals owned did not vary by intestinal perforation status . Salmonella Typhi was isolated from seven ( 9% ) of 74 blood cultures and one ( 2% ) of 47 stool cultures from Kasese patients , and from 15 ( 21% ) of 72 blood cultures and one ( 10% ) of 10 stool cultures from Bundibugyo patients . In total , eight ( 11% ) of 75 Kasese patients and 16 ( 20% ) of 79 Bundibugyo patients tested were culture-confirmed . Two ( 14% ) of 14 patients with intestinal perforation who had a blood culture performed had Salmonella Typhi isolated . Salmonella Typhi isolates from five of the eight culture-confirmed Kasese cases and 13 of the 16 culture-confirmed Bundibugyo cases were further characterized at the U . S . Centers for Disease Control and Prevention in Atlanta . PFGE subtyping revealed four distinct XbaI/BlnI pattern combinations among the five Kasese isolates and six distinct XbaI/BlnI pattern combinations among the 13 Bundibugyo isolates . When compared to the PulseNet Global Salmonella Typhi database , a database of globally distributed Typhi isolates , these pattern combinations were unique with the exception of a single pattern combination , pattern combination A , which was observed in the single chloramphenicol resistant Kasese isolate from 2009 ( Figure 4 ) . Among the 2011 isolates , pattern combination A was the pattern combination most frequently observed and was shared by two Kasese isolates and six Bundibugyo isolates , all of which were chloramphenicol resistant . Two novel pattern combinations , designated B and C , were also shared by isolates from both districts . Eighteen isolates were tested for susceptibility to a panel of antimicrobials that included amoxicillin/clavulanic acid , ampicillin , ceftriaxone , chloramphenicol , ciprofloxacin , nalidixic acid , streptomycin , sulfisoxazole , tetracycline , and trimethoprim-sulfamethoxazole . Two isolates , both from Bundibugyo patients , were pan-susceptible . Fifteen ( 83% ) isolates were resistant to ampicillin , chloramphenicol , and trimethoprim-sulfamethoxazole ( MDR ) , and were also resistant to sulfisoxazole , streptomycin , and tetracycline . A single isolate , from Kasese , was resistant to nalidixic acid and showed intermediate susceptibility to ciprofloxacin; it was fully susceptible to other antimicrobials tested . TUBEX-TF serological testing was positive for 23 ( 35% ) of 65 Kasese patients and 13 ( 45% ) of 29 Bundibugyo patients . Among the 59 patients who had both a blood culture and a TUBEX-TF serological test performed , six ( 10% ) had Salmonella Typhi isolated from blood or stool and tested positive by TUBEX-TF , four ( 7% ) had a blood culture positive for Salmonella Typhi but a negative TUBEX-TF test , 15 ( 25% ) had blood culture negative for Salmonella Typhi but had a positive TUBEX-TF test , and 34 ( 58% ) were blood culture and TUBEX-TF negative . Thirteen ( 7% ) of 182 Kasese and Bundibugyo patients tested were positive for malaria by RDT or blood smear; one ( <1% ) of 124 patients with blood or stool culture and malaria diagnostic results reported was positive for both typhoid and malaria . Water samples were collected from six drinking water taps and two surface water sources in Kasese and three drinking water taps and one surface water source in Bundibugyo . Total coliforms were present in all samples . E . coli were detected in 100 mL of water from four ( 67% ) of six Kasese drinking water taps ( corresponding to a concentration of ≥1 cfu/100 mL ) and in 100 µl of water from the rivers Rwimi and Hima ( corresponding to a concentration of ≥1000 cfu/100 mL ) . In Bundibugyo , E . coli were present in 100 ml of water from the Kirumya river upstream of the Kirumya-Bubukwanga GFS intake , from the GFS outflow tank , and from two taps on the GFS and one tap in the Bundibugyo Town Council municipal water supply . E . coli were also present in 1 ml of water from the two taps on the GFS ( corresponding to a concentration of ≥100 cfu/100 mL ) ; the Bundibugyo Town Council municipal water supply and Kasese tap water were not tested at volumes below 100 mL .
A large and severe typhoid fever outbreak in rural western Uganda persisted from 2008 through 2011 , spread to a neighboring district , and became more refractory to antimicrobial treatment . In 2009 , an investigation suggested that contaminated drinking water was the most likely vehicle of infection , and general prevention measures such as hand washing , improved sanitation , and promotion of household water treatment were recommended [6] . Absent a sustained and widespread intervention campaign , a resurgence of cases with intestinal perforation was investigated in 2011 . Molecular subtyping and epidemiologic evidence from the 2011 investigation indicate that the typhoid outbreak persisted in Kasese and spread to the neighboring district of Bundibugyo . Compared to Salmonella Typhi isolated from Kasese patients over a six-week period in 2009 , of which only one isolate ( 1/21; 5% ) was multidrug resistant [6] , isolates obtained from Kasese and Bundibugyo patients over the three-month period October to December 2011 were more likely to be multidrug resistant . Additionally , an isolate with reduced susceptibility to ciprofloxacin , the current recommended first-line treatment for uncomplicated typhoid , was identified for the first time among outbreak strains . Across the 2009 and 2011 enhanced case finding periods , the frequency of co-trimoxazole and chloramphenicol use were similar ( 29% vs . 26% and 9% vs . 5% , respectively ) , indicating that changes in antibiotic use do not explain the increased frequency of MDR isolates in 2011 . These findings demonstrate that the ramifications of severe , uncontrolled typhoid outbreaks include outbreak strains that become increasingly resistant to lifesaving antibiotics and the spread of disease to neighboring areas . Selective recognition and documentation of patients with intestinal perforation , the most severe complication of typhoid fever , led to an underestimation of the magnitude of the outbreak and an overestimation of the proportion of reported cases with intestinal perforation . Although the overall proportion of cases with intestinal perforation was 82% in Kasese and 20% in Bundibugyo , prospective case finding in district health facilities showed that patients with intestinal perforation represented only 8% of Kasese patients and 3% of Bundibugyo patients , or only 6% of all typhoid cases . Inflation of the intestinal perforation rate as an artifact of retrospective case finding methods was more pronounced in Kasese , where linelists recorded only patients with intestinal perforation , compared with Bundibugyo , where all suspected typhoid cases were included on the linelist . Extrapolating from the 570 Kasese intestinal perforation cases identified from August 1 , 2009 to December 31 , 2011 and the 8% intestinal perforation rate observed through prospective case-finding , we estimate that 7 , 125 cases of typhoid occurred among Kasese residents during this period , giving an estimated annual incidence of 409 cases per 100 , 000 persons . In Bundibugyo , where 59 cases with intestinal perforation were reported from January 1 to December 31 , 2011 and the intestinal perforation rate was 3% , we estimate that there were 1 , 967 typhoid fever cases and an annual incidence of 820 cases per 100 , 000 persons . Although based on a different case definition and different case-finding method , these incidences exceed rates observed in African urban slums of 247 per 100 , 000 [9] , and indicate that intense , sustained typhoid transmission occurs in rural areas of sub-Saharan Africa . Males were disproportionately affected by intestinal perforation; in Bundibugyo , and to a lesser extent in Kasese , this was more pronounced among adults . In both districts , females with intestinal perforation were younger than those without intestinal perforation , and the opposite was observed for male patients . The higher frequency of intestinal perforation in males compared to females has been well-documented in several case-series in Africa [4] , [22] , [23] , Asia [1] , [24] , and the Caribbean [25] , and male sex was identified as a risk factor for intestinal perforation among hospitalized typhoid patients in Turkey [3] . The reasons for this often observed association remain unknown . We found fewer published observations of the influence of age on the association between intestinal perforation status and sex . A single study in South Africa found that typhoid clinical features varied by sex among adults but not children; however , in this study no cases of intestinal perforation occurred among female patients of any age [4] . The differences observed in our study may reflect age and gender-specific care seeking behaviors or treatment adherence . Alternatively , the relatively low rates of intestinal perforation observed in women beyond the age of puberty may indicate that sex hormones play a role in disease pathogenesis; in mouse models of typhoid fever , estrogen decreased the intensity of infection [26] . Certain clinical factors , such as multiple health care visits and taking two or more antibiotics , were associated with intestinal perforation , suggesting that initial treatments were not effective . Inadequate treatment of typhoid was previously described as a risk factor for intestinal perforation among hospitalized typhoid patients [3] . Chloramphenicol was the only antibiotic specifically associated with intestinal perforation , and this may be related to widespread chloramphenicol resistance among outbreak strains . Loss of chloramphenicol susceptibility was previously associated with a high rate of intestinal perforation during a typhoid fever outbreak in Kinshasa , Democratic Republic of Congo , where chloramphenicol was the drug of choice for empiric treatment of typhoid [27] . Unlike in Kinshasa , expansion of chloramphenicol resistance was not associated with a detectable increase in the intestinal perforation rate; this may be because chloramphenicol use was rare among Uganda typhoid patients . For the first time in this epidemic , MDR Salmonella Typhi isolates predominated among the outbreak strains and an isolate resistant to nalidixic acid and with reduced susceptibility to ciprofloxacin was identified . Recently , there have been multiple reports of widespread MDR Salmonella Typhi in East and Central Africa . In the Democratic Republic of Congo , 30% of Salmonella Typhi isolated from 2007–2011 were MDR , and 15% showed nalidixic acid resistance and decreased susceptibility to ciprofloxacin [14]; in an urban area in Kenya , 78% of Salmonella Typhi isolates were MDR and 3% were resistant to nalidixic acid [9]; and in a 2009 outbreak in Malawi , all isolates were MDR and 10% were resistant to nalidixic acid [5] . Suboptimal dosage and duration of therapy , as might occur with poor prescribing practices and poor adherence to therapy , may accelerate the development of antimicrobial resistance . We documented widespread improper antibiotic use among Kasese and Bundibugyo patients; of those who reported taking ciprofloxacin , only 18% completed the recommended 14-day course and half took ciprofloxacin for 5 days or fewer . Development of full ciprofloxacin resistance , alone or in combination with MDR , would further limit treatment options in western Uganda , as there are practical concerns about the use of the three primary alternatives to ciprofloxacin there . The best orally administered alternative to ciprofloxacin , azithromycin , is expensive and not stocked by Ministry of Health-sponsored facilities; another oral therapy , gatifloxacin , has been shown to be effective in areas with widespread nalidixic acid resistance but has been pulled from several markets due to severe side effects in adults and is not licensed in Uganda; the third possibility , ceftriaxone , is stocked at Ugandan hospitals and some health centers but must be administered parenterally . In this resource-limited setting , emergence of ciprofloxacin resistant Salmonella Typhi in the absence of alternative oral therapies would be a devastating blow to the typhoid pharmacopeia , and would likely result in increased rates of complications , including intestinal perforation . Piped drinking water was the probable primary transmission vehicle in both districts . Although we did not isolate Salmonella Typhi from water sources , the presence of E . coli , a marker of fecal contamination , and the epidemiologic evidence are consistent with a waterborne source of typhoid infection . Despite an ongoing typhoid fever outbreak and recommendations to implement safe water interventions , the percentage of patients who reported treating water in the month before they became ill decreased from 22% in 2009 to 14% in 2011; in both districts the most common reason for not treating water was the belief that treatment was not necessary . This is consistent with the limited uptake of point-of-use water treatment observed among at-risk populations in other parts of the world [28] , [29] . Furthermore , uptake in Kasese in 2011 may have been particularly low because the outbreak had passed the acute emergency phase , in which supplies were provided free of charge and treatment could be viewed as a temporary measure . The enduring nature of this typhoid outbreak and emergence of increasingly drug-resistant strains indicate a need for alternative interventions . Providing treated water through the many piped water systems that exist in both districts , and expanding these systems to new areas would provide long-term reductions in risk of typhoid and other waterborne diseases . Typhoid vaccination should be strongly considered as a medium-term intervention that controls the typhoid outbreak while water infrastructure improvements are made . These findings are subject to several limitations . The absence of systematic typhoid surveillance before implementation of laboratory-enhanced case-finding made it difficult to ascertain the true scope of the outbreak . During enhanced case-finding , cases were missed , both because the system did not capture patients who did not seek care at government or private not-for-profit health facilities and because some facilities chose not to participate in case-finding . Missed cases may have differed from reported cases systematically by sub-county of residence , socioeconomic status , clinical history , and other factors . Typhoid fever is a diagnostic challenge , particularly in malaria endemic areas where a large proportion of fevers are attributed to malaria , and we attempted to balance the sensitivity and specificity of the case definition with this in mind . We included abdominal pain to increase specificity; however , by improving specificity , we likely lost sensitivity to detect typhoid patients early in the clinical course , before abdominal pain develops . This may have impacted blood culture positivity , since positivity is highest in the week following symptom onset [30] . The high number of patients taking amoxicillin , ciprofloxacin , and co-trimoxazole likely biased organisms recovered to those resistant to these therapies . The imputation of missing data as negative responses for clinical signs and symptoms and socioeconomic factors could have biased results towards underestimation of frequencies . For example , had missing data been censured , the calculated intestinal perforation rates would have been higher , 12% and 5% in Kasese and Bundibugyo , respectively . Additionally , due to the low positivity of blood cultures , it is possible that many patients with non-typhoidal febrile illness were included in the case definition , and the true proportion with IP may have been higher . Accurate identification of sub-counties of residence was challenged by missing data , multiple spellings for the same location , and reorganization of several western Kasese sub-counties in 2011 . Comprehensive , laboratory-enhanced surveillance for typhoid fever is necessary in outbreak settings to characterize affected populations , monitor disease trends , target interventions , and assess their impact . Reliance on severe complications , such as intestinal perforation , as surrogates for all cases can be misleading when complication rates vary over time due to changes in treatment practices or antimicrobial resistance . Given the many reports of MDR and nalidixic acid resistant Salmonella Typhi in sub-Saharan Africa , improving laboratory diagnostic capacity is crucial for making appropriate , timely treatment recommendations . In areas of high incidence where conventional approaches to safe water have not stopped transmission and drug resistant strains are circulating , novel approaches , such as typhoid vaccination , should be considered .
|
Typhoid fever is an acute febrile illness caused by the bacteria Salmonella Typhi and transmitted through food and water contaminated with the feces of typhoid fever patients or carriers . We investigated typhoid fever outbreaks in two neighboring Ugandan districts , Kasese and Bundibugyo , where typhoid fever outbreaks began in 2008 and 2011 , respectively . In Kasese from August 2009–December 2011 , we documented 709 cases of typhoid fever . In Bundibugyo from January–December 2011 , we documented 333 cases . Salmonella Typhi from Bundibugyo and Kasese had indistinguishable molecular fingerprints; laboratory and epidemiological evidence indicate that the outbreak spread from Kasese to Bundibugyo . Salmonella Typhi isolated during our investigation were resistant to more antibiotics than isolates obtained from Kasese in 2009 . Drinking water in both districts was fecally contaminated and the likely vehicle for the outbreaks . Our investigation highlights that in unchecked typhoid fever outbreaks , illness can become geographically dispersed and outbreak strains can become increasingly resistant to antibiotics . Lasting interventions , including investments in drinking water infrastructure and typhoid vaccination , are needed to control these outbreaks and prevent future outbreaks .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"infectious",
"diseases",
"global",
"health",
"epidemiology",
"biology",
"microbiology",
"public",
"health"
] |
2014
|
Shifts in Geographic Distribution and Antimicrobial Resistance during a Prolonged Typhoid Fever Outbreak — Bundibugyo and Kasese Districts, Uganda, 2009–2011
|
Epistatic interactions between residues determine a protein’s adaptability and shape its evolutionary trajectory . When a protein experiences a changed environment , it is under strong selection to find a peak in the new fitness landscape . It has been shown that strong selection increases epistatic interactions as well as the ruggedness of the fitness landscape , but little is known about how the epistatic interactions change under selection in the long-term evolution of a protein . Here we analyze the evolution of epistasis in the protease of the human immunodeficiency virus type 1 ( HIV-1 ) using protease sequences collected for almost a decade from both treated and untreated patients , to understand how epistasis changes and how those changes impact the long-term evolvability of a protein . We use an information-theoretic proxy for epistasis that quantifies the co-variation between sites , and show that positive information is a necessary ( but not sufficient ) condition that detects epistasis in most cases . We analyze the “fossils” of the evolutionary trajectories of the protein contained in the sequence data , and show that epistasis continues to enrich under strong selection , but not for proteins whose environment is unchanged . The increase in epistasis compensates for the information loss due to sequence variability brought about by treatment , and facilitates adaptation in the increasingly rugged fitness landscape of treatment . While epistasis is thought to enhance evolvability via valley-crossing early-on in adaptation , it can hinder adaptation later when the landscape has turned rugged . However , we find no evidence that the HIV-1 protease has reached its potential for evolution after 9 years of adapting to a drug environment that itself is constantly changing . We suggest that the mechanism of encoding new information into pairwise interactions is central to protein evolution not just in HIV-1 protease , but for any protein adapting to a changing environment .
The interactions between the residues within a single protein ( within-protein epistasis [1] ) often determine the structure and function of the protein [2–4] . Understanding these epistatic interactions is important because they shape the protein fitness landscape and thus guide the evolution of a protein given its genetic background [5 , 6] . At the same time , the environment influences the fitness effects of mutations and their epistatic interactions , and a change in the environment changes the topography of the fitness landscape as well as the epistatic effect of mutations [7] . A typical example of a changing environment for a protein is the introduction of drugs to counteract a pathogen . In that case , a drug-resistance mutation that is beneficial in a drug environment might have a significant fitness cost in the absence of drugs [8 , 9] . Even in the drug environment , resistance mutations can incur fitness costs that are mitigated by compensatory mutations , and these fitness-restoring mutations are selected together with the resistance mutations if the fitness of the resulting protein exceeds that of the wild-type in the drug-environment [10] . Other compensatory mutations act by re-stabilizing the protein and preventing misfolding or proteolysis [11–13] , while sometimes resistance mutations partly compensate each other’s deleterious effects [14] . Moreover , compensatory mutations can appear before resistance mutations if they are not deleterious on their own , paving the way for resistance mutations to appear without incurring significant fitness costs [15–17] . Epistatic interactions have been implicated in the evolution of drug resistance in many pathogens , including influenza , malaria , Escherichia coli , tuberculosis , and Chlamydomonas reinhardtii[14 , 18–20] . To understand the long-term evolution of a protein in a dynamic environment in which the selection pressure is persistent , evidence of epistatic interactions at multiple time points is needed as the fitness landscape changes over time with the environment . Previous investigations of the impact of epistasis on evolutionary outcomes analyzed epistatic interactions in the absence or presence of a selection pressure at a single time point , showing that epistatic interactions can facilitate adaptation [21–28] . In particular , computational analyses of HIV fitness landscapes derived from statistical models showed that these landscapes have high neutrality as well as ruggedness , suggesting high potential for epistatic interactions [23] . Further , an analysis of in vitro virus fitness measurements from more than 70 , 000 patients revealed that epistasis explained more than half of the variance in fitness measurements , highlighting the importance of epistasis in the HIV-1 fitness landscape [24] . While these studies emphasize the importance of epistasis in adaptive evolution , here we analyze how epistatic interactions themselves change over time . We focus on the Human Immunodeficiency Virus-1 ( HIV-1 ) protease , a small protein necessary for the production of mature and infectious HIV-1 particles . The HIV-1 protease is a homodimer of two 99 residue chains , and it cleaves the viral polyprotein into active components that are necessary for virus maturation [29] . Because an inactive protease results in uninfectious viruses , the protease was one of the first targets of anti-retroviral drugs . However , due to its high mutation rate ( about 0 . 3 mutations per genome per generation [30] ) , HIV-1 quickly evolves resistance to those drugs . These resistance and compensatory mutations , as well as the covariation between residues in HIV-1’s protease are well studied [31–33] . The evolutionary trajectory of resistance to protease inhibitors is extremely complex . When the virus is first exposed to the changed fitness landscape containing the inhibitor , viral count drops to undetectable levels [34] , as the fitness peak that the viral population occupied in the no-drug fitness landscape is erased . However , some minority variants in the untreated viral population might still possess replicative capacity in the drug environment ( albeit much reduced ) , forming the seeds of drug resistance [35] . Indeed , most well-known resistance mutations are associated with significant fitness costs [10 , 36] , so that at the onset of resistance the viral populations are likely still significantly smaller than in untreated individuals . However , mutations that compensate for the fitness defects will readily emerge . While these compensatory mutations can either maintain or reduce viral fitness if they occur on their own , they usually restore the fitness cost of the resistance mutation to some extent , and as a consequence form an epistatic pair with the resistance mutation . Note that the residues that are coupled in the treatment landscape are unlikely to be coupled in the absence of treatment , implying that it is the adaptation to the new landscape that forces the interaction . Once compensatory mutations have restored viral fitness , viral populations return to pre-drug levels and can search for more compensating mutations [37] . Exposure to second-line drugs ( given to patients that have evolved resistance ) repeats the cycle , leading to more resistance mutations [38] , followed by compensatory mutations that are epistatically linked with them . As more and more resistance mutations accumulate , loss of thermodynamic stability becomes more and more acute , and compensatory mutations are selected that re-stabilize the protein . While stabilizing mutations are not selected against individually when they occur on their own ( compensatory mutations often have a fitness advantage if they occur in the absence of treatment [39] ) , several stabilizing mutations together can be undesirable as they decrease protein variability and therefore can impede evolvability [40 , 41] . Since epistasis and ruggedness are coupled [25 , 42 , 43] , the repeated emergence of resistance mutations and selection for compensatory mutations that interact epistatically with those resistance mutations results in a protein that finds itself in an increasingly rugged fitness landscape . We hypothesize that a strong selection pressure of treatment increases the epistatic interactions in the long-term evolution of HIV-1 protease , allowing the protein to realize its evolutionary potential . To test this hypothesis , we compare the HIV-1 protease sequences from patients treated with anti-retroviral drugs to protease sequences from untreated patients , collected over a nine year period . This wealth of publicly available data provides an unprecedented opportunity to study the adaptive evolution of a protein in an environment with persistent ( perhaps even increasing ) selection pressures due to the continued introduction of new drugs as opposed to a protein evolving in the absence of these selection pressures . Epistatic interactions between mutations are usually deduced by measuring the fitness effect of the single site mutations as well as the double-mutant . However , obtaining evidence of changes in epistasis over a long enough period of time is challenging when viral fitness has to be assayed . Here we use the mutual ( shared ) information between protein loci ( residues ) as a proxy for epistasis . Briefly , information shared between sites is a measure of covariation between the sites , so that knowing the residue at one site makes it possible to predict the residue at the covarying site with accuracy better than chance [44] . Thus , information constrains the context of a residue , in the same manner that epistasis constrains the fitness effect of a mutation in the presence of another . Using information as a proxy for epistasis is not a new idea ( see , e . g . , [45] and references cited therein , as well as [46] for an application to gene networks ) . However , the relationship between epistasis and information is not one-to-one . As a consequence , it is possible that two residues interact epistatically but show no information . Conversely , if two residues have positive information , they must interact epistatically , at least in a model without spatial effects and infinite population size . Thus , information is a sufficient ( but not a necessary ) condition for epistasis . We have analyzed the relationship between epistasis and information in random fitness landscapes of two interacting loci , and found that in less than 30% of cases would a pair of loci be epistatic while showing negligible information ( see below ) . Furthermore , those pairs that showed little information also predominantly showed little epistasis . Keeping in mind this limitation , studying pair-wise information ( which can be deduced from sequence data ) instead of epistasis ( which cannot ) allows us to use the wealth of time-course sequence data of a protein to study its long-term evolution . Not that residue covariation in the protease can in principle be due to population subdivision rather than epistasis [47] , which would be reflected in a deeply branched phylogenetic structure . However , it turns out that deep branches are rare in HIV-1 phylogeny . Indeed , using sequence data from the HIV Stanford Database , Wang and Lee show that amino acid covariation in HIV-1 protease sequences is largely due to selection pressures and not due to background linkage disequilibrium [48] . We corroborate this and find that protease sequences from the same database assume a star-like phylogeny ( S1 Text ) , supporting the notion that any observed residue covariation is largely due to selection pressures and not due to population subdivision and other phylogenetic factors . We first show how the persistent selective pressure of a drug environment increases residue variability over time , and document changes in physico-chemical properties due to these residue substitutions that suggest non-neutral evolution . We then show that epistatic interactions are enriched over time in the protein undergoing continuous adaptive evolution in the presence of drugs , but not in proteins where the environment remained constant . This leads us to conclude that while selection pressures increase per-site residue variability–and thus reduce information stored at each protein locus on average–the information stored in higher-order interactions increases over time .
To compare the long-term adaptation of the HIV-1 protease in the presence and absence of selection pressure of treatment , we analyzed the protease sequences collected in the years 1998–2006 from patients that did not receive treatment ( untreated group ) , as well as from patients that received treatment ( treated group , see Materials and Methods ) . The per-site entropy ( a measure of sequence variation ) for each of the 99 positions shows that some protease positions are highly variable even in the absence of treatment , thus indicating some degree of neutrality . However , many more loci become variable ( entropic ) upon treatment ( Fig 1 , see S2 Text for sequence logos of untreated and treated protease sequences averaged over all years ) . Moreover , several loci in the protein had higher entropy in 2006 compared to 1998 , hinting at an increase in per-site residue variability over the years , especially in the treated data set ( see also S1 Fig , which shows entropy differences for all years for both the treated and untreated group ) . The increased variability per site might seem counter-intuitive from the point of view of population genetics , where adaptation results in substitutions that lead to reduced diversity from hitchhiking [35] . While such reduced diversity is possible in the short term , we caution that due to the high mutation rate of HIV , pre-treatment diversity can be re-established fairly quickly after drug resistance has emerged . This rebound in virus titer further allows the virus to find mutational paths to resistance . However , it is also possible that some of the observed variability is due to the stochastic mixing ( within the database ) of many populations that each took diverse paths to resistance . Some positions in the untreated group also show higher residue variability at the later time point , and this is likely due to transmission of drug-resistant virus to untreated individuals via new infections [49–51] . However , this background signal remains low due to the reduced selection pressure in the untreated group . A closer look at the protease positions with increasing residue variability reveals that the protein loci that have undergone a marked increase in variation have been previously associated with drug resistance ( Fig 2 top panel ) [52] . Position 63 is the only site that becomes significantly less variable upon treatment . Many of the changes in entropy at each position correlate well with physico-chemical changes ( such as changes in the iso-electric point pI or in the residue weight ) at those positions ( Spearman correlation coefficient between absolute values of entropy difference and pI difference: R = 0 . 738 , p = 2 . 95 × 10−18; Spearman R between absolute values of entropy difference and residue weight difference: R = 0 . 819 , p = 3 . 52 × 10−25 ) suggesting that these changes are adaptive and influence the function of the protein in its new environment ( see S3 Text and Fig 2 , middle and lower panels ) . Leucine at position 10 is substituted by the slightly heavier isoleucine , a compensatory mutation showing a slight increase in residue weight difference ( Fig 2 , bottom panel , L10I is a compensatory mutation as shown in [52 , 53] ) ; lysine is replaced by a more basic arginine residue at position 20 , another compensatory mutation as arginine can form more electrostatic interactions compared to lysine and thus enhances protein stability [52 , 54]; and aspartic acid is substituted by the uncharged asparagine at position 30 , a strong resistance mutation [52 , 55] . While the residue physico-chemical properties discussed here are not exhaustive , the significant positive correlation between changes in these properties and changes in residue variability suggest that the residue variation brought about by treatment is non-random . Information about a protein’s environment is stored not only in individual residues of the protein , but also in the manner in which these residues interact epistatically . We estimate the information the protein stores about its environment in two ways: one that considers explicitly the information stored at each position in the protein independent of other sites ( I1 ) , and the measure I2 that in addition to I1 includes the mutual information between every pair of positions ( see Materials and Methods ) . An increase in entropy per site corresponds to a decrease in per-site information ( I1 ) . We find that the I1 for treated protease sequences is consistently lower that that of untreated sequences , indicating high sequence variability in the treated data ( Fig 3 , top panel ) . The slopes for the treated and untreated I1 data are not significantly different , suggesting that although the treated data has higher sequence variability , this variability does not increase significantly over the years . However , the sum of mutual information of all pairs of positions ( the component of information due solely to pair-wise interactions ) gradually increases , suggesting that epistatic interactions become enriched over time ( Fig 3 , middle panel , slopes for treated and untreated data are significantly different , p ≤ 0 . 001 ) . Fig 4 shows a gradual increase in pairwise mutual information between protease positions due to treatment ( bottom panel ) , while pairwise mutual information in the untreated group remains low over the years ( top panel ) . As discussed , adding the sum of pairwise mutual information to I1 gives I2 , which thus measures the information stored in the protein taking into account the pairwise dependencies between positions in addition to per-site variability ( Fig 3 , bottom panel , slopes for treated and untreated data are not significantly different ) . We find that I2 remains relatively constant over the years and converges to the same value in the treated as well as untreated data despite increase in residue variability and enrichment of epistatic interactions in the treated data . Such a finding may at first appear surprising , but Carothers et al . showed that the functional capacity of biomolecules ( RNA aptamers in their study ) correlated with the information contained in the sequence , and that functionally equivalent biomolecules had similar total information content [56] . Our analysis thus suggests that the similar total information content for proteins in the treated and untreated case reflects similar functional activity , achieved in the treated group by increasing epistasis that compensates for the information loss due to increased sequence variability . The increase in epistasis thus allows the protein to adapt and attain high fitness ( via wild-type level biological activity ) , in the increasingly rugged fitness landscape of treatment . Besides a trend in the overall strength of epistatic interactions in the treated group , we also find that the spatial organization of interactions in the protein is modified . In S5 Text , we show that untreated subjects stored more information in distant pairs early ( 1998 , distant: residue distance ≥ 8Å ) , a trend that is reversed in 2002 . As pairwise mutual information is a proxy for epistasis ( Materials and Methods ) , the significant temporal increase in mutual information suggests that epistatic interactions are crucial for the protease to adapt in a dynamic drug environment . In contrast , the sum of pairwise mutual information remains fairly constant in the drug-free environment ( Fig 3 , middle panel ) . It should be noted that most of the treated data for year 2003 came from two phase-III clinical trials that focused on the 2nd-line anti-retroviral drug tipranavir ( 2900 out of 3399 sequences ) [57] . Resistance to tipranavir requires accumulation of several mutations , more than the mutations required for the 1st-line protease inhibitors , and this higher genetic barrier to resistance makes it suitable for salvage therapy for patients already experiencing resistance to other drugs . The substantial decrease in I1 for the year 2003 thus can be attributed to an increased entropy as a consequence of accumulating a higher number of mutations required for resistance to tipranavir . It is curious that the marked increase in variability in the tipranavir-dominated data set is not associated with a marked increase in pair-wise information ( compared to the adjacent years ) , suggesting that the tipranavir-induced mutations are mostly non-epistatic . Pairwise interactions between amino acids are thought to be sufficient to encode the protein fold , and thus pairwise interactions can be considered as a sufficient source of protein functional information [58] . While there is currently no evidence that higher-order interactions between residues are important , some authors have discussed this issue [59] . Mapping the high information loci ( interactions of a strength of at least 0 . 1 bits ) on the protease structure for the treated group for the first year ( 1998 ) and comparing them to the treated group in 2006 ( Fig 5 ) clearly shows the increase in epistatic connections in the latter time point . The observation that strong selection due to treatment increases residue variability ( thus decreasing I1 ) while increasing epistatic interactions in the protease ( as measured by pairwise mutual information ) is corroborated by a longitudinal analysis of information measures in protease sequences from patients who were untreated at the first time point and received treatment at the second time point ( see S4 Text ) .
It is difficult to predict the evolutionary trajectory of a protein from fitness effects of mutations along with mutation rate and population size , but we can trace the evolutionary history of a protein to understand the processes underlying its long-term evolution . While evolution experiments with bacteria , viruses , and yeast provide direct evidence for evolution-in-action , enabling the study of various aspects of evolutionary dynamics as well as the likelihood of certain evolutionary paths [61–63] , indirect evidence such as sequence data collected over years can be a valuable resource to retrace the evolutionary steps taken by a protein on an adaptive fitness landscape over a long period of time . The sequences of the HIV-1 protease are one such resource that contain the “fossils” of the evolutionary trajectory ( albeit in a statistical form ) of the protein . Although this sequence data comes from isolates collected from different patients over the years , the star-like phylogeny observed for data from each year indicates the absence of population subdivisions and linkage disequilibrium that might have confounded the information-theoretic analysis we present here ( S1 Text ) . It is likely that each sequence in the data represents the major HIV-1 variant circulating in each patient . If the viral populations within the hosts have not yet reached equilibrium ( if such an equilibrium exists in the presence of persistent selection pressures of host immune response and treatment ) , the sequence obtained from treated individuals will not contain all of the resistance and compensatory mutations that might be found at the equilibrium . Thus , the information-theoretic measures computed here represent a lower bound of the true sequence variability and epistatic interactions present in a within-host viral population . We have investigated the long-term evolution of a protein by computational analysis of HIV-1 protease sequences from treated and untreated patients collected over a span of nine years ( 1998–2006 ) , and find that the molecule responds to the selection pressures of treatment by accumulating mutations that confer drug resistance . At the same time , several positions in the protease show considerable neutrality even in the absence of treatment ( Fig 1 ) . Indeed , a complete mutagenesis of the protease showed that several sites are insensitive to mutation in the absence of a selection pressure , and thus appear neutral [64] . Yet , some of those neutral mutations often appear in tandem with known resistance mutations , possibly compounding the fitness effects of the resistance mutations [65] . Even known resistance mutations usually do not confer resistance in isolation , but require compensatory mutations before resistance is achieved [53] . Because on average a mutation destabilizes the protein fold by about 1 kcal/mol , proteins cannot accumulate multiple resistance mutations without running the risk of thermal instability [11] . Yet , HIV-1 protease with resistance to multiple drugs can accumulate more than 10 mutations [66–68] . We suspect that many of these mutations are re-stabilizing the protein , thus compensating for the fitness cost of resistance mutations [12 , 13] . In addition , sequence logos of protease sequences from untreated patients ( S2 Text ) shows that several resistance and compensatory mutations are present at low frequencies prior to treatment , suggesting that pre-existing compensatory mutations may protect the protein from incurring the potentially strong fitness costs of resistance mutations . We show that as sequence variability increases due to treatment , more and more of the variable residues interact with other residues , leading to an increase in the epistatic interactions . Along this evolutionary trajectory , the protein finds itself in a fitness landscape that is increasingly rugged [23] . That the ruggedness of the fitness landscape has significant effects on the evolution of a protein has been discussed extensively [25 , 43 , 69–72] . These studies suggest that epistasis can have either an inhibitory or an accelerating effect on evolutionary trajectories , determined mainly by whether evolution occurs far off the fitness peak or close to it . Computational simulations of evolutionary adaptation reveal that while increased epistasis correlates with high ruggedness in the fitness landscape , evolvability ( the ability to attain the highest fitness peak ) declines beyond a threshold epistasis [25] . The sign of epistasis ( positive vs . negative epistasis ) also can affect the speed of adaptation . Epistasis is positive ( or negative ) when the fitness effect of the double mutant is greater ( or smaller ) than the sum of the fitness effects of individual mutations . Negative epistasis between beneficial mutations is often associated with diminishing returns in climbing a single peak [26 , 73–75] , while positive epistasis is associated with compensatory effects [76] that can enhance crossing of valleys between peaks [25 , 77] . Computational studies further clarify that positive epistasis accelerates evolvability , while negative epistasis promotes robustness [78] . However , if the landscape is too rugged , the reciprocal sign epistasis between mutations may prevent valley crossing [79] . As we cannot determine the sign of epistasis using our information-theoretic methods , we are unable to verify that the epistasis between mutations in the HIV-1 protease is mostly positive , as suggested by Bonhoeffer et al . [80] . We find here that as epistasis between residues in the protease continues to increase ( as revealed by the monotonically increasing sum of pairwise mutual information ) , epistasis is still contributing to adaptation rather than inhibiting it . However , there is little doubt that evolution in a more and more rugged fitness landscape will have a significant impact on evolvability . For example , it is known that some compensatory mutations that repair fitness costs increase viral fitness in the absence of drugs [39] . These mutations are found in individuals who have transmitted drug resistance , but they are not common in the treatment-naive HIV-1 population . One suggestion is that the evolutionary pathway to this polymorphism is simply too complex to emerge de novo [81] , suggesting that the virus has evolved into a region of genetic space from which there is no return . Previous studies of epistatic interactions in adaptive evolution emphasized their importance for adaptation [2 , 14 , 18 , 20 , 21 , 26 , 82] . Our analysis provides statistical arguments about the changes in epistatic interactions during long-term adaptation of a protein in an environment of persistent high selective pressure . Ideally , the present findings should be corroborated by longitudinal studies that collect sequence or fitness data regularly and for an extended period of time . Such studies would significantly contribute to and understanding of the impact of strong selection pressures and continually changing landscapes on adaptive evolution .
The protease sequences for HIV-1 subtype B were collected from the HIV Stanford database [http://hivdb . stanford . edu] on September 17 , 2013 . The database collects sequences from pilot studies and clinical trials that are published with sequence data deposited in GenBank . We only used sequences for which the exact dates for isolate collection are known ( as opposed to guessed from the time of publication or clinical trials ) . We focused our analysis on 18 , 571 sequences from the years 1998 to 2006 , in which more than 300 protease sequences are available for both treated and untreated sequence datasets ( Table 1 ) . Since these data come from different sources , they are not longitudinal . Sequences obtained from patients receiving ≥ 1 protease inhibitors are labeled as treated , while sequences from patients not receiving any protease inhibitors are labeled as untreated . Note that a fraction of new infections have transmitted drug-resistance , and thus even higher sequence diversity ( ≈8% of new cases had transmitted drug-resistance by January 2007 [83] ) . Since the HIV Stanford database does not keep data on transmitted drug-resistance for treatment-naive individuals , we label all sequences from treatment-naive patients as ‘untreated’ but cannot exclude the possibility that they might carry resistance mutations . However , due to their high fitness cost some drug-resistance mutations tend to revert back to wild-type residue in absence of therapy [84] , suggesting that the protease sequence diversity due to resistance mutations should decrease in cases of transmitted drug-resistance in treatment-naive patients . We also obtained longitudinal data ( protease sequences collected from the same patient after a time interval ) that was available from HIV Stanford database for the following: i ) patient was untreated at first as well as second isolate collection ( untreated to untreated ) : 596 sequences , ii ) patient untreated at first isolate collection but treated with protease inhibitors at second isolate collection ( untreated to treated ) : 395 sequences , iii ) patient treated at both isolate collections ( treated to treated ) : 921 sequences , and iv ) patient treated at first isolate collection but untreated at second isolate collection ( treated to untreated ) : 153 sequences . The time between first and second isolate collections ranged from one month to a few years . Positive information between two sites indicates that the two sites are co-varying , but co-variation is not the same as epistasis . To study the relationship between informational co-variation and epistasis , we constructed a population-genetic two alleles-two loci model that we solve using replicator-mutator equations . ( The model can be generalized to two loci with 20 alleles in a straightforward manner . ) We consider four genotypes ( two alleles ‘A’ and ‘a’ , at two loci ) : AA: ( type 0 , the wild type ) , Aa ( type 1 ) , aA ( type 2 ) , and aa: ( type 3 ) that undergo mutation with rate μ per unit time ( see Fig 6 ) . The probability to find each of these genotypes in an infinite population depends on the fitness and probabilities of the other genotypes . In a discrete update scheme , the probability to find type i at time t + 1 is related to the same quantity at time t via p 0 t + 1 = p 0 t w 0 w ¯ F + μ p 1 t w 1 + p 2 t w 2 w ¯ + μ 2 p 3 t w 3 w ¯ ( 1 ) p 1 t + 1 = p 1 t w 1 w ¯ F + μ p 0 t w 0 + p 3 t w 3 w ¯ + μ 2 p 2 t w 2 w ¯ ( 2 ) p 2 t + 1 = p 2 t w 2 w ¯ F + μ p 0 t w 0 + p 3 t w 3 w ¯ + μ 2 p 1 t w 1 w ¯ ( 3 ) p 3 t + 1 = p 3 t w 3 w ¯ F + μ p 1 t w 1 + p 2 t w 2 w ¯ + μ 2 p 0 t w 0 w ¯ ( 4 ) where w ¯ is the mean fitness w ¯ = ∑ i = 0 3 p i t w i , and F is the fidelity of replication F = 1 − 2μ − μ2 . It is easy to show that ∑ p i t + 1 = 1 as long as ∑ p i t = 1 . Eqs ( 1–4 ) can be solved numerically iteratively , but alternatively the fixed point ( the pi in the limit t → ∞ ) can be calculated by solving for the right eigenvector of the associated Markov matrix . We investigate different fitness landscapes by varying the fitness of the mutants , while the fitness of the wild-type is constant at w0 = 1 . Fig 7 shows equilibrium allele frequencies for a landscape where the double mutant has a given fitness w3 , while the intermediate genotypes have zero fitness ( a valley-crossing landscape with reciprocal sign epistasis ) . Armed with the equilibrium probabilities pi , we can calculate the information between loci as follows . First we define p ( A ) and p ( a ) for the first and second locus: p ( 1 ) ( A ) =p0+p1 , p ( 1 ) ( a ) =1−p0−p1p ( 2 ) ( A ) =p0+p2 , p ( 2 ) ( a ) =1−p0−p2 ( 5 ) giving us the marginal entropies of the first and second locus H ( 1 ) =−∑i=a , Ap ( 1 ) ( i ) logp ( 1 ) ( i ) , H ( 2 ) =−∑i=a , Ap ( 2 ) ( i ) logp ( 2 ) ( i ) . ( 6 ) The joint entropy of both loci is then H ( 1 , 2 ) = - ∑ i = a , A ∑ j = a , A p ( 1 , 2 ) ( i , j ) log p ( 1 , 2 ) ( i , j ) . ( 7 ) where p ( 1 , 2 ) ( i , j ) is the joint probability to observe allele i at locus 1 and allele j at locus 2 . The shared entropy ( or information ) is I ( 1 : 2 ) = H ( 1 ) + H ( 2 ) - H ( 1 , 2 ) . ( 8 ) We can then relate this information to the epistasis between loci calculated as [25 , 80] E = log ( w 3 w 0 ) - log ( w 2 w 0 ) - log ( w 1 w 0 ) = log ( w 3 w 0 w 1 w 2 ) ( 9 ) There are other ways of defining epistasis between loci ( see , e . g . , [77] ) , but the qualitative relation between information and epistasis is not affected . An extreme example occurs when w0 = w3 = 1 and w1 = w2 = 0 , that is , when the double mutant has the same fitness as the wild type , but the intermediate genotypes have no fitness . In that case , it is necessary to cross a valley in the fitness landscape to reach the double mutant aa . In this case of reciprocal sign epistasis [63] , E = ∞ , and I ( 1 : 2 ) = - ( 1 - p 0 ) log ( 1 - p 0 ) - ( 1 - p 3 ) log ( 1 - p 3 ) . ( 10 ) If p0 = p3 = 0 . 5 ( full equilibration ) this extreme level of epistasis correspond to 1 bit of information ( the maximum possible ) . Fig 8 shows the changes in genotype probabilities and mutual information as the population adapts from a single-peak landscape ( w0 = 1 and w1 = w2 = w3 ≈ 0 ) to a two-peak fitness landscape landscape ( w0 = w3 = 1 and w1 = w2 ≈ 0 ) . Pairwise mutual information increases as the landscape becomes more rugged . See S7 Text for simulations for a three-loci two-allele model that show that an increase in the sum of mutual information coincides with the increase in the ruggedness of the fitness landscape . To study the general relationship between epistasis and information , we calculated both epistasis and information starting with w0 = 1 ( wild type fitness ) , and random fitness values ( between 0 and 1 ) for the single and double mutants w1 , w2 , and w3 . The equilibrium genotype probabilities were obtained by iterating the replicator-mutator equations until they had stabilized ( 30 , 000 updates of genotype probabilities p0 , p1 , p2 , and p3 using Eqs ( 1–4 ) , with starting genotype probabilities: p0 = 1 , p1 = p2 = p3 = 0 ) . We find that the absolute value of epistasis |E| is positively correlated with information ( Spearman R = 0 . 497 , p < 10−15 , see Fig 9 ) . It is clear that information is a sufficient ( but not necessary ) condition for epistasis . Thus , high information between two loci guarantees epistasis between those two loci , but there may be epistatically interacting loci that are missed when focusing only on information , as some loci can interact epistatically without being informative about each other . In addition , the direction of epistasis ( the sign of E ) cannot be determined solely from information . To ascertain the fraction of epistatically interacting pairs that are missed by an informational analysis , we examined pairs with information below a chosen cut-off ( here I < 0 . 001 bits ) . About 35% of all pairs have information below the threshold , but only 28% of all pairs have non-vanishing epistasis at the same time ( see inset in Fig 9 , which shows the distribution of epistasis among pairs with sub-threshold information ) . This finding leads us to conclude that , at least in the simple model presented here , only about 30% of epistatically interacting pairs are missed by using an information proxy . Furthermore , the majority of those missed pairs have small epistasis , as evidenced by the distribution of epistasis for those pairs with negligible information in the inset of Fig 9 . The information content of biomolecules can be estimated using information-theoretic constructs [44 , 85 , 86] . Information content is different from sequence length: it can be thought of as the amount of information that is stored in the protein sequence about the cellular environment within which it functions ( implying that it is contextual ) . Information content can change when the environment changes even if the sequence remains the same , and is thought to be a proxy for fitness [87] . Because a changed environment usually translates into reduced information content ( reflecting reduced fitness ) , the virus seeks to recover the information and achieve previous fitness levels by evolving drug resistance . The total information content of a protein of length ℓ ( taking into account all correlations between residues ) is given by ( adapting a formula for the “multi-information” of ℓ events due to Fano [88] ) : I ℓ = H max - ∑ i = 1 ℓ H i - ∑ i < j ℓ I ( i : j ) + ∑ i < j < k ℓ I ( i : j : k ) - ⋯ , ( 11 ) where I ( i : j ) is the mutual information between two sites , I ( i : j : k ) the information shared between three sites [88] , and so on . The terms not shown in Eq ( 11 ) are the higher-order corrections I ( i : j : k : m ) etc . , up to I ( i1 : i2 : … : iℓ ) , with alternating signs . Corrections due to interactions between three or more sites are expected to be small , but cannot be estimated using the present data because the samples are too small . H max is the sum of maximum entropy at every site , and thus is equal to log2 ( 20 ) × ℓ ≈ 4 . 32 × ℓ bits , where ℓ is the sequence length [44] . We define the first- and second-order information estimates of Eq ( 11 ) I1 and I2 as follows: I 1 = H max - ∑ i = 1 l H i , ( 12 ) I 2 = I 1 + ∑ i < j ℓ I ( i : j ) . ( 13 ) Thus , I1 measures the information content of the protein without considering any interactions among protein residues , and I2 includes information contained in pairwise interactions between protein positions over and above I1 , but ignores any other higher-order interactions between residues . The second term in Eq ( 13 ) is the sum of pairwise mutual information for all pairs of residues in the protein . Note that if information was only described by I1 , mutations that provide resistance reduce the information in the ensemble , as they increase sequence entropy . Achieving an increase in total information must then occur via correlated mutations . The site-wise changes in physicochemical properties of residues such as the iso-electric point pI and residue weight in the protease are calculated by averaging their values across the subsamples from each year and environment ( treated and untreated ) . The physicochemical properties of residues and their changes are discussed in S3 Text . Small datasets do not correctly estimate the amino acid probabilities due to dataset-dependent observed frequencies of residues: this introduces a bias in the entropy and mutual information calculations ( see , e . g . , [89 , 90] ) . Several priors and estimators have been proposed to estimate entropies from undersampled probability distributions [90] , and our analysis suggests that a sample size of 300 with NSB ( Nemenman-Shafee-Bialek ) entropy bias correction gives reliable estimates of entropy and mutual information values ( see S6 Text ) . Since the number of treated and untreated protease sequences in our dataset is different across years ( Table 1 ) , we sampled ( with replacement ) 10 sets/year of 300 sequences each to account for sample size bias . We calculate bias-corrected entropies and pairwise mutual information for the subsampled datasets , and report the average values ( along with ±1 SD ) . Because several protease sequences had gaps at the sequence ends , we calculated the I1 , I2 , and sum of pairwise mutual information for a truncated sequence length ( from positions 15 to 90 instead of positions 1 to 99 of the protease sequence , see S2 Fig ) . To compare the temporal trends of I1 , I2 , and sum of pairwise mutual information for untreated and treated sequence data , we first fit linear regression models to the yearly data using the R statistical analysis platform [91] and then determine if the slopes of the regression models ( for treated and untreated datasets ) are significantly different or not . Although the data used in this analysis is not longitudinal ( protease sequences are collected from different patients participating in different clinical trials/studies across the years ) , the linear regression model is fit to the average values of the response variables ( I1 , I2 , and sum of pairwise mutual information ) calculated by sampling ten sets of 300 sequences each , and thus reflect the approximate temporal trends of the response variables with respect to the two factors ( untreated and treated ) .
|
Evolution is often viewed as a process that occurs “mutation by mutation” , suggesting that the effect of each mutation is independent of that of others . However , in reality the effect of a mutation often depends on the context of other mutations , a dependence known as “epistasis” . Even though epistasis can constrain protein evolution , it is actually very common . Such interactions are particularly pervasive in proteins that evolve resistance to a drug via mutations that create defects , and that must be repaired with compensatory mutations . We study how epistasis between protein residues evolves over time in a new and changing environment , and compare these findings to protein evolution in a constant environment . We analyze the sequences of the human immunodeficiency virus type 1 ( HIV-1 ) protease enzyme collected over a period of 9 years from patients treated with anti-viral drugs ( as well as from patients that went untreated ) , and find that epistasis between residues continues to increase as more potent anti-viral drugs enter the market , while epistasis is unchanging in the proteins exposed to a constant environment . Yet , the proteins adapting to the changing landscape do not appear to be constrained by the epistatic interactions and continue to manage to evade new drugs .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"organismal",
"evolution",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"evolutionary",
"biology",
"enzymes",
"pathogens",
"enzymology",
"microbiology",
"retroviruses",
"epistasis",
"immunodeficiency",
"viruses",
"viruses",
"rna",
"viruses",
"microbial",
"evolution",
"pharmacology",
"evolutionary",
"adaptation",
"thermodynamics",
"fitness",
"epistasis",
"entropy",
"proteins",
"medical",
"microbiology",
"hiv",
"microbial",
"pathogens",
"hiv-1",
"viral",
"evolution",
"physics",
"biochemistry",
"heredity",
"viral",
"pathogens",
"virology",
"genetics",
"biology",
"and",
"life",
"sciences",
"proteases",
"physical",
"sciences",
"lentivirus",
"drug",
"interactions",
"evolutionary",
"processes",
"organisms"
] |
2016
|
Strong Selection Significantly Increases Epistatic Interactions in the Long-Term Evolution of a Protein
|
Extracellular phosphorylation of proteins was suggested in the late 1800s when it was demonstrated that casein contains phosphate . More recently , extracellular kinases that phosphorylate extracellular serine , threonine , and tyrosine residues of numerous proteins have been identified . However , the functional significance of extracellular phosphorylation of specific residues in the nervous system is poorly understood . Here we show that synaptic accumulation of GluN2B-containing N-methyl-D-aspartate receptors ( NMDARs ) and pathological pain are controlled by ephrin-B-induced extracellular phosphorylation of a single tyrosine ( p*Y504 ) in a highly conserved region of the fibronectin type III ( FN3 ) domain of the receptor tyrosine kinase EphB2 . Ligand-dependent Y504 phosphorylation modulates the EphB-NMDAR interaction in cortical and spinal cord neurons . Furthermore , Y504 phosphorylation enhances NMDAR localization and injury-induced pain behavior . By mediating inducible extracellular interactions that are capable of modulating animal behavior , extracellular tyrosine phosphorylation of EphBs may represent a previously unknown class of mechanism mediating protein interaction and function .
Modification of protein function by phosphorylation controls many aspects of cellular function and signaling [1] . Interestingly , the first evidence for phosphoproteins came from the observation that the secreted milk protein , casein , contained phosphate , suggesting that phosphorylation can occur in the extracellular space [2 , 3] . Recently , protein kinases that mediate the selective phosphorylation of extracellular serine , threonine , and tyrosine amino acids have been identified . For example , extracellular phosphorylation of serine and threonine residues can be mediated by Fam20C [4 , 5] and phosphorylation of extracellular tyrosine residues can be accomplished by vertebrate lonesome kinase ( VLK or PKDCC ) , an essential gene expressed throughout the body , including the nervous system [6 , 7] . Yet despite identification of these kinases , the functional significance of extracellular phosphorylation and whether extracellular phosphorylation of proteins inducibly modulates their function remains largely unexplored . At excitatory synapses , glutamate receptors must be recruited and stabilized at synaptic sites . Of particular importance are interactions that maintain the proper localization of N-methyl-D-aspartate receptors ( NMDARs ) , glutamate receptors that are essential for synaptic plasticity and development [8] . The synaptic localization , function , and signaling of NMDARs are regulated by intracellular scaffolding proteins such as PSD-95 [9] , extracellular interacting proteins such as neuroligin-1 [10] , and the EphB receptor tyrosine kinases ( RTKs ) [11] . While the mechanisms mediating intracellular interactions are well understood , the mechanisms mediating extracellular protein—protein interactions are not . The direct extracellular interaction between the EphB receptor tyrosine kinase and the NMDAR appears to play important roles in the localization , function , and signaling of NMDARs [11] . The EphB family of RTKs consists of 5 members that bind to transmembrane ephrin-B ligands . EphB1–3 are essential for formation of up to 40% of excitatory synapses in the developing hippocampus and cortex , while in the mature brain EphBs are required for normal levels of synaptic NMDARs [12–14] . Ephrin-B binding to EphBs controls the localization and function of synaptic NMDARs by inducing a direct extracellular domain—dependent interaction with the NMDAR [11 , 14–17] . While in vitro binding assays indicate that EphBs bind the GluN1 subunit of the NMDAR via a direct extracellular interaction [11 , 18] , the domain and molecular mechanism mediating the interaction between these 2 proteins remain undefined . Underscoring the importance of the EphB—NMDAR interaction , the EphB—NMDAR interaction has been linked to a number of human diseases that are associated with NMDAR dysfunction . The pathological disruption of the ability of the EphB and NMDAR to interact has been linked to NMDAR dysfunction in Alzheimer disease [19 , 20] and in anti-NMDAR encephalitis [18 , 21] . A common feature of these findings is the disruption of the ability of EphBs to interact biochemically with the NMDAR and a rescue of the defects associated with the disease state by restoring the EphB—NMDAR interaction . The EphB-dependent enhancement of NMDAR activity associated with the EphB—NMDAR interaction is also linked to disease . EphB-dependent enhancement of NMDAR function plays a key role in sensitization of nociception , leading to chronic neuropathic and malignancy-induced pain through an unknown mechanism [22–25] . Because the interaction between EphBs and the NMDAR occurs in the extracellular space , it is thought to be a promising drug target [19] . However , despite the apparent importance of the EphB—NMDAR interaction , the molecular mechanisms controlling direct extracellular interaction between these proteins are unknown . Here we show that the EphB2 receptor tyrosine kinase undergoes ephrin-B ligand-induced extracellular phosphorylation of tyrosine residues ( Y481 and Y504 ) . Sequence analysis indicates that 1 of these amino acids , Y504 , is widely conserved in fibronectin type III ( FN3 ) domains of Eph proteins across phylogeny and is present in all Eph proteins known to interact with the NMDAR . In cortical and spinal cord neurons , ephrin-B—dependent induction of the EphB—NMDAR interaction is mediated by extracellular phosphorylation of Y504 on EphB2 . The charge of Y504 is both necessary and sufficient for EphB—NMDAR interaction and regulates the amount of NMDARs found at synaptic sites . Virally mediated spinal cord expression of EphB2 or a phosphomimetic EphB mutant that induces the EphB—NMDAR interaction increases NMDAR levels in the dorsal horn of the spinal cord and results in mechanical hypersensitivity . Intrathecal injections of a membrane-impermeable ectokinase inhibitor that blocks the EphB—NMDAR interaction in cortical and spinal cord neurons reduces long-term hypersensitivity induced by EphB2 wild type ( WT ) but not pathological pain induced by injection of a phosphomimetic EphB mutant . These findings suggest that extracellular phosphorylation of EphB2 regulates NMDAR synaptic localization and function . Together the data suggest extracellular phosphorylation as a novel , dynamic mechanism that regulates protein—protein interactions at synapses to drive assembly of macromolecular complexes .
EphB receptors interact directly with NMDA-type glutamate receptors through an undefined region of their extracellular domain [11] . The extracellular domain of the EphB receptor consists of a globular domain required for ephrin-B binding , a cysteine-rich domain , and 2 FN3 repeat domains of unknown function ( Fig 1C ) . To study whether the extracellular region of EphB2 undergoes post-translational modification , we took an unbiased mass spectrometry—based approach: liquid chromatography tandem mass spectrometry ( LC-MS/MS ) in combination with receptor immunoprecipitation ( IP ) and phosphopeptide mapping ( S1A–S1D Fig ) . FLAG-tagged EphB2 was expressed in the neuroblastoma cell line NG108 , treated with either ephrin-B1 or control reagents , and immunoprecipitated . After enrichment of phosphopeptides using TiO2 , LC-MS/MS identified the known tyrosine phosphopeptides in the juxtamembrane and kinase domains ( S1 Table ) of EphB2 and 2 tyrosine phosphopeptides ( ELSEYNATAIK [Y481] and AGAIYVFQVR [Y504] ) from the extracellular portion of EphB2 ( Fig 1A–1C ) . Each extracellular peptide was identified in 4 independent experiments , with Mascot scores of 34 and 63 , respectively , and definable separation from the next peptide assigned to that spectrum . Manual inspection of the tandem mass spectrometry ( MS/MS ) spectrum confirmed that the majority of ion signals present are accounted for by the assigned amino acid sequence and ions critical to localization of the site of phosphorylation are present . The 2 phosphopeptides in the extracellular region of EphB2 were both found in the C-terminal FN3 ( cFN3 ) domain ( see Fig 1C for schematic ) and correspond to tyrosine residues Y481 and Y504 . Interestingly , recent analysis of human non—small cell lung cancer cell line also identified Y481 as undergoing phosphorylation ( PhosphoSitePlus; http://www . phosphosite . org ) , further supporting our analysis . Y504 and neighboring amino acid residues in the identified peptide are well conserved among different species ( 79 . 0% identical amino acids ) ( S1E Fig ) , within the Eph family of proteins ( 55 . 4% identical amino acids ) ( Fig 1D ) , and in other FN3-containing molecules ( Figs 1E and S1F ) . In contrast , Y481 and neighboring residues in the identified peptide are less well conserved amongst different species ( 63 . 6% identical amino acids ) ( S1E Fig ) , within the Eph family ( 19 . 6% identical amino acids ) , and in other FN3-containing proteins ( Figs 1D and 1E and S1F ) . Because Y481 is found only in EphB2 and was not conserved among other known NMDAR-interacting Eph family members ( Fig 1D ) , we focused our study on Y504 . Our analysis of the F-strand region of cFN3 domains indicates that Y504 is well conserved in proteins that contain homologous domains [11 , 26] . Therefore , we next examined mass spectrometry ( MS ) databases and asked whether FN3-containing molecules that regulate axon guidance and target recognition might also contain phosphorylated tyrosines in FN3 domains that are similar to the cFN3 domain of EphB2 . Interestingly , phosphorylated tyrosines have been identified in a number of these proteins at homologous residues . Proteins with previously identified phosphorylation sites include Sidekick1 and Sidekick2 as well as DSCAM1 ( S1F Fig ) . These findings suggest that phosphorylation at EphB2 Y504 may be conserved among various species , Eph family proteins , and other synaptic FN3-containing molecules . To begin to determine whether Y504 might be phosphorylated , we generated a polyclonal phospho-specific antibody to tyrosine 504 of EphB2 . Recognition by the phospho-Y504 antibody ( p*Y504 ) was blocked by preabsorption with the immunogenic phospho-Y504 peptide ( S1G Fig ) . In addition , the p*Y504 antibody recognized full-length EphB2 WT and full-length EphB2 Y481F but not nonphosphorylatable EphB2 Y504F , kinase-dead , or intracellular region—truncated EphB2 mutants ( Fig 1F and S1H Fig ) . These results suggest that the antibody selectively recognizes phosphorylated Y504 and that the kinase domain may play a role in the phosphorylation of EphB2 Y504 . To test whether Y504 phosphorylation of EphB2 is enriched in the cortex and spinal cord , we purified synaptosomes from these regions of WT mice . To validate our synaptosome purification , blots were probed for EphB2 , PSD-95 , and GluN1 [27] . The EphB p*Y504 signal was enriched in the synaptosome fraction and migrated at the same molecular weight as EphB2 . To confirm that the signal detected was due to the presence of phosphate groups , the same blots were incubated with calf intestinal alkaline phosphatase ( CIP ) ( Fig 1G ) . Incubation with CIP completely removed the signaling from the p*Y504 antibody but had no effect on the signal from the EphB2 antibody . Thus , EphB2 is likely to be phosphorylated on Y504 at synapses in both the brain and spinal cord ( S1I Fig ) . Finally , to begin to address whether phosphorylation of Y504 on EphB2 might change as synapses mature , we asked whether the phosphorylation of Y504 was developmentally regulated . Synaptosomes were purified from P9 , P15 , and P21 WT mouse brains and probed for Y504 phosphorylation . We found that the levels of p*Y504 were elevated at P9 and P15 and declined with age . Interestingly , the pattern of p*Y504 parallels decreases in the proportion of NMDARs containing the developmentally regulated NMDAR subunit GluN2B ( Fig 1H ) [28 , 29] . Together these findings indicate that EphB2 is phosphorylated on a highly conserved residue in vitro and in the brain and spinal cord at synaptic sites and this phosphorylation is down-regulated as synapses begin to mature . To begin to examine the mechanism of EphB2 phosphorylation , we next asked whether EphBs might be phosphorylated on the cell surface after ephrin-B2 stimulation . Cultured cortical neurons were treated with activated ephrin-B2 for 45–60 minutes and then lysates were probed with the p*Y504 antibody . Consistent with the ephrin-B2-dependent activation of EphB2 , ephrin-B2 treatment led to a significant increase in Y504 phosphorylation ( Fig 2A and 2B ) . Blockade of EphB2 kinase activity after cell lysis had no effect on ephrin-B2–dependent phosphorylation of Y504 ( S2A and S2B Fig ) , suggesting that phosphorylation of EphB2 at Y504 occurred prior to cell lysis . EphB2 can be internalized by clathrin-mediated endocytosis after ephrin-B stimulation [30] . To block internalization of EphB2 , we blocked clathrin-mediated endocytosis with 450 mM hypertonic sucrose and asked whether ephrin-B2 treatment induced phosphorylation of EphB2 at Y504 . Inhibition of endocytosis blocked ephrin-B2–dependent internalization of EphB2 ( S2C and S2D Fig ) but had no effect on ephrin-B2–induced extracellular tyrosine phosphorylation ( p*Y504 ) of EphB2 ( Fig 2A and 2B ) . These findings suggest that EphB2 undergoes ligand-dependent phosphorylation of Y504 on the cell surface of neurons . To further test whether the extracellular domain of EphB2 was phosphorylated on the surface of neurons , we conducted live-cell cell surface staining experiments . Because our Y504 phospho-specific antibody is not suitable for immunostaining , these experiments were conducted with the broad-spectrum and well-characterized phosphotyrosine antibody PY99 . To induce Y504 phosphorylation , day in vitro ( DIV ) 7–16 cultured neurons were stimulated with activated soluble ephrin-B2 for 45 minutes , and live-cell staining was conducted with α-phosphotyrosine ( PY99 ) or control antibodies recognizing intracellular proteins . As expected , after treatment with ephrin-B2 or control reagents , live-cell staining with intracellular α-actin antibodies only showed weak background staining ( S2E Fig ) . In contrast , ephrin-B treatment of cultured neurons resulted in a marked increase in tyrosine phosphorylation staining on the cell surface at endogenous sites of ephrin-B binding compared to the control ( Fig 2C–2F; p < 0 . 01 , ANOVA followed by Dunn's multiple comparison test ) . We next asked whether any phosphotyrosine staining might be due to phosphorylation of Y504 . To test this , cultured neurons were transfected at DIV 3 with either EphB2 WT or EphB2 Y504F . After ephrin-B2 treatment , surface staining with PY99 resulted in a significant increase in phosphotyrosine staining in the EphB2 WT-transfected neurons but not Y504F-transfected neurons ( Fig 2G and 2H; p < 0 . 05 , ANOVA followed by Dunn's multiple comparison test ) . These findings suggest that the majority of the surface phosphotyrosine signal induced by ephrin-B2 treatment can be attributed to phosphorylation at Y504 . To begin to determine whether EphB2 might be phosphorylated by a kinase outside the cell , we blocked kinase activity selectively in the extracellular space by pretreating neurons with a membrane-impermeable ectokinase inhibitor , K252b [31] . In control in vitro experiments , K252b did not block EphB2 kinase activity ( S2A Fig ) . Following 1 hour of K252b treatment of cultured neurons , ephrin-B2–induced surface tyrosine phosphorylation was significantly reduced ( Fig 2C and 2D; p < 0 . 01 , ANOVA followed by Dunn's multiple comparison test ) . Moreover , preincubation with the phosphatase inhibitor vanadate potentiated the ephrin-B—induced increase in surface PY99 staining of ephrin-B2 binding sites ( Fig 2E and 2F; p < 0 . 01 , ANOVA followed by Dunn’s multiple comparison test ) . These findings suggest that ephrin-B2 stimulation of EphBs in neurons results in the tyrosine phosphorylation of the EphB ectodomain in the extracellular space . Both soluble and membrane-attached ectokinases have been identified in several cell types , including neurons [32] . We next sought to determine whether ephrin-B stimulation of neurons might induce a soluble activity sufficient to phosphorylate the extracellular domain of EphB2 . To address this , the medium of cultured neurons was replaced with artificial cerebrospinal fluid ( ACSF ) , a solution that contains only salts and glucose . Activated ephrin-B2 or control reagents were added to the ACSF , and the ACSF was collected after 45–60 minutes . The conditioned ASCF was then tested for an activity that could phosphorylate the ectodomain of EphB2 ( Fig 2I ) . Treated media alone failed to phosphorylate the extracellular region of EphB2 ( Fig 2J ) . Remarkably , the addition of 100 μM ATP , 10 mM magnesium acetate , and 10 mM manganese chloride to the ephrin-B2 treated ASCF resulted in robust phosphorylation of the EphB2 ectodomain ( Fig 2J and 2K; p < 0 . 005 , ANOVA followed by Fisher’s exact test ) . This activity was absent in control ACSF with ATP added or ephrin-B2–treated media without ATP and blocked by heating the ephrin-B2–treated ACSF to 73°C ( Fig 2J and S2F and S2G Fig ) . Taken together with the results from the live-cell staining experiments , these findings indicate that EphB2 Y504 is phosphorylated on the cell surface , likely by an ephrin-B2–induced soluble activity that requires ATP . EphBs are required for normal levels of synaptic NMDARs [14] . EphB2 binds to the NMDAR in both the cortex and the spinal cord [23] via an extracellular domain—dependent interaction that requires ephrin-B [11] . Therefore , we asked whether phosphorylation of EphB2 Y504 might be required for the EphB—NMDAR interaction in cortical and spinal cord neurons . In these experiments , dissociated cortical or spinal cord neurons were treated with soluble activated ephrin-B2 to activate the endogenous EphB—NMDAR interaction . We then conducted experiments to determine whether endogenous EphB2 Y504 was phosphorylated by ephrin-B2 treatment and whether pharmacological blockade of this phosphorylation might block the EphB—NMDAR interaction . To determine whether the EphB2–NMDAR interaction requires Y504 phosphorylation , endogenous extracellular kinase activity was blocked with K252b . Neurons were then treated with activated ephrin-B2 for 45–60 minutes , endogenous EphB2 was immunoprecipitated in a RIPA buffer , and blots were probed for endogenous GluN1 and p*Y504 . Ephrin-B2 treatment of DIV 6–7 cultured cortical and spinal cord neurons effectively induced the EphB—NMDAR interaction and phosphorylation of Y504 ( Fig 3A–3F ) . Pretreatment of neurons for 1 hour with K252b caused a significant decrease in phosphorylation of Y504 ( p < 0 . 001 , ANOVA followed by Fisher’s exact test; Fig 3B and 3E ) . K252b treatment did not alter the ability of ephrin-B2 treatment to stimulate EphB2 intracellular phosphorylation ( Fig 3D and S3A Fig ) , suggesting that K252b does not alter the ability of ephrin-B to activate the EphB kinase . However , K252b treatment did cause a significant decrease in the ephrin-B—induced EphB—NMDAR interaction ( **p < 0 . 01 , ****p < 0 . 001 , ANOVA followed by Fisher’s exact test; Fig 3C and 3F ) . These effects were particularly robust in cortical neurons but resulted in significant decreases in both cortical and spinal cord neurons . These data indicate that blockade of endogenous p*Y504 blocks the ability of EphB2 to interact with the NMDAR and suggest that the extracellular phosphorylation of EphB2 at Y504 is required for the EphB—NMDAR interaction . Induction of phosphorylation requires ATP hydrolysis . If phosphorylation of Y504 is necessary for the EphB—NMDAR interaction , we expect that blocking extracellular ATP hydrolysis should block the interaction . Neurons were treated with the nonhydrolyzable ATP analogue ATPγS ( 1 μM ) for 1 hour before ephrin-B treatment , and the effect on the EphB—NMDAR interaction was determined by co-IP . Bath application of ATPγS significantly reduced the ephrin-B—induced EphB—NMDAR interaction without affecting intracellular tyrosine phosphorylation of EphB2 ( ***p < 0 . 005 , ****p < 0 . 001 , ANOVA followed by Fisher’s exact test; S3B–S3D Fig ) . These findings suggest that phosphorylation of EphB2 at Y504 is required for the EphB—NMDAR interaction and that ephrin-B—dependent induction of this interaction can be blocked by blocking ATP hydrolysis in the extracellular space . Phosphorylation of Y504 appears to be required for the EphB—NMDAR interaction in neurons . Therefore , we tested whether the phosphorylation of Y504 within the cFN3 of EphB2 might be necessary and sufficient for the EphB—NMDAR interaction ( Fig 4A ) . To test this possibility , we generated an EphB2 phosphomimetic mutant ( Y504E ) and a nonphosphorylatable mutant ( Y504F ) and then examined the interaction between EphB2 mutants and GluN2B-containing NMDARs in HEK293T cells by co-IP . The ability of surface EphB2 to bind to ephrin-B2 was not altered by mutation of Y504 ( Fig 4B and S4A–S4D Fig , p > 0 . 05 , ANOVA ) , suggesting that the structure of the extracellular domain remains intact after mutation of this residue . In HEK293T cells , mutations to Y504 modulated the rate of removal of EphB2 from the cell surface; however , these effects were not seen in neurons ( S5A–S5G Fig ) . Mutation of the second phosphotyrosine identified by MS ( Y481 ) had no effect on the ability of EphB2 to interact with the NMDAR , suggesting that this amino acid is not required for the EphB—NMDAR interaction ( S4E and S4F Fig ) . In contrast , mutations of EphB2 at Y504 resulted in profound changes . Expression of the phosphomimetic EphB2 Y504E mutant resulted in a significant increase in EphB2 co-IP when pulled down with GluN1 antibody , while expression of the nonphosphorylatable EphB2 Y504F mutant caused a significant decrease in co-IP compared to WT ( Fig 4C and 4D; p > 0 . 001 , ANOVA followed by Fisher’s exact test ) . Similar effects were observed when antibodies against FLAG ( FLAG-EphB2 or FLAG-EphB2 mutants Y504E or Y504F ) were used to immunoprecipitate GluN1 ( Fig 4E and 4F ) . These results indicate that the charge of this residue within the extracellular domain of EphB2 is necessary and sufficient for the EphB—NMDAR interaction in HEK293T cells . To test the effect of EphB2 Y504 mutants in neurons , DIV 2 cultured cortical neurons were transduced with lentiviruses expressing enhanced yellow fluorescent protein ( EYFP ) -tagged EphB2 WT , EphB2-Y504E , or EphB2-Y504F . Neurons were then challenged with ephrin-B2 treatment at DIV 9 for 45–60 minutes , and the ability of the GluN1 subunit of the NMDAR to co-IP with EphB2 was tested . In the control group , cultured neurons were transduced with EphB2 WT . In these neurons , little EphB—NMDAR interaction was detectable without ephrin-B stimulation , but ephrin-B stimulation induced a significant increase in the co-IP of GluN1 with EYFP-tagged EphB2 WT ( Fig 4G and 4H; p = 0 . 0056 , ANOVA followed by Fisher’s exact test ) . In contrast , neurons transduced with Y504E mutant EphB2 showed robust GluN1 pull-down even in the absence of ephrin-B treatment ( Fig 4G and 4H; p = 0 . 0106 , ANOVA followed by Fisher’s exact test ) , and ephrin-B treatment of neurons transduced with EYFP-tagged EphB2 Y504E did not further increase the EphB—NMDAR interaction ( Fig 4G and 4H; p = 0 . 919 , ANOVA followed by Fisher’s exact test ) . These findings suggest that Y504 phosphorylation is sufficient to induce the EphB2–NMDAR interaction in neurons . Neurons transduced with the Y504F mutant form of EphB2 failed to show GluN1 co-IP with EphB2 , either with or without ephrin-B treatment ( Fig 4G and 4H; p = 0 . 549 , ANOVA followed by Fisher’s exact test ) . These data indicate that phosphorylated EphB2 Y504 is both necessary and sufficient for the EphB—NMDAR interaction in neurons and suggest that extracellular phosphorylation of a tyrosine in the cFN3 domain of EphBs may drive protein—protein interactions . To test whether phosphorylation of specific tyrosine residues and surrounding amino acids in the FN3 domain might provide a mechanism to mediate protein—protein interactions , we examined whether other Eph proteins with sequences similar to that found in EphB2 might interact with the NMDAR . EphB1–3 all interact with the NMDAR and have a similar set of amino acids near Y504 ( XpYVXQVR ) , while the sequences of EphA3 and EphA4 , which do not interact with the NMDAR , differ [11] . Interestingly , the sequence of EphA8 , which was not previously known to interact with the NMDAR , is identical to EphB2 in this region ( Fig 4I ) . To test whether EphA8 can associate with the NMDAR , we generated a FLAG-tagged EphA8 expression construct and coexpressed it along with GluN1 and GluN2B in HEK293T cells . We found that FLAG-EphA8 efficiently coimmunoprecipitates with the NMDAR from HEK293T cell lysates ( Fig 4J and 4K ) . EphA8 is only 49 . 0% homologous to EphB2 outside of this region , while EphA4 , which does not interact with the NMDAR , is 58 . 5% homologous; thus , these findings suggest that tyrosines with surrounding amino acids similar to EphB2 might be used to predict the ability of proteins to interact with the NMDAR . Early in development , EphBs function to control synapse formation [13] , but later in neuronal development , EphBs are key regulators of NMDAR localization and function at synapses [14] . To test whether the extracellular phosphorylation of EphB2 Y504 is required for EphB-dependent NMDAR function and synaptic accumulation , we asked if mutation of EphB2 at Y504 might alter the NMDAR-dependent synaptic currents in cortical neurons ( Fig 5A ) . To avoid effects on synapse development , neurons were transfected at DIV 14 with enhanced green fluorescent protein ( EGFP ) with or without EYFP-tagged EphB2 WT , Y504E , or Y504F constructs [14] . Cultures were infected at DIV 10 with adeno-associated virus transducing channelrhodopsin-2 to enable optical activation [33] ( Penn Vector Core ) . To evaluate the strength of evoked NMDAR currents , we measured both spontaneous and optogenetically evoked synaptic currents in neurons at DIV 21–23 using whole-cell patch-clamp recording at 50 mV . Expression of EYFP-tagged EphB2 WT and Y504E both caused a significant increase in amplitude of the NMDAR-dependent component of excitatory postsynaptic currents ( EPSCs ) compared to the control or the Y504F mutant ( Fig 5B and 5C; NMDAR component measured 30 milliseconds after the evoked EPSC peak; ****p < 0 . 001 , ANOVA followed by Fisher’s exact test ) . In addition , the amplitude of Y504E-expressing neurons was significantly higher than that of WT-expressing neurons ( **p < 0 . 02 , ANOVA followed by Fisher’s exact test ) . To determine whether these effects were due to changes to the NMDAR at synaptic sites , spontaneous miniature excitatory postsynaptic currents ( mEPSCs ) were recorded in the presence of tetrodotoxin and blockers of GABAergic channels . Consistent with previous work [14] , no significant changes in mEPSC frequency were observed between conditions ( S6A and S6B Fig ) . As expected , expression of EYFP-tagged EphB2 WT and Y504E both caused a significant increase in amplitude of the NMDAR-dependent component of mEPSC compared to control or Y504F mutants ( S6C and S6D Fig ) . These changes in mEPSC amplitude are attributable specifically to the increase of synaptic NMDARs because treatment with the NMDAR antagonist D-2-amino-5-phosphonovalerate ( D-APV ) ( 50 μM ) blocked the effects of EphB overexpression and reduced mEPSC amplitude to a similar size in all transfection conditions ( S6E and S6F Fig; NMDAR component measured 15–25 milliseconds after the mEPSC peak; p < 0 . 01; ANOVA followed by Fisher’s exact test ) . Cumulative probability histograms of events from neurons transfected with EphB2 Y504E or EphB2 Y504F indicate that the NMDAR-dependent component of mEPSC amplitude is greater in EphB2 Y504E than in EphB2 Y504F—expressing neurons ( S6F Fig , p < 0 . 0001 , Kolmogorov—Smirnov [K–S] test ) . APV blocks all NMDARs , but previous work suggests that EphBs may preferentially act on GluN2B-containing NMDARs in younger neurons [14] . Therefore , we asked whether mutations to EphB2 at Y504 that enhance ( Y504E ) or prevent ( Y504F ) the EphB—NMDAR interaction might drive or exclude GluN2B-containing NMDAR from synapses . We selectively blocked GluN2B-containing NMDAR with the GluN2B-specific antagonist Ro 25–6981 ( Ro-25 ) and determined the effect on the NMDAR component of mEPSCs in cultured cortical neurons . Whole-cell patch-clamp recordings of mEPSCs revealed that NMDAR-dependent currents in neurons transfected with EphB2 Y504E decreased significantly with selective inhibition of GluN2B-containing NMDARs ( Fig 5D and 5E , p < 0 . 001 , K–S test ) . However , Ro-25 treatment had no significant effect on mEPSC amplitude in Y504F-overexspressing neurons ( Fig 5D and 5F ) . These data suggest that EphB2 preferentially recruits GluN2B-containing NMDARs to synaptic sites . We next asked whether the accumulation of NMDARs at synaptic sites is altered by expression of EphB2 Y504 mutants in mature neurons . To test this , we conducted 2-color super-resolution imaging using stimulated emission depletion ( STED ) microscopy of FLAG-EphB2 or FLAG-EphB2 mutants ( Y504E and Y504F ) and endogenous GluN1 combined with 2-color confocal imaging of vGlut1 and cell-filling EGFP expressed in DIV 21–23 cortical neurons [34] . EphBs are required for normal numbers of dendritic spine synapses and the proper number of NMDARs at synaptic sites [13] . Therefore , we focused on analyzing spine synapses . Synapses were defined by the presence of 1 or more vGlut1 puncta that contacted a spine ( Fig 5G ) . Consistent with our previous findings and physiological data , the percentage of vGlut1-positive spines did not differ between the transfection conditions , with approximately 80% of imaged spines containing vGlut1 puncta ( Fig 5I ) [14] . Using 2-color STED super-resolution imaging , we determined the proportion of vGlut1-positive spines containing FLAG-EphB2 and GluN1 puncta . Approximately 30% of spines contained FLAG-EphB2 WT puncta that colocalized with 1 or more puncta of GluN1 ( Fig 5G , 5H and 5J ) . We next examined the impact of the expression of FLAG-EphB2 Y504E . FLAG-EphB2 Y504E interacts with the NMDAR independent of ephrin-B ligand activation; therefore , we expect that expression of FLAG-EphB2 Y504E would increase the number of synapses where EphB2 and the GluN1 subunit of NMDAR colocalize . Indeed , STED analysis revealed that both the fraction of spines containing colocalized FLAG and GluN1 puncta and the fraction of spines with GluN1 colocalization with EphB2 at synaptic sites were significantly increased in Y504E-expressing neurons compared with EphB2 WT- or Y504F-expressing neurons ( *p < 0 . 05 , **p < 0 . 01 , ANOVA followed by Fisher’s exact test; Fig 5G , 5H , 5J and 5K ) . Interestingly , in these experiments , expression of EphB2 Y504F did not appear to act in a dominant negative manner . Although we knocked down endogenous EphB2 , the lack of an effect of EphB2 Y504F might be due to the presence of other EphB proteins or might reflect the presence NMDAR-interacting proteins such as PSD-95 . Regardless , these findings indicate that phosphorylation of Y504 enables EphB2 to recruit or retain NMDARs to spine synapses . Despite normal levels of NMDARs in EphB triple-knockout mice , in these animals NMDARs are redistributed from synaptic sites to extrasynaptic locations , suggesting that EphBs play an important role in regulating NMDAR localization [14] . To test whether Y504 might regulate the recruitment of NMDARs in vivo , neurons in the dorsal horn of the spinal cord were transduced by intrathecal injection of lentivirus coding for either EYFP-tagged EphB2 WT or the constitutively interacting EYFP-tagged EphB2 Y504E mutant ( Fig 6A and 6B ) . This method results in transduction of approximately 60% of NeuN-positive cells in the region near the injection site [35] . Consistent with our previous findings , injection of the virus resulted in a focal infection of neurons ( NeuN-positive cells , Fig 6C ) within the dorsal aspect of the spinal cord near the injection site ( Fig 6B–6D ) . The superficial nociceptive layers of the dorsal horn are vGlut1-negative and vGlut2-positive [36 , 37] . To determine whether transduction of EYFP-tagged EphB2 results in changes in NMDAR expression within the nociceptive region of the spinal cord , sections were stained for GluN1 and vGlut2 . In mice transduced with EYFP-tagged EphB2 WT or Y504E , the expression of the GluN1 subunit of the NMDAR was significantly up-regulated in the superficial layers of the dorsal horn compared to control and Y504F-injected mice ( p < 0 . 05 , ANOVA followed by Tukey’s range test; Fig 6D and 6E ) . In addition to the increase in NMDAR levels , there was also a significant increase in vGlut2 intensity ( p < 0 . 0005 , ANOVA followed by Tukey’s range test; Fig 6D and 6F ) . These changes in presynaptic vesicle markers are consistent with the known function of EphB2 in the induction of presynaptic terminal formation . Evidence suggests that different domains might mediate the role of EphB2 in synapse formation and the EphB—NMDAR interaction , with the ephrin-B binding domain of EphB2 being essential for induction of presynaptic differentiation [12 , 38] . Consistent with this model , transduction of EphB2 Y504F resulted in a significant increase in the intensity of vGlut2 but not in GluN1 levels ( Fig 6E and 6F ) . Regardless , these findings suggest that Y504 of EphB2 regulates the localization of the NMDAR in the spinal cord . Ephrin-B—EphB signaling to the NMDAR is implicated in pain plasticity , leading to pathological pain states [39 , 40] . Given the impact of overexpression of EphBs on NMDAR levels in the dorsal synaptic region of the spinal cord that is associated with pain , we next examined whether the EphB—NMDAR interaction might be enhanced in models of pain . Consistent with previous reports on the role of ephrin-Bs in induction of pain and pain plasticity [40 , 41] , intrathecal injection of activated ephrin-B2 , which activates EphBs and the EphB—NMDAR interaction [11] ( Fig 7A , 0 . 2 μg ) , induced robust , sustained mechanical hypersensitivity in mice ( p < 0 . 0001 , ANOVA followed by Dunnett’s multiple comparison test; Fig 7B ) . Thus . injection of activated ephrin results in hypersensitivity . Because EphB1 is expressed more highly than EphB2 in the spinal cord and is known to interact with the NMDAR ( S7C Fig ) [23 , 42 , 43] , we tested whether the region homologous to Y504 in EphB2 was also important for the previously described EphB1–NMDAR interaction . EphB1 contains a homologous tyrosine residue at Y502 ( Fig 1D ) . Therefore , we hypothesized that the charge of this residue might regulate the EphB1–NMDAR interaction . Consistent with this model , compared to FLAG-EphB1 , WT co-expression of FLAG-EphB1 Y502E with GluN1 and GluN2B in HEK293T cells resulted in increased coimmunoprecipitation , while the interaction was significantly reduced when FLAG-EphB1 Y502F was coexpressed ( Fig 7C and S7A Fig ) . We next asked whether EphB1 is also phosphorylated at Y502 . HEK293T cells were transfected with either EphB1 WT or EphB1 Y502F , and lysates were probed with the p*Y504 antibody . Consistent with selective phosphorylation of Y502 , only WT EphB1 was recognized by the p*Y504 antibody ( S7B Fig ) . These findings suggest that the ability of both EphB1 and EphB2 to interact with the NMDAR is mediated by phosphorylation of homologous tyrosine residues ( Y502 and Y504 respectively ) . Next , we asked whether injury-induced mechanical hypersensitivity might induce the EphB—NMDAR interaction ( Fig 7D ) . Consistent with a role for the EphB—NMDAR interaction , unilateral plantar incision ( a mouse model sensitive to NMDAR receptor blockade [44] ) resulted in a significant increase in the coimmunoprecipitation of the GluN1 subunit of the NMDAR with EphB1 ( 51 . 3 ± 26 . 4-fold increase , p < 0 . 05 , Mann—Whitney U test , 8 hemispinal cords pooled per sample , n = 3 , Fig 7E and 7F ) . In the spinal cord , detection of phosphorylation of EphB1 and EphB2 was difficult because of the small amounts of tissue available from each animal and levels of EphB expression . However , by pooling samples we were able to observe modest effects on EphB2 Y504 and Y502 phosphorylation in spinal cord samples after injury ( S7C Fig ) . Importantly , the increase in the EphB—NMDAR interaction was only observed in tissue from the same side as the injury , indicating that this effect is likely specific to the injury paradigm . Next , to begin to test whether extracellular phosphorylation might be important for the behavioral response to injury , we tested whether blockade of extracellular kinase activity with the membrane-impermeant drug K252b might rescue injury-induced mechanical hypersensitivity ( Fig 7D ) . We used intrathecal infusion , a method allowing for local infusion of drugs and widely used clinically to allow for spinal cord—selective treatment [45] . Intrathecal injections of K252b ( 1 μg ) at the time of plantar incision significantly reduced mechanical hypersensitivity for up to 24 hours after incision in adult mice ( Fig 7G ) . These results support a model in which inhibiting extracellular phosphorylation can reduce pain signaling and mechanical hypersensitivity behavior ( S8 Fig ) . To begin to test whether the EphB—NMDAR interaction might play a role in pathological pain , we asked whether inducing the EphB—NMDAR interaction with transduction of EYFP-tagged EphB2 WT or Y504E that results in increased levels of NMDAR in the spinal cord ( Fig 6 ) might result in mechanical hypersensitivity ( Fig 7H ) . Intrathecal injection of lentivirus transducing either EYFP-tagged EphB2 WT or Y504E resulted in a long-lasting enhancement of mechanical sensitivity ( Fig 7I and 7J ) . Consistent with in vivo and in vitro immunostaining and physiological data ( Figs 5 and 6 ) , we find that intrathecal injections of the Y504F mutant did not result in a long-lasting increase in mechanical hypersensitivity ( Fig 7I ) . Thus , transduction of EYFP-tagged EphB2 WT or Y504E results in mechanical hypersensitivity and increased levels of GluN1 in regions of the spinal cord linked to pathological pain ( Figs 6 and 7I ) . Injections of lentivirus transducing WT EphB2 and EphB2-Y504E resulted in long-lasting mechanical hypersensitivity , even 2 months after virus injection ( Fig 7I and 7J ) . To test whether these effects are due to phosphorylation of Y504 , we asked next whether inhibition of extracellular phosphorylation might block the effects of EphB2 expression . If mechanical hypersensitivity depends on phosphorylation of Y504 , then blocking extracellular phosphorylation should only reduce or rescue mechanical hypersensitivity in mice transduced with WT but not in those with Y504E EYFP-tagged EphB2 . To avoid potential effects of inhibiting the EphB—NMDAR interaction outside of the spinal cord [45] , intrathecal injections of K252b were made near the site of viral injection . Remarkably , intrathecal injection of K252b resulted in rapid ( <3 hours ) and sustained ( >24 hours ) reversal of mechanical hypersensitivity compared to control levels when injected into mice overexpressing WT EYFP-tagged EphB2 ( p < 0 . 01 , 2-way ANOVA followed by Dunnett’s multiple comparison test; Fig 7J and 7K ) . Importantly , the rescue of mechanical hypersensitivity to control levels in EphB2 WT—transduced mice ( Fig 7K ) was found in animals having experienced increased levels of pain for 2 months . We next conducted a series of controls . Although we did not observe evidence of action of K252b outside of the spinal cord , and intrathecal infusion of drugs is used clinically to avoid side effects seen with other injection methods [45] , the drug used to block the interaction ( K252b ) could act elsewhere in the brain . However , injection of K252b had no effects in control EGFP-expressing mice ( Fig 7K ) . Next , to control for effects of K252b injection outside of the spinal cord and to demonstrate the specific action of this compound on EphB2 with an intact extracellular phosphorylation site , we asked whether K252b might alter the mechanical hypersensitivity in mice expressing the constitutively interacting EphB2 Y504E mutant . Importantly , in mice expressing the EYFP-tagged EphB2 Y504E construct , injection of K252b did not block mechanical hypersensitivity ( Fig 7K ) . Thus , K252b , which inhibits extracellular phosphorylation of Y504 , is acting selectively within the spinal cord in WT EphB2-overexpressing mice to block mechanical hypersensitivity . Taken together , these findings support a model wherein the EphB—NMDAR interaction mediated by extracellular phosphorylation of EphB cFN3 induces amplification in pain signaling , causing mechanical hypersensitivity ( S8 Fig ) .
Phosphorylation of protein ectodomains can regulate Drosophila wing development , neurite outgrowth , osteoclast migration , and aggregation of amyloid beta [46–49]; however , these events appear to occur largely constitutively and are not known to be regulated by cell—cell signaling . In contrast , phosphorylation of Y504 on EphB2 is induced by ephrin-B and results in the recruitment of NMDAR to synaptic sites . The activity that mediates extracellular phosphorylation of EphB2 is present in ACSF of cultured neurons after eprhin-B2 stimulation and requires ATP . Importantly , ample ATP is found at synaptic terminals [50] and ATP can be released into the extracellular space by synaptic or secretory vesicles [51 , 52] . Together with recent work demonstrating that a wide variety of proteins are constitutively extracellularly phosphorylated [4 , 5] our data suggest a model in which , similar to intracellular tyrosine phosphorylation , inducible extracellular tyrosine phosphorylation of specific amino acids can regulate extracellular domain—mediated protein—protein interactions . Intracellular phosphorylation sites enable protein—protein interactions with specific amino acid motifs . Similarly , the conserved region surrounding the Y504 residue on EphB2 appears to define a new extracellular interaction domain . Based on homology , we could accurately predict that EphA8 would also interact with the NMDAR . In contrast , the corresponding regions of EphA3 and EphA4 that do not co-IP with the NMDAR are less homologous to EphB2 [11] , although we have not ruled out the possiblity that they may also become phosphorylated . In addition , mutation of the homologous amino acid on EphB1 ( Y502 ) can drive increased or reduced interactions with the NMDAR . These data suggest that the motif XpYVXQVR may be important for interactions between the extracellular domains of proteins . Phosphoprotein databases provide additional examples of phosphorylated tyrosine residues with unknown function in homologous regions of FN3 domains of proteins linked to synaptogenesis , axon guidance , and target recognition , such as Sidekicks and DSCAML1 ( S1F Fig ) . It will be important to determine whether phosphorylation of the FN3 domain of other proteins also regulates protein—protein interactions and whether extracellular phosphorylation is an underappreciated mechanism contributing to the development and function of the nervous system and synapse . How might extracellular phosphorylation enable protein—protein interaction ? In the intracellular space , changes in the charge of tyrosine residues in specific amino acid motifs control the binding of specific SH2 domain—containing molecules [1] . Interestingly , the EphB2 Y504F constructs fail to interact with the NMDAR but are likely not functioning in a dominant negative manner . These findings are consistent with a model in which the EphB—NMDAR interaction is mediated by charge because phenylalanine is not a positively charged amino acid . The role of charge is difficult to test , as positively charged amino acids are bulky and may affect protein—protein interactions in other ways , such as steric hindrance . However , a dominant negative mutant should be able to be designed using surface charge analysis of the region surrounding Y504 . It is likely that phosphorylation of the Y504 residue would alter the secondary structure or interaction surface of the FN3 domain . Supporting this idea , analysis of the crystal structures of Eph family members indicates that Y504 of EphB2 is exposed and suggests that the cFN3 of Eph family members , but not the N-terminal FN3 ( nFN3 ) domain , might have a pocket for binding to other molecules [53] . Indeed , the crystal structures of ligand-unbound EphA4 and ligand-bound EphA4 indicate that ligand binding results in changes to the structure of cFN3 [54] . These data suggest that phosphorylation of Y504 could modify FN3 structure . Consistent with this model , phosphorylation of Y504 in EphB2 is necessary and sufficient to induce binding to the NMDAR , and we find that homologous FN3 domains and tyrosine residues can be phosphorylated . However , detailed structural analysis will be required to resolve these questions . FN3 domains of other synaptic proteins contain a homologous residue , suggesting that extracellular phosphorylation is a novel mechanism that may mediate many events related to synaptic function and behavior . Extracellular phosphorylation of EphB2 at Y504 stabilizes GluN2B-containing NMDARs on the cell surface and drives the accumulation of GluN2B-containing NMDARs at synaptic sites in young neurons . During development , GluN2B-containing NMDARs predominate at synaptic sites , and the presence of GluN2B-containing NMDARs at synapses can enhance synaptic plasticity [55] . Interestingly , phosphorylation of EphB2 at Y504 is down-regulated as synapses mature , and the proportion of GluN2B-containing NMDARs is reduced . EphB2 regulates a number of other aspects of GluN2B-containing NMDAR function . EphB2 activation can drive the Src kinase—dependent phosphorylation of GluN2B on specific tyrosine residues that prevent AP2-mediated internalization [56] . The kinase-active EphB2 reduces calcium-dependent inactivation of GluN2B-containing ( but not GluN2A-containing ) NMDAR currents [14] . However , in adult animals , EphBs are required for normal synaptic levels of both GluN2A and GluN2B , suggesting a role for additional developmentally regulated mechanisms . It remains to be determined whether the ability of EphB2 to interact with GluN2A-containing NMDAR relies on the same Y504 site or a different mechanism . Further indication of the importance of extracellular modification of the EphB2 receptor is suggested by findings in human disease models . Disruption of the EphB—NMDAR interaction in models of Alzheimer disease [19] or in NMDAR encephalitis [18] that results in reduced surface levels of NMDAR may involve the FN3 domain of the EphB2 protein , further supporting the importance of understanding phosphorylation of these domains . The kinase regulating the phosphorylation of Y504 on EphB2 remains to be identified . Interestingly , Y504 is not phosphorylated on the kinase-dead version of EphB2 ( EphB2 KD ) implicating the kinase domain of EphB2 . It is not likely , though , that the ephrin-B—induced extracellular phosphorylation of EphB2 Y504 is phosphorylated directly by the EphB2 kinase because ( 1 ) p*Y504 requires extracellular ATP , ( 2 ) p*Y504 occurs on the cell surface , ( 3 ) p*Y504 is blocked by a membrane-impermeable kinase inhibitor K252b , ( 4 ) extracellular treatment with phosphatases augments the interaction , ( 5 ) an activity that can phosphorylate Y504 is secreted following ephrin-B treatment , and ( 6 ) although K252b effectively inhibited p*Y504 , it failed to block EphB2 kinase activity . However , while the EphB2 kinase is unlikely directly responsible for phosphorylating Y504 , given that p*Y504 requires ephrin-B activation of EphB2 , it is likely that the EphB2 kinase participates in the signal transduction pathway necessary for p*Y504 . Further work will be needed to identify the genuine Y504 kinase . One interesting set of candidates is the recently identified family of extracellular kinases ( such as Fam20C ) that function in the secretion pathway [46] . A second class of candidates is represented by the tyrosine kinase VLK , which can be secreted from platelets and is an essential gene [6 , 7] . The mechanism of action of this family is unknown , but since these kinases can function extracellularly , they might be good candidates for the genuine p*Y504 kinase , which appears to be secreted . While disorders like Alzheimer disease and NMDAR-encephalitis may be linked to pathological reduction of the EphB—NMDAR interaction [18 , 21] , induction of hypersensitivity and pain can occur by enhancing EphB-dependent effects on NMDAR function [39] . Indeed , increases in the amount of ephrin-B ligand have been reported in the spinal cord in pain models . Consistent with a role for the EphB—NMDAR interaction are the following: ( 1 ) MK801 blocks the effect of ephrin-B1 on pathological pain [40] , ( 2 ) local intrathecal application of EphB-Fc blocks pathological pain [24 , 42] , and ( 3 ) Nav1 . 8+ nociceptive sensory neuron ( DRG ) -specific ephrin-B2 knockout regulates inflammatory and neuropathic pain [57] . These findings suggest that in the spinal cord and periphery , EphB-dependent modulation of NMDAR function drives NMDAR-dependent hyperexcitability of spinal circuits , causing pathological pain associated with traumatic nerve injury and cancer-induced nerve damage [25 , 41 , 58 , 59] . Consistent with these previous findings , we show that a single intrathecal injection of activated ephrin-B2 results in hypersensitivity . In addition to functions in the induction of pathological pain , we find that the EphB—NMDAR interaction also appears to play an important role in maintaining long-lasting mechanical hypersensitivity . Virally induced expression of WT EphB2 in dorsal horn neurons results in sustained mechanical hypersensitivity . Importantly , injection of K252b 2 months after induction by an EphB2-expressing virus rapidly alleviates this mechanical hypersensitivity . These findings suggest that inhibition of extracellular phosphorylation might provide a target suitable to alleviate more chronic pain states . In animals injected with virus transducing the EphB2 Y504E point mutant that induces a constitutive , ligand-independent interaction with the NMDARs , injection of K252b did not influence this established mechanical hypersensitivity . This underscores the specific , local effect of extracellular EphB2 phosphorylation . Consistent with clinical practices [45] , our findings indicate that intrathecally injected K252b is acting selectively within the spinal cord on WT EphB2 overexpressing cells and not likely targeting regions outside of the spinal cord . In addition to blocking EphB2-EYFP—induced mechanical hypersensitivity , K252b reduced incision-induced mechanical hypersensitivity , suggesting a key role for EphB—NMDAR interactions in this context , a notion supported by biochemical findings in our models . Although K252b is not itself a suitable compound for further development , these data suggest that extracellular phosphorylation and inhibition of the EphB—NMDAR interaction may provide potential therapeutic targets for the mitigation of pathological pain . In summary , extracellular phosphorylation of EphB mediates the direct extracellular EphB—NMDAR interaction to direct NMDARs to synaptic sites in a variety of neuronal types and regulates behaviorally relevant events such as pathological pain .
All animal procedures were approved by the Institutional Animal Care and Use Committee of Thomas Jefferson University , the University of Pennsylvania , the University of Texas at Dallas , and the University of Arizona and were in accordance with International Association for the Study of Pain guidelines ( protocol numbers: 01286 and 01797 TJU , 14–04 UTD , 802645 UPenn and 09–115 UA ) . Embryonic day 17 ( E17 ) to E18 Long—Evans rats ( Charles River ) were used for preparing dissociated cortical neural culture as previously described [13 , 14] . Postnatal day 30 wild-type CD1 mice ( Charles River ) were used for synaptosome preparation as previously described [14] . Male ICR mice ( Harlan ) were used for all behavioral studies . Mice were used for behavioral experiments starting at 8–12 weeks of age . For LC-MS/MS analysis , NG108 cells were transfected with FLAG-tagged EphB2 receptor , treated with ephrin-B1 , and lysed . Samples immunoprecipitated with α-FLAG antibody were separated by SDS-PAGE , the gels were stained with Coomassie Blue , and the EphB2 band was excised and digested with trypsin in-gel . After enrichment of phosphopeptides using TiO2 , LC-MS/MS analysis was conducted ( Thermo Fisher Scientific LTQ-Orbitrap ) . Generation of FLAG-tagged full-length EphB2 , truncated EphB2 ( fEphB2 Tr ) , and kinase-dead ( KD; K663R ) EphB2 were previously described [11] . Generation of EphB2-EYFP was previously described [60] . Single amino acid point mutations to Y481 and Y504 were introduced using sequence-specific primers and site-directed mutagenesis ( Stratagene , La Jolla , CA ) . EphB2 expression constructs were generated based on published sequences . Expression of each EphB2 version used was validated using RT-PCR from mouse brain cDNA . In some cases , the RT-PCR products were validated by sequencing . FLAG- and EYFP-tagged EphB2 expression constructs gave similar results in assays tested . Lentiviruses were produced and purified by the Gene Therapy Program at the Penn Vector Core facility of the University of Pennsylvania . HEK293T cells were transfected using the calcium phosphate method . Dissociated cortical neurons were prepared from embryonic day 17 ( E17 ) to E18 rats . Dissociated spinal cord neurons were prepared from E14 . 5 rats . Cultured neurons were transfected at DIV 3 or 14 using Lipofecatmine 2000 . Immunoprecipitations and western blot analysis were performed as previously described with small changes [11] . Cell-surface biotinylations were performed as previously described [14 , 38] . For phosphotyrosine surface staining , live neurons were incubated for 10 minutes at room temperature in PBS with α-phosphotyrosine , then washed and fixed . For synaptic NMDAR staining , cells were fixed and permeabilized with 0 . 1% saponin at room temperature . For immunohistochemical analysis , spinal cord tissues were frozen in OCT compound and cut into sections and then fixed and stained . Images were obtained using Leica TCS SP5 confocal scanning microscopy and Leica TCS SP8 STED microscopy . Analysis was done using NIH ImageJ . All data shown as mean ± SEM . Phospho-specific antibodies against α-EphB2 Y504 were generated against a phosphopeptide of the sequence Ac-CKGLKAGAI-pY-VGQVRA-NH2 ( Covance , Denver , PA ) . Synaptosomes were prepared as previously described [14] . Electrophysiological recordings from DIV 21–23 cultured rat cortical neurons were performed using whole-cell patch methods as previously described [14] . Evoked currents were generated by light stimulation ( 470 nanometers ) of neurons expressing channelrhodopsin-2 . Behavioral testing and drug administration were performed as described previously [35] .
|
The activity of proteins can be finely and reversibly tuned by post-translational modifications . The attachment of phosphate groups to tyrosine residues is one of such modifications . While the existence of extracellular phosphoproteins has been known , the functional significance of extracellular phosphorylation is poorly understood . Here we describe a single extracellular tyrosine whose inducible phosphorylation may represent an archetype for a new class of mechanism mediating protein—protein interaction and regulating protein function . We show that the interaction between EphB2—which occurs upon receptor activation by its ligand ephrin-B—and the N-methyl-D-aspartate receptor ( NMDAR ) depends on extracellular phosphorylation of EphB2 . This interaction regulates the localization of the NMDA receptor to synaptic sites in neurons . In vivo , EphB2 is phosphorylated in response to injury , and the subsequent up-regulation of the interaction between EphB2 and NMDA receptors enhances injury-induced pain behavior and mechanical hypersensitivity in mice . Importantly , our study defines a specific extracellular phosphorylation event as a mechanism driving protein interaction and suggests that extracellular phosphorylation of proteins is an underappreciated mechanism contributing to the development and function of the nervous system and synapse .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"phosphorylation",
"medicine",
"and",
"health",
"sciences",
"protein",
"interactions",
"chemical",
"compounds",
"nervous",
"system",
"enzymes",
"immunology",
"enzymology",
"neuroscience",
"organic",
"compounds",
"clinical",
"medicine",
"tyrosine",
"immunoprecipitation",
"amino",
"acids",
"hypersensitivity",
"research",
"and",
"analysis",
"methods",
"aromatic",
"amino",
"acids",
"spinal",
"cord",
"animal",
"cells",
"proteins",
"protein",
"kinases",
"chemistry",
"precipitation",
"techniques",
"tyrosine",
"kinases",
"biochemistry",
"cellular",
"neuroscience",
"neuroanatomy",
"cell",
"biology",
"post-translational",
"modification",
"anatomy",
"organic",
"chemistry",
"clinical",
"immunology",
"neurons",
"hydroxyl",
"amino",
"acids",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"physical",
"sciences"
] |
2017
|
Extracellular phosphorylation of a receptor tyrosine kinase controls synaptic localization of NMDA receptors and regulates pathological pain
|
Marburg virus ( MARV ) , a zoonotic pathogen causing severe hemorrhagic fever in man , has emerged in Angola resulting in the largest outbreak of Marburg hemorrhagic fever ( MHF ) with the highest case fatality rate to date . A mobile laboratory unit ( MLU ) was deployed as part of the World Health Organization outbreak response . Utilizing quantitative real-time PCR assays , this laboratory provided specific MARV diagnostics in Uige , the epicentre of the outbreak . The MLU operated over a period of 88 days and tested 620 specimens from 388 individuals . Specimens included mainly oral swabs and EDTA blood . Following establishing on site , the MLU operation allowed a diagnostic response in <4 hours from sample receiving . Most cases were found among females in the child-bearing age and in children less than five years of age . The outbreak had a high number of paediatric cases and breastfeeding may have been a factor in MARV transmission as indicated by the epidemiology and MARV positive breast milk specimens . Oral swabs were a useful alternative specimen source to whole blood/serum allowing testing of patients in circumstances of resistance to invasive procedures but limited diagnostic testing to molecular approaches . There was a high concordance in test results between the MLU and the reference laboratory in Luanda operated by the US Centers for Disease Control and Prevention . The MLU was an important outbreak response asset providing support in patient management and epidemiological surveillance . Field laboratory capacity should be expanded and made an essential part of any future outbreak investigation .
Marburg virus ( MARV ) is classified as members of the family Filoviridae , genus Marburgvirus , type species Lake Victoria marburgvirus . A single species has been described which includes several virus strains [1] . Today , the geographic distribution of MARV seems to primarily involve areas in East Africa within 500 miles of Lake Victoria , Zimbabwe , but also western Africa [2] , [3] . MARV is of zoonotic nature with an as yet unidentified reservoir in nature , but with strong cumulative evidence that bats are involved in the zoonotic cycle [4] , [5] as this has also been implicated for Ebola virus [6] . MARV is the causative agent of Marburg hemorrhagic fever ( MHF ) , a disease that was first described in 1967 among laboratory workers in Germany and former Yugoslavia [7]–[9] . Until 1998 , only sporadic MHF cases have occurred in Zimbabwe/South Africa ( 1975 ) and in Kenya ( 1980 & 1987 ) [10]–[12] . The first community-based MHF outbreak was reported in 1998–2000 from the Watsa/Durba region in the Democratic Republic of the Congo ( DRC ) [13] , [14] . In 2004/2005 MARV first appeared in western Africa , Angola , causing to date the largest MHF outbreak on record [15] , [16] . The latest MHF episodes involved 4 reported cases from western Uganda associated with a single mine ( 2007 ) [5] , and two imported cases into the US and the Netherlands , who independently visited the same cave in Uganda ( 2008 ) [17] , [18] ( Table 1 ) . In addition , three laboratory exposures , one of them fatal , have been reported [9] , [19] , [20] . In March 2005 , the National Microbiology Laboratory ( NML ) of the Public Health Agency of Canada ( PHAC ) offered assistance to the World Health Organization ( WHO ) as a partner of the ‘Global Outbreak Alert & Response Network’ ( GOARN ) ( http://www . who . int/csr/outbreaknetwork/en/ ) for the MHF outbreak in Angola . Under GOARN , a Mobile Laboratory Unit ( MLU ) was deployed to Uige , the epicentre of the outbreak , to assist in clinical management and epidemiological surveillance with MARV-specific and limited differential diagnostic capacity . Here we discuss the usefulness of this latest response capacity for the management of viral hemorrhagic fever outbreaks .
Laboratory space was made available for the MLU in the Paediatric Ward of the Uige Provincial Hospital ( Figure 1 ) . Four rooms were used for the laboratory set up to ensure isolation of infectious work from other activities and to separate PCR assay steps to minimize contamination . Two rooms were located on one side of a central hallway; the smaller of the two rooms was accessible by a single door and had no windows or other opening and was utilized for infectious work ( ‘hot room’ ) . The anteroom to this room was used for the preparation for entry to the infectious room and the subsequent disinfection of the worker following infectious work . Opposite these rooms were two additional rooms; one was used for RNA extraction and running the Q-RT-PCR and the other room was utilized as a ‘clean room’ for master mix preparation . Reagents and the laboratory team ( 2–3 members ) were replaced every three weeks; in total NML deployed six teams to Angola to cover the period of April 1 to June 27 , 2005 . Clinical samples were collected by personnel wearing personal protective equipment ( PPE ) including a surgical mask , cap , shield or goggles , gown , apron , gloves ( two pairs ) and boots . Swab samples ( nasal and oral ) were collected using cotton tipped applicators ( AMG Medical , VWR , Mississauga , ON , Canada ) . Applicator tips were stored in 700 µl of Dulbecco's modified essential medium ( DMEM ) or phosphate buffered saline ( PBS ) supplemented with 5% bovine serum albumin ( Invitrogen , Burlington , ON , Canada ) . Whole blood and serum samples were collected using EDTA and serum vacutainer tubes , respectively . For transport , tubes were sealed in plastic bags , surface disinfected with a 1% hypochlorite solution , sealed into a second bag or container and again surface disinfected . Collection of human specimens occurred on an outbreak response protocol and was approved by the local Scientific and Technical Coordination Committee in Uige , Angola . Infectious specimens were manipulated in the field laboratory by personnel wearing Tyvek suits and HEPA filter-equipped powered air purifying respirators , in a room isolated and dedicated for this work ( Figure 1 ) . An aliquot ( 140 µl ) was removed from each sample and inactivated by adding 560 µl of the guanidine thiocyanate lysis buffer AVL . The sample tubes were submerged in 1% hypochlorite solution for 10 minutes and released from the infectious area . All further work was performed with PPE as outlined above . For RNA isolation we used the QIAamp Viral RNA mini kit ( Qiagen , Mississauga , ON , Canada ) . All waste material was treated with 1% hypochlorite solution and incinerated on the same day . Two separate sample aliquots were prepared for transportation to the reference laboratory in Luanda operated by the Special Pathogens Branch of the US Centers for Disease Control and Prevention ( US-CDC ) . Remaining samples were forwarded to the National Institute for Communicable Diseases ( NCID ) , Sandringham , South Africa , and finally shipped to the US-CDC ( Atlanta ) or NML ( Winnipeg ) for further testing . Transportation was carried out in compliance with International Air Transport Association ( IATA ) regulations after prior approval by the appropriate national authorities of the sending or receiving countries . Initially , two quantitative real-time PCR ( Q-RT-PCR ) assays were used that targeted regions of the polymerase ( L ) [MARVLF-TTATTGCATCAGGCTTCTTGGCA , MARVLR–GGTATTAAAAAATGCATCCAA ( AY358025; bp . 13321–133517 ) ] and the glycoprotein ( GP ) genes [MARVGPF–AAAGTTGCTGATTCCCCTTTGGA , MARVGPR–GCATGAGGGTTTTGACCTTGAAT ( AY358025; bp . 6131–6355 ) ] . Later , an assay that targeted the nucleoprotein ( NP ) gene [MARVNPF–TGAATTTATCAGGGATTAAC , MARVNPR–GTTCATGTCGCCTTTGTAG ( AY358025; bp . 967–1146 ) ] was used in place of the GP assay . The switch to an NP target was the result of testing that indicated this target was potentially more sensitive and provided a more distinct melting curve which simplified interpretation . MARV RNA was detected using the Lightcycler RNA Amplification SYBR Green I kit ( Roche , Laval , PQ ) . Briefly , 5 µl of RNA was added to 20 µl of master mix containing 1X SYBR Green I mix , 5 mM MgCl2 , 0 . 6 µM forward and reverse primers and 0 . 5 µl of the enzyme mix . Q-RT-PCR assays were run on Smartcycler thermocyclers ( Cepheid , Sunnyvale , CA ) . A reverse transcriptase step at 50°C for 20 minutes and a 2 minute inactivation step at 94°C were followed by 40 cycles at 94°C for 15 seconds , 50°C for 30 seconds and 72°C for 30 seconds where a single acquisition point was taken . Melt curve analysis was performed to confirm the identity of amplification products . Samples were considered positive if they produced melting point confirmed amplification products in two assays . Amplification products were later confirmed by sequencing at NML ( Winnipeg ) .
The algorithm for the laboratory testing and the rational for positive/negative test results are presented in Figure 2 . Overall , the MLU tested 620 clinical specimens from 388 patients/individuals over an operation period of 88 days . The clinical specimens included mainly oral swabs and EDTA blood/serum samples; the remainder consisted of nasal and conjunctival swabs and breast milk . The sample source and test results of individuals tested are presented in Table 2 . The daily case load of the MLU fluctuated , with the number of individuals analyzed per day varying between 0 and 14 ( Figure 3 ) . This analysis often included multiple samples per individual on a single day and serial surveillance sampling of suspect and confirmed cases . The age and sex distribution of individuals tested were slightly shifted towards females ( 68% ) and the younger age groups , in particular children under the age of 5 years ( by far the largest single age group at 21% ) . The distribution of positive cases clearly demonstrated a larger proportion of females and children among the infected individuals ( Figure 4 ) . A comparison of detection of MARV from oral swabs and EDTA blood was performed on 63 individuals from whom both specimen types were available from the same day . Both samples sources yielded identical test results in 98 . 5% of the individuals with roughly 33% positive and 66% negative for MARV . Cycle threshold ( Ct ) values for most paired samples did not differ markedly indicating similar viral loads in both specimen sources ( Figure 5 ) . Testing on some patients did provide disparate results for blood and swab samples but test results were identical even in these instances . Similarly , for 12 individuals , both oral and nasal swabs specimens were collected which resulted in identical test results and no significant differences in Ct values for the positives . Additionally , 3 breast milk specimens from laboratory-confirmed female MHF cases were analyzed and shown to be positive for MARV ( data not shown ) . We did not experience any evidence for PCR contamination during the entire operation . All controls produced the expected positive and negative results . Nevertheless , all samples tested in Uige were subsequently shipped to Luanda for confirmation at a US-CDC established biosafety level 3 ( BSL3 ) laboratory using a real-time PCR hybridization assay targeting the matrix protein ( VP40 ) gene , an antigen capture enzyme-linked immunorsorbent ( ELISA ) assay and antibody ( IgM and IgG ) detection ELISAs [16] . Overall , the reference laboratory confirmed test results of the MLU in 97 . 5% of all specimens analyzed and in all but one case . The high concordance between field and reference laboratory results supported the on-site report of the MLU results to the ward and the surveillance teams , allowing a turn-around time of <4 hours from sample receiving to laboratory diagnosis . After closing the MLU , further clinical specimens were shipped to Winnipeg for diagnosis via Luanda ( US-CDC ) and Sandringham ( NCID ) . Eventually , all specimens were shipped to the BSL4 laboratories in Atlanta and/or Winnipeg for additional analysis . Sequence analysis of all amplified products and of several virus isolates obtained at the US-CDC [16] and NML ( authors , unpublished data ) demonstrated a high degree of conservation indicating a single or very few introductions into the community , with subsequent human-to-human transmission . Differential diagnostic testing was only performed for malaria ( Plasmodium spp . ) using a real time PCR assay targeting the ssuRNA gene [21] . Test results for 19 individuals demonstrated two groups of patients , mild or asymptomatic ( Ct values >20 ) and symptomatic individuals ( Ct values <20 ) , based on parasitemia levels ( data not shown ) . The value of this diagnostic tool needs to be further evaluated .
Under current filoviral hemorrhagic fever outbreak operation protocols several activities are undertaken where accurate and rapid diagnostic testing can have significant impact: To obtain diagnostic testing , specimens have normally been shipped to an international reference laboratory such as the Institut de recherche pour le développement ( IRD ) , Franceville , Gabon; NCID in Sandringham , South Africa; or the US-CDC in Atlanta , United States resulting in a significant delay ( days to weeks due to shipment issues ) in laboratory diagnosis with limited or no benefit for acute case patient or outbreak management [22]–[24] . Therefore , such operation protocols require a fairly large infrastructure , longer hospitalization periods , and more staff and consequently increase resources and exposure risks . An MLU , providing testing results in a 4 hour turn around , can be an integral part of the outbreak response and simplify lessen many of the efforts needed to quickly contain and control the outbreak . Laboratory testing of a symptomatic individuals during triage will allow the team to quickly assess if the person is a case or not . Confirmed cases can be appropriately isolated and supportive care initiated . Symptomatic individuals with negative test results can be maintained separate from confirmed cases either by releasing to another ward or kept in an observation ward for follow up testing or discharging . In Uige , and to a lesser extent also at previous outbreak locations , the isolation ward was largely unacceptable to the local population and significant resistance was present to have family members admitted [15] , [25] . However a positive test result for MARV was normally sufficient to convince people of the necessity for admission to the ward . Isolating only those individuals who require it will reduce the infrastructure needed for isolation , minimize the hospitalization time for non-cases , reduce the number of staff and consequently reduce the risk of exposure for both staff and non-cases . Cases that can be confirmed or excluded by laboratory testing can significantly contribute to one of the most important outbreak control measures , contact tracing . The current protocols call for the follow-up of contacts of suspected cases for 21 consecutive days . The presence of a field laboratory can help to arrive at a rapid confirmed final diagnosis for each suspected case , thereby decreasing the burden of field teams who may frequently be conducting contract tracing of cases with uncertain diagnosis . Testing in this outbreak found that oral swabs from severely ill or deceased patients were a suitable sample for MARV testing by Q-RT-PCR . This allowed the MLU to safely test samples from corpses of unknown cause and when possible , to release MARV-negative bodies to the family members for traditional and religious burial procedures , a sensitive issue with almost all local communities in endemic areas . The value of swabs from corpses for diagnostic purposes needs to be further evaluated in future outbreaks and perhaps confirmed by other technologies such as immunohistochemistry [26] . Post mortem RNA degradation might render a test falsely negative even so infectious Ebola virus has been detected in blood samples more than a month after blood draw and storage at room temperature [27] . Any test results should take clinical presentation and epidemiology into account . A growing concern is the return of negative and convalescent patients to the community , which may increase with the implementation of more advanced case patient care and the perspective of treatment options in the future [2] , [24] , [28] . These people are often shunned by their families and neighbours and a timely negative test result as provided through the MLU may aid in their re-acceptance and safe re-introduction into the community . In Angola , field diagnostic support was used for the first time in response to a MHF outbreak . Also the first time , the combined operation of a field and reference laboratory allowed for a unique evaluation of field diagnostic capacity under difficult circumstances and proved it to be accurate , efficient and safe in operation . There have been previous attempts to provide field laboratory diagnostics for outbreaks of Ebola hemorrhagic fever . In 1976 during the Zaire ebolavirus outbreak an immunofluorescence assay was used for acute case identification but the results were considered poor [29] . In 2000 during the Ebola outbreak ( Sudan ebolavirus ) the US-CDC operated a laboratory within the Gulu district at St . Mary's Lacor Hospital , Uganda , and used antigen capture and reverse transcription nested PCR ( RT-PCR ) to successfully diagnose infection in suspected patients [30] . In 2003 during the Ebola outbreak ( Zaire ebolavirus ) in Mbomo , The Republic of the Congo , NML together with partners from the IRD , Franceville , Gabon , and the Bundeswehr Institute of Microbiology , Munich , Germany , operated a small field laboratory under the lead of WHO using antigen capture and Q-RT-PCR to diagnose acute cases [31] , [32] . In general , the usefulness of on-site laboratory support during filovirus outbreaks is not really questioned [2] , [24] , and , in particular , the positive experience from this MHF outbreak demonstrate that rapid turn-around RT-PCR diagnostics can clearly aid in surveillance and case management [15] , [25] . PCR-based techniques can be prone to contamination resulting in false positive results . Here we used a technique that did not require opening of tubes largely reducing the risk of contamination . Other concerns have been raised towards the reliability of RT-PCR assays during early disease stages and for survivors in the early convalescent stage , the consequences of false-positive and false-negative results of RT-PCR assays could be dire to outbreak management [30] . Indeed , PCR-based assays , like other diagnostic tests , have weaknesses and do not produce reliable results under all circumstances . Therefore , independent , methodologically different , confirmatory assay such as antigen capture to support RT-PCR should be mandatory . However , nowadays most laboratories depend on PCR detection as their first and most rapid diagnostic methods and there are good reasons to support that choice [33] . If a confirmatory assay is not available or unsuccessful , alternatives for RT-PCR confirmation include sample re-extraction , a second clinical specimen and/or an assay with independent targets ( Figure 2 ) . Nevertheless , any diagnostics should not replace general and common sense precautions in case patient management and on-site laboratory diagnostics should be in close proximity to the ward allowing for continuous interaction between physicians/nurses and laboratory personnel [15] , [25] . Importantly , during this field laboratory deployment , Q-RT-PCR proved to be very sensitive and reliable even in this challenging environment . Patient samples were positive in our testing beginning on the day of onset of symptoms but we did see that detection in swab samples could be delayed by a few hours when compared to blood this early in the course of illness . The collection of appropriate clinical specimens for diagnostic testing has become an increasing problem during filovirus outbreaks . The reasons for this can include the lack of properly trained personnel , fear of personnel to apply invasive procedures , cultural objections to bleeding and any other invasive pre- and post mortem sampling procedure , and insufficient infrastructure for sampling and transportation [22] , [24] , [34] . In that respect , the MHF outbreak in Angola was not different from previous outbreaks . In particular , resistance in the community to bleeding and post mortem invasive procedures , such as cardiac puncture or liver biopsy , and the increasing resistance of aid personnel to apply invasive procedures in the field ( community ) made oral swabs the predominant clinical specimen available for testing . As demonstrated here on paired blood/oral swab samples , in general there was no significant difference in viral load between oral swabs and EDTA blood taken at the same time ( Figure 5 ) . This supported oral swabs as an alternative diagnostic specimen to blood . The few incidences when oral swabs were less suitable than EDTA blood related to early disease stage and early convalescent stage samples . Lower viral loads in oral swabs compared to EDTA blood , at these stages , are likely to explain this discrepancy . Additionally , there are inherent sampling variables associated with oral swabs ( the technique and efficiency of swabbing; moisture level of the oral cavity ) that are not present in a blood draw , which may also have a role in these differences . However , despite the fact that oral swabs seemed to have been an appropriate specimen source for laboratory testing during this outbreak , and oral/nasal swabs are valuable alternatives in cases of resistance in the affected population to invasive procedures , EDTA blood should remain the priority choice for a clinical specimen due to the longer period of detectable viremia , the suitability to serological-based testing , and the value for monitoring potential point of care therapies in future . While this study is not a detailed epidemiologic study , brief mention of some of the data is warranted as it has not been yet published elsewhere . This MHF outbreak was unique in regards to its location , case number and case fatalities , but also showed a large proportion of paediatric cases and cases among woman in the child bearing ages [2] , [24] . Since MARV , as Ebola virus , are usually transmitted through close contact with blood , secretions or excretions from infected patients , family members and medical personnel caring for patients or preparing bodies for burials are considered high risk exposure groups [2] , [34] . It has been proposed that because women provide the majority of in-home care that this was the reason for the preponderance of cases in women [35] . Certainly women provide the majority of care for the children and since , especially early in the outbreak , children less than 5 years of age represented the largest single age group affected may also be reflective of this fact . Furthermore , the detection of MARV in breast milk during this outbreak indicates that breastfeeding might have played a role in virus transmission . This is supported by epidemiological data indicating transmission from infected mothers to their nursing babies followed , after death of the mothers , by virus transmission from the infected babies to wet nurses who subsequently infected their own nursing child ( authors , unpublished observation ) . Other factors may have come into play including the alleged lack of appropriate infection control within the paediatric ward prior to the identification of the outbreak [36] . It is very unlikely that the predilection of women and young children represents a biological predisposition , given that the demographics of the outbreak changed through the course of the outbreak ( i . e . early in the outbreak a very high percentage were paediatric cases whereas later cases became more evenly distributed by age ) , and yet the virus changed very little [16] . Without more detailed epidemiologic data , it remains unclear which of these transmission routes constituted significant mechanisms for virus spread in the Uige outbreak . Offering differential diagnosis significantly increases the value of on-site diagnostics . This is much harder to achieve in the field and requires variable clinical specimen ( in particular blood or stool ) , more manpower and more extensive and continuous supplies . At a minimum , malaria diagnostics ( e . g . commercially available rapid dipstick tests ) and diagnosis for severe gastrointestinal infections should be available . Proper case patient management including intravenous fluid administration would also require blood chemistry and haematology analysis , another capacity that needs to be considered for expansion of a field laboratory response capacity . Most of what constitutes the MLU can be sourced from equipment that most reference laboratories would have access to from their normal compliment of equipment and supplies , however a dedicated MLU would likely require the investment of approximately $100 000 and a weekly deployment cost of $2000 for reagents and supplies . Logistic needs and costs during a mission can be best managed through a close working relationship with other organizations including the WHO and Médecins Sans Frontières ( MSF ) . The greatest challenge to the operation of the MLU was the lack of consistent electrical power and our reliance on portable generators . This necessitated the use of battery backup systems for thermocyclers and did not allow for storage of samples or reagents at freezing temperatures as freeze-thaw cycles could not be avoided . Fortunately , all reagents were relatively stable at 4°C over a three week rotation period before replacement teams replenished the reagents . We were able to efficiently operate the MLU using teams of two members as the workload and workflow rarely justified additional staff . We have since recommended that teams of three be deployed to allow for rest and health issues . In conclusion , the combined operation of a field and reference laboratory in this outbreak allowed for a unique evaluation of field diagnostic capacity under difficult circumstances . Rapid MARV-specific Q-RT-PCR was useful for triage and assessing the need for isolation . The quick turn-around of laboratory diagnosis on the basis of Q-RT-PCR assays significantly improved outbreak response efforts . Therefore we propose: “On-site laboratory diagnosis should become a routine part of any future filovirus outbreak response as it provides all responders with valuable information to help minimize the extent and durations of these events” .
|
A mobile laboratory unit ( MLU ) was deployed to Uige , Angola as part of the World Health Organization response to an outbreak of viral hemorrhagic fever caused by Marburg virus ( MARV ) . Utilizing mainly quantitative real-time PCR assays , this laboratory provided specific MARV diagnostics in the field . The MLU operated for 88 consecutive days allowing MARV-specific diagnostic response in <4 hours from sample receiving . Most cases were found among females in the child-bearing age and in children less than five years of age including a high number of paediatric cases implicating breastfeeding as potential transmission route . Oral swabs were identified as a useful alternative specimen source to the standard whole blood/serum specimens for patients refusing blood draw . There was a high concordance in test results between the MLU and the reference laboratory in Luanda operated by the US Centers for Disease Control and Prevention . The MLU was an important outbreak response asset providing valuable support in patient management and epidemiological surveillance . Field laboratory capacity should be expanded and made an essential part of any future outbreak investigation .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine"
] |
2011
|
The Use of a Mobile Laboratory Unit in Support of Patient Management and Epidemiological Surveillance during the 2005 Marburg Outbreak in Angola
|
Epigenetic variations of phenotypes , especially those associated with DNA methylation , are often inherited over multiple generations in plants . The active and inactive chromatin states are heritable and can be maintained or even be amplified by positive feedback in a transgenerational manner . However , mechanisms controlling the transgenerational DNA methylation dynamics are largely unknown . As an approach to understand the transgenerational dynamics , we examined long-term effect of impaired DNA methylation in Arabidopsis mutants of the chromatin remodeler gene DDM1 ( Decrease in DNA Methylation 1 ) through whole genome DNA methylation sequencing . The ddm1 mutation induces a drastic decrease in DNA methylation of transposable elements ( TEs ) and repeats in the initial generation , while also inducing ectopic DNA methylation at hundreds of loci . Unexpectedly , this ectopic methylation can only be seen after repeated self-pollination . The ectopic cytosine methylation is found primarily in the non-CG context and starts from 3’ regions within transcription units and spreads upstream . Remarkably , when chromosomes with reduced DNA methylation were introduced from a ddm1 mutant into a DDM1 wild-type background , the ddm1-derived chromosomes also induced analogous de novo accumulation of DNA methylation in trans . These results lead us to propose a model to explain the transgenerational DNA methylation redistribution by genome-wide negative feedback . The global negative feedback , together with local positive feedback , would ensure robust and balanced differentiation of chromatin states within the genome .
Epigenetic variation of gene expression is mediated by chromatin marks , such as modifications of histones and DNA . Importantly , these marks and associated gene expression patterns can be inherited over multiple generations in both animals and plants [1 , 2] . Transgenerational epigenetic inheritance , especially the one associated with DNA methylation , is widespread in plants , and that could have a significant impact on evolution [3–5] . The long-term dynamics of DNA methylation has recently been examined genome-wide at single base resolution in the flowering plant Arabidopsis [6 , 7]; by analysing repeatedly self-pollinated wild type Arabidopsis plants , heritable gain and loss of DNA methylation have been detected , although their frequencies are generally low . A complementary approach to uncover the background mechanisms controlling long-term DNA methylation dynamics is to examine the effects of impaired DNA methylation pattern over multiple generations . Factors controlling genomic DNA methylation have been studied extensively in Arabidopsis; and many of these factors constitute positive feedback loops to stabilize epigenetic states . Cytosine methylation in the context of dinucleotide CG is maintained by maintenance methyltransferase MET1 [8 , 9] , while cytosine methylation at non-CG site is mediated by chromomethylases ( CMTs ) [10 , 11] . The CMTs are recruited to chromatin by methylation of histone H3 lysine 9 ( H3K9me ) , and the H3K9 methylase KYP/SUH4 is also recruited to chromatin with non-CG methylation , generating a self-reinforcing positive feedback loop [11–14] . Both H3K9me and non-CG methylation are silent heterochromatin marks normally found in repeats and transposable elements ( TEs ) ; and these marks are rarely detectable in transcribed genes . Exclusion of these marks from transcribed genes depends on the H3K9 demethylase IBM1 ( Increase in BONSAI Methylation 1 ) [13 , 15] . IBM1 removes H3K9me from transcribed genes , generating another positive feedback loop to stabilize active states [13] . In addition , a positive feedback loop is also found in a process called RNA-directed DNA methylation ( RdDM ) . RdDM is a de novo DNA methylation process triggered by double-strand RNA; and factors involved in this process have been extensively studied [16–20] . The final step of RdDM is DNA methylation of both CG and non-CG sites by the de novo DNA methyltransferase DRM2 ( Domains Rearranged Methylase 2 ) , with the RNAi machinery and small interfering RNA ( siRNA ) functioning as upstream factors . Interestingly , production of siRNA also depends on DRM2 [21 , 22] , suggesting another positive feedback that stabilizes the silent state . Genome-wide DNA methylation profiles have been determined in mutants of these and other factors controlling DNA methylation [11 , 23 , 24] , although information for the transgenerational effects of these mutations is limited . Among the Arabidopsis mutants affecting genomic DNA methylation , ddm1 ( decrease in DNA methylation 1 ) is one of the mutations with the strongest effects . Mutant plants show drastic reduction of DNA methylation at both CG and non-CG sites in heterochromatic repeats and TEs [25 , 26] . The DDM1 gene encodes a chromatin remodeling factor , which is necessary for DNA methylation in heterochromatic sequences [10 , 27] . Mutation in its mammalian ortholog Lsh induces loss of DNA methylation , suggesting conserved functions across the animal and plant kingdoms [28 , 29] . A striking feature of the Arabidopsis ddm1 mutant is the progressive accumulation of the developmental defects; initial generations of the ddm1 mutant grow relatively normally , but many types of developmental abnormalities arise after multiple rounds of self-pollinations [30 , 31] . Some of the abnormalities are due to DNA sequence changes , such as insertion mutations of de-repressed endogenous TEs [32–34] or a rearrangement of repeats [35] , but others are due to epigenetic changes in gene expression , which correlate with changes in DNA methylation pattern at the affected loci [36 , 37] . Here we analyze the transgenerational effects of the ddm1 mutation genome-wide , by comparing DNA methylation of the ddm1 mutants before and after the repeated self-pollinations . This analysis revealed ectopic accumulation of non-CG methylation at hundreds of loci; and unexpectedly , this hypermethylation could only be seen after repeated self-pollinations . Furthermore , when ddm1-derived chromosomes with disrupted heterochromatin were introduced into a DDM1 wild type background , de novo accumulation of non-CG methylation was induced in trans . These results lead us to propose a model in which loss of heterochromatin is progressively compensated for through a negative feedback mechanism that leads to heterochromatin redistribution across the genome .
To understand the changes in DNA methylation patterns during self-pollinations of ddm1 mutant genome-wide , we compared DNA methylation before and after the self-pollination of the mutant . We examined DNA methylation in four individuals of ddm1 homozygous mutants segregated in progeny of a heterozygote ( hereafter called 1G for the 1st Generation ) and also four lines of ddm1 plants independently self-pollinated eight times ( hereafter called 9G ) ( S1 Fig ) . In 1G , the ddm1 mutation already induced reduction of DNA methylation in heterochromatic regions [10 , 25 , 26] . Methylation in repetitive sequences , such as transposable elements ( TEs ) ( Fig 1D–1F ) , was much more severely affected than that in low copy sequences , such as genes ( Fig 1A–1C ) . The reduction was found for both CG sites ( Fig 1A and 1D ) and non-CG sites . In non-CG sites , both CHG sites ( Fig 1B and 1E ) and CHH sites ( Fig 1C and 1F ) were affected ( H can be A , T , or C ) . When we compared average DNA methylation of 9G to 1G , two features were noted for both genes and TEs: further decrease of CG methylation and an increased methylation at non-CG sites ( Fig 1 ) . Although the ddm1 mutation immediately induces a drastic loss of DNA methylation in repeats , further reduction of methylation in later generations has been reported for a few CG sites [30] . Our genome-wide analysis revealed that many loci behave in a similar manner ( Fig 2A ) . The progressive reduction of DNA methylation can have significant phenotypic effects; for example , the promoter of the imprinted gene FWA remains methylated in the 1G ddm1 but the methylation is lost stochastically in 9G ddm1 ( Fig 2B ) , generating heritable epialleles that cause late-flowering phenotype [31 , 36 , 38] . The progressive reduction is seen genome-wide for both genes and TEs ( Fig 1A and 1D ) . To compare the features of the regions hypomethylated immediately or gradually , we defined differentially methylated regions ( DMRs; details in Materials and Methods ) . The regions ddm1 affects immediately ( 1G-WT DMRs ) were enriched in dimethylation of histone H3 lysine 9 ( H3K9me2 ) ( Fig 2D left and 2E ) . H3K9me2 is a mark of silent heterochromatin , and these results are consistent with previous reports [10 , 26] . In marked contrast , however , regions affected slowly ( 9G-specific DMRs ) have much lower level of H3K9me2 in wild type ( Fig 2D middle ) . DDM1 gene function is necessary for CG methylation in heterochromatin , but in the long-term DDM1 also has significant effects on CG methylation in less heterochromatic regions including gene bodies ( Fig 2C ) . More counter-intuitively , our genome-wide analysis revealed a large number of genes and TEs ectopically hypermethylated at non-CG sites in the self-pollinated ddm1 lines ( Figs 3A , 3B , 4A and 5A–5E ) . The regions CHG hypermethylated also showed hypermethylation at CHH sites ( Figs 3D , 5A–5D , and S6A Fig ) . In addition , although genic CG methylation tends to decrease progressively from 1G to 9G on average ( Figs 1 and 2 ) , non-CG hypermethylated regions show an increase in CG methylation ( Fig 3D ) . The CG and non-CG hypermethylation was found reproducibly at specific loci ( S8 Fig ) . The affected loci include BONSAI and other sequences we have reported previously [37 , 39] , but the majority of the affected loci could only be detected by whole-genome bisulfite sequencing ( WGBS ) , because that can detect increased non-CG methylation with high sensitivity even at loci already CG methylated . In addition to genes , a large number of TEs showed increase in non-CG methylation ( Figs 3A , 3B , 4E , and S9–S11 Figs ) . A very unexpected feature revealed by WGBS is that non-CG hypermethylation of genes is almost undetectable in the first generation of ddm1 but is specifically and reproducibly seen in the repeatedly self-pollinated ddm1 lines . In Fig 3A and 3B , many black dots can be seen along the vertical axis in the panels for 9G but not for 1G . The non-CG hypermethylation of genes is not a simple extension of the effect seen in the first generation . This feature can only be detected in later generations ( Fig 3C ) . In order to further understand the transgenerational dynamics , we examined four independently self-pollinated 2G ddm1 plants . If the hypermethylation proceeds equally at each self-pollination , the increase from 1G to 2G would be 1/8 or more of the increase from 1G to 9G , provided that the methylation level should saturate at specific level ( the methylation level can not exceed 100% ) . Interestingly , although hypermethylation proceeded in 2G , the difference between 1G and 2G was much less than 1/8 of that between 1G and 9G , suggesting that the increase is slow initially but accelerated in later generations ( S12 and S13 Figs ) . How is this non-CG hypermethylation induced ? Our genome-wide bisulfite analyses revealed that the genes non-CG hypermethylated in the self-pollinated ddm1 tend to have low levels of non-CG methylation already in wild type plants ( Fig 3D ) , suggesting that preexisting small heterochromatin domains may function as seed for further heterochromatin formation . Interestingly , distribution of H3K9me2 around the DMR is asymmetric; it is enriched in the 3’ of the DMRs ( S14 Fig ) . We have previously shown that the BONSAI gene is flanked by insertion of a heterochromatic LINE in the 3’ region [37] ( Fig 4A and S13A Fig ) . The BONSAI hypermethylation in ddm1 is induced in a strain with the LINE insertion but not found in a strain without the LINE insertion [37] . The DNA methylation spreads from the 3’ LINE to the BONSAI region during repeated self-pollination of ddm1 mutants [37] . Spread of non-CG methylation from 3’ to 5’ regions was also noted at other loci ( Fig 5A–5D ) . When the methylation level differs among the four 9G ddm1 plants , plants with stronger signals tended to show relative centroid positions more upstream than plants with weaker signals , suggesting that the signal spreads from 3’ to 5’ ( Fig 5F ) . These observations suggest that common mechanisms may operate in BONSAI and many , even if not all , affected loci . We have previously shown that the de novo non-CG methylation in the self-pollinated ddm1 does not require components of the RdDM machinery , such as RDR2 , DCL3 , and DRM2 [39] . On the other hand , the non-CG methylase CMT3 and H3K9 methylase KYP are necessary for the de novo methylation , suggesting that the ectopic methylation occurs by mechanisms mediated by the heterochromatin marks H3K9me and non-CG methylation [39] . Indeed , the non-CG hypermethylation at the BONSAI locus is associated with ectopic H3K9me ( Fig 4B ) . The self-reinforcing loop of non-CG methylase and H3K9 methylase activities could be the basis for the acceleration of hypermethylation as the generation proceeds ( S13B Fig ) . As the two processes enhance each other , the positive feedback would accelerate the spread of the heterochromatin in later generations [12 , 13] . Increased non-CG methylation has been reported in mutants of the CG methyltransferase gene MET1 [40–42] , which results at least in part from a reduction of full-length IBM1 transcript [43] . The IBM1 gene encodes a demethylase for histone H3K9; and mutation in this gene induces accumulation of H3K9me2 and non-CG methylation in gene bodies . Interestingly , developmental phenotypes of the ibm1 mutation also become progressively stronger during self-pollinations [15] . We compared the regions of non-CG hypermethylation in the ibm1 and self-pollinated ddm1 . Although an overlap can be detected , the majority of the DMRs in ddm1 mutants before and after the self-pollinations were distinct from the DMRs of ibm1 mutants ( Fig 6B and S16 Fig ) . Just as progressive loss of CG methylation in the ddm1 mutant , ibm1 mutant shows progressive accumulation of non-CG methylation in later generations ( Fig 6A , S15 and S16 Figs ) . This is consistent with a recent report [44] and likely accounts for the progressive developmental defects in the ibm1 mutant . We examined DNA methylation patterns of the genes and TEs hypermethylated in the self-pollinated ddm1 lines ( Fig 6C ) . Compared to the ibm1 mutant , the peak in the ddm1 was shifted toward 3’ end . Interestingly , the shift of the peak in the hypermethylation was also found for CG methylation ( S5D Fig ) . Although CG methylation of gene body in wild type peaks around the center ( S5C Fig ) , increase of genic CG methylation in 9G ddm1 was not proportional to the methylation of wild type; instead , the increase of CG methylation was shifted toward 3’ regions ( S5D Fig ) . Together with the observation that CHG-hypermethylated genes tend to show CG-hypermethylation ( Fig 3D ) , these results suggest a link between the ectopic CG methylation and non-CG methylation , as we discussed previously [39] . The bias of the hypermethylation signal toward the 3’ region in 9G ddm1 is especially evident in the hypermethylated TEs; the peak was often located outside of the transcription unit for both CHG and CHH methylations ( Fig 6C , bottom half ) . When different families of TEs are compared , the peak in the downstream region was especially evident in the GYPSY-like LTR retrotransposons ( S10 Fig ) . Generally , these TEs lost DNA methylation in 1G ddm1 , but regained methylation during the self-pollinations ( S5A and S9–S11 Figs ) . The ddm1 mutation can induce increased DNA methylation at hundreds of genes and TEs . The hypermethylation can be a direct consequence of impaired DDM1 function , or alternatively , an indirect effect of disruption of heterochromatin in the mutants . To test these possibilities , we examined the effect of chromosomes introduced from ddm1 into wild type DDM1 background . Chromosomes losing DNA methylation in the ddm1 mutants remain unmethylated even after introduction into wild type DDM1 background [25 , 45] . We examined DNA methylation data of epigenetic recombinant inbred lines ( epiRILs ) [46] . In the epiRILs , a ddm1 mutant plant was crossed to wild type plant twice to segregate DDM1/DDM1 lines with around one quarter of chromosome segments derived from ddm1 . Although remethylation can be induced in regions associated with small RNA , hundreds of DMRs remain unmethylated in the wild type DDM1 background [46 , 47] . Each of these segregating lines have been self-pollinated seven times , which makes most of the genomic regions fixed in ddm1-derived haplotype or wild-type derived haplotype [46] . We examined if the loci exhibiting hypermethylation in the self-pollinated ddm1 lines also showed hypermethylation in some of the epiRILs . We utilized DNA methylation data for the 123 epiRILs , which are based on immunoprecipitation ( IP ) of genomic DNA by anti-methylcytosine antibody . As the context of methylation cannot be distinguished , we examined seven loci that show increased methylation in 9G ddm1 but a relatively low level of methylation at CG sites in wild-type . In six out of the seven loci examined , we could detect hypermethylation in multiple epiRILs , suggesting that the hypermethylation can be induced or maintained in the DDM1 background ( Figs 7A , 7C , 7E and S17 Fig ) . In all of them , the hypermethylation showed a strong positive correlation with the amount of disrupted heterochromatin in each of these lines ( Fig 7 , S17 Fig and S1 Table ) , suggesting that the hypermethylation was induced or maintained in the background of disrupted heterochromatin in other genomic regions . The hypermethylation could be induced de novo or alternatively maintained from the parental ddm1 . The parental ddm1 plant originally used for making epiRILs was already self-pollinated three times ( 4G ) and that plant also show low level of ectopic methylation at some loci ( S17 Fig ) , which may have the potential to be maintained in DDM1 background [37] . Very importantly , however , the hypermethylation was found even in chromosome segments originated from wild type DDM1 ( Figs 7B , 7D , 7F and S18–S23 Figs ) , demonstrating that the hypermethylation could be induced de novo after the initial crosses and subsequent repeated self-pollinations in the background of functional DDM1 . In order to confirm and extend this observation , we used WGBS for an epiRIL with genome-wide reduction of heterochromatic DNA methylation . The epiRIL98 , which contains large amount of chromosomes with reduced DNA methylation , showed CHG hypermethylation in many genes ( Fig 8A ) , which include BONSAI gene ( S24A Fig ) and genes with body methylation ( S24B–S24C Fig ) . In the CHG hypermethylated genes , the CHG methylation level was generally much higher than that of the parental 4G ddm1 plant ( Fig 8B ) , suggesting that the hypermethylation was amplified or induced de novo in the background of functional DDM1 . A large number of CHG hypermethylated genes were found in chromosome regions of wild type haplotype ( Fig 8C and S25 Fig ) , again suggesting that they can be induced de novo . In control epiRILs with much lower levels of disrupted chromatin , the hypermethylation was undetectable , confirming that the disrupted heterochromatin was responsible ( Fig 8A ) . Taken together , these results indicate that the hypermethylation can be induced de novo by trans-acting effects of disrupted heterochromatin .
Here we report short- and long-term effects of the ddm1 mutation . The mutation immediately induces a drastic loss of DNA methylation in heterochromatic regions in the first generation when it becomes homozygous . In later generations , the ddm1 mutation reproducibly induces ectopic accumulation of DNA methylation in hundreds of genes and TEs . This work and previous work [39] suggest that the ectopic methylation occurs by spread of heterochromatin marks mediated by the non-CG methylase CMT3 and H3K9 methylase KYP . Interestingly , this effect was slow in the initial generations but accelerated in later generations , suggesting strong positive cooperativity for the heterochromatin accumulation . That could be explained by the self-reinforcing positive feedback of H3K9me and non-CG methylation [12 , 13] . In addition to the local positive feedback , global negative feedback seems important for the DNA methylation dynamics . The ectopic DNA methylation seems to reflect negative feedback of disrupted heterochromatin in other genomic regions , because the ectopic methylation could also be induced in DDM1 wild type background when the genome contains large amount of chromosomal segments with disrupted heterochromatin ( Figs 7 and 8 ) . How does the negative feedback work ? One possible explanation is that disruption of heterochromatin in the ddm1 mutant results in release of heterochromatin-forming factors such as CMTs and H3K9 methylases , which then become available in other regions . As these factors are normally recruited to heterochromatin , disruption of a large proportion of heterochromatin in the genome would result in increased level of these factors in released conditions , which would induce spread of heterochromatin into normally euchromatic regions and its amplification by the self-reinforcing loop of H3K9me and non-CG methylation ( Fig 9 ) . In the model we proposed , global reduction of heterochromatin induces ectopic non-CG methylation ( Fig 9 ) . That would account for the correlation between the global reduction of methylation and ectopic methylation in epiRILs ( Fig 7A , 7C , 7E and S17 Fig ) . An alternative mechanism would be that ddm1 induces change in a specific locus , such as transcriptional de-repression or repression of a specific gene , and the change is inherited in the DDM1 wild type background and induces the ectopic methylation . For example , ROS1 gene expression is reduced in mutants with reduced DNA methylaiton [48] , which would lead to hypermethylation at specific loci . However , although ROS1 gene expression is reduced in ddm1 , it is expressed almost normally in epiRIL98 , which show strong non-CG hypermethylation ( S26A Fig ) . In addition , DMRs hypermethylated in 9G ddm1 and ros1-dml2-dml3 triple mutant do not overlap much , further suggesting that the hypermethylation in 9G ddm1 is not due to reduced ROS1 expression ( S26B Fig ) . More generally , we could not find a locus consistently derived from ddm1 parent in all of the plants showing the high level of ectopic hypermethylation in the six loci ( S18–S23 Figs ) . Although we cannot exclude the possibility that two or more specific loci redundantly mediate the ectopic methylation , a more parsimonious explanation derived from available data would be that the trans-interaction is mediated by global homeostasis . The de novo methylation in the epiRILs might also be related to mechanisms such as paramutation [49 , 50] , or transchromosomal methylation ( TCM ) [51] . In these phenomena , methylated sequences induce methylation in related sequences . However , the ectopic hypermethylation in the epiRILs is generally much higher than that of the parental ddm1 ( Fig 8B ) , suggesting that even if paramutation-like or TCM-like mechanisms are involved , the effect should be much amplified during self-pollinations of epiRILs; and the degree of the amplification correlates with global disruption of heterochromatin ( Fig 7 and S17 Fig ) , which is due to the ddm1-derived chromosomes . This trans-acting negative feedback could also be understood as a hypersensitive reaction to the challenge by active and proliferating TEs . Our genome-wide analyses revealed that many of the TEs can be targets of the negative feedback ( Fig 3A and 3B and S9–S11 Figs ) . Active TEs often keep parts of heterochromatin , which can function as seeds of the self-reinforcing heterochromatin formation . An increase in non-CG methylation is also seen in mutants of the histone demethylase gene IBM1 . However , targets of IBM1 are generally euchromatic and they do not overlap much with regions hypermethylated in the self-pollinated ddm1 lines ( Fig 6B and S16 Fig ) . An increase in non-CG methylation is also found in the maintenance CG methylase gene MET1 [40–42] . As a mechanism for the met1-induced increase in non-CG methylation , loss of IBM1 function is suggested , as IBM1 transcripts become truncated in the met1 mutant [43] . On the other hand , it has been reported that the main targets of the met1-induced accumulation of H3K9me2 are genes with H3K27me3 , another modification for silent chromatin [52] . The negative feedback of heterochromatin marks comparable to that seen in the self-pollinated ddm1 lines may also operate in met1 mutants . In our analyses , although regions affected by met1 , ibm1 , and self-pollinated ddm1 all differ , significant overlaps are noted ( S27 Fig ) . For these mutants , the local triggers for heterochromatin accumulation appear to be distinct , despite the possible overlap in the downstream mechanisms , including the self-reinforcing loop of non-CG methylation and H3K9me . Heterochromatin homeostasis mechanisms analogous to those we have uncovered in Arabidopsis may also be operating in other eukaryotes . Mice with a disruption of its DDM1 homolog Lsh show global reduction of genomic DNA methylation , but interestingly it is also associated with increased DNA methylation at specific regions [29] . In human cancer , hypomethylation of repeats and TEs is often associated with local hypermethylation of genes , such as tumor suppressor genes [53 , 54] . In Drosophila , an increase in the amount of heterochromatic Y chromosome can results in a release of silencing at multiple loci in trans [55] , suggesting a negative feedback similar to that discussed here . Furthermore , Drosophila modifiers of position effect variegation often function in dosage-dependent manners [56 , 57] , consistent with the pathway proposed in Fig 9 . Positive feedback loops would stabilize and enhance silent and active states [12 , 13 , 21 , 58] , but they carry the risk of going out of control to excess . A global negative feedback mechanism , together with the local positive feedback , would ensure a robust and balanced chromatin differentiation within the genome , as has been discussed for pattern formation during development [59 , 60] . In the context of evolution in plants , a large variation in the amount of repetitive sequences is often noted between related species or even within a species [61–63] . On such occasions , fine-tuning of the amount of trans-acting heterochromatin factors would be especially important , as an imbalance would not only immediately affect gene expression level but also influence the epigenotype in a transgenerational manner .
Isolation of the ddm1-1 and ibm1-4 mutants has been described previously [15 , 25] . Self-pollination of ddm1 lines was described previously [30] . In order to remove heritable effects of the ddm1 mutation , the original ddm1 mutant was backcrossed six times in the heterozygous state . The heterozygous plants were propagated by self-pollination . 1G ddm1 mutant plants were selected from self-pollinated progeny of the heterozygote . 9G ddm1 plants were generated by independently self-pollinating different ddm1 segregants eight times ( S1 Fig ) . Generation of epiRILs has been described previously [64] . The annotations of genes and TEs are based on The Arabidopsis Information Resource ( http://www . arabidopsis . org/ ) . TAIR8 was used for analyzing ChIP chip data ( Fig 2E ) , TEG ( TE gene ) data , and epiRILs data . TAIR10 was used for other analyses . The details of the annotation of TEGs were described in a document in TAIR web ( ftp://ftp . arabidopsis . org/home/tair/Genes/TAIR8_genome_release/Readme-transposons ) . For the 1G and 9G ddm1 plants and their controls , genomic DNA was isolated from rosette leaves using the Illustra Nucleon Phytopure genomic DNA extraction kit , and genome-wide bisulfite sequencing was performed as described previously [65] . Raw sequence data were deposited in the DDBJ ( DNA Data Bank of Japan ) Sequence Read Archive ( DRA; accession nos . DRA002545 , DRA002546 , DRA002548 , DRA002549 , DRA002551 , DRA002554 , DRA002555 , DRA003018 , DRA003019 and DRA003020 ) . The adaptor sequences were clipped out using the FASTX-toolkit ( http://hannonlab . cshl . edu/fastx_toolkit/ ) . Reads were trimmed to 90 nucleotide length ( 45 nucleotide for the data obtained from GEO—GSE39901 ) and mapped to reference genomes ( Release 10 of the Arabidopsis Information Resources ) using the Bowtie alignment algorithm [66] with the following parameters , "-X 500-e 90-l 20-n 1" . Only uniquely mapped reads were used . Clonal reads were removed except one with the best quality . Any read with three consecutive methylated CHH sites were eliminated . The level of methylation of cytosine in a genomic region was calculated using the ratio of the number of methylated cytosine to that of total cytosine . For the three epiRILs and two parental lines , whole-genome bisulfite sequencing was described previously [46] and the data are in GEO ( GSE62206 ) . DMRs ( differentially methylated regions ) were defined by comparing the methylation level of 100-bp windows throughout the genome between two genotypes . The windows with at least 20 sequenced cytosines were used for the comparison . The level of methylation was calculated using the weighed methylation level of each genotype [67] . The windows were selected as DMRs when difference of methylation level was 0 . 5 or more at CG site or 0 . 3 or more at CHG sites . For defining contiguous DMR ( conDMR ) , multiple DMRs were merged if they were adjacent to each other or there was only one gap of the 100-bp window . The centroid of cytosine methylation in conDMR was calculated using the relative position within that region weighed by methylation level of each cytosine . In Fig 5F , we used conDMR of 500 bp or longer and overlapping with genes . Each contiguous DMR was aligned according to the orientation of the corresponding gene . The correlation coefficient between the level and the relative centroid position of DNA methylation was calculated among the four 9G ddm1 plants in each conDMR . To plot DNA methylation patterns over genes or TEGs in ddm1 mutants , #1 samples of each genotype ( Figs 2A , 3A and 3B ) in 1G ddm1 and 9G ddm1 were used . To draw the heatmap of methylation of cytosine , cluster 3 . 0 [68] and Java Treeview [69] were used . 15-day-old seedlings were fixed with formaldehyde and ChIP was performed as described previously [70] , using antibody against H3K9me1 ( CMA316 ) and H3K9me2 ( CMA307 ) [71] . To assure the equal amount of chromatin in each line , input DNA were quantified by quantitative PCR using TaKaRa Dice_Real Time System TP800 and ACT7 primers . Then , input DNA and each sample were diluted according to the estimated input DNA concentrations . Input DNA , mock ( without antibody ) , and ChIP samples were analyzed by PCR . The PCR conditions were as follows: pre-incubation for 2 min at 94°C , 27 cycles at 94°C for 30 sec , 58°C for 20 sec , 72°C for 45 sec and a final extension at 72°C for 4 min . Primers used for the ChIP are listed in S2 Table . In addition to the BONSAI locus , we examined six loci with CHG methylation increased more than 0 . 3 from 1G ddm1 to 9G ddm1 . Three of them were selected for relatively high level of ectopic CHH methylation ( H1 , H2 , H3 ) and three with relatively low CHH methylation ( L1 , L2 , L3 ) . The increase of CHH methylation from 1G ddm1 to 9G ddm1 is more than 0 . 2 for the three H loci , and it is less than 0 . 02 for the three L loci . The lengths of amplicons for the six loci are between 250 bp and 300 bp . ChIP-seq data of various histone modifications [72] in GEO ( GSE28398 ) were used for our analysis . The coordinates were remapped onto TAIR10 annotation using a script in TAIR [73] . Enrichment of histone modification in a DMR was calculated by the density of ChIP-seq reads , and normalized by the mean and the standard deviation of the density of reads in 100 , 000 windows randomly chosen across the genome . The MeDIP-chip data of 123 epigenetic recombinant inbred lines ( epiRILs ) , ddm1 and WT are in GEO ( GSE37284 ) . The regions that were methylated ( M ) in WT and unmethylated ( U ) in ddm1 were selected as targets of ddm1 mutation using the values for HMM ( hidden Markov model ) status ( M ( methylated ) or I ( Intermediate ) or U ( Unmethylated ) ) [46] . Global hypo-methylation index of an epiRIL was calculated as the genome-wide average of the values for HMM status of probes on the chip ( M = 0 , I = 0 . 5 , U = 1 ) in the target regions of ddm1 mutation . The data of inference of inherited haplotypes were shown in the previous study [46] . Following are names of lines numbered 1–6 in Fig 7 and S18–S23 Figs . ( Fig 7A and 7B and S18 Fig ) epiRIL208 epiRIL122 epiRIL98 epiRIL232 epiRIL70 epiRIL114; ( Fig 7C and 7D and S19 Fig ) epiRIL122 epiRIL208 epiRIL114 epiRIL258 epiRIL438 epiRIL508; ( Fig 7E and 7F and S20 Fig ) epiRIL208 epiRIL98 epiRIL438 epiRIL508 epiRIL122 epiRIL114; ( S21 Fig ) epiRIL208 epiRIL73 epiRIL71 epiRIL394 epiRIL98 epiRIL438; ( S22 Fig ) epiRIL508 epiRIL114 epiRIL122 epiRIL438 epiRIL208 epiRIL93; ( S23 Fig ) epiRIL208 epiRIL114 epiRIL556 epiRIL71 epiRIL244 epiRIL98 .
|
DNA methylation is important for controlling activity of transposable elements and genes . An intriguing feature of DNA methylation in plants is that its pattern can be inherited over multiple generations at high fidelity in a Mendelian manner . However , mechanisms controlling the trans-generational DNA methylation dynamics are largely unknown . Arabidopsis mutants of a chromatin remodeler gene DDM1 ( Decrease in DNA Methylation 1 ) show drastic reduction of DNA methylation in transposons and repeats , and also show progressive changes in developmental phenotypes during propagation through self-pollination . We now show using whole genome DNA methylation sequencing that upon repeated selfing , the ddm1 mutation induces an ectopic accumulation of DNA methylation at hundreds of loci . Remarkably , even in the wild type background , the analogous de novo increase of DNA methylation can be induced in trans by chromosomes with reduced DNA methylation . Collectively , our findings support a model to explain the transgenerational DNA methylation redistribution by genome-wide negative feedback , which should be important for balanced differentiation of DNA methylation states within the genome .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Genome-Wide Negative Feedback Drives Transgenerational DNA Methylation Dynamics in Arabidopsis
|
Blood is a dense suspension of soft non-Brownian cells of unique importance . Physiological blood flow involves complex interactions of blood cells with each other and with the environment due to the combined effects of varying cell concentration , cell morphology , cell rheology , and confinement . We analyze these interactions using computational morphological image analysis and machine learning algorithms to quantify the non-equilibrium fluctuations of cellular velocities in a minimal , quasi-two-dimensional microfluidic setting that enables high-resolution spatio-temporal measurements of blood cell flow . In particular , we measure the effective hydrodynamic diffusivity of blood cells and analyze its relationship to macroscopic properties such as bulk flow velocity and density . We also use the effective suspension temperature to distinguish the flow of normal red blood cells and pathological sickled red blood cells and suggest that this temperature may help to characterize the propensity for stasis in Virchow's Triad of blood clotting and thrombosis .
Red blood cells are the major component of blood and with a radius of ∼4 µm and a thickness of ∼1–2 µm are sufficiently large that the effects of thermal fluctuations are typically negligible , i . e . their equilibrium diffusivity is very small ( where f is the viscous drag coefficient for a flat disk with radius 4 µm in water at room temperature [1] ) . However , when suspensions of these soft cells are driven by pressure gradients and/or subject to shear , complex multi-particle interactions give rise to local concentration and velocity gradients which then drive fluctuating particle movements [2]–[4] . Nearly all studies of whole blood to date focus on only the mean flow properties , with few notable exceptions [5] . Since the rheology of suspensions in general is largely determined by the dynamically evolving microstructure of the suspended particles [6] , it is essential to measure both the dynamics of individual cells and the collective dynamics of cells in order to understand how the microscopic parameters and processes are related to larger scale phenomena such as jamming and clotting . We complement the large body of work characterizing the flow of sheared and sedimenting rigid particulate suspensions [7]–[11] and here study the statistical dynamics of pressure-driven soft concentrated suspensions while making connections to human physiology and disease . In particular , we provide quantitative evidence that there is heterogeneity in cellular velocity and density . This heterogeneity may play a role in the slow flow or stasis that can lead to the collective physiological and pathological processes of coagulation or thrombosis , as Virchow noted more than 100 years ago [12] . To investigate the short-time dynamics of flowing red blood cells we develop and use computational image processing [13] and machine learning algorithms to segment and track individual blood cells in videos captured at high spatial and temporal resolution in a microfluidic device ( Figures 1 and 2 and Videos S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 ) . We measure individual cell trajectories comprised of more than 25 million steps across more than 500 , 000 video frames . These measurements enable us to ask and answer questions about the variability of velocity fluctuations at the scale of individual normal and sickled red blood cells with variable shape and rigidity . We quantify the effect of bulk flow velocity and density on the microscopic velocity fluctuations , and the role of collective behavior under pathological conditions which alter these properties . We utilized microfluidic devices with cross-sectional area of 250 µm×12 µm , similar to the devices used to characterize the phase diagram for vaso-occlusion in an in vitro model of sickle cell disease [14] . The 12 µm dimension of the microfluidic channels along one axis confines the cell movements in this direction; indeed the range of motion is already hydrodynamically limited by the Fahraeus effect [15] . The primary advantage of this quasi-two-dimensional experimental geometry is the ability to visualize the cells easily , because any significant increase in the size of the channel in this direction would make the cell tracking impossible . This small dimension changes the dynamics as compared to those of cells moving through large circular channels , owing to the effects of the relatively large shear rates in the narrow dimension and our inability to measure fluctuations along this axis , but our system nevertheless enables the characterization and measurement of the quasi-two-dimensional statistical dynamics of both normal and pathological blood flow with very high time and spatial resolution . We chose a set of device and blood parameters relevant to human physiology and pathology in the microcirculation associated with capillaries and post-capillary venules . We derived our quasi-two-dimensional data from the middle fifth of the 250 µm-high channel , where the narrow 12 µm thickness provides the only significant shearing direction , and this shear rate ( ∼10/sec ) is in the physiological range for the microcirculation [15] .
Figure 3a quantifies the planar fluctuations of individual blood cells in terms of the mean-squared displacement , 〈Δr2 ( τ ) 〉 = 〈 ( rbulk ( τ ) −rcell ( τ ) ) 2〉 where denotes a spatial average , and shows that 〈Δr2 ( τ ) 〉 = Dτ , with an effective diffusion constant D much larger than the equilibrium diffusivity ( ∼0 . 1 µm2/s ) . ( See Videos S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 for examples of this diffusive behavior . ) Thus movement of a cell in relation to the bulk at one instant becomes rapidly decorrelated with its subsequent movement , except over very short times relative to the time of interaction between cells . 〈Δr2 ( τ ) 〉 is roughly isotropic at shorter times , and then anisotropic at longer times with fluctuations parallel to the direction of flow 50% larger than perpendicular to it , a finding which is qualitatively consistent with observations of sheared and sedimenting rigid particulate suspensions [3] , [16] . This diffusive behaviour is itself dynamical in its origin , being driven by the relative flow of fluid and cells and the boundary . To understand this dependence , we also plotted in Figure 3b the evolution of the scaling exponent as a function of the bulk flow velocity ( Vbulk ) and red blood cell concentration for more than 700 different experiments with different blood samples . We find that an increase in Vbulk from rest to about 50 µm/s is associated with a change in dynamics from stationary through sub-diffusive to diffusive . However , over the pathophysiologically relevant range of densities studied ( 15%–45% ) there is no consistent effect on the nature of the statistical cell dynamics . Figure 3b shows significant variation in this dynamical process , and only by combining measurements of a large number of cell trajectories are we able to see that the curve flattens with increasing Vbulk as α approaches 1 . 0 . Further , in Figure 3c we show that 〈α〉∼1 . 0 , providing additional support for the conclusion that the typical flow is diffusive . A diffusive process has a characteristic length scale λ corresponding to the mean free path that a cell travels before an interaction , and a characteristic time scale corresponding to the time between these interactions , typically given by the inverse of local shear rate , at the low Reynolds numbers typical of microvasculature flows in vivo as well as in our experiments ( where Re = O ( 0 . 01 ) ) . Then the effective diffusivity scales as , where C is a dimensionless constant which will depend on microscopic properties such as cell shape and rigidity . There are three length scales in the problem that can determine the effective diffusive length scale λ: cell size , cell separation , and cell distance from the boundary . Different length scales will dominate in different limits of density , geometry , and cell size , as a cell will travel only a fraction of the inter-cellular distance before it interacts with another cell or a boundary . In the unconfined limit where the boundary is infinitely far away , the only characteristic scale is the cell size so that , and . This dilute limit has received the most attention to date [2] , [4] , but is far from the soft , dense , and confined suspensions we study . The two remaining origins for this characteristic scale are: ( i ) the distance between cells ( about 3 µm at a two-dimensional density of 33% ) which is comparable to and even smaller than the cell size; ( ii ) the small height of our channel , 12 µm , which implies that the discoid red blood cells interact with the wall . The cells are typically oriented with their discoid faces perpendicular to the smallest dimension of the channel . The strong local shear ( , where 2h is the channel height ) relative to the wall leads to an effective diffusivity , where . As has previously been shown [4] , [6] , [16] , [17] , a velocity gradient can lead to particle interactions and rearrangements in all three principal directions particularly when the shapes of the particles are non-spherical as here . This is particularly true in our study because the particles ( cells ) are disc-like and deformable , so that the combination of shape anisotropy and the generation of normal forces via tangential interaction in soft contact can lead to diffusive motions in the measurement plane [18] . In Figure 4a , we show this diffusive behaviour for Vbulk >∼50 µm/s . The measured D≈8 µm2/s , and for λ ∼ 3 µm . By sampling over times longer than , our measurements reach far enough into the asymptotic behavior of the dynamics to characterize this diffusive process . Over shorter times , we expect a mixture of diffusive and ballistic dynamics , though this effect in our results is dominated by the fact that extremely small displacements are below our analytic sensitivity and appear as stasis . In addition , cell velocities fluctuate because of the localized spatio-temporal fluctuations in shear rate , i . e . , . ( See Videos S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 . ) These shear rate fluctuations could potentially also contribute to the effective diffusivity of the cells , but here we limit ourselves to the simplest mean field picture that ignores the fluctuations in the shear rate itself . To assess the relative role of microscopic determinants such as cell shape and stiffness on this diffusive process , we investigated the behavior of blood cells from patients with sickle cell disease . Red blood cells from these patients become stiff in deoxygenated environments as a result of the polymerization of a variant hemoglobin molecule [19] , resulting in a dramatic increase in the risk of sudden vaso-occlusive events with a poorly understood mechanism [20] . In Figure 4b , we plot D versus Vbulk for oxygenated and deoxygenated sickle cell blood and see that for a given bulk flow rate , the stiffer cells have a smaller diffusivity . Since , our results therefore imply that Cdeoxygenated<Coxygenated , i . e . , the stiffness of the cells influences the dynamics of a pressure-driven suspension independent of Vbulk , likely due to changes in the nature of the interactions of cells with each other , with the channel walls , or with the plasma velocity gradients . The tangential and normal forces between two fluid-lubricated soft moving objects is a complex function of shape , separation , stiffness , relative velocity , and fluid viscosity . Tangential interactions between soft cells lead to normal forces that push the cells away from each other , thus reducing the friction between them [18] . Since the effective diffusion coefficient of this driven system is inversely proportional to the frictional drag , we expect the diffusion coefficient for the stiffer cells to be smaller than that for soft oxygenated cells when the flow velocity is held constant , as is observed .
Hydrodynamic interactions between red blood cells lead to velocity fluctuations and diffusive dynamics of the individual cells . Changes in Vbulk or cellular stiffness alter D and therefore control the magnitude of velocity fluctuations . Cellular velocity fluctuations are quantified by their mean square , , which may be interpreted in the language of the statistical physics of driven suspensions [16] , [21] as an effective suspension temperature . Just as thermal temperature reflects the mean squared molecular velocity fluctuation , the suspension temperature reflects the mean squared cellular velocity fluctuation . This temperature will then change with Vbulk as well as with particle stiffness . Slower flows will have lower effective suspension temperature , as will flows of stiffer particles . In Figure 5 , we show the measured probability distribution of δV2 for two different flow experiments and see that it has longer tails than an equilibrium Maxwell-Boltzmann distribution owing to the non-equilibrium nature of the system , consistent with observations in physical suspensions [3] , [10] . We may nevertheless use the crude analogy of an effective temperature to characterize “hot” blood flow which has increased 〈δV2〉 and is also less likely to coagulate or “freeze” than is a “cold” blood flow where cells are not fluctuating and local stasis is more likely to arise and to persist . Virchow's Triad characterizes the conditions leading to thrombosis as stasis , endothelial dysfunction , and hypercoagulability [12] and our results offer one possible explanation for why pathological blood with stiffer cells and smaller cellular velocity fluctuations will occlude at flow rates where normal blood will not . In conclusion , we have identified random walk-like behavior for pressure-driven dense suspensions of soft particles in quasi-two-dimensional confinement which we quantify in terms of cellular velocity fluctuations as a function of blood flow rate , shape , and stiffness . Our results suggest that these fluctuations may be involved in the collective pathophysiological processes of occlusion and thrombosis , both of which are strongly heterogeneous in space and time . While simple scaling ideas are suggestive , a well-defined microscopic mechanism for this process remains to be established .
This study was conducted according to the principles expressed in the Declaration of Helsinki . The study was approved by the Institutional Review Board of Partners Healthcare Systems ( 2006-P-000066 ) . All patients provided written informed consent for the collection of samples and subsequent analysis . Videos were captured of blood flowing in microfluidic devices under controlled oxygen concentration . Microfluidic fabrication and blood sample collection and handling are described in detail elsewhere [14] . Blood flowed through channels with cross-sectional dimension of 250×12 µm and was driven by a constant pressure head . A juxtaposted network of gas channels allowed control over the oxygen concentration within the blood channel network . Blood samples were collected in EDTA vacutainers and had hematocrit ranging from 18% to 38% . By changing oxygen concentration in situ , we were able to compare the oxygenated and deoxygenated behavior of the same sample and largely control for any differential contributions of the plasma . Videos were captured at a rate of 60 frames per second , with a resolution of about 6 pixels per micron . ( See Videos S1 , S2 , S3 , S4 , S5 , S6 , S7 , S8 for examples . ) We note that the rapid rate of deoxygenation in our studies results in little change in shape for most cells , consistent with existing understanding of heterogeneous hemoglobin polymerization , while the magnitude of the change in stiffness is expected to be more independent of deoxygenation rate [19] , [22] . We developed morphological image processing algorithms to identify a significant fraction of the cells in captured frames of video . See Figure 2 for examples of the segmentation approach . All software was written in MATLAB ( The MathWorks , Natick , Mass . ) . These algorithms implement marker-controlled watershed segmentation , described in detail in reference 13 . Marker images were computed by identifying annular and filled cells of heuristically-determined sizes and shapes . Annular cells were defined as fillable holes not touching the border . Markers for these annuli were created by subtracting border-contacting high-intensity regions and performing morphologic reconstruction on the result . This reconstruction operation used a marker image that was morphologically opened with a 5 µm line segment oriented in increments of 45 degrees . The reconstruction was then subtracted from the border-cleared image . The final result was dilated using a disk with radius 0 . 2 µm . Filled cells were defined using granulometry with a circular structuring element of radius 2 µm . Markers for these cells were selected using two transformations of this opened image: the distance transformation of the thresholded binary image followed by the h-maxima transformation with a height of 3 . Background pixels were identified by the skeletonization of a thresholded binary image . Previously determined cell markers were added to the binary image . The result was eroded using a disk with radius 0 . 5 µm . The skeletonization of this erosion was the background marker image . Foreground and background markers were used to impose minima on the intensity gradient of the original image after background subtraction and histogram equalization . The watershed transformation was then applied to the gradient of the intensity image . The watershed catchment basins , or blobs , were then filtered heuristically by size , shape , and orientation of the objects' convex hulls . First-pass thresholds were determined empirically by manually segmenting several video frames in Adobe Photoshop . Initial size limits were total convex hull area between 5 and 50 µm2 . A measure of convex hull circularity was calculated by comparing the effective radius based on the object area to the effective radius based on the object's perimeter . A circle has a ratio of 1 . All other objects have ratios less than 1 . The initial circularity threshold was set at 0 . 6 . After an initial filtering process , video frames were re-filtered using thresholds for all morphologic characteristics based on the mean convex hull metrics with allowed variation of twice the standard deviation . We then developed machine learning algorithms to track these segmented cells from frame to frame and to compute velocities for individual cells . For each object segmented in each video frame , potential “child” cells were iteratively identified in the subsequent frame and ranked by changes in size , shape , and displacement . Child cells were reassigned if a better “parent” cell was identified . Maximum changes in x- and y-displacement were calculated based on apparent flow rates . Y displacement was limited to 600 µm/s in either direction , and x displacement was limited to 1200 µm/s . Maximum changes in area , perimeter length , and eccentricity were determined by manual tracking of several video frames in Adobe Photoshop as part of a validation check on the tracking algorithm . Area was initially allowed to vary by 50% , perimeter by 50% , and eccentricity by 60% . After all cells in a frame were tracked or determined to be un-trackable , the median inter-frame displacement was computed for all tracked objects . Any tracking events representing displacements that were five times greater than the maximum of the median or the analytic sensitivity threshold ( 1 µm ) were excluded , and the whole frame was retracked with this tighter displacement threshold . Tracking events which represented the extension of existing trajectories were rejected if they represented a change in cell velocity greater than twice the maximum of the median frame displacement or an analytic sensitivity threshold . After excluding these inconsistent tracking events , the whole video frame was retracked iteratively until no trajectory extensions exceeded this threshold . Our measured cell velocities were based on more than 25 million displacements calculated across more than 500 , 000 video frames . We improved and measured the accuracy of our cell velocity measurements a number of different ways , including manual segmentation by an observer of selected video frames and manual tracking by an observer of selected of cells from frame to frame . Inaccuracies in cell velocity measurements can be separated into two categories: errors in the location of a cell , and errors in the assignment of a tracking event for two identified cells . We took a series of steps to reduce the magnitude and bias of this noise and to ensure that it does not influence our results . We measured projected cell density first by thresholding grayscale intensity images using the MATLAB graythresh function . We then combined this thresholded image with the foreground cell markers calculated by our segmentation algorithm . Under steady state conditions , we would expect this density calculation to be relatively stable . Previous studies have reported a coefficient of variation for hematocrit of 3% due to biological variation , and another 3% due to analytic variation achieved with commonly used automated hematologic analyzers [23] . These automated analyzers work with typical volumes of ( 20 , 000 cells*1/0 . 4 total volume/cell volume*80 µm3 cell volume/cell = 4×106 µm3 ) , which is about 100 times larger than the volume projected in a typical video frame . The relationship between an actual three-dimensional volumetric density and a projected two-dimensional density depends on the orientation of the red blood cells and the depth of the flow chamber in the direction of the projection . Under steady state conditions , our density measure is stable over time with a coefficient of variation typically between 10% and 25% .
|
Viewed from a distance , flowing blood looks like a uniform fluid , but up close the cells in the blood change their position and speed somewhat heterogeneously . These individual cell movements may play a role in the physiology and pathophysiology of nutrient and gas transport , clotting , and diseases where normal processes go wrong . To characterize these random motions , we need to follow individual cells in a very crowded suspension—cells usually occupy more than one-third of the volume in blood . We have developed computer software that can separate individual cells in a crowd and track them as they flow . We use this software to analyze blood flow at the level of the cell and find new and possibly important differences between the blood from healthy patients and the blood from patients with sickle cell disease , a disorder in which blood cells become stiff and often stop flowing . We provide evidence that blood from patients with sickle cell disease shows decreased random cellular motions and suggest that this difference may provide a physical basis for the increased risk of occlusion in sickle cell disease .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"computer",
"science",
"hematology/hemoglobinopathies",
"pathology/pathophysiology",
"physics/condensed",
"matter",
"physiology/cardiovascular",
"physiology",
"and",
"circulation",
"physics/fluids,",
"plasmas,",
"and",
"electric",
"discharges",
"physiology/integrative",
"physiology",
"biotechnology/bioengineering",
"mathematics/statistics",
"biophysics",
"computational",
"biology/systems",
"biology",
"radiology",
"and",
"medical",
"imaging"
] |
2009
|
Statistical Dynamics of Flowing Red Blood Cells by Morphological Image Processing
|
Gram-positive bacteria , including Staphylococcus aureus are endemic in the U . S . , which cause life-threatening necrotizing pneumonia . Neutrophils are known to be critical for clearance of S . aureus infection from the lungs and extrapulmonary organs . Therefore , we investigated whether the NLRP6 inflammasome regulates neutrophil-dependent host immunity during pulmonary S . aureus infection . Unlike their wild-type ( WT ) counterparts , NLRP6 knockout ( KO ) mice were protected against pulmonary S . aureus infection as evidenced by their higher survival rate and lower bacterial burden in the lungs and extrapulmonary organs . In addition , NLRP6 KO mice displayed increased neutrophil recruitment following infection , and when neutrophils were depleted the protective effect was lost . Furthermore , neutrophils from the KO mice demonstrated enhanced intracellular bacterial killing and increased NADPH oxidase-dependent ROS production . Intriguingly , we found higher NK cell-mediated IFN-γ production in KO mouse lungs , and treatment with IFN-γ was found to enhance the bactericidal ability of WT and KO neutrophils . The NLRP6 KO mice also displayed decreased pyroptosis and necroptosis in the lungs following infection . Blocking of pyroptosis and necroptosis in WT mice resulted in increased survival , reduced bacterial burden in the lungs , and attenuated cytokine production . Taken together , these novel findings show that NLRP6 serves as a negative regulator of neutrophil-mediated host defense during Gram-positive bacterial infection in the lungs through regulating both neutrophil influx and function . These results also suggest that blocking NLRP6 to augment neutrophil-associated bacterial clearance should be considered as a potential therapeutic intervention strategy for treatment of S . aureus pneumonia .
Acute pneumonia is a leading cause of childhood mortality ( <5 years of age ) accounting for the death of 920 , 136 children annually [1] , and methicillin-resistant Staphylococcus aureus ( MRSA ) has been implicated in severe life-threatening infections , including necrotizing pneumonia and sepsis [2] . In addition , S . aureus infection is also one of the major causes of secondary pneumonia following influenza infection [3] . Furthermore , S . aureus has developed resistance to multiple antibiotics and effective treatment strategies against this bacterium are limited [2 , 4] . Therefore , S . aureus is a serious threat to human health and novel therapeutic strategies are needed . The lung pathology induced by S . aureus is attributed to virulence factors , the intense inflammatory response , and evasion of host defense mechanisms , including neutrophil-mediated ROS-dependent bacterial killing [5 , 6] . Regarding innate immune responses , nucleotide-binding oligomerization domain-like receptor ( NLR ) pyrin domain-6 ( NLRP6 ) is a recently identified NLR present in the cytosol of innate immune cells [7] . Co-transfection of plasmids containing NLRP6 and apoptosis-associated speck-like protein containing card ( ASC ) in 293T cells led to activation of NF-kB and in COS-7L cells to a synergistic increase in caspase-1 activation and IL-1β secretion [8] . Under homeostatic conditions Levy et al have demonstrated that NLRP6 co-localizes with ASC and caspase-1 to form a complex resulting in IL-18 secretion from intestinal epithelial cells , which is essential to prevent development of dysbiosis [9] . Together these results suggest that NLRP6 can co-localize with ASC for inflammasome formation and activation . However , the mechanisms of NLRP6 inflammasome activation and its role in host defense specifically during pulmonary microbial infection has not been explored . Moreover , it remains unknown whether the NLRP6 inflammasome is activated by microbial infections and whether NLRP6 co-localizes with ASC and caspase-1 during such infections to induce pyroptosis . In a mouse model of systemic infection , Anand et al [7] found that the NLRP6 negatively regulates host defense against Listeria and Salmonella infections as NLRP6 KO mice showed higher survival , decreased bacterial burden , and attenuated pathology compared to WT mice . In contrast , a study reported by Wlodarska et al [10] revealed a positive regulatory role of the NLRP6 inflammasome on immune function during enteric infection with Citrobacter rodentium . In this investigation , NLRP6 KO mice were shown to have an increased C . rodentium burden in the intestine , which correlated with extensive mucosal damage in the KO mice compared with WT controls . However , these results cannot be extrapolated to other bacterial infections in different organs , such as the lung , because Gram-positive bacteria have unique virulence factors and immune evasion strategies and the route of administration of bacteria also dictates host responses . Thus , the role of the NLRP6 inflammasome in pulmonary immunity against Gram-positive infections remains unclear . To this end , we have used NLRP6 KO mice to demonstrate how S . aureus exploits the NLRP6 inflammasome to dampen neutrophil function and enhance pyroptosis and necroptosis to increase mortality during acute bacterial pneumonia . Our results show NLRP6 as a potential therapeutic target for treatment of S . aureus-infected pneumonic patients .
To determine whether NLRP6 is upregulated in the lungs of pneumonic patients , we stained human pneumonic and normal lung tissue sections with anti-NLRP6 antibody and found that NLPR6 was upregulated in key innate immune cells in the lungs , such as neutrophils ( lipocalin-2+ ) , macrophages ( CD68+ ) , and epithelial cells ( CD326+ ) ( Fig 1A ) . Next , we determined if NLRP6 expression is upregulated in S . aureus-induced pneumonia in mice . In line with the results seen in human pneumonic lung sections , NLRP6 was upregulated in neutrophils ( Ly6G+ ) , macrophages ( F4/80+ ) , and epithelial cells ( CD 326+ ) of mouse lungs following S . aureus infection ( Fig 1B ) . Consistent with the immunofluorescence results , the expression of NLRP6 was also increased in lung lysates of human pneumonic patients , S . aureus-infected human cell lines ( THP-1 and HL-60 ) , and mouse bone marrow-derived macrophages ( BMDMs ) ( Fig 1C–1E ) . To investigate whether NLRP6 triggers inflammasome activation during S . aureus infection , we infected BMDMs from WT and NLRP6 KO ( KO ) mice with S . aureus ( MOI: 20 ) and measured the extent of caspase-1 activation at 8 hours post-infection . Both cleaved caspase-1 ( p-20 ) in macrophage lysates and IL-1β levels in culture supernatants following infection were attenuated in the NLRP6 KO samples compared to the WT control ( Fig 1F and 1G ) indicating the activation of the NLRP6 inflammasome by S . aureus . Similarly , IL-18 was also reduced in NLRP6 KO BMDMs after S . aureus infection ( S1A Fig ) . Since IL-1β was sharply reduced in the KO BMDMs , we determined if NLRP3 inflammasome is still intact in these BMDMs . IL-1β production in NLRP6 KO BMDMs after treatment with specific NLRP3 agonists ( ATP and Nigericin ) was comparable with that of WT BMDMs suggesting that NLRP3 is indeed intact in NLRP6 KO macrophages ( S1B Fig ) . S . aureus has several virulence components , such as α-hemolysin ( hla ) , clumping factor B , β toxin , phenol-soluble modulins , and panton-valentine leukocidins [5] that could potentially activate the NLRP6 inflammasome . Because hla has been reported to activate the NLRP3 inflammasome in human and mouse monocytic cells under similar conditions [11 , 12] , we investigated whether hla can also activate the NLRP6 inflammasome in BMDMs and found this to be the case ( Fig 1E and 1F ) . Consistent with the in vitro results , NLRP6 KO mice showed lower levels of IL- 1β in bronchoalveolar lavage fluid ( BALF ) after infection with S . aureus , providing evidence of NLRP6 inflammasome activation in vivo ( Fig 1H ) . However , there was no difference observed in the levels of IL-18 between WT and the KO mice ( S1C Fig ) . ASC is known to be an integral part of NLRP3 inflammasome signaling although it is dispensable for NLRC4 inflammasome activation [13] . We infected BMDMs with S . aureus and performed immunofluorescence assay to observe whether NLRP6 co-localizes with ASC and caspase-1 . We found that NLRP6 co-localized with ASC and caspase-1 during S . aureus infection ( Fig 1I and 1J and S1D Fig ) . To assess the role of the NLRP6 inflammasome in pulmonary host defense against S . aureus , we infected WT , ASC KO , and NLRP6 KO mice intratracheally with a lethal dose of S . aureus ( USA 300 ) ( 2X108 CFUs per mouse ) and observed the survival patterns for 10 days . Although all WT mice died within 3 days , 70% of NLRP6 KO mice survived longer than 10-days post-infection ( Fig 2A ) . Furthermore , ASC KO mice displayed a survival pattern similar to that of the NLRP6 KO mice ( Fig 2A ) . To determine whether the difference in survival is due to differences in the bacterial burden in various organs , we measured the bacterial burden in the lung , BALF , and extra-pulmonary organs after infecting mice with a sub-lethal inoculum ( 5X107CFU ) of S . aureus . As compared to NLRP6 KO mice , WT mice had higher bacterial burdens in the lungs , BALF , and liver at both 12 and 24-hours post-infection ( Fig 2B–2D ) . Accordingly , the total protein in the BALF , which is a measure of pulmonary leakage , was higher in WT mice compared to their NLPR6 KO counterparts ( Fig 2E ) . To determine whether the detrimental effects of NLRP6 are bacterial strain specific , we infected WT and KO mice with a methicillin-susceptible Staphylococcus aureus strain ( MSSA , Newman strain ) and measured the bacterial burden in lungs and BALF at 24-hours post-infection . Consistent with the results seen with the USA 300 strain , the bacterial burden of the methicillin-sensitive strain was also higher in the lungs and BALF of WT mice compared to that of KO mice ( Fig 2F and 2G ) . NLRP6 inflammasome has been implicated in regulating intestinal microbiota [9 , 14] . To determine if observed difference in the phenotype between WT and NLRP6 KO mice is due to difference in gut microbiota , we co-housed WT and NLRP6 KO mice for 4 weeks and infected them with S . aureus . As observed with non-co-housed mice , co-housed NLRP6 KO mice had significantly less bacterial burden in the lungs and BALF as compared to co-housed WT mice ( Fig 2H and 2I ) . These data suggest that NLRP6 regulates pulmonary S . aureus infection independent of microbiota composition . Since we found upregulation of NLRP6 in both hematopoietic ( neutrophils and macrophages ) and non-hematopoietic cells ( epithelial cells ) ( Fig 1B ) , we sought to determine if NLRP6 from each of these compartments is detrimental to bacterial clearance of S . aureus-induced pneumonia . Using bone marrow chimeric mice , we found that WT mice that received KO bone marrow ( KO→WT ) had lower bacterial burdens in the lungs and BALF than WT mice that received WT bone marrow ( WT→WT ) ( Fig 2J and 2K ) . However , KO mice that received WT bone marrow ( WT→KO ) showed no increase in bacterial burden in the lungs and BALF compared to KO mice that received KO bone marrow ( KO→KO ) ( Fig 2J and 2K ) . Together , these results suggest that NLRP6 derived from both hematopoietic and non-hematopoietic compartments is detrimental for bacterial clearance during S . aureus pneumonia . Neutrophils have been shown to be essential for containing pulmonary Staphylococcal infections [15 , 16] . To determine if neutrophils confer augmented host protection in NLRP6 KO mice , we depleted neutrophils in the KO mice prior to infection with a lethal dose of S . aureus and observed survival . We found that depletion of neutrophils reversed the survival benefit observed in the NLRP6 KO mice suggesting that protection is neutrophil-dependent ( Fig 3A ) . Since neutrophils are essential for survival , we investigated if disruption of NLRP6 affects recruitment of neutrophils into alveolar spaces during S . aureus pneumonia . Compared to WT mice , KO mice had more leukocytes , including neutrophils , and macrophages recruited into alveolar spaces ( Fig 3B–3D ) . Furthermore , to measure neutrophil accumulation in the lung parenchyma , we performed a myeloperoxidase ( MPO ) assay and found that KO mice had more MPO activity than WT mice ( Fig 3E ) . Since NLRP6 was found to be upregulated in neutrophils and macrophages in the lungs , we wanted to know whether deletion of NLRP6 affects the function of these cells . To this end , we compared the intracellular killing ability of bone marrow derived-neutrophils ( BMDNs ) and BMDMs isolated from both WT and NLRP6 KO mice following infection with S . aureus ( MOI:10 ) . Our results indicate that neutrophils , but not macrophages , from NLRP6 KO mice had improved killing ability compared to WT cells as depicted by the reduction in intracellular CFUs in KO neutrophils ( Fig 4A and S2A Fig ) . However , the rate of phagocytosis was similar in neutrophils from both groups ( WT and KO ) ( S2B Fig ) . It is known that neutrophils use NADPH oxidase-dependent reactive oxygen species ( ROS ) to kill S . aureus intracellularly [17–19] . To determine if differences in killing ability of WT and NLRP6 KO neutrophils is due to an alteration in ROS production , we compared ROS production by neutrophils from WT and KO mice after infection with S . aureus and found that KO neutrophils produced more ROS than WT neutrophils ( Fig 4B ) . To confirm these in vitro findings , we assessed the expression of NADPH oxidase components in lung homogenates from infected mice by western blotting . The expression of p47phox , p67phox , and gp91phox was increased in the lungs from NLRP6 KO mice compared to those from WT mice , supporting the finding of higher ROS production by KO neutrophils ( Fig 4C ) . IFN-γ has been shown to activate phagocytic cells during intra-pulmonary S . aureus infection [20] . Therefore , we measured the levels of IFN-γ secreted in the BALF of WT and NLRP6 KO mice after infection with S . aureus . Interestingly , we found that IFN-γ production was higher in KO mice compared to that of WT mice ( Fig 4D ) . The production of IFN-γ requires activation of MAPK pathways [21] . Accordingly , we found higher MAPK activity in NLRP6 KO lungs than in WT lungs ( Fig 4E ) . IFN-γ has been shown to induce ROS production in human mast cells during Staphylococcal infection [22] . Based on this observation , we assessed if IFN-γ contributes to the enhanced bacterial killing by neutrophils through induction of ROS . To this end , BMDNs from WT and KO mice were pre-treated with IFN-γ and subsequently infected with S . aureus ( MOI of 10 ) followed by assessment of ROS production . Treatment of neutrophils with IFN-γ increased NADPH oxidase activity and ROS production ( Fig 4F and 4G ) . The activation of phagocytes by IFN-γ involves activation of signal transducer and activator of transcription ( STAT ) proteins [23] . Supporting this observation , higher expression of phospho-STAT3 was found in the lungs of KO mice compared to WT ( Fig 4E ) . Collectively , these results suggest that genetic ablation of NLRP6 increases bacterial killing by neutrophils through increased IFN-γ and ROS production , which are associated with higher NADPH oxidase activity . Our in vivo experiments demonstrate that NLRP6 KO mice exhibit improved bacterial clearance ( Fig 2B–2D ) , higher neutrophil accumulation , as well as enhanced IFN-γ production ( Figs 3C and 4D ) . Moreover , our in vitro experiments using BMDNs demonstrate that IFN-γ enhance bacterial killing by neutrophils through increased ROS production ( Fig 4G ) . Therefore , we examined if blocking IFN-γ hinders bacterial clearance in the KO mice following infection with S . aureus . In this context , we administered anti-mouse IFN-γ antibody ( 100 μg/mouse i . p . ) to one group of the KO mice and a similar volume of isotype antibody to another group of mice 12 hours before infection with S . aureus . Blocking of IFN-γ in the KO mice increased the bacterial burden in the lungs and BALF suggesting that IFN-γ mediates bacterial clearance in the KO mice ( Fig 5A and 5B ) . We next sought to identify the cellular sources of IFN-γ during S . aureus infection . To this end , we performed flow cytometric analysis of lung cells from WT and KO mice following infection with S . aureus and found that NLRP6 KO mice had more IFN-γ-positive NK and CD4+T cells ( Fig 6A–6D ) . In this regard , Nguyen et al reported NK cells as the major source of IFN-γ during S . aureus infection [20] . We also found CD8+T cells and γδT cells produce IFN-γ , although there was no difference between WT and KO mice in the total number of IFN-γ-positive CD8 or γδT cells in the lungs ( S3A–S3D Fig ) . To determine whether increased IFN-γ production in the KO mice is due to higher MAPK activity in NK and CD4 T cells , we isolated these cells from the lungs of WT and KO mice and treated them with MAPK inhibitor prior to infection with S . aureus . However , blocking MAPK did not reduce IFN-γ secretion by NK and CD4 T cells ( S3E Fig and S3F Fig ) . Furthermore , we measured the number of NK and CD4 T cells in the lungs after infection through flow cytometry and found that the KO mice had more NK and CD4 T cells accumulated in the lungs compared to that of WT mice ( Fig 6E ) . These results together suggest that increased IFN-γ observed in the KO mice is due to enhanced numbers of NK and CD4 T cells recruited during infection . Since we found decreased neutrophil recruitment and increased protein leakage in the lungs of WT mice compared to KO mice ( Figs 3C and 2E ) , we hypothesized that cell death in the lungs may be responsible for these results during S . aureus infection . In this regard , LDH , IL-1α , and HMGB-1 are intracellular molecules released exclusively after cell death and are thus termed alarmins [24 , 25] . Therefore , we measured the levels of these alarmins in an in vivo setting and found their expression to be increased in WT mice compared to NLRP6 KO mice . This suggests that NLRP6 enhances inflammatory modes of cell death during S . aureus-induced pneumonia ( Fig 7A–7C ) . We also measured the extent of cell death in BMDNs following infection with S . aureus . Neutrophils from WT mice exhibited increased cell death , as seen by increased cytotoxicity ( LDH release ) , compared to that of KO neutrophils ( Fig 7D ) . Next , to determine the nature of cell death , BMDNs from WT and KO mice were pre-treated with either Ac-YVAD-CMK ( Caspase-1 inhibitor ) or Necrostatin-1s ( Nec-1s , necroptosis inhibitor ) and infected with S . aureus . Pre-treatment of neutrophils with Ac-YVAD-CMK or Nec-1s reduced the cell death in WT neutrophils suggesting that the nature of cell death is both pyroptosis and necroptosis ( Fig 7D ) . In this regard , it is reported that caspase-1 and gasdermin-D mediate pyroptosis during bacterial infection [26–28] . To validate that NLRP6 induces pyroptosis , we assessed the expression of caspase-1 and gasdermin-D in BMDMs from WT and NLRP6 KO mice after infection with S . aureus ( MOI of 20 ) for 8 hours . Both cleaved caspase-1 ( Fig 1F ) and gasdermin-D ( Fig 7E ) expression were higher in WT mice compared to KO mice . Also , we performed immunofluorescence microscopy on BMDMs to quantify caspase-1 and gasdermin-D expression and found increased caspase-1 and gasdermin-D- positive cells in BMDMs from WT mice compared to NLRP6 KO mice ( Fig 7F and 7G; S4A and S4B Fig ) . It has been reported that S . aureus induces pathology in the lungs by a distinct cell death mechanism known as necroptosis [29] . Receptor-interacting-serine-threonine kinase-1 ( RIP1 ) , RIP3 , and mixed lineage kinase-domain like protein ( MLKL ) are the core protein kinases that initiate and execute necroptosis [29–31] . Recently , caspase-8 has been shown to negatively regulate necroptosis during Salmonella infection in an enteritis model [32] . Thus , to explore whether NLRP6 can enhance necroptosis in the lungs during S . aureus infection , we assessed the expression of phospho-MLKL RIP1 , RIP-3 , and caspase-8 in lung homogenates obtained from S . aureus-infected WT and NLRP6 KO mice through western blotting . Interestingly , both phospho-MLKL and RIP-3 were higher in the lungs of WT mice compared to NLRP6 KO mice ( Fig 7H ) . To further confirm this finding at the cellular level , we infected both WT and NLRP6 KO BMDMs with S . aureus for 8 hours and quantified necroptosis using immunofluorescence microscopy . The immunofluorescence assay also revealed more phospho-MLKL and RIP-3-positive cells in WT macrophages compared to NLRP6 KO macrophages after S . aureus infection , confirming that NLRP6 increases necroptosis ( Fig 7I and S4C Fig ) . Since S . aureus has been shown to induce necroptosis in human cells , such as THP-1 cell lines [29] , we examined if activation of the necroptosis pathway occurs in lungs of human patients during pneumonia . Immunofluorescence microscopy conducted on lung sections obtained from pneumonia patients displayed more necroptosis , as evidenced by increased phospho-MLKL and RIP-3 expression , compared to that of healthy control lungs ( Fig 7J ) . We found that NLRP6 enhances both pyroptosis and necroptosis pathways during pulmonary S . aureus infection ( Fig 7A–7I ) . Because pyroptosis has been found to be beneficial for bacterial clearance during intracellular bacterial infections [28] , we examined whether pyroptosis is advantageous during S . aureus infection . Blockade of pyroptosis in mice by intra-peritoneal administration of Ac-YVAD-CMK ( caspase-1 inhibitor ) 12 hours prior to infection with S . aureus resulted in reduced pulmonary leakage , LDH release , and bacterial burden in lungs and BALF ( Fig 8A–8D ) . In addition , blockade of pyroptosis suppressed cytokine secretion and enhanced survival of WT mice ( Fig 8E–8H ) , confirming that pyroptosis is detrimental during S . aureus infection . It should be noted that the bacterial burden in WT mice receiving Ac-YVAD-CMK was still high when compared with NLRP6 KO mice ( Fig 8C ) , suggesting that an NLRP6-dependent , but caspase-1-independent , mechanism also exists to increase susceptibility to S . aureus infection . Next , to confirm that NLRP6-mediated necroptosis is detrimental for host defense , we blocked necroptosis using an MLKL inhibitor ( GW806742X ) 12 hours before infection with S . aureus . There was a decrease in total protein and LDH release in the BALF of WT mice treated with GW806742X , which was comparable to that seen in NRLP6 KO mice ( Fig 9A and 9B ) . Moreover , blocking necroptosis reduced the bacterial burden in lungs and BALF and ameliorated the inflammatory cytokine levels in WT mice ( Fig 9C–9G ) . Also , leukocyte recruitment ( neutrophils and macrophages ) and survival were also increased in WT mice treated with necroptosis inhibitor ( Fig 9H–9K ) . Inhibition of necroptosis using a RIP-1 inhibitor ( Nec-1s ) also reduced total protein leakage and LDH release in the BALF ( S5 Fig ) . Collectively , these results reveal that NLRP6-mediated pyroptosis and necroptosis are detrimental to bacterial clearance and host survival during pulmonary S . aureus infection .
Lung diseases induced by Gram-positive pathogens are an important cause of morbidity and mortality in both immunocompetent and immunocompromised individuals [13 , 33] . Although antibiotics decrease the mortality rates of bacterial pneumonia , the efficacy is somewhat limited due to the substantial number of immunocompromised individuals , growing number of elderly patients , and the rise of multi-antibiotic resistant bacterial strains . Thus , alternative therapeutic approaches , including the manipulation of host signaling events , are needed . However , detailed understanding of the host innate immune response is critical for the design of potential therapeutic interventions . Because the lung is continuously exposed to pathogens and their virulence factors , this organ possesses a multifaceted host defense system . Moreover , a successful immune response in the lung is critical for efficient clearance of microbial pathogens and therefore , the innate immune system possess germline-encoded pattern-recognition receptors . The NOD-like receptors ( NLRs ) , including inflammasomes , are specialized cytosolic pattern-recognition receptors/sensors necessary for clearance of invading cytosolic pathogens . Under normal homeostatic conditions , the NLRP6 inflammasome is highly expressed in intestinal epithelial cells [9 , 14 , 34 , 35] where it co-localizes with ASC and caspase-1 [9] . It is also expressed in immune cells including neutrophils , T-cells , macrophages , and dendritic cells [7] . Despite high expression of NLRP6 in the lower respiratory tract , the role of NLRP6 in lung inflammation has not previously been explored . In the current study , we demonstrate that NLRP6 is upregulated in neutrophils , macrophages , and epithelial cells in the lungs of human pneumonia patients . Furthermore , NLRP6 is upregulated in myeloid and non-myeloid cells in the lungs and co-localizes with ASC in a mouse model of pulmonary S . aureus infection . We also found that the important virulence factor , α-hemolysin , can activate the NLRP6 inflammasome . The immune response to S . aureus is manifested by vascular leakage , neutrophil recruitment into the alveolar space , and upregulation of cytokines and chemokines [5 , 6] . The current study demonstrates that the NLRP6 inflammasome increases susceptibility to S . aureus-induced lung infection . Delving into the mechanisms underlying this , we found that NLRP6 dampens NK cell-mediated IFN-γ secretion thereby hindering ROS-dependent bacterial clearance by neutrophils . Moreover , our study identifies NK cells and CD4-T cells as the primary source of IFN-γ during acute pulmonary S . aureus infection . In agreement with these findings , studies of Listeria and Salmonella infections [7] have also shown that NLRP6 signaling is detrimental to host defense . Nonetheless , in a non-infectious model , the NLRP6 inflammasome was found to be important for epithelial self-renewal , proliferation , and mucus secretion , which were essential for protection against chemical-induced colitis [9 , 34] . Studies from different groups have shown that NLRP6 inflammasome regulates gut microbiota composition [7 , 9 , 14] . NLRP6 KO mice were found to have different microbiota configuration which make them more susceptible to chemical-induced colitis compared to the WT mice [14] . In contrast , recent studies have demonstrated that NLRP6 and ASC-related inflammasome do not regulate gut microbiota composition [36 , 37] . Although it remains debatable whether the NLRP6 inflammasome influence gut microbiome , the reported difference in microflora composition in the KO mice can be nullified by co-housing the mice together with WT for 4 weeks [7] . In addition to colitis , microbiota have been shown to influence various disease conditions such as rheumatoid arthritis [38] , diabetes [39] , inflammatory bowel disease [40] , and colorectal cancer [41] . In our study , however , co-housing of WT with NLRP6 KO mice did not change the resistant phenotype of the KO mice against S . aureus . In this context , similar report has been demonstrated by Anand et al , showing that co-housing does not alter NLRP6 mediated immune mechanism during Salmonella and Listeria infection [7] . It is widely accepted that hematopoietic and non-hematopoietic ( stromal ) cells in the lung produce numerous proinflammatory mediators , including cytokines and chemokines . Although hematopoietic cells secrete chemokines or neutrophil chemo-attractants , including CXCL1/KC and CXCL2/MIP-2 , the stromal cells ( alveolar epithelial cells ) secrete other neutrophil chemo-attractants , such as CXCL5/LIX and CXCL15/lungkine [42] . Our observations indicate that NLRP6 from both cell types contributes to the enhanced susceptibility to S . aureus-induced pneumonia . These conclusions are consistent with previous studies of the role of hematopoietic and non-hematopoietic cells in the context of bacterial infections in the lungs . In this context , NLRP6 in both hematopoietic and non-hematopoietic cells increases susceptibility to Listeria and Salmonella infections [7] . Similarly , CXCL1/KC secreted by both hematopoietic and stromal cells was found to be crucial for bacterial elimination and neutrophil accumulation in the lungs following Klebsiella pneumoniae infection [43] . Nevertheless , it is clear from this investigation that neutrophil accumulation and function are critical for host protection against S . aureus . Pyroptosis and necroptosis are two distinct inflammatory modes of cell death . While pyroptosis is mediated by caspase-1 [26–28] and executed by gasdermin-D [26 , 27] , necroptosis is regulated by RIP1 , RIP3 , and MLKL ( caspase-1 independent ) [29 , 30] . Pyroptosis has been shown to play an essential role in limiting several intracellular bacterial infections such Salmonella typhimurium , Legionella pneumophila , and Burkholderia thailandensis [28] . However , extensive caspase-1 activation and subsequent pyroptosis have also been associated with the severity of several diseases such as myocardial infarction [44] , inflammatory bowel disease [45] , and endotoxic shock [46] . Pertaining to these observations , we report that during S . aureus infection , NLRP6-mediated pyroptosis is detrimental for host survival . Furthermore , blocking pyroptosis reduced the hyper-inflammatory milieu and subsequently increased survival suggesting that pyroptosis triggers exaggerated inflammation during S . aureus infection . This difference in the role of pyroptosis can be attributed to differences in pathogenic properties and life styles of bacterial pathogens . While S . aureus is predominantly an extracellular pathogen , studies have shown that it can also survive intracellularly [47] and can resist anaerobic conditions [48] . Necroptosis induced by S . aureus is responsible for pathology in the lung [29]; however , its relationship with inflammasomes was previously unknown . Although ASC and NLRP3 have been linked with pore-forming toxin-induced necroptosis [31] , the precise role of inflammasomes in the induction of necroptosis is not clear . In this study , we used both in vivo and in vitro experiments to show that NLRP6 mediates necroptosis of immune cells during acute pulmonary S . aureus infection . Moreover , S . aureus exploits NLRP6 to drive necroptosis , which is accompanied by an intense inflammatory response and loss of macrophages and neutrophils . It is possible that reduced cell death in NLPR6 KO mice attributed to higher leukocyte accumulation in the lungs of these mice . Since TNF-α has been shown to induce necroptosis [49 , 50] , the reduction of TNF-α found in the NLRP6 KO mice suggests that NLRP6 can trigger necroptosis via the TNF-α pathway . However , more comprehensive future studies will be needed to identify the detailed molecular mechanisms underlying NLRP6-mediated necroptosis . Future studies are also needed to determine whether other toxins or virulence factors produced by S . aureus can also activate the NLRP6 inflammasome . In conclusion , the present study reveals the detrimental role of NLRP6 during S . aureus pneumonia ( S6 Fig ) . Furthermore , NLRP6 in both hematopoietic and resident lung compartments contributes to S . aureus-induced lung inflammation . Not only does NLRP6 subdue neutrophil function by dampening IFN-γ and ROS production , it also triggers pyroptosis and necroptosis in the lungs that may lead to hyper-inflammation , loss of neutrophils , and mortality . However , future studies are essential to determine whether NLRP6 interacts with other inflammasomes such as NLRP3 and/or NLRC4 to induce pyroptosis and necroptosis during S . aureus infection . Comprehensive studies using specific double- or triple-KO mice would be useful to delineate these interactions in a conclusive manner . Furthermore , extending upon our findings , we propose that functional single nucleotide polymorphisms in human NLRP6 may have effects on host defense mechanisms against gram-positive microbes .
Mouse experiments were conducted in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . Animal protocols were approved by the Institutional Animal Care and Use Committee ( IACUC ) at Louisiana State University ( protocol number 16–072 ) . All animal experiments were performed in a manner to ensure minimal pain and distress . Nlrp6-/- , Asc-/- , and Caspase-1/11-/- were obtained from the Millennium Pharmaceuticals ( Cambridge , MA ) whereas C57Bl/6 mice were obtained from Taconic ( Rensselaer , NY ) and Jackson ( Bar Harbor , ME ) Laboratories . THP-1 ( human monocytic ) cells and HL-60 ( human neutrophil-like ) cells were purchased from ATCC ( Manassas , VA ) . Lysates of human healthy control tissue and pneumonic lung tissue were obtained from Novus Biologicals , CO . Immunofluorescence microscopy of lung sections was performed as described previously [51] . Human lung sections from lungs without evidence of infection or injury ( control ) or from patients who died due to ALI/ARDs ( pneumonic ) were used from BioChain Institute Inc . ( Newark , CA ) . Mouse lung sections were from saline-challenged or S . aureus-infected mice . Lung sections were incubated with anti-NLRP6 ( Abgent , CA ) and antibodies for surface markers including anti-lipocalin Ab for neutrophils ( R&D Systems , MN ) , or anti-CD68 Ab for macrophages ( BioLegend , CA ) , and anti-CD326 Ab for alveolar epithelial cells ( BioLegend , CA ) . For mouse lung sections , we used antibodies for surface markers including anti-Ly6G for PMNs ( BioLegend , CA ) , anti-F4/80 for macrophages ( BioLegend , CA ) , and anti-CD326 for alveolar epithelial cells along with anti-NLRP6 Ab ( Sigma , MO ) . Appropriate Alexa-conjugated secondary antibodies ( Invitrogen , CA ) were used . For detection of necroptosis and pyroptosis through an immunofluorescence assay , antibodies against mouse NLRP6 and ASC ( Sigma , MO ) , p-MLKL , ( Abcam , MA ) , RIP3 ( Cell signaling , MA ) , caspase-1 ( Adipogen , CA ) , and gasdermin-D ( Santa cruz , CA ) were used . Excess antibodies were washed off , and the cells were labeled with secondary antibodies , such as mouse IgG/IgM ( H+L ) Alexa fluor 488 , 568 ( Invitrogen , CA ) . Images were obtained using an Axiocam digital camera ( Zeiss , NY ) connected to a Zeiss Axioskop 2 Plus microscope . BMDMs or lungs were harvested at designated time points and homogenized in PBS containing 0 . 1% Triton X-100 ( phosphatase and protease inhibitor cocktail added ) . After centrifugation the supernatants were used for immunoblotting . Total protein content in the supernatant was measured using a BCA protein assay kit ( Thermofisher , NY ) to ensure that equal amounts of proteins were loaded onto 10% SDS-PAGE gels . Proteins were transferred to polyvinylidene fluoride membrane according to the protocol provided by Bio-Rad . Appropriate primary antibodies against mouse NLRP6 ( Sigma , MO ) , phospho-MLKL ( Abcam , MA ) , RIP3 , RIP1 , P47phox , P67phox , gp91phox , phospho-p38 MAPK , phospho-JNK , phospho-Stat3 , caspase-8 , GAPDH ( Cell Signaling , MA ) , caspase-1 ( Adipogen ) , and gasdermin-D ( Santa Cruz , CA ) were added to the membrane and incubated overnight at 40 C . Appropriate secondary antibodies were used , and the films were developed using ECL plus western blot detection system ( ThermoFisher , NY ) . IL-1β , TNF-α , IFN-γ , IL-1α , and IL-6 were measured in BALF supernatants by ELISA according to the manufacturer’s protocol ( eBioscience , CA ) . Eight to twelve-week-old WT mice ( C57BL/6 background ) were used . Equal age- and gender-matched NLRP6 KO , ASC KO , and Caspase-1/11 DKO mice on the C57BL/6 background were used throughout the experiments . Mice were kept on a 12:12 hour light/dark cycle under specific pathogen-free condition with free access to food and water . All animal experiments were approved by the Institutional Animal Care and Use Committee ( IACUC ) at Louisiana State University . To induce pneumonia , mice were anesthetized using ketamine ( 100 mg/kg ) and xylazine ( 5 mg/kg ) prior to intratracheal inoculation of S . aureus ( USA 300 strain ) . A small midline incision was made on the ventral aspect of the neck and excess fat was separated to expose the trachea . Fifty microliters of bacterial suspension containing 5X107 CFU of log phase S . aureus in isotonic saline was injected into the lungs by piercing trachea using a 28 . 5-gauge needle . At 12- and 24-hours post-infection , mice were euthanized to collect BALF , lungs , and liver for quantification of bacterial burden and leukocyte recruitment . BALF and homogenized organs were serially diluted and plated onto Tryptic soy agar ( TSA ) plates for bacterial enumeration . For survival experiments , we used 2 X 108 CFUs/mouse of S . aureus and observed survival for 12 days post-infection . BALF was collected as described in our previous publication [51] . Briefly , after specific time points , mice were humanly euthanized , and the trachea was exposed . Using a 20-gauge catheter , 0 . 8 ml of PBS ( heparin and dextrose added ) was instilled inside the lungs and collected in a clean tube . The process was repeated a total of four times so that a minimum of 2 . 8–3 ml of BALF was collected from each mouse . Total leukocyte count was performed in a hemocytometer using 10 μl of BALF and the differential count was done under light microscopy using cytospin slides stained with DiffQuik reagent . The remaining cell-free BALF was preserved at -80o C for cytokine analysis . Co-housing experiments were performed as described by Anand et al [7] . In brief , age and sex matched WT and NLRP6 KO mice were co-housed together in 1:1 ratio for 4 weeks . After co-housing , WT and the KO mice were infected with S . aureus ( i . t . ) and euthanized 24 hours post-infection to measure the bacterial burden . Bone marrow chimeras were generated as described previously [51 , 52] . In brief , the recipient mice were lethally irradiated with a 1000-rad inoculum from a cesium source . Bone marrow cells collected from healthy donor mice were injected into recipient mice via tail vein ( 8 million cells per mouse ) . The chimeric mice were kept under 0 . 2% neomycin sulfate treatment for 15 days after transplantation . After 8 weeks of transplantation , the chimeric mice were infected with 5X 107 CFU of S . aureus . The mice were euthanized at 24-hour post-infection to estimate cellular recruitment and bacterial burden in the lungs . For neutrophil depletion , mice were treated with 500 μg of anti-Ly6G antibody ( clone 1A8 , BioLegend , CA ) [53] intraperitoneally 24 and 2 hours prior to infection with lethal inoculum of S . aureus ( 2 X 108 CFUs/mouse ) . For IFN-γ inhibition , mice were treated with 100μg of IFN-γ antibody ( BioXCell , NH ) 12 hours prior to infection with S . aureus . For caspase-1 inhibition , mice were treated with 100 μg of caspase-1 inhibitor ( Ac-YVAD-CMK , Cayman chemical , MI ) 12 hours prior to infection with S . aureus . For MLKL inhibition , 100 μl of 100 μM MLKL inhibitor ( GW806742X , Adipogen , CA ) [31] was injected into each mouse i . p . 12 hours prior to infection with S . aureus . For RIP1 inhibition , mice were treated with 300 μg of necroptosis inhibitor ( Nec-1/necrostatin-1s , Calbiochem , MA ) [29] 18 hours before and at the time of bacterial infection , as published elsewhere [29] . Mice were euthanized to collect BALF and organs 24 hours post infection . For MPO activity assay , lungs obtained from WT and KO mice after infection were homogenized . The supernatants obtained after centrifugation were mixed with 50 mM potassium phosphate buffer ( with 0 . 5% hexadecyltrimethylammonium bromide ) in 1:5 ratio and then centrifuged again . The supernatants were transferred to a 96-well plate . Absorbance was measured using a spectrophotometer at 460 nm after adding hydrogen peroxide/O-dianisidine hydrochloride buffer . The intracellular killing assay was performed as described [54] with slight modification . Briefly , bone marrow neutrophils from NLRP6 KO and WT mice were isolated , infected with S . aureus ( MOI: 10 ) , and treated with gentamicin for 30 minutes at designated time points post-infection ( 30 min , 60 min , 90 min , and 120 min ) to kill extracellular bacteria . Cells were washed several times with sterile PBS to remove excess gentamicin and were then lysed with 0 . 1% triton X to release intracellular bacteria . The lysate was serially diluted with PBS , plated onto the TSA , and incubated at 370 C overnight for bacterial load estimation . Total neutrophils , isolated from bone marrow of WT and NLRP6 KO mice using a magnetic negative selection cell isolation kit ( STEMCELL Technologies , Vancouver , Canada ) , were infected with S . aureus ( MOI 10 ) . Total intracellular ROS production was quantified using a Fluorometric kit ( AA Bioquest , CA ) . The effect of IFN-γ on ROS production was determined by pretreating neutrophils with either 20 ng/ml of recombinant mouse IFN-γ ( R&D Systems , MN ) or an equal volume of PBS for 30 minutes before infection with S . aureus . The total ROS production was quantified after 30 minutes of infection using a spectrophotometer . BMDNs were isolated from WT and NLRP6 KO mice and pretreated with Nec-1s ( 300 μM ) or Ac-YVAD-CMK ( 100 μg/ml ) or DMSO for 30 minutes before infection with S . aureus ( MOI: 20 ) . The percentage of cytotoxicity in BMDNs and LDH release into the alveolar space after S . aureus infection were measured using the Cytotox-ONETM homogenous membrane integrity assay kit ( Promega , WI ) . HMGB-1 was measured using a commercially available kit from Chondrex Inc , WA . Data are represented as mean ± SEM . The Mann-Whitney test was used to compare the bacterial burden between two groups . Student’s t-test was used whenever the data were parametric in nature . We used one-way ANOVA followed by Bonferroni’s multiple comparison test wherever more than two groups were involved . All statistical analyses were performed using GraphPad Prism 7 software . The survival curve was analyzed using Log-rank ( Mantel Cox ) test . All experiments were performed thrice . Significant differences are indicated by * ( p<0 . 05 ) , ** ( p<0 . 01 ) , *** ( p<0 . 001 ) , and **** ( p<0 . 0001 ) .
|
Gram-positive bacteria , including Staphylococcus aureus remain a major cause of acute pneumonia worldwide . Due to emergence of multidrug-resistant strains , alternative strategies for treatment of S . aureus pneumonia are needed . To this end , it may be possible to harness host defenses to eradicate the infection instead of directly targeting the bacteria . Neutrophils are a crucial innate immune cell type and serve as a first line of defense against bacterial lung infection . NLRP6 is a recently identified member of Nod-like receptor family . Nonetheless , the molecular and cellular immunological mechanisms by which the NLRP6 regulates neutrophil-mediated host immunity during acute S . aureus pneumonia remain elusive . We found that NLRP6 gene-deficient/knockout ( KO ) mice demonstrate increased survival and lower bacterial burden in the lungs along with enhanced neutrophil recruitment during acute S . aureus pneumonia . Moreover , neutrophils from NLRP6 KO mice showed increased bactericidal ability compared to those from controls . Similarly , NLRP6 KO mice demonstrated decreased cell death through pyroptosis and necroptosis following infection . Blocking of these cell death mechanisms in WT mice resulted in increased survival and decreased bacterial burden in the lungs following infection . Therefore , our study provides novel insights into the novel mechanisms mediated by NLRP6 , which serves as a negative regulator of neutrophil-mediated host defense during Gram-positive pneumonia .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"blood",
"cells",
"cell",
"death",
"medicine",
"and",
"health",
"sciences",
"immune",
"cells",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"cell",
"processes",
"microbiology",
"staphylococcus",
"aureus",
"pulmonology",
"salmonellosis",
"pneumonia",
"methicillin-resistant",
"staphylococcus",
"aureus",
"bacterial",
"diseases",
"inflammasomes",
"bacteria",
"neutrophils",
"bacterial",
"pathogens",
"immune",
"system",
"proteins",
"infectious",
"diseases",
"white",
"blood",
"cells",
"necrotic",
"cell",
"death",
"animal",
"cells",
"staphylococcus",
"medical",
"microbiology",
"proteins",
"microbial",
"pathogens",
"biochemistry",
"cell",
"biology",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"macrophages",
"organisms"
] |
2018
|
NLRP6 negatively regulates pulmonary host defense in Gram-positive bacterial infection through modulating neutrophil recruitment and function
|
Dishevelled ( Dvl ) proteins are important signaling components of both the canonical β-catenin/Wnt pathway , which controls cell proliferation and patterning , and the planar cell polarity ( PCP ) pathway , which coordinates cell polarity within a sheet of cells and also directs convergent extension cell ( CE ) movements that produce narrowing and elongation of the tissue . Three mammalian Dvl genes have been identified and the developmental roles of Dvl1 and Dvl2 were previously determined . Here , we identify the functions of Dvl3 in development and provide evidence of functional redundancy among the three murine Dvls . Dvl3−/− mice died perinatally with cardiac outflow tract abnormalities , including double outlet right ventricle and persistent truncus arteriosis . These mutants also displayed a misorientated stereocilia in the organ of Corti , a phenotype that was enhanced with the additional loss of a single allele of the PCP component Vangl2/Ltap ( LtapLp/+ ) . Although neurulation appeared normal in both Dvl3−/− and LtapLp/+ mutants , Dvl3+/−;LtapLp/+ combined mutants displayed incomplete neural tube closure . Importantly , we show that many of the roles of Dvl3 are also shared by Dvl1 and Dvl2 . More severe phenotypes were observed in Dvl3 mutants with the deficiency of another Dvl , and increasing Dvl dosage genetically with Dvl transgenes demonstrated the ability of Dvls to compensate for each other to enable normal development . Interestingly , global canonical Wnt signaling appeared largely unaffected in the double Dvl mutants , suggesting that low Dvl levels are sufficient for functional canonical Wnt signals . In summary , we demonstrate that Dvl3 is required for cardiac outflow tract development and describe its importance in the PCP pathway during neurulation and cochlea development . Finally , we establish several developmental processes in which the three Dvls are functionally redundant .
Normal mammalian development is the result of complex signaling networks that regulate and coordinate cell behavior . Wnt signaling controls a broad spectrum of cell fate decisions during embryogenesis and is critical for cell to cell communication in mammalian development . Through the activation of specific target genes , the canonical Wnt pathway tightly regulates cell proliferation , differentiation , adhesion and survival , and controls embryonic patterning [1] , [2] . A non-canonical Wnt planar cell polarity ( PCP ) pathway , parallel to that first discovered in flies , has been described in mammals , where it regulates cell polarity and convergent extension ( CE ) movements . In these coordinated cell movements , cells migrate medially and intercalate , producing an elongation and narrowing of the tissue along the anterior-posterior axis [3]–[7] . Dishevelled ( Dvl ) proteins , of which three have been identified in humans and mice [8]–[13] are highly conserved components of both the canonical Wnt and PCP signaling cascades . They function as essential scaffold proteins that interact with diverse proteins , including kinases , phosphatases and adaptor proteins [14] , [15] . In the canonical Wnt pathway , Dvl transduces the signal activated by Wnt binding to membrane-bound Frizzled ( Fz ) receptors and low-density lipoprotein-related receptor protein ( LRP ) 5/6 co-receptors , causing the stabilization and cytosolic accumulation of the critical mediator , β-catenin . Following the translocation of β-catenin to the nucleus , it then binds with members of the T-cell factor ( TCF ) /lymphocyte enhancer factor ( Lef ) family of transcription factors to regulate the expression of target genes . Dvl is also one of the core components of the PCP signaling pathway , in addition to Fz , Van Gogh/Strabismus ( Vang/Stbm ) , Flamingo/Starry night ( Fmi/Stan ) , Diego ( Dgo ) and Prickle ( Pk ) . The specific , highly controlled , asymmetric arrangement of these proteins allows polarity of the cell to be established within the plane of the epithelium and promotes the rearrangement of the cytoskeletal components of the cell . In the heart , normal development of the cardiac outflow tract requires the addition of cells from the secondary heart field ( SHF ) dorsal to the primary heart tube [16] , [17] . SHF cells , which express the LIM-homeodomain transcription factor Islet1 ( Isl1 ) , migrate from the pharyngeal mesoderm to contribute to the myocardium of the outflow tract [18] . Recently canonical Wnt signaling has been shown to play a major role in the proliferation and expansion of SHF cells [19]–[22] . It has further been demonstrated that the canonical Wnt signaling mediator β-catenin directly activates Isl1 expression [23] . Extensive remodeling of the outflow tract region with the addition of cardiac neural crest ( CNC ) , mesenchyme from the crest of the neural folds , is then required to septate the single vessel to form the aorta and pulmonary artery [24] . Connections must form between the left ventricle and the aorta , as well as the right ventricle and pulmonary artery , to successfully establish both systemic and pulmonary circulation . Defects in the development of the outflow tract region cause phenotypes such as double outlet right ventricle ( DORV ) , where both the pulmonary artery and the aorta connect to the right ventricle , transposition of the great arteries ( TGA ) and persistent truncus arteriosis ( PTA ) , where the outflow tract fails to divide into an aorta and pulmonary artery . Studies of Looptail ( LtapLp ) mice , which carry a missense mutation in Van Gogh-like 2 ( Vangl2 , strabismus ) [4] , [25] , a component of the PCP pathway , have also implicated non-canonical Wnt signaling mechanisms in mammalian heart development . LtapLp/LtapLp homozygous mutants display the outflow tract abnormality DORV [26] . Reports indicate that the PCP pathway , which is necessary for the polarized migration of myocardial cells to the outflow tract septum , is disrupted in these mutants [27] , [28] . Similar outflow tract defects including TGA and PTA have also been observed in mice lacking the non-canonical signals Wnt11 [29] and Wnt5a [30] . Wnt5a has been proposed to function in the outflow tract by stimulating intracellular increases in Ca2+ to regulate cells of the CNC . PCP signaling is also required for the normal development of the auditory sensory organ , the organ of Corti [3] , [5] , [6] , [31] . This structure is comprised of one row of inner and three rows of outer sensory hair cells , interdigitated with supporting cells such as inner phalangeal cells , the inner and outer pillar cells and the Deiters cells ( reviewed in [32] ) . The sensory hair cells have stereociliary bundles uniformly orientated on the apical surfaces , which is the most obvious example of planar cell polarity in mammals . Consistent with this , mouse mutants of core PCP components including Vangl2/Ltap [5] , Fz3 and Fz6 [33] , [34] and Dvl1 and Dvl2 [6] display stereocilia misorientation , indicating that the correct alignment of these stereocilia is dependent on functional PCP signaling [32] . Cellular rearrangements characteristic of CE movements are also required during the development of the organ of Corti [6] , [35] , during which a thicker and shorter postmitotic primordium undergoes integrated cellular intercalation movements to produce extension along the longitudinal axis and narrowing along a perpendicular axis . Therefore signaling via the PCP pathway is responsible for both the polarized extension and the establishment of planar cell polarity in the organ of Corti . The neural plate undergoes narrowing and lengthening attributable to CE movements during mammalian neurulation . When PCP signaling is disrupted , cells of the neuroepithelium fail to intercalate , preventing the neural tube from fusing at the midline [36]–[39] . Severe neural tube closure defects are observed in mice carrying mutations in PCP components including Vangl2/Ltap [4] , [25] , Dvl1 and Dvl2 [6] , [40] , Celsr ( a homolog of flamingo ) [3] and Fz3 and Fz6 [34] . These mutants have shorter and wider neural plates and display craniorachischisis , a completely open neural tube from mid-brain to tail . To elucidate the role of specific Dvls in mammalian development , we have generated mouse knockouts for each of the Dvl genes . Interestingly distinct phenotypes were revealed , suggesting separate functions for the Dvl proteins . Dvl1 knockout mice are viable and fertile , but display social interaction abnormalities and defects in sensorimotor gating [41] . By contrast , incompletely penetrant cardiac outflow tract abnormalities ( 50% ) and rib/vertebral malformations ( >90% ) are observed in Dvl2 knockout mice [40] . However , functional redundancy among the Dvl genes is also suggested from their overlapping expression patterns , as well as their high degree of conservation . In support of this , we have previously shown that Dvl1−/−;Dvl2−/− mutants display craniorachischisis , a completely open neural tube , and abnormalities in the organ of Corti , both novel phenotypes not observed in the single Dvl knockouts [38] , [40] . Here we describe the phenotype of mice deficient in Dvl3 and determine its importance in conotruncal , cochlear and neural tube development . Given the overlapping expression patterns of the Dvls , as well as their high degree of conservation , we further addressed the possibility of functional redundancy and demonstrate that many of the roles of Dvl3 are also shared by Dvl1 and Dvl2 . We attribute several of the developmental functions of Dvl3 to its role in PCP signaling , enhancing our knowledge of this essential pathway in mammalian development and further defining the specific role of individual Dvl genes .
Dvl3 knock out mice were successfully generated ( Figure S1 ) . In order to calculate the frequency of survival of Dvl3−/− mice , the genotypes of pups from Dvl3 heterozygote crosses were analyzed at weaning age ( Figure 1A ) . In a 129S6 inbred background , 97 pups were genotyped . No Dvl3−/− pups survived to weaning , while Dvl3+/+ and Dvl3+/− pups were observed in an approximate 1∶2 ratio ( 34 and 63 respectively ) , as expected from Mendelian ratios . In a mixed genetic background , 121 pups were analyzed , but only 4 out of the expected 30 Dvl3−/− mice ( 13 . 3% ) survived . In contrast , at E18 . 5 Dvl3+/+ , Dvl3+/− and Dvl3−/− genotypes were recovered in normal Mendelian ratios from heterozygous crosses . These data indicate that 100% of Dvl3 homozygotes in a 129S6 inbred background and approximately 87% in a mixed genetic background die shortly after birth . No gross abnormalities were observed in the few adult Dvl3−/− mice that did survive in a mixed background . Dvl3−/− newborn pups had difficulty breathing and were often cyanotic . Examination of hearts from Dvl3−/− mutants at P0 ( Figure 1B ) using scanning electron microscopy ( Figure 1C ) and histochemical analysis ( Figure 1D ) revealed that all had conotruncal abnormalities . More specifically , seven of the eleven mutant hearts displayed PTA , the outflow tract having failed to divide into an aorta and pulmonary artery , and four showed DORV , where both the pulmonary artery and the aorta connected to the right ventricle . We generated an EYFP-tagged Dvl3 transgene using homologous recombination of BACs ( Figure S2 ) and observed expression of Dvl3-EYFP in the embryonic heart during conotruncal development at E9 . 5 ( Figure 1E ) . This transgene fully rescued the lethal defect in Dvl3−/− mutants , providing formal proof that the phenotype of the Dvl3−/− mutants is specifically due to the loss of Dvl3 . From a cross between a Dvl3−/− rescued with the Dvl3 transgene and a Dvl3 heterozygote ( both in a mixed background ) , 31% ( 37/118 ) of pups genotyped at weaning age were Dvl3−/− rescued with the transgene , whereas only 1 . 6% ( 2/118 ) Dvl3−/− mutants survived without the transgene . E18 . 5 hearts collected from Dvl3−/− embryos with the Dvl3 transgene appeared normal and displayed no conotruncal abnormalities ( Figure 1F and 1G ) . Normal development of the outflow tract requires contribution from both the CNC and SHF , so we examined whether a lack of either of these tissues in Dvl3−/− mutants may be responsible for the observed conotruncal defects by lineage tracing experiments using lineage-specific Cre/LoxP recombination and a Rosa-26-lacZ Cre reporter that expresses β-galactosidase only in cells with Cre activity . Wnt1-Cre , expressed in neural crest cells , was used to label the CNC cell population ( Figure 2A–H ) and SHF cells were specifically labeled using Isl1-Cre ( Figure 2I–P ) . Embryos were collected and stained at E10 . 5 , E14 . 5 and E18 . 5 . At each of these stages both CNC and SHF cells were clearly evident in Dvl3−/− mutants , suggesting that the outflow tract defects were not due to an appreciable loss of either of these lineages . Neural tube defects were observed in Dvl2−/− mutants and to a greater extent in Dvl1−/−;2−/− double knockouts , indicating functional redundancy [40] . We examined a potential role for Dvl3 in neural tube closure . To determine expression of Dvl3 and Dvl1 in the developing neural tube we used transgenic mice carrying the EYFP-Dvl3 transgene and another , similarly made ECFP-Dvl1 transgene , which rescues the Dvl1−/−;Dvl2−/− lethal phenotype ( data not shown ) . Expression of both Dvl3 and Dvl1 was observed in the developing neural tube at E9 . 5 ( Figure 3A–C ) , an important time for this developmental process . The colors were artificially changed to red and green for Dvl3 and Dvl1 , respectively , for easy observation of colocalization ( yellow ) . These Dvls appeared to colocalize in many cells ( indicated with white arrows ) , but not in all cells . Homozygous LtapLp;LtapLp mutants of the PCP signaling pathway component Vangl2/Ltap , display craniorachischisis , a completely open neural tube from mid-brain to tail [4] , [25] , whereas no neural tube defects are observed in LtapLp/+ heterozygotes . Although no neural tube defects were observed in any of the Dvl3−/− mutant embryos collected ( data not shown ) , defective neurulation was apparent when we crossed Dvl3 mutants with LtapLp/+ mice . Several Dvl3+/−;LtapLp/+ embryos displayed normal neural tube development ( Figure 3D–F ) , whereas others exhibited neural tube abnormalities such as craniorachischisis ( Figure 3H–J ) and exencephaly , defective closure of the rostral neural tube . Similar phenotypes were observed in Dvl3−/−;LtapLp/+ mutants , with some appearing normal ( Figure 3L–N ) and others displaying craniorachischisis ( Figure 3P–R ) . The frequency of these defects was similar in both Dvl3+/−;LtapLp/+ ( 7/22 , 32% , 5 with craniorachischisis and 2 with exencephaly ) and Dvl3−/−;LtapLp/+ mutants ( 6/16 , 38% , all with craniorachischisis ) . However , craniorachischisis appears to be a more severe phenotype than exencephaly , indicating a more severe phenotype in Dvl3−/−;LtapLp/+ mutants compared to Dvl3+/−;LtapLp/+ mutants . As both Dvl3−/− and LtapLp;LtapLp mutants display cardiac defects but the single heterozygotes do not , we looked for defects in the hearts of the Dvl3+/−;LtapLp/+ double heterozygotes ( Figure 3G and 3K ) . However , all hearts appeared normal , even in the mutants with neural tube defects , while all hearts from Dvl3−/−;LtapLp/+ double mutant mice ( Figure 3O and 3S ) displayed conotruncal defects , as expected for Dvl3−/−mice . A major role for the mammalian PCP pathway in the development of the organ of Corti has previously been described . Cochleae from mouse mutants homozygous for a mutant allele of the core PCP component Vangl2/Ltap ( LtapLp;LtapLp ) , display misorientation of stereociliary bundles in sensory hair cells due to disrupted planar cell polarity and the cochlear ducts are also shortened and wider due to defects in CE cell movements [5] , [6] . We have shown that both Dvl1 and Dvl2 function in PCP signaling in the developing cochlea [6] . To assess whether Dvl3 also functions in this process , we examined cochleae from Dvl3 mutant embryos , with and without an extra LtapLp mutation in Vangl2 . No cochlear defects were observed in either Dvl3+/− ( data not shown ) or LtapLp/+ single heterozygotes in the basal and middle regions ( Figs 4A and 4E ) . However , a mild PCP phenotype was observed in Dvl3−/− cochleae , where the uniform orientation of stereociliary bundles was disrupted in some hair cells in both the base and the middle of the cochlear ducts ( Figure 4B and 4F , respectively ) . The cochleae of Dvl3+/−;LtapLp/+ mutants that had normal neural tube development were unaffected . However , in the Dvl3+/−;LtapLp/+ mutants that showed defective neurulation , a PCP phenotype in the cochlea accompanied the neural tube defect , with the misorientation of several sensory hair cells in both the basal and middle regions ( Figure 4C and 4G , respectively ) . Compared to these two mutants the severity of the phenotype was much increased in Dvl3−/−;LtapLp/+ mutants that showed neural tube defects , in which many hair cells had disrupted orientation ( Figure 4D and 4H ) . Misorientation of stereociliary bundles was also observed in Dvl3−/−;LtapLp/+ mutants that had normal neural tube development , but the phenotype was much weaker than in embryos of the same genotype displaying defective neurulation . In the apical regions of the cochlear ducts , rotated stereociliary bundles were observed in the hair cells of LtapLp/+ ( Figure 4I ) , Dvl3−/− ( Figure 4J ) , Dvl3+/−;LtapLp/+ ( Figure 4K ) and Dvl3−/−;LtapLp/+ ( Figure 4L ) mutants , although the number of cells affected was increased in both Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants . In these combined mutants the degree of rotation also appeared more severe , as several stereociliary bundles were completely reversed ( Figure 4K ) . In addition to rotated stereociliary bundles in the sensory hair cells , strong patterning defects were also observed in Dvl3;LtapLp mutant cochleae . Separately , the patterning in the cochleae of Dvl3−/− and LtapLp/+ single mutants appeared normal ( Figure 4M and 4N ) . However , the inner ears of Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants with neural tube defects , were much smaller than in littermate controls ( Figure 4O and 4P , respectively ) . Upon further dissection , the cochleae of the Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants with craniorachischisis were also much shorter compared to controls ( Figure 4M–P ) . At the cellular level , often in these Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants the normal arrangement of 3 outer hair cell rows and 1 inner hair cell row was disrupted . In the apical region of Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ cochleae , additional rows of both outer and inner hair cells were observed ( Figure 4K and 4L ) . Loss of outer hair cell rows was also detected in these mutants , so that the normal three rows became only two rows ( Figure 4R and 4S ) . In total , this patterning defect was observed in 5–20% of the length of the cochlea . Finally , the location of Dvl3 expression was determined in the inner ear sensory organs , including both the vestibular ( Figure 4T ) and cochlear ( Figure 4U ) end organs , using the EYFP-tagged Dvl3 transgene . Dvl3-EYFP signals were asymmetrically localized in cells in all inner ear sensory regions . In the cochlea , Dvl3 was localized either on the lateral sides of the sensory hair cells , or on the medial side of the surrounding supporting cells , similar to what was observed for Dvl2 [6] . We have previously reported both vertebral and rib malformations in Dvl2−/−mutants and more severe skeletal defects in Dvl1−/−;Dvl2−/− double knockouts [40] . Given the overlapping expression patterns and high homology between the Dvls , we also examined the skeletons of Dvl3−/− mutants . However , no rib or vertebral abnormalities were observed ( data not shown ) . 7/11 Dvl3−/− mutants showed xiphoid bifurication , but this was also observed in several wild type controls . Redundancy between Dvl1 and Dvl2 was evident from our studies of Dvl1−/−;Dvl2−/− double mutants , which displayed novel phenotypes not observed in the single Dvl knockouts [38] , [40] . To address possible redundancy between Dvl3 and the other two Dvls we examined the phenotypes of Dvl1;Dvl3 and Dvl2;Dvl3 double mutants . Both Dvl1+/−;Dvl3+/− and Dvl1−/−;Dvl3+/− mutants survived to adulthood and were fertile . Normal development was also observed in Dvl1−/−;Dvl3−/− embryos until E12 . 5 ( Figure 5A and 5B ) , but these mutants died soon after of unknown causes , normally between E13 . 5 and E15 . 5 . Importantly , no neural tube defects were observed . At mid-gestation Dvl2+/−;Dvl3+/− and Dvl2+/−;Dvl3−/− double mutants also appear normal ( Figure 5D and G ) , however a striking phenotype was observed in Dvl2−/−;Dvl3+/− mutants ( Figure 5H ) . These mutants do not appear to survive beyond E9 . 5 and display craniorachischisis , pericardial effusion and abnormal looping of the heart , as well as severe posterior truncation . No Dv2−/−;Dvl3−/− embryos have been recovered from litters collected from E8 . 5 onwards , suggesting lethality earlier in development . Defects are observed in Dvl2+/−;Dvl3−/− double mutants at later developmental stages . At E18 . 5 these mutants appear shorter along the anterior-posterior ( A–P ) axis , show an abnormal head shape and a truncated snout , as well as a shortened and kinked tail ( Figure 5K and L ) . Craniorachischisis was also observed in two Dvl2+/−;Dvl3−/− mutants . One of these with the neural tube defect also showed several other severe phenotypes , including gastroschisis and absence of tail ( Figure 5M and 5N ) . Since both Dvl2−/− and Dvl3−/− mutants both display conotruncal defects , we determined whether these Dvls have redundant or distinct functions in heart development . As common spatial and temporal expression patterns may indicate similar functions , we first confirmed that Dvl3 ( Figure 1E ) and Dvl2 ( Figure 6A ) shared comparable expression patterns in the heart at E9 . 5 during conotruncal development . We first examined the hearts of Dvl2+/−;Dvl3+/− double heterozygotes to see whether these mutants had similar defects as the homozygotes , which would suggest redundant functions for Dvl2 and Dvl3 . We observed that Dvl2+/−;Dvl3+/− mice often survive to adulthood and are fertile . However , in an inbred background , conotruncal abnormalities were seen in 11/28 Dvl2+/−;Dvl3+/− hearts examined at E18 . 5 ( Figure 6B ) . These defects included 9 DORV , 1 PTA and 1 TGA ( Figure 6C–F ) . The hearts of Dvl2+/−;Dvl3−/− mutants were also examined to determine whether an extra loss of one copy of Dvl2 would worsen the phenotype of Dvl3−/− mutants . Three Dvl2+/−;Dvl3−/− hearts collected at P0 ( Figure 6I and 6J ) had similar morphology to Dvl3−/− hearts ( Figure 6G and 6H ) , 1 with DORV and 2 with PTA . However , four Dvl2+/−;Dvl3−/− hearts collected appeared to have a worse phenotype than Dvl3−/− mutants with varying degrees of severity and all displaying PTA ( Figure 6K–N ) . Two were slightly smaller and had lost the characteristic fist shape heart morphology , becoming less tapered towards the bottom ( Figure 6K and 6L ) . The other two Dvl2+/−;Dvl3−/− hearts were much reduced in size compared to Dvl3−/− mutants and had severely altered morphology , with a teardrop shape ( Figure 6M and 6N ) . The embryos that the hearts were collected from were all alive , apart from the two severest phenotypes ( Figure 6M and 6N ) , which were dead . After sectioning , it appeared that both right and left ventricles were still present in all samples . The generation of fully functional Dvl transgenes ( here and in [6] ) allowed us to use a genetic approach to further determine redundancy of function between the Dvl proteins during development . We determined whether an extra copy of either Dvl1 or Dvl2 , in the form of Dvl1-ECFP and Dvl2-EGFP BAC transgenes ( which we will refer to here as Dvl1TG and Dvl2TG , respectively ) was able to rescue the lethal Dvl3−/− phenotype . Importantly , western blot analysis revealed that none of the Dvl transgenes were overexpressed compared to protein levels from wild type alleles ( Figure 7A–C ) . As shown above , Dvl3−/− mutants ( which still have 2 copies of the Dvl1 allele and 2 copies of the Dvl2 allele ) cannot survive . Surprisingly , we found that addition of the Dvl1TG rescued the Dvl3−/− phenotype such that Dvl3−/−;Dvl1TG mutants ( now Dvl3−/− with 3 copies of Dvl1 and 2 copies of Dvl2 ) survived . To test whether this was a complete rescue , we crossed these viable Dvl3−/−;Dvl1TG mutants to Dvl3+/− mice and genotyped the progeny at weaning ( Figure 7D ) . Eighteen Dvl3−/−;Dvl1TG rescued animals were recovered , compared to 16 that would be expected from normal Mendelian ratios if the Dvl1TG is fully able to rescue the Dvl3−/− phenotype . As these crosses were performed in a mixed background , 4 Dvl3−/− mice also survived with out an extra copy of another Dvl gene . Interestingly , in a similar cross using the Dvl2TG , we found that the Dvl3−/− lethal phenotype was also rescued by an extra copy of Dvl2 . We crossed these viable Dvl3−/−;Dvl2TG mutants with Dvl3+/− mice and genotyped the offspring at weaning ( Figure 7E ) . Seventeen Dvl3−/−;Dvl2TG mice were recovered , out of 19 expected from a full rescue , indicating approximately 90% rescue of the Dvl3−/− lethal phenotype with additional Dvl2 . As 50% of Dvl2−/− mutants die perinatally , we used a similar strategy to determine whether additional copies of Dvl1 or Dvl3 could rescue Dvl2−/− lethality . From crosses between Dvl2−/−;Dvl3TG mutants and Dvl2+/− mice , 24 Dvl2−/−;Dvl3TG mutants survived , out of 25 expected from a full rescue , indicating approximately 96% rescue of the Dvl2−/− phenotype with an extra copy of Dvl3 ( Figure 7F ) . However , from crosses between Dvl2−/−;Dvl1TG mutants and Dvl2+/− mice , only 10 Dvl2−/−;Dvl1TG mutants were recovered , out of 18 expected from Mendelian ratios if the Dvl1TG was able to rescue the Dvl2−/− phenotype . As 50% of Dvl2−/− mutants survive without additional Dvl copies , this indicates that adding the Dvl1TG was not able to rescue Dvl2−/− mutants ( Figure 7G ) . To confirm that the BAC transgenes were actually behaving as wild type alleles and that ectopic or inappropriate expression levels were not responsible for rescuing the mutant phenotypes , we genetically reduced the Dvl gene dosage by introducing a knock-out Dvl allele and again tested the rescue ability of the transgene . Dvl2+/−;Dvl3+/− mutants were crossed with Dvl3−/−;Dvl2TG mutants and the surviving progeny were genotyped ( Figure 7H ) . Twelve Dvl2+/+;Dvl3−/−;Dvl2TG mutants with both wild type alleles of Dvl2 ( a total of 3 copies of Dvl2 ) survived , comparable to 9 expected from normal Mendelian ratios , indicating full rescue . However , no Dvl2+/−;Dvl3−/−;Dvl2TG mutants were recovered , even though 9 were expected from normal Mendelian ratios if the Dvl2TG could still rescue . Thus , correction of Dvl2 dosage from three copies to two copies of Dvl2 in Dvl2+/−;Dvl3−/−;Dvl2TG mutants ( 1 wild type allele and the transgene ) eliminated the ability of the Dvl2TG to rescue the Dvl3−/− mutant phenotype , supporting the conclusion that ectopic expression or over-expression of the transgene was not responsible for the rescue of the Dvl3−/− phenotype . Similar cochlear defects have been observed in both Dvl1−/−;Dvl2−/− [6] and Dvl3−/− mutants ( Figure 4 ) . Therefore , we examined whether all three Dvls have redundant functions in the developing organ of Corti . The inner ears of double Dvl mutants were studied to determine whether the further loss of an additional Dvl gene would worsen the phenotype observed in Dvl3−/− mutants . Only the cochlea from Dvl1−/−;Dvl3+/− , Dvl2+/−;Dvl3+/− and Dvl2+/−;Dvl3−/− mutants were examined as other double Dvl mutants were lethal too early in development . The organ of Corti from Dvl1−/−;Dvl3+/− mutants appeared normal ( data not shown ) , as did those from the Dvl2+/− ( Figure 8A , F , K ) and Dvl3+/− ( not shown ) single heterozygote samples . However , several of the sensory hair cells in Dvl2+/−;Dvl3+/− double heterozygotes had rotated stereociliary bundles ( Figure 8B , G , L ) . This phenotype appeared worse in the Dvl2+/−;Dvl3−/− mutants with the loss of another Dvl3 allele ( Figure 8C , H , M ) and mild patterning defects were also observed in one of these mutants ( Figure 8H ) . In the Dvl2+/−;Dvl3−/− mutants that displayed craniorachischisis , the phenotype was even more severe ( Figure 8D , I , N ) . Throughout the whole cochlea duct the hair cells did not appear to be fully developed , even in the basal region . Additionally , it is hard to see the typical shape of stereociliary bundles in most of the hair cells and the differentiation of these cells is apparently delayed ( Figure 8N ) . Strong patterning defects were also observed as the cochleae of the Dvl2+/−;Dvl3−/− mutants were much shorter compared to Dvl3−/− controls ( Figure 8P , Q ) . Functional redundancy of the Dvls in cochlea development could only be examined in the Dvl3−/− mutants rescued with the Dvl2-EGFP transgene . The organ of Corti from Dvl3−/−;Dvl2TG mutants still displayed misorientation of stereocilia in several of the sensory hair cells ( Figure 8E , J , O ) , similar to the phenotype described for Dvl3−/− mutants ( Figure 4 ) , indicating that increasing the copy number of Dvl2 was not able to restore the correct alignment of stereocilia in the Dvl3−/− mutants . Previous studies from our lab [6] , [38] and data presented above are consistent with a role for the Dvls in the PCP pathway during neurulation and cochlear development , while a role for Dvls in Wnt signaling is not as clear . Therefore , BATgal mice , which carry a β-catenin-responsive LacZ gene , were used to determine whether canonical Wnt signaling was functional or interrupted in the Dvl mutants during mid-gestation . Global canonical Wnt signaling patterns appeared largely unaffected in each of the single Dvl mutants compared to their wild type littermates at either E9 . 5 ( Figure 9A–C ) or E11 . 5 ( Figure 9D–F ) . Canonical Wnt signaling was also generally unchanged in the double Dvl mutants Dvl1−/−;Dvl2−/− , Dvl1−/−;Dvl3−/− , Dv2+/−;Dvl3+/− , Dvl2+/−;Dvl3−/− and Dvl2−/−;Dvl3+/− at E9 . 5 ( Figure 9G–I ) . However , subtle abnormalities were observed . For example , TCF activity appears reduced in the somites of Dvl1−/− mutants ( Figure 9A ) and in the limb bud of Dvl2−/− mutants ( Figure 9B ) at E9 . 5 , and the pattern of staining appears altered in both the somites and pharyngeal region of Dvl2−/− mice at E11 . 5 ( Figure 9E ) .
A large body of evidence demonstrates that Dvl proteins function in highly conserved pathways in both vertebrates and invertebrates . It is known that in Drosophila [42] , [43] , Xenopus [37] , [44]–[46] , mouse [6] , [38] and mammalian cell lines [47]–[49] , Dishevelled proteins mediate their effects on the canonical and non-canonical Wnt pathways via highly conserved protein domains . As essential components of both the canonical Wnt and PCP pathways , Dvl proteins are required for many developmental processes . A recent publication has addressed the possibility of redundancy between the three Dvl proteins in mammalian cell lines in vitro [47] . We further aim to determine the action of the Dvl proteins during development in vivo . Having previously described the roles of Dvl1 and Dvl2 in mammalian development [38] , [40] , [41] , we now describe the specific developmental roles of Dvl3 and additionally use well-established genetic approaches to determine the pathways disrupted in vivo that account for the phenotypes observed in the Dvl mutant mice and to demonstrate redundancy between the Dvls during development in vivo . A summary of the phenotypes of single Dvl and double Dvl mutants is shown in Figure 10 . Dvl3 knockout mice die perinatally with cardiac outflow defects , indicating an important role for Dvl3 in conotruncal development . This is particularly interesting as approximately 30% of all congenital heart diseases are due to defects in this region [50] , [51] . The loss of Dvl3 causes a similar conotruncal phenotype to those observed with loss of Dvl2 [40] , suggesting redundant functions for these homologous genes in outflow tract development . This was supported both by the similar expression patterns of Dvl2 and Dvl3 found during conotruncal development and the observation of cardiac abnormalities in double Dvl2+/−;Dvl3+/− heterozygotes , as compared to normal hearts of Dvl2+/− and Dvl3+/− single heterozygotes . The increased severity of the cardiac phenotype of the double Dvl2−/−;Dvl3+/− and Dvl2+/−;Dvl3−/− mutants , compared with the single Dvl2−/− and Dvl3−/− mutants , further indicates redundancy between these Dvls in heart development . Finally , with a genetic approach we found that an extra copy of either Dvl1 or Dvl2 , using either the Dvl1-ECFP or Dvl2-EGFP BAC transgene , could rescue lethal Dvl3−/− heart phenotype , demonstrating the ability of both Dvl1 and Dvl2 to compensate for the loss of Dvl3 to enable normal development . Our data suggests that Dvl1 shares redundant functions with Dvl2 and Dvl3 in cardiac development . However we found that an additional copy of Dvl1 is not able to rescue the Dvl2−/− cardiac defects . Although the Dvl proteins appear to have the same expression patterns and redundant functions , we have not yet addressed the possibility of differences in protein expression level . There may a certain threshold of Dvl protein required for normal heart development , but under this threshold there is insufficient signaling to permit normal development . Interestingly , a recent paper by Lee et al . demonstrated that in a number of mammalian cell lines all three Dvls are present , but that the relative levels of expression of the three Dvls differ greatly [47] . Normal development of the outflow tract requires the addition of cells from both the CNC and the SHF , and a lack of contribution of either of these tissues is often the cause of defects in the conotruncal region . Further , Dvl2 has previously been found to be important for controlling proliferation of CNC cells , via activation of Pitx2 [52] . However , both CNC cells and SHF cells appeared to be present in the absence of Dvl3 , suggesting that the heart defects are not due to a complete lack of either of these tissues . Further studies to determine whether these cells are properly situated in relation to the surrounding tissues , and whether they are indeed still functional , are in progress . Dvl proteins function in both the canonical and non-canonical PCP pathways , therefore disruption of either or both of these pathways could be the cause of the phenotypes observed in Dvl3−/− mice . Roles for both pathways in heart development have been reported . Canonical Wnt signals appear essential for the proliferation of SHF cells [19]–[23] , whereas the PCP pathway is thought to be necessary for the polarized migration of myocardial cells required in the outflow tract septum ( Henderson et al . , 2006; Phillips et al . , 2005 ) . Dvl3+/−;LtapLp/+ double heterozygotes displayed no abnormal heart phenotype , suggesting that Dvl3 may function through the canonical and not the PCP pathway during heart development . Conversely , it is possible that the level of PCP signaling was not reduced below the hypothetical threshold required for the phenotype . Further experiments will be required to distinguish these possibilities . We have shown here that Dvl1 and Dvl3 colocalize in the developing neural tube with similar expression patterns to that of Dvl2 as we described previously [38] , and provide further evidence for a similar functional role for Dvl3 in neurulation as Dvl1 and Dvl2 [38] . Although none of the Dvl single knockouts display neural tube closure defects , disruption of either Dvl1 and Dvl2 or Dvl2 and Dvl3 genes ( Dvl1−/−;Dvl2−/− , Dvl2−/−;Dvl3+/− , Dvl2+/−;Dvl3−/− ) results in incomplete neurulation , suggesting a dosage sensitive redundant role for all three Dvls in this process . Interestingly , Dvl1−/−;Dvl3−/− mutants did not display neural tube defects , suggesting that the three Dvls are not functionally equivalent . Neural tube defects appear in single Dvl mutants when crossed with LtapLp mice , ( Dvl2−/−;LtapLp/+ , Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ ) , indicating genetic interaction with the Dvls and the PCP component , Vangl2 and therefore that the Dvls signal through the PCP pathway to promote neural tube closure . It is interesting that both exencephaly and craniorachischisis are observed in Dvl3;LtapLp mutants . Craniorachischisis has been observed in many mutants with disrupted PCP signaling , whereas exencephaly is often associated with defective ciliogenesis ( reviewed in [53] ) . Recently a connection between cilia and PCP signaling has been suggested as the ciliary proteins Inversin [54] and Bardet-Biedl Syndrome protein-4 ( BBS4 ) [55] were shown to influence PCP signaling and CE movements . Additionally , disruption of the PCP effectors inturned and fuzzy in Xenopus laevis demonstrated a link between ciliogenesis , PCP signaling and Hedgehog signaling [56] . Localization of both dishevelled and inturned near the basal apparatus of cilia further suggests a role for PCP components in regulating ciliogenesis [56] . The polarity defects observed in Dvl3−/− cochleae were much more severe with the introduction of an additional single LtapLp mutation , indicating that Dvl3 genetically interacts with Vangl2 and functions in the PCP pathway to regulate cell polarity in the organ of Corti . The CE movements driven by the PCP pathway that are required for the extension and thinning of the developing organ of Corti were also disrupted in Dvl3;LtapLp mutants , resulting in shortened organs of Corti relative to controls . Furthermore , the asymmetric localization of Dvl3 during establishment of stereocilia orientation is consistent with the asymmetric localization of Dvl2 [6] and other core PCP components in mammals [34] , [57] , [58] . The polarized asymmetric arrangement of PCP proteins in the fly wing is also required for the uniform orientation of hairs [59] , [60] . The sensory hair cells in this region are surrounded by the cellular processes of various types of supporting cells and tight cell contacts are formed between the cells ( reviewed in [32] ) . Dvl3 appears to be localized on the cellular boundary formed between the sensory hair cells and the projections of the supporting cells , making it difficult to determine whether it is expressed on the lateral side of the hair cell , or on the medial side of the supporting cell . Dvl2 has previously been shown to localize at the lateral side of the hair cell [6] and as our evidence suggests redundant functions for the Dvl proteins , it seems likely that Dvl3 is also expressed here . Redundancy between the three Dvls in the developing organ of Corti was implied from the similarity of phenotype in Dvl3−/− mutants and Dvl1−/−;Dvl2−/− double mutants [6] , [38] . Further , a mild PCP phenotype was observed in Dvl2+/−;Dvl3+/− double heterozygotes , despite normal development in the single heterozygotes and the defect in Dvl2+/−;Dvl3−/− double mutants was much more severe than in the single Dvl3−/− mutant . Interestingly , despite redundancy between the Dvls , addition of the Dvl2-EGFP BAC transgene was not able to rescue the rotated stereocilia phenotype in Dvl3−/− mutants . We again propose that a certain threshold of Dvl protein level may be required for normal development and the relative expression levels of the three Dvls in the developing organ of Corti influences the cochlear phenotype and the ability of a Dvl transgene to rescue this phenotype . Interestingly , the Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants that showed the CE/PCP-phenotype in the neural tube also displayed the PCP-phenotype in the organ of Corti . However , the misorientation phenotype was less severe in both Dvl3+/−;LtapLp/+ and Dvl3−/−;LtapLp/+ mutants than previously found in LtapLp/LtapLp mutants , in which 95% cells in the two outer hair cell rows are misorientated [6] , possibly due to the remaining Dvl1 and Dvl2 alleles in these mutants . It was also noted that Dvl2+/−;Dvl3−/− mutants with defects in the neural tube had a much more severe phenotype in the organ of Corti than those with normal neurulation , indicating a strong correlation between the PCP-phenotypes in these two tissues . We have previously shown that both Dvl1 and Dvl2 play roles in somite segregation , causing skeletal malformations in mice that lack these genes [40] . Given the high homology between the three Dvl genes , we examined the skeletons of Dvl3−/− mice . However , no vertebral or rib malformations were observed . Some mutants did display xiphoid bifurication , although we are unsure of the significance of this phenotype as it was also seen in several wild type controls . Severe skeletal defects involving truncation of the A–P axis were , however , observed in Dvl2−/−;Dvl3+/− double mutants , supporting previous evidence to suggest redundant roles of the Dvls in somite formation . A similar phenotype of lack of caudal somites and absence of tail bud formation is seen in mice homozygous for a null Wnt3a allele ( Wnt3aneo ) [61] , and a less severe phenotype appears in the hypomorphic Wnt3a allele mutant , vestigial tail , which shows loss of caudal vertebrae causing a shortening of the tail [62] . Mice lacking Wnt5a are also shortened along the A–P axis with a phenotype similar to Dvl2+/−;Dvl3−/− mutants , and share outgrowth defects in the developing face and lack of tail [63] . Wnt3a is classically considered to stimulate canonical Wnt signals , whereas Wnt5a is normally associated with non-canonical Wnt mechanisms , indicating that further investigation is needed to determine which pathway or pathways Dvl is required to signal through for the normal development of these structures . We grossly examined Wnt signaling in both single and double Dvl mutants using the TOPflash reporter of TCF activity . Interestingly we found that global Wnt signaling was largely unaffected in even the most severely affected mutants , suggesting that only a low level of Dvl is sufficient for functional canonical Wnt signals . However , our data also suggests subtle abnormalities in Wnt signaling . Precise determination of these abnormalities will require careful sectioning of specific tissues and examination of various cell types throughout development . The three highly homologous Dvl proteins shared in mammals have very similar broad expression patterns in development . This study completes the initial characterization of the specific , individual roles of each of these proteins and also establishes functional redundancy and overlap in a number of developmental processes . Dvl3 is required for the development of the cardiac outflow tract and signals in the PCP pathway to regulate CE in the developing neural tube and cochlea , as well as cell polarity in the organ of Corti . Dvl1 and Dvl2 are redundant with each other [6] , [38] , [40] as well as with Dvl3 in a number of these developmental processes .
All animal care and experiments were performed under protocols approved by the NHGRI/NIH and UCSD Animal Care and Use Committees . LtapLp mutants were originally acquired from Jackson Laboratory , Wnt1-Cre mice were a kind gift of Dr . Andrew McMahon , Harvard and Isl1-Cre mice were a generous gift from Sylvia Evans , UCSD . We generated the Dvl3−/− and Dvl3-EYFP mouse mutants as described in Figures S1 and S2 , respectively . Genotyping of the Dvl3 mutants was performed with the following primers , Dvl3 forward 5′-TCCGATGAGGATGATTCCACC-3′ , Dvl3 reverse 5′-TGAGGCACTGCTCTGTTCTGT -3′ , Dvl3 knockout 5′-TTGGCCCACAATGGAGATGCCC-3′ , NLpgk neo forward 5′-AGGCTTACCCGCTTCCATTGCTCA-3′ . The PCR primers used to distinguish between the transgene and wild type Dvl3 allele were Dvl3lnt2 forward 5′-GGACGCAGGAGATCTTTGAA-3′ and Dvl3Int2 reverse 5′-CATAGCTGGGGTTGAAGCTC-3′ which amplify a band of 155 bp for the wild type allele and 189 bp for the transgene , due to the presence of the LoxP site . Genotyping for the following mouse mutants have previously been described; Dvl1 [41] , Dvl2 [40] , Dvl2-EYFP transgene [6] , Wnt1-Cre [64] , Isl1-Cre [18] , Rosa-26-lacZ Cre [65] and BAT-gal [66] . After fixing in 3% aldehyde solution ( 1 . 5% paraformaldehyde , 1 . 5% glutaraldehyde ) in 0 . 1 M phosphate buffer pH 7 . 5 , E18 . 5 hearts were stored in 100% ethanol following dehydration through a graded ethanol series . Hearts were critically point dried , mounted and coated with 300 Angstrom gold-palladium . A Cambridge Instrument Stereoscan 360 scanning electron microscope ( Scripps Institute of Oceanography Analytical Facility ) was used to view the prepared samples . P0 hearts were dissected into 10% buffered formalin , dehydrated , embedded in paraffin wax , sectioned at 8 µm thickness and stained with hematoxylin and eosin using standard methods . To label the CNC and SHF cell populations in Dvl3−/− hearts , Dvl3+/− mice carrying Wnt1-Cre [64] and Isl1-Cre [18] , respectively , were crossed with Dvl3+/− mice carrying the Rosa-26-lacZ Cre reporter gene [65] . Hearts dissected from embryos collected at E14 . 5 and E18 . 5 were stored in 30% sucrose in PBS at 4°C overnight , and then embedded in 7 . 5% gelatin , 15% sucrose . Sections ( 20 µm ) were cut at −24°C on a cryostat and fixed in 2% formaldehyde , 0 . 2% glutaraldehyde , 0 . 02% NP-40 and 0 . 01% sodium deoxycholate . After washing , the slides were stained with 1 mg/ml X-gal in 35 mM potassium ferrocyanide , 35 mM potassium ferricyanide , 2 mM MgCl2 , 0 . 02% NP-40 , 0 . 01% sodium deoxycholate at 37°C overnight . After re-fixing in 4% paraformaldehyde , the slides were counterstained with Nuclear Fast Red ( Vector Lab ) , according to the manufacturer's protocol . Whole embryos from the above crosses were also collected at E10 . 5 and stained for β-galactosidase activity using the same procedures . All images were captured using a Spot 2 camera mounted on a Leica DMR light microscope . To visualize canonical Wnt signaling , Dvl mutant mice were crossed with mice carrying the BAT-gal reporter gene [66] and embryos collected at various stages of development . Staining for β-galactosidase activity was performed as described above . To visualize Dvl3 expression , Dvl3-EYFP embryos or cochleae were fixed in 4% paraformaldehyde for 30 minutes or overnight at 4°C . To examine the stereocilia in the sensory hair cells , organs of Corti were stained with fluorescein- rhodamine-conjugated phalloidin ( Molecular Probes ) . Both native YFP signal and fluorescent staining was observed using an Olympus FV 1000 or a Zeiss LSM 510 confocal scanning microscope . To measure the severity of the patterning defect , the whole cochlea duct was scanned and the cells numbered for length reference . The number of cells in the defective regions ( where there were no longer three outer hair cell rows ) was then calculated as a percentage of total cells .
|
Multi-gene families , comprising a set of very similar genes with shared nucleotide sequences , are common in mammals . Individual family members may be expressed in different places and perform separate functions . Alternatively , the genes may have redundant functions , but distinct dosage requirements . Mammals share three Dishevelled ( Dvl ) family members and while the roles of Dvl1 and Dvl2 have been described previously , the functions of Dvl3 have remained elusive . Here , we show that the lack of Dvl3 in mice affects the formation of the heart , neural tube , and inner ear . We further show that the defects in these tissues are much more severe when the mice are deficient in more than one Dvl family member , indicating redundant functions for these genes . Congenital heart disease affects approximately 75 in every 1 , 000 live human births , and approximately 30% of these diseases are due to disruptions in the outflow tract , the region affected in mice lacking Dvl genes . Neural tube defects , similar to those observed in the Dvl mutants , are also common in humans . The animal models described here provide useful tools to elucidate the genetic mechanisms that underlie these abnormalities and may provide novel ways of treating these disorders in the future .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"developmental",
"biology/morphogenesis",
"and",
"cell",
"biology",
"cell",
"biology/cell",
"signaling",
"cell",
"biology/developmental",
"molecular",
"mechanisms",
"developmental",
"biology/molecular",
"development",
"molecular",
"biology",
"genetics",
"and",
"genomics",
"cardiovascular",
"disorders"
] |
2008
|
Murine Dishevelled 3 Functions in Redundant Pathways with Dishevelled 1 and 2 in Normal Cardiac Outflow Tract, Cochlea, and Neural Tube Development
|
Pore-forming toxins are potent virulence factors secreted by a large array of bacteria . Here , we deciphered the action of ExlA from Pseudomonas aeruginosa and ShlA from Serratia marcescens on host cell-cell junctions . ExlA and ShlA are two members of a unique family of pore-forming toxins secreted by a two-component secretion system . Bacteria secreting either toxin induced an ExlA- or ShlA-dependent rapid cleavage of E-cadherin and VE-cadherin in epithelial and endothelial cells , respectively . Cadherin proteolysis was executed by ADAM10 , a host cell transmembrane metalloprotease . ADAM10 activation is controlled in the host cell by cytosolic Ca2+ concentration . We show that Ca2+ influx , induced by ExlA or ShlA pore formation in the plasma membrane , triggered ADAM10 activation , thereby leading to cadherin cleavage . Our data suggest that ADAM10 is not a cellular receptor for ExlA and ShlA , further confirming that ADAM10 activation occurred via Ca2+ signalling . In conclusion , ExlA- and ShlA-secreting bacteria subvert a regulation mechanism of ADAM10 to activate cadherin shedding , inducing intercellular junction rupture , cell rounding and loss of tissue barrier integrity .
Multicellular organisms have developed barriers to protect their internal body from microbial invasion . Mucosae constituted of single-layered epithelial cells are the favoured tissue barriers that bacteria may cross . Pathogenic bacteria have engineered different weapons to traverse these borders , either going through the cells ( transcellular route ) or at cell-cell junctions ( paracellular route ) . Opportunistic pathogens , like Pseudomonas aeruginosa or Serratia marcescens , can only cross epithelial barriers when tissues are damaged or proliferate after physical trauma or biological insults . Once across the epithelial layer , bacteria may cross the endothelium , translocate into the vascular system and disseminate into the body . Exolysin ( ExlA hereafter ) is a pore-forming toxin recently identified in a subset of strains from Pseudomonas aeruginosa species , generally called PA7-like strains , from the name of the first-identified strain of this category [1–11] . ExlA gene is located in the same locus as exlB , forming a Two-Partner Secretion System , in which ExlB is required for ExlA secretion in the extracellular milieu [12] . PA7-like strains have been isolated from patients with various infection types , such as acute or chronic pneumonia , urinary infections , burns , otitis , and recently from a patient suffering from hemorrhagic pneumonia . The PA7-like strains isolated so far do not possess a Type III secretion system ( T3SS ) and their virulence potential on cellular models is mainly correlated with the level of secreted ExlA [3 , 13] . In the mouse lungs , ExlA-secreting strains induced major injuries of the alveolo-capillary barrier , leading to pulmonary hemorrhages and allowing bacterial dissemination in the body . Osmotic protection assays revealed that the inner diameter of the pore formed by ExlA in the host plasma membrane is approximately 1 . 6 nm , likely allowing the trafficking of small molecules[12] . Pore formation ultimately provokes cell death by plasma membrane rupture , as observed by microscopy , or monitored by lactate dehydrogenase ( LDH ) release and propidium iodide ( PI ) incorporation [3 , 12] . Homologous ExlA proteins were identified in related Pseudomonas species , such as P . protegens , P . entomophila and P . putida [12] . Importantly , a 35% identity was found with ShlA pore-forming toxin from Serratia marcescens [3] , which displays the same domain organization and a comparable size ( 172 kDa for ExlA and 165 kDa for ShlA ) , and shares a similar secretory pathway [14] . ShlA forms pores of 1–2 nm , is cytolytic for various cell types and induces hemorrhagic pneumonia in humans and in infected mouse lungs [15–17] . Hence , ShlA is related to ExlA in several structural and functional aspects . Other related toxins have been identified based on sequence homology and secretion pathway , including HpmA from Proteus mirabilis , HhdA from Haemophilus ducreyi , PhlA from Photorhabdus luminescens , EthA from Edwardsiella tarda and ChlA from Chromobacterium violaceum [14 , 18–22] , for which very little information is available . Altogether , these toxins constitute a unique family of pore-forming toxins , for which neither the 3-dimensional structure , nor the potential oligomerization and mechanisms of pore formation are known , and importantly , for which the mechanism of toxicity remains elusive . Here , we show that ExlA and ShlA have the capacity to disrupt the cell-cell junctions of epithelial and endothelial cells , using an indirect mechanism . The pore formed by either toxin in the host cell triggers a Ca2+ influx , which activates A Disintegrin And Metalloproteinase domain-containing protein 10 ( ADAM10 ) , a transmembrane metalloprotease , whose natural substrates are transmembrane proteins , including some cadherins . ExlA- and ShlA-dependent ADAM10 activation rapidly leads to E- or VE-cadherin cleavage in epithelial or endothelial cells , respectively . As cadherins are major determinants of intercellular adhesion , cadherin cleavage induces cell-cell junction breakdown and loss of tissue integrity .
We previously reported that T3SS-positive P . aeruginosa strains incubated with endothelial cells induce VE-cadherin cleavage , which is mediated by a protease ( LasB ) released by the T2SS [23] . LasB cleaves VE-cadherin in the middle of its extracellular domain , preventing its adhesive activity . We further showed that E-cadherin , located at epithelial cell-cell junctions , was resistant to LasB proteolytic activity [23] . As cadherins are required for tissue integrity , we tested the ExlA-secreting isolate CLJ1 for its capacity to cleave E-cadherin by incubation with the human alveolar cell line A549 ( Fig 1A , left ) . CLJ1 induced a rapid and dramatic decrease in full-length E-cadherin levels , which paralleled the onset of a C-terminal E-cadherin fragment of 30 kDa . As previously reported [23] , two ExlA-negative P . aeruginosa strains injecting ExoS , T and Y toxins through their T3SS , did not degrade E-cadherin , even at longer time points ( Fig 1A , left ) . We also tested an ExlA-negative T3SS-positive strain injecting ExoU , a toxin endowed with phospholipase activity , known to induce membrane permeabilisation to nuclei dyes [24] . A549 cell incubation with the ExoU-positive strain did not induce E-cadherin cleavage ( S1 Fig ) . Thus , membrane permeabilisation by a phospholipase is not sufficient to promote cadherin cleavage . In similar experiments , VE-cadherin was rapidly degraded after incubation of primary human endothelial cells ( HUVECs ) with CLJ1 ( Fig 1A , right ) . The onset of a C-terminal VE-cadherin fragment of 30 kDa was also observed concomitantly to the decrease of full-length VE-cadherin . As previously noticed , VE-cadherin was partially degraded due to LasB activity [23] , however at much longer time points ( Fig 1A , right ) . To examine whether the rapid E- and VE-cadherin cleavage activity was dependent upon ExlA , we used the ExlA-secreting IHMA87 strain [12 , 13] that can be manipulated genetically , as opposed to CLJ1 . Hence , we tested the isogenic mutant IHMA87ΔexlA and its complemented counterpart IHMA87ΔexlA/exlA for their ability to degrade the cadherins ( Fig 1B ) . IHMA87 similarly induced E-cadherin and VE-cadherin cleavages , albeit more slowly than CLJ1 , while IHMA87ΔexlA did not , even at longer time points . The complemented strain recovered this ability . Thus , the cleavage of both types of cadherin is ExlA-dependent . We previously reported that Type IV pili ( “pili” hereafter ) facilitated the ExlA-dependent toxicity of IHMA87 towards A549 cells , probably by enhancing bacterial adhesion [12] . We thus tested the effect of a mutant devoid of pili ( IHMA87ΔpilA ) on cadherin cleavage ( S2 Fig ) . The lack of pili had no effect on E-cadherin cleavage and a very partial effect on VE-cadherin . Next , we investigated whether other P . aeruginosa factors were needed for ExlA action on cadherins . As our attempts to purify a functional ExlA protein were unsuccessful , we used an Escherichia coli strain ectopically expressing ExlB-ExlA , and devoid of other P . aeruginosa factors . The E . coli-exlBA strain induced the cleavage of both E- and VE-cadherins , suggesting that ExlA alone can induce cadherin proteolysis ( Fig 1C ) . We then examined the fate of E-cadherin at cell-cell junctions by confocal videomicroscopy during infection using A549 expressing E-cadherin fused to GFP ( Fig 1D ) . We observed a loss of E-cadherin-GFP labelling at cell-cell junctions when cells were incubated with CLJ1 or IHMA87 , but not when incubated with IHMA87ΔexlA or PAO1F . This feature is in agreement with the above results by Western blot , except that the kinetics differ due to different experimental conditions . Taken together , the results show that ExlA-dependent E-cadherin cleavage promotes E-cadherin loss from cell-cell junctions . To assess cell membrane permeability , the non-permeant dye propidium iodide was added to the cell medium . Membrane disruption , as monitored by propidium iodide incorporation into cell nuclei , occurred much later than E-cadherin loss ( Fig 1D ) . Thus , ExlA effects on E-cadherin could be observed earlier than those leading to cell death . To definitely prove that ExlA induces barrier disruption , a bacterial transmigration assay was performed using Transwell filters . The bacterial transmigration across A549 monolayers were significantly decreased when IHMA87ΔexlA was used , compared to IHMA87 ( S3 Fig ) . To investigate the action of ExlA-secreting bacteria on cadherin cleavage in vivo , we infected mice by inhalation of a suspension of CLJ1 bacteria . IHMA87 was not used in this assay , as it is not a potent infective agent in this model [13] . Mice were euthanized at 18 h . p . i . and lung extracts were analysed by Western blot ( Fig 1E ) . The data revealed that CLJ1-infected lungs contained significantly lower amounts of E- and VE-cadherins than the mock-infected lungs . Likewise , CLJ1 induces the cleavage of both cadherins in vivo . These results also show that ExlA can promote cadherin cleavage from a different species . ADAMs are transmembrane metalloproteases , modulating cell-cell and cell-matrix interactions . They are major ectodomain sheddases , releasing a variety of cell-surface proteins . ADAM10 is an ubiquitous member of this family , for which several protein substrates have been identified , including E- and VE-cadherins [25–28] . ADAM10 cleaves both cadherins in their extracellular membrane-proximal domains , releasing the extracellular domain in the supernatant . The transmembrane domain is then rapidly cleaved by the γ-secretase and the cytoplasmic domain is targeted to the proteasome for degradation [26 , 28] . In our experiments , the transient C-terminal cleavage fragment ( 30 kDa ) induced by CLJ1 co-migrated with that produced by cell incubation with ionomycin , a potent inducer of ADAM10 activity [25 , 29] ( Fig 2A ) . ADAM17 activation by PMA did not produce a similar cleavage . Thus , these preliminary results suggested that ExlA could activate ADAM10 , as has been reported for Staphylococcus aureus pore-forming toxin Hla [30 , 31] . To further evaluate the role of ADAM10 in ExlA-dependent cadherin cleavage , we first used protease inhibitors: GM6001 , a wide-spectrum Zn-metalloprotease inhibitor and GI254023X , a highly specific ADAM10 inhibitor [32] . Both inhibitors prevented E- and VE-cadherin cleavages after infection with CLJ1 or IHMA87 ( Fig 2B ) . To definitely prove the role of ADAM10 in this mechanism , we generated ADAM10-deficient A549 cells , using CRISPR/Cas9 technology . The absence of ADAM10 was verified by FACS analysis ( Fig 2C , right ) . ADAM10-deficient A549 cells were resistant to E-cadherin cleavage after incubation with CLJ1 or IHMA87 ( Fig 2C , left ) . To evaluate the VE-cadherin cleavage in primary cells , we knocked down ADAM10 in HUVECs using siRNA , as it is not possible to generate gene deficiency in primary cells using the CRISPR/Cas9 system . ADAM10-surface expression was dramatically decreased in ADAM10 siRNA-treated HUVECs ( Fig 2D , right ) . VE-cadherin cleavage was prevented in ADAM10 siRNA-treated HUVECs infected with CLJ1 and IHMA87 ( Fig 2D , left ) . Altogether , these findings demonstrate that ADAM10 is the executer of ExlA for cadherin cleavage and is absolutely required for this activity . To examine whether ADAM10 was required for ExlA-dependent plasma membrane rupture , we tested the necrotic activity of ExlA in ADAM10-deficient A549 cells ( Fig 3 ) . LDH release was independent of the presence of ADAM10 , showing that ADAM10 is not involved in cell death and that ADAM10 is not the membrane receptor of ExlA . As previously described , high cytosolic Ca2+ concentrations induce cadherin cleavage through activation of ADAM10 [26 , 33] , which was further confirmed here using the calcium ionophore ionomycin ( Fig 2A ) . Therefore , we tested the capacity of ExlA-secreting strains to induce intracellular Ca2+ elevation in A549 cells and HUVECs . We used Fluo3-AM , a cell-permeant fluorescent probe , to monitor intracellular Ca2+ together with Draq7 , a non-permeant fluorescent probe , to visualize necrotic cells after plasma membrane rupture . Cell fluorescence were recorded on both channels by videomicroscopy and their intensities measured on individual cells . No Fluo3 signal was obtained in uninfected conditions and inconsistent signals were detected when cells were infected with PAO1F ( Fig 4A–4D ) , as previously reported [34] . In contrast , sharp increases of Fluo3 intensities were observed when either cell types were infected with CLJ1 ( Fig 4E and 4F ) ; this increase takes place earlier in HUVECs . Similar observations were made with IHMA87 infection ( Fig 4G and 4H ) , except that Ca2+ rise was not consistent for each A549 cell and less marked in general . When IHMA87ΔexlA was used , no Ca2+ elevation was noted ( Fig 4I and 4J ) , while Fluo3 signal increases were observable within IHMA87ΔexlA::exlBA-infected cells ( Fig 4K and 4L ) . Hence , ExlA triggers a massive Ca2+ elevation in host cells , because of Ca2+ entry through the pore formed in the plasma membrane . ADAM10 may thus be activated by this process . The increase of Fluo3 intensity was consistently followed by a sharp drop of fluorescence that was concomitant to Draq7 incorporation into the cells ( Fig 4E–4H , 4K and 4L ) , suggesting that the Fluo3 diffused in the extracellular milieu when the plasma membrane was ruptured . The inactive form of ADAM10 ( pro-ADAM10 ) , linked to its pro-domain , has been shown to interact with calmodulin , a cytosolic protein displaying high affinity for Ca2+ [29 , 35] . It was hypothesized that this interaction prevents ADAM10 pro-domain cleavage , and subsequent ADAM10 activation and export to the plasma membrane [35 , 36] . The currently accepted model proposes that when cytosolic Ca2+ concentration is increased , calmodulin and Ca2+ interact , which dissociates calmodulin from pro-ADAM10 , as its affinity is lower for the protease . ADAM10 hence becomes available for activation by furin and is eventually ready to process its own substrates [37 , 38] . To test this hypothesis in our system , we first depleted Ca2+ in the extracellular medium , but the absence of Ca2+ triggered a striking over-secretion of ExlA by the bacteria ( S4 Fig ) that masked the results on ADAM10 activation . Therefore , we employed a chemical compound , trifluoperazine ( TFP ) , that exhibits high affinity for calmodulin and dissociates calmodulin from its other interactants [29 , 35] . TFP treatment of A549 resulted in a dose-dependent induction of E-cadherin cleavage , confirming that calmodulin prevents ADAM10 activation in uninfected A549 cells ( Fig 5A ) . To further demonstrate the role of intracellular Ca2+ in ADAM10 activation , we pre-incubated A549 cells with BAPTA-AM , a cell permeant Ca2+ chelator . The chelating action of BAPTA-AM in these cells was tested by Ca2+ imaging using Fluo3-AM probe ( S5 Fig ) . Fluo-3 signal inhibition by BAPTA-AM was total in CLJ1-infected cells and almost complete when cells were infected with IHMA87 . BAPTA-AM strongly decreased CLJ1-induced E-cadherin cleavage and its action was partial in IHMA87-infected ( Fig 5B ) , consistent with the differential ability of BAPTA-AM to sequester intracellular Ca2+ in the two conditions . In conclusion , cytosolic Ca2+ is the messenger allowing ADAM10 activation in infected cells . Interestingly , BAPTA-AM also significantly diminished LDH release induced by CLJ1 or IHMA87 ( Fig 5C ) , suggesting that Ca2+ is also involved in ExlA-induced membrane permeability , independently of cadherin cleavage . As mentioned above , the pore-forming toxin ShlA is homologous to ExlA in terms of sequence identity ( 35% ) , secretion pathway and domain organization [12] . We thus examined whether ShlA had the same capacity to promote cadherin cleavage through Ca2+ influx and ADAM10 activation . To address this question , we used a ShlA-secreting strain of S . marcescens , Db11 , and its isogenic shlB mutant , 21C4 , which is impaired in ShlA secretion [39 , 40] . Db11 rapidly cleaved E-cadherin with C-terminal cleavage fragment of similar size as ExlA , whereas 21C4 did not induce E-cadherin degradation ( Fig 6A left ) . Similar data were obtained for VE-cadherin ( Fig 6A right ) . Incubation of A549 ADAM10-/- cells with Db11 did not induce E-cadherin cleavage ( Fig 6B ) . E-cadherin degradation was only partially prevented in Db11-infected A549 cells by cell pre-incubation with BAPTA-AM ( Fig 6C ) . However , BAPTA-AM incompletely chelated intracellular Ca2+ ( S5 Fig ) . Db11 increased intracellular Ca2+ in A549 cells and HUVECs , followed by Draq7 incorporation ( Fig 6D and 6E ) , while no Fluo3 or Draq7 signals were detected with 21C4 ( Fig 6F and 6G ) . Therefore , we conclude that ShlA has the same capacity as ExlA to induce E-cadherin cleavage through Ca2+ influx and ADAM10 activation .
Cell-cell junctions of mucosal and vascular barriers play an essential role in impeding the spread of bacterial pathogens in the body ( for a review on pore-forming toxin’s action on tissue barriers , see [41] ) . Here , we employed ExlA- and ShlA-secreting bacteria , and their respective non-secreting mutants , to decipher how the two toxins disrupt adherens junctions . Our results demonstrate that one of the primary effects of the cytotoxic action of ExlA and ShlA is that they induce cleavage of E- and VE-cadherins , which disrupts epithelium and endothelium integrity . The massive Ca2+ influx triggered by pore formation initiates a cascade of events leading to junction disruption . The model proposed in Fig 7 includes the following steps: in uninfected conditions , calmodulin is associated with pro-ADAM10 , preventing its maturation . Calmodulin has a high affinity for Ca2+ . When Ca2+ enters the cell through the ExlA or ShlA pore , it interacts with calmodulin , which presumably alters the conformation of calmodulin and causes its dissociation from ADAM10 . Then , the free pro-ADAM10 protein is activated by furin and translocates to the plasma membrane where it cleaves the cadherins . As cadherins are required for cell-cell adhesion , their cleavage is closely followed by cell retraction , owing to actin-cytoskeleton centripetal forces . Similar data were obtained with CLJ1 and IHMA87 , except that the latter was less effective in most assays . The ExlA proteins in these two strains share 99 . 6% identity , suggesting that the milder effect of IHMA87 may be related to a reduced capacity to secrete ExlA [13] . Ca2+ influx started earlier in HUVECs than in A549 cells , suggesting either that the pores are formed more rapidly or that a larger number of pores are formed in HUVECs . This effect could be explained by a higher expression of receptors for ExlA and ShlA at the surface of endothelial cells compared to epithelial cells . However , as no receptor has yet been identified for these two toxins , we cannot currently confirm this hypothesis . For ExlA , the receptor appears to be relatively ubiquitous as ExlA-secreting bacteria are toxic for most tested cell types except erythrocytes [13] . In contrast , ShlA targets erythrocytes , epithelial cells , fibroblasts [17] and endothelial cells ( this study ) , but not myeloid cells . The somewhat different target host cell suggests that the cellular receptors are not identical for the two toxins . Interestingly , the LDH response was unmodified in the absence of ADAM10 , when cells were exposed to ExlA- or ShlA-secreting bacteria . Thus , unlike for Hla [31] , ADAM10 is not the receptor of either ExlA or ShlA , nor it is involved in ExlA-induced cell death . Unlike most bacterial pore-forming toxins , ExlA and ShlA are highly unstable in aqueous solution and are biologically inactive when purified . The instability of ExlA and ShlA and the fact that Type IV pili are required for ExlA-related toxicity of the IHMA87 strain [12] suggest that these toxins must be delivered close to host cells where they can rapidly insert into the plasma membrane . The mechanism of insertion into the membrane and the toxin’s oligomerisation capacity remain to be determined . Because of their role in maintaining tissue barriers , cadherins are frequent targets for bacterial pathogens . Cadherins can be cleaved by ADAM10 , which is activated by the calcium imbalance caused by the pore-forming action of the toxins , as previously demonstrated for Hla from S . aureus and pneumolysin from Streptococcus pneumoniae [30 , 31 , 42] and here for ExlA and ShlA . This mechanism of toxicity might also be used by other pore-forming toxins that are known to promote Ca2+ influx , such as streptolysin O ( SLO ) from Streptococcus pyogenes , α-toxin from Clostridium septicum , aerolysin from Aeromonas hydrophila , and listeriolysin O from Listeria monocytogenes [43–46] . Alternatively , E-cadherin can also be cleaved by HtrA bacterial proteases , released by Helicobacter pylori , enteropathogenic E . coli , Shigella flexneri and Campylobacter jejuni [47 , 48] . As mentioned above , VE-cadherin is cleaved by P . aeruginosa LasB protease [23] . E-cadherin is also the target of Clostridium botulinum neurotoxin-A , preventing cadherin homotypic adhesion [49] . Finally , an intermediate mechanism was found for the metalloprotease BFT from Bacteroides fragilis , which activates the γ-secretase , promoting E-cadherin cleavage [50] . Thus , bacteria employ several independent but related pathways to target the junctions . Along the intercellular space , the cluster of E-cadherin molecules forming the adherens junctions , is located closer to the basal side of the cell layer than the tight junctions . The tight junctions seal intercellular gaps on the apical side to restrict the diffusion of molecules ( e . g . proteases ) present on the apical side of the cell layer . This restriction is less effective in endothelia , where the tight junctions tend to be less developed and where they are mixed with adherens junctions [51] . Bacterial proteases therefore have limited access to E-cadherin in most infection conditions , and the indirect action of pore-forming toxins on E-cadherin appears to be an efficient alternative mechanism to the secretion of proteases directly targeting the junctions . Because ADAM10 can efficiently cleave adhesive and signalling receptors , its activity is tightly regulated in eukaryotic cells . The subversion mechanism of ADAM10 activation by ExlA and ShlA is probably shared by several pore-forming toxins . This may have functional consequences beyond cellular adhesion because of the numerous substrates of ADAM10 , even when cell death is prevented by induction of the membrane repair machinery .
All protocols in this study were conducted in strict accordance with the French guidelines for the care and use of laboratory animals [52] . The protocols for mouse infection were approved by the animal research committee of the institute ( CETEA , project number 13–024 ) and the French Ministry of Research . Anesthesia was performed using a mixture of xylazine/ketamine and euthanasia by CO2 inhalation . The strains and the plasmids used in this study are described in S1 File . Bacteria were grown in liquid LB medium at 37°C with agitation until the cultures reached an optical density value of 1 . 0 to be in the exponential growth phase . All conditions of infection were performed at MOI of 10 , unless indicated in the legend . HUVECs were prepared from anonymized human umbilical cords ( coming from the Groupe Hospitalier Mutualiste de Grenoble ) as previously described [53] and grown in supplemented EBM2 medium ( Lonza ) . A549 ( ATCC ) , A549-E-cadherin GFP and A549 ADAM10-/- cells ( both generated in this study ) were grown in DMEM medium , supplemented with 10% foetal calf serum ( all from Life Technologies ) . Cells were tested for the presence of mycoplasma before freezing vials . Each experiment started from a mycoplasma-free frozen vial . A549 cells were tested for the presence of E-cadherin junctional labelling . A549 E-cadherin GFP cells were obtained after transfection with pCDNA3 . 1-E-cadherin-GFP ( Addgene ) using Lipofectamine 2000 ( Life Technologies ) . Positive cells were selected by antibiotic treatment and fluorescent cells were sorted by MoFlo flow cytometer ( Beckman Coulter ) . When indicated , cells were pretreated 30 min before infection with TFP at indicated concentrations , 0 . 1 μg/mL phorbol 12-myristate 13-acetate ( PMA ) , 5 μM ionomycin , 10 μg/mL GM6001 , 5 μM GI254023X or 25 μM BAPTA-AM ( all from Sigma ) . BAPTA-AM was washed out before the infection step . For videomicroscopy experiments , cells were seeded at 150 , 000 cells per well on Labtek II 8-chambered ( Thermo Fisher Scientific ) coverslips and used 48-h later to obtain highly confluent monolayers . Medium was replaced with non-supplemented EBM-2 medium one hour before infection . Cells were infected and immediately observed by videomicroscopy . For siRNA experiments , HUVECs were seeded at 5 . 104 cells/well in P12 plate . The next day , cells were transfected with anti-sense ADAM10 siRNA ( UAACAUGACUGGAUAUCUGGG ) using Lipofectamine RNAiMAX ( Life Technologies ) . Three days after transfection , cells were analysed by FACScalibur flow cytometer ( Becton Dickinson ) after staining with ADAM10 antibody ( R&D Systems ) and anti-mouse-Alexa 488 antibody ( Molecular Probes ) . ADAM10 gene knockout was generated by using guide RNA oligos specifying the human ADAM10 gene ( forward: 5’-CACCGGATACCTCTCATATTTACAC; reverse: 5’-AAACGTGTAAATATGAGAGGTATCC ) . These oligos were designed by using the tool available at http://crispr . mit . edu/ . Oligos were obtained from Eurofins Genomics and then cloned into the pSpCas9 ( BB ) -2A-GFP vector ( #48138 , Addgene ) . The constructs or the empty vector were transfected into the A549 cells using the Neon Transfection System ( Life Technologies ) . After 24 h , transfected cells were selected by flow cytometry using the GFP signal and were distributed directly in 96-well plates using a BD FACSAria flow cytometer . After 10 days of culture , cells were stained with ADAM10-phycoerythrin antibody ( R&D Systems ) for 30 min at room temperature . The samples were analysed using a BD Accuri C6 flow cytometer . ADAM10-negative clones were selected and the ADAM10 gene was sequenced . To further determine if the selected clones had mutations in both alleles , Guide-it sgRNA in vitro Transcription coupled to Guide-it Genotype Confirmation Kits ( Clontech Laboratories ) were used following the manufacturer recommendations . One clone with both allelic mutations ( A549 ADAM10-/- ) was used in this study and confirmed using BD FACScalibur flow cytometer after staining with ADAM10 antibody ( R&D systems ) and anti-mouse-Alexa488 antibody ( Molecular Probes ) for 1 h at 4°C for each antibody . Pathogen-free BALB/c female mice ( 8–10 weeks ) were obtained from Harlan Laboratories and housed in the institute animal care facility . Bacteria from exponential growth ( OD = 1 . 0 ) were centrifuged and resuspended in sterile PBS at 0 . 85x108 per mL . Mice were anesthetized by intraperitoneal administration of a mixture of xylazine ( 10 mg . Kg-1 ) and ketamine ( 50 mg . Kg-1 ) . Then , 30 μL of bacterial suspension ( 2 . 5x106 ) were deposited into the mouse nostrils . Mice were euthanized by CO2 inhalation at 18 h . p . i . and lungs were withdrawn and homogenized in 2 mL of PBS using a Polytron . The number of animals required for the study was deduced from previous work [23] . Randomization was performed by Harlan Laboratories . The experiment did not required blinding . Cells and tissues were lysed in Triton X-100 and protein concentration of the lysates was determined with a Micro-BCA kit ( Pierce ) using BSA as standard . Protein extracts were then separated by SDS-PAGE , transferred onto Hybond ECL membrane ( Amersham Biosciences ) and revealed with E-cadherin-Cter ( BD transduction laboratories ) , VE-cadherin ( Santa Cruz ) or β-actin ( Sigma ) antibodies . Luminescent signals were revealed using a ChemiDoc ( BioRad ) . Only unsaturated signal intensities are presented . To examine ExlA secretion , the cell supernatants were collected at 2 h . p . i . and concentrated by trichloroacetic acid ( TCA ) precipitation . Briefly , 100 μL of 2% Na deoxycholate were added to 10 mL of cellular supernatant and incubated 30 min at 4°C . Then , 1 mL of TCA were added and the mixture was incubated overnight at 4°C . After centrifugation for 15 min , at 15 , 000 g and 4°C , the pellet was resuspended in 100 μL of electrophoresis loading buffer . The ExlA antibody [3] was used to reveal the western blots . LDH release in the supernatant was measured using the Cytotoxicity Detection Kit from Roche Applied Science , following the recommended protocol . Briefly , cells were seeded at 2 . 5 104 in 96-well plates two days before , and infected in non-supplemented EBM2 . At different post-infection times , 30 μL of supernatant were mixed with 100 μL of reaction mix and OD was read at 492 nm . OD values were subtracted with that of uninfected cells and Triton-solubilized cells were used to determine the total LDH present in the cell culture . For fluorescence time-lapse microscopy , cells were imaged using a confocal spinning-disk inverted microscope ( Nikon TI-E Eclipse ) equipped with an Evolve EMCCD camera . The optical sectioning was performed by a Yokogawa motorized confocal head CSUX1-A1 . Images were acquired using an illumination system from Roper Scientific ( iLasPulsed ) with a CFI Plan Fluor oil immersion objective ( 40X , N . A . 1 . 3 ) . Z-series were generated every 10 min using a motorized Z-piezo stage ( ASI ) by acquiring 25 z-plane images with a step size of 1 μm . Microscope was controlled with MetaMorph software ( Molecular Devices ) . Temperature , CO2 , and humidity control was performed using a chamlide TC system ( TC-A , Quorum technologies ) . Solid-state 491 and 561 nm lasers ( iLas , Roper Scientific ) and ET 525/50M ( Chroma ) and FF01-605/54 ( Semrock ) emission filters were used for excitation and emission of EGFP and propidium iodide fluorescence , respectively . Calcium imaging was performed using Fluo3-AM probe ( Molecular probes ) . Cells were seeded at 2 . 5 104 in Labtek 8-chambered ( Thermo Fisher Scientific ) one or two days before for A549 cells or HUVECs , respectively . Before experiments , cells were washed twice with PBS . Then cells were loaded with 3 μM Fluo3-AM in PBS or EBM2 containing 2 . 5 mM probenecid and 40 mg/ml pluronic acid for 1 . 5 h at room temperature ( A549 ) or 0 . 5 h at 37°C ( HUVECs ) in the dark . Cells were washed twice in EBM2 . Infection was performed in EBM2 medium with 0 . 15 μM Draq7 ( Abcam ) . Labtek chambers containing the infected cells were placed in an incubator equilibrated at 37°C and 5% CO2 located on a IX71 Olympus microscope controlled by the CellR Olympus system , automated in x , y , z axis and driven by Xcellence software ( Olympus ) . Fluo3 was excited at 480/40 nm , and emission was collected at 535/50 nm . Draq7 dye was excited at 620/60 nm , and emission was collected at 700/75 nm . Images were captured with a Hammamatsu Orca-ER camera and a 40X ( N . A . 1 . 35 ) oil objective . Acquisitions were generated every 5 min . The cells used for analysis were chosen at the initial step of recording . Data on lung cadherin content ( Fig 1D ) passed the normality test ( Shapiro-Wilk's test ) , but not the equal variance test . Therefore , significance was evaluated on ranks , using a two-sided Mann-Whitney’s test . LDH data ( Fig 5C ) passed the normality and equal variance tests; significance was thus analysed by a two-sided Student’s t-test . Statistics were performed using SigmaPlot software .
|
Pore-forming toxins are the most widespread toxins delivered by pathogenic bacteria and are required for full virulence . Pore-forming toxins perforate membranes of host cells for intracellular delivery of bacterial factors , for bacterial escape from phagosomes or in order to kill cells . Loss of membrane integrity , especially the plasma membrane , has broad implications on cell and tissue physiology . Here , we show that two members of a unique family of pore-forming toxins , secreted by Pseudomonas aeruginosa and Serratia marcescens , have the capacity to disrupt cell-cell junctions of epithelial and endothelial cells , hence breaching two major tissue barriers .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"toxins",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"pathogens",
"endothelial",
"cells",
"microbiology",
"cadherins",
"toxic",
"agents",
"toxicology",
"pseudomonas",
"aeruginosa",
"epithelial",
"cells",
"enterobacteriaceae",
"serratia",
"cellular",
"structures",
"and",
"organelles",
"bacteria",
"bacterial",
"pathogens",
"pseudomonas",
"animal",
"cells",
"cell",
"adhesion",
"medical",
"microbiology",
"microbial",
"pathogens",
"serratia",
"marcescens",
"biological",
"tissue",
"cell",
"membranes",
"intracellular",
"membranes",
"host",
"cells",
"cell",
"biology",
"anatomy",
"virology",
"epithelium",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"organisms"
] |
2017
|
Pseudomonas aeruginosa ExlA and Serratia marcescens ShlA trigger cadherin cleavage by promoting calcium influx and ADAM10 activation
|
Analyses of genome reduction in obligate bacterial symbionts typically focus on the removal and retention of protein-coding regions , which are subject to ongoing inactivation and deletion . However , these same forces operate on intergenic spacers ( IGSs ) and affect their contents , maintenance , and rates of evolution . IGSs comprise both non-coding , non-functional regions , including decaying pseudogenes at varying stages of recognizability , as well as functional elements , such as genes for sRNAs and regulatory control elements . The genomes of Buchnera and other small genome symbionts display biased nucleotide compositions and high rates of sequence evolution and contain few recognizable regulatory elements . However , IGS lengths are highly correlated across divergent Buchnera genomes , suggesting the presence of functional elements . To identify functional regions within the IGSs , we sequenced two Buchnera genomes ( from aphid species Uroleucon ambrosiae and Acyrthosiphon kondoi ) and applied a phylogenetic footprinting approach to alignments of orthologous IGSs from a total of eight Buchnera genomes corresponding to six aphid species . Inclusion of these new genomes allowed comparative analyses at intermediate levels of divergence , enabling the detection of both conserved elements and previously unrecognized pseudogenes . Analyses of these genomes revealed that 232 of 336 IGS alignments over 50 nucleotides in length displayed substantial sequence conservation . Conserved alignment blocks within these IGSs encompassed 88 Shine-Dalgarno sequences , 55 transcriptional terminators , 5 Sigma-32 binding sites , and 12 novel small RNAs . Although pseudogene formation , and thus IGS formation , are ongoing processes in these genomes , a large proportion of intergenic spacers contain functional sequences .
Obligate bacterial symbionts possess the smallest cellular genomes [1] . Due to their extreme genome reduction , they retain only small fractions of ancestral gene sets . Comparisons of complete genome sequences within clades have revealed that genomes of obligate symbionts are stable , with few rearrangements and no uptake of novel genes over millions of years of evolution ( e . g . , [2]–[5] ) . The principal changes in these genomes are rapid sequence evolution combined with the ongoing erosion and loss of genes due to a mutational bias toward deletions [6]–[7] . This removal of nonfunctional sequences is a unidirectional process leading to ever-shrinking gene sets . Protein-coding genes are relatively easy to recognize in these genomes , based on lengths of undisrupted open reading frames ( ORFs ) and on their clear homology to proteins encoded in other bacterial genomes . Functional inferences based on such homology have yielded insights into symbiont roles within hosts by establishing the specific metabolic capabilities that are retained in symbionts ( e . g . , [8]–[10] ) . In contrast , the intergenic spacers in these small genomes are more enigmatic . On account of the ongoing gene erosion and loss , many spacers consist , in whole or part , of inactivated pseudogenes in varying stages of decay [11] . At the same time , some sequences within intergenic spacers represent functional elements that are retained for their roles in gene regulation [12] . These are of particular interest because regulatory processes are the least understood aspects of symbiont genomes , which have lost most ancestral regulatory mechanisms [13]–[14] . Discriminating between decaying genes and functional elements within spacer regions is difficult in that inert sequences can be of varying sizes and base compositions and there is no single model for recognizing functional motifs . In Buchnera aphidicola , which has coevolved for over 150 million years with its aphid hosts , intergenic spacers occupy about 15% of the genome [15] , which is within the typical range for bacterial genomes [6] , [7] . These spacers contain a mixture of neutral sequences and functional elements , but the latter are largely unrecognized and undefined . One way in which the functional relevance of a sequence can be assessed is through experimental disruption; however , obligate symbionts including Buchnera cannot be cultured in the laboratory , limiting most experimental approaches for linking sequences to functions . However , comparative genome analysis can reveal functional sequences in these genomes , since sequences under either purifying ( negative ) or positive selection will exhibit distinct patterns of evolution . But this approach relies on genomic sequences that are of appropriate levels of divergence in order to manifest meaningful signals in such comparisons . In the past , available genomes for Buchnera corresponded to four very distantly related aphid species that were too divergent for alignment of most intergenic spacers [16] . Conversely , multiple genomes available from Buchnera of a single host species , the pea aphid ( Acyrthosiphon pisum ) , are nearly identical ( <0 . 3% divergence ) , precluding the discrimination of sequences under different functional constraints [17] . In this study , we compare genomes of eight Buchnera of varying divergence times to monitor the origination and decay of intergenic spacers from previous protein-coding regions and to detect evolutionarily conserved elements within spacers . Two Buchnera genomes were newly sequenced and annotated for this study to permit analyses at evolutionarily relevant levels of sequence divergence . An advantage of Buchnera species for this phylogenetic footprinting approach is the complete conservation of gene order and orientation , allowing orthology of spacers to be assigned with confidence even when sequence divergence is high .
The genomes of Buchnera-Ak and Buchnera-Ua were completed at high depth of coverage ( >100X , Table S1 ) , and their general features were compared with previously sequenced Buchnera genomes ( Table 1 ) . Both genomes contained a single chromosome and two plasmids , pLeu and pTrp , as observed in other Buchnera of aphid species within the subfamily Aphidinae . The relative depths of coverage of chromosomal sequences and plasmids by Illumina reads suggest that both plasmids are present at 2 to 3 copies per chromosome in Buchnera-Ak and at about 0 . 3 to 0 . 4 copies per chromosome in Buchnera-Ua . Since Buchnera cells are highly polyploid [18] , plasmids can be present in lower copy number than the main chromosome and still present in every cell . The deep sequence coverage revealed several polymorphic sites within both Buchnera-Ak and Buchnera-Ua , including both single nucleotides changes and single base indels and totaling 13 and 15 sites in the two species ( Table S2 ) . Polymorphisms were unexpected since both samples were derived from lab cultures descended from a single founding female and bottlenecked every few generations during rearing . Also , intrastrain polymorphism was not previously detected in deep sequencing of Buchnera-Ap strains [17] , although it was recently detected in Buchnera-Sg [19] . To achieve our aim of analyzing IGS to detect gene decay and conserved functional elements , we needed a phylogeny for our strains . We reconstructed phylogenies based on sequences of proteins inferred from the sequenced genomes; these were consistent both with those based on 16S rRNA sequences for a larger set of species ( Figure S1 ) and with current knowledge of aphid phylogenetic relationships [20]–[21] . Relationships were unambiguous except that the branching order of the 3 Buchnera-Ap genomes could not be resolved unequivocally due to the recency of their divergence . We also estimated divergence dates within the Aphidinae using the consensus phylogeny ( Figure 1 ) . Using the two new genome sequences for Buchnera-Ak and Buchnera-Ua , we were able to detect and reconstruct gene content changes on lineages within the clade containing Buchnera-Ap strains , Buchnera-Ak , Buchnera-Ua , and Buchnera-Sg , corresponding to Buchnera of the subfamily Aphidinae ( Table S3 ) . These changes were scored as either losses ( no gene remnants detectable ) or inactivations ( recognizable pseudogenes remain ) ( Figure 1 ) . Compared to the Buchnera of other Aphidinae species , Buchnera-Ua has lost more genes due to both deletions ( up to 5 kb ) and to the inactivation of genes that are intact in other Buchnera genomes . The major deletions include genes underlying the membrane-associated complex involved in oxidative stress ( rnfABCDGE ) , DNA repair ( mutT , mutS , mutL ) , pyrimidine biosynthesis ( pyrC , pyrD , pyrBI ) , pantothenate biosynthesis ( panBC ) , spermidine biosynthesis ( speED ) , ornithine biosynthesis ( argECB , argA , argD ) and a transcriptional dual regulator ( hns ) . Differences in gene content between Buchnera-Ak and Buchnera-Ap are fewer ( Table S3 ) . Notably the major regulator of metA in methionine biosynthesis , metR , is intact in Buchnera-Ak and a pseudogene in Buchnera-Ap , suggesting that Buchnera-Ak may retain substrate-based activation of metE transcription , as observed previously in Buchnera-Sg which also retains intact metR [14] . An uninterrupted copy of Asparaginase I ( ansA ) was also identified in Buchnera-Ak , which is responsible for the interconversion of aspartate and ammonia into asparagine . In light of the current picture of metabolic cooperation between Buchnera and A . pisum [22] , this gene could influence amino acid metabolism and nitrogen recycling in A . konodi . Also , Buchnera-Ak has undergone inactivation of the regulator hns , for which the ORF is intact in Buchnera-Ap . The majority ( 42/69 ) of pseudogenes identified in these four genomes are restricted to single genomes and are caused by mutations in homopolymeric runs , as observed previously for pseudogene formation in strains of Buchnera-Ap [17] . In several instances , the identical gene has been inactivated independently in different lineages and through different mutations ( Table S3 ) . As one example , the gene murF , which is involved in biogenesis of the cell envelope , is intact only in Buchnera-ApTokyo but is inactivated by homopolymeric frameshifts in Buchnera-Ap strains 5A and Tuc7 , Buchnera-Ak , Buchnera-Ua , Buchnera-Sg . This situation implies that there were four independent inactivations of this gene in the different lineages . The genes neighboring murF , murE and mraY are also inactivated by homopolymeric frameshifts in Buchnera-Sg . A second such example involves the global regulator hns mentioned above , which is intact in the three closely related Buchnera-Ap genomes but is deleted or pseudogenized in the other sequenced Buchnera genomes , implying three independent losses . Our comparisons reveal fourteen other cases in which orthologs , spanning a variety of functional categories , are independently inactivated in different Buchnera genomes , indicating that these genes are prone to repeated inactivation ( Table S3 ) . Orthologous IGSs were compared for Buchnera of the four species of Aphidinae , and a strong association was detected between the IGS lengths of Buchnera-Ap and those of each of the others ( Figure 2 ) . The relationship among spacer lengths is even stronger for more closely related species pairs , such as Buchnera-Ap and Buchnera-Ak . A phylogenetic footprinting approach [23]–[24] was utilized to identify sequence blocks within IGSs that might be conserved as functional elements , such as riboswitches , binding sites affecting transcription , transcriptional terminators , and sRNAs . Buchnera-Sg , Buchnera-Ua , Buchnera-Ak , and Buchnera-Ap contain from 580 to 621 genes , and total of 537 orthologous IGS regions were identified for these species . Of these , we focused our analysis on the 336 IGSs with alignments of at least 50 nucleotides and with no zero or negative length spacers ( overlapping genes ) . The IGSs in these genomes have an average base composition of only 14–16% G+C , whereas the mean G+C contents of coding regions range from 25 . 3–27 . 3% . These spacers can be divided according to whether they are flanked by genes transcribed in tandem ( 225 IGSs ) , convergently ( 42 IGSs ) , and divergently ( 70 IGSs ) , respectively ( Figure S2 ) . The average sequence identity for the three categories is similar , ranging from 51 . 3–53 . 9% , but the average alignment length is substantially greater for IGSs of divergently oriented genes , at 288 nucleotides , compared to 160 and 136 nucleotides for the tandem and convergent categories . The correlation of spacer lengths is the greatest for the divergent category and lower for the other two categories; for the species pair Buchnera-Sg–Buchnera-Ap , the pairwise correlation of IGS lengths are 0 . 78 for tandem , 0 . 80 for convergent , and 0 . 87 for divergent IGSs ( Pearson's r ) . To detect whether the IGSs contain conserved regions of possible functional significance , we examined the occurrence of perfectly conserved k-mers of at least five nucleotides in the 4-taxon alignments ( hereafter referred to as “conserved k-mers” ) . The length distribution of conserved k-mers was compared to that of the randomly shuffled alignments . The two distributions depart , with an increased incidence of conserved k-mers for the observed IGS alignments than would be expected by chance ( Figure 3 ) . The elevation of conserved k-mers relative to the random expectation was far more pronounced for IGSs than for protein-coding regions . This observation can be understood as the result of the saturation of synonymous sites across the evolutionary divergence represented by these four Buchnera species . Even though many proteins have high conservation of amino acid sequences , synonymous site differences reduce the number of conserved k-mers to only slightly higher than the number expected by chance . The pattern observed in coding regions , an excess of ( 3n−1 ) k-mers , is a result of this saturation at synonymous ( third ) positions ( Table 2 ) . Despite the large synonymous divergences among Buchnera strains , we find that k-mers≥6 nucleotides make up 26% of the alignment corresponding to the 473 coding sequences ( 126 , 973 nt/487 , 127 nt ) ; however , such k-mers constitute only 8 . 4% of the alignment corresponding to the IGSs ( 5 , 068 nt/60 , 673 nt ) . This difference reflects the fact that coding regions are generally more conserved , due to strong conservation at nonsynonymous sites , than IGS regions . However , when conservation does occur in IGS regions , it often involves runs of 6 or more nucleotides , which is not usually true in coding regions . The excess of conserved k-mers was also more pronounced for k-mers conserved across the 4-taxon alignment than for k-mers conserved exclusively within the shallower alignments of 3 or 2 taxa ( Figure 3 ) . The opposite would be expected if the excess of conserved k-mers in IGS regions merely reflected recent ancestry reflected in high sequence similarity even in neutral regions . Instead , the robust conservation even at deeper divergences , which are fully saturated for changes at neutral sites , favors a major role of purifying selection in maintaining these sequences . For IGSs , the excess of conserved k-mers was observed across all length categories , from 5 to 22 nucleotides . It is possible that the excess of conserved k-mers is merely the by-product of low-complexity , A−T-rich intergenic spacers , however two lines of evidence oppose this view . ( 1 ) Of the 336 IGS regions , 232 contained at least one conserved k-mer of at least 5 nucleotides , whereas in shuffled alignments with the same base compositions , only 137 IGSs contained a conserved k-mer of at least 5 nucleotides . ( 2 ) Although many ( 320/775 ) k-mers are entirely composed of As and Ts , longer k-mers tend to be more GC rich , and the overall distribution of GC compositions of k-mers departs from that of IGSs as a whole ( Figure S3 ) . This suggests that most IGSs of length greater than 50 nucleotides contain functional regions that are highly constrained . Considering the three IGS categories based on orientation of flanking genes , the excess of conserved k-mers was greater for IGSs between divergently transcribed genes , at 1 . 44 k-mers per 100 nucleotides compared to IGSs between tandem or convergently transcribed genes ( 1 . 07 and 1 . 26 IGS per 100 nucleotides ) . Because the IGSs between divergently transcribed genes were also longer on average , they had substantially more conserved k-mers per IGS . There is additional evidence of some type of functional element within the majority of the 232 IGSs with conserved k-mers . First , 78 IGSs contained one or more perfectly conserved , or nearly perfectly conserved , Shine-Dalgarno ( SD ) sequence ( consensus AGGAG ) for the four taxon alignment ( n = 88 ) , indicating that such alignments can be used to improve annotation of SD sequences . In addition , we detected 55 putative transcriptional terminators . Of these , most occur in orthologous locations of validated ( n = 11 ) or predicted ( n = 25 ) transcriptional terminators in E . coli K12 ( EcoCyc and TransTermHP ) . Four of the conserved IGSs corresponded to previously identified RpoH ( heat shock sigma factor , or Sigma-32 ) binding sites upstream of dnaKJ , grpE , groESL , and ibpA [13] with near-perfect conservation among the four taxa . In addition , our analyses detected a putative RpoH binding site upstream of rpoD . Four regions with conserved k-mers contained sRNAs previously annotated in Buchnera and correspond to the 4 . 5S RNA component of the signal recognition particle ( ffs ) , the catalytic subunit of RNAse P ( rnpB ) , transfer-messenger RNA ( ssrA ) , and 5S ribosomal RNA ( rrf ) . Using RNAz another 12 putative sRNAs were detected , including two with positional orthology to tpke11 and sraA of E . coli K12 . Thus , of the 232 IGSs with at least one conserved k-mer , a total of 129 encode regions for which a function can be inferred based either on their homology sequences in E . coli or on their structural features . Because RNAz cannot be used to analyze sequences of >75% A+T content and SD sequences are quite short ( 5 nt ) , 93 IGS alignments with k-mer blocks ( see Materials and Methods ) were reanalyzed for additional functional elements . The IGS alignments were trimmed down to the boundaries of the conserved regions and analyzed with RNAalifold to detect secondary structural properties . Of the 93 IGSs examined , 61 had significant secondary structures ( t-test , p<0 . 05 ) , and many formed conserved hairpin structures although some contained internal bulges ( Figure 4 ) . Some of these may represent transcriptional terminators; however , it is possible , particularly among the longer conserved regions with more complex structural predictions , that these regions are expressed as sRNAs . Moreover , 22 of these 93 conserved IGSs are found between divergently arranged CDS , raising the possibility that the conserved regions act as binding sites or possibly leader sequences influencing transcription or translation of the mRNAs . Further analyses of all the IGSs , including several conserved in only three taxa due to deletions in individual Buchnera lineages , revealed another 12 conserved transcriptional terminators and 10 well-conserved Shine-Dalgarno sequences . Of the total of 67 transcriptional terminators , 29 were also retained in Buchnera-Bp and/or Buchnera-Cc . The majority of the transcriptional terminators shared by all six genomes , however , do not show absolute conservation of the primary sequence . When the IGSs of Buchnera-Bp and Buchnera-Cc are considered with those from the four other genomes , only 281 of the 537 orthologous spacers are conserved , and only 25 k-mers≥6 nt are identified . This includes k-mers that coincide with IGSs containing putative RNA secondary structures ( n = 12 ) , Sigma-32 binding sites ( n = 3 ) and a single transcriptional terminator ( n = 1 ) .
The genomes of obligate bacterial symbionts are highly reduced in size and gene repertoires due to a combination of factors . First , the symbiotic lifestyle in a nutrient-rich host environment renders numerous genes superfluous , allowing the inactivation of many previously functional regions [10] . In addition , the dynamics of symbiont transmission to new hosts involve severe restrictions in population size and impose clonality , thereby reducing the efficacy of selection and fostering the accumulation of deleterious mutations [25]–[27] . When combined with the pervasive mutational bias in bacteria in which deletions outnumber insertions , regions that are not under strong selective constraints erode and are eventually lost , leading to small and compact genomes [6]–[7] . Due to difficulties in defining the variety or size of functional elements that might potentially occur within intergenic spacers , the most common comparative methods , such as Ka/Ks ratio tests , are not useful for differentiating spacers ( or those portions of spacers ) that are functional from those that are inert . This problem is further compounded by the extreme AT-richness of symbiont genomes , which can cause erroneous results from motif-finding and structural algorithms . Therefore , we tested the degree of conservation for a series of short sequences ( k-mers ) across orthologous regions from Buchnera aphidicola of varying degrees of phylogenetic relatedness ( Figure S1 ) . To enhance the strength and validity of these tests , we generated complete Buchnera genome sequences for two aphid species ( Buchnera-Ua and Buchnera-Ak ) , which provided information at intermediate levels of relatedness . Based on the presence of identical k-mers within orthologous regions across genomes , intergenic spacers ( IGSs ) contain an excess of conserved k-mers relative to protein-coding regions , indicating most IGSs contain some type of functional elements . Because these analyses require that k-mers be identical , many of the functional regions within IGSs are considerably longer than the associated k-mers but do not show perfect conservation along their entire lengths . Also , we found that conserved k-mers are often located near one another in the same IGS ( k-mer blocks ) , suggesting that they are parts of the same functional element . Orthologous IGSs exhibit not only sequence conservation , as reflected in the elevated numbers of identical k-mers , but also substantial conservation of length across Buchnera genomes ( Figure 2 ) . The similarity in spacer length among Buchnera lineages could be attributable either to selection on the functional elements within spacers or simply to shared ancestry , such that genomes retain ancestral spacer lengths due to lack of time for mutations affecting length to occur . However , the latter explanation is excluded by the observation that DNA from inactivated functional elements is largely eliminated across the time scales corresponding to divergence of these lineages ( up to 70 MYA ) . For example , along the lineage leading to Buchnera-Ua , DNA for 36 genes or pseudogenes was eliminated , with only 13 pseudogenes recognizable in the genome . Thus , the conservation of spacer lengths is largely attributable to functional constraints . Taken as a whole , our analyses established that at least 201 of the 336 IGSs of at least 50 nucleotides in length encode functional elements ( Figure 5 ) . Some of those remaining might harbor functional sequences that do not rely on conserved motifs; for example , the standard sigma-70 binding sites ( RpoD sites ) have a relatively weak and AT-rich consensus sequence in E . coli ( TTGaCann [15]–[19] nnTAtAaT ) . However , many IGSs probably consist of decaying pseudogenes . We note that the IGSs that do not contain conserved k-mers have more often undergone changes in length during the divergence of Buchnera-Ap and Buchnera-Ak ( Figure 6 ) , as expected for functionally inactive regions ( Fisher's r-to-z transformation , p<0 . 0001 ) . Our analyses focused on a cluster of Buchnera corresponding to aphids in the Aphidinae and including the focal species , Buchnera-Ap . This depth of comparison provided sufficient divergence to detect conserved elements not due to recent shared ancestry . Searching for conserved IGS regions in the more distant genomes of Buchnera-Cc and Buchnera-Bp did reveal some of the same elements . However , these were relatively few due to the reduction in number of clearly identifiable orthologous IGSs ( reflecting divergence in gene repertoires ) and to the lack of strict conservation of sequence for stretches corresponding to k-mer lengths . The intermediate level of comparison enabled by the newly sequenced genomes was critical to detecting conserved elements . In addition to the variation observed in intergenic spacers , the process of genome reduction is also expected to cause differences in the gene catalogs and the pseudogene contents of these genomes . Among Buchnera from aphids in the subfamily Aphidinae , the newly sequenced Buchnera-Ua encodes the fewest protein-coding genes . Certain of these gene losses in Buchnera-Ua may reflect changes in its nutritional ecology , related to the host plant , or a greater reliance upon the host or presence of an additional symbiont . The composition of phloem sap ingested by U . ambrosiae feeding on one of its host plant ( Tithonia fruticosa ) contains very high amounts of arginine ( 25% of free amino acids ) [28]; elevated arginine in the diet potentially has led to relaxation of selection for the maintenance of ornithine biosynthesis , resulting in the loss of that pathway in Buchnera-Ua . Alternatively these gene losses ( argECB , argA , argD ) may be influenced by increased metabolic cooperation between U . ambrosiae and Buchnera-Ua in that the activity of an aphid-derived ornithine aminotransferase ( EC 2 . 6 . 1 . 13 ) , involved in analogous functionality , was recently demonstrated to be up-regulated in the bacteriocytes of A . pisum [22] . The Buchnera-Ua genome also has lost the genes for pantothenate biosynthesis ( panBC ) , possibly due to the transfer from dependence on Buchnera for pantothenate provisioning to dependence on another bacterial symbiont , Hamiltonella defensa , which is universally present in U . ambrosiae and closely related Uroleucon from North America . This Uroleucon-associated strain of H . defensa contains panBCD and appears to be a stable coevolving symbiont of this clade of Uroleucon species , along with Buchnera ( [29] , P . Degnan unpublished ) . Ongoing gene erosion in Buchnera has resulted in 15 convergent cases of gene inactivation and loss along independent lineages of the Aphidinae . Although several of these events involve highly degraded or deleted genes ( e . g . , ansA , hflD , hns ) , nine involve inactivating mutations generated by an indel in a homopolymeric tract . Such mutations are common in endosymbiont genomes due to their highly biased base compositions and are commonly interpreted as ‘recent’ gene inactivations [17] , [30] . In fact , between 10 and 70% of disrupted genes identified in Buchnera genomes result from indels occurring within homopolymeric repeats ( Table 1 ) . However , it has been demonstrated that mRNAs for an inactivated locus in Buchnera of the aphid Rhopalosiphum padi can be corrected by transcriptional slippage to yield functional proteins [31] . This phenomenon has been suggested to potentially play a role in regulating gene expression [31] . Therefore , while some convergent gene loss may be the result of independent inactivation events , reflecting low functional constraint , it is plausible that some of these mutations provide an alternative means of gene regulation in Buchnera . Many of the spacers that do not contain functional elements are pseudogenes in various stages of decay , including some newly identifiable on the basis of comparisons between Buchnera-Ap , Buchnera-Ak and Buchnera-Ua ( Figure 5 ) . Although the symbiosis between Buchnera and aphids has existed for more than 150 million years and the ancestral Buchnera already had a highly reduced genome [11] , the loss of genes has been ongoing during this period , even among strains confined to a single aphid host species , as observed for the Buchnera-Ap strains ( Fig . 2 in [17] ) . The continuous production of new pseudogenes , and the resulting new intergenic spacers , is perhaps surprising given the long co-evolution and functional interdependence of the symbiont and host . However , Buchnera genomes are not nearly the smallest genomes found in symbiotic bacteria [32]–[33] , implying that symbiotic bacteria are able to endure and compensate for continued gene loss .
Two aphid species , Acyrthosiphon kondoi ( blue alfalfa aphid ) and Uroleucon ambrosiae ( brown ambrosia aphid ) , were selected based on availability and on evolutionary relationships to aphid species for which Buchnera genome sequences were already available . Isofemale lines of A . kondoi and U . ambrosiae str . UA002 were established from single parthenogenic females collected in Tucson , AZ . The A . kondoi was collected from Medicago sativa ( alfalfa ) on 21 March 2007 and then reared on M . sativa in a growth chamber at 20°C under a 16 h light/8 h dark regime . The U . ambrosiae was collected from Encelia farinosa ( brittlebush ) on 18 March 2006 and maintained on Tithonia rotundifolia ( Mexican sunflower ) under ambient greenhouse conditions at ∼25°C with natural lighting . Preparation of purified DNA from Buchnera of A . kondoi ( Buchnera-Ak ) and Buchnera of U . ambrosiae ( Buchnera-Ua ) was performed as in Charles and Ishikawa [34] , with the following modifications . Buchnera cells from A . kondoi were isolated by initially homogenizing whole insects in Buffer A-250 ( 250 mM EDTA pH 8 . 0 , 35 mM Tris pH 8 . 0 , 25 mM KCl , 10 mM MgCl2 , 250 mM sucrose; as in [35] ) and filtering through a 100 µm nylon filter . The cells were centrifuged at 1 , 500× g and 4°C for 10 minutes , resuspended in fresh Buffer A-250 and then serially filtered through 20 , 11 , and 8 µm filters . Intact Buchnera cells were collected by a final centrifugation for 25 minutes at 1 , 500× g and 4°C . The DNA sample prepared for 454-pyrosequencing was immediately extracted using the PureGene Tissue Core Kit B ( Qiagen ) . The sample prepared for Illumina sequencing was further purified using a Percoll density gradient prior to DNA extraction . Genomic DNA from Buchnera-Ua was isolated in a similar fashion as above but with the following changes: Buffer A-100 was used ( 100 mM EDTA pH 8 . 0 , 35 mM Tris pH 8 . 0 , 25 mM KCl , 10 mM MgCl2 , 250 mM Sucrose ) , only 100 , 20 and 11 µm filters were used , intact cells were treated with DNAse prior to DNA extraction , and the Percoll density gradient was not used . Genomic DNA was submitted for standard 454 pyrosequencing although different sequencing strategies were employed depending on the particular template . Buchnera-Ak was sequenced with a half run of GS-FLX , and Buchnera-Ua was sequenced with one and a half runs of GS-Titanium . To correct any potential artifacts introduced by 454 sequencing , samples were also sequenced on the Illumina platform: Buchnera-Ak using a high-AT , amplification-free library for 60 cycles , and Buchnera-Ua using a standard library for 36 cycles . Pyrosequencing reads were assembled using Newbler ( v1 . 1 . 03 . 24 for Buchnera-Ak , v2 . 0 . 00 . 19 for Buchnera-Ua ) with default parameters and exported for Consed ( v19 ) [36] . Contigs were binned by Blast [37] similarity to the Buchnera-ApTokyo genome [15] , and contaminating scaffolds were removed ( e . g . , aphid ) . The contigs were then ordered and merged in Consed using Sanger sequencing reads produced by amplified PCR products spanning the gaps ( as in [38] ) . Illumina reads were used to correct the consensus using the Perl script “addSolexaReads . pl” , which implements cross_match and is distributed with Consed . The entire genomes were manually screened , and any remaining assembly artifacts corrected . Gene predictions were performed with Prodigal ( v1 . 10 ) [39] , tRNA-scan ( v1 . 23 ) [40] and Blast ( v2 . 2 . 24+ ) . Functions of identified genes were inferred from E . coli K12 orthologs and metabolic descriptions in EcoCyc [41] . Pseudogenes were identified as sequences showing significant homology to known genes but that possessed one or more inactivating mutations that resulted in truncation to <80% of the length of known homologs . Inactivations resulting from frameshifts in homopolymeric tracts were counted separately . We note that previous studies and genome annotations used criteria for pseudogene designation that were more [15] or less [17] stringent than that applied here , often resulting in differences in pseudogene counts between studies . Sequence polymorphisms were recognized as sites that had multiple reads for each of two alternative bases or for an insertion/deletion in both the Illumina and 454 datasets . These genomes have been deposited in GenBank ( Buchnera-Ak: CP002645–CP002647 and Buchnera-Ua: CP002648–CP002650 ) . A total of eight Buchnera genomes , including the two newly determined genome sequences for Buchnera-Ak and Buchnera-Ua were included in analyses . Previously available sequences include Buchnera that are obligate symbionts of Acyrthosiphon pisum ( strains Tokyo , 5A , Tucson ) , Schizaphis graminum , Cinara cedri and Baizongia pistaciae; these symbionts are designated Buchnera-ApTokyo , Buchnera-Ap5a , Buchnera-ApTuc7 , Buchnera-Sg , Buchnera-Cc , and Buchnera-Bp ( Figure S1 ) . Given the high similarity among the Buchnera genomes from A . pisum [17] , our analyses often used only one of the three genomes , which was subsequently designated Buchnera-Ap . Complete Buchnera genomes were aligned with MAUVE [42] , identifying shared orthologs . Conserved protein-coding genes were individually aligned with MAFFT [43] based on their amino acid sequences and then reverse-translated . For sets of orthologs containing one or more pseudogenes , intact genes were aligned as above , and pseudogenes were added secondarily and aligned manually . Estimates of pairwise divergence were calculated with PAML for each gene ( codeml runmode = −2 ) [44] . Additionally , a phylogeny based on concatenated amino acid alignments was estimated in RAxML [45] utilizing 20 random sets of 50 proteins common to all eight Buchnera genomes ( 325 CDS ) , with dates estimated with r8s [46] . Best topologies and 100 bootstrap replicates for each data set were calculated using the CPREV amino acid substitution matrix and the gamma model of rate heterogeneity . Divergence dates were estimated for each of the 20 phylograms by cross-validating ages calculated from 10 replicates using the Penalized Likelihood ( PL ) method and Truncated Newton ( TN ) algorithm with a gamma rate distribution . The twenty tree topologies generated for each data set were virtually identical , except for the placement of the three Buchnera-Ap genomes , which differed in very few amino acid residues [17] . Maximum-likelihood estimates indicate a divergence between the genera Uroleucon and Acyrthosiphon occurring 41±7 MYA and the split between the two Acyrthosiphon species occurring 27±8 MYA ( Figure 1 ) . These ages represent averages from the 20 tree topologies using a mean fixed divergence time of 65 MYA between the Aphidini and Macrosiphini [47] . Estimates using a constrained range ( 50–70 MYA ) or fixed minimum or maximum divergence dates were not markedly different . For comparative purposes a 16S ribosomal RNA ( rRNA ) phylogeny of Buchnera aphidicola was generated using available sequences in GenBank . Briefly , full-length sequences were downloaded ( >1 , 200 nt ) , aligned with MAFFT and analyzed with MrBayes ( v3 . 1 . 2 ) [48] . The topology and posterior probabilities were co-estimated from two independent runs , each run with four chains progressing for 10 million generations and using a burn-in equal to 5% of the saved trees . Likelihood model parameters were set to six nucleotide rate categories , which varied according to the gamma distribution and included a proportion of invariant sites . Intergenic spacers ( IGSs ) were considered to be orthologous for different Buchnera genomes if flanked by orthologous genes in genomes being compared . Orthologous Buchnera IGSs were aligned with MAFFT and compared across species to identify sequence blocks containing conserved elements possibly indicating functional roles ( transcriptional terminators , sRNAs , etc . ) . These analyses were conducted at two phylogenetic levels: across Buchnera of Aphidinae ( Buchnera-Sg , Buchnera-Ua Buchnera-Ak , Buchnera-Ap ) and across all sequenced Buchnera ( including Buchnera-Cc and Buchnera-Bp ) . Although these comparisons only included the Tokyo strain of Buchnera-Ap , the patterns were virtually identical when the other Buchnera-Ap strains are included . We searched IGS alignments for perfectly conserved nucleotide stretches ( k-mers ) of greater than five nucleotides in length across the genomes of interest . The results were compared to random sequences that were obtained for each IGS alignment by shuffling with Multiperm ( v0 . 9 . 4 ) [49] . Conserved IGSs were then analyzed by a variety of means to identify possible functional elements , which were then collated with the k-mer results . Computational detection of Shine-Dalgarno sequences ( ribosome binding sites ) was carried out using RBSfinder [50] and of rho-independent transcriptional terminators using TransTerm-HP ( v2 . 07 ) [51] with default parameters . Buchnera IGSs were interrogated directly for confirmed E . coli K12 sRNAs ( n = 83 ) using Blastn and positional homology . Subsequent identification of potentially novel sRNAs was performed in a two-step process . First , all IGSs alignments were analyzed with RNAz ( v1 . 0 ) [52] for conserved thermodynamically stable RNAs . However , RNAz only works reliably for sequences of >25% G+C , and the mean G+C% of Buchnera IGS is ∼15% . Therefore , remaining highly conserved IGSs lacking identifiable functional domains and “k-mer blocks” ( ≥2 k-mers less than 30 nt apart or k-mers ≥10 nt ) were analyzed with RNAalifold ( v1 . 8 . 4 ) [53] for signatures of conserved secondary structure . Then individual alignments were randomly shuffled 100 times with the Perl script “rnazRandomizeAln . pl” distributed with RNAz , and the presence of possible secondary structures was determined with RNAalifold . A t-test was used to determine if the predicted free energy ( kcal/mol ) of the actual alignment was significantly less than the distribution of values calculated for the shuffled alignments . We note that the high A+T contents of Buchnera sequences were problematic for some motif finders , such as AlignAce ( data not shown ) . All of the identified conserved elements are listed with their genome coordinates in the supporting information ( Table S4 ) .
|
Endosymbiotic associations , such as that between Buchnera aphidicola and its aphid hosts , persist for millions of years and result in substantial changes to symbiont genomes . Most notably , symbionts exhibit reductions in genome size , the accumulation of slightly deleterious mutations , and rapid rates of evolution of gene sequences . Genomic studies have enabled the identification of the specific metabolic contributions of bacteria to their hosts; however , such reconstructions have had limited success recognizing the functional elements present in intergenic regions . Non-coding regions can result from reductive evolution and genome decay , i . e . through the inactivation of previously functional regions; but they may contain any of several classes of regulatory control elements that have been maintained over the history of the symbiosis . We produced complete sequences for two additional Buchnera genomes to identify stretches of intergenic DNA that have been conserved for more than 65 million years . Many of these regions contain identifiable elements involved in the regulation of transcription , and several are predicted to encode small non-coding RNAs , riboswitches , and new regulatory binding sites . Despite the continuous inactivation and erosion of sequences , and the loss of most ancestral regulatory mechanisms , a large portion of intergenic spacers in Buchnera maintains functional sequence elements .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genomics",
"genome",
"evolution",
"microbial",
"evolution",
"genetics",
"and",
"genomics",
"biology",
"evolutionary",
"biology",
"genomic",
"evolution",
"comparative",
"genomics",
"evolutionary",
"genetics",
"microbiology"
] |
2011
|
Sequence Conservation and Functional Constraint on Intergenic Spacers in Reduced Genomes of the Obligate Symbiont Buchnera
|
People make numerous decisions every day including perceptual decisions such as walking through a crowd , decisions over primary rewards such as what to eat , and social decisions that require balancing own and others’ benefits . The unifying principles behind choices in various domains are , however , still not well understood . Mathematical models that describe choice behavior in specific contexts have provided important insights into the computations that may underlie decision making in the brain . However , a critical and largely unanswered question is whether these models generalize from one choice context to another . Here we show that a model adapted from the perceptual decision-making domain and estimated on choices over food rewards accurately predicts choices and reaction times in four independent sets of subjects making social decisions . The robustness of the model across domains provides behavioral evidence for a common decision-making process in perceptual , primary reward , and social decision making .
The most basic tenet in standard economic theory is that individuals know their preferences , and those preferences in turn determine their choices . In this approach , preference and choice are one and the same thing . However , this fundamental assumption is at odds with an extensive experimental literature documenting noisy and inconsistent choices , sometimes referred to as preference reversals [1–4] . Based on these experimental findings , theorists have developed random utility maximization models and experimentalists routinely apply probabilistic choice models to fit their data , often without considering the source of the noise in the decisions [5–7] . One prominent conjecture is that individuals do not know their preferences , but instead must construct their preferences at the time of choice . The promise of neuroeconomics has been to use knowledge from neuroscience and psychology to improve our models of economic decision making [8–19] . One obvious direction for improvement is to develop dynamical models that can jointly predict choices and reaction times ( RT ) . RTs are routinely ignored in economics experiments , presumably because they are irrelevant in the static models that dominate the field . However , we know from the literature on sequential sampling models ( SSM ) that RTs can be used to more accurately predict choice behavior . So there is hope that dynamical models will prove to be a useful alternative to their static counterparts when RT data or other process measures are available . Another direction for improvement is to develop models that can be applied in different domains . Currently we have separate models for risk aversion , temporal discounting , social preference , etc . Because these models take different inputs and result in arbitrarily scaled utilities , there has been little attempt to unify them . However , recent findings from neuroeconomics indicate that value coding in the brain is subject to adaptive coding . This suggests that the range of inputs to the decision mechanism in one task may be very similar to that in a different task , regardless of the actual range of values being considered . For example , an item with a subjective value rating of 5 out of 10 may be encoded at the same level as a $50 reward in a $0–100 experiment or 10 points in a 0–20 point experiment . Thus it may be possible to predict choice probabilities and RTs across tasks if the range of values is accounted for and exogenous factors like time pressure are not too different . The model we use to explore these possibilities is a SSM closely related to the drift-diffusion model ( DDM ) , which has received a great deal of attention for its ability to fit choices , RTs , eye-movements , and measures of neural activity during certain types of decision making [10 , 12 , 15 , 16 , 20–30] . While most applications of SSMs have been in the perceptual decision-making domain , they have also been applied to some value-based tasks [15 , 16 , 22–25 , 28 , 31] . Here we build off of this work and explore ( a ) whether a SSM can capture social decisions , and ( b ) whether the parameters fit to a binary food choice task can help predict behavior in four separate social-decision tasks . It is important at the outset to address some unique aspects of this work . First , one might assume that individual variability must surely make our attempt to explain out-of-sample choice behavior an exercise in futility . While it is likely true that individual subjects will vary in terms of their parameters , those variations will average out in sufficiently large datasets . Many applications of behavioral models are concerned with aggregate behavior and that is our focus here as well . Accounting for individual variability in behavior is certainly a promising direction , but it is not the aim of this paper . Second , as alluded to above , we do not wish to claim that there are universal parameters that will precisely predict data in every new dataset . Rather we think of these as “baseline” parameters meant to capture the fact that , in the aggregate , experimental subjects will tend to put in similar levels of effort to make quick and accurate decisions , regardless of the task in front of them . Factors like time pressure , performance improvement or degradation over time , and experimental manipulations will certainly alter these parameters , but hopefully in predictable ways . Investigating such alterations is also separate from the aims of this paper . Instead , a primary goal here is to address the criticism that SSMs involve too many free parameters and so could conceivably fit nearly any set of choice and RT data ( [32] , but see [20 , 33] ) . We address this criticism by estimating the parameters in one dataset and then assessing how well the model performs with these parameters in other datasets . Thus , this is not a demonstration of how well we can fit the model to any new dataset , but rather how well the model can perform out-of-sample with minimal or no fitting .
Task 1 provides behavioral data from a task using the Dictator game where 30 subjects in the role of the dictator made 70 binary decisions between two allocations of money , each one specifying an amount for the dictator and an amount for the receiver ( Fig 1 ) . For each choice there was a “selfish” option and a “fairer” option . Compared to the fairer option , the selfish option gave more money to the dictator and less money to the receiver ( see Methods ) . Task 2 is a behavioral task consisting of 69 subjects who faced 100 different binary choice problems . These decisions again involved choosing between two allocations of money , but this time there was , apart from the dictator and the receiver , a third subject involved ( a “co-dictator” ) who always received the same amount as the dictator . As in Task 1 , the selfish option always resulted in more money for the dictators and less money for the receiver . An important feature of Task 2 is that while the game consists of three players , there are only two unique sets of payoffs for the dictator to consider . Existing models of social preference would predict that there should be more weight put on the dictator’s payoff compared to the standard Dictator game from Task 1 . For example , the widely-used Fehr-Schmidt model predicts that with the extra partner , the weight on the receiver’s payoff should decrease by half [39] . However , as detailed below , our model predicts no change from Task 1 . This is because the dictator and co-dictator’s payoffs are identical and we assume that consistently redundant information is ignored in the choice process . So as in Task 1 there are only four payoffs to consider , two for the dictators and two for the receiver . Task 3 ( n = 30 ) was identical to Task 2 , except that in Task 3 we removed the redundant column of information so that there were only four payoffs on the screen , two for the dictators and two for the receiver . We anticipated no difference in behavior between Tasks 2 and 3 . Task 4 involves the behavioral data from a previously published fMRI study on the Ultimatum game [36] . In the Ultimatum game , one subject ( the proposer ) offers some division of a fixed amount of money to another subject ( the responder ) . The responder can either accept the offer , or he can reject the offer , leaving both players with nothing . In this game , subjects often reject unfair offers , contrary to economic models assuming pure self-interest . Rejection behavior in these games directly implies that receivers of unfair offers have a negative valuation of the proposer’s payoff . In this study , 18 subjects were asked to respond to 16 Ultimatum offers , each from a total pie of 20 Swiss Francs ( ~$25 ) , and subjects received offers of 4 , 6 , 8 , and 10 Swiss Francs . In a SSM , there exists a latent variable , which we have termed the relative decision value ( RDV ) . Others have referred to this variable simply as “information” or “net evidence” . In our version of the model , the RDV begins each new decision with a value of zero . As the decision maker accumulates evidence , the RDV evolves until it reaches a threshold value of +/- b , at which point he/she chooses the corresponding option . For each step in time ( here 1 ms . ) , the RDV changes by an amount equal to the drift rate plus Gaussian white noise with standard deviation σ . Because the RDV has arbitrary units , we are always free to set the drift rate , the barrier height ( b ) , or the noise ( σ ) equal to 1 . To allow comparison with our previous results on food choice , we fix the decision barriers to one ( i . e . b = 1 ) . A critical question in applying SSMs to value-based decisions is how to handle the drift rate . The drift rate represents the average rate of net evidence accumulation for one option over the other . In a standard perceptual experiment , there would be several conditions with different levels of discriminability . A separate drift rate would be fit to each condition and they would reflect the strength of the perceptual evidence , i . e . the task difficulty . In value-based decision making , discriminability is determined not by perceptual qualities of the items , but by the strength of preference for one item over the other . In a typical experiment every decision problem is unique so there can be a continuum of preference levels . In principle one could try to bin certain trials together and fit separate drift rates to different strengths of preference ( indeed this is how we visualize the model fits ) . However , which trials “fit together” can depend on parameters of the model and so the modeler may be forced to make arbitrary decisions about how to bin the data . Moreover , this approach discards valuable information about how changes in preference lead to changes in drift rate . Our approach is to instead let drift rate vary continuously as a function of the strength of preference . This way we fit all of the data at once , using just one drift parameter d , which multiplies the underlying value difference between the two options . For example , a choice between a $1 item and a $2 item would have a drift rate of d , while a choice between a $1 item and a $5 item would have a drift rate of 4d . More concretely , our model can be written as follows: Vt=Vt−1+d ( x−y ) +εt where V is the RDV , x and y are the underlying values for option X and option Y respectively , and ε is Gaussian noise with mean zero and variance σ2 . V starts each trial at 0 and evolves over time until it reaches +/-1 . Recent food-choice experiments have used eye-tracking data to demonstrate that visual fixations play an important role in the evidence accumulation process [15 , 16] . These results led to the introduction of the attentional DDM ( aDDM ) where evidence is accumulated more quickly for an option when it is being looked at than when it is not . This effect is captured by a fixation-bias parameter θ that discounts the value of the unlooked-at option when calculating the current drift rate , and it leads to a substantial choice bias for options that happen to capture more looking-time over the course of a trial . The estimated bias parameter value in that study ( which has been validated in multi-option choice in [16] ) was θ = 0 . 3 , meaning that the values of the unlooked-at food items are discounted by a factor of roughly a third during comparison . Here we hypothesize that , much like the unlooked-at food items , other peoples’ payoffs receive less weight than one’s own payoffs during the choice process . When subjects make choices in an experiment that involves their own monetary payoff and another player’s monetary payoff , they have to compare these payoffs across alternatives to make a decision . For example , if subject i earns xi in alternative X and yi in alternative Y the subject needs to compare these values . Likewise , the subject needs to compare the monetary payoffs xj and yj for subject j in the two alternatives . We can parsimoniously capture these comparisons in an extended DDM as follows: Vt=Vt−1+ds[ ( xi−yi ) ±θs ( xj−yj ) ]+εt In this model ds denotes the drift-rate multiplier and θs measures the subject’s discount rate for the other player’s payoff . This aspect of the model is inspired by Decision Field Theory , where attentional weights dictate the influence of different attributes on the decision [23 , 40] . Note , that the discounted payoff difference θs ( xj−yj ) for the other player may enter the relative decision value either positively or negatively depending on how the subject values the other player’s payoff . In principle , it is possible to estimate the drift rate ds , the standard deviation of the error term σs , and the discount rate θs for each of our four social-preference experiments . Instead , we would like to pose a different question . Using parameters fit to one task ( a food-choice experiment ) , how well can the aDDM capture choices and RTs in separate , random groups of subjects in social-decision experiments ? Because the social-decision datasets do not include eye-tracking data , we need to make a few extensions to the model . First , we assume ( based on [16] ) that on average subjects allocate equal attention to both choice options . Second , we assume that consistently redundant information is ignored in the choice process ( we test this assumption in Tasks 2 & 3 ) . Third , we incorporate adaptive coding [41 , 42] to account for the fact that each experiment has different units and ranges of value ( see Methods ) . Finally , we assign a weight of θ = 0 . 3 to the other subject’s payoff . There are a couple of motivations for this choice . Initially we hypothesized that subjects in these tasks would be focused on themselves and thus discount the other subjects’ payoffs in the same way that unattended items are discounted in the food-choice aDDM . We later confirmed this hypothesis with ex-post model fits of the θ parameter to each dataset . However , we cannot rule out that this may be a coincidence . In order to capture the stochastic nature of the aDDM we simulate the model 1000 times for each choice problem presented to the subjects and create 95% confidence intervals using bootstrapped samples equal in size to the actual datasets . We then compare the simulated choice and RT curves to the real data . Note that the inputs to these simulations are only the monetary values of the options X and Y for each trial ( e . g . Fig 1 ) . No data regarding the subjects’ choices or RTs in the social-decision tasks at the individual or group levels were used to generate predictions for these tasks . Figs 2–5 show the predicted choice and RT curves from the aDDM and the actual data . The model does a remarkable job of predicting both the shape and the absolute levels of the choice probability and RT curves in each of the four tasks , as evidenced by the consistent overlap between the 95% confidence intervals of the model and standard error bars of the data . In S1 Fig we also show that the model accurately predicts the entire RT distributions for each dataset [20] . In addition to the figures , the predictive accuracy of the aDDM can also be assessed in quantitative terms . To do so we calculate the mean error magnitudes between the model and the data ( Table 1 ) as well as traditional goodness-of-fit tests for all four tasks ( see S1 Table ) . In Task 1 ( Fig 2A-2B ) we see that the probability of choosing the selfish option generally tends to increase with the dictator’s own payoff gain ( compared to the other option ) , and to decrease with the receiver’s payoff loss from the dictator’s selfish option . In other words , while dictators are more likely to choose the selfish option if it earns them more money , they are less likely to choose an option that reduces the receiver’s earnings . Looking at the RTs , we see that the quickest decisions generally occur when the selfish option leads to the most money for the dictator and to the smallest loss for the receiver ( Fig 2C-2D and S2 Fig ) . This decrease in RT is about a third of a second or 12% compared to the slowest decisions . Furthermore , in a mixed-effects regression of log ( RT ) on |p-0 . 5| , where p is the mean group probability of choosing the selfish option , we find a highly significant effect ( p = 0 . 0005 ) . The average error magnitudes between the model and the data ( as a function of the dictator’s payoff gain and the receiver’s payoff loss , respectively ) are only 3 . 1% and 2 . 7% for the choice curves ( Fig 2A–2B ) and 5 . 6% and 6 . 1% ( of the highest mean RT ) for the RT curves ( Fig 2C and S2 Fig ) . In Task 2 we hypothesized that adding the co-dictator without adding additional payoffs to consider would not increase the dictator’s focus on that payoff and thus would not change the relative weight between the dictators and the receiver . In other words we predicted that the dictator would treat the decision problem as if there were only two players: the dictator and the receiver . Contrast this with an alternative scenario where the dictator’s and co-dictator’s payoffs are different; in that case the co-dictator’s payoff would receive a weight of 0 . 3 , just like the receiver’s payoff . Note that in the extreme , the model makes the same prediction for any number of co-dictator or receiver subjects , as long as there are not additional different payoffs to consider . While counterintuitive , this “scope insensitivity” effect has been well documented in the literature on contingent valuation [43] . As expected , in Task 2 ( Fig 3 ) the probability of choosing the selfish option steadily increases with the dictators’ payoff gain , while there is no linear trend as a function of the receiver’s payoff loss . In this task there was systematic variation in the size of the tradeoff between money for the dictator and money for the receiver . For example , trials with losses to the receiver of 10 and 30 had dictator gains of 5 and 10 , while trials with losses to the receiver of 20 and 40 had dictator gains of 17 and 23 . Thus the selfish option was more appealing in the latter trials and was more likely to be chosen , as seen in Fig 3B . So , what initially looks like random noise in Fig 3B is actually due to systematic variations in the dictator’s gains and is well-predicted by the model ( mean error magnitude: 5 . 6% for the dictator’s gain ( Fig 3A ) , 3 . 4% for receiver’s loss ( Fig 3B ) ) . Fig 3 also shows that RTs increase as both the dictators’ and the receiver’s payoff differences between the two options go to zero ( i . e . for the most similar options ) and that the model captures these trends well ( mean error magnitude: 3 . 1% for the dictator’s payoff gain ( Fig 3C ) , 5 . 2% for receiver’s loss ( Fig 3D ) ) . The correspondence between predicted and empirically measured RT is also high when using the model to combine dictator and receiver payoffs into a single utility measure ( S2 Fig ) . In Fig 3 we also present simulations of the model with θ = 0 . 15 to demonstrate how the model misfits the data if we assume that the receiver’s payoff receives only half of the weight from the normal Dictator game ( Task 1 ) , as would be the case , for example , in the Fehr-Schmidt model [39] . Below , we present data on explicit tests of the Fehr-Schmidt model . As expected , behavior in Task 3 closely matched behavior in Task 2 . Overall subjects chose the selfish option on 64% of trials in Task 2 and 65% of trials in Task 3 ( t = 0 . 173 , p = 0 . 86 ) . At a more detailed level , we observed similar choice and RT curves ( Fig 4 and S2 Fig ) ( choice: mean error magnitude: 9 . 1% for the dictator’s gain ( Fig 4A ) , 5 . 6% for receiver’s loss ( Fig 4B ) , RT: mean error magnitude: 6 . 5% for the dictator’s gain ( Fig 4C ) , 8 . 1% for receiver’s loss ( Fig 4D ) ) . In Task 4 we observe the standard behavioral trend that subjects almost always accept “fair” or “almost fair” offers of 40–50% of the total pie , but start to reject lower offers ( Fig 5 ) , with an average indifference point in the 20–30% offer range . Again , the model provides accurate predictions for both choices ( mean error magnitude: 5 . 7% ) and RT ( mean error magnitude: 6 . 3% ) .
Taken together , our results show that social decisions display relationships between choice probability and RT that are consistent with a dynamical decision process and that these relationships can be quantitatively captured with a sequential sampling model . Additionally , we demonstrated that these four particular datasets were accurately predicted by parameters from a previous food-choice model . These results are consistent with the notion that there is a common computational framework by which value-based decisions are made . More broadly , the established success of SSMs in describing memory and perception suggests that these models are capturing fundamental aspects of decision making . Namely , a decision-maker evaluates options at the time of choice by accumulating evidence for them until one of the options is judged as being sufficiently better than the other ( s ) . In perceptual decision making the stochastic evidence is primarily exogenous while in value-based decision-making the stochastic evidence is generated internally . Nevertheless , similar evidence-comparison processes seem to apply in these different cases . It is important to highlight that the particular model we have used here is simplified in the sense that it does not include additional parameters for variability in drift rate or starting point across trials or subjects . One strength of SSMs is that they predict not just mean RT , but entire RT distributions . While our model captures these distributions reasonably well , additional parameters are likely required to fully capture every facet of the data . While our model may sacrifice some amount of precision in these distributional analyses , we believe this is outweighed by the benefits of simplicity and robustness across different choice environments . The benefits of this robustness are apparent from our ability to predict behavior across the different datasets . Of course , factors like time pressure , performance improvement or degradation over time , and experimental manipulations will likely alter the parameters of the model . It also remains an open question whether these same parameters will predict behavior outside of the lab where focus and its discounting effects on choice can be affected by many uncontrolled factors such as distraction and advertising . For instance , we know already from a large body of work that putting subjects under time pressure causes faster and less accurate decisions , reflected by a change in parameters of the SSM ( e . g . [22 , 46 , 47] ) . This means that if one cannot directly control for time pressure in the field environment , the parameters measured in the laboratory may not predict field behavior accurately . Studying how environmental/experimental factors influence these parameters is an important line of investigation to see how far we can stretch these models . It is also important to highlight that the particular model we have used here is just one of the many SSMs in the literature , including the drift-diffusion model , linear ballistic accumulator model , leaky competing accumulator model , Ornstein-Uhlenbeck model , biophysical attractor models , and urgency-gating models [12 , 27 , 45 , 46 , 48–50] . While there is a good deal of interest and work being done to distinguish between these models , many of them are identical under certain parameter restrictions and otherwise tend to yield only subtle differences in RT distributions [45 , 51 , 52] . In other cases , the data themselves must have specific characteristics to allow comparisons between different SSMs . For example , it has been reported that the urgency-gating model is identical to the drift-diffusion model in situations with constant evidence , as in the datasets we test here [50] . Thus it is impossible to distinguish between the urgency-gating model and the aDDM in this setting . However , it is worth noting that the urgency-gating model would presumably not predict the effects of attention on the weighting of self and other payoff , and thus would be equivalent to the aDDM with θ = 1 . As can be seen from Fig 6 , that model would clearly fail to capture our subjects’ behavior . In principle it is possible to test specific features of different SSMs against each other in value-based tasks , but to do so would likely require considerably more data , as these sorts of model-fitting exercises are typically performed with ~1000 trials per subject , whereas here our tasks ranged from 16 to 100 choices per subject , spread over a continuum of “conditions” ( i . e . subjective-value differences ) . Even with smaller datasets like the ones reported here , the close relationship between choice probabilities and RT provides additional possibilities to test and constrain the models . We tested several alternative models , both in terms of the subjective value function and in terms of the SSM framework . Specifically we compared our aDDM with Fehr-Schmidt and Bolton-Ockenfels social-preference models and an Ornstein-Uhlenbeck SSM [39 , 44 , 46] . While each of these models adequately captured some aspects of some tasks , none were able to match the predictions of the aDDM across tasks and measures ( choice probability and RT ) . Furthermore , the Fehr-Schmidt and Bolton-Ockenfels models do not predict RT at all without inserting them into a SSM such as our aDDM . RTs have been woefully understudied in economics , potentially because of a perceived lack of formal models that make use of them . SSMs provide a clear and precise link between RT and the choice problem . Longer RTs are a direct result of smaller differences in subjective value between choice options , with the longest RTs coinciding with a subject’s indifference point ( see S2 Fig ) . This follows from the fact that smaller value differences lead to slower net evidence accumulation and consequently a higher latency to cross a choice threshold . The measurement of indifference points is the basis for inferring preferences and our model implies that using RTs can improve the estimation of preferences , which should be of great interest to the field . Furthermore , the fact that people consistently spend the most time on decisions that matter the least , i . e . where the subjective value difference between the choice options is smallest , has important normative implications for economics [15 , 22 , 25 , 28 , 31 , 53] ( but see [54 , 55]for a more nuanced discussion of this issue ) . The proposed role of focusing effects in social decisions suggests novel interventions , namely the possibility of changing subjects’ behaviorally revealed social preferences by guiding their focus to other players’ payoffs . Such experimental manipulations have been shown to alter choice patterns ( e . g . [56] ) . This possibility stands in sharp contrast to the orthodox view in economics that takes preferences as given and unchangeable [57] . It also stands in contrast to the prevailing social preference models in economics ( e . g . [39 , 44 , 58 , 59] ) that neglect the focusing effects underlying other-regarding behaviors . There is indeed evidence suggesting that fixations to other subjects’ payoffs is predictive of other-regarding behavior [60] , which provides support for our approach . Interestingly , these results are also consistent with findings from individuals with autism and amygdala damage [61 , 62] , who are impaired both in making eye contact and in social behavior . Based on these findings , it has been argued that making eye contact is a way of directing focus to social , rather than selfish outcomes . While manipulating their attention has been successful in improving the social behavior of autistic subjects , it has yet to be shown that such manipulations can alter social preferences in normal subjects . However , manipulating social distance through the wording of the instructions or identity of the subjects has been shown to change social behavior , possibly by focusing subjects’ more on others’ outcomes [63 , 64] . The SSM framework , as a general model of decision making , provides a means of quantifying the effects of such manipulations on choice behavior .
Subjects gave informed written consent before participating in each of the four studies . All studies were approved by the local ethics committee ( Zurich , Switzerland; KEK-ZH 2010–0327 ) . There are four parameters that are standard in all DDMs: the standard deviation of the Gaussian noise ( σ ) , the drift parameter ( d ) , the decision barrier height , and a non-decision time . Because the relative decision value has arbitrary units , we are always free to set one of the first three parameters in the DDM equal to 1 . Following the convention from our previous papers , we fix the decision barriers at +/- 1 . The first parameter ( σ ) dictates the amount of noise added to the average drift rate at each millisecond time step , and its value is taken directly from the food-choice experiments ( σ = 0 . 02 ) . The second parameter is the drift parameter d that multiplies the value difference between the options , in order to determine the average drift rate . Because our formulation of the DDM has fixed decision barriers at +/- 1 , the d parameter must be adjusted to the range of values in order to ensure the equivalence of the model across experiments with different units and/or ranges of value . For example , if in a certain experiment the payoffs are expressed in dollars rather than in cents , the nominal payoff range in the experiment with the dollar representation is smaller by a factor of 100 although the real payoffs are identical . It is therefore necessary to adjust the drift parameter d by a factor of 100 because otherwise the model would make wildly different predictions for two identical decisions . Similarly , a dollar difference in value has very different consequences if we are choosing between snack foods or luxury cars [41 , 42] . Thus , if different experiments involve different payoff ranges there is the need to adjust the drift parameter such that the product of the drift parameter d and the payoff range v¯ remains constant across experiments . Note , that because d⋅v¯ is constant across all experiments this does not introduce any degree of freedom in choosing the value of d in the social-preference experiments examined here . In the social-preference experiments the subjects made choices sequentially and thus could not know the precise range of values that they would face . However , subjects could immediately see the order of magnitude of their payoffs , so we set the value of d such that the product of d and the order of magnitude of v¯ is always equal to 0 . 002 ms-1 , the value from the food choice studies . See S2 Table for details . From the binary food-choice study we know that fixation durations do not depend on the values of the options so we can assume that , on average , each option is focused on for half of the trial . Therefore , each option ( e . g . option X ) accumulates evidence at a rate of d·x for half of the trial , and 0 . 3·d·x for the other half of the trial , meaning that the average evidence accumulation rate for each option is [0 . 5·d·x + 0 . 5·0 . 3·d·x] = 0 . 65·d·x . In the context of the food-choice model this means that the average relative drift rate is equal to 0 . 65·d· ( x−y ) . Again , we apply the behavioral insights from the food-choice study ( i . e . that fixations do not depend on the values of the options ) to the social-preference studies by assuming that each option is fixated on for half of the trial . Therefore , the final parameterized social DDM can be written as: Vt=Vt−1+0 . 65d[ ( xi−yi ) ±0 . 3 ( xj−yj ) ]+ε where x and y are the amounts received by the decision maker ( subscript i ) and the other player ( subscript j ) for options X and Y . As before , the RDV evolves over time ( in increments of 1 ms ) until it reaches a value of +1 or -1 . Note that in this extended version of the model , the factor of 0 . 3 appears twice , once to maintain the discount on the non-fixated choice options , and a second time to maintain the asymmetry between self and other payoffs . Another way to implement the model would have been to omit the asymmetry between self and other for the non-fixated option , but this alternative model provided worse fits to every dataset . In our previous work [15] the non-decision time was not fit to the data , but instead directly measured from the food-choice data by calculating the average difference between RT and total fixation time . This difference was 355ms . In our simulations of the model we add this “non-decision time” to all the simulated RTs . To simulate each model we used the given utility function to calculate the subjective value for each trial’s choice options ( X and Y ) , and then used the difference between those utilities to determine the model’s drift rate . The model can be written as: Vt=Vt−1+d ( U ( X ) −U ( Y ) ) +εt The remaining parameters in the model ( d , σ and non-decision time ) remained the same as in the aDDM . To test alternative SSM implementations we utilized the Ornstein-Uhlenbeck model . For a given choice problem we assume that the input to the model is the difference in subjective values between options X and Y , denoted I = U ( X ) −U ( Y ) , where Ui ( X ) = xi ± 0 . 3xj as in the main aDDM . Then the Ornstein-Uhlenbeck model can be written as the following differential equation: dV= ( λV+kI ) dt+σdW where V is the amount of accumulated evidence at time t , dW is independent white noise ( Wiener ) , and λ , k , σ are free parameters . The process proceeds in steps of dt = 1ms . The parameters used to simulate the model [46] are: λ = 2 . 3 , k = 0 . 11ms−1 , σ = 0 . 6 and a non-decision time of 400ms .
|
One critical question that concerns all disciplines involved in the study of human decision-making is whether different types of decisions are made in different ways , or whether there exists a common decision mechanism that underlies human choices . If the latter , what are the properties of that mechanism ? Here we characterize a dynamical model of decision making that was initially fit to subjects making food choices but was later able to accurately predict choices and reaction times of separate groups of subjects making social decisions . The robustness of the model across different subjects , tasks , and environments supports the idea that the brain uses a consistent process for making decisions .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[] |
2015
|
A Common Mechanism Underlying Food Choice and Social Decisions
|
Successful treatment of aspergillosis caused by Aspergillus fumigatus is threatened by an increasing incidence of drug resistance . This situation is further complicated by the finding that strains resistant to azoles , the major antifungal drugs for aspergillosis , have been widely disseminated across the globe . To elucidate mechanisms underlying azole resistance , we identified a novel transcription factor that is required for normal azole resistance in Aspergillus fungi including A . fumigatus , Aspergillus oryzae , and Aspergillus nidulans . This fungal-specific Zn2-Cys6 type transcription factor AtrR was found to regulate expression of the genes related to ergosterol biosynthesis , including cyp51A that encodes a target protein of azoles . The atrR deletion mutant showed impaired growth under hypoxic conditions and attenuation of virulence in murine infection model for aspergillosis . These results were similar to the phenotypes for a mutant strain lacking SrbA that is also a direct regulator for the cyp51A gene . Notably , AtrR was responsible for the expression of cdr1B that encodes an ABC transporter related to azole resistance , whereas SrbA was not involved in the regulation . Chromatin immunoprecipitation assays indicated that AtrR directly bound both the cyp51A and cdr1B promoters . In the clinically isolated itraconazole resistant strain that harbors a mutant Cyp51A ( G54E ) , deletion of the atrR gene resulted in a hypersensitivity to the azole drugs . Together , our results revealed that AtrR plays a pivotal role in a novel azole resistance mechanism by co-regulating the drug target ( Cyp51A ) and putative drug efflux pump ( Cdr1B ) .
Aspergillosis is one of the most common infectious diseases by mold , with Aspergillus fumigatus being the most frequently causative fungus with high mortality ( more than 60% ) . Despite the importance of this disease , approved antifungals for treatment of aspergillosis are quite limited . Together with difficulties in early diagnosis , today aspergillosis is a major threat especially for immunocompromised patients . Further worsening the outlook for successful antifungal therapy , azole resistant A . fumigatus strains have been found with increasing incidence worldwide [1 , 2] . Azoles are an essential antifungal drug for therapy for chronic aspergillosis . In fact , azole resistant A . fumigatus strains exhibit a poor clinical outcome [3] . Therefore it is an urgent issue to unravel the molecular mechanisms underlying azole resistance and to develop new antifungal drugs or agents that are able to reverse azole resistance . Mutations in the cyp51A gene locus have been frequently reported in azole resistant A . fumigatus strains . This locus encodes lanosterol-α-14-demethylase , the target enzyme of azole drugs [1 , 3] . Inhibition of the Cyp51A enzyme results in a significant change in sterol profile in A . fumigatus cells , where ergosterol deficiency as well as accumulation of toxic intermediates is thought to cause the antifungal effect [4 , 5] . Several reports have revealed that G54 and M220 are hot spots for azole resistance mutations in the Cyp51A protein [6] . Resistant mutations tend to arise during prolonged treatment of chronic aspergillosis with azole drugs [3 , 7 , 8] . In addition to resistance acquired during therapy , environmentally-derived resistance mutations have emerged as a major source of resistant organisms . Azole resistant strains with a combination of a tandem repeat in the promoter region of cyp51A and amino acid mutation ( s ) ( TR34/L98H and TR46/Y121F/T289A ) have been isolated from patients regardless of azole therapy history [9–11] . This pre-acquired azole resistant strain limits the alternatives for effective drug therapy of aspergillosis . Recently emerging mutations were described worldwide and range across isolates from Europe [10–13] , Asia [14–16] , and North America [17] . To date , one surveillance report provided data that azole resistant strains were up to 38% of the isolates from aspergillosis patients in the Netherlands , about 90% of which are related to Cyp51A [18] . Apart from alterations in cyp51A , a mutation in hapE [19] and overexpression of cdr1B [20] or cyp51B [21] were reported to be responsible for azole resistance in a clinical setting . Importantly , cdr1B encodes an ATP-binding cassette ( ABC ) transporter sharing high sequence similarity with Saccharomyces cerevisiae Pdr15 and Pdr5 , Candida albicans Cdr4 and Cdr1 , and Candida glabrata Cdr1 and Pdh1 . This shared homology is suggestive of a role of Cdr1B in drug efflux . The fact that deletion of the cdr1B resulted in azole sensitive phenotype further supported the important role in azole resistance in A . fumigatus [20 , 22] . The transcription factors involved in azole resistance in fungi have been intensively investigated in S . cerevisiae . The well-studied Pdr1 and its paralog , Pdr3 , are zinc finger transcription factors that regulate the pleiotropic drug response in the yeast [23] . Pdr1 and Pdr3 serve as transcriptional activators and repressors by binding to pleiotropic drug response elements ( PDREs ) that are found in the promoter regions of target genes [24] . The targets include the ABC transporters encoded by PDR5 , PDR10 , and PDR15 [25] . In filamentous fungi including plant and animal pathogens , only one regulatory protein involved in azole resistance has been studied . This protein was designated SrbA and encodes a homologue of the Sterol Regulatory Element-Binding Proteins ( SREBPs ) . The SREBP is a transcription factor with a basic Helix-Loop-Helix ( bHLH ) DNA-binding domain and is well conserved in mammalian cells , where it regulates cholesterol synthesis and uptake [26] . Roles of the A . fumigatus SREBP , SrbA , were recently demonstrated [27 , 28] . SrbA was shown to regulate expression of multiple genes related to ergosterol biosynthesis pathway , including erg3 , erg24A , and erg25A . A mutant strain defective in srbA showed hypersensitivity to azoles . Further experiments revealed that SrbA exerts a direct regulation on these genes , suggesting that this factor is a central regulator of ergosterol biosynthesis in the fungus [29 , 30] . Importantly , the srbA mutant was unable to grow under hypoxic condition . This is because SREBPs sense sterol level in the cells as an indirect indicator of oxygen . The srbA mutant was unable to adapt to hypoxic conditions that the fungi are thought to encounter at infection sites . This inability of the mutant to grow under hypoxia is also thought to lead to a strong attenuation in virulence in a mouse infection model . Whereas Candida species have no obvious SREBP orthologs , Cryptococcus neoformans has a SREBP , Sre1 , which is also involved in azole sensitivity , hypoxia growth , and virulence [31 , 32] . These findings highlighted that ergosterol biosynthesis is crucial for not only drug resistance but also for adaptation to hypoxia and virulence . Besides SREBPs , however , current knowledge about the transcription factors responsible for azole resistance and sterol biosynthesis is limited particularly in pathogenic fungi . In the present study , we identified a novel transcription factor involved in azole resistance mechanism in Aspergillus fungi . From a genetic screen in Aspergillus oryzae designed to identify transcriptional regulators of azole drug resistance , we found a novel Gal4-type Zn2-Cys6 zinc finger domain-containing transcription factor designated AtrR . Deletion of the atrR gene in A . oryzae and Aspergillus nidulans resulted in hypersensitivity to azole drugs . Similarly , deletion of the atrR gene in A . fumigatus resulted in hypersensitivity to azole antifungals , inability to grow under hypoxic conditions , attenuated virulence , and decreased expression of cyp51A as well as the ABC transporter-encoding gene cdr1B . Even in a clinical azole resistant strain ( cyp51A G54E ) , deletion of atrR led to an azole-hypersensitive phenotype . All our data indicate that AtrR plays an essential role in azole resistance in A . fumigatus . Importantly , this is the first identification of a regulatory protein coordinating expression of not only an azole drug target ( Cyp51A ) but also for the putative drug efflux transporter ( Cdr1B ) .
To identify the transcription factors ( TFs ) responsible for azole resistance in filamentous fungi , we utilized a transcription factor-overexpressing library of Aspergillus oryzae that was previously constructed in the Noda Institute for Scientific Research [33] First , we searched for A . oryzae transcription factors with a DNA-binding domain related to S . cerevisiae Pdr1 and Pdr3 , resulting in 5 candidate proteins ( designated TF1 to TF5 ) ( S1A Fig ) . To evaluate a role of these TFs in azole resistance , strains overexpressing each TF were tested for growth in the presence of azoles . Only A . oryzae strain expressing TF3 showed a significant hyper-resistance to clotrimazole ( Fig 1A ) . TF3 corresponds to AO090026000614 which is a protein with 894 amino acids containing two characteristic motifs designated as GAL4-like Zn ( II ) 2/Cys6 binuclear cluster DNA-binding domains ( IPR001138 ) and a fungal specific transcription factor domain ( IPR007219 ) ( S1B Fig ) . Based on the screening design to find transcription factors homologous to ScPdr1 and/or ScPdr3 that regulate ABC transporter gene expression , we named this novel TF AoAtrR ( A . oryzae ABC-transporter regulating transcription factor ) . To further characterize the role of AoAtrR , the corresponding gene was deleted in A . oryzae host strain ( RIB40 ) ( S2A and S2B Fig ) , and azole resistance was examined . The AoatrR mutant showed hyper-sensitivity in the presence of 0 . 01 μg/mL miconazole , whereas the host strain was able to grow under this condition ( Fig 1B ) . These results indicated that AoAtrR is involved in azole resistance in A . oryzae . The closely related species Aspergillus nidulans has an orthologous protein ( designated AnAtrR; amino acids sequence identity: 79 . 9% ) ( S1B Fig ) , and the AnatrR deletion mutant strain ( S2C and S2D Fig ) also clearly exhibited azole sensitive phenotype ( Fig 1C ) . As a result , the AtrR-family protein appears to be a novel TF responsible for azole resistance mechanisms in Aspergillus species . This AtrR-family TF is widely conserved in filamentous fungi including important plant pathogenic fungi ( Fusarium graminearum , Magnaporthe oryzae , and Collectotrichum orbiculare ) and human pathogenic fungi ( Aspergillus fumigatus , Ajellomyces dermatitidis , and Coccidioides immitis ) ( Fig 1D ) . To expand our understanding of molecular mechanism underlying azole resistance with clinically important fungi , we searched for an AtrR homologue in the human pathogenic fungus A . fumigatus . The AfAtrR in A . fumigatus ( hereafter , designated AtrR to simplify descriptions ) is homologous to AoAtrR and AnAtrR , and the deletion mutant and the complementing strains ( designated ΔatrR and Co-atrR ) were constructed in the A . fumigatus host strain ( Af293 ) ( S3 Fig ) . A disc-diffusion assay revealed that ΔatrR was hypersensitive to fluconazole and miconazole , whereas the mutant showed no distinguishable sensitivity to amphotericin B and micafungin ( Fig 2A ) . In addition to the medical azoles , ΔatrR showed hypersensitivity to widely used azole fungicides , bromuconazole , tebuconazole , difenoconazole , and propiconazole ( S4 Fig ) . Growth inhibition assay with varied concentrations of drugs in plate media indicated that ΔatrR was highly susceptible to itraconazole and miconazole ( Fig 2B ) . The complemented strain showed comparable drug sensitivity to the WT , indicating that these phenotypes were caused by deletion of the atrR gene . We assumed that hypersensitivity to azoles in ΔatrR was associated with a defect in expression of the target protein Cyp51A . In fact , the expression level of cyp51A gene was quite low in the ΔatrR strain ( less than 5% of that in the WT ) , whereas cyp51B expression was also reduced ( approximately 50% of the WT ) ( Fig 3A ) . Since SrbA was previously characterized to regulate cyp51A [29] , we also constructed a null mutant of srbA and compared the two mutants ( ΔatrR and ΔsrbA ) for their relative effects on cyp51A and cyp51B expression levels . We found that the expression levels of cyp51A and cyp51B were largely comparable between ΔsrbA and ΔatrR ( Fig 3A ) . Although not large , there was a slight increase in expression level for srbA in ΔatrR mutant and a slight decrease in atrR expression in ΔsrbA , suggesting effects of AtrR and SrbA on each other’s mRNA expression ( Fig 3A ) . To determine if the heterologous expression of cyp51A was sufficient to restore azole resistance in mutants lacking either atrR or srbA , we introduced an ectopic cyp51A gene regulated by the thiA promoter ( PthiA-cyp51A ) into the ΔatrR and ΔsrbA backgrounds . When the thiA promoter was maximally active ( absence of thiamine ) , the resulting expression level of Cyp51A was able to partially suppress loss of either transcription factor gene in both strains . ΔatrR+PthiA-cyp51A and ΔsrbA+PthiA-cyp51A showed smaller inhibition halos against miconazole compared to those under the repressed condition ( presence of thiamine ) ( Fig 3B ) . This suggested that azole susceptibility in ΔatrR and ΔsrbA are partly a result of decreased cyp51A expression . Hypersensitivities to fluconazole , miconazole , and itraconazole were also seen in the ΔsrbA strain ( Fig 2A and 2B ) . The ΔatrR strain was slightly more sensitive to fluconazole and itraconazole compared to a mutant lacking the srbA gene . However , both of these factors are required for normal azole resistance in A . fumigatus . In order to confirm the relative role of these genes across strains , these mutations were produced in a different genetic background ( Afs35 ) . The double ΔatrR ΔsrbA mutation was also produced in this same strain . The expression level of cyp51A was comparable between the ΔatrR , ΔsrbA , as well as ΔatrR ΔsrbA , supporting the required function of both AtrR and SrbA ( Fig 3C ) . To globally identify the genes responsive to AtrR , we performed transcriptome analysis using RNA-sequencing . Compared to the WT , 13 and 53 genes were up-regulated and down-regulated , respectively , in the ΔatrR strain , showing AtrR-dependent genes . We also determined SrbA-dependent genes in a similar fashion , resulting in 21 up-regulated genes and 51 down-regulated genes . Comparing each group of up- and down-regulated genes between the ΔatrR and ΔsrbA strains , we found that 9 genes were commonly negatively regulated by AtrR and SrbA whereas 18 genes were positively regulated ( Fig 4A , Table 1 , S1 and S2 Tables ) . This comparative transcriptome analysis revealed that AtrR and SrbA were required for the expression of erg3B , erg24A , and erg25A as well as cyp51A genes , all of which are related to ergosterol biosynthesis . Interestingly , the other genes related to ergosterol biosynthesis were only modestly affected by deletion of either atrR or srbA genes ( Fig 4B ) . Involvement of AtrR and SrbA in the transcriptional regulation of the erg3B , erg24A , erg25A , and cyp51A genes was further verified by real-time RT PCR analysis ( Fig 4C ) . As a previous report demonstrated that SrbA regulates genes related to the ergosterol biosynthesis pathway [30] , our data indicate that AtrR also functions in regulating this pathway . As both AtrR and SrbA have a role in transcriptional regulation of the ergosterol biosynthesis pathway , we wanted to assess if AtrR also contributed to hypoxic growth and virulence as previously shown for SrbA [27 , 28] . We tested growth under hypoxic conditions and found that growth of both ΔatrR and ΔsrbA mutants , constructed in either the Af293 or Afs35 genetic backgrounds , was strongly inhibited under this condition ( Fig 5A and 5B ) . These results indicate a crucial role for AtrR in adaptation to hypoxia . To confirm the participation of AtrR in the hypoxic response , the transcriptional regulation of erg3B , erg24A , erg25A , and cyp51A upon oxygen limitation ( 1 h ) was compared in ΔatrR and ΔsrbA strains . The induction of these genes upon oxygen limitation was diminished or eliminated in the both mutants ( Fig 5C ) . These results indicated that AtrR also plays an important role in hypoxia adaptation in A . fumigatus physiology as found for SrbA . We also examined the role of AtrR in pathogenesis using a murine infection model . In this model for aspergillosis , virulence of the ΔatrR strain was significantly reduced compared to the WT and Co-atrR strains ( Fig 6A ) . The reduction in virulence was also seen in the Afs35-background ΔatrR as well as ΔsrbA and ΔatrR ΔsrbA ( Fig 6B ) . These results indicated that AtrR was required for full pathogenicity of A . fumigatus . This is again reminiscent of the role of SrbA in virulence , where ΔsrbA showed significantly attenuated virulence in immune-compromised mice [27] . To determine the role of AtrR in drug-induced gene expression in the A . fumigatus azole response , we performed further RNA-sequencing analysis following azole challenges . WT and ΔatrR were cultivated in PDB for 20 h , and then fluconazole ( final concentration , 64 μg/mL ) or miconazole ( 2 μg/mL ) were added . After 2 h , the mycelia were harvested , and the transcriptomes were determined by RNA-sequencing . Of 7015 genes that showed >10 reads per kilobase of exon per million mapped reads ( RPKM ) in WT without drug treatment , 301 genes were reduced to <33% in ΔatrR compared to the WT ( Fig 7A ) . Upon fluconazole and miconazole treatment , 56 and 25 genes were upregulated in WT , respectively . Among them , 19 genes were commonly responsive to fluconazole and miconazole , 6 genes of which were dependent on AtrR . The azole-responsive AtrR-dependent genes included cyp51A , erg24A , erg24B , erg25A as well as hyd1 and Afu3g00820 . These results again confirmed that AtrR regulated multiple genes related to the ergosterol biosynthesis pathway ( Table 2 ) . While atrR and srbA expression was slightly affected in response to fluconazole ( 1 . 22- and 1 . 62-fold , respectively ) and miconazole ( 1 . 27- and 1 . 37-fold ) treatment , real-time RT-PCR analysis suggested that the effect of azoles on the expressions were only subtle ( Fig 7B ) . By comparing amino acid sequences , A . fumigatus was found to possess 12 PDR-type ABC transporters . Among them , cdr1B expression was highest in WT according to the transcriptome data , whereas the expression level was reduced to less than 10% in the ΔatrR strain ( Table 2 ) . On the contrary , expressions of some transporter genes ( i . e . Afu3g01400 and AtrI ) were slightly higher in the ΔatrR compared to the WT , which suggests a negative role for AtrR . We then verified the expression profiles for the cdr1B by real-time RT PCR analysis , showing that the basal cdr1B expression was reduced in the ΔatrR strain ( approximately 25% relative to the WT ) , and there was no induced expression in ΔatrR ( Fig 8A ) . In a sharp contrast , the ΔsrbA strain showed WT-level basal expression and upregulation of cdr1B . This result clearly indicated that AtrR is responsible for cdr1B expression while SrbA is not involved in control of this transporter . Furthermore , we found that when treated with propiconazole , a representative of an azole fungicide , cdr1B expression was upregulated . This induction was also dependent on AtrR ( Fig 8B ) and supported a crucial role of AtrR in cdr1B expression in A . fumigatus . To examine the mode of AtrR interaction with the promoters of cyp51A and cdr1B , we carried out chromatin immunoprecipitation ( ChIP ) analysis using an epitope-tagged allele of atrR . A 3x hemagglutinin ( HA ) tag was placed at the carboxy-terminus of the chromosomal atrR gene ( S5A Fig ) . This allele was still able to confer azole resistance indicating that AtrR function was maintained ( S5B Fig ) . ChIP experiments were carried out as described [30] with immunoprecipitated chromatin examined for enrichment of fragments corresponding to the promoter regions of cyp51A , cdr1B and act1 . We found the strongest level of enrichment for the cdr1B promoter with easily detectable recovery of the cyp51A promoter ( Fig 9A ) . ChIP experiments using the act1 promoter did not produce any enrichment above background when comparing ChIP reactions carried out on strains lacking the HA tag versus a strain expressing the tagged AtrR protein . This specific enrichment for the cdr1B and cyp51A promoters was further confirmed by quantitative real-time PCR ( Fig 9B ) . These data are consistent with AtrR acting at sites upstream from both cdr1B and cyp51A . Specificity of this interaction is supported by the lack of enrichment of the act1 promoter region . To compare the physiological importance of the atrR , srbA , cdr1B , and cyp51A genes , we generated a series of isogenic strains lacking these important drug resistance loci in the Afs35 background . Representative isolates were then assessed for their drug resistance by determining their minimum inhibitory concentration ( MIC ) for a variety of antifungal drugs . The MICs of fluconazole , itraconazole , voriconazole , and miconazole were lowered in ΔatrR and ΔsrbA compared to the WT ( Afs35 ) ( Table 3 ) . The MICs of voriconazole and miconazole were more depressed in the ΔatrR strain than in the corresponding ΔsrbA mutant , which are consistent with the results of growth inhibition assay using the Af293 background strains ( Fig 2B ) . The double deletion mutant ( ΔatrR ΔsrbA ) exhibited increased susceptibility for azoles compared to either of single mutants . In the Δcdr1B and Δcyp51A , the MICs of itraconazole , voriconazole , and miconazole were lowered ( 50%-87 . 5% ) , compared to the WT . These drug susceptibility tests confirmed that the transcription factors AtrR and SrbA , the ABC transporter Cdr1B , and the target enzyme Cyp51A were required for normal azole resistance . Importantly , simultaneous loss of the transcription factors AtrR and SrbA caused the most profound azole sensitivity of all the genetic alterations tested . Considering the important role of AtrR in azole resistance , we wanted to evaluate the contribution of the atrR gene to azole resistance in a clinical isolate . We deleted the atrR gene from the genome of the clinical isolate IFM 61567 that contains a known itraconazole resistance-conferring mutation in its cyp51A gene ( Cyp51A G54E ) . We obtained three independent deletion mutant strains ( IFM 61567 ΔatrR No . 2-4 ) and compared the MICs of antifungals with the starting strain ( Table 3 ) . The parental strain IFM 61567 showed resistance to itraconazole and posaconazole , but retained susceptibility to voriconazole and miconazole . All the three ΔatrR deletant strains were highly susceptible to itraconazole ( MIC: 0 . 125~0 . 25 ) and posaconazole ( MIC: ≤0 . 06~0 . 06 ) , as well as to voriconazole ( MIC: 0 . 06 ) and miconazole ( MIC: 0 . 06 ) , which revealed these strains are not resistant to azoles , even if judged from a clinical endpoint . We also used a disc-diffusion assay to confirm azole susceptibility for these mutants upon fluconazole , miconazole , and itraconazole treatment ( Fig 10A ) . Together , these different assays demonstrated that the ΔatrR strains were more sensitive to these azoles compared to the parental resistant strain ( IFM 61567 ) . We confirmed that the expression levels of cdr1B , cyp51A , and cyp51B were markedly decreased in the ΔatrR derivatives ( Fig 10B ) , as was observed earlier for ΔatrR in Af293 background . Collectively , our results showed that the atrR gene deletion led to azole sensitization even in this clinically isolated multi-azole resistant strain with a G54E mutation in Cyp51A .
The ABC transporters in the PDR sub-family are one of the main players implicated in azole resistance in fungi [34] . It has been demonstrated in many fungi , including plant and human pathogens , that certain ABC transporter genes were up-regulated in response to azole treatment , and loss of genes encoding PDR-type ABC transporters resulted in hypersensitivity to azoles [35–37] . According to phylogenetic analysis , the A . fumigatus PDR sub-family ( also called ABCG family ) contains 12 ABC transporters [38] . Among these include the already characterized Cdr1B , AbcA , AtrF , and AtrI , deletion mutants of which showed decreased susceptibility to azoles [20 , 22 , 39 , 40] . Our RNA-seq analysis showed that only a few ABC transporter genes were upregulated upon miconazole and fluconazole treatment , and that Cdr1B seemed to be the only transporter regulated by AtrR ( Table 2 ) . Studies of Cdr1B revealed a clinical relevance as an azole resistant strain overexpressing cdr1B has been isolated from a patient [20] . In that report , genetic evidence was provided that azole resistance was caused by cdr1B overexpression . However , the mechanism of that overexpression was not assessed in the paper . It is possible that elevated activation via AtrR led to the increased cdr1B transcription . In a model filamentous fungus Neurospora crassa , Cdr4 , an ortholog of Cdr1B or S . cerevisiae Pdr5p , plays a major role in azole tolerance [41] . Although direct interaction with the promoter region was not investigated , two transcription factors were characterized to be involved in regulating the cdr4 expression . The putative transcription factor CCG-8 , which has been identified as a clock-controlled gene , positively regulates cdr4 as well as the azole target gene erg11 , an ortholog of A . fumigatus cyp51A [42] . Ketoconazole challenge induced cdr4 and erg11 expression in the WT , whereas the induction levels were partly decreased in the ccg-8 deletion mutant . The CCG-8 ortholog ( Afu5g09420 ) in A . fumigatus has not been characterized so far . A role of the bZip transcription factor ADS-4 was also investigated with regard to azole resistance in N . crassa [43] . A null mutant lacking ads-4 showed hypersensitivity to itraconazole and ketoconazole . ADS-4 was involved in regulating cdr4 as well as erg5 encoding a sterol C-22 desaturase . The ortholog of ads-4 ( Afu1g16460 ) was also characterized in A . fumigatus with its deletion strain , which showed a slightly higher sensitivity to azoles compared to the parental strain . Our RNA-seq data revealed that expression of the two transcription factors ( CCG-8 and ADS-4 ) was not affected by atrR deletion in A . fumigatus ( Table 2 ) , which supports the idea that AtrR is an independent regulator of cdr1B and cyp51A expression . The molecular mechanism of how AtrR functions will be an essential goal of future study . In the pathogenic yeasts , C . albicans and C . glabrata , the molecular mechanisms underlying transcriptional regulation of ABC transporters ( functioning as drug efflux pump ) have been more intensively studied . The azole resistance-related ABC transporter Cdr1 ( an ortholog of A . fumigatus Cdr1B and S . cerevisiae Pdr5p ) is directly regulated by a Zn2-Cys6 transcription factor Pdr1 in C . glabrata [44] . In a clinical setting , azole resistant strains with elevated CDR1 expression were commonly recovered . In most of the strains , CDR1 overexpression was attributed to constitutively activated Pdr1 that possessed a gain-of-function mutation [45 , 46] . As stated earlier , Candida has no ortholog of AtrR . Indeed , Candida Pdr1 and A . fumigatus AtrR only share sequence homology in a Zn2-Cys6 domain at the N-terminus . The most notable distinctive role between the ABC transporter regulating transcription factors , Pdr1 and AtrR , is that AtrR expression is not substantially upregulated in response to azole treatment ( Fig 7B ) , whereas Pdr1 itself is induced by azole treatment [47] . Collectively our results suggest that the transcriptional regulator of azole resistance-related ABC transporters , particularly for a Cdr1-type transporter , is evolutionarily distinct between yeasts and filamentous fungi . Ergosterol biosynthesis is a crucial process that is specific to fungi . Antifungal drugs such as azoles and polyenes target this biosynthetic pathway or the ergosterol molecule directly . The SREBP SrbA is a well-studied transcription factor that regulates ergosterol pathway-related genes including erg3 , erg24 , erg25 , and cyp51A [27 , 30] . Deletion of srbA resulted in a disturbed sterol profile and hypersusceptibility to azole drugs . Intriguingly , a deletion mutant of atrR revealed strikingly similar phenotypes to those of the srbA deletion mutant . In addition to the regulation of erg genes and azole susceptibility , hypoxia adaptation and virulence attenuation were phenocopied between the srbA and atrR null mutants . This striking overlap argues that AtrR and SrbA closely function to regulate the ergosterol biosynthesis pathway . Recent studies provided information on the contributors in the SREBP pathway including the Golgi Dsc E3 ligase complex members ( DscA-E ) , a rhomboid family protease RbdB ( also reported as RbdA ) , and an aspartyl peptidase SppA [48–51] . Although these contributors are essential for SrbA function , transcription levels of these genes including srbA were not largely affected by deleting AtrR ( Table 2 ) . This suggested that AtrR did not act upstream via the SREBP pathway at a transcriptional level but rather may have its own independent regulatory input . The possibility that AtrR interacts with SrbA to function as a transcriptional regulator complex over the genes related to ergosterol biosynthesis , thus far , cannot be ruled out . Previous studies by others revealed that SrbA binds to the promoter region of the cyp51A gene ( also known as erg11A ) [27 , 30] . ChIP-seq data indicated that SrbA-enrichment peaks resided in the cyp51A promoter region at the site 302 bp upstream from the initiation codon [30] . Accordingly , a putative SrbA binding site similar to the mammal SRE binding motif ( TCACNCCAC ) was found in this same region . Notably , the peak of SrbA-enrichment and the putative binding site are contained within a 34 bp sequence that had been identified as a tandem repeat sequence in azole resistant strains [11 , 13] . In our ChIP data for cyp51A with AtrR-3HA , a slightly more intense band was exhibited by probing with -69 to -415 region than with -395 to -697 ( Fig 9A ) . This suggested that the binding motif of AtrR in the cyp51A promoter is more likely to be present at the region from -69 to -415 rather than -395 to -697 . Gal4-type Zn2-Cys6 transcription factors typically bind to inverted CGG triplets as homodimers [52] . We therefore searched for CGG-Nx-CCG motifs in the cyp51A promoter region ( to -1000 ) . Interestingly , we found three regions with the motifs , CGG-N10-CCG , CGG-N18 ( 9 ) -CCG , CGG-N17 ( 12 ) -CCG , at -949 to -934 , -796 ( -787 ) to -773 , and -314 ( -309 ) to -292 , respectively . The last one is located in the 34 bp of tandem repeat-related sequence and thus is adjacent to the putative SrbA-binding site . From these data , we anticipate that AtrR and SrbA could cooperatively function to regulate cyp51A expression at the proximal sites in the promoter region . Additionally , three motifs ( CGG-N5-CCG , CGG-N21-CCG , and CGG-N20-CCG ) were found in the cdr1B promoter at the sites ( -723 to -713 , -537 to -511 , and -495 to -470 ) . For the other 51 AtrR-dependent genes identified in RNA-sequencing analysis , the potential CGG-Nx-CCG motifs were widely detected in the promoter region ( -1000 to 0 ) of several genes ( S1 Table ) . Determination of the accurate binding sequences for AtrR is crucial to allow detailed understanding of the molecular function of AtrR . As molecular oxygen is a critical component for several biochemical processes , fungi must have the ability to adapt to hypoxic conditions . Previous studies showed that ergosterol biosynthesis is one of the fungal metabolic pathways most responsive to oxygen level [27 , 32 , 53 , 54] . Several enzymes in the fungal ergosterol biosynthesis pathway , such as the cytochrome P450 Erg3 and Erg11/Cyp51 , require molecular oxygen as well as heme . Heme biosynthesis is also an oxygen-requiring pathway . The SREBP pathway coordinates oxygen levels with sterol and heme biosynthesis since these pathways are responsive to levels of this gas [28 , 30 , 55] . In the present study , AtrR was found to be involved in hypoxia adaptation as seen before for SrbA . In fact , the expression levels of cyp51A , erg3B , erg24A , and erg25A were induced in response to oxygen limitation in an AtrR- and SrbA-dependent manner ( Fig 5B ) . Although the mechanism is still unknown , AtrR is responsible for sterol regulation under oxygen-limited condition , likely in association with SREBP pathway . Sterol regulation of S . cerevisiae and Candida species involves a Gal4-type Zn2-Cys6 transcription factor Upc2 . Interestingly , these yeasts lack an SREBP pathway . A recent finding in the lipid degrading yeast Yarrowia lipolytica that retains an intact SREBP , called YlSre1 , revealed that YlSre1 only played a minor role while Upc2 played a major role in sterol regulation , hypoxia adaptation and azole resistance [56] . Therefore it was proposed that Upc2 has replaced the role of SREBPs in sterol regulation in these yeasts . From a comparison of the amino acid sequences , it is evident that AtrR is not an orthologue of Upc2 . Several lines of data have supported an intimate relationship between hypoxic adaptation and pathogenicity of pathogenic fungi . 1 ) The murine lung infected by fungi is a hypoxic environment at the infection site , which was revealed by a hypoxic indicator [57] , and 2 ) SREBP mutants of C . neoformans ( Sre1 ) and A . fumigatus ( SrbA ) which are unable to grow under hypoxic condition showed an avirulent phenotype [27 , 32] . We demonstrated that AtrR is also required for both hypoxic adaptation and A . fumigatus virulence . As AtrR is a fungal specific transcription factor , it would be a potential drug target , inactivation of which could suppress A . fumigatus in vivo growth during the infection process . One of the most noteworthy results in the present study is that deleting the atrR gene effectively sensitized the clinically isolated multi azole-resistant strain ( Fig 10A and Table 3 ) . Loss of atrR from this clinical strain reduced expression levels of cyp51A and cyp51B to the low level seen in the other laboratory lineage ( Af293-background ) ( Fig 10B ) . This , in turn , likely reduced the amount of target proteins , Cyp51A and Cyp51B , in the cells . Even though the strain carried a G54E substitution in the Cyp51A protein , the reduction in gene transcription contributed to the resulting strains showing hypersensitivity to azoles in contrast to the resistant parental strain IFM 61567 . Alternatively , accumulation of toxic intermediates at multiple AtrR-regulated steps ( Erg3B , Cyp51A , Erg24A , and Erg25A ) in the ergosterol biosynthesis pathway could also contribute to reacquisition of azole sensitivity . Even when cyp51A expression was heterologously elevated in an atrR background , this was insufficient to suppress the azole-susceptibility of this mutant strain ( Fig 3B ) . The same result was also observed in srbA mutant background ( Fig 3B ) , and is consistent with earlier work [29] . These results suggested that defective cyp51A expression was not the only reason for the azole-sensitivity in atrR or srbA deletion mutants . If judged by MIC values caused by loss of AtrR ( itraconazole: 0 . 125–0 . 25; voriconazole: 0 . 06; posaconazole: ≦0 . 06–0 . 06 ) , the resultant deletion mutants would be curable in a conventional azole drug therapy . This fact clearly illuminates a path toward development of new azole-sensitizing agent that could overcome Cyp51A-related azole resistance mechanisms . In this study , we characterized a transcriptional factor designated AtrR that is responsible for a novel A . fumigatus intrinsic azole resistance mechanism . AtrR crucially governs adaptation to azole treatment in the sense that both target ( cyp51A ) and efflux pump ( cdr1B ) are transcriptionally co-regulated by AtrR . The experiment using a clinical azole-resistant isolate revealed a potential clinical relevance of AtrR . Our findings provide the first direct evidence that AtrR serves as a critical coordinator of both ergosterol biosynthesis and levels of a plasma membrane ABC transporter that can efflux azole drugs , the major antifungal drug in current use against pathogenic fungi .
The principles that guide our studies are based on the Guidelines for Proper Conduct of Animal Experiments formulated by Science Council of Japan in June 1 , 2006 . All protocol used in this study were approved by the institutional animal care and use committee of Chiba University ( Permit Number: DOU25-207 , DOU26-228 , and DOU28-345 ) . All efforts were made to minimize suffering in strict accordance with the principles outlined by the Guidelines for Proper Conduct of Animal Experiments . Animals were clinically monitored at least daily and humanely sacrificed if moribund ( defined by lethargy , dyspnea and weight loss ) . All stressed animals were sacrificed by cervical dislocation . A . oryzae ΔligD::sC ( niaD− ) [58] and A . nidulans Δku70 ( pyrG− , pyroA− , biA1− , ku70::argB ) strains ( derived from the strains , RIB40 and ABPU1 , respectively ) were used for a generation of deletion mutants of the AtrR-family genes . For identification of the AoatrR , the A . oryzae strains overexpressing a TF gene were used , which were obtained from a genetic library prepared by Noda Institute for Scientific Research , Kikkoman Corporation [33] . Briefly , the transcription factor genes of interest were cloned into an overexpression vector ( pAPLTN ) [59] , which drives the expression under the amyB promoter system in the presence of maltose as a sole carbon source . A . fumigatus strains Af293 and Afs35 ( akuA::loxP ) were used to generate the following deletion strains and atrR complemented strain: ΔatrR , Co-atrR , ΔsrbA , Δcyp51A , Δcdr1B , ΔatrR , srbA , Af293+PthiA-cyp51A , ΔatrR+PthiA-cyp51A , and ΔsrbA+PthiA-cyp51A . The clinical strain IFM 61567 was isolated in 2011 in Japan from a patient with itraconazole treatment . The mutation in cyp51A gene in the strain was identified as a procedure described previously [60] . The fungal strains used in this study are listed in Table 4 . All strains were routinely cultivated in potato dextrose agar ( PDA ) , potato dextrose broth ( PDB ) , 0 . 1% yeast extract-containing glucose minimal medium ( YGMM ) , or glucose minimal medium ( GMM ) at 37°C . For A . oryzae or A . nidulans culture , Czapek-Dox ( CD ) medium was used , which contained 1% maltose or 1% glucose , 68 . 8 mM ( NH4 ) 2SO4 , trace elements ( 1 μg/ml FeSO4·7H2O , 8 . 8 μg/ml ZnSO4·7H2O , 0 . 4 μg/ml CuSO4·5H2O , 0 . 15 μg/ml MnSO4·4H2O , 0 . 1 μg/ml Na2B4O7·10H2O , 0 . 05 μg/ml ( NH4 ) 6Mo7O24·4H2O ) and appropriate supplements such as 1 mg/ml uridine , 0 . 02 μg/ml biotin , 2 . 5 μg/ml pyridoxine . To collect conidia of each A . fumigatus strain , PDA was used . The antifungal chemicals were commercially obtained as follows: fluconazole , miconazole , itraconazole , propiconazole , bromuconazole , tebuconazole , difenoconazole , clotrimazole ( Wako Pure Chemical Industries , Osaka , Japan ) , and amphotericin B ( Sigma-Aldrich Co . , St . Louis , MO , USA ) . Micafungin was a generous gift from Dr . Keietsu Abe , Tohoku University . For the deletion of the AoatrR gene in A . oryzae , approximately 1-kb fragments upstream and downstream of the coding region of the gene were amplified by PCR using primers YESAoAtrRFw + AoAtrRPtrARv and PtrAAoAtrRFw + AoAtrRYESRv , respectively . The selectable marker gene , ptrA , was isolated by digesting pPTRI ( TaKaRa Bio , Otsu , Japan ) with NheI . Then , the PCR-amplified two DNA fragments , the ptrA gene , and a yeast vector , pYES2 ( Invitrogen , Tokyo , Japan ) , digested with EcoRI and BamHI , were assembled in S . cerevisiae BY4741 ( MATa , his3Δ1 , leu2Δ0 , met15Δ0 , ura3Δ0 ) as described previously [61 , 62] . The resulting plasmid DNA was digested with KpnI and NotI to prepare a deletion construct , which was used for deletion of AoatrR . Similarly , for the deletion of the AnatrR gene in A . nidulans , DNA fragments upstream and downstream of the coding region of the gene were PCR-amplified using primers YESAnAtrRFw + AnAtrRPtrARv and PtrAAnAtrRFw + AnAtrRYESRv , respectively . The PCR-amplified two DNA fragments , the ptrA gene , and pYES2 digested with EcoRI and BamHI were assembled in S . cerevisiae , and used for deletion of AnatrR . Southern blot analyses demonstrated that the transformation cassettes had integrated homologously at the targeted loci and the target ORF was replaced with a selectable marker gene . To construct the A . fumigatus deletion mutants and the complemented strain , plasmids for each purpose were generated . DNA manipulation was performed according to standard laboratory procedures . To amplify DNA fragments from the genome , Prime STAR HS ( TaKaRa Bio ) was used . To prepare gene replacement cassettes for A . fumigatus , the fragments and plasmids were constructed by one-step fusion PCR and GeneArt Seamless Cloning and Assembly Kit ( Invitrogen ) , respectively . Primers used in the present study were listed in S3 Table . To generate a replacement cassette for atrR gene , the 5′- and 3′-flanking regions were obtained using the primers AtrR-5 . UP and AtrR-5 . LP ( hph ) ( for the 5′-region ) and AtrR-3 . UP ( hph ) and AtrR-3 . LP ( for the 3′-region ) . These flanking regions and an hph fragment , hygromycin B resistant marker that was amplified from pCB1004 plasmid , were fused by one-step fusion PCR , which were then used for transformation . For complementation of atrR gene , the atrR fragment containing both upstream and downstream sequences were PCR amplified using the primers AtrR-5 . UPHind and AtrR-3 . LP . The PCR fragment was digested by HindIII and XmaI and ligated into HindIII- and HpaI-digested pKIS518 ( derived from a binary vector pPZP-HYG2 , in which the hygB gene was removed by BamHI digestion , and a ptrA cassette is ligated at KpnI and EcoRV cites between the right and left borders of T-DNA ) , resulting in a plasmid pKIS518-atrR , which was used for Agrobacterium tumefaciens-mediated transformation . To generate replacement cassettes for the srbA , cyp51A , and cdr1B genes , the 5′- and 3′-flanking regions were obtained using the primers srbA-U-F ( pUC119E ) and srbA-U-R ( ptrA ) ( for the 5′-region of srbA ) and srbA-D-F ( ptrA ) and srbA-D-R ( pUC119B ) ( for the 3′-region of srbA ) , cyp51A-U-F ( pBC-ph-s ) and cyp51A-U-R ( ptrA ) ( for the 5′-region of cyp51A ) and cyp51A-D-F ( ptrA ) and cyp51A-D-R ( pBC-ph-x ) ( for the 3′-region of cyp51A ) , and cdr1B-U-F ( pUC119B ) and cdr1B-U-R ( ptrA ) ( for the 5′-region of cdr1B ) and cdr1B-D-F ( ptrA ) and cdr1B-D-R ( pUC119E ) ( for the 3′-region of cdr1B ) , respectively . These flanking regions and a ptrA fragment , pyrithiamine resistant marker that was amplified from pPTRI plasmid ( TaKaRa Bio ) , were fused into pUC119 or pBC-phleo by GeneArt cloning system , resulting in plasmids pUC119-srbA::ptrA , pBC-phleo-cyp51A::ptrA , and pBC119-cdr1B::ptrA . The cassettes used for transformation were amplified from these plasmids . For generation of a plasmid pBC-phleo-PthiA::cyp51A , promoter fragment ( approximately 800bp ) of A . fumigatus thiA ( Afu6g08360 ) and the cyp51A ORF and terminator ( approximately 450bp ) fragment were obtained using primers PthiA-F ( pBC-ph-s ) and PthiA-R ( cyp51A ) ( for the PthiA ) and RT-cyp51A-F and cyp51A+T-R ( pBC-ph-x ) ( for the cyp51A ORF+ter ) , respectively . These fragments were fused into pBC-phleo by GeneArt cloning system , resulting in a plasmid pBC-phleo-PthiA::cyp51A , which was used for transformation to obtain strains , Af293+PthiA-cyp51A , ΔatrR+PthiA-cyp51A , and ΔsrbA+PthiA-cyp51A . A . fumigatus transformation was performed according to a protoplast-polyethylene glycol transformation method for Aspergillus [63] . Precise recombination and integration were confirmed by Southern Blot analysis and/or PCR of the genomic DNA , and in case of gene deletion the absence of mRNA of the target gene was confirmatively verified using real-time RT-PCR analysis . To generate a complemented strain for the atrR gene , A . tumefaciens strain EHA105 carrying the pKIS518-atrR was used for transformation of A . fumigatus ΔatrR strain by the method described elsewhere [64] . Paper-disc diffusion assay: The conidia of each strain of interest were mixed with 20 mL of GMM ( final concentration , 104 conidia/mL ) , and they were pored into a petri dish to solidify . A paper-disc was placed on center of the plate , and 10 μl of drug solution was dropped on it ( fluconazole: 10 mg/mL; miconazole: 10 mg/mL; amphotericin B: 0 . 25 mg/mL; micafungin: 0 . 02 mg/mL ) . The plates were incubated at 37°C for 48 h before photographed . Instead of mixing conidia with pre-solidified GMM , the strains were inoculated by streaking conidia on a GMM plate when multiple strains were investigated . Colony growth inhibition test: The 104 conidia of each strain were inoculated on the center of GMM plates containing 0 to 5 μg/mL of itraconazole or miconazole . The plates were incubated at 37°C for 70 h . The colony diameter was measured from three independent plates , and the mean values for plates with drugs were compared with those without drugs . MIC test: MICs of each strain against antifungal drugs were investigated as described previously [65] . Tests were performed in triplicates using micafungin , amphotericin B , flucytosine , fluconazole , itraconazole , voriconazole , miconazole , and posaconazole in RPMI 1640 medium ( pH 7 . 0 ) at 35°C . This was performed according to the Clinical and Laboratory Standards Institute reference broth microdilution method , document M38-A2 , with partial modifications in term of using the dried plate for antifungal susceptibility testing of yeasts method ( Eiken Chemicals , Tokyo , Japan ) . The mycelia were cultured in GMM , YGMM , or PDB at 37°C and harvested at time-points ( 18~24 h ) appropriate for the applications . The mycelia were frozen in liquid nitrogen , and total RNA was isolated using the FastRNA Pro Red Kit ( MP Biomedicals , Santa Ana , CA , USA ) . To obtain cDNA pools from the total RNA , reverse transcription was performed using the ReverTra Ace qPCR RT Master Mix with gDNA remover ( Toyobo , Osaka , Japan ) . Real-time RT-PCR was performed using SYBR Green detection as described previously [63] . The Thunderbird SYBR qPCR Mix was used for reaction mixture preparation ( Toyobo ) . The primer sets are listed in S3 Table . The relative expression ratios were calculated by the comparative cycle threshold ( Ct ) ( ΔΔCt ) method . The actin gene was used as a normalization reference ( internal control ) . Each sample was tested in triplicate . RNA-sequencing analysis was performed as described previously [66] . Briefly , mRNA libraries and cDNA libraries were prepared by Illumina TruSeq RNA Sample Prep Kit v2 according to the standard protocols ( Illumina , San Diego , CA , USA ) . Each total RNA sample ( 1 μg ) was enriched for mRNA using oligo ( dT ) -tagged beads . The library construction involved cDNA synthesis , end repair , A-tailing , adapter ligation , and amplification . The mean length for each library was approximately 280 to 300 bp . Sequencing was performed in a pair-end 50 base mode on a Miseq system ( Illumina ) . In this study , we performed two independent experimental settings: one containing Af293 , ΔatrR , and ΔsrbA ( total 3 samples ) and the other containing Af293 and ΔatrR with DMSO , fluconazole , or miconazole treatment ( total 6 samples ) . These read data were deposited to the DDBJ Sequence Read Archive under accession No . PRJDB5273 . To compare expression levels for each gene , the sequences were analyzed using CLC genomics workbench ( CLC Bio , Aarhus , Denmark ) . The sequence reads were trimmed on the program software . The only reads with quality values higher than Q30 were used for mapping . The A . fumigatus Af293 genome data retrieved from NCBI ( http://www . ncbi . nlm . nih . gov/bioproject/PRJNA14003 ) was used as the template for mapping . From the mapping data , RPKM values were calculated . Heat map presentation was performed using Cluster 3 . 0 and Java TreeView ver 1 . 1 . 6r2 according to the instructions . The conidia of each strain were inoculated on GMM plate , and the plate was incubated at 37°C in anaeropack system ( Mitsubishi Gas Chemical , Tokyo , Japan ) where initial concentration of oxygen was set at 2% ( hypoxia condition ) . After 96 h , oxygen in the pack appeared to be completely consumed as estimated from no more growth of all colonies on the plate . The plate for normoxia was incubated at 37°C in a regular incubator without gas control ( 21% oxygen ) for 72 h . The mouse model of pulmonary aspergillosis was performed according to [67] with slight modifications . A . fumigatus strains Af293 , ΔatrR , and Co-atrR were used to infect immunosuppressed mice ( 12 mice per group ) . Outbreed male ICR mice ( 5w , body weight , 20 to 24 g ) were housed in sterile cages with sterile bedding and provided with sterile feed and drinking water containing 300 μg/mL tetracycline hydrochloride to prevent bacterial infection . Mice were immunosuppressed with cyclophosphamide at a concentration of 200 mg per kg of body weight , which was administered intraperitoneally on days −4 , −2 , 1 , 3 , 5 , and 7 prior to and post-infection ( day 0 ) . The A . fumigatus conidia used for inoculation were grown on PDA for 5 days prior to infection . Fresh conidia were harvested in PBS+0 . 01% Tween 20 and filtered through a Falcon Cell Strainer ( Corning Inc . , NY , USA ) . Conidial suspensions were spun for 5 min at 3 , 000 × g , washed twice with PBS+0 . 01% Tween 20 , counted using a hemocytometer , then resuspended at a concentration of 106 conidia/μl . Mice were anesthetized by ketamine and xylazine and infected by intratracheal instillation of 2 . 5 × 107 conidia in 25 μl of PBS . Mice were weighed and visually inspected every 24 h from the day of infection . The endpoint for survival experimentation was identified when a 30% reduction in body weight was recorded , at which time the mice were sacrificed . The statistical significance of comparative survival values was calculated using log rank analysis using the Prism statistical analysis package . For A . fumigatus strains Afs35 , ΔatrR , ΔsrbA , and ΔatrR ΔsrbA , immunosuppressed ICR mice ( 5w , body weight , 20 to 24 g ) were used as described above ( 9 to 11 mice per group ) . Mice were immunosuppressed with cyclophosphamide at a concentration of 150 mg per kg of body weight , which was administered intraperitoneally on days −4 , −1 , 3 , 6 , 9 , and 12 prior to and post-infection ( day 0 ) . Cortisone acetate ( Sigma-Aldrich Co ) was also administrated subcutaneously at a concentration of 200 mg per kg of body weight on day -1 . Mice were anesthetized by ketamine and xylazine and infected by intratracheal instillation of 3 × 105 conidia in 30 μl of PBS . Mice were weighed and visually inspected every 24 h from the day of infection . The endpoint for survival experimentation was identified when a 30% reduction in body weight was recorded , at which time the mice were sacrificed . For ChIP experiments , a strain possessing 3x HA-tagged allele of atrR was constructed . A cassette consisting of 750 bp of the C-terminus of atrR fused to a 3x HA tag , a trpC terminator followed by a hygromycin resistance gene and 750 bp of the 3’ untranslated region of the atrR gene was transformed into Afs35 cells with selection for hygromycin resistance . Resistant clones were confirmed to contain a homologous integration by PCR ( Designated SPF89 strain ) . 1×106 spores/mL of A . fumigatus strains AfS35 ( atrR ) and SPF89 ( atrR-3xHA ) were grown in 50 mL of Sabouroud dextrose medium in 250 mL shaking flask cultures for 16 h . Cross-linking was done in ( 0 . 4 M Sucrose , 10 mM Tris-HCl , pH 8 . 0 , 1 mM EDTA , adding 1 mM PMSF and 1% formaldehyde just before use ) for 15 min under shaking ( 100 rpm ) at 30°C . Crosslinking was stopped by adding 2 . 5 mL of 2 M glycine , and continued shaking incubation for 10 min . Mycelia were collected and dried using vacuum filtration and rinsed with sterile ddH2O and frozen immediately with liquid nitrogen and stored at −80°C . Approximately 100 mg of frozen mycelia were ground to a fine powder in a chilled mortar and pestle with liquid nitrogen added . Powder was transferred to 0 . 5 mL of ChIP lysis buffer ( CLB: 50 mM HEPES pH 7 . 5 , 150 mM NaCl , 1 mM EDTA , 1% Triton X-100 , 0 . 1% Deoxycholate ( Sigma D6750 ) , 0 . 1% SDS , 1 mM PMSF , 1× fungal proteinase inhibitor cocktail ( Sigma ) ) . Each sample was vortexed for 2 min and then split into 3*130 μl volume in AFA Fiber Pre-Slit Snap-Cap [6 x 15mm] microTUBE [Covaris] ( 130μl sample per tube ) . Chromatin was sheared with E220 Focused-ultrasonicator [Covaris] under following conditions: Peak Incident Power ( W ) : 175; Duty Factor: 10%; Cycles per burst: 200; Treatment Time: 420 sec; Temperature: 7°C; Sample volume: 130 μl; in the presence of E220 –Intensifier ( pn500141 ) . Tubes from each sample were then pooled and centrifuged at 10 , 000 g for 5 min at 4°C . Supernatant was transferred into new tube . 25 μL was reserved as input control ( IC ) fraction for reverse crosslinking to verify sonication and control for ChIP and qPCR . The sheared chromatin was incubated with HA . 11 ( 16B12 ) mouse monoclonal antibody ( Covance ) at 1:150 dilution overnight ( 12 h ) on a nutator at 4°C . This sample was further incubated with 30 μL of washed Protein G Dynabeads ( Life Technologies ) for another 12 h . Washing , reverse-crosslinking and purification of ChIP-ed DNA was performed as described in Chung et al . [30] . Semiquantitative PCR was initially performed across the cyp51A and cdr1B promoters to serve as initial verification of enrichment of these promoter regions upon ChIP with HA . This PCR was done using Phusion DNA polymerase ( NEB ) under following conditions: 95°C for 30 s followed by 25 cycles of 95°C for 15 s , 60°C for 10 s and 72°C for 10 s . The primers used for the reaction were cyp51A-ChIP-F1 and cyp51A-ChIP-R1 , cyp51A-ChIP-F2 and cyp51A-ChIP-R2 , and cyp51A-ChIP-F3 and cyp51A-ChIP-R3 ( for the cyp51A promoter ( -69 to -415 , -395 to -697 , and -677 to -995 , respectively ) ) , cdr1B-ChIP-F1 and cdr1B-ChIP-R1 , cdr1B-ChIP-F2 and cdr1B-ChIP-R2 , and cdr1B-ChIP-F3 and cdr1B-ChIP-R3 ( for the cdr1B promoter ( -75 to -393 , -376 to -691 , and -675 to -982 , respectively ) ) , and act1-ChIP-F and act1-ChIP-R ( for the act1 promoter ( -134 to -454 ) ) ( S3 Table ) . Real-time PCR was performed in triplicate for each separate ChIP experiment using primers designed for regions as identified above as enriched in preliminary analysis , under the following conditions: 1 cycle of 95°C for 30 s followed by 40 cycles of 95°C for 15 s and 60°C for 30 s on an MyiQ2 BioRad machine . 1 μl of ChIP-ed or input ( diluted 25-fold to bring it to 1% ) DNA was used in 20 μl total volume reaction using SYBR green master mix ( BioRad ) and 0 . 4 μM of each primer . Percent input method was used to calculate the signal of enrichment of the promoter region for each gene ( http://www . thermofisher . com/us/en/home/life-science/epigenetics-noncoding-rna-research/chromatin-remodeling/chromatin-immunoprecipitation-chip/chip-analysis . html ) .
|
Better survival of chronically ill patients has produced an increased number of cases involving fungal infectious disease . This is partly due to a larger number of immunocompromised patients . Meanwhile , the catalogue of antifungal drugs remains quite limited and the appearance of resistant isolates is becoming more common . The filamentous fungus Aspergillus fumigatus is the main causative pathogen for deep-seated aspergillosis . For this fungus , resistance to azoles , which are the most commonly utilized drug for treatment of aspergillosis , has exhibited an alarming increase linked with elevated mortality . We know relatively little of the molecular details underpinning azole resistance in A . fumigatus other than alterations in the target enzyme . Here , we identified and characterized a novel transcription factor AtrR that is required for wild-type azole resistance in this fungus as well as other Aspergillus fungi , such as Aspergillus oryzae and Aspergillus nidulans . We discovered that AtrR co-regulates the target and pump , both of which are essential for azole resistance . More importantly , deletion of the atrR gene could sensitize a clinically isolated multi-azole resistant strain and exhibited a virulence defect in a mouse model . This success in overcoming an existing resistance mutation provides a new avenue for the prevention and treatment of aspergillosis .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"aspergillus",
"fumigatus",
"medicine",
"and",
"health",
"sciences",
"chemical",
"compounds",
"pathology",
"and",
"laboratory",
"medicine",
"aspergillus",
"fungal",
"genetics",
"gene",
"regulation",
"pathogens",
"regulatory",
"proteins",
"microbiology",
"dna-binding",
"proteins",
"organic",
"compounds",
"fungi",
"transcription",
"factors",
"pharmacology",
"fungal",
"pathogens",
"mycology",
"antimicrobial",
"resistance",
"heterocyclic",
"compounds",
"azoles",
"proteins",
"medical",
"microbiology",
"gene",
"expression",
"microbial",
"pathogens",
"chemistry",
"molds",
"(fungi)",
"biochemistry",
"organic",
"chemistry",
"genetics",
"microbial",
"control",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"organisms"
] |
2017
|
A Novel Zn2-Cys6 Transcription Factor AtrR Plays a Key Role in an Azole Resistance Mechanism of Aspergillus fumigatus by Co-regulating cyp51A and cdr1B Expressions
|
Basement membranes ( BMs ) are thin sheet-like specialized extracellular matrices found at the basal surface of epithelia and endothelial tissues . They have been conserved across evolution and are required for proper tissue growth , organization , differentiation and maintenance . The major constituents of BMs are two independent networks of Laminin and Type IV Collagen in addition to the proteoglycan Perlecan and the glycoprotein Nidogen/entactin ( Ndg ) . The ability of Ndg to bind in vitro Collagen IV and Laminin , both with key functions during embryogenesis , anticipated an essential role for Ndg in morphogenesis linking the Laminin and Collagen IV networks . This was supported by results from cultured embryonic tissue experiments . However , the fact that elimination of Ndg in C . elegans and mice did not affect survival strongly questioned this proposed linking role . Here , we have isolated mutations in the only Ndg gene present in Drosophila . We find that while , similar to C . elegans and mice , Ndg is not essential for overall organogenesis or viability , it is required for appropriate fertility . We also find , alike in mice , tissue-specific requirements of Ndg for proper assembly and maintenance of certain BMs , namely those of the adipose tissue and flight muscles . In addition , we have performed a thorough functional analysis of the different Ndg domains in vivo . Our results support an essential requirement of the G3 domain for Ndg function and unravel a new key role for the Rod domain in regulating Ndg incorporation into BMs . Furthermore , uncoupling of the Laminin and Collagen IV networks is clearly observed in the larval adipose tissue in the absence of Ndg , indeed supporting a linking role . In light of our findings , we propose that BM assembly and/or maintenance is tissue-specific , which could explain the diverse requirements of a ubiquitous conserved BM component like Nidogen .
Basement membranes ( BM ) are specialized thin extracellular matrices underlying all epithelia and endothelia , and surrounding many mesenchyme cells . This thin layer structure , which appears early in development , plays key roles in the morphogenesis , function , compartmentalization and maintenance of tissues [1] . All BMs contain at least one member of the Laminin , Type IV Collagen ( Col IV ) , proteoglycan Agrin and Perlecan , and Nidogen ( Entactin ) families . Nidogen is a 150-kDa glycoprotein highly conserved in mammals , Drosophila , Caenorhabditis elegans ( C . elegans ) and ascidians [2 , 3] . Nidogens have been proposed to play a key role in BM assembly by providing a link between the Laminin and Col IV networks and by integrating other ECM proteins , such as Perlecan , into this specialized extracellular matrix [4–7] . While invertebrates possess only one Nidogen , two Nidogen isoforms , Nid1 and Nid2 , have been identified in vertebrates . The individual knock out of either Nid1 or Nid2 in mice does not affect BM formation or organ development [8–10] . In fact , these Nid1 or Nid2 null animals appear healthy , fertile and have a normal life span . However , simultaneous elimination of both isoforms results in perinatal lethality , with defects in the lung , heart and limb development that are not compatible with postnatal survival [11 , 12] . In addition , BM defects are only observed in certain organs , which strongly suggests tissue-specific roles for Nidogens in BM assembly and function [11] . Like in mice , loss of the only Nidogen-encoding gene in C . elegans , NID-1 , is viable with minor defects in egg laying , neuromuscular junctions and position of longitudinal nerves , but no defects in BM assembly [13–15] . Altogether , these studies reveal that Nidogen may play important roles in specific contexts , consistent with its evolutionary conservation . However , the different requirements for Nidogens in BM assembly and organogenesis in mice and C . elegans suggest that new functions may have arisen in vertebrates . The isolation of mutants in Nidogen in other organisms will help to shed light on the role of the Nidogen proteins in vivo and its conservation throughout evolution . All Nidogens comprise three globular domains , namely G1 , G2 and G3 , one flexible linker connecting G1 and G2 , and one rod-shaped segment , composed primarily of epidermal growth factor repeats , separating the G2 and G3 domains [4 , 16 , 17] . A number of studies using recombinant fragments of Nidogens have provided a wealth of information on the structure and binding properties of the different Nidogen domains in vitro . Thus , key roles have been proposed for the globular domains G3 and G2 in mediating interactions of Nidogen with the Laminin network and with the Collagen IV network , respectively [4 , 7 , 17–20] . Despite this , the relevance of these interactions in vivo remains to be established . Furthermore , some of the predictions from cell culture and in vitro experiments do not hold when tested in model organisms . For example , deletion of the G2 domain in C . elegans is viable and does not affect organogenesis [14] . Furthermore , it has been shown that Ndg1 and Ndg2 do not form molecular cross-bridges between the Laminin and Collagen IV networks in the epidermal BM of human skin [21] . These results in animal models are inconsistent with a role for Nidogen as a generally essential linker between the Collagen IV and Laminin networks , leaving open the question of whether in vivo Nidogen functions at all as a linker . Drosophila BMs are analogous to the vertebrate ones [22] . They cover the basal surface of all epithelia and surround most organs and tissues , including muscles and peripheral nerves . Even though their composition might vary according to tissues and developmental stages , all Drosophila BMs contain Col IV , Laminin , Perlecan and Nidogen . However , in contrast to the three Col IV , sixteen Laminins and two Nidogens found in humans , Drosophila only produces one Col IV , two distinct Laminins and one Nidogen ( Ndg ) . The reduced number of ECM components , which limits the redundancy among them , and their high degree of conservation with their mammalian counterparts , makes Drosophila a perfect model system to dissect their function in vivo . Drosophila Col IV has been identified as a homolog of mammalian Type IV Collagen , which is a long helical heterotrimer that consists of two α1 chains and one α2 chain encoded by the genes Collagen at 25 C ( Cg25C ) and viking ( vkg ) , respectively [23–25] . The C terminal globular non-collagenous ( NC1 ) domain and the N terminal 7S domain interact to form the Col IV network [26] . Loss of function mutations in either of the two Col IV genes in flies affect muscle development , nerve cord condensation , germ band retraction and dorsal closure , causing embryonic lethality [27] . In addition , mutations in Col IV have been associated with immune system activation , intestinal dysfunction and shortened lifespan in the Drosophila adult [28] . Finally , while Col IV deposition in wing imaginal discs and embryonic ventral nerve cord ( VNC ) BMs is not required for localization of Laminins and Nidogens , it is essential for Perlecan incorporation [29 , 30] . The Drosophila Laminin αβγ trimer family consists of two members comprised of two different α subunits encoded by Laminin A and wing blister , one β and one γ subunits encoded by Laminin B1 and Laminin B2 , respectively [31] . Same as Col IV , Laminin trimers can also self-assemble into a scaffold through interactions of the N-terminal LN domains located in their short arms [32] . Elimination of Laminins in Drosophila affects the normal morphogenesis of most organs and tissues , including the gut , muscles , tracheae and nervous system [33 , 34] . In addition , abnormal accumulation of Col IV and Perlecan was observed in Laminin mutant tissues [33] . Perlecan , encoded by the trol ( terribly reduced optic lobes ) gene , is subdivided into five distinct domains . Interactions with Laminins and Col IV occur through domains I and V ( reviewed in [35] ) . Mutations in trol affect postembryonic proliferation of the central nervous system , plasmatocytes and blood progenitors [36–38] . Loss of trol also affects the ultrastructure and deposition of Laminins and Col IV in the ECM around the lymph gland [38] . Altogether , these results suggest that BM components Laminin , Col IV and Perlecan are all essential for proper development . In addition , they also reveal a hierarchy for their incorporation into BMs that seems to be tissue-specific and required for proper BM assembly and function . In this context , however , the role of Ndg in Drosophila morphogenesis and BM assembly has remained elusive . This may be in part due to the lack of mutations in this gene . In this work , we have dissected the role of Ndg in Drosophila . Using a newly generated anti-Ndg antibody , we have shown that Ndg accumulates in the BMs of embryonic , larval and adult tissues . By isolating several mutations in the single Drosophila Ndg gene , we find that while , similar to C . elegans and mice , Ndg is not required for overall organogenesis or viability , it is required for fertility . Also similar to the tissue-specific defects in mice and C . elegans , we find that the BMs surrounding the larval fat body and flight muscles of the notum are disrupted in the absence of Ndg . Furthermore , we observed uncoupling of laminin and Collagen IV in the fat body of Ndg mutants , indeed supporting a role of Ndg as a linker between the two networks . In addition , we have performed a thorough functional analysis of the different Ndg domains in vivo , which , on one hand , supports an essential requirement of the G3 domain for Ndg function and , on the other hand , uncovers a new key role for the Rod domain in regulating Ndg incorporation into BMs . Finally , we find that BM assembly is not universal but differs depending on the tissue and propose that this could explain the diverse requirements of a ubiquitous conserved BM component like Nidogen .
Previous analysis has shown that , during embryogenesis , Ndg is expressed in multiple mesodermal cell types , such as visceral mesoderm , somatic muscle founder cells , a subset of pericardial and cardial cells and at the edges of the visceral mesoderm [39–41] . Here , we decided to further analyse Ndg expression in embryonic , larval and adult tissues . In order to do this , an antibody against a peptide encoded by exon 7 was developed ( see Materials and Methods ) . We found that in addition to the pattern described previously , similar to Laminins [33] , Ndg was also detected in the BM surrounding most embryonic tissues in stage 16 embryos , including muscles , ventral nerve cord ( VNC ) and gut ( Fig 1A , 1A’ , 1B and 1B’ ) . However , in contrast to Laminins , Ndg was not enriched at muscle attachment sites ( Fig 1A ) . In addition , a careful analysis of Ndg expression in stage 13 embryos revealed a dotted pattern along the visceral mesoderm , which differs from the continuous line observed around the muscles or the VNC ( Fig 1C and 1C’ ) . At this stage , caudal visceral mesodermal cells migrate over the visceral mesoderm . In fact , using a marker for these cells , croc-lacZ [42] , we found that Ndg accumulated around them as they migrate ( Fig 1C and 1C’ ) . In this case , Ndg seems to be organized in track-like arrays , similar to the distribution of laminins around migrating hemocytes . Ndg was also found in migrating hemocytes , as visualized using a version of Nidogen tagged with superfolder GFP ( sGFP ) , expressed from a duplication of the Ndg genomic region ( Fig 1D , [43] ) . Finally , Ndg was also found at high levels in chordotonal organs ( Fig 1A and 1A’ , asterisk ) . These results suggest that as it is the case for Laminins , Ndg can be deposited and/or assembled in different patterns throughout embryogenesis . In addition , and similar to the other BM components , Ndg was found in the BMs that surround most larval tissues , including fat body , imaginal discs , tracheae , salivary glands , midgut , mature muscles and heart ( Fig 1E and Fig 2 ) , as well as in the follicular epithelium of the adult ovary ( Fig 1F ) . A recent study has shown that macrophages are the major producers of BM components in the Drosophila embryo [30] . To investigate the cellular origin of Nidogen in the developing fly , we designed a GAL4-driven , UAS-controlled short hairpin against sGFP to eliminate sGFP-tagged Nidogen without disrupting normal function of endogenous untagged Nidogen ( Fig 2A ) . This approach ( isGFPi , for in vivo sGFP interference ) is similar to iGFPi ( in vivo GFP interference , [29 , 44] and iYFPi [45] , which we previously used to show that fat body adipocytes are the major source of Collagen IV and Perlecan in the larva . We found here that isGFPi knock down of Ndg . sGFP driven by Cg-GAL4 , which drives expression in fat body and blood cells ( Cg>isGFPi ) , reduced the presence of Ndg . sGFP in the whole animal . Ndg . sGFP signal was largely reduced or undetectable in most tissues , including fat body itself , imaginal discs , tracheae , midgut and heart ( Fig 2B ) . Deposition of Ndg . sGFP was only partially reduced in the VNC , the imaginal ring of the salivary gland and body wall muscles , and was not visibly affected in myoblasts ( Fig 2B ) , suggesting that some larval tissues besides the fat body could produce their own Ndg . To test this in the case of myoblasts , we induced isGFPi with muscle-specific Mef2-GAL4 [46] , and found that Ndg . sGFP disappeared from notum myoblasts ( Mef2>isGFPi , Fig 2B ) , proving that they produce their own Nidogen . These results show that , as it is the case for Collagen IV , fat body and blood cells are the main source of Ndg in the larva , but also that exceptions to this rule exist . We next decided to assess the origin of the three other core BM components by performing the same assay for sGFP-tagged Laminin B1 , GFP-tagged Col IV and YFP-tagged Perlecan ( S1A Fig ) . We found that , similar to Nidogen , fat body and blood cells are the main source of the BM components of all tissues , except myoblasts and partially the VNC ( S1A Fig ) . Component-specific exceptions were tracheal cells with regard to Perlecan , the imaginal rings of the salivary glands for Perlecan and Laminin , and finally imaginal discs for Laminin ( S1A Fig ) . Indeed , LanB1 knock down in the wing disc under control of en-GAL4 reduced presence of Laminin in the corresponding region of the disc , proving the ability of imaginal discs to secrete part or all of their Laminin ( S1B Fig ) . In all , our results show that although the four core BM components are largely produced by fat body adipocytes and blood cells during larval stages , other tissues may also be able to provide them . Consistently , Ndg mRNAs have been detected in muscle founder cells [39] , while Laminin mRNAs [47] and Laminin protein secretion [48] have been shown in the developing VNC . The Drosophila genome contains a single Ndg gene . To analyse Ndg requirements during development , we isolated a series of deficiencies uncovering the gene Ndg ( Fig 3A , S2 Fig; see Materials and Methods ) . These deficiencies were all homozygous lethal . However , as they removed other genes ( S2 Fig ) , we could not draw conclusions from this result . Therefore , we next took advantage of the CRISPR-Cas9 technology to isolate a series of specific Ndg alleles that would allow us to study Ndg function ( Fig 3A; see Materials and methods ) . To generate Ndg null alleles , embryos were injected with Cas9 mRNA and a combination of four sgRNAs designed against the 5’UTR exon ( sgRNA1 ) , exon 3 ( sgRNA2 and sgRNA3 ) and exon 8 ( sgRNA4 ) . Two mutant lines in which the intervening Ndg sequence between sgRNA1 and sgRNA4 had been deleted partially ( Ndg1 ) or completely ( Ndg2 ) were isolated ( Fig 3A ) . Gene CG3422 , contained between exons 9 and 10 of the Ndg gene was not perturbed . Both mutations are predicted to be Ndg null alleles because of absence of a transcription start site . In fact , qRT-PCR , using different primers along the Ndg gene ( Fig 3A ) , showed no mRNA expression in Ndg1 homozygous mutant larvae compared to wild type controls ( Fig 3B ) . Furthermore , consistent with Ndg1 and Ndg2 being null alleles , staining with our Ndg antibody could not detect presence of the protein in larval or embryonic tissues ( Fig 3C and 3D , S3D Fig ) . As domain G3 has been postulated to be critical for binding of Ndg to Laminins , we also isolated Ndg mutant alleles in which this domain and the adjacent Rod domain were eliminated ( ΔRod-G3 alleles ) in order to analyze its function in the context of the whole organism . In this case , transgenic lines stably expressing an sgRNAs against exon 5 were generated and crossed to flies expressing Cas9 ( see Materials and Methods ) . Three mutant alleles , NdgΔRod-G3 . 1 , NdgΔRod-G3 . 2 and NdgΔRod-G3 . 3 , were selected ( Fig 3A ) . Two of them , NdgΔRod-G3 . 1 , NdgΔRod-G3 . 2 , were deletions of five and eight base pairs that resulted in frame-shifts generating stop codons eight and seven amino acids after the shift , respectively . In the other one , NdgΔRod-G3 . 3 , six base pairs were replaced by seven different ones , generating a frame-shift and a stop codon right after the shift . As expected , no staining using the antibody generated in this study was detected in NdgΔRod-G3 . 1 homozygous embryos ( Fig 3D and S3 Fig ) . All CRISPR/Cas9 Ndg mutant alleles we generated were homozygous viable with no obvious morphological abnormalities ( Fig 3E ) . These data show that , in contrast to double Nid1 Nid2 knock out mice and similar to C . elegans Nid-1 mutants , Ndg is dispensable for viability in Drosophila . In addition , similar to C elegans [14] , elimination of Ndg results in reduced fertility in flies ( S4A Fig ) . However , in contrast to the phenotype observed when removing Laminins , Col IV or Perlecan [22 , 49] , no defects were observed in the shape of eggs laid by Ndg mutant females ( S4B Fig ) . Once shown that Ndg mutant flies are viable , we decided to analyze the effects of Ndg loss in the BMs of the fly . We could not detect defects in most of the BMs we analyzed , including those present in the embryo , larval epidermis , imaginal discs , salivary glands , gut , muscles , VNC and the follicular epithelium in the ovary . However , we did observe a clear defect in the BM surrounding the larval fat body ( Fig 4A ) . The larval fat body is an organ formed by large polyploid cells ( adipocytes ) covered by a BM that separates it from the hemolymph [50] . This BM contains , besides Ndg , the other three major components of BMs: Col IV , Laminins and Perlecan . Using tagged versions of these proteins , we found that the BM surrounding the fat body adipose tissue of Ndg1 mutant larvae showed many holes , in contrast to the continuous appearance of the BM in wild type controls ( Fig 4A ) . This phenotype was also observed when we knocked down Ndg expression using Cg-GAL4 ( Ndgi , Fig 4A ) and in the fat body of transheterozygous Ndg1/Ndg2 and Ndg1/Df ( 2R ) BSC281 larvae ( Fig 4B ) . Furthermore , loss of BM integrity was additionally displayed by transheterozygous Ndg1/NdgΔRod-G3 . 1 fat body ( Fig 4B ) , indicating a strong requirement of the Rod and G3 domains for this Ndg function . Confirming that this phenotype reflected a loss of Ndg function , the Ndg . sGFP transgene ( see Fig 2B ) rescued the integrity of fat body BMs in Ndg1 mutants ( Fig 4C ) . We next investigated whether adipose tissue physiology was affected in Ndg mutants . Similar to human adipose , a major role of insect fat body is storage of neutral lipids . We stained fat body adipocytes with neutral lipid dye BODIPY and found that the lipid content in Ndg1 and Ndg2 mutant adipocytes was reduced , with some cells clearly presenting fewer and smaller lipid droplets than controls ( Fig 4D and 4E ) . Quantification of lipid droplet diameter confirmed a significant reduction in droplet size in Ndg mutant larvae ( Fig 4E ) . This result indicates a mild effect of Ndg loss in the physiology of these cells , suggesting inefficient lipid adsorption or intracellular metabolism . In addition to fat body BM defects observed in Ndg1 mutant larvae , we further discovered BM integrity defects in the flight muscles of the notum in Ndg mutant flies ( S4C Fig ) . In addition , while flies appeared to fly normally and negative geotaxis climbing assays did not show differences with the wild type ( not shown ) , Chill Coma Recovery Time ( CCRT ) assays [51] showed increased recovery times after cold exposure in flies lacking Ndg , suggesting mild behavioral or motor defects ( S4D Fig ) . In summary , our results show that although Ndg is not critically essential for fly development and assembly of most BMs , it is necessary for the integrity of BMs around specific tissues , such as the larval fat body and the adult flight muscles , and for appropriate fertility and fitness of the fly . In order to better understand the function of Ndg , we performed a functional analysis of the different Ndg domains . To do that , we generated transgenic flies capable of expressing GFP-tagged versions of the protein as well as mutant variants lacking one or several domains ( Fig 5A ) and tested their ability to localize to the BM of wing discs when expressed in fat body and blood cells . We carried out this localization analysis in the wing disc because Ndg present at wing disc BMs is produced by fat body and blood cells ( Fig 2B ) and because successful incorporation into the BM of a mutant GFP-tagged Ndg protein after secretion can be easily discerned in this tissue through confocal imaging . All the mutant variants we generated retain the signal peptide of full length Ndg to ensure correct secretion . Confirming secretion of all of them into the hemolymph , GFP signal was detected in the kidney-like pericardial cells ( S5C Fig ) , known to filter the hemolymph and concentrate proteins present in it [52] . When expressed in the fat body and blood cells under control of the Cg-GAL4 driver , full length Ndg ( NdgFL . GFP ) was able to localize to the BMs of imaginal discs , as expected ( Fig 5B ) . A similar analysis of the localization properties of the deletion constructs showed that no single domain of the protein was capable by itself to confer localization to BMs , suggesting cooperative interactions among domains are required for BM localization ( Fig 5B ) . In addition , analysis of the localization of proteins in which a single domain was deleted ( NdgΔG1 , NdgΔG2 , NdgΔG3 and NdgΔRod ) indicated that the only domain absolutely required for BM localization was the Rod domain , as NdgΔRod was uncapable of localizing to the BM of imaginal discs or other larval tissues ( Fig 5B , S6 Fig ) . However , the Rod domain was insufficient to drive protein localization on its own ( NdgRod ) , but required the presence of the G2 or the G3 domains ( NdgG2Rod and NdgRodG3 , respectively; Fig 5B ) . This result is also supported by our analysis of the localization of Ndg in our NdgΔRod-G3 mutant embryos using an antibody raised against the G2 domain of Ndg [34] ( S3 Fig ) . Staining with this antibody showed that the mutant protein NdgΔRod-G3 , lacking the Rod and G3 domains , did not localize to embryonic BMs , but it was still present in the chordotonal organs ( S3B Fig ) . Altogether , our results show that the Rod domain is required but not sufficient for Ndg BM localization . They also show that G2+Rod and Rod+G3 are minimal alternative units capable of conferring BM localization to the Ndg protein , with G1 enhancing G2+Rod dependent-localization . As mentioned in the introduction , analysis of the binding properties of Ndg domains and crystal structure have suggested that the G3 domain binds to Laminin , whereas the G2 domain binds to Col IV , with no clear function ascribed so far to the G1 domain . To investigate this in vivo , we assayed the localization abilities of the NdgΔG1 , NdgΔG2 and NdgΔG3 mutant proteins in the BMs of fat bodies where the expression of either Laminins or Col IV had been knocked down . We found that knock-down of LanA or Cg25C , encoding a Laminin α chain and Col IV α1 , caused a marked reduction in the incorporation of full length NdgFL . GFP ( Fig 6A–6C ) . In addition , while knocking down LanA had no effect on NdgΔG3 localization , LanA loss caused a marked reduction in the localization of both NdgΔG1 and NdgΔG2 ( Fig 6A and 6B ) . Conversely , knocking down Cg25C resulted in a strong reduction in NdgΔG3 localization and significant but not as drastic effects on the localization of NdgΔG1and NdgΔG2 ( Fig 6A and 6C ) . Altogether , these results show that localization directed by the G3 domain depends on Laminins , whereas localization by the G1 and G2 domains depends on Col IV . Next we tested the ability of the Ndg mutant proteins lacking the G1 , G2 or G3 domains , all three capable of localizing to BMs , to rescue the fat body BM defects observed in Ndg1 mutant larvae . Overexpression of the mutant variants NdgΔG1 or NdgΔG2 was able to rescue integrity of the fat body BM ( Fig 6D ) , as imaged with Cg25C . RFP [53] . In contrast , expression of the mutant form NdgΔG3 failed to rescue BM integrity ( Fig 6D ) , indicating that G3 is a key domain for Ndg function , while the G1 and G2 domains may function in a partially redundant way . This is supported by our results showing that Ndg1/NdgΔRod-G3 . 1 transheterozygous mutant larvae show fat body BM defects indistinguishable from those found in Ndg1 homozygotes ( Fig 4A and 4B ) . The localization and rescue properties of the different domains of Ndg suggest that Ndg may indeed act as a linker between Laminin and Collagen IV , as originally proposed . Confirming this , simultaneous imaging of Collagen IV and Laminin in fat body BMs shows that in the Ndg1 mutant Laminin and Collagen IV appear separate from each other when the broken BM is observed at high magnification ( Fig 6E; co-localization analysis in Fig 6G ) . Conversely , Perlecan and Collagen IV were still highly co-localized in the damaged BM of Ndg mutant fat body ( Fig 6F and 6G ) . In all , these results are consistent with a function of Ndg as a linker of the Col IV and Laminin networks ( Fig 6H ) . This linker function would depend on binding to Laminin through G3 and to Col IV through either G1 or G2 . We have previously shown that Drosophila Laminins are critical for proper assembly of other ECM components in the BM of embryonic tissues [33] . Furthermore , a recent study has uncovered a temporal hierarchy of expression of BM components in the Drosophila embryo , with Laminins being expressed first , followed by Col IV and then Perlecan [30] . This seems to be critical for proper formation of the BM around the embryonic VNC . Thus , while elimination of Laminins affects both Col IV and Perlecan deposition , Laminin incorporates in the absence of any of these two components and Perlecan requires Col IV [30] . The requirements of these BM proteins for Ndg incorporation into embryonic BMs are still unknown . Here , we decided to investigate this by analysing Ndg expression in embryos devoid of the other BM components . We found that depletion of LanB1 results in a strong reduction of Ndg accumulation in the gut , muscles and VNC ( Fig 7A , 7A’ , 7B and 7B’ ) . However , elimination of Col IV ( Fig 7C and 7C’ ) or Perlecan ( Fig 7D and 7D’ ) did not prevent Ndg deposition into embryonic BMs , except in the VNC midline pores of embryos lacking Col IV , where it is very much reduced . Next , we tested the requirements of Laminins , Col IV and Perlecan for Ndg incorporation into the BM of the larval fat body . To this end , we analysed the expression of the transgene Ndg . sGFP in the fat body of larvae where we had knocked down expression of BM components under the control of Cg-GAL4 driver . We found that the knock down of Laminins or Col IV , but not Perlecan , caused a reduction in the amount of Ndg in fat body BMs ( Fig 7E and 7F and S7 Fig ) , consistent with our functional analysis of the different Ndg domains ( Fig 6 ) . We additionally decided to analyze the mutual requirements of the remaining components of the adipose tissue BM . We found that loss of Col IV resulted in a strong reduction in Laminin levels and in a depletion of Perlecan ( Fig 7E and 7F and S7 Fig ) . This is in agreement with previous results showing that knocking down vkg with hsp70-GAL4 , which is a heat shock inducible promoter , reduced the presence of Nidogen and Laminin in fat body BM [54] . In contrast , and similar to the loss of Ndg ( Fig 4 ) , absence of Laminins led to holes in the BM without apparent reduction in Col IV or Perlecan levels ( Fig 7E and S7 Fig ) . We also noted that reduction of Col IV and to a lesser extent of Laminins resulted in changes in adipocyte morphology and cell rounding . Finally , knock down of Perlecan did not affect the presence of any of the other components , consistent with the notion that it is a terminal BM component ( Fig 7E and 7F ) [29] . In summary , these results show that Ndg incorporation into embryonic and fat body BMs depends on both Laminin and Collagen IV . They also suggest a model for the assembly and maintenance of the adipose tissue BM in which Ndg is not essential for the incorporation of other components , but reinforces the connection between the Laminin and Col IV networks , thus allowing correct formation of the BM or preventing its rupture ( Fig 7G ) . To finally ascertain whether Nidogen incorporation had a wider stabilizing role on BMs despite limited phenotypic defects in the mutants , we tested genetic interactions with other conditions compromising BM functionality . LanA216 and LanA160 are two homozygous lethal EMS-induced LanA alleles that in combination produce animals viable until pupal stages [55] . While LanA216/LanA160 3rd instar larvae showed an elongated VNC , no defects in VNC condensation were observed in LanA216/+ , LanA160/+ or Ndg1 animals . In contrast , we found that Ndg1 mutants heterozygous for LanA216/+ or LanA160/+ showed VNCs that were significantly more elongated than those found in Ndg1 LanA216/+ or LanA160/+ larvae ( Fig 8A and 8B ) . In addition , we found that Ndg interacted genetically with Perlecan . Thus , while single knock down of either Ndg or Perlecan in the whole fly , using actin-GAL4 , produced normal-looking pupae and viable adults , the double knock down of these genes caused a significant decrease in the size of pupae , which were unable to develop to adulthood ( Fig 8C and 8D ) . This genetic interaction was exacerbated when knock down was driven at 30°C , a temperature at which GAL4-driven transgene expression is higher [56 , 57] . In summary , these results prove that Nidogen interacts genetically with Laminins and Perlecan , suggesting a more general role of Nidogen in maintaining BM stability and consistent with its remarkable evolutionary conservation .
BMs are thin extracellular matrices that play crucial roles in the development , function and maintenance of many organs and tissues [58] . Critical for the assembly and function of BMs is the interaction between their major components , Col IV , Laminins , proteoglycans and Ndg [59] . Both the ability of Ndg to bind laminin and Col IV networks and the crucial requirements for Laminins and Col IV in embryonic development [60 , 61] anticipated a key role for Ndg during morphogenesis . However , experiments showing that elimination of Ndg in mice and C . elegans are compatible with survival casted doubt upon the crucial role for Ndg in organogenesis as a linker of the crucial Laminin and Col IV networks within the BM . Here , we have isolated mutations in the single Drosophila Ndg gene and found that , as it is the case in mammals and C . elegans , Ndg is not generally required for BM assembly and viability . However , Ndg mutant flies display mild motor or behavioral defects . In addition , similar to mammals , we show that the Nidogen-deficient flies show BM defects only in certain organs , suggesting tissue-specific roles for Ndg in BM assembly and maintenance . Finally , our functional study of the different Ndg domains challenges the significance of some interactions derived from in vitro experiments while confirming others and additionally revealing a new key requirement for the Rod domain in Ndg function and incorporation into BMs . Results from cell culture and in vitro experiments led to propose a crucial role for Ndg in BM assembly and stabilization . Recombinant Ndg promotes the formation of ternary complexes among BM components [62] . In addition , incubation with recombinant Ndg or antibodies interfering with the ability of Ndg to bind Laminins results in defects in BM formation and epithelial morphogenesis in cultured embryonic lung , submandibular glands and kidney [63 , 64] . However , elimination of Ndg in model organisms has shown that Ndg is not essential for BM formation per se but required for its maintenance in some tissues . Thus , while the early development of heart , lung and kidney prior to E14 is not affected in Nidogen-deficient mice , defects in deposition of ECM components and BM morphology were observed at E18 . 5 [11] . Similarly , whereas BM components localized normally in Nidogen-deficient mice during the early stages of limb bud development , this BM breaks down at later stages [12] . In contrast , removal of Ndg does not impair assembly or maintenance of any BM in C . elegans [14] . Here , we show that in Drosophila , as it is the case in mammals [11 , 12] , different BMs have different requirements for Ndg . Thus , while elimination of Ndg in Drosophila does not impair embryonic BM assembly or maintenance , it results in discontinuity of the BM in fat body and flight muscles . The basis for this tissue-specificity of Ndg requirements is currently unknown . Recent experiments have shown that there is a tissue-specific hierarchy of expression and incorporation of BM proteins in the Drosophila embryo , with Laminins being expressed first followed by Col IV and finally Perlecan [30] . Laminins and Col IV can reconstitute polymers in vitro that resemble the networks seen in vivo [32] [65] . In this context , Laminins and Col IV could self-assemble into networks in the embryo as they are produced , being this sufficient to assemble a BM capable of sustaining embryonic development in the absence of the two subsequent components , Ndg and Perlecan . We also show here that , while fat body and blood cells are the source of the majority of the proteins in larval BMs , there are notable exceptions , a fact that highlights a diversity in the origins of BM components in different tissues . Thus , fat body produces entirely all its BM , the larval heart receives it all from the hemolymph , imaginal discs produce a portion of their Laminins and similarly for tracheae with respect to Perlecan . These differences in the source of BM components for different tissues ( incorporated vs . self-produced ) may impose different assembly mechanisms , a possibility to study in more detail in the near future . In addition , although BM components are universally present in numerous tissues and organs , they are diverse depending on tissue and developmental stage ( reviewed in [66] ) . This heterogeneity arises from variations in protein subtypes , such as the two alternative Laminin α chains or the numerous Perlecan isoforms . Heterogeneity may also stem from differences in relative amounts of each component and posttranslational modifications thereof . In this respect , it is possible that BM assembly of the Drosophila fat body and adult flight muscles of the notum is such that is more dependent on Ndg function for its formation and stability than BMs found in other tissues . Finally , dynamics of BMs can orchestrate organ shape changes . Reciprocally , the associated tissues can control properties of BMs by , for instance , expressing a specific repertoire of ECM receptors or remodeling factors . In this context , it is also possible that fat body or adult flight muscles sculpt BMs with properties demanding a high requirement of Ndg function . We find here that Ndg mutant flies are less fertile and behave differently with respect to wild type in ChillComa Recovery Time assays . The physiological mechanisms underlying the response in insects to critical thermal limits remain largely unresolved . The onset and recovery of chill coma have been attributed to defects in neuromuscular function due to depolarization of muscle fiber membrane potential [67] . Interestingly , flight muscle fiber membrane is strongly depolarized upon exposure to low temperatures in Drosophila [67] . In this context , the defects we observed in the BM of adult flight muscles in the absence of Ndg could be behind the defective response of Ndg mutant flies to chill coma recovery assays . Altogether , these results show that , though not critical for survival , Ndg is required for overall fitness of the fly . All Nidogen proteins consist of three globular domains , G1 to G3 , and two connecting segments , one Rod domain separating G2 and G3 and a flexible linker between G1 and G2 . Crystallographic and binding epitope analyses using recombinant domains of the mouse Nidogen-1 protein have demonstrated high affinity binding of domain G2 to Col IV and Perlecan , of domain G3 to the Laminin γ1 chain and Col IV , and no activity for the Rod domain [4–7 , 68] . In addition , recent physicochemical studies analyzing the solution behavior of full length purified Nidogen-1 confirmed the formation of a high affinity complex between the G3 domain of Nidogen-1 and the Laminin γ1 chain , and excluded cooperativity effects engaging neighboring domains of both proteins [69] . However , little is known about the functional meaning of the binding abilities of Ndg on its localization and function in BM assembly in vivo . In fact , mutant C . elegans animals carrying a deletion removing the entire G2 domain of NID-1 are viable and show no defects on Ndg or Col IV localization in BMs [14] . These results demonstrate that , despite the strong sequence conservation between C . elegans and mammalian G2 domains , C . elegans NID-1 localization appears to occur independently of this domain . Here , we show that , as it is the case in C . elegans , the Drosophila G2 domain is not essential for neither Ndg localization nor function . A possible explanation for this result is that although some of the modules present in BM components are conserved , there might be variations in sequence and structure that might be sufficient to confer binding specificity to the different proteins . For instance , the IG3 domain of mouse Perlecan , which binds to a β-barrel in the G2 domain of Nidogen , is strikingly conserved in all mammals , but not in Drosophila or C . elegans [70 , 71] . This result suggests that either the Perlecans present in these organisms are too distant in evolution from the mouse proteins for these domains to be conserved or that Perlecans may only bind Nidogen in mammals . Previous studies aimed to characterize the biological significance of the Nidogen-Laminin interactions have targeted the Nidogen-binding module of the Laminin γ1 chain , showing that this domain is required for kidney and lung organogenesis [63] [72] . However , the role of the Nidogen G3 domain has not yet been addressed directly . Here , we show that the G3 domain is essential for Ndg localization , supporting a role for Nidogen-Laminin interactions on Ndg function . In addition , in contrast to what has been shown in mammals ( see above ) , our results unravel a key role for the Rod domain in Nidogen localization . Again , an explanation for this result could hinge on variations in Nidogen between species . In fact , one of the major differences between Drosophila and mammalian Nidogen lies on the Rod domain . Thus , while vertebrates have four EGF repeats and one or two thyroglobulin repeats , Drosophila and C . elegans have 12 and 11 EGF repeats , respectively . Alternatively , conclusions derived from in vitro studies may not be always applicable to the circumstances occurring in the living organism . Furthermore , the appearance of new in vitro studies combining different techniques has revealed the existence of multiple Nidogen-1/Laminin γ1 interfaces , which include , besides the known interaction sites , the Rod domain [68] . Different BM assembly models have been proposed over the last thirty years . Based upon biochemical studies and rotary shadow electronic microscopic visualization , the BM assembly model firstly proposed that Collagen IV self-assembles into an initial scaffold , followed by Laminin polymerization structure attachment mediated by Perlecan [73 , 74] . However , more recent studies have postulated a contradicting model for in vivo systems . The most widely endorsed model states that the polymer structure is initiated by a Laminin scaffold built through self-interaction , bridged by Nidogen and Perlecan and finally completed by another independent network formed by Col IV self-interaction [4] . Here , we studied in detail the hierarchy of BM assembly in the Drosophila larval fat body . Thus , while the requirements for Drosophila Laminins in the incorporation of other ECM components into BMs are preserved between tissues , this is not the case for Collagen IV . For instance , absence of Col IV does not completely prevent deposition of Laminin in the fat body , but remarkably reduces it ( Fig 7E ) ; in contrast , no such drastic effect has been observed in wing discs or embryonic BMs [29 , 30] , suggesting that Collagen IV does not affect Laminin incorporation in these other tissues to the same degree or that it does not affect it at all . In addition , we found that BM assembly in Drosophila also differs from that in mammals and C . elegans . In this case , the divergences may arise during evolution , when different organisms might have incorporated novel ways to assemble ECM proteins to serve new specialized functions . Nidogen has been proposed to play a key role in BM assembly based on results from in vitro experiments and on its ability to serve as a bridge between the two most abundant molecules in BMs: Laminin and Type IV Collagen . However , phenotypic analysis of its knock out in mice and C . elegans have called into question a general role for Nidogen in BM formation and maintenance . Here , we show that although Ndg is dispensable for BM assembly and preservation in many tissues , it is absolutely required in others . These differences on Ndg requirements stress the need to analyze its function in vivo and in a tissue-specific context . In fact , we believe this should also be the case when analyzing the requirements of the other ECM components for proper BM assembly , as we show here they also differ between species and tissues . One has to be cautious when inferring functions of different BM proteins or their domains based on experiments performed in vitro or in a tissue-specific setting . This might be especially relevant when trying to apply conclusions derived from these studies to our understanding of the pathogenic mechanisms of BM-associated diseases or to the development of innovative therapeutic approaches .
Standard husbandry methods and genetic methodologies were used to evaluate segregation of mutations and transgenes in the progeny of crosses [75] . The following stocks were used: The FTG , CTG and TTG balancer chromosomes , carrying twist-Gal4 UAS-2EGFP , were used to identify homozygous NdgΔRod-G3 mutants [76] . For the generation of Ndg deficiencies the following stocks were used ( all from Bloomington Drosophila Stock Center ) : Mi{ET1}NdgMB12298 , w; BlmN1/TM3 , Sb1 [77] , w; Sco/Sm6aP ( hsILMiT ) 2 , 4 , w; Gla/CyO , Df ( 2L ) BSC172 [29] and Df ( 2R ) BSC281 . w; tub-GAL80ts , y v; UAS-trol . RNA TRiP . JF03376 , y v; Ndg . RNAiTRiP . HMJ24142 , y v sc; UAS-LanB2 . RNAiTRiP . HMC04076 , y v; UAS-LanA . RNAiTRiP . JF02908 , w; UAS-GFP . S65T ( BDSC 1522 ) , w; en2 . 4-GAL4 UAS-mCherry . NLS ( BDSC 38420 ) and y v sc; UAS-EGFP . shRNA ( BDSC 41560 ) are from Bloomington Drosophila Stock Center . w; Ndg . sGFPfTRG . 638 , w; LanB1 . sGFPfTRG . 638 , w; UAS-LanB1 . RNAiVDRC . v23121 , w; UAS-trol . RNAiVDRC . v24549 and w; UAS-Cg25C . RNAiVDRC . v28369 were obtained from Vienna Drosophila Resource Center . y w;vkgG454 . GFP , w trolZCL1700 . GFP and w trolCPTI-002049 . YFP were from Drosophila Genomics Resource Center . w;UAS-vkg . RNAiNIG . 16858R-3 was from National Institute of Genetics . Other strains used were: Df LanB1/CTG [33] , Df ( 3R ) BSC524/CTG [78] , trolnull/FMZ [79] , w; UAS-Cg25C . RFP3 . 1 [45] and croc-lacZ [42] . w;LanA160/TM6B , and w; LanA216/TM6B are gifts from Luis Garcia-Alonso . w; UAS-secreted . GFP is a gift from Fujian Zhang . Lines generated in this study are: w; sGFPRNAi . attP40 , w; Ndg1 , w; Ndg2 , NdgΔRod-G . 3 . 1 , NdgΔRod-G . 3 . 2 , NdgΔRod-G . 3 . 3 . w; UAS-NdgFL . GFP , w; UAS-NdgG1 . GFP , w; UAS-NdgG2 . GFP , w; UAS-NdgG3 . GFP , w; UAS-NdgRod . GFP , w; UAS-NdgRodG3 . GFP , w; UAS-NdgG1L . GFP , w; UAS-NdgG1LG2 . GFP , w; UAS-NdgΔG3 . GFP , w; UAS-NdgΔG2R . GFP , w; UAS-NdgΔG2 . GFP , w; UAS-NdgΔG1 . GFP , w; UAS-NdgG2R . GFP , w; UAS-NdgL . GFP and w; UAS-NdgΔR . GFP . The description of all lines used in this study is available in Supplementary information S2 Table . The GAL4-UAS system was used to drive expression of transgenes and RNAi constructs in larval fat body and hemocytes ( blood cells ) under control of Cg-GAL4 ( BDSC 7011 ) or BM-40-SPARC-GAL4 ( gift from Hugo Bellen ) , and ubiquitously with act-GAL4 . For Collagen IV knock down experiments ( vkgi and Cg25Ci ) , thermosensitive GAL4 repressor GAL80ts was used to prevent embryonic lethality . Cultures were grown at 18°C for 6 days , followed by transfer of cultures to 30°C ( L2 stage ) and dissection two days later ( L3 stage ) . For the structure/function analysis of Ndg domains , deletion of specific domains of Ndg was achieved by PCR-amplifying plasmid PDONR211-Ndg with the appropriate combinations of the following primers: NdgG1 . GFP-Forward: TGAGAACGAGGACCCAGCTTTCTTGTACAAAG NdgG1 . GFP-Reverse: AAGCTGGGTCCTCGTTCTCAATGGGAGCCAC NdgG2 . GFP-Forward: GGTCAGCGGAGCTAATGATCAACCTATCCGAGTG NdgG2 . GFP-Reverse: GATCATTAGCTCCGCTGACCAGGATCACCGAG NdgG3 . GFP-Forward: CAGCGGACAGCGTCCCATTTCGGTGGCCC NdgG3 . GFP-Reverse: AAATGGGACGCTGTCCGCTGACCAGGATCAC NdgRod . GFP-Forward: GACATTACGTGACCCAGCTTTCTTGTAC NdgRod . GFP-Reverse: AAGCTGGGTCGACATTACGTCCGTTTAG NdgRodG3 . GFP-Forward: GGTCAGCGGAAACGATGGTACCGCCGATTG NdgRodG3 . GFP-Reverse: TACCATCGTTTCCGCTGACCAGGATCACCGAG NdgG1L . GFP-Forward: GTCCTGCCTGTACGACCCAGCTTTCTTGTACAAAG NdgG1L . GFP-Reverse: CTGGGTCGTACAGGCAGGACTTTCCATTGCC NdgG1LG2 . GFP-Forward: AAATGGGACGCAGGCAGGACTTTCCATTGCC NdgG1LG2 . GFP-Reverse: CTGGGTCGTAATCGTTGCAGGCATTCGATTCGGG NdgΔG3 . GFP-Forward: GACATTACGTGACCCAGCTTTCTTGTAC NdgΔG3 . GFP-Reverse: AAGCTGGGTCGACATTACGTCCGTTTAG NdgΔG2R . GFP-Forward: GTCCTGCCTGCGTCCCATTTCGGTGGCCCAG NdgΔG2R . GFP-Reverse: AAATGGGACGCAGGCAGGACTTTCCATTGCC NdgΔG2 . GFP-Forward: GTCCTGCCTGAACGATGGTACCGCCGATTG NdgΔG2 . GFP-Reverse: TACCATCGTTCAGGCAGGACTTTCCATTGCC NdgΔG1 . GFP-Forward: GGTCAGCGGAGAGCAGAACGTGAGGTCTCCC NdgΔG1 . GFP-Reverse: CGTTCTGCTCTCCGCTGACCAGGATCACCGAG NdgG2R . GFP-Forward: GGTCAGCGGAGCTAATGATCAACCTATCCG NdgG2R . GFP-Reverse: GATCATTAGCTCCGCTGACCAGGATCACCG NdgL . GFP-Forward: GGTCAGCGGAGAGCAGAACGTGAGGTCTCC NdgL . GFP-Reverse: CGTTCTGCTCTCCGCTGACCAGGATCACCG NdgΔR . GFP-Forward: GAATGCCTGCCGTCCCATTTCGGTGGCCCA NdgΔR . GFP-Reverse: AAATGGGACGGCAGGCATTCGATTCGGGGG The resulting PCR reactions were incubated with 10 units of DMT enzyme ( TransGen Biotech , Beijing , China ) at 37°C for 1 hour to digest the original templates . After digestion , PCR products were transformed into DMT competent cells ( TransGen Biotech , Beijing , China ) . Colonies were validated by sequencing . Transgenic lines were obtained through standard P-element transgenesis [82] . The Mi{ET1}NdgMB12298 transposon was used in a Blm mutant background to generate deficiencies by imprecise excision of the transposon . In these mutants , homologous recombination DNA reparing enzymes are compromised , thus increasing the events of non-homologous recombination DNA repair . Non-homologous recombination DNA repair increases the chances of generating DNA deficiencies [77] . We selected 132 Blm mutant males carrying the Mi{ET1}Ndg[MB12298] transposon and crossed them to w; Gla/CyO females . The offspring of this cross rendered a 110 EGFP negative males that were crossed to the Df ( 2R ) BSC281 deficiency . 6 out of the 110 males did not complement the deficiency and were selected for further molecular characterization with the following primers from the Ndg genomic region . PCR primers were used as follows: ( 5’-3’ ) Ndg primer1-Forward: GTGTGGACTCGGTGTGACTG Ndg primer1-Reverse: ACTTCGAACAGCCAGACTCC Ndg primer2-Forward: CCTTCGGCAGTAAGTTGCTC Ndg primer2-Reverse: GTGCTGTTGGACAGACAACG Ndg primer3-Forward: CGATCAAGCGGCGCAATATC Ndg primer3-Reverse: CCAACATGCCACAATGGGTG Ndg primer4-Forward: GTCTGAGTGGTTTCGGCAC Ndg primer4-Reverse: TTTGCTTAAAGTGGGTGTTGC Ndg primer5-Forward: CCATTGTGGCATGTTGGATA Ndg primer5-Reverse: TGTTTCGAAGGCGATACTCA Ndg primer6-Forward: AAACTGAAAAAGCGGGGAAT Ndg primer6-Reverse: TTAATCAGTGCACCGCAGAG Ndg primer7-Forward: GATGAAGGAGGCAAAGCAAG Ndg primer7-Reverse: TTTTCATCTGCAGTGCGTTC Ndg primer8-Forward: GAGGAGCAGATACCCCAACA Ndg primer8-Reverse: CAGTGCCGTCATATTTGGTG Ndg primer9-Forward: GGATTCAGAGGCGATGGATA Ndg primer9-Reverse: GACCAGTTCCGTCCAGGTTA Ndg primer10-Forward: TTTCTGCCAGTTTTCGCTTT Ndg primer10-Reverse: CGTGTTGTTGGATTGTGGAG Ndg primer11-Forward: GTGCTGTGCCTCAGATGAAA Ndg primer11-Reverse: GGGAACCCAATGTGCTTAGA Ndg primer12-Forward: TTACCTTCACGCACGATCAG Ndg primer12-Reverse: GGCTGCGGCATTAGAGATAC All deficiencies eliminated the 5’ UTR and the first exon of the Ndg gene and at least two adjacent genes: Obp46a and CG12909 ( S2 Fig ) . Four sgRNAs were designed for generating Ndg null mutant lines [83] . sgRNAs and cas9 mRNA were injected into w1118 embryos . Ndg deletions in the germ line Ndg1 and Ndg2 were selected by sequencing by Beijing Fungene Biotechnology ( Beijing , China ) . sgRNA1: GAGAGATACACAAGTCAGGAAGG sgRNA2: CCAGCCCTTTCCGCTGGAATATGC sgRNA3: GCGGCCTTCTACTCGAACGTGG sgRNA4: GCCATTTGCAAGTGGGACTCGG For assessment of Ndg mRNA expression in Ndg1 mutants was assessed by quantitative real-time PCR . RNA was extracted using TRIzol reagent ( Life technologies , USA ) . cDNA was synthesized from 2 μg of RNA with PrimeScript RT-PCR Kit ( Takara , Kyoto , Japan ) . Analysis was performed in a CFX96 Touch system ( Bio-Rad , California , USA ) using iTaq Universal SYBR Green Supermix ( Bio-Rad ) . rp49 was used as a reference for normalization . Three experiments per genotype were averaged . The following intron-spanning pairs of primers were used: Ndg primerA-Forward: GAGCAGTACGAGCAGCT Ndg primerA-Reverse: CGAGTAGAAGGCCGCTAT Ndg primerB-Forward: ATCCATATCCTGAGGAGCAGAT Ndg primerB-Reverse: GGTGCAGGTGTAGCCAT Ndg primerC-Forward: AGTGCCGTTCGACCAATT Ndg primerC-Reverse: GACAATCAGGAAGTCAGAGT Ndg primerD-Forward: GACTCAGCAAAGGATACCAT Ndg primerD-Reverse: CAGTCCGACCAGAACAGTT rp49 primer-Forward: GGCCCAAGATCGTGAAGAAG ′ rp49 primer-Reverse: ATTTGTGCGACAGCTTAGCATATC To generate Ndg mutants carrying a deletion of the rod and G3 domains , one single guide ( sgRNA ) target was designed in the 5th exon of Ndg: sgRNA5: GGGGAATGCCGATGCCCCTATGG The sgRNAs were cloned in the PCFD3 vector as previously described in [84] and http://www . crisprflydesign . org/plasmids/ . Transgenic gRNA flies were created by the Best Gene Company ( Chino Hills , USA ) using either y sc v P{nos-phiC31\int . NLS}X; P{CaryP}attP2 ( BDSC 25710 ) or y v P{nos-phiC31\int . NLS}X; P{CaryP}attP40 ( BDSC 25709 ) . Transgenic lines were verified by sequencing by Biomedal Company ( Armilla , Spain ) . Males carrying the sgRNA were crossed to females either act-Cas9 or nos-Cas9 and the progeny was screened for the v+ch- eye marker . To identify CRISPR/Cas9-induced mutations , genomic DNA was isolated from flies and sequenced using the following primers: ( 5’-3’ ) Ndg primerg5-Forward: GCGAAGTTTGGGAGAACGGA Ndg primerg5-Reverse: ACAGTATCTCACTCAGATCGGC For generation of anti-Nidogen antibody , rabbits were immunized with epitope CTYVQEFDGERNADLIPC by Bio-med Biotechnology ( Beijing , China ) . Embryos , fat bodies , wing imaginal discs and ovaries were stained using standard procedures and mounted in DAPI-Vectashield ( Vector Laboratories , Burlingame , California ) . The following primary antibodies were used: rabbit anti-Ndg ( 1:2000 , this study ) , chicken anti-betagalactosidase ( 1:500 , AbCam , Cambridge , UK ) , chicken anti-GFP ( 1:500 , AbCam ) , rabbit anti-Ndg ( 1:100 , [34] ) . Secondary antibody is IgG conjugated to Alexa-555 , IgG conjugated to Alexa-488 and Alexa 549 ( 1:200 , Life technologies ) . For lipid droplet staining , L3 larvae were turned inside out and fixed in 4% PFA for 20 minutes , washed twice in PBS and then incubated in a 1:1000 dilution in PBS of 1 mg/ml BODIPY 493/503 stock ( Life Technologies ) for 30 minutes , followed by two 10-min washes in PBS and mounting in DAPI-Vectashield ( Vector Laboratories ) . Confocal images were obtained using a Leica ( Wetzlar , Germany ) SP2 microscope or a Zeiss ( Oberkochen , Germany ) LSM780 microscope equipped with a Plan-Apochromat 63X oil objective ( NA 1 . 4 ) . Eggs and pupae were imaged in a Leica M125 stereoscope . All images were processed with Adobe Photoshop and ImageJ . For quantification of lipid droplet diameter ( Fig 4E ) , confocal micrographs of 3rd instar larval fat body stained with BODIPY were analyzed with the automated particle detection tool of Nikon NIS-Elements AR 5 . 0 software . Sixteen adipocytes per genotype were analyzed and only particles larger than 3 μm in diameter were counted . For quantification of egg laying ( S4A Fig ) , five 2-day old virgins were transferred to fresh vials daily for ten days and the eggs laid on each vial counted . Three such experiments were conducted per genotype . For calculation of egg aspect ratio ( S4B Fig; [85] ) length and width of eggs were measured on images using the line tool in FIJI-ImageJ . Aspect ratio is defined as egg length divided by width . In chill coma recovery time assays ( S4D Fig; [51] ) , 2-day old females were placed into 10 mL tubes . These tubes were submerged into an ice-water bath for 2 hours , resulting in paralyzed flies . The amount of time required for a fly at room temperature to stand after becoming paralyzed in this way was measured . For quantification of fluorescence intensity of different Ndg . GFP constructs in fat body BM ( Fig 6B and 6C ) , GFP signal was measured on 4–6 confocal images per genotype using FIJI-ImageJ . Each measurement represents mean value intensity inside a 500 μm2 square drawn on a flat portion of BM of an individual fat body cell , avoiding measuring intensity in cell contacts . For colocalization analysis ( Fig 6G ) , 63x confocal images of fat body were analyzed . Pearson’s correlation coefficients after automated Costes thresholding ( R coloc ) were calculated with the FIJI-ImageJ plugin Colocalization Threshold . Each data point in the graph represents one image containing several fat body cells , like those in Fig 6D . VNC length ( Fig 8B ) was measured on confocal images using the segmented line tool of FIJI-ImageJ . For quantification of pupal length ( Fig 8D ) , stereoscope images of pupae were measured using the line tool of FIJI-ImageJ . Each data point in the graph represents one pupa . Graphpad Prism software was used for graphic representation and statistical analysis . For measurements of lipid droplet diameter ( Fig 4E ) , a non-parametric Mann-Whitney test was used . For statistical comparisons of fluorescence intensity in Fig 6B and 6C , unpaired Student’s t tests were used in LanAi+NdgΔG3 . GFP , Cg25Ci+NdgFL . GFP , Cg25Ci+NdgΔG1 . GFP and Cg25Ci+NdgΔG2 . GFP experiments ( data passed D’Agostino & Pearson normality tests and F-tests for equal variance ) . Student’s t tests with Welch’s correction were used for LanAi+NdgFL . GFP , LanAi+NdgΔG1 . GFP and LanAi+NdgΔG2 . GFP experiments ( data passed D’Agostino & Pearson normality tests , but not F-tests for equal variance ) . A non-parametric Mann-Whitney test was used in Collagen IVi+NdgΔG1 . GFP experiment ( data did not pass D’Agostino & Pearson normality test ) . For comparisons of VNC length in Fig 8B and pupal length in Fig 8D , we performed non-parametric Mann-Whitney tests . For egg production curves in S4A Fig , we conducted non-parametric Kolmogorov-Smirnov tests . For comparison of aspect ratio in S4B Fig , we performed unpaired two-tailed Student’s t tests . For comparison of chill coma recovery time in S4C Fig , Student’s t-tests with Welch’s correction were used . Significance of statistical tests is reported in graphs as follows: **** ( p < 0 . 0001 ) , *** ( p < 0 . 001 ) , ** ( p < 0 . 01 ) , * ( p < 0 . 05 ) , n . s . ( p > 0 . 05 ) .
|
Basement membranes ( BMs ) are thin layers of specialized extracellular matrices present in every tissue of the human body . Its main constituents are two networks of laminin and Type IV Collagen linked by Nidogen ( Ndg ) and proteoglycans . They form an organized scaffold that regulates organ morphogenesis and function . Mutations affecting BM components are associated with organ dysfunction and several congenital diseases . Thus , a better comprehension of BM assembly and maintenance will not only help to learn more about organogenesis but also to a better understanding and , hopefully , treatment of these diseases . Here , we have used the fruit fly Drosophila to analyse the role of Ndg in BM formation in vivo . Elimination of Ndg in worms and mice does not affect survival , strongly questioning its proposed linking role , derived from in vitro experiments . Here , we show that in the fly , Ndg is dispensable for BM assembly and preservation in many tissues , but absolutely required in others . Furthermore , our functional study of the different Ndg domains challenges the significance of some interactions between BM components derived from in vitro experiments , while confirming others , and reveals a new key requirement for the Rod domain in Ndg function and incorporation into BMs .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"blood",
"cells",
"invertebrates",
"caenorhabditis",
"animals",
"collagens",
"animal",
"models",
"developmental",
"biology",
"caenorhabditis",
"elegans",
"drosophila",
"melanogaster",
"model",
"organisms",
"experimental",
"organism",
"systems",
"embryos",
"drosophila",
"research",
"and",
"analysis",
"methods",
"embryology",
"lipids",
"animal",
"studies",
"proteins",
"fats",
"animal",
"cells",
"life",
"cycles",
"insects",
"arthropoda",
"biochemistry",
"eukaryota",
"cell",
"biology",
"protein",
"domains",
"nematoda",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"larvae",
"organisms"
] |
2018
|
Dissection of Nidogen function in Drosophila reveals tissue-specific mechanisms of basement membrane assembly
|
During Alzheimer's Disease , sustained exposure to amyloid-β42 oligomers perturbs metabolism of ether-linked glycerophospholipids defined by a saturated 16 carbon chain at the sn-1 position . The intraneuronal accumulation of 1-O-hexadecyl-2-acetyl-sn-glycerophosphocholine ( C16:0 PAF ) , but not its immediate precursor 1-O-hexadecyl-sn-glycerophosphocholine ( C16:0 lyso-PAF ) , participates in signaling tau hyperphosphorylation and compromises neuronal viability . As C16:0 PAF is a naturally occurring lipid involved in cellular signaling , it is likely that mechanisms exist to protect cells against its toxic effects . Here , we utilized a chemical genomic approach to identify key processes specific for regulating the sensitivity of Saccharomyces cerevisiae to alkyacylglycerophosphocholines elevated in Alzheimer's Disease . We identified ten deletion mutants that were hypersensitive to C16:0 PAF and five deletion mutants that were hypersensitive to C16:0 lyso-PAF . Deletion of YDL133w , a previously uncharacterized gene which we have renamed SRF1 ( Spo14 Regulatory Factor 1 ) , resulted in the greatest differential sensitivity to C16:0 PAF over C16:0 lyso-PAF . We demonstrate that Srf1 physically interacts with Spo14 , yeast phospholipase D ( PLD ) , and is essential for PLD catalytic activity in mitotic cells . Though C16:0 PAF treatment does not impact hydrolysis of phosphatidylcholine in yeast , C16:0 PAF does promote delocalization of GFP-Spo14 and phosphatidic acid from the cell periphery . Furthermore , we demonstrate that , similar to yeast cells , PLD activity is required to protect mammalian neural cells from C16:0 PAF . Together , these findings implicate PLD as a potential neuroprotective target capable of ameliorating disruptions in lipid metabolism in response to accumulating oligomeric amyloid-β42 .
Perturbations in glycerophosphocholine ( GPC ) metabolism are linked to the pathogenesis of Alzheimer's Disease ( AD ) with the accumulation of choline-containing lipids in AD patients associated with accelerated cognitive decline [1]–[4] . Soluble amyloid-β42 ( Aβ42 ) oligomers can increase hydrolysis of structural membrane lipids by activating cytosolic phospholipase A2 ( cPLA2 ) , a Group IVa PLA2 that preferentially hydrolyzes arachidonic acid from the sn-2 position of 1-O-alkyl-2-arachidonoyl- and 1-O-acyl-2-arachidonoyl- glycerophospholipids [2] , [5] , [6] . Sustained activation results in arachidonic acid signaling cascades , as well as the intraneuronal accumulation of choline-containing second messengers [1] , [2] , [5] . The fate and functions of these GPC second messengers are of particular interest . Recent evidence points to the accumulation of specific choline-containing metabolites [1] , [2] , [5] . Underlying mechanisms have yet to be fully elucidated . PC ( O-16:0/2:0 ) platelet activating factor ( PAF ) species are elevated in AD brain and human neurons exposed to Aβ42 [1] . Of these elevated species , 1-O-hexadecyl-2-acetyl-sn-glycerophosphocholine ( C16:0 PAF ) , but not its immediate precursor 1-O-hexadecyl-sn-glycerophosphocholine ( C16:0 lyso-PAF ) , is implicated in Aβ42 toxicity [1] . Rising intraneuronal concentrations of C16:0 PAF activate an endoplasmic reticulum ( ER ) -stress signaling cascade leading to the hyperphosphorylation of tau that ultimately compromises neuronal viability [1] . Molecular and pharmacological approaches designed to promote the hydrolysis of C16:0 PAF to C16:0 lyso-PAF and/or block downstream signaling are sufficient to inhibit Aβ-mediated neurotoxicity [7]–[10] . These findings underscore the importance of this lipid species in Alzheimer's disease and emphasize the rationale for identifying key targets involved in shielding its toxic effects and/or promoting its hydrolysis and inactivation . Unbiased approaches exploiting the cross-species conservation of biochemical pathways between man and the budding yeast Saccharomyces cerevisiae have proven to be successful in elucidating the mode-of-action of various compounds [11] . Further , S . cerevisiae has also been used as a model organism to delineate key aspects of eukaryotic lipid metabolism and to investigate various neurodegenerative diseases [12] , [13] . Despite conservation of the PAF metabolic pathway genes and detection of PAF species in yeast [14] , the function of the different species in yeast is currently unclear , although PAF species are suggested to play a role in cell cycle progression [15] , [16] . Similar to the effects of PAF on mammalian cells , yeast cells treated with PAFs and structurally related alkylacylglycerophosphocholine analogs induce disruptions in lipid metabolism and reduce viability [17] , [18] . Interestingly , S . cerevisiae strains harboring deletions for enzymes involved in phosphatidic acid ( PA ) metabolism , including phospholipase D ( SPO14 ) and glycerol 3-phosphate acyltransferase ( SCT1 ) , exhibit increased susceptibility to PAF species and other choline-containing lipids suggesting an essential role for PA in mediating the toxic effects of PAFs [17] . Whether the toxic effects of alkylacylglycerolipids are solely dependent upon SPO14 and similar pathways impinging upon PA metabolism or whether other aspects of cellular metabolism are involved has not been systematically assessed at a genome-wide level . Moreover , it remains unclear why PAF second messengers accumulate in AD tissue without compensatory metabolism . To identify additional requisite proteins and/or pathways which serve to regulate the cytotoxic affects of C16:0 PAF , we performed genome-wide yeast chemical genomic screens with both C16:0 PAF ( pathogenic in AD ) and C16:0 lyso-PAF ( non-pathogenic in AD ) . We found that the two PAF species identify largely distinct chemical genetic interactions and that the deletion mutant that exhibited the greatest differential sensitivity to the pathogenic C16:0 PAF species was YDL133w , a previously uncharacterized ORF encoding a putative transmembrane protein with unknown function . Upon subsequent investigation we determined that Ydl133w physically interacts with Spo14 and is required for PLD catalytic activity in S . cerevisiae . Importantly , Ydl133w , here in referred to as Srf1 ( Spo14 Regulatory Factor 1 ) , represents the only reported regulator of Spo14 required for PLD catalytic activity in mitotic cells . We report that though C16:0 PAF does not impact global PLD activity , it does cause the delocalization of Spo14 and PA from the periphery . Importantly our observations can be extended from the yeast model system as PLD activity in mammalian cells was found to confer protection against the toxic effects of C16:0 PAF .
Previous studies have determined that , similar to neuronal cells [1] , S . cerevisiae are sensitive to C16:0 PAF [17] . To determine whether C16:0 PAF and C16:0 lyso-PAF differentially impact the growth of S . cerevisiae and to identify an appropriate working concentration range for these lipids in subsequent studies we performed liquid growth curve analysis using wild type haploid yeast cultured with increasing concentrations of C16:0 PAF , C16:0 lyso-PAF or ethanol ( carrier control ) . As expected , C16:0 PAF inhibited wild type haploid yeast growth in liquid culture in a concentration-dependent manner whereas C16:0 lyso-PAF was found to be comparatively less toxic at similar concentrations ( Figure 1 ) . Although both lipids impact viability at higher concentrations , the distinct effects of these two lipids at lower concentrations have not previously been appreciated and parallels their toxicity to neuronal cells [1] , [19] . To identify critical proteins and/or pathways which are involved in regulating the cytotoxic effects of C16:0 PAF , the yeast haploid deletion mutant array ( DMA ) was robotically pinned onto agar plates containing either ethanol or a sublethal concentration of C16:0 PAF ( 120 µM ) . To facilitate the identification of potentially AD-relevant pathways we also screened with equimolar concentrations of C16:0 lyso-PAF ( 120 µM ) to distinguish between pathways that regulate AD-associated ( C16:0 PAF ) and non-associated ( C16:0 lyso-PAF ) phenotypes . The screen was performed in triplicate and 90 strains displaying putative increased sensitivity to C16:0 PAF or C16:0 lyso-PAF were further subjected to quantitative liquid growth curve measurements in the presence of C16:0 PAF , lyso-PAF , or ethanol ( Table S1 ) . A reduced concentration of 40 µM was employed in liquid cultures as this was found to cause moderate growth inhibition in wild type cells , thereby permitting quantitative analysis of the sensitivity of the individual strains to C16:0 PAF and C16:0 lyso-PAF relative to ethanol treatment using logistic growth curve analysis ( LGCA , see Material and Methods and Text S1 for details ) . This rigorous methodology accounts for the repeated measurements of individual liquid cultures and also exploits full , sigmoidal growth curves , since it does not assume exponential growth . Furthermore , we also normalize for plate and plate-treatment effects . LGCA revealed 13 deletion mutants that were hypersensitive to at least one PAF species , compared to the wild type response: ten strains were hypersensitive to C16:0 PAF , five strains were hypersensitive to C16:0 lyso-PAF , and two overlapping strains were hypersensitive to both ( Bonferroni corrected p-value <0 . 04; Figure 2A , Table S1 ) . As expected , the spo14Δ mutant was hypersensitive to C16:0 PAF as has previously been described [17] . sct1Δ mutants have also been reported to be hypersensitive to C16:0 PAF [17] , and though it did not make our stringent cut-off , the sct1Δ mutant was ranked 13th in sensitivity to C16:0 PAF ( Table S1 ) . Using LGCA we also assessed the differential sensitivity of each deletion mutant to C16:0 PAF versus C16:0 lyso-PAF . At the 40 µM treatment level , wild type cells exhibited approximately the same growth inhibition when exposed to either PAF species and we defined differential sensitivity as a departure from the status quo . Eleven deletion mutants exhibited differential sensitivity to one of the PAF species ( Bonferroni-corrected p-value <0 . 04 , Figure 2B and Table S1 ) . Of the two mutants that were identified in both screens , agp2Δ cells were equally sensitive to both lipids whereas taf14Δ displayed greater sensitivity to C16:0 PAF . Though LGCA identified ccs1Δ as being sensitive to C16:0 PAF its differential sensitivity was not significant . The differential sensitivity of four strains was striking . Namely , nup84Δ cells displayed the greatest differential sensitivity to C16:0 lyso-PAF , whereas srf1Δ , snf6Δ and spo14Δ were significantly more sensitive to C16:0 PAF than C16:0 lyso-PAF . The largely distinct chemical genetic profiles for C16:0 PAF and C16:0 lyso-PAF indicates that these related alkylacylglycerophospholipids impact upon distinct cellular pathways in yeast . Our results indicate that at the 40 µM treatment level , Srf1 is pivotal for buffering the effects of C16:0 PAF . The biological function ( s ) of Srf1 is unknown but it is predicted to be a transmembrane protein . Therefore we sought to decipher its cellular function by identifying proteins that interact with Srf1 . As traditional tandem affinity purification ( TAP ) protocols were not successful in purifying Srf1-TAP [data not shown and ref . 20] , [21] , we utilized a less stringent single step affinity purification approach based on the modified chromatin immunopurification ( mChIP ) technique [22] . Though this technique was developed for improving the purification of insoluble chromatin associated proteins , it is also applicable to other subclasses , including membrane associated proteins [23] . Using mChIP we successfully purified Srf1-TAP and identified five co-purifying proteins by mass spectrometry , of which the largest number of peptides correspond to Spo14 ( Figure 3 ) . The physical interaction between Srf1 and Spo14 , combined with the sensitivity of the corresponding deletion mutants to C16:0 PAF [Figures 2 , 4 and ref . 17] suggest Srf1 may work in a complex with Spo14 to regulate PA metabolism . To determine if Srf1 and Spo14 function together or work in parallel pathways , we compared the sensitivity of the srf1Δspo14Δ double mutant to C16:0 PAF with that of the single mutants ( Figure 4A ) . In agreement with our chemical genomic screen , srf1Δ cells display greater sensitivity to C16:0 PAF than spo14Δ cells . Interestingly , deletion of both SRF1 and SPO14 did not result in an additive increase in C16:0 PAF sensitivity but rather , the double mutant and spo14Δ exhibited similar sensitivity to C16:0 PAF . This indicates that spo14Δ is epistatic to srf1Δ and is in agreement with the hypothesis that Spo14 and Srf1 are in a complex and not in parallel pathways . In light of the physical interaction between Srf1 and Spo14 , one interpretation of this result is that Spo14 activity is misregulated in the absence of Srf1 . To further examine this possibility , we overexpressed SPO14 in wild type , srf1Δ and spo14Δ strains ( Figure 4B ) . Overexpression of catalytically active ( SPO14 and GFP-SPO14 ) [24] , but not catalytically inactive ( GFP-SPO14K–H ) [24] , SPO14 was sufficient to rescue the spo14Δ strain from C16:0 PAF-mediated toxicity , whereas overexpression of SPO14 was not found to have any effect in srf1Δ cells . Our findings suggest that Srf1 functions either downstream of Spo14 in mediating an aspect of PA-dependent signaling or directly upon the regulation of Spo14 function . In consideration of the physical interaction between Spo14 and Srf1 ( Figure 3 ) , it is more likely that Spo14 and Srf1 act in concert to mediate choline hydrolysis and PA production . Although dispensable in mitotic cells , Spo14 is strictly required for the progression of S . cerevisiae through meiosis [25] . This observation provides a simple approach to measure the effects of potential Spo14 interacting partners upon its catalytic activity during meiosis . Indeed , GCS1 , which encodes an indirect regulator of Spo14 catalytic activity is essential for sporulation [26] . Therefore , we assessed whether deletion of SRF1 would result in impaired spore formation . In contrast to spo14Δ diploids that failed to sporulate , srf1Δ diploids displayed only minor impairments in sporulation which were associated with an increased frequency of dyads ( Table 1 ) . The modest effect of SRF1 on sporulation is in agreement with previous reported genome-wide studies [27] , and suggests that Srf1 may have only a minor or no impact on Spo14 activity during meiosis or that Srf1 may function via PLD-independent mechanism during sporulation . Despite the limited impact on sporulation , there is a possibility that Srf1 may regulate PLD activity in mitotic cells . Therefore , we sought to examine whether Srf1 could modify Spo14 catalytic activity or localization in mitotic cells . The former possibility was directly assessed by measuring PLD activity in particulate fractions prepared from wild type and mutant strains using a previously described methodology employing a fluorescently labeled phosphatidylcholine derivative as a PLD substrate [28] , [29] . Production of PA and phosphatidyl butanol ( PBt ) , a product of transphosphatidylation , was evident in wild type particulate preparations but was completely absent in srf1Δ , spo14Δ and srf1Δspo14Δ mutant strains ( Figure 5A ) . This result indicates that Srf1 may contribute to particulate-associated PLD catalytic activity in mitotic cells . As we have demonstrated that Spo14 physically interacts with Srf1 , a predicted transmembrane protein , we sought to determine whether the deletion of SRF1 promotes the loss of Spo14 from the particulate fraction . To test this , particulate and cytosolic fractions were prepared from strains transformed with either an empty vector control or a plasmid expressing HA-tagged SPO14 [24] . The absence of PLD activity in srf1Δ strains is not a consequence of altered partitioning of Spo14 catalytic activity between particulate and cytosolic fractions in these cells as catalytic activity was absent from both fractions ( Figure 5B ) . Furthermore , western blot analysis demonstrates that HA-Spo14 remains associated with the particulate fraction independent of Srf1 ( Figure 5C ) . Interestingly , HA-Spo14 protein levels are consistently reduced in srf1Δ mutants ( ∼30% less HA-Spo14 as determined by densitometry ) . However , the absence of detectable PLD activity cannot be fully explained by the exclusion of Spo14 from the particulate fraction or a reduction in Spo14 protein levels ( Figure 5C ) thereby further implicating a biological role for Srf1 in regulating Spo14 catalytic activity during mitosis . Our findings implicating Srf1 in both buffering against the toxic effects of C16:0 PAF and in regulating mitotic PLD activity , underscores the importance of the PLD pathway in protecting yeast cells from C16:0 PAF . Therefore , we next sought to address the effects of C16:0 PAF on the subcellular localization and activity of Spo14 . As had been previously shown [24] , [30] , in untreated or vehicle treated wild type ( WT ) cells , GFP-Spo14 displays modest peripheral and diffuse cytosolic localization ( Figures 6A and S1 ) . Deletion of SRF1 was not observed to grossly affect the subcellular localization of GFP-Spo14 in untreated or vehicle treated cells . However , addition of C16:0 PAF resulted in the rapid loss of GFP-Spo14 at the cell periphery with a concomitant accumulation of GFP-Spo14 at discrete foci or intracellular aggregates in 61±7% of wild type cells ( Figures 6A , 6B and Figure S1 ) . However , treatment with C16:0 PAF resulted in significantly fewer foci in the srf1Δ background ( 9±2% ) suggesting Srf1 plays a role in the intracellular trafficking of Spo14 under C16:0 PAF treatment ( Figures 6A , B S1 ) . As a secondary method to explore the impact of C16:0 PAF on PLD localization we looked at the impact of C16:0 PAF on wild type cells expressing GFP-Q2; GFP tagged to the PA-binding domain of the transcription factor Opi1 [31] . GFP-Q2 localizes to both the periphery and nucleus in wild type cells , however in spo14Δ cells the peripheral signal is lost indicating PA and hence PLD activity is no longer concentrated at the periphery [31] . Treatment of wild type cells expressing GFP-Q2 with C16:0 PAF resulted in the loss of GFP-Q2 from the periphery ( Figure 6C ) , mirroring the effect of C16:0 PAF upon GFP-Spo14 ( Figure 6C ) . Together our studies support C16:0 PAF-mediated changes in the subcellular localization of GFP-Spo14 which are at least in part dependent upon Srf1 expression . Interestingly although C16:0 PAF treatment consistently resulted in the formation GFP-Spo14 aggregate structures; we never detected localization of GFP-Q2 in a similar structure . This suggests that , potentially , GFP-Spo14 aggregates that form upon PAF treatment are no longer catalytically active or alternatively that the PA produced is not accessible to GFP-Q2 . Therefore we also assessed whether the C16:0 PAF-dependent changes in the subcellular localization of GFP-Spo14 and PA were the result of changes in Spo14 catalytic activity , partitioning within subcellular compartments or expression levels . Addition of C16:0 PAF was not observed to impact the catalytic activity of PLD localized to particulate fractions ( Figure 6D ) which is in agreement with previous in vitro findings [28] . Furthermore , C16:0 PAF treatment did not dissociate PLD activity ( Figure 6D ) or GFP-Spo14 ( Figure 6E ) from the particulate fraction , which suggests the C16:0 PAF-dependent GFP-Spo14 aggregate structures are likely still associated with a membranous compartment . Though the production of PA has previously been suggested to buffer PAF toxicity , our work further suggests that the localization of PA production is also likely of importance in regulating the deleterious effects of C16:0 PAF . Our chemical genomic study clearly shows that PLD activity is essential for regulating the toxic effects of C16:0 PAF . To investigate whether this observation could be extended to higher eukaryotes we investigated the role of PLD activity in conferring C16:0 PAF resistance to the murine neuroblastoma cell line N2a , a neural cell line previously used to study PLD and Aβ effects [32] , [33] . N2a cells were treated with C16:0 PAF or vehicle in the presence or absence of a small molecule inhibitor of PLD activity that targets both PLD1 and PLD2 with equal LC50s [34] . Treatment of N2a cells in serum-free medium with 0 . 1% EtOH ( PAF vehicle ) and 0 . 1% DMSO ( PLD inhibitor vehicle ) or with EtOH and 5 µM PLD inhibitor did not impact upon cell viability ( Figure 7 , inset ) . As expected , addition of 1 µM C16:0 PAF for 24 h resulted in reduced cell survival in comparison to treatment with either vehicle ( Figure 7 ) . However , treatment with C16:0 PAF in the presence of the PLD inhibitor resulted in a significant decrease in cell survival in comparison to treatment with C16:0 PAF alone or vehicle . These findings further support the role of PLD activity in regulating against C16:0 PAF mediated toxicity in mammalian neuroblastoma cells and indicate that a conserved mechanism for dealing with elevated levels of C16:0 PAF may exist within eukaryotes .
In this study a chemical genomic screen was employed to identify the key regulators involved in buffering the toxic effects of C16:0 PAF and C16:0 lyso-PAF , lipid species previously shown to be elevated in neurons in response to oligomeric Aβ42 [1] . We identified ten deletion mutants that were sensitive to C16:0 PAF and five deletion mutants that were sensitive to C16:0 lyso-PAF as compared to wild type ( Figure 2A , Table S1 ) . The dramatically different effects on growth ( Figure 1 ) and the minimal overlap in mutants with sensitivity to either lipid suggests that C16:0 PAF , C16:0 lyso-PAF and potentially other PAF species , impinge upon distinct cellular pathways in yeast , which parallels the distinct PAF-mediated effects that have been reported in mammalian systems [1] , [35] . Such a distinction is important in light of recent evidence that aberrant metabolism , in part , underlies Aβ42 neurotoxicity with C16:0 PAF , but not C16:0 lyso-PAF or other PAF species [1] , [2] , [5] . Our unbiased chemical genomic approach identified the deletion mutant of SRF1 as having the most significant differential sensitivity to C16:0 PAF ( Figure 2B ) . We identified a robust interaction between Srf1-TAP and Spo14 ( Figure 3 ) , whose deletion mutant is also hypersensitive to C16:0 PAF [Figures 2B , 4 and ref . 17] . The identification of a physical interaction between Srf1 and Spo14 is striking as only two other proteins , neither with roles in PLD function , have been reported to co-purify Spo14 in high-throughput TAP studies [20] . Furthermore , biochemical assays determined that Srf1 is required for PLD activity in mitotic cells ( Figures 5 and 6 ) . A role for Srf1 in mitotic PLD activity is also supported by genome-wide synthetic lethal genetic screens which revealed that deletion mutants of both SPO14 and SRF1 display genetic interactions with the sec14-bypass mutants CKI1 and KES1 [36] . However , in contrast to Spo14 [24] , [37] , [38] , Srf1 is not essential for sporulation ( Table 1 ) which suggests Srf1 is not regulating PLD activity in meiosis . Our results clearly show that Srf1-TAP can co-purify Spo14 suggesting a model where Spo14 and Srf1 form a complex in mitotic cells that is required for PLD activity and to buffer the toxicity of C16:0 PAF ( Figure 8 ) . How is Srf1 regulating PLD activity ? It is unlikely that the impact of Srf1 on Spo14 protein stability ( Figures 5C and S1 ) could explain the complete absence of PLD activity in srf1Δ cells . Indeed , if Srf1 was only regulating Spo14 protein levels , then overexpression of SPO14 should have rescued the C16:0 PAF hypersensitivity of srf1Δ cells ( Figure 4B ) . Additionally , it is unlikely that Srf1 is regulating PLD activity through Spo14 localization as we found that Spo14 remained associated with the particulate fraction and localized to the plasma membrane in the absence of Srf1 ( Figures 5C Figure 6A , and Figure S1 ) . How else could Srf1 be regulating Spo14 ? Similar to other eukaryotic PLD enzymes , Spo14 catalytic activity can be regulated by numerous mechanisms aside from changes in expression and localization . The binding of phosphoinositol phosphates , fatty acids , indirect interactions with ADP ribosylation factors ( ARFs ) , and phosphorylation have all been demonstrated to regulate PLD activity [reviewed in 39] . Hence Srf1 may be regulating Spo14 through one of these established mechanisms . Alternatively , though we do not detect hydrolysis of phosphatidylcholine in the absence of Srf1 , it is possible that some catalytic activity , possibly against other lipid targets , is still present but misregulated . Indeed this could explain why srf1Δ cells display greater sensitivity to C16:0 PAF than spo14Δ cells ( Figure 2 and Figure 4 ) . A different explanation for this phenomenon could be attributed to the mislocalization of Spo14 in the absence of Srf1 upon C16:0 PAF treatment ( Figure 6 ) . Namely , that in the absence of Srf1 , the interaction of Spo14 with other cellular factors is perturbed thereby potentially serving to titrate away other cellular factors important in the response to PAF . Deletion of SPO14 in combination with deletion of SRF1 would thereby alleviate C16:0 PAF toxicity to that level which is observed solely in the absence of PLD activity . The exact mechanism of how Srf1 regulates Spo14 activity will require further investigation with recombinant proteins to confirm direct interaction and reconstitute the complex activity . However , our work clearly shows that Srf1 is a novel interactor and regulator of Spo14 PLD activity in mitotic cells and together Srf1 and Spo14 are necessary to buffer the toxic effects of C16:0 PAF ( Figure 8 ) . Our yeast chemical genomic study and murine cell culture work indicate that the role of PLD activity in buffering the cytotoxic effects of C16:0 PAF is potentially conserved across species . How is PLD buffering the toxic effect of this GPC ? One possibility is that PLD is rapidly inactivating C16:0 PAF through choline hydrolysis . Indeed , human PLD has been shown to be capable of hydrolysing lyso-PAF species [40] . However a simpler explanation is that PA ( or downstream diacylglycerol ( DAG ) ) isoforms signal inhibition of C16:0 PAF toxicity . Indeed , expression of the E . coli DAG kinase , which converts DAG to PA , has been shown to suppress the toxicity of lyso-PAF and PAF in wild type yeast [17] . While C16:0 PAF treatment is not inhibiting the PLD catalytic activity ( Figure 6B and [28] ) , it is causing the delocalization of GFP-Spo14 and PA concentrations from the cell periphery ( Figure 6 and Figure S1 ) . Intriguingly the C16:0 PAF-mediated delocalization of GFP-Spo14 is dependent on Srf1 ( Figure 6 ) . This suggests that the localized generation of PA ( or PAF hydrolysis ) may be required to buffer the toxic effects of C16:0 PAF . One possibility is that delocalization of GFP-Spo14 and decreased PA levels from the periphery may induce a transcriptional response that is necessary to survive C16:0 PAF exposure . Indeed PA has been shown to play a direct role in the transcriptional regulation of phospholipid biosynthetic genes through the transcriptional repressor Opi1 ( [31] , and reviewed in [39] . Further , our chemical genomic screen supports this hypothesis as several genes with established roles in transcription were identified as being differentially sensitive to C16:0 PAF , including two members of the SWI/SNF chromatin remodeler complex , SNF6 and TAF14 , and the transcriptional regulator UME6 ( Figure 2 ) . Although PAF has been implicated in mediating changes in gene expression , particularly those involved in responses to inflammation [41] , the mechanism ( s ) by which these transcriptional changes occur in response to PAF are not clearly understood , nor is it known if a PAF-mediated transcriptional response is contributing to Aβ-induced neuronal toxicity . An alternative , but not mutually exclusive hypothesis , is that PLD is buffering C16:0 PAF toxicity through membrane trafficking events . Our identification of an interaction between Srf1-TAP and eisosome component Pil1 [42] , suggest that PLD activity may impact sites of endocytosis . However , localization of GFP-Spo14 in either untreated or PAF treated cells ( Figure 6 ) are not reminiscent of the eisosome patches found beneath the plasma membrane [42] nor does recent genetic epistatic miniarray profiles of plasma membrane mutants implicate Spo14 in eisosome function [43] . Alternatively , despite yeast PLD's relatively minor role in vesicle budding from the Golgi and membrane trafficking [reviewed in 39] , yeast PLD may become essential for lipid membrane trafficking upon C16:0 PAF exposure . It has recently been established that PLD1 is a negative regulator of presenilin by two independent mechanisms [32] , [33] . Presenilins are a key component of the AD-associated γ-secretase complex , responsible for cleaving the amyloid precursor protein ( APP ) to Aβ . PLD1 , but not PLD2 , facilitates both APP and presenilin-1 intracellular trafficking and cell surface accumulation [33] , [44] , however PLD1 also interacts with presenilin inhibiting γ-secretase activity [32] , thus reducing Aβ42 production . Despite this controversy , it has been suggested that inhibiting PLD1 represents a novel therapeutic approach to reducing APP and presenilin presentation at the plasma membrane and thus retard the rate of Aβ42 production [44] . In this study , our unbiased approach suggests that such inhibition may be counterproductive with respect to associated GPC metabolic defects . As intraneuronal C16:0 PAF levels are elevated following exposure to soluble Aβ42 oligomers [1] it may be that PLD1 can inhibit the underlying C16:0 PAF ER-stress pathway by reducing Aβ42 production and slowing the rate of PAF accumulation . Thus careful dissection of the impact of PLD1 on Aβ42 production and downstream GPC-mediated signaling is warranted . Here , the discovery that PLD is required to buffer the neurotoxic effect of C16:0 PAF suggests that therapeutic strategies modulating PLD activity may be effective in ameliorating the progression of Alzheimer's Disease pathology .
The yeast strains used in this study are listed in Table 2 . The MATa deletion mutant array was purchased from OpenBiosystems ( catalog no . YSC1053 ) . Deletion strains and TAP tagged SRF1 made for this study were designed using a standard PCR-mediated gene insertion technique [45] , [46] . Plasmids pME962 ( SPO14 LEU2 2 µ ) , pME940 ( HA-SPO14 LEU2 CEN ) , pME1096 ( GFP–SPO14 LEU2 2 µ ) and pME1130 ( GFP-SPO14K–H LEU2 2 µ ) were kind gifts of J . Engebrecht [24] . Plasmid expressing GFP-Q2 was a kind gift of C . Loewen [31] . Cells were grown in standard YPD or SD medium supplemented with amino acids [47] , unless otherwise described . C16:0 PAF and C16:0 lyso-PAF ( L100-0025 and L101-0025 , Cedarlane Canada ) and resuspended in ethanol . The MATα haploid deletion mutant array was robotically pinned in duplicate onto YPD +200 mg/L G418 plates at a density of 1536 colonies per plate using the SingerRotor HAD ( Singer Instrument Company Limited ) and grown for 3 days at 25°C . These plates were pinned onto YPD containing either ethanol , 120 µM C16:0 PAF , or 120 µM C16:0 lyso-PAF . Plates were incubated at 25°C and pictures taken using a Biorad Imager at 15 and 40 hours . The sensitivity of each mutant was assessed by comparing colony sizes on the treated plates to the ethanol control plates by eye and by a computer based method using AlphaEaseFC V4 . 0 . 0 ( Alpha Innotech Corporation ) as described [48] . The screen was performed in triplicate and any interactions identified at least 2 out of 3 times or at least once for both lipids were confirmed by quantitative growth curves . Multiple-drug resistance ( MRD ) genes based on published literature [49] , [50] were removed from our data set . Liquid growth curves were obtained for 91 strains ( 90 haploid knockouts plus wild type ) , spanning 24 plates . Each curve consisted of 45 OD readings taken at intervals of 25 minutes using Multiskan Ascent Plate Reader ( Thermo Electron Corporation ) and Ascent Software Version 2 . 6 . On each plate , 6 strains were examined ( wild type plus 5 others ) and each strain was represented in 9 independent wells ( 3 replicates of the 3 treatments: ethanol , C16:0 lyso-PAF , and C16:0 PAF ) . In total , 1247 growth curves were analyzed ( see Text S1 for more details ) . A four-parameter logistic growth model was fit to the growth curves [51]: where x is time and y is the OD reading , a proxy for cell density or population size . A is the starting point of growth or minimum OD reading , B is the carrying capacity or maximum OD reading , xmid is the time at 50% of total growth , and scal is approximately the time taken to go from 50 to 75% of growth . To simultaneously account for the repeated measurement of individual wells and for the systematic effects of treatment and gene deletion on growth , we fit a mixed effects logistic growth model using the R package nlme [52] , [53] ( R code available upon request ) . Each of the four growth parameters ( A , B , xmid , scal ) could therefore be modeled with a combination of fixed gene deletion and/or treatment effects and random well effects ( see Text S1 for details ) . The model was fit to the ensemble of 54 curves obtained for each plate separately . To identify chemical-genetic interactions , we focused on the xmid parameter . From the fitted model , we combined fixed effect estimates and predicted random well effects to produce a value of xmid for each growth curve , which was treated as derived data for downstream analysis . There was evidence of non-negligible plate effects as well as interaction effects between plates and the treatments C16:0 lyso-PAF and C16:0 PAF , indicating the need for inter-plate normalization ( see Text S1 for details ) . Due to the inclusion of wild type replicates on all plates for all 3 treatments , a normalization model could be fit to the wild type xmid values to obtain estimates of plate and plate-by-treatment interaction effects . These were then used to remove plate-related artifacts from all of the xmid values , i . e . including those for deletion mutants . After normalization , we fit the following model for xmid: where is the estimated/predicted and normalized value of xmid for one growth curve , corresponding to the deletion of gene g and the treatment t . The typical xmid for wild type in the ethanol reference condition is given by αwt and αg and αt are the individual effects , respectively , of deleting gene g or of treatment t . The primary parameter of interest is αg*t which captures the interaction between deletion g and treatment t . Mutants deemed hypersensitive to a single PAF species were identified based on tests of the null hypotheses that αg*L-PAF = 0 or αg*PAF = 0 . Mutants deemed differentially sensitive were identified based on a test of the null hypothesis that αg*PAF−αg*L-PAF = 0 . There are 90 potential gene deletions g and 3 parameters of interest ( αg*L-PAF , αg*PAF , αg*PAF−αg*L-PAF ) , for a total of 270 tests . We did a global Bonferroni correction , i . e . multiplied p-values by 270 , and thresholded at 0 . 04 to obtain hits ( 5 C16:0 lyso-PAF hypersensitive , 10 C16:0 PAF hypersensitive , 11 C16:0 PAF vs . C16:0 lyso-PAF differentially sensitive ) . Modified Chromatin Immunoprecipitation ( mChIP ) and mass spectrometry to identify co-purifying proteins was performed as previously described [22] from 2 . 1 L YPD culture of YKB2270 ( OD600∼0 . 9 ) using 300 µL of Dynabeads ( Invitrogen ) coated with rabbit IgG ( I5006 , Sigma ) . Cells were grown in YPD or minimal media at 30°C to mid-log phase and resuspended to an OD600 of 0 . 1 and dot assays were performed by spotting 5 µL of five-fold serial dilutions ( OD600 = 0 . 1 , 0 . 01 , 0 . 001 , 0 . 0001 ) onto YPD or minimal media selection plates containing the specified concentrations of ethanol , C16:0 lyso-PAF or C16:0 PAF as indicated . Strains were grown overnight in YPD at 30°C . The following day cells were pelleted at 800 g for 5 min and washed in sterile water . An OD600 of 2 . 0 was resuspended in YP-acetate and incubated at 25°C for 3 days prior to microscopic examination of sporulation efficiency . The number of spores per ascus was enumerated in a minimum of 300 cells from three independent experiments and expressed as a percentage of total cells ± standard error . For localization of GFP-Spo14 ( pME1096 ) and GFP-Q2 overnight cultures of yeast cells grown at 30°C in YPD medium were re-suspended at a final OD600 of 0 . 2 and allowed to reach mid-log phase prior treatment and image acquisition . Similarly , wild type ( YAM282-2 ) and srf1Δ ( YKB2472 ) cells expressing GFP-Spo14 under the control of a copper inducible promoter were grown at 30°C and induced with 3 µM CuSO4 for 2 h to induce GFP-Spo14 expression prior to imaging or extract preparation . Cells were briefly centrifuged ( 800 g for 3 min ) , resuspended in a minimal volume of growth media , spotted onto glass slides and coverslipped prior to imaging . Images were acquired using a Leica DMI 6000 florescent microscope ( Leica Microsystems GmbH , Wetzler Germany ) , equipped with a Sutter DG4 light source ( Sutter Instruments , California , USA ) , Ludl emission filter wheel with Chroma band pass emission filters ( Ludl Electronic Products Ltd . , NY , USA ) and Hamamatsu Orca AG camera ( Hamamatsu Photonics , Herrsching am Ammersee , Germany ) . Images were acquired and analyzed using Velocity Software ( Perkin Elmer ) . Overnight cultures of yeast strains cultured in YPD were diluted to an OD600 of 0 . 2 in YPD and allowed to reach mid-log growth prior to harvesting . Yeast cells were pelleted at 800 g for 5 min and washed with ice cold sterile water . Cytosolic and particulate extracts were prepared essentially as described previously [28] . Briefly , cells pellets were resuspended in 200 µL of lysis buffer ( 20 mM HEPES , 150 mM NaCl , 2 mM EDTA and protease inhibitor cocktail ( Sigma , P-8215 ) ) and lysed by vortexing with glass beads . Glass beads and intact cells were first removed by brief centrifugation . Particulate and cytosolic fractions were collected and separated by centrifugation at 13 000 rpm for 15 min at 4°C . Particulate fractions were resuspended in an equal volume of lysis buffer . Protein concentration was determined using Bradford reagent . Cellular and particulate extracts were added in equal volume to a reaction mixture ( 500 mM octylglucoside , 400 mM NaCl , 60 mM HEPES ( pH 7 . 0 ) and 1% v/v n-butanol ) containing 200 µM BODIPY labeled glycerophosphocholine ( 2-decanoyl-1- ( O- ( 11- ( 4 , 4-difluoro-5 , 7-dimethyl-4-bora-2a , 4a-diaza-s-indacene-3-propionyl ) amino ) undecyl ) -sn-glycero-3-phosphocholine , Invitrogen , D-3771 ) in the absence of exogenous PIP2 , which was not observed to affect measured PLD activity ( data not shown ) , as previously described [28] . Reactions were incubated at 30°C for 40 min prior to spotting on TLC plates ( EMD chemicals , 5626-6 ) . Products were visualized under UV light and images captured using Quantity One 4 . 6 . 1 software ( Biorad ) following separation in chloroform/methanol/water/acetic acid ( 45:45:10:1 ) . 10 µg of total protein from particulate and cytosolic fractions were separated by SDS-PAGE on 5% SDS-polyacrylamide gels . Proteins were transferred to nitrocellulose at 0 . 8 mA/cm2 for 2 h prior to blocking overnight in 5% skim milk in TBS-T . Standard Western blotting procedures were performed using α-HA ( Roche , 11667149 ) , α-GFP ( Roche , 11874460001 ) and peroxidase-conjugated goat α-mouse IgG ( BioRad , 170-6516 ) . N2a cells were maintained in DMEM/F12 media containing 10% fetal bovine serum , L-glutamine and penicillin/streptomycin . Cells were plated at 1 . 85×104 cells per well in 24 well plates prior to treatment . Treatment with vehicle ( 0 . 1% DMSO + 0 . 1% Ethanol ) , the small molecule PLD inhibitor N- ( 2-{4-[2-oxo-2 , 3-dihydro-1H-benzo ( d ) imidazol-1-yl]piperidin-1-yl}ethyl ) -2-naphthamide ( VUO155056 , Avanti Polar Lipids , 857370 ) + 0 . 1% EtOH , C16:0 PAF ( PAF , 1 µM ) + 0 . 1% DMSO , or the PLD inhibitor in the presence of C16:0 PAF ( PAF + Inh ) were performed in serum free complete DMEM/F12 supplemented with 0 . 025% bovine serum albumin ( BSA ) for 24 h . Cell survival was assessed using LIVE/DEAD reagent ( Invitrogen , L3224 ) . Viable cells in each treatment condition were quantified and standardized to vehicle only treated cells . Data represent the results from the measurement of 4 fields of view from 3-4 independent wells per condition ± SEM ( n = 12-16 ) . Statistical analyses were one-way analysis of variance ( ANOVA ) followed by post-hoc Student-Newman-Keuls multiple comparisons test or Student's t test where only two experimental groups were analyzed .
|
Accelerated cognitive decline in Alzheimer's patients is associated with accumulation of choline-containing lipids . One of these lipids , C16:0 platelet activating factor ( PAF ) , is specifically elevated in brains of Alzheimer's patients . As elevated exposure to C16:0 PAF ultimately leads to neuronal death , it is crucial to identify underlying mechanisms that mitigate the toxic effects of this lipid . In this study we exploit the conserved biology between humans and baker's yeast to identify key genes that are essential to buffer the toxic effects of C16:0 PAF . We found that Srf1 , or Spo14 Regulatory Factor 1 , the previously uncharacterized protein Ydl133w , is essential for mitigating the toxic effects of C16:0 PAF in yeast . We determine that Srf1 interacts with yeast phospholipase D ( PLD ) Spo14 and is required for PLD activity in mitotic cells . Hence we discovered a novel regulator of PLD in yeast . Further , we extend our studies to higher eukaryotes demonstrating that PLD is required to buffer the neurotoxic effect of C16:0 PAF . Our study suggests that therapeutic strategies modulating PLD activity may be effective in ameliorating Alzheimer's Disease pathology associated with disruptions in lipid metabolism .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"and",
"genomics/functional",
"genomics",
"geriatrics/dementia",
"biochemistry/chemical",
"biology",
"of",
"the",
"cell",
"genetics",
"and",
"genomics/disease",
"models",
"genetics",
"and",
"genomics/gene",
"function",
"biochemistry/cell",
"signaling",
"and",
"trafficking",
"structures",
"microbiology",
"chemical",
"biology"
] |
2011
|
Srf1 Is a Novel Regulator of Phospholipase D Activity and Is Essential to Buffer the Toxic Effects of C16:0 Platelet Activating Factor
|
Bacterial pathogens often manipulate host immune pathways to establish acute and chronic infection . Many Gram-negative bacteria do this by secreting effector proteins through a type III secretion system that alter the host response to the pathogen . In this study , we determined that the phage-encoded GogB effector protein in Salmonella targets the host SCF E3 type ubiquitin ligase through an interaction with Skp1 and the human F-box only 22 ( FBXO22 ) protein . Domain mapping and functional knockdown studies indicated that GogB-containing bacteria inhibited IκB degradation and NFκB activation in macrophages , which required Skp1 and a eukaryotic-like F-box motif in the C-terminal domain of GogB . GogB-deficient Salmonella were unable to limit NFκB activation , which lead to increased proinflammatory responses in infected mice accompanied by extensive tissue damage and enhanced colonization in the gut during long-term chronic infections . We conclude that GogB is an anti-inflammatory effector that helps regulate inflammation-enhanced colonization by limiting tissue damage during infection .
Horizontal gene transfer ( HGT ) is an important contributor to the genetic and phenotypic diversity of enteric bacteria . Plasmids and genetic regions called pathogenicity islands ( PAI ) can provide a fitness advantage and the ability to colonize and expand into novel host niches [1] . HGT has been a major driver of evolution of various Salmonella enterica serovars , giving rise to pathogens that infect a wide range of cold and warm-blooded animal hosts . S . enterica is a facultative intracellular pathogen that invades and replicates within host cells . S . enterica serovar Typhimurium ( S . Typhimurium ) causes food poisoning in humans characterized by diarrhea , abdominal pain and fever . The acquisition of two pathogenicity islands , Salmonella Pathogenicity Island-1 ( SPI-1 ) and SPI-2 , is considered a major factor in the evolution of S . Typhimurium pathogenesis . These genomic islands encode two different type III secretion systems ( T3SS-1 and T3SS-2 ) that deliver effector proteins into host cells to manipulate host machinery leading to bacterial uptake , growth , and dissemination . T3SS-1 promotes invasion of epithelial cells , macrophage apoptosis and recruitment of phagocytes [2] , [3] whereas T3SS-2 is primarily required for intracellular replication , dissemination , and disease associated with systemic spread [4] . SPI-1 and SPI-2 mutants are severely attenuated for virulence in mice infections [5] indicating that their concerted activities profoundly influence the host-pathogen interaction . Acquisition of phage genes by lysogenic conversion contributed to the genetic diversity of Salmonella by providing an extended repertoire of virulence determinants that have integrated into ancestral regulatory networks of the bacterial cell [6] . GogB is a secreted effector encoded in the Gifsy-1 prophage found in some S . enterica strains and is a substrate of both T3SS-1 and T3SS-2 [7] . We showed previously that GogB is a chimeric protein consisting of an N-terminal canonical leucine-rich repeat domain ( LRR ) and a C-terminal domain with similarity to known proteins . Salmonella translocates GogB into the host cytoplasm , however its function and host cell target ( s ) were not known . The LRR domain in GogB resembles other LRR-containing effectors that function as a novel class of E3 type ubiquitin ligases called NELs [8] , [9] . These include the Salmonella effectors SspH1 , SspH2 , and SlrP , and the IpaH family of proteins from Shigella . The host ubiquitination system regulates a wide array of cellular processes including growth and differentiation , gene expression , and regulation of the host inflammatory response [10] . As such , host ubiquitination is a well-known target of bacterial as well as viral proteins [11] , [12] , [13] , [14] . For example , the Shigella effector OspG binds to the Skp , Cullin , F-box containing complex ( SCF ) component UbcH5 to inhibit IκBα degradation and NFκB activation . In Legionella , interaction of the F-box protein AnkB with SCF ligase promotes intracellular replication within the host cell [12] . Viral pathogens among the Poxviridae family also target the SCF complex indicating a widespread mechanism of manipulating host cells during infection [14] , [15] . As well , Salmonella exploits host ubiquitination pathways to regulate the temporal and spatial activity of SopE , SptP , and SopB effector proteins [16] , [17] , [18] . Other NEL effectors act as ubiquitin ligases for certain host cell targets [9] . The SCF complex is a multi-protein E3 ubiquitin ligase catalyzing the ubiquitination of proteins fated for degradation . The E3-type SCFβ-TrcP ligase complex regulates the NFκB pathway by targeting ubiquitinated IκBα for degradation by the proteasome [19] . The F-box domain is a ∼50 amino acid motif that mediates protein–protein interactions , often in concert with other protein-protein interaction platforms such as LRRs . The F-box motif interacts directly with the SCF protein Skp1 to bring specific protein targets to the E3 ligase [20] . Humans have at least 38 genes encoding different F-box proteins , although most of their functions and substrates are not known [21] . In this study , we identified the host cell target of the Salmonella effector GogB to be F-box only protein 22 ( FBXO22 ) . We mapped the FBXO22 interaction domain of GogB and showed that this interaction targets GogB to the SCF ubiquitin ligase complex to dampen the host inflammatory response by inhibiting IκBα degradation and NFκB activation . Consistent with this , a gogB mutant was hyper-inflammatory in the murine gut during chronic infection with an accompanying increase in tissue pathology and bacterial burden in gut tissue . We conclude that GogB is an anti-inflammatory effector that dampens host inflammatory responses following colonization in order to limit tissue damage and to balance bacterial colonization levels during chronic infection .
Our previous work with GogB revealed an N-terminal LRR domain with similarity to the LRR-containing Salmonella effectors SspH1 , SspH2 , and SlrP and Shigella IpaH7 . 8/9 that have been characterized as NEL proteins [7] , [8] , [9] . The NEL domain has a conserved catalytic cysteine residue in the C-terminus that mediates E3 ubiquitin ligase activity in vitro [8] , [22] . In order to identify potential host cell targets of GogB , we immunoprecipitated GogB-HA that had been delivered into HeLa cells by Salmonella via the T3SS ( Fig . 1A ) . We identified three bands that co-purified only with GogB-HA and mass spectrometry analysis identified these proteins as lactate dehydrogenase , α-actin , and FBXO22 , an uncharacterized ∼40 kDa human F-box-containing protein component of the SCF ubiquitin ligase complex . To determine whether the precipitated proteins specifically interacted with GogB , the pull-down assay was repeated and we confirmed that GogB specifically targeted FBXO22 , whereas α-actin was a non-specific contaminant of the pull-down ( Fig . 1B ) . F-box proteins function as the substrate recognition module of the E3-type SCF ubiquitin ligase complex by binding to the adapter protein Skp1 [23] thereby targeting substrates for degradation [21] . To explore the GogB-FBXO22 interaction further , we tested whether GogB also interacts with Skp1 . A pull-down assay with GST-GogB mixed with lysates from either RAW264 . 7 or HeLa cells confirmed that Skp1 also co-purified with GogB ( Fig . 1C ) . The Skp1 binding domain on GogB was identified by separately expressing the GogB N-terminal LRR domain ( amino acids 1–253; gogB-NT ) and the C-terminus encoding residues 254–497 ( gogB-CT ) and purifying each as GST fusions . GST pull-downs with these GogB fragments showed that Skp1 bound to GogB-CT but not to GogB-NT or GST alone ( Fig . 1C ) . To verify this interaction in host cells , GogB-NT and GogB-CT ( the C-terminal GogB fragment fused to the T3SS secretion signal [24] ) , were expressed as HA-tagged proteins in a ΔgogB mutant and verified to be secreted by the T3SS ( Fig . S1 ) . Similar to the GST pull-downs , GogB and GogB-CT but not GogB-NT delivered to host cells by the Salmonella T3SS interacted with a complex containing both Skp1 and FBXO22 as denoted by the presence of both proteins in lanes containing GogB and GogB-CT ( Fig . 1D ) . Interestingly , we only observed FBXO22 co-purifying with GogB when delivered directly to host cells but not in GST pull downs with cell lysates indicating that targeting FBXO22 by GogB is specific to the infected host cell . These results demonstrate that GogB interacts with the SCF ubiquitin ligase complex and that the C-terminal domain of GogB mediates GogB-Skp1 and GogB-FBXO22 interactions . To determine whether Skp1 is required for the binding of GogB to FBXO22 , we performed RNA interference experiments to knockdown the expression of Skp1 in HeLa cells . Lysates from infected HeLa cells shows that Skp1 expression was knocked down only in cells transfected with the Skp1 siRNA but not control siRNA . Infection of these transfected cells with ΔgogB complemented with the full-length GogB-2HA showed that FBXO22 was precipitated by GogB-2HA in the absence or presence of Skp1 ( Fig . 1E ) . Pull-down assays were also performed using cells infected with ΔgogB as a control for non-specific binding . These results denote that GogB interacts with Skp1 and FBXO22 using different binding domains and that a GogB-FBXO22-Skp1 complex is formed in the host cells as shown by the presence of all three proteins in the pull-down assay using GogB-HA-infected cells . Whether FBXO22 is required for the binding of GogB to Skp1 remains to be determined as our attempts to knock down FBXO22 expression were not successful . F-box proteins typically bind to Skp1 through a ∼50-amino acid F-box motif . We examined whether GogB also binds to Skp1 through a similar domain . Bioinformatic analysis of GogB identified an F-box-like domain between amino acids 270–334 . Alignment of the GogB F-box-like motif with sequences from human F-box proteins β-Trcp1 , Skp2 , and the bacterial F-box secreted effector GALA1 from Ralstonia solanacearum [25] showed a conserved leucine-proline ( LP ) at positions 270–271 ( Fig . 2A ) . Mutation of these residues in human Skp2 F-box abolished its interaction with Skp1 [26] . When we tested a GogB mutant ( Δ264–352 ) lacking the F-box-like domain it was severely compromised in its ability to interact with Skp1 ( Fig . 2B ) . Although the amounts of GogB-CT and GogB LP270AA bound to the HA-affinity beads were lower compared to that of full-length GogB ( lane 1 ) , this did not affect the interaction of these proteins with Skp1 . In comparison , GogB-NT and GogB Δ264–352 bound to the affinity beads at similar amounts compared to GogB-CT or GogB LP270AA but these proteins did not precipitate Skp1 , demonstrating that the binding of Skp1 is dependent on the F-box-like domain found at the C-terminus of GogB . To determine if the conserved leucine-proline residues in the GogB F-box-like domain were involved , we made alanine substitutions in full-length GogB and GogB-CT . Introducing L270A and P271A mutations in full-length GogB did not abolish binding to Skp1 ( Fig . 2B ) . Although L47A and P48A mutations in GogB-CT appeared to reduce its interaction with Skp1 , the amount of GogB-CT bound to the affinity beads is lower . We conclude from these results that GogB contains an F-box-like domain that is essential for Skp1 binding and that multiple determinants in this domain ( perhaps including the conserved LP motif ) are involved in this interaction . The SCF−βTRcP1 complex is important for NFκB activation during the host inflammatory response because it targets the IκBα inhibitor protein for ubiquitination and degradation [19] , [27] . Because other LRR-containing effector proteins are involved in modulating NFκB activity during infection , we investigated a potential role for GogB in the NFκB pathway through its interaction with the SCF ligase complex . To do so , we infected RAW264 . 7 cells with wild type and gogB-deficient Salmonella and monitored levels of IκBα . Western blot analyses showed that the amount of IκBα was lower in cell populations infected with a gogB mutant compared to wild type Salmonella ( Fig . 3A ) . Non-virulent salmonellae have been shown to suppress the NFκB pathway by inhibiting ubiquitination of IκBα , thereby modulating the epithelial tissue response to inflammation [28] . To determine whether the GogB-dependent decrease in IκBα was due to ubiquitination and subsequent degradation , the level of poly-ubiquitinated IκBα was determined by infecting macrophages with wild type Salmonella , ΔgogB , and ΔgogB complemented with plasmids encoding GogB or GogB-Δ264–352 in the presence or absence of the proteasome inhibitor MG-132 to trap ubiquitinated IκBα molecules . In the absence of proteasome inhibition it was difficult to isolate ubiquitinated IκBα presumably due to its rapid degradation . However , in the presence of the proteasome inhibitor MG-132 , cells infected with ΔgogB or bacteria expressing the non-functional GogBΔ264–352 mutant had higher levels of poly-ubiquitinated IκBα in IκBα pull-downs compared to the wild type strain . Complementation of ΔgogB with full-length GogB returned the levels of poly-ubiquitinated IκBα to that in wild type ( Fig . 3B ) . These results suggested that GogB plays an essential role in down-regulating the host inflammatory response during Salmonella infection by inhibiting poly-ubiquitination of IκBα and thus NFκB-dependent gene expression . To test the latter , we measured NFκB-dependent expression of luciferase in RAW264 . 7 cells co-transfected with pNFκB-luc and pCMV-βgal reporter plasmids . Transfected cells were infected with wild type Salmonella containing an empty vector or the ΔgogB mutant and ΔgogB complemented with full-length gogB , or the mutant variants gogB-NT , gogB-CT , gogB LP270AA , gogB-CT LP47AA , and gogBΔ264–352 . Macrophages infected with the gogB mutant had >10 fold higher luciferase levels compared to cells infected with wild type Salmonella ( Fig . 3C ) . Deletion of the C-terminal domain previously shown to be the active fragment , or deletion of the F-box motif inhibited this NFκB-luciferase activity to ∼4-fold that of wild type . Complementation of ΔgogB with the full-length GogB or the Skp1-interacting domain , GogB-CT , resulted in NFκB-luciferase activity similar to that of wild type . However , mutation of the conserved leucine-proline residues in the F-box-like domain of GogB resulted in only a slight increase in levels of NFκB activity compared to wild type Salmonella denoting that other residues in the F-box motif may be needed to inhibit host inflammatory response , in keeping with our in vitro data with Skp1 interaction using the same GogB mutant . These results show that GogB partially blocks NFκB activation during Salmonella infection by inhibiting IκBα degradation through its interaction with Skp1 and that the GogB F-box-like domain is essential for this inhibitory activity . These data also support the idea that other regions of GogB , in addition to the F-box-like domain are involved in inhibiting NFκB activity . The functional relevance of the GogB-mediated decrease in NFκB activity during Salmonella infection was examined first by characterizing the inflammatory response of infected macrophages . RAW264 . 7 cells were infected with wild type Salmonella , ΔgogB , or complemented strains and interleukin-1β ( IL1β ) levels in culture supernatants were measured . Similar to the NFκB luciferase assay experiments , macrophages infected with wild type Salmonella or ΔgogB complemented with full-length GogB or GogB-CT secreted less IL1β compared to ΔgogB or the mutant complemented with GogB-NT or GogBΔFbox ( Fig . 3D ) . Previous work showed that the adapter protein Skp1 is essential for NFκB signaling [19] . To examine whether the increased activity of NFκB in cells infected with the gogB mutant was also dependent on Skp1 or whether it was signaled through an alternate pathway , we introduced the pNFκB-luc and pCMVβgal reporter plasmids into HeLa cells and then knocked down Skp1 expression by RNA interference ( Fig . 3E and 3F ) . Thus , wild type S . Typhimurium and the gogB mutants are unable to activate NFκB in the absence of Skp1 . Luciferase assays showed that knockdown of Skp1 expression in host cells resulted in similar levels of NFκB activity upon infection of cells with ΔgogB and complemented strains compared to the wild-type SL1344/pWSK129 . In contrast , treatment of HeLa cells with control siRNA showed a 2 to 3 fold increase in NFκB activity in cells infected with ΔgogB and strains complemented with gogB-NT or strains with mutations in the F-box motif , gogB Δ264–352 , gogB LP270AA and gogB-CT LP47AA ( Fig . 3E ) . Similar to the luciferase assays performed using macrophages , ΔgogB complemented with the full-length GogB or GogB-CT had similar NFκB activity as wild-type cells containing an empty plasmid . Interestingly , HeLa cells infected with GogB mutants complemented with gogB LP270AA and gogB-CT LP47AA showed a higher NFκB level compared to wild-type Salmonella . These results show that differences in NFκB activation in cells infected with wild-type Salmonella and ΔgogB are abolished in the absence of Skp1 . Overall , our results demonstrate that GogB plays a role in blocking the host inflammatory response through inhibition of NFκB activation , IκBα ubiquitination and degradation and that this modulation requires the interaction of GogB with Skp1 of the SCF ubiquitin ligase . Our results demonstrating the role of GogB in modulating the host immune response through its interaction with the SCF ligase led us to examine the role of this effector during acute and chronic animal infections . We first performed gentamicin protection assays to determine whether deletion of gogB impaired Salmonella replication in macrophages and epithelial cells , which it did not , as invasion at 2 h post-infection and replication after 20 h were similar between the strains with and without GogB ( Fig . 4A ) . This suggested that the role of GogB might manifest in the context of animal infection . To test this we used competitive acute infections of susceptible C57BL/6 mice infected with a mixture of wild type and ΔgogB Salmonella . Mutant and wild type bacteria were recovered in equal numbers from gut tissues and systemic organs after three days ( Fig . 4B ) indicating that deletion of gogB does not significantly affect Salmonella fitness over acute infection periods in genetically susceptible mice . The Natural resistance-associated macrophage protein ( Nramp ) -1 controls intracellular Salmonella growth by limiting the transport of divalent cations across the bacterial vacuole [29] . Given that gogB-deficient Salmonella were not attenuated for acute infection of susceptible mice we tested whether the GogB-mediated modulation of the host inflammatory response contributed to chronic infection . We infected groups of Nramp1+/+ mice ( 129S1/SvImJ ) with wild type or ΔgogB Salmonella and measured bacterial colonization in systemic and intestinal sites at 4 , 7 , 39 , and 60 days after infection ( Fig . 5 ) . After four days , both wild type and ΔgogB colonized the mice , however , mice infected with gogB-deficient Salmonella had moderate but significantly higher bacterial load in all tissues ( p<0 . 05 ) , with the largest effect seen in the cecum . The ΔgogB-infected mice had enlarged mesenteric lymph nodes ( MLN ) yet we did not observe outward signs of acute illness such as hunching , ruffled fur or piloerection ( data not shown ) . At time points greater than 7 days there was no difference between the number of wild type and mutant bacteria recovered from the spleen or liver . However , what was striking was the high degree of cecal colonization in mice infected with gogB-deficient Salmonella throughout the long-term infection coupled with greater bacterial load in the MLNs at day 60 ( Fig . 5 ) . Although the bacterial load differences in the MLN at day 60 did not reach statistical significance , all the ΔgogB-infected mice displayed a similar high degree of colonization whereas colonization by wild type Salmonella was more variable . From days 21 to 49 , resistant mice are able to clear Salmonella or eventually progress to a chronic carrier state [30] . Indeed at day 60 , very low numbers of wild type bacteria were recovered from the cecum whereas ΔgogB mutants were present at up to 5–6 log greater numbers , demonstrating a profound difference between wild type and ΔgogB to persist in the mouse cecum ( Fig . 5 ) . To gain preliminary insight into the connection between bacterial load and inflammation , we treated mice with dexamethasone as an immunosuppressant prior to infection with Salmonella and monitored bacterial load in the cecum at day 4 after infection ( Fig . S2 ) . In these experiments , the bacterial load of the gogB mutant was drastically reduced in immunosuppressed mice and similar to the levels of wild-type bacteria in the cecum during dexamethasone treatment . These data suggested to us that the gogB mutant does not have an inherent increased growth rate leading to increased inflammation , but rather benefits from a heightened inflammatory response that is suppressible by chemical blockade . Histopathological analysis revealed a marked difference in tissue inflammation and integrity of the cecal tissue in mice infected with wild type and ΔgogB at day 4 ( Fig . 6A and 6B ) and at day 60 ( Fig . 6C and 6D ) . Tissue samples from wild type-infected mice were characterized by mild to moderate infiltration of mononuclear cells and polymorphonuclear leukocytes ( PMN ) particularly at the surface epithelium and submucosa , minimal necrosis and edema , and minimal pathological changes in the mucosa . In comparison , increased colonization by ΔgogB was accompanied by moderate to severe pathology with dramatic changes in the gross morphology of the cecum due to increased influx of mononuclear cells and PMNs , granulomatous lesions , ulceration and increased frequency of cryptic abscesses . From recent work , intestinal inflammation provides a selective advantage for Salmonella by creating an environment that enhances the pathogen's growth and dissemination via metabolic fitness [31] , [32] . To examine the inflammatory markers associated with colonization and pathology , we measured the levels of pro- and anti-inflammatory cytokines in the cecum and MLN by quantitative RT-PCR . At day 4 , expression of the NFκB target genes , TNFα , IL1β and IL12p40 were significantly upregulated in the cecum of ΔgogB-infected mice compared to wild type , while expression of TGFβ1 and the anti-inflammatory cytokines IL4 and IL10 were similar in both groups of mice ( Fig . 7A ) . The expression profiles of pro- and anti-inflammatory cytokines in the MLNs were similar in both groups of infected mice except for MIP2 ( Fig . 7B ) . NFκB activity is repressed by TGFβ1-mediated stabilization of IκBα [33] and the expression of TGFβ1 in the MLN from both groups of mice corresponded with an overall lower expression of NFκB target genes compared to the cecum . The higher expression of macrophage inflammatory protein 2 ( MIP-2 ) chemokine in the cecum and MLN of ΔgogB-infected mice is consistent with the increased recruitment of phagocytes . At day 60 , the levels of IL12p40 and MIP2 remained higher in the cecum of ΔgogB-infected mice ( Fig . 7C ) whereas the other inflammatory markers had normalized in the MLN between both groups of mice . Overall , these data indicate that GogB plays an important role in dampening the host immune response and limiting tissue destruction during Salmonella infection in the gut .
In this study , we characterized the role of the LRR-containing T3SS effector GogB in Salmonella pathogenesis . Due to its similarity in the LRR domain with members of the NELs , we hypothesized that GogB may be involved in subverting the host ubiquitination process . We identified FBXO22 as the host cell target of GogB that facilitates binding to host Skp1 . Skp1 is an adapter protein linking diverse F-box proteins and Cullin-1 to form the SCF ligase complex [23] . Accordingly , we found that GogB interacted with Skp1 , which was mediated by an Fbox-like domain in the C-terminal part of GogB . Thus , GogB could be functioning in a similar manner to the Legionella effector AnkB ( Lpp2082 ) previously identified as a prokaryotic F-box protein and to the ankyrin repeat proteins found in poxvirus that contains an F-box motif at the C-terminal domain [12] , [13] , [14] . Whether or not the N-terminal LRR in GogB interacts with other host cell proteins or forms a complex with other effectors through this domain has yet to be determined , although we did show that this LRR domain is not necessary for targeting GogB to either Skp1 or FBXO22 , or for the downstream inhibition of NFκB activation . Recently , one target of the SCFFBXO22 complex was identified as the histone demethylaase KDM4A [34] . However , another study showed that KDM4A turnover was instead coordinated by an SCFFbxL4 complex [35] . Therefore , whether or not GogB is involved in KDM4A turnover remains an open question that could be the focus of follow up work . To our knowledge , our report is the first to implicate a role for FBXO22 in the regulation of the NFκB pathway , thus broadening the function of this newly described human F-box protein . In the human F-box protein Skp2 , the last 30 amino acids of the LRR domain fold over the F-box-Skp1 interface , suggesting a possible mechanism by which the GogB LRR may be performing a regulatory function in the full-length protein [26] . The Skp2 F-box has three helices , with H1 folding orthogonally to the H2-H3 anti-parallel helices that interdigitate with the C-terminus of Skp1 [26] . Structure predictions for GogB show its F-box domain to be a combination of coiled-coils and helices . Given that the GogB F-box was necessary for Skp1 interaction , it seems this motif may perform a similar function to the Skp2 F-box in binding to Skp1 . In the case of FBXO22 and GogB , it is possible that both proteins bind to Skp1 but the presence of GogB may prevent proper function of the SCFFBXO22 ligase , perhaps in conjunction with another protein ( s ) . On this point , some of our data is consistent with the GogB-FBXO22 interaction being dependent on another molecule since FBXO22 co-purified with GogB only when this effector was delivered into host cells by bacteria and not when GogB was mixed with host cell lysates . This might implicate another bacterial-delivered effector or host protein in the interaction of GogB with FBXO22 . Alternatively , it is increasingly being shown that host-dependent modification of bacterial effectors can dictate their biological function [36] , [37] . As such , the GogB-FBXO22 interaction may be enhanced or stabilized by a host-dependent modification process following translocation . These possibilities will be the subject of work to follow . Gut inflammation is important for colonization by Salmonella [31] , [32] . A typical Salmonella infection elicits transient inflammation during the early stages of infection but does not lead to destruction of the intestinal epithelium [38] . The ability to elicit intestinal inflammation is dependent , in part , on both T3SS encoded in Salmonella and their associated effectors [39] , [40] , [41] . The SPI-1 T3SS is better known for its role in regulating the early colonization of the intestine , whereas the SPI-2 T3SS has typically been studied in the context of systemic infection . Although the role of these two T3SS in distinct stages of infection may be less disparate than previously thought , it is interesting that GogB is one of only a few effector substrates that is delivered by both T3SS-1 and T3SS-2 systems in Salmonella [7] indicating that this protein might be involved in regulating immune processes from an early stage of the host-pathogen interaction . A major finding of biological significance in this study was that the GogB-mediated down-regulation of the host inflammatory response during chronic mouse infection limits tissue damage and host pathology . In the absence of GogB , mice were colonized to greater levels especially in the chronically infected cecum accompanied by increased inflammation and tissue damage . Salmonella reside within macrophages in the MLN , which serves as a reservoir for subsequent re-seeding of the liver and spleen [42] , [43] . Our data on the levels of pro-inflammatory cytokine expression at days 4 and 60 post-infection showed that inflammation was prominent only in the cecum and not in the MLN . This might indicate that a subpopulation of ΔgogB Salmonella that are able to traverse the epithelial barrier and reach the MLN , liver and spleen retain their ability to suppress systemic host inflammation possibly through the concerted action of other anti-inflammatory effectors . The increased inflammation in gut tissue elicited by bacteria lacking GogB along with higher cecal colonization in Nramp1-positive mice by the gogB mutant supports recent data showing that Salmonella takes advantage of intestinal inflammation for metabolic fitness by using tetrathionate as a respiratory electron acceptor [31] . What was unclear however was how the pathogen might temper this host inflammatory response once infection is established to limit the tissue damage that arises from an unchecked immune response . The acquisition of the type III effector GogB with anti-inflammatory activity appears to benefit Salmonella by limiting host tissue damage during of colonization , while maintaining a degree of inflammation that supports competition with the host microbiota . In this way , GogB appears to balance the pathogen's short-term need for inflammation-enhanced colonization while moderating this immune response to limit tissue damage in the longer term .
All experiments with animals were conducted according to guidelines set by the Canadian Council on Animal Care . The local animal ethics committee , the Animal Review Ethics Board at McMaster University , approved all protocols developed for this work . All bacterial strains used in this study are listed in Table S1 . S . enterica serovar Typhimurium strain SL1344 was used in this study and all mutant strains were derivatives of Salmonella SL1344 . The generation of the gogB mutant was described previously [7] . Salmonella was grown in Luria-Bertani ( LB ) broth or on agar plates with appropriate antibiotics . E . coli strain BL21 ( DE3 ) was used for expression and purification of recombinant GST-tagged GogB and truncated mutants . RAW264 . 7 and HeLa cells were routinely cultured in Dulbecco's Modified Eagle Medium ( DMEM ) supplemented with 10% fetal bovine serum ( FBS ) and grown at 37°C and 5% CO2 . The plasmids and primers used in this study are listed in Table S1 . The generation of the plasmid pWSK129-gogB-2HA encoding a hemagglutinin ( HA ) -tagged gogB that is expressed under its native promoter was described previously [7] . To generate the epitope-tagged mutant gogB-NT1–253 , the region encoding amino acid residues 1 to 253 of gogB was PCR amplified using the plasmid pWSK129-gogB-2HA as a template and the product was digested with SalI and BglII and cloned into the corresponding sites of pWSK129-2HA in which the gogB coding region was removed . Strand-overlap PCR was used to construct the HA-tagged mutant gogB-CT254–497 . The gogB native promoter and the secretion signal sequence encoding the first 29 amino acids at the N-terminus were amplified by PCR . A second PCR was performed to amplify the gogB region encoding the C-terminus from amino acid residues 254 to 497 . The resulting products were fused together by a third round of PCR , digested with SalI and BglII and cloned into the corresponding sites of pWSK129-2HA . The conserved leucine ( L270 ) and proline ( P271 ) in GogB , and L47 and P48 in GogB-CT254–497 were substituted with alanine residues using the QuikChange site-directed mutagenesis kit ( Stratagene ) and the plasmids pWSK129 gogB-2HA and pWSK129 gogBCT254–497-2HA as templates . In-frame deletion of the identified F-box motif in GogB was done by strand-overlap PCR in which the upstream gogB promoter and the region encoding amino acid residues 1 to 264 were amplified . A separate PCR was performed to amplify the region spanning amino acid residues 353 to 497 . The resulting products were fused together by PCR , digested with SalI and BglII and cloned into the corresponding sites of pWSK129-2HA . The plasmids encoding epitope-tagged GogB and GogB mutants were transformed into ΔgogB . An in-frame marked deletion of gogB and sspH2 or sseL to produce a double gene knockout strain was performed by lambda red recombination . To generate GST-tagged gogB and the truncated mutants gogBNT1–253 and gogBCT254–497 , the corresponding regions were amplified by PCR using the pWSK129 templates described above ( Table S1 ) . The PCR products were digested with SalI and BamHI and cloned into the corresponding sites of the plasmid pGEX-6P-1 . RAW264 . 7 or HeLa were seeded into 24-well plates at 5×105 cells/ml 18 hr prior to infection . Overnight bacterial cultures were opsonized with 20% normal human serum and used to infect macrophages at a multiplicity of infection ( MOI ) of 10 or 20 bacteria per cell . To prepare highly invasive bacteria for infection of HeLa cells , overnight cultures were subcultured in LB and grown for 2 hr at 37°C with shaking . Bacteria were washed with PBS and resuspended in DMEM/FBS and used to infect epithelial cells at MOI of 10 or 20 bacteria per cell . After 30 min of infection , the media was replaced with DMEM/FBS containing 100 µg/ml gentamicin to kill extracellular bacteria and incubated for 1 . 5 hr . At 2 hr post-infection ( PI ) , cells were lysed with PBS/1% Triton X-100 to release intracellular bacteria and the number of colony-forming units ( CFU ) was determined by plating serially-diluted bacteria on LB agar plates supplemented with 50 µg/ml streptomycin . Infected cells were washed with PBS and incubated with media containing 10 µg/ml gentamicin for longer time points . Bacterial numbers were expressed as the ratio of intracellular CFU at 20 h to 2 h . The procedure for purification and solubilization of recombinant GogB , GogB-NT and GogB-CT GST fusion proteins was performed as described previously with several modifications [44] . E . coli BL21 ( DE3 ) were transformed with plasmids encoding GST alone or GST fusion proteins and grown in 1 liter LB with 100 µg/ml ampicillin to OD600∼0 . 7 at 37°C . Protein expression was induced with 0 . 1 mM isopropyl β-D-1-thiogalactopyranoside ( IPTG ) for 24 hr at 16°C . Cells were harvested , resuspended in lysis buffer containing 40 mM Tris pH 8 . 0 , 500 mM NaCl , 1 mM EDTA , 50 µg/ml lysozyme and protease inhibitor cocktail ( Roche ) . DTT was added to a final concentration of 5 mM then cells were lysed by sonication . A final concentration of 1% Triton X-100 was added to the lysate and incubated on ice for 20 min . The lysates were clarified by centrifugation and the supernatant was mixed with glutathione sepharose 4B matrix beads ( GE Healthcare ) previously equilibrated with 40 mM Tris pH 8 . 0 , 500 mM NaCl ( TBS ) . The mixture was incubated overnight at 4°C in a rotator-shaker then the column was washed with ten bed volume of TBS . Proteins were eluted by incubating beads overnight with elution buffer ( 40 mM Tris pH 8 . 0 , 500 mM NaCl , 1% Triton X-100 , 5 mM DTT and 50 mM reduced glutathione ) at 4°C . Proteins were dialyzed against TBS with 1% Triton X-100 , 5 mM DTT to remove excess glutathione and concentrated using an Amicon centrifugal filter ( Millipore ) . Alternatively , GST beads coupled to recombinant proteins were washed with TBS and stored at 4°C in storage buffer ( 40 mM Tris pH 7 . 5 , 500 mM NaCl , 1% Triton X-100 , 5 mM DTT , 10% glycerol ) until further use in GST pull-down assays . Purified recombinant GST-tagged GogB , GogB-NT , GogB-CT or GST were coupled to GST affinity beads . HeLa or RAW264 . 7 cells were lysed with lysis buffer containing 40 mM Tris pH 8 , 200 mM NaCl , 1% Triton X-100 and protease inhibitor cocktail ( TBST ) . Lysates were clarified by centrifugation and the supernatant was incubated with GST affinity beads overnight at 4°C with end-on-end mixing . The beads were washed under stringent conditions and resuspended in SDS sample buffer . Bound proteins were resolved by SDS-PAGE and analyzed by Western blot . Cell-based co-immunoprecipitation assays were performed by infecting HeLa or RAW264 . 7 cells with wild type SL1344 , ΔgogB or a gogB mutant strain containing a plasmid that encodes HA-tagged gogB or gogB mutants . At 20 h post-infection , cells were washed with PBS and lysed with TBST . Lysates were incubated with anti-HA affinity beads ( Roche ) overnight at 4°C with end-on-end mixing . The beads were washed under stringent conditions and resuspended in SDS sample buffer . Bound proteins were resolved by SDS-PAGE and analyzed by Western blot . RAW264 . 7 or HeLa cells were seeded into 96-well plates and after 16 h , cells were transfected with the reporter plasmid pNFκB-luc and the control plasmid pCMVβgal ( Clontech ) using Fugene HD reagent ( Roche ) according to manufacturer's recommendations . Twenty-four hours after transfection , cells were infected as described above . At 20 h post-infection , luciferase activity was measured using the Luc-Screen System ( Applied Biosystems ) and β-galactosidase levels were measured using Galacto-star system following the manufacturer's protocols . Luciferase signals were normalized to β-galactosidase levels and to colony forming units ( CFUs ) enumerated for each strain at 20 hr post-infection . Data were expressed as fold activation relative to cells infected with wild-type Salmonella containing an empty plasmid ( mean ± SEM , n≥3 ) . The short-interfering RNA ( siRNA ) sequence used to knockdown Skp1 expression was published previously [13] . HeLa cells were seeded into 96 well plates and after 16 h , cells were co-transfected with 2 pmol of the Skp1 siRNA ( GCA AGU CAA UUG UAU AGC AGA A and UUC UGC UAA UAC AAU UGA CUU GC ) , the reporter plasmid pNFκB-luc and the control plasmid pCMVβgal using Lipofectamine 2000 transfection reagent ( Invitrogen ) . For control experiments , HeLa cells were co-transfected with a negative control siRNA and the reporter plasmids . At 24 h post-transfection , cells were infected at an MOI of 20 and the luciferase activity was measured as described above . Western blot analysis of HeLa cell lysates was performed to confirm knockdown of Skp1 expression using anti-Skp1 antibody and anti-GAPDH as loading control . For co-immunoprecipitation assays , HeLa cells were seeded into 24 well plates and after 16 h , cells were transfected with 80 pmol Skp1 or control siRNA using Lipofectamine 2000 reagent for 24 hr . Cells were then infected with either ΔgogB or ΔgogB complemented with pgogB-2HA at an MOI of 20 . At 20 hr post-infection , cells were lysed and mixed with anti-HA affinity beads overnight at 4°C . Bound protein complexes were resolved by SDS-PAGE and analyzed by Western blot using rabbit anti-Skp1 , mouse anti-FBXO22 , and HRP-conjugated rat anti-HA . Lysates from infected cells were normalized by protein content and analyzed by Western blot to determine knockdown of Skp1 expression . Mouse anti-GAPDH antibodies were used as loading control . RAW264 . 7 cells were seeded in 96 well plates and infected as described above . At 20 h post-infection , the culture supernatants were assayed for interleukin-1-beta ( IL1β ) levels using an ELISA kit ( eBioscience ) . Data was expressed as the average IL1β concentration ( pg/mL ) normalized to CFUs obtained for each strain at 20 h post-infection . Assays were done in three separate experiments ( n = 3 ) . Statistical analyses were performed using a two-tailed student's t test . RAW264 . 7 cells were infected with wild type SL1344 and the ΔgogB strain as described above . At various time points , cells were lysed with PBS/1% Triton X-100 and the lysates were normalized for total protein content . Proteins were resolved by SDS-PAGE and IκBα levels were analyzed by Western blot . Immunoblotting was performed by resolving proteins or cell lysates in 8 or 10% SDS-PAGE and transferring to a PVDF membrane . Blots were analyzed using the following antibodies: rabbit anti-Skp1 ( Novus Biologicals ) , mouse anti-FBXO22 ( Abcam ) , rabbit anti-β-actin ( Imgenex ) , mouse anti-ubiquitin FK2 ( Enzo Life Sciences ) , rabbit anti-GST ( Bethyl Laboratories ) , rat anti-HA-HRP ( Roche ) , rabbit anti-IκBα ( Calbiochem ) , and mouse anti-GAPDH ( Novus Biologicals ) . RAW264 . 7 cells were seeded in 6 well plates at 2×106 cells/well and infected with wild type SL1344 , ΔgogB/pgogB , ΔgogB/pgogB Δ264–352 and ΔgogB strains at an MOI of 20 bacteria per cell in the presence of 10 µM MG-132 ( Sigma ) to inhibit proteasome degradation of ubiquitinated IκBα . Control wells were infected with the same strains in the presence of 0 . 2% DMSO . At 30 min post-infection , cells were washed with PBS and fresh media containing 10 µM MG-132 was added to the infected cells and incubated for 3 . 5 hr . Cells were then lysed with buffer containing 40 mM Tris pH 7 . 5 , 150 mM NaCl , 1% Triton X-100 , 1 mM EDTA , 10 µM MG-132 , 10 µM N-ethylmaleimide and protease inhibitor cocktail ( Roche ) . Lysates were normalized for protein content and mixed with IgA Dynabeads ( Invitrogen ) previously coupled to rabbit anti-IκBα following manufacturer's instructions . The mixtures were incubated overnight at 4°C with end-on-end mixing . Beads were then washed three times with TBS/1% Triton X-100 followed by TBS and resuspended in SDS sample buffer . Protein samples were resolved by SDS-PAGE and analyzed by Western blot using HRP-conjugated mouse anti-ubiquitin FK2 ( Enzo Life Sciences ) and rabbit anti-IκBα ( Calbiochem ) . Protocols for the infection of experimental animals were approved by the Animal Research Ethics Board at McMaster University and in accordance to guidelines from the Canadian Council on the Use of Laboratory Animals . Competition assays were performed as previously described [45] . Groups of five female C57BL/6 mice were orally inoculated with a mixed inoculum of wild type Salmonella and ΔgogB resistant to chloramphenicol ( ∼1×106 CFU per strain ) . At 72 h post-infection , the bacterial loads were determined by plating tissue homogenates from the liver , spleen , and cecum . The competitive index ( CI ) was computed as the ratio of mutant/wild type CFU in the output versus input . 129/svImJ mice that are resistant to Salmonella due to a homozygous Nrampr allele were used as experimental animals for chronic Salmonella infections . Groups of five female 129/svImJ mice were orally inoculated with 1×107 CFU of either wild type Salmonella or the ΔgogB strain . After 4 , 7 , 39 , and 60 days of infection , bacterial loads were determined by plating tissue homogenates of the liver , spleen , mesenteric lymph nodes ( MLN ) , and cecum . In some experiments , mice were immunosuppressed prior to infection using dexamethasone in the drinking water as described previously [46] . Data were plotted as geometric means of log-transformed CFU per mg of tissue . Statistical analyses were performed using a nonparametric , two-tailed Mann-Whitney t test . A portion of the distal cecal tip of experimental animals was fixed in 4% paraformaldehyde ( PFA ) for 72 h , then transferred to 70% ethanol for 48 h prior to paraffin embedding , sectioning , and hematoxylin-eosin staining . Methodology for histology scoring was performed as previously described [47] . Nrampr mice were orally infected with 107 CFU of wild type and ΔgogB Salmonella . At day 4 post-infection , total RNA was extracted from the cecum and MLN using Trizol reagent ( Invitrogen ) following the manufacturer's protocol . After determining the purity of the extracted RNA , 5 µg of RNA was reverse-transcribed to cDNA using random hexamers and Superscript III reverse transcriptase ( Invitrogen ) according to manufacturer's protocols . RT-PCR was performed in a Lightcycler 480 ( Roche ) using the Lightcycler 480 SYBR I Master Mix ( Roche ) and specific primers listed in Table S2 . RT-PCR was performed in triplicate with the following settings: pre-incubation at 95°C for 5 min and 60 cycles of amplification at 95C° for 10 s , 55°C for 10 s , and 72°C for 20 s . Melting curve analysis was done at 95°C for 5 s , 65°C for 1 min and at 97°C followed by cooling to 40°C for 10 s . Data analysis was performed using the Lightcycler 480 version 1 . 5 software ( Roche ) . Relative fold change in expression was calculated using the formula based on the Pfaffl method: Ratio = [ ( Etarget ) ΔCt target ( control-treated ) ]/[ ( Eref ) ΔCt ref ( control-treated ) ] where E is the primer efficiency determined for specific primers and ΔCt is the difference in the mean crossing thresholds from control uninfected and infected mice . 18S rRNA was used as the reference gene .
|
Bacterial pathogens have evolved sophisticated ways to subvert the innate defenses of their host . One way in which pathogens do so is by blocking or dampening the inflammatory response that is triggered once a microorganism is detected by the innate immune system . In this way , the microorganism can limit the activation of innate defenses in the host to promote its own colonization and dissemination . In this work we found that the enteric human pathogen Salmonella enterica serovar Typhimurium limits the activation of innate immune defenses in the host by using a bacterial protein called GogB to interfere with NFκB activation . NFκB is a key human transcription factor involved in the expression of pro-inflammatory cytokines during infection . In this infection situation , Salmonella delivers GogB to the infected cell where it interferes with ubiquitination of the NFκB inhibitor protein called IκBα to prevent translocation of NFκB to the nucleus where it would normally activate pro-inflammatory gene expression . The anti-inflammatory property of GogB is important for the bacteria to reach optimal infection densities in host tissues and to actively limit the tissue damage associated with prolonged inflammatory responses .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"inflammation",
"immunity",
"microbial",
"pathogens",
"biology",
"microbiology",
"host-pathogen",
"interaction",
"microbial",
"growth",
"and",
"development",
"bacterial",
"pathogens"
] |
2012
|
GogB Is an Anti-Inflammatory Effector that Limits Tissue Damage during Salmonella Infection through Interaction with Human FBXO22 and Skp1
|
Plus-stranded RNA viruses replicate in infected cells by assembling viral replicase complexes consisting of viral- and host-coded proteins . Previous genome-wide screens with Tomato bushy stunt tombusvirus ( TBSV ) in a yeast model host revealed the involvement of seven ESCRT ( endosomal sorting complexes required for transport ) proteins in viral replication . In this paper , we show that the expression of dominant negative Vps23p , Vps24p , Snf7p , and Vps4p ESCRT factors inhibited virus replication in the plant host , suggesting that tombusviruses co-opt selected ESCRT proteins for the assembly of the viral replicase complex . We also show that TBSV p33 replication protein interacts with Vps23p ESCRT-I and Bro1p accessory ESCRT factors . The interaction with p33 leads to the recruitment of Vps23p to the peroxisomes , the sites of TBSV replication . The viral replicase showed reduced activity and the minus-stranded viral RNA in the replicase became more accessible to ribonuclease when derived from vps23Δ or vps24Δ yeast , suggesting that the protection of the viral RNA is compromised within the replicase complex assembled in the absence of ESCRT proteins . The recruitment of ESCRT proteins is needed for the precise assembly of the replicase complex , which might help the virus evade recognition by the host defense surveillance system and/or prevent viral RNA destruction by the gene silencing machinery .
Plus-stranded ( + ) RNA viruses replicate in the infected cells by assembling viral replicase complexes consisting of viral- and host-coded proteins in combination with the viral RNA template . Although major progress has recently been made in understanding the functions of the viral replication proteins , including the viral RNA-dependent RNA polymerase ( RdRp ) and auxiliary replication proteins , the contribution of host proteins is poorly documented [1] , [2] , [3] , [4] . Genome-wide screens to identify host factors affecting ( + ) RNA virus infections , such as Brome mosaic virus ( BMV ) , Tomato bushy stunt virus ( TBSV ) , West Nile virus and Droshophila virus C , in yeast and animal model hosts led to the identification of host proteins including ribosomal proteins , translation factors , RNA-modifying enzymes , proteins of lipid biosynthesis and others [2] , [3] , [5] , [6] , [7] , [8] , [9] . The functions of the majority of the identified host proteins in ( + ) RNA virus replication have not been fully revealed . TBSV is a small ( + ) RNA virus that infects a wide range of host plants . TBSV has recently emerged as a model virus to study virus replication , recombination , and virus - host interactions due to the development of yeast ( Saccharomyces cerevisiae ) as a model host [10] , [11] , [12] , [13] . Systematic genome-wide screens covering 95% of yeast genes have led to the identification of over 100 host genes that affected either TBSV replication or recombination [5] , [7] , [14] , [15] . Moreover , proteomics analysis of the highly purified tombusvirus replicase complex revealed the presence of the two viral replication proteins ( i . e . , p33 and p92pol ) and 6–10 host proteins in the replicase complex [16] , [17] , [18] . These host proteins have been shown to bind to the viral RNA and the viral replication proteins [1] , [17] , [19] . The auxiliary p33 replication protein has been shown to recruit the TBSV ( + ) RNA to the site of replication , which is the cytosolic surface of peroxisomal membranes [20] , [21] , [22] . The RdRp protein p92pol binds to the essential p33 replication protein that is required for assembling the functional replicase complex [12] , [22] , [23] , [24] . Genome-wide screens for host factors affecting TBSV replication in yeast [5] , [7] has led to the identification of seven ESCRT proteins involved in multivesicular body ( MVB ) /endosome pathway [25] , [26] . The identified host proteins included Vps23p and Vps28p ( ESCRT-I complex ) , Snf7p and Vps24p ( ESCRT-III complex ) ; Doa4p ubiquitin isopeptidase , Did2p having Doa4p-related function; and Vps4p AAA-type ATPase [5] . The identification of ESCRT proteins supports the idea that tombusvirus replication could depend on hijacking of ESCRT proteins , thus promoting efforts to test their roles in TBSV replication in this paper . Recruitment of ESCRT proteins for TBSV replication might facilitate the assembly of the replicase complex , including the formation of TBSV-induced spherules and vesicles in infected cells [27] . Induction of membranous spherule-like replication structures in infected cells might be common for many plus-stranded RNA viruses [28] . The endosome pathway is a major protein-sorting pathway in eukaryotic cells , which down- regulates plasma membrane proteins via endocytosis; and sorts newly synthesized membrane proteins from trans-Golgi vesicles to the endosome , lysosome or the plasma membrane [29] , [30] , [31] . The ESCRT proteins in particular have a major role in sorting of cargo proteins from the endosomal limiting membrane to the lumen via membrane invagination and vesicle formation . Defects in the MVB pathway can cause serious diseases , including cancer , defect in growth control and early embryonic lethality [29] , [30] , [31] , [32] . In addition , various viruses , such as enveloped retroviruses ( HIV ) , ( + ) and ( − ) RNA viruses ( such as filo- , arena- , rhabdo and paramyxoviruses ) usurp the MVB pathway by redirecting ESCRT proteins to the plasma membrane , leading to budding and fission of the viral particles from infected cells [25] , [26] . The first step in the endosome pathway is the monoubiquitination of cargo proteins , which serves as a signal for proteins to be sorted into membrane microdomains of late endosomes [30] , [31] , [32] . The ubiquitinated cargo protein is bound by Vps27p ( Hrs protein in mammals ) , which in turn recruits Vps23-containing ESCRT-I complex . Then , the ESCRT-I-complex recruits ESCRT-II complex , which in turn recruits the large ESCRT-III complex . The proposed role of the ESCRT-III complex is grouping the cargo proteins together in the limiting membranes of late endosomes and deforming the membranes that leads to membrane invagination into the lumen [33] , [34] . Then , Vps4p recycles the ESCRT proteins , whereas Doa4p recycles the ubiquitin , leading to budding of multiple small vesicles into the lumen of endosome and to MVB formation . Fusion of the limiting membrane of MVB with the lysosome/vacuole will release the lipid/protein content of MVB into the lysosome , or alternatively , by fusing to the plasma membrane ( exocytosis ) , it releases its content outside the cell [30] , [31] , [32] . In this paper , we show that ESCRT proteins previously identified in a genome-wide screen in yeast for affecting TBSV replication are involved in tombusvirus replication in plants . We demonstrate that Vps23p , which is a key ESCRT-I adaptor protein recognizing ubiquitinated cargo proteins , and Bro1p accessory ESCRT protein bind to the TBSV p33 replication protein , which could be critical for p33 replication protein to recruit Vps23p and Bro1p , allowing TBSV to usurp additional ESCRT proteins for replication . We also show that functional ESCRT proteins are needed for optimal replicase activity and protection of the viral RNA template within the tombusvirus replicase from a ribonuclease in vitro . These data are consistent with the model that TBSV co-opts ESCRT proteins for its replication .
To test the roles of ESCRT proteins in tombusvirus replication in a plant host , we have expressed dominant negative mutants of selected ESCRT factors in N . benthamiana . This approach has been facilitated by the availability of dominant negative mutants of ESCRT genes in mammals and yeast [25] , [35] . Expression of dominant negative mutants might inhibit the function of the endogenous ESCRT proteins in N . benthamiana , albeit the number of VPS23 and other ESCRT genes are not known in N . benthamiana , whose genome is not yet sequenced completely . To this end , we cloned ten ESCRT genes , including AtVPS4 , AtVPS24 , AtVPS36 , AtBRO1 , and two genes for AtVPS23 , AtVPS28 , and AtSNF7 from Arabidopsis thaliana , a model plant with known sequence , which is not infected by TBSV or the closely related Cucumber necrosis virus ( CNV ) . First , we have made dominant negative mutants of two AtVPS23 via deletion of the N-terminal UEV ( ubiquitin E2 variant ) domain . We have found that co-expression of CNV genomic ( g ) RNA with either AtVPS23-1dn or AtVPS23-2dn in N . benthamiana leaves from the constitutive 35S promoter via agroinfiltration led to inhibition of CNV gRNA replication in the infiltrated leaves ( down to 25–30% , Fig . 1A , lanes 5–8 ) . Expression of the wt AtVPS23-1 in N . benthamiana did not inhibit CNV RNA accumulation when compared with the samples based on agroinfiltration with the empty vector ( Fig . 1A , lanes 3-4 versus 1-2 ) . Altogether , inhibition of CNV gRNA replication by the dominant negative Vps23p mutants supports the idea that Vps23p play a significant role in tombusvirus replication in a plant host . To test the effect of additional dominant negative mutants in the ESCRT pathway , we generated a dominant negative mutant of AtVPS4 by changing the highly conserved K178 to A , which has been shown to inhibit the ATPase activity required for disassembly and release of the ESCRT proteins from the endosomal membranes , resulting in strong inhibition of ESCRT functions [25] , [35] . We have found that co-expression of CNV with AtVPS4 ( K178A ) in N . benthamiana leaves led to dramatic inhibition of CNV gRNA replication in the infiltrated leaves ( down to 4% , Fig . 1B , lanes 5–6 ) , whereas expression of the wt AtVPS4 in N . benthamiana inhibited CNV RNA accumulation only by 21% ( Fig . 1B , lanes 3–4 ) . In addition , expression of one AtVPS24 gene and two AtSNF7 genes with C-terminal deletions , which are known to interfere with proper ESCRT-III functions [36] , inhibited CNV gRNA accumulation by 89–95% in the infiltrated leaves ( Fig . 1B ) . Expression of AtBRO1 mutant inhibited CNV replication to a lesser extent in N . benthamiana ( by ∼60% Fig . 1C , lanes 11–14 ) , whereas AtVPS28 and AtVPS36 mutants did not significantly affect CNV replication ( Fig . 1C , lanes 1–10 ) . Altogether , these data support the model that several ESCRT components are involved in tombusvirus replication in plants . Since the ESCRT proteins are involved in membrane bending/invagination [33] , [34] , it is possible that they are used by tombusviruses during the assembly of the membrane-bound replicase complexes [1] . Accordingly , we observed the formation of the characteristic spherule-like structures ( which likely serve as the sites of tombusvirus replication ) in reduced number in plants actively replicating TBSV repRNA and expressing the dominant negative mutants AtVPS4 ( K178A ) and AtSNF7-1 , respectively , in N . benthamiana leaves when compared with the control samples ( Fig . S1 and Protocol S1 ) . To exclude the possibility that the above inhibitory effect of the ESCRT protein mutants on tombusvirus replication is due to unwanted cytotoxic effect of the expressed ESCRT proteins , we tested the replication of a distantly related RNA virus , Tobacco rattle virus ( TRV ) , in similarly treated N . benthamiana leaves . These experiments revealed the lack of inhibition of TRV RNA accumulation in leaves expressing the dominant negative ESCRT proteins ( Fig . 1A-B , lower panels ) , suggesting that the inhibitory effects of these mutant proteins are specific to CNV RNA replication . To test if the above dominant negative ESCRT mutants affect the activity of the tombusvirus replicase , we isolated the membrane-bound tombusvirus replicase from N . benthamiana expressing selected mutated ESCRT proteins . The tombusvirus replication proteins as well as the plus-stranded ( + ) DI-72 replicon ( rep ) RNA were expressed from separate expression plasmids in the above plants . The activity of the isolated tombusvirus replicase was tested in vitro on the co-purified repRNA ( Fig . 2 ) . These experiments revealed that the expression of dominant negative mutants of AtVPS4 , AtVPS24 , and AtSNF7-1 factors in plants inhibited the activity of the isolated replicase by ∼70–75% ( Fig . 2 , lanes 3–8 ) , whereas AtVPS23-1dn had ∼40% inhibitory effect ( lanes 9–10 ) . Expression of the full-length AtVPS24 and AtVPS4 did not show as much inhibitory effect on the activity of the isolated replicase as the AtVPS24 and AtVPS4 ( K178A ) dominant negative mutants ( Fig . S2 ) . Western blot analysis revealed that expression of dominant negative mutants of AtVPS4 , AtVPS24 , and AtSNF7-1 factors did not affect TBSV p33 level , while inhibited the accumulation of p92pol replication protein ( Fig . 2 , bottom panel ) , which could be partially responsible for the reduced activity of the TBSV replicase . The lesser inhibitory effect of the ESCRT dominant negative mutants on the tombusvirus replicase activity ( Fig . 2 ) than on viral RNA accumulation ( Fig . 1 ) is likely due to the uncoupled expression of the p33 and p92pol viral replication proteins and the viral replicon RNA from separate plasmids in plants used for the replicase assay , while expression of the p33/p92pol is coupled to viral RNA level in the viral RNA accumulation assay . To further test the possible roles of ESCRT proteins in the assembly of the tombusvirus replicase complex , we used yeast model host , since yeast strains with deletion of ESCRT genes are available . To find out if the activity of the tombusvirus replicase is inhibited in vps23Δ yeast , we isolated the pre-assembled tombusvirus replicase from yeast cells expressing the wt p33 and p92pol replication proteins and a TBSV repRNA , followed by testing for the replicase activity in vitro . Under the assay conditions , the pre-assembled tombusvirus replicase in the membrane-enriched fraction uses the co-purified repRNA as template for RNA synthesis , which is measured by denaturing PAGE analysis . We found that the pre-assembled tombusvirus replicase from vps23Δ yeast supported TBSV RNA synthesis at ∼40% of the level obtained with similar amount of replicase from the wt yeast ( Fig . 3A , lanes 3–4 versus 1–2 ) . These results have demonstrated that the tombusviral replicase was less active when formed in vps23Δ yeast , suggesting that Vps23p could be involved in the assembly of the viral replicase . In the second assay , we expressed wt p33 and p92pol replication proteins in yeast lacking vps4 , snf7 or vps24 , followed by isolation of the membrane fraction carrying these viral proteins . Although the viral replication proteins associate with the membranes , they cannot form active replicase in the absence of the viral template [12] , [24] , [37] . Then , we added the DI-72 ( + ) RNA to the isolated membrane fraction to assemble the functional tombusvirus replicase in vitro [23] , followed by replicase activity assay . In this assay , the tombusvirus replicase supports complete cycle of viral RNA synthesis in vitro [23] . These experiments revealed that the yeast extract prepared from vps4Δ , snf7Δ or vps24Δ yeast supported TBSV replication only at ∼20% level when compared to the wt yeast ( Fig . 3B ) . Overall , these data demonstrated that the effect of ESCRT proteins on the tombusvirus replicase is similar in yeast and plant extracts , supporting an important role for ESCRT proteins in tombusvirus replication . Replication of the TBSV RNA , including ( − ) - and ( + ) -strand synthesis , takes place in a membrane-bound replicase complex that provides protection against ribonucleases [23] . Since our model proposes a role of Vps23p/ESCRT proteins in facilitating the precise assembly of the tombusvirus replicase , we predicted that the tombusvirus replicase might become more sensitive to a ribonuclease if assembled in the absence of an ESCRT factor . To test this model , we isolated the membrane-bound replicase from vps23Δ or vps24Δ yeast strains expressing wt p33/p92pol/repRNA , which was followed by RNase A nuclease treatment that should destroy the unprotected viral RNA . Then , we used strand-specific Northern blot analysis to estimate the amount of protected ( − ) repRNA , which is associated with the replicase [22] , [23] in the samples . These experiments revealed that only ∼20% of the ( − ) repRNA survived the treatment when the membrane was derived from vps23Δ yeast in contrast with 46% in the wt control samples ( Fig . 3C ) . In addition , we have tested the repRNA in the membrane-fraction from vps24Δ yeast for ribonuclease sensitivity , since Vps24p is an important ESCRT-III factor affecting TBSV replication [5] . The protected ( − ) repRNA was only 8% in these samples ( Fig . 3C ) . The simplest interpretation of the enhanced sensitivity of the ( − ) repRNA within the viral replicase assembled in the absence of Vps23p or Vps24p is that the viral replicase complex ( possibly the whole spherule ) assembles less precisely in the absence of recruitment of ESCRT proteins , thus making the viral RNA within the membrane-bound replicase-complex more accessible to a ribonuclease . To confirm the increased sensitivity of the tombusvirus replicase complex to ribonucleases when ESCRT factors are inhibited in N . benthamiana , we have developed a novel approach for targeted degradation of minus-stranded RNA replication intermediate via RNA interference ( RNAi ) . We chose the ( − ) repRNA as a target , since it has been shown to be always part of the membrane-bound replicase complex and it is protected from ribonucleases [22] , [23] . We introduced two microRNA171 ( miR171 ) sequences to the repRNA in such a way that the miR171 sequences were active targets to the RNAi machinery only when present in the ( − ) repRNA ( Fig . 4 ) . The repRNA carrying the miR target sequences ( called DI-miR , Fig . 4A ) accumulated to ∼60% level of the wt repRNA ( DI-72 ) lacking the miR171 target sequence ( Fig . 4B , lanes 1-2 versus 3–4 ) in the control plants . On the other hand , DI-miR RNA accumulated only to 10% and 21% in plants expressing the dominant negative ESCRT-III factors , AtVPS24 , and AtSNF7-1 , respectively ( Fig . 4B , lanes 5–12 ) . Expression of AtVPS4 ( K178A ) decreased DI-miR accumulation moderately when compared with the control plants ( 48% for DI-miR RNA versus 59% for DI-72 RNA , Fig . 4B ) . The greatly reduced accumulation of DI-miR repRNA in comparison with DI-72 repRNA can be explained with increased sensitivity of ( − ) DI-miR RNA to the RNAi machinery when dominant negative ESCRT-III factors are expressed . This , in turn , supports the model that the tombusvirus replicase complexes are assembled less precisely in these plants , making them more accessible to targeted ribonucleases . Overall , these data support the role of ESCRT-III proteins in the precision/quality of viral replicase assembly . Since the above experiments demonstrated that ESCRT proteins affect the activity of the tombusvirus replicase , we wanted to test if interaction between p33 replication co-factor and the ESCRT proteins occurs that could facilitate the recruitment of ESCRT factors for tombusvirus replication . We performed the split-ubiquitin assay , a variant of the yeast two-hybrid approach [38] , which can detect protein-protein interaction on the surface of cellular membranes , where p33 is normally localized [22] , [27] with selected yeast ESCRT proteins . We found that Vps23p ESCRT-I and Bro1p accessory ESCRT factors interacted with p33 in the split ubiquitin assay ( Fig . 5A–B ) , whereas the other ESCRT proteins did not ( not shown ) . Additionally , we found that the N-terminal UEV domain of the yeast Vps23p was sufficient for interaction with p33 ( Fig . 5C ) . Also , the UEV domains from two Arabidopsis and two Nicotiana homologues of Vps23p interacted with the p33 replication co-factor ( Fig . 5C ) . The interaction between p33 and either Vps23p or Bro1p was much weaker than the interaction between p33 and Ssa1p , the yeast heat shock protein 70 , which is a resident protein in the tombusvirus replicase ( Fig . 5A–B ) [17] , [37] , [39] . The weak interaction with p33 suggests that Vps23p and Bro1p might only interact with p33 in a temporary fashion or only a small portion of p33 is involved in these interactions . To further demonstrate that the interaction between p33 and Vps23p as well as Bro1p can take place in yeast cells , we co-expressed p33 replication protein tagged with FLAG and 6xHis ( termed p33HF ) with either the UEV domain of Vps23p or Bro1p . In this experiment , the HA-tagged UEV domain of Vps23p and Bro1p were expressed from the original chromosomal locations and the native promoters . Purification of p33HF on a FLAG-column , followed by Western analysis revealed that UEV-HA ( Fig . 6A , lane 2 ) and Bro1-HA ( Fig . 6B , lane 2 ) were co-purified with p33HF . Similar purification experiments on the FLAG-affinity column with p33H tagged with 6xHis only resulted in only minor amounts of nonspecifically-bound UEV-HA or Bro1-HA ( Fig . 6A–B , lane 1 ) , demonstrating that specific interaction between p33 and the UEV domain as well as Bro1p occurs in yeast . To test the subcellular compartment where p33 - Vps23p interaction takes place , we co-expressed the 6xHis-tagged p33 with Vps23p tagged with green-fluorescent protein ( GFP ) from its native promoter and chromosomal location in combination with Pex13p tagged with red fluorescent protein ( RFP ) , a marker for the peroxisomal membrane [40] . Laser confocal microscopy analysis revealed that Vps23p-GFP was present in the cytosol in the absence of p33 ( Fig . 7B ) . However , 15 min after the induction of p33 from the CUP1 promoter , we observed the partial re-distribution of Vps23p-GFP to the peroxisomal membrane ( Fig . 7A ) . The co-localization of Vps23p-GFP and Pex13p-RFP in cells expressing p33 is in agreement with the model that p33 is involved in re-targeting , at least temporarily , Vps23p to the peroxisomes , the sites of TBSV replication , at the beginning of replication .
Host factors likely play key roles during the assembly of viral replicases in infected cells [1] , [2] . In this work , we have shown that a set of ESCRT proteins is critical for optimal tombusvirus replication and the assembly of the membrane-bound tombusvirus replicase complex . We found that over-expression of dominant negative mutants of ESCRT-III and Vps4p in plants reduced the accumulation of tombusvirus RNA by 10–20-fold ( Fig . 1 ) , inhibited the tombusvirus replicase in vitro ( Fig . 2 ) and reduced the number of spherules formed in infected cells ( Fig . S1 ) . This inhibition seems to be specific for tombusviruses , since the distantly related TRV RNA accumulation was not inhibited in these plants . The inhibitory effect on tombusvirus replication by the over-expressed dominant negative ESCRT mutants seems to be direct , since the activity of the tombusvirus replicase was also reduced when isolated from these plants ( Fig . 2 ) . Similarly , the activity of the tombusvirus replicase assembled in vitro was inhibited when we used the cellular membrane fraction from yeast lacking ESCRT-III or Vps4p proteins ( Fig . 3B ) . In addition , the replicase from vps24Δ yeast was more sensitive to RNase treatment than the replicase preparation obtained from wt yeast , suggesting that Vps24p ESCRT-III protein is important to assemble RNase-insensitive replicase complexes . Moreover , DI-miR repRNA carrying the miR171 target sequence in the ( - ) strand RNA replicated poorly in N . benthamiana leaves expressing dominant negative ESCRT-III mutants ( Fig . 4B ) . These data suggest that the ( - ) strand repRNA in the replication intermediate within the viral replicase complex became more accessible to ribonuclease cleavage when the replicase was assembled in the presence of dominant negative ESCRT-III mutants . Altogether , the presented data support a role for ESCRT-III and Vps4p proteins in the formation of active tombusvirus replicase in plants and in yeast as well . Intriguingly , the role of ESCRT proteins seems to control the quality of the replicase complex assembly , making the viral RNAs within replicase complex more protected from ribonucleases . Based on these observations , we propose that ESCRT proteins help tombusviruses hide from host defense recognition and/or avoid the attack by the host defense during viral replication . We also show that the recruitment of the ESCRT factors for virus replication is likely driven by interaction between the auxiliary p33 replication cofactor and Vps23p ESCRT-I protein and Bro1p accessory ESCRT protein . These interactions seem to be important for tombusvirus replication in yeast ( Fig . S3 ) as well as in plant cells ( Fig . 1 ) . The interaction between p33 and Vps23p depends on the N-terminal UEV domain in Vps23p and p33 , which is monoubiquitinated [18] . The ubiquitination of p33 may play a role in interaction with Vps23p since it has been shown that Vps23p binds to monoubiquitinated proteins [30] , [41] . By binding directly to Vps23p or Bro1p , p33 might be able to recruit additional ESCRT factors for tombusvirus replication as discussed below . Interestingly , Human immunodeficiency virus ( HIV ) and other enveloped retroviruses co-opt ESCRT components through direct interaction with Tsg101 , the human homologue of Vps23p , and with Alix , a homologue of Bro1p [41] , [42] . Tsg101 and Alix play redundant roles in this process [25] . HIV gag protein also interacts with Nedd4 E3 ubiquitin ligase protein that could complement Tsg101 or Alix during HIV budding [42] . We have shown previously that p33 can bind to Rsp5p , the yeast homologue of Nedd4 [18] , [43] . This indicates that different viruses seem to exploit the ESCRT proteins through co-opting Vps23p ( Tsg101 ) , Bro1p ( Alix ) and/or Rsp5p ( Nedd4 ) via direct protein-protein interactions . In addition , the ESCRT machinery is also recruited for cell division to separate the daughter cells from the mother cells through the process of cytokinesis . It has ben shown that the midbody resident protein Cep55 interacts with Tsg101 and Alix to recruit additional ESCRT proteins [41] . Interestingly , these cellular processes , such as MVB biogenesis , cytokinesis and HIV virion budding as well as the spherule formation during the assembly of the tombusvirus replicase , are based on topologically similar membrane invaginations ( membrane deformation occurring away from the cytosol ) . Co-opting ESCRT proteins for these processes could be critical , since Snf7p and other ESCRT-III proteins have been shown to be involved in membrane deformation in vivo and in vitro as well [33] , [34] . Thus , the interaction of HIV gag protein , Cep55 midbody protein and the tombusvirus p33 replication protein with members of the endosomal pathway shows close parallel and mechanistic similarities , albeit the ESCRT proteins would be used for different processes , such as either virus budding , cytokenesis or viral RNA replication . Another novel feature is that Vps23p , in the presence of p33 replication co-factor , is shown to re-localize temporarily to the peroxisomal membrane , which represents the place of tombusvirus replicase assembly [20] , [22] , [27] ( Fig . 7 ) . After brief recruitment , Vps23p seems to be released from the replicase , because we did not find Vps23p in the highly-purified functional replicase complex [17] and it was not co-localized with a peroxisomal marker at latter time points after induction of p33 expression ( not shown ) . The relatively weak interaction between Vps23p and p33 as well as the low percentage of ubiquitinated p33 [18] could be useful during virus replication to optimize the number of Vps23p recruited into each replicase complex . Indeed , based on replicases of other plus-strand RNA viruses [28] , it is predicted that 100-to-200 p33 molecules are likely needed for the formation of a single replicase complex , whereas only a few Vps23p molecules should be recruited temporarily for each replicase complex . In addition , it is likely that Vsp23p and Bro1p and possibly Rsp5p could play complementary roles in recruiting additional ESCRT factors as shown in case of HIV gag for virion budding [33] , [34] . Weak interactions between Vps23p - p33 and Bro1p - p33 could help recycling these host proteins that should not be present in the fully assembled replicase complex [17] . On the contrary , host factors that are permanent residents in the replicase , such as heat shock protein 70 ( the yeast Ssa1p protein ) [17] , bind more efficiently to p33 as shown in Fig . 5 . We propose that p33 - Vps23p interaction is important for the optimal assembly of the tombusvirus replicase , because the membrane-bound tombusvirus replicase preparation obtained from vps23Δ yeast supported low TBSV repRNA replication ( Fig . 3A ) . Further support on the role of p33 - Vps23p interaction in replicase assembly comes from data obtained using a membrane-enriched fraction containing the viral replicase prepared from vps23Δ yeast ( Fig . 3C ) . The protection of the ( - ) repRNA associated with the replicase complex is likely due to the repRNA becoming inaccessible as part of the assembled replicase complex [23] . We found that the viral ( − ) repRNA within the replicase complex obtained from vps23Δ yeast was more sensitive to RNase treatment than the replicase preparation obtained from wt yeast . Thus , similar to vps24Δ yeast discussed above , the viral replicase , which is located inside the spherules , assembles less precisely in vps23Δ than in wt yeast . Overall , these data are compatible with the model that p33 - Vps23p interaction could be important during the replicase assembly process and/or affect the structure of the replicase complexes , which could determine the accessibility of the repRNA to RNases during replication . However , the effect of VPS23 deletion ( Fig . 3 ) or over-expression of a dominant negative mutant of Vps23 homologue in N . benthamiana plants ( Fig . 1 ) was not as detrimental to virus replication as deletion of ESCRT-III or VPS4 in yeast or over-expression of dominant negative mutants of ESCRT-III or Vps4p homologues in plants . This could be due to the redundant roles likely played by Vps23p , Bro1p and possibly Rsp5p in recruitment of ESCRT-III/Vps4p factors , based on the similar redundancy documented for subversion of ESCRT-III/Vps4p by HIV gag's interaction with Tsg101/Alix/Nedd4 [42] . Also , we do not know if the dominant negative mutants were able to block completely the function of every VPS23 gene , since the number of VPS23 genes is yet not known in N . benthamiana . It is likely that the most important aspect of recruitment of Vps23p and Bro1p for tombusvirus replication is the possibility to co-opt additional ESCRT proteins , including the ESCRT-III factors and Vps4p . Accordingly , the in vitro activity of the tombusvirus replicase is 3-5-fold lower when obtained from plants expressing dominant negative ESCRT-III/Vps4p ( Fig . 2 ) or from vps24Δ , snf7Δ , or vps4Δ yeast strains ( Fig . 3B ) . Since the ESCRT-III factors are involved in grouping the cargo proteins together in the membrane and they have been shown to deform the membrane [33] , [34] , we propose that these proteins could be useful to affect the formation/structures of the spherules for virus RNA replication . Indeed , TBSV replication induces the formation of spherules ( Fig . S1 ) [27] , which are topologically related to multivesicular bodies ( MVB ) , since both require membrane invagination into the lumen , away from the cytosol . This is opposite to the regular intracellular vesicle formation , which buds into the cytosol . Collectively , usurping a partial set of ESCRT factors by tombusvirus replication proteins might facilitate the optimal formation of active viral replicase complexes within the membranous spherules . Moreover , recruitment of Vps4p AAA ATPase [25] , [35] , could help re-cycling of the ESCRT factors from the replicase after the assembly . Thus , it is possible that the expression of dominant negative mutants of Vps24p and Snf7p ESCRT-III or Vps4p factors inhibit tombusvirus replication in plants by interfering with the proper assembly of the replicase complex . Accordingly , the ( − ) repRNA located within the viral replicase complex became more accessible to targeted ribonuclease cleavage when the replicase was assembled in the presence of dominant negative ESCRT-III mutants in plant leaves ( Fig . 4B ) . Based on the data presented here , we propose that the ESCRT machinery is recruited for tombusvirus replication in a unique way . The first step in recruitment of Vps23p is the ubiquitination of small percentage of p33 ( Ub-p33 ) ( Fig . 8 , step 1 ) [18] . Then , the ubiquitinated p33 ( Ub-p33 ) binds to the adaptor protein Vps23p ( step 2 ) or to Bro1p accessory ESCRT protein , followed by the recruitment of ESCRT-III proteins , Snf7p and Vps24p . The ESCRT-III proteins could then help the optimal assembly of the replicase complex , facilitate the grouping of p33 molecules together in the membrane and/or promote the formation of viral spherules by deforming the membrane ( membrane invagination ) ( step 3 ) . Then , Doa4p deubiquitination enzyme is predicted to remove ubiquitin from Ub-p33 , whereas Vps4p AAA ATPase could recycle the ESCRT proteins ( step 4 ) . Altogether , we suggest that these events could promote the optimal and precise assembly of the TBSV replicase complex , resulting in TBSV RNA replication , including ( − ) and ( + ) RNA syntheses , in a protected microenvironment and then the regulated release of the progeny ( + ) RNAs into the cytosol . Importantly , the precisely assembled replicase complex provides protection from recognition by the host defense surveillance system and/or viral RNA destruction by the gene silencing/RNAi machinery . Similar events might also take place in case of other ( + ) RNA viruses , which are also known to deform membranes and/or form spherules during replication .
A . thaliana VPS4 ( At2g27600 ) [44] was amplified by PCR using primers #2746 and #2749 ( Table S1 ) and A . thaliana total DNA as template . This product was digested with BamHI and XhoI and ligated into similarly digested pGD [45] to generate plasmid pGD-35S-AtVPS4 . To create the mutant version pGD-35S-AtVPS4 ( K178A ) , two PCR reactions were done using primer pairs #2746/#2748 and #2747/#2749 and A . thaliana total DNA as template . These PCR products were digested with NheI , ligated together and the product was re-amplified with primers #2746/#2749 . The obtained PCR product was digested with BamHI and XhoI and cloned into pGD . The 3′ terminal portion of the two VPS23 homologues from A . thaliana , namely At3g12400 ( aa 181–398 , named AtVPS23-1dn in Fig . 1 ) and At5g13860 ( aa 170–368 , named VPS23-2dn in Fig . 1 ) as well as the full length At3g12400 ( AtVPS23-1 ) [44] , were amplified with primers #2843/#2671; #2844/#2845; and #2750/#2671 , respectively . The VPS23 PCR products were digested with BamHI and SalI and cloned into BamHI/SalI-digested pGD-L . To construct pGD-L , the leader sequence from Tobacco etch virus was PCR-amplified using plasmid pTEV-7DA [46] with primers #2915/#2916 . The PCR product was digested with BglII and BamHI and cloned into BamHI-digested pGD to generate pGD-L . The ESCRT-III homologues from A . thaliana , namely VPS24 [At5g22950 ( aa 1–153 , named VPS24dn in Fig . 1 ) , At2g19830 ( aa 1–152 , named AtSNF7-1dn ) and At4g29160 ( aa 1–152 , named SNF7-2dn ) were amplified with primers pairs #2846/#2847; #2850/#2851; and #2852/#2853 , respectively , digested with BamHI and SalI and cloned into pGD . In addition , the 5′ terminal portion of two VPS28 homologues ( At4g05000 , aa 1–105 and At4g21560 , aa 1–104 ) , and the 3′ terminal portion of a VPS36 homologue ( At5g04920 , aa 175–440 ) [44] were amplified with primers #2867/#2868 , #2869/#2870 and #2871/#2872 , respectively and cloned into pGD using BamHI and SalI . The BRO1 homologue from A . thaliana ( At1g15130 ) was identified based on sequence similarity to yeast BRO1 and RIM20 and human AIP1/ALIX . A portion of At1g15130 ( aa 179–846 ) was amplified with primers #2883/#2884 and cloned into pGD using BamHI and SalI . Plasmid pGD-35S-20Kstop ( expressing a full length CNV RNA , but not the p20 protein ) was created by PCR using primers #532/#720 and pK2/M5 20K stop [47] as template . The PCR product was digested with BamHI/XhoI and ligated into BamHI/XhoI-digested pGD . A . tumefaciens strain C58C1 transformed with pGD-35S-20Kstop or one of the pGD constructs expressing various ESCRT genes were co-infiltrated onto N . benthamiana leaves at OD600 = 0 . 1 and 0 . 5 , respectively . Agroinfiltrations of N . benthamiana and analysis of viral RNA accumulation in the infiltrated leaves 2 . 5 days after infiltration were done as described [19] , [48] . To launch TRV replication in N . benthamiana , we used the TRV plasmids pTRV1 and pTRV2 [49] . A . tumefaciens transformed with plasmids expressing TRV1/TRV2 and one of the dominant negative ESCRT proteins were co-infiltrated into leaves using bacterial cultures at OD600 = 0 . 1 for TRV and OD600 = 0 . 5 for the ESCRT dominant negatives as described above for CNV . DI-72 ( containing a TRSV ribozyme sequence at the 3′ end ) was amplified from pYC/DI72sat [13] using primers #532/#1069 ( Table S1 ) . The PCR product was digested with BamHI and SacI and cloned into pGD to generate pGD/DI72sat . CNV p33 and p92 were amplified from plasmids pGBK-His33 or pGAD-His92 [12] with primers #1794/#1403 and #1794/#952 , respectively . The PCR products were digested with BamHI and XhoI and cloned into pGD-L ( described above ) to generate pGD-L-p33 and pGD-L-p92 . The plasmid pYC/DI72sat/2xmiR171 was designed to express a modified DI-72 repRNA containing two miR171 target sites [50] , between regions I and II and regions II and III , from a 35S promoter in N . benthamiana plants . The orientation of both copies of miR171 sequence was to allow cleavage of the target when present in the ( − ) strand of the repRNA . To generate this expression plasmid , we assembled PCR products in three steps . First , we PCR-amplified region I of DI- 72 with primers #532 and #3369 from pYC/DI72sat [12] , [13] followed by digestion of the PCR product with XbaI and gel purification of the PCR product . Second , the 3′ portion of DI-72 containing regions III , IV and the TRSV satellite ribozyme sequence [12] , [13] was PCR-amplified with primers #3367 and #1069 , followed by digestion with PstI and gel purification . Third , PCR was performed with primers #532 and #313 on pYC/DI72sat to obtain region II of DI-72 , followed by digestion of the PCR product with XbaI and PstI and gel purification . The three different PCR products were ligated and used as template for a final PCR with primers #532 and #1069 . The resulting PCR product was digested with BamHI and SacI and cloned into pGD plasmid resulting pYC/DI72sat/2xmiR171 . N . benthamiana leaves were agroinfiltrated as described [19] , [48] . A . tumefaciens cultures containing different plasmids were combined as follows: pGD-L-p33 ( OD600 = 0 . 35 ) , pGD-L-p92 ( OD600 = 0 . 15 ) , pGD-DI72sat ( OD600 = 0 . 15 ) , pGD-p19 [to express p19 suppressor of gene silencing [48] , OD600 = 0 . 15 ) and the above plasmids expressing the dominant negative A . thaliana ESCRT proteins ( OD600 = 0 . 4 ) . Agroinfiltrated leaves were collected after 2 . 5 days . For analysis of repRNA accumulation , total RNA was extracted as described [5] , [13] . The DI-72 repRNA was detected with a labeled RNA probe complementary to RIII/RIV ( + ) [5] , [13] . For the analysis of the tombusvirus replicase activity , leaf samples ( 250 mg ) were ground in liquid nitrogen and mixed with 2 ml buffer A ( 50 mM Tris-HCl pH 8 . 0 , 15 mM MgCl2 , 10 mM KCl , 2 mM EDTA , 20% glycerol , 0 . 3% plant protease inhibitor cocktail , 80 mM β-mercaptoethanol ) [51] . The mixture was passed through a 10 ml syringe fitted with cheesecloth to trap cell debris . The clarified extract was centrifuged at 300 g for 5 min to pellet additional cell debris . The supernatant was collected and centrifuged at 21 , 000 g for 20 min to pellet membranes . The pellet was washed in 1 ml of buffer B+1 . 2 M NaCl ( 50 mM Tris-HCl pH 8 . 0 , 10 mM MgCl2 , 1 mM EDTA , 6% glycerol , 0 . 3% plant protease inhibitor cocktail , 80 mM β-mercaptoethanol ) [51] , centrifuged again and the membrane pellet was finally resuspended in 250 µl buffer B ( no NaCl ) . 20 µl of the plant membrane fractions ( containing active viral replicase ) were used for in vitro replicase assays as described [7] , [24] . Replication assays in yeast was performed as described [52] . Accumulation of DI-72 repRNA was measured by Northern blot using RNA probes complementary to region III-IV of DI-72 and to the 18S ribosomal RNA [5] , [52] . In vitro replicase assay with membrane-enriched fraction was done as described previously [7] , [24] . Note that the amount of p33 protein in each sample was adjusted to comparable levels . The in vitro replicase assembly assay with yeast extracts was done as described previously [23] . For the RNase protection assay , the membrane-enriched fraction from wt , vps23Δ and vps24Δ yeast strains expressing wt p33/p92pol/repRNA was obtained as described [12] . Then , 25 µl of the membrane-enriched preparation dissolved in buffer E [200 mM sorbitol , 50 mM Tris-HCl pH 7 . 5 , 15 mM MgCl2 , 10 mM KCl , 10 mM β-mercaptoethanol , 1% proteinase inhibitor mix ( Sigma ) ] was digested with 1 µl of 20 µg/ml RNase A for 5 min . After the treatment , the RNA was extracted with phenol-chloroform , ethanol precipitated , recovered by centrifugation and analyzed in 5% polyacrylamide denaturing gel , blotted and hybridized with a 32P-labeled ( + ) DI-72 RNA probe [12] . S . cerevisiae strains BRO1::6xHA-KanMX4 and UEV::6xHA-KanMX4 were generated by homologous recombination using strain BY4741 as background . PCR was performed using plasmid pYM-14 ( EUROSCARF ) [53] as template and primers #2493/#2494 and #2492/#2491 ( Table S1 ) , respectively . The PCR products were transformed to BY4741 and recombinant yeast colonies were selected in YPD plates supplemented with G418 . Recombinant yeast strains were transformed with plasmid pGBK-33HFH or pGBK-His33 [12] and grown in minimal media supplemented with 2% glucose at 29°C . 300 µl of pelleted yeast were used to purify p33 with anti-FLAG M2 agarose as described previously [18] , except that the washing steps were performed at room temperature . P33 was detected with anti-FLAG antibody ( 1/5 , 000 dilution ) and AP-conjugated anti-mouse antibody ( 1/5 , 000 ) . Bro1-6xHA and UEV-6xHA proteins were detected with anti-HA antibody from rabbit ( Bethyl; 1/10 , 000 dilution ) and AP-conjugated anti-rabbit ( 1/10 , 000 ) followed by NBT-BCIP detection . The split-ubiquitin assay was based on the Dualmembrane kit 3 ( Dualsystems biotech ) . pGAD-BT2-N-His33 has been described previously [18] . pPR-N-VPS23 and pPR-N-ScUEV were generated by PCR using yeast genomic DNA and primer pairs of #2252/#2046 and #2252/#2292 ( Table S1 ) , respectively . The PCR products were digested with EcoRI and NheI and cloned into pPR-N-RE [18] . pPR-C-BRO1 was obtained by PCR using yeast genomic DNA and primers #2053/#2054 . The product was digested with BamHI and NheI and cloned into pPR-C-RE [18] . A . thaliana UEV-1 ( At3g12400; aa1-186 ) and UEV-2 ( At5g13860; aa1-175 ) were amplified from genomic DNA with primers #2669/#2670 and #2984/#2985 respectively , digested with BamHI and SalI and cloned into pPR-N-RE digested with BamHI/SalI or BglII/SalI . Nicotiana sp homologues of Vps23-UEV were amplified from N . benthamiana or N . tabacum genomic DNA with primers #2986 and #2987 ( based on accession # EB680173; nt 208-750 ) , digested with BglII and SalI and cloned into pPR-N-RE digested with BglII/SalI . Yeast strain NMY51/vps4Δ::URA3 was created by homologous recombination using the URA3 gene , which was amplified from plasmid pCM189 [54] with primers #2446 and #2447 . The PCR product was transformed into yeast strain NMY51 ( Dualsystems ) and the recombinants selected on Ura- plates [18] . NMY51 or NMY51/vps4Δ::URA3 were transformed with pGAD-BT2-N-His33 and pPR constructs . Transformed colonies were selected in Trp-/Leu- plates . Yeast colonies were re-suspended in a small volume of water and streaked onto Trp-/Leu-/His-/Ade- plates to score interactions . S . cerevisiae strain DKY79 ( VPS23-GFP , vps27Δ , vps4Δ ) , expressing the GFP tagged Vps23p from the native promoter and chromosomal location [55] was transformed with pYC-CUP-Flag33 and/or pGAD-pex13-RFP . To create pYC-CUP-Flag33 , Flag-tagged p33 was amplified from pGBK-33HFH with primers #2450/#992B ( Table S1 ) , digested with NcoI/PstI and cloned into similarly digested pGBK-His33/CUP1 [56] . The resulting plasmid was used as template for a PCR with primers #2753/#1403 . The product was digested with NheI/XhoI and cloned into SpeI/XhoI digested pYC2/CT ( Invitrogen ) generating pYC-CUP-Flag33 . PEX13 ORF was amplified by PCR using primers #1277/#1278 and pGAD-pex13-CFP [22] as template . RFP ORF was amplified from genomic DNA of a pex3-RFP yeast strain [40] using primers #2691/#2663 ( Table S1 ) . Both PCR products were digested with BglII , ligated and reamplified with primers #1277 and #2663 . This product was digested with HindIII and BamHI and cloned into similarly digested pGAD H- [12] to generate pGAD-pex13-RFP . The transformed yeast strains were grown at 29°C in minimal media supplemented with 2% glucose . Yeast cells were imaged with Olympus FV1000 confocal laser scanning microscope [20] within 15–45 minutes after addition of 50 mM CuSO4 to induce p33 expression .
|
Plus-stranded RNA viruses , which are important pathogens of humans , animals and plants , replicate in infected cells by assembling viral replicase complexes consisting of viral- and host-coded proteins . In this paper , we show that a group of host factors called ESCRT proteins ( endosomal sorting complexes required for transport ) play important roles in tombusvirus replication . The expression of dominant negative mutants of ESCRT factors inhibited virus replication in the plant host , suggesting that tombusviruses co-opt selected ESCRT proteins for the assembly of the viral replicase complex . In addition , we show direct interaction between the viral p33 replication protein and Vps23p ESCRT-I and Bro1p accessory ESCRT factors . The interaction with p33 leads to the recruitment of Vps23p to the peroxisomes , the sites of tombusvirus replication . We also showed that the viral RNA within the viral replicase complex became more sensitive to ribonuclease in the absence of ESCRT factors , suggesting that the protection of the viral RNA is compromised within the replicase complex assembled in the absence of ESCRT proteins . Intriguingly , the host ESCRT factors also affect the budding of several enveloped viruses , intracellular transport of proteins and cytokinesis . Overall , this work demonstrates that a plus-stranded RNA virus uses the endosomal sorting pathway in a unique way .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"virology/mechanisms",
"of",
"resistance",
"and",
"susceptibility,",
"including",
"host",
"genetics",
"virology/viral",
"replication",
"and",
"gene",
"regulation",
"virology"
] |
2009
|
A Unique Role for the Host ESCRT Proteins in Replication of Tomato bushy stunt virus
|
Genetic analyses of human lice have shown that the current taxonomic classification of head lice ( Pediculus humanus capitis ) and body lice ( Pediculus humanus humanus ) does not reflect their phylogenetic organization . Three phylotypes of head lice A , B and C exist but body lice have been observed only in phylotype A . Head and body lice have different behaviours and only the latter have been involved in outbreaks of infectious diseases including epidemic typhus , trench fever and louse borne recurrent fever . Recent studies suggest that body lice arose several times from head louse populations . By introducing a new genotyping technique , sequencing variable intergenic spacers which were selected from louse genomic sequence , we were able to evaluate the genotypic distribution of 207 human lice . Sequence variation of two intergenic spacers , S2 and S5 , discriminated the 207 lice into 148 genotypes and sequence variation of another two intergenic spacers , PM1 and PM2 , discriminated 174 lice into 77 genotypes . Concatenation of the four intergenic spacers discriminated a panel of 97 lice into 96 genotypes . These intergenic spacer sequence types were relatively specific geographically , and enabled us to identify two clusters in France , one cluster in Central Africa ( where a large body louse outbreak has been observed ) and one cluster in Russia . Interestingly , head and body lice were not genetically differentiated . We propose a hypothesis for the emergence of body lice , and suggest that humans with both low hygiene and head louse infestations provide an opportunity for head louse variants , able to ingest a larger blood meal ( a required characteristic of body lice ) , to colonize clothing . If this hypothesis is ultimately supported , it would help to explain why poor human hygiene often coincides with outbreaks of body lice . Additionally , if head lice act as a reservoir for body lice , and that any social degradation in human populations may allow the formation of new populations of body lice , then head louse populations are potentially a greater threat to humans than previously assumed .
Lice are extremely well-adapted ectoparasites that are usually host-specific [1] . Three recognized taxa of lice feed on humans: head lice ( Pediculus humanus capitis ) , body lice ( Pediculus humanus humanus ) , and pubic lice ( Pthirius pubis ) , feed on humans . As one of the most intimate parasites of humans , lice have been widely used as a genetic model to infer host evolutionary history by providing genetic date independent of host data [1] , [2] . Several nuclear and mitochondrial DNA sequences have previously been used in population genetic studies of human lice . Of these , the nuclear DNA sequences , EF-1α and 18S rDNA , discriminated lice into two subgroups , lice from Sub-Saharan Africa and lice worldwide[3] . In each subgroup , the head lice were genetically different from the body lice [3] . However , Leo et al . suggested that 18S rDNA phylogeny was concordant with the phylogenies from mitochondrial genes , but the EF-1α phylogeny was concordant neither with the mitochondrial phylogenies nor with the 18S rRNA phylogeny [4] . Furthermore , the mitochondrial DNA markers , partial COI and cytB classified the lice into three deeply divergent clades ( Clades A , B , and C ) , and each having unique geographical distribution . Clade A includes both head and body lice and is worldwide in distribution . Clade B consists only of head lice from America , Australia and Europe , and Clade C consists only of lice from Ethiopia and Nepal [5] . More variable genetic markers , such as internal transcribed spacers ( ITS ) of ribosomal DNA and microsatellite DNA , were also used to deduce the louse phylogeny . However , the ITS that was chosen was not useful to study the populations structure of human lice because some of the lice had more than one copy of ITS2 in their genome [6] . A subsequent microsatellite DNA-based study has suggested that human head and body lice are genetically distinct [7] , however recent studies contradict this hypothesis [5] , [8] . Taken together , the population structure of human lice is complex and still unclear . The previously used genetic markers were mostly mitochondrial and nuclear genes that were too conserved to generate more information of genetic diversity of studied louse isolates . So far , no genetic marker has been found that can discriminate among individual human lice . While being used as a suitable genetic model to study the evolutionary history of humans , lice have long been associated with infectious diseases . Of the three types of lice associated with humans , body lice can be a serious public health problem because they are known vectors of Rickettsia prowazekii , Bartonella quintana , and Borrelia recurrentis , which cause epidemic typhus , trench fever and relapsing fever in humans , respectively [9] . However , medical interest in louse-borne diseases had waned for more than 30 years until 1997 , when an outbreak of infection of louse-transmissed R . prowazekii and B . quintana occurred among the displaced population of Burundi [10] , [11] . Body lice have long been recognized as human parasites and although typically prevalent in rural communities in upland areas of countries close to the equator , high incidence of louse-borne infections are also increasingly found in the homeless in developed countries [9] , [12] , [13] . In contrast , head lice represent a major economic and social concern throughout developed nations , because head louse infestations are often associated with school-aged children . Faster evolving molecular markers are needed in order to epidemiologically survey the vectors of these bacterial infections and to address more recent population-level questions , [8] , [14] . Among these fast-evolving genetic markers , intergenic spacers are promising for individual discrimination of lice because they are under less evolutionary pressure , and are more variable than coding genes . These factors make intergenic spacers useful for understanding the population genetics of lice . Highly variable intergenic spacers are useful for strain-typing many bacteria , including louse-transmitted R . prowazekii and B . quintana [15] , [16] as well as other pathogenic bacteria [17] . Additionally , intergenic spacer sequences for individual discrimination of lice , are now publicly available [14] . In this study , we used four highly variable intergenic spacers that were selected from the genomic sequence to study the genotypic distribution of a large collection of lice of worldwide origins .
Two hundred and eighty-four human lice collected from Russia , France , Portugal , Mexico , USA , UK , Morocco , Algeria , Peru , Thailand , Australia , Rwanda , and Burundi were included in this study . Lice were collected by experienced entomologists from patients who had only one type of infestation ( head or body ) and classified according to the site where they were found . Among them , only 97 lice were tested with four nuclear intergenic spacers , other 110 and 77 lice were tested with two intergenic spacers , respectively , due to limited DNA quantity . The strain information , including origin , the body part where they were removed ( body or head ) , and the year when it was collected are given in Figure 1 and Figures S1 and S3 . In addition , to estimate the utility of multi-spacer typing ( MST ) of louse populations , we also studied two body lice from our laboratory colony ( Culpeper strain ) per year , collected in 1998 , 1999 , 2000 , 2003 , 2004 and 2009 . From 1998 to 2009 , our louse colony went through 132 generations . All lice were stored at −20°C until processed further . Before DNA isolation , each louse was rinsed twice in sterile water for 15 minutes and cut lengthwise in half . Then , total genomic DNA of each half louse was extracted using the QIAamp Tissue kit ( QIAGEN , Hilden , Germany ) as described by the manufacturer . The extracted genomic DNA was stored at −20°C until PCR amplification . The nuclear intergenic spacers were randomly selected from the genomic sequence of Pediculus humanus humanus UDSA strain ( http://phumanus . vectorbase . org/index . php ) and were identified with flanking genes which exhibited >40% sequence identity with homologous genes in the Vectorbase [14] , [18] . Primers used for amplification and sequencing of these intergenic spacers were chosen from the flanking genes using the Primer 3 . 0 software ( http://frodo . wi . mit . edu/cgi-bin/primer3/primer3_www . cgi ) and are listed in Table 1 . All Primers used in this project were obtained from Eurogentec ( Seraing , Belgium ) . As the testing of all intergenic spacers on all louse samples was labor intensive , a panel of 16 lice from a wide range of origins was first used to test for the validity and the presence of polymorphisms for each of the intergenic spacers . Subsequently , the intergenic spacers with successful amplification and sequencing from the 16 tested louse samples , were finally used as markers in order to genotype the remaining strains . PCR amplification of each intergenic spacer was carried out in a PTC-200 automated thermal cycler ( MJ Research , Waltham , MA , USA ) . 1 µl of each DNA preparation was amplified in a 20 µl reaction mixture containing 10 pM of each primer , 2 mM of each nucleotide ( dATP , dCTP , dGTP and dTTP ) , 4 µl of Phusion HF buffer , 0 . 2 µl of Phusion polymerase enzyme ( Finnzymes , Espoo , Finland ) and 12 . 4 µl of distilled water . The following conditions were used for the amplification: an initial 5 min of denaturation at 95°C , followed by 35 cycles of denaturation for 1 min at 94°C , an annealing time of 30 sec at 56°C , and an extension cycle for 1 min at 72°C . The amplification was completed by an extension period of 5 min at 72°C . PCR products were purified , using the MultiScreen PCR filter plate ( Millipore , Saint-Quentin en Yvelines , France ) , as recommended by the manufacturer . PCR products were then sequenced in both directions , with the same primers used for PCR amplification , using BigDye Terminator version 1 . 1 cycle sequencing ready reaction mix ( Applied Biosystems , Foster City , CA ) . Sequencing products were resolved using an ABI 3100 automated sequencer ( Applied Biosystems ) . Sterile water was used as a negative control in each assay . In order to compare the discriminatory power of intergenic spacers with genes , as well as to compare their phylogenetic organizations , the mitochondrial gene , cytB ( cytochrome b ) was amplified and sequenced from those louse samples when there was DNA to perform the PCR experiments . The primers used for this experiment were CytbF1 ( 5′-GAGCGACTGTAATTACTAATC-3′ ) and CytbR1 ( 5′-CAA CAA AAT TAT CCG GGT CC-3′ ) [19] . Nucleotide sequences were obtained using Sequencher 4 . 8 ( Gene codes Corp , Ann Arbor , MI , USA ) . The primers used to amplify intergenic spacers were selected based on flanking gene sequences . The sequences from the coding sequence fragments were not used in the analyses . For each intergenic spacer , and cytB , a genotype was defined as a sequence exhibiting a unique mutation . Each genotype was confirmed to be unique by BLASTn search in all the obtained sequences [20] . Multiple sequence alignments were carried out using the Clustal W software [21] . Phylogenetic analysis of the lice that were studied was obtained using the neighbor-joining and maximum parsimony methods within the MEGA 3 . 1 software with complete deletion [22] and using the maximum likelihood method in PhyML 3 . 0 with SH-like approximate likelihood-ratio test and HKY85 substitution model [23] , [24] . For this purpose , sequences of the selected intergenic spacers were concatenated . The discriminatory power ( D ) of each intergenic spacer , and cytB , was calculated with the Hunter and Gaston's formula [25] . DNA sequences obtained from the S2 and S5 spacers , the PM1 and PM2 spacers and the cytB gene were deposited in GenBank under accession numbers EU928781-EU928862 , EU913096-EU913223 , and GU323324-GU323334respectively .
Twenty-two nuclear intergenic spacers were initially selected from the genomic sequences and preliminary tested on 16 louse samples ( Table 1 ) . However , due to non-specific amplification , or low sequencing quality , 18 intergenic spacers were removed from this study . Finally , four intergenic spacers , hereafter termed S2 , S5 , PM1 , and PM2 , were used as typing markers in this study ( Table 1 ) . Through amplification and sequencing , 165–185 bp of S2 and 156–189 bp of S5 were obtained from 207 louse samples and133–155 bp of PM1 and 323–328 bp of PM2 were obtained from 174 louse samples . Sequences from the different genotypes of the four intergenic spacers have been deposited in the EMBL/GenBank database with access numbers: EU928781-EU928862 for S2 and S5 and EU913096-EU913223 for PM1 and PM2 . Two hundred and seven lice were differentiated into 84 and 49 genotypes based on intergenic spacers S2 and S5 , respectively . Concatenation of S2 and S5 sequences differentiated the 207 lice into 148 genotypes . One hundred and seventy-four lice were differentiated into 25 and 62 genotypes based on intergenic spacers PM1 and PM2 , respectively . Concatenation of PM1 and PM2 sequences differentiated the 174 lice into 77 genotypes . Further concatenation of S2 , S5 , PM1 , and PM2 , discriminated a panel of 97 lice into 96 MST genotypes . Except for two lice collected from French homeless people which shared the MST genotype 89 , the other 90 lice exhibited unique MST genotypes based on the concatenation of four intergenic spacers . Sequences from each of the four intergenic spacers S2 , S5 , PM1 , and PM2 were identical among the 12 body lice from our laboratory colony collected over 12 years . The genotypes obtained were: 8 , 6 , 18 , and 39 for S2 , S5 , PM1 and PM2 , respectively . A partial cytB gene sequence was amplified and sequenced from 170 lice . A 316 bp fragment was obtained from each louse after sequence correction and assembling . The cytB sequences were used to classify the 170 lice into 11 genotypes . The body lice sampled from our laboratory colony over 132 generations exhibited identical cytB sequences ( genotype 4 ) . The discriminatory power ( D ) of the intergenic spacers , PM1 , PM2 , S2 , and S5 was respectively 0 . 6988 , 0 . 8406 , 0 . 9677 , and 0 . 8913 . The D value of cytB was 0 . 6445 . The D value of concatenation of intergenic spacers varied from 0 . 9123 for concatenation of PM1 and PM2 to 0 . 9945 for concatenation of S2 and S5 . A D value of 0 . 9998 was reached by combined use of the four intergenic spacers . The dendrograms of studied lice inferred by the methods of neighbor-joining , maximum parsimony , and maximum likelihood exhibited similar phylogenetic organizations ( Figures 1 – 3 , Figures S1 – S6 ) . The 148 genotypes of intergenic spacers S2 and S5 , were grouped into 3 clusters , C1 , C2 , and C3 ( Figure S1 ) . Each cluster included both head and body lice . In addition , genotypes 96 and 101 consisted of two body lice and two head lice , respectively ( Figure S1 ) . A subcluster ( Burundi and Rwanda subcluster ) in cluster C3 was comprised of 29 lice from Burundi and Rwanda and one louse from Russia . The other 24 lice collected in Rwanda and Burundi were grouped into cluster C2 . The majority of French lice , including those collected from homeless people , were grouped into a sub-clade within cluster C1 ( Figure S1 ) . The 77 nuclear intergenic spacer-genotypes , for PM1 and PM2 , were grouped into 2 distinct clusters ( Figure S3 ) . Cluster C1 included 148 lice collected from Russia , Mexico , France , UK , USA , and Portugal as well as one louse from Rwanda ( Figure S3 ) . A subcluster in cluster C1 contained 31 lice from France and 2 lice from Portugal , which is hereafter referred to as the “French subcluster” . The 19 lice collected from French homeless individuals were tightly grouped with French lice in cluster C1 ( Figure S3 ) . Cluster C2 was comprised of 25 lice collected from Rwanda and Burundi . Genotypes 6 , 29 , and 32 , were observed in both head and body lice . Based on the concatenation of the four intergenic spacers PM1 , PM2 , S2 , and S5 , the 97 lice were discriminated into 96 MST genotypes and grouped into two clusters ( Figures 1 and 2 ) . Cluster C1 included 75 lice from Russia , France , Mexico , and Portugal , and Cluster C2 , the Burundi and Rwanda cluster , contained 22 lice from Burundi and Rwanda ( Figure 1 ) . Lice collected from the French homeless individuals were tightly grouped with French lice into two subclusters within C1 ( Figure 1 ) . The 11 genotypes of cytB from 170 lice were grouped into 2 clusters ( Figure 3 ) . Cluster C1 included 111 lice from France , Russia , UK , USA , Mexico , Portugal , Burundi , and Rwanda ( Figure 3 ) , and corresponded to Type A in a study by Light et al [5] . Cluster C2 included 59 lice from the UK , USA , and Mexico , and corresponded to the Type B reported by Light et al . [5] . Genotypes 4 and 6 comprised both head and body lice .
In this study , MST based on four highly variable intergenic spacers selected from the genomic sequence of a body louse , classified 97 lice into 96 MST genotypes . To date , MST appears to be the most sensitive discriminatory genotyping system of human lice , allowing for discrimination of individuals . In addition , MST helped us to address several important debates associated with human lice . One of the ongoing debates is whether head and body lice are separate species or two subspecies within Pediculus humanus [3] , [4] , [7] , [8] , [26] . To address this issue , most of the previous studies have used mitochrondrial or nuclear genes to evaluate and compare the genetic variability of human head and body lice . Studies based on the mitochondrial gene COI [26] , the mitochondrial genes cytB and ND4 and nuclear genes EF-1α and RPII [27] , the mitochondrial genes cytB and COI [28] , or the nuclear gene 18S rDNA [4] , supported the hypothesis that human head and body lice are conspecific . Using previously published sequence data , by reticulated networks , gene flow , population genetics , and phylogeny analysis , Light et al . [8] also observed that human head and body lice are conspecific . However , a recent study performed by Leo et al . [7] , in which microsatellites were used as genetic markers , concluded that human head and body lice are two distinct species . These studies made opposite conclusions by using different genetic markers . The low discriminatory power of previously used markers limited their ability to provide convincing evidence whether head and body lice are subspecies of one species or two distinct species . Our genotypic and phylogenetic analyses using MST did not support the hypothesis that human head and body lice are separate species . For instance , genotype 32 , which was a concatenation of the intergenic spacers PM1 and PM2 , was comprised of 44 head lice from the USA and the UK as well as one body louse from Europe ( Figure S3 ) . Genotypes 6 and 29 were comprised of both head and body lice collected in France ( Figure S3 ) . Genotypes 96 and 101 , a concatenation of the intergenic spacers S2 and S5 , also were comprised of both head and body lice ( Figure S1 ) . Phylogenetic organizations of head and body lice based on each of the four intergenic spacers , and on concatenation of both , support the hypothesis that head lice were grouped with body lice in the same clusters or subclusters ( Figure 1 , Figures S1 and S2 ) . The changing tree topography observed among spacers may be related to differences in selection pressure that their flanking genes undergo . The genotypic distribution of 170 lice based on partial cytB gene sequences , and the phylogenetic organization of 11 cytB genotypes , also demonstrated that head and body lice shared the same cytB genotypes and were grouped in the same cluster ( Figure 2 ) , which further confirmed the hypothesis that human head and body lice are conspecific [3] , [27] , [28] . In our study , although only two clusters were observed based on partial cytB gene sequences , cluster C1 contained both head and body lice from worldwide origins , and cluster C2 included only head lice from America and Europe ( Figure 3 ) . This result did not contradict the previous observation [19] of three deeply divergent clades of human lice , as our study did not include lice from either Ethiopia or Nepal . However , the phylogenetic organization of cytB sequences was significantly simpler than those based on intergenic spacers . Three and two clusters were respectively obtained from the concatenations of the intergenic spacers S2 and S5 ( Figure S1 ) , and the intergenic spacers PM1 and PM2 ( Figure S3 ) . Additionally , two clusters were generated from the concatenation of the four intergenic spacers PM1 , PM2 , S2 and S5 ( Figure 1 , Figures S1 and S2 ) . Each cluster was comprised of several subclusters , such as the French subcluster , including the majority of French lice , and the Rwanda/Burundi cluster , which also consisted of lice collected from sub-Saharan Africa ( Figure 1 , Figures S1 and S2 ) . This discrepancy of phylogenetic organizations obtained from intergenic spacers and cytB sequences resulted , at least partially , from the high variability of intergenic spacers , which enabled individual discrimination of human lice . In addition , these differences may also be explained by the fact that the louse samples incorporated in each phylogenetic analysis were different due to limited DNA available for such experiments . Furthermore , louse genomic DNA may be highly recombined , which would in turn result in distinct phylogenetic organization from different markers [8] . Thus , collecting more louse samples with wide origins , especially lice from Ethiopia and Nepal , and subjecting them to MST analysis , would help to further clarify the relationship between head and body lice . Human head and body lice are strict obligate human ectoparasites that differ in several aspects of their morphology , physiology and life histories . Head lice are mostly found on the head and attach their eggs to the base of hair shafts , whereas body lice reside in clothing and attach their eggs to clothing fiber , a life history strategy that probably arose when humans first began wearing clothes [27] . By comparison with body lice , head lice have been described as having shorter and broader antennae , shorter legs , more marked indentations between successive abdominal plates , and as being larger and more deeply pigmented [29] , [30] . However , such morphological differences have been determined on a small number of lice and may not hold at the species level [29] . Body lice also take a larger blood meal , lay higher numbers of eggs and develop faster than head lice [29] , [31] , [32] . In addition , body lice are more resistant to environmental conditions , can stay alive for longer period of time outside the host , are able of transmitting infectious diseases , and are mostly found in adults whereas head lice are essentially found in children [29] . Despite various genetic differences [1]–[8] , detailed above , head and body lice have been shown to be able to interbreed [30] . Lice are extremely well adapted ectoparasites , which are usually host-specific by co-speciation with their host [2] , [28] , [33] . Thus , lice have become a good genetic model for studying specific aspects of human evolution , including addressing when our human ancestors began to wear clothing . Very recently , a study based on sequence analysis of COI and cytB from human head and body lice suggested direct contact between modern and archaic humans [28] . More recently , Light et al . [34] verified this hypothesis by using both nuclear and mitochondrial genes [34] . However , these studies were based on conserved mitochondrial or nuclear genes , which provided limited genetic variability of studied lice . In our study , we also tested the use of highly variable intergenic spacers for strain-typing of human lice to explore human evolutionary history . Concatenation of these highly variable intergenic spacer sequences classified some lice from Rwanda and Burundi into a basal cluster or subcluster and grouped other lice collected in Rwanda and Burundi with lice from North Africa , Europe , USA , and Asia , which supports the hypothesis that human beings originated in Africa ( Figure 1 , Figures S1 and S2 ) . Thus , highly variable intergenic spacer sequences could be used to study the evolution history of human lice and its host . It might be argued that , due to fast evolution and high polymorphism , intergenic spacers may not be able to fully reflect long-term dynamic changes of populations . However , we observed that MST was a valuable tool for tracing distinct louse populations , and was not biased by mutations that might arise within a single population over time , for at least 132 generations . Nevertheless , we recommend using a combination of coding genes and intergenic spacers because coding genes are conserved enough to highlight evolutionary relationships , and the intergenic spacers are variable enough to identify fine-scale genetic variability . While lice may present a valuable model to study its host evolution , human head and body lice cause serious health and social problems . Head lice are common worldwide , infesting millions of school children every year and the resistance of Pediculus humanus capitis to insecticides is spreading [35] . Body lice are less prevalent parasites , associated mainly with those living in poor conditions , but are potentially more harmful because they are known vectors of at least three bacterial pathogens in humans: R . prowazekii , B . quintana , and B . recurrentis . There have been several outbreaks of louse-borne R . prowazekii infections in Burundi and Rwanda jails in 1997 and 2001 , and sporadic R . prowazekii infections were also recently reported [9] . Epidemiological surveys of these louse-borne diseases are also very important for us to understand and potentially combat these diseases . In addition , recent evidence has been brought that head lice are potential vectors of B . quintana [36] , [37] , and their role in the epidemiology of epidemic typhus has been questioned [38] . Other studies have identified two endosymbiotic bacteria that have co-evolved in head and body lice [39]–[41] . However , whether these symbionts have any influence on louse behavior , development and/or competence as disease vectors is as yet mostly unknown . Based on phylogenetic analysis of the four intergenic spacers , S2 , S5 , PM1 , and PM2 as well as a concatenation of them , the head and body lice collected from Rwanda and Burundi tightly grouped together to form clusters as well as subclusters ( Figure 1 , Figures S1 and S2 ) . In addition , the lice collected from homeless people in France grouped tightly with those collected in non-homeless French people , which suggested louse populations migrate between homeless people and non-homeless people in France and homeless people are known to be at high risk for louse-borne diseases [9] , [11] , [12] . MST may ultimately be a good tool for performing surveys associated with louse transmission and louse-borne diseases . In addition , our MST analysis demonstrated that head and body lice collected in Rwanda and Burundi in 1997 , 2001 , 2003 , and 2008 , were closely grouped ( Figure 1 , Figures S1 and S2 ) . Thus , the outbreak of louse-borne R . prowazekii infection that happened in 1997 and 2001 opens up the possibility that lice in this region may still pose a risk for the transmission of R . prowazekii to humans . However , clear separation of African lice ( collected from Rwanda and Burundi ) from other lice was not recovered by intergenic spacers S2 and S5 , likely due to recent recombination of nuclear DNA [8] . As mentioned above , head lice are different from body lice morphologically and physiologically . It is possible that these phenotypic differences are controlled by a single mutation or potentially a regulatory gene ( or genes ) governing , for example , the volume of ingested blood . This is the simplest explanation to understand the genetic data showing that lice have exactly the same origin . Under certain conditions of low hygiene , a head louse infestation can transform into a massive infestation ( Figure 4 ) . Certain head lice could colonize clothing ( Figure 5 ) , and produce a body louse variant by purifying selection or allotropism , which can in turn generate an epidemic of body lice ( Figure 6 ) . Several previous observational studies had also suggested that head lice could become body lice when raised in appropriate conditions [42]–[44] . If this scenario is true , the body louse reservoir is not autonomous and actually depends upon head lice . Previous work has shown that all body lice arose from mitochondrial Type A [5] , which suggests that only that genotype has the ability to evolve into the body louse niche . This also makes it possible to understand the difficulties to eradicate body lice in a community , especially when the patients are surrounded by other individuals that are infested by head lice . In our clinical work in Marseilles , France , despite 10 years of attempts to minimize human louse populations , body lice continually reappear and may be due to the persistence of head louse populations [45] , [46] . Recent work demonstrating the presence of B . quintana in head lice [36] , [37] suggested that they might also transmit infectious diseases , which supports our hypothesis presented in Figure 6 , giving them a greater opportunity to ingest circulating bacteria [29] , [31] , and that head lice are rarely collected and tested , even when present , in outbreaks of louse-borne infections , may explain why head lice have long been considered to be free from human pathogens . In conclusion , by strain-typing of human head and body lice using both coding sequences and highly variable intergenic spacers , our data supports the hypothesis that human head and body lice belong to the same species . Based on genotypic and phylogenetic analyses , we also hypothesize that head lice may transform into body lice [47] and cause outbreaks of louse-borne diseases . However , more efforts on the genetic studies of head and body lice are necessary to link their genetic difference with morphological and physiological diversity . Whole genome sequencing of head lice and comparative genomics between head and body lice would be useful to address these questions . In addition , due to its high resolution and reasonable phylogenetic classification , MST based on highly variable intergenic spacer sequences may be helpful for the epidemiological survey of louse-borne diseases .
|
While being phenotypically and physiologically different , human head and body lice are indistinguishable based on mitochondrial and nuclear genes . As protein-coding genes are too conserved to provide significant genetic diversity , we performed strain-typing of a large collection of human head and body lice using variable intergenic spacer sequences . Ninety-seven human lice were classified into ninety-six genotypes based on four intergenic spacer sequences . Genotypic and phylogenetic analyses using these sequences suggested that human head and body lice are still indistinguishable . We hypothesized that the phenotypic and physiological differences between human head and body lice are controlled by very limited mutations . Under conditions of poor hygiene , head lice can propagate very quickly . Some of them will colonize clothing , producing a body louse variant ( genetic or phenetic ) , which can lead to an epidemic . Lice collected in Rwanda and Burundi , where outbreaks of louse-borne diseases have been recently reported , are grouped tightly into a cluster and those collected from homeless people in France were also grouped into a cluster with lice collected in French non-homeless people . Our strain-typing approach based on highly variable intergenic spacers may be helpful to elucidate louse evolution and to survey louse-borne diseases .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"genetics",
"and",
"genomics/genomics",
"microbiology/medical",
"microbiology",
"infectious",
"diseases/bacterial",
"infections"
] |
2010
|
Genotyping of Human Lice Suggests Multiple Emergences of Body Lice from Local Head Louse Populations
|
Several vertebrate microRNAs ( miRNAs ) have been implicated in cellular processes such as muscle differentiation , synapse function , and insulin secretion . In addition , analysis of Dicer null mutants has shown that miRNAs play a role in tissue morphogenesis . Nonetheless , only a few loss-of-function phenotypes for individual miRNAs have been described to date . Here , we introduce a quick and versatile method to interfere with miRNA function during zebrafish embryonic development . Morpholino oligonucleotides targeting the mature miRNA or the miRNA precursor specifically and temporally knock down miRNAs . Morpholinos can block processing of the primary miRNA ( pri-miRNA ) or the pre-miRNA , and they can inhibit the activity of the mature miRNA . We used this strategy to knock down 13 miRNAs conserved between zebrafish and mammals . For most miRNAs , this does not result in visible defects , but knockdown of miR-375 causes defects in the morphology of the pancreatic islet . Although the islet is still intact at 24 hours postfertilization , in later stages the islet cells become scattered . This phenotype can be recapitulated by independent control morpholinos targeting other sequences in the miR-375 precursor , excluding off-target effects as cause of the phenotype . The aberrant formation of the endocrine pancreas , caused by miR-375 knockdown , is one of the first loss-of-function phenotypes for an individual miRNA in vertebrate development . The miRNA knockdown strategy presented here will be widely used to unravel miRNA function in zebrafish .
MicroRNAs ( miRNAs ) have a profound impact on the development of multicellular organisms . Animals lacking the Dicer enzyme , which is responsible for the processing of the precursor miRNA into the mature form , cannot live [1–3] . MiRNA mutants have been described only for Caenorhabditis elegans and Drosophila , reviewed in [4] . From these studies , it is clear that invertebrate miRNAs are involved in a variety of cellular processes , such as developmental timing [5 , 6] , apoptosis [7 , 8] , and muscle growth [9] . Analysis of conditional Dicer null alleles in mouse has indicated a general role for miRNAs in morphogenesis of the limb , skin , lung epithelium , and hair follicles [10–13] . Overexpression studies in mouse have implicated specific vertebrate miRNAs in cardiogenesis and limb development [14 , 15] . In zebrafish , embryos lacking both maternal and zygotic contribution of Dicer have severe brain defects [2] . Strikingly , the brain phenotype of maternal-zygotic Dicer zebrafish can be restored by injection of miR-430 , the most abundant miRNA in early zebrafish development . Despite all these studies describing functions for miRNAs in development , no vertebrate miRNA mutant has been described to date . Genetically , it is challenging to obtain mutant miRNA alleles in zebrafish , because their small size makes them less prone to mutations by mutagens , and for many miRNAs , there are multiple alleles in the genome or they reside in families of related sequence . Temporal inhibition of miRNAs by antisense molecules provides another strategy to study miRNA function . 2′-O-methyl oligonucleotides have been successfully used in vitro and in vivo to knock down miRNAs [16–18] . Morpholinos are widely applied to knock down genes in zebrafish development [19] and have recently been used to target mature miR-214 in zebrafish [20] . However , off-target phenotypes are often associated with the use of antisense inhibitors . Here , we show that morpholinos targeting the miRNA precursor can knock down miRNAs in the zebrafish embryo . Several independent morpholinos can knock down the same miRNA , and these serve as positive controls to filter out off-target effects . Morpholinos can block miRNA maturation at the step of Drosha or Dicer cleavage , and they can inhibit the activity of the mature miRNA . We show that inhibition of miR-375 , which is expressed in the pancreatic islet and pituitary gland of the embryo [21] , results in dispersed islet cells in later stages of embryonic development , whereas no effects were observed in the pituitary gland . The morpholino-mediated miRNA knockdown strategy presented here , is an extremely fast and well-controlled method to study miRNA function in development .
Since it is difficult to obtain a genetic mutant for a miRNA in zebrafish , we looked for alternative strategies to deplete the embryo of specific miRNAs . Antisense molecules such as 2′-O-methyl and locked nucleic acid ( LNA ) oligonucleotides have been used to inhibit miRNAs in cell lines [16 , 18 , 22] , Drosophila embryos [23] , and adult mice [17] . We tried to use these molecules to inhibit the function of endogenous miRNAs in the zebrafish embryo . Although they can be used to suppress the effects of miRNA overexpression [24] , injection of higher concentrations required to obtain good knockdown of endogenous miRNAs resulted in toxic effects , when injecting 1 nl solution at a concentration of approximately 10 μM and 50 μM for LNA and 2′-O-methyl oligonucleotides , respectively ( unpublished data ) . Therefore , we switched to morpholinos because these are widely used to inhibit mRNA translation and splicing in zebrafish embryos [19] , and have also been shown to target miRNAs in the embryo [2 , 20 , 24] . We injected 1 nl of 600 μM morpholino solution with a morpholino complementary to the mature miR-206 in one- or two-cell–stage embryos . Subsequently , embryos were harvested at 24 , 48 , 72 , and 96 hours postfertilization ( hpf ) , and subjected to in situ hybridization and Northern blotting ( Figure 1A and 1B ) . This analysis showed that the mature miRNA signal is suppressed up to 4 d after injection of the morpholino . The knockdown effect was specific for this miRNA; parallel in situ analysis of the same embryos with a probe for miR-124 did not show any effects on expression of this miRNA ( Figure 1B ) . Thus , miRNA detection can be specifically and efficiently suppressed during embryonic and early larval stages of zebrafish development using morpholinos antisense to the mature miRNA . The zebrafish embryo can be used to monitor the effect of miRNAs on green fluorescent protein ( GFP ) reporters fused to miRNA target sites [24] . To determine the effect of a morpholino in this assay system , we constructed a GFP reporter for miR-30c and tested it in the presence and absence of a mature miR-30c duplex . Injected miR-30c silences this GFP reporter , which is in line with previous reports using similar strategies in the embryo ( Figure 1C ) [2 , 20 , 24] . Co-injection of the miR-30c duplex and a morpholino targeting mature miR-30c rescues the reporter signal , whereas injection of a control morpholino did not reverse the silencing by miR-30c . These data indicate that a morpholino can block the activity of a mature miRNA duplex in a functional assay . There are three possible explanations for the observed reduction in the detection signal for a miRNA that is targeted by a morpholino . First , the hybridization of a morpholino could disturb isolation of the miRNA . Second , the morpholino could destabilize the miRNA . Third , the morpholino could inhibit the maturation of the miRNA . To examine the effect of a morpholino on the isolation of a mature miRNA , we incubated a mature miR-206 duplex and a control duplex ( miR-205 ) with a morpholino against miR-206 in vitro . After isolation , samples were analyzed by Northern blotting for the presence of miR-206 and miR-205 . We could still detect miR-206 , indicating that there is no effect of the morpholino on the RNA isolation procedure ( Figure 1D ) . However , when morpholino and miRNA duplex were incubated together in vitro and loaded on a denaturing gel without isolation , we observed a decrease in the signal for miR-206 , indicating that the morpholino can bind to the miRNA in vitro and still does so in the denaturing gel . Next , we wanted to know whether a morpholino could affect the stability of a mature miRNA in vivo . Therefore , we injected a mature miR-206 and a control duplex ( miR-205 ) together with a morpholino against miR-206 in the embryo . After incubation for 8 h , RNA was isolated and subjected to Northern blot analysis to probe for injected miR-206 and injected miR-205 . In contrast to the data obtained for endogenous miR-206 , there was no decrease observed in the amount of injected miR-206 in the morpholino-injected embryos ( Figure 1D ) ( endogenous miR-206 is not yet expressed at this stage ) . Since these data show that there is no effect of a morpholino on miRNA isolation or stability , we conclude that morpholinos deplete the embryo of miRNAs by inhibiting miRNA maturation . If this is the case , then we expect morpholinos targeting other regions of the miRNA precursor to act as well as the morpholinos designed against the mature miRNA , and this is indeed what we find ( see next section ) . Injection of antisense oligos in embryos might result in off-target effects . Thus , phenotypic data retrieved from antisense knockdown experiments should be treated with caution . In Drosophila , 2′-O-methyl oligo–mediated knockdown of embryonically expressed miRNAs caused defects that clearly differed from the phenotype of the corresponding knockout fly [9 , 23] . In sea urchin experiments , off-target effects of morpholino knockdowns are well documented , though low incubation temperatures favor off-target interactions [25] . To filter out off-target effects , we sought a control strategy that would allow us to compare effects of morpholinos with independent sequences targeted to the same miRNA . Because our data on morpholinos targeting the mature miRNA suggested that miRNA biogenesis might be affected , we designed morpholinos targeting the Drosha and Dicer cleavage sites of the precursor miRNA ( Figure 2A ) . We decided to test this strategy on miR-205 , since it is expressed relatively early , and there are only two , but identical , copies in the fish genome . Four different morpholinos were designed to inhibit miR-205 biogenesis: two targeting the Drosha cleavage site complementary to either the 5′ or 3′ arm of the stem , and two morpholinos similarly targeting the Dicer cleavage site ( Figure S1 ) . These morpholinos were injected under similar conditions as described for miR-206 and compared to the morpholino targeting mature miR-205 . Interestingly , all five morpholinos induced complete or near-complete loss of miR-205 ( Figure 2B ) . Many miRNAs are highly expressed during later stages of embryonic development [21] . Therefore , we tested how long the effect of the morpholinos would last . Although for this series of morpholinos the knockdown is best at 24 hpf , the effect is still significant up to 72 hpf ( Figure 2C ) . Next , we tested a similar series of morpholinos against the miR-30c precursor and analyzed miR-30c expression by Northern blotting ( Figure S2 ) . However , we only observed knockdown for the morpholino targeting mature miR-30c , but not for the other four morpholinos targeting the miR-30c precursor . This could be because miR-30c resides in a family of closely related species , with more sequence variability in the regions outside of the mature miRNA . The precursors of the family members might not all be targeted by these morpholinos ( Figure S2 ) . Thus , not all miRNAs are equally prone to knockdown by morpholinos that target the miRNA precursor . To investigate the effect of morpholinos on exogenously introduced pri-miR-205 , we injected mRNA derived from a GFP construct with pri-miR-205 in the 3′ UTR . Again , we could not detect mature miR-205 derived from this construct after targeting by morpholinos ( Figure 2D ) . Interestingly , the miR-205 precursor also could not be detected in the embryos co-injected with morpholinos , whereas pre-miR-205 could be detected in the absence of morpholinos ( Figure 2D ) . Because pri-miR-205 was cloned in the 3′ UTR of GFP , we monitored GFP fluorescence after injection of this construct . In the presence of a morpholino , GFP fluorescence increased ( Figure 2E ) , suggesting accumulation of the primary miRNA . Therefore , we performed reverse transcriptase PCR ( RT-PCR ) on 8-h-old embryos injected with GFP-pri-miR-205 and a control mRNA ( luciferase ) ( Figure 2F ) . In the presence of a morpholino , the GFP-pri-miR-205 mRNA level is higher compared to control embryos that were not injected with morpholinos . This experiment confirms the GFP data and shows that morpholinos targeting the miRNA precursor inhibit Drosha cleavage . Next , we tested whether processing of the pre-miRNA might also be inhibited by morpholinos . Therefore , we injected a miR-205 precursor in the one-cell–stage embryo . Northern analysis showed that the precursor was processed into mature miRNA in the embryo ( Figure 2G ) . However , co-injection of the overlap loop and non-overlapping loop morpholinos blocked processing completely . There was only a little effect of morpholinos targeting the Drosha cleavage site , probably because they only partially overlap the precursor . A similar analysis was performed for miR-375 , which is expressed in the pancreatic islet and pituitary gland [21] , and has two copies in the zebrafish genome , which differ in the regions outside the mature miRNA . Overlap loop and loop morpholinos were designed for both miR-375–1 and miR-375–2 , and a morpholino against the miRNA star sequence could be used to target both copies of miR-375 simultaneously ( Figure 3A ) . The efficacy of all morpholinos was assessed by determining their effect on injected pri-miR-375–1 or pri-miR-375–2 transcripts ( Figure 3B ) . As expected , each morpholino targeted the transcript to which it was directed . However , the star miR-375 morpholino did not knock down miR-375 completely . In addition , morpholino oligonucleotide ( MO ) miR-375 did not interfere with processing of miR-375 from pri-miR-375–1 , possibly because this primary transcript forms a more stable hairpin . In all cases , the lack of a signal for mature miR-375 coincided with the absence of pre-miR-375 , which could be detected in the absence of a complementary morpholino . Next , all morpholinos were injected separately and in combination , and embryos were subjected to Northern blotting to determine endogenous miR-375 expression at 24 and 48 hpf ( Figure 3C ) . In contrast to the results obtained by in situ hybridization ( see last section ) , the morpholino to mature miR-375 only slightly decreased the expression of miR-375 . However , MO miR-375 could inhibit the activity of a mature miR-375 duplex in a GFP-miR-375-target reporter assay ( Figure 3E ) . The morpholinos targeting only one copy of miR-375 reduced miR-375 expression , with the strongest effect for the morpholinos targeting pri-miR-375–1 . However , simultaneous injection of morpholinos targeting pri-miR-375–1 and pri-miR-375–2 completely knocked down mature miR-375 , indicating that both transcripts are expressed . To further determine the contribution of each transcript to mature miR-375 accumulation , we performed in situ hybridization for pri-miR-375–1 and pri-miR-375–2 ( Figure 3D ) . Both transcripts could not be detected in wild-type embryos . However , pri-miR-375–1 was detected in the pancreatic islet and the pituitary gland in embryos injected with the miR-375–1 loop morpholino and the morpholino to miR-375 star . Similarly , pri-miR-375–2 was only detected in embryos injected with the miR-375–2 loop morpholino , the morpholino to miR-375 star and mature miR-375 . Thus , both transcripts are expressed in the pituitary gland and the pancreatic islet , similar to miR-1 in the developing mouse heart [15] . Together , this indicates that these morpholinos inhibit primary miRNA processing and result in primary miRNA accumulation , as we described for miR-205 . In conclusion , our data demonstrate that morpholinos targeting the miRNA precursor can interfere with primary miRNA processing at either the Drosha or Dicer cleavage step and that morpholinos targeting the mature miRNA can inhibit their activity in a functional assay . Taken together , our data show that different morpholinos targeting the same miRNA may serve as positive controls for miRNA knockdown phenotypes in the embryo . To identify functions for individual miRNAs in zebrafish embryonic development , we knocked down a series of 11 conserved vertebrate miRNAs ( Table S1 ) and analyzed their expression after morpholino knockdown . Injected embryos were monitored phenotypically by microscopic observation until four days postfertilization ( dpf ) . Knockdown of most miRNAs resulted in loss of in situ staining for the respective miRNA . However , we could not observe gross morphological malformations after knockdown of these miRNAs ( Figure 4A ) . Therefore , we analyzed embryos injected with morpholinos against miR-182 , miR-183 , or miR-140 in more detail , because we could easily stain the tissues that express these miRNAs ( Figure 4B ) . Embryos injected with morpholinos against miR-182 or miR-183 , which are expressed in the lateral line neuromasts and hair cells of the inner ear , were treated with DASPEI , which stains hair cells . Embryos injected with a morpholino against miR-140 , which is expressed in cartilage , were subjected to Alcian Blue staining , a cartilage marker . However , staining of these specific cell types that express the miRNA did not uncover any defects upon knockdown ( Figure 4B ) . In conclusion , knockdown of many miRNAs does not appear to significantly affect zebrafish embryonic development , at least not to the extent that can be visualized by the methods used in these examples . MiR-375 is known to be expressed in the pancreatic islet and the pituitary gland , and was first isolated from pancreatic beta cells [21 , 26] . This miRNA is conserved in vertebrates and may regulate insulin secretion by inhibiting myotrophin [26] . We injected a morpholino against mature miR-375 into the one-cell–stage embryo . This morpholino effectively knocked down miR-375 in the first 4 d of development ( Figure 5A ) , and it could also block the activity of an injected miR-375 duplex , as monitored by its effect on a GFP reporter silenced by miR-375 ( Figure 3E ) . During the first 5 dpf , there was no clear developmental defect except for a general delay in development . At around 7 dpf , approximately 80% of the injected embryos died . Next , we analyzed the development of both the pituitary gland and the pancreatic islet , by in situ hybridization with pit1 and insulin markers . This analysis revealed no change in the formation of the pituitary gland ( Figure 5B ) . However , analysis of insulin expression showed a striking malformation of the islet cells in 3-d-old morphant embryos ( Figure 5B ) . Wild-type embryos have a single islet at the right side of the midline , whereas the miR-375 knockdown embryos have dispersed insulin-positive cells . The effect is sequence specific , because a morpholino complementary to the mature miR-375 morpholino inhibited the pancreatic islet phenotype ( Figure 5E ) . The pancreatic islet consists of four cell types , α , β , δ , and PP , expressing glucagon , insulin , somatostatin , and pancreatic polypeptide , respectively . Insulin is the first hormone expressed , and somatostatin co-localizes partially with insulin , whereas glucagon-expressing cells are distinct [27] . A more detailed analysis using somatostatin and glucagon as marker genes revealed a similar pattern of scattered islet cells in the miR-375 morphant ( Figure 5C ) . In zebrafish , insulin is first expressed at the 12-somite stage in a few scattered cells located at the midline , dorsal to the yolk [28] . Insulin-positive cells migrate posteriorly and converge medially to form an islet by 24 hpf . To look at the development of the pancreatic islet in time , we collected MO miR-375 and noninjected control embryos at different stages , and investigated the expression of insulin ( Figure 5D ) . At the 16-somite stage , insulin-positive cells are scattered at the midline in both noninjected and MO miR-375–injected embryos , and a presumptive islet is formed by 24 hpf . Subsequently , when the insulin-positive islet is moving to the right side of the embryo in later stages , the islet breaks apart and insulin-positive cells become scattered in morphant embryos ( Figure 5D ) . Also , in later stages , the phenotype persists , although miR-375 is re-expressed at approximately 5 dpf in morpholino-injected embryos ( Figure 5A ) . Next , we analyzed the effect of all miR-375 control morpholinos described in the previous section , by staining for insulin ( Figure 6A ) . Both the dispersion phenotype and the knockdown were striking for embryos injected with MO miR-375 . Injection of the overlap loop and loop morpholinos targeting pri-miR-375–1 also resulted in scattered insulin-positive cells at 72 hpf , although the effect was weaker compared to MO miR-375 . The miR-375–2 loop and overlap loop morpholinos hardly induced any scattering of insulin-positive cells , whereas the effect was very strong in embryos injected with morpholinos to pri-miR-375–1 and −2 simultaneously . The effect of the miR-375 star morpholino on insulin-positive cells was moderate compared to MO miR-375 . To further prove the specificity of the pancreatic islet phenotype , we injected two control morpholinos against let-7 and miR-124 and analyzed these for miR-375 and insulin expression . None of these control morpholinos showed loss of miR-375 expression or abnormal development of the islet cells ( Figure 6A ) . Next , we analyzed miR-375 knockdown embryos with markers staining the endocrine or exocrine pancreas ( Figure 6B ) . Similar to insulin staining , islet1 expression showed dispersed islet cells in embryos of 48 hpf and 72 hpf , but not 24 hpf . Embryos injected with MO miR-375 exhibited delayed development of the exocrine pancreas , liver , and gut as shown by ptf1a and foxa2 staining . At 72 hpf , these markers showed a similar pattern in MO miR-375–injected embryos as in noninjected embryos at 48 hpf . However , co-injection of miR-375–1/2 loop morpholinos did not delay development of the exocrine significantly , but these embryos still displayed the scattered insulin-positive cells ( Figure 6A ) . This shows that loss of miR-375 mainly results in malformation of the endocrine pancreas , whereas surrounding tissues that do not express miR-375 are not affected .
Functional data on miRNAs in vertebrate development have been obtained mainly from overexpression studies and analysis of conditional Dicer knockouts . For example , the role of miR-430 in zebrafish brain morphogenesis has become clear from experiments that rescued Dicer null mutants by injection of an miRNA duplex that mimicked a miR-430 family member [2] . MiRNA expression can be conveniently studied in zebrafish embryos . However , dissecting miRNA function by disrupting miRNA genes is difficult in zebrafish , because the miRNA is too small to efficiently search for mutations by a target-selected mutagenesis approach [29] . In addition , it is unclear what such point mutations would do to processing or function of the miRNA . It has been shown previously that morpholinos can target miRNAs in the zebrafish embryo [20 , 24] . In a recent study , mature miR-214 was targeted by a morpholino in zebrafish , and this resulted in a change in somite shape , reminiscent of attenuated hedgehog signaling [20] . Although the phenotype could be rescued by simultaneous inhibition of a negative regulator of hedgehog signaling , no positive control morpholinos were reported that could mimic the phenotype . In addition , data were lacking that showed an effect of the morpholino on endogenous miR-214 levels . The results in this paper show that morpholinos targeting the miRNA precursor form a reliable and efficient tool to deplete the embryo of miRNAs during the first 4 d of development , when most organ systems are formed and miRNAs are expressed . We have shown that miRNA expression can be inhibited by targeting the mature miRNA , the precursor miRNA or the primary miRNA . Our data show that such morpholinos can inhibit miRNA processing at the Drosha cleavage step or the Dicer cleavage step , probably by steric blocking , although the exact mechanism is unclear . In addition , morpholinos targeting the mature miRNA can inhibit their activity , probably by preventing binding to a target mRNA . We used morpholinos targeting the mature miRNA for a set of 13 conserved vertebrate miRNAs to identify their developmental functions . By microscopic analysis , we could not observe clear defects associated with loss of 11 of these miRNAs during the first 4 d of embryonic development , although in situ hybridization revealed specific loss of most knocked-down miRNAs . Because all the targeted miRNAs are expressed in very specific tissues and we did not investigate most morphants in much detail by marker analysis , we may have missed subtle defects . In addition , many miRNAs reside in families of related sequence ( e . g . , let-7 and miR-182 ) , and these should possibly be targeted simultaneously by different morpholinos to obtain a biological effect . Furthermore , in those instances in which miRNAs of unrelated sequence target a similar set of mRNAs when expressed in the same tissue [21] , removing only one miRNA might not have a profound impact on transcript levels or expression . Finally , microarray analysis and computational predictions have shown that a single miRNA may regulate hundreds of mRNAs [30 , 31] , but that some miRNAs act as a backup for mRNAs that are already repressed transcriptionally [32] . Thus , knockdown of such miRNAs might not dramatically affect gene expression , but ensure robustness of protein interaction networks as for example miR-7 in Drosophila [33] . In zebrafish , there are two copies of miR-375 , and in human and mouse only one copy has been identified [34] . To verify the miR-375 knockdown phenotype , we designed control morpholinos targeting both precursors simultaneously ( MO miR-375 star ) and separately . Complete knockdown was only observed in those instances in which both miR-375 copies were targeted simultaneously . This also led to scattered islet cells , proving the specificity of the phenotype . However , knockdown with miR-375–1/2 loop morpholinos did not delay development as seen in the knockdown with the mature miR-375 morpholino . This shows the strength of using control morpholinos and excludes the delayed development as a relevant miR-375 loss-of-function phenotype . A moderate version of the phenotype was also observed in embryos injected with a morpholino specifically targeting miR-375–1 . Thus , a reduction in the level of miR-375 already disturbs islet integrity . Similar to mouse miR-1 [15] , miR-375 copies survived evolution and are expressed similarly in time and space , probably to ensure the high intracellular concentration of miR-375 necessary to repress many weakly binding targets . In a forward genetic screen , several mutants were identified with improper development of the endocrine pancreas [35] . These mutants fall into three classes: ( 1 ) mutants with severely reduced insulin expression; ( 2 ) mutants with reduced insulin expression and abnormal islet morphology; and ( 3 ) mutants with normal levels of insulin expression and abnormal islet morphology . However , in all of these mutants , islet cells do not merge into an islet from their first appearance at approximately the 14-somite stage . Our miR-375 knockdown phenotype differs from this , because in the first instance , an islet is formed at approximately 24 hpf , but in later stages , the islet falls apart into small groups of cells . This rules out a general role for miR-375 in early endocrine formation as is seen for Wnt5 [36] , but rather indicates a role in maintenance of tissue identity , which is assumed to be a general function of miRNAs in development [21] . It is as yet unclear which miR-375 targets are involved in the phenotype . Work in cell lines has implicated miR-375 in insulin secretion by targeting myotrophin [26] . The zebrafish homolog of myotrophin also contains a seven-nucleotide seed match to miR-375 ( unpublished data ) , but future studies should reveal whether this target or many other predicted targets are relevant to the phenotype . The specific expression of miR-375 in the pancreatic islet and its implication in insulin secretion make it a candidate drug target in diabetes , e . g . , to influence insulin levels in the blood . However , our data show that if miR-375 is used as a drug target , developmental side effects need to be taken into account .
Morpholinos were obtained from Gene Tools LLC ( http://www . gene-tools . com ) and dissolved to a concentration of 5 mM in water . Morpholinos were injected into one- or two-cell–stage embryos at concentrations between 200 μM and 1 , 000 μM , and per embryo , one nl of morpholino solution was injected . RNA oligos ( Table S2 ) were obtained from Sigma ( http://www . sigmaaldrich . com ) and dissolved to a concentration of 100 μM in distilled water . Oligos were annealed using a 5x buffer containing 30 mM HEPES-KOH ( pH 7 . 4 ) , 100 mM KCl , 2 mM MgCl2 , and 50 mM NH4Ac . Typically , 1 nl of a 10 μM miRNA duplex solution was injected . All morpholino sequences used in this study are listed in Table S1 . The miR-30c and miR-375 reporter constructs were made by cloning two annealed oligos containing two perfectly complementary miRNA target sites into pCS2 ( Clontech , http://www . clontech . com ) containing a gfp gene between BamHI and ClaI restriction sites . A construct containing pri-miR-205 was made by amplifying a genomic region ( 801 base pairs ) containing the miR-205 precursor ( miR-205-hairpinF ggcattgaattcataaCCTCTTACCTGCATGACCTG; miR-205-hairpinR ggcatttctagaGTGTGTGCGTGTATTCAACC ) . The resulting PCR fragment was cloned between XbaI and EcoRI restriction sites of PCS2GFP . Pri-miR-375–1 and pri-miR-375–2 constructs were made by amplifying genomic regions containing miR-375–1 and miR-375–2 precursors ( WKmiR-375–1F-pCS2 gcccgggatccTGTGTCTTGCAGGAAAAGAG; WKmiR-375–1R-pCS2 attacgaattcTCAAACTCTCCACTGACTGC; and WKmiR-375–2F-pCS2 gcccgggatccGCCCTCCCATTTGACTC; WKmiR-375–2R-pCS2 attacgaattcAATGAGTGCACAAAATGTCC ) , and cloning of the resulting PCR fragments into the BamHI and EcoRI sites of pCS2 . mRNA was synthesized using SP6 RNA polymerase . Luciferase mRNA was derived from pCS2 containing luciferase between BamHI and EcoRI sites . In situ hybridization was performed as described previously [37] . LNA probes for miRNA detection were obtained from Exiqon ( http://www . exiqon . com ) and labeled using terminal transferase and DIG-11-ddUTP . cDNA clones for pri-miR-375–1 , pri-miR-375–2 , pit1 , insulin , somatostatin , and glucagon were used for antisense DIG-labeled probe synthesis by T7 or Sp6 RNA polymerase . For Northern blotting , total RNA was isolated from ten embryos per sample using Trizol reagent ( Invitrogen , http://www . invitrogen . com ) . RNA was separated on a 15% denaturing polyacrylamide gel . Radiolabeled DNA probes complementary to miRNAs or 5S RNA ( atcggacgagatcgggcgta ) were used for hybridization at 37 °C . Stringency washes were done twice for 15 min at 37 °C using 2 × SSC 0 . 2% SDS . Alternatively , DIG-labeled LNA probes were used for hybridization at 60 °C and stringency washes were performed at 50 °C with 2x SSC 0 . 1% SDS for 30 min and 0 . 5x SSC 0 . 1% SDS for 30 min . For RT-PCR , RNA was isolated with Trizol , treated with DNAse ( Promega , http://www . promega . com ) and subsequently purified again using Trizol . cDNA was made with a poly dT primer . Primers used for amplification were miR-205-hairpinF and miR-205-hairpinR , and lucF ( ATGGAAGACGCCAAAAACATAAAG ) and lucR ( ATTACATCGATTTACACGGCGATCTTTCC ) . For Alcian Blue staining , embryos were fixed for 1 h at room temperature in 4% PFA in PBS , rinsed for 5 min in 50% MeOH , and stored overnight in 70% MeOH at 4 °C . Next , embryos were incubated for 5 min in 50% MeOH and for 5 min in 100% EtOH . Embryos were stained at room temperature with Alcian Blue ( Sigma ) for 90 min with continuous shaking . Subsequently , embryos were rinsed in 80% , 50% , and 25% EtOH for 2 min each and two times in water containing 0 . 2% Triton and neutralized in 100% Borax solution . Finally , embryos were incubated for 60 min in digest solution ( 60% Borax solution , 1 mg/ml colleganase-free and elastase-free trypsin , 0 . 2% trypsin ) and stored in 70% glycerol . Staining of the hair cells was done by incubating live embryos for 5 min in a 200 μM solution of Daspei ( Sigma ) in + chorion . After rinsing twice in + chorion , embryos were anesthetized using MS222 and mounted in methylcellulose .
|
The striking tissue-specific expression patterns of microRNAs ( miRNAs ) suggest that they play a role in tissue development . These small RNA molecules ( ∼22 bases in length ) are processed from long primary transcripts ( pri-miRNA ) and regulate gene expression at the posttranscriptional level . There are hundreds of different miRNAs , many of which are strongly conserved . Vertebrate embryonic development is most easily studied in zebrafish , but genetically disrupting miRNA genes to see which miRNA does what is technically challenging . In this study , we interfere with miRNA function during the first few days of zebrafish embryonic development by introducing specific antisense morpholino oligonucleotides ( morpholinos have been used previously to interfere with the synthesis of the much larger mRNAs ) . We show that morpholinos targeting the miRNA precursor can block processing of the pri-miRNA or directly inhibit the activity of the mature miRNA . We also used morpholinos to study the developmental effects of miRNA knockdown . Although we did not observe gross phenotypic defects for many miRNAs , we found that zebrafish miR-375 is essential for formation of the insulin-secreting pancreatic islet . Loss of miR-375 results in dispersed islet cells by 36 hours postfertilization , representing one of the first vertebrate miRNA loss-of-function phenotypes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"developmental",
"biology",
"danio",
"(zebrafish)",
"vertebrates",
"molecular",
"biology"
] |
2007
|
Targeted Inhibition of miRNA Maturation with Morpholinos Reveals a Role for miR-375 in Pancreatic Islet Development
|
Sister chromatid cohesion , which is mediated by the cohesin complex , is essential for the proper segregation of chromosomes in mitosis and meiosis . The establishment of stable sister chromatid cohesion occurs during DNA replication and involves acetylation of the complex by the acetyltransferase CTF7 . In higher eukaryotes , the majority of cohesin complexes are removed from chromosomes during prophase . Studies in fly and human have shown that this process involves the WAPL mediated opening of the cohesin ring at the junction between the SMC3 ATPase domain and the N-terminal domain of cohesin's α-kleisin subunit . We report here the isolation and detailed characterization of WAPL in Arabidopsis thaliana . We show that Arabidopsis contains two WAPL genes , which share overlapping functions . Plants in which both WAPL genes contain T-DNA insertions show relatively normal growth and development but exhibit a significant reduction in male and female fertility . The removal of cohesin from chromosomes during meiotic prophase is blocked in Atwapl mutants resulting in chromosome bridges , broken chromosomes and uneven chromosome segregation . In contrast , while subtle mitotic alterations are observed in some somatic cells , cohesin complexes appear to be removed normally . Finally , we show that mutations in AtWAPL suppress the lethality associated with inactivation of AtCTF7 . Taken together our results demonstrate that WAPL plays a critical role in meiosis and raises the possibility that mechanisms involved in the prophase removal of cohesin may vary between mitosis and meiosis in plants .
The timely establishment and dissolution of sister chromatid cohesion is essential for the proper segregation of chromosomes during cell division , as well as the repair of DNA damage and the control of transcription ( reviewed in [1]–[4] ) . Four proteins form the core cohesin complex: Structural Maintenance of Chromosome ( SMC ) proteins 1 ( SMC1 ) and 3 ( SMC3 ) , Sister Chromatid Cohesion ( SSC ) protein 3 ( SCC3 ) , and an α-kleisin , either SCC1 which is part of the mitotic cohesion complex , or REC8 that functions during meiosis . Studies in several organisms have shown that cohesin complex components and the general mechanisms of cohesin action are conserved across species; however variations in complex member composition and the mechanistic roles of some complex members have been observed between some species and between mitosis and meiosis ( reviewed in [3] , [5]–[7] ) . Cohesin complexes are recruited to the chromatin by the Scc2/Scc4 complex throughout the cell cycle , with most of the complexes loaded onto chromosomes during telophase/G1 [8]–[10] . Prior to S-phase cohesin association with the chromatin is dynamic and regulated in part by a complex which has been referred to by several names , including “releasin” , the “antiestablishment” and/or the “antimaintenance” complex [11] , [12] . This complex consists of the Wings apart-like protein ( Wapl ) and the Precocious dissociation of sisters protein 5 ( Pds5 ) [13]–[17] . In vertebrates sororin is also part of the complex [18] , [19] . The Ctf7/Eco1-dependent acetylation of SMC3 inhibits Wapl and results in the stable association of cohesin with chromosomes [20]–[24] . Cohesin is subsequently removed from chromosomes in steps [25] . While the specific details vary somewhat depending on the organism being studied , the general process appears to be relatively conserved . In higher eukaryotes , arm cohesin is removed during mitotic prophase in a Polo-like kinase , cyclin-dependent kinase and Wapl dependent process that involves opening of the cohesin ring at the junction between the SMC3 ATPase domain and the N-terminal winged-helix domain ( WHD ) of SCC1 [26]–[29] . Centromeric cohesin is protected by the Shugoshin ( Sgo1 ) -protein phosphatase 2A ( PP2A ) complex , which binds and dephosphorylates cohesin , protecting it from Wapl [30]–[32] . At the metaphase to anaphase transition the metallo-proteinase separase is activated and cleaves the SCC1 subunit of centromere-localized cohesin , allowing the cohesin ring to open and the sister chromatids to disjoin [33] . Meiotic cohesin is removed in three steps: a prophase step , followed by the separase dependent cleavage of chromosome arm associated REC8 at anaphase I; finally centromere associated REC8 is cleaved by separase at anaphase II [34]–[36] . The importance of Wapl in controlling mitotic sister chromatid cohesion has been known for some time , but it is only recently that we have begun to understand how specifically Wapl helps facilitate the interaction of cohesin with chromosomes . Wapl was first identified in Drosophila melanogaster as a protein involved in the regulation of heterochromatin organization , with mutant flies containing parallel sister chromatids with loosened cohesion at their centromeres [37] . More recently structural studies on Wapl and its role ( s ) in sister chromatid cohesion during mitosis have been conducted in several organisms , including fungi , fly and vertebrates [38]–[40] . Wapl proteins from different species contain a conserved C-terminus with more divergent N-terminal domains . The divergent N-terminus appears to be a primary Pds5 binding region , while the C-terminus contains cohesin-binding determinants . While a number of similarities exist between the yeast and vertebrate proteins , structural and binding differences have also been identified . These results , along with the observation that wapl mutants in different organisms can exhibit different phenotypes , indicate that there is still much we do not understand about Wapl and how its structure is related to its function . Furthermore , while the effect of Wapl inactivation on mitosis has been studied in several organisms , little is known about the role of the protein during meiosis . In the current study , we have characterized WAPL in the model organism Arabidopsis thaliana . We show that while AtWAPL plays a critical role in facilitating sister chromatid separation during meiosis , it appears to have a more minor role in somatic cells . AtWAPL mutations resulted in reduced male and female fertility but had little effect on plant growth . Meiotic defects , including alterations in chromosome structure and the separation of homologous chromosomes and sister chromatids was observed in most meiocytes . The removal of cohesin from meiotic chromosomes during prophase was blocked in Atwapl mutants resulting in chromosome bridges , broken chromosomes and the uneven segregation of chromosomes . Finally , we show that AtWAPL mutations can partially suppress the lethality associated with inactivation of the cohesin establishment factor , AtCTF7 .
In order to determine if the two predicted Arabidopsis WAPL genes are in fact involved in controlling sister chromatid cohesion , we characterized T-DNA insertion lines that were available in the Arabidopsis Stock Center . Two lines were characterized for AtWAPL1 ( Atwapl1-1 and Atwapl1-2 , Figure 1B ) and one line for AtWAPL2 ( Atwapl2 , Figure 1B ) . Plants homozygous for the individual insertion lines displayed normal vegetative growth , development and fertility when compared with wild type plants . The high degree of similarity between AtWAPL1 and AtWAPL2 raised the possibility that the two genes share overlapping functions . Therefore , we crossed Atwapl2 with both Atwapl1-1 and Atwapl1-2 . Plants double homozygous for both combinations ( Atwapl1-1wapl2 and Atwapl1-2wapl2 ) were isolated and studied . Plants homozygous for both the Atwapl1-2 and Atwapl2 mutations displayed normal vegetative growth and development , but a reduction in fertility . Average seed set/slique in Atwapl1-2wapl2 plants ( 43 . 7±5 . 1 , n = 32 ) is lower than wild type ( 53 . 7±4 , n = 42 , p<0 . 0001 ) . Plants containing the Atwapl1-1wapl2 double mutant combination showed a more pronounced phenotype . Specifically , the plants grew somewhat slower than wild type plants ( Figure 2A ) and produced shorter siliques , which contained fewer seeds ( 37 . 5±6 . 7 , n = 45 , p<0 . 0001 ) than Atwapl1-2wapl2 sliques . Further analysis of both double mutant combinations identified similar alterations in reproduction , including aborted pollen and ovules prior to fertilization and embryo defects in approximately 25% of the fertilized seed , with higher numbers of aborted pollen , ovules and seed consistently observed in Atwapl1-1wapl2 plants ( Figure 2B , C ) . The Atwapl1-2 and Atwapl2 T-DNA insertions are in the first exon and intron , respectively , while the Atwapl1-1 insert is located in intron 6 ( Figure 1B ) . In order to investigate the differences we observed between the two double mutant combinations and determine if the T-DNA insertions result in complete inactivation of the genes we examined AtWAPL1 and AtWAPL2 transcriptional patterns in both wild type and mutant plants . Transcripts for both genes were detected in roots , leaves , buds and sliques of wild type plants; little to no transcript for either gene was detected in stems ( Figure 3A ) . While both genes are active , AtWAPL1 transcripts were more abundant than those for AtWAPL2 in all tissues examined , with the highest overall levels observed in roots ( Figure 3A ) . Analysis of WAPL transcript levels by qPCR with primers located downstream of the T-DNA inserts in the different double mutant backgrounds indicated that the Atwapl1-1 mutation effectively results in complete inactivation of the gene . In contrast , RNA corresponding to sequences downstream of the Atwapl1-2 T-DNA insert were detected at levels approximately 75% of wild type ( Figure 3B ) . The weak phenotype associated with Atwapl1 . 2wapl2 plants may be due to the production of reduced levels and/or a partially functional protein from the Atwapl1 . 2 allele . Low levels ( >10% wild type ) of truncated Atwapl2 transcripts were also detected downstream of the Atwapl2 T-DNA insert . This raised the possibility that a small amount of truncated WAPL2 protein may also be produced . The truncated protein would be missing at least the first 136 amino acids of the protein , including a stretch of highly conserved amino acids ( Figure S1 ) . Because the Atwapl1-1wapl2 mutant combination resulted in the most severe phenotype , we confined our more detailed analyses to Atwapl1-1wapl2 plants . Anthers of Atwapl1-1wapl2 plants contain less pollen than wild type plants ( 229±21 . 3 , n = 15 verses 458±23 . 8 , n = 10 , p<0 . 0001 ) and 28% of the pollen ( n = 2752 ) that is produced is not viable , appearing green and shriveled when analyzed by Alexander stain ( Figure 2B ) . Analysis of seed development in Atwapl1-1wapl2 plants revealed that 28% of the ovules ( n = 1689 ) abort prior to fertilization , while 23% of the seed ( n = 2022 ) that is produced is shrunken and shriveled . Examination of cleared ovules from developmentally staged siliques of Atwapl1-1wapl2 plants identified defects beginning after the Megaspore Mother Stage ( Figure 4 A , D ) . Approximately 16% of ovules examined ( n = 409 ) arrest at FG1 with one nucleus ( Figure 4E ) . Approximately eight percent of the ovules arrest at FG2 ( Figure 4F ) . In most instances the arrested nuclei persisted throughout ovule development and were observed in siliques with normal FG7 ovules . The presence of aborted ovules and reduced numbers of pollen in Atwapl1-1wapl2 plants suggested that AtWAPL plays an important role in meiosis . To investigate this possibility further we analyzed DAPI ( 4′ , 6-diamidino-2-phenylindole ) stained meiotic chromosomes in Atwapl1-1wapl2 plants . Early stages of meiosis appeared relatively normal in the mutant . As observed in wild type meiocytes , chromosomes began to condense as fine thin threads during leptotene ( Figure 5A , E ) and homologous chromosome co-alignment and pairing occurred during early to mid zygotene ( Figure 5B , F ) . In wild type meiocytes homologous chromosomes are fully synapsed by the beginning of pachytene ( Figure 5C ) . While most late zygotene/pachytene stage meiocytes exhibited normal synapsis , in 15% of the Atwapl1-1wapl2 pachytene meiocytes ( n = 135 ) the chromosomes co-aligned but did not synapse completely ( Figure 5G ) . In addition , four to six brightly stained chromocenters are typically observed in wild type meiocytes , while in the mutant we observed three or fewer heterochromatin regions in 60% of the Atwapl1-1wapl2 pachytene cells , suggesting that abnormal association of heterochromatic regions may occur in the mutant . Desynapsis occurs during diplotene ( Figure 5D ) with five bivalents appearing at diakinesis in wild type meiocytes ( Figure 5I ) . The five bivalents align at the equatorial plane at metaphase I ( Figure 5J ) . Segregation of homologous chromosomes and then sister chromatids at anaphase I and anaphase II , respectively , results in the presence of four sets of five individual chromosomes at the cell poles by telophase II ( Figures 5K , L and Q , R ) . Early diplotene appeared relatively normal in the mutant ( Figure 5H ) . However , alterations were observed by diakinesis in essentially all cells . Specifically meiocytes were observed in which the chromosomes condensed into either one or two large intertwined masses of chromatin ( Figure 5M , n = 25 ) . The chromosomes continued to appear primarily as one intertwined mass as they further condensed and moved to the cell equator; five normal appearing individual bivalents were never observed ( Figure 5N , n = 23 ) . While some ( <20% ) normal cells were observed at the metaphase I-anaphase I transition , most cells contained stretched chromosomes that did not separate properly ( Figure 5O , n = 57 ) . Chromosome bridges and lagging chromosomes were observed by late anaphase I and telophase I ( Figure 5P , n = 31 ) , respectively in the majority of meiocytes . In most cells ( 68% , n = 31 ) “sticky” chromosome masses were observed at one or both poles at metaphase II ( Figure 5U ) ; however in approximately 30% of the meiocytes individual chromosomes appeared to align normally . Twenty or more chromosomes/chromosome fragments were typically observed scattered around most ( 62% , n = 26 ) anaphase II and telophase II cells ( Figure 5V ) . Ultimately , a mixture of polyads ( 6% ) , tetrads ( 26% ) containing a mixture of shrunken and mis-shaped microspores with varying amounts of DNA ( Figure 5W , X ) , and relatively normal appearing tetrads were observed ( n = 506 ) . One of the earliest defects observed in the meiocytes of Atwapl1-1wapl2 plants is a reduced number of heterochromatin regions , suggesting that AtWAPL is important early in prophase I , possibly in controlling heterochromatin structure . In order to investigate this possibility , fluorescence in situ hybridization ( FISH ) experiments were conducted using a 180 bp repetitive centromere fragment as a probe on meiocytes of wild type and Atwapl1-1wapl2 plants . Eight to ten centromere signals were observed in meiocytes during leptotene in both wild type ( mean = 9 . 2±0 . 71 , n = 26 ) and Atwapl1-1wapl2 ( mean = 9 . 0±1 . 2 , n = 29 ) plants ( Figure 6 A , E ) . Four to six signals were normally observed in wild type meiocytes ( mean = 5 . 4±0 . 5 , n = 25 ) during zygotene as homologous chromosomes pair ( Figure 6B ) . Alterations were first observed at zygotene when approximately 50% of the Atwapl1-1wapl2 meiocytes observed ( n = 30 ) were found to contain clusters of condensed signals ( Figure 6F ) . At pachytene wild type and Atwapl1-1wapl2 meiocytes contained on average 4 . 8±0 . 35 ( n = 8 ) and 3 . 04±1 . 3 ( n = 84 ) centromere signals , respectively with 50% of Atwapl1-1wapl2 meiocytes showing one or two clusters of signals ( Figure 6 C , G ) . Five pairs of centromere signals corresponding to the five bivalents are visible at diakinesis and early metaphase I in wild type meiocytes , followed by ten signals during anaphase I/telophase I and 20 during meiosis II ( Figure 6D , I–L , n = 48 ) . In contrast , centromere signals continued to cluster together at late diplotene and diakinesis ( Figure 6H ) in 60% of the Atwapl1-1wapl2 meiocytes examined ( n = 24 ) . Individual centromere signals could however be observed within the condensed chromatin at metaphase I ( n = 15 ) ( Figure 6M ) . While some normal anaphase I cells were observed , more than ten centromere signals were observed beginning at anaphase I in 65% of the Atwapl1-1wapl2 meiocytes observed ( n = 27 ) , suggesting that either centromere cohesion is lost prematurely or never properly formed in these cells . Approximately 35% of the cells proceed normally through the remainder of meiosis . However , in most cells centromere signals of varying intensities were observed that associated with mis-segregated chromosomes and chromosome fragments at telophase I ( Figure 6N ) and chromosomes scattered around the cells during meiosis II ( Figure 6O , P , n = 24 ) . Results from our chromosome spreading suggested that defects in homologous chromosome pairing and synapsis may also exist in the mutant . To investigate this possibility further we performed FISH using a telomere-derived fragment that also strongly labels a region proximal to the centromere of chromosome 1 [42] . Two strong chromosome 1 signals with weaker telomere signals were observed during leptotene in both wild type ( n = 17 ) and Atwapl1-1wapl2 ( n = 24 ) meiocytes ( Figure 7A , E ) . One strong signal was observed in wild-type meiocytes starting at zygotene and extending through diplotene ( mean = 1 . 02±0 . 17 , n = 36 ) ( Figure 7B–D ) . Cells with either one or two chromosome 1 signals were observed during these stages in Atwapl1-1wapl2 plants . While most cells resembled wild type meiocytes and contained one signal ( mean = 1 . 19±0 . 40 ) during zygotene , pachytene and diplotene ( Figure 7F–H ) , approximately 20% of the nuclei observed ( n = 139 ) contained two widely spaced chromosome 1 signals throughout prophase ( Figure 7I–L ) . Therefore , a small but significant fraction of meiocytes do not undergo normal pairing and synapsis . Meiotic prophase was investigated further by analyzing the distribution of ASY1 and ZYP1 . ASY1 is a meiosis-specific protein that is intimately associated with chromosome axes during prophase I . Differences were not observed in ASY1 labeling between wild type and Atwapl1-1wapl2 meiocytes ( Figure S2 ) . In both wild type and Atwapl1-1wapl2 meiocytes ASY1 appears as diffuse foci during G2 , forming thin threads that co-localize with the developing univalent axes during leptotene . It is associated with the axes of the synapsed chromosomes during pachytene and disappears from chromosomes at diplotene . Subtle alterations were however observed in ZYP1 distribution in approximately 25% of the meiocytes . ZYP1 , an axial element protein , appears at zygotene as foci . ZYP1 signals extend during pachytene producing a continuous signal between the synapsed homologous chromosomes [43] . The majority ( 77% ) of Atwapl1-1wapl2 pachytene meiocytes examined ( n = 30 ) resembled wild type and exhibited continuous ZYP1 signals . However , 23% of the meiocytes exhibited more diffuse ZYP1 labeling patterns and contained pachytene chromosomes that exhibited discontinuous and/or unpaired ZYP1 signals ( Figure S3 ) . Therefore , while ASY1 and ZYP1 appear to load normally on Atwapl1-1wapl2 meiotic chromosomes , some meiocytes do not undergo complete synapsis . The observed alterations in chromosome structure and the “sticky” nature of meiotic chromosomes suggested that Atwapl1-1wapl2 plants may be defective in the release of cohesin during prophase . In order to investigate this possibility , we performed immunolocalization experiments on Atwapl1-1wapl2 and wild type meiocytes with antibodies to SYN1 , the Arabidopsis homolog of REC8 [44] . Cohesin labeling appeared normal in Atwapl1-1wapl2 plants during early stages of prophase I . At interphase SYN1 exhibited diffuse nuclear labeling with the signal decorating the developing chromosomal axes beginning at early leptotene and extending into zygotene . During late zygotene and pachytene the protein lined the chromosomes ( Figure 8A , B , G , H ) . A large amount of SYN1 is released from wild type meiotic chromosomes during diplotene ( Figure 8C , n = 9 ) and diakinesis ( Figure 8D , n = 7 ) as the chromosomes condense . By prometaphase I SYN1 is barely detectable on wild type chromosomes ( Figure 8E , n = 14 ) . In contrast , strong SYN1 labeling was consistently observed from diplotene into anaphase I in the mutant ( Figure 8I–L ) . SYN1 was observed on “sticky” metaphase I chromosomes ( Figure 8K , n = 5 ) and stretched bivalents during anaphase I ( Figure 8L , n = 10 ) . While 20% of metaphase II meiocytes ( n = 25 ) showed faint SYN1 signals , the majority of meiocytes did not , suggesting the protein is removed during telophase I and interphase II . As part of our studies to better define meiotic stages in the mutant and further characterize chromosome behavior , we performed immunolocalization studies using β-tubulin antibody on wild type and Atwapl1-1wapl2 meiocytes . No significant differences in β-tubulin labeling were observed between wild type and mutant plants during interphase and prophase I . Wild type spindles exhibit a bipolar configuration during metaphase I and anaphase I ( Figure 9A , B ) , with radial spindles forming between the two groups of chromosomes at telophase I ( Figure 9C ) . Two bipolar spindles , which are perpendicular to each other , are then observed during metaphase II and anaphase II ( Figure 9D , E ) , with radial microtubules again forming between the four separated nuclei during telophase II . While normal bipolar spindles were formed during metaphase I and metaphase II in approximately 35% of Atwapl1-1wapl2 meiocytes , the majority of cells showed abnormal spindle configurations . For example , cells in which spindle microtubules passed over the chromosomes were observed ( Figure 9F , n = 20 ) . During anaphase I spindles were commonly stretched and not well defined ( Figure 9G , n = 31 ) , with alterations being observed in the radial spindles during telophase I ( Figure 9H , n = 14 ) and interphase II . Two types of alterations were commonly observed during meiosis II . Approximately 30% of metaphase II cells contained parallel spindles ( Figure 9I , J , n = 12 ) , while another 30% of the cells lacked metaphase II spindles altogether and instead contained random microtubule networks ( Figure 9K , L , n = 13 ) . A large number of additional alterations , including cells lacking metaphase I spindles , stretched metaphase II spindles , and cells with four bipolar or parallel spindles were observed at lower frequencies ( Figure S4 ) . The siliques of Atwapl1-1wapl2 plants contain approximately 25% aborted seed ( n = 2022 ) , suggesting defects in embryo and/or endosperm development . In order to investigate this possibility we examined cleared seeds in siliques of self-fertilized Atwapl1-1wapl2 plants and found that 23% of the seed contained abnormal embryos ( n = 31 siliques ) . Alterations in embryo development were observed as early as the two cell stage when instead of the typical vertical division of the apical cell , 9% the mutant embryos ( n = 61 ) performed a horizontal division ( Figure 10A , E ) . Alterations in the suspensor were also observed early in development in approximately 5% of the seeds ( n = 39 ) . Suspensors with either two cells instead of a file of four cells and suspensors with abnormal shapes were observed ( Figure S5A ) . A common alteration at later stages involved embryos exhibiting altered division planes and shapes ( Figure 10G , n = 10 ) . Another common defect involved either abnormal or uncontrolled division in cells destined to become the suspensor hypophysis ( Figure 10G , n = 14 ) . In early cotyledon stage siliques , both normal-appearing and abnormal embryos that were either arrested or delayed were observed at several developmental stages , including: dermatogen , globular and early heart stages ( Figure 8H , S5 ) . Shrunken seeds with no trace of an embryo were also observed . Alterations in embryo development could result from the wapl mutations directly affecting cellular division in the embryo or from fertilization events involving abnormal gametes . Results from an analysis of embryo development in reciprocal crossing experiments and the analysis of wapl1-1wapl2+/− and wapl2wapl1-1+/− plants suggest that the embryo defects may result from multiple factors . When wapl1-1wapl2 was used as the female in crosses with wild type pollen 2 . 9% of the seed was defective ( n = 105 ) with no sign of embryo development , similar to wild type crossing experiments ( 2% , n = 125 ) . In contrast , when wild type females were crossed with wapl1-1wapl2 pollen , 12 . 3% of the embryos ( n = 173 ) were defective , exhibiting altered divisional planes and defective suspensors . An additional 2 . 8% of the seed showed no sign of embryo development , similar to wild type . Embryo defects were also observed in self fertilized wapl1-1wapl2+/− ( 8 . 6% , n = 214 ) and wapl2wapl1-1+/− ( 7 . 1% , n = 197 ) plants . These results clearly show that alterations associated with wapl1-1wapl2 pollen are sufficient to produce embryos with altered divisional planes and defective suspensors . However , the frequency of defective embryos is doubled when both the sperm and egg carry the wapl mutations , suggesting a synergistic effect . Further experiments are required to better define the underlying basis for the defects . The fact that Atwapl1-1wapl2 plants grow and develop normally , albeit slightly slower than wild type suggested that WAPL does not play a major role in nuclear division in somatic cells . In order to determine if WAPL mutations have an effect on mitotic cells we examined root tips of Atwapl1-1wapl2 plants . The majority of mitotic figures observed in the root tips of Atwapl1-1wapl2 plants ( n = 120 ) appeared normal , with ten pairs of chromosomes condensing at the metaphase plate and then segregating at anaphase/telophase ( Figure 11A–C ) . Altered mitotic figures were however observed in approximately 20% of the cells , with most of the alterations resembling those observed in meiotic cells . The most common alterations were the presence of “sticky chromosomes” at metaphase ( Figure 11A , D ) that failed to segregate properly at anaphase ( Figure 11B , E ) resulting in chromosome bridges , lagging chromosomes and possibly chromosome fragments at telophase ( Figure 11C , F ) . Immunolocalization using antibody to SMC3 [35] was performed on root tips of Atwapl1-1wapl2 plants to determine if cohesin is released normally during mitotic prophase . SMC3 displayed a diffuse labeling pattern during interphase in both wild type and Atwapl1-1wapl2 plants ( Figure 12A , E ) . The chromosome bound SMC3 signal gradually decreased during prophase and was absent from the chromosomes by metaphase in both wild type and Atwapl1-1wapl2 plants ( Figure 12B , F ) . Although weak SMC3 signals were sometimes observed , chromosome bound SMC3 signal was never observed ( n = 20 ) during anaphase and telophase ( Figure 12C , D , G , H ) , even on “sticky” metaphase chromosomes or chromosome bridges during anaphase and telophase ( Figure 12F–H ) . Therefore , mitotic cohesin complexes appear to be removed normally during mitosis . However , we can not rule out the possibility that small amounts of cohesin remain on the chromosomes leading to the mitotic alterations we observe . Finally , we investigated the possible genetic interaction between AtWAPL and AtCTF7 by crossing Atwapl1-1wapl2 plants with plants heterozygous for a T-DNA insertion in AtCTF7 [45] . We were particularly interested in determining if inactivation of WAPL can suppress the dramatic affect of Atctf7 mutations . AtCTF7 is an essential gene with ctf7 mutations causing female gametophyte lethality [45] . Plants homozygous for Atctf7 mutations can however be recovered at very low frequencies [46]; the plants are dwarf , completely sterile and display multiple developmental alterations ( Figure 13A ) . PCR genotyping was used to first identify plants triple heterozygous for the three mutations and then Atwapl1-1wapl2ctf7+/− plants were identified in F2 populations of several different crosses . Atwapl1-1wapl2ctf7+/− plants resembled Atwapl1-1wapl2 plants , displaying relatively normal vegetative growth and reduced fertility ( Figure 13C ) . Atwapl1-1wapl2ctf7+/− anthers ( n = 16 ) produce on average 234±18 . 2 pollen and 41% ( n = 1642 ) of the pollen produced was not viable ( Figure 13B ) . Likewise , 43% of the ovules in siliques ( n = 21 ) of Atwapl1-1wapl2ctf7+/− plants abort prior to fertilization and 52% of the seed produced ( n = 2036 ) is shrunken and shriveled . Ultimately Atwapl1-1wapl2ctf7+/− plants produce on average17 . 9±3 . 3 viable seeds per silique ( n = 23 ) . Plants homozygous for mutations in all three genes ( Atwapl1-1wapl2ctf7 ) were readily obtained from selfed Atwapl1-1wapl2Ctf7+/− plants . The vegetative growth of Atwapl1-1wapl2ctf7 plants is relatively normal , with the growth rate and overall size of the plants resembling that of wild type ( Figure 13A ) . Further , while Atctf7 plants are completely sterile , Atwapl1-1wapl2ctf7 plants produce some viable pollen and seed ( Figure 13B , C ) . Atwapl1-1wapl2ctf7 plants produce 47±15 . 5 viable pollen/anther ( n = 16 ) and approximately 10 . 8±4 . 3 normal seeds/silique ( n = 23 ) . Fewer ovules appear to be fertilized in Atwapl1-1wapl2ctf7 plants; however those that are fertilized develop into viable seed .
Our results show that while AtWAPL is not critical for nuclear division in somatic cells , it is required for the proper release of cohesin from meiotic chromosomes during prophase . Most Atwapl1-1wapl2 male meiocytes observed at metaphase I/early anaphase I contained “sticky chromosomes” that displayed strong SYN1 labeling . SYN1 is undetectable on the chromosomes of wild type meiocytes beginning at pro-metaphase I [44] . The formation of chromosome bridges at anaphase I and ultimately mis-segregated chromosomes at telophase I is likely due to the prolonged presence of chromosome arm cohesin in Atwapl1-1wapl2 meiocytes . While some Atwapl1-1wapl2 metaphase II chromosomes showed faint cohesin signals , the majority did not . This suggests that arm-associated cohesin complexes normally removed by WAPL during prophase are instead removed during telophase I/interphase II in the mutant , potentially through the action of separase . Although we did not specifically analyze meiosis in megasporocytes , the fact that a relatively large number of female gametophytes arrest at FG1 or FG2 suggests that inactivation of AtWAPL affects both male and female meiosis . Little is known about the role of WAPL in meiosis . Drosophilia wapl mutants exhibit meiotic alterations , specifically in the segregation of nonexchange X chromosomes and the loosening of adhesion between sister chromatids in heterochromatic regions [37] . In budding yeast inactivation of Wpl does not appear to affect spore formation and viability [49] . The chromosomal alterations we observe during meiosis in Atwapl1-1wapl2 plants resemble those caused by depletion of WAPL during mitosis in human cell cultures and flies . Depletion of Wapl in human cell lines blocks the removal of cohesin during prophase resulting in poorly resolved sister chromatids [16] . Likewise , mitotic chromosomes in wapl flies also show prolonged arm cohesion that delay/block the resolution of sister chromatids at anaphase [37] . While yeast wpl/rad61 cells display increased steady-state levels of cohesin , Wpl/Rad61 does not play a critical role in the removal of cohesin complexes during mitotic prophase [11] , [13] . Rather , most mitotic cohesin complexes are removed from yeast chromosomes at anaphase by separase . Interestingly , 65% of Atwapl1-1wapl2 meiocytes contained more than the expected ten centromere signals at metaphase I/anaphase I . This suggests that while the removal of arm cohesin is delayed , centromere cohesion either is not established properly or is prematurely released . The aggregation of centromere sequences we observe during prophase indicate that there are alterations in heterchromatin structure , suggesting that meiotic chromosome centromere cohesion may in fact not form properly in the mutant . This is similar to the situation in Drosophila wapl neuroblasts in which the largely heterochromatic chromosomes 4 and Y display a precocious loss of cohesion , while the other chromosomes maintain arm cohesion and arrest at prometaphase [37] . Finally , Wpl appears to be important for controlling chromosome condensation in budding yeast where inactivation of Wpl results in increased compaction of chromosome arms in S/G2 [49] . Our results show that inactivation of AtWAPL results in the aggregation of heterochromatin regions in particular centromeres . Therefore , WAPL plays a common role in controlling chromosome structure . We show here that AtWAPL mutations suppress the lethality associated with ctf7 mutations in Arabidopsis . This is similar to similar to the situation in yeast [13] , [20] , [21] , [23] , [24] . Inactivation of AtCTF7 results in embryo lethality [45]; however for reasons that are not understood , homozygous Atctf7 mutant plants can be obtained at very low frequencies [46] . Atctf7 plants are dwarf , exhibit severe developmental abnormalities and are completely sterile . They also display mitotic defects , alterations in double strand break repair and the premature dissociation of cohesin from meiotic chromosomes , which leads to the early separation of sister chromatids [46] . Plants triple homozygous for the Atwapl1-1wapl2ctf7-1 mutations display normal vegetative growth and produce small numbers of viable seed . The growth rate and overall size of Atwapl1-1wapl2ctf7-1 plants is similar to that of wild type , indicating that inactivation of AtWAPL suppresses most , if not all of the effects associated with CTF7 inactivation in somatic cells . Furthermore , inactivation of WAPL restores some fertility to Atctf7-1 plants . The overall fertility of Atwapl1-1wapl2ctf7-1 plants is significantly lower than that of Atwapl1-1wapl2 plants but similar to that of Atwapl1-1waplCtf7+/− plants . Therefore , meiotic chromosomes are much more sensitive to the level and distribution of cohesin than somatic cells in plants . Our results indicate that AtWAPL most likely functions during meiosis in a manner similar to that proposed for Wapl in mitotic cells in vertebrates . Prior to DNA replication cohesin has been shown to bind the chromatin in a reversible manner that is normally not able to establish sister chromatid cohesion [13] , [16] , [17] , [23] , [24] , [50] . This reversible binding is controlled , in part , through interactions between Wapl , Pds5 and the cohesin complex . Stable cohesin binding to the chromosomes and the establishment of cohesion , which occurs during DNA replication , involves the inactivation of this Wapl-dependent anti-establishment activity through the Eco1/Ctf7-dependent acetylation of critical lysine residues in SMC3 [20]–[23] , [51] . In animal cells , acetylation of SMC3 facilitates the recruitment of sororin and displacement of Wapl to help create a stable cohesin complex [15] , [18] . A sororin ortholog has not been detected in yeast where SMC3 acetylation appears to directly inactivate the Wpl releasing activity and result in tight binding of cohesin to the chromosomes [22] , [24] , [52] . Most closely related to our work are studies in vertebrate cells that have shown that Wapl is involved in the non-proteolytic removal of cohesin from the arms of mitotic chromosomes as part of the prophase pathway [25] . This process , which involves the mitotic kinases Polo-like kinase ( Plk1 ) and Auora B [17] , [28] , [41] , [53] , [54] , involves opening of the cohesin ring at the junction between SMC3 and the SCC1 WHD [26] , [27] . Plk1 and Auora B have been shown to phosphorylate multiple sites on Sororin , which leads to the disassociation of Sororin from acetylated cohesin complexes [55] . SA2/SCC3 is also phosphorylated by Plk1 [28] , which likely also alters the interaction of Wapl with cohesin . Finally , structural studies on Wapl from fungi and human have generated partial structures of Wapl , which have provided further insights into how Wapl exerts its' anti-maintenance activity and the residues important for interactions between Wapl , Pds5 and cohesin [38] , [39] , [56] . A number of features are shared between the fungal and human Wapl proteins; however , several structural and mechanistic differences were also identified . These structural differences are likely related to the fact that Sororin plays an important role in the Wapl-dependent opening of the cohesin ring in vertebrates but not in yeast . The removal of cohesin from meiotic chromosomes in Arabidopsis involves a prophase step [57] , which we show here is dependent on WAPL . This suggests that the process may also involve the phosphorylation of SCC3 . Further studies are required to test this hypothesis and determine if an Aurora or Polo-like kinase is involved in this process . Likewise , a sororin ortholog does not appear to be present in the Arabidopsis genome , suggesting that acetylation of SMC3 may directly interfere with WAPL binding in plants . However , further experiments are necessary to determine if Arabidopsis SMC3 is actually acetylated by CTF7 and if this affects WAPL binding . Furthermore , while five potential PDS5 orthologs are present in the Arabidopsis genome , a role for the proteins in controlling sister chromatid cohesion has not yet been established . Therefore , additional studies are needed to further characterize the roles of WAPL , PDS5 and CTF7 in plants and further define the specifics of how they control the association of cohesin with chromatin . These studies will help us to better understand the apparent differences in how cohesin interacts with chromosomes in meiotic and somatic cells and determine the specific reason ( s ) meiotic and mitotic plant cells respond so differently to Atwapl mutations .
Arabidopsis thaliana , Columbia ecotype , was used for crossing , transcript analysis and microscopic studies . Plants were grown in Metro-Mix 200 soil ( Scotts-Sierra Horticulture Products; http://www . scotts . com ) or on germination plates ( Murashige and Skoog; Caisson Laboratories; www . caissonlabs . peachhost . com ) in a growth chamber at 22°C with a 16-h-light/8-h-dark cycle . Arabidopsis T-DNA lines were obtained from Arabidopsis Biological Resource Center . Leaves were collected from rosette-stage plants grown on soil and used for DNA isolation and genotyping . Approximately 24 d after germination , buds were collected and staged for microscopy studies . For transcript analysis all samples were harvested , frozen in liquid N2 , and stored at −80°C until needed . A description of the molecular characterization of AtWAPL1 and AtWAPL2 is provided in Text S1 . Sequences of primers used in this study are given in Table S1 . Male meiotic chromosome spreads were performed on floral buds fixed in Carnoy's fixative ( ethanol∶chloroform∶acetic acid: 6∶3∶1 ) and prepared as described previously [58] . Chromosomes were stained with using DAPI and observed under Olympus BX51 epifluorescence microscope system . Images captured using a Spot camera system and processed using Adobe Photoshop . In order to study mitosis Arabidopsis seeds were sterilized and plated on MS agar plates . Seven days after plating the root tips from the seedlings were excised , fixed and prepared as described previously [58] , with the exception that digestion was conducted for 15 min . Immunolocalization studies were performed on 4% paraformaldehyde fixed cells as previously described [59] . Meiotic stages were assigned based on the chromosome structure and morphology as well as the developmental stages of the surrounding anther cells . Primary antibodies ( 1∶500 dilutions ) used in this study ( SYN1 , SMC3 , ASY1 , ZYP1 , β-tubulin ) have been described [43] , [44] , [60] , [61] . The slides were incubated overnight at 4°C , and then washed for 2 h with eight changes of wash buffer . The slides were then incubated overnight with Alexa 488 labeled goat anti-rabbit secondary antibody ( 1∶500 ) or Alexa Fluor 594 labeled goat anti-mouse secondary antibody ( 1∶500 ) overnight at 4°C and again washed and stained with DAPI . FISH was conducted on inflorescences fixed in Carnoy's solution for 1 h at room temperature after replenishing the fixative . FISH was performed on meiotic spreads as previously described [62] , [63] with the following change: samples were treated with a solution of freshly prepared 70% formamide in 2× SSC for 2 min at 80°C and dehydrated through a graded ethanol series ( 70% , 90% , 100% ) of 5 min for each incubation at −20°C . The slides were then dried at room temperature before adding the probe . The 180-bp pericentromeric repeat [64] was amplified , purified , labeled with Roche High Prime fluorescein and was used at a concentration of 5 ugml−1 . Telomere-repeat sequences were detected by hybridization with the 5′-end fluorescein isothiocyanate-labeled oligonucleotide probe , ( CCCTAAA ) 6 at 5 ugml−1 . Slides were counterstained with DAPI and observed under epifluorescence microscope as described above . Total RNA was extracted from stems , buds , roots , leaves and siliques of wild-type plants to examine WAPL expression patterns , and from inflorescences of wild-type , Atwapl1-1wapl2 and Atwapl1-2wapl2 plants to measure WAPL transcript levels in mutant plants . Total RNA was extracted from with the Plant RNeasy Mini kit ( Qiagen , Hilden , Germany ) , treated with Turbo DNase I ( Ambion ) and used for cDNA synthesis with an oligo ( dt ) primer and a First Strand cDNA Synthesis Kit ( Roche ) . Real time PCR was performed with SYBR-Green PCR Mastermix ( Clontech ) and the amplification was monitored on a CFXsystems ( Biorad ) . Expression was normalized against β-Tubulin-2 . At least three biological replicates were performed , with two technical replicates for each sample . Primers used in this study are presented in ( Table S1 ) . Whole-mount clearing was used to determine the embryo phenotypes [65] , [66] . Sliques from wild-type and mutant plants were dissected and cleared in Hoyer's solution containing lactic acid∶chloral hydrate∶phenol∶clove oil∶xylene ( 2∶2∶2∶2∶1 , w/w ) . Embryo development was studied microscopically with a Olympus BX51 microscope equipped with differential interference contrast optics . Female gametophyte analysis was performed as described in [67] . Whole anther morphology was analyzed by staining with Alexander staining [68] .
|
Wapl has been shown to play an integral role in the removal of cohesin from chromosomes during mitotic prophase . While Wapl's role appears to be conserved between yeast , fly and animal cells , structural and possible mechanistic differences have also been identified . As part of a study to better understand the protein and its role ( s ) we have characterized Wapl in plants . We show that Arabidopsis contains two copies of WAPL that share overlapping functions . Inactivation of the individual genes has no effect . Plants containing mutations in both genes growth normally but exhibit reduced fertility due to alterations in meiosis . Cohesin removal from chromosomes during meiotic prophase is blocked in wapl mutant plants resulting in unresolved bivalents and uneven chromosome segregation . In contrast , cohesin appears to be removed normally in mitotic cells . These results demonstrate that WAPL plays a critical role in removing cohesin from meiotic chromosomes . They also suggest that the mechanism involved in prophase removal of cohesin may vary between mitosis and meiosis in plants . Finally , wapl mutations suppress ctf7-associated lethality and restore normal growth and partial fertility to ctf7 mutant plants , suggesting that sister chromatid cohesion is not essential for plant growth and development .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"meiosis",
"plant",
"growth",
"and",
"development",
"cell",
"cycle",
"and",
"cell",
"division",
"plant",
"cell",
"biology",
"cell",
"processes",
"brassica",
"cytogenetic",
"analysis",
"developmental",
"biology",
"plant",
"science",
"model",
"organisms",
"plants",
"arabidopsis",
"thaliana",
"research",
"and",
"analysis",
"methods",
"chromosome",
"biology",
"plant",
"genetics",
"plant",
"and",
"algal",
"models",
"cell",
"biology",
"genetics",
"biology",
"and",
"life",
"sciences",
"molecular",
"cell",
"biology",
"cytogenetics",
"organisms"
] |
2014
|
Arabidopsis thaliana WAPL Is Essential for the Prophase Removal of Cohesin during Meiosis
|
We develop a quantitative single cell-based mathematical model for multi-cellular tumor spheroids ( MCTS ) of SK-MES-1 cells , a non-small cell lung cancer ( NSCLC ) cell line , growing under various nutrient conditions: we confront the simulations performed with this model with data on the growth kinetics and spatial labeling patterns for cell proliferation , extracellular matrix ( ECM ) , cell distribution and cell death . We start with a simple model capturing part of the experimental observations . We then show , by performing a sensitivity analysis at each development stage of the model that its complexity needs to be stepwise increased to account for further experimental growth conditions . We thus ultimately arrive at a model that mimics the MCTS growth under multiple conditions to a great extent . Interestingly , the final model , is a minimal model capable of explaining all data simultaneously in the sense , that the number of mechanisms it contains is sufficient to explain the data and missing out any of its mechanisms did not permit fit between all data and the model within physiological parameter ranges . Nevertheless , compared to earlier models it is quite complex i . e . , it includes a wide range of mechanisms discussed in biological literature . In this model , the cells lacking oxygen switch from aerobe to anaerobe glycolysis and produce lactate . Too high concentrations of lactate or too low concentrations of ATP promote cell death . Only if the extracellular matrix density overcomes a certain threshold , cells are able to enter the cell cycle . Dying cells produce a diffusive growth inhibitor . Missing out the spatial information would not permit to infer the mechanisms at work . Our findings suggest that this iterative data integration together with intermediate model sensitivity analysis at each model development stage , provide a promising strategy to infer predictive yet minimal ( in the above sense ) quantitative models of tumor growth , as prospectively of other tissue organization processes . Importantly , calibrating the model with two nutriment-rich growth conditions , the outcome for two nutriment-poor growth conditions could be predicted . As the final model is however quite complex , incorporating many mechanisms , space , time , and stochastic processes , parameter identification is a challenge . This calls for more efficient strategies of imaging and image analysis , as well as of parameter identification in stochastic agent-based simulations .
In early development , tumors grow up to 1–2mm in diameter , nourished by the nutrients and oxygen provided by the existing vasculature . Either 2D or 3D cell culture systems are utilized as biological models to study that phase , or aspects usually occurring in later phases of tumor growth and development . Current 2D cell culture approaches are only of limited use to investigate tumor progression in these stages , as they neglect crucial histo-morphological and functional features of these avascular micro-metastases or inter-capillary micro-regions of solid in vivo tumors . During the last decades , great effort has been undertaken to generate biological 3D models that describe the early phases of tumor development in a tissue context more accurately . They can thus serve as intermediate systems between traditional 2D cell culture and complex in vivo models ( [3 , 4] ) . Of these approaches , Multicellular Tumor Spheroids ( MCTS ) offer easy handling and fast generation , even for larger batches , and automation ( [5 , 6] ) . MCTS as a model system can be well characterized and have been shown to reproduce the spatial organization and micro-environmental factors of in vivo micro tumors , such as relevant gradients of nutrients and other molecular agents and deposition of Extracellular Matrix ( ECM ) ( see Fig 1 ) ( [7 , 8] ) . Furthermore , gene expression studies revealed substantial differences in both the baseline profiles and profiles after stimulation between 2D and MCTS cultures . The latter is decisively closer to patients profile gene expression ( [9–11] ) . Consequently , MCTS have now been established as experimental systems for both basic research and high throughput screening of clinically relevant drugs ( reviewed by: [12 , 13] ) . As already explained in [14] the organization of the different cell phenotypes within a MCTS ( growing , quiescent and dead ) is supposed to be radial , and to be controlled by different factors: growth promoters ( GP ) , viability promoters ( VP ) , growth inhibitors ( GI ) and viability inhibitors ( VI ) . In the case of spheroids the promoters are mainly delivered from the growth medium surrounding the tumor ( with exception of ECM ) while inhibitors are generally assumed to be produced by the tumor itself . As a consequence , the local composition and interplay of those factors favor different transitions between cell phenotypes at different distances from the tumor border . To understand the dynamics of avascular tumor growth , several mathematical models were proposed linking the growth kinetics on the multi-cellular level ( radius/volume in time ) with mechanisms on cell or subcellular scales ( cell growth , contact-inhibition , nutrient limitation etc . ) . They can be classified in two main approaches: ( 1 ) continuum models of the different components densities that evolve in time and space following PDEs ( see e . g . [15–18] ) , ( 2 ) agent-based models that describe each cell individually and how it grows , divides , moves and dies ( see e . g . [19–21] ) . When cells are modeled as agents , oxygen , nutrients and/or growth factors or inhibitors are often modeled by continuum models [1 , 2 , 22–31] or with simplified assumed profiles [32 , 33] . Hybrid models on the other hand combine within the same framework the two model types for the cells , depending on the tumor zone ( e . g . [26] ) . Despite the large variety of models , identification of a plausible mechanistic model and its quantitative parameterization able to quantitatively explain a large set of data and predict the outcome of experiments that were not used to calibrate the model remains a difficult task . Issues are the large number of parameters and the lack of validation of the underlying mechanisms . Different models have successfully been fitted to the growth dynamics of cell populations ( e . g . [2 , 22] consider only the population size but not the diameter , [1 , 34] consider both ) relying on different mechanisms but leading to the same growth curves . For example the transition from exponential to linear radius growth phase can be due to contact-inhibition or nutrient-limitation . Based on the growth curves alone , model selection could not be made , indicating that development of a mathematical model only relying on growth curves is insufficient . In this paper we pursue a quantitative image-based approach based on bright field micrographs . This is in-line with a recent trend in large-scale simulations of brain tumors based on magnetic resonance imaging ( MRI ) [35 , 36] following inspiring work by Swanson and co-workers [37] . Histological information has been used recently by Frieboes et . al . [38] , who included histological staining measurements in a partial-differential-equation tissue model of Non-Hodgkin lymphoma growth , and by Macklin et . al . [39–41] who developed a multiscale model , mimicking cells as individual agents subject to forces , to predict ductal carcinoma growth in individual patients , including histological information . In this paper we study mathematical model development and model parameterization by comparison with experimental data for different oxygen and glucose concentrations in the medium for the non-small cell lung cancer ( NSCLC ) cell line SK-MES-1 . These data consist in the growth kinetics and the corresponding spatial staining patterns for nuclei , different cell states and cell environment , namely , HOECHST for cell nuclei , KI67 for proliferation , TUNEL for dead cells , collagen IV for ECM . We study in how far mechanisms that have not been directly assessed can be inferred by simultaneous matching of simulation results with experimental results on many experimental observables . Our strategy is stepwise: we first develop a model for one growth condition only , and then expand the model to capture additional growth conditions after verifying that the previous ( simpler ) model stage was incapable of explaining the added growth data . For this , we perform many computer simulations with the “previous” model varying each model parameter within its physiological range . Finally we arrive at a model that can almost completely be projected to the experimentally derived scheme on spheroid growth ( compare Fig 1 to reference [8] ) . By such a stepwise strategy involving experiments , imaging , image analysis and modeling , an order mechanism during liver regeneration could be identified [42] , indicating that such a strategy may be powerful in unveiling interplaying mechanisms in multi-cellular organization .
In order to study the influence of environmental conditions on the growth dynamics of tumor spheroids ( Fig 2 ) , SK-MES-1 cells were cultivated in-vitro as multi-cellular tumor spheroids under different nutriment conditions , with the hanging drop method . Then , at different points in time , spheroid size was determined with bright field microscopy and some of the spheroids were frozen , cryosectioned , stained and imaged with fluorescence microscopy ( Fig 3 ) . For a detailed description see the Materials and Methods section . Our objective is to explain the experimentally observed growth pattern for different glucose and oxygen medium concentrations within one mathematical model . For this purpose we first searched for a minimal model ( in the sense specified in the abstract ) explaining the experimental tumor growth observations for medium concentrations of [G] = 25mM and [O] = 0 . 28mM , and then stepwise extended this model to capture the other growth conditions ( for illustration of the stepwise model development strategy , see Fig S3 in S1 Document ) . We based our choice of possible control mechanisms upon prior knowledge guided by published information and own experiments . We have chosen the condition of maximum glucose and oxygen medium concentration as the similarity of the growth kinetics for [G] = 25mM and [O] = 28mM versus [G] = 5mM and [O] = 28mM suggests that for the former condition neither glucose nor oxygen may be limiting ( see also Discussion below ) . This line of argument is supported by findings of Freyer and Sutherland for another cell type at almost the same oxygen and glucose medium conditions ( compare with [1] ) . We fit at each model development step all parameters again . So the fits shown in this article were the best we could obtain for the respective model . However , due to the large search space and duration of simulation of at least one day ( reference computer: Intel ( R ) Xeon ( R ) processor X5680 3 . 33GHz 12M cache 6-core and 144 GB DDR3-RAM 1333 MHz ) it cannot completely be excluded that further parameter searches could give additional slight improvements . In order to promote readability , we enumerate the model at each development level . Usually we performed for each parameter set only a single simulation i . e . , a single realization of the stochastic growth process . This can be justified by observing that the growth process , that starts with about 10000 cells as in the experiment , is self-averaging such that the variations for different realizations of the growth process for the same parameters are negligible ( Fig S11 in S1 Document ) . Our basic model considers each cell individually within an agent-based model i . e . each individual cell is represented by an agent . Molecules , which finally are glucose , oxygen and lactate , as well as extra-cellular matrix and “waste” material released by dying cells , are represented by their local concentrations . We use the term “molecules” in what follows to generically describe these environmental factors that affect the cells . The model is three-dimensional . In the following sections we will introduce briefly the main model components . A detailed description of the model as well as the biological processes mimicked can be found in the material and method section and in the supporting information ( S1 Document ) . A number of parameters related uniquely to cells or molecules were either taken or estimated from literature ( e . g . molecular diffusion coefficients , consumption rates , cell cycle time distribution ) . Others could be inferred from the data presented in this article either directly ( e . g . initial conditions , cell size ) or by sensitivity analysis ( e . g . cellular division , cell death and lysis rates ) ( see Table S1 in S1 Document ) . The identification of the main mechanisms coupling the cellular and molecular kinetics and their parameterization ( see Fig 5 ) was subject to the model comparison with data explained below . The labeling patterns in Figs 3 and 4 confirm a border distance-dependent “zonation” as discussed in the introduction ( compare also [8] and Fig 1 ) . In the following we will study the influence of cell-cell-contacts , extra-cellular matrix and metabolic compounds ( nutrients/metabolites ) to infer the corresponding model mechanisms . For the latter ( metabolism ) we will compare four different hypotheses ( model 1–4 ) . A summary of the equations used to calculate the transition rates and probabilities depicted in Fig 5 for all models studied in the following sections can be found in Table 2 .
In this paper we inferred a mathematical model of tumor spheroid growth for the non-small cell lung cancer cell line SK-MES-1 from image data of growing tumor spheroids . Cell nuclei , proliferating cells , extra-cellular matrix and dying cells ( by either necrosis or apoptosis ) were labeled at different points in time and under different oxygen and glucose medium concentrations . The model was built by an iterative procedure , which we propose as a general template for modeling tissue organization processes . We started by developing a minimal model for one growth condition only , then stepwise extending this model by further mechanisms whenever the previous simpler model turned out to be insufficient to reproduce the experimental observations for an additional growth condition . Before adding a new mechanism to an existing model version we verified by extensive computer simulations ( usually hundreds of runs ) , that within the parameter range for each parameter of the existing model no satisfying agreement between model and data could be achieved . Minimal is here to be understood as sufficient to explain the data and containing as least mechanisms as possible , whereby the building blocks of the model were chosen from those mechanisms that have already been described somewhere for any cell population . A similar iterative strategy was pursued for liver regeneration after drug induced damage predicting a previously unrecognized and subsequently validated order mechanisms [65] . We studied four different combinations of glucose and oxygen in the medium . To explain the growth kinetics , the proliferation , ECM , and cell death for the condition with high glucose and high oxygen medium concentration ( [G] = 25mM , [O] = 0 . 28mM ) , the second with intermediate concentration of glucose and high oxygen concentration ( [G] = 5mM , [O] = 0 . 28mM ) , we needed to assume that the cell cycle progression is possible only above a critical local production rate of ATP ( = mM/h ) . A second necessary condition was , that the local density of extra-cellular matrix had to be higher than a critical value ( 0 . 003 ) . This is in accordance to literature , where dependence of cancer progression on the ECM has been shown for skin cancer [66] , breast cancer [67] and NSCLC [68] , where Collagen IV can regulate crucial cell signaling . If both conditions ( enough ATP and ECM ) were fulfilled , cells could reenter the cell cycle after a cell division . Here , cells , which were closer to the spheroid surface and thus needed less energy in order to expand , had an increasing chance to continue proliferation and not to become quiescent . Interestingly the decision whether a cell in a certain condition became quiescent , had to be stochastic . This introduced some heterogeneity in subsets of cells in the same conditions . A deterministic scenario could not have explained the smooth transition from proliferating to quiescent zones . The production rate of ATP depended on the local oxygen and glucose concentrations . Thereby , the ratio between both dictates to which extent a cell is in the aerobic Krebs cycle or the anaerobic lactate fermentation . Warburg stated in [69] that all cancer cells suffer from an injured respiration and thus have an exclusively anaerobic metabolism . In opposition , Zu and Guppy [70] disproved this hypothesis due to the lack of evidence and rather claimed the metabolism in cancer cells to be functional , but mainly glycolytic due to hypoxia . Here we come to a partially different conclusion: if cells are sufficiently supplied with glucose ( independent from the oxygen supply ) , the metabolism will remain glycolytic ( 90% ) , and only if the glucose supply is getting short , the metabolism will favor the aerobic Krebs cycle ( see Fig 8 ) . Besides lactate acidity ( > 20mM ) , the depletion of carbon sources to maintain a critical ATP production ( = 900mM/h ) and not hypoxia were the main reasons of death . The latter was also recently suggested by Kasinskas et al . [71] , while , in contrast to our assumptions , they excluded lactate as source of acidity and instead assumed it to be an important secondary metabolic resource . However , either growth adverse or death promoting effects were described for high lactate concentrations [59] . So here further clarification of the dominating role of lactate would be necessary . The functional forms of the oxygen and glucose consumption rates were inferred from experimental findings of Freyer , Sutherland and co-workers in EMT6/Ro cells . The lactate and ATP production rates were then directly derived from those rates by the single assumption that cells transform the consumed glucose in an optimal way with respect to ATP output . For wide ranges of glucose and oxygen concentrations the ATP production rates remain stable between 80…130 × 10−17 mol/cell/s or 1000…1700mM/h respectively , assuming a reference cell volume of 2700μm . In literature values can be found between 4 . 6…15 . 3 × 10−17 mol/cell/s ( [72–75] ) . The difference could be either due to differences in energy needs between different cell types , or to the model simplification that glucose in our model is exclusively used for metabolism . Interestingly and importantly , the model , despite only having been calibrated with two of the four growth conditions , were subsequently able to correctly and quantitatively predict the growth phase of the other two growth conditions ( [G] = 1mM , [O] = 0 . 28mM and [G] = 25mM , [O] = 0 . 07mM , respectively ) . This indicates that the model did capture the functionalities necessary to explain the data for different glucose and oxygen conditions . To further permit independent validation of our model , we performed additional simulations for other glucose and oxygen medium concentrations ( Fig S12 in S1 Document ) . However , all growth curves showed saturation and partially even shrinkage after some time . The saturation phase could be largely captured by adding the potential effect of a waste produce being released in the extracellular space from cells undergoing lysis . Shrinkage could be added if dying cells at the border detach and enter the growth medium; however , we did not consider this process , as it was not observed in the experiments ( for example , for A549 cells , another NSLC cell line , a massive detachment of cells from the spheroid could be observed in the experiments ) . Interestingly , model simulations with a lysis rate of 0 . 35/h , a typical value in-vivo , turned out to be incompatible with the in-vitro data . A lysis rate of a few hours as observed in-vivo would lead to a very fast removal of dying cells and thus almost no dead cells in the tumor center , in sharp contast to the in-vitro experiments . We obtained a much smaller value of about 0 . 01/h by comparison of model simulation results and the spatial cell death and proliferation profiles i . e . , only such a small lysis rate permits the occurrence of a “necrotic core” as observed in the in-vitro experiments . For such a low lysis rate we found that the apoptosis—if present—would need to be very slow , as it affects also cells in the viable rim in order to agree with the experimental observation of only very little dead cells in the viable rim . For this reason , apoptosis could be neglected in explaining the experimental results in this paper . The small value of the lysis rate , even though surprising on a first view , may be explained by noticing that stromal cells ( such as e . g . macrophages ) digesting dead cells are not present in-vitro . Hence lysis might be expected to be slower in-vitro than in-vivo . Contact inhibition seems to be a crucial element . Suppressing contact inhibition with varying combinations of the other mechanisms in each case leads to complete failure of match between data and model simulations ( see Fig S6 in S1 Document , where the parameters of model 4 has been used ) . This observation supports the view expressed previously in the paper that a mechanical growth inhibition plays an important role in multicellular spheroids . We moreover tested the possibility that cells may actively migrate towards the necrotic zone by necrotaxis ( Figs S9 , S10 in S1 Document ) . As to keep a sufficiently large necrotic core as experimentally observed the lysis rate had to be small , significant migration could not be observed . On the other hand , if the lysis rate was chosen large , then significant migration of cells could be observed but the necrotic core was too small , as cells in the center were too quickly eliminated by lysis . In the latter case , the necrotic core with increasing migration rate became smaller ( Figs S9 , S10 in S1 Document ) . We concluded that migration driven by morphogens towards the central necrosis in SK-MES-1 cells is small . Interestingly the final model emerging from this stepwise , image-guided inference strategy closely resembles the hypothesis on growth control of MCTS by growth promoters ( GP ) , growth inhibitors ( GI ) , viability promoters ( VP ) and inhibitors ( VI ) ( Fig 11 ) . In order to permit validation of our model , we simulated a number of predictions ( Fig S13 in S1 Document ) . We predicted the spatial temporal growth dynamics for [G] = 1mM , [G] = 3mM , O2 = [0 . 28mM] , and [G] = 25mM , [O] = 0 . 07mM . In this context we would like to remind that our model was able to predict the growth kinetics ( L ( t ) ) for [G] = 1mM , [O2] = 0 . 28mM , and [G] = 25mM , [O2] = 0 . 08mM correctly . In order to quantify the goodness of the fits shown in Figs 7 , 9 and 10 we calculated the log-likelihoods for subsets of curves ( see Table 3A ) as well as the ensemble of all curves ( see Table 3B ) . We assumed that the measurement error is additive , normally distributed as well as independent and identically distributed ( i . i . d . ) . Accordingly , the likelihood of the measured data mean μ given the parameter θ and the corresponding model prediction x ( θ ) is given by L ( θ ) = ∏ i 1 2 π σ i 2 e x p ( ( x ( θ ) i - μ i ) 2 2 σ i 2 ) , where the index i runs over the data points . The uncertainty of the data points is determined by the standard deviation , σ . Accordingly , points with large uncertainties σ are weighted less . Despite some deviations in single profiles , the final model considering ATP , lactate and waste is found to be the most likely to explain the experimental data . On the other hand , an increase of the likelihood correlates with the increasing number of model parameters and the risk of over-fitting . Especially model 2 and 3 have a small relative difference in log-likelihood . As a measure for model quality accounting for the number of parameters we used the Akaike information criterion ( AIC ) ( see Table 3D ) , which also confirms the final model to be the best choice . We note that the number of parameters had no influence on the ordering of the models as the absolute differences of their log-likelihoods ΔlnL is many orders of magnitude larger than the difference in number of parameters Δk ( see Table 3B and 3C ) . To conclude , we would like to stress the key message demonstrated in this paper: quantitative comparison of spatial profiles observed from cells states in different experimental conditions and time-points , generates information so rich that one may infer even molecular control mechanisms and parameters of spatio-temporal growth and death patterns . Hence , careful imaging , image processing and image analysis may serve as an important source of information to infer mechanistic knowledge on tissue growth and organization processes . Such an approach would gain to be more fully explored . Our model is hybrid . It integrates as separate components the cell and molecules , and as functional components a mechanical form of contact inhibition , a metabolic component comprising oxygen , glucose , ATP , lactate , and waste , several of the molecules acting as morphogens . It would be interesting to see in how far the same model can capture the growth behavior of other cell lines and of other cell types . We think that the possible imaging techniques and image analysis software in combination with modeling could permit a screening of growth dynamics and subsequent quantitative classification of multicellular spheroids . For example , EMT6/Ro cells ( [14] ) show a very similar but not equal growth phenotype as SK-MES-1 cells: ( 1 ) detachment of cells is rare ( as opposed to , for example , A549 cells , that reveal significant detachment in-vitro ( and in-vivo ) ) , ( 2 ) under sufficient oxygen supply , the growth of the outer spheroid diameter remains unaffected ( or almost unaffected ) by glucose , while ( 3 ) reduction of oxygen from 0 . 28mM to 0 . 07mM reduces growth dramatically in SK-MES-1 cells even if glucose medium stays high , while in EMT6/Ro cells no reduction is observed as long as nutrient medium concentration stays high: this demarcates a difference between the EMT6/Ro and SK-MES-1 cells . On the other hand , glucose affects the size of the necrotic core . Reduction from 16 . 5mM glucose to 0 . 8mM glucose medium concentration in EMT6/Ro cells ( at 0 . 28mM oxygen ) increases the necrotic core ( as one can infer from comparing the cell count with the diameter , see [34] ) , which can also be observed in SK-MES-1 cells if glucose is reduced from 25mM to 5mM ( at 0 . 28mM O2 ) . However , in SK-MES-1 cells the necrotic core for richer glucose ( [G] = 25mM vs . 5mM ) occurs later but at about 24 days is about the same size . Models can provide a quantitative framework to test how far such differences can be attributed to parameters with the same model , or whether “another” model needs to be used by adding or dropping mechanisms . For example in the first case , can the same model be used to capture a wide range of cell lines with regard to their MCS growth behavior by only adjusting its parameters—indicating only quantitative changes , or , in the 2nd case , does one need to implement mechanisms for one cell type that are not observed within the physiological range of parameters for another cell type ? Given how much multicellular spheroids are still in use as biological model system , we think it would be of fundamental interest to do such an analysis as a community effort , even though this might be considered as on the first view as a “step-back” as the growth dynamics of multi-cellular spheroids could have been measured 20—30 years ago . Modern technology could largely permit automated analysis if pipelines were constructed for that purpose , hence avoiding the largely manual and tedious analysis applied for the work in this paper . Adding more and more cell lines would permit to refine the model one starts with , and zoom into the so far still highly simplified representation of mechanisms without the threat of having a far to large number of fit parameters that cannot be controlled . In this way , identification of necessary model components and adjustment of parameters linking the components could be achieved .
NSCLC cell line SK-MES-1 used in this study was obtained from ATCC ( Manassas , VA , USA ) and cultivated in a humidity controlled incubator at 37 C and 5% CO2 in 150cm2 tissue culture dishes ( TPP ) in DMEM ( Dulbecco’s modified Eagle’s medium , LONZA , Verviers , Belgium ) supplemented with 10% FCS ( fetal calf serum , Southern America , GIBCO , Germany ) and 1% Penicillin/Streptomycin ( Biochrom AG , Berlin , Germany ) . Cells were used between passages 10 and 30 and passaged at a split ratio of 1:4 to 1:6 . Cultures were routinely tested for mycoplasm contamination as described by Stacey and Doyle 1997 and always found to be negative . Additional medium for the test cultures was DMEM w/o Glucose ( GIBCO , Germany ) supplemented with 10%FCS and 1mM Glucose ( Carl Roth GmbH , Germany ) , and DMEM with 1 . 0 g/L glucose w/o L-Glutamine ( LONZA , Verviers , Belgium ) supplemented with 10% FCS and 25mM L-Glutamin ( SIGMA , Germany ) . Additionally , cells were kept in a humidity controlled incubator at 37 C and 5% CO2 and either normal atmospheric 20% O2 ( corresponding to 0 . 28mM ) or 5% O2 ( corresponding to 0 . 07mM ) . This choice permits comparison to classical work in literature ( e . g . [12] ) and takes into account that lung is rich in oxygen . NSLC cells originate from lung epithelium having at least partially direct contact to the inhaled air so are at least initially not limited to the blood oxygen level . The images acquired ( see above ) are raw data . They consist of a set of pixels with position ( x , y ) ∈ N 2 and color intensities for the three color channels red , green and blue defined by I c h a n n e l [ x , y ] ∈ [ 0 , 1 ) , I c h a n n e l : N 2 → R , ( 12 ) where channel ∈ {red , green , blue} . In the following , the tools used to preprocess the raw images ( e . g . to reduce noise ) and to identify or segment objects ( e . g . cell nuclei , spheroid border ) will be introduced . Then the preprocessed images have been analyzed quantitatively as described subsequently . The cellular automaton model , which extends on previous work ( [43 , 47] ) , is defined by a set of rules: Glucose and oxygen are among the main metabolites of most biological cells . In normal tissue they are provided mainly by the vascularization . Tumor spheroids as avascular tumors ( in-vivo ) are mainly fed by nutrients diffusing into the interior from the border . Above a certain size they display regions lacking glucose ( hypo-nutrition ) and/or oxygen ( hypoxia ) . This occurs if the nutrients entering the tumor via its borders are consumed completely before reaching the center . We modeled the diffusion and uptake of glucose and oxygen by a reaction-diffusion equation: ∂ t u = ∇ · ( D u ( σ ) ∇ u ) + r u ( σ , u ) , ( 19 ) where Du is the molecule diffusion coefficient and ru ( σ , u ) denotes the reaction term , u ∈ {G , O} . The diffusion coefficient was chosen differently in the nutrient medium than inside the tumor spheroid . The reaction term mimicked the cells consumption with the choice r u ( σ , u ) = - q u ( u ) σ = - q u ( u ) ∑ k δ ( r - r k ) . ( 20 ) The cellular dynamics was modeled by a master equation describing the time evolution of the probability of the multi-cellular configuration . The molecular dynamics was modeled by a system of deterministic partial differential equations .
|
We here present how to parameterize a mathematical agent-based model of growing MCTS almost completely from experimental data . MCTS show a similar establishment of pathophysiological gradients and concentric arrangement of heterogeneous cell populations as found in avascular tumor nodules . We build a process chain of imaging , image processing and analysis , and mathematical modeling . In this model , each individual cell is represented by an agent populating one site of a three dimensional un-structured lattice . The spatio-temporal multi-cellular behavior , including migration , growth , division , death of each cell , is considered by a stochastic process , simulated numerically by the Gillespie algorithm . Processes on the molecular scale are described by deterministic partial differential equations for molecular concentrations , coupled to intracellular and cellular decision processes . The parameters of the multi-scale model are inferred from comparisons to the growth kinetics and from image analysis of spheroid cryosections stained for cell death , proliferation and collagen IV . Our final model assumes ATP to be the critical resource that cells try to keep constant over a wide range of oxygen and glucose medium concentrations , by switching between aerobic and anaerobic metabolism . Besides ATP , lactate is shown to be a possible explanation for the control of the necrotic core size . Direct confrontation of the model simulation results with image data on the spatial profiles of cell proliferation , ECM distribution and cell death , indicates that in addition , the effects of ECM and waste factors have to be added to explain the data . Hence the model is a tool to identify likely mechanisms at work that may subsequently be studied experimentally , proposing a model-guided experimental strategy .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"carbohydrate",
"metabolism",
"cell",
"death",
"chemical",
"compounds",
"oxygen",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"carbohydrates",
"glucose",
"metabolism",
"organic",
"compounds",
"glucose",
"oxygen",
"metabolism",
"cellular",
"structures",
"and",
"organelles",
"extracellular",
"matrix",
"chemistry",
"biochemistry",
"cell",
"biology",
"organic",
"chemistry",
"apoptosis",
"monosaccharides",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"metabolism",
"chemical",
"elements"
] |
2016
|
Inferring Growth Control Mechanisms in Growing Multi-cellular Spheroids of NSCLC Cells from Spatial-Temporal Image Data
|
Although Plasmodium vivax infection is a frequent cause of malaria worldwide , severe presentations have been more regularly described only in recent years . In this setting , despite clinical descriptions of multi-organ involvement , data associating it with kidney dysfunction are relatively scarce . Here , renal dysfunction is retrospectively analyzed in a large cohort of vivax malaria patients with an attempt to dissect its association with disease severity and mortality , and to determine the role of inflammation in its progression . A retrospective analysis of a databank containing 572 individuals from the Brazilian Amazon , including 179 patients with P . vivax monoinfection ( 161 symptomatic malaria , 12 severe non-lethal malaria , and 6 severe lethal disease ) and 165 healthy controls , was performed . Data on levels of cytokines , chemokines , C-reactive protein ( CRP ) , fibrinogen , creatinine , hepatic enzymes , bilirubin levels , free heme , and haptoglobin were analyzed to depict and compare profiles from patients per creatinine levels . Elevated creatinine levels were found predominantly in women . Vivax malaria severity was highly associated with abnormal creatinine increases , and nonsurvivors presented the highest values of serum creatinine . Indication of kidney dysfunction was not associated with parasitemia levels . IFN-γ/IL-10 ratio and CRP values marked the immune biosignature of vivax malaria patients , and could distinguish subjects with elevated creatinine levels who did not survive from those who did . Patients with elevated serum creatinine or severe vivax malaria displayed indication of cholestasis . Biomarkers of hemolysis did not follow increases in serum creatinine . These findings reinforce the hypothesis that renal dysfunction is a key component in P . vivax malaria associated with clinical severity and mortality , possibly through intense inflammation and immune imbalance . Our study argues for systematic evaluation of kidney function as part of the clinical assessment in vivax malaria patients , and warrants additional studies in experimental models for further mechanism investigations .
Malaria is an infectious disease known for millennia . Although substantial investments have been made in the last couple of decades , with slight improvements regarding control of disease transmission in some endemic territories , the disease is still responsible for over 200 million cases annually , with almost 500 , 000 deaths only in 2015 , especially among children [1] . Plasmodium vivax has a wide geographical distribution and was historically associated with milder disease presentations , while Plasmodium falciparum has been commonly related with increased severity and mortality [1–3] . Nevertheless , severe cases caused by Plasmodium vivax monoinfection have been frequently reported in recent years [1 , 4] and increasing interest resulted in several studies focused on the details of the vivax malaria clinical characteristics and pathogenesis [5–8] . Severe P . vivax cases have been associated with impaired immune response and inflammation-driven hepatic damage [5 , 9] , as well as kidney involvement , among other symptoms [5 , 10–14] . While the mechanisms driving inflammatory damage in some key organs such as liver are quite well described in vivax malaria patients [4–6 , 9] , scarce data are available with regard to the mechanisms of kidney injury during acute and severe disease [10] . In individuals with P . falciparum infection , acute kidney injury ( AKI ) has been linked to acute tubular necrosis ( ATN ) [4 , 15] , associated with altered cytoadhesion of infected erythrocytes [16–20] . Whether these mechanisms are common in P . vivax-related AKI or whether there are unique features that distinguishes the kidney damage caused by infection with different Plasmodium species is unknown . In some reports , vivax malaria patients have been described to exhibit oliguria or anuria [10–13] , and in those cases , hemodialysis treatment has been employed as the kidney failure had already established . The use of a widely available and easily measured laboratory parameter that could faithfully readout kidney damage during acute malaria has potential to optimize clinical management . In the present study , the relevance of serum creatinine levels in evaluation of patients with P . vivax malaria was tested and associated with the systemic inflammatory profile and clinical outcomes . These analyzes were aimed to examine how much of the parasite-host relationships , and especially systemic inflammation , would play a role in the settlement of vivax malaria-associated AKI .
Written informed consent was obtained from all participants or their legally responsible guardians , and all clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki . The project was approved by the institutional review board of the Faculdade de Medicina , Faculdade São Lucas , Rondônia , Brazil , where the study was performed . The present study is a retrospective assessment of a databank containing epidemiological , immunological , and clinical data from 572 subjects recruited between 2006 and 2007 from the Brazilian Amazon ( Rondônia , Brazil ) , as a part of a finalized project that described determinants of susceptibility to vivax malaria and from which several investigations have been already reported [5 , 9 , 21–28] . The details of the recruitment , diagnosis and clinical definitions were published previously [5 , 21–27] . The primary investigations included passive case detection , from individuals who sought care at the municipal hospital in Buritis ( Rondônia , Brazil ) or at Brazilian National Foundation of Health ( FUNASA ) diagnostic centers . In addition , active case detection was performed and included home visits with interviews , clinical evaluations and specimens sampling for laboratory assays in the municipalities of Buritis and Demarcação ( Rondônia , Brazil ) . Blood collection for nested polymerase chain reaction ( PCR ) and other measurements ( cytokines , chemokynes , hepatic enzymes , creatinine , fibrinogen , bilirubin levels , free heme and haptoglobin ) were performed at the time of study enrollment , meaning that specimens were collected at diagnosis , during acute phase of disease in malaria patients , before treatment initiation . For the present study , only patients with P . vivax monoinfection ( n = 179 ) and healthy controls ( n = 165 , from which 152 had all the epidemiological data available ) were included . The exclusion criteria for the present study were: asymptomatic P . vivax monoinfection , documented P . falciparum or HIV infections , tuberculosis , cancer , or use of immunosuppressant drugs . Noteworthy , as there were no consensus to define severe vivax malaria , we adapted the criteria used to define severe disease caused by P . falciparum infection , as published previously by our group [5] . Clinical , demographic and epidemiological characteristics of the participants included in the current study are described in Table 1 , S1 Table and S2 Table . Data on several mediators were selected in order to assess the overall immune response; and all the biomarkers contained in the databank were analyzed . Plasma levels of cytokines IL-1β , IL-4 , IL-6 , IL-10 , IL-12p70 , IFN-γ , tumor necrosis factor ( TNF ) -α , and of chemokines CCL2 , CCL5 , CXCL9 , and CXCL10 were measured using the Cytometric Bead Array—CBA ( BD Biosciences Pharmingen , San Diego , CA , USA ) , according to the manufacturer’s protocol . Circulating concentrations of total heme in plasma samples were estimated using the QuantiChrom Heme Assay Kit ( BioAssay Systems , Hayward , CA ) , according to the manufacturer’s protocol . The following procedure was performed to distinguish heme bound to hemoglobin vs . non-hemoglobin heme , i . e . free heme [27] . Plasma absorbance was measured in the UV-VIS range at λ = 576nm ( peak of absorption for hemoglobin ) and λ = 630nm using a using NanoDrop spectrophotometer ( NanoDrop Technologies ) . The following formula: [hemoglobin-heme] = ( 66 x λ = 576nm ) - ( 80 x λ = 630nm ) was applied and referred to a calibration curve of purified hemoglobin ranging from 320–5 μM hemoglobin-heme equivalents . Free heme was calculated according to the following formula: [Free heme] = [Total heme]-[hemoglobin-heme] . The measurements of aspartate amino-transferase ( AST ) , alanine amino-transferase ( ALT ) , total bilirubin , direct bilirubin , creatinine , fibrinogen and C-reactive protein ( CRP ) were performed at the Pharmacy School of the Federal University of Bahia and at the clinical laboratory of Faculdade São Lucas . Serum creatinine levels from the 179 malaria vivax patients were considered as elevated or normal considering thresholds established by gender , as previously described [29] . Elevated creatinine levels were defined in female patients as values above 1 . 24mg/dL and in male individuals as higher than 1 . 29mg/dL [29] . Elevated total bilirubin levels were defined as higher than 1 . 5mg/dL whereas high indirect or direct bilirubin levels infer concentrations above 1 . 2mg/dL or 0 . 3mg/dL , respectively [30] . The median values with interquartile ranges ( IQR ) were used as measures of central tendency . The Fisher’s exact test was used to compare frequencies between the study groups . Continuous variables were compared between the study groups using the Mann-Whitney U test ( 2-group comparisons ) , or the Kruskall-Wallis test with Dunn’s multiple comparisons ad hoc test ( between 3 or more groups ) . Hierarchical cluster analyzes were performed using the Ward’s method with bootstrap ( 100X ) . A p-value below 0 . 05 was considered statistically significant after adjustments for multiple comparisons using the Holm’s Bonferroni test . The statistical analyzes were performed using Graphpad Prism 6 . 0 ( GraphPad Software Inc . , San Diego , CA , USA ) , and JMP 12 . 0 ( SAS , Cary , NC , USA ) .
Among vivax malaria patients , those categorized as having elevated creatinine levels ( n = 89 ) had similar age , but were more frequently female than individuals with normal levels ( n = 90 ) ( Table 1 ) . Overall , patients with P . vivax malaria were younger than uninfected individuals ( median age: 32yrs , IQR: 24–47 vs . 39yrs , IQR: 25–51 , P = 0 . 0096 ) , but were undistinguishable with regard to gender distribution ( male frequency: 55 . 9% vs . 47 . 4% , P = 0 . 151 ) . Patients with severe P . vivax infection recalled less previous malaria episodes and lived for shorter time in the endemic area than individuals with mild infection ( S1 Table ) . Vivax malaria patients exhibiting elevated serum creatinine levels presented sweating , diarrhea , dehydration , abdominal pain , hepatomegaly , jaundice and disorientation more frequently than those with normal creatinine values ( S2 Table ) . Among vivax malaria patients , the frequency of individuals presenting with severe disease was significantly higher in patients with elevated creatinine levels than in those who had normal values ( 16 . 85% vs . 3 . 33% , P = 0 . 0027 , Table 1 ) . Following a similar pattern , frequency of hospitalization was also elevated in patients with elevated creatinine levels vs . those with normal values ( P = 0 . 0003 , Table 1 ) . Moreover , all patients who did not survive ( n = 6 ) were among the group of individuals presenting with the highest creatinine levels ( Fig 1A ) ; two of them presented with anuric renal failure while the other four presented respiratory failure as the major presentation at admission ( see also in [5] ) . Furthermore , within the group of patients with abnormally high creatinine values , nonsurvivors presented with even higher levels when compared with survivors ( median: 2 . 4mg/dL , IQR: 1 . 925–2 . 55 vs . 1 . 36mg/dL , IQR: 1 . 31–1 . 41 , P < 0 . 0001 , Fig 1B ) . Circulating levels of several cytokines , chemokines and inflammatory parameters were compared between the groups of P . vivax malaria patients presenting with or without elevated creatinine values . Patients who displayed elevated creatinine levels exhibited higher CRP values than that in those who did not ( median: 21 . 30ng/mL , IQR: 9 . 45–38 . 45 vs . 11 . 20ng/mL , IQR: 7 . 125–29 . 58 , respectively , P = 0 . 0057 , Table 2 and Fig 2A and 2B ) . Neither IL-10 ( median: 9 . 69pg/mL , IQR: 6 . 43–36 . 83 vs . 19 . 70pg/mL , IQR: 6 . 50–59 . 54 , P = 0 . 0665 ) or IFN-γ ( median: 132 . 0pg/mL , IQR: 35 . 0–333 . 5 vs . 85 . 48pg/mL , IQR: 40 . 15–321 . 5 , P = 0 . 5326 ) levels were significantly different between the study groups ( Table 2 and Fig 2A and 2B ) . Interestingly , values of IFN-γ/IL-10 ratio , which have been shown previously to correlate with systemic inflammation in vivax malaria [5] , were increased in patients who had high creatinine levels compared with those who had not ( median: 5 . 720 arbitrary units [AU] , IQR: 1 . 675–25 . 21 vs . 3 . 334AU , IQR: 1 . 181–6 . 558 , P = 0 . 0306 , Table 2 , Fig 2A and 2B ) . Circulating levels of all other cytokines , chemokines and acute phase proteins could not distinguish the two groups of patients stratified accordingly to serum creatinine levels , and are described in detail in Table 2 . Patients categorized as having elevated creatinine levels presented with higher total bilirubin concentrations than those who had normal creatinine values ( median: 1 . 3mg/dL , IQR: 0 . 965–2 . 1 vs . 0 . 91mg/dL , IQR: 0 . 687–1 . 725 , respectively , P = 0 . 0109 , Table 2 , Fig 2B ) . A similar pattern was observed with regard to direct bilirubin fraction ( P = 0 . 0054 , Table 2 , Fig 2B ) , while no difference was found in indirect bilirubin levels between these study groups . Counterintuitively , when only patients with elevated indirect bilirubin levels ( >1 . 2mg/dL ) were analyzed ( n = 43 ) , higher values of this parameter were detected among the group of patients with normal serum creatinine ( median: 2 . 33mg/dL , IQR: 1 . 6–2 . 7 vs . 1 . 53mg/dL , IQR: 1 . 4–2 . 2 , P = 0 . 0453 , Table 2 ) . In addition , circulating concentrations of both free heme and haptoglobin were undistinguishable between the groups of patients presenting with high indirect bilirubin levels and elevated or normal creatinine values ( Table 2 ) . Furthermore , concentrations of hepatic aminotransferases were also similar between these subpopulations ( Table 2 ) . Hierarchical cluster analysis of the z-score normalized circulating levels of the markers evaluated confirmed that individuals with elevated creatinine levels exhibited a distinct inflammatory profile compared to malaria patients with normal creatinine values or to uninfected healthy controls ( S3 Table and Fig 2A ) . Additional analyses were performed to depict the overall systemic inflammatory profile between vivax malaria patients who survived and those who did not . Nonsurvivors presented higher CRP values than those who survived ( median: 34 . 43ng/mL , IQR: 16 . 43–50 . 7 vs . 21 . 3ng/mL , IQR: 9 . 4–38 . 4 vs . 11 . 23ng/mL , IQR: 7 . 125–29 . 58 , P = 0 . 0122 , Table 3 ) . In addition , values of IFN-γ/IL-10 ratio were increased among the nonsurvivors ( median: 18 . 63AU , IQR: 8 . 29–31 . 55 vs . 5 . 075AU , IQR: 1 . 664–22 . 69 vs . 3 . 334AU , IQR: 1 . 181–6 . 558 , P = 0 . 0282 , Table 3 ) . Circulating levels of all the other cytokines , chemokines and acute phase proteins could not distinguish these two groups that considered survival among patients with elevated serum creatinine , and are described in detail in Table 3 . Hepatic aminotransferases and indirect bilirubin values were not significantly different as well . Nevertheless , nonsurvivors presented higher values of total bilirubin ( median: 2 . 15mg/dL , IQR: 1 . 2–3 . 175 vs . 1 . 3mg/dL , IQR: 0 . 95–1 . 9 vs . 0 . 91mg/dL , IQR: 0 . 687–1 . 725 , P = 0 . 0250 , Table 3 ) and direct bilirubin ( median: 1 . 1mg/dL , IQR: 0 . 325–1 . 675 vs . 0 . 5mg/dL , IQR: 0 . 3–0 . 8 vs . 0 . 4 mg/dL , IQR: 0 . 2–0 . 7 , P = 0 . 0133 , Table 3 ) when compared with the group of survivors . Moreover , when nonsurvivors were compared only with patients with normal creatinine levels , values of CRP ( median: 34 . 4ng/mL , IQR: 16 . 43–50 . 7 vs . 11 . 20ng/mL , IQR: 7 . 125–29 . 58 , respectively , P = 0 . 0121 ) and IFNγ/IL-10 ratio ( median: 18 . 63AU , IQR: 8 . 29–31 . 55 vs . 3 . 334AU , IQR: 1 . 181–6 . 558 , respectively , P = 0 . 0129 ) were increased in nonsurvivors whereas IL-12p70 concentrations were elevated in those with normal creatinine levels ( median: 10 . 0pg/mL , IQR: 5 . 093–15 . 14 vs . 20 . 45pg/mL , IQR: 15 . 73–30 . 30 , respectively , P = 0 . 0333 ) ( Table 3 ) . Hierarchical cluster analysis of circulating inflammatory markers from vivax malaria patients considering mortality and creatinine levels showed a distinct biosignature among nonsurvivors when the group is assessed separately ( Fig 3A ) . Among the biomarkers examined , statistically significant differences between the study groups were observed in values of total and direct bilirubin , CRP and IFNγ/IL-10 ratios ( Fig 3B , Table 3 ) . Noteworthy , parasitemia levels could not distinguish these groups , considering either abnormal creatinine elevation or mortality ( Table 1; Fig 3B ) .
Malaria and its complications have been extensively studied . However , kidney dysfunction related to P . vivax infections has not been completely explored . Some studies have reported cases of severe vivax malaria associated with AKI and described some of its classical clinical symptoms [10–13] , whilst other studies have already depicted the immune response profile in severe vivax cases [5 , 6 , 9 , 26 , 28] . The present study adds on the current knowledge as it now addresses the associations of serum creatinine concentrations with a distinct inflammatory profile which hallmarks subpopulations at higher risk of mortality linked to P . vivax infection . This profile was especially highlighted by a strong signature composed by IFN-γ/IL-10 ratio values and CRP levels , measurements which are directly related with the degree of systemic inflammation over anti-inflammatory mechanisms . Intense systemic inflammatory responses , although non-specific , have been reportedly associated with settlement of different forms of AKI [31–33] . Our findings depict in more detail the relationships between inflammation and AKI in vivax malaria and generate hypotheses to be tested in future pathogenesis studies , rather than propose screening for cytokines in point-of-care settings . In the study population , malaria severity was highly associated with abnormal creatinine increases . Indeed , almost 85% of the patients with severe disease presentation and more than 90% of the hospitalized patients exhibited elevated creatinine levels ( Table 1 ) . Whilst these results highlight the link between severe vivax malaria and kidney dysfunction , it is also worth highlighting that elevations in creatinine levels were not always reflected in increased disease severity , as multiple patients exhibiting elevated serum creatinine presented with nonsevere vivax malaria . The inflammatory profile observed in the group of nonsurvivors , among those with the highest levels of serum creatinine , was even more intense and unbalanced than that in the groups of malaria patients who survived with or without serum creatinine elevation ( Fig 3B ) . However , it remains unknown whether P . vivax genetic diversity [34] would be responsible for different outcomes in those groups , with more virulent strains causing severe disease presentations . Interestingly , parasitemia levels , previously considered as a factor associated with AKI in vivax malaria patients [12 , 13] , were not associated with increases in creatinine levels in the present study ( Table 1 , Fig 3B ) . In addition , values of CCL2 and CCL5 , chemokines , which are known to be produced by kidney tubular cells during injury [31] , were found increased only in the group of nonsurvivors ( Fig 3A ) . Despite the non-significant differences in the levels of these inflammatory chemokines , it is possible that the readouts are indicating altered homeostasis in nephron’s vessels . Hence , the overall results presented here corroborate with the idea that disease severity in vivax malaria , especially considering kidney involvement , is more of a case of inflammatory imbalance and parasite-host interactions through immune activity [5 , 9] than heavy parasitism alone . One of the factors that can influence serum creatinine levels is hemolysis [35] , which hallmarks malaria . Although total bilirubin levels were significantly higher in patients with elevated creatinine values when compared with those with normal creatinine levels , there was no significant difference in indirect bilirubin values between those groups . Moreover , subjects with abnormally high indirect bilirubin values were most frequently not part of the group of individuals with elevated serum creatinine levels ( Table 2 ) . The opposite tendency should have been expected if mainly hemolysis was responsible for the elevation in serum creatinine levels . Hence , these results dissociate the relationship between malaria-associated hemolysis and serum creatinine elevation . The vivax malaria patients who did not survive presented with the highest levels of creatinine in the study population , as well as with elevation of total bilirubin concentrations more dependent on the direct fraction . These results suggest an association between kidney injury and some degree of liver abnormalities and cholestasis , in line with previous reports [12–14] . Furthermore , these results could also suggest and reinforce the idea that elevations in total bilirubin levels may reflect in decreased kidney function , as hyperbilirubinemia would be associated with prerenal failure by altered water balance [36 , 37] and ATN [37 , 38] . Hence , elevated bilirubin levels may play a part in the settlement of acute kidney injury in severe vivax malaria , but probably not as the main cause , similarly to the one previously suggested in P . falciparum infections [38] . With regard to liver damage , hepatic aminotransferase values exhibited wide distribution in both groups . Additional studies are warranted to better define the relationships between kidney and liver function in severe vivax malaria . This study presents some limitations . Information regarding the nutritional condition of the patients were not collected in the original cohort from which this study is based on . There were no follow up data on eventual dialysis or other procedures that may have been conducted during the hospitalization of severe vivax malaria patients . However , previous reports from the same population have shown that malarial treatment improved creatinine levels and inflammation [5] . Moreover , malarial treatment and renal replacement therapy have been shown previously to improve overall patient condition in other population as well [39] . AKI severity categorization was also not included in the original cohort . The RIFLE/AKIN classifications would have helped to distinguish and stratify patients based on their risk or stage of AKI , as well [40 , 41] . However , as delayed access to healthcare is known as a major problem in the region [23] , interfering with the estimation of baseline renal function , these classifications would have little impact in the present study . Despite these limitations , serum creatinine evaluation per se was indeed effectively associated with worst clinical conditions and outcomes ( Table 1 , S1 Table , S2 Table , Fig 1 ) . Hence , overall creatinine evaluation , bearing in mind that there are other potentially more sensitive and specific diagnostic tools [42 , 43] , is a relevant part of physicians’ arsenal for a more efficient detection of kidney dysfunction in vivax malaria . Considering that serum creatinine assessment is widely available , its associations with worst conditions and outcomes makes it of importance for patients in certain settings . In conclusion , the systematic analyzes of multiple inflammatory and clinical biomarkers with creatinine levels argue that renal dysfunction might be a key event in P . vivax malaria severity . Overall , disease severity was associated with elevated creatinine levels , with nonsurvivors presenting with the highest values of serum creatinine , which reinforces the hypothesis that kidney injury is highly associated with mortality in vivax malaria . The detailed inflammatory profile was depicted for each subgroup of patients stratified per creatinine levels , revealing a distinct biosignature as well as evidencing the complexity of the mechanisms leading to disease severity . Our investigation demonstrated an association between systemic inflammation and kidney dysfunction , but not an exclusive effect . We predict that the systemic inflammation observed in more severely ill malaria patients could be contributing to kidney dysfunction through different mechanisms . In addition , the overall evaluation of biomarkers suggest that liver abnormalities and hyperbilirubinemia could also play a role in severe vivax malaria-associated kidney dysfunction . The identification of key factors driving the pathogenesis of this type of disease presentation and experimental models still are necessary to guide future studies and approaches .
|
Severe clinical presentations of Plasmodium vivax malaria are not completely understood . Multi-organ involvement is described in severe vivax cases , however data associating it with kidney dysfunction are relatively scarce , in part because the clinical signs only appear late during kidney injury . We analyzed biomarkers of renal function in groups of patients from the Brazilian Amazon with different presentations of vivax malaria to determine its associations with disease progression . Inflammatory biomarkers were also analyzed to assess inflammation related to kidney dysfunction . The results indicate that severe disease presentation in these patients was associated with abnormal serum creatinine elevations and exacerbated systemic inflammatory response . The highest levels of creatinine were observed in nonsurvivors . Biomarkers of hemolysis did not directly follow increases in serum creatinine . These readouts suggest that kidney dysfunction probably influences vivax malaria severity and mortality . As P . vivax is a widely distributed species of Plasmodium in the world , and severe cases are increasingly being reported , it is important to better understand the role of kidney injury in these presentations , especially considering that it may affect clinical outcomes .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"inflammatory",
"diseases",
"heme",
"medicine",
"and",
"health",
"sciences",
"parasite",
"groups",
"body",
"fluids",
"pathology",
"and",
"laboratory",
"medicine",
"plasmodium",
"immunology",
"tropical",
"diseases",
"bile",
"biomarkers",
"parasitic",
"diseases",
"parasitology",
"apicomplexa",
"signs",
"and",
"symptoms",
"kidneys",
"inflammation",
"proteins",
"creatinine",
"immune",
"response",
"biochemistry",
"bilirubin",
"diagnostic",
"medicine",
"anatomy",
"post-translational",
"modification",
"physiology",
"biology",
"and",
"life",
"sciences",
"malaria",
"renal",
"system"
] |
2018
|
Distinct inflammatory profile underlies pathological increases in creatinine levels associated with Plasmodium vivax malaria clinical severity
|
Most cancer cells harbor multiple drivers whose epistasis and interactions with expression context clouds drug and drug combination sensitivity prediction . We constructed a mechanistic computational model that is context-tailored by omics data to capture regulation of stochastic proliferation and death by pan-cancer driver pathways . Simulations and experiments explore how the coordinated dynamics of RAF/MEK/ERK and PI-3K/AKT kinase activities in response to synergistic mitogen or drug combinations control cell fate in a specific cellular context . In this MCF10A cell context , simulations suggest that synergistic ERK and AKT inhibitor-induced death is likely mediated by BIM rather than BAD , which is supported by prior experimental studies . AKT dynamics explain S-phase entry synergy between EGF and insulin , but simulations suggest that stochastic ERK , and not AKT , dynamics seem to drive cell-to-cell proliferation variability , which in simulations is predictable from pre-stimulus fluctuations in C-Raf/B-Raf levels . Simulations suggest MEK alteration negligibly influences transformation , consistent with clinical data . Tailoring the model to an alternate cell expression and mutation context , a glioma cell line , allows prediction of increased sensitivity of cell death to AKT inhibition . Our model mechanistically interprets context-specific landscapes between driver pathways and cell fates , providing a framework for designing more rational cancer combination therapy .
Oncogene-targeted small molecule kinase inhibitors , like imatinib for BCR-ABL [1] , have transformed chemotherapy . However , such precision medicine approaches are not always efficacious . In some cases , mutation-matched patients do not respond to the drug [2] , or alternatively , resistance stochastically develops [3] . Monotherapy can even activate the target pathway , depending on cellular context [4] . Combination therapy is a logical and clinically-proven path forward [5] , but rationalizing even just the choice of combinations from among the at least 28 FDA-approved [6] targeted small molecule kinase inhibitors , notwithstanding important questions related to the dozens of traditional chemotherapeutics , monoclonal antibodies , dose , timing and sequence , remains challenging for basic and clinical research . Approaches are needed that consider quantitative , dynamic , and stochastic properties of cancer cells given a context . Computational modeling can help fulfill this need , since simulations often are much quicker and less expensive than the explosion of experimental conditions one would need to assay the drug combination space . Some of the first approaches define transcriptomic signatures that specify tumor subtypes and suggest drug vulnerabilities [7] . Big data and bioinformatic statistical approaches are at the forefront of predicting drug sensitivity , with penalized regression and other machine learning methods linking mutation or expression biomarkers to drug sensitivity [8 , 9] . Such statistical modeling approaches largely cannot extrapolate predictions into untrained regimes with confidence . However , this is often a major task-of-interest; for example , drug combination predictions usually only have data for drugs used alone . They also usually cannot account for dose and dynamics , central to combinations . Alternatively , mechanistic computational models based on physicochemical representations of cell signaling have an inherent ability to account for dose and dynamics . They also have greater potential for extrapolation , since they represent physical processes . Multiple mechanistic models exist for almost every major pan-cancer driver pathway identified by The Cancer Genome Atlas ( TCGA ) : receptor tyrosine kinases ( RTKs ) , Ras/ERK , PI3K/AKT , Rb/CDK , and p53/MDM2 [10] . An integrative model that accounts a multi-driver context has not yet been built , but will likely be needed to address drug combination prediction problems with such modeling approaches . Here , we constructed the first mechanistic mathematical model integrating these commonly mutated signaling pathways . We use the MCF10A cell line–a non-transformed mammalian cell line with predictable phenotypic behaviors . We tailor the model to genomic , transcriptomic , and proteomic data from MCF10A cells and train the model using a wealth of literature resources as well as our own microwestern blot and flow cytometry experiments to refine biochemical parameters and phenotypic predictions . We use the model to explore stochastic proliferation and death responses to pro- and anti-cancer perturbations . This mechanistic , biological context-tailored mathematical model depicting several major cancer signaling pathways allows us to probe the mechanisms that underlie how noisy signaling dynamics drive stochastic proliferation and death fates in response to various perturbations , and gain insight into their biochemical mechanisms .
An overarching goal of this work is a foundation that integrates canonical biochemical knowledge of cancer driver pathways with cell-specific context to make quantitative predictions about how drugs , microenvironment signals , or their varied combinations influence cell proliferation and death ( Figs 1A and S1; S1 and S2 Tables ) . Three main considerations guided us . First is scope . TCGA identified pan-cancer driver pathways [10]: multiple RTKs; proliferation and growth ( Ras/Raf/MEK/ERK; PI-3K/PTEN/AKT/mTOR ) ; cell cycle; DNA damage . These interface with expression and apoptosis . See Fig 1A for an overview of the model ( see S1 Fig for a detailed kinetic scheme ) and lists of proteins included in each submodel . Second is inter- and intra-tumoral heterogeneity . Experimental accessibility defines each , which predominantly consists of copy number variation , mRNA expression and mutations , which the foundation integrates . Intra-tumoral heterogeneity is genetic [11] and non-genetic [12] , so we require accounting for such noise . Third is formalism . Signaling , cell fate , and drug action are all inherently quantitative , stochastic and dynamic phenomena . Predictions for a multitude of contexts and drugs , many of which would not have yet been calibrated for , is facilitated by mechanistic as opposed to empirical models [13] , so we take a mechanistic chemical kinetics approach . We present the model as follows ( Fig 1B ) . In “unit testing” we adapt and develop submodels for individual pathways ( see Supplement ) in isolation , requiring that each satisfies a set of experimentally observed biological behavior . We focus on a widely-studied non-transformed cell line , MCF10A [14] to leverage literature data and establish a baseline for “normal” cell behavior before describing transformed contexts with genetic instability and multiple poorly understood mutations . We then connect these submodels together via established signaling mechanisms to perform “integrated unit testing” , which is iterative . Finally , we analyze the model to reason about emergent signaling mechanisms underlying biological observations . Each submodel unit is now constrained by a variety of biological observations and data , so we move to integrated unit testing . We do not imply that submodels are “correct” . Rather , the presented characterization simply increases confidence that submodels reproduce expected cellular behavior , which of course is predicated upon the considered data and assumptions . We first evaluate integrated behavior in the reference state—serum starvation . The chemotherapeutic etoposide induces DNA damage . We treated serum-starved MCF10A cells with etoposide for 24 or 48 hours , but found negligible induction of cell death over control ( Fig 4A-top ) . This is consistent with etoposide-induced cell death being amplified by DNA replication [42] , which we also observed ( EGF + insulin , Fig 4A-bottom ) . We postulated a pharmacodynamic model where etoposide-induced DNA damage is strongly dependent on S-phase cyclins ( see S1 Supplemental Methods ) that reproduced the observed cell death results ( Fig 4A–4C ) . Very few simulated cells without growth factors show a p53 response or active cycling ( Cyclin A ) , and therefore very little cell death ( cleaved PARP ) . With EGF and insulin most simulated cells show a robust and sustained p53 pulsing response , which leads to cell cycle arrest ( through p21 and Chk1 ) and cell death . We noticed that etoposide-induced cell death in the presence of EGF and insulin plateaus over time in experiments and simulations at around 60% , as opposed to increased cell killing over time . We reasoned that this was due to p21/Chk1-induced cycle arrest . In simulations where p53 has limited ability to upregulate p21 , etoposide is predicted to cause greater cell death ( Fig 4D ) . This result is supported by studies showing that cells with mutated p21 are more sensitive to DNA damaging agents [43–45] . Furthermore , several cancer cell lines from the Genomics of Drug Sensitivity in Cancer Project with p21 mutations exhibit increased sensitivity to etoposide [9] . Thus , our model recapitulates the complex dynamic response to a common chemotherapy and predicts features of its context-dependent efficacy . The mechanisms by which oncogene-targeted drugs cause cell death through loss of survival signaling are not as well understood as their cytostatic effects . We measured cell death dynamics ( assayed via Annexin-V/PI flow cytometry ) in response to small molecule inhibitors of ERK and AKT pathways ( downstream mediators of many oncogene-targeted drugs ) alone or in combination , with serum-starvation or EGF + insulin ( Fig 5A ) . As before , serum-starvation induces cell death , which is attenuated by EGF + insulin . Inhibition of either ERK or the AKT pathway alone increases cell death 72 hours post treatment , but negligibly at 48 hours . Combining the two inhibitors revealed synergy in cell death induction at 48 hours , and much more cell death at 72 hours . The integrated model captures these cell death dynamics reasonably well ( Fig 5A—compare far right to far left ) . We used the integrated model to reason about what mechanisms may underlie the synergy between and overall impacts of ERK and AKT inhibition for cell death . Both ERK and AKT can phosphorylate and inactivate the pro-apoptotic function of BAD , ERK can phosphorylate BIM causing similar inhibition of pro-apoptotic function , and AKT-mediated deactivation of FOXO downregulates BIM levels ( Fig 5B ) . Simulations where either BAD- ( blue arrows ) or BIM-dependent ( red arrows ) mechanisms were computationally knocked-out predicted BIM to account for most ERK and AKT inhibition effects ( Fig 5C ) . Independent experimental data corroborate this prediction; MCF10A cell detachment-induced death , which depends on such survival signaling mechanisms , has been observed to be predominantly controlled by BIM rather than BAD [46 , 47] . The model suggests several contributing mechanisms . One reason is expression levels; BAD is lowly expressed compared to anti-apoptotic BCL2 family proteins ( ~14 , 000 vs . ~61 , 000 molecules per cell ) , and thus is unlikely to exert significant control over cell death . Another is the cascade network structure for BIM regulation . If only AKT is inhibited , phosphorylation by ERK blocks the accumulation of active BIM . If only ERK is inhibited , the amount of BIM in the cell , even if completely unphosphorylated , is largely insufficient to drive cell death in the absence of further strong death signals . However , when both pathways are inhibited , the production and accumulation of active BIM proceeds uninhibited , thus creating a potent apoptotic signal . Furthermore , the inhibition of these activities must be coordinated over relatively long periods of time , with duration of AKT inhibition having enough overlap with that of ERK to cause both the accumulation and activation of BIM . Why do some cells die early and others die late following ERK and AKT inhibition ? We used simulation data to gain insight into this question . First , we asked whether initial total protein levels ( sum across all forms of a protein ) were correlated with time-to-death after ERK and AKT inhibition . In line with conclusions from prior studies focused on extrinsic death pathways [12] , time-to-death was not significantly correlated with the initial levels of any single protein ( Fig 5D ) . This suggests that stochastic processes post-inhibitor treatment largely dictate variable cell death fate . We performed lasso regression using time-averaged protein levels as predictors of early or late death ( pre/post 40 hours ) over 400 individual cell simulations . This analysis identified time-averaged BIM and BCL2 of the top explanatory variables . We note that the temporal dependence of explanatory variables reveals a dynamic signature of the pathway . Next , we trained a support vector machine ( 200 cells for training , 200 for validation ) and found that average BIM and BCL2 levels over a 40-hour time course ( or as long as a cell lived ) are highly predictive for death timing ( Fig 5E ) . Initial levels , or even those averaged over the first 8 hours post-stimulus , were moderately predictive at best . As one might expect based on the respective pro- and anti-apoptotic functions of BIM and BCL2 , respectively , stochastic expression time courses that tend towards high BCL2 and low BIM are the most protective ( Fig 5F ) . Interestingly , these predictors of stochastic death phenomena were not equally applicable to different death stimuli such as the extrinsic ligand TRAIL ( S6A Fig ) , thought to depend more on stochastic activation dynamics of Caspase 8 [12] , which highlights the treatment-specificity of stochastic cell fate determinants . Overall , these results suggest that intrinsic death of individual cells induced by inhibition of survival signals may be inherently unpredictable prior to treatment . Such a claim is inherently difficult to prove conclusively , since there always might exist alternative measurable predictors and/or methods of analysis that render such prediction possible . Moreover , in our specific case , we have not yet incorporated detailed stochastic mechanisms of protein degradation , which when included may provide additional informative co-variates . EGF and insulin regulate MCF10A proliferation . We measured ( BrdU incorporation/flow cytometry ) how these two ligands influence cell cycle progression of serum-starved MCF10A cells and how the ERK and AKT pathways were involved ( Fig 6A ) . Insulin induces negligible cell cycle entry on its own but synergizes with EGF ( Fig 6A ) , which is also seen on the level of cyclin D expression ( Fig 6C ) . The ERK and AKT pathways are both essential to drive EGF + insulin-induced cell cycle progression ( Fig 6A ) . After tuning the dependencies of transcriptional processes downstream of the ERK and AKT pathways , the integrated model reproduces these data ( Fig 6B and 6C ) . What underlies the synergy between EGF and insulin with regards to cell cycle progression ? Prior work has suggested that the dynamics of the ERK pathway can determine cell proliferation fate [48] . However , data and simulations suggest that ERK pathway activation dynamics are essentially identical for EGF or EGF + insulin ( Fig 6D ) . Insulin is a very strong activator of AKT but not ERK signaling ( Fig 6D ) . When cells are treated with EGF + insulin , there are negligible differences in acute ( <30 min ) AKT activation dynamics , and significant differences only develop over long ( hours ) time scales . Thus , the mitogenic synergy between EGF and insulin seems associated with prolonged AKT pathway activity over long time scales . Simulations suggest that increasing mitogen-induced cyclin D expression cannot be accomplished by only increasing the activity of one pathway , but rather roughly equal amounts of both time-integrated ERK and AKT activity are needed to drive more robust mitogenic responses ( Fig 6E ) . This interpretation cannot rule out coincident factors such as increased insulin-induced glucose uptake , but does suggest that the coordinated activation of both ERK and AKT for several hours post-mitogen treatment seem important to drive cell cycle entry , rather than acute activation dynamics following mitogenic stimulus . These observations suggested to us that stochastic differences from cell-to-cell in AKT activity dynamics might be predictive for EGF + insulin-induced cell cycle entry . Surprisingly , in simulation results , neither the initial AKT activity nor dynamics post-stimulus contained significant discriminatory information related to subsequent cell cycle entry decisions of individual simulated cells ( Fig 6F , bottom ) . Rather , the ERK activity dynamics were much more predictive of subsequent stochastic cell cycle entry ( Fig 6F , top ) . This is consistent with prior live-cell imaging work that showed a strong correlation between time-integrated ERK activity and resultant S-phase entry decisions [34] , albeit in a setting of much higher cell confluence than our experiments here entail and with additional mitogenic factors present ( hydrocortisone and cholera toxin ) . Why would AKT activity dynamics control synergy between EGF and insulin but not be predictive of stochastic cell fate ? AP1 and cMyc control cyclin D expression in this model . AP1 ( cFos-cJun ) activity has a sharp response to time-integrated ERK activity , relative to the less sharp dependence of cMyc expression on AKT activity . This sharpness difference is due to positive autoregulation of cJun expression by AP1 [49] , and is likely also influenced by the already high expression of cMyc in serum-starved MCF10A ( ~48 , 000 molecules/cell , compared to 0 and ~3 , 000 for cFos and cJun ) , which is one of the few MCF10A alterations ( they also have CDKN2A/ARF loss , which is captured in our gene copy number data ) [50] . We also wondered whether higher AKT activity lowers the threshold of ERK activity needed to cause cell cycling; however , we found this not to be the case ( S7D Fig ) . Could stochastic cell cycle entry be predicted by protein abundance fluctuations prior to EGF + insulin treatment ? We again applied lasso regression using initial conditions for total protein levels in individual simulated cells to identify explanatory variables for cell cycle entry , and then trained a support vector machine classifier . In contrast to predicting death in individual cells , here we found that initial CRaf and BRaf levels were fairly predictive of whether a cell would enter the cell cycle by 24 hours post EGF + insulin stimulus ( Fig 6G ) . Initial total ERK and AKT ( active + inactive ) levels were not predictive at all ( Fig 6G ) . We conclude that so long as AKT activity is prolonged , ERK activity fluctuations likely control stochastic cell proliferation fate in response to acute growth factor stimulus . These ERK activity fluctuations across a cell population seem heavily influenced by initial Raf protein levels . The model contains multiple oncogenes and tumor suppressors , and is calibrated to a non-transformed epithelial cell state . We wondered what simulations would predict the transformation potential to be for each gene . To test this , we systematically altered expression of each RTK , proliferation and growth submodel gene ( 10-fold up for oncogenes , 10-fold down for tumor suppressors ) , simulated proliferation response to EGF + insulin ( emulate a growth factor-containing microenvironment ) , and ranked nodes by this metric ( Fig 7A ) . Several expected high-ranking results are observed , such as Ras , Raf , ErbB2/HER2 and EGFR , those with well-established transforming viral analogs such as cFos , cJun and cCbl , and more recently recognized translation control with EIF4E and EIF4E-BP1 [51] . Gene products along the PI3K axis ( PIK3Cα , PTEN , β-Catenin , cMyc ) are lower ranked , which seems counter-intuitive since these are also frequent cancer drivers . However , MCF10A cells , while non-transformed , are immortal and fast growing in culture indefinitely , and one of the reasons ( as stated above ) is cMyc overexpression . The cMyc protein is downstream biochemically of PI3K/AKT , its overexpression occurs as a resistance mechanism in response to PI3K inhibitor therapy , and PI3K and cMyc alterations tend to be mutually exclusive [52–54] . Thus , we interpret these results as being consistent with the MCF10A context being naturally less reliant on these PI3K axis gene products to drive increased proliferation . One paradoxical result is the low ranking of MEK ( MAP2K1/2 ) , despite being functionally surrounded by highly ranked genes . This result is actually consistent with widespread clinical data that seldom report MEK alterations ( ~<5% TCGA pan-cancer in cBioPortal ) as compared to Ras/Raf alterations , the explanation for which remains unclear . What mechanisms might explain MEK’s poor simulated transforming potential ? In simulations , free C/B-Raf—those that are available to transmit signals from active Ras—are reduced , and most of this difference ends up sequestered by MEK itself , which seems to reduce active ERK ( Fig 7B–7D ) . Simulations where MEK has artificially high affinity for MEK phenocopy this result . We interpret this as a consequence of stoichiometry and expression levels , since both C-Raf and B-Raf are relatively lowly expressed ( ~29 , 000 mpc and ~2 , 800 mpc , respectively ) as compared to MEK ( MAP2K1~147 , 000 mpc; MAP2K2~378 , 000 mpc ) , so although this sequestration effect might be relatively small , it can account for a non-negligible proportion of the total Raf pool . Thus , the prediction from these simulations is that MEK typically has poor transforming potential because increasing its levels it tends to sequester Raf from increased signaling as opposed to allowing for increased signal transmission . This of course does not fully explain observations related to the very low frequency of MEK activating mutations , which convolve protein structural robustness , and also potentially genome organization factors . Nevertheless , such a mechanism would not be possible to uncover from genomics or network model-based approaches because it is inherently based in quantitative biochemistry that our model captures from its formulation . Future work will have to perform a thorough genetic screen to experimentally validate these simulation-derived predictions regarding the transformation potential of single gene modifications . Thus far we have focused on tailoring the model to MCF10A cells , and have parameterized the model to reflect expression levels , protein dynamics , cell death and cell cycling percentages following various stimuli . Does our model possess the ability to be predictive of cell fate outcomes when tailored to an alternative cellular context ? To test this , we used data from U87 cells , a glioma cell line . Flow cytometry experiments of U87 cells treated with a MEK inhibitor , an AKT inhibitor , or dual kinase inhibition indicated pronounced sensitivity of U87 cells to AKT , but not MEK , inhibition ( Fig 8A ) ; this result was not found for MCF10A cells , which were found to be minimally sensitive to either inhibitor used alone ( Fig 5A , grey bars ) . We thus sought to test whether the model tailored to U87 expression data could predict this increased sensitivity to AKT inhibition . To test this , we quantified mRNA levels in serum-starved U87 cells and tailored the model with the same setup pipeline procedure as for the MCF10A-tailored model ( Fig 2D ) . In addition , because U87 cells are PTEN-deficient [55–57] , we set the PTEN translation rate constant to zero , essentially removing PTEN from the system . U87 cells are also thought to possess a mutation in CDKN2A; however , since the incorporation of this mutation would negligibly affect cell death responses to kinase inhibition , we excluded it from this analysis . However , future iterations of this model application to U87 should certainly incorporate CDKN2A mutation , in addition to more generally capturing the spectrum of mutations in this and other cell lines . We then simulated our U87 cell model under the above experimental conditions , recording percentage of cell death at 48 hours post stimulus ( Fig 8B , red bars ) and compared this to flow cytometry data ( Fig 8A ) . Remarkably , the U87-tailored model predicted with reasonable accuracy cell death responses to the different stimuli , capturing the increased sensitivity to AKT inhibition compared to our MCF10A-tailored model simulations . We noted that the U87-tailored model is unable to capture basal cell death under the “No Stim” condition ( Fig 8B ) . We suspect this highlights a limitation in the current initialization procedure , specifically in the tuning of basal caspase 8 activity to reflect balanced levels of intrinsic death signals in the absence of any stimuli . Future work must more carefully consider the phenotypic requirements of initialization across cell types . Simulated time courses suggest that the increased sensitivity to AKT inhibition in U87 cells may be driven by increased accumulation of FOXO ( Fig 8C , left ) , and subsequently the pro-apoptotic BIM ( Fig 8C , right ) , when compared to MCF10A cells . A proposed model-derived mechanism behind this observation lies in the faster degradation rate of phosphorylated FOXO [58–60] . In U87 cells , the presence of the PTEN-null mutation , in addition to higher basal AKT levels , results in higher basal AKT activity . This results in increased FOXO phosphorylation , which causes it to be more rapidly degraded . In turn , when AKT is inhibited and FOXO becomes unphosphorylated , its degradation is slowed and it accumulates , driving BIM upregulation and cell death . Future experimental work will have to test these mechanistic interpretations .
We developed the first pan-cancer driver network model of cell proliferation and death regulation that has a biological context mechanistically parameterized by multi-omics data . The model describes fractional cell kill and partial cytostasis in response to treatment with a variety of anti-cancer drugs , conditioned upon an expression and microenvironmental context . We first developed sets of unit test operations that build confidence in individual submodels , and then evaluated the ability of the model to reproduce cell biological behavior as an integrated whole , which we subsequently analyzed to gain insight into biochemical mechanisms of stochastic cell proliferation , death and transformation . We then applied the model to a glioma cell context , which was able to predict differential sensitivity to AKT inhibition . While this model is complex and larger than any previous such signaling model , some may argue it still does not account for enough biology . We have predominantly focused on a single non-transformed cell line with no mutations . Impeding future work here is that most mutations are not functionally well-understood . We do not incorporate detailed mechanisms of post-translational Myc regulation , pathways outside of pan-cancer scope , tumor metabolism , hypoxia signaling or immune response . We do not include VEGFR , which although an important receptor tyrosine kinase in cancer , predominantly plays a tissue level role in angiogenesis as opposed to a signaling role inside tumor cells themselves , the focus of this study [61 , 62] . We also have not considered cell-to-cell communication from physical contact , autocrine , or paracrine signaling—so called population effects—which will be important to consider moving forward . In our experiments , we minimized cell confluence to reduce such effects . Nevertheless , MCF10A are strongly adherent to one another , and the presented data , results and phenotypes , and therefore our model implicitly account for these population effects . Future experiments may focus on better teasing these components apart . Yet others may argue that our first model draft is too large to be useful . Symmetric critiques often reflect striking an appropriate balance , but both views have validity . Rule-based modeling [63] and algorithmic model building [64] could help modelers capture more biology . Open and FAIR [65] data that is well-annotated and available by computational query ( e . g . LINCS consortium ) will be needed to drive such modeling forward . We certainly do not claim that our present model is unique , neither in terms of structure nor parameterization . Several variants of submodels exist [22 , 29 , 30 , 66] , as well as efforts to integrate them with expression data [67–70] , most notable the whole cell model of mycoplasma [71] . These classes of models are well-known to exhibit parametric non-identifiability but yet are still able to make valid and precise biological predictions [72] . The fact that these models stand on the shoulders of decades of cell and molecular biology research and are formulated to the extent possible based on physicochemical principles facilitates predictive potential despite the inherent uncertainty . Increased ability to algorithmically compare model structure and predictions to a wide net of publicly available experimental data will help reduce model uncertainty . Regardless , model complexity is driven by the questions and data at hand , and here the ultimate goal of cancer precision medicine coupled with omics data for tailoring gives a rational basis for the model size and complexity . The fact that when we tuned the model to the U87 glioma cell context , it could predict increased sensitivity of cell death to AKT inhibition , suggests that the level of detail may be an appropriate foundation . To study stochastic cell fate with our model we combined penalized regression ( identify predictive features ) with machine learning ( relate predictive features to cell fates ) . Like prior work [12] , we found that ERK+AKT inhibitor-induced apoptosis was not predictable from initial protein levels . Rather , the couplet of time-averaged total BIM and BCL2 levels following death stimulus were highly predictive . Thus , apoptosis seems inherently unpredictable at the time of the death stimulus , and rather evolves stochastically over the treatment time course . It is difficult to prove such a claim conclusively , as a combination of different initial measurements with non-linear models could provide such predictive power . However , the results do suggest a strong stochastic element of cell death that is hard to predict from initial protein levels alone . To test the BIM and BCL2 dynamics predictions experimentally , the ability to track cells over time from the onset of perturbation to the final cell fate is needed . If one could engineer cell lines to express fluorescent protein tagged ( yet still functional ) copies of all BIM and BCL2 isoforms from each allele , and monitor their responses by live-cell imaging , then this prediction could be tested . However , construction of such lines remains a challenging proposition . In contrast to the inherent unpredictability of apoptosis based on initial protein levels , S-phase entry in simulations was predictable from initial C-Raf and B-Raf levels . Why is this the case for S-phase entry and not apoptosis ? There are several potential factors in the model . First is the relatively low expression levels of both Raf isoforms , giving larger expression noise . Second is the bottleneck role of Raf in propagating signals from RTKs to S-phase entry , coupled with enzymatic signal amplification as opposed to stoichiometric titration as in apoptosis signaling . Third is the predominant stochastic control of S-phase entry by ERK signaling dynamics , as opposed to AKT signaling dynamics . This may be due in part to the downstream positive feedback of cJun expression , which amplifies noise . Small fluctuations in Raf levels therefore might be amplified to all or nothing cJun and consequently cell cycle entry responses . Our model can naturally account for drug dose , promiscuity , scheduling , kinetics and pharmacodynamics . These hallmarks of pharmacology are as yet difficult to rationalize with genomic approaches . One potential application of ours and similar models is improving upon genomic-based approaches to cancer precision medicine . For example , patient transcriptomic data could be used to tailor the model as described here , giving an in silico test bed to explore a plethora of single or combination therapy options , including dosing and timing . Given advances in patient-derived experimental systems from biopsies [73] , perturbation data can be used to further refine patient-specific mechanistic models and test drug predictions prior to clinical administration . Such predictions would have engrained within important properties of clinical response such as the promiscuity of many targeted kinase inhibitors [74] and dynamic feedback mechanisms in cancer cells important for resistance [75] . Moreover , given inference of subclonal evolution patterns of a patient’s tumor [11] , multiple instantiations of the model could be generated to find a cross-section of therapies , or perhaps even order of therapies , that could most efficiently control an ( epi ) genomically heterogeneous tumor . Alternatively , virtual clinical trials could be conducted for a particular drug or combination , given a cohort of patients defined by multi-omics data ( e . g . transcriptomics ) . Simulation results could stratify patients for inclusion or exclusion in trials—an exceedingly important task given the niche efficacy of targeted drugs from single genetic biomarkers . Further simulations could optimize dosing and timing strategies . With now hundreds of FDA-approved anti-cancer therapeutics ( including dozens of targeted drugs ) , and the realization that combination therapy is almost always required for clinical response , computational approaches such as this that help prioritize drugs , their doses , and their scheduling for particular patients will become increasingly important in drug development and personalized oncology .
Our model is the amalgamation of several already-published models from the literature into an integrated system of ordinary differential equations ( ODEs ) . The model can be subdivided into six principal submodels: ( 1 ) receptor tyrosine kinase ( RTK ) [22 , 23 , 76–78] , ( 2 ) proliferation and growth ( which includes Ras-Raf-MAPK plus PI3K-AKT-mTOR pathways ) [23 , 29 , 30] , ( 3 ) cell cycle [40] , ( 4 ) apoptosis [37] , ( 5 ) DNA damage response [35] , and ( 6 ) gene expression ( the genes/proteins involved in each submodel are noted in main text Fig 1 ) . The model is composed of ~1200 total species , which includes all genes , mRNAs , lipids , proteins , and post-translationally modified protein/protein complexes . There are a total of 141 genes across all submodels that can be reduced into 102 functionally unique proteins , which we term “protein conglomerates” ( see S1 Table for mapping ) . These 102 immediate gene products can become post-translationally modified in various ways to make up the remaining ~1100 species in the model . We opted to represent the majority of reactions as elementary steps to the extent possible , using mass action kinetics . However , sometimes we employ Michaelis-Menten or Hill representations to capture events where the mechanisms are not sufficiently elucidated . Kinetic parameters are taken from original source models , previous models , experimental studies , or estimated to fit empirical observations ( see S2 Table for rate laws and rate constant sources ) . See Supplementary Methods for a more detailed description of the model structure and formulation .
|
Cancer is a complex and diverse disease . Two people with the same cancer type often respond differently to the same treatment . These differences are primarily driven by the fact that two type-matched tumors can possess distinct sets of mutations and gene expression profiles , provoking differential sensitivity to drugs . Over the past few decades , we have seen a shift away from more broadly cytotoxic drugs to more targeted molecules therapies; but how to match a patient with a specific drug or drug cocktail remains a difficult problem . Here , we build a mechanistic ordinary differential equation model describing the interactions between commonly mutated pan-cancer signaling pathways—receptor tyrosine kinases , Ras/RAF/ERK , PI3K/AKT , mTOR , cell cycle , DNA damage , and apoptosis . We develop methods for how to tailor the model to multi-omics data from a specific biological context , devise a novel stochastic algorithm to induce non-genetic cell-to-cell fluctuations in mRNA and protein quantities over time , and train the model against a wealth of biochemical and cell fate data to gain insight into the systems-level , context-specific control of proliferation and death . One day , we hope models of this kind could be tailored to patient-derived tumor mRNA sequencing data and used to prioritize patient-specific drug regimens .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"cell",
"death",
"medicine",
"and",
"health",
"sciences",
"diabetic",
"endocrinology",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"messenger",
"rna",
"hormones",
"erk",
"signaling",
"cascade",
"insulin",
"endocrinology",
"biochemistry",
"rna",
"signal",
"transduction",
"biochemical",
"simulations",
"cell",
"biology",
"nucleic",
"acids",
"apoptosis",
"biology",
"and",
"life",
"sciences",
"computational",
"biology",
"cell",
"signaling",
"signaling",
"cascades",
"cyclins"
] |
2018
|
A mechanistic pan-cancer pathway model informed by multi-omics data interprets stochastic cell fate responses to drugs and mitogens
|
Since cholera appeared in Africa during the 1970s , cases have been reported on the continent every year . In Sub-Saharan Africa , cholera outbreaks primarily cluster at certain hotspots including the African Great Lakes Region and West Africa . In this study , we applied MLVA ( Multi-Locus Variable Number Tandem Repeat Analysis ) typing of 337 Vibrio cholerae isolates from recent cholera epidemics in the Democratic Republic of the Congo ( DRC ) , Zambia , Guinea and Togo . We aimed to assess the relationship between outbreaks . Applying this method , we identified 89 unique MLVA haplotypes across our isolate collection . MLVA typing revealed the short-term divergence and microevolution of these Vibrio cholerae populations to provide insight into the dynamics of cholera outbreaks in each country . Our analyses also revealed strong geographical clustering . Isolates from the African Great Lakes Region ( DRC and Zambia ) formed a closely related group , while West African isolates ( Togo and Guinea ) constituted a separate cluster . At a country-level scale our analyses revealed several distinct MLVA groups , most notably DRC 2011/2012 , DRC 2009 , Zambia 2012 and Guinea 2012 . We also found that certain MLVA types collected in the DRC persisted in the country for several years , occasionally giving rise to expansive epidemics . Finally , we found that the six environmental isolates in our panel were unrelated to the epidemic isolates . To effectively combat the disease , it is critical to understand the mechanisms of cholera emergence and diffusion in a region-specific manner . Overall , these findings demonstrate the relationship between distinct epidemics in West Africa and the African Great Lakes Region . This study also highlights the importance of monitoring and analyzing Vibrio cholerae isolates .
Since 1817 , seven cholera pandemics have plagued humans worldwide [1] . During the current pandemic , the disease first appeared on the African continent when Vibrio cholerae was imported by migrants traveling to the Guinean capital of Conakry in 1970 [1] . Following this importation of the bacterium , cholera cases have been reported every year in Africa , and many regions in Sub-Saharan Africa have been deemed cholera endemic [2] . Although Africa currently has the highest incidence of cholera globally , the disease affects the continent in a heterogeneous manner . African cholera outbreaks primarily cluster at certain hotspots including ( 1 ) the African Great Lakes Region and ( 2 ) West Africa , stretching from Cameroon and the Lake Chad region along the coast to Guinea [3–6] . Cholera outbreaks commencing in the African Great Lakes Region have been found to spread to neighboring countries , such as Sudan in 2006 and Kenya in 2009 [5] . Likewise , epidemics have progressively moved along the West African coast , as observed in 2003–2007 when outbreaks spread from Liberia and Sierra Leone to Guinea [6] . Indeed , cholera appears to spread in a highly dynamic manner that poses a significant public health threat at a regional level across Africa . To design effective public health strategies to combat the disease , it is critical to understand the mechanisms of cholera emergence and diffusion in a region-specific manner . Epidemiological analysis of outbreaks is critical to identify hotspots and patterns of disease spread . However , molecular biological methods can provide further insight into the relationship between pathogenic strains and epidemic populations [7] . Indeed , isolate typing is useful to differentiate between different isolates , identify clusters , establish phylogeny , and track bacterial transmission . Lam et al . [8] have recently shown that MLVA ( Multi-Locus VNTR ( Variable Number Tandem Repeat ) Analysis ) represents a highly discriminatory technique to distinguish between closely related seventh pandemic isolates . They have also emphasized that the method is best applied for outbreak investigations or to identify the source of an outbreak . Our research group has recently demonstrated that MLVA-based analysis of clinical V . cholerae isolates combined with an epidemiological assessment was instrumental in deciphering the origin of the 2012 Guinean epidemic [9] . In the current study , we applied MLVA-based typing of 337 V . cholerae isolates from recent cholera epidemics in Sub-Saharan Africa to assess the relationship between outbreaks . We identified 89 unique MLVA haplotypes across our isolate collection . When coupled with corresponding epidemiological data , we revealed the short-term divergence and microevolution of these V . cholerae populations to provide insight into the dynamics of cholera outbreaks and the relationship between distinct epidemics in West Africa and the African Great Lakes Region .
Overall , we analyzed 337 V . cholerae isolates derived from epidemics in the Democratic Republic of the Congo ( DRC ) , Guinea , Togo and Zambia . Our panel included six isolates from environmental samples and 331 clinical samples . A total of 237 V . cholerae isolates from epidemics in the DRC occurring in 2008 ( 3 isolates ) , 2009 ( 108 isolates ) , 2011 ( 60 isolates ) , 2012 ( 44 isolates ) and 2013 ( 22 isolates ) were provided by the INRB ( French acronym for the National Institute of Biomedical Research ) , Kinshasa , DRC . Of the 60 DRC isolates from 2011 , two were isolated from environmental water samples collected from Lake Tanganyika by staff at the Centre de recherche en Hydrobiologie ( in Uvira , RDC ) and analyzed at the Hôpital Générale de Référence in Uvira by Hilde de Boeck . The Guinean reference laboratory of the Public Health National Institute ( INSP—Institut National de Santé Publique ) , with support from the AFRICHOL Consortium ( http://www . africhol . org/ ) , provided 36 V . cholerae isolates collected throughout Guinea during the 2012 epidemic as previously described [9] . The National Institute of Hygiene in Lomé , Togo provided 35 V . cholerae isolates from Togo , which corresponded to epidemics in 2010 ( 13 isolates ) , 2011 ( 10 isolates ) and 2012 ( 12 isolates provided by the bacteriology laboratory of the National Institute of Hygiene in Togo via the project AFRICHOL ) . A total of 27 V . cholerae clinical isolates from the 2012 Zambian epidemic were analyzed; the isolates were collected by the staff at the Cholera Treatment Center at the Mpulungu Health Centre during the CHOLTIC project together with the Institute of Tropical Medicine in Antwerp , Belgium . Two environmental isolates were also collected during the CHOLTIC project in association with the Department of Fisheries , Lake Tanganyika Research Unit . The Mpulungu Health Centre performed the initial characterization of the V . cholerae environmental samples . Regarding the isolates from Zambia , suspected cholera cases were routinely cultured to test for the presence of V . cholerae . The study , including a waiver of written consent , was approved by the University of Zambia Biomedical Research Ethics Committee , the Institute of Tropical Medicine institutional review board , and the ethical committee of the University of Antwerp , Belgium ( study registration number B300201317249 ) . Patients , children’s parents and/or legal guardians were informed and approved via oral consent before enrollment into the study . Oral consent was registered by the ward nurse , and participant samples received a study ID number to anonymize the data . Concerning the Guinean isolates , the sampling of suspected cholera cases for culture confirmation of V . cholerae is included among the routine procedures in accordance with the policies of the Ministry of Health Guinean . The Ministry of Public Health and Public Hygiene , Conakry ( Ministre de la Santé Publique et de l’Hygiène Publique ) approved the use of these V . cholerae isolates for research and publication purposes . In Togo , to confirm suspected cholera cases , patient samples are routinely cultured to test for the presence of V . cholerae . The directorate of the National Institute of Hygiene , Lomé , Togo ( the laboratory director and the head of the bacteriology department ) approved the study of these isolates for research purposes , including the comparison of these Togolese isolates with other African V . cholerae samples . In the Democratic Republic of the Congo , samples from suspected cholera cases are routinely collected and analyzed for the presence of V . cholerae in the framework of the epidemiological surveillance program of the Ministry of Health . The ethics committee at the University of Kinshasa approved the analysis of these isolates for research purposes . All data analyzed in the study were anonymized . The isolates were subcultured and inoculated into Vibrio cholerae Enrichment Broth vials ( Bio-Rad ) . The Bio-Rad vials were subsequently expedited ( 2–3 days ) at ambient temperature to L'Hôpital d'Instruction des Armées Laveran in Marseille , France . In Marseille , the strains were recultivated on non-selective trypticase soy agar medium ( Difco Laboratories/BD ) for 24 hours at 37°C . Suspected V . cholerae colonies were identified via Gram-staining , oxidase reaction and agglutination assessment with V . cholerae O1 polyvalent antisera ( Bio-Rad ) . For DNA extraction , an aliquot of cultured cells ( approximately 50 colonies ) was suspended in 500 μL NucliSENS easyMAG lysis buffer ( bioMérieux , Marcy l'Etoile , France ) . Total nucleic acid was extracted from V . cholerae cultures using a NucliSENS easyMAG platform ( bioMérieux ) according to the manufacturer’s instructions . Nucleic acid concentration and 260/280 ratio were measured using a NanoDrop 3300 fluorospectrometer ( Thermo Scientific , Villebon sur Yvette , France ) . The supernatants ( 100 μL ) were stored at -20°C for downstream applications . Genotyping of the V . cholerae isolates was performed via MLVA of six VNTRs , including five previously described assays and a novel VNTR assay , VCMS12 , specifically designed for this study to improve the discriminative power of the analysis ( Table 1 ) [9–11] . The VCMS12 assay was designed based on the reference strain El Tor N16961 ( GenBank accession numbers AE003852 . 1 and AE003853 . 1 ) using Perfect Microsatellite Repeat Finder ( currently unavailable ) . VCMS12 is located within the cholera toxin A subunit promoter region at position 1568189 on chromosome 1 of El Tor N16961 . This polymorphic tandem heptanucleotide repeat region has been previously identified by Naha et al [12] . Specific primer pairs were subsequently designed using Primer3 ( http://simgene . com/Primer3 ) . The fluorescent-labeled primers ( Table 1 ) were purchased from Applied Biosystems . Each VNTR locus was amplified separately . DNA amplification was carried out by preparing a PCR mix containing the following components: 0 . 375 μL of each primer ( 20 μM ) , 1 X LightCycler 480 Probes Master ( Roche Diagnostics ) and approximately 100 ng of template DNA . The PCR mix was brought to a total volume of 30 μL with H2O . PCR was performed using a LightCycler 480 System ( Roche Diagnostics ) . All PCRs were performed using the thermal cycling conditions as follows: 95°C for 5 min; followed by 30 cycles of 95°C for 30 sec , 58°C for 30 sec and 72°C for 45 sec; 72°C for 5 min . Aliquots of the PCR products were first diluted 1:30 in sterile water . Next , 1 μL of the diluted PCR reaction was aliquoted into a solution containing 25 μL Hi-Di Formamide 3500 Dx Series ( Applied Biosystems ) and 0 . 5 μL GeneScan 500 LIZ Size Standard ( Applied Biosystems ) . The fluorescent end-labeled PCR amplicons were separated via capillary electrophoresis using an ABI PRISM 310 Genetic Analyzer ( Applied Biosystems ) with POP-7 Polymer ( Applied Biosystems ) . Finally , amplicon size was determined using GeneMapper v . 3 . 0 software ( Applied Biosystems ) . The MLVA results were exported to Microsoft Excel 2008 v . 12 . 2 . 0 . Allele numbers were derived directly from the fragment sizes , and MLVA types were determined from the combined profile of alleles ( i . e . , each unique combination of six allele numbers was assigned a novel MLVA type number ) . The absence of allele at a given locus ( i . e . , no amplification , despite repeated attempts ) was assigned “999” for further analyses . No amplification was observed for 22 of the 2022 examined alleles . To perform the Minimum Spanning Tree ( MST ) analysis , the isolates were further assigned into 10 epidemic populations as follows: DRC 2008 , DRC 2009 A ( collected in January-May ) , DRC 2009 B ( collected in July-November ) , DRC 2011 , DRC 2012 , DRC 2013 , Guinea 2012 , Zambia 2012 , Togo 2010 , Togo 2011 and Togo 2012 . The DRC 2009 epidemic isolates were sub-grouped by time period as two distinct epidemics were observed in the region during field investigations . The maps were generated using QGIS version 2 . 4 . 0-Chugiak with shapefiles obtained from DIVA-GIS ( http://www . diva-gis . org/gdata ) . Based on allelic profiles the evolutionary relationship between all 337 isolates was assessed with the MST algorithm in BioNumerics ( Applied Maths , Sint-Martens-Latem , Belgium ) using the default settings according to the manufacturer’s recommendations . The MST was constructed using a categorical coefficient as previously described [13] . To identify clonal complexes and founder MLVA types among the 331 clinical isolates ( i . e . , excluding the six environmental isolates ) , MLVA-types were compared at each of the six VNTR loci and genetic relatedness between the strains was assessed using goeBURST version 1 . 2 . 1 ( http://goeburst . phyloviz . net/ ) [14] . The goeBURST algorithm identifies mutually exclusive groups of related MLVA types in a population . The algorithm also predicts the presumed founder ( s ) of each clonal complex and any single locus variant ( SLV ) and double locus variant ( DLV ) derivatives . The primary founder of a group is defined as the MLVA type that has the greatest number of SLVs . goeBURST then constructed a spanning forest in which each MLVA type is a node and two MLVA types are connected if they are SLVs . An MLVA cluster was defined as a group of isolates that share identical alleles at five of the six loci with at least one other member of the group . Accordingly , singletons were defined as MLVA types having at least two allelic mismatches with all other MLVA types . The number of re-samplings for bootstrapping was set at 1000 . Population genetic analyses were performed on each set of epidemic isolates for which there were 10 or more isolates per set; therefore , the three DRC 2008 isolates were excluded from the statistical analyses . Fst ( F-statistics; also known as fixation indices ) and p-values were calculated pair-wise for all epidemic populations of the 328 clinical isolates using the Nei ( 1987 ) method implemented in GenoDive 2 . 0b25 [15] as were the allelic diversity of all clusters and loci . Several field investigations were conducted in the DRC , Guinea and Togo by members of our team ( Renaud Piarroux , Stanislas Rebaudet , Berthe Miwanda , Didier Bompangue , Aaron Aruna Abedi , Jean-Jacques Depina and Sandra Moore ) . Field investigations were performed at sites affected by cholera , in which the data collected included number of cases/deaths , laboratory results , the locales affected and the spatiotemporal evolution of the epidemics . Further details of the field investigations are provided in the corresponding studies of the specific epidemics ( concerning the Togo field investigation , please see S1 Text ) [9 , 16] . In Zambia , strains collected in the context of the national surveillance system and the CHOLTIC project with the objective of comparing strains collected in the vicinity of Lake Tanganyika ( at sites of varying distances from the coast ) . Strains were collected in Mpulungu , Northern Province , between the 12th and 24th of August 2012 , during an outbreak that occurred at this time . The following data corresponding to the collected strains was also compiled: patient name , sex , age , date seen at facility , date of symptom onset , laboratory results ( if any were performed ) and date of patient discharge . No corresponding field investigation was performed during this outbreak .
A total of 337 V . cholerae isolates from recent cholera epidemics in the DRC , Zambia , Guinea and Togo were subjected to MLVA . Each country is localized on the map of Africa in Fig 1 . Analysis of the six VNTRs yielded 89 MLVA types . The VNTR loci and epidemic populations ( grouped by country and year of isolation ) corresponding to each MLVA type are outlined in S1 Table . A MST was constructed using the combined MLVA data to assess the relationships between the 337 V . cholerae isolates and the epidemic populations . On a continental scale , the MST revealed strong geographical clustering with isolates from the African Great Lakes Region , including the DRC and Zambia , clustering together . Furthermore , the isolates collected in the West African countries of Guinea and Togo formed a separate group ( Fig 2 ) . All but seven clinical isolates from the DRC and Zambia 2012 were linked by one- or two-VNTR variations . Likewise , all clinical isolates from Guinea 2012 and Togo were linked by one- or two-VNTR divergences . At a country-level scale our analyses revealed several distinct clonal groups , most notably ( 1 ) DRC 2011/2012 , ( 2 ) DRC 2009 , ( 3 ) Zambia 2012 and ( 4 ) Guinea 2012 . We used goeBURST to identify a potential founder MLVA haplotype for each MLVA cluster . Each epidemic complex was characterized by a central founder MLVA haplotype and closely related derivative SLV or DLV MLVA haplotypes , which branched from the founder . In contrast , all six of the typed environmental isolates were found to be singletons , unrelated to the main clinical epidemic isolate clusters ( Fig 2 ) . Our analysis showed that the DRC 2011 and DRC 2012 isolates grouped together as one discrete complex ( Fig 2 ) . During this two-year period in the DRC , cholera was caused by the extensive expansion and diversification from a single MLVA haplotype . The isolate found at the beginning of the 2011/2012 epidemic in Kisangani , Orientale Province in March 2011 was haplotype #67 , which was designated the founder of the DRC 2011/2012 complex . MLVA type #67 and a SLV of this haplotype were the only types identified during the first week of the outbreak . This MLVA cluster then diversified in parallel with the spatiotemporal spread of the epidemic [16] , as the most distant MLVA haplotypes within this cluster were identified in distant provinces in 2012 . These findings correlate with an epidemiological report of the cholera epidemic that struck the DRC in 2011 . This epidemic aggressively diffused from the onset point in Kisangani , Orientale Province across the country in less than 130 days . Strikingly , outbreaks followed the Congo River and quickly reached non-endemic zones in the West that had not experienced an epidemic for approximately 10 years [16] . Kisangani and the Congo River are localized on the detailed map of the DRC ( Fig 1 , lower right ) . Interestingly , the predicted founder of the 2011/2012 DRC epidemic , persisted in the country over the course of several years , as haplotype #67 was represented in isolates collected in the DRC during the 2009 , 2011 , 2012 and 2013 epidemics . Only one DRC isolate ( MLVA type #129 ) collected in 2011 did not belong to the major DRC 2011/2012 MLVA cluster . Instead , haplotype #129 was a SLV of the DRC 2009 haplotype #116 cluster . In stark contrast , the V . cholerae non-O1 strain isolated at the same period from a water sample in Uvira , South Kivu was a genetically unrelated singleton ( MLVA type #40 ) ( Fig 2 ) . Uvira is located on the northern shores of Lake Tanganyika as indicated in Fig 1 ( lower right panel ) . Overall , the panel of DRC 2009 isolates displayed a high level of genetic diversity . In fact , four DRC 2009 clinical isolates ( MLVA types #39 , #43 , #21 and #108 ) collected in February and March 2009 were designated distantly related singletons ( Fig 2 ) . The MST was then analyzed in further detail considering the date of sample isolation . The isolates collected during the first half of the year were highly diverse ( indicated as “DRC 2009 A” in pink; Fig 2 ) . In contrast , 63 of 66 isolates ( 95 . 5% ) collected from July to November of 2009 in Katanga Province formed a tight clonal complex ( indicated as “DRC 2009 B” in red; Fig 2 ) . This bottleneck effect was concomitant with an epidemic rebound in Katanga Province after a complete lull in cholera transmission in May and June 2009 . In July 2009 , cholera first appeared in Kalemie , a city located on the shore of Lake Tanganyika , and then spread throughout the rest of the province ( based on field investigations in the DRC; Renaud Piarroux ) . MLVA types #110 and #116 were designated potential founders of this DRC 2009 B MLVA complex . Interestingly , MLVA type #110 was the first isolate collected on February 5 , 2009 , and isolates harboring this haplotype were collected up to November 20 , 2009 . Notably , MLVA type #110 was also found in August 2009 in Uvira , a city located approximately 360 km north of Kalemie on the shore of Lake Tanganyika ( all sites are labeled in the lower right panel of Fig 1 ) . Therefore , we hypothesize that this strain likely persisted in the region following the early-2009 outbreaks and a subsequently gave rise to the late-2009 DRC epidemic . In Zambia , all clinical isolates from the 2012 epidemic formed a restricted clonal complex , which derived from the predicted founder MLVA #30 . Once more , the two non-O1 environmental isolates collected from the shores of Lake Tanganyika in Mpulungu , Northern Province were singletons ( MLVA haplotype #13 ) unrelated to the clonal complex ( Fig 2 ) . The Guinea 2012 clinical isolates formed a solid clonal complex , with an MST of closely related derivative isolates that stemmed from the founder haplotype ( MLVA type #47 ) ( Fig 2 ) . In a previous study , our group has shown that the Guinea 2012 epidemic appears to be due to the importation of a toxigenic clone from Sierra Leone . Using MLVA typing , we have demonstrated progressive genetic diversification of the strains from the founder type correlated with spatiotemporal epidemic spread [9] . The founding MLVA type was also the first and only MLVA type indentified during the onset of the epidemic , on Kaback Island ( Fig 1 , lower left ) , Guinea in February 2012 [9] . In contrast , the two Guinean environmental strains ( MLVA types #1 and #130 ) isolated from water samples at the site of the initial outbreak ( Kaback ) were unrelated singletons ( Fig 2 ) . Our data showed that the Togo isolates represent a diverse set of MLVA haplotypes , as they were only related to the Guinean isolate MLVA types by a single DLV ( Fig 2 ) . Most of the isolates collected in Togo were designated singletons . To provide statistical power to the observed relationships between 328 clinical isolates , population genetic analyses were performed . The six unrelated environmental isolates and the three isolates from DRC in 2008 were excluded from this analysis . Taking each MLVA VNTR locus in turn , this analysis revealed that there were 5 , 17 , 6 , 6 , 17 and 5 alleles , for the VNTR loci denoted VC1 , VC4 , VC5 , VC9 , LAV6 and VC12 , respectively ( Table 2 ) . The discriminatory power of the six tested VNTRs was calculated via an index of genetic diversity ( IOD ) analysis ( Nei , 1987 ) using GenoDive 2 . 0b25 . The IOD and PCR product size range for each VNTR is outlined in Table 2 . Accordingly , the corrected indices of diversity per locus were 0 . 758 , 0 . 927 , 0 . 54 , 0 . 512 , 0 . 871 and 0 . 579 for VC1 , VC4 , VC5 , VC9 , LAV6 and VCMS12 , respectively . The overall corrected IOD for the six VNTRs combined was 0 . 698 . The two most variable VNTRs were located on the small chromosome , which correlates with the observations reported by Lam et al . [8] ( Table 2 ) . The IOD ( based on Nei , 1987 ) per population was calculated to determine the extent of genetic diversity of each designated population . The epidemics with the highest degree of genetic diversity were DRC 2009 A ( IOD = 0 . 545 ) and DRC 2013 ( IOD = 0 . 586 ) . In contrast , the epidemic isolates derived from Togo 2012 ( IOD = 0 . 152 ) displayed the lowest diversity . Epidemics in DRC 2011 ( IOD = 0 . 218 ) , DRC 2012 ( IOD = 0 . 242 ) , Zambia 2012 ( IOD = 0 . 206 ) and Guinea 2012 ( IOD = 0 . 222 ) also showed relatively low gene diversity ( Table 3 ) . Pairwise differentiation analysis was performed to understand the statistical relationship between the epidemic populations . The Fst values for all pairs of populations were calculated considering all 328 V . cholerae clinical isolates . All Fst and p-values are outlined in Table 4 . The closest statistically significant relationship was between the epidemics of DRC 2011 and DRC 2012 ( Fst = 0 . 125 , p = 0 . 001 ) , which is coherent with the rapid diversification and expansion of an epidemic clone [16] . The DRC 2013 epidemic was closest related to the early DRC 2009 isolates ( Fst = 0 . 107 , p = 0 . 001 ) , which suggests that the 2013 epidemic was likely due to an expansion of strains circulating in the country during pervious epidemics . The pairwise analysis also demonstrated a statistically signification relationship between the early and late 2009 isolates from the DRC ( 2009 A and 2009 B , respectively ) ( Fst = 0 . 276 , p = 0 . 001 ) . These observations further support the hypothesis that the late 2009 epidemic came about following a bottleneck of the early 2009 DRC epidemic isolates .
Overall , our results provide novel insight into the epidemic phenomena of cholera in West and Central Africa . At the sub-regional level , MST analysis revealed two distinct African clusters: ( 1 ) the African Great Lakes Region group , comprising DRC and Zambia isolates , and ( 2 ) the West African cluster with isolates from Togo and Guinea . At the country level , the epidemic V . cholerae populations from DRC 2011/2012 , Zambia 2012 , Guinea 2012 and late-2009 DRC were each designated tight complexes . The expansion of the isolate populations coincided with the progression of each epidemic , as the founder MLVA types corresponded to isolates collected at outbreak onset and more distantly related derivatives represented isolates found later during the epidemic . Analysis of isolates from the DRC revealed that certain strains appear to remain in circulation in the country over a period of several years and eventually engender explosive outbreaks with diversification of founding isolates , as observed in 2011 . This phenomenon is distinct from that observed in Togo , where isolates were grouped into a loosely connected quasi-complex without a founder MLVA type . The Togo isolate results rather indicate that when outbreaks occur , the isolates fail to diversify or diffuse throughout the country . A field assessment of outbreaks in Togo has revealed that the country is vulnerable to importation of cholera cases from neighboring countries , although outbreaks are then quickly extinguished ( UNICEF field investigation ( Sandra Moore ) and personal communication with Dr Adodo Yao Sadji ) . From 2010 to week 48 of 2014 , the country only recorded 551 suspected cholera cases [17–21] . We hypothesize that the Togo isolates rather represent the descendants of a much larger epidemic cluster from a neighboring country experiencing severe cholera epidemics , such as the nearby countries of Ghana and Nigeria . From 2010 to week 48 of 2014 , Ghana reported 48 , 546 suspected cholera cases [17–21] . During that same five-year period , Nigeria signaled a staggering 110 , 904 suspected cases [17–21] . Isolates from these affected countries would have to be analyzed to test this hypothesis . As V . cholerae is autochthonous in the coastal aquatic ecosystem , it has been widely presumed that cholera epidemics are triggered by environmental factors that promote growth of local bacterial reservoirs [22] . However , all six of the environmental isolates collected from a range of countries were genetically unrelated singletons . We acknowledge that additional environmental isolates of V . cholerae should be included in the panel to affirm the relationship ( or lack of ) between clinical outbreak strains and those found in water bodies . Indeed , to examine the diversity of environmental strains , efforts should be made to collect further samples . Nevertheless , our preliminary analysis of the environmental samples correlates with two recent reviews elucidating the environmental determinants of cholera outbreaks in Africa . These reviews by Rebaudet et al . [5 , 6] found that at least 76% of cholera cases in Sub-Saharan Africa occurred in non-coastal regions located over 100 km from the coast . From 2009 to 2011 , annual incidence rates of cholera were three times higher in inland Africa compared with the coastal region . In fact , toxigenic V . cholerae isolates have only been recovered from the environment during an outbreak , when patient-derived contamination of water sources is expected [5 , 6] . Our findings are also consistent with the phylogenic assessment of an extensive panel of seventh pandemic V . cholerae isolates . Mutreja et al . [23] have revealed that a specific V . cholerae El Tor clonal lineage appears to be responsible for the current pandemic . Their study demonstrated that the seventh pandemic is monophyletic and originated from a single ancestral clone that has radiated globally in distinct waves [23] . Lineages of the current pandemic appear to emerge , diversify and eventually become extinct [23] . Notably , we observed a similar phenomenon at a smaller scale with the DRC 2009 , DRC 2011/2012 , Guinea 2012 and Zambia 2012 epidemics . As isolates from the African Great Lakes Region were not included in the seventh pandemic phylogeny , whole-genome sequencing and phylogenic analysis of this panel of African strains would provide even further insight into the mechanisms of cholera epidemics in the region . Together , MLVA and whole-genome sequencing-based phylogeny represent complementary approaches to better understand epidemic dynamics . MLVA is useful to elucidate the short-term microevolution of clonal complexes , while sequence-based phylogeny enables the identification of distant ancestors and related strains at a global level . Concerning the limitations of the study , our findings would be bolstered by increased isolate sampling of several years in these and neighboring affected countries . Indeed , we could not verify that MLVA type #67 isolates found in the DRC persisted in the country throughout the 2010 epidemic , as few isolates were collected in 2010 due to a lack of funding for the epidemiological surveillance and prevention of cholera . Likewise , the analysis of the Togolese epidemics would benefit from additional isolate sampling in neighboring countries . Finally , although environmental V . cholerae samples are difficult to isolate , this study would be enhanced by including additional isolates found in water bodies located in cholera-endemic areas . Further studies should address the detailed mechanisms of cholera in the zones where cholera appears to persist . If the cholera dilemma in Africa can be narrowed down to a few locales , secondary affected areas ( such as perhaps Togo and Guinea ) may be largely protected by targeted interventions in cholera epicenters such as the DRC , Ghana and Nigeria . Therefore , with a clear understanding of cholera dynamics in the region , public health resources would be most effectively and efficiently applied . Overall , our results show that cholera is indeed a regional public health dilemma in Africa . With varying dynamics in each country , certain strains are able to persist in a given region over a period of several years and occasionally spread into non-endemic areas or neighboring countries . Indeed , several elements play a role in cholera epidemics including climate , geography , economy , hygiene , sanitation , access to potable water and population movement , as addressed in the corresponding epidemiological reports ( concerning the Togo field investigation , please see S1 Text ) [9 , 16] . Therefore , public health strategies should be optimized according to the dynamics and scale of cholera epidemics in each region . These findings also demonstrate the importance of monitoring the circulation of the bacterium among human populations , which appear to serve as the principal reservoir of toxigenic V . cholerae . Combined with classical epidemiological investigations , MLVA represents a rapid and discriminatory tool to track outbreak evolution at an epidemic and regional level .
|
Cholera is caused by the toxigenic bacterium Vibrio cholerae . Since cholera was imported into the West African country of Guinea in 1970 , cases have been reported on the continent every year . In Sub-Saharan Africa , cholera occurs in a heterogeneous manner; outbreaks primarily cluster at certain hotspots including the African Great Lakes Region and West Africa . To gain further insight into the mechanisms by which cholera outbreaks emerge and diffuse , we performed genetic analyses of 337 Vibrio cholera isolates from the Democratic Republic of the Congo ( DRC ) , Zambia , Guinea and Togo . Isolates from both patients and environmental samples were examined . Our findings demonstrate the relationship between distinct epidemics in West Africa and the African Great Lakes Region . For example , certain strains in the DRC have circulated in the region over a period of several years , occasionally giving rise to expansive epidemics . We also found that the six environmental isolates in our panel were unrelated to the epidemic isolates . Such insight into the country- and region-specific dynamics of the disease is critical to implement optimized public health strategies to control and prevent cholera epidemics . This study also highlights the importance of analyzing Vibrio cholerae isolates to complement epidemiological studies .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[] |
2015
|
Relationship between Distinct African Cholera Epidemics Revealed via MLVA Haplotyping of 337 Vibrio cholerae Isolates
|
Recent advances in sequencing technologies have enabled the production of massive amounts of data on somatic mutations from cancer genomes . These data have led to the detection of characteristic patterns of somatic mutations or “mutation signatures” at an unprecedented resolution , with the potential for new insights into the causes and mechanisms of tumorigenesis . Here we present new methods for modelling , identifying and visualizing such mutation signatures . Our methods greatly simplify mutation signature models compared with existing approaches , reducing the number of parameters by orders of magnitude even while increasing the contextual factors ( e . g . the number of flanking bases ) that are accounted for . This improves both sensitivity and robustness of inferred signatures . We also provide a new intuitive way to visualize the signatures , analogous to the use of sequence logos to visualize transcription factor binding sites . We illustrate our new method on somatic mutation data from urothelial carcinoma of the upper urinary tract , and a larger dataset from 30 diverse cancer types . The results illustrate several important features of our methods , including the ability of our new visualization tool to clearly highlight the key features of each signature , the improved robustness of signature inferences from small sample sizes , and more detailed inference of signature characteristics such as strand biases and sequence context effects at the base two positions 5′ to the mutated site . The overall framework of our work is based on probabilistic models that are closely connected with “mixed-membership models” which are widely used in population genetic admixture analysis , and in machine learning for document clustering . We argue that recognizing these relationships should help improve understanding of mutation signature extraction problems , and suggests ways to further improve the statistical methods . Our methods are implemented in an R package pmsignature ( https://github . com/friend1ws/pmsignature ) and a web application available at https://friend1ws . shinyapps . io/pmsignature_shiny/ .
Cancer is a genomic disease . As we lead a life , DNA within our cells acquires random somatic mutations , mainly caused by DNA replication errors and exposures to mutagens such as chemical substances , radioactivities and inflammatory reactions . Although most mutations are harmless ( called “passenger mutations” ) , a small portions of mutations at some specific sites in cancer genes ( “driver mutations” ) affect cell growth , causing autonomous proliferation , tissue invasion , and contributing to oncogenesis [1] . Cancer genome studies typically focus on identifying driver mutations , to help understand the mechanism of cancer development . However , passenger mutations can also yield important information , because they often show patterns ( “mutation signatures” ) which can provide insights into the forces that cause somatic mutations . For example , classical studies of mutation patterns revealed that C > A mutations are abundant in lung cancers in patients with smoking history , and these are caused by benzo ( a ) pyrene included in tobacco smoke [2] . Also , C > T and CC > TT mutations are abundant in ultraviolet-light-associated skin cancers , and these are caused by pyrimidine dimers as a result of ultraviolet radiation [3] . The potential for classical studies to yield insights into somatic mutation processes was limited in several ways . Due to limited sequencing throughput , most classical studies focused on a few cancer genes , such as TP53 , where high mutation frequencies could be expected . They then contrasted mutation pattern profiles among different cancer types , aggregating mutations across multiple individuals within the same cancer type to yield sufficient mutations for analysis . However , since many of the mutations in cancer genes are driver mutations causing cell proliferation , the resultant mutation profiles are a biased representation of the underlying mutation process . Furthermore , the paucity of mutation data made it effectively impossible to assess variation in mutation patterns among individuals . Recent advances in high-throughput sequencing provide new opportunities to investigate sample-by-sample mutation signatures in an unbiased way using genome-wide somatic mutation data . For example , a large-scale study using 21 breast cancer samples identified an association of C > [AGT] mutations at TpC sites , which was later proved to be caused by APOBEC protein family [4–6] , and a novel phenomenon called kataegis [7] . Moreover , a landmark study of 7 , 034 primary cancer samples , representing 30 different cancer classes , has provided the first large-scale overview of mutation signatures across a large number of cancer types [8] . This has lead to great hopes that detection of novel mutation signatures and associated mutagens can lead to identification of novel mutagens and prevention of cancer . To make the most of these opportunities requires the development of efficient and effective statistical methods for analyzing mutation signatures in vast amounts of somatic mutation data . Current statistical approaches [9 , 10] are excellent starting points , and have helped generate the new insights noted above . However , we argue here that these existing methods have two important limitations , caused by the fact that they use an unconstrained model for each “mutation signature” . First , although using an unconstrained model might appear to be a good thing in terms of flexibility , in practice it can actually reduce flexibility , because the price of using an unconstrained model is that one must limit the domain of mutation signatures considered . For example , most recent analyses of mutation signatures consider only the immediate flanking 5′ and 3′ bases of each substitution to be part of the signature , even though it is known that more distal bases—and particularly the next flanking base on each side—can contain important contextual information [11] . These recent analyses take this approach because , in the unconstrained model , incorporating the more distal bases into the signature very substantially increases the number of parameters , making estimated mutation signatures unstable . Secondly , and just as important , the unconstrained model means that each signature is a probability distribution in a high-dimensional parameter space , which can make signatures difficult to interpret . In this paper , we present a novel probabilistic approach to mutation signature modelling that addresses these limitations . In brief , we first simplify the modelling of mutation signatures by decomposing them into separate “mutation features” . For example , the substitution type is one feature; flanking bases are each another feature . We then exploit this decomposition by using a probabilistic model for signatures that assumes independence across features . This approach substantially reduces the number of parameters associated with each signature , greatly facilitating the incorporation of additional relevant sequence context . For example , our approach can incorporate the two bases 3′ and 5′ of the substitution and transcription strand biases using only 18 parameters per signature , compared with 3 , 071 parameters per signature with current approaches . We demonstrate the benefits of this simplification in data analyses . These benefits include more stable estimation of signatures from smaller samples , refinement of the detail and resolution of many mutation signatures , and possibly identification of novel signatures . Assuming independence among features in a signature may initially seem unnatural . However , its use here is analogous to “position weight matrix models” which have been highly successful for modelling transcription factor binding motifs . Indeed , an important second contribution of our paper is to provide intuitive visual representations for mutation signatures , analogous to the “sequence logos” used for visualizing binding motifs . Finally , we also highlight the close connection between mutation signature models and the “mixed-membership models” , also known as “admixture models” [12] or “Latent Dirichlet Allocation” models [13] that are widely used in population genetics and document clustering applications . These connections should be helpful for future elaboration of computational and statistical methods for cancer mutation signature detection . Software implementing the proposed methods is available in an R package pmsignature ( probabilistic mutation signature ) , at https://github . com/friend1ws/pmsignature . The core part of the estimation process is implemented in C++ by way of the Rcpp package [14] , which enables handling millions of somatic mutations from thousands of cancer genomes using a standard desktop computer . In addition , a web-based application of our method is available at https://friend1ws . shinyapps . io/pmsignature_shiny/ .
The term “mutation signature” is used to describe a characteristic mutational pattern observed in cancer genomes . Such patterns are often related to carcinogens ( e . g . , frequent C > A mutations in lung cancers with smoking histories ) . What constitutes a mutational pattern varies among papers . The simplest approach is to consider 6 possible mutation patterns , corresponding to 6 possible substitution patterns ( C>A , C>G , C>T , T>A , T>C , T>G; the original base is often fixed to C or T to remove redundancy of taking complementary strands ) . However , in practice we know that DNA context of a substitution is often important , and so it is common to go the next level of complexity , and include the immediate 5′ and 3′ flanking bases in the mutation pattern . This results in 96 ( 6 × 4 × 4 ) patterns . Further incorporating the strand ( plus or minus ) of each substitution extends this to 192 patterns [8 , 9] . Mathematically , mutation signatures have previously been characterized using an unconstrained distribution over mutation patterns [9 , 10] . Thus , if the number of mutation patterns considered is M then each mutation signature is characterized by a probability vector of length M ( which must sum to 1 , so M − 1 parameters ) . A problem with this approach is that it requires a large number of parameters per mutation signature . As noted above , even accounting only for immediately flanking bases gives M = 96 . Furthermore , M increases exponentially if we try to account for additional context: to take account of up to n bases 5′ and 3′ to the mutated site ( henceforth referred to as the −n position and the +n position , respectively ) gives M = 6 × 42n . Having a large number of parameters per signature causes two important problems: i ) estimates of signature parameters can become statistically unstable; ii ) signatures can become difficult to interpret . The first contribution of this paper is to suggest a more parsimonious approach to modelling mutation signatures , with the benefit of producing both more stable estimates and more easily interpretable signatures . In brief , we substantially reduce the number of parameters per signature by breaking each mutation pattern into “features” , and assuming independence across mutation features . For example , consider the case where a mutation pattern is defined by the substitution and its two flanking bases . We break this into three features ( substitution , 3′ base , 5′ base ) , and characterize each mutation signature by a probability distribution for each feature ( which , by our independence assumption , are multiplied together to define a distribution on mutation patterns ) . Since the number of possible values for each feature is 6 , 4 , and 4 respectively this requires 5 + 3 + 3 = 11 parameters instead of 96 − 1 = 95 parameters . Furthermore , extending this model to account for ±n neighboring bases requires only 5 + 6n parameters instead of 6 × 42n − 1 . For example , considering ±2 positions requires 17 parameters instead of 1 , 535 . Finally , incorporating transcription strand as an additional feature adds just one parameter , instead of doubling the number of parameters . Since the aim of a mutation signature is , in some ways , to capture dependencies among features , the independence assumption may seem counter-intuitive . However , the idea is exactly analogous to the use of a “position weight matrix” ( PWM ) to represent motifs in sequence data . In this analogy , a motif is analogous to a mutation signature , and each location in the motif is analogous to a “feature” . Just as we use a probability vector for each feature , a PWM defines a probability vector for each location in a motif , and these probabilities at each location can be multiplied together to yield a probability distribution on sequences . Even though a PWM cannot capture complex higher-order dependencies , some of which likely do exist in practice , it has been a highly successful tool for motif analysis—likely because it can capture the most important characteristics of transcription factor binding sites ( that some locations will show strong preference for a particular base , whereas others will not ) , and also because it can be represented in an easily interpretable way via sequencing logos [15] . For similar reasons—in addition to the empirical demonstrations we present later—we believe our mutation signature representation will prove useful for mutation signature analysis . Fig 1 illustrates the way that our new representation of signatures can simply capture a previously identified signature [9 , 16] and provides an easily interpretable visualization of the signature that is analogous to sequencing logos [15] . We particularly note how the key elements of this mutation signature are more immediately visually apparent than with visualizations of the full vector of probabilities used by existing approaches . Suppose each somatic mutation has L mutation features , m = ( m1 , m2 , ⋯ , mL ) , where each ml can take Ml discrete values . Also , let M: = ( M1 , ⋯ , ML ) . For example , when taking account of 6 substitution patterns and ±2 flanking sites , M = ( 6 , 4 , 4 , 4 , 4 ) . See S1 Table for other examples . Suppose we have observed mutations in I sampled cancer genomes , and let Ji denote the number of observed mutations in the i-th cancer genome . Further , let xi , j = ( xi , j , 1 , ⋯ , xi , j , L ) , ( i = 1 , ⋯ , I , j = 1 , ⋯ , Ji ) denote the observed mutation feature vector for the j-th mutation of i-th cancer genome , where xi , j , l ∈ {1 , ⋯ , Ml} . Our model assumes that each mutation arose from one of K possible mutation signatures . Each cancer sample has its own characteristic proportion of mutations of each signature type ( which might depend on lifestyle , genetic differences , etc . ) . g We let qi , k denote the proportion of signature k in sample i , so qi = ( qi , 1 , qi , 2 , ⋯ , qi , K ) ∈ΔK , ( i = 1 , ⋯ , I ) where Δ S = { ( t 1 , ⋯ , t S ) : t s ≥ 0 ∀ s , ∑ s = 1 S t s = 1 } denotes the S-dimensional simplex of non-negative vectors summing to 1 . Further , each mutation signature is characterized by parameter vectors Fk: = ( fk , 1 , … , fk , L ) , where fk , l is a probability vector for the l-th feature in the k-th signature . That is , fk , l = ( fk , l , 1 , … , fk , l , Ml ) ∈ ΔMl . Our generative model for the observed mutations {xi , j} in each cancer sample can now be described as a two-step process . This generative model is summarized in Fig 2 . This model is essentially a “mixed-membership model” , also known as an “admixture model” [12] or “Latent Dirichlet Allocation” [13] . For example , the membership proportions for each sample are analogous to admixture proportions in an admixture model; the mutation signatures are analogous to populations , and the mutation signature-specific parameters are analogous to population-specific allele frequencies . The key parameters in this model are the membership proportions for each sample , qi , and the mutation signature parameters , Fk . We estimate these parameters by maximizing likelihood using an EM algorithm . A simulation study demonstrates that the estimation method can reproduce the mutation signature very accurately provided enough mutations and samples are available ( see S1 Text ) . See Methods for more detailed models , parameter estimation , further discussion on relationships with mixed membership models , how to select K , etc . The intrinsic composition of genome sequence , if unaccounted for , can undesirably influence estimated mutation signatures . For example , since the di-nucleotide CpG is underrepresented in most genomic regions ( other than promoters ) , a signature with substitutions from a C base can have weaker signals of G base at the +1 position . In previous work [10] , this background problem was dealt by explicitly incorporating “mutation opportunity” coefficients into the model . Here , to reduce the influences of intrinsic sequence composition on our signature estimates , we introduce a special “background signature” { F 0 } ∈ Δ M 1 × ⋯ × M L , which is designed to capture biases in intrinsic genome sequence composition and is calculated from the composition of consecutive nucleotides of the human genome sequence ( See Methods for the detail ) . Here we compare our new “independent model” for mutation signatures , which assumes independence among mutation features , with the “full model” , which corresponds to existing approaches . We compare mutation signatures obtained by the two approaches and investigate the robustness of each approach by down-sampling experiments . The data consist of 14 , 717 somatic substitutions collected from a study of 26 urothelial carcinomas of the upper urinary tract ( UCUT ) [18] . The original study identified a novel mutation signature in these data: T > A substitutions at CpTpG sites with a strong transcription strand specificity caused by aristolochic acids ( AA ) . We consider a mutation pattern to consist of the substitution pattern , the ±2 flanking bases , and the transcription strand direction . Thus each signature is characterized by 18 parameters in our independent model , and by 3 , 071 parameters in the full model . After analyzing the data with various numbers of mutation signatures K , we selected K = 3 signatures for these analyses ( see S2 Text ) . The inferred APOBEC signature under the independent model shows a clear depletion of G base at the −2 position , which is consistent with the previous study [9] and results in the next subsection ( Fig 3A and 3B ) . In contrast , for the full model , this tendency is rather mild ( Figs 3C , 3D , and S1 ) . The inferred AA mutation signature has no clear characteristics at the −2 position . These results suggest that our independent model has the potential to identify signatures in more detail and with less data than existing approaches based on the full model . To investigate this further we performed down-sampling experiments . Using the mutation signatures obtained using all 14 , 717 substitutions as a gold standard , we assessed performance of the proposed method on down-sampled data consisting of r% of the original data , where r = ( 1% , 2 . 5% , 5% , 10% , 25% , 50% ) . To measure robustness we used the cosine similarity on the full dimensional vector space , which allows comparison between the full model and the independent model . We repeated each down-sampling experiment 100 times for each model . The results ( Fig 3E and 3F ) confirm that the results of the independent model are substantially more robust to reductions in data size than the full model . Indeed , mutation signatures inferred using the independent model with only 10% of the data remain highly similar to the signatures inferred from the full data; by comparison the full model shows a much larger drop-off in similarity , especially in the APOBEC signature where even using 50% of the data gives a substantial drop-off in similarity . Both methods found the AA signature easier to recover than the APOBEC signature . We believe that this is because the number of T > A substitutions at GpTpC sites are far more frequent in this dataset . To provide a more comprehensive practical illustration of our method , we applied it to somatic mutation data from 30 cancer types [8] . We applied the method to each cancer type separately to assess similarity of estimated signatures across cancer types . For each cancer type we selected the number of signatures K by fitting the model with increasing K and examining the log-likelihood , bootstrap errors , and correlation of membership parameters . The selected values of K are given in S2 Table . Also , we simply removed somatic mutations located in an intergenic region to include transcription strand biases as mutation features . Finally , we merged similar mutation signatures across different cancer types ( when their Frobenius distances were < 0 . 6 , where the Frobenius distance between two matrices ( F1 , F2 ) is Tr ( ( F 1 - F 2 ) ( F 1 - F 2 ) t ) ( Tr means the trace of square matrices ) . Figs 4 and 5 summarize the results . In total , we identified 27 mutation signatures . Many of these signatures show reassuring similarities with signatures identified in previous studies . However signatures from our independence model , because of its ability to effectively and parsimoniously deal with both ±2 flanking base context and strand bias , are often more refined , highlighting additional details or features not previously evident . By comparing the composition of nucleotides and cancer types exhibiting the signatures with results of previous studies , we were able to associate many of the detected signatures with known mutational processes . In addition , as we reviewed these signatures and compared them with previous work , we noticed connections that , while not directly related to our new model , appear novel and noteworthy . The remainder of this section provides a comprehensive discussion of these findings . Signatures 1 and 8 ( C > A at TpCpT and C > T at TpCpG , respectively ) observed in colorectal and uterine cancers appear likely to be associated with deregulated activity of the error-prone polymerase Pol ϵ . In previous analyses of these data [8] , the signature for Pol ϵ dysfunction was represented by a single signature ( their “signature 10” , see S2 Fig ) . In contrast our new approach uses two signatures . Since these signatures are highly correlated , and appear connected by a single biological mechanism , we certainly do not argue that inferring them as a single signature is “wrong” . However , splitting them into two signatures does help highlight certain features . Specifically , signature 1 shows a transcription strand bias whereas signature 8 does not , and this is true for both colorectal and uterus cancers ( S3C and S3D Fig ) . This strand bias may be connected with the enrichment of C >A at TpCpT mutations in leading strands of replication forks observed by [19] . Although replication strand bias is different from transcription strand bias , these two biases may be connected through the fact that replication origins prefer transcription start sites [20] . These signatures also illustrate the ability of our model to help highlight sequence context effects beyond the immediate flanking bases . Specifically , both signatures 1 and 8 show an elevated frequency of the T base at position −2 , and signature 1 also shows slightly elevated frequency of the T base at position +2 ( Figs 6B , 6C , S3C , and S3D ) . A previous study of Pol ϵ [19] found that a nonsense mutation R23X of TP53 is enriched in cancers with Pol ϵ defects . In fact , the pattern of this mutation is C > T at TpTpCpGpA , closely matching signature 8 . This illustrates that the inclusion of ±2 bases into signatures may be helpful for identifying underlying mechanisms . Signature 2 ( C > A at [CT]pCpT ) is observed solely in low grade gliomas , and appears related to , but slightly different from , the signature previously detected in the same cancer types ( “signature 14” , [9] ) . Indeed , the corresponding signature in the previous study shows very complex patterns ( C > A at NpCpT or C > T at GpTpN ) . Further investigation revealed that this signature is driven by a single sample with an extremely high mutation rate ( see S4A and S4B Fig ) , and signature 2 disappeared when we removed this sample ( S5 Fig ) . It may be that the complex low-grade-glioma specific signature detected in the previous study is driven by the same single sample . We suggest that these signatures should be treated with caution until validated in additional samples . Signature 4 ( C > A at CpCpG ) observed in kidney clear cell carcinomas , lung adenocarcinomas and melanomas seems to correspond to the “signature R2” detected in the same cancer types ( plus lung squamous carcinomas ) in [9] ( see their Supplementary Figures ) . Again our analysis highlights additional contextual information , with a strikingly elevated frequency of base C at the −2 position ( S6A , S6B and S6C Fig ) . However , for each cancer type , only a few samples support this signature ( see S6D , S6E and S6F Fig ) , and the corresponding signature could not be validated in the previous study: most somatic mutations corresponding to that signature could not be validated by re-sequencing or visual inspection of BAM files using genomic viewers . Again , further investigation yielded a potential explanation for this finding: this signature largely matches that of a putative artifact caused by oxidation of DNA during acoustic shearing [21] , and we conclude that this signature , and the corresponding signature in previous work , are likely artefactual . Although not of direct biological interest , identifying artefactual signatures could be helpful in removing false positive mutations . Signature 13 ( T > [AGT] at TpCpN sites ) was observed in 12 cancer types , and is surely related to the activity of the APOBEC family . The 12 inferred signatures were highly consistent among cancer types except for B-cell lymphoma ( see S3A Fig ) , highlighting the robustness of our approach . Almost all of them show enrichment of A and T and depletion of G base at the −2 position ( Figs 6A and S3A ) , consistent with the UCUT data above and previous analyses [9] . The estimated transcribed strand specificities varied among cancer types , suggesting that there is not consistent strand-specificity in APOBEC signatures ( and the observed variation may be due to estimation errors ) . Signatures 15 and 16 may also be related to APOBEC , although the estimated signatures are sufficiently different from 13 that they were not merged into a single cluster by our specified clustering criteria . Signatures 10 , 11 , 12 , 19 and 21 provide further examples of our method refining previously observed signatures , highlighting strand biases and/or sequence context effects , particularly 2 bases upstream of the substitution . Signature 10 ( C >T at [CT]pCpC ) was observed in head and neck cancers and melanomas , and probably relates to ultraviolet light . Consistent strand specificities among the two cancer types ( S3E Fig ) matches previous results [9] , but our analysis additionally highlights elevated abundance of T at the −2 position ( Figs 6D and S3E ) . Signature 11 ( C > T at GpCp[CG] ) appears in small-cell lung cancers and stomach cancers , and seems to be the same as “signature 15” in the previous study , whose function remains unclear . Again our analysis highlights elevated abundance of G at the −2 position ( Figs 6E and S3F ) . Signatures 12 ( C > T at [CG]pCp[CT] ) , 19 ( T > C at GpTpN ) and 21 ( T > [CG] at CpTpT ) observed in pilocytic astrocytomas , stomach cancers and oesophagus cancers , respectively , agree well with those detected in the same cancer types in the previous study [9] . However our analysis again refines these signatures , highlighting a strand bias in all three , and sequence context effects at the −2 position in Signatures 12 and 21 . One signature , Signature 20 , appears not to match any signatures in the previous analysis [9] and represents a potentially novel signature . This signature ( T > C at [AC]pTpN ) is observed in thyroid cancers , and shows a very strong strand specificity , which could be due to transcription-coupled nucleotide excision repairs . This signature may have been too weak for previous methods to detect , perhaps because the mutation ratio of thyroid cancer is low , possibly reflecting improved sensitivity of our more parsimonious model . The remaining signatures largely recapitulate previous results . Signature 3 and 5 ( C > A at NpCpN ) observed in head-and-neck cancers and three types of lung cancers are probably associated with tobacco smoking . The estimated signature in each cancer type shows higher mutation prevalence on the template strand ( S3B Fig ) , which is consistent with the previous study [2 , 9] . Signature 6 ( C > A at NpCp[AT] ) observed in neuroblastomas matches the pattern detected in the same cancer type in the previous study . Signature 7 ( C > T at NpCpG sites ) was observed in 25 out of 30 cancer type , and arguably relates to deamination of 5-methyl-cytosine . Signature 9 ( C > T at NpCp[CT] ) was observed in melanomas and glioblastomas , and is probably associated with a chemotherapy drug , temozolomide . Signature 18 ( T > C at ApTp[AG] ) observed in liver cancers has been shown to be more common in Asian cases than in other ancestries [16] , though the source of this signature is still not clear . In this signature , we observe a very strong strand specificity as shown in [9 , 16] , suggesting a possible role for transcription-coupled nucleotide excision repairs .
In this paper , we presented new methods for inferring and visualizing mutation signatures from multiple cancer samples . The new methods exploit simpler , more parsimonious , models for mutation signatures than existing methods . This improves stability of statistical estimation , and easily allows a wider range of contextual factors ( e . g . more flanking bases ) to be incorporated into the analysis . In addition , we provide a new intuitive way to visualize the inferred signatures . We have also emphasized the connection between mutation signature detection , and the use of mixed-membership models in other fields , particularly admixture analysis and document clustering . This connection naturally raises the possibility of improving the proposed approach by learning from experiences in those other fields . For example , in admixture analysis , [22] found that the use of a correlated prior on allele frequencies improved sensitivity to detect population clusters; this suggests that it might be fruitful to consider a correlated prior distribution on signatures , to allow that some signatures—perhaps in different cancers—may be similar to one another ( though not identical ) . More generally , introducing certain prior distributions or penalty terms , such as sparsity-promoting penalties [23 , 24] and determinantal point process priors [25 , 26] could improve both accuracy and interpretation . Further , as the scale of cancer genome data becomes large , more sophisticated computational approaches for estimating parameters may become necessary . We can potentially borrow a number of computational techniques such as those using EM-algorithm [27 , 28] , sequential quadratic programming [29] , Gibbs sampling [12 , 30] and variational methods [13 , 31 , 32] . Finally , to address the problem of determining the number of signatures , it may be fruitful to extend the framework to the Hierarchical Dirichlet processes [33] . Although we have focused on point substitution mutations in this paper , many other types of mutations occur in cancer genomes , including insertions , deletions , double nucleotides substitutions , structural variations and copy number alterations [34 , 35] . Our framework could incorporate these additional mutation types , by summarizing them using appropriate mutation features . In some cases , choice of appropriate features may need investigation . For example , longer deletions could be represented by the length of deletion and the adjacent bases; for short deletions ( a few bases ) it may be fruitful to include the actual deleted bases as part of the features . We have detected a number of mutation signatures having transcription strand biases , which are naturally considered to be associated with transcription activities . Therefore , to further understand the effect of transcription activities on mutational mechanisms , we can include gene expression or RNA polymerase II occupancies to mutation features , so that the relationships of strand biases and transcription activities will be clarified . Also , it may be interesting to devise a probabilistic model for mutation signatures somewhere between complete independence and non-independence assumption , for example , using ideas analogous to those in [36] that uses a Markovian structure for transcription factor binding sites . This may help improve the modelling flexibility of mutation signatures while keeping the number of parameters moderate . Although we believe our new methods already provide useful gains compared with existing approaches , the methods are perhaps even more important for their future potential to incorporate other contextual data , including epigenetic data , into mutation signature analysis . This is important , because local mutation densities are closely related to a number of genomic and epigenetic factors , such as GC content , repeat sequences , chromatin accessibility and modifications , and replication timing [37–40] . A recent study found that epigenetic information in the cell types of origin of the corresponding tumors is the most predictive [41] for local mutation densities . A growing range of epigenetic data from many cell types are now available , and it will be interesting to integrate these epigenetic factors into mutation signature analysis to help understand how these epigenetic factors influence DNA damage and repair mechanisms . Our work here provides a straightforward way to do this: epigenetic data can be simply added as features to the mutation signature . This has the potential to improve accuracy of signature detection ( e . g . S7 Fig ) , and to produce novel biological insights . We believe that the value and impact of our work , and specifically our proposed approach to modelling mutation signatures via independent features , will grow as more and more features are incorporated into the analysis .
The parameters {fk , l} and {qi} must be estimated from the available mutation data {xi , j} . Here we adopt a simple approach that uses an EM-algorithm to maximize the likelihood . Let gi , m denote the number of mutations in the i-th sample that have mutation feature vector m . In the E step of the EM algorithm , we calculate values of auxiliary variables θi , k , m defined as θ i , k , m = q i , k ∏ l = 1 L f k , l , m l ∑ k ′ = 1 K q i , k ′ ∏ l = 1 L f k ′ , l , m l . ( 2 ) Then , in the M-step , we update the parameters {fk , l} and {qi , k} as f k , l , p = ∑ m : m l = p g i , m θ i , k , m ∑ p ′ ∑ m : m l = p ′ g i , m θ i , k , m , ( 3 ) q i , k = ∑ m g i , m θ i , k , m ∑ k ′ ∑ m g i , m θ i , k ′ , m . ( 4 ) We use the R package SQUAREM [42] to accelerate convergence of this EM algorithm ( SQUAREM implements a general approach to accelerate the convergence of any fixed-point iterative scheme such as an EM algorithm ) . To address potential problems with convergence to local minima , we apply the EM algorithm several times ( 10 times in this paper ) using different initial points , and use the estimate with the largest log-likelihood . See S3 Text for the derivation of the above updating procedures . Here , we describe how the background mutation signature is obtained in the case where mutation features are the substitution patterns , the ±2 flanking bases , and the transcription strand . Since the majority of the data used in this paper is exome sequencing data , and since we consider transcription strand as a mutation feature , we use the exonic regions of the human genome reference sequence to obtain the background mutation signature . First we calculate the frequencies of 5-mers with transcription strands of the corresponding exon , where we take complement sequences and flip the strand for those whose central bases are A or G . Then , assuming alternated bases are equally likely from each central base C and T , the frequency of each mutation feature is derived directly from those of the 5-mers and transcription strands . Finally , the probability of each mutation feature is derived by normalizing each frequency to sum to one . We use the non-parametric bootstrap [43] to calculate standard errors for parameter estimates . This involves resampling somatic mutations from the empirical distribution of the original data {xi , j} for each cancer genome . For each of 100 such bootstrap samples , we re-fitted the model , using parameters obtained for the original data as initial points . We then used sample standard errors of the inferred mutational signatures as estimates of parameter standard errors . Determining an appropriate number of mutation signatures K is a challenging task . One approach is to utilize some statistical information criteria such as AIC [44] or BIC [45] . In the population structure problems , for example , the Bayesian deviance [12] and cross-validation [46] have been suggested . One previous study on mutation signature problems [10] utilized BIC . On the other hand , the problem of using these statistical information criteria is that most of them are based on the likelihood , where slight deviations between the specified probabilistic models and the reality sometimes leads to additional ( possibly spurious ) mutation signatures being selected to compensate for those deviations . In this paper , instead of utilizing a statistical information criteria , we adopt the following strategy: These strategies are not claimed as optimal , but appeared to provide satisfactory results in our applications here . The development of automated and practical approaches for choosing K is a possible area for future development . Previous approaches to mutation signature modelling in [8 , 9] are a special case of our framework . Specifically , they correspond to combining all possible combinations of mutation features into a single “meta-feature” , which takes M1 × M2 × … × ML possible values . Thus , instead of having L features with M = ( M1 , … , ML ) , existing approaches have one feature with M = ( M1 × … × ML ) ( see S1 Table ) . The resulting model allows for arbitrary distributions on the M1 × … × ML feature space , and we call the resulting model the “full model” . The full model can represent complicated dependencies in a single signature . For example , a situation where C > A is frequent at ApCpG sites and C > T is frequent at TpCpA sites could be represented with one signature . This may be desirable in some settings and not in others . However , when many mutation contextual factors are taken into account and the number of free parameters becomes huge , estimated results can be unstable and unreliable . Furthermore , there is a risk of over-interpreting the complex features of estimated signatures . Our model is closely related to mixed-membership models that have been adopted in other applications , such as document classification and population structure inference problems . In this subsection , we outline these relationships , slightly abusing notation to contrast the relationships . In the topic model [13 , 27] , which are a form of mixed-membership models frequently used in document classification problems , each document is assumed to have K different “topics” in varying proportions ( qi ∈ ΔK ) , where each topic is characterized by a word frequency ( a multinomial distribution on a set of words W ( fk ∈ ΔW ) . And each word is assumed to be generated by one of K multinomial distributions ( topics ) . The detailed generative process of the j-th word in the i-th document xi , j is: Actually our “full model” ( L = 1 ) is essentially the same as a topic model . On the other hand , in population structure inference problems [12 , 47] , each individual is assumed to be an admixture of K ancestries in varying proportions , where each ancestry is characterized by the allele frequency at each SNP locus . Each SNP genotype of an individual is assumed to be generated by the two step model: first , an ancestry ( “population” ) is chosen according to the admixture proportion for each individual , and then the SNP genotype is generated according to the allele frequency of the selected ancestry at that locus . The relationships among the mutation signature models , topic models and population structure models are summarized in Table 1 . As pointed out by [48] , there is a close relationship between mixed-membership models and nonnegative matrix factorization , which has been successfully used in the previous studies for mutational signature problems [7–9] . In fact , the proposed method can be seen as non-negative matrix factorization with additional restrictions . See S4 Text for details of the relationship between the proposed approach and nonnegative matrix factorization .
|
Somatic ( non-inherited ) mutations are acquired throughout our lives in cells throughout our body . These mutations can be caused , for example , by DNA replication errors or exposure to environmental mutagens such as tobacco smoke . Some of these mutations can lead to cancer . Different cancers , and even different instances of the same cancer , can show different distinctive patterns of somatic mutations . These distinctive patterns have become known as “mutation signatures” . For example , C > A mutations are frequent in lung caners whereas C > T and CC > TT mutations are frequent in skin cancers . Each mutation signature may be associated with a specific kind of carcinogen , such as tobacco smoke or ultraviolet light . Identifying mutation signatures therefore has the potential to identify new carcinogens , and yield new insights into the mechanisms and causes of cancer , In this paper , we introduce new statistical tools for tackling this important problem . These tools provide more robust and interpretable mutation signatures compared to previous approaches , as we demonstrate by applying them to large-scale cancer genomic data .
|
[
"Abstract",
"Introduction",
"Result",
"Discussion",
"Methods"
] |
[] |
2015
|
A Simple Model-Based Approach to Inferring and Visualizing Cancer Mutation Signatures
|
The galactosaminogalactan ( GAG ) is a cell wall component of Aspergillus fumigatus that has potent anti-inflammatory effects in mice . However , the mechanisms responsible for the anti-inflammatory property of GAG remain to be elucidated . In the present study we used in vitro PBMC stimulation assays to demonstrate , that GAG inhibits proinflammatory T-helper ( Th ) 1 and Th17 cytokine production in human PBMCs by inducing Interleukin-1 receptor antagonist ( IL-1Ra ) , a potent anti-inflammatory cytokine that blocks IL-1 signalling . GAG cannot suppress human T-helper cytokine production in the presence of neutralizing antibodies against IL-1Ra . In a mouse model of invasive aspergillosis , GAG induces IL-1Ra in vivo , and the increased susceptibility to invasive aspergillosis in the presence of GAG in wild type mice is not observed in mice deficient for IL-1Ra . Additionally , we demonstrate that the capacity of GAG to induce IL-1Ra could also be used for treatment of inflammatory diseases , as GAG was able to reduce severity of an experimental model of allergic aspergillosis , and in a murine DSS-induced colitis model . In the setting of invasive aspergillosis , GAG has a significant immunomodulatory function by inducing IL-1Ra and notably IL-1Ra knockout mice are completely protected to invasive pulmonary aspergillosis . This opens new treatment strategies that target IL-1Ra in the setting of acute invasive fungal infection . However , the observation that GAG can also protect mice from allergy and colitis makes GAG or a derivative structure of GAG a potential treatment compound for IL-1 driven inflammatory diseases .
Aspergillus fumigatus is an opportunistic fungus that causes infections under specific conditions , of which secondary immunodeficiency is by far the largest risk factor for the development of invasive infections [1] . In order to initiate an effective host response against Aspergillus , recognition of conserved pathogen associated molecular patterns ( PAMPs ) by specific pattern recognition receptors ( PRRs ) is required . A . fumigatus has a complex cell wall consisting of polysaccharides that play essential biological functions in fungal cell biology and host-pathogen interactions . Some of these polysaccharides are recognized by various PRRs expressed on human immune cells [2] . However , A . fumigatus employs various strategies to evade immune recognition . Aspergillus expresses surface molecules that shield PAMPs or can modulate TLR responses [3] . Several surface molecules and PAMPs of A . fumigatus have been characterized as being capable of modulating or suppressing the immune response . Rodlets and melanin , that are present on the conidial surface , shield PAMPs that elicit pro-inflammatory host responses [4] , [5] . In addition , β-glucan , α-glucan and galactomannan ( GM ) have been shown to modulate the host immune response [6] . Another cell wall component of A . fumigatus that is capable of modulating the immune response is galactosaminogalactan ( GAG ) [7] . GAG is not expressed on Aspergillus conidia , but is exposed when conidia start to germinate and was found to be present in the extracellular matrix that surrounds Aspergillus hyphae in aspergilloma isolated from patients and in experimental murine invasive aspergillosis [8] . Furthermore , GAG has been shown to serve as an adhesin of Aspergillus [9] , [10] and to shield β-glucan moieties on the cell wall [9] . This polysaccharide that is shed into the host environment during Aspergillus vegetative growth induces immunosuppressive effects that results in diminished neutrophil recruitment , which predisposes mice to A . fumigatus infection [7] . However , the mechanism through which GAG induces immunosuppressive effects as well as its capacity to induce similar immunosuppressive effects on the human immune response were unknown . Therefore , we investigated whether GAG can be immunosuppressive in the human host response against A . fumigatus , and we have systematically addressed the possible mechanisms responsible for the anti-inflammatory property of GAG .
To investigate whether GAG can exert immunomodulatory effects in humans , we tested whether GAG induces the production of pro- and/or anti-inflammatory cytokines in human PBMCs . GAG did not induce the proinflammatory cytokines TNFα , IL-6 , IL-8 , IFN-γ , IL-17 , IL-5 and IL-9 ( Figure 1A ) , neither did it induce the anti-inflammatory cytokine IL-10 ( Figure 1A ) . To determine whether GAG modulates Aspergillus-induced innate monocyte-derived cytokines , PBMCs were stimulated for 24 hours with Aspergillus conidia ( these morphotypes of A . fumigatus were selected because they do not contain GAG that would interfere with the study of the immunological function of GAG ) in the presence or absence of GAG . The presence of GAG did not have a significant effect on the production of the innate cytokines TNFα and IL-6 , or the anti-inflammatory cytokine IL-10 ( Figure 1B ) . However , when the production of the characteristic T-helper cytokines IL-17 , IL-22 and IFN-γ induced by A . fumigatus was investigated , the IL-17 and IL-22 responses were significantly reduced in the presence of GAG ( Figure 1C ) . To determine whether the effects of GAG are specific for Aspergillus-driven T-helper ( Th ) responses , or whether GAG has a general ability to modulate human Th responses , the effects of GAG on cytokine-driven Th responses were studied . GAG significantly decreased the proinflammatory Th cytokine production induced by the cytokine combinations IL-1β/IL-23 and IL-12/IL-18 that induce IL-17/IL-22 and IFN-γ , respectively ( Figure 1E ) . Thus , GAG can inhibit human proinflammatory Th cytokine production induced by Aspergillus and cytokine cocktails . Human Th cytokine responses such as IL-17 and IL-22 production are highly dependent on the IL-1 receptor pathway [11] , [12] . To investigate whether the observed modulation of Th cytokines by GAG was due to an interaction of GAG with the IL-1 pathway , we determined the capacity of GAG conditioned medium ( culture supernatants of PBMCs that were exposed to 10 µg/ml GAG for 24 hours ) to reduce IL-1β bioactivity . Indeed it was shown that GAG significantly reduced the bioactivity of IL-1β while culture supernatants of unstimulated PBMCs did not ( Figure 2A ) . Bioactivity of the IL-1 signalling pathway is dependent on IL-1 receptor agonists ( IL-1α and IL-1β ) and IL-1 receptor antagonists [13] . One of the natural inhibitors of the IL-1 signalling is the interleukin-1 receptor antagonist ( IL-1Ra ) ; therefore the ability of GAG to induce IL-1Ra was tested . IL-1Ra concentrations in the supernatant of the cells stimulated with GAG were significantly increased , whereas GAG induced none of the IL-1 receptor agonists , IL-1α or IL-1β ( Figure 2B ) , showing that GAG has the capacity to modulate immune responses by blocking the IL-1 receptor pathway . To demonstrate that IL-17 , IL-22 , and IFN-γ production by human PBMCs is indeed dependent on the IL-1 receptor pathway and that IL-1Ra can inhibit the production of these Th cytokines , Th1 and Th17 inducing stimuli were studied in the presence or absence of IL-1Ra . Addition of IL-1Ra reduced IL-17 , IL-22 , and IFN-γ induction by both Aspergillus conidia and by IL-1/IL-23 and IL-12/IL-18 cytokine combinations ( Figure 3A ) . To determine whether the immunosuppressive effect of GAG was mediated through the induction of IL-1Ra , PBMCs were stimulated with IL-1β/IL-23 and GAG in the presence or absence of neutralizing anti-IL-1Ra antibodies . GAG reduced IL-17 and IL-22 levels significantly , which was not observed in the presence of neutralizing anti-IL-1Ra antibodies , demonstrating that the inhibitory effects of GAG on Th cytokine production are dependent on IL-1Ra ( Figure 3B ) . The in vitro stimulations described above suggest that the immunomodulatory effects of GAG are due to inhibition of IL-1 bioactivity by inducing IL-1Ra . To assess whether this has relevant consequences in vivo , we measured IL-1Ra transcription in the lungs of mice infected with Aspergillus with or without the administration of GAG . Induction of Il1ra was increased in the presence of GAG ( Figure 4A ) . To determine which cells are responsible for the induction of Il1ra , we isolated macrophages , neutrophils and epithelial cells from the lungs of naïve mice . Macrophages and neutrophils , but not epithelial cells , expressed Il1ra after stimulation with Aspergillus in the presence of GAG ( Figure 4B ) . Interestingly , not all microbiological stimuli can prime for increased GAG-induced Il1ra , since pre-stimulation with LPS did not increase Il1ra induction by GAG ( Figure 4B ) . To investigate the significance of IL-1Ra in vivo and to determine whether the effects induced by GAG are dependent on IL-1Ra , we studied the effects of GAG in wild type ( WT ) and Il1ra−/− mice with invasive aspergillosis . Il1ra−/− mice were highly resistant to invasive aspergillosis , as indicated by long-term survival ( Figure 4C ) and reduced fungal burden ( Figure 4D ) . Administration of GAG resulted in increased protein levels of IL-1Ra in the lungs of wild-type mice during infection ( Figure 4E ) . In line with previous observations , GAG increased the susceptibility to invasive aspergillosis in WT mice but not in Il1ra−/− mice ( Figure 4D ) . Il1ra−/− mice had increased Mpo expression ( Figure 4F ) and PMN influx in their respiratory tract ( Fig . 4G ) . As expected , administration of GAG reduced inflammatory PMN recruitment in WT but not in Il1ra−/− mice ( Figure 4G ) . These data demonstrate that IL-1Ra has an important role in invasive aspergillosis , and support the concept that the induction of IL-1Ra by GAG may have important clinical consequences . In ABPA , GAG administration decreased PMN recruitment , but not eosinophilic infiltration in the BAL and lung of allergic mice ( Figure 4H ) , a finding consistent with decreased Th17 but not Th2 cell responses in the draining lymph nodes ( Figure 4I ) . To address whether GAG would have similar effects on human Th2 responses , PBMCs isolated form healthy subjects were pre-incubated for 1 h with GAG and subsequently stimulated with Aspergillus conidia for 7 days . Similar to mice , IL-17 production decreased in the presence of GAG , but not Th2 cytokines such as IL-5 and IL-13 ( Figure 4J ) . Thus , GAG has the potential to ameliorate Th17-dependent immunopathology in ABPA . Since IL-1Ra treatment can be beneficial for autoinflammatory diseases such as chronic granulomatous disease CGD colitis in humans [14] , we investigated whether GAG could be beneficial in experimental DSS-induced colitis in mice with chronic granulomatous disease ( CGD ) . The administration of GAG resulted in the amelioration of clinical signs of colitis ( weight loss and stool consistency ) ( Figure 5A-B ) and of inflammatory lesions ( Figure 5C ) in both wild-type and CGD mice . However the protective effects of GAG were most apparent in CGD mice . GAG induced IL-1Ra and , consistently , reduced IL-1β and IL-17 ( Figure 5D ) . Concomitantly , there was an increased production of IL-10 , an anti-inflammatory cytokine that plays an important role in the protection of colitis [15] , [16] . The effects of GAG on CGD colitis were similar to those of IL-1Ra administration ( unpublished data ) .
In the original report describing GAG [7] , it was shown that GAG has anti-inflammatory effects in mice . However , the mechanism through which GAG elicits its immunomodulatory effects remained a question at that time . In the present study , we demonstrate that GAG induces its anti-inflammatory effects by inducing the potent anti-inflammatory cytokine IL-1 receptor antagonist . IL-1Ra can inhibit the activation of the IL-1 pathway by binding to the IL-1R type 1 receptor and prevents recruitment of the IL-1R accessory protein that is required for signalling . It has been repeatedly shown that IL-1 is an essential proinflammatory cytokine of the innate immunity . A deficient IL-1 pathway is also detrimental for the host , since it is an important protective pathway required to fight infection [17] . Thus the IL-1 axis is a potent pro-inflammatory pathway that needs to be tightly regulated , and IL-1Ra is a crucial player in this regulation . Therefore , it is rather surprising that the role of IL-1Ra in invasive fungal infection has not been studied in detail to date . We observed that the absence of IL-1Ra completely protects mice from developing invasive pulmonary aspergillosis , underscoring the importance of the IL-1 pathway in clearance of an acute invasive Aspergillus infection . The observation that GAG induces IL-1Ra in vivo identifies GAG as a potent anti-inflammatory molecule that suppresses the IL-1 pathway , subsequently resulting in increased susceptibility to invasive aspergillosis . The relevance of the IL-1 pathway in aspergillosis is underscored by the fact that polymorphisms IL-1 gene cluster polymorphisms are associated with susceptibility to develop in invasive pulmonary aspergillosis [18] , and that dectin-1 knockout mice display increased fungal burden and mortality during invasive aspergillosis , which is dependent on IL-1 [19] . One of the major risk factors that increases susceptibility to invasive aspergillosis is neutropenia [20] , and neutrophils are crucial for clearing invasive germinating and hyphal forms of Aspergillus infection [21] . GAG has been shown to inhibit neutrophil recruitment to the lung , which is at least partly due to neutrophil apoptosis [7] . We observed that in the presence of GAG , IL-1Ra increased during invasive aspergillosis , which correlated with decreased PMN recruitment , and therefore increased fungal burden . In contrast , Il1ra−/− mice displayed increased neutrophil influx when exposed to Aspergillus , which could explain the resistance of Il1ra−/− mice to invasive aspergillosis , since they can rapidly and efficiently clear Aspergillus conidia due to their increased neutrophil response . In addition to the induction of IL-1Ra by GAG in vitro and in vivo , we observed that the inhibitory effects of GAG on the proinflammatory Th cytokine response in human PBMCs could be restored in the presence of a neutralizing antibody against human IL-1Ra . Furthermore , the increased susceptibility to invasive aspergillosis induced by GAG is not observed in Il1ra−/− mice . These observations strengthen the hypothesis that the anti-inflammatory properties of GAG are dependent on IL-1Ra . The anti-inflammatory properties of GAG were present at a concentration of 10 µg/ml , which is a relevant concentration in vivo , since Aspergillus can secrete GAG in a concentration of 50 µg/ml ( data not shown ) . The finding that antibodies against GAG are present in human serum [8] suggests that there is an adequate exposure of GAG to trigger the immune system . In the present study we were also able to demonstrate that these antibodies do not inhibit the effect of GAG , since we observed significant effects of GAG on IL-1Ra induction and inhibition of IL-17 in the presence of human serum that contained measurable concentrations of antibodies against GAG ( data not shown ) . The relevance of GAG is highlighted by its presence in the extracellular matrix in aspergilloma resected from patients and mice with aspergillosis [8] . It is therefore expected that GAG plays a role in the immunological synapse between host immune cell and the mycelium , not only by inducing anti-inflammatory responses through IL-1Ra but also by shielding β-glucan from recognition , which has been proposed previously [9] . It must be taken into account that in the setting of chronic inflammation in which neutrophils and increased Th17 responses are detrimental for the host , IL-1Ra plays a protective role , due to its significant capability to suppress the IL-1 signaling pathway . This hypothesis is in line with the observation that Il1ra−/− mice develop spontaneous destructive arthritis that is IL-1 and Th17 dependent [22] . The importance of IL-1Ra in controlling IL-1 mediated proinflammatory responses in humans is underlined by a disease called deficiency of IL-1Ra ( DIRA ) . This disease is characterized by the absence of IL-1Ra and severe Th17 mediated responses with neutrophil influx in the skin and bones of these patients , subsequently resulting in severe skin inflammation and osteomyelitis [23] . Therefore , the timing of IL-1Ra induction is of utmost importance to protect the host from infection and overwhelming inflammation . Chronic allergic aspergillosis is associated with excessive inflammation , with increased production of IL-1 and IL-22 [24] . We have demonstrated in the present study that administration of GAG induces IL-1Ra and is able to decrease IL-22 production . Therefore we investigated the effect of GAG in a murine model of allergic bronchopulmonary aspergillosis ( ABPA ) . We observed that the administration of GAG reduces the amount of neutrophils , but not eosinophils in ABPA . Additionally , Th17 responses were downregulated , but not Th2 responses . It is therefore tempting to speculate that administration of GAG can be beneficial in the setting of chronic allergic inflammation that is associated with excessive neutrophil-driven inflammation by reducing Th17 dependent pathology by inhibiting the IL-1 pathway . In addition , GAG also protected CGD mice from experimental colitis . Therefore , we envisage a model in which GAG on the one hand might be detrimental for the host in the setting of an acute infection , and on the other hand could be beneficial for the host during chronic inflammation driven by IL-1 . Next to the identification of GAG or IL-1Ra as a therapeutic target for invasive aspergillosis , it is the first time that a polysaccharide produced by a human pathogen has been identified as an inducer of IL-1Ra by cells of the innate immunity without inducing proinflammatory responses , and which has been demonstrated to have therapeutic capacity in IL-1 mediated disease . The search of the sensing and signal transduction cascade activated by this polysaccharide will now be the center of future research . The data presented here brings new questions into light and opens opportunities for future research . First , one of the most interesting observations is the complete protection of IL-1Ra knockout mice to invasive pulmonary aspergillosis . This opens new treatment strategies that target IL-1Ra in the setting of an acute invasive fungal infection . Second , the significant induction of IL-1Ra by GAG makes GAG or a derivative structure of GAG a potential treatment compound for IL-1–mediated diseases , such as joint , bone and muscle diseases and even very common inflammatory diseases such as diabetes and gout [25] . Previously , we have shown that mitogenic stimulation of monocyte derived macrophages and lymphocytes by αCD3/αCD28 coated beads , or recombinant cytokine-induced IL-17 and IFN-γ production is inhibited in the presence of live A . fumigatus [26] . Although these changes in cytokine responses were attributed to changes in tryptophan and kynurenine , it is tempting to speculate that GAG secretion by live A . fumigatus could have attributed to the decreased IL-17 production . In conclusion , our results demonstrate that GAG has potent anti-inflammatory effects in mice and humans that can be explained by the capability of GAG to induce IL-1Ra . These observations help to explain one of the immune-evasive mechanisms of A . fumigatus . Moreover , inhibition of GAG or IL-1Ra might prove beneficial in the treatment of acute invasive pulmonary aspergillosis , and GAG might be exploited for treatment of IL-1–mediated inflammatory diseases .
All studies with human blood samples were conducted in the Radboud University Nijmegen Medical Centre and the use of healthy volunteers was approved by the institutional ethics review board . Peripheral venous blood samples from healthy volunteers were obtained after written informed consent was provided . All animal studies were conducted within the University of Perugia and were performed according to the Italian Approved Animal Welfare Assurance A–3143–01 . Legislative decree 245/2011-B regarding the animal license was obtained by the Italian Ministry of Health lasting for three years ( 2011–2014 ) . Infections were performed under avertin anesthesia and all efforts were made to minimize suffering . Galactosaminogalactan ( GAG ) was isolated from A . fumigatus culture supernatant and purified from the urea-soluble fraction as previously described [7] . Lyophilized GAG was resuspended in 10 mM HCl at 2 mg/ml and used in a final concentration of 10 µg/ml . Before using GAG in stimulation experiments it was incubated with polymixin B to neutralize potential contamination of lipopolysaccharide . A clinical isolate of Aspergillus fumigatus V05-27 , which has been previously characterized was used for stimulations [27] . Conidia and hyphae were prepared and heat-killed as previously described [6] . A concentration of 1×107/ml was used in the experiments , unless otherwise indicated . Recombinant human IL-1β , IL-23 , IL-12 and IL-18 were purchased from R&D Systems ( Minneapolis , MN , USA ) and were used in end concentrations of 100 ng/ml , 50 ng/ml , 10 ng/ml and 50 ng/ml respectively . Recombinant human ( rh ) IL-1Ra ( Amgen , Inc . , Thousand Oaks , CA , USA ) was used to antagonize IL-1β signalling at a final concentration of 10 ng/ml . Anti-humanIL-1Ra ( R&D ) was used to block IL-1Ra in a final concentration of 10 µg/ml , and was compared to isotype control . PBMCs were isolated as described previously [28] . The cells were counted using a particle counter ( Beckmann Coulter , Woerden , The Netherlands ) and the cell number was adjusted to 5×106/ml . PBMCs were plated in 96-well round-bottom plates ( Corning , NY , USA ) at a final concentration of 2 , 5×106/ml and in a total volume of 200 μl . Cells were pre-stimulated for 1 hour with medium or 10 µg/ml GAG . Following prestimulation , the PBMCs were stimulated with culture medium , heat killed A . fumigatus conidia ( 1×107/ml ) , IL-1β/IL-23 ( 100 and 50 ng/ml respectively ) or IL-12/IL-18 ( 10 and 50 ng/ml respectively ) . These experiments were also performed in the presence of 10 μg/ml anti-IL-1Ra antibody or isotype control . Plates were incubated at 37°C , 5% CO2 for 24 hours , 48 hours or 7 days . 7 day cultures were supplemented with 10% human serum . In this serum we detected anti-GAG antibodies as described previously [7] . After incubation , culture supernatants were collected and stored at -20°C until cytokine measurements were performed . The murine cell line NOB-1 responds to both human or mouse IL-1 by production of IL-2 , furthermore these cells are unresponsive to other cytokines like tumor necrosis factor ( TNF ) , colony stimulating factors ( CSFS ) , IL-3 , IL-5 , IL-6 and IFN-γ [29] . NOB-1 cells were plated in 96 well flat-bottom plates at a final density of 1×106 cells/ml and were stimulated for 24 hour using culture supernatants of unstimulated PBMCs or PBMCs stimulated in presence of GAG ( GAG conditioned medium ) . After 24 hours of incubation at 37°C , 5% CO2 the culture supernatants of the NOB-1 cells were collected and IL-2 production by the NOB-1 cells was measured by ELISA ( R&D systems ) . Cytokines were measured using commercially available ELISAs ( R&D Systems ) ( Biolegend , San Diego , CA , USA ) ( Sanquin , Amsterdam , The Netherlands ) according to the protocols supplied by the manufacturer . IL-1α , IL-1β , TNF-α , IL-6 , IL-8 , IL-1Ra and IL-10 were measured in culture supernatants of 24 hour cultures , and IL-5 , IL-9 , IL-13 , IL-17 , IL-22 and IFN-γ were measured in culture supernatants of 7 day cultures . Female , 8- to 10-weeks old , BALB/c ( wild-type , WT ) mice were purchased from Charles River ( Calco , Italy ) . Breeding pairs of homozygous Il1ra−/− mice on the BALB/c background , were kept under specific-pathogen free conditions at the breeding facilities of the University of Perugia , Perugia , Italy . Experiments were performed according to the Italian Approved Animal Welfare Assurance 229-2011-B . Viable conidia from the A . fumigatus Af293 strain were obtained as described [30] . For infection mice were anesthetized in a small plastic cage , containing 3% Isofluoran ( Isofluran Forene Abbot Scandinavia AB , Solna ) before intranasal ( i . n . ) instillation of a suspension of 2×107 resting conidia/20 µl saline . Mice were treated with 250 μg/kg i . n . of GAG the day of infection and on days 1 to 3 post infection . Mice were monitored for survival , fungal growth ( colony forming unit/organ , mean ± SEM ) , as described [31] , histopathology , myeloperoxidase ( Mpo ) and Il1ra mRNA expression in lung cells and IL-1Ra production . For histology , sections ( 3–4 ìm ) of paraffin-embedded tissues were stained with periodic acid-Schiff ( PAS ) reagent . For allergy , mice received an i . p . and s . c . injection of 100 µg of A . fumigatus culture filtrate extract ( CCFA ) dissolved in incomplete Freund's adjuvant ( Sigma ) followed by two consecutive intranasal injections ( a week apart ) of 20 µg CCFA . A week after the last intranasal challenge , mice received 107 Aspergillus resting conidia and evaluated a week later ( 16424201 ) . GAG ( 250 μg/kg i . n . ) or vehicle alone was administered daily , for a week , in concomitance with the Aspergillus infection . Lungs were filled thoroughly with 1 ml aliquots of pyrogen-free saline through a 22-gauge bead-tipped feeding needle introduced into the trachea . The lavage fluid was collected in a plastic tube on ice and centrifuged at 400 g , 4°C , for 5 min . For differential BAL cell counts , cytospin preparations were made and stained with May- Grünwald Giemsa reagents ( Sigma-Aldrich ) . At least 200 cells per cytospin preparation were counted and the absolute number of each cell type was calculated . Photographs were observed using a BX51 microscope ( Olympus , Milan , Italy ) and images were captured using a high-resolution DP71 camera ( Olympus ) . Mice received either regular drinking water ( control ) or 2 . 5% dextran sulfate sodium ( DSS ) in drinking water for 7 days and then allowed to recover by drinking water alone for an additional 7 days . GAG was given intraperitoneally ( 1 mg/kg ) daily for a week . Weight changes were recorded daily , and the day after the 7-days of rest mice were killed and tissues were collected for histology and cytokine analysis . Colonic sections were stained with H&E [32] . To assess colitis severity , stool and histological scores were used that recently were introduced and proven sensitive to experimental therapy [33] . Purified peritoneal CD11b+ Gr-1+ polymorphonuclear neutrophils ( PMNs ) ( >98% pure on FACS analysis ) were obtained as described [34] . Lung epithelial cells were isolated as described [35] Murine macrophages were isolated from total lung cells after 2 hours plastic adherence at 37°C . PMNs , epithelial cells and macrophages were exposed to unopsonized Aspergillus conidia at the ratio of 1∶1 or LPS ( 10 ng/ml ) at 37°C for 1 hour in the presence of different concentrations ( 1 or 20 µg/ml ) of GAG for 18 hours before the assessment of Il1ra mRNA expression . The differences between the various stimulations were analyzed with the Wilcoxon signed rank test ( p-value of <0 . 05 was considered statistically significant ) . All experiments were performed at least twice and data represent cumulative results of all experiments performed and are presented as mean +/− standard error of the mean ( SEM ) unless otherwise indicated . Data was analyzed using GraphPad Prism v5 . 0 .
|
Aspergillus fumigatus is an opportunistic pathogenic fungus that primarily causes infections in the immunocompromised host . It is known that Aspergillus employs various strategies to evade immune recognition by the host's immune system . Recently , galactosaminogalactan ( GAG ) , a new component of the Aspergillus cell wall , was discovered to have potent anti-inflammatory effects in mice making them more susceptible to Aspergillosis . In the current study we found that this anti-inflammatory property of GAG was due to its capacity to induce the potent anti-inflammatory cytokine interleukin-1 Receptor antagonist . This cytokine interferes with IL-1 signalling and thereby can reduce IL-1–induced immune responses such as T-cell responses . We also found that the induction of this anti-inflammatory cytokine by GAG correlates with increased fungal burden , and mice deficient for this cytokine were protected against aspergillosis . Additionally , we show that the capacity of GAG to induce the natural regulator of IL-1 signalling could be used in the treatment of IL-1–mediated disease such as allergy and colitis . Our study provides new insights on the immunoregulatory activity of GAG and opens up possibilities to exploit the anti-inflammatory potential of GAG as a therapy for inflammatory diseases .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"biochemistry",
"infectious",
"diseases",
"mycology",
"medical",
"microbiology",
"immunology",
"biology",
"microbiology",
"pathogenesis"
] |
2014
|
A Polysaccharide Virulence Factor from Aspergillus fumigatus Elicits Anti-inflammatory Effects through Induction of Interleukin-1 Receptor Antagonist
|
Phenotype switching is commonly observed in nature . This prevalence has allowed the elucidation of a number of underlying molecular mechanisms . However , little is known about how phenotypic switches arise and function in their early evolutionary stages . The first opportunity to provide empirical insight was delivered by an experiment in which populations of the bacterium Pseudomonas fluorescens SBW25 evolved , de novo , the ability to switch between two colony phenotypes . Here we unravel the molecular mechanism behind colony switching , revealing how a single nucleotide change in a gene enmeshed in central metabolism ( carB ) generates such a striking phenotype . We show that colony switching is underpinned by ON/OFF expression of capsules consisting of a colanic acid-like polymer . We use molecular genetics , biochemical analyses , and experimental evolution to establish that capsule switching results from perturbation of the pyrimidine biosynthetic pathway . Of central importance is a bifurcation point at which uracil triphosphate is partitioned towards either nucleotide metabolism or polymer production . This bifurcation marks a cell-fate decision point whereby cells with relatively high pyrimidine levels favour nucleotide metabolism ( capsule OFF ) , while cells with lower pyrimidine levels divert resources towards polymer biosynthesis ( capsule ON ) . This decision point is present and functional in the wild-type strain . Finally , we present a simple mathematical model demonstrating that the molecular components of the decision point are capable of producing switching . Despite its simple mutational cause , the connection between genotype and phenotype is complex and multidimensional , offering a rare glimpse of how noise in regulatory networks can provide opportunity for evolution .
Life is constantly challenged by environmental change . Survival requires that organisms match their phenotype to current conditions . In predictable environments , this occurs by phenotypic acclimation [1] . In unpredictable environments , where the quality of environmental information is unreliable , stochastic phenotype switching allows organisms to hedge their evolutionary bets [2] . Bet hedging spreads the risk of being maladapted in the current environment among variable offspring , each of which has a chance of being fit in some future environment [3–5] . Bet-hedging strategies are widespread in nature , with examples from humans through to microbes [6–10] . Such strategies appear to be common in bacterial pathogens and commensals; when confronted with the challenge of survival in the face of an unpredictable host immune response , these bacteria have evolved mutational mechanisms that allow stochastic modulation of phenotype ( contingency loci [11] ) . Typically , these involve short repetitive DNA sequences that affect expression of genes involved in critical interactions with the host . By virtue of their repeated nucleotide motifs , contingency loci are prone to polymerase slippage , resulting in localized hypermutation . The sophistication of contingency loci begs questions as to their evolutionary origins . Given that pathogens and commensals are often derived from environmental reservoirs , it is likely that the ancestral state was a gene , or gene network , that was subject to environmental regulation . A recent experiment in which a plant saprophyte subject to fluctuating environmental conditions evolved , de novo , the capacity for stochastic phenotype switching [12] offers opportunities to understand how such switches originate . In this previous study , an experimental population of Pseudomonas fluorescens SBW25 was propagated under fluctuating conditions that mimicked essential aspects of the host immune response [13] . After multiple environmental reversals in which evolving populations experienced repeated bottlenecks and phenotypic exclusion , genotypes able to stochastically switch between different colony states were identified in two of 12 replicate lines . This work focuses on the first of these lineages . The capacity to stochastically switch evolved after nine mutations ( Fig . 1A ) . The first eight facilitated bouts of adaptation to different states of the fluctuating environment . The ninth , a single nucleotide change in carB ( c2020t; encoding a subunit of CarAB , EC 6 . 3 . 5 . 5 ) , gave rise to a genotype referred to as 1B4 ( see Fig . 1A for details of evolutionary series genotype names ) that switched at high frequency between two colony types ( translucent and opaque; Fig . 1B ) . Further investigation showed that differences in colony type were attributable to the varying proportion of cells with different capsulation states ( Cap- and Cap+ , respectively; Fig . 1C ) . In addition , allelic replacement studies showed the carB mutation alone was sufficient to generate stochastic colony and capsule switching in the wild-type background ( Fig . 1D ) . Here we describe empirical and theoretical efforts to understand how a single mutation in a gene deeply enmeshed in central metabolism generates such a striking phenotype . The switch-causing mutation reduces concentrations of intermediates in the pyrimidine biosynthetic pathway , thus exposing a pre-existing epigenetic switch ( Fig . 2 ) . Central to the switch is the latter portion of the pyrimidine biosynthetic pathway , which we postulate includes a cell-fate decision point . Together , these components possess functional and regulatory connectivities that a simple mathematical model shows are sufficient to generate stochastic switching .
The transition between capsulation states in genotype 1B4 resembles phase or antigenic variation in pathogens , which is typically underpinned by contingency loci [11] . Initial hypotheses predicted that the mechanism of switching would be mutational , involving either high-frequency gain and loss of the carB mutation or amplification and collapse of the carB locus . Both hypotheses were rejected ( see S1 Text ) . Attention therefore turned to the likelihood that the carB mutation established an epigenetic switch . In order to identify genes with a role in capsule switching , 1B4 was subjected to transposon mutagenesis [12 , 15] . From a pool of ~69 , 000 transposon mutants , 157 were deficient in their capacity to switch between capsulation states . While six showed an increase in the proportion of Cap+ cells , 151 showed loss ( or reduction ) of Cap+ ( S1 Table ) . Thirty-eight mutants were implicated in biosynthesis of a capsular polymer ( see below ) . Of the remaining 119 mutants , 38 contained insertions in potential regulators of capsule biosynthesis ( see below ) and 81 contained insertions in genes involved in a wide range of cellular processes , the most prominent of which were transport and central metabolism ( 22 mutants ) , cell division and DNA processing ( 11 mutants ) , RNA processing ( ten mutants ) , and nucleotide biosynthesis ( five mutants ) . Nineteen transposon mutants contained insertions in the wcaJ-wzb locus ( pflu3658–3678 ) , a region of the genome that resembles the colanic acid biosynthetic locus of Escherichia coli ( Fig . 3 ) [16] . Deletion of the entire wcaJ-wzb locus ( giving 1B4-ΔwcaJ-wzb ) produced translucent colonies ( Fig . 1E ) comprised of Cap- cells . These phenotypes were indistinguishable from those of 19 transposon mutants with insertions in wcaJ-wzb ( previously reported in [12] ) . A further 19 mutants contained insertions in genes encoding enzymes required for biosynthesis of guanosine diphosphate ( GDP ) –fucose , uridine diphosphate ( UDP ) –glucose , and UDP—glucuronic acid [12] , three of the four building blocks for colanic acid ( see S2 Fig ) [16] . These include insertions in galU ( EC 2 . 7 . 7 . 9 ) , pgi ( EC 5 . 3 . 1 . 9 ) , pflu5986 ( EC 5 . 4 . 2 . 8/5 . 4 . 2 . 2 ) , and udg ( EC 1 . 1 . 1 . 22 ) [17] . The usual genes encoding enzymes for biosynthesis of the fourth precursor ( UDP-galactose ) are absent in SBW25 , indicating that the 1B4 capsule polymer is structurally different than colanic acid . Indeed , analysis of the monosaccharide composition of extracellular polysaccharide ( EPS ) in SBW25 , 1A4 , 1B4 , and two transposon mutants indicates that the 1B4 capsule polymer contains six monosaccharides: fucose , glucose , glucuronic acid , galactouronic acid , and two unidentified monosaccharides ( S2 Table ) . Only three of these monosaccharides are present in colanic acid , which consists of repeating units of fucose , galactose , glucose , and glucuronic acid [18] . On the basis of the genetic organisation of the wcaJ-wzb locus , the similarity of these enzymes to functionally characterized enzymes from E . coli , the requirement for colanic acid-like precursors , and the monosaccharide analysis , we conclude that the 1B4 capsule polymer is related to , but distinct from , colanic acid . Henceforth , the 1B4 capsule polymer will be referred to as the colanic acid-like polymer ( CAP ) . Epigenetic switches are often underpinned by differences in gene expression . Total mRNA was extracted from SBW25 , 1A4 , 1B4-Cap- , and 1B4-Cap+ cells and sequenced . Four pairwise comparisons ( SBW25 versus 1A4 , 1A4 versus 1B4-Cap- , 1A4 versus 1B4-Cap+ , and 1B4-Cap- versus 1B4-Cap+ ) allowed differences in transcription to be detected ( S2–S5 Tables ) . Each comparison revealed differential expression of ~60% of all coding sequences , including expression of genes encoding polymer biosynthetic enzymes , cell surface components , metabolic enzymes , and cell division genes . Of note is that transcription of wcaJ-wzb , the CAP biosynthetic locus , was significantly greater in 1B4-Cap+ than 1A4 or 1B4-Cap- . To corroborate this , a promoterless lacZ gene was transcriptionally fused to 1B4 wcaJ . Growth of 1B4-wcaJ-lacZ on media containing X-gal gave rise to colonies with blue sectors comprised of predominantly Cap+ cells ( Fig . 1F ) , indicating that wcaJ transcription is indeed increased in Cap+ cells . The 1B4 transposon mutagenesis screen identified three regulators of CAP biosynthetic genes ( Pflu4939 , BarA-GacA , and Pflu3654-Pflu3657; S1 Table ) . The first of these is Pflu4939 , a probable transcriptional regulator with homology to MvaT , a global regulator of virulence gene expression in other pseudomonads [19] . Three independent transposon insertions in pflu4939 ( and the ~10 bp immediately upstream ) led to increased capsule production , suggesting that Pflu4939 is a negative regulator of CAP biosynthetic gene transcription . Conversely , insertions in loci encoding the other two regulatory systems caused elimination or reduction of capsule formation , indicating that these are activators of CAP biosynthetic gene transcription . The second regulatory pathway , identified by 24 transposon insertions , is the BarA-GacA two-component signal transduction relay system . BarA-GacA is known to regulate a broad range of virulence and stress response systems in various gram-negative bacteria [20] . The third regulatory system is identified by 14 insertions in pflu3654-pflu3657 , a genomic region directly upstream of the CAP biosynthetic genes ( see Fig . 3 ) , most of which has no significant database matches . Notably , when a transcriptional fusion of gfpmut3 and the genomic region upstream of pflu3655-pflu3657 was inserted into the 1B4 genome , expression of green fluorescent protein ( GFP ) in Cap+ cells was observed ( see S2 Fig and S2 Text ) . This demonstrates that this region contains a promoter sequence transcribed in response to the capsule ON signal . None of the above regulators bears homology to the Rcs phosphorelay that controls colanic acid gene transcription in the Escherichia , Erwinia , Klebsiella , and Rhizobium genera [21] . Indeed , a search of the SBW25 genome using translated Basic Local Alignment Search Tool nucleotides ( TBLASTN , a tool used to search a query protein against a translated nucleotide database ) revealed no loci with homology to E . coli rcsA . Thus far , phenotypic observation of 1B4 colonies , combined with the results of transposon mutagenesis and transcription analyses , indicates colony switching to be the result of bistable CAP expression , which is at least partially effected at the transcriptional level . To understand the epigenetic events by which the carB mutation influences CAP expression , we next studied the pyrimidine biosynthetic pathway , which provides a molecular link between carB and CAP ( Fig . 2 ) . In brief , CarB is the product of the second gene of the carAB operon , which encodes the two subunits of carbamoyl phosphate synthetase ( CPSase , EC 6 . 3 . 5 . 5 ) . The product of CPSase , carbamoyl phosphate ( CP ) , is the precursor for synthesis of arginine and pyrimidines . The pyrimidine biosynthetic pathway concludes with synthesis of UTP , which is further metabolised by PyrG ( to produce cytidine triphosphate [CTP] for DNA synthesis ) and/or GalU ( to produce UDP-glucose , a precursor of CAP biosynthesis ) . The importance of the pyrimidine biosynthetic pathway in 1B4 capsule switching is corroborated by the isolation of a transposon mutant ( JG176 ) carrying an insertion in one of the pathway genes , ndk ( encoding nucleoside diphosphate kinase , EC 2 . 7 . 4 . 6 ) . This mutant is intriguing because it generates uniformly opaque colonies—a rare phenotype in the mutagenesis screen . Cre-mediated excision of the transposon left a 189 bp inactivational insertion in ndk , creating JG176ΔCre and eliminating the possibility of polar effects . The mutant showed elevated levels of capsulation compared to 1B4 ( Fig . 4A ) . A similar increase was observed upon deletion of the entire ndk reading frame from 1B4 ( 1B4-Δndk; Fig . 4A ) . Deletion of ndk from wild type also resulted in an increase in capsulation ( SBW25-Δndk; Fig . 4A ) , demonstrating that a reduction in ndk activity plays a role in capsule bistability; hypercapsulation as a result of ndk inactivation indicates that capsule production is a response to perturbations in the supply of nucleotide precursors . Notably , SBW25-Δndk has a lower capsulation level than 1B4-Δndk ( p < 0 . 01 ) , highlighting that multiple mutations in the pyrimidine pathway can have additive effects on capsule bistability . Given that production of nucleoside triphosphates ( NTPs ) is an essential intracellular process , the viability of the ndk deletion strains indicates that SBW25 possesses at least one alternative enzyme with nucleoside diphosphate kinase ( NDK ) activity; adenylate kinase [22] , pyruvate kinase [23] , and polyphosphate kinase [24] have been shown to function as alternative NDKs in other species . While each enzyme showing NDK activity uses UDP , CDP , GDP , and adenosine diphosphate ( ADP ) as substrates for NTP production , differing affinities of the enzymes for each substrate can carry large consequences for intracellular UTP , CTP , GTP , and adenosine triphosphate ( ATP ) concentrations [23] . Changing NDK activity through ndk deletion is therefore expected to have unpredictable consequences for all nucleoside diphosphate ( NDP ) and NTP pools in SBW25 and 1B4 cells . Reduced growth of 1B4 compared to that of the immediate precursor genotype ( 1A4 ) indicates that the carB mutation causes a reduction in CPSase function ( S3 Fig ) . In line with this , direct measures of pyrimidine intermediates in 1A4 and 1B4 demonstrated statistically significant reductions in intracellular concentrations of UDP and UTP as a consequence of the c2020t carB mutation ( Fig . 4B , S7 Table ) . Similar measurements in SBW25 and SBW25-Δndk revealed that deletion of ndk leads to significant reductions in UMP , UDP , and UTP ( Fig . 4B , S7 Table ) . Thus , we conclude that function-reducing mutations in pyrimidine biosynthetic genes lead to reductions in pyrimidine pathway intermediates ( Fig . 4B ) , which in turn lead to increases in capsulation ( Fig . 4A ) . ( For concentrations of other intracellular metabolic intermediates in a range of genotypes , see S8 Table . ) Given the above results , it is likely that the greater the loss in pathway functionality , the greater the reduction in pyrimidine pathway intermediates . However , genotypes containing a mutation in both carB and ndk ( 1B4-Δndk and JG176ΔCre ) showed an increase in UMP , UDP , and UTP compared to 1B4 ( Fig . 4B ) , despite relative capsulation increases ( Fig . 4A ) . This result was confirmed by a second , independent set of measurements performed using an alternative protocol ( see S8 Table ) . The observations in these double-mutant genotypes are likely indicative of the complexity of regulatory systems underpinning central metabolism ( see also Results section above ) [25] . Results to this point have demonstrated that the carB mutation causes ( i ) a reduction in concentrations of pyrimidine intermediates and ( ii ) capsule switching . Next , we investigated whether reduction of one or more pyrimidine intermediates causes ( rather than merely correlates with ) capsule switching . If the relationship were causal , then switching should be reduced or eliminated by restoration of the concentration of relevant pyrimidine intermediates to ancestral levels . Thus , the effect of adding increasing amounts of exogenous uracil—which is taken up by cells and enters the pyrimidine biosynthetic pathway as UMP ( see Fig . 2 ) —on capsule switching was investigated . Uracil addition gradually decreased capsule switching , and a concentration of 2 mM uracil significantly reduced the proportion of capsulated cells ( two-sample t test , p < 0 . 001; Fig . 5A ) . By way of control , the effect of adding increasing amounts of exogenous arginine , the biosynthesis of which also depends on carB , was investigated and found to have no effect on the proportion of capsulated cells ( Fig . 5A ) . Additionally , quantification of the effect of adding guanine hydrochloride on capsulation revealed no direct role for purines—intracellular levels of which are tightly coordinated with pyrimidines [14]—in capsule switching ( S3 Text ) . Together , the above findings not only demonstrate a causal relationship between reduction of pyrimidine biosynthetic intermediates and capsule switching , but they also show that the switching phenotype can be eliminated through provision of an intermediate ( UMP ) several enzymatic steps downstream of carB . This indicates that capsule switching is dependent on reduced availability of UMP and/or downstream products . Work to this point has shown that the likelihood of cells becoming capsulated is negatively related to the supply of pyrimidine pathway intermediates , particularly downstream of UMP . In principle , supply-reducing mutations elsewhere in the pyrimidine pathway should effect a similar reduction in pyrimidine concentrations , leading to switching . In order to investigate whether mutations in genes other than carB can give rise to capsule switching , an evolution experiment was performed in which we sought to evolve new switching genotypes from 1A4 . Each of these genotypes was expected to contain the first eight mutations of the evolutionary series ( Fig . 1A ) plus an independently evolved switch-causing mutation . A set of 36 independent replicate populations of 1A4 ( the nonswitching predecessor of 1B4 ) was subjected to one round of selection under the regime that delivered 1B4 . At the conclusion of this “re-evolution” experiment , cells from each population were plated on agar and screened for the presence of rapidly sectoring colonies . From 36 replicates , six populations produced switchers . A capsule counting assay showed that each switcher genotype produced populations of cells with distinctly different ratios of Cap+ to Cap- cells ( Fig . 5B ) , suggesting that the underlying mutations had quantitatively different effects on concentrations of pyrimidine intermediates and the likelihood of switching . Indeed , analysis of intracellular metabolite levels in two of the re-evolved switchers , Re1_4 and Re1_5 , revealed that the switch-causing mutations in these genotypes cause a multitude of metabolic effects ( S8 Table ) , including alteration of the concentration of pyrimidine intermediates . In order to see whether the switch-causing mutations lay upstream or downstream of the entry point of uracil , the effect of uracil supplementation on the frequency of switching was examined . Five of the six re-evolved switchers showed a response to uracil that was typical of 1B4; however , the sixth ( Re1_4 ) was unresponsive ( Fig . 5B ) . On sequencing the carB gene , each of the five uracil-responsive switchers was found to contain a carB mutation ( Table 1 ) . Re1_4 did not contain a carB mutation . Genome resequencing of Re1_4 revealed the final mutation as a single , nonsynonymous change in pyrH ( R123C ) . PyrH encodes uridylate kinase ( EC 2 . 7 . 4 . 22 ) , the enzyme responsible for the phosphorylation of UMP to UDP . As the UDP/UMP ratio of Re1_4 was decreased 28-fold relative to that of 1A4 ( see S8 Table ) , this mutation was deleterious to PyrH function . Given that pyrH is six genes along the pyrimidine pathway from carB , this finding reduced the number of genes and regulatory connections under consideration: from this point on , attention focused on pyrH and downstream genes . To define a set of genes necessary for switching , the expression of each gene downstream of pyrH was manipulated . We reasoned that manipulations of any candidate gene that led to a change in the behaviour of the switch indicated the involvement of that component in the circuit . Thus , carB , pyrH , ndk , galU , and pyrG were overexpressed in turn in 1B4 , and the ratio of Cap+ to Cap- cells determined for each ( Fig . 5C ) . Although no longer a central focus , carB was included in the overexpression set by way of positive control; given that 1B4 switching results from a loss of carB function , successful overexpression of wild-type carB is expected to reduce capsulation . Identifying the point in the pyrimidine pathway at which the switch decision occurs is central to understanding the switch mechanism . Logic suggests this is likely to be the UTP bifurcation point ( see Fig . 2 ) , and the results shown in Fig . 5C are consistent with this prediction . First , overexpression of carB , pyrH , and ndk ( the three enzymes immediately prior to the bifurcation ) significantly lowered 1B4 capsulation levels , indicating that sufficiency prior to the UTP bifurcation directs pathway resources away from capsulation . Second , overexpression of PyrG and GalU ( two enzymes immediately beyond the UTP decision point ) had opposing effects on 1B4 capsulation: pyrG overexpression lowered capsulation , while galU overexpression increased capsulation . The ability to both decrease and increase capsulation through manipulation of these genes demonstrates that these components encompass key elements of the switch . In addition to increasing 1B4 capsulation levels , galU overexpression led to cell chains . Each chain , composed of 2–10 cells , was exclusively Cap+ or Cap- ( Fig . 5E ) . This is in contrast to the phenotype of the Cap- galU transposon mutant ( JG114ΔCre , S1 Table ) , which is low in UDP-glucose ( S8 Table ) and produces short cells ( Fig . 5F ) . The fact that an increase in GalU inhibits cell division ( producing cell chains ) and a reduction in GalU stimulates cell division ( producing short cells ) indicates that GalU , or downstream pathway components , play a role in regulation of cell division . Indeed , such a finding has recently been reported in E . coli [26] and Bacillus subtilis [27] , in which UDP-glucose appears to function as a “metabolic sensor” that couples nutrient availability to cell division . Introduction of the carB mutation from SBW25 to E . coli B REL606 did not generate any obvious cellular- or colony-level phenotypic switching ( S4 Fig ) . In order to directly observe capsule switching events , a genotype carrying a transcriptional fusion of gfpmut3 and the region directly upstream of pflu3655—a probable regulator of CAP expression ( see Fig . 3 and S1 Table ) —was constructed in 1B4 ( see S2 Text ) . Microscopic analysis of the resulting genotype , 1B4-CAP-GFP , revealed a 100% correlation between GFP and capsule expression; Cap+ 1B4-CAP-GFP cells fluoresce , while Cap- cells do not ( S2 Fig ) . The growth of single 1B4-CAP-GFP cells into microcolonies was observed microscopically . First , the development of microcolonies founded by noncapsulated cells was examined ( S2A Fig ) . The switch from Cap- to Cap+ was found to be dependent on population size; Cap- to Cap+ switch events were only observed once a population size of 1870 ± 660 cells was exceeded ( n = 58 microcolonies ) . In accordance with this , the distribution of the population size at which the first switch event is observed is Gaussian ( S2B Fig ) . This is in contrast to a Poisson distribution , which would have supported a random process independent of population size . However , once a minimum threshold population size is reached , Cap- cells switch to Cap+ at random , and , thereafter , the expression of CAP genes is heritable ( S1 Video ) . In contrast to Cap- founder cells , Cap+ founder cells were observed to systematically switch to the Cap- form within a very short time frame ( S2 Video ) . To investigate the cost of capsule production , we measured the growth rate of Cap- and Cap+ cells using image analysis . Using a conservative estimate , Cap+ cells were found to grow more slowly ( growth rate of 0 . 0113 min-1 and SE = 2 . 5 x 10-4 min-1 , or one division every ~61 . 3 min; n = 39 ) than Cap- cells ( growth rate of 0 . 0116 min-1 and SE = 6 . 10 x 10-5 min-1 , or one division every ~59 . 8 min; n = 17 ) . This demonstrates a cost for capsule synthesis of ~3% . While genetic and biochemical data define a minimal set of components involved in capsule switching , specific molecular details remain unknown . Nonetheless , we present a simple mathematical model in order to show that the bifurcation of the UTP pool into CTP and UDP-glucose synthesis can , with a minimal set of plausible assumptions regarding regulation ( outlined in S5 Fig ) , give rise to capsule switching . For simplicity , the model assumes a molecule—hereafter referred to as UXP—whose level is proportional to UTP as the central signal molecule for capsule biosynthesis . In reality , the signal could be any one , or a combination , of many possibilities ( e . g . , pyrimidine pathway intermediates , biosynthetic enzymes , and transcription factors ) . Shown in Fig . 6A , the growth-capsulation model postulates that capsule bistability arises as a result of differential utilization of the UTP pool at the bifurcation point ( see Fig . 2 ) : Cap- prevails when UTP is preferentially channelled by PyrG into DNA metabolism ( and ultimately cell division ) , while Cap+ results from utilization of UTP by GalU for capsule biosynthesis . The likelihood of a cell entering the Cap+ state increases as intracellular UXP levels decline . This we model as a positive regulator of CAP biosynthesis whose activity is inversely proportional to the availability of UXP . For example , the regulator may have high affinity for UXP , which ensures that at high UXP concentrations the regulator is inactive . However , as UXP concentrations decline , the regulator becomes free , resulting in high-level expression of CAP biosynthetic genes . As previously noted , the 1B4 transposon mutagenesis screen identified two genetic loci encoding transcriptional regulators required for maximal CAP biosynthetic gene transcription: pflu3655-pflu3657 and barA/gacA ( S1 Table ) . It is plausible that the activity of one of these regulators is sensitive to changes in concentrations of pyrimidine intermediates . Experimental observations show that the Cap+ state is heritable ( S1 Video , [12] ) . This “phenotypic memory” depends on maintenance of low UXP levels , which we model as a positive feedback loop whereby the synthesis of CAP draws UTP into the capsule production . Although UDP from UDP-glucose is recycled back through UMP , commitment to CAP synthesis ensures a reduction of the UXP pool . Duration of the Cap+ state depends on molecular noise inherent in the mechanism of regulation ( e . g . , uneven splitting of intracellular molecules upon cell division , strength of positive regulator-promoter binding , and rate of CAP degradation ) . To model this , interactions among components were simulated using a stochastic simulation algorithm [28] . For a switch to Cap- , UTP utilization must shift from CAP biosynthesis in favour of nucleotide metabolism . The likelihood of this occurring increases with increasing UXP levels . We model this as a time-delayed , autoregulatory feedback loop that restricts the activity period of capsule synthesis . For example , CAP may inhibit its own biosynthesis . In order to demonstrate that the components of the model are capable of generating switching , we used the chemical equations in S5 Fig to simulate changing concentrations of CAP in a single cell over a 48-h period . Further , to investigate the effect of UXP levels on capsulation state , the simulation was performed with differing UXP production rates . Fig . 6B shows that at low UXP production rates , a cell typically has two widely separated levels of CAP expression over 48 h , representing the Cap+ ( higher ) and Cap- ( lower ) states . Notably , both the high and low CAP levels are maintained—in a probabilistic manner—over several cell divisions ( i . e . , are heritable ) . This phenotypic dichotomy is typically eliminated at higher UXP production rates; Fig . 6C shows CAP expression maintained at a relatively constant basal level . Further examples of simulated CAP expression at high , low , and intermediate UXP levels are provided in S6 Fig . This simple model demonstrates stochastic capsule switching in which both states persist for a duration that is long enough to allow epigenetic heritability ( sufficient to generate colonies that are visibly Cap+ or Cap- ) but short enough to enable each state to give rise to the other . Furthermore , UXP concentration in the system determines the probability of switching . While the model relies on stochasticity in biochemical reactions , it is expected that other biological factors—such as cell division , which appears to be linked to GalU ( and thus CAP ) levels—will also contribute to heterogeneous populations [29] . In addition to simulating stochastic switching , the model also encompasses a circuitry ( of reactions , reaction rates , and number of molecules ) that allows the level of stochasticity to be tuned . To demonstrate this , the 23 reaction rates ( determined by the equations in S5 Fig ) were altered and the resulting time spent in Cap- and Cap+ states determined . The value of each reaction rate , one at a time , was decreased by a factor of ten , and the simulation run ten times . This process was then repeated but with a 10-fold increase in each rate . The fraction of time for which the network exhibited the Cap- phenotype was determined under all 47 sets of conditions . The results ( shown in S9 Table and S7 Fig . ) reveal phenotypes that are Cap- between 11 . 6% and 90 . 4% of the time , indicating tuning over a wide range of reaction rates . Thus , while certain specific molecular details of the switch remain to be determined , the growth-capsulation model is consistent with experimental observations to date . The above model postulates that all machinery necessary for switching is present in the ancestral SBW25 genome but that distance from the Cap+ “threshold”—due to sufficiency of UXP under normal laboratory culture—means that colonies appear uniformly translucent . Upon discovery of a mutation that reduces flux through the pyrimidine pathway , the model predicts that the population comes closer to the Cap+ “threshold” and thus the possibility that extrinsic noise places more of the population above the threshold . Accordingly , there is reason to expect the occasional occurrence of Cap+ cells in colonies of the ancestral type that are otherwise visibly translucent . Indeed , a capsule counting assay revealed the proportion of capsulated cells in SBW25 populations to be 0 . 0016 ( ~1 capsule per 625 cells ) compared with 0 . 084 ( ~1 in 12 cells ) in 1B4 populations ( Fig . 1D ) . If the capsulated cells observed in SBW25 and 1B4 are a result of the same switch mechanism , then it is expected that Cap+ cells of both genotypes will respond to manipulations of the pyrimidine biosynthetic pathway in a similar manner . Thus , as was done for 1B4 ( Fig . 5C ) , carB , pyrH , ndk , galU , and pyrG were each sequentially overexpressed in SBW25 . A capsule counting assay performed on each genotype ( Fig . 5D ) revealed that , as seen for 1B4 , overexpression of ndk resulted in a significant reduction in SBW25 capsulation ( no capsules observed in 4 , 500 cells ) . No statistically significant change in capsulation was observed for any other gene , possibly a result of the low number of capsulated cells assayed . Taken together , these results demonstrate that , as predicted by the growth-capsulation model , the capsule switch machinery is present and functional in SBW25 .
Rapid switching of colony morphology is a striking phenotypic innovation . In bacterial pathogens , such behaviour is reflective of phase variation in which mutable ( contingency ) loci control the expression of epitopes on the surface of bacterial cells that interact with the host immune response [11 , 30] . Variable expression of such epitopes among members of a population allows the risk of elimination by the immune system to be spread among offspring , each of which has some chance of avoiding detection . Similarly , evolution of the capacity in 1B4 to switch between opaque and translucent colonies enabled this genotype to succeed in the face of fluctuating selection wrought via an experimental regime that mimicked fundamental features of the host immune response [12] . The ancestral P . fluorescens SBW25 genotype lacks the capacity to switch colony phenotype under standard laboratory conditions . The fact that switching emerged with relative ease ( in two independent lines and with few mutations ) suggests a predisposition for evolution of this trait . Our initial prediction was to find modification to an existing tract of nucleotides controlling expression of capsule production that increased the likelihood that this tract underwent mutation . We were therefore surprised to find the cause of switching to be a mutation in carB . We were further surprised to find that this mutation , when transferred back into the original ancestor ( SBW25 ) , was sufficient to generate switching behaviour [12] . While the proximate cause of switching in 1B4 is a mutation in carB , the ultimate cause resides in interactions among cellular components ( genes , enzymatic products , and metabolic intermediates ) surrounding the bifurcation of the UTP pool into CTP and UDP-glucose . At any point in a metabolic pathway where a single , inevitably limiting resource is partitioned into two or more components , there must necessarily be some decision-making mechanism in order to allow cells to match changes in substrate availability with cellular demands [31] . The pre-existence of such a decision-making mechanism—one that allows cells to partition the UTP pool preferentially towards DNA metabolism or polysaccharides—is central to the emergence of switching . Our data show that such a switch exists , that it is probabilistic , responsive to the level of pyrimidine pathway intermediates , and determines whether or not cells commit to a normal cell cycle or to a phase during which protective CAP is produced . The net effect of the carB mutation is to induce a shortage in the supply of pyrimidine intermediates , thus bringing the population closer to the threshold level at which the switch is likely to be “thrown . ” A decision to enter the Cap+ state and thereby use already scarce pyrimidine resources to synthesize a nonessential polymer may at first appear paradoxical . However , pyrimidine nucleotides are not consumed by the capsule biosynthetic process but instead are recycled back into the pyrimidine biosynthetic pathway as UMP ( see Fig . 2 ) . Further , the channelling of UTP into CAP precursor biosynthesis ensures that levels of core pyrimidine nucleotides ( UMP , UDP , and UTP ) are kept low , making cells less likely to enter the Cap- state ( in which pyrimidine resources are diminished by splitting between daughter cells ) . Thus , CAP biosynthesis might be viewed as a mechanism for conserving scarce pyrimidine resources until such a time that intracellular conditions are optimal for cell division . Although the 1B4 carB mutation causes visible switching between colony types , it is clear that the underlying switch ( and capacity to toggle between capsulation states ) exists in the ancestral genotype . Such a cell cycle checkpoint—underpinned by a stochastic switch—is understandable from the perspective of bacterial survival . In natural environments , resources are often limiting . If the supply of essential components ( e . g . , nucleotides ) to complete the cell cycle is insufficient , then it makes little sense to commit to cell division . An alternate strategy would be to reduce growth rate and invest in protective polymers , thus allowing cells to persist with the chance that intracellular nucleotide concentrations might rise , eventually allowing a normal round of cell division . The switch being probabilistic provides organisms with the possibility to hedge their evolutionary bets . This is in line with the fact that a decision to commit to the cell cycle versus polymer production is necessarily based on knowledge of the present state of the environment . However , the present state will not always accurately predict the future state , and the cost of making a wrong decision ( i . e . , cell division when nucleotides turn out to be insufficient or polymer production when nucleotides turn out to be abundant ) is likely to be high . A stochastic switch would ensure that some cells commit to a strategy that is at odds with the current environment on the off chance that in some future state of the environment that strategy is optimal . In many regards , capsule switching is reminiscent of both the sporulation process in B . subtilis [32]—a cell-fate decision that has recently been shown to also be dependent on UDP-glucose levels [33]—and the phenotypic switch to bacterial persister cells [8 , 32] . Capsule bistability also shares features with the E . coli lactose switch in which expression of lacZYA is either ON or OFF , depending on the presence of a metabolized inducer . Theoretical studies show that at high or low inducer levels the population is uniformly lacZYA ON or OFF respectively , while at intermediate inducer levels the population bifurcates , giving rise to both steady states [34] . In terms of 1B4 colony switching , high pyrimidine levels result in the CAP OFF state ( producing uniform Cap- colonies as in SBW25 and 1A4 ) , and lower pyrimidine levels result in bistable CAP expression and thus colony bistability . Even lower pyrimidine levels push the population towards a majority CAP ON and opaque colonies ( as in the ndk transposon mutant , JG176ΔCre ) . If our conclusions concerning the existence of a cell cycle checkpoint are correct , then such a checkpoint is likely conserved across many bacteria . Recent work in E . coli and B . subtilis has shown that UDP-glucose and downstream polysaccharide biosynthetic enzymes ( OpgH and UgtP , respectively ) are part of a signal cascade that influences cell-fate decisions in response to nutrient availability [26 , 27] . At the molecular level , these enzymes bind to FtsZ , inhibiting assembly of the cell division spindle . Further , it has been noted that disruptions in E . coli UDP-glucose—through inactivation of pgm ( S1 Fig . ) —alter the timing of cell division , resulting in short cells [35 , 36] . Disruptions in pgi , pgm , or galU showed a similar , short-cell phenotype in 1B4 ( S1 Table ) . Further , overexpression of galU resulted in long cells in both 1B4 and SBW25 ( Fig . 5E ) . The conservation of these effects on cell division strongly indicates that GalU , UDP-glucose , and/or downstream enzymes influence cell division in SBW25 . Thus , we propose that UDP-glucose may act not only as a nutrient sensor but also as a nucleotide sensor during progression through the cell cycle . Evolution works with the building blocks at its disposal , typically via small modifications [37–39] . While the jump in phenotypic space associated with the emergence of colony switching in 1B4 is large , the innovation is underpinned by a minor genetic change that has resulted in apparent modification of the threshold for flipping an existing epigenetic switch; selection appears to have taken advantage of molecular noise to generate an adaptive phenotype . Whether such a starting position is typical for the emergence of a classical contingency locus is unclear , but it is possible that phenotypic switches with an epigenetic basis [40 , 41] are the evolutionary forerunners of genetic switches and possibly even developmentally controlled regulation [42] . Although the molecular bases of 1B4 colony switching currently reside in a regulatory circuit deep in central metabolism , subsequent fluctuating selection could conceivably result in the switch being accommodated in a more specific part of the capsule machinery . Indeed , selection against negative pleiotropic effects associated with the defect in carB is likely to be a potent force driving such a process .
Bacterial strain details are provided in S2 Text . Unless otherwise stated , strains were grown for 16–24 h at 28°C in shaken 30 mL glass microcosms containing 6 mL King’s Medium B ( KB ) [43] , with appropriate supplements . Where stated , uracil , L-arginine hydrochloride , and/or guanine hydrochloride ( Sigma-Aldrich ) were added to the medium . Cells were plated on lysogeny broth ( LB ) containing 1 . 5% agar . Antibiotics were used at the following concentrations: tetracycline ( 10 μg mL-1 , Tc ) , kanamycin ( 100 μg mL-1 , Km ) , gentamicin ( liquid medium 10 μg mL-1 , agar plates 20 μg mL-1; Gm ) , and nitrofurantoin ( 100 μg mL-1 , NF ) . Cell microscopy was performed using a Zeiss Axiostar Plus light microscope coupled with fluorescence lighting ( HBO 50/AC ) where required . Fluorescence microscopy samples were incubated for ~16 h on KB plates containing 10 μg mL-1 calcofluor ( Fluorescent Brightener 28 , Sigma-Aldrich ) . A dissection microscope was used for colony images . Microscopy images were cropped and processed in Preview as indicated in figure legends . 1B4 cultures were grown from glycerol stocks ( 24 h ) . Cultures were mixed thoroughly , diluted 100 times in fresh KB , and grown overnight again ( 24 h ) . Bacteria were then diluted 104 times and plated on a gel pad ( 1% agarose in KB ) . The preparation was sealed with a glass coverslip using double-sided tape ( Gene Frame , Fischer Scientific ) . A duct was cut through the pad centre to allow oxygen diffusion . Time-lapse videos of microcolonies were captured in phase contrast with an automated inverted microscope ( IX81 , Olympus ) using a 100x/NA 1 . 35 objective ( Apo-ph1 , Olympus ) . Images were acquired with an Orca-R2 CCD camera ( Hamamatsu ) . Fluorescence excitation was achieved with a mercury vapour light source ( EXFO X-Cite 120Q ) . GFP was imaged with a 485 ( 20 ) /520 ( 28 ) -nm filter set using a dichroic beam splitter at 500 nm ( Semrock ) . The growth rates of Cap- and Cap+ cells were determined by image analysis using MATLAB routines . For each strain , 3–5 replicate KB microcosms were inoculated from glycerol stocks . After a 24-h incubation and a 1:1000 transfer , the proportion of Cap+ cells was determined in each . Cells were diluted , transferred to a microscope slide , and stained with 1:10 diluted India ink , and a cover slip was added . After 1 min , each preparation was photographed under phase contrast 40x or 63x magnification . Capsule expression was recorded manually for 500 cells per preparation ( ≤100 cells assayed per photograph ) . Average proportions of Cap+ cells were determined and statistical analyses performed . See also S2 Text . Gene deletions were constructed in the SBW25 background by pUIC3-mediated two-step allelic exchange as described elsewhere [44] . The carB mutation was constructed in E . coli REL606 using a pKOV-mediated two-step allelic exchange as described in [45] . For further details of genetic constructs , see S2 Text . Wild-type wcaJ was PCR-amplified and ligated into pUIC3 [46] immediately upstream of promoterless lacZY ( see S2 Text ) . The construct was used to transform E . coli DH5α-λpir and transferred to 1B4 via triparental conjugation ( with a helper strain carrying pRK2013 ) . A successful transconjugant was purified , giving 1B4-wcaJ-lacZ . Single 1B4-wcaJ-lacZ colonies were grown at 28°C for 56 h on an LB+NF+Tc+X-gal ( 60 μg mL-1 ) plate prior to microscopic analysis . The carB , pyrH , ndk , galU , and pyrG genes were PCR-amplified using primers with restriction sites ( see S2 Text ) . Error-free PCR products were ligated into pSX [47] . Alongside empty pSX , each construct was used to transform chemically competent 1B4 and SBW25 cells [48] . Successful transformants were selected and purified on LB+Gm plates , and insert presence was checked by PCR . Three independent genotypes were constructed for each gene-strain combination . Capsule counting assays were performed as previously described . 1B4 was subjected to random mutagenesis as described [15] . Approximately 69 , 000 transposon mutants from 41 independent conjugations were screened on LB+Km , on which 1B4 mutants typically form translucent colonies with opaque sectors . In selected strains , the bulk of the transposon was deleted [15] , leaving 189 bp at the insertion site ( “-ΔCre” genotypes ) and eliminating polar effects . Extracellular polysaccharide ( EPS ) was isolated from SBW25 , 1A4 , 1B4 , JG176Δcre , and JG114ΔCre . Each genotype was grown on KB agar ( 28°C , 48–96 h ) . Cellular material was resuspended in 12 mL of 1 M NaCl to give an OD600 of ~3 . 5 , vortexed for 40 min , and centrifuged ( 30 min , 4 , 168 g ) . EPS was precipitated from supernatants by the addition of 3 volumes isopropanol . Following resuspension of pelleted ( 40 min , 4 , 168 g ) EPS in 0 . 5 mM CaCl2 , RNAse ( 0 . 1 mg mL-1 ) and DNAse ( 1 . 2 units mL-1 ) were added , and samples incubated at 37°C overnight . The next day , samples were supplemented with citrate buffer ( pH 4 . 8 , 50 mM ) and , to eliminate any cellulose , treated with cellulase ( ICN Biomedicals; 0 . 15 mg mL-1 , 2 h at 50°C ) . After addition of 0 . 5 mg mL-1 Proteinase K , EPS was precipitated by the addition of 3 volumes isopropanol and pelleted ( 40 min , 4 , 168 g ) . EPS pellets were dissolved in dH2O ( 50°C , overnight ) . Finally , samples were dialysed against dH2O for 48 h ( SnakeSkin dialysis tubing , Thermo Scientific ) . The Callaghan Research Institute ( New Zealand ) performed EPS analysis . Samples were freeze-dried and weighed , and a colorimetric total sugar assay was performed in duplicate for each EPS isolation . A Re1_4 colony was grown overnight in KB . Cap+ and Cap- fractions were separated by centrifugation , and genomic DNA was isolated from each fraction using the CTAB method . Equal quantities of each isolated DNA were mixed , and whole genome resequencing was performed ( Illumina; Massey University , New Zealand ) . Point mutations were identified by aligning 36-bp sequence reads to the SBW25 genome [49] via SOAP ( short oligonucleotide alignment program [50] ) and ELAND ( Illumina ) . Insertions and deletions were identified by analysing genomic regions with unusual coverage and BLAST ( Basic Local Alignment Search Tool ) analysis of discarded sequences . SBW25 , 1A4 , and 1B4 colonies were grown overnight in KB , diluted 1:1000 into 20 mL KB , and incubated for 24 h . Total RNA was harvested from each; for SBW25 and 1A4 , 100 μL culture was mixed with 900 μL KB , pelleted , and resuspended in 1 mL of RNAlater ( Ambion ) . For 1B4 , 100 μL was separated into Cap+ and Cap- by centrifugation , and each was resuspended in 1 mL RNAlater . All mRNA extractions proceeded using a RiboPure Bacteria Kit ( Ambion ) . Two mRNA preparations from separate colonies were pooled for each strain . Normalized mRNA-seq library preparation , followed by 100-bp single-end sequencing , was performed by the Australian Genome Research Facility ( Brisbane , Australia; accession number GSE48900 ) . The DEseq R package [51] was applied to identify differentially expressed genes . Intracellular nucleotides were extracted and measured by two independent methods . The first used liquid chromatography mass spectrometry ( LC-MS ) to measure intracellular concentrations of UMP , UDP , and UTP in five replicates of six genotypes ( Fig . 4B , S7 Table ) . For each sample , single colonies were grown in 16-h glycerol-SA [52] cultures ( shaking , 26°C ) , vortexed , diluted 1:100 or 1:1000 in glycerol-SA , and grown for a further 24 h . Cultures were vortexed , diluted in glycerol-SA to give an OD600 of 0 . 1 ( total volume 10 mL in 100-mL flask ) . Flask cultures were grown until the midexponential phase , when a 1-mL sample was rapidly collected by centrifugation ( 15 sec , 16 , 000 g ) . Pellets were immediately frozen in liquid nitrogen until further processing . The remaining culture was used to determine the number of cells per sample ( cfu mL-1 ) . Samples were extracted by cold extraction [53] and submitted to LC-MS , and the data were analysed as previously described [54] . The second method involved radiolabelling followed by thin-layer chromatography and was used to measure 22 metabolites in triplicate across eight genotypes ( S8 Table ) . Cultures were grown exponentially in glycerol-SA with aeration and labelled with [33P]-orthophosphate ( Perkin-Elmer ) from optical density of 0 . 1 at 600 nm [52] . For determination of nucleotide pools in Cap+ and Cap- fractions , cultures were separated by centrifugation and propagated in fresh media for at least three generations before sampling . Charcoal-binding nucleotides were extracted , separated by two-dimensional thin-layer chromatography , and quantified [55 , 56] . Six independent switcher genotypes were isolated from 1A4 according to the evolutionary protocol [12] . Each switcher was purified and its carB gene sequenced . Simulations were performed by incorporating the stochastic Gillespie algorithm into a MATLAB program [28] ( S5 Fig , S1 Code , and S2 Code ) . Initial parameter values were assigned randomly ( with the proviso that UTP outnumbered activator molecules ) , and reaction rates were chosen to promote the equivalent of several cell divisions between switch events , reflecting observed heritability . UTP production rates were adjusted from low to high ( 1 , 000-fold difference ) to simulate the intracellular state of a switcher and ancestral cell , respectively ( S6 Fig ) . Reaction rates were sequentially increased and decreased by a factor of 10 to investigate the role of stochasticity in switching ( S7 Fig and S9 Table ) . Simulations were run for 200 , 000 sec ( 55 . 55 h ) before removal of the initial 10 , 000 sec , allowing escape from any restrictive initial conditions . To detect differences in Cap+ levels or nucleotide concentrations between two strains , two-sample t tests ( parametric or Welch ) or , where normality assumptions were violated , Wilcoxon rank-sum tests were applied . To detect differences in Cap+ levels across three strains ( while testing overexpression genotypes ) , one-way ANOVA or , where normality assumptions were violated , Kruskal Wallis tests were used . All analyses were performed in R [57] . On the graphs , * = 0 . 05 < p < 0 . 01 , ** = 0 . 01 < p < 0 . 001 , and *** = p < 0 . 001 .
|
Phenotype switching—the ability to switch rapidly between phenotypic states—is an evolutionary survival strategy commonly used by organisms in the face of unpredictable environmental conditions . However , little is known about how phenotype switches emerge and function in their early evolutionary stages . A previous study observed the evolutionary emergence of colony morphology switching in Pseudomonas fluorescens populations in response to fluctuating selection . Here we describe the underlying molecular basis of this colony switching , providing the first account of the mechanism behind a real-time evolved phenotype switch . We show that colony switching in this instance is underpinned at the cellular level by high frequency ON/OFF expression of colanic acid-like capsules in response to varying levels of a metabolite . Biochemical assays revealed that capsule switching results from mutations that reduce concentrations of intermediates in a central metabolic pathway—the pyrimidine biosynthetic pathway . Of key importance is the partitioning of these metabolic resources between polymer production ( leading to capsulation ) and cell division ( leading to noncapsulation ) ; this bifurcation marks a decision point whereby cells with low metabolite levels divert resources towards polymer production , increasing the likelihood of switching to the capsulated state . As a greater proportion of cells become capsulated , colony switching emerges . These findings show that , while colony switching evolved with relative ease , the underlying molecular mechanism is surprisingly complex .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Bistability in a Metabolic Network Underpins the De Novo Evolution of Colony Switching in Pseudomonas fluorescens
|
The gammaherpesviruses , including Epstein-Barr virus ( EBV ) and Kaposi's sarcoma-associated herpesvirus ( KSHV ) , establish latency in memory B lymphocytes and promote lymphoproliferative disease in immunocompromised individuals . The precise immune mechanisms that prevent gammaherpesvirus reactivation and tumorigenesis are poorly defined . Murine gammaherpesvirus 68 ( MHV68 ) is closely related to EBV and KSHV , and type I ( alpha/beta ) interferons ( IFNαβ ) regulate MHV68 reactivation from both B cells and macrophages by unknown mechanisms . Here we demonstrate that IFNβ is highly upregulated during latent infection , in the absence of detectable MHV68 replication . We identify an interferon-stimulated response element ( ISRE ) in the MHV68 M2 gene promoter that is bound by the IFNαβ-induced transcriptional repressor IRF2 during latency in vivo . The M2 protein regulates B cell signaling to promote establishment of latency and reactivation . Virus lacking the M2 ISRE ( ISREΔ ) overexpresses M2 mRNA and displays uncontrolled acute replication in vivo , higher latent viral load , and aberrantly high reactivation from latency . These phenotypes of the ISREΔ mutant are B-cell-specific , require IRF2 , and correlate with a significant increase in virulence in a model of acute viral pneumonia . We therefore identify a mechanism by which a gammaherpesvirus subverts host IFNαβ signaling in a surprisingly cooperative manner , to directly repress viral replication and reactivation and enforce latency , thereby minimizing acute host disease . Since we find ISREs 5′ to the major lymphocyte latency genes of multiple rodent , primate , and human gammaherpesviruses , we propose that cooperative subversion of IFNαβ-induced IRFs to promote latent infection is an ancient strategy that ensures a stable , minimally-pathogenic virus-host relationship .
The gammaherpesviruses ( γHVs ) establish life-long latent infection in memory B lymphocytes . The human γHVs Epstein-Barr virus ( EBV ) and Kaposi's sarcoma-associated herpesvirus ( KSHV ) are the causes of infectious mononucleosis and Kaposi's sarcoma ( KS ) , respectively [1] , [2] . γHV latency is a cofactor in the development of lymphomas , sarcomas , and carcinomas . Viral reactivation and neoplasms increase in immune compromised individuals , highlighting the need for immune surveillance to prevent severe disease [2] . Mechanisms of immune control of latent EBV and KSHV are not completely understood due to their human-specific host range . Murine gammaherpesvirus 68 ( MHV68 ) is closely related to the human γHVs and provides a genetic model to study γHV-immune interactions that regulate pathogenesis [3] , [4] . We previously uncovered an unexpected role for type I ( alpha/beta ) interferons ( IFNαβ ) during MHV68 latency [5] . IFNαβ are a family of antiviral cytokines whose expression is triggered by cellular sensors of viral nucleic acid that activate interferon regulatory factor ( IRF ) family transcription factors [6] . IRFs bind to interferon stimulated response elements ( ISREs ) in IFN gene promoters to trigger expression of IFNαβ . IFNαβ signaling via its heterodimeric receptor ( IFNAR1/2 ) induces a large family of interferon-stimulated genes ( ISGs ) that inhibit viral replication by multiple mechanisms . Once virus infection has been cleared , the IFNαβ-induced transcriptional repressor IRF2 exerts a negative feedback role to terminate IFNαβ expression and prevent inflammatory pathology [7] . Many viruses antagonize IFNαβ expression or ISG function to maximize replication [8] . However , the interactions between latent viruses and IFNαβ are largely unexplored . We found that mice lacking the IFNαβ receptor ( IFNAR1-/- ) exhibit increased MHV68 reactivation from latency in both splenic B cells and peritoneal macrophages [5] . This was unexpected since viral molecules that trigger IFNαβ production should be largely absent during latency , when infectious virus is undetectable using classical virologic assays . In addition , known antiviral functions of IFNαβ are critical during acute viral infection , but are thought to be dispensable once replication is controlled [6] . One clue to the mechanism of IFNαβ function during MHV68 latency came from our observation that the MHV68 latent gene M2 is specifically upregulated in splenocytes from IFNAR1-/- mice [5] . M2 is required for establishment of latency in splenic B cells following mucosal infection and is essential for reactivation from B cells [9] . While the precise function of M2 is not known , it interacts with B cell signaling molecules including fyn and vav1 , resulting in efficient entry of infected B cells into a germinal center ( GC ) reaction [10]–[13] . This suggests that M2 is a functional analog of the human γHV B cell signaling mimics LMP2A and K1 of EBV and KSHV , respectively [14] . M2 also promotes differentiation into plasma B cells , the main cell type that supports reactivation of MHV68 , EBV , and KSHV [15] . Thus , M2 plays important roles in both establishment of latency and reactivation . Upregulation of M2 in IFNAR1-/- mice suggested that latency and reactivation are directly regulated by IFNαβ-dependent modulation of M2 expression . Here we show that latent MHV68 infection triggers sustained , IFNαβ-driven expression of IRF2 , which binds an ISRE present in the M2 promoter . A mutant virus lacking the M2 ISRE ( ISREΔ ) exhibits uncontrolled replication and increased host lethality late in acute infection . During latency , ISREΔ overexpresses M2 mRNA , and displays increased viral load and aberrantly high reactivation . These phenotypes were absent in mice lacking B cells , IRF2 , or IFNAR1 . Thus , we demonstrate that MHV68 subverts IFNαβ-dependent IRF2 signaling to silence expression of a viral B cell signaling mimic , thereby preventing viral replication and reactivation . This demonstrates that viral promoters can cooperate with IFNαβ-induced host transcription factors to directly mediate the antiviral effects of IFNαβ . To our knowledge , this is the first example of viral cooperation with the IFNαβ system . We hypothesize that evolution of IFNαβ-responsive viral promoters provides a selective advantage , by curtailing replication and expansion of the latently-infected reservoir prior to severe host pathology , and by ensuring that reactivation occurs only when the microenvironment of the latent cell favors productive replication . Given the conservation of ISREs in latent promoters of EBV and KSHV [16]–[18] , we propose that this cooperative approach is a general regulatory strategy that arose during γHV-host coevolution .
We found a consensus ISRE in the M2 intron ( Figure 1A ) . Functional intronic ISREs have been reported , suggesting that this ISRE regulates the M2 promoter [19] , [20] . To determine whether the M2 ISRE binds host IRFs , we incubated M2 ISRE probes with nuclear proteins from splenocytes of latently-infected mice in electromobility shift assays ( EMSA ) . As a control , we mutated four residues essential for IRF binding ( ISREΔ , Figure 1A ) [21] . M2 ISRE and M2 ISREΔ probes formed distinct complexes with nuclear proteins ( Figure 1B ) . Only complexes formed with M2 ISRE were specific , since formation was inhibited with excess unlabeled M2 ISRE but not M2 ISREΔ probe ( Figure S1 ) . Two different antisera specific for IRF2 super-shifted M2 ISRE-bound complexes but not those bound to M2 ISREΔ ( Figure 1B ) . IRF2 is an essential component of these complexes , since they are not formed using nuclear extracts from IRF2-/- mice ( Figure 1C ) . To determine whether IRF2 binds to M2 ISRE in vivo we used chromatin immunoprecipitation ( ChIP ) from splenocytes of latently-infected mice ( Figure 1D ) . Anti-IRF2 antisera enriched DNA within one kilobase of the M2 ISRE , but not adjacent control regions . Interestingly , in two of three experiments , we also detected IRF2 binding to a region in the nearby M4 gene . Analysis of this region revealed a second consensus ISRE ( M4-ISRE , Figure 1D ) supporting the specificity of the assay . Thus , IRF2 binds the M2 ISRE during latent infection in the spleen . IRF2 is generally a transcriptional repressor , is constitutively expressed at low levels in many cell types including lymphocytes , and is upregulated by IFNαβ [21] , [22] . IFNAR1-/- mice display increased reactivation and upregulation of M2 [5] , suggesting that IFNαβinduces IRF2-dependent repression of M2 during latency . However , others have reported that IFNαβproteins are not detectable during acute MHV68 infection in the lung [23] . Therefore , we determined kinetics of IFNβand IRF2 expression during MHV68 infection in the spleen of wildtype , IFNAR1-/- , and IRF2-/- mice . Under these conditions , IRF2-/- mice experience no lethality , clear acute infection , and establish latency with no evidence of persistent lytic replication ( not shown and Table 1 ) . IFNβ and IRF2 transcripts were strongly induced in a time-dependent fashion during acute infection ( Figure 2A , D ) . Both transcripts were more highly induced during latent infection ( 16–28 days post infection ( dpi ) ) than at the peak of acute infection ( 4–9 dpi ) . Full induction of both transcripts required IFNAR1 ( Figure 2C , F ) , confirming that extracellular IFNαβ proteins are produced and functional . Consistent with the repressive role of IRF2 , IFNβ was significantly elevated during latent infection in IRF2-/- mice ( Figure 2B ) . These data demonstrate sustained expression of IFNβ and IRF2 at the major site of MHV68 latency . We generated two independent mutant viruses lacking the IRF contact residues in M2 ISRE ( ISREΔ1 and ISREΔ2 ) and a repaired marker rescue ( MR ) virus ( Figure 3A , B ) . ISREΔ1 replication was identical to MHV68 in murine embryonic fibroblasts ( MEFs ) ( Figure 3C , D ) or bone marrow-derived macrophages ( BMM ) ( Figure 3E , F ) , and ISREΔ1 replication was inhibited normally by pretreatment of cells with IFNβ ( Figure 3D , F ) . Thus , the M2 ISRE is not required for viral replication or inhibition by IFNβ in vitro . To determine whether the M2 ISRE regulates acute infection in vivo , we infected mice with MHV68 and ISREΔ1 and quantified viral titer in lung and spleen ( Figure 4 ) . At 4 dpi , replication of MHV68 and ISREΔ1 in lungs of wildtype mice of two genetic backgrounds was identical , indicating that the M2 ISRE is not required for early acute infection . In contrast , at 9 dpi , we observed a 20- to 30-fold increase of ISREΔ1 replication in both lung and spleen ( Figure 4A , D , G , J ) . Increased replication of ISREΔ1 persisted at 12 dpi ( 3- to 7-fold upregulated ) , but no infectious virus of either strain was detectable in spleen at 16 , 21 , or 28 dpi , indicating that clearance of ISREΔ1 acute infection is not delayed ( See Methods ) . Increased replication of ISREΔ1 was specific for the ISREΔ mutation , since it was observed during infection with ISREΔ2 and was restored to MHV68 levels in MR virus infection ( Figure S2 ) . Thus the M2 ISRE represses viral replication at late times of acute infection , suggesting that deletion of M2 ISRE allows MHV68 to bypass some component of the host response . To test whether the host control mechanism uncovered by M2 ISRE deletion requires IFNαβ , we compared MHV68 and ISREΔ1 replication in IFNAR1-/- mice . Replication of both MHV68 and ISREΔ1 was significantly upregulated in IFNAR1-/- mice compared to wildtype mice , but in the absence of IFNAR1 no difference in replication of MHV68 and ISREΔ1 was observed ( Figure 4B , E , H , K ) . As a control for specificity of the IFNAR1 signaling pathway , we infected mice lacking the IFNγ receptor ( IFNGR1-/- ) . These mice displayed increased early replication ( 4 dpi ) , but no difference was observed between MHV68 and ISREΔ1 at this time point . However , as observed in wildtype mice , replication of ISREΔ1 was increased 15- to 30-fold at 9 dpi in lung and spleen of IFNGR1-/- mice , and remained elevated 6- to 17-fold at 12 dpi ( Figure 4C , F , I , L ) . Thus , the M2 ISRE functions as a repressor of MHV68 replication at late times of acute infection , and acts by a mechanism that seems to require functional IFNαβ , but not IFNγ , signaling . However , it is possible that the high level of replication of both viruses in IFNAR1-/- mice may obscure the contribution of the M2 ISRE to replication . Since IFNαβ likely controls MHV68 replication by multiple mechanisms , we quantified replication of MHV68 and ISREΔ in IRF2-/- mice as a more specific test of the requirement of IFNAR1-dependent signaling in regulating MHV68 replication via the M2 ISRE . While the replication of ISREΔ1 was increased ∼100-fold relative to MHV68 in IRF2+/+ mice , in IRF2-/- littermates MHV68 replication and lytic gene expression rises precisely to the level of ISREΔ1 , and the two viruses are statistically identical ( Figures 5A , B ) . Importantly , titers in IRF2-/- mice are >10-fold lower than the maximum observed in IFNAR1-/- mice ( Figure 4B , E , H , K ) , suggesting that IRF2-independent increases in replication of ISREΔ virus should be evident if they existed . The absence of increased replication of ISREΔ in IRF2-/- mice suggests that the M2 ISRE functions solely in response to IFNαβ-dependent IRF2 to decrease replication . The observation that ISREΔ1 replicates at higher levels than MHV68 during acute infection was unexpected , since M2 is dispensable for acute replication in vitro and in vivo [9] . All known functions of M2 are B-cell-specific and include inducing B cell entry into and egress from the GC reaction , and triggering B cell differentiation into plasma cells , the predominant cell type supporting viral reactivation in vivo [10] , [15] . However , latently-infected B cells are detectable in the lung early during acute MHV68 infection [24] . Thus , we reasoned that increased replication of ISREΔ during late acute infection may be due to premature reactivation in infected B cells . To test this hypothesis , we infected B cell deficient mice ( μMT-/- ) with MHV68 and ISREΔ1 and quantified replication ( Figure 5C , D ) . MHV68 replication increased approximately 20-fold in µMT-/- mice , which we speculate may be due to redirection of MHV68 virions to a purely lytic infection in the absence of B cells as targets for latency . Alternatively , B cells may exert an indirect antiviral effect on MHV68 replication . However , this effect must still require the M2 ISRE for function since ISREΔ replication is identical in wildtype and µMT-/- mice . ISREΔ1 replication was indistinguishable from MHV68 replication in lungs of µMT-/- mice at 4 , 9 , and 12 dpi ( Figure 5D and data not shown ) . Neither MHV68 nor ISREΔ1 virus was detected in spleen of µMT-/- mice , consistent with a critical role for B cells in spread ( [25] and data not shown ) . Thus , the increased replication phenotype of ISREΔ1 requires B cells , suggesting that newly infected B cells require IFNAR1- and IRF2-dependent repression of M2 expression to prevent premature viral reactivation . To determine if the M2 ISRE regulates M2 expression during latency , we quantified M2 mRNA in spleens at times when lytic replication is absent ( Figure 6 ) . M2 mRNA was significantly upregulated at both early ( 16 dpi ) and later ( 28 dpi ) times during latent ISREΔ1 infection ( Figure 6A–C ) . Upregulation was specific for M2 mRNA and was not observed for viral M3 or M9 transcripts ( Figure 6D–I ) . Importantly , in IRF2-/- mice M2 transcript expressed by MHV68 increased precisely to the level observed in ISREΔ1 infection , while M3 and M9 expression efficiency were unaltered . Thus , M2 expression is specifically repressed during latency by an M2 ISRE- and IRF2-dependent mechanism . When we compared the kinetics of IFNβ , IRF2 , and M2 expression in the spleen , we detected elevated IFNβ and IRF2 mRNA in spleen by 4 dpi ( Figure S3 ) . At this timepoint no M2 mRNA is detectable , likely because virus has not yet reached the spleen ( Figure 4 ) . From 9–28 dpi with MHV68 , M2 mRNA is present in the spleen at a low but constant level . In contrast , M2 expression is upregulated ∼3–4-fold at all time points in ISREΔ infection , indicating that the M2 ISRE reduces , but does not completely silence , M2 expression ( Figure S3 ) . Thus , our data indicate that M2 expression is controlled by at least two promoter elements: a 5′ promoter proximal to the transcription start site [20] and the intronic ISRE we report here ( Figure 1 ) . M2 overexpression is sufficient to drive viral reactivation from plasma B cells [15] . To determine whether repression of M2 by IRF2 decreases MHV68 reactivation , we performed ex vivo reactivation assays with splenocytes and peritoneal exudate cells ( PECs ) from mice infected with MHV68 , ISREΔ1 , or ISREΔ2 ( Figure 7 , Figure S4 , Table 1 ) . Splenocytes from wildtype mice infected with either ISREΔ1 or ISREΔ2 showed a significant four-fold increase in reactivation compared to MHV68 at both 16 and 28 dpi ( Table 1 and Figure 7A , C ) . At 16 dpi , increased reactivation was solely attributable to increased reactivation efficiency , since the frequency of latently-infected cells was equivalent ( ∼1∶1400 ) in mice infected with either virus . However , at later times , increased reactivation during ISREΔ infection was a composite effect of both increased numbers of latently-infected cells ( Figure S4 ) and increased efficiency of reactivation ( Table 1 ) . Increased reactivation efficiency of ISREΔ required IFNAR1 and IRF2 , since in IFNAR1-/- and IRF2-/- mice MHV68 and ISREΔ mutant viruses reactivated with identical frequencies that are increased relative to those observed in MHV68-infected wildtype mice ( Figure 7B–D , Table 1 ) . Importantly , in this assay , <10% of latently-infected cells reactivate ( Table 1 ) , permitting sufficient upward dynamic range for IFNAR1-independent or IRF2-independent effects of the M2 ISRE to be observable if they existed . Increased reactivation of ISREΔ is likely B cell-specific , since reactivation of MHV68 , ISREΔ1 , and ISREΔ2 from PECs was identical under all conditions ( Table 1 and Figure 7E–H ) . The major latent cell in the spleen is the B cell , while in PECs most latent virus resides in macrophages [26] . Taken together , these genetic data strongly suggest that MHV68 reactivation from B cells is repressed by IFNαβ-driven , IRF2-mediated repression of M2 expression , acting through the M2 ISRE .
We have uncovered a previously unappreciated mode of interaction between viruses and IFNαβ: rather than evading this antiviral system , MHV68 directly cooperates with it to silence replication during establishment of latency . This strategy relies on IFNαβ-induced IRF2 to regulate critical cell differentiation decisions following viral infection of B cells ( Figure 9 ) . Shortly after B cell infection , the M2 ISRE is either unoccupied or may be bound by transactivating IRFs ( IRF”X” , Figure 9A ) . M2 expressed at this time drives B cells into a GC reaction , resulting in expansion of latently-infected memory and plasma cells . As replication peaks , IFNαβ induces IRF2 , which binds the M2 ISRE and represses M2 transcription . M2 silencing would decrease entry of infected B cells into the GC ( Figure 9B ) , reducing overall latent load [10] . M2 is sufficient to promote differentiation into plasma cells [15] and the majority of viral reactivation is derived from plasma cells [15] . Thus , IRF2-dependent M2 repression is also expected to decrease reactivation . When we perturb this regulatory switch using either IRF2-/- mice or ISREΔ virus , the result is a substantial increase in viral replication during late acute infection ( Figure 4 ) , which we attribute to premature reactivation from newly infected B cells ( Figure 5 ) driven toward plasma cell differentiation by overexpression of M2 ( Figure 6 ) . This increase in reactivation is still evident during ISREΔ latency , and results in increased latent load over time ( Figure 7 ) . Additional IFNαβ-dependent mechanisms exist to control MHV68 replication and reactivation , since the M2 ISREΔ mutation does not fully recapitulate the dysregulation of these processes observed in IFNAR1-/- mice [5] . While the simplest mechanism that is consistent with our genetic and biochemical data involves IFNαβ-induced IRF2 binding to the M2 ISRE to reduce M2 expression in infected B cells during latency expansion in the spleen , other interpretations are possible . For example , it is also conceivable that B cells and IRF2 exert M2 ISRE-dependent control of viral replication and reactivation in a trans-acting manner , rather than directly in the infected B cell . We propose the term “cooperative subversion” to describe this regulatory approach . The cooperative nature of the strategy is evident by the lack of IFNAR1-dependent inhibition of ISREΔ replication we observe late in acute infection ( Figure 4 ) , which correlates with increased lethality in a moderately immune compromised ( IFNγ-/- ) host ( Figure 8 ) . This indicates that a primary function of the M2 ISRE is to reduce viral replication in cooperation with the host IFNAR1 signaling pathway . To our knowledge this is the first element identified in a herpesvirus genome that directly engages a host ISG to decrease viral replication . We hypothesize that this mechanism provides a developmental switch to MHV68: once it has established a latent load that assures life-long persistence , viral gene expression shifts to a pattern that will prevent host pathology and lethality , minimize the risk of B cell transformation , and reduce viral antigen presentation . This also provides a potential strategy for γHVs to target reactivation to periods of localized immune quiescence during long-term latency , by permitting high level M2 expression only in microenvironments where IFNαβ secretion has decreased , when replication is more likely to be productive . Thus , we believe this mechanism is simultaneously cooperative and subversive , but must be contrasted with overt IFNαβ evasion strategies that are well documented during lytic infection with many viruses [8] . The γHVs are uniquely suited to cooperation with the host immune response . They need minimal replication to establish latency , and instead rely on virus-driven proliferation of B cells to seed the host . Indeed , the frequency of MHV68 latency is independent of inoculum dose [28] , and replication-defective MHV68 can establish latent infection [24] , [29] . γHVs can rely on lifelong transmission to spread to a new host , obviating the need for high-level persistent replication; asymptomatic reactivation at mucosal surfaces is instead the rule [30] . Since γHVs rely on the health of the host to promote spread , it is likely a selective advantage for them to cooperate with the host immune response to prevent unbridled amplification of latent B cells and predisposition of the host to neoplasia . Consistent with this hypothesis , there is evidence that the dominant T cell epitopes of EBV are selected for conservation , rather than evasion [31] . Our data demonstrate that cooperation with innate antiviral cytokines may also function during acute and latent infection . Whether this attenuating , cooperative effect of the M2 ISRE is the primary function selected during virus-host coevolution is not directly discernable from our studies . We found unexpectedly high-level IFNβ expression during latency , at times where we detect no infectious virus ( Figure 2 ) . In subsequent studies , we found IFNβ-producing cells in the spleen of latent mice at 48 dpi , and significant upregulation of ISGs at 90 dpi ( data not shown , manuscript in preparation ) . It is not clear what viral triggers or host sensors lead to this sustained IFN production . However , these data demonstrate that the IFNαβ-driven response is not limited to acute infection , but extends well into latency . Our data also indicate that host genes may not directly mediate all “antiviral” effects of IFNαβ , but that viral elements may be required to repress replication in response to IFNαβ . By 28 dpi , nearly half of the antiviral effect of IFNαβ in the spleen is mediated by the M2 ISRE , since MHV68 reactivation in IFNAR1-/- mice is upregulated 7 . 9-fold yet deletion of the M2 ISRE alone upregulates reactivation 3 . 7-fold ( Table 1 , 129S2 background , 28 dpi ) . Interestingly , the M2 ISRE has no apparent function during latency in peritoneal macrophages , a cell type where latency and reactivation are also independent of M2 ( Figure 7 ) [9] . This confirms that the ISREΔ mutation does not dysregulate latency and reactivation in all cell types , and indicates that IFNAR1-dependent pathways active in latent macrophages remain to be defined . Our data indicate a novel , direct antiviral function for IRF2 during γHV latency . Although IRF2-/- mice have immune defects that disrupt control of some acute viral infections [7] , [32]–[34] , they clear MHV68 replication , establish latency at frequencies nearly identical to IRF2+/+ littermates , and regulate viral reactivation from the peritoneal compartment normally . This indicates that the immune modulatory functions of IRF2 are not essential for an effective response to MHV68 infection . In IRF2+/+ littermates , ISREΔ virus replicates to 100-fold higher titer in the lung , expresses significantly more M2 mRNA , and reactivates with enhanced efficiency relative to MHV68 . Importantly , we found that MHV68 replication , M2 expression , and reactivation rises precisely to that of ISREΔ in IRF2-/- mice ( Figures 5 , 6 , 7 , Table 1 ) . Thus , the M2 ISRE can only repress M2 expression and restrain MHV68 replication and reactivation in a host that expresses IRF2 . This confirms that these phenotypes are not the result of generalized immune defects during latency in IRF2-/- mice , but almost certainly require the interaction between IRF2 and the M2 ISRE observed in Figure 1 . Since IRF2 has oncogenic properties [35] , our data raise the question of whether γHV-induced IRF2 may play a tumor-promoting role during γHV latency . Our data indicate that M2 expression is regulated by two distinct promoter elements . The first identified M2 promoter is located 5′ to the M2 transcription initiation site , is functional in murine B cells , and binds undefined transcription factors [20] . Our data indicate that the M2 ISRE can decrease the firing rate of this 5′ promoter when it is occupied by IRF2 . IRF2 generally functions as a transcriptional repressor , but its function is modulated by several posttranslational modifications , including proteolysis ( which exposes a transcriptional transactivation domain ) [36] , sumoylation [37] , acetylation [38] , phosphorylation [39] , and interaction with other IRFs and co-factors [40] . The modification state of IRF2 and levels of other IRFs that compete for binding to the M2 promoter is likely to be dynamically regulated . Importantly , IRF2 binding to the M2 ISRE does not completely silence the M2 locus , thereby allowing expression of reduced M2 levels in the face of the host IFNαβ response . We speculate that during establishment and early expansion of latency , IRF2-mediated reduction in M2 expression is required to prevent untimely viral reactivation , which can be triggered directly by M2 overexpression [15] . A spatiotemporally regulated balance of IRFs likely determines the expression of M2 at distinct stages of the MHV68 latent life cycle . It is noteworthy that multiple intrinsic and extrinsic stimuli that induce γHV reactivation activate IRFs . Toll-like receptor ( TLR ) stimulation with multiple viral and bacterial molecules triggers reactivation of KSHV and MHV68 [41] , [42] . IRFs that are activated downstream of TLR stimulation include IRFs 1 , 3 , 5 , and 7 [43] , [44] . In addition , DNA damage both activates IRF5 and induces MHV68 reactivation [45] , [46] . These data suggest that TLR- or stress-induced IRF activation may serve to displace IRF2 from the M2 ISRE , inducing M2 expression and reactivation . However , there is no evidence for a transactivator bound to the M2 ISRE at the time points we assess , since M2 transcript levels are upregulated to the same extent when either the ISRE or IRF2 is deleted ( Figure 6 ) . It has been reported that overexpression of M2 may impair IFNαβ-induced signaling pathways [47] . Although this function has not been confirmed in lymphocytes expressing physiologic levels of M2 , this strategy may enable sustained M2 expression once reactivation is induced , by acting as a negative feedback loop to prevent IFNαβ-driven IRF2 expression and silencing of M2 transcription during latency establishment or reactivation . Complex but poorly understood relationships exist between the human γHVs and IRFs . KSHV encodes viral IRF homologs ( vIRFs ) that modulate function of host IRFs [48] , [49] . vIRF3 is expressed during KSHV latency , when it antagonizes IRF5 and p53 and enhances transactivation by IRFs 3 and 7 [49] , [50] , and is required for latent cell proliferation and survival [51] . EBV interacts with numerous IRFs to regulate latent promoters . IRF2 binds to the EBNA-1 Qp promoter during restricted latency programs I and II [18] , [52] . Although IRF2 represses Qp in some EBV-infected B cell lines [18] , other reports indicate that it can upregulate Qp-driven EBNA-1 expression [52] , [53] . In addition , the promoter of EBV LMP1 , a viral CD40 signaling mimic , contains an ISRE and is induced by IRF7 and repressed by IRF5 ( Table 2 ) . The consequences of EBV and KSHV promoter-IRF interactions for the infected cell in vivo are unknown . Several of the IRFs implicated in regulating KSHV and EBV genes ( including IRFs 2 , 5 , and 7 ) are induced by IFNαβ [21] . Little attention has been given to the possibility that IRF-induced or -repressed EBV and KSHV latent gene expression may be responsive to the inflammatory environment of the infected cell . We demonstrate that a latent gene controlling B cell differentiation and reactivation is repressed by IFNαβ . Interestingly , LMP1 and M2 are located in homologous regions of the viral genome [3] , and our work here demonstrates that like LMP1 , M2 is regulated by a conserved ISRE . While the importance of IRF-mediated LMP1 regulation in vivo is unknown , our data suggest that EBV-infected cells may utilize this circuit to fine-tune the balance between latent B cell proliferation and reactivation in response to host inflammation . Intriguingly , we find consensus ISREs in the 5′ regions of the major lymphocyte immortalization genes of EBV , KSHV , the primate γHV herpesvirus saimiri , and two newly sequenced rodent γHVs [54] , suggesting that subversion of host IRFs to regulate the switch between lytic and latent infection is an evolutionarily ancient invention ( Table 2 ) . Future studies with MHV68 will enable dissection of the dynamics of viral promoter/IRF interactions that may permit rational intervention to manipulate the balance between viral latency , reactivation , and oncogenesis .
This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health . Mice were handled according to all applicable institutional , state , and federal animal care guidelines , under animal care protocols approved by the Purdue University ( animal welfare assurance #A3231-01 , protocol #06-115 ) and Wake Forest University Animal Care and Use Committees ( animal welfare assurance #A3391-01 , protocol #A11-007 ) . Veterinary technicians or laboratory staff assessed animal health at least once daily . Moribund mice were humanely euthanized . All cells were maintained in DMEM containing 10% fetal bovine serum ( DMEM/10 ) . MEFs were harvested from embryonic d15–18 129S2 and C57BL6/J mice . Bone marrow was harvested from the femur of 129S2 mice and differentiated in vitro to produce BMM [55] . Bone marrow was cultured for four days on 100 mm polystyrene dishes in 10 ml endotoxin-free DMEM/10 supplemented to contain 20% vol/vol L929 cell supernatant , 5% vol/vol horse serum , 2 mM L-glutamine , and 1 mM sodium pyruvate . On day four of differentiation , 10 ml of endotoxin-free DMEM/10 supplemented to contain 10% vol/vol L929 cell supernatant , 5% vol/vol horse serum , 2 mM L-glutamine , and 1 mM sodium pyruvate was added to each plate . On day seven of differentiation , cells were detached from dishes using PBS ( Ca2+/Mg2+-free , 1 mM EDTA ) and scraping . Wildtype and recombinant virus stocks were generated from wildtype MHV68 propagated as a bacterial artificial chromosome ( BAC ) [56] . To generate infectious virus stocks , column-purified BAC DNA was transfected into BALB/3T12 cells ( ATCC CCL-164 ) stably transduced with cre recombinase to permit deletion of bacterial sequences . Viruses were passaged at low multiplicity of infection for three generations in BALB/3T12-cre cells prior to use . Organs were disrupted with 1 mm silica beads using a Minibeadbeater 16 ( Biospec Products ) in 1 ml DMEM/10 . Viral titers were determined by plaque assay on BALB/3T12 monolayers [5] . For in vitro growth curves , 6×104 MEFs or BMM were plated in 48 well tissue culture treated plates . Immediately after plating , MEFS were incubated with 500 U/ml rIFNβ ( PBL laboratories ) overnight ( ∼18 hours ) prior to infection . BMMS were kept at 37°C for 24 hours prior to overnight treatment with IFNβ . Cells were infected with 0 . 1 plaque-forming units ( PFU ) MHV68 or ISREΔ per cell in an inoculum volume of 0 . 1 ml DMEM/10 for one hour at 37°C . Inocula were aspirated , cells were washed twice with 37°C PBS , and incubated in DMEM/10 . Plates were frozen at −80°C at indicated time points . Plates were frozen and thawed twice prior to plaque assay . Recombinant viruses were generated using BAC mediated mutagenesis as described [56] . The M2 locus ( nt . 3791–4700 ) relative to Genbank Accession U97553 . 2 , [3] was replaced with a kanamycin resistance cassette to generate M2-Kan BAC . Two M2 homology arms ( 5′ arm: nt 3301–3790 , 3′ arm: nt 4701–5201 ) were amplified by PCR and cloned on either side of a kanamycin resistance cassette in allelic exchange mating vector pGS284 to generate pGS284/M2/Kan . MHV68 BAC was mated to pGS284/M2/Kan by cross streak on LB agar plates , and the expected genomic configuration of kanamycin-resistant clones ( M2-Kan BAC ) arising from host strain intersections were confirmed using a minimum of four restriction endonucleases that yield diagnostic fragment lengths . M2-Kan BAC was mated to pGS284 containing the entire M2 locus ( generated by PCR using 5′ homology arm sense and 3′ homology arm antisense primers ) that was mutagenized via PCR to encode the M2 ISREΔ mutations as indicated in Figure 1 ( pGS284/M2/ISREΔ ) . Kanamycin sensitive recombinants were identified by replica plating and expected genomic configuration confirmed by restriction digest . Two independent ISREΔ mutant clones were generated using independent stocks of wildtype MHV68 BAC . Mating of ISREΔ1 to pGS284/M2/Kan and subsequent replacement of the M2 locus by mating to pGS284 containing the wildtype M2 sequence generated a genetically repaired marker rescue virus , M2-MR . All PCR amplified homology arms and mutagenized sequences were confirmed by DNA sequencing over the entire length of the construct , and resulting mutant viral BAC DNA was directly sequenced to confirm incorporation or repair of mutations . To generate infectious virus stocks , column-purified BAC DNA ( 4 µg ) was transfected using Fugene HD ( Roche ) into BALB/3T12 cells that were stably transduced with cre recombinase to permit deletion of the BAC backbone . BAC sequence elimination was confirmed after three passages in BALB/3T12-cre using indirect fluorescence for EGFP expressed from the BAC locus . Age- and sex-matched mice ( 7-12 weeks of age ) were used for all experiments . Wildtype , IFNAR1-/- , and IFNGR-/- mice on 129S2 ( old designation , 129/SvPas ) background have been described [57] . Wildtype , IFNAR1-/- , IFNGR-/- mice on C57BL6/J background were obtained from Dr . Herbert Virgin ( Washington University ) . Dr . Stephanie Vogel ( University of Maryland ) donated IRF2-/- mice on C57BL6/J background [58] . B cell deficient ( μMT-/- ) and IFNγ-/- BALB/c ( strain C . 129S7 ( B6 ) -Ifngtm1Ts/J ) were purchased from Jackson laboratories and are the only genotypes used in this study that were not derived from in-house breeding . Isoflurane-anesthetized mice received 100 PFU intranasally in 40μL of DMEM/10 . For viral pneumonia induction ( Figure 8 ) , IFNγ-/- BALB/c received 4×105 PFU intranasally . Mice were humanely euthanized in Isoflurane prior to tissue harvest . Nuclear extracts were prepared from splenocytes of latently-infected C57BL6/J mice 28–35 dpi using Pierce NE-PER kit and protein concentration was determined using Bio-Rad RC/DC Kit . For EMSA , probes used were M2 ISRE: 5′-TTACCTGAAAACGAAACCTCATCA-3′ and M2 ISREΔ: 5′-TTACCTGGAACCTGAACCTCATCA-3′ . 32P-labeled complementary oligonucleotides were hybridized to generate double stranded ( ds ) probes . Ds probes were separated from free radiolabeled dUTP by size exclusion chromatography using Sephadex G-50 columns ( Roche ) . Radiolabeled , ds probes were incubated with nuclear extracts and resolved on acrylamide gel [59] . Five µg of protein was incubated in a reaction with 1X binding buffer ( 40 mM KCl , 20 mM HEPES pH 7 . 6 , 1 mM MgCl2 , 1 mM EGTA , 0 . 5 mM DTT ) , 0 . 32 mg/ml poly dI-dC ( Sigma-Aldrich ) , 0 . 02 mg/ml plasmid pgL4 . 10 , 4 mM AMP ( Sigma-Aldrich ) and 1×105 cpm of radiolabeled probes in a total volume of 12 . 5 µl at room temperature for 30 minutes . Complexes were resolved on 6% nondenaturing acrylamide-20 mM TBE gel at 4°C . For supershift , two µg of gel-shift certified antisera raised against mouse IRF2 ( Santa Cruz #H229 and #C19 ) were added to gel shift reactions . For competition assays , 32P-labeled M2 ISRE probe and nuclear extract were incubated with increasing concentrations of ds unlabeled M2 ISRE or M2 ISREΔ probes . Dried gels were exposed to storage phosphorimager plates and images analyzed using Bio-Rad PDQuest software . For ChIP , splenocytes ( 6×107 ) from latently-infected 129S2 mice were fixed in 1% formaldehyde , washed in PBS , and sheared using a Misonix S3000 Sonicator . Resulting chromatin had an average length of 500–1000 base pairs . Chromatin was incubated overnight with two µg anti-IRF2 ( H229 , Santa-Cruz Biotech ) and immunoprecipitated with protein A/G sepharose . Immunoprecipitated DNA was reverse-crosslinked , phenol/chloroform extracted , ethanol precipitated , and amplified using conditions , PCR primers , and thermal cycling parameters detailed in Supporting Protocol S1 . Control ( no antibody , or irrelevant rabbit antiserum ) immunoprecipitated chromatin yielded no amplicons for any primer set ( not shown ) . Total RNA was isolated from intact organs ( during lytic infection ) or erythrocyte-depleted splenocytes ( during latent infection ) by silica bead disruption in Trizol ( Invitrogen ) and subjected to RNA cleanup ( Qiagen RNAeasy Kit ) and DNAse treatment ( Ambion Turbo DNAse Kit ) . Total RNA ( 1 . 5 µg ) was used for cDNA synthesis ( Invitrogen Superscript Kit ) followed by real time PCR on an ABI 7300 using primers for host HPRT , IFNβ , IRF2 or viral M2 , M3 , or M9 genes . The primers for housekeeping gene HPRT , IFNβ , IRF2 or viral gene M2 span exon-intron junctions , and all amplicons were resolved on agarose gel electrophoresis to confirm predicted size . Amplicons for M2 were sequenced and confirmed that ISREΔ mutations did not alter M2 splicing . M3 and M9 are unspliced viral transcripts; parallel reactions performed in the absence of reverse transcriptase indicated that samples were free from contaminating viral genomic DNA . For quantitation of viral episome number , DNA was harvested from erythrocyte-depleted splenocytes and quantitative PCR analysis performed using primers specific to GAPDH or v-cyclin ( ORF72 ) genomic DNA . Detailed analysis and normalization equations are described in Supporting Protocol S2 . Frequencies of viral genome positive and reactivating cells were determined as described [5] . Briefly , on the indicated day post infection mice were euthanized and spleen and peritoneal exudate cells ( PECs ) removed . Spleens were homogenized to single-cell suspensions , erythrocytes hypotonically lysed , and cell viability and concentration determined . Cells were serially diluted and plated immediately on indicator MEFs for the purposes of assessing viral reactivation or were cryopreserved in 10% DMSO . To determine the frequency of cells reactivating lytic viral replication , freshly explanted cells were serially diluted and plated in 96-well tissue culture plates seeded with 104 C57BL6/J MEFs per well . Twenty-four replicates of each cell dilution were plated . Cells were co-cultured for 21 days , and viral reactivation was scored by visual inspection for cytopathic effect ( CPE ) . To control for possible persistent lytic viral replication in vivo , the extent of preformed lytic virus in explanted cell populations was quantitated by mechanical disruption of parallel cell samples using 0 . 5 mm silica beads prior to plating on indicator MEFs . Such mechanical disruption kills >99% of cells but has minimal effect on infectious virus . Under the infection conditions used in these experiments , no significant virus persistence was observed in any genotype of mice infected with MHV68 or ISREΔ viruses . To determine the frequency of explanted cells that harbored viral genome , cryopreserved cells were thawed , counted , and serially diluted in 96-well thermal cycling plates . Cells were lysed by overnight incubation with proteinase K . Single-copy-sensitivity nested PCR was performed using primers specific for MHV68 ORF72 . Amplicons were visualized by agarose gel electrophoresis . Twelve replicates of each cell dilution were analyzed in separate PCR reactions . Statistical analyses and nonlinear regression were performed using GraphPad Prism 5 . 0 ( GraphPad Software , San Diego , CA ) . Data from limiting dilution viral genome and viral reactivation assays were fitted to a sigmoidal dose-response curve by nonlinear regression to determine the concentration of explanted cells required to achieve 63% viral DNA-positive PCR reactions or CPE-positive reactivation wells . This cell number was defined according to the Poisson distribution as the reciprocal frequency of viral latency or viral reactivation , respectively , as listed in Table 1 .
|
Herpesviruses establish life-long infection in a non-replicating state termed latency . During immune compromise , herpesviruses can reactivate and cause severe disease , including cancer . We investigated mechanisms by which interferons alpha/beta ( IFNαβ ) , a family of antiviral immune genes , inhibit reactivation of murine gammaherpesvirus 68 ( MHV68 ) . MHV68 is related to Epstein-Barr virus and Kaposi's sarcoma-associated herpesvirus , human gammaherpesviruses associated with multiple cancers . We made the surprising discovery that during latency , MHV68 cooperates with IFNαβ to inhibit its own replication . Specifically , a viral gene required for reactivation has evolved to be directly repressed by an IFNαβ-induced transcription factor , IRF2 . Once virus replication has triggered sufficient IFNαβ production , expression of this viral gene is reduced and reactivation efficiency decreases . This strategy safeguards the health of the host , since a mutant virus that cannot respond to IRF2 replicates uncontrollably and is more virulent . Viral sensing of IFNαβ is also potentially subversive , since it allows MHV68 to detect periods of localized immune quiescence during which it can reactivate and spread to a new host . Thus , we highlight a novel path of virus-host coevolution , toward cooperative subversion of the antiviral immune response . These observations may illuminate new targets for drugs to inhibit herpesvirus reactivation or eliminate herpesvirus-associated tumors .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"animal",
"models",
"animal",
"models",
"of",
"infection",
"model",
"organisms",
"viral",
"immune",
"evasion",
"immunity",
"virology",
"innate",
"immunity",
"viral",
"persistence",
"and",
"latency",
"immune",
"defense",
"biology",
"microbiology",
"mouse",
"pathogenesis"
] |
2011
|
A Gammaherpesvirus Cooperates with Interferon-alpha/beta-Induced IRF2 to Halt Viral Replication, Control Reactivation, and Minimize Host Lethality
|
Influenza virus infection remains a public health problem worldwide . The mechanisms underlying viral control during an uncomplicated influenza virus infection are not fully understood . Here , we developed a mathematical model including both innate and adaptive immune responses to study the within-host dynamics of equine influenza virus infection in horses . By comparing modeling predictions with both interferon and viral kinetic data , we examined the relative roles of target cell availability , and innate and adaptive immune responses in controlling the virus . Our results show that the rapid and substantial viral decline ( about 2 to 4 logs within 1 day ) after the peak can be explained by the killing of infected cells mediated by interferon activated cells , such as natural killer cells , during the innate immune response . After the viral load declines to a lower level , the loss of interferon-induced antiviral effect and an increased availability of target cells due to loss of the antiviral state can explain the observed short phase of viral plateau in which the viral level remains unchanged or even experiences a minor second peak in some animals . An adaptive immune response is needed in our model to explain the eventual viral clearance . This study provides a quantitative understanding of the biological factors that can explain the viral and interferon kinetics during a typical influenza virus infection .
Despite vaccines and antiviral agents , influenza A virus infection remains a major public health problem worldwide . Seasonal and pandemic influenza results in approximately 3 to 5 million cases of severe illness and approximately 250 , 000 to 500 , 000 deaths worldwide [1] . Influenza viruses primarily infect and replicate in epithelial cells [2] . The immune response to influenza virus infection plays an important role in controlling the virus within a host . The nonspecific innate immune response provides the first line of defense , which reacts immediately upon infection and involves generating a variety of chemotactic , proinflammatory and antiviral cytokines [3] . An important cytokine produced during the innate immune response is type I interferon ( mainly IFN-α/β ) . IFN-α/β has been shown to stimulate resistance to infection in the neighboring cells by inducing the expression of many IFN-stimulated gene products , including antiviral proteins , such as protein kinase R , PKR [4] . Depletion of key IFN signaling proteins in mice results in greater mortality , accompanied by systemic ( as opposed to respiratory-restricted ) infection [5] . In addition , IFN is able to activate immune system cells , such as natural killer ( NK ) cells , during the early stage of infection , which can destroy infected cells [6]–[10] . The secretion of IFN-α/β by infected epithelial cells is also important for the initiation of the antigen-specific adaptive immune response [11] , [12] , which in mice takes approximately 5 days to begin in the lung [13] . The adaptive immune response mainly consists of cytotoxic CD8+ T cells eliminating infected cells and antibodies neutralizing the virus [11] . It is important for clearing the virus and provides immunity against future influenza virus infections . Because of limited information about influenza pathogenesis and the host immune response in humans , various animal models , such as mice , ferrets , and horses [14]–[17] , have been used to obtain a better understanding of the biological mechanisms underlying viral control . A number of mathematical models have been developed to study the dynamics of influenza virus infection and immune responses [13] , [18]–[28] ( also see recent reviews in [29]–[31] ) . By fitting a simple viral dynamic model to the data derived from 6 experimentally infected human volunteers , Baccam et al . [20] showed that target cell limitation can explain the kinetics of influenza A virus infection in humans . Both innate [18] , [20] , [28] and adaptive immune responses [21] , [22] , [24] have also been incorporated into the basic model to evaluate the effect of immune responses on viral control . In a recent study , Miao et al . [13] quantitatively investigated the innate and adaptive immune responses to primary influenza A virus infection in mice . They compared the half-life of infected epithelial cells and free virus before and during a virus-specific immune response ( about 5 days post-infection ) . Lee et al . [27] developed a two-compartment model to study the contributions of different factors , such as antigen presentation and activation of naive T and B cells , CD4+ T cell help , CD8+ mediated cytotoxicity , and antibody , to the control of influenza A virus infection . These studies provide a quantitative understanding of the host immune response in controlling virus replication . The relative contributions of target cell availability and immune responses to viral control remain unclear . In a recent study , Saenz et al . [19] estimated the numbers of viral-antigen-positive cells in the lungs of ponies at days 2 . 5 , 4 . 5 , and 5 . 5 after challenge with equine influenza virus ( EIV ) . The result indicated that up to 5% of bronchiole cells were infected at any one time , yielding an estimated total cell loss of about 27% by the end of the infection . This suggests mechanisms for viral control in addition to target cell depletion [20] , and motivates the development of a model that includes a strong innate immune response to explain the clearance of virus during infection [19] . However , the model in [19] is unable to capture a number of important features of the viral kinetics observed in 6 ponies , e . g . , the viral peak in most of the ponies , the rapid and substantial viral decline after the peak ( 2 to 4 log decline within 1 day ) , and a short plateau phase in which the viral load remained unchanged or even experienced a minor second peak in some ponies [19] . In this study , we develop mathematical models based on several possible biological mechanisms that attempt to explain all of these observations . Our objective is to investigate which biological parameters can give rise to the viral load change observed during an uncomplicated influenza virus infection .
The data we studied were from an experimental challenge of 6 unvaccinated ponies infected with EIV A/eq/Kildare/89 ( H3N8 ) [16] . Nasal secretions ( NS ) were collected daily for 10 days post-challenge and number of copies of influenza virus RNA per milliliter ( ml ) was quantified . Blood samples were also collected to quantify the fold changes in cytokine expression including IFN for days 1 through 5 post-challenge compared to the day prior to challenge . We used both the viral load and the IFN fold change data in this study . High antibody titers were detected by the single radial haemolysis ( SRH ) assay 14 days post-challenge in the horses . Upon infection , the viral load increased rapidly and reached its peak at day 2 for all ponies . There was a wide variation in the peak level . The highest was approximately 108 copies of viral RNA/ml of NS ( pony 2 ) , while the lowest was 104 copies/ml of NS ( pony 6 ) . After the peak , the viral load experienced a rapid and substantial decline ( about 2 to 4 logs within 1 day ) . All the ponies had a viral plateau and some experienced a minor but obvious second peak . After the viral plateau/second peak , there was a second viral decline starting around day 6 . In 4 out of the 6 ponies , the viral load decreased to below the detection limit by day 8 . The rest of the ponies had undetectable viral load at day 9 . During the infection , IFN expression increased substantially reaching a peak on day 2 in 5 of the 6 ponies , followed by a rapid decrease to the pre-infection level [16] , [19] . The peak of IFN-fold change ranged from approximately 1 ( pony 3 ) to more than 10 ( pony 6 ) . We developed a model to study the within-host dynamics of EIV infection in horses . It is described by the following system of equations ( 1 ) The model has five variables: target cells ( T ) , productively infected cells ( I ) , uninfected cells that are refractory to infections ( R ) because of IFN-induced antiviral effect [32] , free virus ( V ) , and IFN ( F ) . The term βVT represents the rate of infection when virus encounters susceptible target cells . IFN induces an antiviral effect and enables uninfected cells to become refractory to infection at rate . Cells in the refractory state revert back to the susceptible state at rate ρ . Infected cells are assumed to die at per capita rate δ . Prior to the emergence of the antigen-specific adaptive immune response , we assume δ is a constant δI . This rate ( δ ) becomes δA = δm- ( δm-δI ) e−σ ( t-μ ) after the adaptive immune response emerges , where μ is the time at which the adaptive immune response emerges , δm is the maximum death rate of infected cells in the presence of an adaptive immune response , and σ determines how fast the death rate increases from δI to the saturation rate δm . Because we only model the dynamics for a few days after the adaptive immune response emerges , we modify the time-varying death rate to δA = δIeσ ( t-μ ) without using the maximum constant δm . In this way , the number of parameters introduced is reduced by 1 . Another method that explicitly includes the adaptive immune response as an additional variable in the model was also examined and the results are mentioned in the Discussion section . In the early stage of influenza virus infection , NK cells can be activated by IFN to induce cytolysis of infected epithelial cells and play an important role in the innate immune response [6] , [7] , [8] , [9] , [10] . Here , we assume the number of activated NK cells is proportional to the level of IFN and use the mass action term to represent the killing by NK cells . Note that killing by NK cells is an important , but not the only factor leading to the loss of infected cells . Cytokines or proteins released by other cells such as macrophages [33] during the innate immune response can also promote increased lung epithelial apoptosis following influenza virus infection [34] , [35] . Infected cells are assumed to produce virus at rate p and free virus is cleared at rate c per virion . As in the previous models by Baccam et al . [20] and Saenz et al . [19] , loss of virions due to infection has been neglected . Since an infected cell may produce as many as 20 , 000 virions [36] , the loss of one virion to produce an infected cell can be neglected . IFN is secreted by infected cells at rate q and decays at rate d . A schematic diagram of Eq . ( 1 ) is shown in Figure 1 . Variables and parameters are summarized in Table 1 . We fixed some parameters and estimated the rest by fitting the model to both the viral load and IFN data . The lifespan of infected cells prior to the emergence of the adaptive immune response , 1/δI , was fixed to 0 . 5 days [31] , [37] , which is the value used in previous modeling studies [19] , [21] . Because no CD8+ T cell data were obtained in this experiment , we chose the time at which the adaptive immune response emerges ( μ ) according to the second viral decline . For example , we chose μ = 7 days for pony 1 and μ = 6 days for pony 2 . A similar method has been used previously in analyzing acute HCV infection kinetics in chimpanzees [38] . We also included a delayed adaptive immune response explicitly in the model and obtained similar results ( see Discussion ) . The initial population of epithelial cells in the equine respiratory tract was fixed at T0 = 3 . 5×1011 cells [39] . We assume all such cells are target cells , as used in Saenz et al . [19] , although H3N8 viruses prefer to infect α 2 , 3 sialic acid glycan-expressing cells [40] and thus the number of target cells could be less than assumed here . We include sensitivity test to a number of parameters including the initial number of target cells below . We set the initial population of infected cells and refractory cells to 0 , and the initial IFN fold change to 1 , i . e . , no change , as given in the data set . The remaining parameters were estimated from data fitting . Note that some parameters , such as the infection rate constant β and the viral production rate p , do not have physiological values because they are in the unit of ml of nasal secretions . We fit the model to both the viral load and IFN data of each pony using the commercial software package Berkeley Madonna ( Version 8 . 3 . 18 ) . The obtained parameter values were based on the best nonlinear least squares fit of the model equations to the data set , i . e . , the program minimized the root mean square ( RMS ) between data points and the corresponding model predictions , given by ( 2 ) where the number of viral load and IFN fold change measurements for an individual pony are denoted by nV and nF , respectively . Viral load data is given by and the analogous value given by our model is Vi . Similarly , the measured IFN fold change is and the corresponding model prediction is Fi . The first data point below the detection limit ( 100 copies/ml of NS ) was assumed to be 1 copy/ml of NS . Other values , such as half of the detection limit , can also be used [41] , which will affect the estimate of the parameter σ in this study . There are also other approaches to incorporating left-censored measurements [42] . We did not include the viral load data under the detection limit after the first undetectable data point . Equal weights for both viral titer and IFN data were employed because they are approximately in the same range . Using different weights or normalized data ( each value is divided by the maximum ) generates a similar fit , although the estimates of parameter values can be different . The target cell limited model was used in [20] and described by the following equations: dT/dt = −βVT , dI/dt = βVT-δI , and dV/dt = pI-cV . Assuming tpeak is the time at which the viral load achieves its peak , we have pI = cV at t = tpeak . Thus , I ( tpeak ) = cV ( tpeak ) /p . Because target cells are nearly depleted around the peak of infection in this model [20] , we assumed T≈0 for a short time period after tpeak , and solved for I ( t ) . This assumption was also used in [23] to obtain an approximation for the decay after the peak using the model with an eclipse phase . The solution is . Substituting this into the V ( t ) equation and solving for V ( t ) , we have . Thus , the predicted viral load reduction 1 day after the peak is . As c is typically much larger than δ ( Table 2 ) , this ratio is mainly determined by the value of δ . For δ in the range of ( 0 , 4 . 5 ) day−1 , which covers most of the estimates in the literature [20] , the ratio is always greater than 0 . 01 for any positive value of c . This implies that for any value of δ<4 . 5 day−1 , the target cell limited model generates <2 log decline within 1 day after the peak . The actual viral load reduction predicted by the model should be less than this approximation because we assumed T≈0 over the interval [tpeak , tpeak+1] . Numerical results show that to obtain a 3 log decline within 1 day after the peak , c should be >12 day−1 and δ needs to be >8 day−1 . To attain a 4 log decline , c should be >18 day−1 and δ needs to be >10 day−1 . To statistically compare the best fits using model 1 ( Eq . ( 1 ) ) and model 2 ( setting κ to 0 in model 1 , i . e . , no killing of infected cells by NK cells ) , we performed an F-test . An F-test is used to compare two nested models used to fit the same data set to determine whether the model with more parameters statistically improves the fit . The improvement is considered to be statistically significant if the p-value is less than 0 . 05 . We begin with the calculation of the F-value as follows:where RSS is the sum of squared residuals between model predictions and data . The RMS value generated from Berkeley Madonna is the root of the mean squared residuals . Hence , RSS = n• ( RMS ) 2 , where n is the number of data points . The subscripts 1 and 2 represent model 1 and model 2 , respectively . The degree of freedom associated with RSS is df = n-m , where m is the number of fitted parameters . Note that μ , the time at which the adaptive immune response emerges , was counted as a fitted parameter although we fixed it according to the second viral decline . To compute the p-value , we calculated the F distribution evaluated at the F-value with ( df2-df1 , df1 ) degrees of freedom . Comparison between models was performed individually for all the ponies .
We fit the predicted values of V ( t ) and F ( t ) in Eq . ( 1 ) to the viral load and IFN ( fold change ) kinetic data , respectively , of each pony . The best fits , shown in Figures 2 ( red solid ) and 3 ( blue solid ) , indicate that Eq . ( 1 ) agrees with both the viral load and IFN data well . Parameter values corresponding to the best fits are given in Table 2 . Note that the estimates of some parameters , such as the infection rate β and the viral production rate p , have large variations . This is expected because there is a large variation ( up to 4 logs ) in the peak viral load of the 6 ponies . We also fit the model to the average data of the 6 ponies ( Figures 2 and 3 ) . The average data show similar kinetic changes of viral titer and IFN , and the best-fit model agrees well with the data . For comparison , we also plotted the best fits ( dashed lines in Figures 2 and 3 ) of the Saenz et al . model [19] to the same viral load and IFN data . Our model improves the viral load data fits in several aspects . First , our fits capture the viral peak in all 6 ponies . Second , the fits achieve the rapid and substantial viral decline within 1 day after the peak in all ponies . Third , the fits generate a period of viral plateau and/or a second peak . Lastly , our fits generate the rapid second viral decline to below the detection limit in all 6 ponies . Detailed explanations and possible biological mechanisms for these viral load changes are given below . The viral loads in all 6 ponies experienced a 2 to 4 log decline within 1 day after the peak [16] , [19] . Similar viral declines were also observed in 6 volunteers experimentally infected with influenza A virus [20] . What causes such a rapid and substantial viral decline within a short period of time ? The data fits using both the target cell limited model in [20] and the modified model in [19] did not capture this feature . In fact , using the target cell limited model we can derive an approximation of the viral load reduction 1 day after the peak ( see Materials and Methods ) . For most of the estimates of the infected cell death rate in the literature , the target cell limited model cannot generate a >2 log decline within 1 day after the viral peak . This suggests that other factors not included in the target cell limited model may be responsible for this dramatic viral decline . We tested different models based on several possible biological mechanisms ( see below ) and found that the model shown in Eq . ( 1 ) can reproduce the viral load change observed in the 6 ponies . The rapid viral decline after the peak is mainly due to the combination of two factors: the decline of target cells because of their conversion to the refractory class ( in Eq . 1 ) by IFN's antiviral effect , and the killing of infected epithelial cells ( in Eq . 1 ) , possibly mediated by IFN activated NK cells during the innate immune response . We plotted the changes of uninfected target cells ( solid blue ) , infected cells ( solid green ) , refractory cells ( dashed red ) , and total cells ( dotted black ) in Figure 4 . The number or percentage of infected epithelial cells is low compared to the prediction of the target cell limited model [20] . In contrast with the predictions of the Saenz et al . model [19] , the level of uninfected target cells remains high ( >1010 cells ) for all the ponies during the entire infection course . The reversion of cells from the refractory to the susceptible class ( ρR ) prevents uninfected target cells from decreasing to a very low level . This suggests that in addition to target cell depletion , cytolysis of infected cells mediated by IFN activated cells such as NK cells during the innate immune response may be responsible for the viral decline during the early stage of influenza virus infection . To further test if a model that only includes the refractory class without NK cell-mediated infected cell killing ( in Eq . 1; referred to as model 2 ) can explain the first rapid viral decline , we fit model 2 to the same experimental data ( dashed lines in Figure 5 for viral load and Supporting Figure S1 for IFN fold change ) . We found model 2 cannot generate the rapid viral load decline after the peak . We also tested a model assuming that IFN only reduces the viral production rate ( i . e . , assuming and replacing p with in Eq . ( 1 ) ; this is referred to as model 3 ) and found this model could not generate the first rapid viral decline either and yielded dynamics very similar to model 2 ( dotted lines in Figure 5 ) . Thus , the cell-mediated lysis of infected cells during the innate immune response plays a critical role in generating the first rapid viral decline in our model . We calculated the error between modeling predictions and experimental data ( RMS ) for different models . The RMS values are given in Table 3 . Model 1 generated the smallest error for each pony . We compared the best fits of using model 1 and model 2 by performing an F-test , which determines which one of the two nested models provides a better data fit from a statistical standpoint ( Materials and Methods ) . The results given in Table 3 show that model 1 provides significantly better fits for ponies 2 and 3 ( with the p-value<0 . 05 ) . For the other ponies , the F-test shows that there is a statistical trend supporting model 1 ( with the p-value from 0 . 1 to 0 . 4 ) . We also compared the best fits using the modified Akaike Information Criterion ( AICc ) ( Supporting Text S1 ) . Model 1 is supported over model 2 for each pony ( Table S4 ) . We did not statistically compare the fits of model 1 with the Saenz et al . fits [19] because the objective functions minimized during data fitting are different . Saenz et al . [19] incorporated the percentage of infected cells in their fitting . We did not include this because the data of the percentage of infected cells were from a different study [43] . The errors listed in Table 3 and the fitted curves ( Figures 2 and 3 ) show that our fits improve those using the Saenz et al . model . The phenomenon of bimodal viral titer peaks in most ponies [16] was also observed in other studies with influenza virus infection [44] , [45] , [46] . The target cell limited model [20] and the Saenz et al . model [19] cannot generate bimodal virus titer peaks . Adding the effect of IFN and a time delay in its production into the target cell limited model was shown to be able to generate bimodal peaks [20] . However , the fits obtained by Baccam et al . [20] using this model did not agree well with the data . Our fits using model 1 generated an obvious bimodal behavior ( Figure 2 ) . The level of IFN peaked around day 2 and then declined rapidly ( Figure 3 ) , concordant with the emergence of viral plateau/second peak ( Figure 2 ) . Thus , the viral plateau and the second viral titer peak can be explained by the loss of the IFN-induced antiviral effect ( in Eq . 1 ) . Increased availability of susceptible cells due to reversion from the refractory state ( ρR in Eq . 1 ) can also contribute to the viral plateau/second peak . From our data fits we estimated that the rate ( ρ ) at which refractory cells ( R ) revert from the refractory to the susceptible state is on average 2 . 6 per day . The reversion rate is also important in preventing uninfected target cells from decreasing to a very low level . Sensitivity tests of the model predictions to a number of parameters , including and ρ , are given below . We examined the sensitivity of the predicted viral load of pony 1 to several parameters , including , ρ , κ , and p ( Figure 6 ) . More sensitivity tests of the predicted viral load and IFN to other parameters and contour plots are presented in Supporting Figures S2 , S3 , S4 , S5 , S6 , S7 , S8 . Sensitivity tests show that the IFN's antiviral efficiency ( ) and the reversion rate ( ρ ) are important in generating the viral plateau and the second peak ( Figure 6A , B ) . A large value of can also yield a rapid first viral decline . However , this will eliminate the viral plateau and the second peak ( Figure 6A ) . Increasing the infected cell killing rate constant alone will decrease the first viral peak and increase the second peak ( Figure 6C ) . A large value of the viral production rate p ( Figure 6D ) or the infection rate β ( Figure S2 ) can achieve the first viral peak . However , they will significantly reduce the time for the viral titer to reach the peak . These sensitivity tests suggest that the cell-mediated lysis of infected cells ( κ ) and the IFN's antiviral effect ( ) during the innate immune response are the major factors responsible for the first rapid viral decline and subsequent viral plateau/second peak . Since the initial number of target cells of H3N8 virus infection could be less than 3 . 5×1011 cells ( T0 ) , the estimate of total epithelial cells in the equine respiratory tract [39] , we reduced it from T0 to 75% or 50% of T0 . The simulation in which the other parameters are assumed to be unchanged shows that a small initial number of target cells can delay the time to reach the first viral peak , reduce the magnitude of the peak viremia , and eliminate the viral plateau ( Figure S2 ) . However , data fitting using 75% and 50% of T0 still generates good fits to the experimental data ( see Figure S2 for the fit to the viral load data of pony 1 ) .
The biological factors responsible for viral control during influenza virus infection remain unclear . Earlier work [20] suggested that the viral decline after the peak could be explained by a limitation in the availability of target cells . However , a recent study by Saenz et al . [19] estimated that <5% of epithelial cells are infected at any one time and that the total epithelial cell loss is <30% by the end of the infection . They modified the target cell limited model by including an IFN-induced antiviral state of uninfected cells [19] . However , their modified model is still essentially a target cell limited model — uninfected target cells move to the refractory class , causing the depletion of susceptible cells and hence the viral titer declines after reaching the peak . Numerical simulations also confirmed this prediction ( Figure 3 in [19] ) . As we analytically showed in Materials and Methods , the target cell limited model cannot generate a rapid and substantial viral decline after the peak unless a very large death rate of infected cells is chosen . However , only increasing the death rate of infected cells will decrease the first peak and eliminate the viral plateau/second peak , which is observed in all the 6 ponies . In this paper , we developed a new model ( Eq . ( 1 ) ) and showed that cytolysis of infected cells mediated by cytokines and cells such as NK cells during the innate immune response , can explain the rapid viral decline after peak . During an early stage of infection , NK cell activity contributes to a rapid termination of many virus infections , including influenza , before the onset of the adaptive immune response [11] , [47] , [48] , [49] , [50] . Several studies in mice have illustrated that depletion of NK cells resulted in increased morbidity and mortality from influenza infection [51] , [52] , [53] . In humans , severe/lethal 2009 H1N1 influenza virus infection in 3 cases was associated with reduction of NK cells rather than effector CD8+ T cells [54] , and influenza vaccination led to increased levels of NK cells with activation markers CD56 and CD69 [55] . NK cells are not only responsible for producing antiviral cytokines , but they are also directly involved in destroying virus-infected cells via the recognition by the natural cytotoxicity receptors ( NCR ) NKp46 ( NCR1 in mice [6] ) and NKp44 [7] , [8] , [9] , [10] . Gazit et al . [6] showed that influenza virus infection was lethal in mice when the NK receptor NCR1 was knocked out . In our model , we assumed that the level of activated NK cells is proportional to that of IFN , whose levels were measured in the study [16] . There is evidence supporting that NK cells have similar dynamics to IFN and virus during influenza virus infection . For example , an experimental study on murine influenza virus infection [56] showed that the effector cells with the properties of NK cells had very similar dynamics to the IFN level changes , i . e . , peaked at 1–2 days post-infection and decreased to low levels by day 6 . In mice that were inoculated intranasally with the mouse-adapted strain of human influenza A/PR/8/34 ( H1N1 ) virus , the timing of viral peak and subsequent decline was consistent with that of NK cell-mediated cytolysis [57] . Another study [58] also showed that the peak of NK cells occurred within the first several days after influenza virus infection in mice , consistent with the timing of IFN production . In addition to the killing by IFN activated NK cells , high expression of cytokines during the innate immune response may also lead to infected cell death [34] . For example , influenza A virus-stimulated apoptosis was shown to be enhanced by IFN α/β and by increased expression of the antiviral protein PKR [35] . Macrophage-derived TRAIL ( tumor necrosis factor-related apoptosis-inducing ligand ) also plays an important role in promoting epithelial cell apoptosis [33] . We used IFN as a proxy of the innate immune response to model the cell-mediated lysis of infected epithelial cells and the antiviral effect . This may not be accurate because a number of other cytokines are involved in the innate immune response . Dendritic cells ( DCs ) and macrophages produce large amounts of antiviral and immunostimulatory cytokines in response to influenza virus infection [2] , [4] , [59] , [60] , [61] . We assumed that IFN is secreted by epithelial cells once they are infected . Other cells , such as monocytes , macrophages , and plasmacytoid DCs , can also contribute to IFN production [4] , [37] , [62] . Further , there may exist a time delay in IFN production , as observed in pony 1 ( Figure 3 ) in which viral titer/infected cells peaked at day 2 post-infection while IFN peaked at day 3 post-infection . A similar time lag was observed in mice with influenza virus infection [63] . Moltedo et al . [63] showed that the initiation of lung inflammation ( generation of IFNs , cytokines , chemokines , etc ) did not begin until almost 2 days after infection , when virus replication reached its peak . This delay may be mediated by the influenza-encoded NS1 protein [63] , which can act to block IFN production in influenza infected cells [48] , [64] , [65] . The burst of IFN production after day 2 might be explained by activation of plasmacytoid DCs or other uninfected cells in the lung , which are activated to a degree that correlates with viral titer or number of infected cells . Future comprehensive models may wish to take macrophages , DCs and other cytokines into account . However , more complicated models should be accompanied with appropriate data for model verification . After the rapid post-peak decline of viral titer , we observed a plateau phase and/or the second viral peak . Although a number of models have been developed to study within-host influenza virus dynamics , very few models can generate the second peak . As the innate immune response weakens ( Figure 3 shows that a rapid IFN decay was observed in all ponies even when the viral load was still high ) , the killing of infected cells ( ) lapses in our model . Thus , the level of infected cells can remain unchanged for a while or even increase . This can explain the viral plateau and the second viral increase . Another factor leading to the second peak is the augmented availability of target cells . The rapid IFN decay significantly reduces the conversion of susceptible cells to the refractory class . Because cells are most likely unable to maintain the antiviral state for a long time without continued IFN signaling , those cells that are already in the refractory class will revert back to the susceptible state and become the target of virus infection again . This will enhance the viral production . Some other factors may also contribute to the second peak . For example , when virus spreads to a previously uninvolved site in the lung or respiratory tract as discussed in [20] , viral infection and production will increase and may lead to a second viral load increase . After reaching the second peak around day 6 post-infection , the viral titer underwent a rapid second viral decline to below the detection limit . We showed that this second viral decline can be generated by the emergence of an adaptive immune response ( Figure 2 ) , which usually arises 4 to 7 days post-infection [11] . Without introducing an adaptive immune response in the model , the virus will not be cleared in ponies with a plateau/second peak . Because CD8+ T cell were not measured for these ponies , we assumed an increasing death rate of infected epithelial cells , δA , after the second peak . We have also examined a model with an explicit adaptive immune response by adding another variable X , representing cytotoxic T lymphocytes ( CTL ) , with dX/dt = rX , where r is the net expansion rate . We assumed the CTL-mediated killing of infected cells is −kXI in addition to δII in the model . In order for the adaptive immune response to remain at a very low level during the first several days , r should be very small . However , such a low-level adaptive immune response cannot generate the rapid second viral decline . This problem can be resolved by using a larger r and a time delay for the emergence of the adaptive immune response . However , this method is almost the same as what we did in the main text: increasing the death rate of infected cells several days after infection . In addition to CD8+ T cells , antibodies neutralizing free virions may also be involved in viral clearance . Increasing either the infected cell death rate δ , as shown in our study , or the viral clearance rate c can generate the same second viral decline to below the detection limit . Thus , from the comparison between model predictions and the data , we cannot determine if the viral clearance is mainly caused by CD8+ T cells or neutralizing antibodies . However , in the experiment [16] from which we studied the data , no anti-influenza antibodies were detected by the SRH assay 7 days post-challenge in any of the ponies . Low levels of antibodies were detected by ELISA on day 7 for 3 of the 6 ponies . Although such antibodies may exist at low levels before day 7 , they may not be the major factor responsible for viral clearance because the infection was already resolved by day 7 in ponies 5 and 6 . Likewise , we cannot estimate the duration of the eclipse phase in which infected cells have not started to produce virions because the model with and without an eclipse phase both fit the experimental data well ( Supporting Text S1 , Table S1 , Table S2 , Figures S9 , and S10 ) . Although target cells are not depleted , we predict a decline of target cells as well as the total number of epithelial cells during infection ( Figure 4 ) . The reason for the decline is that we did not include generation/proliferation of epithelial cells . This is not important for the short time period of infection we studied . Consistent with the other studies [19] , [20] , [21] , including the regeneration of target epithelial cells in our model does not improve the fits of the model to the data set . This is also supported by the observation in humans that regenerating respiratory epithelium cells appeared only in 3 out of 14 subjects after 5–14 days post-infection [66] , whereas virus infection is usually resolved within 7–10 days [67] . Once the virus is cleared , generation/proliferation will increase epithelial cells to the pre-infection level . In summary , by fitting mathematical models to the viral load and IFN data we illustrate that both the innate and adaptive immune responses are needed to explain the viral load change during influenza virus infection . The first post-peak viral decline ( about 2 to 4 logs within 1 day ) can be explained by the lysis of infected epithelial cells , mediated by cytokines and cells such as NK cells , during the innate immune response . The subsequent viral plateau/second peak is generated in our model by the loss of the IFN-induced antiviral effect and the increased availability of target cells as cells lose their antiviral state . An adaptive immune response is needed in our model to explain the eventual viral clearance . A detailed and quantitative study of the within-host dynamics of virus , cells , and cytokines may provide more information for future research in influenza pathogenesis , treatment , and vaccination .
|
Influenza , commonly referred to as the flu , is a contagious respiratory illness caused by influenza virus infections . Although most infected subjects with intact immune systems are able to clear the virus without developing serious flu complications , the mechanisms underlying viral control are not fully understood . In this paper , we address this question by developing mathematical models that include both innate and adaptive immune responses , and fitting them to experimental data from horses infected with equine influenza virus . We find that the innate immune response , such as natural killer cell-mediated infected cell killing and interferon's antiviral effect , can explain the first rapid viral decline and subsequent second peak viremia , and that the adaptive immune response is needed to eventually clear the virus . This study improves our understanding of influenza virus dynamics and may provide more information for future research in influenza pathogenesis , treatment , and vaccination .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"nonlinear",
"dynamics",
"mathematics",
"theoretical",
"biology",
"viral",
"transmission",
"and",
"infection",
"population",
"modeling",
"virology",
"immunology",
"biology",
"computational",
"biology",
"microbiology",
"viral",
"load",
"immune",
"response"
] |
2012
|
Modeling Within-Host Dynamics of Influenza Virus Infection Including Immune Responses
|
Polycomb group ( PcG ) proteins are transcriptional repressors of genes involved in development and differentiation , and also maintain repression of key genes involved in the cell cycle , indirectly regulating cell proliferation . The human SCML2 gene , a mammalian homologue of the Drosophila PcG protein SCM , encodes two protein isoforms: SCML2A that is bound to chromatin and SCML2B that is predominantly nucleoplasmic . Here , we purified SCML2B and found that it forms a stable complex with CDK/CYCLIN/p21 and p27 , enhancing the inhibitory effect of p21/p27 . SCML2B participates in the G1/S checkpoint by stabilizing p21 and favoring its interaction with CDK2/CYCE , resulting in decreased kinase activity and inhibited progression through G1 . In turn , CDK/CYCLIN complexes phosphorylate SCML2 , and the interaction of SCML2B with CDK2 is regulated through the cell cycle . These findings highlight a direct crosstalk between the Polycomb system of cellular memory and the cell-cycle machinery in mammals .
Polycomb group ( PcG ) proteins are key developmental regulators that maintain epigenetic silencing of genes [1] and determine the expression patterns of homeobox genes during embryonic development . In Drosophila five different PcG complexes have been described: Polycomb Repressive Complex 1 ( PRC1 ) and 2 ( PRC2 ) [1] , Pho Repressive Complex ( PhoRC ) [2] , Polycomb repressive deubiquitinase ( PR-DUB ) [3] , and dRING associated factors ( dRAF ) [4] . PRC2 methylates lysine 27 of histone H3 ( H3K27 ) [5] , [6] , whereas PRC1 compacts chromatin [7] , and catalyzes the deposition of ubiquitination at H2AK119 [8] , contributing to the establishment of a chromatin environment that is repressive for transcription . PRC1- and PRC2-mediated repression in Drosophila is partially dependent on the presence of PhoRC [9] . Research on PcG function has mostly focused on components of the PRCs and their role in transcriptional repression . However , mutations in several other PcG genes display strong homeotic phenotypes in Drosophila , and the products of these genes are likely to have important roles in gene regulation and epigenetic memory . One of these less studied proteins is the product of the Sex comb on midleg ( scm ) gene . SCM is required for the recruitment and repressive function of PRC1 and PRC2 [9] , and interacts with SFMBT , a component of PhoRC [10] . SCM contains two malignant brain tumor ( MBT ) repeats [11] , a domain of unknown function ( DUF3588 ) , an SPM/SAM domain , and two zinc fingers [12] . It associates in substoichiometric amounts with the PRC1 complex [13] , likely through the interaction of its SPM domain with that of polyhomeotic ( PH ) [14] . However , the absence of PRC1 does not affect SCM localization to target genes , suggesting that SCM may function upstream ( or independently ) of PRC1–2 [9] . Like other PcG proteins , SCM exerts a repressive effect on target genes , which requires both its MBT and SPM domains [15] , [16] . MBT domains bind preferentially to mono- and di-methylated lysine residues in histone tails , which might facilitate their recruitment to chromatin [17] , [18] . Despite its biochemical and genetic connections with PcG complexes , very little is known about the cellular function ( s ) of SCM , especially in humans . There are four homologues of SCM in mammals: SCMH1 and SCML2 comprise the MBT repeats , the DUF3588 domain , and the SPM domain , whereas SCML1 only presents the C-terminal SPM domain and SCML4 comprises the DUF3588 and SPM domains [19]–[22] . Similar to Drosophila SCM , SCMH1 is a substoichiometric component of PRC1 [23] , interacts with homologues of PH [22] , and its hypomorphic mutation in mice results in homeotic transformations , defective spermatogenesis , and premature senescence of embryonic fibroblasts [24] . Other studies have suggested a role for SCMH1 and PRC1 in geminin ubiquitination , and showed that SCMH1 itself is ubiquitinated [25] . The SCML2 gene is deleted in a subset of medulloblastomas [26] , suggesting a role in tumor suppression . In addition to the regulation of developmental genes , PcG proteins impinge on other cellular functions , such as the cell cycle or the DNA damage response [27] . Both PRC1 and PRC2 repress the Ink4a/Arf locus [28] , restricting the expression of p16INK4a . This is a member of the INK4 family of proteins , which blocks CDK4 and CDK6 by inhibiting the interaction with their cyclin partner . Another family of inhibitors , the Kip family , establishes a ternary complex with the CDK/Cyclin , locking it in an inactive conformation . The regulation of these inhibitors occurs at both the transcriptional and protein level . Several mechanisms are responsible for the degradation of p21 or p27 at different phases of the cell cycle [29] , modulating their stability and their inhibitory actions . Interestingly , PRC1 has been recently shown to directly regulate the stability of geminin , Mdm2 , and p53 [25] , [30] , [31] . The regulation of these proteins can indirectly impact on cell-cycle progression and on the levels of CDK inhibitors , suggesting that the functions of PcG are not restricted to transcriptional regulation . This idea is further supported by the recent report of the direct regulation of CYCB by PRC1 components in Drosophila [32] . However , an analogous role in vertebrates remains unexplored . Here we uncover a new function for SCML2B , one of the isoforms of SCML2 , in the regulation of the cell cycle . Our results show that SCML2B contributes to the formation of repressive CDK2/CYCE/p21 complexes and stabilization of p21 in early G1 , leading to reduced kinase activity and controlling the progression through the G1/S border . SCML2 is itself differentially phosphorylated by CDK2 and CDK1 during the cell cycle , suggesting further levels of crosstalk regulation . These findings reveal a role of a PcG protein in modulating the cell-cycle machinery in mammals .
We purified SCML2 from nuclear extract of HeLa-S3 cells using conventional chromatography ( Figure S1D ) . SCML2B is the most abundant isoform present in this fraction , and during the purification , SCML2A and SCML2B separated across the MonoS column , indicating that they form distinct complexes . The putative SCML2B-containing complex eluted with an apparent molecular weight of 150–200 kDa during size exclusion chromatography . Polypeptides co-eluting with SCML2B in the final chromatographic step were identified by mass spectrometry of the fraction showing the peak of SCML2 signal ( Figure 1D and Figure S1E , arrow ) . SCML2B was the most abundant protein identified , and the absence of peptides mapping to the SPM domain ( which is exclusive to SCML2A ) confirmed that SCML2B is part of at least one complex that does not contain SCML2A . In addition , we detected two cyclins ( CYCE2 and CYCB2 ) and two CDKs ( CDK1 and CDK2 ) , along with the CDK/CYC inhibitor p27 ( Figure 1E ) . We excised different regions containing the bands that show a similar elution profile to SCML2B , and the results from mass spectrometry indicated that CYCE2 , CYCB2 , CDK1 , CDK2 , and p27 could indeed be forming a complex with SCML2B ( Figure S1E and Tables S1 , S2 , S3 , S4 ) . In order to confirm whether these proteins form a stable complex with SCML2B , we subjected the material from this step of purification ( hydroxyapatite chromatography , Figure S1D ) to size exclusion chromatography and found that CDK2 , CYCE2 , CYCB2 , and p27 co-eluted with SCML2B ( Figure 1E ) . The apparent molecular weight of this complex was 170 kDa , in agreement with the sum of the sizes of CDK2 , CYCE2 , p27 , and SCML2B . In contrast , although USP3 and DDX17 were detected in the purified material by mass spectrometry ( Figure 1D ) , they either eluted as a monomer from the size exclusion column ( Figure 1E , USP3 ) or were not detectable in the co-elution ( DDX17 ) , suggesting that they do not form a stable complex with SCML2B . We confirmed that SCML2 and CDK2 interact by reciprocal immunoprecipitation in nuclear extract from HeLa cells ( Figure 1F ) . The antibody for SCML2 is targeted to a region between the DUF and the SPM domains . As this region mediates the interaction with CDK2/CYCE complexes ( see below ) , the pull-down of SCML2 may partially disrupt the interaction with CDK2 , explaining why this immunoprecipitation is not very efficient . The interaction can also be detected in nuclear extracts from HCT116 cells ( Figure S1F ) and other cell types ( unpublished data ) . The potential interaction with other proteins present in this fraction was tested by immunoprecipitation from nuclear extracts , and we failed to detect an interaction with Septin-9 , hnRNPD0 , or NONO ( unpublished data ) . Thus , our biochemical purification uncovered the existence of a complex between the PcG protein SCML2B and a core component of the cell-cycle machinery , the CDK/CYC/p27 complex . We next analyzed the interaction of SCML2B with CDK/CYC complexes and p27 in vitro employing recombinant proteins purified from either bacteria ( His-SCML2B , and p27 ) or insect cells ( HA-CDK2/His-CYCE , HA-CDK2/His-CYCA , and CDK1/His-CYCB ) ( Figure S2A–C ) . Recombinant SCML2B interacted with the CDK2/CYCE complex ( Figure S2D , left ) and with p27 ( Figure S2D , middle ) separately , and the interaction with a preformed CDK2/CYCE/p27 complex was stronger ( Figure S2D , right ) . SCML2B also interacted with CDK2/CYCA and CDK1/CYCB ( Figure S2E and F ) as well as with p21 , either alone or in a complex with CDK2/CYCE ( Figure S3B ) , suggesting a role for SCML2B in the function of these cell-cycle regulators . We mapped the domains in SCML2B responsible for the interaction employing the different fragments of SCML2B depicted in Figure S3A . GST-p21 pull-down indicated that two regions in SCML2B mediate interactions with p21 or the CDK2/CYCE/p21 complex: the MBT repeats at the N-terminus of the protein , and a region between the DUF3588 and the SPM domains , predicted to be unstructured ( Ran ) ( Figure S3C ) . The MBT-DUF fragment was more efficiently pulled down by GST-p21 than the MBT repeats alone , indicating that the DUF domain may be important to structurally favor the interaction ( Figure S3C ) . The results thus far suggested that SCML2B might bind to p21/p27 and CDK/CYC complexes in a cooperative manner . Indeed , substoichiometric and stoichiometric amounts of p21 stimulated the interaction between SCML2B and CDK2/CYCE in a dose-dependent manner ( Figure S3D , lanes 2 and 3 ) , but excess p21 had the opposite effect ( Figure S3D , lane 4 ) . Several sites of interaction have been described between CDK2/CYCE and p21 , and it has been postulated that more than one p21 or p27 molecule can bind to the CDK2/CYCE complex to achieve full repression [33] , although one molecule of p21 is sufficient to repress CDK/CYCE complexes [34] , in line with the crystal structure of the CDK2/CYCA complex in the presence of the inhibitory domain of p27 [35] . We cannot rule out that additional binding surfaces are present in regions of p27 outside of the inhibitory domain , but our results indicate that excess p21 blocks the binding sites of CDK2/CYCE within SCML2B . The incubation of CDK2/CYCE with increasing amounts of SCML2B also resulted in the stimulation of the interaction with GST-p21 ( Figure S3E ) . A similar effect was detected when the complexes were pulled down by GST-p21 ( Figure S3F , compare lanes 1–2 and 5–6 ) . To further confirm that p21/p27 and SCML2B bind cooperatively to CDK/CYC complexes , we reconstituted the complex stepwise with recombinant proteins and subjected it to size exclusion chromatography . Only a small part of CDK2 co-eluted with SCML2B in the absence of p27 ( Figure S4 , left ) , indicating that the interaction with CDK2/CYCE alone is weak . The addition of p27 resulted in a change in the migration of both SCML2B and CDK2 that now co-eluted with p27 ( Figure S4 , right ) , further supporting that the binding of SCML2B to CDK complexes is stimulated by p21/p27 . In summary , our in vitro experiments show that SCML2B directly interacts with CDK2/CYCE complexes and that the presence of p21/p27 is required to stabilize the interaction . Next , we sought to determine the functional consequences of SCML2B association with CDK/CYC/p21-p27 . We monitored CDK/CYC kinase activity towards histone H1e in vitro , following its phosphorylation at residue T146 with a phospho-specific antibody ( Figure S5A ) . The analysis of the phosphorylation of a single residue allows the determination of the kinetic parameters of a single reaction , avoiding measuring different events at the same time . In this way we avoid confounding effects due to a mixture of reactions with different parameters being measured in the same experiment . All the recombinant CDK/CYC complexes tested were active towards H1eT146 , and in each case , H1e phosphorylation was inhibited by the addition of increasing amounts of p27 , as expected ( Figure S5B ) . We then compared the kinase activity of CDK2/CYCE when associated with either SCML2B or p27 or both , using reconstituted complexes that were fractionated by size exclusion chromatography ( Figure S4 ) . The presence of SCML2B had no effect on the activity of CDK2/CYCE in the absence of p27 , but resulted in a significant inhibition of p27-containing CDK2/CYCE complexes ( Figure S5C–E ) , even considering that p27 alone reduced the activity of CDK2/CYCE ( Figure S5C , compare bar 1 with 4 ) . A detailed analysis of H1e phosphorylation using a mixture of CDK2/CYCE and p27 showed that the activity of this complex follows Michaelis-Menten kinetics and that the presence of SCML2B almost abolished the residual activity of the CDK2/CYCE/p27 complex ( Figure 2A ) . This effect was not limited to p27 as SCML2B also enhanced the inhibitory effect of p21 on CDK2/CYCE kinase activity ( Figure 2B ) . These results , together with the in vitro interaction experiments , suggest that SCML2B has an inhibitory effect on CDK/CYC through the stabilization of their interaction with p27 and p21 . The addition of SCML2B reduced the Vmax of the reaction without significantly changing the affinity for H1e , indicating that even if SCML2B is itself a substrate of CDK2 ( see below ) , it is not competing out H1e under the reaction conditions , where H1e is present at ≥2-fold molar excess versus SCML2B . As SCML2B interacts in vivo ( Figure 1 ) , and in vitro ( Figures S2 and S3 ) , with CDK/CYC complexes that function during different phases of the cell cycle , we next tested whether SCML2B expression is itself subjected to cell-cycle regulation . After release from a double thymidine block ( Figure S6A ) , SCML2A and SCML2B protein levels showed a similar profile , with small fluctuations during the cell cycle in HeLa and U2OS cells ( Figure S6B and unpublished data ) . Both SCML2 isoforms were expressed at higher levels in S compared to early G1 and G2/M . SCML2 exhibited slightly altered gel mobility in G2/M ( see below ) , suggesting cell-cycle–dependent posttranslational modification ( s ) of the protein . We compared the mobility of SCML2 in extracts obtained from asynchronous HeLa cells ( As ) , or HeLa cells arrested in G0 by serum starvation ( SS ) , G1/S by double thymidine block ( Thy ) , or mitosis with nocodazole ( NCZ ) , and found that the mobility of either SCML2A or SCML2B was slowest in the case of extracts prepared from cells arrested in mitosis ( Figure S6C ) . The mobility of each of the SCML2 isoforms increased upon treating the extracts with Antarctic phosphatase ( Figure 3A ) , suggesting that SCML2A and SCML2B are phosphorylated during mitosis . Consistent with this , two proteomic studies reported that SCML2 is phosphorylated at several Ser and Thr residues ( shown schematically in Figure 3B ) in cells arrested in mitosis after nocodazole treatment [36] , [37] . These results indicate that SCML2 might itself be regulated through its phosphorylation mediated by its interaction partners , CDK/CYC . We confirmed that both CDK2/CYCE and CDK1/CYCB phosphorylate SCML2B in vitro , while Aurora kinase A was ineffectual ( Figure 3C ) . Mass spectrometry of the products of in vitro phosphorylation reactions revealed that CDK2/CYCE and CDK1/CYCB targeted similar sites on SCML2B ( Table S5 and Figure S7A ) , some of which are preferentially phosphorylated during mitosis in vivo ( Table S5 ) [36] , [37] . Analysis of their sequence using the Phosida software [38] revealed that S267 and T305 are embedded within a consensus CDK2 and CDK1 target , and S511 and S590 are in the context of a CDK1 motif . The target sites of CDK phosphorylation were concentrated between the MBT repeats and the DUF3588 domain and in two Ser/Thr-rich stretches in the Ran region , which is important for the interaction with CDK/CYC/p21-p27 complexes ( Figure 3B ) . Next , we analyzed the phosphorylation status of SCML2B during different phases of the cell cycle in vivo . To this end , we induced expression of transgenic Flag-One-STrEP-tagged SCML2B ( FS-SCML2B ) in stably transfected 293T-REx cells , and analyzed its phosphorylation status by mass spectrometry after arresting the cells in mitosis or S phase ( Figure S7B ) . By comparing the signal from in vitro dephosphorylated and untreated peptides , we observed that SCML2B was preferentially phosphorylated in mitosis at several residues ( Figure 3D , red ) , including some of the in vitro CDK/CYC targets ( S499 and S511 ) ( Table S5 ) . In contrast , other residues displayed a similar level of phosphorylation in mitosis and S phase ( Figure 3D , yellow ) , while the phosphorylation in a Ser/Thr stretch in the N-terminal region was higher in S phase ( Figure 3D , blue ) . We confirmed that T305 , S511 , and S590 are phosphorylated by CDK1/2 in vivo , as treatment of cells with increasing concentrations of Roscovitine ( an inhibitor for CDK1 and CDK2 ) for 8 h reduced the levels of phosphorylation of these residues ( Figure 3E ) . In contrast , phosphorylation of S267 was not affected by the inhibition of CDK1/2 . The treatment with Roscovitine did not induce major changes in the cell-cycle distribution of the cells ( Figure S7C ) . These data confirmed that SCML2B is highly phosphorylated during mitosis and that this phosphorylation is partly mediated by CDK1/2 . We analyzed the effect of the phosphorylation of SCML2 on its interaction with p21 in vitro . The pull-down of SCML2B by GST-p21 was not changed when SCML2B was previously phosphorylated by CDK2/CYCE ( Figure 3F , left ) . CDK1 and CDK2 target S/T-P motifs that then become substrates for isomerization of the Pro by Pin1 [39] . Several of the residues of SCML2 phosphorylated by CDK in cells and in vitro are adjacent to Pro ( T305 , S511 , S590 ) and reside within flexible regions that mediate the interaction with CDK2/CYCE and p21/p27 ( Figure S3 ) . The addition of Pin1 to the kinase reaction did not change the levels of phosphorylation of SCML2 by CDK2/CYCE ( Figure 3G ) , as has been described for other substrates of Pin1 [40] . In contrast , the presence of Pin1 partially impairs the interaction of phosphorylated SCML2 with p21 ( Figure 3F ) . These results suggest that Pin1 recognizes the phosphorylated residues in SCML2 , inducing a conformational change that reduces the binding to p21 , potentially restricting the actions of SCML2 during the cell cycle . Both p21 and p27 regulate cell-cycle progression into S phase , and p21 is also involved in the transition to mitosis [29] . Phosphorylation of p21 and p27 by CDK2/CYCE complexes is required for their proteasome-mediated degradation that , in turn , allows cells to progress from G1 to S phase . Because our biochemical data showed that SCML2B interacts with CDK/CYC/p21-p27 and enhances the inhibitory effect of p21 or p27 on the kinase activity , and is itself subjected to phosphorylation as a function of the cell cycle , we reasoned that SCML2B could regulate the cell cycle in vivo . We transiently transfected U2OS cells with a control siRNA or two different siRNAs specific for SCML2: siRNA#1 and siRNA#2 ( Figure 4A ) . Immunofluorescence analysis indicated that siRNA#2 was more effective than siRNA#1 ( Figure 1C ) . None of the siRNAs affected the expression of CDK2 , but knockdown of SCML2 elicited a variable decrease in the levels of p27 and a strong reduction in the levels of p21 ( Figure 4A ) . As the knockdown of SCML2 consistently destabilized p21 , we decided to focus on the role of this inhibitor , although we do not rule out a potential contribution of p27 to the functions of SCML2 . Consistent with SCML2 having an impact on cell-cycle progression , knockdown of both SCML2A and SCML2B led to a decreased proportion of cells in G1 phase ( 5%–10% ) , which coincided with an increased proportion of cells in S phase and , to a lesser extent , in G2/M ( Figure 4B ) . Overexpression of CYCE elicits a decrease in G1 phase of around 10%–20% in different cell types [41] , [42] , indicating that the changes observed upon SCML2 knockdown are highly significant . There was no detectable increase in cellular apoptosis upon treatment with any of the siRNAs ( unpublished data ) . To verify if the effects of SCML2 on the cell cycle were mediated by regulation of p21 and CDK/CYC complexes as suggested by our biochemical experiments above , we performed double knockdown experiments . Knockdown of p21 elicited a decrease in the proportion of cells in G1 , and depletion of both p21 and SCML2 had an additive effect ( Figure 4C ) . If the effect of SCML2 in G1/S progression were solely mediated through the function of p21 on CDK2/CYCE , a similar effect would be expected between the single and double knockdowns . However , the siRNA against p21 results in a stronger depletion of p21 than the reduction of SCML2 alone ( Figure 4E ) , and this may be affecting other functions of p21 , such as inhibition of PCNA activity [43] , [44] . In the absence of p21 , other members of the Kip family of inhibitors , such as p27 , can compensate for the inhibition of the CDK/CYC complexes ( see below ) , but not for these additional functions . We noted that knockdown of SCML2 in the absence of p21 induces a greater reduction in the levels of p27 than knockdown of SCML2 alone ( Figure 4E ) , and this could result in an additive acceleration of passage through G1 through the combined regulation of CDK2/CYCE activity and other p21-dependent processes such as PCNA activity . The double knockdown of p21 and SCML2 did not allow us to firmly conclude if SCML2 functions in G1/S progression , and the effect of SCML2 depletion in the cell cycle is reminiscent of the overexpression of CYCE . Thus , we decided to analyze the effect of a simultaneous reduction of SCML2 and CYCE2 , the CYCE homologue detected in our initial purification ( Figure 1D and E ) . Depletion of CYCE2 alone did not have a significant effect on the cell-cycle distribution in U2OS cells , but it partially rescued the effect of SCML2 knockdown ( Figure 4D ) . The double knockdown of CYCE2 and SCML2 did not change the levels of p21 or p27 when compared to the knockdown of SCML2 alone ( Figure 4E ) . The changes in the cell-cycle profile upon SCML2 knockdown ( Figure 4B ) are similar to the effects of CYCE overexpression . Together with the decreased p21 protein levels ( Figure 4A ) and the effect of the double knockdown of SCML2 and CYCE2 ( Figure 4D ) , these data suggest a role for SCML2 in delaying the progression through G1 via regulation of CDK2/CYCE activity . To confirm this point , we knocked down SCML2 in U2OS cells , arrested them in mitosis , and then monitored their progression through G1 after release ( Figure S8A ) . In the case of control siRNA , the cells progressed from mitosis to G1 in ∼4 h and began S phase in 12–14 h ( Figure S8A ) . Although SCML2 knockdown did not affect the exit from mitosis ( Figure S8A ) , by 16 h a larger proportion of cells were in S phase compared to the control ( Figure 4F ) . In fact , entrance into S phase occurred significantly faster ( 2 to 4 h earlier than control cells ) when the levels of SCML2 were reduced , particularly in the case of siRNA #2 ( Figure 4G ) . Similar results were obtained by monitoring the entry into S phase using EdU staining ( Figure S8B ) . Again , the magnitude of the acceleration of G1 passage is similar to what has been previously reported upon overexpression of CYCE1 [45] . We further analyzed the function of SCML2B during the progression from G1 to S phase . Thus , we arrested U2OS cells in mitosis with nocodazole and analyzed if depletion of SCML2 or p21 affected the levels of cyclins and their inhibitors during G1 progression . In control-treated cells , the levels of SCML2 increased during G1 ( Figure 5A ) , confirming that its expression is highest in S phase ( Figure S6B ) . Interestingly , the increase in SCML2 levels paralleled the accumulation of p21 , and was not detected in the absence of p21 ( Figure 5A ) . Correspondingly , the accumulation of p21 during G1 was blocked when SCML2 was knocked down ( Figure 5A–B ) . Additionally , CYCE2 levels increased prematurely , and CYCB2 reduction after mitosis exit was impaired in the absence of SCML2 ( Figure 5A ) . In control-treated cells and SCML2-depleted cells , the levels of p27 remained constant or slightly decreased during G1 progression . In contrast , the levels of p27 increased in G1 when p21 was depleted , suggesting that it may compensate for its loss ( Figure 5A ) . The changes in the levels and accumulation of p27 could further explain the accumulative decrease in the percentage of cells in G1 observed upon double knockdown of SCML2 and p21 ( Figure 4C ) . These data suggest that SCML2 modulates the accumulation of p21 and CDK2/CYCE2 complexes in G1 . Accordingly , overexpression of SCML2B induced a slight increase in p21 in control cells growing asynchronously , and SCML2B also rescued the decrease in p21 upon SCML2 knockdown ( Figure S9A ) . Overexpression of SCML2A ( to higher levels than those attained with SCML2B ) had no effect on the levels of p21 in control cells , and only partially restored the levels of p21 in the absence of endogenous SCML2 ( Figure S9A ) , suggesting that it can potentially contribute to p21 regulation , at least in an overexpression setting . Further , upon exit from mitosis , overexpression of SCML2B was sufficient to restore the accumulation of p21 during G1 in the absence of the PRC1-associated SCML2A isoform ( Figure 5C ) . The effect is not complete , as the kinetics of p21 accumulation are delayed when SCML2B is overexpressed compared to control cells ( Figure 5D ) , suggesting that additional indirect effects may be contributing to the regulation of G1 progression by SCML2 . Although our original purification of the SCML2B complex did not recover p21 , the experiments above suggest a functional link between the two . To verify that p21 associates with SCML2B in vivo , we first fractionated nuclear extracts from HeLa cells and analyzed co-elution by size exclusion chromatography . The resulting profiles demonstrated that both p21 and p27 associated with SCML2B , CYCE2 , and a fraction of CDK2 ( Figure S9B ) . CDKs and cyclins were also detected in other fractions , but p21 and p27 peaked in the same high molecular weight fraction as SCML2B , which , together with our in vitro results ( Figure 2 and Figures S2 and S3 ) , suggested that SCML2 binds to p21/p27 and CDK/CYC complexes co-operatively . In this sense , the pull-down of p21 in nuclear extracts only recovers a very small amount of SCML2 ( Figure S1F ) , indicating that in cells SCML2 interacts mainly with CDK2/CYCE . Even if the interaction of SCML2 with p21 is not direct , we reasoned that p21 could be modulating the binding of CYC/CYC and changing through the cell cycle , similar to the effect observed in vitro . Indeed , CDK2 co-precipitated with FS-SCML2B in asynchronously growing 293 cells ( most of which are in G1 ) . This interaction was less pronounced in cells arrested in S phase with thymidine or in mitosis with nocodazole ( Figure 6A ) , suggesting that the association of SCML2B with CDK2 occurred mainly in G1 . A detailed analysis of the interaction in extracts from U2OS cells upon release from mitosis revealed that the binding of SCML2 to CDK2 is high at the exit of mitosis ( 30 min after release from the mitotic arrest ) ( Figure 6B–C ) . The percentage of SCML2 pulled down by CDK2 decreases during early G1 , and then peaks again coinciding with the onset of p21 accumulation and association of p21 to CDK2 ( Figure 6B–C ) . During the G1/S transition the binding of SCML2 to CDK2 is progressively reduced ( Figure 6B–C ) , in parallel with the onset of p21 accumulation ( Figure 5A ) . Next , we analyzed this interaction in cells depleted for SCML2 or p21 , quantifying the amount of these proteins pulled down by CDK2 , and normalizing by the input and the levels of CDK2 in the elution . The percentage of SCML2 pulled down in cells depleted for p21 was calculated compared to control-treated cells , showing that the depletion of p21 decreases its association to CDK2 ( Figure S10A , upper panel ) . The pull-down of p21 is also reduced when SCML2 is depleted ( Figure S10A , lower panel ) . While the association of p21 increases around 4 h in control-treated cells , this effect is only seen at later time points ( 10–12 h ) in the absence of SCML2 ( Figure S10A , lower panel ) , confirming that the association of p21 with CDK2 is delayed in the absence of SCML2 . As a whole , these results are in line with the in vitro experiments ( Figure S3 ) showing that low amounts of p21 stimulate the interaction of SCML2B with CDK2/CYCE complexes . At 4 h after exit from mitosis the presence of low amounts of p21 and SCML2 may co-operate in the binding to CDK2/CYCE complexes . Because of this initial interaction , p21 accumulates and , later in G1 , inhibits CDK2 , independent of SCML2 . Our results also suggest that there are additional functions for SCML2 in the exit from mitosis . The results presented above indicated that in the absence of SCML2 , the interaction of p21 with the accumulating CDK2/CYCE is not established in a timely manner , resulting in a premature activation of these complexes . To address this possibility , we measured the kinase activity of CYCE2-associated complexes in cells treated with siRNA against SCML2 ( si#2 ) , growing asynchronously or released from mitotic arrest . Using H1e as a substrate , we could not detect a substantial difference in the activity of CYCE2 in asynchronously growing U2OS cells in the absence or presence of SCML2 ( Figure 6D , As ) . In contrast , in cells arrested in mitosis and released for different times , we detected a faster and stronger activation of CYCE2-asssociated kinases ( Figure 6D , 8–10 h ) , in agreement with the data showing faster entry into S phase upon SCML2 knockdown , as measured by PI and EdU staining ( Figure 4F and Figure S8B , respectively ) . Interestingly , in control cells an increase in the association of CYCE2 and p21 was observed around 6–8 h after release , prior to the increase in CDK2/CYCE2 activity and before the maximal levels of p21 are achieved ( Figure 6D ) . Again , the data suggest that the presence of SCML2 potentiates the initial interaction of p21 with CDK2/CYCE2 to establish an effective inhibition of the complex . In contrast , in cells with reduced levels of SCML2 , the association of CYCE2 with p21 did not change during G1 as the accumulation of p21 is impaired . As a consequence , the inhibition of CDK2/CYCE2 complexes is also blocked ( Figure 6D ) . Our results suggest that SCML2B contributes to the stabilization of p21 upon exit from mitosis and to an efficient inhibition of CDK/CYCE . Consistent with this possibility , treatment with the proteasome inhibitor MG132 reversed the reduction in p21 protein levels caused by SCML2 knockdown , while only a limited effect was detected for p27 ( Figure 7A–B ) . SCML2 knockdown had no effect on the levels of unrelated proteins , USP7 and PR-Set7 ( Figure S10B ) . Proteasome inhibition did not affect USP7 protein levels , although it did result in increased expression of PR-Set7 , as previously reported [46] ( Figure S10B ) . The half-life of p21 decreased from ∼22 min in control cells to ∼20 min after knockdown of SCML2 , and this effect was rescued by an siRNA-resistant version of SCML2B ( Figure 7C , top ) . Given that SCML2 specifically interacts with CDK2 during a narrow temporal window in G1 progression , we speculate that a stronger effect could be detected at this time . As expected , the half-life of p21 was greatly increased in synchronized cells 8 h after release from mitosis ( half-life over 60 min ) , independent of the presence or absence of SCML2 ( Figure 7C , bottom ) . However , at 5 h after mitosis knockdown of SCML2 resulted in much less stable p21 ( half-life of ∼28 min versus ∼56 min in siRNA control-treated cells; Figure 7C , middle ) . Importantly , overexpression of SCML2B alone completely rescued this phenotype ( Figure 7C , middle ) , reinforcing the notion that the effect of SCML2 proteins on p21 stability is independent of PRC1 . Taking into account the time frame for CDK2/CYCE2 activation , our results suggest that the doubling of the half-life of p21 is important to establish an inactive complex around 6 h after release , and to avoid premature activation ( Figure 6D ) . Thus , SCML2B is required for stabilization of p21 in early G1 , when p21 levels are limiting , and stimulates the formation of an inactive complex . PcG proteins play an essential role in the modulation of self-renewal and differentiation of embryonic stem ( ES ) cells . p21 has been proposed to mediate the induction of differentiation upon treatment with Nutlin , an inhibitor of MDM2 that increases the levels of p53 [47] . In contrast , other differentiating agents such as retinoic acid do not induce the expression of p21 [47] . We analyzed whether the induction of differentiation affected the interaction of SCML2 with CDK2 and if this effect was dependent on the induction of p21 . Pull-down of CDK2 in control H9 ES cells shows a weak interaction between SCML2 and CDK2 ( Figure 8A and Figure S10C ) . Treatment of H9 cells with 15 µM Nutlin for 3 d increased the amount of SCML2 pulled down by CDK2 ( Figure 8A ) , along with the induction of p21 and differentiation of the cells as assessed by reduced levels of Nanog ( Figure 8B ) . In contrast to p21 , the levels of p27 remained undetectable under these conditions . Treatment of cells with 30 µM retinoic acid for 3 or 5 d reduced the levels of p21 and increased the levels of p27 ( Figure S10D ) , as previously reported [47] , while inducing a strong decrease in the levels of Nanog ( Figure S10D ) , confirming the differentiation of the cells . The treatment with retinoic acid slightly increased the interaction between SCML2 and CDK2 ( Figure S10C ) , although the effect is smaller than the one observed with Nutlin treatment ( Figure 8A ) . These results suggest that SCML2 may have an effect in the regulation of the cell cycle during the differentiation of ES cells , and this effect is likely modulated by the induction of p21 , independent of differentiation itself . In conclusion , our findings show that SCML2B associates with CDK/CYC/p21-p27 complexes and promotes the p21-mediated inhibition of CDK/CYC in vitro . In vivo , SCML2B contributes to the stabilization of p21 in early G1 , and fosters both the accumulation of p21 and the establishment of an inactive CDK2/CYCE complex with p21 and SCML2B , even when limiting amounts of p21 are present . As a consequence , SCML2B inhibits the premature activation of CDK2/CYCE complexes , prolongs the duration of G1 , and contributes to proper cell-cycle progression ( Figure 8C ) . Our data further suggest a role for SCML2 on the regulation of the cell cycle of ES cells during differentiation , in coordination with p21 .
PcG proteins ensure the epigenetic repression of lineage-specific genes that is necessary for the correct development of complex organisms [2] . Consequently , they are indispensable for the maintenance of pluripotency [48] and for the proper onset of differentiation programs [2] . The transition from pluripotent to specialized cells requires that the differentiation programs and the proliferation capacity of these cells be coordinately regulated [49]: as cells become specialized , their proliferation potential is progressively reduced , until they stop dividing when terminal differentiation is reached . In some cases , adult stem cells with a pluripotent state present a slow proliferation rate , allowing the generation of a pool of stem cells that can be expanded when needed . Therefore , it is not surprising that PcG proteins are involved in the transcriptional regulation of gene networks controlling the cell cycle , exerting a pivotal role in the interplay between differentiation and proliferation [50] . For example , in Drosophila , the cyclin A gene is repressed by PcG proteins through a Polycomb Response Element within its promoter , linking the stable repression of cyclin A with the differentiation process [51] . In mammals , both PRC1 and PRC2 bind to and repress the INK4a/Arf locus [52]–[54] , which encodes several proteins involved in cell-cycle regulation . Recently it has been shown that CBX7 , a component of PRC1 , represses the CYCE1 gene [55] . Most of the evidence for cell-cycle regulation by PcG proteins arose from studies of indirect effects through transcriptional repression of specific genes . Only very recently , PSC , a component of PRC1 , has been reported to affect the stability of CYCB in Drosophila [32] . Interestingly , this activity of PSC does not depend on the PRC1 complex . In mammals , PRC1 has been proposed to regulate the stability of MDM2 and/or p53 [30] , [31] , which can indirectly modulate the cell-cycle machinery . To our knowledge , the binding and regulation of CDK/CYC complexes by SCML2B represent the first direct biochemical link between the Polycomb axis and cell-cycle progression in mammals . The presence of two isoforms potentially allows SCML2 to coordinate PRC1 function with cell-cycle regulation: SCML2A associates with PRC1 through its SPM domain and regulates its recruitment to chromatin ( Bonasio et al . , submitted ) , while SCML2B is present in the soluble nuclear fraction where it modulates the activation of CDK2/CYCE complexes . The concerted regulation of the expression of both isoforms may establish a dual activity on transcription and the cell cycle , similar to the dual function of PSC in Drosophila [32] . A role for PRC1 in the DNA damage response is also becoming prominent [27] , supporting the function of PcG as an essential axis to control cell fitness through cell division and differentiation . Consistent with our in vitro and biochemical observations , we show that SCML2B functions in vivo during the G1/S checkpoint , slowing the progression of cells through G1 . SCML2B interacts with CDK2/CYCE complexes at early G1 , when p21 is in limiting amounts . Increased levels of p21 ensure CDK/CYC inhibition and most likely impair the interaction of SCML2B with CDK/CYC complexes by blocking the binding surfaces on SCML2B , which would explain why excess amounts of p21 do not result in increased SCML2B binding ( Figure S3 ) . Additionally , most of the CDK2/CYCE complexes are bound by p21 or p27 in cells [56] , [57] ( Figure 6 and Figure S10 ) , and the small pool of free CDK2/CYCE is critical for progression from G1 to S phase . This is in good agreement with the recently revisited model for cell-cycle progression . In this model , sequential increases of CDK activity determine the progression into S phase or mitosis [58] . At early G1 , CDK2/CYCE complexes should be tightly controlled to avoid CDK activity from rising above the threshold required for progression into S phase , and therefore the binding of SCML2B to a small fraction of total CYCE could be relevant for G1/S progression [56] , [57] . In this context , we propose that SCML2B promotes the interaction of p21 with the small pool of free CDK2/CYCE . Our data show that SCML2 mainly interacts with CDK2/CYCE in vivo , and this binding in early G1 promotes p21 stabilization , preventing the premature activation of the complex and controlling progression through G1 , which is accelerated upon SCML2 depletion ( Figure 6 ) . Reciprocally , the presence of p21 is necessary for the function of SCML2 , indicating that the cooperative binding of CDK2/CYCE , p21 , and SCML2B observed in vitro is also relevant in cells . This is also consistent with the increased interaction of SCML2 with CDK2 in differentiated versus control ES cells . The cell cycle in ES cells presents a very short G1 phase , with no checkpoints controlling G1/S transition . This is due to high and constitutive activation of CDKs , with undetectable levels of Ink4a and Kip CDK inhibitors [59] . Overexpression of p21 can arrest the cell cycle of human ES cells inducing their differentiation [60] , and elevated p21 levels drive the differentiation of cells upon treatment with Nutlin , correlating with the increased interaction between SCML2 and CDK2 . Our results suggest a role for SCML2 in the coordination of differentiation of ES cells and the modulation of the cell cycle , a system where the restoration of the G1 checkpoint is essential to allow cells to differentiate . Additionally , we show that SCML2 is itself a target of the kinase activity of CDK/CYC and that its phosphorylation occurs preferentially during mitosis . Although our results show that phosphorylation by itself is not enough to affect the interaction of SCML2B with p21 , the presence of Pin1 leads to a decreased in vitro interaction between these proteins . In this way , a regulatory feedback loop can be established to restrict the functions of SCML2 to early G1 . Other PcG proteins are also regulated by phosphorylation by the CDKs , as is the case of EZH2 ( a PcG protein that is part of the PRC2 complex ) [61]–[64] . Taking into account that the binding of is maximal upon exit from mitosis , we cannot rule out that SCML2B plays additional roles in other phases of the cell cycle , such as mitosis . A detailed analysis of the phosphorylation of SCML2 in cells will be required to elucidate the interplay between the phosphorylation and the association to CDK2-containing complexes . Proteins that regulate cell-cycle progression are often targets of mutations and epimutations during cancer development and progression . Indeed PcG proteins often act as tumor suppressors . For example , removal of PRC1 components from Drosophila eye imaginal discs leads to increased proliferation and tumor-like phenotypes , possibly via deregulation of the Notch or JAK/STAT signaling pathways [65] , [66] . In mammals , MEL18 and CBX7 have also been proposed to act as a tumor suppressor in prostate cancer [55] , [67] . The effects of SCML2B depletion on the cell cycle reported here suggest that SCML2B may also function as a tumor suppressor . In its absence , cells enter S phase too early , leading to faster proliferation , and potentially increasing genome instability . Although a possible role for the SCML2 gene in cancer has already been suggested [26] , [68] , the lack of functional information on the proteins that it encodes has hampered the elucidation of its mechanism of action . Single base and frame-shift mutations have been reported in different kinds of cancers in several databases ( http://dcc . icgc . org/web/ ) , but these alterations have remained unexplored . Indeed , the absence of scm in Drosophila induces an increase in proliferation similar to that associated with defects in PRC1 components [10] , [66] , and human SCML2 and other MBT-containing genes are focally deleted in medulloblastomas [26] and breast cancer [69] . Deregulation of the expression of SCML2 has also been observed in acute myeloid leukemia and in several T-cell malignancies [68] . In light of these studies and our new observations , we propose that SCML2 contributes to the regulation of the cell cycle . Based on our results we propose that SCML2 exerts this regulatory function through its nucleoplasmic isoform , SCML2B , which stabilizes p21 and reinforces its inhibitory activity on CDK2/CYCE , avoiding a premature activation of the complex and thereby delaying the entry of cells in S phase .
HeLa , U2OS , K562 , and 293T-REx cells were grown in DMEM with 10% FBS , penicillin ( 100 IU/ml ) , streptomycin ( 100 µg/ml ) , and glutamine ( 300 µg/ml ) . Human H9 ( WA09 ) embryonic stem cell lines were obtained from WiCell Research Institute and grown in TeSR medium ( Stem Cell Technologies ) with GELTREX ( Life Technologies ) . For differentiation experiments , H9 cells were individualized by treatment with accutase ( Millipore ) and then treated with 15 µM Nutlin or 30 µM retinoic acid ( Sigma ) in DMEM/F12 medium ( Invitrogen ) supplemented with 20% Knockout Serum Replacement ( Invitrogen ) , 1 mM L-glutamine , 0 . 1 mM nonessential amino acids , 55 µM β-mercaptoethanol , and 2 . 5 ng/ml bFGF ( R&D ) . Cytosolic and nuclear extracts were prepared as previously described [70] . The nuclear pellet was extracted by solubilization in 50 mM Tris , pH 7 . 5 , 8 M Urea , and 1% Chaps . Transfection of the siRNA for human SCML2 ( #1 5′CCAAACGATCTCCTCAGCAAA , #2 5′CAGTATGTATTGCTACGGTTA , #3 5′GTTATATAGCTGTGTACCTGA , #4 5′CAGGAGATATTTATACTACGA ) or for human p21 and CYCE2 ( SMARTPool L-003471-00-0005 and L-003214-00-0005 , from Dharmacon ) was performed using lipofectamine RNAimax ( Invitrogen ) according to the manufacturer's instructions . Cotransfection of the different siRNA together with the empty pINTO vector or the pINTO-FSH-SCML2A , pINTO-FSH-SCML2B was performed using Lipofectamine 2000 ( Invitrogen ) according to the manufacturer's instructions . 293T-REx cells were transfected with the pINTO-FS-SCML2B [71] plasmid using PEI , and clones were selected in the presence of 5 µg/ml blasticidin ( InvivoGen ) and 100 µg/ml Zeocin ( Invitrogen ) . SCML2B expression was induced with 1 µg/ml doxycycline for 24 h . For proteasome inhibition , cells were treated with 5 µM MG132 ( or DMSO in the control cells ) for 4 h . For the determination of the half-life of p21 , cells were incubated with 25 µg/ml cycloheximide ( Sigma ) for 10/20/40/80/120 min . Rabbit antibody against SCML2 was generated using a GST fusion protein of a central region of SCML2 , and affinity purified . The antibodies against CDK2 ( Santa Cruz , sc-163 ) , CYCE2 ( Epitomics , 1775-1 ) , CYCB2 ( Santa Cruz , sc-22776 ) , p21 ( Calbiochem ) , p27 ( BD , 610242 ) , histone H1T146-Phospho ( Abcam , ab3596 ) , USP7 ( Bethyl , A300-033A ) , RNA polymerase II and PRSET7 ( custom made ) , and CYCE and CYCB ( kindly provided by Dr . Michele Pagano ) were used for Western blot analysis and immunoprecipitation . The purification of SCML2B is schematically depicted in Figure S1D . Briefly , nuclear extracts obtained from HeLa S3 cells were loaded onto a p11 column , and the bound material was eluted with increasing salt concentrations . SCML2 elutes mainly at 0 . 3 M KCl . After dialysis , this fraction was then loaded onto a cation exchange DE52 resin . SCML2 remains in the flow-through , which was loaded onto an anion exchange CM-sepharose column . Bound material was eluted with 1 M NaCl , dialyzed , and loaded onto another anion exchange column , SP-sepharose . Step elution with increasing NaCl concentration recovered SCML2 in the 0 . 3 M fraction . This material was dialyzed and subjected to a strong cation exchange chromatography using a MonoQ column . SCML2 was recovered in the flow-through and then loaded onto a Heparin affinity column . The bound material was eluted using a NaCl gradient ( 0 . 05 to 0 . 6 M ) . The fractions containing SCML2 were pooled , dialyzed , and then subjected to a strong anion exchange chromatography using a MonoS column . The bound material was eluted with a NaCl gradient ( 0 . 05 to 0 . 6 M ) . At this step the two isoforms of SCML2 separated in two different pools . The pool containing SCML2B was dialyzed and loaded onto a HiTrap SP-sepharose anion exchange column . A step elution with 1 M NaCl was performed to concentrate the sample , which was then fractionated on a Superdex200 size exclusion column . The fractions containing SCML2B ( 150–200 kDa ) were dialyzed and then fractionated by affinity chromatography using a Heparin-5PW column . After elution with a NaCl gradient ( 0 . 04 to 0 . 6 M ) , SCML2B-containing material was dialyzed and then loaded onto a Hydroxyapatite column . Bound material was eluted with a phosphate gradient ( 0 . 01 to 0 . 5 M ) . The peak fraction for SCML2B was analyzed using SDS-PAGE and silver staining , and mass spectrometry was performed both in solution and from gel-excised bands . Further fractionation was performed using a Superdex200 size exclusion column , and protein elution was followed by Western blot analysis . Proteins were incubated for 10 min at 30°C and fractionated on a Superdex 200 column ( GE Healthcare ) in 50 mM Tris , pH 7 . 5 , 200 mM NaCl , and 10% glycerol . Nuclear extract from HeLa cells was fractionated on a Superose 6 column ( GE Healthcare ) in 50 mM Tris , pH 7 . 5 , 200 mM NaCl , and 10% glycerol . Histone H1e from calf thymus ( 14–155 , Millipore ) was incubated with the indicated CDK/CYC complexes at 30°C in 50 mM Tris , pH 7 . 5 , 10 mM MgCl2 , 1 mM DTT , 5 mM β-glycerophosphate , 1 mM sodium orthovanadate , and 1 mM ATP . The reaction was stopped after 30′ for end-point assays or after 2′ for the kinetics assay by addition of sample buffer . Dephosphorylation assays were carried out for 2 h at 37°C with Antarctic phosphatase ( New England Biolabs ) . Recombinant His-SCML2B was incubated in vitro in the presence of CDK2/CYCE , CDK1/CYCB , Aurora kinase A , or without kinase at 30°C in 50 mM Tris , pH 7 . 5 , 10 mM MgCl2 , 1 mM DTT , 5 mM β-glycerophosphate , 1 mM sodium orthovanadate , 1 mM ATP , and 1 µCi of γ32P-ATP ( Perkin-Elmer , 3000 Ci/mmol ) . The activity of CYCE2 in nuclear extracts was measured using complexes immunoprecipitated with an anti-CYCE2 antibody ( Epitomics ) . Briefly , the antibody was coupled to Dynabeads ( Invitrogen ) in the presence of 1 mg/ml BSA . Nuclear extracts ( 100 µg ) were incubated with the beads in 50 mM Tris , pH 7 . 5 , 175 mM NaCl for 1 h at 4°C . The beads were washed 3 times with 50 mM Tris , pH 7 . 9 , 200 mM NaCl , and 0 . 05% Igepal CA630 ( Sigma-Aldrich ) ; once with 50 mM Tris , pH 7 . 9 , 100 mM NaCl; and then resuspended in 50 µl of the same buffer . We incubated 5 µl of the immunoprecipitated material with 150 ng of histone H1e , and the reaction was carried out as described above . For double thymidine block , cells were incubated for 16 h with 2 . 5 mM thymidine , released into thymidine free medium for 9 h , and incubated again for 16 h with 2 . 5 mM thymidine . Arrest in mitosis was performed with 0 . 04 µg/ml nocodazole for 16 h , and then cells were released for 10 min . Arrest in S phase was performed with 2 . 5 mM thymidine for 16 h . For serum starvation , cells were incubated for 24 h in the absence of serum . Cells were trypsinized , fixed in 60% ethanol , and incubated with 0 . 09 mg/ml RNase A and 35 µg/ml propidium iodide overnight at room temperature . PI staining was detected in a FACscalibur ( BD ) flow cytometer , and the data were analyzed with FlowJo software . Progression into S phase was measured by incubating U2OS cells with EdU for 5 min at 37°C . EdU incorporation was measured using the Click-iT kit ( Invitrogen ) , following the manufacturer's instructions . After induction of SCML2B with 1 µg/ml doxycycline for 24 h , cells were incubated in complete medium , arrested in the presence of 1 µg/ml doxycycline with either 2 . 5 mM thymidine or 0 . 04 µg/ml nocodazole for 16 h , or treated in the presence of 1 µg/ml doxycycline together with different concentrations of roscovitine for 8 h . Nocodazole-treated cells were then released for 10 min in complete medium . The cells were collected and nuclear extracts were obtained as described above . Nuclear extract was diluted to 1 mg/ml in 50 mM Tris , pH 7 . 9 , 200 mM NaCl , and incubated with Strep-Tactin ( IBA ) beads for 1 h at 4°C . The beads were washed with 50 mM Tris , pH 7 . 9 , 500 mM NaCl , and 0 . 1% Igepal CA630 ( Sigma-Aldrich ) , and then with 50 mM Tris , pH 7 . 9 , and 200 mM NaCl . SCML2B and associated proteins were eluted with 2 mM Biotin in 50 mM Tris , pH 7 . 9 , and 200 mM NaCl . For mass spectrometry analysis , SCML2B was separated by SDS-PAGE , the gel was silver stained ( SilverQuest , Invitrogen ) , and the band was excised . Total SCML2B from in vitro phosphorylation reactions and excised SCML2B bands from Strep pull-downs were digested with trypsin . Phospho-peptides were identified and the position of the phosphorylation was determined by MSMS . For quantification of the phosphorylation in each peptide , the samples were divided in two , one was mock treated , and the other was dephosphorylated with lambda phosphatase ( New England Biolabs ) , following the manufacturer's instructions . To control for loading and experimental variability , an internal standard of unphosphorylated 15N-His-SCML2B tryptic digest was added to each sample prior to sample division and phosphatase treatment . The area of the peak corresponding to the unphosphorylated peptide was measured , and the percentage of phosphorylation was calculated as follows:
|
The processes of development and differentiation require an exquisite coordination of the gene expression program with the proliferation of the cells . The Polycomb group of proteins are important development regulators and most research to date has focused on their involvement in maintaining epigenetic silencing of genes during development and the self-renewal and differentiation of stem cells . Up to now , we've seen that Polycomb proteins influence the transcriptional status of cell-cycle regulators via chromatin modifications . Here we describe a transcription-independent function for a human Polycomb group protein in regulating the cell cycle . We show that the Polycomb group protein SCML2 directly regulates the progression of cells from G1 into S phase by cooperating with p21 to restrain the activation of CDK2/CYCE complexes in early G1 . This function is carried out by the “B” isoform of SCML2 that does not interact with the Polycomb complex PRC1 . Further , SCML2B phosphorylation is regulated through the cell cycle and is partly dependent on CDK1 and CDK2 . These findings highlight a direct crosstalk between the Polycomb system of cellular memory and cell-cycle machinery in mammals , providing insight into novel functions of the mammalian Polycomb system .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2013
|
Polycomb Protein SCML2 Regulates the Cell Cycle by Binding and Modulating CDK/CYCLIN/p21 Complexes
|
During nervous system development , neuronal cell bodies and their axodendritic projections are precisely positioned through transiently expressed patterning cues . We show here that two neuronally expressed , secreted immunoglobulin ( Ig ) domain-containing proteins , ZIG-5 and ZIG-8 , have no detectable role during embryonic nervous system development of the nematode Caenorhabditis elegans but are jointly required for neuronal soma and ventral cord axons to maintain their correct position throughout postembryonic life of the animal . The maintenance defects observed upon removal of zig-5 and zig-8 are similar to those observed upon complete loss of the SAX-7 protein , the C . elegans ortholog of the L1CAM family of adhesion proteins , which have been implicated in several neurological diseases . SAX-7 exists in two isoforms: a canonical , long isoform ( SAX-7L ) and a more adhesive shorter isoform lacking the first two Ig domains ( SAX-7S ) . Unexpectedly , the normally essential function of ZIG-5 and ZIG-8 in maintaining neuronal soma and axon position is completely suppressed by genetic removal of the long SAX-7L isoform . Overexpression of the short isoform SAX-7S also abrogates the need for ZIG-5 and ZIG-8 . Conversely , overexpression of the long isoform disrupts adhesion , irrespective of the presence of the ZIG proteins . These findings suggest an unexpected interdependency of distinct Ig domain proteins , with one isoform of SAX-7 , SAX-7L , inhibiting the function of the most adhesive isoform , SAX-7S , and this inhibition being relieved by ZIG-5 and ZIG-8 . Apart from extending our understanding of dedicated neuronal maintenance mechanisms , these findings provide novel insights into adhesive and anti-adhesive functions of IgCAM proteins .
The structural organization of an adult nervous system depends on two genetically separable processes . First , during development - the wiring phase - the soma and axonal/dendritic extensions of neurons need to be accurately positioned . This process depends on the precisely orchestrated activity of a multitude of well-characterized and dynamically acting guidance and signaling systems [1] , [2] , [3] . Second , during postembryonic life , dedicated maintenance factors ensure that neuronal soma , axon and dendrites maintain their precise position in neuronal ganglia and fascicles [4] . These maintenance factors counteract the various forms of mechanical and physical stress exerted onto a nervous system [4] . The need for such maintenance mechanisms , and the specific maintenance factors involved , were first identified in the nematode C . elegans . The removal of a number of distinct molecules was found to result in no apparent effect on the initial positioning of neurons and fibers during embryonic development; yet the absence of these molecules affected the maintenance of the positioning of neuronal soma and fibers . These molecules include the L1CAM-type adhesion molecule SAX-7 [5] , [6] , [7] , the extracellular matrix protein DIG-1 [8] , a specific splice form the FGF receptor EGL-15 , called EGL-15A [9] and ZIG-3 and ZIG-4 , members of a family of small , secreted two-Ig domain proteins [10] , [11] ( Figure 1A ) . While SAX-7 , DIG-1 , EGL-15A and the ZIG proteins appear to be solely dedicated to a maintenance role , other proteins , such as the basement membrane protein SPON-1/Spondin or UNC-70/β–Spectrin function both during the embryonic neuronal wiring phase and postembryonically in maintenance [12] , [13] . How these maintenance factors functionally interact with one another has been unclear . In this paper , we describe the function of two previously uncharacterized ZIG proteins , ZIG-5 and ZIG-8 , in maintaining neuron soma position . We tie their function specifically to the function of SAX-7 , the C . elegans ortholog of the L1CAM family of vertebrate adhesion molecules . In C . elegans , SAX-7 exists in two splice forms , a short splice form ( SAX-7S ) and a long splice form ( SAX-7L ) ( Figure 1B ) . Intriguingly , several studies have shown that the short isoform , SAX-7S , is more adhesive than the long isoform SAX-7L [6] , [7] , [14] . We show here that the two ZIG proteins ZIG-5 and ZIG-8 serve to prevent the SAX-7L isoform from interfering with cellular adhesion .
Loss of the C . elegans L1CAM ortholog sax-7 affects the maintenance of neuron soma position in the main head ganglia of C . elegans , as well as the positioning of axons in the ventral nerve cord ( VNC ) [5] , [6] , [7] , [14] . Loss of two members of the zig gene family ( zig-3 and zig-4 ) also affects the maintenance of axon positioning in the ventral nerve cord , but do not affect neuron soma position [10] , [11] . To test whether zig genes may phenocopy the sax-7 effect on the maintenance of soma position in head ganglia , we analyzed deletion alleles of all presently known , eight zig gene family members . Visualizing head neuron position either with gfp reporters or by dye labeling showed no defects in any zig single mutant strain ( Figure 1C , 1D ) . Since zig genes may act redundantly , we generated double mutant combinations of all six neuronally expressed zig genes ( that is all zig genes except muscle-expressed zig-6 and zig-7; Aurelio and Hobert , 2002 ) . This double mutant analysis led us to discover striking neuronal soma displacement defects in head ganglia of zig-5 ( ok1065 ) zig-8 ( ok561 ) double null mutant animals ( Figure 1C , 1D ) . This defect can be observed both with cell-type specific gfp reporters ( Figure 1C ) as well as with dye filling of sensory neurons ( 47% animals affected; n = 150 ) . zig-5 zig-8 double mutants also display postembryonic axon position defects in the VNC ( Figure 2 ) . The C . elegans VNC is composed of unilaterally positioned motoneuron axons , located on the right side of the VNC and of axons of bilaterally symmetric neurons that extend along the left and right side of the VNC [15] . The left and right side of the VNC are separated by a midline structure , initially made of neuronal cell bodies , later of a hypodermal ridge [16] . In zig-5 zig-8 double mutants , the axons of bilaterally symmetric neurons become aberrantly positioned across the midline during larval life ( Figure 2A ) . Similar VNC axon defects can also be observed in zig-3 and zig-4 single mutant animals [10] , [11] . Yet the cellular specificity of the VNC axon flip-over defects is broader in zig-5 zig-8 double than in zig-3 and zig-4 single or double mutants , since HSN axons are affected only in zig-5 zig-8 double mutants ( Figure 2B ) . Other than these neuronal morphology defects , zig-5 zig-8 double null mutant animals appear healthy , fertile , locomote normally and do not display other obvious morphological abnormalities . Also , the organization of muscle and epidermal tissues is normal in these mutants ( assessed using specific reporters; data not shown ) , indicating that zig-5 and zig-8 function to maintain the integrity specifically of the nervous system . The zig-5 zig-8 double mutant phenotype can be rescued by genomic DNA clones that encompass these genes ( Table 1 ) and can be phenocopied by RNAi ( Table 2 ) . Both zig-5 and zig-8 have previously been reported to be expressed in many neuronal and non-neuronal cell types in the head of the worm [zig-5: [10] , zig-8: [17]] , as well as the PVT tail neuron which extends its axons along the ventral nerve cord into the nerve ring [10] . The PVT neuron bears a critical role in conveying the function of zig gene family members in controlling maintenance of axon position in the ventral nerve cord [10] , [11] . However , laser ablation of PVT does not affect head neuron position ( 0/57 PVT-ablated animals showed defects ) , and we therefore surmise that secretion of ZIG-5 and ZIG-8 from its many cellular sources in the head is required for maintaining cell body position . Even though we consider this model the most parsimonious based on the expression patterns of zig-5 and zig-8 in many head neurons , we have not been able to experimentally corroborate this notion as we observed no rescue of the mutant phenotypes by driving expression of zig-5 and/or zig-8 under control of variety of different heterologous promoters ( tested promoters: neuronal unc-14 , osm-6 , sra-6 , muscle myo-3 , hypodermis dpy-7; injected at different concentrations from 0 . 1 to 125 ng/µL ) and with different co-injection markers ( pRF4 , ttx-3::mCherry ) . Heterologous expression of zig-5 and zig-8 ( under unc-14 , myo-3 and dpy-7 promoter; 3 lines each ) do not induce defects in a wild-type background , indicating that the lack of rescue through heterologous expression is not the result of overexpression . Since rescue with genomic clones is also just partial , it is conceivable that the correct dosage of zig-5 and zig-8 is critical for their function , yet difficult to achieve in transgenic animals . We note that we have also have problems expressing ZIG-5 and ZIG-8 in heterologous tissue culture cells . As in the case of sax-7 mutants [5] , [6] , [7] , [14] , the neuronal defects of zig-5 zig-8 double null mutants appear to be reflective of maintenance rather than developmental defects . First , the neuronal soma position defects of zig-5 zig-8 double mutants manifest themselves only postembryonically , long after birth and initial placement of neuronal cell bodies in the embryo ( Figure 3 ) . That is , animals in early larval stages appear completely indistinguishable from wild type and the phenotype manifests itself fully only in adult animals ( Figure 3 ) . Likewise , the axons of PVP and PVQ are initially positioned normally along the ventral midline during embryogenesis , but later become displaced during larval growth ( Figure 2B ) . Second , the soma position defect can be evoked also in wild-type animals upon postembryonic knockdown of zig-5 and zig-8 by RNAi ( Table 2 ) . Third , the soma and axon displacement defects of the zig-5 zig-8 double null defects can be suppressed through prevention of locomotion of the animal , achieved either by genetic means or through drug treatment ( Figure 2B , Figure 3B ) . A similar suppression of maintenance defects is a hallmark of all previously described maintenance mutants [6] , [8] , [10] , [11] and indicates that ZIG-5 and ZIG-8 , like other maintenance factors , serve to counteract the effects of physical movement , enabling neuronal soma and axonal projection to appropriately maintain their position . The zig-5 zig-8 double null mutant phenotype , both in terms of the head soma positioning and ventral cord axon positioning defects , is similar to the null phenotype of sax-7 ( Figure 4A , 4B ) . We set out to test whether these loci act in the same pathway by examining whether combinations of null alleles show similar phenotypes ( which would argue for acting in the same pathway ) or enhance each other's phenotype ( which would argue for acting in separate pathways ) . While the null phenotype of sax-7 is completely penetrant for the soma positioning defect , the axon positioning defect is only partially penetrant , therefore allowing to do the genetic interaction test of sax-7 and the zig genes . We find that the sax-7 ( nj48 ) null mutant phenotype is not enhanced in zig-5 zig-8; sax-7 triple mutant animals , suggesting that these genes act in a similar genetic pathway ( Figure 4C ) . To further investigate the genetic interaction of zig-5 , zig-8 and sax-7 , we considered different isoforms of the sax-7 locus . sax-7 produces two distinct splice forms , a longer isoform , sax-7L , that displays the canonical , L1CAM-type 6 Ig/5 FnIII-domain architecture and a shorter isoform , sax-7S , which lacks the first two Ig domains [6] , [14] ( Figure 1A ) . Cell aggregation assays in tissue culture as well as transgenic rescue experiments with the two different isoforms have firmly established that the short isoform , sax-7S , is more adhesive than the long isoform , sax-7L [6] , [7] , [14] . Based on structural studies of various IgCAM proteins , including L1CAM family members themselves , the first 4 Ig domains of SAX-7L are expected to be configured in a horseshoe conformation in which Ig1 and Ig2 fold back onto Ig3 and Ig4 ( Figure 1B ) and it is this horseshoe configuration that is thought to engage in homophilic interactions [18] , [19] , [20] , [21] , [22] , [23] , [24] . It is therefore curious that the SAX-7S form , as well as a mutant version of SAX-7L that is unable to adopt the horseshoe configuration ( through shortening of the hinge region between Ig2 and Ig3 , “SAX-7L ( Δ11 ) ”; Figure 1B ) , is more adhesive than the presumably horseshoe-configured , wild-type SAX-7L protein [6] , [7] , [14] . Further , consistent with the first two Ig domains being dispensable for effective homophilic adhesion , two alleles , eq2 and nj53 , that exclusively disrupt the sax-7L form , but not the sax-7S form [6] , [14] , neither affect soma nor axon position ( Figure 4B; see Figure 1A for schematic presentation of the sax-7 isoform specific alleles ) . Unexpectedly , we find that the two sax-7L-specific alleles each completely suppress the soma displacement defect of zig-5 zig-8 double mutants: While three quarters of zig-5 zig-8 double mutants display soma positioning defects , almost no triple mutant animals do ( Figure 4A , 4B ) . Similarly , the sax-7L isoform specific alleles also suppress the VNC axon defects of zig-5 zig-8 double mutants ( Figure 4C ) . These results indicate that zig-5 and zig-8 function is not required ( i . e . their loss produces no phenotype ) as long as the sax-7L isoform is not present . In other words , in wild-type animals , SAX-7L has the potential to disrupt cell adhesion , but this disruptive ability is counteracted by zig-5 and zig-8 . This disruptive function is uncovered in zig-5 zig-8 mutants . To further probe the disruptive activity of SAX-7L , we tested whether expression of SAX-7L above its usual level in an otherwise wild-type background may cause soma and axon positioning defects . Using a pan-neuronal unc-14 driver to express sax-7L , we indeed find such transgenic animals display severe soma and axon position defects that are virtually indistinguishable from a complete loss of sax-7L/S function or loss of zig-5 and zig-8 function ( Figure 4D shows the soma defect; the axon defect is 58% penetrant , n = 86 ) . The sax-7L overexpression-induced defects are maintenance defects as they are only evident in late larval and adult stages , but not earlier , after hatching ( Figure 4D ) . Overexpression of sax-7S using the same driver does not induce any defects and neither does the sax-7L isoform when it is converted into a more adhesive form through shortening of the hinge region between Ig2 and Ig3 ( sax-7L ( Δ11 ) , as shown in Figure 1B ) ( Figure 4D ) . This result can be interpreted to mean that overexpression of SAX-7L overwhelms the ability of ZIG-5/ZIG-8 to convert SAX-7L to a more adhesive form , thereby revealing the disruptive function of SAX-7L . However , the overexpression effect of SAX-7L is not alleviated in animals that carry arrays with extra copies of zig-5 and zig-8 , but this experiment is not easily interpretable in light of the issues that we discuss above with transgenic zig-5 and zig-8 expression . We considered the possibility that the disruptive function of SAX-7L , which is revealed upon removal of zig-5 and zig-8 , lies in opposing the adhesive function of SAX-7S . To test this possibility , we examined whether overexpression of SAX-7S may overcome the antagonistic function of endogenous SAX-7L and therefore may abrogate the need for zig-5 and zig-8 . To this end , we generated transgenic animals that overexpress sax-7S in a zig-5 zig-8 double mutant background . We find that pan-neuronal sax-7S expression completely rescues the soma positioning defects of zig-5 zig-8 double mutants ( Figure 4B ) . Expressing sax-7S from a muscle specific promoter does not rescue the zig-5 zig-8 defects ( Figure 4B ) . Neuronal overexpression of the long isoform sax-7L also does not rescue the zig-5 zig-8 defects ( Figure 4B ) . However , if the sax-7L isoform is converted into a more adhesive form through shortening of the hinge region between Ig2 and Ig3 ( sax-7L ( Δ11 ) , as shown in Figure 1B ) , sax-7L becomes able to rescue the zig-5 zig-8 double null mutant phenotype ( Figure 4A , 4B ) . We therefore conclude that providing additional copies of non-horseshoe-configured sax-7 can compensate for the loss of zig-5 and zig-8 . This compensatory effect is not simply caused by the addition of unspecific “glue” , since SAX-7S is not able to rescue the neuronal soma position defects of animals lacking the dig-1 maintenance factor ( Figure 4E ) . In conclusion , our results suggest that ZIG-5 and ZIG-8 exert their activity on neuronal architecture through their genetic interactions with SAX-7 . All genetic interaction tests are schematically summarized in Figure 5 . The head neuron position defects described above concern neurons in the most populated ganglia of the head , the two lateral ganglia which containing about 50 neurons . In the smaller ventral head ganglion , the adhesive function of SAX-7 may be regulated in a different manner . This is because the soma positioning defects of the two ventral ganglion neuron types AIY and AVK , that are observed in sax-7 null mutant animals [7] , are not phenocopied in zig-5 zig-8 double null mutant animals ( 0/50 animals show defects ) . Yet , as in neurons of the lateral ganglia , the sax-7 defect again is more efficiently rescued by sax-7S compared to sax-7L [7] . Conceivably , another combination of zig genes may act to promote sax-7 gene function in this cellular context . We also examined the reverse and asked whether zig gene function can generally be explained zig genes affecting sax-7 gene function . To this end we turned to zig-3 zig-4 double mutants in which head soma position is unaffected , but positioning of axons in the ventral nerve cord fails to be maintained [11] . Similar axon maintenance defects are observed in animals lacking sax-7 [7] , but are not observed in the sax-7L-specific allele eq2 ( Figure 4C ) . In this case , eq2 is not able to suppress the zig-3 zig-4 double mutant phenotype ( Figure 4C ) . However , as mentioned above , eq2 can suppress the axon positioning defects of zig-5 zig-8 double mutants ( Figure 4C ) . In conclusion , the interaction of zig genes and sax-7 depends on the type of ZIG proteins that evoke the defects and it depends on cellular context , possibly because the adhesive substrate for specific neuronal ensembles may differ at distinct locations in the worm .
Our analysis has revealed the function of two previously unstudied , secreted ZIG proteins , ZIG-5 and ZIG-8 , thereby further broadening the concept of the requirement of specific factors for maintaining neuronal anatomy . Together with the previously characterized zig-3 and zig-4 genes , four of the eight presently known zig genes have now been specifically implicated in nervous system maintenance . The spectrum of activities of these four zig genes is partially overlapping ( in the VNC ) , but also distinct ( soma positioning in head ganglia ) . Given these precedents , it is conceivable that the remaining zig genes also have functions in maintaining nervous system architecture , perhaps affecting distinct subset of ganglia or even just individual neurons which have so far escaped attention . Moreover , we have provided first hints toward the mechanism , but also diversity of ZIG protein function , by demonstrating that two ZIG family members ( zig-5 and zig-8 , but not zig-3 or zig-4 ) genetically interact with an IgCAM protein , the L1CAM protein SAX-7 . These interaction results are summarized in Figure 5 . Given the involvement of L1CAMs in various neurological diseases , a detailed understanding of this family of proteins is much warranted [25] . We have provided here novel and unexpected insights into mode of regulation of the adhesive activities of the L1CAM protein SAX-7 . We found that the likely horseshoe-configured SAX-7L isoform of SAX-7 , previously shown to provide less homophilic adhesiveness than the shorter isoform SAX-7S [6] , [7] , has in fact an anti-adhesive activity in an in vivo context . This anti-adhesive activity is counteracted in wild-type animals by the two ZIG proteins ZIG-5 and ZIG-8 , and thereby revealed only by either removal of ZIG-5 and ZIG-8 or by overexpression of SAX-7L . SAX-7L may engage SAX-7S in multimers in cis that are not able to engage in homophilic interactions in trans . ZIG-5 and ZIG-8 may be able to break up those complexes , for example by opening the SAX-7L horseshoe configuration , thereby converting SAX-7L into a more adhesive state and/or freeing up SAX-7S , which can then engage in homophilic trans interactions . In vitro cell aggregation assays , which we have not been able to undertake so far due to our inability to heterologously produce ZIG proteins , may address these possibilities in the future . The in vivo studies on SAX-7 protein function have yielded results that are unexpected if one considers that several in vitro studies have provided evidence for horseshoe-configured IgCAMs engaging in homophilic interactions [21] , [22] , [23] , [24] . The SAX-7 case argues for additional and alternative types of homophilic interactions of IgCAM molecules that are not only independent of a horseshoe configuration but also more adhesive than the horseshoe configuration . SAX-7S may be able to engage in other versions of previously described IgCAM interactions , such as the zipper mechanism proposed for the IgCAM superfamily member NCAM [26] . It is also conceivable that the adhesive mechanisms of IgCAM proteins may have diverged in the course of evolution and that the SAX-7/ZIG adhesion pathway may not be phylogenetically conserved . Interestingly , while the activity of zig-5 and zig-8 can be explained entirely through their effect on sax-7 ( as best evidenced by the complete suppression of the zig-5 zig-8 double mutant phenotype by sax-7L-specific alleles ) , an interaction with sax-7 is not the only way the ZIG family members operate . First , the genetic interactions of zig-5 and zig-8 with sax-7 are apparent in some ganglia and axonal tract , but not others . And second , the mutant phenotype of zig-3 and zig-4 , even though similar to those of zig-5 and zig-8 in the context of the VNC , shows no genetic interaction with the sax-7L isoform . The existence of other maintenance factors , such as the extracellular matrix proteins DIG-1 , F-spondin or the FGF receptor EGL-15 all point to a diversity of mechanisms involved in maintenance of nervous system integrity [4] . On a mechanistic level , our findings support the hypothesis [4] that maintenance of tissue integrity is controlled through a finely tuned and tightly regulated balance of adhesive and anti-adhesive forces ( Figure 5 ) . The function of SAX-7L may lie in modulating the strongly adhesive function of SAX-7S at stages and in tissues where high adhesiveness is not desired . This anti-adhesiveness of SAX-7L may in turn be eliminated by ZIG proteins when high degree of adhesiveness is required , such as during postembryonic life to maintain neuron position . It will be interesting to see whether modulation of the homophilic interaction of IgCAM proteins through interaction with other Ig domain proteins is a common theme in maintenance of neuronal positioning . The wealth of Ig domain proteins , many of them secreted , in vertebrate and invertebrate genomes [27] , suggests that many such modulatory interactions remain to be discovered .
The nature and structure of all zig and all sax-7 mutant alleles were previously described in detail [6] , [11] , [14] and some of them are schematically shown in Figure 1A . Genotyping was done by PCR . sax-7 transgenes are described in [7] . Transgenes used to rescue zig-5 zig-8 double mutant phenotype were obtained by injecting fosmid WRM063cG06 containing the zig-5 gene ( injected at 20 ng/µL along with ttx-3::mCherry at 100 ng/µL ) and the YAC Y39E4B , containing the zig-8 gene ( injected at 10 ng/µL along with pRF4 rol-6 ( su1006d ) at 100 ng/µL ) . The following gfp transgenes were used to score anatomy , using a Zeiss Axioplan 2 microscope: oyIs14: Is[sra-6::gfp] , hdIs29: Is[sra-6::DsRed2; odr-2::gfp] oxIs12 Is[unc-47::gfp] , bwIs2 Is[flp-1::gfp] ( described in [11] . Cell body position was examined in three- to five-days old adults , i . e . worms that have lived for 3 to 5 days after the L4 to young adult molt . The position of the soma of ASI and ASH neurons is normally posterior to the nerve ring , and was scored defective when the cell body of at least one neuron was anterior to the nerve ring , on top of it , or contacting it . Ventral nerve cord anatomy was examined in freshly hatched L1 larvae ( <30 min post-hatch ) and in L4 larvae . The axons of the pairs of bilateral neurons examined are normally separate and lie on either side of the ventral midline . An axon was scored defective when a segment of its length was located on the opposite side of the ventral nerve cord and contacted the axon on the other side . For the paralysis on levamisole experiment , embryos were placed on plates containing 50 µM levamisole and seeded with OP50 bacteria , were allowed to reach the L4 and adult stages , and were examined as above . All phenotypes were scored as percent animals defective and results are shown with error bars representing the standard error of proportion . Statistical significance was calculated using the z-test to compare the proportion of abnormal animals of two genotypes . When using the same control for multiple comparisons , the P value was multiplied by the total number of comparisons . oyIs14 L4 hermaphrodites were placed on bacteria harboring plasmids to express dsRNA corresponding to the genes zig-5 and zig-8 ( J . Ahringer library ) , or the empty vector ( L4440 ) . The inserts for the zig-5 and zig-8 plasmids were verified by sequencing . The two bacterial strains containing the zig-5 and the zig-8 plasmids were grown separately overnight . Bacterial cultures were concentrated by centrifugation , mixed and added to RNAi plates . A day later , these animals were transferred onto fresh plates containing the RNAi bacteria . F1 animals were scored for the morphology of the chemosensory neurons at days 3–5 of adulthood . Animals were also placed on RNAi plates at the L1 , L4 and young adult stages , and their neuroanatomy was examined when they reached 3–5 days of adulthood . Similar experiments were carried out in the rrf-3;oyIs14 , eri-1;lin-15b;oyIs14 genetic backgrounds , as well as with RNAi of zig-5 in zig-8 mutant background , and vice versa , but failed to elicit stronger defects .
|
The structure of nervous systems is determined during embryonic development . After this developmental patterning phase , active maintenance mechanisms are required to uphold the structural integrity of the nervous system . This concept was revealed through the genetic elimination of factors in the nematode Caenorhabditis elegans , which left the initial establishment of the nervous system during embryogenesis unperturbed , but subsequently resulted in postembryonic defects in its structural integrity . The extent to which such maintenance mechanisms exist , the nature of the players involved , and the mechanisms through which they operate are subjects of active investigation . In this study , we reveal two novel , previously uncharacterized maintenance factors encoded by the zig-5 and zig-8 genes . Both genes are predicted to encode small secreted immunoglobulin domains . We show that the two proteins operate by counteracting the anti-adhesive effects of a specific isoform of the SAX-7 Ig domain protein , the C . elegans homolog of L1CAM , a human protein involved in various neurological diseases . This study therefore provides novel mechanistic insights into nervous system patterning and may help to better understand the function of an important human disease gene .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"animal",
"models",
"caenorhabditis",
"elegans",
"cellular",
"neuroscience",
"animal",
"genetics",
"model",
"organisms",
"genetics",
"biology",
"neuroscience",
"genetics",
"and",
"genomics"
] |
2012
|
The Secreted Immunoglobulin Domain Proteins ZIG-5 and ZIG-8 Cooperate with L1CAM/SAX-7 to Maintain Nervous System Integrity
|
Current methods for studying the genetic basis of adaptation evaluate genetic associations with ecologically relevant traits or single environmental variables , under the implicit assumption that natural selection imposes correlations between phenotypes , environments and genotypes . In practice , observed trait and environmental data are manifestations of unknown selective forces and are only indirectly associated with adaptive genetic variation . In theory , improved estimation of these forces could enable more powerful detection of loci under selection . Here we present an approach in which we approximate adaptive variation by modeling phenotypes as a function of the environment and using the predicted trait in multivariate and univariate genome-wide association analysis ( GWAS ) . Based on computer simulations and published flowering time data from the model plant Arabidopsis thaliana , we find that environmentally predicted traits lead to higher recovery of functional loci in multivariate GWAS and are more strongly correlated to allele frequencies at adaptive loci than individual environmental variables . Our results provide an example of the use of environmental data to obtain independent and meaningful information on adaptive genetic variation .
The genetic basis of environmental adaptation in natural and agricultural populations is a topic of growing interest and urgency . Conventionally , the search for adaptive genes involves testing for associations of genomic markers with either ecologically relevant traits measured in common garden experiments [1] [2] [3] [4] or with environmental variables [5] [6] [7] [4] [8] . These two approaches reflect the assumption that traits , environment and genotype are correlated due to natural selection , as is indeed expected under local adaptation [9] [10] [11] . In practice , observations and measurements are subject to error and may not accurately reflect the actual variables involved in adaptation [6] . At best therefore , empirical data on traits and environment provide independent approximations of the parameters defining ecological adaptation , offering limited power to detect causative genes when used in isolation . An obvious improvement would be to combine both types of data to better approximate the adaptive process . One example is to identify the most probable selective forces from a set of environmental variables based on their correlation with traits of interest and use these variables in association mapping , as was done recently in Arabidopsis thaliana [7] . Although attractive , the reliance on single variables means that this method cannot account for more complex relations between traits and the environment and makes limited use of the independent information provided by trait and environmental data . An alternative approach , which we explore here , is to extract information from ecological data by modeling traits as a function of multiple environmental variables [12] [13] and to use the resulting trait prediction , conjointly with the observed trait , in a bivariate analysis of genetic association . The reasoning behind this idea is as follows . We start from the usual assumption that individuals from different geographic locations express location-specific , genetically determined trait values that are optimal with respect to some combination of environmental conditions in their native habitat . Furthermore , as in other studies on environmental association , we assume that clinal variation in selective forces causes corresponding differences in gene frequencies across the landscape . Under these assumptions , the value of a trait and its defining selective environment can be treated as two correlated aspects of an individual’s phenotype with a shared genetic basis . In the same way , observed variation in an adaptive trait and a function of environmental variables explaining part of this variation can be treated as two genetically correlated characteristics that are effectively repeated measurements of the underlying selective environment . As has been shown for other genetically correlated traits , such repeated measurements may be combined to increase the power to detect common causative loci by testing for genetic associations with both traits simultaneously using a multi-trait mixed model ( MTMM ) [14] [15] . We propose that testing for genetic loci with an effect on both observed and predicted traits provides more power to detect genes of adaptive significance than mapping on individual traits or environmental variables separately . In addition , environmentally predicted traits may be used in univariate association mapping to map adaptive loci in individuals for which only environmental data is available . We will refer to these two applications of predicted traits as bivariate- and univariate Environmentally predicted Trait Mapping ( ETM ) throughout the paper . We demonstrate the potential of bivariate ETM by computer simulations and evaluate its performance using phenotypic and high-density SNP data from a published association study on flowering time in Arabidopsis thaliana [1] . Flowering time is known to affect fitness in A . thaliana [16] and shows strong geographic variation [17] , making it an ideal trait for our purposes . Moreover , its genetic and molecular basis is well understood [18] [19] . We compare the power of bivariate ETM to recover known flowering genes to that of conventional univariate association methods using single traits or environmental variables . In addition , we use univariate ETM to map flowering genes in individuals without available phenotypic data [7] , an approach that may offer potential for allele mining germplasm collections for adaptive variation .
ETM first models the observed phenotype as a function of environmental data , producing a combination of the environmental variables which we call the predicted phenotype . The trait prediction model is fit on the set of accessions for which both phenotypic and environmental data are available , but the resulting prediction can be extended to the accessions for which there are only environmental data . In case of non-constant prediction , bivariate ETM then performs multitrait association mapping on the observed and predicted phenotype , using all available accessions . In univariate ETM we perform single trait association mapping for the accessions with missing phenotypic data . As proof of concept , we simulated a simple scenario in which an adaptive trait is modeled as a linear function of a random subset of ten out of 30 simulated environmental variables ( Materials and Methods ) . The frequency of the causative SNP was set to be a monotone function of the true adaptive trait . The observed trait was then defined as the sum of a SNP effect and polygenic and residual noise . Four trait prediction methods were implemented: linear model ( LM ) prediction with backward variable selection , elastic nets ( EN ) [20] , random forests ( RF ) [21] and canonical correlation analysis ( CCA ) [22] . For comparison , we also performed bivariate analysis using the trait and the most correlated environmental variable , as well as univariate GWAS on the trait alone . Bivariate mapping was performed both using a test for a common marker effect ( ‘common’ ) and a test whether there is any marker effect ( ‘full’ ) , described in the Materials and Methods ( see also [14] ) . We first simulated a scenario where the heritability is 0 . 5 and the causative SNP explains 5% of the phenotypic variance; correlations between true and observed environmental variables was set to 0 . 8 . For both types of tests , bivariate ETM using predicted traits shows a clear gain in power over univariate mapping ( Fig 1 ) . Bivariate analysis using the environmental variable most correlated to the observed trait performs well in the test for any marker effect , but poorly when testing for a common marker effect , especially at lower significance thresholds . For the four prediction methods the two types of tests perform similarly . Using the test for a common marker effect , CCA showed the highest increase in power ( e . g . 0 . 80 at a −log10 ( p ) threshold of 5 , versus 0 . 64 for univariate mapping ) . Other methods perform similarly with power ranging between 0 . 68–0 . 73 at the same threshold , and achieving larger gains over univariate mapping at higher −log10 ( p ) thresholds . There is a clear relationship across simulated traits between the significance of ETM and correlation between the predicted trait and the simulated true adaptive trait ( S1 Fig ) : ETM is most powerful for simulations where this correlation is large . At lower prediction accuracy the difference with univariate p-values decreases , thus giving smaller differences in power at low −log10 ( p ) thresholds . Similar differences between methods are observed in 8 additional scenarios with heritabilities 0 . 2 , 0 . 5 and 0 . 8 and the causative SNP explaining 2% , 5% and 10% of the phenotypic variance ( S2a–S2i Fig ) . As expected , the advantage of ETM increases for larger proportions of variance explained . In S2a–S2i Fig we also compared bivariate ETM with univariate mapping on the predicted traits , the latter having lower power for most prediction methods , except for low heritabilities . For CCA , univariate mapping also performs well for higher heritabilities . Next , we modified the scenario of Fig 1 in the following ways: by lowering the correlations between true and observed environmental variables to 0 . 5 ( S3 Fig ) , by introducing genetic correlations between the trait and some of the environmental variables ( S4 and S5 Figs ) , and by removing the association between the environmental variables and the causative SNP ( S5 and S6 Figs ) . In the first case , the larger measurement errors in the observed environmental variables leads to a decrease in power of ETM , which however is still more powerful than univariate mapping ( S3 Fig ) . We then performed simulations where the polygenic component of the trait is correlated with the environmental variables defining the true adaptive trait , reflecting the presence of adaptive loci elsewhere on the genome . When the SNP explains 5% of phenotypic variance ( as in the main scenario ) , differences among methods become smaller , in particular between CCA and ETM with the correlated variable ( S4 Fig ) . When the SNP does not affect the phenotype , p-values appear randomly distributed on the unit interval ( S5 Fig ) , indicating that ETM adequately corrects for population structure . In our last scenario ( S6 Fig ) , neither the SNP under consideration nor the polygenic effect was related to the environmental variables . In this case ETM has lower power than univariate mapping , as the SNP is only associated with one the two variables . The largest loss in power then occurs in the test for a common effect , while also the test for any marker effect is affected due to less degrees of freedom [14] . Given the similar performance of the two tests we chose to present all subsequent results for the common marker effect only . We consider this test to be conceptually more appropriate for the detection of loci associated with both the observed trait and its selective environment , which are expected to be positively correlated . We used the statistical methods described above to predict flowering time variation among 149 Arabidopsis thaliana accessions [1] , using public data for 61 environmental variables ( S1 File ) . These predictions will be used in bivariate and univariate ETM below . As expected [23] [17] , flowering time is strongly correlated with variables related to latitude such as day length , potential evapotranspiration and temperature ( S7 Fig ) . Spring and summer day length are most correlated with flowering time [7] , each explaining 40% of variation compared to 29% for latitude itself . The importance of these variables is reflected in the trait predictions ( S8–S11 Figs ) , where day length is among the most important variables for all prediction methods . The contribution of other variables varies between methods , with the LM and RF prediction assigning relatively high importance to precipitation variables not strongly correlated with latitude ( S8 and S10 Figs ) . The highest correlation between the predicted trait and any single environmental variable , summer day length in all cases , ranges between 0 . 71–0 . 84 for LM , RF and CCA but is notably higher for EN ( r = 0 . 98 ) ( S8–S11 Figs ) . The EN-predicted trait may therefore offer little advantage over day length when used in bivariate ETM . Notwithstanding the differences between methods , trait predictions are highly correlated among themselves ( r = 0 . 78–0 . 88 ) and with the observed trait ( r = 0 . 84 ( CCA ) to r = 0 . 68 ( EN ) ) , suggesting that ETM performance will be similar for different prediction methods . For the different methods , we measured the cumulative success in recovering 240 known flowering genes ( S2 File ) as a function of the number of evaluated candidate genes . We thereby assume that GWAS results are used to create a list of candidate SNPs or genes of a certain length as a basis for further evaluation ( see S12 Fig for recovery as a function of p-values for comparison ) . SNPs were sorted by increasing p-value and candidates were defined as genes overlapping with or being closest to any of the top 2000 SNP positions , evaluated successively in order of significance ( approximately 1% of all SNPs ) . We compared univariate association mapping on observed flowering time , bivariate ETM and bivariate analysis using the most correlated trait ( Summer day length ) . Significance of enrichment was calculated as the probability of recovering the observed number of flowering genes by chance ( see Materials and Methods ) . All methods result in significant enrichment but recover only a modest number of genes , yielding 27 flowering genes at most ( Fig 2 , left ) . Maximum significance of enrichment ranged from 5 ⋅ 10−3 to 4 . 9 ⋅ 10−6 and was achieved after evaluating varying numbers of genes ( Fig 2 , right ) . Bivariate ETM outperforms univariate trait mapping over the entire range , with a maximum difference in recovery of 9 flowering genes at 621 evaluated genes ( Fig 2 , left ) . ETM based on LM and CCA trait prediction performs particularly well , with high and sustained recovery and peaks of maximum significance of enrichment of 4 . 9 ⋅ 10−6 and 1 . 3 ⋅ 10−5 respectively . Overall , the recovery curves for EN prediction and summer day length are similar , as expected based on the high correlation between the two variables . For all prediction methods ETM p-values showed some inflation , which also occurred in univariate mapping of the predicted traits , the individual environmental variables and to a lesser extent the observed trait ( S13–S14 Figs ) , and therefore does not appear to be an artifact of our method . Inflation largely disappeared in univariate analyses with a multi-locus mixed model [24] ( S15 Fig ) , suggesting that inflation is due to large effects of a small number of loci , insufficiently captured by the kinship matrix . Considering the top 400 candidate genes for each method , univariate mapping on observed flowering time recovers 2 flowering genes within the first 16 , with probabilities of 7 . 2 ⋅ 10−3 , but the total of 4 recovered genes does not represent a significant enrichment ( p = 4 . 1 ⋅ 10−1 ) . Bivariate ETM , by contrast , recovers 9–13 flowering genes within the first 400 candidates ( p = 5 . 6 ⋅ 10−3 − 2 . 6 ⋅ 10−5 ) , with all prediction methods providing higher enrichment than summer day length ( 7 genes , p = 4 . 5 ⋅ 10−2 ) . The four types of bivariate ETM all recover the genes SVP , GA1 , DFL2 , LDL1 , SPA2 , FPF1 , DOG1 , within the first 400 candidates ( Table 1 ) . The latter four genes are only recovered by univariate mapping after considering at least 100 additional genes . Although different bivariate ETM analyses identify different sets of genes , overlap is relatively high . Considering the top 400 candidate genes of each prediction method , an average of 249 ( 220–282 ) genes is shared between prediction methods ( Fig 3 ) , compared to an average of 199 between bivariate ETM and univariate mapping . Bivariate ETM and standard association mapping thus recover different genes . These differences are unlikely to be due to chance , as shown by the fact that bivariate ETM ( LM prediction ) with a simulated trait equally correlated with the observed trait ( i . e . r = 0 . 81 ) identifies only 5 unique genes compared to univariate association mapping ( Fig 3 ) . Environmental prediction of trait values offers the possibility of association mapping when phenotypic data is incomplete . Traits of interest can be predicted across geographic space using geographic information and association mapping may then be performed on any set of georeferenced individuals for which genotypic data are available . Fig 4 shows geographic maps of predicted flowering time obtained by the four different prediction methods . Although the importance of latitude is evident , in all cases the predicted trait surface clearly reflects the effect of variables that are not strongly correlated with latitude . We compared the performance of univariate ETM to that of ( univariate ) association mapping on summer day length and latitude , for a dataset of 478 genotyped and georeferenced accessions for which no flowering time data was available and whose range of predicted trait values did not exceed that observed for the 149 phenotyped individuals . Recovery of known flowering genes is somewhat lower compared to bivariate ETM ( Fig 5 ) . Although performance is only slightly higher compared to random permutations , maximum enrichment is significant in all cases . Differences in performance between methods are small , but ETM has higher recovery and enrichment within the first 400 genes compared to mapping the two environmental variables individually . Within these top 400 candidates , SVP , CRP , SPA2 , DOG1 , PIE1 and FRI are recovered by more than one method ( Table 2 ) and for each , ETM with LM prediction requires fewer candidate genes to be evaluated compared to mapping the two environmental variables , although the best prediction method differed between genes . FRI is a well studied , major flowering locus in A . thaliana[23] [25] , which together with the gene FLC affects the latitudinal cline in flowering time [26] [17] [27] [28] . FLC ranks 617 and 627 using RF and day length respectively , but is not recovered at all by latitude . The relatively weak recovery of FLC , FRI , SVP and DOG1 with latitude is surprising since all have been reported to show allelic variation with latitude [29] . This suggests that predicted traits used in ETM may be better correlated with the underlying gene frequency at these loci than latitude itself . We confirmed this by estimating the geographic frequencies of the SNP distinguishing the two functional haplotypes at FLC and FRI [16] and of the SNPs with the lowest p-values at SVP and DOG1 , and correlating these to the different variables including latitude ( Fig 6 ) . In each case , the best trait prediction ( i . e . yielding highest r2 with SNP frequency ) has a higher correlation with SNP frequency than either summer day length or latitude . In fact , our data provides no evidence for a latitudinal trend for either FRI or FLC , while the correlation with predicted flowering time is weak but significant ( p < 1 ⋅ 10−9 ) .
We have explored the use of environmentally predicted traits for genome-wide mapping of genes underlying adaptive trait variation . This is basically an extension of the concept of phenotype to include the environment . That idea is not new , in the sense that it has been implicit in most studies relating environment to gene frequency . The novelty of our approach lies in the fact that this extension is made explicit and is used in conjunction with the observed trait of interest to obtain a better approximation of the selection gradient responsible for trait variation . Although this may seem counter-intuitive at first , its merit becomes apparent when considering that information on correlated environmental variables can be used to reduce the effect of experimental error in the same way as correlated traits can [30] [14] [31] . We thereby take advantage of so-called latent variables , which are factors indirectly related to the trait of interest and that are generally considered a source of spurious associations [32] . Although selective forces determining trait variation may sometimes shape allele frequencies at non causal loci ( e . g . those affecting an unmeasured adaptive trait ) , independent estimates of these selective forces can at the same time help to find true associations , particularly when combined with the trait itself . Bivariate ETM is designed to detect genes whose frequencies correlate with selective forces that have shaped a trait of interest . These are likely to affect the target trait directly , although they may also be genes affecting correlated adaptive traits . In our case an average of 87% of the top 2000 SNPs for bivariate ETM had p-values below 0 . 05 for flowering time itself . Since our primary aim is to find genes related to adaptation however , any gene that is affected by the same selective environment is of interest , regardless of its causal relation to the trait . The success of this approach does require that traits and the environment provide complementary estimates of underlying selective forces , something that may not always be the case . The result that enrichment for known flowering genes is higher for bivariate ETM than for univariate mapping on the trait itself , and that this is not observed for randomly simulated variables with the same correlation to the observed trait , suggests that predicted and observed traits indeed complement each other . One thing to observe , is that our definition of recovery as the closest gene to a detected SNP , deviates from Atwell et al . ’s decision to consider SNPs within 20kb of their candidate genes [1] . Our criterion was chosen to avoid calling multiple genes per evaluation position and reflects the fact that in the Arabidopsis genome , LD is estimated to decay within 10kb on average [33] . Another application of environmental trait prediction is the mapping of adaptive genes in individuals with missing phenotypic information . It offers potential for mining the growing genomic data available for many species without the need for complete phenotypic data , and exploiting the wealth of publicly available geographic and environmental data . Our results on mapping flowering genes in unphenotyped individuals are encouraging in the sense that more genes are found than expected at random . On the other hand , the improvement achieved over single environmental variables such as latitude is rather modest . This probably reflects the fact that environmental variables related to latitude are the dominant selective agents affecting flowering time , making it hard to improve over the use of well chosen single environmental variables . Nonetheless , at several genes with known association with latitude , estimated gene frequencies are more strongly correlated with predicted flowering time than with latitude . This observation provides evidence that mapping on predicted traits has the potential of producing more relevant association results than single environmental variables chosen a priori . In conclusion , we have provided evidence that integrating environmental and phenotypic data can improve our ability to map genes of adaptive significance . We have thereby explored several statistical methods for modeling traits as a function of the environment . We do not consider our results conclusive with respect to the best prediction method and more work remains to be done in that respect . Alternatives such as sparse multivariate methods [34] may be worth exploring . In addition , it is conceivable to integrate prediction into the MTMM step of our approach , and target the combination of environmental variables with the highest genetic rather than phenotypic correlation . This however implies an optimization problem for which no algorithms currently seem to be available . Alternatively , bivariate MTMM could be replaced by multivariate MTMM , including all environmental variables individually ( as well as the observed trait ) , but state-of-the art approaches [15] currently cannot perform GWAS on more than 10 traits . Another issue is that of inflation , which may affect the distribution of p-values in any GWAS study due to confounding of the polygenic background with population structure [35] [36] or the occurrence of large effect loci [24] . Although we adopt the standard MTMM approach of correcting for population structure by a marker-based kinship matrix it is clear that for traits like flowering time there is a certain degree of residual inflation . The fact that inflation for most traits was adequately controlled in a univariate multi-locus mixed model ( MLMM ) , suggests there is scope for the development of a multi-locus version of MTMM . In terms of application , it will be interesting to test the added value of our approach for traits that are more weakly correlated with known environmental factors , such as is the case for disease or drought resistance . We hope that the present work may serve as a first step in moving adaptation mapping beyond the traditional univariate analysis of traits and environmental variables and towards a more integrated use of all available data .
We used two datasets from two highly cited examples of trait association and environmental association in A . thaliana [1] [7] . The first set consisted of 199 phenotyped accessions of which we retained 149 individuals with available Eurasian geographic coordinates and no missing data for any of the included traits . We reduced data on flowering time measured at 10 , 16 and 22 degrees Celsius to a single principal component explaining 90 percent of total variation , which was used in all subsequent analyses , unless stated otherwise . The second set consisted of 948 georeferenced accessions , sampled across Eurasia , of which we excluded 39 accessions with non-terrestrial coordinates . High-density Single Nucleotide Polymorphism ( SNP ) data , using the Affymetrix 250K SNP-tiling array was available for both studies [29] . SNP positions and gene annotations were based on version 10 of the Arabidopsis genome annotation ( TAIR10 ) . A list of 240 mapped candidate genes for flowering time was obtained from [1] and [37] , complemented with a subset of genes derived from the list of known Arabidopsis flowering genes available from the Prof . Coupland lab ( MPIPZ , Cologne , Germany; https://www . mpipz . mpg . de/14637/Arabidopsis_flowering_genes ) . SNP positions with the highest frequency differentiation at functional variants of the flowering genes FLC and FRI were identified based on 85 accessions for which functional haplogroups were available [16] . We compiled georeferenced climatic , soil and vegetation data from a variety of public sources ( S1 File ) , resulting in a final set of 61 environmental variables with a spatial resolution ranging from 0 . 5 to 50 km . Remote sensing data were mosaicked , time averaged and converted to GIS raster layers with custom R scripts , using functions from the programs cdo [38] , MRT [39] and the package Raster [40] . Average day length for different seasons was calculated from latitude [41] . Visualization of geographic data and assignment of environmental variables to sample locations was done using the QGIS software [42] . Estimates of continuous allele frequencies across the landscape were produced using the program SCAT [43] . Our ETM procedure can be summarized as follows . First we predict the observed phenotype as a function of environmental data . Below we describe four possible prediction methods , but in principle any method can be used here . Provided this prediction is not constant we then perform bivariate GWAS on the observed and predicted phenotype ( bivariate ETM ) , or univariate GWAS on the predicted phenotype alone ( univariate ETM ) . In the case of bivariate ETM , we consider the test for a common marker effect ( details given below ) , but the test for any marker effect is possible as well . For all methods ( bivariate/univariate ETM , univariate mapping ) SNPs were ordered by their significance and the 2000 SNPs with lowest p-values were considered as candidate SNPs . We assigned each of these SNPs to the gene ( s ) overlapping with its position or to the closest gene in the case of non-genic SNPs . This criterion differs from that used by Atwell et al . ( 2010 ) [1] , who assigned genes within a 20kb window around each SNP as candidates . Our criterion was designed to minimize the number of genes evaluated per SNP , without requiring arbitrary decisions on relevant window size ( See S16 Fig for a comparison of results using different criteria ) . We counted how many out of the 240 known flowering genes were recovered as a function of the number of unique genes considered when going down the ordered list of candidate genes . At each point , enrichment was calculated as the hypergeometric probability of finding ( at least ) the number of unique flowering genes , given the number of genes evaluated so far , the total of flowering genes ( 240 ) and the total of 29 , 477 genes assigned to any of the SNPs . We simulate traits and environmental variables for a fixed set of n = 300 accessions taken from the regmap , of which we randomly selected 100 Swedish , 100 French , 50 German and 50 Czech accessions . Each simulation consists of k = 30 simulated environmental variables and 1 simulated trait . Each simulation starts by drawing a Gaussian n × k matrix XT , containing the true ( unobserved ) environmental variables at the locations of origin of the accessions . XT specifies what we will call the true environment . First we randomly draw a subset S ⊂ {1 , … , k} , containing s = 10 environmental variables , which will later form the environmental gradient . We will use the notation XT ( S ) for the submatrix of XT with columns defined by S . To model confounding with population structure , the variables in XT contain polygenic components , such that their heritabilities are 0 . 5 . Specifically , XT is the sum of Genv and Eenv , which are drawn from zero mean matrix variate normal distributions ( see e . g . [15] ) . Genv is simulated together with the column ( n × 1 ) vector Gtrait , such that ( Genv , Gtrait ) is matrix variate normal with column covariance matrix VG and row covariance given by a marker-based kinship matrix K . Gtrait is the polygenic signal in the observed trait yO ( defined below ) . VG is the ( k + 1 ) × ( k + 1 ) covariance matrix of ( Genv , Gtrait ) . The off-diagonal elements of VG are chosen such that for each pair of variables in S , the genetic correlation is 0 . 5 . Also the genetic correlations between environmental variables from the complement of S are set to 0 . 5 , while it is zero for all variables j ∈ S and j′ ∈ Sc . The correlation between Gtrait and the columns of Genv ( S ) is either 0 or 0 . 5 . In the latter case , this reflects the assumption that Gtrait is to a certain extent adaptive . The correlation between Gtrait and the columns of Genv ( Sc ) is always 0 . The row and column covariance matrices of Eenv are both diagonal . Given the outcome of XT we then simulate XO , the observed environmental variables , by adding random Gaussian errors with variance chosen as to achieve a correlation of 0 . 80 , for each corresponding pair of columns in XT and XO . We then define the environmental gradient as yT = βXT ( S ) , where β1 , … , βs are drawn independently from a uniform distribution on the interval [−1 , 1] . For simplicity we assume that yT is the ( unobserved ) adaptive phenotype , although more complex relations between environmental gradients and phenotypes can be expected in nature . The vector f of causal allele frequencies at each simulated location , is defined as f ( y T ) = e λ y T / ( 1 + e λ y T ) with λ = 3 , and hence has a correlation of 1 with yT . A corresponding genotypic vector g is formed by sampling a single allele for each location from a Bernoulli distribution with probability f . Finally , we simulate the vector of observed phenotypes yO = βsnp g + Gtrait + Etrait , where βsnp represents the SNP-effect on the trait , Gtrait is the polygenic effect defined above , and Etrait is residual noise . We performed the following sets of 2000 simulations: The main set ( Fig 1 ) , where βsnp and the variance of Etrait are chosen such that the SNP explains 5% of the phenotypic variance , while Gtrait and Etrait explain respectively 45% and 50% , i . e . the heritability of the observed trait is 0 . 5 . The correlations between Gtrait and Genv ( S ) are set to 0 . In S2a–S2i Fig , we repeated the simulations from the main set , for heritabilities of 0 . 2 , 0 . 5 and 0 . 8 , and the causal SNP explaining 2% , 5% and 10% of the phenotypic variance . In S3 Fig , we repeated the simulations from the main set , lowering the correlations between true and observed variables to 0 . 5 . In S4 Fig , we repeated the simulations from the main set , the correlations between Gtrait and Genv ( S ) being 0 . 5 . In S5a and S5b Fig , we repeated the simulations from the main set , the correlations between Gtrait and Genv ( S ) being 0 . 5 . Additionally , the SNP effect ( βsnp ) was set to 0 , and Gtrait explained 50% of the variance . In S6 Fig , we repeated the simulations from the main set , but sampled the vector g of SNP scores randomly from independent Bernoulli ( 0 . 5 ) distributions , i . e . independent of any environmental variable . In all cases , ETM p-values from simulations yielding constant trait predictions were set to their corresponding univariate GWAS p-values .
|
Finding genes involved in adaptation to the environment has long been of interest to evolutionary biologists and ecologists . Most commonly , researchers look for loci whose differences in allelic state correlate with differences in a particular trait or environmental variable such as temperature . The implicit assumption behind such methods is that natural selection by the environment will shape variation in adaptive traits through associated changes in allele frequencies . This means that both environmental and phenotypic variation are relevant for detecting adaptive genes , although we have incomplete knowledge of how the two types of variation relate to adaptation . Here we present a method that aims to identify adaptive genes by combining phenotypic and environmental data . We first predict trait variation from a set of environmental variables as a way to extract the most biologically relevant information from the environment and then look for genes associated with both the predicted and observed trait . Using simulations and published data from the model plant Arabidopsis thaliana , we show that this approach may find adaptive genes more effectively compared to existing methods . We also demonstrate that predicted traits can be used to identify relevant loci in individuals for which no phenotypic data is available .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Genome-Wide Association Analysis of Adaptation Using Environmentally Predicted Traits
|
Gene duplication with subsequent interaction divergence is one of the primary driving forces in the evolution of genetic systems . Yet little is known about the precise mechanisms and the role of duplication divergence in the evolution of protein networks from the prokaryote and eukaryote domains . We developed a novel , model-based approach for Bayesian inference on biological network data that centres on approximate Bayesian computation , or likelihood-free inference . Instead of computing the intractable likelihood of the protein network topology , our method summarizes key features of the network and , based on these , uses a MCMC algorithm to approximate the posterior distribution of the model parameters . This allowed us to reliably fit a flexible mixture model that captures hallmarks of evolution by gene duplication and subfunctionalization to protein interaction network data of Helicobacter pylori and Plasmodium falciparum . The 80% credible intervals for the duplication–divergence component are [0 . 64 , 0 . 98] for H . pylori and [0 . 87 , 0 . 99] for P . falciparum . The remaining parameter estimates are not inconsistent with sequence data . An extensive sensitivity analysis showed that incompleteness of PIN data does not largely affect the analysis of models of protein network evolution , and that the degree sequence alone barely captures the evolutionary footprints of protein networks relative to other statistics . Our likelihood-free inference approach enables a fully Bayesian analysis of a complex and highly stochastic system that is otherwise intractable at present . Modelling the evolutionary history of PIN data , it transpires that only the simultaneous analysis of several global aspects of protein networks enables credible and consistent inference to be made from available datasets . Our results indicate that gene duplication has played a larger part in the network evolution of the eukaryote than in the prokaryote , and suggests that single gene duplications with immediate divergence alone may explain more than 60% of biological network data in both domains .
Genomic sequence data provides substantial evidence for the abundance of duplicated genes in all organisms surveyed: at least 40% of genes in two prokaryotes [1 , 2] and 15%–90% of genes in eukaryotes [3–5] appear to be products of gene duplication . This suggests that gene duplication is a key mechanistic driving force behind the evolution of complex organisms [6] . In particular , the fact that the number of interactions shared by paralogous proteins decreases with sequence similarity in Saccharomyces cerevisiae [7 , 8] indicates that gene duplication might shape the topology of protein networks . In theory , the evolutionary fate of gene duplicates can differ: ( D1 ) one copy may become silenced ( nonfunctionalization ) ; ( D2 ) both copies are very similar in sequence , and one is functionally redundant to the other [9]; ( D3 ) both copies are mutationally compromised , and one or more subfunctions of the single progenitor are altered ( subfunctionalization ) ; or ( D4 ) one copy may acquire a novel function preserved by natural selection , while the other copy retains the original function ( neofunctionalization ) . The strength of ( D3 ) is that it does not rely on the sparse occurrence of beneficial mutations , but on more frequently occurring loss-of-function mutations in regulatory regions [10 , 11] . Alternatively , based mostly on the assumption that the number of protein pairs that may acquire a novel function is large , several studies [7 , 12 , 13] promoted the relative importance of ( D4 ) , as well as the formation or degeneration of functional links between proteins in general ( link turnover ) . The structure of protein interaction networks ( PINs ) derives from multiple stochastic processes over evolutionary time scales , and a number of mechanisms have been proposed to capture aspects of network growth [12 , 14–16] . These models correspond to our limited understanding of network evolution , and there is no consensus as to which mechanisms are required to produce “realistic” models for biological PINs [17] . What is required is to be able to fit to biological network data a model ( or mixture of models ) of growing networks that reproduce more accurately the properties of real biological networks than simple preferential attachment [18] or duplication models [16] . For duplication–attachment models of network growth , Wiuf et al . [19] developed a full likelihood approach; this class of models , however , does not adequately explain the structure of most biological network data . The analysis of PINs is notoriously difficult because measurements of PINs are subject to considerable levels of noise [20 , 21] , and in their present guise , offer only an incomplete description of the true interaction network [22] . Interaction datasets are also highly averaged , not only over technical aspects such as the experimental protocol , but also over the precise cellular conditions under which interactions take place , interaction strength , and individual variation . In this work , we develop an approximate , likelihood-free Monte-Carlo inference technique based on approximate Bayesian computation ( ABC ) [23–26] for inference on biological protein network data . Previously , an approximate composite likelihood approach has been proposed , using only the degree sequence to test whether or not simple scale-free models offer an adequate description of PIN data [27] . Owing to the complexity of PINs , we take multiple features of the data into account , which characterize PINs more fully . Our likelihood-free approach allows us to reliably compare more complex models of network evolution in order to study the relative importance of aspects of gene duplication and subsequent interaction divergence in prokaryotic and eukaryotic network evolution . Within the limits of the model and the available data , we find evidence for different dynamics in PIN evolution between the prokaryotic and eukaryotic domains as represented by H . pylori and P . falciparum , respectively . The degree sequence [18] , as well as the frequency profile of motif counts [28] are widely used to analyze protein network data . Our analysis shows that the degree sequence barely captures evolutionary footprints of PINs relative to other statistics . It also suggests that motif counts are extremely variable over the modelled evolutionary history of PINs , and thus inference based on these alone is fragile . Only the simultaneous analysis of many global aspects of PIN data rendered our evolutionary study credible and consistent .
To study the relative importance of aspects of duplication divergence in network evolution between different domains , we simulated the evolutionary history of PINs with a mixture of duplication divergence with parent–child attachment ( DDa ) and preferential attachment ( PA ) ; see Box 1 . At each step , the network either grows according to DDa with probability 1 − α or PA with probability α . More precisely , let Gt be a network with t nodes ( proteins ) , v a new node , u a randomly chosen parent node in Gt , δDiv the divergence probability , δAtt the parent–child attachment probability , and let θ = ( δDiv , δAtt , α ) . Then the probability of Gt+1 conditional on Gt and u is The terms PA ( u , v ) and DDa ( u , v , δDiv , δAtt ) correspond to the probabilities of moving to the new configuration under PA and DDa , respectively . They are explained in Box 1 and defined fully in Protocol S1 . By repeated application of the mechanism in Equation 1 , we grew PINs to the approximate number of open reading frames in the respective genomes ( H . pylori: 1 , 500 , and P . falciparum: 5 , 300 ) . We chose this mixture evolution model for a number of reasons . DDa agrees with aspects of genome evolution by gene duplication [29] . Several studies [7 , 8 , 10 , 30] found a rapid divergence of the interaction profiles of duplicate genes , indicating that duplication and subsequent divergence might be adequately modelled in a single step . Importantly , DDa may relate to subfunctionalization [31]: as a rule , at least one edge disappears , and the duplicates share the pleiotropy of the parent node [10 , 32] . Also , the model does not disagree with purifying selection that maintains the ancestral function at both duplicates [9 , 33 , 34] , because , occasionally , all ancestral edges are retained . The second component of the mixture model , first introduced in [18] , is a generic local growth mechanism based on PA that may explain some characteristics of networks , in particular the approximate power-law decay of the node degrees . In the present context , it captures effects of network growth which are not specifically related to ( D1–D3 ) . Such effects are likely present in network evolution; Middendorf et al . [35] showed that PINs simulated by DDa alone may underrepresent tree-like subgraphs , whereas these are more accurately generated by PA . Also , horizontal gene transfer is a major force in prokaryote evolution . It is plausible to model such transfer with an attachment process , although no particular model has been proposed in the literature . Overall , in the mixture model ( Equation 1 ) , network evolution proceeds by repeated node addition . Apart from rate homogeneity over all proteins , there are thus no further assumptions on the evolutionary clock of our model; a property that is particularly desirable because evolutionary events such as duplication or interaction divergence are generally unavailable or difficult to estimate reliably . Since link turnover is suspected to operate on a different time scale than duplication divergence , extending the model ( Equation 1 ) with preferential link rewiring [12] would imply further assumptions on the evolutionary clock; potentially , phylogenetic data could help to fit such birth and death models of network evolution . The evolution parameters are abstract quantities that subsume a number of complex biological processes [36] . The parameter δDiv may , for example , relate to mutations and insertions or deletions on the sequence level , but also to novel posttranslational modifications or translocations into a different cellular compartment of one interaction partner . Notably , δDiv is associated with immediate divergence and thus differs from divergence probabilities obtained from sequence data , since the latter are usually inferred over a time interval [37] . The parameter δAtt represents the probability of link formation between duplicates . In this study , the mixture parameter α is of particular interest; we ask whether and to what extent , despite high incompleteness , the PIN topology of representatives from the prokaryotic and eukaryotic domains contain evolutionary footprints that may be related to a model that captures hallmarks of network evolution by ( D1–D3 ) . To account for incomplete data , random subnetworks of order N are chosen from the simulated networks that are grown to approximately the number of open reading frames in the respective genomes . Here , N is the number of proteins with observed interactions in the two datasets ( H . pylori: 675 and P . falciparum: 1 , 271 ) . The PIN datasets generated by Equation 1 and subsequent subsampling are dominated by stochastic effects ( Figure S1 ) ; nevertheless , different parameters leave distinguishable imprints on simulated PINs ( Figure S2 ) . The Bayesian paradigm is a powerful probabilistic framework for making inference on complex stochastic systems and allows all sources of uncertainty to be accounted for [38] . We applied this paradigm to estimate the posterior density p ( θ| D℘ ) of θ , given a real PIN dataset D℘ . Bayes' Theorem relates p ( θ| D℘ ) to the likelihood p ( D℘|θ ) and the prior of θ , p ( θ ) , via where ∝ denotes “proportional to . ” In the absence of substantial prior information on θ , we use a uniform prior . The increased flexibility of Equation 1 comes at a computational cost and prohibits likelihood calculations that have been formalized by Wiuf et al . [19] for only very simple evolution models . ABC confers computational tractability by circumventing the problem of evaluating the likelihood directly [23–26] and relies instead on the simulation of networks and the computation of network summaries . All ABC algorithms have in common to approximate first the data D℘ by a set of summaries S℘ D℘ , for example ND and DIA ( see Box 2 for a glossary of summary statistics and their abbreviations in the text ) in the case of protein networks , and then proceed through several steps to sample parameter values from an approximate posterior density; see Materials and Methods for details . One approach is to sample from the prior density ( noninformative in our case ) and accept the proposed value , given that certain criteria are fulfilled . However , as suggested by Figure S2 , only a small range of parameter values generate data with summaries close to S℘ D℘ . Consequently , we anticipate that generating candidate parameters from the prior will be highly inefficient . Likelihood-free inference ( LFI ) within Markov Chain Monte Carlo ( MCMC ) [25] improves efficiency of standard ABC by exploiting knowledge of the current parameter value to make an educated guess on the next one . The details of algorithm ABC-MCMC are outlined in Material and Methods . It is guaranteed to eventually generate a series of correlated samples from where ɛ is the tolerance according to the distance function d , and S℘θ is the set of summary statistics calculated on simulated data with parameter θ . If ɛ is large , then Equation 3 will roughly equal the prior . On the other hand , if ɛ is very small , then the estimator ( Equation 3 ) is too variable . In the latter case , MCMC may become inefficient or even fail [25 , 26] . If ɛ is small and the set of summaries captures all aspects of the protein network sufficiently well , then In order to achieve an approximation of the posterior for inference on protein networks , we modified ABC-MCMC to our algorithm LFI; see also Materials and Methods . Our evolutionary analysis of real PIN datasets centres on a comparison of two representatives from the prokaryotic and eukaryotic domain . We obtained descriptions of the PINs of H . pylori and P . falciparum from the Database of Interacting Proteins ( http://dip . doe-mbi . ucla . edu ) . We first investigated LFI with different sets of summaries on simulated data as outlined in Protocol S1; based on the test results , we selected the set of summaries WR + DIA + CC + + FRAG for LFI . We successfully applied LFI on the H . pylori PIN . Figure 3 presents the MCMC chains for the divergence parameter δDiv ∈ [0 , 1] , and the estimated posterior p ( δDiv| D℘ ) . Similar good convergence was obtained for the attachment probability δAtt and the mixture parameter α , and the 80% credible intervals ( i . e . , the inner range of values of a random variable that attains 80% posterior mass ) are presented in Table 1 . Technically important , the Markov chain resulting from algorithm LFI did not get stuck and did not sit in the tails for relatively small threshold values ɛmin . We could not reproduce our results without averaging over an ensemble of B = 50 simulated PIN datasets during burn-in , nor without tempering of ɛ and Σ as described in Materials and Methods . Based on our theoretical considerations with smd ( θ ) and cv ( θ ) and our test results , we believe approximation ( Equation 4 ) has been achieved , but note that ultimate evidence cannot be provided since evaluating the likelihood is not feasible to date . We repeated the LFI analysis on the P . falciparum PIN with the same set of summaries; importantly , these capture global aspects of PIN data simultaneously . The posterior distribution of θ for P . falciparum is summarized in Table 1 . Notably , the DDa component obtained more weight in the posterior mixture model DDa + PA relative to H . pylori . This suggests , first , that duplication–divergence shapes the global structure of protein networks in a way distinguishable from preferential attachment , and that the difference is also evident when incompleteness of present PIN data is taken into account . Second , gene duplication and interaction divergence might play a larger role in eukaryotic than in prokaryotic protein network evolution , pointing to either discontinuous ( i . e . , likely to be adaptive ) or continuous ( i . e . , unlikely to be adaptive ) taxonomical differences , as already suggested from the extent [42] , the size [43 , 44] , and the complexity [45] of protein families . We found that the lower 80% quantile of 1 − α is larger than 0 . 6 in both investigated species . Genomic and expression data indicate that repeated single gene duplications with immediate subfunctionalization are a driving force in the evolution of higher organisms [10 , 11 , 32 , 46 , 47] . Since , on average , DDa mimics duplication with subfunctionalization ( see also Box 1 ) , our results emphasize the potential importance of single gene duplications with immediate subfunctionalization in the evolution of the eukaryote . Moreover , we prove in Protocol S1 that DDa may describe any protein network topology due to complementary , random interaction divergence . The precise mechanisms of evolution are less clear for the prokaryote; in particular it is unclear whether horizontal gene transfer is adequately modelled with PA [48] , and we caution against interpreting DDa + PA as a model of vertical versus horizontal gene transfer . Nevertheless , the prevalence of duplication divergence in prokaryotic evolution is also indicated from the protein repertoire itself [5 , 49 , 50] . In particular , the phylogenetic distributions of protein families over 41 bacteria are consistent with our findings: 60% of protein families in these prokaryotes can be explained by gene duplications alone [50] . The role of duplication divergence in evolution of protein networks across domains we promote here must be considered within the limits of our model and the data . However , we note that our analysis is based on several global features of the network data , which are more reliable than local aspects ( Figure S4 ) . More importantly , LFI allows us to take the stochasticity of the evolutionary process and the incompleteness of available network data into account . Also , the credible intervals of δDiv and δAtt for the P . falciparum PIN overlap with parameter estimates obtained from sequence data of S . cerevisiae . The study of Wagner [37] indicates a mean divergence probability around 35%–42% and a mean attachment probability around 1%–2% within the first 25 million years after a duplication event in this species . Given the number of limitations in both approaches , further work will be required to combine genomic with network data for a detailed reconstruction of the evolution of complex cellular units . Importantly , fitting a model of network evolution that includes link turnover as a case of neofunctionalization might put our conclusions into perspective . The complexity of PIN data suggests that LFI on biological network data may be highly influenced by the choice of summaries . Table 2 summarizes that for different combinations of four or more summaries , the respective posterior means and 80% credible intervals coincided with those obtained by WR + DIA + CC + + FRAG . Thus , based on many aspects of PINs , inference on θ was consistent . Based on the H . pylori PIN , we found that the approximate posterior ( Equation 4 ) was not identifiable from single summary statistics . Using ND only , it is possible to choose ɛmin small , ɛmin ≤ 0 . 35; but Table 2 shows that the inferred 80% credible interval on θ is very wide . Considering the parameters δDiv , δAtt , and α pairwise , as in Figure 4 , illustrates that ND alone leads to two-dimensional high-density regions that are inconsistent with those obtained by four or more summaries . Similarly , LFI based on several other single summary statistics allowed small threshold values ɛmin , but did not lead to a reliable and consistent estimation of θ ( unpublished data ) . This indicates that many evolutionary histories may explain single aspects of PINs almost perfectly without representing the full topology , reflecting the complex nature of biological network data . Our findings relating to ND are particularly worrisome because the degree sequence is a standard descriptor of protein networks , and often kept fixed when generating randomized networks for a significance analysis on aspects of PIN data [28 , 51 , 52] . PA alone generates tree-like networks , whereas DDa occasionally produces triangles . Surprisingly , LFI with TRIA included in the set of summary statistics did not aid inference in that convergence took longer and fewer samples were accepted without tightening the credible intervals . Taken together with the fact that other motif counts have a similar high variation over the evolutionary history ( unpublished data ) , this suggests that the extreme variability of motif counts in simulated data reduces their usefulness for inference on biological network data . Aspects of the complete , unobserved PINs are easily predicted from the observed networks , once MCMC output is available . Here , we discuss the true network size R , by means of its posterior predictive distribution; as outlined in Materials and Methods . The posterior predictive distribution of R for H . pylori and P . falciparum is displayed in Figure 5 . De Silva et al . [22] proposed a simple estimator of the network size based on the sampling fraction ρ of proteins that are present in the dataset . Applied to H . pylori ( P . falciparum ) , the estimate is R′ = 5 , 636 ( 43 , 835 ) . This is consistent with the posterior predictive distribution obtained by LFI based on WR + DIA + CC + + FRAG in the sense that Pr ( ½ ≤ R/R′ ≤ 2| D℘ ) ≥ 0 . 80 . The fact that current PINs are largely incomplete hampers inference [22 , 53] . Within our Bayesian framework , we compared the effect of different network order and different levels of incompleteness of PIN datasets on protein network inference ( H . pylori: 675 , ρ = 0 . 45; and P . falciparum: 1 , 271 , ρ = 0 . 24 ) . We found large variability associated with predictions of the true network size ( see Figure 5 ) ; notably , the P . falciparum posterior network size was more diffuse than the one of H . pylori . In order to see whether the large variability arises from the approximative nature of LFI , we repeated LFI based on WR + DIA + CC + + FRAG for relaxed choices of ɛmin . Figure 5 shows that tightening the threshold values results in more reliable predictions , and that this effect is negligible when twice as much network data are available . This suggests that aspects of the structure of the true networks remain highly uncertain under the model ( Equation 1 ) when incompleteness is large . Instead , the credible intervals of all evolution parameters θ are tighter for P . falciparum than for H . pylori , even though our model accounts for incompleteness . This indicates that the power of LFI to uncover the evolutionary history of PIN datasets increases with network order irrespective of the levels of incompleteness , essentially because the resolution of the network summaries increases . We further analysed how the degree of incompleteness affects LFI by randomly withholding more network data of the P . falciparum PIN ( ρ = 0 . 17 , 0 . 12 , 0 . 06 ) ; see Materials and Methods for details . Briefly , for PINs with ρ ≥ 0 . 17 , LFI using WR + DIA + CC + + FRAG was possible , and the parameters were distinguishable in terms of the errors between the real and associated simulated summaries . Table 3 summarizes the 80% credible intervals of all parameters for LFI based on WR + DIA + CC + + FRAG for different ρ . As expected , highly increased incompleteness implied larger credible intervals . More importantly , randomly omitting 500 proteins from the available PIN of 1 , 271 proteins did not significantly affect LFI . This is further illustrated with the posterior densities of the mixture parameter α , Figure 6 . PINs from different species have attracted much attention in molecular systems biology . Apart from their suspected role in modulating and underpinning the molecular machinery of complex phenotypes , their evolutionary properties are increasingly being investigated using a range of evolutionary and statistical approaches . We showed that it is possible to draw evolutionary inferences from large-scale , incomplete network data when models of randomly growing graphs are conditioned on many , carefully chosen aspects of the networks . Using a likelihood-free approach that relies on comparing summaries of real network data to simulated PINs , we were able to study more complex models of network evolution at increased confidence than had previously been possible [19] . Our results have important implications for the analysis of protein network topology . Due to its elusive complexity , the topology of a PIN is commonly summarized by the degree sequence [18] , as well as the frequency profile of motif counts [28] . An extensive sensitive analysis showed that the degree sequence has very little power to distinguish among different parameters relative to other statistics ( Figures 1B and S4B ) , and fails to infer the parameters correctly ( Figure 4 ) . We found that the number of triangles is extremely variable over the evolutionary history of simulated PINs ( Figures S1B and S4A ) and did not help inference , suggesting that motif counts are risky descriptors of PINs . Instead , if four or more network summaries are combined , then our method yields ( i ) consistent estimates as well as tight credible intervals on biological data , and ( ii ) accurate estimates on simulated test data where , by definition , the model is correct . The fact that a reliable , consistent analysis requires the combination of several summaries that capture global aspects of the networks , of which WR is computationally very expensive , renders an implementation targeting the S . cerevisiae PIN dataset extremely challenging . We used our computational inference scheme to estimate the potential role of aspects of duplication divergence in different domains from large-scale biological network data of H . pylori and P . falciparum , complementing a number of efforts to uncover the mechanisms that underlie the evolutionary history of complex organisms from sequence data [1–3] , protein structures [4] , or gene families within a wider context [54] . Here , the evolutionary history of PINs was modelled with a mixture of randomly growing graphs that ( i ) agrees in particular with evolution by single gene duplications and immediate divergence , and ( ii ) puts minimal assumptions on the time of evolutionary events , because these are difficult to estimate reliably . Crucially , our approach fully deals with incomplete network data and the stochasticity of the underlying evolutionary process . Inference of the evolutionary parameters improves with an increasing order of the PIN data , irrespective of the levels of incompleteness ( Figure 6 and Table 3 ) . Within the limits of our evolutionary model and the available data , gene duplication and interaction divergence appear to play a dominant , distinguishably larger part in the evolution of the protein network of the eukaryote P . falciparum ( Table 1 ) . Our results emphasize the potential importance of duplication divergence in the evolution of networks across domains . Based on our sensitivity analysis of network summaries , our study suggests , in line with two other recent studies [55 , 56] , that more information could be inferred from combining global aspects of interaction networks than is presently appreciated . The opportunities arising from LFI to computational statistics on complex systems are large . Our results emphasize that choosing a set of appropriate summaries is central to maintaining the approximate character of LFI . We proposed the standardized mean derivative and measures of scaled variation to compare the power of summaries one by one . Although ABC-MCMC failed on network data , algorithm LFI enabled efficient and consistent inference . LFI might prove useful in other biological contexts when prior information is relatively vague , and when the underlying model is complex and highly stochastic .
For clarity of exposition , we first outline algorithm ABC-MCMC [25] and then present algorithm LFI , which achieves approximation ( Equation 4 ) in protein network inference . Let S℘ = {S1 , … , Sk…SK} be the chosen set of summary statistics , and let ɛ > 0 be a threshold value . Let S℘ D℘ , respectively S℘θ , denote the set of summary statistics calculated on the observed network D℘ , respectively a network simulated with parameter θ , and choose some initial parameter value . Then do the following: ABC-MCMC1 If now at θ , propose a move to θ′ according to a proposal density q ( θ → θ′ ) . ABC-MCMC2 Generate a dataset from θ′ and compute S℘θ′ . ABC-MCMC3 Calculate Here , d ( S℘ D℘ , S℘θ′ ) ≤ ɛ denotes that the distance between the kth observed and simulated summary statistics is less than ɛ for all k [23] . The summaries are standardized over the values of the sampled summaries . Different choices of d are possible [24] . Here , 1 denotes the indicator function . ABC-MCMC4 Accept θ′ with probability h and otherwise stay at θ , then return to ABC-MCMC1 . ABC-MCMC is guaranteed to eventually sample from p ( θ|d ( S℘ D℘ , S℘θ ) ≤ ɛ ) , [25] . We now present algorithm LFI . Let ɛt = {ɛ1 , t , … , ɛk , t , … , ɛK , t} be the vector of threshold values at iteration t , one for each summary statistic , and let ɛk , min be the final , preset threshold value for the kth summary statistic after cooling . Similarly , let Σt be the variance of the proposal density at iteration t , and let Σmin be the final , preset variance after cooling . LFI0 If ɛk , t ≥ ɛk , min , update ɛk , t; if Σt ≥ Σmin , update Σt . LFI1 If now at θ , propose a move to θ′ according to a Gaussian density , centred at θ with diagonal covariance matrix Σt and restricted to the interval [0 , 1] , i . e . , qt ( θ→θ′ ) ∝ N ( θ , Σt ) 1[0 , 1] , appropriately normalized . LFI2 During the preset , empirically determined burn-in phase , go to LFI2′ . Else , generate B = 1 PIN according to the mixture model ( Equation 1 ) with parameter θ′ and grown to the number of open reading frames in the genome of the observed PIN . Take a subnetwork of order that equals the order of the observed PIN . Compute the summaries , put sk , θ′ := Sk , θ′ for all Sk ∈ S℘ and go to LFI3 . LFI2′ Perform LFI2 with B = 50 and compute the sample mean S̄k , θ for all Sk ∈ S℘; in the case of ND and WR , compute the pointwise sample mean . Put sk , θ′: = S̄k , θ and go to LFI3 . LFI3 Calculate In our case , the prior is uniform , and p ( θ′ ) /p ( θ ) is one . The distance function dk for the kth summary statistic may depend on k ( see below ) . LFI4 Accept θ′ with probability h and otherwise stay at θ , then return to LFI0 . LFI fulfils the detailed balance equations for the same reasons as [25] , and hence is guaranteed to eventually sample from Tempering scheme . We temper the acceptance threshold ɛt with an exponential cooling scheme , starting at some initial temperature ɛ0 and cooling at the next iteration to ɛt+1 = γɛt , until a minimal temperature ɛmin is reached . In all cases , the minimal temperature is reached in about 750 iterations . Tempering reduces the number of accepted parameters as the number of iterations increases . We employ a similar exponential cooling scheme on Σt , in which the minimal temperature is reached in about 800 iterations . In practice , convergence depends on suitable tempering; we chose ɛmin and γ for all summary statistics , such that the empirical acceptance probabilities were not too low , and such that the Gelman-Rubin ( GR ) statistic was well below 1 . 2 [41] , as further detailed in Protocol S1 . Choice of distance function d . Our distance function in LFI3 is inspired by the Chebyshev distance proposed in [23] ( and outlined in ABC-MCMC3 ) . Notably , since the reliability of PIN summaries differs largely , we combine and do not standardize the summaries; our approach requires K different tempering schemes . For ND and WR , dk is chosen to capture major pointwise differences in the summaries . Given Sk , D℘ and Sk , θ′ ( or S̄k , θ′ ) , we compute the common node degrees ( or distances ) , and for these values , sum the absolute differences of the associated frequencies , cutting off the tails of these distributions . Initial values . One approach to investigate whether a Markov chain has not yet converged is to start multiple chains at overdispersed initial values . We have started four Markov chains at the initial values ( 0 . 9 , 0 , 0 ) , ( 0 . 7 , 0 . 13 , 0 . 23 ) , ( 0 . 5 , 0 . 26 , 0 . 46 ) , and ( 0 . 3 , 0 . 4 , 0 . 7 ) . The first and the latter initial values represent unrealistic models to check that the chains move toward the support of the distribution . The other two initial values interpolate between these two extremes . Within-reach distribution ( WR ) . Given a network D℘ and two connected nodes i and j , consider the shortest path from i to j as their distance d D℘ ( i , j ) . The ( random ) number wrk ( i ) of nodes in distance less than or equal k from i is then wrk ( i ) ≔ #{j|d D℘ ( i , j ) ≤ k} , and the WR is defined as where the normalization constant C is the sum of all node pairs in each component in D℘ . Mean derivative of summary statistics . In order to analyze the information content of summaries for protein networks , we follow the approach recently proposed by K . Heggland and A . Frigessi [39] . Consider one summary statistic S ( θ , G ) , evaluated on simulated data G generated with parameter θ . Heggland and Frigessi argue that “if for fixed θ , the variance in S ( θ , · ) is large compared with the derivative of its expectation , it will be more difficult to detect genuine changes at θ in S ( · , G ) . ” We adopt a variant of their approach , modified to the settings of this paper . Networks Gb , b = 1 , … , 50 , are generated for each value of θ , and the mean statistics are computed ( note that Gb is a different realization of the mixture model ( Equation 1 ) for the same values of θ ) . The parameter θ has L = 3 dimensions , and we integrate over all directional ( absolute ) mean derivatives to obtain a measure of the overall sensitivity to changes in θ: Here , h > 0 and lD is the L-dimensional vector that has h in dimension D and zero otherwise . Since we wish to compare summary statistics , we divide the measure in Equation 7 by the mean of the summary statistic and define the standardized mean derivative: Note that the average cluster coefficient is an observed probability , which is already normalized , and we utilize Equation 7 directly to compute its mean derivative . For the node degree distribution and the WR distribution , we compute the common support of S̄ ( θ + hl ) and S̄ ( θ − hl ) , apply Equation 7 pointwise , and sum these values to give smd ( θ ) . We chose h = 0 . 025 as an approximation to h → 0 , which we regard as sufficiently accurate to delineate differences between summary statistics . Variation of summary statistics . Consider a summary statistic S ( θ , Gb ) evaluated on simulated data Gb generated with parameter θ , and the corresponding mean statistic S̄ ( θ ) as in Equation 6 . We consider the absolute error distribution S ( θ , Gb ) − S̄ ( θ ) , b = 1 , … , 1 , 000 , scaled appropriately: These values yield a relative error histogram for fixed S℘ and θ , and we employed the biweight kernel to estimate the density of standardized variation . In the case of CC , ND , and WR , we normalized as detailed above . Aspects or quantities of PINs can be predicted within the Bayesian framework . The posterior predictive distribution of such a quantity , e . g . , the network size R , may be estimated directly from the MCMC output: where denotes a posterior sample from the set of accepted parameters θ after convergence in the MCMC run . We are left to approximate p ( R | ) by repeatedly generating PINs according to and calculating R , i . e . , We have chosen B = 50 again , and took I = 500 samples from the MCMC output . Out of 1 , 271 proteins in the P . falciparum PIN dataset , we randomly chose subgraphs of order n = 900 , 600 , and 300 to mimic increased incompleteness . For each Markov chain in an LFI simulation , such a subgraph was taken as the observed PIN dataset . Consequently , the four chains within one LFI simulation are fitted to slightly varying observations , making inference harder .
|
The importance of gene duplication to biological evolution has been recognized since the 1930s . For more than a decade , substantial evidence has been collected from genomic sequence data in order to elucidate the importance and the mechanisms of gene duplication; however , most biological characteristics arise from complex interactions between the cell's numerous constituents . Recently , preliminary descriptions of the protein interaction networks have become available for species of different domains . Adapting novel techniques in stochastic simulation , the authors demonstrate that evolutionary inferences can be drawn from large-scale , incomplete network data by fitting a stochastic model of network growth that captures hallmarks of evolution by duplication and divergence . They have also analyzed the effect of summarizing protein networks in different ways , and show that a reliable and consistent analysis requires many aspects of network data to be considered jointly; in contrast to what is commonly done in practice . Their results indicate that duplication and divergence has played a larger role in the network evolution of the eukaryote P . falciparum than in the prokaryote H . pylori , and emphasize at least for the eukaryote the potential importance of subfunctionalization in network evolution .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] |
[
"mathematics",
"plasmodium",
"computational",
"biology",
"evolutionary",
"biology",
"eubacteria"
] |
2007
|
Using Likelihood-Free Inference to Compare Evolutionary Dynamics of the Protein Networks of H. pylori and P. falciparum
|
Invasive infections associated with non-typhoidal Salmonella ( NTS ) serovars Enteritidis ( SE ) , Typhimurium ( STm ) and monophasic variant 1 , 4 , [5] , 12:i:- are a major health problem in infants and young children in sub-Saharan Africa , and currently , there are no approved human NTS vaccines . NTS O-polysaccharides and flagellin proteins are protective antigens in animal models of invasive NTS infection . Conjugates of SE core and O-polysaccharide ( COPS ) chemically linked to SE flagellin have enhanced the anti-COPS immune response and protected mice against fatal challenge with a Malian SE blood isolate . We report herein the development of a STm glycoconjugate vaccine comprised of STm COPS conjugated to the homologous serovar phase 1 flagellin protein ( FliC ) with assessment of the role of COPS O-acetyls for functional immunity . Sun-type COPS conjugates linked through the polysaccharide reducing end to FliC were more immunogenic and protective in mice challenged with a Malian STm blood isolate than multipoint lattice conjugates ( >95% vaccine efficacy [VE] versus 30–43% VE ) . Immunization with de-O-acetylated STm-COPS conjugated to CRM197 provided significant but reduced protection against STm challenge compared to mice immunized with native STm-COPS:CRM197 ( 63–74% VE versus 100% VE ) . Although OPS O-acetyls were highly immunogenic , post-vaccination sera that contained various O-acetyl epitope-specific antibody profiles displayed similar in vitro bactericidal activity when equivalent titers of anti-COPS IgG were assayed . In-silico molecular modeling further indicated that STm OPS forms a single dominant conformation , irrespective of O-acetylation , in which O-acetyls extend outward and are highly solvent exposed . These preclinical results establish important quality attributes for an STm vaccine that could be co-formulated with an SE-COPS:FliC glycoconjugate as a bivalent NTS vaccine for use in sub-Saharan Africa .
Non-typhoidal Salmonella ( NTS ) are important human pathogens worldwide where infection in healthy adults normally results in self-limiting gastroenteritis . They can also cause fulminant invasive disease ( e . g . , bacteremia , septicemia , meningitis ) , particularly in hosts with immunological immaturity , immunosenescence , or immunosuppression [1] . In sub-Saharan Africa , hospital-based surveillance of pediatric patients admitted with fever or suspect focal bacterial infections ( e . g . , meningitis ) revealed invasive NTS ( iNTS ) to be the major pathogen in infants and toddlers following Haemophilus influenzae type b ( Hib ) and Streptococcus pneumoniae ( prior to implementation of Hib and pneumococcal conjugate vaccines ) [2 , 3] . iNTS case fatality rates of 15–30% are typical and most isolates are resistant to multiple antibiotics [1 , 3] . Microbiological analyses of sub-Saharan iNTS isolates revealed that serogroup B serovars S . Typhimurium ( STm ) and 1 , 4 , [5] , 12:i:- ( that share the same phase 1 flagella type ) , and S . Enteritidis ( SE ) , a serogroup D Salmonella , comprise 80–90% of isolates . Genomic sequencing of African STm isolates revealed the emergence and spread of a dominant rare multi-locus sequence type ( MLST ) 313 ( rather than MLST 19 prevalent elsewhere globally ) . Sequencing also demonstrated genomic DNA loss and many pseudogenes including homologs found in S . Typhi or S . Paratyphi A [4 , 5] . The bulk of sub-Saharan Africa iNTS disease occurs in children ≤5 years of age [1–3 , 6] . The putative protective role of antibodies against iNTS disease in young children is supported by the observation that a higher proportion of infant cases occur after the first six months of life; by which time , maternal antibodies have waned [3 , 7] . Given the limited number of serovars associated with disease burden , a vaccine immunoprophylaxis strategy is epidemiologically feasible . The success of field trials with Vi capsular polysaccharide ( CPS ) vaccines in typhoid endemic areas has established the paradigm for subunit vaccines comprised of Salmonella surface molecules to protect against invasive Salmonella infections . The O-polysaccharide ( OPS ) of lipopolysaccharide ( LPS ) and the flagellin subunit protein of flagellar filaments ( H antigen ) constitute prominent Salmonella cell surface structures and bear serovar-specific epitopes . Serogroups A , B and D OPS have a common →2 ) -α-D-Manp- ( 1→4 ) -α-L-Rhap- ( 1→3 ) -α-D-Galp- ( 1→ trisaccharide backbone motif ( antigen O12 ) that can be variably ( 1→6 ) glucosylated at galactose ( antigen O1 ) . An immunodominant dideoxy hexose linked α- ( 1→3 ) to mannose distinguishes these serogroups and is an abequose ( antigen O4 ) for serogroup B . Group B OPS can also undergo O-acetylation at abequose C2 ( generating antigen O5 ) or C2/3 of rhamnose as a consequence of phage lysogeny [8–11] . Studies in animal models have established that Salmonella OPS is an important virulence factor and is a target of protective antibodies for defense against invasive infection [12–16] . Unconjugated Salmonella OPS molecules are poorly immunogenic . However , covalent linkage to proteins improves their immunogenicity and enables development of OPS-based vaccines [17] . We documented that conjugates of SE COPS linked to the homologous serovar phase 1 flagellin protein are immunogenic and protect mice against fatal infection with a Malian SE blood isolate , and that antibodies against NTS flagellin proteins have bactericidal activity [18 , 19] . We report here the development of a glycoconjugate vaccine comprised of STm COPS linked to the phase 1 flagellin protein from the same serovar , with exploration of the role that COPS O-acetyls provide in protective immunity .
The strains used in this study are described ( S1 Table ) . All strains were maintained in Hi-Soy ( HS ) bacteriological media ( 5 g/L sodium chloride , 10 g/L soytone [Teknova , CA] , 5 g/L Hy-yest [Sigma Aldrich , MO] ) at 37°C . Growth and preparation of bacteria for in-vitro analyses and in-vivo infection was conducted as described [18] . Growth media for all guaBA mutants were supplemented with guanine; kanamycin was additionally supplemented for CVD 1925 ( pSEC10-wzzB ) ( S1 Table ) . STm CVD 1925 ( with or without pSEC10-wzzB ) ( S1 Table ) and CVD 1943 ( S1 Table ) were grown in fully chemically defined media ( CDM ) in a fermenter . STm D65 ( S1 Table ) was grown in shake flasks . Fermentation conditions were as follows: 50 mL of CDM supplemented with 0 . 004% guanine was inoculated with 3–5 colonies from an HS agar plate and grown for 12–18 h at 37°C in a shake flask with agitation at 80 rpm . This culture was then used to inoculate 500 mL of CDM supplemented with 0 . 004% guanine that was grown under equivalent conditions for 8–10 h . Four liters of CDM containing 0 . 025% guanine was then inoculated to an OD600 nm of 0 . 15 with the 500 mL shake flask and maintained in a Biostat A-plus fermenter ( Sartorius , Germany ) culture , for 18–24 h at 400 rpm , 5 LPM ambient air , with an adjustment to pH 7 using 28% ammonium hydroxide . STm phase 1 flagellin proteins ( FliC ) were purified as described from culture supernatants of CVD 1925 [20] . Recombinant CRM197 produced in E . coli was obtained from Fina Biosolutions ( Rockville , MD ) . FliC and CRM197 were confirmed for integrity and removal of residual host cell protein by SDS-PAGE with Coomassie Brilliant Blue staining , and endotoxin levels by Limulus Amebocyte Lysate ( LAL ) assay using the Endosafe PTS system ( Charles River Laboratories , MA ) . COPS was harvested directly from LPS in the cell biomass and conditioned growth media of fermentation ( CVD 1925 ( pSEC10-wzzB ) , CVD 1943 ) or shake flask ( STm D65 ) cultures by reducing the culture pH to 3 . 5–3 . 7 with glacial acetic acid , and incubation at 100°C for 4 h in glass bottles submerged in a boiling water bath . Post-hydrolysis supernatants were separated from insoluble material by centrifugation at 10k x g / 4°C for 30 min using a GS3 Rotor in a Sorvall RC5 refrigerated centrifuge . The supernatant fraction was brought to 1 M NaCl and filtered by tangential flow microfiltration through a 0 . 2 μm hollow-fiber filter ( GE , NJ ) at 4 . 5 psi transmembrane pressure ( TMP ) passing the full volume through followed by flushing with an equivalent volume of 1 M NaCl . The 0 . 2 μm cleared 1 M NaCl permeate was then concentrated 10-fold on a 30 kDa Hydrosart TFF membrane ( Sartorius , Germany ) at 14 psi TMP and diafiltered against 35 diavolumes of 1 M NaCl , followed by 10 diavolumes of 50 mM Tris pH 7 . The retentate fraction in 50 mM Tris pH 7 was then passed through 3 x 3 mL Sartobind NanoQ anion exchange membranes ( Sartorius , Germany ) linked in series using an AKTA Purifier ( GE , NJ ) at 10 mL/min in 50 mM Tris pH 7 . The flow-through fraction was brought to 25% ammonium sulfate and incubated overnight at 4°C . Precipitated material was removed by centrifugation at 10k x g / 4°C for 30 min using a GS3 rotor in a Sorvall RC5 refrigerated centrifuge followed by filtration through a 0 . 2 μm Stericup vacuum filter unit ( Millipore , MA ) . Filtrates were then concentrated 10-fold by TFF with a Slice 200 TFF device using a 10 kDa Hydrosart membrane ( Sartorius , Germany ) at 7 . 5 psi TMP , and diafiltered against 10 diavolumes of de-ionized water . TFF retentates were lyophilized and stored at -20°C until use . For assessment of residual polysaccharide O-acetyls after exposure to different pH levels , 1925wzzB-COPS was incubated in 50 mM HEPES pH 7 , 50 mM HEPES pH 8 , 50 mM sodium borate pH 9 , or 50 mM sodium borate pH 10 for 2 days at room temperature ( RT ) . Complete de-O-acetylation was accomplished by incubation at pH 12 for 3 h at 37°C with pH adjustment and maintenance with 0 . 1M sodium hydroxide . Preparation of de-O-acetylated ( dOAc ) 1925wzzB-COPS for use as antigen in ELISA and vaccine preparation was accomplished by incubation in 50 mM sodium borate pH 10 for 2 days at RT . High performance liquid size-exclusion chromatography ( HPLC-SEC ) analyses were performed with a Biosep SEC4000 column ( Phenomenex , CA ) on an Alliance 2795 ( Waters , MA ) run at 1 mL/minute with PBS pH 7 . 4 . Absorbance at 280 nm and 252 nm were monitored with a 2487 dual-UV detector ( Waters , MA ) and refractive index with a 2414 refractive index detector ( Waters , MA ) . Monosaccharide composition analyses were accomplished by depolymerization of purified polysaccharides in 1 M trifluoroacetic acid for 4 h at 100°C , followed by lyophilization , reconstitution in deionized water and filtration through a 0 . 2 μm syringe filter . Depolymerized samples were analyzed by high performance anion-exchange chromatography coupled with pulsed amperometric detection ( HPAEC-PAD ) using a CarboPac PA10 column run on a Dionex ICS4000 ( Thermo Scientific , MA ) at 0 . 010 mL/minute in 18 mM KOH and were compared to commercially available purified monosaccharide standards ( Sigma Aldrich , MO ) prepared under similar conditions . Analyses for O-acetylation were conducted by the method of Hestrin as described , with acetylcholine chloride standards ( Sigma Aldrich , MO ) [21] . Protein levels were assessed by bicinchoninic acid assay ( Thermo-Pierce , MA ) per the manufacturer’s instructions using purified bovine serum albumin ( Sigma Aldrich , MO ) as standards . Endotoxin levels were measured ( LAL assay ) , and removal of nucleic acid was confirmed by absorbance at 260 nm . Size exclusion chromatography with multi angle light scattering ( SEC-MALS ) to determine the absolute molecular weight for purified COPS was performed using an Agilent 1100 HPLC system with an 8-angle Heleos detector and a Optilab T-rEX refractive index detector ( Wyatt Technologies , CA ) . Fractionation was performed using TSKgel G4000 and 5000PWxl ( Tosoh Biosciences , OH ) in series with PBS + 0 . 02% sodium azide as the buffer at a flow rate of 0 . 5 mL/min . Analysis was performed using Astra 6 . 2 software ( Wyatt Technologies , CA ) . The differential refractive index ( dn/dC ) necessary for the calculations was experimentally determined from purified , lyophilized COPS solubilized in the equilibration buffer , according to the protocol provided by the manufacturer . Resorcinol assays for carbohydrate concentration were conducted as described [18] . Polysaccharides for NMR analyses were prepared by lyophilization and reconstitution in D2O . NMR spectra were recorded at 25°C on an 800 MHz ( 800 . 27 MHz for protons ) Bruker Avance-series NMR spectrometer equipped with four frequency channels and a 5 mm triple-resonance z-axis gradient cryogenic probehead . A one-second relaxation delay was used , and quadrature detection in the indirect dimensions was obtained with states-TPPI phase cycling; initial delays in the indirect dimensions were set to give zero- and first-order phase corrections of 90° and –180° , respectively . Data were processed using the processing program nmrPipe on Mac OS X workstations . The 1H , 13C HSQC experiment was collected to monitor changes in the 13C and 1H resonances for O-acetylated and de-O-acetylated polysaccharides . Overnight bacterial cultures were adjusted to an OD600 of 1 . 0 and then 2 mL of culture was centrifuged at maximum speed for 2 min at 4°C . The supernatant was removed and the pellet resuspended in 100 μL lysis buffer ( 0 . 1 M Tris-HCl , pH 6 . 8 , 2% SDS , 10% Glycerol , 4% 2-mercaptoethanol ) . The sample was boiled at 95–100°C for 10 min to lyse the cells . Proteins were digested by adding 25 μg Proteinase K . The sample was incubated at 60°C for 1 h . The sample was boiled for 10 min and then allowed to cool on ice . 20 μl of the sample was electrophoresed on 4–15% Mini Protean TGX stain-free gels ( BioRad Laboratories , CA ) with the CandyCane Glycoprotein ladder ( Life Technologies , CA ) . LPS was stained using Pro-Q Emerald 300 LPS Gel Stain ( Life Technologies , CA ) as per the manufacturer’s instructions . Crude LPS extracts were made from several Malian isolates of STm ( D65 , A13 , D23580 , P142 , Q65 and S42 ) ( S1 Table ) as well as SE R11 ( S1 Table ) . Overnight bacterial cultures were normalized to an OD600 of 0 . 2 , from which a 1 mL aliquot was centrifuged at 4°C for 10 min at 13 , 200 rpm . Pellets were resuspended in 100 μL of lysis buffer , vortexed vigorously and heated at 100°C for 10 min . Samples were cooled on ice followed by the addition of 25 μg of proteinase K . Samples were then incubated at 60°C for 1 h followed by 100°C for 10 min . LPS preparations were diluted with 100 μL of 2 x Laemmli sample buffer ( Bio-Rad , Hercules , CA ) . Two volumes ( 10 μL and 2 μL ) from each LPS sample were separated by electrophoresis on neutral pH , 1 . 5 mm , 4–12% Bis-Tris gels ( Life Technologies , CA ) . The gels were wet transferred overnight at 4°C to methanol-activated polyvinylidene difluoride ( PVDF ) membranes and subsequently blocked with PBS + 0 . 05% Tween-20 pH 7 . 4 ( PBST ) + 10% Omniblok ( AmericanBio , MA ) . To detect LPS , the membranes were then incubated with monoclonal IgA to STm O5 [clone Sal4 , kind gift from Dr . Nicholas Mantis , Wadsworth Institute , NY] ( 1:1 , 000 ) or monoclonal IgG to Salmonella core polysaccharide ( sc-52219 , Santa Cruz Biotechnology , CA ) ( 1:50 ) diluted in PBST + 10% Omniblok and incubated for 1 h at room temperature . Membranes were washed with PBST and incubated with either CruzFluor ( CFL ) -488-labeled anti-mouse IgA ( for O5; 1:100 ) or CFL-647-labeled anti-mouse IgG ( for core; 1:100 ) diluted in PBST + 10% Omniblok and incubated for 1 h at room temperature . All CFL-labeled antibodies were purchased from Santa Cruz Biotechnology , CA . Membranes were again washed and visualized using the Chemi-Doc MP system ( Bio-Rad , CA ) . Eight to 10 week old female CD1 mice ( Charles River Laboratories , MA ) were injected intramuscularly ( IM ) in the right gastrocnemius at 0 , 28 and 56 days with either sterile PBS ( pH 7 . 4 ) , 2 . 5 μg of STm FliC , or 2 . 5 μg polysaccharide conjugated to either STm FliC or CRM197 . Sera were obtained before vaccination and three weeks after the final immunization . Four weeks after the final immunization ( day 84 ) , immunized mice were challenged intraperitoneally ( IP ) with STm D65 ( LD50 = ~2 x 104 colony forming units [CFU] ) and monitored daily for 14 days after challenge while recording overall health , weight loss , and mortality . Mice that reached a moribund state ( lethargy , non-responsiveness , dehydration , piloerection , and/or 48 h of sustained ≥20% weight loss ) were euthanized and recorded as dead . Vaccine efficacy ( VE ) was calculated as [ ( proportional mortality in controls ) - ( proportional mortality in vaccine group ) ]/ ( proportional mortality in controls ) . Sera from mice immunized with CVD 1931 have been previously described [23] . Group sizes for challenge experiments were determined by the minimal number of mice required to provide ≥90% power to detect a significant difference for mortality rates in controls and vaccinees of ≥90% and ≤30% respectively ( one-sided Fisher’s exact test , α = 0 . 025 ) . Immunization , protection studies , and data analysis were done by two investigators blinded to the group allocations . All animal studies were performed in facilities that are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care and were in compliance with guidelines for animal care established by the US Department of Agriculture Animal Welfare Act , US Public Health Service policies , and US federal law . All animal experiments were in compliance with study protocols ( 0715008 and 0812010 ) approved by the University of Maryland School of Medicine Institutional Animal Care and Use Committee . Bacterial clinical isolates data had been de-identified and were analyzed anonymously . Titration of serum IgG from STm vaccine-immunized mice , monoclonal antibodies ( mAbs ) ( anti-S . Typhimurium O4 IgG [SC5223 , Santa Cruz Biotechnology , CA] ) and anti-S . Typhimurium O5 IgA ( Sal4 ) , or polyclonal anti-SE COPS sera ( described previously [18] ) was accomplished using an enzyme-linked immunosorbent assay ( ELISA ) . Briefly , 96-well , medium-binding , microtiter plates ( Greiner Bio-One , NC ) were coated with either COPS antigens ( 1925wzzB-COPS , dOAc-1925wzzB-COPS , or SE COPS ) or COPS conjugates ( STm-COPSKDO:CRM197 , or dOAc-STm-COPSKDO:CRM197 ) at a concentration of 5 μg polysaccharide/mL and incubated overnight at 4°C . Plates were washed with PBST and blocked with PBS + 10% Omniblok non-fat , dry milk for 2 h at 37°C . Serum samples and monoclonal antibodies were serially diluted in PBST + 10% Omniblok , transferred to blocked ELISA plates , and incubated for 1 h at 37°C . Plates were washed , and incubated for 1 h at 37°C with horseradish peroxidase ( HRP ) -labeled anti-mouse IgG ( for O4 and mouse serum; 1:1 , 000 ) ( KPL , MD ) or HRP-labeled anti-mouse IgA ( for O5; 1:500 ) ( Southern Biotech , AL ) . After washing , substrate ( 3 , 3’ , 5 , 5’-tetramethylbenzidine , KPL , MD ) was added , and the plates were incubated on a rocker at ambient temperature for 15 min in darkness . The reaction was stopped with the addition of 1 M H3PO4 , and the absorbance at 450 nm was recorded using an Ascent microplate reader ( Thermo Scientific , MA ) . Endpoint titers , represented as ELISA units ( EU ) per mL , were defined and calculated as previously described [18] . Analyses were conducted with positive sera as plate controls with acceptance criteria of < 15% variance between plates . Proportional levels of epitope-specific antibody in sera were calculated by subtracting the serum IgG titers for various COPS antigens as follows: O1 , 12 , core antibody levels were defined as the SE COPS titer; O4 antibody levels were calculated as ( dOAc-1925wzzB-COPS titer ) – ( SE COPS titer ) ; O-acetyl specific antibody levels were calculated by ( 1925wzzB-COPS titer ) – ( dOAc-1925wzzB-COPS titer ) . Any negative values were assigned a titer of 0 . The relative levels of each antibody population were calculated as a percentage of the sum of all three calculated titers . Assays were conducted as described [24] . Briefly , the assay was prepared by first combining 25 μl of baby rabbit complement ( BRC ) ( Pel-Freez Biologicals , AR ) , 15 μl of saline , and 50 μl of antibody sample and incubating with 10 μl of diluted bacteria ( 100–350 CFU ) . Negative controls contained the respective bacteria and complement only . Individual mouse sera were heat-inactivated at 56°C for 20 min prior to use in the assay . Antibody samples were adjusted prior to addition such that each sample contained an equivalent number of total anti-1925wzzB-COPS IgG EU in the 100 μl assay volume . Viable bacteria were determined after plating on rich media agar . Assays were conducted as described [25] . HL-60 cells were purchased from the American Type Culture Collection and were maintained in 1x RPMI-1640 medium supplemented with 10% [v/v] heat-inactivated fetal bovine serum prior to use . Briefly , 10 μl of bacterial suspension ( ~700–1000 CFU ) was added to 25 μl of antibody in each well for opsonization at 37°C for 15 min in a 5% CO2 incubator , at which point 25 μl of BRC and 4 x 105 HL-60 cells in 40 μl of media were added to each well . The 96 well plate was incubated at 37°C ( no CO2 ) with shaking agitation at 160 rpm for 45 mins . Antibody samples were adjusted prior to addition such that each sample contained an equivalent number of total anti-1925wzzB-COPS IgG EU in the 100 μl assay volume . Negative controls were performed with buffer in place of sera . Viable bacteria were determined after plating on rich media agar . All statistical analyses were performed using GraphPad Prism v6 ( GraphPad Software , CA ) . No animals were excluded from analysis . For ELISA analyses , the majority of data did not meet the criteria for Gaussian distribution . Therefore , all comparisons between groups were conducted using non-parametric tests: either a two-tailed Mann-Whitney U test for unpaired samples or two-tailed Wilcoxon signed rank test for paired samples ( α = 0 . 05 for both tests ) . No adjustment was made for multiple comparisons . Survival analysis after active immunization was assessed by the log-rank test . Comparisons of either SBA or OPA fitted curves were made using a nonlinear regression analysis ( extra sum-of-squares F test , α = 0 . 05 ) . P values of ≤ 0 . 05 were considered statistically significant .
STm COPS for use as a vaccine antigen was purified from CVD 1925 ( pSEC10-wzzB ) , a recombinant strain engineered to express long-chain OPS due to overexpression of wzzB , a member of the polysaccharide co-polymerase family ( Fig 1A ) [36 , 37] . CVD 1925 ( pSEC10-wzzB ) COPS ( 1925wzzB-COPS ) demonstrated a single , sharply-defined population determined to be ~19 . 8 kDa by SEC-MALS ( Fig 1B ) . HPAEC-PAD analysis confirmed glucosylation at ~12% of OPS repeats based on the ratio with rhamnose ( S4 Table ) . NMR analyses indicated variable O-acetylation at abequose C2 and C2/3 of rhamnose ( Fig 2 and Table 1 ) [38] . 1925wzzB-COPS was recognized by monoclonal antibodies specific for the O4 and O5 antigens as well as polyclonal sera recognizing O1 , 12 , and core epitopes ( derived from mice immunized with an SE-COPS:FliC conjugate ) ( Fig 1E ) . A comparable pattern of glucosylation ( S4 Table ) and O-acetylation ( Fig 2 and Table 1 ) was found for the OPS of STm strain D65 , a previously described Malian ST313 blood isolate used herein for challenge studies [23] . D65 COPS demonstrated a bimodal size distribution with a population equivalent in size to 1925wzzB-COPS , as well as a higher molecular weight species ( Fig 1B ) . Polysaccharide O-acetyl groups are stable at neutral pH but labile under alkaline conditions . To assess the site-specific susceptibility of O-acetyls to base treatment , residual O-acetylation in 1925wzzB-COPS was assessed after exposure to different pH conditions . O-acetylation was maintained at pH 7–8 but lost at approximately equivalent levels from abequose and rhamnose at ≥pH 9 ( Figs 1C , 1D and 2 and Table 1 ) . ELISA analysis confirmed marked O5 loss at pH 10 , while maintaining O1 , 4 , 12 and core antigenicity ( Fig 1E ) . Characterization of the immune response to bacterial surface polysaccharides after exposure to the whole organism provides important information for development of subunit vaccines based on these antigens . Alkaline treatment of COPS is also effective for analyzing antibody responses to polysaccharide O-acetyls , as other attributes such as size and backbone structure remain unchanged [10 , 39] . Accordingly , we measured antibody titers against native and pH-10 , de-O-acetylated ( dOAc- ) 1925wzzB-COPS as well as SE COPS in sera from mice immunized with CVD 1931 , an attenuated vaccine strain derived from STm strain D65 that mediates robust protection against wild-type D65 challenge [23] . In these sera , anti-COPS IgG recognized primarily STm-specific O-epitopes ( 1925wzzB-COPS vs . SE COPS , P < 0 . 05 ) , and IgG titers were generally higher for native compared to dOAc-1925wzzB-COPS ( P = 0 . 09 , Fig 3A ) . These data suggest that antibodies to O-acetylated epitopes are induced under conditions where protection is achieved . To confirm that COPS from strain D65 was representative of other circulating STm strains in Mali , we assessed LPS from isolates obtained in different years from l’Hôpital Gabriel Touŕe in Bamako , Mali . Western blots performed with an anti-O5 mAb revealed comparable LPS banding patterns and intensities for all STm strains analyzed , including Malawian STm D23580 ( Fig 3B ) . Conjugates of STm COPS and FliC were generated by different conjugation methods . We initially produced lattice-type conjugates ( S4A Fig ) by multipoint conjugation between random 1925wzzB-COPS hydroxyls and amino groups on ADH-derivatized FliC using CDAP ( STm-COPSLat:FliC ) [40] . Maximal linkage between polysaccharide and protein by CDAP requires pH 9–10 conditions during conjugation [40] . Accordingly , while efficient formation of high molecular weight conjugates was observed ( S4B Fig ) , we found marked loss of polysaccharide O-acetyls after conjugation ( Table 2 ) . Mice immunized with this conjugate generated higher geometric mean titers ( GMTs ) of anti-1925wzzB-COPS IgG compared to controls administered PBS; however , the GMTs were low , and no difference was seen when the sera were screened against dOAc-1925wzzB-COPS ( Fig 4A ) . Challenge with 1x105 or 5x105 CFU of STm D65 produced 70% and 100% mortality , respectively , in PBS controls , whereas immunization with the STm-COPSLat:FliC conjugate provided 43% and 30% protection against mortality following these challenge doses ( Fig 4D ) . To produce a conjugate formulation that retained OPS O-acetyls , sun-type conjugates ( S4A Fig ) were generated by functionalization of the KDO carbonyl at the reducing end of 1925wzzB-COPS with an aminooxy thiol reagent . This yielded a free thiol that was then coupled to maleimide-derivatized protein lysines ( STm-COPSKDO:FliC ) . This approach allowed the entire conjugation to ensue at neutral pH . Conjugates generated by this method maintained O-acetylation levels comparable with the native polysaccharide ( Table 2 ) . Mice immunized with STm-COPSKDO:FliC manifested robust anti-1925wzzB-COPS IgG titers with a GMT ~1 , 000-fold higher than was seen for the lattice conjugate ( Fig 4B ) . Importantly , we found that the GMT of antibody to native 1925wzzB-COPS was ~10-fold higher than was seen for antibodies directed against dOAc-1925wzzB-COPS and thus was similar to the profile found for sera from CVD 1931-immunized mice ( Fig 3A ) . Immunization with this conjugate also induced high anti-FliC titers in all mice comparable to titers achieved after immunization with unconjugated FliC ( Fig 4C ) . Challenge with both low ( 1x105 CFU ) and high ( 5x105 CFU ) doses of STm D65 was sufficient to cause >90% mortality in unimmunized controls , with mice given the higher dose succumbing more rapidly ( Fig 4E ) . Alternatively , mice immunized with STm-COPSKDO:FliC were protected against fatal infection at both of these challenge doses ( 95–100% VE ) . To determine the functional relevance of O-acetyl groups in the context of equivalent conjugate architecture and the absence of other STm antigens , we assessed the immunogenicity and protection imparted by native and dOAc-1925wzzB-COPS thioether-linked sun-type conjugates with CRM197 ( [dOAc-]STm-COPSKDO:CRM197 ) . Antigenicity analyses of the conjugated COPS confirmed retention of the O4 antigen in both formulations but marked loss of O5 in the dOAc-1925wzzB-COPS conjugate ( S5 Fig ) . Sera from mice immunized with STm-COPSKDO:CRM197 demonstrated high titers of anti-COPS IgG against the native polysaccharide ( Fig 5A ) that were similar to the GMT achieved by sun-type conjugates of native COPS with FliC ( Fig 4B ) . The IgG GMT for these sera was ~10-fold lower when assessed with dOAc-1925wzzB-COPS and ~1 , 000-fold lower for SE COPS , which shares Salmonella O-antigens 1 and 12 and the core polysaccharide , indicating that the anti-COPS IgG immune response was largely STm-specific with a bias towards O-acetylated epitopes . Mice immunized with dOAc-STm-COPSKDO:CRM197 demonstrated a broader range of serum IgG titers against 1925wzzB-COPS with a GMT ~10-fold lower than that achieved with the O-acetylated STm-COPSKDO:CRM197 conjugate . The IgG GMT of the dOAc-STm-COPSKDO:CRM197 sera was comparable for both native and dOAc-1925wzzB-COPS . A greater proportion of these sera demonstrated antibody titers against SE COPS than was found for sera from mice immunized with STm-COPSKDO:CRM197 . The proportional level of epitope-specific IgG in the sera of individual mice was generally consistent with the trend seen for the overall group GMT; however , some exceptions were evident ( S6 Fig ) . To assess the efficacy of these two vaccine preparations , conjugate-immunized mice and PBS controls were challenged in equal proportions with 1x106 or 5x106 CFU of STm D65 , resulting in >95% mortality in PBS controls; mice receiving the higher dose succumbed more rapidly ( Fig 5B ) . Mice immunized with the STm-COPSKDO:CRM197 conjugate were 100% protected against fatal infection with either challenge dose . Mice immunized with dOAc-STm-COPSKDO:CRM197 were significantly protected against challenge but at a lower level than mice immunized with STm-COPSKDO:CRM197 ( 63–74% VE versus 100% VE , P < 0 . 05 ) . The basis for the differential protective efficacy achieved by the CRM197 conjugates synthesized with native- and dOAc-1925wzzB-COPS was unclear since differences were found in both total anti-COPS IgG GMTs and O-epitope specificity patterns . Thus , analyses were undertaken to distinguish whether a qualitative difference ( e . g . , specificity , avidity ) in the anti-COPS immune response or rather a quantitative difference in the anti-COPS IgG titer accounted for the higher protection achieved with the STm-COPSKDO:CRM197 conjugate . Serum bactericidal antibody ( SBA ) and opsonophagocytic antibody ( OPA ) assays were performed using equivalent anti-1925wzzB-COPS IgG ELISA units ( EU ) . For these analyses , we selected sera that demonstrated different O-antigen specificity profiles ( Fig 5C ) . Sera that contained antibodies predominantly against O-acetyl epitopes were identified by high reactivity with 1925wzzB-COPS but low reactivity with dOAc-1925wzzB-COPS . Conversely , sera manifesting antibody titers that were equivalent for native- and dOAc-1925wzzB-COPS but negligible for SE COPS were presumed to be directed primarily against O4 . Remarkably , bactericidal activity among the different sera was indistinguishable when the anti-COPS IgG EU were equivalent ( Fig 5D and 5E ) . Additionally , comparison of the anti-1925wzzB-COPS IgG titers induced by the two COPSKDO:CRM197 vaccines for mice that survived infection at the more restrictive higher challenge dose versus those that succumbed to infection , indicated that protection against challenge correlated with higher anti-polysaccharide antibody levels ( S7 Fig ) . Due to their multiple glycosidic linkages , polysaccharides are flexible and have the potential to assume different conformations . The conformational properties of STm OPS could possibly become altered upon acetylation , producing epitopes unique to O-acetylated polysaccharide repeats , and thereby influencing the pattern of antibody induction . To address this possibility , we conducted in silico-enhanced sampling molecular dynamics ( MD ) simulations to determine the accessible conformations of STm OPS and their relative frequencies upon O-acetylation . For these analyses , we modeled a base O:4 , 12 3-repeat molecule ( Fig 6A ) as well as variants that were either O-acetylated on rhamnose and abequose , glucosylated at the central or terminal repeat unit , or a combination thereof . Although slight differences were evident , we found that the all of these STm polysaccharides maintained the same dominant conformation ( “11111111” ) for at least 90% of the in silico simulations ( Table 3 ) . The modeled polysaccharides also demonstrated comparable patterns of total occupied 3D space based on the cumulative sum and high degree of overlap between the conformational volumes for the individual monosaccharide residues in the polysaccharide chain ( S5 and S6 Tables ) . Representative native and O-acetylated saccharides in the dominant conformation “11111111” ( Fig 6B and 6C ) underwent 3D rendering , which displays the total volume occupied by each of these polysaccharide structures ( Fig 6D and 6E ) . Examination of the abequose and adjacent OPS repeat rhamnose O-acetyls in the “11111111” conformation indicates that they are highly solvent accessible ( S7 Table ) and are in close spatial proximity ( Fig 6F and 6G ) .
The native COPS purified from our STm reagent strain was incompletely O-acetylated at abequose and rhamnose . Analyses of the OPS from Malawian STm strain D23580 and STm clinical isolates from Kenya found they were also incompletely O-acetylated at both abequose and rhamnose [11 , 38] . Variability in OPS O-acetylation patterns among STm clinical isolates may have implications , however , for selection of a representative OPS hapten . Antibodies against O-acetyls may not bind efficiently to the underlying monosaccharide structure and binding by OPS backbone-specific antibodies may be reduced in the context of O-acetylation [41] . The close spatial proximity of the abequose and rhamnose O-acetyls also raises the possibility of unique epitopes formed by their combination . Sun-type glycoconjugates with 1925wzzB-COPS induced antibodies against both O-acetyl and OPS backbone epitopes and imparted high levels of protection against fatal challenge in mice . O-acetyls are important immune determinants for several bacterial polysaccharide vaccines including S . Paratyphi A COPS-based glycoconjugates wherein the rhamnose O-acetyls in the trisaccharide backbone were found to be essential for induction of bactericidal antibodies [10 , 42–44] . However , their importance to STm immunity was heretofore unclear . Protection by passive transfer of anti-O4 IgG mAbs in mice has been reported for both an O:4 , 5 , 12 STm as well as strain D23580 that is variably O-acetylated on both rhamnose and abequose [13 , 45] . The functional activity of anti-O5 IgG mAbs is not well established; however , a monoclonal anti-O5 IgA protected mice against oral but not intraperitoneal STm challenge [13 , 46] . We found that the functional bactericidal activity of anti-O-acetyl antibodies may not differ greatly from antibodies recognizing other STm OPS epitopes . This notion is in agreement with a recent report that found D23580 COPS conjugated to CRM197 induced antibodies that bound the homologous OPS more efficiently than COPS lacking O-acetyls or acetylated only at abequose , but maintained bactericidal activity against STm isolates that expressed all these OPS forms [47] . Protection may thus depend upon reaching a critical titer of antibody that is sufficient to bind the infecting isolate OPS , irrespective of epitope specificity . This is supported by our observation that higher anti-OPS titers correlated with survival in mice immunized with the CRM197-based COPS conjugates , wherein the COPS molecule was the sole protective antigen and the two vaccine constructs differed only in the presence or absence of polysaccharide O-acetyls . It is therefore reasonable that a partially O-acetylated polysaccharide would constitute the preferred polysaccharide vaccine antigen , as it could induce immunity against a wide range of structures , and thereby ensure broad coverage . We also found that conjugate architecture contributed to immunogenicity . The lattice-type conjugate of 1925wzzB-COPS and FliC synthesized with an ADH linker ( STm-COPSLat:FliC ) was less immunogenic and imparted lower protection than the sun-type conjugate generated with thiol-aminooxy and GMBS linkers ( STm-COPSKDO:FliC ) . Although the lattice-type conjugate suffered from loss of OPS O-acetyls , the sun-type dOAc-STm-COPSKDO:CRM197 , which similarly lacked O-acetyls , was immunogenic and protective . Differences in protein carrier function can influence the immunogenicity of glycoconjugates [48]; however , this was likely not a contributing factor here as sun-type conjugates of 1925wzzB-COPS ( either FliC or CRM197 ) induced equally robust anti-COPS IgG levels . It is conceivable that the ADH linker may have exerted epitopic dominance , as can occur when the hapten is small and the linker highly immunogenic [49 , 50] . The superior immunogenicity of conjugates generated by linkage through the polysaccharide reducing end is more likely accounted for , however , by minimal perturbation of OPS epitopes and the formation of a structure better able to cross-link B-cell receptors . Our finding that STm OPS O-acetyls are immunodominant epitopes is in agreement with a prior report which found that mice immunized with an O:1 , 4 , 5 , 12 STm isolate displayed higher anti-LPS antibody titers compared with an isogenic derivative lacking abequose O-acetylation [51] . We also found that there was marked heterogeneity in the humoral response to conjugates generated with de-O-acetylated COPS compared to immunization with conjugates generated with the native O-acetylated polysaccharide . As other antigenic determinants were not altered upon de-O-acetylation , it is presumed that the disparity is due to loss of the O-acetyl groups . The high variance in anti-COPS antibody titers for the de-O-acetylated COPS conjugates is similar to our prior findings in mice immunized with SE COPS and FliC conjugates [18] . It was suggested that the differential immunogenicity of O-acetylated STm OPS may be accounted for by unique conformational epitopes [41 , 51] . The in silico analyses conducted herein assessed preferred polysaccharide conformations defined by changes in the orientation around the glycosidic linkages [32] . This method also models the total 3D space occupied by all polysaccharide conformations , thus approximating the relative antigenic shape seen by the immune system . Our analyses suggest no alteration of STm OPS conformational properties upon O-acetylation . They further indicate a single dominant conformation in which the O-acetyls extend outward from the polysaccharide backbone and are highly solvent exposed . It is conceivable that the physicochemical properties of O-acetyls ( e . g . , size , partial charge , hydrophobicity ) may provide a putative stabilizing interaction between the polysaccharide and the B-cell receptor and thus account for their pronounced immunogenicity . Further structural analyses will be needed to confirm these findings and assess the biophysical interactions between polysaccharide O-acetyls and antibodies . We have reported previously that antibodies directed against NTS flagellin proteins mediated functional bactericidal activity in-vitro , and that passive transfer of a monoclonal antibody specific for STm FliC could protect mice against fatal infection with D65 [19] . By using flagellin as the carrier for STm COPS , protective immunity could thus be provided by both hapten and carrier components . The immunologic memory induced by glycoconjugate vaccines includes T cell responses to the carrier protein . It is thus conceivable that the anti-polysaccharide immune response in vaccinated individuals may thus be additionally boosted after natural S . Typhimurium infection due to T helper memory responses against flagellin peptides . Development of a vaccine to control iNTS infections in sub-Saharan Africa remains an important global health priority . Further studies will also need to address a bivalent vaccine formulation of the sun-type STm COPSKDO:FliC with a comparably optimized SE COPS:FliC glycoconjugate . We elected to use FliC as the carrier protein to ensure coverage against serovar 1 , 4 , [5] , 12:i:- that lacks phase 2 flagella expression [3] . It will also be important to determine whether cross-protection can be achieved against other sub-Saharan African group B serovars ( e . g . , S . Stanleyville ) that have the same O type , but different H type , and constitute a low but nevertheless significant proportion of pediatric iNTS disease [3] .
|
In sub-Saharan Africa , invasive non-typhoidal Salmonella ( NTS ) infections with serovars Enteritidis ( SE ) and Typhimurium ( STm ) are widespread in children , where up to 30% of cases are fatal . There are several licensed typhoid vaccines but no NTS vaccines . We previously reported that conjugates of SE lipopolysaccharide-derived core and O-polysaccharide ( COPS ) and the homologous serovar flagellin protein were effective vaccines against fatal infection in mice with a Malian SE invasive clinical isolate . We document herein the development of a promising STm vaccine candidate consisting of STm COPS conjugated to STm phase 1 flagellin ( FliC ) . STm-COPS:FliC glycoconjugates induced robust immune responses and protected mice against fatal challenge with a Malian STm blood isolate . We additionally found that while COPS O-acetyl groups were strongly immunogenic , antibodies against non-O-acetyl polysaccharide backbone epitopes could also mediate bactericidal activity . Computational modeling further indicated that STm COPS assumes a single dominant conformation in which the O-acetyl groups are highly solvent exposed and thus available targets for the immune system . These results pave the way toward clinical studies of a bivalent NTS vaccine formulation .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"bacteriology",
"medicine",
"and",
"health",
"sciences",
"immune",
"physiology",
"enzyme-linked",
"immunoassays",
"pathology",
"and",
"laboratory",
"medicine",
"pathogens",
"immunology",
"microbiology",
"vaccines",
"bacterial",
"diseases",
"enterobacteriaceae",
"infectious",
"disease",
"control",
"antibodies",
"immunologic",
"techniques",
"bacteria",
"bacterial",
"pathogens",
"research",
"and",
"analysis",
"methods",
"monoclonal",
"antibodies",
"immune",
"system",
"proteins",
"infectious",
"diseases",
"microbial",
"physiology",
"proteins",
"medical",
"microbiology",
"antigens",
"immunoassays",
"microbial",
"pathogens",
"salmonella",
"biochemistry",
"bacterial",
"physiology",
"polysaccharides",
"physiology",
"flagellin",
"biology",
"and",
"life",
"sciences",
"glycobiology",
"organisms"
] |
2017
|
Development of a glycoconjugate vaccine to prevent invasive Salmonella Typhimurium infections in sub-Saharan Africa
|
Neural circuits in the medial entorhinal cortex ( MEC ) encode an animal’s position and orientation in space . Within the MEC spatial representations , including grid and directional firing fields , have a laminar and dorsoventral organization that corresponds to a similar topography of neuronal connectivity and cellular properties . Yet , in part due to the challenges of integrating anatomical data at the resolution of cortical layers and borders , we know little about the molecular components underlying this organization . To address this we develop a new computational pipeline for high-throughput analysis and comparison of in situ hybridization ( ISH ) images at laminar resolution . We apply this pipeline to ISH data for over 16 , 000 genes in the Allen Brain Atlas and validate our analysis with RNA sequencing of MEC tissue from adult mice . We find that differential gene expression delineates the borders of the MEC with neighboring brain structures and reveals its laminar and dorsoventral organization . We propose a new molecular basis for distinguishing the deep layers of the MEC and show that their similarity to corresponding layers of neocortex is greater than that of superficial layers . Our analysis identifies ion channel- , cell adhesion- and synapse-related genes as candidates for functional differentiation of MEC layers and for encoding of spatial information at different scales along the dorsoventral axis of the MEC . We also reveal laminar organization of genes related to disease pathology and suggest that a high metabolic demand predisposes layer II to neurodegenerative pathology . In principle , our computational pipeline can be applied to high-throughput analysis of many forms of neuroanatomical data . Our results support the hypothesis that differences in gene expression contribute to functional specialization of superficial layers of the MEC and dorsoventral organization of the scale of spatial representations .
Spatial cognition emerges from interactions between specialized neuronal populations in the hippocampal-entorhinal system [1] . The medial entorhinal cortex ( MEC ) is of particular importance for cognitive functions that rely on estimation of spatial position and orientation [2] . Neurons in each layer of the MEC represent distinct information , have differing connectivity , and can be distinguished by their morphological and biophysical properties [3–7] . For example , layer II has a relatively high density of neurons with grid firing fields , whereas deeper layers contain a higher proportion of neurons with firing also modulated by head direction [3 , 8] . Further topographical organization is present orthogonal to cell layers along the dorsoventral axis in that the scale of spatial representations , local and long-range connectivity , synaptic integration and intrinsic electrophysiological properties all vary with dorsoventral position [9–15] . While this specialization of encoding and cellular properties is well established , the extent to which molecular specialization defines neuronal populations within the MEC or contributes to their distinct functions is not clear . Insights into molecular substrates for topographical organization in other brain regions have been gained through large-scale analysis of differences in gene expression [16–21] . Our understanding of the architecture and functions of MEC may benefit from similar approaches . While detailed anatomical and histochemical studies have shown that certain genes , including reelin , calbindin [22] and some cadherins [23] , identify cell populations associated with particular layers of the MEC , we know very little about the identity , laminar or dorsoventral organization of the vast majority of genes expressed in the MEC . This is a difficult problem to address for the MEC as its borders with adjacent structures are ambiguous , it has a dorsoventral as well as a laminar organization and its similarity to other cortical structures is unclear . As a result , key questions about its molecular organization are currently unanswered . For example , are the layers and borders of the MEC unambiguously delineated by coordinated expression patterns of multiple genes ? Are genes differentially expressed along the laminar and dorsoventral axes ? Do genome-wide laminar or dorsoventral differences in gene expression lead to mechanistic predictions regarding the organization of functional properties in the MEC ? The MEC is ontogenetically distinct from both the 3-layered hippocampus , and from neocortex [24] , with which it shares a similar laminar organization , but does this similarity to neocortex reflect a common molecular organization ? The topographical organization of the MEC extends to pathological signatures of common disorders in which it is implicated . Layer II exhibits neuronal loss [25] in patients with mild to severe Alzheimer’s disease ( AD ) and altered excitability in animal models [26] . Layer II is also affected in individuals with schizophrenia where there is evidence of abnormalities in cell size , organization and RNA expression [27 , 28] . In contrast , epilepsy is primarily associated with loss of layer III neurons in humans [29] and in animal models [30] . Yet , the mechanisms that predispose different cell populations in the MEC to particular disorders are not known . Given the evidence of genetic associations with these diseases [31–33] , it is possible that the molecular profiles of particular MEC neurons confer vulnerability . However , testing this hypothesis requires knowledge of the laminar and dorsoventral organization within the MEC of genes that are causally involved in disease . To better understand the molecular basis for its function and pathology , we aimed to establish a genome-wide approach to define the laminar and dorsoventral organization of the MEC transcriptome . Large-scale investigations into gene expression patterns have previously used either in situ hybridization ( ISH ) or RNA sequencing ( RNA-Seq ) to identify genes with differential expression in the neocortex [16 , 34] , hippocampus [17] and sub-cortical structures including the striatum [20] . While transcriptomic approaches such as RNA-Seq provide a robust platform for quantification of transcripts , accurate isolation of cell populations at the resolution of cell layers is challenging [16] , limiting the current applicability of this approach . In contrast , ISH is more useful for identifying patterns of gene expression because the precise location of transcripts can be examined . The Allen Brain Atlas ( ABA ) , a high-throughput ISH database , contains brain-wide data for over 20 , 000 genes and has been used to identify laminar borders , and to distinguish regions and cell types in the somatosensory cortex [34] , hippocampus [17] , and cerebellum [35] . However , in its currently accessible form ABA data is searchable at best at a resolution of 100 µm [34] and is further limited in its utility for automated comparison of laminar gene expression because the magnitude of error in alignment accuracy of brain sections is comparable to the width of narrow individual cortical layers . Small alignment errors can therefore easily lead to incorrect assignment of genes to layers , thereby confounding systematic analysis of differences between layers . To address these issues , we established a computational pipeline for registration and automated analysis of ISH data at a resolution of approximately 10 µm , enabling us to compare precise spatial expression patterns of over 80% of genes in the ABA dataset [34 , 36] . We combined analysis of this high spatial resolution data with RNA-Seq analysis of gene expression in dorsal and ventral regions of the MEC . We demonstrate that while very few genes are uniquely expressed in the MEC , differential gene expression defines its borders with neighboring brain structures , and its laminar and dorsoventral organization . We propose a new molecular basis for distinguishing the deep layers of the MEC and provide evidence that at a molecular level deep layers of the MEC are relatively similar to those of neocortex . Superficial layers are substantially more divergent between neocortex and MEC . Analysis of genes with differential expression suggests roles in layer-specific and dorsoventral specialization of calcium-ion binding molecules , ion channels , adhesion molecules and axon guidance-related molecules . We find that differential laminar expression patterns do not extend to genes directly implicated in disease , but selective expression of related genes may provide a context that confers vulnerability to pathology in neurodegenerative diseases such as AD . Our data establish a genome-wide framework for addressing the organization of circuit computations and pathology in the MEC .
To be able to systematically compare expression of genes in the adult mouse MEC at laminar resolution we extended the precision with which the localization of expressed genes in the ABA dataset can be compared . To achieve this we implemented methods to warp ISH images and their corresponding processed expression images , in which pixel intensity is used to represent the relative total transcript count [34] , into a standard reference frame ( see Materials and Methods , Fig . 1A-B and S1A Fig . ) . Our re-registered ABA data set contains at least 1 image meeting our quality criteria and that contains the MEC for 16 , 639 genes ( 81 . 2% of genes in the ABA sagittal dataset ) ( S1B Fig . ) . Non-linear registration provides a striking improvement in the spatial resolution at which the localization of gene expression can be compared ( Fig . 1C ) . Prior to re-registration , exploration of the organization of gene expression is confounded by variability in the shape and size of brain sections used in different ISH experiments . For example , averaging ISH expression images for 1000 genes prior to registration results in diffuse images without any laminar organization ( Fig . 1C ) . In contrast , following re-registration of the same images , neocortical layers are clearly distinguishable from one another ( Fig . 1C ) . Other landmarks , for example the white matter border with the striatal region , the hippocampal pyramidal layer and layer I of the piriform cortex , also become clearly identifiable ( Fig . 1C ) . Within the MEC , the resolution is such that multiple layers and a dorsoventral organization can be recognized . Thus , our computational pipeline for image registration enables high-resolution comparison of gene expression between layers and along the dorsoventral axes of the MEC , as well as with other brain regions . To validate gene expression data extracted from the ABA we compared mean pixel intensity values across dorsal and ventral MEC with RNA-Seq data acquired from the same regions ( see Materials and Methods , S1A Fig . ) . Our RNA-Seq analysis detected 20 , 106 of the 38 , 553 genes ( 52 . 7% ) in the Ensembl mouse database ( release 73 ) , including 15 , 496 protein-coding genes . In comparison , of the 16 , 639 genes from the ABA dataset for which we successfully registered images , our analysis revealed that 9 , 873 ( 59 . 3% ) are expressed in the MEC , including 8 , 941 genes that we could identify in the Ensembl database ( Fig . 1D ) . Of the registered Ensembl genes , 8 , 064 ( 90 . 2% ) were also detected by RNA-Seq , indicating a high degree of consistency between the two approaches ( Fig . 1D ) . This is supported by a significant positive correlation between RNA-Seq transcript FPKM ( fragments per kilobase of exon per million fragments mapped [37] ) and mean pixel intensity of ABA images ( r = 0 . 40 , p < 2 . 2 × 10-16 ) ( Fig . 1E ) . It is possible that the 877 genes that appear to be expressed in ABA data , but are not detected by RNA-Seq , are false positives , while the 3 , 297 genes detected using RNA-Seq that are not detectable in the ABA data , may reflect false negatives in the ISH data , for example due to errors in probe design , staining or image processing . A further 6 , 463 genes detected using RNA-Seq that are not in the re-registered ABA dataset ( Fig . 1D ) include 1 , 939 pseudogenes and 791 long intergenic non-coding RNAs , as well as 1 , 855 protein-coding genes . Together , these data demonstrate the potential of combining advanced image processing tools for high resolution alignment and analysis of ISH data sets with RNA-Seq . RNA-Seq enables quantification of thousands of transcripts over a large dynamic range , while automated analysis of ISH data reveals gene expression at laminar resolution that can be quantified and compared within and across brain regions . Previous investigation using microarrays to compare tissue harvested from multiple brain regions has shown that gene expression in the entorhinal area is most similar to that of neocortex and hippocampus and least similar to non-telencephalic regions [38] . However , it is not clear if there are individual genes that distinguish these brain regions , whether differences show laminar or dorsoventral specificity or if this pattern applies to the MEC specifically rather than the entorhinal area as a whole . To first determine whether gene expression in the MEC alone shows a relationship to other brain areas similar to that established by microarray analysis , we used the re-registered ABA dataset to isolate ISH gene expression data from several brain regions ( Fig . 2A ) . Consistent with microarray data [38] , we found that gene expression in MEC correlates most strongly with neocortex ( r = 0 . 958 , p < 2 . 2 × 10-16 ) , amygdala ( r = 0 . 936 , p < 2 . 2 × 10-16 ) and the hippocampal pyramidal layer ( r = 0 . 942 , p < 2 . 2 × 10-16 ) and more weakly with the caudate putamen ( r = 0 . 887 , p < 2 . 2 × 10-16 ) ( Figs . 2B , S2A-B ) . The relatively high correlations between the MEC and the other regions suggests that differential expression of relatively few genes is likely to underlie functional differences between these areas . To investigate whether the expression of single genes could distinguish MEC from other regions , we identified genes with at least 4-fold higher mean pixel intensity in MEC compared with each of the other regions ( S2C Fig . , Materials and Methods ) . This analysis revealed 118 genes that are expressed at higher levels in MEC than neocortex , 93 for piriform cortex and 54 for the hippocampus , compared with 318 for the amygdala and 1 , 162 for the caudate putamen . These numbers decrease as the threshold difference in mean intensity between MEC and the other regions is increased ( S2D Fig . ) . Section-wide average images reveal that the expression within the MEC of these genes is often not uniform , but can be concentrated in specific layers ( e . g . MEC vs neocortex ) or dorsoventrally organized ( MEC vs piriform cortex and amygdala ) ( Fig . 2C ) . They also reveal that genes selectively enriched in MEC compared with one region are , on average , also strongly expressed in other regions ( Fig . 2C ) . Nevertheless , 3 genes could be identified as uniquely enriched in the MEC compared with the other 5 regions ( Fig . 2C ) , although expression of each followed a laminar organization and did not mark the MEC as a whole . We also asked if combinations of expressed genes might better distinguish the MEC from other regions . However , we found that only in a minority of pairs of genes with converging expression in MEC ( 14/456 ) does expression fully colocalize to the same laminar and dorsoventral regions ( S2E-F Fig . ) . Together , these data further validate our quantification of MEC gene expression and indicate that few , if any , individual genes or pairs of genes are likely to distinguish the MEC as a whole from other brain regions . Thus , specific attributes of the MEC are unlikely to be a product of highly specific expression of a few genes . Instead , our data are consistent with combinatorial expression of larger sets of genes defining differences between cell populations in the MEC and other brain areas ( c . f . [39] ) . Our data also highlight a limitation of regional comparison of gene expression in that genes which are co-expressed in a given brain region may not be colocalized to the same cell layer or dorsoventral area . Therefore , to better understand the laminar and dorsoventral organization of gene expression in the MEC , and the relationship between the organization of the MEC and neocortex , with which it has the most similar overall gene expression [38] , we took advantage of our pipeline for large-scale comparison of gene expression to analyze expression at laminar and sub-laminar resolution . Borders of the MEC , which we consider here as the region also previously referred to as the caudal entorhinal field [4 , 40 , 41] , have typically been defined on the basis of classical cytoarchitectonic criteria , chemoarchitecture and connectivity [11 , 40 , 42] . However , because these criteria don’t always converge , ambiguity exists regarding the definition and location of the borders with adjacent regions including the parasubiculum [43 , 44] as well as with more ventral structures ( c . f . [42 , 45 , 46] ) . We therefore sought to determine whether ISH data , and in particular our re-registered ABA sagittal data set , would enable a clearer resolution of dorsal and ventral MEC borders , which can be viewed unambiguously in the sagittal plane . We focused initially on identifying genes that delineate the dorsal border of the MEC . In some atlases this region is considered as the retrosplenial or perirhinal region [47] , and in others as the ectorhinal region [34 , 36] . However , cytoarchitectonic , histological and electrophysiological studies in rats suggest that part of this region corresponds to the superficial layers of the parasubiculum [43 , 48] . By comparing relative pixel intensity between the dorsal MEC and adjacent regions ( see Materials and Methods , Fig . 3A ) , we identified a number of genes with expression that appears to stop at the dorsal border of the MEC ( e . g . Wdr16 and Fabp5 ) ( Fig . 3B ) . For some of these genes , expression is only absent within a wedge-shaped region before resuming in more dorsal cortical areas ( e . g . Nov ) , a pattern which is highly consistent across different medio-lateral sections ( Figs . 3C , S3A-B ) . We therefore asked if there are genes that are expressed in the wedge-shaped region , but not the adjacent regions . We identified 9 such genes ( see Materials and Methods ) , including Igfbp6 and Kctd16 ( Fig . 3D ) . All of these genes are also expressed in the parasubicular region medial to the MEC ( Fig . 3E ) . Of these , 7/9 also have sparse expression in superficial parts of MEC ( Fig . 3D ) . These observations support the view that the parasubiculum extends to wrap around the dorsal border of MEC [43 , 48] . At the ventral aspect of the MEC , cytoarchitectonic analysis delineates a border with the medial entorhinal field [40 , 47] . We asked whether differential gene expression supports the presence of this ventral border and whether it can clarify its position . By analyzing gene expression in regions either side of the approximate location of this border ( Fig . 3F ) , we identified genes with expression that drops off sharply ( Fig . 3G , S3C-D Fig . ) . These genes include apparently layer-specific genes such as Nef3 , Cutl2 and Col5a1 in layers II , III and V/VI respectively . We also identified genes with the converse pattern of high ventral expression and low MEC expression , including Kctd6 , Sema3c and LOC241794 ( Fig . 3H ) . These expression patterns are consistent across different mediolateral sections ( S3C-D Fig . ) . Thus , the ventral border of the MEC can be identified by genes with sharply increased or reduced expression . Cytoarchitectonic and developmental studies indicate that the MEC is a type of periarchicortex ( paleocortex ) , a transitional structure between 6-layered neocortex and 3-layered archicortex , with 5 cytoarchitecturally distinct cell body layers [24 , 40 , 45] . However , while layer II and III are easily distinguished by their cytoarchitecture and connectivity ( cf . [40] ) , differentiation of cell populations within layers V and VI is less well established , although there is evidence that the cell bodies and dendrites of distinct cell types are differentially distributed within these layers [4 , 6 , 40 , 49] . Layer IV corresponds to the cell free lamina dissecans [40] . To better define the laminar organization of the MEC , and to be able to compare its structure to other cortical regions , we therefore asked if differential gene expression distinguishes the superficial from deep layers or clarifies laminar borders within the deep layers ( Figs . 4 and S4 ) . We first identified 159 genes specifically expressed in layers II , III or V/VI ( see Materials and Methods and S4A-C Fig . ) . We refer to these genes , which show no consistent expression in other layers , as layer-specific genes ( see Materials and Methods ) . We initially analyzed deep layers ( V—VI ) together because their divisions and borders are not easily distinguished by cytoarchitectonic criteria . Since genes with layer-specific expression patterns are of particular interest as neuroscience tools for isolating laminar functions , we examined their likely validity . Almost all layer-specific genes could be detected in our RNA-Seq analysis ( 149/159 with mean FPKM ≥ 0 . 1 ) and 62/159 had substantial levels of expression ( mean FPKM ≥ 10 ) . We also identified a further set of 622 genes , which we define as strongly differentially expressed ( DE ) ( see Materials and Methods , S4B Fig . ) . These genes are expressed at higher levels in at least one layer than another , but are not necessarily exclusive to one layer . Both layer-specific and DE genes show consistent expression patterns across mediolateral sections ( S6 Fig . ) . Only 37 of the layer-specific genes and 144 of the DE genes are amongst the 1000 most viewed genes in the ABA [50] . Thus , layers of the MEC can be distinguished by layer-specific and DE genes , many of which have received little previous attention suggesting they may represent new targets for future exploration . Further examination of genes specific to the deep layers revealed three separate divisions of layers V and VI . First , a narrow zone at the deep border of layer IV is distinguished by expression of 5 genes , including Etv1 , Grp and Nts ( Figs . 4A , S4D ) . A second narrow zone of cells that is adjacent to the white matter is delineated by the expression of 8 genes , including Jup and Nxph4 ( Figs . 4A , S4D ) . Finally , the wide intervening region is distinguished by 20 genes , including Thsd7b , Cobll1 and Col5a1 ( Figs . 4A , S4D ) . Because layer V has been suggested to have a narrow superficial and wider deep zone [40] , we refer here to the two more superficial subdivisions as layer Va ( narrow ) and Vb ( wide ) , and we refer to the layer bordering the white matter as layer VI [40] . This delineation of layers Va , Vb and VI is supported by patterns of expression within the deep layers for the larger set of DE genes ( layer Va ( n = 24 ) , Vb ( n = 55 ) and VI ( n = 13 ) ; S4E Fig . ) . A further 27 DE genes are expressed in both layer Va and VI , but not Vb ( S4E Fig . ) . Thus , patterns of gene expression enable differentiation of divisions within the deep layers . Previous cytoarchitectonic and electrophysiological studies have indicated that within layer II a subset of cells are clustered in islands [51–54] . In mice , neurons within islands express calbindin , whereas neurons outside islands express reelin [22 , 53 , 54] . Cells in these islands are of particular interest as they differ from reelin-positive cells in both their electrophysiology and projection targets [22 , 53 , 54] . Taking all DE genes within MEC , we found 30 genes that within layer II are predominantly expressed in apparent islands ( Figs . 4A , S4F ) . Of these , just 8 are also specific to layer II within the MEC ( S4G Fig . ) , including the calcium-binding protein , calbindin ( Calb1 ) . A further 19 are strongly expressed in the MEC deep layers but not layer III . The island genes include 13 genes that are expressed in the wedge-shaped patches of presumed parasubiculum adjacent to dorsal MEC , 3 of which we identified earlier ( e . g . Mrg1 , S4F Fig . ) . We also identified 37 genes with the converse , ‘Inter-island’ , pattern ( S4F Fig . ) . Of these genes 23 are specific to layer II , including Reln ( reelin ) and Il1rapl2 ( S4G Fig . ) . A further 11 are also strongly expressed in the MEC deep layers , in particular in layer Va . The remaining layer II-specific genes do not appear to be uniquely expressed in either island or inter-island regions ( S4G Fig . ) . Thus , differential gene expression distinguishes cell populations within layer II , shows that cells within and outside islands may be distinguished from cells in other layers by expression of common genes and provides evidence of similarities between the layer II island cells and parasubiculum . What is the relationship between laminar organization of the MEC and other regions of cerebral cortex ? While MEC has greatest similarity in gene expression to neocortex and also shares a similar laminar structure , classic ontogenetic evidence indicates that the two regions are developmentally distinct [24] . However , it is unclear if these ontogenetic differences are associated with later molecular differences between specific layers of the mature cortices . To address this we first systematically examined overall expression in visual cortex and somatosensory ( SS ) cortex of genes with layer-specific expression in MEC . When we averaged expression of all genes selectively expressed in layers V/VI of MEC we found them to have a similar laminar organization in neocortex ( Fig . 5A ) . In contrast , mean expression in neocortex of genes localized specifically to layer II or III of the MEC has a less distinct laminar organization ( Fig . 5A ) . To quantify these differences we measured the distribution of gene expression intensity as a function of distance from the corpus callosum to the pial surface ( Fig . 5B , Materials and Methods ) . We found that the three groups of MEC layer-specific genes have differing expression patterns in neocortical regions ( Figs . 5C , S5A; Mixed Model Analysis , F = 12 . 3 , p < 0 . 001 ) . To assess the degree to which the laminar organization of each group of MEC layer-specific genes is maintained in neocortex , we first calculated the ratio of their expression in deep layers ( V and VI ) to superficial layers ( II-IV ) . We then calculated the difference between these ratios and their expected values of 1 for deep genes , and zero for superficial genes . This difference was significantly smaller for deep layer-specific genes compared with superficial layer-specific MEC genes ( Fig . 5D; MANOVA , p = 0 . 002 and p = 0 . 004 for effect of MEC layer-specific group in visual and SS cortex respectively ) . A similar relationship is apparent when we consider the expression patterns of individual genes ( S5B , C , D Fig . ) . Around 92% of deep layer-specific MEC genes are expressed in the neocortex and 61–64% of all these genes are enriched in deep visual and SS cortex , respectively ( S5C Fig . ) . In contrast , 77% of superficial layer genes are expressed in SS or visual cortex , but just 28–43% are enriched in superficial layers ( S5C Fig . ) . Thus , our analysis of layer-specific genes supports the idea that deep layers of MEC have greater similarity to neocortical regions than superficial layers . To investigate whether the relationship between layers of the MEC and neocortex extends beyond layer-specific genes , we took all genes in the re-registered ABA data set and examined the correlations in pixel intensity between layers in different cortical regions ( Fig . 5E ) . MEC deep layers together correlate most strongly with neocortical layer VI ( r = 0 . 96 , 0 . 94 , p < 2 . 2 × 10-16 ) , while layer II and III of MEC are more strongly correlated with neocortical layer V ( r = 0 . 93–0 . 95 , p < 2 . 2 × 10-16 ) than II or III ( r = 0 . 90–0 . 91 , p < 2 . 2 × 10-16 ) ( Fig . 5E ) . To establish whether these correlations differ from those between neocortical regions , we investigated correlations between visual and SS cortices across all genes . We found that all corresponding layers correlated strongly ( r > 0 . 96 , p < 2 . 2 × 10-16 ) ( Fig . 5E ) . Similarly , when we examined expression patterns of SS cortex layer-enriched genes [34 , 36] , we found a similar laminar organization of expression between SS and visual cortex ( S5E Fig . ) . Thus , while laminar organization of gene expression is maintained between neocortical regions , gene expression within superficial layers of MEC , in particular , diverges from corresponding layers of neocortex . Given the overall similarity between gene expression in deep layers of MEC and neocortex , we examined possible relationships between particular deep layers in each region . Of genes specifically expressed in particular deep layers we found that MEC layer VI-specific genes are almost always also expressed in layer VIb of neocortical regions ( n = 7/8; Figs . 5A , S5D ) ( c . f . [55] ) . Meanwhile , layer Vb-specific genes are more commonly expressed in layer VIa of neocortex than layer V ( S5D Fig . , n = 15 vs 7 / 19 ) . Moreover , MEC deep layers together correlate most strongly with neocortical layer VI ( r = 0 . 96 , 0 . 94 , p < 2 . 2 × 10-16 ) , and more weakly with layer V ( r = 0 . 94 , 0 . 91 , p < 2 . 2 × 10-16 ) ( Fig . 5E ) . This is consistent with our observation that MEC layer Vb genes are more commonly expressed in layer VIa of neocortex than layer V , as the layer Vb region occupies the majority of the area of the MEC deep layers . Thus , at the level of gene expression MEC layers Vb and VI can be considered most closely related to neocortical layers VIa and VIb , respectively . In summary , our analysis provides molecular evidence for an organization in which deep layers of MEC and neocortex implement similar gene expression programs , whereas superficial layers of MEC and neocortex express more diverse sets of genes . Our analysis suggests specialized gene expression in different layers of the MEC . If this reflects an underlying functional organization then it could be reflected in the functions associated with layer-specific or DE genes . Given known differences in electrical intrinsic properties , morphology , connectivity and organization of cells between MEC layers [4 , 6 , 40 , 45] , we hypothesized that genes involved in cell excitability and communication might be differentially expressed across layers . To test this , we focused initially on DE genes as their greater number gives more statistical power in identifying over-represented gene attributes . We identified Gene Ontology ( GO ) annotations and pathways that are overrepresented amongst DE genes ( n = 722 Ensembl-identified genes ) relative to all genes expressed in the MEC ( n = 9 , 057 Ensembl-identified genes ) . To reduce redundancy and identify diverse functions of interest , we clustered enriched terms into groups . Consistent with our prediction , genes associated with neuronal projections ( n = 75 , padj = 1 . 85 × 10-9 ) , particularly synapses ( n = 45 , padj = 4 . 57 × 10-5 ) , and those involved in calcium ion binding ( n = 61 , padj = 2 . 72 × 10-7 ) , cell adhesion ( n = 54 , padj = 9 . 48 × 10-8 ) , and axon guidance ( n = 23 , padj = 3 . 14 × 10-4 ) are overrepresented amongst DE genes ( Fig . 6A ) . We also found strong enrichment of genes involved in ion channel activity ( n = 46 , padj = 7 . 05 × 10-7 ) and synaptic transmission ( n = 25 , padj = 2 . 36 × 10-4 ) ( Fig . 6A ) . Amongst ion transport-related genes , cation channel activity ( n = 34 , padj = 3 . 82 × 10-5 ) is particularly enriched whereas anion channel activity is not . We asked if attributes enriched among DE genes were also identifiable amongst layer-specific genes . In addition to being significantly overrepresented amongst all DE genes , cell adhesion , axon guidance and calcium ion binding-related genes were also significantly overrepresented amongst the group of layer-specific genes ( Fig . 6B ) . Given critical roles of these genes in neuronal signaling , these data support the idea that laminar differences in gene expression within the MEC support laminar organization of computations within MEC microcircuits . Are genes within the functional groups that are overrepresented amongst DE genes enriched in particular layers or are they distributed across layers ? Comparison of expression patterns for individual genes revealed genes with enriched expression in each layer ( Fig . 6C , Materials and Methods ) . This analysis highlights a number of genes of potential functional importance . For example , ion channel-related genes include the potassium channel subunits Kcna4 and Kcnmb4 , which control excitability and are enriched in layers II and Vb respectively , while axon guidance/adhesion-related genes enriched in layer II include Lef1 , Lhx2 and Dcc as well as the ephrin receptor gene Epha4 ( Fig . 6C ) . A possible role for the latter genes could be to control guidance of axons to newborn granule cells in the dentate gyrus [56] . Cell adhesion-related genes are also selectively expressed and significantly overrepresented in all layers and include several cadherins and protocadherins ( Fig . 6B-C ) . These data suggest that subsets of each functional group of DE genes are expressed in each layer . Together these data reveal candidate categories of genes that are most likely to distinguish the functions of different layers within the MEC . Our analysis also identifies molecules with highly specific laminar expression that could contribute to particular electrical and synaptic properties . Topographical organization of intrinsic features along the dorsoventral extent of the MEC has received considerable interest because the characteristics of grid cells vary systematically along this axis [9 , 10 , 12 , 13 , 57] . The extent to which gene expression parallels this organization is not currently known . We took two approaches to addressing this issue , one using our re-registered ABA dataset , with its advantage of high spatial resolution , and the other using RNA-Seq analysis , which enables quantification across a wide dynamic range and the ability to test the reproducibility of gradients . This combined approach therefore enabled us to question not only dorsoventral differences in gene expression , but also their laminar organization . We first calculated the ratio of pixel intensity between dorsal and ventral regions in images from the re-registered ABA dataset ( Fig . 7A ) . We defined genes with at least 20% more expression in the dorsal than ventral area as being expressed higher dorsally ( D>V ) and those with at least 20% more in the ventral area as being expressed higher ventrally ( V>D ) ( see Materials and Methods , S7A Fig . ) . As a result , we identified 3 , 188 D>V genes compared with 1 , 352 V>D genes ( Fig . 7B ) . We next used RNA-Seq analysis to compare gene expression from microdissected regions of dorsal and ventral MEC . This also identified genes with dorsoventral differences in their expression ( Fig . 7C ) , of which 1 , 467 D>V genes and 1 , 198 V>D genes satisfied our criteria of 20% more expression in one of the areas than the other . Of these genes 452 and 347 , respectively , had statistically significant differences in expression ( Cuffdiff 2 [58]: FDR < 0 . 05 ) across 4 replicate samples ( Fig . 7C ) . To establish whether similar populations of dorsoventrally expressed genes are identified by RNA-Seq and in the re-registered ABA dataset , we correlated the ratio of dorsal to ventral expression determined by each method . First , to avoid confounds from genes with different expression between layers , we focused on genes expressed in only one layer . We found that measures of differential expression are strongly correlated between ABA and RNA-Seq datasets for layer II , III and V/VI ( Fig . 7D , LII: slope = 0 . 85 , r = 0 . 74 , p = 0 . 0007 , LIII: slope = 1 . 5 , r = 0 . 97 , p = 0 . 0002 , LV/VI: slope = 2 . 4 , r = 0 . 77 , p = 0 . 0038 ) . Second , we compared gene expression for all genes found to be significantly differentially expressed across biological replicates in RNA-Seq data . We again found a significant correlation between the datasets ( S7B Fig . , r = 0 . 52 , p < 2 . 2 × 10-16 ) . Do dorsoventral differences in gene expression manifest differently across layers ? Average images indicate that layer II has the strongest D>V pattern , while the deep layers have the strongest V>D pattern ( Fig . 7B ) . To test this , we compared the average ratio of ventral to dorsal expression for all layer-specific genes . We found significant differences in the ventral to dorsal ratio for deep layer-specific genes compared to layer-II or III-specific genes ( 1-way ANOVA F = 7 . 47 , p = 0 . 0008 . Post-hoc Tukey’s HSD LII vs . Deep p = 0 . 0016 , LIII vs Deep: p = 0 . 010 ) , with deep layers enriched for a V>D expression pattern ( Fig . 7E ) . Indeed , while 20 . 6% of layer II and 14 . 3% of layer III-specific genes show significant D>V expression , only 1 . 8% of deep layer genes do ( Figs . 7D , S7C ) . Together our data provide convergent evidence for systematic organization of gene expression along the dorsoventral axis of the MEC , identify dorsally and ventrally enriched sets of genes , and suggest differences in the laminar organization of dorsoventral gradients . Are the roles of genes with differential dorsoventral expression related to the cellular and system-level organization of function in the MEC ? Taking all significant D>V and V>D genes identified by RNA-Seq , we investigated their possible functions using a GO and pathway analysis . By using clustering to distinguish enriched terms into key groups of interest ( see Materials and Methods ) , we found that D>V genes are enriched for a number of attributes , particularly axon ensheathment ( n = 15 , padj = 2 . 08 × 10-7 ) and channel activity ( n = 32 , padj = 2 . 81 × 10-7 ) ( Fig . 8A ) . We next used the re-registered ABA data set to examine the expression patterns of the identified gene groups . The D>V pattern found using RNA-Seq is replicated for the majority of axon ensheathment- ( n = 9/12 ) and channel activity-related genes ( n = 19/24 ) that show expression in the re-registered ABA dataset ( Fig . 8B ) . We then investigated the layers in which gradients are strongest . Axon ensheathment genes show consistent D>V gradients in the superficial ( n = 11/12 ) and deep ( n = 9/12 ) layers ( Fig . 8B ) . In contrast , genes involved in channel activity are more likely to show D>V gradients in the superficial ( n = 22 / 24 ) than in the deep ( 9/24 ) layers ( Fig . 8B ) . In contrast to D>V genes , genes with V>D expression in the RNA-Seq data set are most strongly enriched for the neuroactive ligand-receptor pathway ( n = 24 , padj = 4 . 04 × 10-13 ) , which is related to G-protein coupled receptor activity ( n = 35 , padj = 2 . 2 × 10-9 ) , and for the extracellular region ( n = 44 , padj = 2 . 16 × 10-9 ) ( Fig . 8C ) . Of the 11 identified neuroactive ligand-receptor pathway genes that are expressed in our re-registered ABA data set , 7 have consistent overall V>D patterns in the ABA ( Fig . 8D ) while 3 show no detectable expression . A total of 15/24 extracellular region-related genes are consistent with ABA data ( Fig . 8D ) , with a further 12 showing no detectable expression . A possible reason for the discrepancies between ABA and RNA-Seq measures is that in the ABA analysis of the overall V>D gradient , V>D gradients that are only present in the deep layers may go undetected . This is because in some ABA images , ventral deep layers become narrower towards the medial border of the MEC and therefore ventral gene expression may be overshadowed by expression in the superficial layers or may not be present in the image . In support of this , we found that most V>D gradients found using RNA-Seq could be detected in ABA data when gene expression was specifically measured in the deep layers ( Neuroactive: 8/11 , Extracellular: 20/24 , Fig . 8D ) . Pathological changes in the MEC have been observed in a number of neuro-developmental and neurodegenerative disorders . Whereas layer III appears to be most consistently affected in epilepsy patients and in animals models of epilepsy [29 , 30] , cell number and disrupted organization within layer II are consistently reported in Alzheimer’s disease ( AD ) [51 , 59] , as well as in Huntington’s ( HD ) and Parkinson’s disease ( PD ) [60] , schizophrenia [61] and autism [62] . This laminar specificity suggests that particular features of these layers , whether genetic or network-based , hard-wired or experience-driven , confer vulnerability . One possibility is that genes with mutations causally linked to particular disorders have layer-enriched expression . Alternatively , broadly expressed causal genes might cause specific pathology in layers with enriched expression of genes that confer vulnerability . To address whether normal adult gene expression in the mouse provides insight into vulnerability , we first explored the laminar expression patterns of genes involved in signaling pathways that are disrupted in disease . Images showing the average expression pattern of genes involved in KEGG neurodegenerative disease pathways ( AD , HD and PD ) [63] indicate high expression in layer II , particularly in dorsal regions ( Fig . 9A ) . To test whether this reflects significant enrichment of neurodegenerative disease pathway genes in layer II , we took all DE genes that exhibit high expression in layer II ( see Materials and Methods ) and compared representation of disease-related genes to their representation in the MEC as a whole . We found that AD , PD , and HD pathway genes are all overrepresented amongst layer II-enriched genes ( Fig . 9C , Exact Fisher Test with Benjamini-Hochberg correction: Log2 Fold Enrichment > 1 . 19 , padj < 0 . 024 ) . Thus , basal gene expression may confer vulnerability of layer II in neurodegenerative diseases . In the absence of KEGG pathway information , we used several database resources to identify genes related to schizophrenia [64 , 65] , autism [66] and epilepsy [67] ( see Materials and Methods ) . Average images reveal weak , if any laminar organization for schizophrenia and epilepsy-related genes , with some evidence of layer II enrichment for autism-related genes . However , an enrichment analysis shows that schizophrenia-related genes are overrepresented amongst layer III-enriched genes ( Fig . 9C ) , while both autism- and schizophrenia-related genes are enriched amongst RNA-Seq defined D>V genes ( Fig . 9C ) , suggesting that pathology related to these diseases could show dorsoventral differences . Since layer II enrichment of AD pathway genes corresponds with layer II vulnerability to AD , we explored whether genes with variants that have been established to confer increased risk of AD show layer-specific expression . AD possesses several key genetic risk factors , namely APP , PSEN1 , and PSEN2 [68] , but meta-analyses of genome-wide association studies have also shown that ApoE , ABCA7 , Clu , Bin1 , Cd33 , Cd2ap , Epha1 , Ms4a6A-E , Picalm , Sorl1 , Ptk2b , NME8 , FERMT2 , CASS4 , Inpp5d , Dsg2 , Mef2c and Cr1 are strongly associated with late-onset AD [69–71] . We found that 16 out of the 20 of these genes that are in our re-registered ABA data set are expressed in the MEC . However , none are specifically expressed in layer II and only a minority show strong differential expression across layers ( Fig . 9D ) . Pathology in layer II is therefore unlikely to be the result of layer-specific expression of AD risk genes . It could instead reflect enriched expression of signaling pathways linked to neurodegeneration in AD ( Fig . 9A ) . Indeed , further analysis of the laminar expression patterns of AD pathway genes ( Fig . 9E , Materials and Methods ) reveals that almost all those with moderate laminar enrichment show highest expression in layer II , and that many ( n = 16 / 26 ) are mitochondrion-associated genes , suggesting that cells in layer II may have higher energy demands than cells in other layers . This is consistent with the strong cytochrome oxidase staining observed in layer II [48] . Given that mitochondrial dysfunction is a feature of neurodegenerative disease [72] and that genes related to metabolism are altered in the MEC of patients with mild cognitive impairment [73] and AD [74] , layer II vulnerability could be due to or compounded by enriched expression in layer II of signaling pathways that confer vulnerability to AD pathology .
By developing a pipeline for automated comparison of brain sections at 10 µm resolution we were able to identify genes whose expression pattern delineates the borders and layers of the MEC ( Fig . 3 and Fig . 4 ) . Validation of this pipeline against RNA-Seq data indicates that relative expression levels estimated with the two approaches are consistent ( Fig . 1 and Fig . 7 ) . Recent work using double ISH labeling validates the layer-specific expression patterns we find for cadherins in the MEC [23] , while other well-characterized genes such as reelin and calbindin [22 , 75] also have expected expression patterns . Our analysis identifies a further 767 genes with layer-specific or enriched expression and 799 genes with dorsoventral expression . Nevertheless , our current analysis is limited by the availability of genes in the ABA data set ( 20 , 495 / 38 , 553 Ensembl genes , most of which are protein-coding ) , by the likelihood of false negative data in the ABA ISH data where true gene expression has been missed ( estimated 3 , 297 / 14 , 054 by comparison with RNA-Seq data ) and by limitations in image processing and registration accuracy that prevent us making use of the entire ABA data set . While our analysis is restricted to sections in parasagittal planes containing the MEC , it could be extended to include other brain regions through additional planes and to other species including humans [76] . In principle our approach could also be extended to analysis of images from three-dimensional datasets obtained using different methods [77 , 78] . Our results resolve dorsal and ventral borders of the MEC , provide molecular evidence for laminar divisions of its deep layers , and identify numerous new molecular markers for the well-established separation of the superficial layers . While dorsal and ventral borders can be distinguished unambiguously in sagittal images , medial and lateral borders are better resolved in horizontal sections , so demarcation of these borders may require use of additional horizontal data sets . Delineation of deep layers is of particular interest as they are believed to relay hippocampal output to neocortex ( c . f . [4] ) , but their organization and functional properties have received relatively little attention . We distinguish a narrow region deep to the laminar dissecans as layer Va , consistent with that described by [45] . We also identify a distinct division of the deeper layers into Vb and VI , a narrow region of cells that appears continuous with neocortical layer VIb ( Fig . 4 ) . We suggest that the divisions previously reported within layer V [6 , 40 , 79] correspond to the superficial layer Va and a deeper layer Vb that we identify here . Definitive laminar delineations within the deep layers will require analysis of shared gene expression , dendritic morphology and axonal connectivity . Our results also identify new markers for island cells and , to our surprise , suggest their similarity to neurons in the parasubiculum . It will be interesting to establish whether this similarity extends to functional properties [43] . Our analysis and data sets provide a resource for future functional investigation of laminar organization of functions in the MEC . This includes identification of markers for distinguishing cell populations ( Fig . 4 ) , particularly for layer III and the deep layers , for which there are currently few specific markers . Our delineation of layers Va , Vb and VI identifies several genes in each layer whose promoters may be usable for generation of driver lines to target that cell population . We also identify common expression patterns between MEC and neocortex that may underlie shared functional roles ( Fig . 5 ) . For example , the cortical layer VI-specific immunoglobulin heavy chain gene , TIGR accession TC146068 [34] , also shows expression in MEC layer Vb . It is unlikely that this similarity in expression between deep layers of MEC and neocortex reflects biased selection of exemplar genes [80] as it is present when considering all expressed genes as well as those with laminar selectivity ( Fig . 5 ) . Instead , our analysis demonstrates that deep layers of MEC show greater similarity to corresponding cortical layers than do more superficial layers . Because our analysis includes the majority of protein-coding genes ( Fig . 1 ) , it also leads to novel predictions about gene expression underlying specialized function . As well as identifying candidates for electrophysiological differences between neurons from different layers [6] , many cell adhesion and axon guidance molecules are enriched amongst patterned genes ( Fig . 6 ) . Of particular interest are cell adhesion-related genes such as Cdh13 , Lef-1 and Dcc that show a similar expression pattern to Reln , which marks the subset of excitatory layer II cells that project to the dentate gyrus [22] . One possibility is that these genes play roles in forming connections with new born granule cells . Genome-wide views of cortical organization can inform investigation of disease mechanisms by identifying convergent expression of molecular components of disease pathways [34 , 81] . We found no evidence of layer-specific expression of genes causally implicated in disease pathology ( Fig . 9 ) . Instead , our analysis suggests that differential gene expression may underlie layer-specific pathology by predisposing specific cell populations to disease-causing mechanisms . For example , enriched expression of energy-related genes may reflect susceptibility of this layer to degeneration in AD . These functional and pathological predictions should be testable in future experimental studies . The dorsoventral organization of the resolution of grid firing fields and the corresponding organization of excitable and synaptic properties of layer II stellate cells [9 , 10 , 12 , 13] suggests that cellular mechanisms for grid firing may be identifiable by comparison of key features of dorsal and ventral MEC circuits . However , until now there has been little evidence for molecular differences that could underlie this organization ( cf . [12] ) . We provide converging evidence from re-registered ABA data and from RNA-Seq data for systematic coordination of gene expression along the dorsoventral axis of the MEC . Consistent with key roles of superficial layers in the generation of grid fields , D>V gradients were most often found in layer II and III ( Fig . 7 ) . We also found evidence for genes with the opposite V>D pattern of expression , but these were most prominent in deeper layers , suggesting that control of dorsoventral differences by molecular pathways differs across layers . A potential caveat of our analysis is that dorsoventral differences in gene expression could reflect differences in the proportions of certain cell types . While approaches such as transcriptomic analysis of isolated cells will be required to resolve this , our finding that many layer-specific genes are not significantly differentially expressed along the dorsoventral axis , while dorsoventral genes have continuous rather than all or nothing changes in intensity ( Fig . 8 ) , argues for gradients reflecting coordination of gene expression levels within populations of a single neuron type . By taking a genome-wide approach to differences in gene expression we obtained unbiased estimates of gene functions that are enriched among dorsoventral genes . Strikingly , we found enrichment among genes with higher dorsal expression of axon ensheathment and ion channel activity ( Fig . 8 ) . This is in accordance with previous evidence for dorsoventral differences in synaptic transmission and ionic conductances [12 , 13 , 15] , and in immunolabelling for myelin [48] . Enrichment of 10–20 genes associated with each function indicates that the corresponding cellular differences may involve coordinated control of gene expression modules . For example , our analysis extends candidates for dorsoventral differences in excitability from HCN and leak K+ channels [12] , to include non-selective cation channels such as Trpc5 [82] and voltage-dependent potassium channels such as Kcnq3 [83] and Kcnk1 ( Twik1 ) [84] . Similarly , we identify myelin-related genes such as Mbp and Plp1 , as well as related adhesion molecules such as Cntn2 ( Tag-2 ) , as candidates for dorsoventral differences in coordination of axon ensheathment [85] . Future gene manipulation studies will be required to establish causal roles of these genes in dorsoventral tuning of cell properties and of spatial firing . They may also provide insight into the role of topographic gene expression in the development and maintenance of topographical connectivity between the MEC and hippocampus . Additional investigation will also be required to establish whether dorsoventral coordination of transcription is complemented by similar coordination of translational and post-translational mechanisms . Neurons in the MEC encode representations of space [9] that are critical for spatial learning and memory [86] . An unresolved question is whether this computation requires a specialized cortical circuit , or whether it is an example of a generic computation to which canonical cortical circuits can easily be adapted . Evidence for the former comes from findings that in layer II , which contains the highest density of cells with grid firing fields , excitatory stellate cells are only able to communicate indirectly via inhibitory interneurons [87–89] , whereas in other cortical regions excitatory layer II principal neurons synapse with one another [90] . Consistent with this view our molecular analysis suggests considerable divergence between superficial layers of MEC and neocortex . In contrast , deeper layers of MEC appear much more similar to neocortex . Together with the dorsoventral organization of ion channel and axon ensheathment genes , our findings suggest that specialization important for spatial circuits is particularly striking within the superficial layers of the MEC . The functions within the MEC of the individual genes and functional gene groups that we identify as having laminar and dorsoventral organization have for the most part not been investigated and likely will be important targets for future exploration .
All animal experiments were carried out according to guidelines laid down by the University of Edinburgh’s Animal Welfare Committee and in accordance with the UK Animals ( Scientific Procedures ) Act 1986 . Brains were rapidly extracted from 13 male 8-week-old C57Bl/6JolaHsd mice and maintained in modified oxygenated artificial cerebrospinal fluid ( ACSF ) of the following composition ( mM ) : NaCl 86 , NaH2PO4 1 . 2 , KCl 2 . 5 , NaHCO3 25 , CaCl2 0 . 5 , MgCl2 7 , glucose 25 , sucrose 75 ) , at approximately 4ºC . One 400 µm thick sagittal slice containing the right MEC was cut from each brain using a Leica Vibratome VT1200 system [91] . Dorsal and ventral regions were microdissected under a dissection microscope ( S1A Fig . ) , with care taken to avoid inclusion of ventral entorhinal cortical regions , parasubicular or postrhinal regions and subicular regions . Tissue sections were collected into separate RNase-free eppendorf tubes before being quickly frozen on dry ice . Frozen tissue was stored at -80ºC for several weeks before RNA extraction . We compared RNA from dorsal and ventral MEC of 4 groups of mice . To minimize the effects of inter-animal variability and variability in the dissection , whilst maintaining sufficient power to detect dorsoventral differences , samples were pooled with 3 or 4 mice in each group . RNA was extracted using RNeasy Lipid Tissue Mini Kit ( Qiagen Cat:74804 ) . RNA integrity was assessed using a Agilent 2100 Bioanalyzer . All sample RINs were between 7 . 1 and 8 . 5 . cDNA was synthesized and amplified using the Ovation RNA-Seq System V2 ( NuGEN Cat:7102 ) using 120 ng of starting material for each sample . The samples were fragmented and sequenced by the Ark-Genomics facility using Illumina HiSeq with multiplexed paired-end analysis on two lanes . Raw data were processed using Casava 1 . 8 . Sequenced fragments were aligned using TopHat v2 . 0 . 8 . After sequencing and alignment , absolute RNA expression and differential expression were computed using Cuffdiff 2 software on the output BAM files [58] . We chose Cuffdiff 2 to ensure accurate counting of transcripts in the presence of alternatively splicing . Reported gene expression therefore reflects the summed expression of all transcripts/isoforms of a gene . Cuffdiff 2 was run on the Edinburgh Compute and Data Facility ( ECDF ) [92] cluster on 4 cores each with 2GB of RAM . The reference genome used was Ensembl 73 , downloaded 12th Nov 2013 . Transcripts were classified as expressed if their mean fragments per kilobase of exon per million fragments mapped ( FPKM ) [37] across samples ≥ 0 . 1 ( c . f . [16] ) in at least one of the dorsal or ventral regions ( Fig . 1D ) . We also only considered transcripts for inclusion if Cuffdiff 2 analysis revealed them to have a minimum number of 10 alignments in a locus ( default value ) . Transcripts were only tested for differential expression if mean FPKM across samples ≥ 1 . The steps for processing and extraction of data from ABA images are summarized in S1 Fig . and described in detail below . Code used in this section is available at https://github . com/MattNolanLab/Ramsden_MEC . Image download from ABA . Images were downloaded from the ABA database using the application programming interface ( API: http://www . brain-map . org/api/index . html ) . Since the ABA sagittal reference atlas begins at 3 . 925mm laterally , and as the MEC is located between approximately 3 . 125 and 3 . 5 mm laterally , images between 0 and 1400 μm ( refers to distance from most lateral point ) were selected for download for each image series . Two files were downloaded for each image: an ISH image file and a corresponding expression image file . Images were downloaded using the API files: http://www . brain-map . org/aba/api/imageseries/[enterimageseries] . xml and http://www . brain-map . org/aba/api/image ? zoom=3& top=0& left=0& width=6000& height=5000&mime=2&path=[path specified in xml file] . Images were approximately 500KB each . Approximately 120 , 000 images were downloaded in total and they were stored on a cluster provided by ECDF [92] . Preprocessing and cerebellar segmentation . ABA images , of variable dimensions , were first downsized by a factor of 1 . 25 and pasted onto the center of a new image of 1200 ( width ) x 900 pixels ( height ) using the Python Image Library . ISH images were then processed to improve image segmentation and registration ( S1A Fig . : steps 2–5 ) . No further changes were made to the expression image files until application of a segmentation mask ( step 6 ) . Image preprocessing proceeded as follows . ( a ) Background subtraction was carried out on the ISH images using ImageJ [93] ( S1A Fig . ) , with radius set to 1 pixel as this is approximately the size of a cell at the chosen resolution . ( b ) Images were thresholded using the ImageJ ‘Min_error’ automatic thresholding method such that all visible objects in the image , including anatomical features and cells with very low staining , were retained . The aim of this step was to minimize gene expression-specific information in the images whilst retaining anatomical detail to facilitate image registration based on landmark features . ( c ) To aid feature extraction , a smoothing filter ( ImageJ ) was applied to the images to smooth them prior to processing . Because the cerebellum could impair performance of the registration algorithm we developed an automated segmentation workflow to remove the cerebellar region from images prior to registration ( S1A Fig . ) . ( d ) An edge detection algorithm was applied to background-subtracted images ( FeatureJ Edge detection [94] ) . This image was thresholded to provide two outlined regions: the forebrain and cerebellum . These regions occasionally featured internal gaps caused by very low pixel intensity brain regions . We used an ImageJ algorithm to identify the two regions as objects ( defined by the complete perimeter ) and to fill in any such gaps ( ImageJ/Process/Binary/Fill Holes ) . These regions could then be detected as separate objects , using the ImageJ particle analysis tool ( ImageJ/Analyze/AnalyzeParticles ) , and only the largest object , corresponding to the forebrain , was subsequently included in the segmentation mask . This mask could then be applied to the ISH and corresponding expression images . Segmentation failed for images with low ISH labeling ( because of edge detection failures ) , where the cerebellum and forebrain overlapped ( due to mounting errors ) , and where erroneous staining prevented typical boundary detection . It was not feasible to examine all images and check those in which segmentation had failed , so we developed a method for automatically detecting successful segmentation . For each image within an image series , we used a binary support vector machine ( SVM ) classifier with a linear kernel to classify image masks based on success . To classify images , it is first necessary to extract features of the image that represent the patterns found in them . We used the VLFeat toolbox in Matlab [95] and custom-written code [96] to extract scale invariant feature transform ( SIFT ) features [97] from the images . The toolbox extracts SIFT features at 4 different scales to provide a spatial histogram that contains information about the positioning of features in space ( PHOW features ) . For each image a feature histogram containing 4000 values was generated for input to the classifier . The SIFT feature library was provided by [96] . We trained the SVM classifier on PHOW feature vectors from 800 correctly segmented images and 250 poorly segmented images and used a further 800 positive and 250 negative images for validation and tuning of the regularization parameter . The classifier was able to separate positive and negative validation images with over 98% accuracy with a tuned regularization parameter . We therefore used the SVM model with the same parameters to obtain a score for all remaining ( ~120 , 000 ) images that estimated their chance of success . The majority of images were assigned positive scores but we flagged any image with a score below 1 ( 11% of images ) as being potentially erroneous . Generation of reference images . To enable the extraction of information from 2D ISH images with precision , we generated reference images for five planes covering the medio-lateral extent of the MEC and its borders ( S1A Fig . ) . The central image ( C ) was our primary data extraction image , images in the adjacent lateral plane ( L1 ) supplemented this information , while images in more lateral ( L2 ) and in medial ( M1 and M2 ) planes were used for reference but not for data extraction . Reference images were generated using hand-selected ISH images that were chosen based on ( 1 ) relatively uniform expression in the MEC , ( 2 ) good tissue quality , and ( 3 ) medium ISH staining intensity . Approximately 15–20 images were chosen for each of the 5 reference images ( See Figs . 1A , S1A ) . Pre-processed images were rigidly aligned using an ImageJ plugin “Align Image by Line ROI” ( http://fiji . sc/Align_Image_by_line_ROI ) [93] . Given images in which the user has marked 2 corresponding points on each image , this plugin finds an optimal transformation ( translation , rotation , scale ) in closed form that aligns the images into the same location . Images then underwent group registration using a Matlab library , the Medical Image Registration Toolbox [98] ( Fig . 1A ) . Images were registered by two-dimensional non-linear deformation to one another , with the aim of finding the group match with the greatest similarity . We chose to group register 15–20 images to capture a sufficient degree of variance without requiring excessive memory ( ∼ 5GB RAM ) or time ( ∼ 20 hours ) . A Gaussian filter with a window size of 13 and a standard deviation of 3 was applied iteratively three times to each , followed by contrast enhancement , to enhance image structures at the relevant spatial scale . We chose to use cubic β-splines to represent the possible class of transforms and mutual information as the similarity measure , as it is relatively resistant to differences in contrast . The output of group registration is a series of transforms that correspond to each image . We generated reference images by applying these transforms and then calculating the median of the transformed images ( S1A Fig . ) . Classifying images based on their medio-lateral location . The images downloaded from the Allen Brain Atlas could be assigned either to one of the five reference image groups or to a sixth group for images not containing MEC . To ensure the ~120 , 000 images were appropriately classified based on medio-lateral extent , we used classification to identify , for each image series , the image most similar to our central reference image , ImrefC . We used a Support Vector Machine ( SVM ) library for Matlab [95] with a linear kernel and binary classification . The SVM provided a score for each image across all image series reflecting the chance that the image was approximately in the same medio-lateral plane as ImrefC . We could then compare scores for all images that had been downloaded for a given image series and choose the image with the highest score . Images from each image series that corresponded to more medial and lateral reference images could then be identified based on their relative medio-lateral location ( calculated using the ABA API database xml file corresponding to the relevant image series using the position and referenceatlasindex xml tags ) , since images within image series were always separated by 100 , 200 or 400 μm . The procedure for the SVM followed several stages . ( a ) A Gaussian filter was applied to preprocessed ISH images , with a window size of 15 and a standard deviation of 3 . ( b ) We trained the SVM classifier with PHOW features extracted from images manually classified for 501 genes into mediolateral groups . ( c ) We optimized the regularization parameter of the SVM and tested performance with a further 483 genes . After training , all the remaining images were run through the SVM and assigned a score . The highest scoring image from each image series was then assigned to an ImrefC folder for manual inspection . Any images that did not belong in ImrefC were removed . Images that were distantly located from ImrefC images were placed in a No-ML folder . We manually checked this folder and any images that appeared to match a reference image were moved to the appropriate folder . Records were kept of all movements and this process of checking continued until we were satisfied that images had been assigned to a reference with ∼95% accuracy . Registration gene images to a reference image . Each pre-processed ISH image was Gaussian-filtered and contrast-enhanced to facilitate extraction of large anatomical features and then 1-to-1 registered to its respective reference image , also Gaussian filtered , using the MIRT toolkit ( S1A Fig . ) . 1-to-1 registration is unlikely to be as accurate as group registration but group registration would be unfeasible , in terms of memory and time required , for the number of images involved ( ∼ 20 , 000 ) . Images were registered using cubic β-splines with mainly default settings , with the exception of the transformation regularization weight , which sets a limit on the scale of the deformation . We increased this from the default value of 0 . 01 to 0 . 1 to prevent large deformations . MI was used as a similarity measure because of its invariance to differences in contrast . The output of the algorithm was a transform describing the deformation of all points . Apply registration transformation to expression images . The transform , calculated using the thresholded images , was also applied to the original ISH images ( for visual assessment of registration success ) , as well as to the expression images ( for extracting pixel intensity ) . All images underwent the same procedure . Image quality check . To assess the accuracy of registration we used several measures automatically collected from all images: an MI-related score from the registration algorithm , a cross-correlation score on the final image , and a classification score from a classifier trained on poorly registered images . The MI score of the final deformation for each image reflects its similarity to the reference image . There is a clear distinction between the distribution of scores before and after registration ( S1A Fig . ) . To determine how well these scores represent correct alignment , we chose a random sample of 100 images and manually rated their registration accuracy , then plotted their scores against the final mutual information result . This allowed us to set a threshold so that we could flag images that were potentially poorly registered . A total of 17% of images in the central plane were flagged compared with 14% in the L1 plane . The MI score reflects registration accuracy across the whole image and therefore could overestimate the accuracy of registration in the MEC . Therefore as a second test we used the Matlab function ( normxcorr2 ) to cross-correlate the MEC-containing region of a registered image with a larger region of the respective reference image . This cross-correlation function provides both a normalized maximum fit score and the location of the maximum fit , thereby enabling us to estimate the offset between the posterior MEC border within the registered image and within the reference image . We validated scores by using the image set used for validation of MI analysis . The majority of images were given a high manual rating of 5 and had a cross-correlation offset of near zero . A total of 7% of central images were flagged as having an offset that was potentially too large . To detect image flaws including holes in the tissue created by bubbles , and aberrant detection of the pial surface as an RNA-expressing cell , which artificially increases the mean pixel intensity of the image , we used an image classifier . We decided to use image features to capture erroneous elements in the expression images that could subsequently be detected using an SVM . Features were extracted from the region including the MEC and immediate surround . We then trained a binary SVM classifier with a radial basis function kernel on all the images that we had visually assessed as having significant errors ( n=∼ 50 ) against high quality images ( n = 300 ) . We used the LIBSVM package in Matlab for this [99] . We used cross validation to optimize the regularization parameter , C , and the hyperparameter of the radial basis function , gamma . We then tested all images with the classifier , giving a probability estimate that each image was erroneous . We again compared the probability estimate with visual assessment of a subset of the images used previously to estimate accuracy for the MI score , and flagged images with scores greater than 0 . 13 . The classifier distinguishes images with large registration errors and pial surface errors from high-quality images ( S1A Fig . ) . However , this method is naive to anatomy and not particularly sensitive to minor misalignment errors as it extracts features from the entire MEC region that are scale and alignment-invariant . 29% of images were flagged based on these results . Images were assigned error statuses and defined as not meeting the quality criteria if they had poor MI scores or were flagged by at least 2 of the other error measures . Error statuses were updated for all visually assessed images . In summary , 15 , 447 / 20 , 032 ( 77% ) genes had images meeting these quality criteria in the central plane and 12 , 814 ( 64% ) had images meeting the criteria in the L1 plane ( S1B Fig . ) . Extraction of pixel intensities from ABA images . Custom python scripts were written for all analysis of 8-bit ABA expression TIFF images . Expression images were used instead of raw ISH images because stages of processing that control noise , background illumination and contrast invariance across the images have already been performed as part of development of the ABA [34 , 36 , 100] ( see ABA Informatics Data processing white paper ) . In addition , pixel intensity information represents overall expression level of individual cells that have been detected as expressing the gene of interest and should not contain structural information present in brightfield images that is not gene-specific , such as densely fibrous regions , Genes were classified as being expressed if the mean expression in either the custom-defined dorsal or ventral region was ≥ 1 ( scale up to 255 ) . All data presented are based on pixel intensity values from 8-bit grayscale expression images . Expression images have been shown with a 16-color lookup table for visualization purposes . Average images are shown either based on absolute intensity or intensity normalized by the mean of the MEC region , as indicated . Regional gene expression was estimated by manually outlining regions using Bezier lines ( ImageJ ) , using the selections to create a binary mask ( black on white pixels ) that could be imported using custom Python or Matlab scripts , then using the mask to select elements of the original expression images . Some genes are represented more than once in the ABA dataset , either because multiple probes have been used to detect different transcripts ( n = 351 / 20 , 334 genes ) , or where ISH experiments with a single probe have been replicated ( n = 1 , 011 genes ) . Where multiple probes are used to target a single gene we analyze images for each probe separately , but in population analyses we report the gene once whether only 1 probe or all probes were detected . For replications we found that relative intensities across different regions were similar in each image set , but overall image intensities could vary . We therefore analyzed average images generated by obtaining the mean relative pixel intensity across regions of interest for each image series and multiplying this by the mean pixel intensity of each whole image averaged across all relevant image series . We extracted pixel intensity information from central and adjacent lateral images , which showed highly similar patterns of gene expression . Comparisons in mean pixel intensity between corresponding MEC layers of central and adjacent lateral images showed correlations of at least 0 . 945 , which was higher than correlations with non-corresponding layers ( < 0 . 938 ) . When relative mean pixel intensities were compared between corresponding layers we found correlations of at least 0 . 66 , compared with < 0 . 21 for non-corresponding layers . Comparison of ABA and RNA-Seq expression . To compare the results from ABA data and RNA-Seq data , we first used the Ensembl Biomart database to match Ensembl data from RNA-Seq to Entrez IDs and Gene symbols . To correlate ABA and RNA-Seq expression , we took the mean pixel intensity of the dorsal and ventral regions , averaged across the ImrefC and ImrefL1 planes ( where images were available ) and compared this to the mean FPKM of RNA-Seq dorsal and ventral samples ( S1A Fig . ) . We describe below methods used for analyses associated with each main figure . Pearson correlation coefficients and linear regression analyses were performed using the statistics linear regression package , lm , in R . We define absolute intensity as the pixel intensity measurement in 8-bit images ( range 0–255 ) , and relative intensity as a measure reflecting the ratio of pixel intensities between two or more regions , for example layers or brain regions . ABA regional comparisons and combinatorial analysis ( Fig . 2 ) . The neocortical , hippocampal , caudate putamen , amygdala , piriform and MEC regions were manually outlined using the central composite reference image and Allen Mouse Reference Atlas [34 , 36] as guides . Genes with a mean pixel intensity in MEC ≥ 5 were defined as being MEC-enriched if their relative intensity compared to the comparison regions was ≥ 0 . 8 . Genes with mean pixel intensity in MEC < 5 and ≥ 2 were defined as being MEC-enriched if their relative intensity was ≥ 0 . 99 . In both cases the mean intensity of the compared brain region also had to be < 5 . The proportion of MEC-enriched genes also expressed in the other brain regions was calculated by finding genes that did not satisfy these criteria . MEC-unique genes were identified using the same thresholds applied in comparison to all regions . We identified pairs of genes with overlapping expression by first finding all genes that are enriched in the MEC relative to at least one other brain region and pairing them with each other . We then identified those pairs for which at least one of the pair was present in all five MEC-enriched lists . Images of genes were overlaid using ImageJ and manually inspected for degree of overlap . To aid visual assessment , genes with MEC expression < 5 were only included if expression was restricted to a particular subregion , as uniform expression at this intensity appears to most likely reflect non-specific ISH staining or uneven illumination across the tissue . Detection of borders ( Fig . 3 ) . To identify genes defining the dorsal and ventral borders of MEC , we outlined regions dorsal and ventral to the approximate location of the borders using the central reference image ( Fig . 3A , 3F ) . All gene images meeting the quality criteria in the re-registered data set with mean pixel intensity < 15 in one region and with a differential pixel intensity of > 15 between the regions were selected for manual validation . We supplemented this search with an ABA differential expression search [34 , 36] ( Target structure: Parasubiculum , Contrast structure: Medial Entorhinal Cortex ) , with an expression threshold of 3 . 5 , which identified an additional 8 genes that demarcated the dorsal border of the MEC . Identification of layer-specific and differentially expressed ( DE ) genes ( Fig . 4 ) . Genes were defined as layer-specific if they show no consistent expression in other MEC layers or differentially expressed ( DE ) if they show substantially higher expression in at least one layer than in another . Exact criteria for identification of layer-specific and DE genes are as follows . ( 1 ) Comparison of laminar regions within the re-registered ABA dataset . Layers were defined using the composite central and adjacent lateral reference images ( S4A Fig . ) . We primarily used data from the central reference image , only using information from the lateral plane when no central image was available . Since layers possess multiple types of cells that may themselves differentially express genes , we compared both average expression across layers , and the distribution of high-intensity expression . For each image a high-intensity pixel was defined as having intensity ≥ mean + 2 x S . D of all pixels in the MEC . The absolute pixel intensity ( Lxabs = Lxrel x 3 x MECmean , where x refers to layer ) and proportion of high-intensity pixels were then calculated for each of the layers II , III and V/VI . Relative laminar mean intensity ( Lxrel ) and relative proportion of high-intensity pixels ( Lxprop ) were calculated by dividing layer measures by their sum . We also identified the first ( Lmax = maximum of ( LIIrel + LIIprop , LIIIrel + LIIIprop , LVrel + LVprop ) ) and second highest expressing layers ( Lmid ( not Lmin or Lmax ) ) . We then calculated an absolute mean pixel intensity difference between these layers ( Lmaxdiff = ( ( Lmaxrel-Lmidrel ) 2 x Lmaxabs / 255 ) x 30 ) and the joint expression of the two highest expressing layers ( Lmaxabsjoint = ( ( Lmaxrel + Lmidrel−Lminrel ) 2 x Lmaxabs / 255 ) x 30 ) . These values acted to penalize low absolute pixel intensities . We did not initially distinguish layers V and VI as the border between these layers is not clearly defined in the sagittal plane . Only genes with a mean overall pixel intensity ≥ 1 and mean pixel intensity ≥ 2 in at least one layer were evaluated . To quantify laminar differences in expression we calculated patterning scores ( PS ) as follows: We assigned weights to the measures relative intensity ( wrel = 0 . 4 ) , relative proportion of high-intensity pixels ( wprop = 0 . 5 ) and absolute pixel intensity ( wabs = 1- ( wrel + wprop ) ) . If the image had no high-intensity pixels , wrel = 0 . 9 . We calculated a PS for single layer enrichment: PSsingle = wrel x Lmaxrel + wprop x Lmaxprop + wabs x min ( [Lmaxabs , 1] ) We calculated a PS for joint layer enrichment: PSjoint = wmean x ( Lmaxrel + Lmidrel ) + wprop x ( Lmaxprop + Lmidprop ) + wabs x min ( [Lmaxabsjoint , 1] ) Genes with PSsingle ≥ 0 . 65 or PSjoint ≥ 0 . 88 ( n = 1 , 314 ) were marked as candidates for DE genes , including the subset of layer-specific genes . ( 2 ) Cross correlation of genes within the registered data set . We also used the re-registered data set to find genes with similar patterns of intensity to genes identified through differential laminar expression , but that may have been missed in the previous analysis due to their occupation of very small areas . Using a SciPy cross-correlation function ( Alistair Muldal:https://github . com/oleg-alexandrov/projects/blob/master/fft_match/norm_xcorr . py ) , we used Nxph4 as the seed gene to find other layer VI genes and Mrg1 as the seed to find other potential island genes . To identify genes only expressed in the narrow VI layer , we compared a small dorsal region and searched only images with mean pixel intensity ≥ 1 and < 5 that had a sum of squares difference ( SSD ) of zero with the target gene . We checked images for the 30 genes with the highest cross-correlation . For island genes , we searched using a small dorsal region including layer II and checked the top 50 genes with mean intensity ≥ 1 and SSD = 0 . ( 3 ) Identification of genes through ABA search tools . Since our aim was to make this resource as comprehensive as possible , we extended our search beyond our re-registered data set to make use of ABA differential expression tools and knowledge of cortex-enriched genes [34 , 101] . We initially identified strongly differentially expressed ‘seed’ genes through manual exploration using the ABA differential expression tool , Fine structure Annotation and Anatomical Gene Expression Atlas ( AGEA; http://mouse . brain-map . org/agea ) ( total checked = 922 ) . Taking at least 8 genes with the strongest expression for each layer , we used the ABA NeuroBlast tool to identify all other genes with an expression correlation of at least 0 . 5 with any one of these ‘seed’ genes in the retrohippocampal ( RHP ) region . This provided us with over 4 , 000 potential DE genes , but no clear indication of their laminar expression profile . We visually assessed all those that did not have an image meeting the quality criteria in our re-registered ABA data set ( n = 959 ) . For the identification of layer-specific genes we also scanned all those that our analysis suggested had borderline differential expression ( n = 163 ) or that had a NeuroBlast expression correlation greater than 0 . 7 ( n = 1 , 288 ) . Given the potential for images to be poorly registered in both ABA and our re-registered data set , we also visually inspected differentially expressed neocortical genes ( n = 302 ) using lists provided by [101] ( http://www . nature . com/nrn/journal/v8/n6/suppinfo/nrn2151 . html ) and those annotated as having a ‘high’ specificity score in the Somatosensory cortex Annotation on the ABA website ( http://help . brain-map . org/download/attachments/2818169/SomatosensoryAnnotation . xls ? version=1&modificationDate=1319171046372 ) [34 , 36] to determine whether they also showed laminar specificity in MEC . ( 4 ) Visual validation . To minimize the number of false positives in the data , all candidate layer-specific and DE genes were validated through visual inspection . Layer-specific genes were confirmed as showing consistent expression in a single layer using the original ISH image and , where possible , images in more lateral and medial planes . Genes with DE expression had to show consistently higher expression in at least one layer than in another . For ABA re-registered data , 703 / 1314 candidates could be added to the set of DE genes . An additional 4 genes with layer-specific expression were identified using cross-correlation and an additional 36 DE including 10 layer-specific genes using the cortex-enriched lists referred to above . The NeuroBlast data identified an additional 121 DE genes , 13 of which were layer-specific . During manual validation of DE genes , we also recorded particular patterns of gene expression , including island or inter-island expression and specific laminar expression within the deep layers . To generate final lists of layer-specific genes , we included all genes visually validated as layer-specific , independent of PS score . For DE genes , we included all layer-specific genes , all genes in the re-registered ABA data set that had a single layer PS ≥ 0 . 65 or joint PS ≥ 0 . 88 and that were visually assessed as strongly differentially expressed , and all genes acquired using ABA tools and cortex layer-enriched lists that we validated as showing differential expression . See S4B Fig . Layer-specific and DE genes showed consistent expression patterns across mediolateral sections ( S6 Fig . ) . Analysis of laminar similarities and differences between MEC and neocortex in ABA data ( Fig . 5 ) . Taking the neocortical region used in our analysis of regional expression , we associated all pixel intensities for each gene image with a normalized location relative to the corpus callosum ( or subicular border for MEC ) and the nearest point along the pial surface . For all MEC layer-specific genes ( S4C Fig . ) , we calculated mean pixel intensities at different normalized locations throughout the three cortical regions ( Fig . 5B ) . We plotted histograms by binning the distances into 20 regions . Statistically significant differences in the laminar gene list expression patterns were detected using Mixed Model Analysis in SPSS ( v21 ) with an unstructured covariance matrix . Fixed effects were the list of genes , location and their interaction . Random effects were the list of genes and location with image series as subject . To test whether deep and superficial layer-specific expression patterns correspond between MEC and neocortex , we used genes previously identified as having ‘high’ specificity in SS cortical layers [34] ( S5E Fig . ) to divide the cortical regions into deep ( layers V/VI ) and superficial ( II-IV ) regions ( cf . [101] ) . We also used these genes to estimate an approximate border between visual and SS cortex . For each gene with mean pixel intensity ≥ 2 , we calculated the ratio of pixel intensity in the deep region to the superficial region . If gene expression in MEC and visual or SS cortex corresponds , we would expect this ratio to be 1 for MEC deep layer-specific genes and 0 for MEC superficial layer genes . To test this prediction , we subtracted the ratio for each gene from the expected value and performed a MANOVA using SPSS ( v21 ) with type of MEC specificity as the between-subjects variable and the visual and SS results as dependent variables . Post-hoc tests were performed with Tukey’s HSD . The percentage of genes that are enriched in superficial or deep regions was calculated by including genes with a deep: superficial ratio of less than 0 . 4 or greater than 0 . 6 ( 50% difference ) . Genes with mean intensity < 2 in the neocortical region were not included in the analysis . To calculate the correlations between cortical layers across all ABA genes , we used the estimated laminar boundaries described previously to generate binary array masks corresponding to each layer . Pixel intensities were averaged across all pixels within laminar boundaries then Pearson correlation coefficients calculated . Functional analysis for genes with laminar organization ( Fig . 6 ) . We used the GOElite gene ontology tool [102] with recent versions of the GO OBO database and gene annotation file for mus musculus ( 24/1/2014 ) and Kyoto Encyclopaedia of Genes and Genomes ( KEGG ) database [63] to extract all enriched terms . For the reference set for DE genes , all DE genes and genes with weak differential or uniform expression and threshold mean intensity ≥ 2 in MEC ( n = 9 , 057 Ensembl identifiers ) were included . This ensured we were not simply selecting for brain-enriched genes but for genes differentially expressed within the MEC . Terms were defined as enriched if associated with a p value < 0 . 05 after a one-sided Fisher overrepresentation test followed by Benjamini-Hochberg false discovery rate adjustment for multiple tests . To reduce redundancy and identify clusters of meaningful GO and KEGG terms we calculated a kappa similarity measure ( described in [103] ) to identify terms sharing a higher proportion of genes than chance . We then used hierarchical clustering ( adapted from code written by Nathan Salomonis: http://code . activestate . com/recipes/578175-hierarchical-clustering-heatmap-python/ ) on the kappa similarity matrix to cluster terms with at least 5 genes and fewer than 100 genes into groups . Within each cluster we then extracted all kappa similarity scores . Only terms with less than modest overlap ( kappa < 0 . 7 ) with a more significant term and fewer than 100 genes were presented in the summary figure ( Fig . 6A ) . Clusters were labelled using their most significant term . To establish the significance of the over or under representation of layer-specific genes in the selected lists we used R implementation of a 2-way Exact Fisher test ( fisher . test ) followed by multiple corrections analysis ( p . adjust . M ) . In heatplots , data are shown for images in the plane corresponding to the central reference image , where available , or for the adjacent lateral plane . Investigating dorsoventral differences in gene expression ( Fig . 7 ) . For analysis of RNA-Seq data Cuffdiff 2 [58] was used to identify all differentially expressed genes with an FPKM of at least 1 and difference in expression of at least 20% ( log2 ( 1 . 2 ) = 0 . 2630 ) . Differences in the direction of differential expression between genes with different layer-specific expression ( Fig . 4 ) were detected using ANOVA followed by post-hoc Tukey’s HSD tests , performed in R . For analysis of ABA images dorsal and ventral regions were manually outlined using the two primary custom reference atlases . Average pixel intensities within each region and across central and the adjacent lateral sections containing MEC were calculated for each ABA image series . Only ABA images with a threshold mean intensity ≥ 2 in the dorsal or ventral region were included in the differential expression analysis . We used a threshold log fold enrichment of 0 . 2630 ( increase of 20% ) to define differential dorsoventral expression , the same value used for RNA-Seq . For comparison between Cuffdiff 2 analysis and ABA analysis , Cuffdiff 2 results were matched to ABA Entrez values or gene official symbols using the Ensembl Biomart tool . Two comparisons were made: one specific to ABA layer-specific genes , and the other only including genes determined to be significantly differentially expressed according to RNA-Seq data . Functional analysis for dorsoventral genes ( Fig . 8 ) . A similar method was used to that described above for laminar genes . For the reference set we included all genes with an FPKM ≥ 1 in MEC that had a sufficient number of alignments in a locus ( measured using Cuffdiff 2 ) to be tested for differential expression ( Ensembl identifiers = 13 , 954 ) . To estimate the presence of gradients in MEC in re-registered ABA data , we divided the MEC region in both the central and adjacent lateral planes into 5 subregions along the dorsoventral axis . For each layer and subregion , we calculated the mean pixel intensity and used these values to calculate linear regression gradients ( SciPy stats . linregress ) for each layer and across the whole MEC . For heatmaps , genes were sorted according to this gradient and color-coded according to whether their dorsoventral difference was sufficiently large for them to be defined as D>V or V>D-expressed in the analysis performed in Fig . 6 . Disease analysis of layer-patterned genes ( Fig . 9 ) . Genes involved in pathways for Alzheimer’s disease , Huntington’s disease and Parkinson’s disease were acquired from the KEGG pathway database [63] . This tool contains manually drawn pathways of molecules known to be perturbed , either through environment or genes , in certain diseases , as well as molecules that are therapeutic markers or diagnostic markers . Autism-related genes include mouse genes for which the human homolog has a score of 1–4 ( high-low evidence ) or syndromic in the SFARI AutDB database ( n = 238 ( Ensembl ) ) . Schizophrenia genes include the results of a computational analysis of meta-analysis results from [64 , 104] ( n = 175 ( Ensembl ) ) . Epilepsy genes include those associated with any form of epilepsy in the Disease database ( n = 90 ( Ensembl ) ) [67] . Genes that have been causally associated with early or late-onset AD were obtained from key reviews and meta-analyses in the AD literature [69 , 72 , 105] . To generate lists of layer-enriched genes we took all DE genes and identified those visually validated as being enriched in particular layers . We computed the significance of over or underrepresentation of disease genes amongst layer-enriched genes using Fisher’s Exact Test in R , as described for gene ontologies in Fig . 6 . To determine the associated ontology terms of all AD pathway genes with at least moderate layer enrichment we used a PSsingle threshold of 0 . 60 .
|
Higher brain functions such as spatial cognition are carried out in specialized brain areas . Within a specialized brain area nerve cells with different functions are organized in layers and gradients . It is possible that this topographical organization reflects underlying differences in molecular organization of the brain . However , systematic comparison of the expression patterns of tens of thousands of genes at the resolution of layers and borders is challenging . Here we develop a new computational pipeline that addresses this problem . We apply this pipeline to analysis of the medial entorhinal cortex ( MEC ) , a brain structure that is important for spatial cognition . Our analysis shows that the MEC is highly organized at a molecular level , identifies related groups of genes that might underlie functional specialization , and implicates energy-related genes in vulnerability of certain neuronal populations to neurological disorders including Alzheimer’s disease . Our computational pipeline may have general utility for high-throughput and high-resolution analysis of brain anatomy . Our results support the notion that molecular differences contribute to functional specialization of higher cognitive circuits .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Laminar and Dorsoventral Molecular Organization of the Medial Entorhinal Cortex Revealed by Large-scale Anatomical Analysis of Gene Expression
|
Recorded potentials in the extracellular space ( ECS ) of the brain is a standard measure of population activity in neural tissue . Computational models that simulate the relationship between the ECS potential and its underlying neurophysiological processes are commonly used in the interpretation of such measurements . Standard methods , such as volume-conductor theory and current-source density theory , assume that diffusion has a negligible effect on the ECS potential , at least in the range of frequencies picked up by most recording systems . This assumption remains to be verified . We here present a hybrid simulation framework that accounts for diffusive effects on the ECS potential . The framework uses ( 1 ) the NEURON simulator to compute the activity and ionic output currents from multicompartmental neuron models , and ( 2 ) the electrodiffusive Kirchhoff-Nernst-Planck framework to simulate the resulting dynamics of the potential and ion concentrations in the ECS , accounting for the effect of electrical migration as well as diffusion . Using this framework , we explore the effect that ECS diffusion has on the electrical potential surrounding a small population of 10 pyramidal neurons . The neural model was tuned so that simulations over ∼100 seconds of biological time led to shifts in ECS concentrations by a few millimolars , similar to what has been seen in experiments . By comparing simulations where ECS diffusion was absent with simulations where ECS diffusion was included , we made the following key findings: ( i ) ECS diffusion shifted the local potential by up to ∼0 . 2 mV . ( ii ) The power spectral density ( PSD ) of the diffusion-evoked potential shifts followed a 1/f2 power law . ( iii ) Diffusion effects dominated the PSD of the ECS potential for frequencies up to several hertz . In scenarios with large , but physiologically realistic ECS concentration gradients , diffusion was thus found to affect the ECS potential well within the frequency range picked up in experimental recordings .
The number of ions exchanged between neurons and the extracellular space ( ECS ) during a brief period of activity ( i . e . , due to the integration of synaptic input and generation of a few action potentials ) is typically too small to evoke significant changes in extracellular ion concentrations . In models of short-term electrical signalling of neurons , the ion concentrations of the main charge carriers ( e . g . , K+ , Na+ , Cl- ) are therefore commonly assumed to remain effectively constant . This assumption often holds also at longer time scales , due to the work done by neuronal and glial uptake mechanisms in maintaining ion concentrations close to baseline levels . However , during periods of intense neural signalling , the uptake mechanisms may fail to keep up , and ion concentrations in the ECS may change by several millimolars [1–5] . For example , the extracellular K+ concentration can increase from a typical baseline level of around 3 mM and up to levels between 8 and 12 mM during non-pathological conditions [4 , 6–8] . Ion-concentration shifts in the ECS will change neuronal reversal potentials and firing patterns [9–12] , and too large deviations from baseline levels can lead to pathological conditions such as hypoxia , anoxia , ischemia , epilepsy and spreading depression [9 , 13–15] . One of the most common experimental methods for investigating neural activity is the measurement of electrical potentials with extracellular electrodes . Commonly , it is assumed that extracellular potentials predominantly reflect transmembrane cellular current sources , including synaptic currents and currents through active and passive membrane mechanisms in neurons and glial cells [16–19] . However , in scenarios where ECS concentration gradients become sufficiently large , electrical currents carried by diffusing ions in the ECS could in principle also give measurable effects on the extracellular electrical potentials ( cf . , liquid junction potentials [20–22] ) . In support of this , local ion-concentration changes in the ECS are indeed often accompanied by slow local negative potential shifts , which can be on the order of a few millivolts [1 , 3 , 13 , 23–27] . Whereas K+ buffering currents through the glia-cell membranes are believed to be the main source of these slow potential shifts [3 , 7] , it has been estimated that also diffusive currents along extracellular concentration gradients could contribute by shifting ECS potentials by up to 0 . 4 mV [3] . As ion concentrations in the ECS typically vary on the time scale of seconds [3 , 4 , 28] , it is nevertheless a priori unclear whether diffusion-evoked potential shifts would be picked up by the electrode measurement systems applied in most experiments , which typically have cut-off frequencies of about 0 . 1–0 . 2 Hz or higher ( see e . g . , [29 , 30] ) . In most computational studies of ECS potentials , diffusive currents in the ECS are assumed to be negligible compared to the currents propelled by the electrical field ( hereby termed field currents ) . This is , for example , an underlying assumption in volume-conductor theory which has been the basis for estimating ECS potentials from cellular current sources [18 , 31–36] , and in estimation of current-source density ( CSD ) which predicts transmembrane neural current sources from recordings of extracellular potentials [29 , 34 , 37–40] . Another series of theoretical studies have aimed to incorporate possible effects of diffusion in the complex impedance environment of the extracellular medium [41–44] , and have suggested that such effects may account for the 1/f-scaling observed for the LFP-power spectrum at low frequencies [41] . In neither of the above mentioned studies , however , ionic diffusion was explicitly modelled . The reason why diffusive effects are often neglected in models of extracellular fields , may be that the task of modelling it is challenging . This is because the study of diffusion requires an explicit tracking of all present ions and their spatiotemporal dynamics: i . e . , keeping track of only the electric currents and net electric charges is not sufficient . Existing electrodiffusive models have typically been based on the Poisson-Nernst-Planck ( PNP ) formalism [45–51] . The PNP formalism explicitly models charge-relaxation processes , which occur at spatiotemporal scales on the order of nanometers and nanoseconds . This requires an extremely high spatiotemporal resolution , which makes PNP models computationally expensive and unsuited for predictions at the tissue/population level [52] . However , a series of modelling schemes have been developed that circumvent the charge relaxation processes , essentially by replacing Poisson’s equation by the constraint that the bulk solution is electroneutral [28 , 52–58] . The electroneutrality condition is a physical constraint valid at a larger spatiotemporal scale , and thus allows for a dramatic increase in the spatial and temporal grid sizes in the numerical simulations . One of these simpler models were previously developed by our group [28 , 57] , and is here referred to as the Kirchhoff-Nernst-Planck ( KNP ) scheme . The KNP scheme is a means of deriving the local potential in the intra- and extracellular bulk solution from the constraint that Kirchhoff’s current law should be fulfilled for all finite volumes ( the sum of currents into a finite subvolume of bulk solution should be zero ) . In the current work we have developed a hybrid modelling formalism that allows us to compute electrodiffusive ion dynamics in the ECS surrounding active neurons . The formalism is briefly summarized in Fig 1 . First , it utilizes the NEURON simulator [59 , 60] , which is a standard tool for simulating morphologically complex neurons , to simulate the activity of a neural population and its exchange of ions with the ECS ( Fig 1A ) . Second , it utilizes the KNP formalism [28 , 57] to compute the dynamics of ion concentrations and the electrical potential in the ECS surrounding the neurons ( Fig 1B ) . The KNP scheme accounts for all electrical currents entering an ECS subvolume in the system ( i . e . , transmembrane ionic currents , transmembrane capacitive currents , diffusive currents through the ECS , and field currents through the ECS ) , as well as for concentration-dependent variations in the ECS conductivity ( see Methods ) . It computes the ECS potential from the constraint that all currents into a ECS subvolume should sum to zero ( Fig 1C ) . In this way , the KNP-scheme accounts for effects of ionic diffusion on the ECS potential , and thus differs from previous simulation schemes for computing ECS potentials based on output from standard neuron simulators such as NEURON ( e . g . [61] ) . We have here used the hybrid scheme to model a small tissue element consisting of a population of ten pyramidal neurons embedded in ECS ( Fig 1A ) . Motivated by the layered structures of cortex and hippocampus , we assumed lateral homogeneity , so that all spatial variation occurred in the vertical direction . As neuronal model , we used a well established multicompartmental model of pyramidal cells [62] . The ten neurons were centered at the same depth level of the tissue . Although the choice of neuronal model was somewhat arbitrary , and the small tissue element was too simple to represent any particular biological system , the model gave rise to biologically realistic variations in ECS concentrations , and we regard it as a meaningful scenario for which we could explore how diffusive currents in the ECS can influence extracellular potentials . In simulations that evoked large , but not pathologically large , concentration gradients in the ECS , we found that diffusion gave rise to a detectable 1/f2 power law in the low-frequency part of the power spectral density ( PSD ) of the ECS potential . Furthermore , we found that diffusion influenced the PSD for frequencies as high as 1–10 Hz . This quantitative prediction was , of course , specific to the particular model setup used here . Although the relative effects of diffusion may be smaller in many realistic , more complex scenarios ( see Discussion ) , we regard our findings as an important demonstration that in general , diffusive currents can not by default be assumed to have a negligible impact on ECS potentials . The article is organized as follows: In the Results section , we use the KNP scheme to explore the role of diffusive currents on electrical potentials in the ECS surrounding a population of pyramidal neurons . In the Discussion section , we discuss possible implications that our findings will have for the interpretation of data from extracellular recordings . The Discussion also includes an overview of the assumptions made in the presented model , and on how the framework can be expanded to allow for more thorough investigations of concentration-dependent effects on ion dynamics in neural tissue . A detailed derivation of the KNP-formalism is postponed to the Methods section ( which is found at the end of the article ) .
Ten pyramidal neurons were simulated by running ten independent simulations on a single neuron model . As neuron model , we used a well established model developed for cortical layer 5 pyramidal cells [62] . Each neuron was driven by uncorrelated Poissonian input spike trains ( with the same statistics for all neurons ) through 10 , 000 synapses . Synapses were uniformly distributed over the membrane area ( sections with equal membrane area had the same expected number of synapses ) , and synaptic weights were tuned so that the average single-neuron action potential ( AP ) firing rate was about five APs per second ( this is within the range of typical firing frequencies observed for cortical neurons [63] ) . As illustrated in Fig 1 , a piece of tissue was subdivided vertically into 15 depth intervals ( here referred to as ECS subvolumes ) , which we could picture as spanning from the bottom to the top layer of a layered structure such as cortex or hippocampus . The neurons were positioned so that they occupied the 13 interior subvolumes . The output from all neural segments contained in a specific subvolume were summed , and this gave the total output into the given subvolume . In the neuronal output signal we kept separate track of the different kinds of transmembrane currents , including ( i ) the net Na+ current , ( ii ) the net K+ current , ( iii ) the net Ca2+ current , ( iv ) non-specific ionic currents , and ( v ) the capacitive current . For simplicity , we assumed that all unspecified ionic currents in the model [62] ( such as leakage currents , synaptic currents , and currents through non-specific active ion channels ) were carried by a single , non-specified anion species X- . We chose to use an anion , because many of the non-specified currents are likely to be mediated largely by Cl- ( for further comments on this choice , see Methods and Discussion ) . The output from the neural population into three selected ECS subvolumes is shown in Fig 2 for the first seven seconds of the simulation . For example , Fig 2A shows the currents into the subvolume ( n = 3 ) containing the somata . Here , we clearly see the brief Na+ ( Fig 2A1 ) and K+ ( Fig 2A2 ) current pulses associated with neuronal AP firing . The current amplitudes were about -30 nA ( inward , depolarizing current ) for Na+ and 30 nA ( outward , repolarizing current ) for K+ . Generally , the subvolume containing the somata received a higher influx/efflux of ions ( Fig 2A ) compared to the subvolumes containing the apical trunk ( Fig 2B ) and apical branches ( Fig 2C ) . These differences have two explanations: First , the somata subvolume contained a larger proportion of the total neuronal membrane area , which generally enhanced the ionic exchange in this subvolume . ( Similarly , currents are larger in Fig 2C compared to Fig 2B because the subvolume where the apical dendrites branched out contained a larger membrane area than subvolumes containing a part of the apical dendritic trunk . ) Secondly , the somata also had a higher density of Na+ and K+ channels than the dendrites . Accordingly , almost all exchange of Na+ and K+ between the neurons and the ECS occurred in the soma subvolume ( compare somatic output in Fig 2A1 and 2A2 to the total neuronal output current in Fig 2D1 and 2D2 ) . For the other ions ( Ca2+ and X- ) , the dendrites contributed with a larger proportion of the total output . As we just saw , the neurodynamics fluctuated vividly on the millisecond time scale . However , the input statistics was the same throughout the simulation , so that the slow time-scale neurodynamics was essentially stationary ( see Methods ) . To illustrate this , we split the seven seconds of neural simulations shown in Fig 2 into five 1 . 4 second time intervals , and averaged the total transmembrane current ( IM ) over the five respective intervals . Fig 3 shows how the ( temporally averaged ) transmembrane sources were distributed across tissue depth . The spatial profile of IM was essentially independent of which 1 . 4 second interval of activity it was averaged over . The main current source ( positive transmembrane current , i . e . , net positive charge leaving the neurons ) was found in the soma subvolume ( n = 3 ) . The main current sinks ( negative transmembrane current , i . e . , net positive charge entering the neurons ) were found in subvolumes containing proximal apical dendrites ( n = 5 , 6 ) and distal , branching apical dendrites ( n = 12 , 13 , 14 ) . We note that the transmembrane current profile summed to zero across depth , meaning that the sinks and sources balanced each others out ( no neuron can be a net current sink nor source ) . Of course , the neurodynamics and source/sink configurations seen in Figs 2 and 3 depended in a complex way on the particular neuronal morphology and the subcellular distribution of membrane mechanisms and synapses used in the simulations . The main objective of this work was , however , not to analyze these dependencies , but rather to explore how the ECS potential surrounding the neuronal population depended on whether diffusion was included in the simulations of the ECS dynamics . We investigated this for the particular scenario summarized in Figs 2 and 3 , which was used in all simulations shown in the following , but with 84 seconds of simulated neurodynamics , and not only the seven seconds depicted in the figures . We note that the neuron model by Hay et al . exhibited a rich repertoire of firing properties , including the occasional dendritic Ca2+ spikes seen in Fig 2B3 . We refer to the original work for further details on the model properties [62] . In the following , the focus will be on how the simulated ECS potential ( surrounding this given system ) depend on whether ECS diffusion is accounted for . Knowing the neuronal output to each ECS subvolume , we used the KNP-formalism to compute the resulting dynamics of ionic concentrations and the electrical potential in the ECS . Typically , ECS potentials are thought to mainly originate from various transmembrane current sources [16 , 17] . Here , we explored whether diffusive currents in the ECS could constitute an additional source . Fig 4A1–4A4 illustrates the dynamics of the ECS potential in two selected subvolumes ( soma , n = 3 , solid line; apical dendrite , n = 13 , dashed line ) due to the neuronal activity shown in Fig 2 . Similarly , Fig 4B and 4C show the field currents and diffusive currents ( respectively ) from subvolume n = 3 to n = 4 ( solid line ) and from subvolume n = 13 to n = 14 ( dashed line ) . For simplicity , we in the following discussion refer to the current from n = 3 to n = 4 as the current out from the soma subvolume , and the current from n = 13 to n = 14 as the current out from the apical dendrite subvolume . The first column ( 1 ) of Fig 4 shows the time course of these variables over the full simulation , while the remaining columns ( 2–4 ) show the time course over selected , shorter ( 40 ms ) time intervals , which include only a few neuronal APs . Red curves represent the scenario without diffusion in the ECS simulations , while blue curves represent the scenario with ECS diffusion included . When we explore the extracellular AP signatures ( panels A2–4 ) , we see that they had the same time course as the field currents ( panels B2–4 ) , while diffusive currents varied little at this fast time scale ( panels C2–4 ) . Diffusive currents thus had no impact on the fast temporal dynamics , and the AP signatures resembled those previously studied in models based on volume-conductor theory , where diffusive currents are neglected [34] . Somatic AP generation was due to an inward ( depolarizing ) current into the neuron followed by an outward ( repolarizing current ) . Since the sum of transmembrane currents over the neuron as a whole ( all ionic + capacitive currents ) must be zero at all times , the dendritic branches experienced the opposite current configuration during the APs ( outward currents followed by inward currents ) . Therefore , AP signatures in the apical ECS subvolume ( dashed lines in Fig 4A2–4A4 ) had the opposite temporal profiles compared to what we observed in the soma subvolume ( solid lines in panels A2–4 ) . Although the AP signatures were of the same order of magnitude in the soma and apical subvolumes , ECS field currents out of the soma subvolume were generally much larger than field currents between neighboring dendritic subvolumes ( panels B2–4 ) . The explanation lies in the spatial distribution of transmembrane inward and outward currents , and the rather unique role played by the soma . For example , a local inward current to the soma returned to the ECS in a widespread manner , i . e . , it was distributed over the entire dendritic tree . Neighboring dendritic subvolumes therefore had similar AP signatures , implying that the ECS voltage differences ( and therefore the field currents ) between them were small . The diffusive currents varied at a much slower time scale compared to field currents ( Fig 4C ) . This was due to the slow time scale at which ion concentrations varied ( as we shall explore further below ) . The diffusive current out of the soma region reached a peak value after around 30 seconds , after which it decreased slowly . The concentration build-up was slower in the subvolumes containing apical dendrites , and diffusive currents were smaller there , and still increasing at the end of the simulation ( panel C1 ) . In the early part of the simulation , when diffusive currents were small , the ECS potential V was close to identical in the cases with and without diffusion ( panel A2 ) . However , as diffusive currents built up , they did have an effect on V , which was shifted to more negative values in the simulation with diffusion included compared to case without ECS diffusion ( panel A3–A4 ) . Towards the end of the simulation , diffusion had shifted V by about -0 . 2 mV in the soma subvolume . In the following , we shall explore this process in further detail . Diffusive currents in the ECS are proportional to concentration gradients in the ECS . To gain insight in the slow dynamics of the diffusive currents , we must therefore investigate the ECS ion-concentration dynamics . In our simulations , ECS concentrations varied due to ionic output from the neurons . Fig 5 shows how the ECS concentration varied over the tissue depth at selected time points . The deviations from the initial concentrations became gradually larger throughout the 84 second simulation , illustrating the slow time scale of ion-concentration dynamics in the ECS . When diffusion was not included in the ECS simulations ( Fig 5A ) , ionic transports were solely due to electrical migration , and were not biased towards following concentration gradients of distinct species . In this case , the ECS concentration profiles predominantly reflected the distribution of neuronal sources . For example , somatic AP generation caused a sharp decrease in the Na+ concentration and a corresponding increase in the K+ concentration in the soma subvolume , while the Na+ and K+ concentration changes were relatively small outside this subvolume ( Fig 5A2 and 5A3 ) . We note that the ion-concentration changes in the soma subvolume were unphysiologically high in the no-diffusion case . However , this was of no concern in the current study , since ion concentrations had negligible impact on the ECS dynamics in the case where diffusion was not included . ( In this case the only effect on the ECS potentials came from the concentration dependence of the ECS conductivity , see Methods , Eq 11 . However , for the present case the conductivity changes were found to be too small to have a visible impact on V in the simulations , see Discussion ) . With diffusion included in the ECS simulation , the ion-concentration gradients across the depth of the piece of tissue became smoother ( Fig 5B ) . For example , a fraction of the K+ expelled during somatic AP firing diffused out of the soma subvolume , and distributed across the entire tissue volume . In this case , the K+ concentration in the soma subvolume increased from a baseline level of 3 mM to slightly above 10 mM during the 84 second simulation , accompanied by a similar reduction in the Na+ concentration . These concentration shifts were within the range that can be expected under non-pathological physiological conditions ( for K+ , the limiting concentration between non-pathological and pathological conditions is typically estimated to be between 10 and 12 mM [7] ) . The buildup of ECS concentration gradients explains the temporal development of the diffusive current that we observed in Fig 4C1 . Early in the simulation , the diffusive current out of the soma subvolume ( i . e . , from n = 3 to n = 4 ) increased in an approximately linear fashion with time . This was because the local ion concentration in the soma subvolume ( n = 3 ) increased in an approximately linear fashion due to the high neuronal output/input in this subvolume . As the ion-concentration gradients built up , diffusion from n = 3 to n = 4 increased , and the concentration increase in n = 3 became sublinear . Eventually , diffusion tended to smoothen out the ECS ion-concentration gradients ( Fig 5A ) , and after about 30 s , diffusion between n = 3 and n = 4 experienced a slight decrease . A similar process took place over the entire tissue depth , but was slower further away from the soma , as the transmembrane ionic exchange was smaller there . In the apical dendrites ( i . e . , diffusion from n = 13 to n = 14 ) , the diffusive current still increased in a close to linear fashion at the end of the 84 second simulation ( Fig 4C1 ) . Due to the slow nature of diffusive currents , we proceeded to investigate the slow time scale dynamics of the ECS potential . To do this , we took the time series of V ( plotted for selected subvolumes in Fig 4 ) , and split it up in five equal time intervals of 16 . 8 second duration ( adding up to the total simulation time of 84 seconds ) . Next , we took the temporal average of V in these five intervals and obtained a ( very ) low-pass filtered version of the ECS potential . The results are displayed in Fig 6 showing how the low-pass filtered V was distributed across the tissue depth in the cases without ( Fig 6A2 ) and with ( Fig 6B2 ) diffusion included in the ECS simulations . We first investigate the ECS voltage gradients obtained in the case where ECS diffusion was not included in the simulations ( Fig 6B ) . In this case , there was an ECS voltage drop ( of about 1 . 3 mV ) from the soma subvolume to the subvolumes containing the apical dendrites . The drop in V was consistent with the neuronal source/sinks configurations that we observed earlier ( Fig 3 ) : Since the main neuronal current source ( transmembrane current entering the ECS ) was found in the soma subvolume ( n = 3 ) , while the sinks ( transmembrane current leaving the ECS ) were located higher up along the apical dendrites , there had to be an ECS current in the positive z-direction ( corresponding to a negative voltage gradient in this direction ) to close the current loop between the sources and sinks . Similar V profiles have been seen experimentally where sustained voltage profiles which vary by a up to several mV at spatial scales of millimeters have been seen in cortex [1 , 3] , hippocampus [26] and in the spinal cord [23] . We also note that the neuronal current sources/sinks were effectively constant at this slow time scale ( Fig 3 ) , meaning that they were essentially the same in all the five different time intervals in Fig 6 . We would then a priori expect the ECS current to be constant over time as well . Without extracellular diffusion , this would in turn imply that also the ECS voltage gradient should remain constant throughout the simulation , which is indeed what is observed in Fig 6A2 ( lines are on top of each others ) . With diffusion included in the ECS simulations , the situation became more complex ( Fig 6C ) . The gross features of the ECS voltage gradient resembled what we saw in Fig 6B . The similarity was not surprising , since the neuronal sources were identical in the two cases . However , with diffusion included , the ECS potential gradients no longer remained constant throughout the simulation ( Fig 6C ) . The time-dependent variations were most pronounced in the soma subvolume where the ECS potential decreased by about 0 . 2 mV over the time course of the simulation . This shift in V was caused by diffusive currents along the ion-concentration gradients that built up during the simulation , and was the same shift that we previously observed in Fig 4A4 . A detailed physical interpretation of the diffusion-induced shifts in the ECS potential is provided in the following subsection . To obtain a more thorough understanding of the interplay between the potential V and diffusive currents , we next plotted the ECS fluxes of all ion species ( K+ , Na+ , Ca2+ , and X- ) in the cases without and with extracellular diffusion ( Fig 7 ) . Also here , the focus was on the long time-scale dynamics , and we compared the time-averaged fluxes taken over five 16 . 8 second time intervals ( same procedure as used for V in Fig 6 ) . In the rightmost column in Fig 7 , we have also plotted the total electrical ECS current associated with the ionic fluxes ( the definition is given in the caption of Fig 7 ) . When ECS diffusion was not included in the simulations , all ion transport in the ECS were due to the electrical field ( Fig 7A ) . In that case , most of the transports were mediated by the most abundant ion species in the ECS , which in our simulation were Na+ and X- . Due to the negative potential gradient between the subvolumes containing the soma and apical dendrites ( Fig 6C ) , the positively charged Na+ ions were driven away from the soma subvolume , while the negatively charged X- ions were driven towards the soma subvolume . Both these ion fluxes amounted to a net electrical current away from the soma subvolume , i . e . , a positive current in subvolumes above the somata ( n > 3 ) and a negative current in subvolumes below the somata ( n < 3 ) . In simulations including extracellular diffusion we plotted the ECS flux densities due the electrical field ( jf ) and diffusion ( jd ) separately ( Fig 7B1 and 7B2 ) , as well the total flux density ( jf + jd , Fig 7B3 ) . As AP firing evoked a decrease/increase of Na+/K+ in the soma subvolume , ECS diffusion drove Na+ into this subvolume , while it drove K+ out of this subvolume ( Fig 7B2 ) . As these two cation fluxes were oppositely directed , the net diffusive charge transport ( id/F ) was smaller than the charge transported by Na+ and K+ separately . However , the diffusive fluxes still gave rise to a net electrical transport of the same order of magnitude as the field-driven current , especially around the soma subvolume ( compare current densities in panels B1 and B2 in Fig 7 ) . The ionic fluxes in the ECS differed quite significantly between the cases with and without ECS diffusion ( compare flux densities in panels A with B3 in Fig 7 ) . However , the net electrical current in the system were identical in the two cases ( compare current densities in panels A and B3 ) . This can be understood from basic electric circuit theory: As the neuronal transmembrane sources/sinks were identical in the two cases , the same had to hold for the net extracellular current . Otherwise , the current loop would not be completed . This leads to the following key insight: Since the net electrical current density ( itot = if + id ) was independent of whether diffusion was present in the model or not , an increase in id had to be accompanied by a corresponding decrease in if , and vice versa . A time-dependent variation of diffusive currents therefore by necessity evoked a time-dependent variation of the field currents ( Fig 7B1 ) ) . As if was proportional to the voltage gradient , this in turn implied that the ECS voltage gradients varied with time , as observed in Fig 6 . So far , we have demonstrated that diffusive currents can have quite substantial effects on ECS potentials , at least on a slow time scale . As a next inquiry , we would like to know the frequency range in which diffusion can be expected to have an effect on recorded ECS potentials , and in particular whether diffusion can be expected to affect experimental LFP recordings where the low-frequency cut-off typically ranges from 0 . 1 Hz to 1 Hz ( see e . g . , [29 , 30] ) . We limited this study to ECS potentials recorded in the soma subvolume , where the diffusive effects were most pronounced in our model . Fig 8 shows the power spectral densities ( PSDs ) of the ECS potential recorded outside the somata ( n = 3 ) , where V was obtained as in the above simulations in Figs 4–7 ) . To explore the development of the PSDs over the time course of our simulation , we split the 84 second time series of V into four 21 second intervals , and computed the PSD for these time intervals separately . A first observation is that the PSDs for the simulations without ( red lines ) and with ( blue lines ) ECS diffusion differed dramatically for the lowest frequencies , where the presence of diffusion boosted the PSD by up to several orders of magnitude . Contrarily , for the highest frequency components the PSDs were close to identical in the cases without and with diffusion ( red and blue lines overlap ) . This was as expected from our previous analysis where we saw that diffusion was important for the slow , but not the fast system dynamics ( Fig 4 ) . The cross-over frequency for which the diffusion contributed negligibly to the PSD , was for all four time intervals depicted in Fig 8 seen to be in the frequency range between 1 and 10 Hz . Extracellular diffusion was thus found to have effect on the PSD for frequency components well within the range typically considered in recordings of LFPs in vivo [29 , 30] . The PSDs obtained with no ECS diffusion ( red lines ) were quite constant throughout the simulation , while the PSDs obtained with ECS diffusion included ( blue lines ) were generally higher for the earliest time intervals ( compare panels A and D ) . To provide a hand-waving explanation to the latter , we start by noting that the contribution of diffusion to the local PSD essentially depended on the absolute value of the temporal variation of local ion concentration ( i . e . , on | c ˙ k | , see S1 Appendix ) , which in turn depended on two competing processes . The first process was the local neuronal output of ion species k , which was roughly constant at the long timescale considered here . The second process was ECS transportation of ion species k out from/into the local region . Generally , these two processes had opposing effects on the local ion-concentration dynamics ( i . e . , when neurons expelled K+ into a given subvolume , ECS transports tended to drive K+ out from that subvolume ) . Early in the simulation , ECS concentration gradients ( and thus ECS diffusive transports ) were small , and the time development of the local concentration was approximately proportional to the neuronal output . At a later stage , ECS concentration gradients had built up , and the competing diffusive process had increased . Then local concentrations changed more slowly with time . In the rather complex scenario studied so far , transmembrane and extracellular currents interacted ( as is , of course , the case in real brain tissue ) . However , diffusive fluxes and currents in the ECS can in principle exist even without on-going neuronal sources , provided that there are concentration gradients present in the ECS . To improve our understanding of diffusion-generated potentials , we explored them also in such a simplified scenario . For simplicity , we used the same simulation as above ( Figs 2–7 ) to generate reasonable ECS concentration gradients needed in the simplified scenario . However , this time we turned off the neuronal current sources midways in the simulation ( i . e . , after 42 seconds ) , and analyzed the ECS dynamics in last 42 seconds of the simulation when the ECS dynamics was solely due to diffusion along the concentration gradients that had built up during the first 42 seconds of the simulation . For this scenario , only the simulations with ECS diffusion included gave non-trivial results ( when extracellular diffusion was not included , the ECS voltage gradient instantly turned to zero when the neuronal current sources were removed , and the extracellular ion fluxes immediately stopped ) . This can be easily understood from the current conservation laws upon which the KNP formalism was based , stating that the sum of currents into an ECS compartment should be zero ( Fig 1C ) . In the simplified scenario , there were no transmembrane sources after 42 seconds , and with no diffusive currents between ECS subvolumes , the field currents ( and thus voltage differences ) between ECS subvolumes must by necessity also be zero . The simulations with ECS diffusion included are shown in Fig 9 . Panels A2–5 show the ECS concentration profiles at selected time points after the neuronal sources were turned off at t = 42 s . Initially ( i . e . , at t = 42 s ) , ionic concentrations of Na+ , K+ , Ca2+ and X- in the soma subvolume had been shifted by approximately -5 . 1 mM , 6 . 0 mM , -0 . 1 mM and 0 . 7 mM , respectively , relative to the baseline concentrations . We note that these shifts fulfilled the requirement of local electroneutrality , i . e . , did not correspond to any net change in the local charge density: Σk ( zkck ) = ( −5 . 1 + 6 . 0 − 2 × 0 . 1 − 0 . 7 ) mM = 0 . Here zk and ck are the valence and concentration , respectively , of ion species k . The deviations from baseline concentrations were smaller outside the soma subvolume , and the concentration gradients out of the soma subvolume were quite steep . Diffusive currents along these gradients gave rise to a diffusion potential , which at t=42 s peaked in the soma subvolume where V was about -0 . 17 mV ( Fig 9B2 ) . Diffusion-evoked voltage gradients like this are well understood , and have been observed in many systems with spatial variation in ion composition [3 , 20–22] . With no neuronal sources present , the ECS concentration gradients were gradually smoothed over time ( i . e . , for t> 42 s ) . Consequently , the ECS voltage gradients decayed . At the end of the simulation ( i . e . , for t = 84 s ) , V was about -0 . 05 mV in the soma subvolume . The PSD corresponding to this decay process is depicted by the black lines in Fig 9B3 and 9B4 . Since the concentration gradients became gradually smoother , the power was generally higher during the first 21 s after the neurons were turned off ( Fig 9B3 ) than in the proceeding 21 s ( Fig 9B4 ) . In both cases , the PSDs were very close to a 1/f2 power law ( the fitted power-law coefficients were 1 . 998 in panel B3 and 2 . 02 in B4 ) . This so-called Brownian-noise power law essentially follows from an exponential decay of local ion concentrations , and can be derived analytically ( see S1 Appendix ) . For comparison , we also show the PSDs of the simulation with neuronal sources included ( the red and blue lines in Fig 9B3 and 9B4 are the same as in Fig 8C and 8D , respectively ) . Also in the presence of neuronal sources , the electrodiffusive ECS process roughly followed a 1/f2 power law for low frequencies where diffusion dominated ( blue line and black line close to parallel for f < 10 Hz ) . Comparing the blue and black lines , we further note that the removal of neuronal current sources at t = 42 s increased the low-frequency components of V , especially during the first 21 second time interval after the time of the sources offset ( Fig 9B3 ) . To explain this , we may recall that the diffusive power spectrum is proportional to the absolute value of the temporal variation of local ion concentration ( | c ˙ k | ) . As argued above , this value depends on the balance between two competing processes , i . e . , the local neuronal output of ion species k and the ECS transports of ion species k out from/into the local region . The observation in Fig 9B3 simply implies that the local concentration approached the baseline levels faster when the neuronal sources were turned off ( black line ) than it diverged from the baseline level in the case when the neuronal sources were kept on ( blue line ) . In reality , transmembrane current sources and ECS transport processes do interact , and the correct electrodiffusive PSD is predicted by the blue line in Fig 9B3 and 9B4 . Likewise , the predicted maximum frequency that will be affected by diffusion is in the frequency range 1–10 Hz where the red and blue lines in Fig 9B3 and 9B4 merge . However , we still believe that the study of the simplified decay process process ( with neuronal sources turned off ) provide useful insights to how ECS diffusion can affect the PSD . Firstly , the simplified ‘decoupled’ model nicely illustrated that ECS diffusion gave rise to a 1/f2 contribution to the PSD , as we saw above . Secondly , we propose that the crossing point between the PSD obtained for the diffusive process alone ( with concentration gradients representable for what one typically see in the system ) and the PSD obtained from neurodynamics when diffusion was not included ( black vs . red line in Fig 9B3 ) may serve as a crude estimate of the maximum frequencies for which diffusion can be expected to influence the PSD . For example , the crossing point between the red and black line in Fig 9B3 was found in the frequency range 1–10 Hz , which agreed with the frequency range where the blue and red lines merged . We will provide further arguments for the usefulness of the simplified scenario further in the Discussion . So far , we conclude that in the current model , diffusive processes affected ECS potentials for frequencies up to several hertz .
The simplified model set-up used here have several limitations . Firstly , V was computed as an averaged value over a large ECS volume , and comparison between this and experimental recordings of V with point-like electrodes with small contacts is not straightforward . Secondly , brain tissue contains many types of neurons , which are distributed with somata in different depth layers ( see , e . g . , [18 , 69 , 78] ) , whereas we only included one . Thirdly , we did not include synaptic connections between neurons . Such connections could induce a level of synchrony in the neuronal firing , which likely would influence the power spectrum of the ECS potential [16 , 35] . Fourthly , we assumed that spatial variations in the electric potential and ion concentrations occurred only in one spatial dimension . This is clearly not strictly true , and some aspects of the estimated power spectra are likely to depend on the three-dimensional nature of the real system . Fifthly , the presently used multicompartmental neuronal model [62] ( together with most other available multicompartmental models ) does not include ionic uptake mechanisms such as Na+/K+-pumps . Such mechanisms , along with glial uptake mechanisms [28 , 79] , would generally act to maintain the ECS ion concentrations closer to the baseline levels than what we predicted with our model . These shortcomings are discussed in further detail below . In the current section we discuss the predictions that we made regarding diffusion-generated electric potentials , and to which degree these can be expected to reflect realistic experimental scenarios . To clarify the discussion , we start by labeling the three situations that we have studied P1 , P2 , and P3 , respectively . We refer to Fig 9B3 and 9B4 , where all three situations ( P1–P3 ) are represented . The blue line ( P1 ) represents ECS dynamics surrounding an active neuronal population in the realistic scenario described by the full electrodiffusive formalism . The red line ( P2 ) represents the situation where ECS diffusion was neglected so that the ECS potential was given exclusively by the distribution of transmembrane sources ( cf . , standard volume-conductor theory [32] ) . Finally , the black line ( P3 ) represents the situation where the neuronal sources had been turned off , so that the ECS potential was driven exclusively by concentration gradients in the ECS . The concentration gradients could in principle be imposed as an initial condition in the system , independent of the neural model , meaning that P3 and P2 were essentially independent processes . As we shall see below , this independence is useful for analyzing our results , and for comparing them to previous studies . A common starting point for the estimation of the current-source density CSD ( x , y , x ) from the ECS potential V ( x , y , z ) is [37 , 39]: ∇ ( σ ∇ V ) = - C S D ( 1 ) The left hand side is the divergence of the ECS currents , and an implicit assumption in this equation is that only electric currents driven by the electrical field is present in the ECS , i . e . , solely Ohmic current densities given by if = −σ∇V . If also diffusive ECS currents were accounted for , the corresponding equation would be: ∇ ( σ ∇ V ) - ∇ i d = - C S D , ( 2 ) where the diffusive current density is a function of ionic concentrations in the ECS ( see Methods , Eq 9 ) . The use of Eq 1 for predicting the CSD could thus lead to a misinterpretation of diffusive ECS currents ( if present in the real system ) as neuronal current sources . An example where experimental recordings seems to disagree with the standard CSD theory ( Eq 1 ) was reported recently by Riera et al . [30] who found that the estimated instantaneous current-source density ( CSD ) from recorded ECS potentials did not sum to zero over the volume of the barrel column . According to the standard CSD-theory , this would indicate the presence of a non-zero current-source monopole on a mesoscopic ( cell population ) scale . The possible origin of these apparent current monopoles was later debated [30 , 103–106] . A non-negligible diffusive source term , cf . Eq 2 could be one ( of several ) possible explanations of this discrepancy between experiments and original CSD-theory . The model presented here was a simplified one , both in terms of using a 1D geometry and in terms of neglecting several neuronal and glial mechanisms that would likely contribute to the generation of the LFP . A future ambition is to expand this framework to a 3D model that also accounts for more of the complexity of neuronal tissue , and includes effects of neuronal and glial ionic uptake mechanisms ( ion pumps ) . A 3D version of the KNP framework could ideally be combined with existing , comprehensive simulators of large neuronal networks such as the Blue Brain simulator [69] . We believe that such a framework would be very important for the field of neuroscience as it not only would be useful for exploring how diffusive currents can have an impact on ECS potentials , but also to simulate various pathological conditions related to ion-concentration dynamics in neural tissue [9 , 13–15 , 101 , 107] .
What we have here coined the Kirchhoff-Nernst-Planck formalism , was originally developed for computing the intra- and extracellular dynamics of ion concentrations and the electrical potential during astrocytic K+ buffering [28] . In the current application , it was only applied in the ECS ( the intracellular space was handled with the NEURON-simulator ) . For simplicity , we assumed that spatial variation only occurred in one spatial direction ( z-direction ) , and thus that we had lateral homogeneity of all state variables . In the current work , the KNP formalism was used to predict the extracellular ion-concentration dynamics and electrical potential surrounding a small population of ten pyramidal cells . The neural simulation used in this study was briefly introduced in the Results section , but is presented in further detail here . Simulations on the pyramidal cell model by [62] was run the NEURON/Python simulation environment [60] . The ECS dynamics was computed separately with the KNP formalism , using the neuronal output/input as an external input time series . The KNP model was implemented in MATLAB ( http://se . mathworks . com/ ) . The MATLAB code ( along with the neuronal input time series ) will be made publicly available at ModelDB ( http://senselab . med . yale . edu/modeldb ) .
|
When electrical potentials are measured in the extracellular space ( ECS ) of the brain , they are interpreted as a signature of neural signalling . The relationship between the ECS potentials and the underlying neuronal processes is often studied with the aid of computer models . The ECS potential is typically assumed not to be affected by diffusive currents in the ECS , and existing models therefore neglect diffusion . However , there may be scenarios where this assumption does not hold . Here , we present a new computational model which explicitly models ion-concentration dynamics in the ECS surrounding a neural population , and which allows us to quantify the effect that diffusive currents have on the ECS potential . Using this model , we simulate a scenario where a population of pyramidal neurons is active over a long time , and produces large , but realistic concentration gradients in the ECS . In this scenario , diffusive currents are found to influence the ECS potential at frequency components as high as ten hertz . Unlike previously believed , we thus predict that there are scenarios where recorded local field potentials ( LFPs ) are likely to contain signatures not only of neural activity , but also of ECS diffusion .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"nervous",
"system",
"membrane",
"potential",
"ions",
"electrophysiology",
"neuroscience",
"simulation",
"and",
"modeling",
"computational",
"neuroscience",
"neuronal",
"dendrites",
"physical",
"chemistry",
"research",
"and",
"analysis",
"methods",
"animal",
"cells",
"chemistry",
"cellular",
"neuroscience",
"neuronal",
"morphology",
"cell",
"biology",
"anatomy",
"synapses",
"physiology",
"neurons",
"single",
"neuron",
"function",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"physical",
"sciences",
"computational",
"biology",
"neurophysiology"
] |
2016
|
Effect of Ionic Diffusion on Extracellular Potentials in Neural Tissue
|
Extreme differences in allele frequency between West Africans and Eurasians were observed for a leucine-to-valine substitution ( Leu372Val ) in the human intestinal zinc uptake transporter , ZIP4 , yet no further evidence was found for a selective sweep around the ZIP4 gene ( SLC39A4 ) . By interrogating allele frequencies in more than 100 diverse human populations and resequencing Neanderthal DNA , we confirmed the ancestral state of this locus and found a strong geographical gradient for the derived allele ( Val372 ) , with near fixation in West Africa . In extensive coalescent simulations , we show that the extreme differences in allele frequency , yet absence of a classical sweep signature , can be explained by the effect of a local recombination hotspot , together with directional selection favoring the Val372 allele in Sub-Saharan Africans . The possible functional effect of the Leu372Val substitution , together with two pathological mutations at the same codon ( Leu372Pro and Leu372Arg ) that cause acrodermatitis enteropathica ( a disease phenotype characterized by extreme zinc deficiency ) , was investigated by transient overexpression of human ZIP4 protein in HeLa cells . Both acrodermatitis mutations cause absence of the ZIP4 transporter cell surface expression and nearly absent zinc uptake , while the Val372 variant displayed significantly reduced surface protein expression , reduced basal levels of intracellular zinc , and reduced zinc uptake in comparison with the Leu372 variant . We speculate that reduced zinc uptake by the ZIP4-derived Val372 isoform may act by starving certain pathogens of zinc , and hence may have been advantageous in Sub-Saharan Africa . Moreover , these functional results may indicate differences in zinc homeostasis among modern human populations with possible relevance for disease risk .
Zinc homeostasis is critically important for human health . Similarly to iron , zinc has manifold functions in the body , such as in the immune system [1] , aging [2] , DNA repair [3] , signaling [4] and in diseases such as diabetes [5] and cancer [6] . On the molecular level , zinc acts as a co-factor in hundreds of metallo-enzymes as well as in hundreds of DNA-binding proteins ( e . g . zinc finger proteins ) . Zinc homeostasis is tightly regulated by 10 zinc efflux transporters and 14 zinc influx transporters ( encoded by the SLC30A and SLC39A gene families , respectively ) . ZIP4 ( SLC39A4 ) is the most important intestinal zinc uptake transporter and is expressed at the apical membrane of enterocytes [7] , [8] . Loss-of-function mutations in ZIP4 cause acrodermatitis enteropathica [9] , [10] [MIM 201100] , a congenital disease characterized by extreme zinc deficiency if left untreated without supplemental zinc [11] , [12] . Fittingly , it was recently reported that the loss of expression of this gene in a ZIP4 intestine-specific knockout mouse caused systemic zinc deficiency , leading to disruption of the intestine stem cell niche and loss of intestine integrity [13] . The single nucleotide polymorphism ( SNP ) c . 1114C>G ( rs1871534 ) in the ZIP4 gene ( SLC39A4; NM_130849 . 2 ) results in the substitution of leucine for valine at amino acid 372 ( Leu372Val ) in the human ZIP4 transporter . This non-synonymous SNP is one of the most markedly differentiated genetic variants in the genome in terms of allele frequency differences between populations [14]–[16] , according to data from HapMap [17] , the Human Genome Diversity Panel ( HGDP ) [18] and the 1000 Genomes Project [16] . Extreme population differentiation is a signature of local positive selection [15] , [19]–[21] , but genomic scans for targets of natural selection based on other criteria , such as extended long haplotypes [22]–[24] or selective signatures in the allele frequency spectrum [25] , have failed to identify ZIP4 as a candidate gene for positive selection . To date , whether this variant has evolved under positive selection or neutrality , and its potential functional significance , has not been examined . In the work reported here , we had three main objectives: ( i ) to investigate evolutionary explanations for the extreme population differentiation of the ZIP4 Leu372Val polymorphism by use of coalescent simulations; ( ii ) to test for functional differences in cellular zinc transport between the alleles of the Leu372Val polymorphism using a heterologous expression system; and ( iii ) to discuss potential selective forces behind this possibly adaptive event and their implications for zinc homeostasis in modern humans . We have extensively characterized the extreme geographical differentiation of the Leu372Val substitution and provide evidence that it has been subject to a nearly complete but mild selective sweep in Sub-Saharan Africa . Our simulations show how the extreme pattern of population differentiation , yet absence of other classical signatures of positive selection , can be explained by directional selection accompanied by the effects of a recombination hotspot near the polymorphic adaptive site . Additionally , our data demonstrate in vitro functional differences between the two human polymorphic alleles at codon 372 of the human ZIP4 transporter in surface protein expression , basal intracellular levels of zinc and zinc uptake . We hypothesize that the reduction in intracellular zinc levels mediated by the Val372 allele may have been advantageous in Sub-Saharan Africa , possibly by restricting access of a geographically restricted pathogen to this micronutrient , and that other possible secondary consequences for disease risk and health may result from the differential activity of the ZIP4 alleles .
Five common non-synonymous SNPs are known in the human ZIP4 gene ( Table 1 ) : Glu10Ala ( rs2280839 ) , Ala58Thr ( rs2280838 ) , Ala114Thr ( rs17855765 ) , Thr357Ala ( rs2272662 ) and Leu372Val ( rs1871534 ) . However , only the latter two SNPs show elevated levels of population differentiation in the 1000 Genomes Phase1 sequencing data when comparing the Yoruba from Ibadan , Nigeria ( YRI ) with either the Han Chinese from Beijing , China ( CHB ) or Utah residents of Northern and Western European origin ( CEU ) . As shown in Figure 1A and 1B , their FST values fall above the 99 . 999 th percentile of the genome-wide FST distributions between CEU-YRI ( with FST values for rs2272662 and rs1871534 of 0 . 48 and 0 . 98 , respectively ) and between CHB-YRI ( with FST values of 0 . 51 and 0 . 98 , respectively ) . We therefore verify that the Leu372Val substitution encoded by SNP rs1871534 is the non-synonymous polymorphism exhibiting the most extreme allele frequency differences in the human ZIP4 gene . Next , we genotyped the 51 populations from the Human Genome Diversity Panel ( HGDP ) and compiled additional allele frequencies for this position in worldwide populations from the Alfred database [26] , [27] . Additionally , we obtained new data from a Pygmy population from Gabon and North African populations of Western Sahara , Morocco , and Libya . These new data confirm that the Leu372 variant is the most common allele outside of Africa , and provide a more detailed picture of the geographical allele frequency distributions of this non-synonymous polymorphism ( Figure 1C and Table S1 ) . Overall , the Val372 variant showed the highest frequencies in Sub-Saharan Africa , with populations such as the Ibo or the Yoruban people exhibiting the most extreme derived allele frequencies worldwide ( 0 . 99 and 0 . 96 , respectively ) . Interestingly , two presumably early-branching groups in Sub-Saharan Africa , the Pygmy and the San people , showed opposing trends in the derived allele frequency ( 0 . 94 and 0 . 0 , respectively ) . Even though the small sample size from the San ( only six individuals ) means that a population frequency of up to 0 . 221 cannot be excluded ( with p = 0 . 05 based on assuming Hardy-Weinberg equilibrium and a binomial approach ) , such divergent tendencies in these two Sub-Saharan populations are maintained . Given the elevated levels of population differentiation of the SNP rs2272662 , we also genotyped the HGDP panel for the Thr357Ala polymorphism . However , compared with the Leu372Val substitution , the derived allele at this non-synonymous SNP displayed intermediate frequencies worldwide ( Figure S1 and Table S1 ) and less extreme allele frequency differences between populations . Given the allele frequency differences observed in the Leu372Val polymorphism between the two early human branches in Africa and the uncertainty associated with the low coverage of the Neanderthal genome draft sequence [28] , we resequenced the corresponding orthologous positions for rs1871534 and rs2272662 in an additional Neanderthal sample , labeled SD1253 and excavated at El Sidrón site in Spain [29] . The two positions were amplified in a multiplexed reaction , along with a diagnostic Neanderthal mitochondrial DNA ( mtDNA ) fragment , to monitor contamination in the PCR reaction . For the L16230-H16262 diagnostic mtDNA fragment , 64 clones were generated ( Figure S2 ) , all of which show the Neanderthal-specific 16234T-16244A-16256A-16258G haplotype [28] . This again supports the very low level of contamination in this particular sample . For the orthologous positions of the human rs1871534 and rs2272662 SNPs , 19 and 14 sequences were successfully obtained , respectively . With the exception of one clone in the second position , all sequences showed the previously inferred ancestral alleles , in agreement with the reads present for the Vindija individuals 33 . 16 ( one read for each position ) , 33 . 25 ( two for rs1871534 and none for rs2272662 ) and 33 . 26 ( two and one , respectively ) ( Figure 2 ) . The successful resequencing of this Neanderthal individual , together with published reads from additional Neanderthals [28] and from the Denisovan individual [30] , strongly suggests that the Leu372 variant ( encoded by the C allele in rs1871534 ) is the ancestral human form , which is also in agreement with the chimpanzee state ( Figure 2 ) . Together with the extreme population differentiation pattern , these results suggest that a selective sweep may have taken place in Sub-Saharan Africa , where the derived variant is nearly fixed . Next we examined the complete genomic region around ZIP4 ( Figure 3 ) in the 1000 Genomes sequencing data . Whereas we found a cluster of three strongly elevated FST scores between CEU and YRI in the neighboring SNPs rs1871535 ( intronic ) , rs1871534 and rs2272662 ( further suggesting directional selection in a specific geographical region ) , in both populations there was a clear absence of extreme values in neutrality statistics such as Tajima's D or Fay and Wu's H ( Figure S3 ) . Notably , no other polymorphism in the flanking region of the human ZIP4 displays the high levels of population differentiation of the Leu372Val substitution . Interestingly , in both African and non-African populations there is a recombination hotspot in the ZIP4 gene , which could have reduced any signature of selection on the surrounding linked variation , thereby explaining the apparent lack of significant departures from neutrality . To further investigate this possibility , we carried out coalescence simulations under a variety of recombination and selection scenarios using a well-established demography [31] . As shown in Figure 3D , the observed values for FST and most of the different neutrality statistics cannot be explained by neutral evolution or positive selection with a constant recombination rate . Instead , this atypical pattern of extreme population differentiation , yet seemingly neutral Tajima's D and other neutrality statistics , showed a higher recovery in simulations with directional selection on the derived allele in Sub-Saharan African populations in the context of the observed recombination landscape , including the hotspot ( Figure 3D and 3E , Figure S4 ) . In a more formal evaluation of the results , we quantified the empirical probability for each scenario and neutrality test as well as for different combinations of tests by using composite scores encompassing at least three complementary signatures of positive selection: ( i ) site frequency spectrum , ( ii ) population differentiation , and ( iii ) haplotype structure . The scenario of “weak selection ( s = 0 . 005 ) + hotspot” is the most likely among the different ones tested ( Table S2 ) . Moreover , all the empirical likelihoods calculated for the different composite scores indicate that the proposed scenario of “weak selection ( s = 0 . 005 ) + hotspot” is more likely than the neutral scenario ( Table 2 ) . Therefore , our simulation results indicate that the atypical patterns of selection in the ZIP4 gene can indeed be explained by positive selection having acted upon the Val372 allele in Sub-Saharan African populations and that recombination has erased further accompanying signatures of the selective sweep . Selection coefficients lower than the ones tested ( 3 . 0% , 1 . 0% , 0 . 5% ) further dilute the signal of selection in the site frequency spectrum based neutrality tests ( results not shown ) , but require such long duration times of the sweep that would substantially predate the population split between African and Eurasian populations . We observed that the Leu372Val polymorphism affects a highly conserved amino acid ( Figure 4 ) and that the same codon position has been altered in acrodermatitis patients carrying missense mutations Leu372Arg [32] and Leu372Pro [8] . Moreover , both PolyPhen [33] and SIFT [34] algorithms predict functional effects for the Leu372Val substitution ( see Table 1 ) . These observations led us to test the Leu372Val polymorphism for a possible functional change in the ZIP4 transporter , using transiently transfected HeLa cells . To be able to control for possible haplotypic effects between the two most highly differentiated non-synonymous SNPs in the ZIP4 transporter , we also considered variation at the Thr357Ala polymorphism in the functional analyses . Furthermore , we introduced the pathological mutations Leu372Arg and Leu372Pro in the Ala357 background of the human ZIP4 gene and analyzed them as well . The pathological impact of the Leu372Pro mutation on ZIP4 protein biology and function has already been evaluated in the mouse ZIP4 protein [10] , but not the Leu372Arg mutation . Besides providing confirmation of their impact in the context of the human gene , the use of these pathological mutations provided us with an extreme phenotype to which to compare the phenotype associated with the ZIP4 non-synonymous polymorphisms . In all cases , functional analyses were carried out to determine effects on expression , subcellular localization , and zinc transport . As shown in Figure 5 , human ZIP4 proteins carrying the Leu372Pro and Leu372Arg mutations showed an absence of surface protein expression ( P<0 . 001 , one way ANOVA versus the Ala357-Leu372 isoform ) , consistent with the known causal role of these variants in the zinc deficiency disorder , acrodermatitis enteropathica . Interestingly , the derived Val372 variant also showed significantly decreased surface expression , but to a much lesser extent , and independently of the Thr357Ala substitution ( P<0 . 05 in both Ala357 and Thr357 backgrounds; one way ANOVA versus the Ala357-Leu372 isoform ) . Overall , the Leu372Val substitution had a highly significant effect on surface expression ( ANOVA , p = 0 . 00021 ) , while there was no effect ascribable to the Thr357Ala replacement ( p = 0 . 579 ) . Western blot analysis of all isoforms revealed a remarkable decrease in detection of the Ala357-Pro372 isoform ( Figure S5A ) . However , the reduced expression of this isoform was not due to a defect in the construct sequence but to a higher protein degradation rate , as shown in Figure S5B . Further analysis showed that the Ala357-Leu372 and Ala357-Val372 isoforms do not differ in protein degradation rate . Therefore , the differences in the surface expression experiment must be due to a different trafficking pattern of these variants . In this sense , co-localization of ZIP4 with calnexin ( a protein present in the lumen of the endoplasmic reticulum ) indeed showed that those proteins presenting lower surface expression were partially retained in the endoplasmic reticulum ( Figure S6 ) . Zinc transport analysis of the different ZIP4 isoforms was performed in two ways . First , we quantified basal zinc content with FluoZin-3 in HeLa cells overexpressing the various ZIP4 variants during a 24-hour period ( Figure 6A ) , and second , we recorded intracellular zinc uptake upon perfusion with an external solution containing 200 µM Zn2+ ( Figure 6B ) . Our results show that basal zinc content in cells overexpressing pathological variants Pro372 and Arg372 did not differ from surrounding non-transfected HeLa cells . On the contrary , all common ZIP4 variants ( Ala357 , Thr357 , Leu372 and Val372 ) promoted increased intracellular zinc levels . However , and in agreement with their reduced surface expression , Val372 variants ( in both Ala357 and Thr357 backgrounds ) presented lower basal zinc content compared to Leu372 ( P<0 . 01 and P<0 . 05 , respectively; one way ANOVA versus the Ala357-Leu372 isoform; Figure 6A ) . As shown in Figure 6B , cells overexpressing the pathological Leu372Arg and Leu372Pro mutations did not uptake zinc , consistent with their inability to traffic to the plasma membrane . Zinc uptake mediated by the Val372 variants was also consistent with their reduced membrane expression; i . e . the Val372 variants in both Ala357 and Thr357 backgrounds presented significantly lower maximum transport ( Tmax ) compared to the Leu372 variant ( P<0 . 01 in each case; Figure 6B ) . However , the time to reach half-maximal transport ( t1/2 ) showed no significant difference , indicating that transport kinetics were not markedly different among the four common variants ( Figure 6 ) . Overall , these results support the idea that the Val372 variant does not disturb the kinetics of the ZIP4 transporter but leads to lower zinc uptake transport due to reduced surface expression .
Our study was triggered by the observation of extreme population differentiation between Sub-Saharan African and non-African populations involving the Leu372Val polymorphism in the ZIP4 gene , unaccompanied by any other signals of a classic hard sweep , such as long extended haplotype homozygosity , in either population ( Figures S3 , S7 and S8 ) . By interrogating and compiling allele frequencies in more than 100 worldwide human populations , we further characterized the extreme population differentiation of the Leu372Val polymorphism and confirmed that this result was not an artifact of allele switching [15] . Given the worldwide distribution of the human derived and ancestral alleles ( confirmed by sequencing a Neanderthal and phylogenetic conservation ) , we conclude that this sweep must have taken place within Africa , probably in Sub-Saharan Africa , and not outside the African continent . Notably , the extreme population differentiation of the Leu372Val polymorphism represents the top fourth region within the global genome-wide FST distribution between CEU-YRI obtained from the 1000 Genomes Project data . The only CEU-YRI FST values that are more extreme all involve well-known examples of local geographical adaptation in humans: the SLC24A5 and SLC45A2 genes ( with an FST of 0 . 9826 and 0 . 9765 , respectively ) , which have been associated with light skin pigmentation in Europeans; and the DUFFY gene ( with an FST of 0 . 9765 ) , which provides resistance to the malaria pathogen Plasmodium vivax . Moreover , with the notable exception of DUFFY FY*O allele [35] , [36] , most of the extreme FST values obtained when comparing Africans with non-Africans are usually attributed to local adaptation outside of Africa . Our detection of such a rare signature of natural selection in the African continent is therefore quite remarkable . Interestingly , it is congruent with a recent study that has found only limited evidence for classical sweeps in African populations , which is likely due to a combination of limitations of the currently used methodology and specific characteristics of African population history [37] . Notably , we observed a nearly complete but mild selective sweep for the Val372 variant in Africa , which involves three SNPs with extremely high population differentiation , whereas most other commonly used tests for selection show values not even close to genome-wide significance . Our coalescent simulations indicate that this unusual pattern might be explained by local positive selection in combination with an observed recombination hotspot of moderate strength . At approximately 7 cM/Mb , the recombination rate is only around 7-fold higher than the genomic background , but the hotspot is extended over 3–4 kb . Therefore , a similar number of recombination events may accumulate over time corresponding to a more typically sized hotspot of 1 kb and a recombination rate of around 25 cM/Mb . To our knowledge , this is the first example of a nearly complete selective sweep that is obscured by the effect of a recombination hotspot . It is compatible with earlier theoretical observations that instances of weaker selection in the presence of recombination may not always have an influence on polymorphism statistics [38] and with the observed effect of recombination on the partial sweep around the malaria-related β-globin gene [39] . Because of the unclear effects of the recombination hotspot , it was not possible to estimate the age of the sweep using linkage disequilibrium decay related methods ( e . g . [40] ) . It is likely that a mild selection pressure would have needed a long time to reach the extreme population differentiation values observed , indicating this may be an ancient event . The fact that the high frequency of the Val372 allele is restricted to Sub-Saharan African populations suggests that the selection process started after the Out of Africa expansion of modern humans ( i . e . sixty thousand years ago ) . Alternatively , it is also possible that the bottleneck in the Out of Africa expansion did not sample the Val372 allele , which in turn could explain its absence in most non-African populations . This implies that the Out of Africa event is not a hard upper limit for the age of the selection process . Other more complex evolutionary scenarios cannot be entirely ruled out , and could warrant a more detailed investigation . For example: ( i ) selection acting on standing genetic variation , in the sense that the Val372 variant was already segregating when it came under the influence of local selection; ( ii ) additional directional selection against the Val372 allele in non-African populations; ( iii ) selection favoring the Leu372 variant on multiple , geographically independent origins mostly in non-African populations , in addition to positive selection on the Val372 variant in Africa; and ( iv ) ‘gene surfing’ of any of the two variants on the wave of a population range expansion [41] . However , we consider it is unnecessary to invoke such complex scenarios in preference to the simpler one we propose based on coalescent simulations . Moreover , back-and-forth migrations between Sub-Saharan African , Northern African and Middle Eastern populations after the first Out-of-Africa wave of migration [42] could easily explain the observed low-intermediate allele frequencies in Middle Eastern populations without invoking additional selection events . In the absence of additional linked functional variants in the region , we infer that directional selection has acted on the ZIP4 gene . This conclusion is supported by: ( i ) the disease phenotype of acrodermatitis enteropathica , which involves extreme and potentially lethal zinc deficiency and is caused by , among others , diverse mutations at amino acid position 372 in ZIP4 [43]; ( ii ) the absence of cellular zinc transport in Leu372Arg and Leu372Pro acrodermatitis mutants; ( iii ) the finding that the Val372 variant leads to reduced zinc transport at the cellular level; and finally ( iv ) the conservation of this amino acid position across diverse species ( Figure 4 ) . Furthermore , we infer that the Leu372Val substitution was the functional site targeted by selection due to its location in the predicted center of selection ( highest FST ) , and since it is the only putative functional polymorphism in the ZIP4 gene . Of the other two polymorphic variants with somewhat high allele frequency differences between populations , the Thr357Ala substitution ( rs2272662 ) does not show any functional effect and the intronic rs1871535 cannot be associated with any known regulatory function ( according to information on DNAse I hypersensitivity clusters , CpG Islands and transcription factor binding sites available from the ENCODE data ( http://genome . ucsc . edu/ENCODE [44] ) . Therefore , both rs1871535 and rs2272662 are likely to be neutral . Other non-synonymous polymorphisms with intermediate allele frequencies in the ZIP4 gene ( Glu10Ala , Ala58Thr , and Ala114Thr ) have very low FST scores and are therefore not considered candidate variants for selection . Our functional results in transfected HeLa cells indicate that the Val372 form of the ZIP4 receptor has lower relative cell surface expression , despite no expected differences in mRNA expression and protein synthesis . Interestingly , we found that this decreased expression translated into reduced zinc transport of the derived Val372 variant at the cellular level . That is , we observed differences in the maximal transport ( Tmax ) with no significant differences in the transport kinetics ( T1/2 ) between Leu372 and Val372 . The functional results observed in transfected HeLa cells are likely to be transferable to other epithelial cells , in accordance with independent experiments showing an effect of acrodermatitis variants at position 372 on surface expression ( in CHO cells ) and on zinc transport ( in HEK293 cells ) when using mouse cDNA [10] . However , the critical function of ZIP4 in knockout studies has been shown to primarily affect intestinal zinc uptake [13] . In contrast to the Leu372Pro and Leu372Arg acrodermatitis mutations , which served as controls and showed an almost complete absence of zinc transport , both the Leu372 and Val372 variants must be capable of carrying out zinc transport in the normal range of concentrations , given their high frequency in the healthy population . The consequences of this difference in zinc transport at the organ and organismal level are currently unclear , although there is a strong indication that this variant may indeed be phenotypically relevant . For example , a similar non-synonymous mutation in the porcine homologue of ZIP4 leads to non-pathogenic reduced tissue concentrations of zinc in piglets [45] . Could the concept of “nutritional immunity” [46] , [47] involving zinc explain a putative selective force in Sub-Saharan Africa ? According to this hypothesis , the human host restricts access to certain micronutrients , so that pathogens become less virulent . This is a well-known mechanism of immune defense mediated by iron metabolism [48] , and there are indications that zinc metabolism could have a similar function [47] , [49] . For example , hypoferremia and hypozincemia are both part of the acute phase response to infection and both seem to be influenced by a different zinc transporter from the same family , ZIP14 [50] . We speculate that the selective force behind the extreme FST pattern of the Leu372Val substitution may be related to pathogens or infectious diseases . It is known that decreased zinc uptake mediated by ZIP4 leads to decreased zinc concentrations in the major organs , as shown in a mouse knockout model [13] . While the phenotypic effect of the Val372 allele in humans is currently unknown , we conjecture that the in vitro difference may indeed translate into physiological differences , possibly leading to a slightly decreased uptake of dietary zinc . Fittingly , there is suggestive evidence that African genetic ancestry may involve lower serum levels of zinc [51] , as African-American children have a fourfold risk of zinc deficiency compared to Hispanic children . This result would suggest that African ancestry may be associated with lower serum zinc levels , although these results may be biased due to differences in lifestyle , socio-economic status etc . , and this observation would need to be confirmed by controlled studies . Alternatively , lower zinc concentrations mediated by the Leu372Val substitution in the enterocyte cells could facilitate early diarrheal episodes during a digestive infection in order to reduce the pathogen load on the luminal surface [52] , [53] . Similarly , the lower level of expression of the ZIP4 isoform carrying the Val372 variant could also be advantageous if any parasite uses the ZIP4 receptor to enter enterocytes . Furthermore , the selective force may be related to pre-historic differences in dietary zinc due to lifestyle or to local levels of zinc concentrations in soil and the food chain . No large-scale ethnic comparisons related to serum or tissue zinc concentrations are available . To our knowledge , rs1871534 has not been tested in case-control studies in African populations related to one of the numerous existing infectious diseases like malaria , trypanosmias or Lhassa fever . It is therefore possible that important evidence for a possible selective force has been missed . In future research , the inclusion of additional cell lines , and genotype-phenotype association studies in diverse ethnic populations may help to clarify further phenotypic consequences of this non-synonymous polymorphism . Genotype-phenotype association studies should involve African-American or East African populations in which the Val372 allele is segregating at intermediate frequencies . Candidate phenotypes and traits to interrogate could be serum zinc concentrations , zinc content in hair and nails , serum zinc concentrations after controlled zinc supplementation , and a range of disease traits , especially diseases with an elevated risk in different populations , for example , diverse types of cancer in African Americans . As this SNP was not included in the commonly used Affymetrix and Illumina SNP arrays with up to one million variants ( although it is included in several of the latest arrays ) , potential clinically relevant associations may have been missed . Interestingly , common polymorphisms in other zinc transporters show genome-wide associations with disease traits , such as a non-synonymous variant in the zinc efflux transporter ZnT8 ( SLC30A8 ) and diabetes incidence [54] , as well as a regulatory variant in the zinc influx transporter ZIP6 ( SLC39A6 ) and survival in esophagal cancer [55] . The identification of a high-frequency derived allele polymorphism in the ZIP4 zinc transporter gene ( SLC39A4 ) , combined with a more complete picture of worldwide allele frequencies and in-depth coalescent simulations , is consistent with a long lasting selective event in Sub-Saharan Africa driven by a moderate selection coefficient . This event did not leave the typical footprint of a selective sweep with long haplotypes or detectable neutral deviations in the allele frequency spectrum of the surrounding region , most likely because of the presence of a moderate recombination hotspot . Through functional experiments we have verified the Leu372Val substitution as the likely causal site . Given that two functionally different alleles of this key component of cellular zinc uptake are distributed so divergently across worldwide populations , our results may point to functional differences in zinc homeostasis among modern human populations with possible broader relevance for health and disease .
The G and C alleles at rs1871534 ( Leu372Val ) have been swapped in various public sources such as HapMap ( http://www . hapmap . org ) or dbSNP ( http://www . ncbi . nlm . nih . gov/SNP ) that report conflicting allele frequencies in populations with a similar geographical origin . This situation led us to repeat the genotyping of this SNP in the Human Genome Diversity Panel ( HGDP-CEPH ) [18] . We also genotyped rs2272662 ( which causes the Thr357Ala substitution ) because , within the ZIP4 gene , it shows the second highest allele frequency differences between CEU and YRI HapMap populations and allele frequencies were not available at the worldwide level . The rs1871534 and rs2272662 loci were genotyped in the H971 subset [56] of the HGDP-CEPH [18] , representing 51 worldwide populations , and in an additional population from Africa: Pygmies from Gabon ( N = 39 ) [57] . We also genotyped rs1871534 in North African populations from Western Sahara ( Saharawi , N = 50 ) , Morocco ( Casablanca , N = 30; Rabat , N = 30; Nador , N = 30 ) and Libya ( Libyans , N = 50 ) . Genotyping was performed using Taqman assays C__11446716_10 and C__26034235_10 on an Applied Biosystems Light Cycler ( 7900HR ) , according to standard protocols . Additional genotypes for rs1871534 were obtained from the Alfred database ( http://alfred . med . yale . edu ) [26] , [27] . Informed consent was obtained for all human samples analysed and genotyping analyses were performed anonymously . The project obtained the ethics approval from the Institutional Review Board of the local institution ( Comitè Ètic d'Investigació Clínica - Institut Municipal d'Assistència Sanitària ( CEIC-IMAS ) in Barcelona , Spain . The El Sidrón Neanderthal sample SD1253 has been used in many paleogenomic studies due to its high endogenous DNA content and low contamination levels [28] , [58]–[62] , attributable in part to having been extracted using an anti-contamination protocol [63] . In addition , it has the advantage of having been dated to 49 , 000 years ago [64] , prior to the arrival of modern humans to Europe . The two orthologous positions for rs1871534 and rs2272662 were amplified using a two-step PCR protocol [59] in a multiplexed reaction along with a diagnostic Neanderthal mitochondrial DNA ( mtDNA ) fragment . After visualizing the PCR products in a low-melting temperature agarose gel , the bands were excised , purified and cloned using the TOPO-TA cloning kit ( Invitrogen ) . Inserts of the correct size were sequenced on an ABI3730 XL capillary sequencer ( Applied Biosystems ) . Simultaneous coalescent simulation of recombination hotspots and selection were carried out using Cosi v1 . 2 [31] , [65] . As the underlying neutral demography , we used the best-fit model of Shaffner et al . [31] , [65] with slight modifications ( Table S3 ) , similar to a previously used approach [66] . In particular , the migration frequencies were set to zero and the time points of the European and African population bottlenecks were moved back to 3 , 300 generations before present to accommodate the long sweep times resulting from the lowest selection coefficient we used ( 0 . 5% ) . The sweep was shifted back 350 generations to retain the final population expansions with the advantage of ( i ) a better approximation to the fitted model , and ( ii ) the generation of sufficient singletons when compared to the 1000 Genomes Phase1 data . Subsequent thinning of the simulated data was performed by removing 48% of singleton positions across all populations to account for the underestimation of singletons in 1000 Genomes data . This correction step yielded a much improved ( although not perfect ) unfolded site frequency spectrum as displayed by the derived allele frequencies ( DAF ) and a FST distribution that closely matched the empirical data from 1000 genomes ( Figure S9 ) . Specifically , we compared the empirical FST and DAF distributions from the 1000 genomes data against the original demographic “best-fit” model [31] and two models adapted to allow for different selective sweeps ( the one from [66] and that applied in the current study ) . As seen in Figure S9 , our modified model matched the empirical data as well as or better than the other demographic models . For each subsequent simulation , we used either the recombination landscape including hotspots from the YRI population provided by the 1000 Genomes Consortium and based on HapMap 2 trio data ( http://1000genomes . org ) or alternatively a constant recombination rate of 8 . 17×10−9 , which was calculated as the mean recombination rate in the 100 kb window surrounding ZIP4 . Simulations had a length of 100 kb , were run in 500 replicates for each scenario and sample sizes were set to 176 chromosomes for Sub-Saharan Africans and 194 chromosomes for Europeans . Regions under positive selection were modeled using a single causal variant that rose to an allele frequency of 0 . 98 corresponding approximately to that observed today in YRI . We simulated three different selection coefficients ( 0 . 5% , 2% and 3% ) that led to different durations of the sweep: 2 , 938 generations ( ∼60 , 000–85 , 000 years for generation times of 20 and 29 years , respectively;[67] ) , 1 , 469 generations ( ∼30 , 000–43 , 000 years ) , or 458 generations ( ∼10 , 000–13 , 000 years ) . Empirical probabilities and likelihoods for the different selection statistic values observed in ZIP4 were estimated under each simulated selection scenario ( see Table 2 ) . Firstly , the empirical percentile in which each observation was found was estimated for each test ( FST , dDAF , Tajima's D , Fay and Wu's H , Fu Li's D and XP-EHH ) and scenario ( neutral + constant recombination , neutral + hotspot recombination , low selection + constant recombination , medium selection + constant recombination , high selection + constant recombination , low selection + hotspot recombination , medium selection + hotspot recombination , high selection + hotspot recombination ) . This percentile was then subtracted from one if it was higher than 0 . 5 and multiplied by two to mimic a two-tailed test . Thus , if the observed value was found at the median of the simulated distribution , it yielded a probability of one . By contrast , if it was found in a tail of the distribution , it yielded a probability close to zero . For each scenario , we computed the combined empirical probability for several set combinations of observed neutrality test values by multiplying each corresponding empirical probability ( Table S2 ) . Each combination contained at least one neutrality statistic capturing each of the three main signatures of selection explored ( population differentiation , haplotype structure or the site frequency spectrum ) . Next , empirical likelihoods were estimated as the ratio of the combined empirical probability under each selection scenario over the same probability under neutrality only for the hotspot recombination landscape observed in ZIP4 ( Table 2 ) . Likelihoods for the different combinations of statistics containing dDAF in Table S2 were nearly identical to the equivalent combinations obtained with FST ( data not shown ) . As a conservative decision given the high correlation between FST and dDAF , we do not present the likelihood of any combination including both statistics . It is important to point out that any of the currently available human demographies in combination with coalescent simulators have relatively severe limitations mainly ( i ) in terms of the number of included populations ( e . g . African populations ) ( ii ) the accuracy and timing of the demographic events and ( iii ) the option to include selective sweeps as well as a defined recombination landscape . Therefore it is clear that the complexities of possible evolutionary scenarios ( as discussed in the main text ) are beyond what can be modeled by current approaches . Neutrality tests on simulated and the 1000 Genomes population data were performed as described by Pybus et al . [68] and using the 1000 Genomes Selection Browser ( http://hsb . upf . edu ) . Briefly , Tajima's D , Fu and Li's D and Fay and Wu's H were calculated using a sliding window approach with 30 kb windows and approximately 3 kb offset . FST [69] and XP-EHH [70] between CEU and YRI were calculated for each polymorphic position . Human ZIP4 cDNA encoding the long isoform of the protein and the Ala357 and Leu372 variants was cloned into pcDNA 3 . 1 ( + ) expression vector together with a hemagglutinin ( HA ) tag at the carboxyl terminus as described previously [71] . The Leu372Pro and Leu372Arg mutants , as well as the Thr357Ala and Leu372Val polymorphisms , were introduced via site-directed mutagenesis following standard conditions ( QuikChange II XL; Stratagene; see Table S4 for complete human cDNA and primers used in the mutagenesis ) . The six human ZIP4 isoforms obtained ( i . e . Ala357-Leu372 , Ala357-Val372 , Thr357-Leu372 , Thr357-Val372 as well as Ala357-Pro372 and Ala357-Arg372 ) were confirmed by sequencing with the ABiPrism 3 . 1 BigDye kit before their use in transfection experiments . HeLa cells were cultured in DMEM plus 10% FBS and , subsequently , each of the various ZIP4 forms were transiently transfected using polyethyleneimine as the transfection reagent ( PolySciences ) . For the cell surface expression experiments , live cells were incubated with anti HA ( 1∶1000 ) in DMEM without serum for 1h at 37° before fixation with 4% paraformaldehyde . After blocking for 30 min ( 1% BSA , 2% FBS in PBS ) , cells were incubated with a secondary antibody ( 1∶2000 ) for 45 min in the blocking solution . For the total cell expression experiments , cells were permeabilized with 0 . 1% Triton in PBS for 10 min after fixation . Following blocking for 30 min ( 1% BSA , 2% FBS in PBS ) , cells were incubated in the blocking solution with anti HA ( 1∶1000 ) for 1 h 30 min , washed with PBS and incubated with the secondary antibody ( 1∶2000 ) for 45 min . Images were acquired using an inverted Leica SP2 confocal microscope with a 40×1 . 32 Oil Ph3 CS objective . Expression was quantified by measuring chemiluminescence with a plate reader ( 24-well plates ) using peroxidase-linked anti-mouse antibody ( GE Healthcare ) as a secondary antibody and SuperSignal West Femto reagent as a substrate ( Thermo scientific ) . Data are presented as the ratio between surface expression and total expression of the transporter . Statistical significance was tested using standard ANOVA . Cells were transiently transfected with the various ZIP4 isoforms plus empty ECFP vector for 24–36 h . Cytosolic Zn2+ signal was determined in CFP-positive cells loaded with FluoZin3 2 . 5 µM ( Invitrogen ) in a solution containing 140 mM NaCl , 5 mM KCl , 1 . 2 mM CaCl2 , 0 . 5 mM MgCl2 , 5 mM glucose , 10 mM HEPES , 300 mosmol/l , pH 7 . 4 for 20 min . Cytosolic [Zn2+] increases are presented as the difference with respect to the basal signal of emitted fluorescence ( 510 nm ) after adding 200 µM ZnSO4 in a continuous perfusion bath . The kinetics of the various isoforms were calculated using a sigmoidal non-linear regression . In the same set of experiments , basal cellular Zn2+ content was estimated as the difference in FluoZin intensity between transfected cells and non-transfected cells before adding Zn2+ to the bath . Flourescence intensity was measured using an Olympus IX70 inverted fluorescence microscope , controlled by Aquacosmos software ( Hamamatsu ) .
|
Zinc is an essential trace element with many biological functions in the body , whose concentrations are tightly regulated by different membrane transporters . Here we report an unusual case of positive natural selection for an amino acid replacement in the human intestinal zinc uptake transporter ZIP4 . This substitution is recognized as one of the most strongly differentiated genome-wide polymorphisms among human populations . However , since the extreme population differentiation of this non-synonymous site was not accompanied by additional signatures of natural selection , it was unclear whether it was the result of genetic adaptation . Using computer simulations we demonstrate that such an unusual pattern can be explained by the effect of local recombination , together with positive selection in Sub-Saharan Africa . Moreover , we provide evidence to suggest functional differences between the two ZIP4 isoforms in terms of the transporter cell surface expression and zinc uptake . This result is the first genetic indication that zinc regulation may differ among modern human populations , a finding that may have implications for health research . Further , we speculate that reduced zinc uptake mediated by the derived variant may have been advantageous in Sub-Saharan Africa , possibly by reducing access of a geographically restricted pathogen to this micronutrient .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"mutation",
"haplotypes",
"genomics",
"adaptation",
"genetic",
"mutation",
"genetic",
"polymorphism",
"natural",
"selection",
"genetics",
"population",
"genetics",
"evolutionary",
"selection",
"biology",
"comparative",
"genomics",
"evolutionary",
"biology",
"evolutionary",
"immunology",
"population",
"biology",
"evolutionary",
"processes",
"evolutionary",
"genetics",
"gene",
"function"
] |
2014
|
Extreme Population Differences in the Human Zinc Transporter ZIP4 (SLC39A4) Are Explained by Positive Selection in Sub-Saharan Africa
|
Metazoans protect themselves from environmental toxins and virulent pathogens through detoxification and immune responses . We previously identified a small molecule xenobiotic toxin that extends survival of Caenorhabditis elegans infected with human bacterial pathogens by activating the conserved p38 MAP kinase PMK-1 host defense pathway . Here we investigate the cellular mechanisms that couple activation of a detoxification response to innate immunity . From an RNAi screen of 1 , 420 genes expressed in the C . elegans intestine , we identified the conserved Mediator subunit MDT-15/MED15 and 28 other gene inactivations that abrogate the induction of PMK-1-dependent immune effectors by this small molecule . We demonstrate that MDT-15/MED15 is required for the xenobiotic-induced expression of p38 MAP kinase PMK-1-dependent immune genes and protection from Pseudomonas aeruginosa infection . We also show that MDT-15 controls the induction of detoxification genes and functions to protect the host from bacteria-derived phenazine toxins . These data define a central role for MDT-15/MED15 in the coordination of xenobiotic detoxification and innate immune responses .
In nature , organisms encounter environmental insults , such as chemical toxins , secreted microbial virulence factors and invasive pathogens , that threaten their ability to survive and reproduce . As a result , metazoans have evolved protective pathways to counter these challenges . For example , gene families such as cytochrome P450s ( CYPs ) , glutathione-s-transferases ( GSTs ) , and UDP-glucuronosyltransferases ( UDPs ) detoxify xenobiotic small molecule toxins and are conserved from nematodes to humans [1] . Likewise , innate immune defenses provide protection from invasive pathogens [2] . Recent publications have suggested that recognition of xenobiotic toxins is involved in the activation of immune response pathways [3] , [4] . From an evolutionary perspective , it is logical that hosts respond to threats encountered in the wild at least in part through surveillance pathways that monitor the integrity of core cellular machinery , which are often the targets of xenobiotic small molecules or microbe-generated toxins . These studies predict that organisms may integrate detoxification and immune responses as a means to respond rapidly to such challenges , but the mechanisms underlying this coordinated host response have not been reported . Our research group and others use bacterial and fungal pathogenesis assays in the nematode Caenorhabditis elegans to investigate mechanisms of immune pathway activation in intestinal epithelial cells [2] . Genetic analyses of C . elegans that are hypersusceptible to bacterial infection have revealed that the nematode mounts defense responses through evolutionarily conserved innate immune pathways . For example , the C . elegans NSY-1/SEK-1/PMK-1 Mitogen Activated Protein ( MAP ) kinase pathway , orthologous to the ASK1 ( MAP kinase kinase kinase ) /MKK3/6 ( MAP kinase kinase ) /p38 ( MAP kinase ) pathway in mammals , is required for protection against pathogens [5] . C . elegans animals carrying loss-of-function mutations in this pathway have defects in the basal and pathogen-induced expression of immune effectors and are hypersusceptible to killing by bacterial and fungal pathogens [5]–[7] . We previously used a C . elegans pathogenesis assay as a means to identify small molecules that protect the host during bacterial infection [8] . One of the compounds identified in this screen , a small molecule called RPW-24 , extended the survival of nematodes infected with the human bacterial pathogen Pseudomonas aeruginosa by stimulating the host immune response via the p38 MAP kinase PMK-1 pathway [9] . A genome-wide microarray analysis of animals exposed to RPW-24 revealed that , in addition to inducing the transcription of putative immune effectors , this molecule also strongly upregulated Phase I and Phase II detoxification enzymes ( CYPs , GSTs and UDPs ) , suggesting that RPW-24 is a xenobiotic toxin to C . elegans . Consistent with this hypothesis , RPW-24 caused a dose dependent reduction of nematode lifespan on nonpathogenic food and delayed development of animals that were exposed starting at the first larval stage . Here we sought to use RPW-24 as a tool to characterize mechanisms of p38 MAP kinase PMK-1 pathway activation in C . elegans . We found that activation of PMK-1-regulated pathogen response genes is genetically linked to the induction of genes involved in the detoxification of small molecule toxins . We show that the evolutionarily conserved Mediator subunit MDT-15/MED15 is required for the induction of the p38 MAP kinase PMK-1-mediated immune effectors as well as non-PMK-1-dependent detoxification genes by RPW-24 . These data demonstrate that the host response to a xenobiotic involves coordination of detoxification and innate immune responses via the Mediator subunit MDT-15 . Moreover , loss of MDT-15 function has important physiological effects on the ability of an animal to mount protective immune responses , resist bacterial infection and survive challenge from lethal bacterial toxins .
To investigate mechanisms of immune activation in C . elegans , we generated a GFP transcriptional reporter for the immune response gene F08G5 . 6 . F08G5 . 6 is a putative immune effector that contains a CUB-like domain [6] and is transcriptionally induced by exposure of C . elegans to several bacterial pathogens , including P . aeruginosa [6] , [10] . We chose F08G5 . 6 for these studies because it is upregulated more than 100-fold by RPW-24 in a manner that requires the p38 MAP kinase PMK-1 [9] . pF08G5 . 6::GFP was induced in the C . elegans intestine during P . aeruginosa infection and GFP expression was also robustly upregulated in pF08G5 . 6::GFP animals following exposure to RPW-24 when animals were feeding on nonpathogenic E . coli ( Figure 1A ) . When three components of the p38 MAP kinase PMK-1 signaling cassette were individually knocked down by RNAi [tir-1 [11] , pmk-1 [6] and atf-7 [12]] , pF08G5 . 6::GFP induction by RPW-24 was entirely abrogated ( Figure 1A ) . The level of F08G5 . 6 induction during bacterial infection is dependent upon the virulence of the invading pathogen ( Figure S1 ) . We exposed C . elegans to several P . aeruginosa strains , each of which was previously shown to have a different pathogenic potential toward nematodes [13] and used qRT-PCR to determine the expression levels of F08G5 . 6 in these animals . In general , more pathogenic P . aeruginosa strains caused significantly greater induction of F08G5 . 6 , suggesting that some aspect of P . aeruginosa virulence , rather than a structural feature of the bacteria itself , causes the activation of F08G5 . 6 . Consistent with these data , it was previously shown that C . elegans primarily responds to virulence-related cues to mount its innate immune defenses towards P . aeruginosa [14]–[16] . To identify genes that regulate the p38 MAP kinase PMK-1 pathway in response to RPW-24 , we screened a library of RNAi clones corresponding to 1 , 420 genes expressed in C . elegans intestinal epithelium ( approximately 9% of the genome , Table S1B ) for gene inactivations that abrogated the RPW-24-mediated induction of pF08G5 . 6::GFP . We specifically focused on intestinally expressed genes because of the recognized role for intestinal cells in coordinating the host response to ingested pathogens [2] , [17] , and because P . aeruginosa and RPW-24 induce F08G5 . 6 expression in the intestine ( Figure 1A ) . Our initial screening effort identified 153 genes that , when inactivated by RNAi , diminished or abrogated the induction of pF08G5 . 6::GFP by RPW-24 . We took several steps to identify specific regulators of p38 MAP kinase PMK-1-dependent immune effectors among these 153 gene inactivations . First , we noticed that knockdown of many of these genes markedly slowed nematode growth . To eliminate genes that simply reduced GFP reporter expression as a consequence of pleiotropic effects on worm growth and development , we determined if these 153 gene inactivations also affected induction of the C . elegans immune reporter irg-1::GFP [14] . irg-1 is strongly upregulated in intestinal epithelial cells during P . aeruginosa infection or by an E . coli strain that expresses the bacterial virulence factor Exotoxin A ( ToxA ) , but via a pathway independent of p38 MAP kinase PMK-1 signaling [14]–[16] . Moreover , RPW-24 does not cause the induction of irg-1::GFP and mutation of the zip-2 gene , which encodes the transcription factor that regulates irg-1 expression , does not affect the RPW-24-mediated induction of F08G5 . 6 ( data not shown ) or alter the ability of RPW-24 to extend the survival time of nematodes infected with P . aeruginosa [9] . We therefore discarded the genes that , when inactivated , reduced the induction of irg-1::GFP by E . coli expressing ToxA , reasoning that they were unlikely to be specific regulators of the p38 MAP kinase PMK-1-dependent pathogen response genes . Using this approach , we selected 56 of the 153 genes for further study . In a tertiary screen , we determined the effects of these 56 gene inactivations on pF35E12 . 5::GFP , a second immune reporter that is also strongly induced in the intestine by RPW-24 in a PMK-1-dependent manner [9] , [10] . 29 of the 56 RNAi clones reduced or eliminated the induction of both the pF35E12 . 5::GFP and pF08G5 . 6::GFP reporters by RPW-24 ( Table S1A ) . Validating the screen , the 29 clones we identified as putative regulators of the p38 MAP kinase PMK-1-dependent genes included the three known components of the p38 MAP kinase PMK-1 pathway that were present in the screening library , which suggested that the screen could identify additional , unrecognized components of this signaling pathway . To confirm further the results of the screen , we used RNAi to knockdown the expression of a representative sample of the 29 genes identified , and tested the induction levels of F08G5 . 6 and F35E12 . 5 by RPW-24 with qRT-PCR ( Table S1A ) . For all six genes tested , we verified that inactivation of the gene by RNAi dramatically reduced the induction levels of the p38 MAP kinase PMK-1-regulated genes F08G5 . 6 and F35E12 . 5 . One of the strongest hits from our RNAi screen was mdt-15 , which encodes a subunit of the Mediator complex homologous to mammalian MED15 ( 78% sequence identity ) [18]–[20] . Knockdown of mdt-15 eliminated all visible expression of the pF08G5 . 6::GFP and pF35E12 . 5::GFP reporters , and reduced expression of these genes by at least two orders of magnitude in response to RPW-24 ( Figure 1B and Table S1 ) . MDT-15 was previously found to regulate the transcription of detoxification genes , including cytochrome P450s , glutathione-s-transferases , and UDP-glucuronosyltransferases [19] , gene classes that are strongly induced by RPW-24 [9] . To study the role of mdt-15 in the regulation of p38 MAP kinase PMK-1 gene activation , we crossed the pF08G5 . 6::GFP reporter into the hypomorphic mdt-15 ( tm2182 ) allele , which was previously shown to recapitulate many of the phenotypes observed in mdt-15 ( RNAi ) animals [19] , [21] . As in mdt-15 ( RNAi ) animals , we observed no induction of GFP when mdt-15 ( tm2182 ) ; pF08G5 . 6::GFP animals were exposed to RPW-24 ( Figure 1B ) . An extrachromosomal array containing wild-type mdt-15 under its own promoter partially restored RPW-24-induced GFP expression in mdt-15 ( tm2182 ) ;pF08G5 . 6::GFP animals ( Figure 1B ) . To determine if MDT-15 is required for the induction of other RPW-24-induced genes , we used NanoString nCounter gene expression analysis to generate transcription profiles of 118 C . elegans genes with known involvement in immune , stress and detoxification responses ( Table S2 ) . As in our microarray analysis [9] , we found that RPW-24 caused robust transcriptional changes in wild-type nematodes . 40 of the 118 genes in the NanoString codeset were induced at least 4-fold or greater . 25 of these 40 genes are putative immune effectors upregulated during pathogen infection ( shown in Figure 2 ) and 13 are genes putatively involved in the detoxification of small molecule toxins ( discussed below ) . Of note , we had previously observed that 31 of these 40 genes , including 28 of the 31 most strongly induced , were also upregulated in whole genome Affymetrix GeneChip microarray analysis of wild-type animals exposed to RPW-24 versus DMSO [9] . Of the 25 pathogen-induced genes upregulated by RPW-24 , 21 have been shown to be induced during P . aeruginosa infection [6] , [7] , [22] , [23] . The RPW-24-induced expression of 13 of the 25 putative immune effectors was significantly reduced in pmk-1 ( km25 ) loss-of-function mutants compared to wild-type controls ( Figure 2 top panel ) in accord with the previously determined role for p38 MAP kinase PMK-1 in regulating both pathogen-induced [6] and RPW-24-induced [9] expression of putative immune effector genes . Consistent with a role for MDT-15 in the expression of PMK-1 activated genes , the NanoString analysis revealed that the RPW-24-dependent induction of 10 of the 13 PMK-1-dependent genes was abrogated in mdt-15 ( RNAi ) animals compared to controls ( Figure 2 bottom panel ) . We used qRT-PCR to show that mdt-15 was knocked down by RNAi in this experiment and to confirm that mdt-15 depletion caused a dramatic reduction in the RPW-24-induced expression of three pmk-1-dependent immune genes ( Figure S2A ) . Further , we found that the RPW-24-mediated induction levels of these three immune genes was reduced in the mdt-15 ( tm2182 ) mutant , which recapitulated our findings in mdt-15 ( RNAi ) animals ( Figure S2B ) . Knockdown of mdt-15 reduced the expression of the top five most strongly upregulated p38 MAP kinase PMK-1-dependent immune effectors by several orders of magnitude ( C32H11 . 1 , F35E12 . 5 , F08G5 . 6 , F49F1 . 7 , and F49F1 . 1 ) ( Figure 2 top panel ) . These five genes were still induced by RPW-24 in pmk-1 ( km25 ) loss-of-function animals , albeit to levels markedly lower than in wild-type animals ( Figure 2 top panel ) , but their induction was entirely abrogated by knockdown of mdt-15 ( Figure 2 bottom panel ) . These data suggest that MDT-15 coordinates inputs from PMK-1 and other immune signaling pathway ( s ) to modulate the expression of these p38 MAP kinase PMK-1-dependent putative immune effectors . We also identified a requirement for mdt-15 in the induction of five putative immune effectors ( clec-206 , C29F7 . 2 , K05B2 . 4 , cdr-1 , T28C12 . 4 ) that are not transcriptional targets of the PMK-1 pathway ( Figure 2 , compare top and bottom panels ) . Thus , MDT-15 is required for the induction of p38 MAP kinase PMK-1-dependent immune genes and a second group of defense effectors that are independent of the p38 MAP kinase PMK-1 signaling pathway . The p38 MAP kinase PMK-1 pathway plays an important role in the regulation of putative immune effectors during P . aeruginosa infection [6] . To determine whether MDT-15/MED15 is also involved in the regulation of these genes during bacterial infection , we infected C . elegans with P . aeruginosa PA14 for 8 hours and used qRT-PCR to compare the induction levels of several immune response genes in mdt-15 ( RNAi ) and control animals . The basal and pathogen-induced expression of three p38 MAP kinase PMK-1-dependent immune effectors ( C32H11 . 1 , F08G5 . 6 and F35E12 . 5 ) was reduced in mdt-15 ( RNAi ) animals by one to three orders of magnitude ( Figure 3 ) . We also tested the induction levels of three genes whose transcription is activated during P . aeruginosa infection in a manner independent of PMK-1 ( irg-1 , irg-2 and F01D5 . 5 ) and found that PMK-1-independent immune effectors were induced in mdt-15 ( RNAi ) animals during P . aeruginosa infection to levels comparable to that observed in wild-type animals ( Figure 3 ) . The expression levels of irg-1 and irg-2 were higher under basal conditions in mdt-15 ( RNAi ) animals compared to L4440 controls ( Figure 3 ) . Together , these data support the observations from our NanoString experiments that MDT-15 is required for the expression of some , but not all , immune genes that are activated in response to an environmental insult . One possible explanation for these observations is that the Mediator subunit MDT-15 is a general regulator of transcription that non-specifically affects the expression of a large number of genes . In the NanoString experiment , however , we identified 10 genes that were induced by RPW-24 in mdt-15 ( RNAi ) animals to levels similar to the wild-type controls ( Figure 2 bottom panel ) . Also of note , our secondary screen demonstrated that mdt-15 ( RNAi ) had no effect on the induction of irg-1::GFP by ToxA . We further wondered if the defects in expression of the PMK-1 targets in mdt-15 ( RNAi ) animals were due to direct transcriptional regulation of the p38 MAP kinase PMK-1 pathway components by MDT-15 . However , we found that the mRNA levels of pmk-1 , tir-1 and sek-1 in mdt-15 ( RNAi ) animals were not different from the wild-type control ( Figure S2C ) . Taken together , the data in this section indicate that MDT-15 is required for the transcriptional activation of immune effectors controlled by the p38 MAP kinase PMK-1 , and at least one other pathway . However , MDT-15 is not required for the induction of all defense-related genes . To determine if MDT-15 acts genetically upstream or downstream of the p38 MAP kinase PMK-1 in coordinating the induction of immune effectors during P . aeruginosa infection , we utilized the MAP kinase phosphatase VHP-1 , which is a negative regulator of the p38 MAP kinase PMK-1 [24] . Knockdown of vhp-1 in animals carrying the F08G5 . 6::GFP p38 MAP kinase PMK-1 immune reporter caused increased induction of GFP expression during P . aeruginosa infection in a manner dependent on pmk-1 expression ( Figure 4A ) . This induction of F08G5 . 6::GFP by vhp-1 ( RNAi ) was entirely suppressed in the mdt-15 ( tm2182 ) partial loss-of-function allele ( Figure 4A ) . We used qRT-PCR to confirm this observation ( Figure 4B ) . We found that RNAi knockdown of vhp-1 in wild-type C . elegans caused constitutive activation of F08G5 . 6 , F35E12 . 5 and C32H11 . 1 in a manner that required mdt-15 when nematodes were growing on their normal food source , E . coli OP50 ( Figure 4B ) . During P . aeruginosa infection , we also observed a requirement for MDT-15 in the vhp-1 ( RNAi ) -mediated induction of F08G5 . 6 and F35E12 . 5 , but not C32H11 . 1 . The significance of this discrepancy is unclear , but may be explained by the observations of Taubert et al . who found that only 52% of mdt-15 targets were misregulated in the hypomorphic mdt-15 ( tm2182 ) allele , which encodes a truncated form of the MDT-15 protein [19] , [21] . Indeed , we also observed that mdt-15 ( RNAi ) was more effective than mdt-15 ( tm2182 ) at reducing the expression of the six mdt-15 targets we studied following exposure to RPW-24 , including C32H11 . 1 ( Figures S2A , S2B and S2D ) . In any case , these data suggest that MDT-15 acts downstream of the p38 MAP kinase PMK-1 to coordinate the induction of at least some immune effector genes . The experiments in the preceding sections show that the Mediator subunit MDT-15/MED15 is necessary for the induction of the p38 MAP kinase PMK-1-regulated genes , both in response to RPW-24 and during P . aeruginosa infection . We therefore reasoned that mutation or RNAi-mediated knockdown of mdt-15 would result in enhanced susceptibility to P . aeruginosa infection . Initial P . aeruginosa pathogenesis assays showed a modest , but significant and reproducible , enhanced susceptibility to infection in mdt-15 ( RNAi ) ( Figure S3A ) and mdt-15 ( tm2182 ) ( Figure S3B ) animals compared to controls . However , both mdt-15 ( tm2182 ) and mdt-15 ( RNAi ) animals have reduced brood sizes and varying degrees of sterility . Sterile animals are more resistant to P . aeruginosa infection than wild-type animals , due in part to daf-16-dependent induction of stress response genes [25] . To eliminate this potentially confounding effect , we made all animals in the P . aeruginosa pathogenesis assay sterile by knocking down cdc-25 . 1 , a technique that has been used previously in C . elegans bacterial pathogenesis assays [26] . Under these conditions , we found that mdt-15 ( tm2182 ) animals were markedly hypersusceptible to P . aeruginosa infection compared to control animals ( Figure 5 ) . Moreover , injection of the mdt-15 gene under control of its own promoter partially rescued the enhanced susceptibility to P . aeruginosa phenotype of mdt-15 ( tm2182 ) animals ( Figure 5 ) . Knockdown of mdt-15 by RNAi also caused a hypersusceptibility to P . aeruginosa phenotype in C . elegans fer-15 ( b26 ) ;fem-1 ( hc17 ) sterile animals [6] , [27] , and as predicted , to a greater degree than wild-type animals that were not made sterile by cdc-25 . 1 ( RNAi ) ( Figure S3C ) . We also found that the ability of RPW-24 to extend survival of P . aeruginosa-infected wild-type and sterile nematodes was significantly attenuated in mdt-15 ( RNAi ) animals compared to controls , suggesting that MDT-15 is required for the immunostimulatory activity of RPW-24 ( Figure S3A and S3C ) . The degree of lifespan extension by RPW-24 during P . aeruginosa infection was reduced in mdt-15 ( tm2182 ) compared to controls , but this difference did not reach statistical significance ( p = 0 . 09 ) ( Figure S3B ) . As discussed above , the gene expression defects of mdt-15 targets were more severe in mdt-15 ( RNAi ) animals than in the hypomorphic mdt-15 ( tm2182 ) allele [19] , [21] , which may account for this observation . One caveat concerning the observation that mdt-15 depleted animals are hypersusceptible to P . aeruginosa infection is that mdt-15 ( RNAi ) and mdt-15 ( tm2182 ) animals have a reduced lifespan when grown on the normal laboratory food source E . coli OP50 compared to wild-type controls [18] , [19] . Several observations indicate , however , that MDT-15 is an important modulator of nematode survival during bacterial infection . First , we have shown above that mdt-15 ( RNAi ) animals fail to upregulate p38 MAP kinase PMK-1-dependent immune effectors both in response to a xenobiotic toxin and during P . aeruginosa infection , but retain the ability to induce other immune genes in response to pathogens and following exposure to the bacterial toxin ToxA . In addition , mdt-15 ( RNAi ) animals do not respond to the immunostimulatory effects of RPW-24 . We have shown previously that C . elegans with mutations in the ZIP-1 and FSHR-1 immune pathways , which act in parallel to the p38 MAP kinase PMK-1 cassette , are hypersusceptible to P . aeruginosa infection , but retain the ability to respond to RPW-24 [9] . That mdt-15 ( RNAi ) animals are blind to the immunostimulatory effects of RPW-24 suggests a specific role of MDT-15 in regulating p38 MAP kinase PMK-1 pathway activity . It is also important to note that despite their reduced lifespan , mdt-15 ( RNAi ) animals are not sensitive to all environmental insults . For example , animals deficient in mdt-15 are sensitive to the toxin fluoranthene , but not β-naphthoflavone , and are not more sensitive to high temperatures than wild-type animals [19] . We previously demonstrated that RPW-24 is a xenobiotic toxin [9] . MDT-15 is known to coordinate protection from the toxin fluoranthene and regulate the transcriptional induction of CYPs [19] . We found that the RPW-24-mediated induction of the 13 detoxification genes in the NanoString codeset was nearly entirely abrogated by RNAi knockdown of mdt-15 ( Figure 6A top panel and Table S2 ) . This result was confirmed for three detoxification genes by qRT-PCR ( Figures S2B and S2D ) . To determine if MDT-15 is required to protect C . elegans from the toxic effects of RPW-24 , we studied the development of wild-type , pmk-1 ( RNAi ) and mdt-15 ( RNAi ) in the presence of the xenobiotic RPW-24 , an assay that has been used previously to assess the toxicity of small molecules [19] . We found that RPW-24 slowed the development of control and pmk-1 ( RNAi ) animals to similar degree , and that mdt-15 ( RNAi ) animals were markedly delayed in the presence of RPW-24 ( Figures 6B and S4 ) . These data show that MDT-15 controls the induction of detoxification genes following exposure to RPW-24 and is required to resist the toxic effects of this xenobiotic . Given that MDT-15 is required for the regulation of detoxification genes and the p38 MAP kinase PMK-1 pathway , we wondered if the detoxification machinery in C . elegans is regulated by the p38 MAP kinase PMK-1 . We found , however , that 12 of the 13 RPW-24-induced detoxification genes were upregulated in pmk-1 ( km25 ) null mutants to levels comparable to that in wild-type animals exposed to RPW-24 ( Figure 6A bottom panel ) . We used qRT-PCR to confirm this observation for three cytochrome P450 genes ( Figure S2D ) . Thus , the MDT-15-dependent xenobiotic detoxification program is induced in a manner independent of the p38 MAP kinase PMK-1 . Many xenobiotic toxins , including RPW-24 , induce an avoidance response wherein C . elegans leave a lawn of bacterial food , to which they are otherwise attracted , if it contains a toxic compound [3] , [9] . We therefore wondered if MDT-15 is required for the avoidance behavior induced by RPW-24 . However , pmk-1 ( km25 ) and mdt-15 ( tm2182 ) animals left the lawn of E . coli containing RPW-24 as readily as wild-type animals ( Figure S5 ) . We also observed a similar phenotype when we knocked down the expression of pmk-1 and mdt-15 in the neuronally-sensitive RNAi strain TU3311 ( data not shown ) . Thus , animals lacking the function of MDT-15 and PMK-1 are still able to recognize RPW-24 as a toxin . In summary , the data in this and the preceding section suggest that MDT-15 , but not PMK-1 , controls the induction of genes that are required to resist the toxic effects of RPW-24 , and that neither are required for avoidance of RPW-24 . We have shown that MDT-15 regulates the C . elegans detoxification response to the xenobiotic RPW-24 . We therefore hypothesized that MDT-15 might also be required for protection from lethal secreted toxins produced by pathogenic bacteria . To address this question , we used an assay that allows the specific study of C . elegans killing by secreted low molecular weight toxins of P . aeruginosa . When P . aeruginosa is grown on high osmolarity media , phenazine toxins produced by the bacteria are lethal to nematodes [28] , [29] . In contrast to the “slow killing” infection assay , which was used in the assays described above , wild-type nematodes exposed to these “fast killing” conditions die within a few hours via a process that does not require live bacteria [28] , [29] . Wild-type C . elegans at the fourth larval stage of development , but not young adult animals , are exquisitely sensitive to the phenazine toxins produced by P . aeruginosa in this assay and are killed within 6 hours of exposure . We took advantage of the inherent resistance of young adult animals to test whether mdt-15 ( tm2182 ) animals are hypersusceptible to the phenazine toxins produced by P . aeruginosa . As expected , both wild-type and mdt-15 ( tm2182 ) L4 animals were rapidly killed by P . aeruginosa in the “fast kill” assay ( Figure S6A ) . P . aeruginosa carrying deletions of both phenazine biosynthetic operons ( Δphz ) does not make these toxins and accordingly had reduced ability to kill both wild-type and mdt-15 ( tm2182 ) animals ( Figure S6A ) . In contrast to L4 animals , we observed that almost no young adult , wild-type animals were killed after six hours of exposure to the phenazine toxins compared with 98% death of L4 staged animals ( Figures 7A and S6A ) , which reproduces the findings of others [28] , [29] . In contrast to wild-type animals , young adult mdt-15 ( tm2182 ) animals were dramatically susceptible to P . aeruginosa in this assay and this pathogenesis required the secretion of phenazine toxins , as P . aeruginosa Δphz was markedly less pathogenic toward mdt-15 ( tm2182 ) young adults ( Figure 7A ) . Moreover , mdt-15 ( tm2182 ) animals were not simply hypersusceptible to the high osmolarity conditions of this assay because we observed no mortality over the course of the assay in mdt-15 ( tm2182 ) mutants exposed to “fast kill” media containing the normal nematode food source E . coli OP50 ( Figure 7A ) . We confirmed that phenazine toxins are lethal to mdt-15-depleted animals by supplementing “fast kill” growth media with both phenazine-1-carboxylic acid and 1-hydroxyphenazine in the absence of pathogen [28] . These two particular phenazines are toxic to wild-type nematodes [28] and , as expected , the mixture of phenazine-1-carboxylic acid and 1-hydroxyphenazine rapidly killed L4 animals ( Figure S6B ) . Young adult wild-type animals were resistant to the lethal effects of these molecules ( Figure 7B ) . However , mdt-15 ( RNAi ) young adult animals were dramatically susceptible to phenazine-mediated killing ( Figure 7B ) . We also found that pmk-1 ( km25 ) loss-of-function mutants were more susceptible to P . aeruginosa in the “fast killing” assay than wild-type animals , although were less susceptible than mdt-15 ( tm2182 ) animals . This lethality , however , was mediated by factors other than phenazine toxins , since there was no difference in pathogenicity of the wild-type and Δphz P . aeruginosa toward pmk-1 ( km25 ) mutants ( Figure 7A ) . Likewise , pmk-1 ( RNAi ) animals were not more susceptible to phenazines-1-carboxylic acid and 1-hydroxyphenazine than L4440 RNAi control animals ( Figure 7B ) . Together , these data define a role for MDT-15 , but not PMK-1 , in the protection from bacterial-derived phenazine toxins .
Detecting and countering environmental threats is central to the ability of organisms to survive and reproduce in the wild . We examined the C . elegans response to the xenobiotic RPW-24 , which is able to induce a host immune response that is protective for animals infected with the lethal bacterial pathogen P . aeruginosa [9] . In an RNAi screen with RPW-24 , we identified a number of genes , including mdt-15/MED15 , which are required for induction of p38 MAP kinase PMK-1-dependent immune effectors . mdt-15 encodes a subunit of the highly conserved Mediator complex that controls the activation of a variety of genes involved in the response to external stress . We demonstrate that: ( i ) MDT-15 is required for the induction of p38 MAP kinase PMK-1-dependent immune effectors following exposure to a xenobiotic toxin , as well as during infection with P . aeruginosa , ( ii ) MDT-15 controls the expression of some p38 MAP kinase PMK-1-independent immune effectors , but not all defense genes , ( iii ) , MDT-15 functions downstream of the p38 MAP kinase PMK-1 cascade to control the induction of at least two immune effectors , ( iv ) the induction of xenobiotic detoxification genes and protection from the toxic effects of RPW-24 requires MDT-15 , but not the p38 MAP kinase PMK-1 , and ( iv ) MDT-15 is necessary for protection during P . aeruginosa infection and from phenazine toxins secreted by this organism . The Mediator complex is strongly conserved from yeasts to humans , and is required for transcription by physically interacting with and directing the activity of RNA polymerase II [30] , [31] . Although initial studies from yeast described the Mediator complex as a general regulator of transcription [32] , it is now becoming clear that the individual subunits of this complex , of which there are at least 26 , play important roles in translating inputs from cell signaling pathways to specific outputs [30] , [31] , [33] . For example , the regulation of chemotherapeutic resistance in cancer cells requires MED12 [33] , MED23 channels MAP kinase signaling activity to coordinate cell growth [34] , and MED15 serves as a regulatory node in lipid metabolism by directing the activity of the transcriptional activator SREBP ( sterol regulatory element binding protein ) [20] . Interestingly , MED15's role as a lipid sensor is strongly conserved . In C . elegans , the MED15 homolog MDT-15 interacts with SBP-1 , the SREBP homolog , and serves as an important regulator of lipid metabolism [18]–[20] . Likewise , the MED15 homolog in yeast ( called GAL11p ) serves a similar function [35] . Taubert et al . showed in a C . elegans genome-wide microarray analysis that MDT-15 coordinates the transcription of Phase I and Phase II detoxification genes , and is required to resist the lethal effects of a xenobiotic toxin [19] and agents that induce oxidative stress [21] . Consistent with these observations , we found that MDT-15 is necessary to mount a detoxification response toward and provide protection from RPW-24 , a xenobiotic toxin . Our studies extend the known functions of this conserved Mediator subunit to innate immunity in demonstrating that MDT-15 plays a critical role in controlling immune and detoxification pathway activation in a manner that is important for protection from bacterial infection and from the secreted phenazine toxins of P . aeruginosa . We propose that the Mediator subunit MDT-15 links innate immune activation to xenobiotic detoxification responses as part of a general strategy to ensure survival in harsh environments . In their natural habits , nematodes encounter numerous threats from ingested pathogens and xenobiotic toxins [1] . Thus , coupling xenobiotic detoxification to innate immune activation may have been selected for by pathogens that secrete soluble toxins during infection . Whether the protection from phenazine toxins mediated by MDT-15 requires detoxification genes or occurs via another mechanism is not known , but it is interesting to note that MDT-15's function as a regulator of host protection may be evolutionarily conserved . In yeasts , the MDT-15 homolog Gal11p coordinates a protective cellular response following xenobiotic exposure , which is manifest as the upregulation of drug efflux pumps [36] . The data presented in this study imply the existence of surveillance mechanisms that monitor for xenobiotic toxins ( or their effects ) and signal through MDT-15/MED15 to simultaneously activate protective detoxification and innate immune responses . Whether such activation occurs in the context of xenobiotic-induced organelle dysfunction is not known . McEwan et al . [15] and Dunbar et al . [16] found that C . elegans monitor the integrity of its translational machinery as a means to detect pathogen invasion . Inhibition of translation by the secreted bacterial toxin ToxA leads to an antibacterial response via known immune pathways involving the transcription factor ZIP-2 and the p38 MAP kinase PMK-1 . Several observations suggest , however , that MDT-15-mediated regulation of immune activation and detoxification gene induction by RPW-24 does not occur via a mechanism involving translation inhibition . First , we previously found that a loss-of-function mutation in the zip-2 gene did not affect the ability of RPW-24 to extend the survival of animals infected with P . aeruginosa and genome-wide transcriptional analysis of animals exposed to RPW-24 did not suggest that this compound is an inhibitor of translation [9] . Second , the transcriptome analysis of animals exposed to ToxA did not show an abundance of Phase I and II detoxification genes [15] . Finally , we show here that the ToxA-responsive , ZIP-2-regulated genes , irg-1 and irg-2 , are induced in mdt-15 ( RNAi ) animals during P . aeruginosa infection to levels comparable to that observed in wild-type animals . Melo et al . recently studied the behavioral response of C . elegans to xenobiotic toxins known to disrupt the function of the mitochondria , the ribosome , and the endoplasmic reticulum [3] . Inhibiting these essential processes by the action of small molecules or through targeted gene disruptions triggered an avoidance response , which required serotonergic and JNK signaling pathways . They proposed that organisms monitor disruption of core metabolic processes as a means to detect pathogen invasion and challenges from xenobiotic toxins . Based on the data reported here , however , it is not clear , whether immune signaling is an integral part of xenobiotic-elicited avoidance behavior , at least with respect to RPW-24 . We found that the function of neither MDT-15 nor PMK-1 was required for the avoidance of RPW-24 , indicating that the behavioral component of this protective response occurs upstream of MDT-15 , or via separate mechanism altogether . It will be interesting to determine the mechanism by which MDT-15 activates immune and detoxification responses in C . elegans . In our genetic screen for regulators of the p38 MAP kinase PMK-1 pathway , we identified a number of genes that are involved in fatty acid biosynthesis , including mdt-15 , fat-6 , fat-7 , elo-5 , acs-19 and C25A1 . 5 . Indeed , MDT-15 is known to control the expression of fat-6 and fat-7 [18]–[20] . It is therefore possible that a fatty acid signaling molecule or membrane component is required for p38 MAP kinase PMK-1 activity . We found , however , in epistasis analyses with the MAP kinase phosphatase VHP-1 that MDT-15 functions downstream of PMK-1 to coordinate the expression of F08G5 . 6 and F35E12 . 5 . Thus , for at least a subset of immune genes , MDT-15 likely also physically interacts with sequence-specific regulators , such as ATF-7 , a transcription factor that is the downstream signaling target of the p38 MAP kinase PMK-1 pathway [12] , to coordinate protective host responses mounted following exposure to xenobiotic toxins .
C . elegans were grown on standard NGM plates with E . coli OP50 [37] unless otherwise noted . The previously published C . elegans strains used in this study were: N2 Bristol [37] , pmk-1 ( km25 ) [5] , AY101 [acIs101[pDB09 . 1 ( pF35E12 . 5::GFP ) ; pRF4 ( rol-6 ( su1006 ) ) ] [10] , XA7702 mdt-15 ( tm2182 ) [19] , [21] , CF512 fer-15 ( b26 ) ;fem-1 ( hc17 ) [38] , and AU0133 [agIs17 ( pirg-1::GFP; pmyo-2::mCherry ) ] [14] . The C . elegans strains created for this study were: AU0307 [agIs44 ( pF08G5 . 6::GFP::unc-54-3′UTR; pmyo-2::mCherry ) ] , AU0316 [mdt-15 ( tm2182 ) ; agIs44] , AU0325 [mdt-15 ( tm2182 ) ; agEx116 ( mdt-15;pmyo-3::mCherry ) ] , AU0326 [mdt-15 ( tm2182 ) ; agEx117 ( mdt-15;pmyo-3::mCherry ) ] , AU0327 [mdt-15 ( tm2182 ) ; agEx118 ( mdt-15;pmyo-3::mCherry ) ] and AU0323 [mdt-15 ( tm2182 ) ; agIs44; agEx114 ( mdt-15;pmyo-3::mCherry ) ] . The strain carrying agIs44 was constructed by PCR amplification from N2 genomic DNA of an 851 bp region upstream of the start codon of the F08G5 . 6 gene ( primers GACTTGTCAAATGAACAATTTTATCAAATCTCA and CGCCTAGGTGTCAATTGATAATGAATA ) and ligated to the GFP coding region and unc-54-3′UTR sequences amplified from pPD95 . 75 using published primers , and a previously described protocol [39] . The agIs44 construct was transformed into N2 animals with the co-injection marker pmyo-2::mCherry using established methods [40] . A strain carrying the pF08G5 . 6::GFP::unc-54-3′UTR and pmyo-2::mCherry transgenes in an extrachromosomal array was irradiated , and strains carrying the integrated array agIs44 were isolated . AU0307 was backcrossed to N2 five times . The mdt-15 rescuing arrays agEx116 , agEx117 and agEx118 contain a 4 . 8 kb mdt-15 genomic fragment , which includes 707 bp upstream and 1075 bp downstream of the mdt-15 coding region , amplified from N2 genomic DNA ( primers GGAGTATCAGAAGCTCACGATGCTC and CCAAATAATACTAACCACCACATATCTTCCATT ) . This mdt-15 genomic fragment was transformed into N2 animals or AU0316 with the co-injection marker pmyo-3::mCherry using established methods . RNAi clones presented in this study were from the Ahringer [41] or Vidal [42] RNAi libraries unless otherwise stated . The atf-7 [12] and the pmk-1 [5] RNAi clones have been previously reported . All RNAi clones presented in this study have been confirmed by sequencing . The P . aeruginosa strain PA14 were used for all studies , unless otherwise indicated . The P . aeruginosa strains used in Figure S1 have been previously described [13] and were ( in order of descending virulence toward C . elegans ) : CF18 , PA14 , MSH10 , S54485 , PA01 , PAK , 19660 , and E2 . The P . aeruginosa PA14 phenazine null mutant ( Δphz ) lacks both the phzA1-G1 and phzA2-G2 operons and has been previously described [43] . The BL21 E . coli strain that expresses the bacterial toxin Exotoxin A ( ToxA ) has been previously described [15] . 1 , 420 RNAi clones that correspond to genes expressed in the C . elegans intestine based on their annotation in Wormbase ( www . wormbase . org ) in April , 2008 were selected from the Ahringer [41] or Vidal [42] RNAi libraries ( see Table S1B ) . RNAi clones were pinned into 1 . 2 ml of LB plus 100 µg/ml carbenicillin in 96-well culture blocks ( Corning Incorporated ) and grown overnight at 37°C with shaking at 950 RPM in a Multitron II Shaking Incubator ( Appropriate Technical Resources ) . 40 µL of the 10× concentrated overnight culture were added to each well of a 24-well plate containing RNAi agar medium and grown overnight at room temperature . The following day , 50–100 L1 staged AU0307 animals , which carry the agIs44 transgene , were added to each well and allowed to grow until they were at the L4 or young adult stage . Worms were then transferred to new 24-well screening plates containing 1 mL of “slow kill” media supplemented with 70 µM RPW-24 and seeded with E . coli OP50 food . Animals were dried on the screening plates for several hours at room temperature and then incubated overnight at 20C . The L4440 vector and pmk-1 RNAi clones were included on each of the screening plates as the negative and positive controls , respectively . Animals were scored for GFP expression and rated on a subjective scale from 0 ( no GFP expression in response to RPW-24 ) to 3 ( RPW-24-mediated induction of GFP expression equivalent to L4440 ) . Exposure of the C . elegans transcriptional reporter irg-1::GFP to an E . coli strain that expresses ToxA was performed as previously described [15] . “Slow killing” P . aeruginosa infection were performed as previously described [9] , [44] . In all of these assays , the final concentration of DMSO was 1% and RPW-24 was used at a concentration of 70 µM , unless otherwise indicated . The propensity of wild-type C . elegans to leave a lawn of bacteria supplemented with RPW-24 was assayed using a previously described protocol [3] , [9] with minor modifications . Rather than adding the toxin on top of the small lawn of food , 20 µg of RPW-24 was mixed with E . coli OP50 , which was spotted onto NGM plates . To assess the toxicity of RPW-24 , we assayed the development of animals exposed to vector control ( L4440 ) , pmk-1 ( RNAi ) and mdt-15 ( RNAi ) in the presence of 70 µM RPW-24 , as previously described [9] . P . aeruginosa “fast kill” pathogenesis assays were conducted with late L4 and early young adult animals ( picked 1–3 hours after the L4 molt ) obtained from timed egg lays as described [28] , [29] . For the killing assay using toxic phenazines , 50 µg/ml phenazine-1-carboxylic acid ( PCA ) and 5 µg/ml 1-hydroxyphenazine in DMSO were added to modified “fast kill” media ( 1% bacto-peptone , 1% glucose , 1% NaCl , 150 mM sorbitol , 1 . 7% bacto agar , 5 µg/ml cholesterol and 50 mM sodium citrate , pH 5 ) [28] . These phenazine concentrations correspond to the amount of PCA and 1-hydroxy-phenazine that are produced under “fast kill” conditions [28] . E . coli OP50 was used as the food source . The modified “fast kill” media pH 5 . 0 plus 1% DMSO was used as the control condition . These assays were incubated at 21–23°C . Synchronized L1 staged C . elegans N2 animals were grown to L4/young adult stage on the indicated RNAi strain , transferred to assay plates and incubated at 25°C for 24 hours . To prepare the assay plates , 70 µM RPW-24 or DMSO was added to 20 mL “slow killing” media [44] in 10 cm petri dishes seeded with E . coli OP50 . N2 and pmk-1 ( km25 ) animals were raised on E . coli OP50 and exposed to the above conditions for 18 hours at 20°C . For qRT-PCR studies of nematodes infected with P . aeruginosa PA14 or the indicated strain of P . aeruginosa , 20 mL of “slow killing” media containing either DMSO or 70 µM RPW-24 was added to 10 cm petri dishes . Plates were seeded with either 75 µL of E . coli OP50 or P . aeruginosa , each from cultures grown for 15 hours at 37°C . The plates were incubated for 24 hours at 37°C and 24 hours at 25°C . L4/young adult animals were added to the assay plates and incubated at 25°C for eight hours . RNA was isolated using TriReagent ( Molecular Research Center , Inc . ) and analyzed by NanoString nCounter Gene Expression Analysis ( NanoString Technologies ) using a “codeset” designed by NanoString that contained probes for 118 C . elegans genes ( Table S2 ) . Probe hybridization , data acquisition and analysis were performed according to instructions from NanoString with each RNA sample normalized to the control genes snb-1 , ama-1 and act-1 . For the qRT-PCR studes , RNA was reverse transcribed to cDNA using the Retroscript kit ( Life Technologies ) and analyzed using a CFX1000 machine ( Bio-Rad ) with previously published primers [6] , [18] . The qRT-PCR primers for the mdt-15 , pmk-1 , tir-1 and sek-1 genes were designed for this study and are available upon request . All values were normalized against the control gene snb-1 . Fold change was calculated using the Pfaffl method [45] . Nematodes were mounted onto agar pads , paralyzed with 10 mM levamisole ( Sigma ) and photographed using a Zeiss AXIO Imager Z1 microscope with a Zeiss AxioCam HRm camera and Axiovision 4 . 6 ( Zeiss ) software . For comparisons of GFP expression in the F08G5 . 6::GFP transgenic animals , photographs were acquired using the same imaging conditions . Differences in survival of C . elegans animals in the P . aeruginosa pathogenesis assays were determined with the log-rank test in each of two biological replicates . Differences were considered significant only if the p value was less than 0 . 05 for both replicates . In the manuscript , data from one experiement that is representative of both replicates is shown and the sample sizes for these experiments are given in Table S3 . To determine if the increase in survival conferred by RPW-24 treatment was different in one population compared to another , we examined the difference in the effect of RPW-24 treatment on the hazard in each group using a Cox proportional hazard model ( Stata13 , Stata , College Station , TX ) from two biological replicates , as previously described [9] . Fold changes in the qRT-PCR analyses were compared using unpaired , two-tailed student t-tests . Accession numbers for genes and gene products are given for the publically available database Wormbase ( http://www . wormbase . org ) . The accession numbers for the principal genes mentioned in this paper are: atf-7 ( C07G2 . 2 ) , C32H11 . 1 , cyp-35A1 ( C03G6 . 14 ) , cyp-35B2 ( K07C6 . 3 ) , cyp-35C1 ( C06B3 . 3 ) , F35E12 . 5 , F08G5 . 6 , F01D5 . 5 , fshr-1 ( C50H2 . 1 ) , irg-1 ( C07G3 . 2 ) , irg-2 ( C49G7 . 5 ) , mdt-15 ( R12B2 . 5 ) , nsy-1 ( F59A6 . 1 ) , pmk-1 ( B0218 . 3 ) , sek-1 ( R03G5 . 2 ) , skn-1 ( T19E7 . 2 ) , tir-1 ( F13B10 . 1 ) , and zip-2 ( K02F3 . 4 ) . Other accession numbers are given in Figure 2 , Figure 6 , Table S1 and Table S2 .
|
Metazoans respond to environmental threats in part through conserved pathways that coordinate protective transcriptional responses . During infection with an invasive pathogen , for example , innate immune pathways regulate the secretion of antimicrobial immune effectors . Likewise , exposure to toxic molecules leads to the induction of detoxification mechanisms that protect the host from the deleterious effects of these compounds . Here we find that a conserved transcriptional regulator MDT-15/MED15 links xenobiotic detoxification and immune responses in a manner that is important for protection during bacterial infection . We also show that MDT-15/MED15 is necessary for the host to resist the lethal effects of secreted toxins produced by pathogenic bacteria . Rapid coordination of these protective host responses through MDT-15/MED15 may therefore be part of a conserved survival strategy in the wild .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"invertebrates",
"innate",
"immune",
"system",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"caenorhabditis",
"immunology",
"microbiology",
"animals",
"animal",
"models",
"bacterial",
"diseases",
"caenorhabditis",
"elegans",
"model",
"organisms",
"pseudomonas",
"infections",
"research",
"and",
"analysis",
"methods",
"infectious",
"diseases",
"pathogenesis",
"immune",
"system",
"immunity",
"host-pathogen",
"interactions",
"nematoda",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2014
|
The Evolutionarily Conserved Mediator Subunit MDT-15/MED15 Links Protective Innate Immune Responses and Xenobiotic Detoxification
|
Populations of human cytomegalovirus ( HCMV ) , a large DNA virus , are highly polymorphic in patient samples , which may allow for rapid evolution within human hosts . To understand HCMV evolution , longitudinally sampled genomic populations from the urine and plasma of 5 infants with symptomatic congenital HCMV infection were analyzed . Temporal and compartmental variability of viral populations were quantified using high throughput sequencing and population genetics approaches . HCMV populations were generally stable over time , with ∼88% of SNPs displaying similar frequencies . However , samples collected from plasma and urine of the same patient at the same time were highly differentiated with approximately 1700 consensus sequence SNPs ( 1 . 2% of the genome ) identified between compartments . This inter-compartment differentiation was comparable to the differentiation observed in unrelated hosts . Models of demography ( i . e . , changes in population size and structure ) and positive selection were evaluated to explain the observed patterns of variation . Evidence for strong bottlenecks ( >90% reduction in viral population size ) was consistent among all patients . From the timing of the bottlenecks , we conclude that fetal infection occurred between 13–18 weeks gestational age in patients analyzed , while colonization of the urine compartment followed roughly 2 months later . The timing of these bottlenecks is consistent with the clinical histories of congenital HCMV infections . We next inferred that positive selection plays a small but measurable role in viral evolution within a single compartment . However , positive selection appears to be a strong and pervasive driver of evolution associated with compartmentalization , affecting ≥34 of the 167 open reading frames ( ∼20% ) of the genome . This work offers the most detailed map of HCMV in vivo evolution to date and provides evidence that viral populations can be stable or rapidly differentiate , depending on host environment . The application of population genetic methods to these data provides clinically useful information , such as the timing of infection and compartment colonization .
Human cytomegalovirus ( HCMV ) is a β-herpesvirus with seroprevalence of 30–90% in the United States [1] . In healthy individuals , primary HCMV infection is usually asymptomatic or causes a mild febrile illness . However , in the immuno-compromised or immune-naïve host , HCMV infection can lead to more severe outcomes , such as graft rejection or death . In particular , an estimated 0 . 7% of live-born infants per year ( 30 , 000 in the U . S . ) are diagnosed with congenital CMV infection , and nearly 20% exhibit permanent neurologic sequelae [2] . HCMV has been shown to be highly polymorphic both among and within human hosts [3]–[6] , and previous studies have used inter-host or inter-species divergence data to infer the evolutionary path of HCMV ( and other herpesviruses ) over long timescales ( thousands to millions of years ) [7]–[10] . These studies have shown that the virus is well-adapted to its natural host , using an impressive array of immune evasion and modulation genes to establish life-long infections [10] , [11] . However , the possibility that the virus evolves within human hosts on much shorter timescales has been poorly studied . Importantly , it has recently been shown that HCMV genomic and genetic populations are highly polymorphic , with levels of intrahost diversity comparable to those observed in RNA virus populations [5] , [6] . Because variation is the substrate upon which selection can act , it is possible that intrahost variation of HCMV populations allows for rapid evolution within the human host across relatively short timescales . To better understand this process , we used high throughput sequencing to sample HCMV genomic populations from the urine and plasma of 5 infants with symptomatic congenital HCMV infection at multiple time points during the first year of age . This approach allowed us to monitor genome-wide evolution of the virus and to determine the mechanisms that contribute to pathogen evolution . We find that HCMV populations can evolve slowly within the same tissue compartment , or rapidly when moving between compartments . In order to characterize the mode and tempo of this differentiation , we constructed detailed maps of the in vivo demographic history of HCMV populations , and tested for signatures of positive selection . In total , this work provides the most complete insight to date into the evolutionary paths of a DNA virus as well as the mechanisms that guide these routes in human hosts .
Our previous work showed significant variability within HCMV populations sampled from the urine of congenitally infected neonates - variation that could potentially enable rapid evolution of the viral population [6] . To study the evolution of HCMV in human hosts , HCMV genomic populations were examined within fluid specimens from 5 infants with symptomatic congenital infection . Specimens were longitudinally collected from the urine ( patients B101 , MS1 and MS2 ) , plasma ( patient M103 ) or both the urine and plasma ( patient B103 ) during the first year after birth ( 14 specimens in total ) ( Table 1 ) . Patients B101 , M103 and B103 are unrelated , while patients MS1 and MS2 are monozygotic , monochorionic twins . A previously described high throughput sequencing strategy [6] was used to study the viral populations . Briefly , HCMV genomic DNA was specifically amplified through a series of PCR reactions , the amplified DNA pools were sequenced , and the sequence output was analyzed for variants within and among viral populations . Due to improvements in high throughput sequencing technology since our initial report , the frequencies at which variants are detected has improved from ≥1 . 9% [6] to ≥1 . 02% and the error rate after filtering has improved from 6 . 7% [6] to 5 . 0% ( Figure S1 ) . An average of 854 megabases of sequence information was produced for each of the 14 specimens ( Table S1 ) . The sequence data covered an average of 85% of the genome to an average depth of 1921 genome equivalents ( Table S1 ) . The high depth and coverage allows for quantitative description of the evolution of the HCMV in vivo populations . Initially , the populations were studied separately by quantifying the intrahost variation of each population . Consistent with previous work , thousands of single nucleotide polymorphisms ( SNPs ) were observed in each population ( Table 2 ) . On average , ∼5 , 400 SNPs were identified per population , which accounts for ∼3% of the positions in the viral genomes sequenced . Similar levels of variability existed at both the nucleotide and amino acid levels . To quantify the variability of the populations , nucleotide diversity values were estimated from the sequence data . These values were comparable to previous work [6] , with a mean nucleotide diversity value of 0 . 12% across all populations . There was a negative correlation between nucleotide diversity of the viral population and time of sample collection , though the correlation was not significant ( data not shown ) . However , a significant correlation was found between nucleotide diversity and the compartment from which the population was sampled ( P = 0 . 046 , Mann-Whitney ) , with urine populations exhibiting lower diversity levels than plasma populations ( Figure 1 ) . To better understand the patterns of sequence variability , the HCMV genomic populations were studied as pairs across time or compartments . Eight pairings were created for host-matched specimens , and the frequencies of all SNPs were tracked across the pairings ( Figure 2 , Figure S2 ) . A complex picture of HCMV evolution emerged from this analysis . For most SNPs , the frequency was either high or low and remained stable between paired specimens ( seen as black bands at the top or bottom of the panels ) . This pattern was observed in all pairings but was most apparent in longitudinal tracking of SNPs in urine populations ( Figure 2 , Panel A ) . Indeed , for the 36 , 131 SNPs tracked for all specimens , 31 , 919 ( 88% ) changed in frequency by ≤10% ( median change = 1 . 14% ) . This pattern suggests that the frequencies of the majority of SNPs in these HCMV populations are stable . Figure S2 shows that the populations with the most stable SNPs trajectories ( slopes≈0 ) also appear to have the largest population sizes as measured by viral load ( e . g . urine compartment populations ) ( Table 1 ) . This pattern is consistent with the relationship of drift and population size , in which the effect of drift on the frequencies of neutral SNPs is inversely proportional to population size [12]–[14] . In contrast , subsets of SNPs were observed to rise from low frequency to high frequency or vice versa . This pattern is most apparent when comparing populations sampled across compartments ( for example Figure 2 , Panel C ) . Lastly , specimens were collected at 3 time points from patients MS1 and MS2 . In these populations , 0 . 5% of SNPs changed frequency by ≥50% between each time point , moving from low to high to low or vice versa ( seen as “V” or inverted “V” patterns in Figure 2 , Panel D; Figure S2 , Panel H ) . These patterns suggest that HCMV populations are dynamic and do not completely stabilize in congenitally infected patients during the first year after birth . A consensus sequence was then called for each sampled population . The consensus sequences of two populations of interest were aligned and SNPs identified between the sequences ( Table 3 and Figure 2 , Panels E–H ) . Phylogenetic trees were also constructed from all consensus sequences in this study ( Figure 3 ) . Longitudinal specimens collected from a single compartment of a single host were closely related as indicated by few SNPs between consensus sequences and short branch lengths on the phylogenetic tree . For example , 49 consensus sequence SNPs were identified between the two B103 urine populations , 20 of these SNPs were non-synonymous . Thus , the consensus sequences differed by only 0 . 02% at the nucleotide level ( π ) and 0 . 04% at the amino acid level ( πAA ) ( Table 3 ) . A similar result was observed for longitudinally-sampled plasma populations . For the MS1 and MS2 twins , an average of 149 SNPs ( 0 . 08% difference ) was identified between urine specimen consensus sequences at the same time point and all MS1 and MS2 sequences clustered together on the phylogenetic tree . In contrast , for infant B103 , an order of magnitude increase in the prevalence of consensus sequence SNPs was identified between urine and plasma populations at the same time point . Comparing the consensus sequences of the urine and plasma populations , 1 , 602 SNPs ( ∼0 . 9% difference at the nucleotide and amino acid levels ) were identified at 1 week , and 1 , 771 SNPs ( ∼1 . 2% difference ) were identified at 6 months . Previously , we have shown that HCMV consensus sequences from different hosts differ by ∼1 . 1% [6] . Further , the B103 plasma sequences are clearly distant from the B103 urine sequences on the phylogenetic tree ( Figure 3 ) . Intriguingly , the urine samples appear as a divergent group on the tree , with B101 sequences being an outlier in this grouping , while the plasma sequences from the two unrelated hosts are closely clustered . This result is consistent with convergent evolution acting on HCMV plasma populations , though the conclusion is tentative given the small sample size . Taken together , these results suggest that HCMV collected from different compartments of a single host are as divergent , or even more divergent , than HCMV collected from unrelated hosts , but that sequences sampled longitudinally from the same compartment of the same host or from monozygotic , monochorionic twins are similar . In the above analysis , comparisons were made between consensus sequences , which incorporate only high frequency SNPs of the populations . To determine if similar levels of differentiation were present for comparisons of all SNPs , FST [15] , [16] was estimated for the specimen pairings . FST is a statistic bounded by 0 and 1 that increases with increasing variation in SNP frequencies between populations [17] , and is a commonly to measure differentiation between two populations of the same species [18] . In this study , FST was used to summarize population differentiation either across time or compartments . FST values of the viral populations in longitudinal populations sampled from urine or plasma compartments were low ( average = 0 . 13 , range [0 . 05–0 . 23] ) ( Figure 3 , Table 3 ) . Thus , SNP frequencies were generally stable within compartments during the sampling time frame , giving statistical support to the observations from Figure 2 . It is noted that the longitudinal FST values for MS1 and MS2 populations ( which were sampled over the longest timespan ) rose with the last collection time points ( Figure 3 ) . It is unclear whether this trend continued or reached a plateau as no later specimens were available . The highest FST values , and thus the largest population differentiation , were observed for infant B103 when comparing populations sampled from different compartment specimens ( urine and plasma ) collected at the same time point ( Figure 3 , Table 3 ) . FST was equal to 0 . 42 at 1 week and 0 . 45 at 6 months . These values are similar to those estimated between HCMV populations of two unrelated hosts ( specifically , urine specimens from infants U01 and U04 that were described previously [6] ) – thus , within host compartment differentiation can be as great as between host differentiation ( Figure 3 ) . . These results corroborate the consensus sequence analysis , and suggest that HCMV populations rapidly differentiate between compartments of a single host , but are generally stable over time within the same compartment of a single host or between monozygotic , monochorionic twins . These data led us to explore the mechanisms of both the population stability and rapid differentiation of the HCMV populations observed in this study . From population genetics , it is known that both demography and selection can lead to large changes in SNP frequencies in relatively short timespans [19] . Demographic events are changes of population size and structure , such as population bottlenecks , expansions and splits , which can lead to stochastic changes in allele frequencies . In contrast , selection is mediated by variations in fitness associated with different alleles and can cause deterministic changes in allele frequency . The high throughput sequence data was analyzed to infer the demographic history of the viral populations using the statistical framework described previously [20] . This method estimates the parameters of a demographic model by solving the appropriate diffusion equations . This approach has been used to study the demographic history of higher organisms , such as the migration of humans out of Africa [20] . The maximum likelihood values for the parameters of the demographic models were estimated , and uncertainties of the estimates were calculated through non-parametric bootstrapping ( Table S2 ) . Using this method , best fit demographic models were developed for the viral populations sampled from the 5 hosts in this study ( Figure 4 ) . All population sizes within the model are reported relative to their ancestral population ( which is set to 1 ) . In support of modeling accuracy , there was good agreement between the relative population sizes as calculated from the models and as measured empirically with qPCR ( Tables 1 and S2 ) . The demographic history was first inferred for the B103-sampled viral populations given the rapid inter-compartment differentiation observed . In the best-fitting B103 demographic model ( Figure 4A ) , an ancestral population experienced a bottleneck at time 0 and seeded the plasma compartment . The bottleneck reduced population size by approximately 99 . 7% . The resultant population split 10 weeks later , when 0 . 41% of the plasma population colonized the urine compartment . The urine population then expanded nearly 2000-fold by the time of the initial collection ( 1 week postnatally ) . The urine population continued to expand , growing 4-fold by the second collection at 6 months . In the plasma compartment , the viral population stayed relatively stable throughout the model , increasing in size by approximately 22-fold from the initial bottleneck to the first collection and decreasing by 45% by the 6 months collection . The model also includes asymmetrical migration between the compartments , with 9-fold higher flow of alleles from the urine to the plasma before the first collection and 15-fold higher flow before the second collection ( Table S2 ) . The B101 and M103 best-fit models appear to agree qualitatively with that of B103 with respect to the demographic histories of in vivo viral populations ( Figures 4B and 4C ) . B101 and M103 viral populations also experienced bottlenecks at time 0 that reduced population size by 99 . 1% and 99 . 8% , respectively ( Table S2 ) . For B101 , the population experienced a second bottleneck approximately 11 weeks after the initial bottleneck , which reduced population size by 97 . 6% . The population then expanded and increased in size by ∼1100-fold by the first urine collection ( 7 months ) followed by an additional increase of ∼1 . 4-fold by the 10 month collection . For M103 , the viral population experienced a more modest expansion after the initial ( and only detected ) bottleneck , increasing in size by ∼140-fold and 2 . 4-fold by the plasma collections at 1 . 5 and 5 months after birth , respectively . Next , a best-fit demographic history was inferred from the MS1 and MS2 sequence data ( Figures 4D–E , Table S2 ) . This model differed from the previous models in several substantial ways . First , the initial bottleneck did not occur at time 0 , but rather at 0 . 466 months after time 0 ( i . e . the earliest event inferred in the model ) . Second , the previous models of populations sampled from the urine compartments ( B101 and B103 ) contained 2 bottlenecks in the population histories , while the best-fit model of MS1 and MS2 contained 3 bottlenecks , with the first bottleneck shared between the twins . Of note , the first 2 bottlenecks of the MS1 and MS2 model occurred sequentially , with contiguous confidence intervals for the timing of the bottlenecks ( Table S2 ) . In addition , these bottlenecks were less severe than those from other models , but the net effect of the 2 sequential bottlenecks was to reduce population size by 95% and 92% in the MS1 and MS2 lineages , respectively . Together , these reductions in population size are comparable to those inferred in the other models for a single bottleneck . Possible causes for the disagreements between the models are relayed in the Discussion . From the demographic model , some or all of the differentiation across compartments may be explained by a bottleneck effect associated with plasma populations colonizing the urine compartment , such as shown in Figure 4A . However , positive selection may also contribute to rapid differentiation due to the fixation of advantageous alleles as the virus moves to a new host environment . To test for evidence of positive selection within HCMV populations , the population branch statistic ( PBS ) [21] was employed . PBS measures localized increases in population differentiation between two closely related populations and a third outlier population [21] , with larger PBS values indicative of a higher probability that positive selection has locally altered SNP frequencies . The test has high power to detect recent positive selection , particularly when the selected SNP rises from standing variation [21] , an important consideration for this work . An HCMV population sampled from an unrelated host ( described previously [6] ) was used as the outlier in this study , and simulations under the demographic models in Figure 4 were used to determine the 5% significance threshold ( Figure S5 ) . The results of this analysis show that the effect of positive selection in shaping HCMV evolution is variable and dependent on context . Very few putative targets of positive selection were identified in the compartment-matched , longitudinal specimens from all infants ( Figures 5 and S3 , Tables 4 and S2 , S3 , S4 , S5 ) . In these populations , an average of 10 total SNPs and 3 non-synonymous SNPs were identified in the screen . UL7 , UL13 and UL73 ( encoding gN ) were identified as putative targets of positive selection in several populations , suggesting that variation in these genes can affect viral fitness among different hosts . Additionally , few targets were identified when screening for beneficial SNPs associated with the urine compartments of monozygotic twins ( MS1 and MS2 ) ( Figure S3 , Table 4 and S6 , S7 , S8 , S9 , S10 , 11 ) , where on an average 12 total SNPs and 2 non-synonymous SNPs were identified in each host at the 3 collection time points . In contrast , the assay yielded significantly different results when studying positive selection associated with movement of virus across compartments . The B103 populations sampled from the 1 week urine and plasma specimens were used to estimate the effect of positive selection associated with colonization of a new host compartment . 114 SNPs detected in the B103 urine compartment were identified as putative targets of positive selection , including 31 in non-coding regions and 83 in coding regions ( Figure 5 , Tables 4 and S13 ) . The coding SNPs were located in a total of 34 ORFs . Comparison of screens of positive selection ( Table 4 ) showed that the urine compartment colonization result was a significant outlier ( Grubbs test , P = 6 . 8×10−4 ) . These data suggest that positive selection is an important contributor to the rapid evolution of HCMV populations associated with colonization of distinct tissue compartments . Many of the SNPs detected in the screen of selection associated with urine compartment colonization showed positional clustering ( Figure S4 ) , suggesting that only a few SNPs were positively selected and the remainder were swept to high frequency via hitchhiking [22] , [23] . To identify which SNPs associated with compartment colonization may be targets of selection and not simply linked neutral variants , the coding SNPs were classified as synonymous or non-synonymous ( the presumed targets of positive selection ) . Of the 83 coding SNPs , 26 were non-synonymous mutations located in 16 ORFs ( Tables 4 and S12 ) . Within this group , several targets have been shown to exhibit strain or inter-host variability , such as RL5A [24] and UL55 ( encoding gB ) [25] . Thus , these data extend previous studies of HCMV polymorphisms by showing that differentiation can exist within hosts as well as between hosts , and that differentiation can be driven by positive selection . Because selection was observed in the urine compartment ( i . e . , kidney ) , these results suggest that the RL5A and UL55 SNPs offer a fitness advantage in this compartment , possibly due to their effects on cell tropism as suggested previously [26] . Positively selected nonsynonymous SNPs were also localized to US2 , US3 , and US7 , which are contained within the US2 through US11 block of ORFs that are important for immune evasion by interference of the antigen presentation machinery [27] . Many ORFs of unknown function were identified in this screen , some with multiple nonsynonymous SNPs , such as UL124 . However , the functional effects of these identified SNPs are still unknown . Lastly , it was observed that positively selected SNPs ( Tables S4 , S5 , S6 , S7 , S8 , S9 , S10 , S11 , S12 , S13 , S14 ) were both pre-existing ( i . e . , identified in the ancestral population ) and newly arising ( i . e . , not identified in the ancestral population ) . The relative contribution of each in driving differentiation was calculated using genome wide data . The analysis showed that both pre-existing and novel SNPs contribute to HCMV evolution ( Figure 6 ) , though there was a significant difference between the relative contributions of the two classes ( Chi-square , P = 1 . 4×10−11 ) . Specifically , pre-existing SNPs were more frequently identified in longitudinal plasma populations , while novel SNPs were more prevalent in the cross-compartment comparisons . In total , these results suggest that both pre-existing and de novo mutations contribute with varying degrees to the evolution of HCMV within human hosts .
HCMV has been shown to exhibit significant inter-host sequence divergence as well as high levels of intra-host variability . Here we provide a detailed study of the evolution of HCMV in human hosts using specimens collected across tissue compartments and time . Furthermore , we provide evidence of rapid evolution of HCMV populations associated with movement of the virus between host compartments . The rapid divergence can best be explained by a demographic history that includes population bottlenecks and expansions , as well as positive selection across a subset of loci . This study shows that the evolution of HCMV within human hosts can proceed in two ways . HCMV populations are relatively stable within tissue compartments . For example , the urine populations of B101 and B103 exhibited only 52 and 49 consensus sequence SNPs , respectively , and low FST values . This result is in agreement with previous studies that have monitored HCMV genetic sequences within patients [5] , [28] . In addition , HCMV populations were stable across hosts , albeit for the unique case of monozygotic , monochorionic twins when sampled from the same compartment . In these cases , demographic modeling suggested stable population size and structure , and the trajectory of many SNPs is consistent with the effects of drift . In contrast , HCMV populations can also rapidly evolve during colonization of different compartments within a host . Significant HCMV inter-host differentiation is well documented , both at the genetic and genomic levels [3] , [6] , [24] , [29] , [30] . Here we show that a similar level of differentiation can be observed within different compartments of a single host ( here represented by the plasma and urine ) . However , HCMV can be found in many other organs [4] and bodily fluids ( e . g . , saliva , breast milk , and semen ) . While the large differentiation between compartments and the patterns of bottlenecks and expansions observed in our study may be unique to plasma and urine , any movement between other compartments would likely involve similar demographic histories . Moreover , each compartment may have unique selective pressures as reported here for the plasma and urine compartments . It is therefore probable that most compartments will show levels of differentiation similar to those reported in this study . One reason that the plasma and urine compartments were chosen for study is because they represent the circulating and shed HCMV populations , respectively . Plasma virus is likely limited to the individual ( with the important exceptions of transfusion/transplantation-associated and congenital infections ) whereas virus shed into urine is excreted and possibly infecting other hosts . The results in this work suggest that there are large differences between viral populations isolated from these two sources . The potential implications of these findings are numerous . For example , do studies of secreted virus , such as from the urine and/or saliva , provide useful information about circulating virus or infection-associated diseases ? In addition , how do SNPs associated with shedding viral populations ( e . g . , urine populations ) affect tropism for mucosal surfaces , the presumed route of most inter-host infections ? The evidence for positive selection when crossing into the urine compartment suggests that the viral population has adapted for increased fitness , but most likely increased fitness in renal epithelial cells . Thus , it is not known how compartmentalization affects inter-host infectivity . Alternatively , because at least a portion of the differences between populations could be explained by the neutral demographic history , some of the SNPs would have arisen stochastically and may have little or no effect on viral fitness . Therefore , it will be interesting to determine the phenotypic differences between patient-derived HCMV populations from different compartments . Demographic histories of HCMV populations were generated from population data in this study . These demographic models may provide a partial explanation for the population differentiation observed across compartments . The B103 model suggests that the urine and plasma populations split 3 months prior to the initial collection . It was also observed that these populations are highly differentiated , as measured by FST ( Figure 2 , Table 3 ) . From the B103 demographic model , at least a part of this differentiation may be explained by the strong bottleneck and large population expansion that occurs as the virus colonized the urine compartment . The bottleneck would reduce population size , leading to decreased diversity due to the stochastic elimination of ancestral SNPs ( which likely explains the results shown in Figure 1 ) . A decreased population size would also increase the speed and probability of fixation of neutral SNPs [13] , [14] . Secondly , DNA replication associated with population expansion could introduce new mutations to the urine compartment population . The demographic model generated from B103 data also suggests a novel mechanism for reinfection . In the model , there was asymmetric migration of alleles between host compartments with a higher flow from the urine to the plasma . In this study , the flow of alleles was relatively low and the plasma population size was fairly stable . One could speculate that after a collapse of the plasma population , due the effects of the adaptive immune response or antiviral therapy , migration between compartments would provide a source of highly differentiated virus . Thus , a clinical manifestation of this phenomenon could be intrahost ( i . e . self ) reinfection by virus sourced from various compartments . If this is true , effective HCMV vaccine design will need to account for not only plasma-associated epitopes but also epitopes sourced from other compartments . Future studies are needed to test the validity and prevalence of this mechanism of reinfection , and its potential effect on anti-HCMV clinical strategies . An important disagreement was noted between the B101 and B103 models and those of MS1 and MS2: the best-fit model of B101 and B103 urine populations contained 2 bottlenecks separated by ∼2 . 5 months , while the MS1 and MS2 model contained 3 bottlenecks , separated by 0 . 2 months and 2 . 2 months . One possible explanation for the disagreement is that in some hosts , HCMV undergoes 2 bottlenecks en route to the urine compartment , while in other hosts there are 3 bottlenecks . However , there is no clear biological mechanism to explain this difference . Alternatively , the MS1 and MS2 models appear to infer events further back in time than those of the other models . Thus , the first bottleneck in this model could reflect an earlier event for which the other datasets had insufficient power to detect . A third explanation is that the MS1 and MS2 models have higher-resolution than those inferred from the other datasets . To generate these models , one compares the spectrum of neutral allele frequencies of the experimental dataset to the spectra that would be expected from populations that have experienced various demographic events ( i . e . bottlenecks and/or expansions ) . The best fit model is one for which the experimental spectrum most closely matches the expected spectrum . Unfortunately , models of various population histories could generate similar allele frequency spectra , limiting the ability to fully resolve these models . However , because MS1 and MS2 viral populations were derived from the same ancestral population , two allele frequency spectra were estimated from populations that evolved in parallel from the same ancestor , providing more information about the evolutionary history of the populations and allowing similar models to be resolved . Thus , it is proposed that all HCMV urine populations have experienced at least 3 bottlenecks prior to colonization of the kidney . However , two bottlenecks may not be resolved by data from a single host due to the very close timing of the bottlenecks ( ∼1 week ) . The net effect of both bottlenecks is reduction of population size by >90% ( Table S2 ) . If three bottlenecks do indeed occur during the course of HCMV congenital infections , the biological cause of these reductions in population size is not known . We hypothesize that the first bottleneck results from the movement of the virus from the maternal compartment to the placenta . In all models , the timing of this event was ∼13–18 weeks gestational age and appears to agree with the known epidemiology and pathology of symptomatic HCMV congenital infections [31]–[33] . In the MS1 and MS2 models , the first bottleneck is shared and followed by a ∼1 week period in which the viral population expanded ∼4-fold before splitting and entering the separate , second bottlenecks of the two infant lineages . We propose that these events reflect replication of the virus in the twin's shared placenta , followed by independent infection of the fetal circulatory compartments . Replication of HCMV in placental cytotrophoblasts has been demonstrated experimentally [34] , and the data presented here offer a novel method to support the hypothesis that HCMV replicates within the placental compartment . We further hypothesize that the third bottleneck in the history of the urine populations represents movement of the virus from the plasma to the renal compartment , which occurred ∼9–11 weeks after the initial fetal infection event . Whether the timing and extent of the bottlenecks is consistent across HCMV congenital infections or differs between symptomatic and asymptomatic infections still needs to be studied . Nevertheless , our work demonstrates that viral demographic histories can be clinically informative for understanding viral infections . From the data in this study , it is tempting to speculate on the relative contribution of demography and selection to the evolution of HCMV populations . For example , 1602 SNPs were observed between the B103 urine and plasma compartment consensus sequences at 1 week of age ( Table 3 ) , of which 114 were putative targets of positive selection ( Table 4 ) . One could propose that ∼7% of SNPs rose in frequency due to positive selection ( either directly or via linkage ) , and the remainder were governed by genetic drift . However , caution must be used with these estimates . For example , a demographic model was first fit to the data , and then the signatures of positive selection were detected , which may result in over-fitting the demographic model [35] . Second , the test of positive selection may have excluded SNPs because of missing data due to incomplete coverage of the outlier group . Third , the time course of selection may overlap poorly with the time course of sample collection . For example , a SNP may have arisen from very low frequency to become fixed due to strong selection . A changing selective environment may have reduced or eliminated the selective advantage , allowing the SNP to drift to a lower frequency in the population . If the timing of sampling does not overlap the timing of selection , the SNP would not be identified as a target of positive selection . Thus , we are limited in our ability to accurately quantitate the relative contribution of demography and selection to the evolutionary patterns observed in this study . However , it is very reasonable to state based on the data as a whole that both demography and positive selection influence the intrahost evolution of HCMV , and thus both need to be accounted for in future studies of the natural history of HCMV and other viral infections .
The use of specimens from subjects B101 , B103 , and M103 was approved by the University of Massachusetts Medical School and Baystate Medical Center Institutional Review Boards . Subjects MS1 and MS2 clinical specimens were obtained from neonates with congenital HCMV infection during the course of routine clinical care at the University of Minnesota Medical School . Protocols for collection of HCMV isolates from congenitally infected infants were approved by the University of Minnesota Institutional Review Board . Informed consent was obtained from subjects' parents for study of HCMV . Patients identified at the University of Minnesota Medical Center or University of Massachusetts Memorial Health Center were evaluated for HCMV infection on the basis of signs and symptoms suggesting congenital infection . Patients MS1 and MS2 were monochorionic , monozygotic twins with clinical evidence of congenital CMV consisting of thrombocytopenia , transaminitis , and , for MS1 , a small gestational age phenotype . Congenital infection was confirmed for all patients by urine HCMV positive cultures before 3 weeks of age and/or PCR detection of HCMV DNA . No patients were treated with antiviral drugs . Serial specimens were collected at times described in Table 1 . Specimens were stored at −80°C until DNA purification . Total DNA was purified using a Qiagen Blood and Tissue Kit using the standard protocol . A set of primer pairs were constructed that spanned the entire HCMV genome . Details of the primers sequences and primer design strategy have been described [6] . The conditions for PCR were as follows: 1× PfuUltra II PCR buffer , 0 . 25 mM each dNTP ( NEB ) , 0 . 25 uM each primer ( IDT DNA ) , 0 . 5 uL PfuUltra II Polymerase ( Agilent ) and 1M betaine . A touchdown PCR was run on an Eppendorf Mastercycler ep gradient S with the following program for all reactions: 98°C for 2 min , 5 cycles of 98°C for 30 s , 63°C ( decreasing by 1°/cycle ) for 30 s , 72°C for 2 min , followed by 25 cycles of 98°C for 30 s , 58°C for 30 s and 72°C for 2 min , with a 10 min final extension at 72°C . Approximately 10 , 000 genomes copies , based on qPCR results , were used for each amplification reaction . All amplified products were size-selected on agarose gels and gel purified . After amplification of the HCMV genome , all amplicons were quantified on a Nanodrop 2000 , pooled in equimolar proportions and used as substrate in Illumina sequencing . Quantitative PCR was performed using primers and probes described previously [36] . The DNA in pooled amplicons was sheared by sonication on a Sonic Dismembrator 550 ( Fisher ) until the median size was ∼350 basepairs ( bp ) . The DNA library was prepared as described [37] . Briefly , DNA was end-repaired using the End-Repair Enzyme Mix ( NEB ) , and A-tailed using the ATP and Klenow ( exo− ) ( NEB ) . Adapters with appropriate barcodes were ligated onto the modified DNA ends . The library was then size selected on a 2% agarose gel , to produce a library with a median size of 350 bp+/−50 bp . The library was amplified with Illumina primers ( P/N 1003454 ) ( www . illumina . com ) . Once prepared , the libraries were combined in appropriate ratios and submitted for paired-end sequencing on the Illumina GAII . A Toledo strain amplicon set generated from a BAC clone was included as an internal control for measuring error rates . The raw sequence images were processed through Illumina Pipeline 1 . 8 to generate sequence data . The consensus sequence of each sample was called as described [6] . Variants from the consensus sequence were called using the variant filtering protocol as described previously [6] . The variants were filtered to remove errors using parameter thresholds . Variants with parameters below the thresholds were discarded . The thresholds were: minimum basecall quality ( ≥33 ) , mapping quality ( ≥89 ) , and depth ( ≥15 ) . Variant frequency was also used as a threshold . In our prior work , a variant frequency threshold of ≥1 . 9% was used . However , due to reduced error rates associated with Illumina sequencing , a lower variant frequency threshold of ≥1 . 02% could be used to achieve a 5% false positive rate ( Figure S1 ) . Unfolded SNP frequency spectra were generated using the HCMV reference sequence ( Strain Merlin , Ref Seq ID: NC_006273 ) as an outgroup for the intrahost populations under study . Whole genome alignments were generated using the Vista whole genome aligner hosted on the Vista server [38] . Maximum likelihood trees were constructed using PhyML software [39] with an HKY85 substitution model , transition/transversion rates estimated from the data , and 100 bootstrap replicates . A dataset of only synonymous mutations for each viral population was created . These datasets were used as neutral datasets for subsequent analysis of demographic history and were analyzed with the program dadi ( Diffusion Approximation for Demographic Inference ) [20] to build demographic models and yield maximum likelihood estimates of the parameters of the models . SNP frequency data from the pooled sequencing data was projected down into 15 bins , which is the minimum sequencing depth required for SNP calling . For the M103 and B101 populations , samples were collected at two time points from a single host and a single compartment . The models built from these data used the “frozen” function of dadi , in which the model was constructed with the joint frequency spectrum from both time points and the early time point sample was “frozen” ( no input of additional mutations and drift ceased ) after the collection time point . Samples collected from B103 were collected at multiple time points and two compartments , and samples from MS1 and MS2 were collected at 3 time points and two hosts . In these cases , the models depicted in Figure 4 reflect the best-fit model that was constructed from the two divergent populations at the final time point . Specifically , the models were built from the 6 month urine and plasma samples from B103 and the 11 month urine samples from MS1 and MS2 . However , models were also built using the earlier time point data from different compartments or hosts ( for example , 1 week urine and plasma from B103 ) ( data not shown ) . The earlier time point data were used to build a framework of the model , which eased the building of the later , more complex model . Many models were tested , including a standard neutral model and models excluding migration or alternating the number of bottlenecks . The alternate models were found to be worse fits to the experimental data than the models depicted in Figure 4 , as measured by log likelihoods and AIC model comparisons . A fit of the models are depicted in Figure S5 . 95% confidence intervals of the parameters were estimated using nonparametric bootstrapping . In the resulting models , all times are reported in units of 2N generations . However , the time between the collection time points was known ( e . g . , 1 week to 6 months of age ) . Timespans were used to convert all times within the models to calendar time . Migration rates in the model are given in units of M = 2Nem , where Ne is the effective population size and m is the number of migrants per generation . Ne was estimated from the variance in SNP frequency in time sampled populations using the method of Jorde and Ryman [40] . The estimate of Ne for all samples was approximately 1000 ( mean = 961 , range [478–1450] ) ( Table S3 ) and this value was used to convert parameters from the model to units of migrants per generation . To test for evidence of positive selection , we employed the population branch statistic ( PBS ) as described [21] . The PBS statistic uses a log transformation of the FST statistic to identify regions of the genome that have an elevated level of differentiation between closely related populations . Positive selection acting on an allele would be predicted to locally increase FST at the site of the selection , and thus , lead to an increase in the PBS statistic in this region . Importantly , FST-based tests show high power to detect selection from populations with high levels of standing variation [41] , [42] . This was an important characteristic for a test of selection because we have previously shown that HCMV populations collected from congenitally infected neonates are highly variable . We also employed a test of selection that searched for localized regions of reduced variability and an excess of surrounding high frequency alleles [43] . This pattern is expected when selection has recently acted upon a new mutation from a population with low mutation rates . This test identified fewer targets of selection within these data , but there was high concordance between this test and the PBS statistic ( data not shown ) . To determine the significance of the PBS statistic , 10 , 000 simulations were run for a 5 , 000 bp region under the inferred demographic histories of the populations ( Figure 4 ) using the forward simulator sfs_code [44] . Simulation of a full length HCMV genome ( ∼235 , 000 bp ) was computationally intensive and required a prohibitively long time to complete . For this reason , sequence length in the simulation was set to be similar to an average HCMV gene with surrounding regulatory elements , following a previously described approach [21] . The 10 , 000 iterations were analyzed as a complete dataset , and each position contained at least 1 SNP . The SNP at each position ( 5 , 000 in total ) with the maximum PBS value was identified from the simulations ( Figure S6 ) . The results from the simulations were then compared to the experimental data to determine the 5% significance threshold of the PBS values and the corresponding P values of all SNPs with significant ( P<0 . 05 ) PBS scores . Raw sequencing reads from Illumina sequencing will be deposited in the Sequence Read Archive ( http://www . ncbi . nlm . nih . gov/Traces/sra/sra . cgi ) . Data will be available upon publication .
|
The large , dsDNA virus Human cytomegalovirus ( HCMV ) is the most genetically complex viral pathogen of humans . HCMV populations are highly variable , which may allow the virus to evolve in human hosts on short timescales . We tested this hypothesis by longitudinally sampling HCMV populations from the urine and/or plasma of congenitally infected infants . We found that HCMV is generally stable within a compartment , but rapidly evolves when crossing host compartments . In fact , HCMV sampled from two compartments of the same host is as different as HCMV collected from unrelated hosts . We used mathematical modeling and population genetic analysis to show that both a bottleneck ( i . e . , a reduction in population size ) associated with compartment colonization as well as positive selection are necessary to explain the observed differences between compartments . We also conclude from these data that fetal infection in these patients occurred between 13–18 weeks gestational age , consistent with the timing of symptomatic congenital HCMV infections . This study is the most detailed investigation of DNA virus evolution in human hosts to date , provides a framework for the study of other viral infections using similar techniques , and will aid in the development of new antiviral therapies and vaccines .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2013
|
Rapid Intrahost Evolution of Human Cytomegalovirus Is Shaped by Demography and Positive Selection
|
Basal gene expression levels have been shown to be predictive of cellular response to cytotoxic treatments . However , such analyses do not fully reveal complex genotype- phenotype relationships , which are partly encoded in highly interconnected molecular networks . Biological pathways provide a complementary way of understanding drug response variation among individuals . In this study , we integrate chemosensitivity data from a large-scale pharmacogenomics study with basal gene expression data from the CCLE project and prior knowledge of molecular networks to identify specific pathways mediating chemical response . We first develop a computational method called PACER , which ranks pathways for enrichment in a given set of genes using a novel network embedding method . It examines a molecular network that encodes known gene-gene as well as gene-pathway relationships , and determines a vector representation of each gene and pathway in the same low-dimensional vector space . The relevance of a pathway to the given gene set is then captured by the similarity between the pathway vector and gene vectors . To apply this approach to chemosensitivity data , we identify genes whose basal expression levels in a panel of cell lines are correlated with cytotoxic response to a compound , and then rank pathways for relevance to these response-correlated genes using PACER . Extensive evaluation of this approach on benchmarks constructed from databases of compound target genes and large collections of drug response signatures demonstrates its advantages in identifying compound-pathway associations compared to existing statistical methods of pathway enrichment analysis . The associations identified by PACER can serve as testable hypotheses on chemosensitivity pathways and help further study the mechanisms of action of specific cytotoxic drugs . More broadly , PACER represents a novel technique of identifying enriched properties of any gene set of interest while also taking into account networks of known gene-gene relationships and interactions .
Large-scale cancer genomics projects , such as the Cancer Genome Atlas [1] , the Cancer Genome project [2] , and the Cancer Cell Line Encyclopedia project [3] , and cancer pharmacology projects , such as the Genomics of Drug Sensitivity in Cancer project [2] , have generated a large volume of genomics and pharmacological profiling data . As a result , there is an unprecedented opportunity to link pharmacological and genomic data to identify therapeutic biomarkers [4–6] . In pursuit of this vision , significant efforts have been invested in identifying the genetic basis of drug response variation among individual patients . For instance , a recent study performed a comprehensive survey of genes with basal expression levels in cancer cell lines that correlate with drug sensitivity , revealing potential gene candidates for explaining mechanisms of action of various drugs [7] . While significant efforts have focused on specific genes that interact with compounds and confer observed cellular phenotypes , there has been relatively little progress in studying the synergistic effects of genes . These effects are key factors in comprehensively deciphering the mechanisms of action of compounds and understanding complex phenotypes [8] . Similarly , pathways , which comprise a set of interacting genes , have emerged as a useful construct for gaining insights into cellular response to compounds . Analysis at the pathway level not only reduces the analytic complexity from tens of thousands of genes to just hundreds of pathways , but also contains more explanatory power than a simple list of differentially expressed genes [9] . Consequently , an important yet unsolved problem is the effective identification of pathways mediating drug response variation . Although the associated pathways for certain drugs have been studied experimentally [10–12] , in vitro pathway analysis is costly and inherently difficult , making it hard to scale to hundreds of compounds . Fortunately , a growing compendium of genomic , proteomic , and pharmacologic data allows us to develop scalable computational approaches to help solve this problem . Although statistical significance tests and enrichment analyses can be naturally applied to compound-pathway association identification ( e . g . , by testing the overlap between pathway members and differentially expressed genes ) , these approaches fail to leverage well-established biological relationships among genes [13–16] . Even when analyzing individual genes , molecular networks such as protein-protein interaction networks have been shown to play crucial roles in understanding cellular drug response [8 , 17–20] . Therefore , we propose to combine molecular networks with gene expression and drug response data for pathway identification . However , integrating these heterogeneous data sources is statistically challenging . Moreover , networks are high-dimensional , incomplete , and noisy . Thus , our algorithm needs to accurately and comprehensively identify pathways while exploiting suboptimal networks . Here , we present PACER , a novel , network-assisted algorithm that identifies pathway associations for any gene set of interest . PACER first constructs a heterogeneous network that includes pathways and genes , pathway membership information , and gene-gene relationships such as protein-protein physical interaction . It then applies a novel dimensionality reduction algorithm to this heterogeneous network to obtain compact , low-dimensional vectors for pathways and genes in the network . Pathways that are topologically close to the given set of genes ( e . g . , drug response-related genes ) in the network are co-localized with those genes in this low-dimensional vector space . Hence , PACER ranks each pathway based on its vector’s proximity to vectors representing the given genes . We used the proposed algorithm to discover chemosensitivity-related pathways , by applying it to genes whose basal expression level correlates with drug sensitivity . We evaluated PACER’s ability to identify compound-pathway associations with two “ground truth” sets built from compound target data [7] and LINCS differential expression data [21] . When comparing PACER to state-of-the-art methods that ignore prior knowledge of interactions among genes , we observed substantial improvement of the concordance with the chosen benchmarks . Even though we developed PACER and tested its ability to identify compound-pathway associations , the algorithm is applicable to any scenario in which one seeks to discover pathways related to a pre-specified gene set of interest , while utilizing a given gene network .
We obtained a large-scale compound response screening dataset from Rees et al . [7] , which spans 481 chemical compounds and 842 human cancer cell lines encompassing 25 lineages . These 481 compounds were collected from different sources including clinical candidates , FDA-approved drugs and previous chemosensitivity profiling experiments . Area under the drug response curve ( AUC ) was used by the authors of that study to measure cellular response to individual compounds . We also obtained gene expression profiles for these cell lines from the Cancer Cell Line Encyclopedia ( CCLE ) project [22] , profiled using the GeneChip Human Genome U133 Plus 2 . 0 Array . Since these expression measurements were done in each cell line without any drug treatment , they are referred to as “basal” expression levels . In contrast , the expression profiling of a cell line was performed after treatment with a drug in certain studies such as LINCS L1000 [21] and CMAP [23] . We obtained the SMILE specification of each drug from PubChem [24] . We obtained a collection of six human molecular networks from the STRING database v9 . 1 [25] . These six networks include experimentally derived protein-protein interactions , manually curated protein-protein interactions , protein-protein interactions transferred from model organism based on orthology , and interactions computed from genomic features such as fusion-fusion events , functional similarity , and co-expression data . There are 16 , 662 genes in the network . We used all of the STRING channels except “text-mining” and used the Bayesian integration method provided by STRING . Since our approach can deal with different edge weights , we did not set a threshold to remove low-confidence edges . We referred to this integrated network as the “STRING-based molecular network” . To test whether genes that are highly correlated with many compounds tend to have higher degrees in the network , we formed two groups of genes . One group contained genes that are correlated with over 100 compounds , and the other group contained the remaining genes . We then used the Wilcoxon signed-rank test to test whether the degrees of genes in these two groups were from the same distribution . We obtained a collection of 223 cancer-related pathways from the National Cancer Institute ( NCI ) pathway database [26] . These manually curated pathways include human signaling and regulatory pathways as well as key cellular processes . PACER integrates pathway information with the STRING-based molecular network described above by constructing a heterogeneous network of genes and pathways . An edge exists between two genes if they are connected in the network . An edge exists between a pathway and a gene if the gene belongs to the pathway . There are no direct pathway-pathway edges in the heterogeneous network . Formally , let A denote the weighted adjacency matrix of the STRING-based molecular network with n genes ( or proteins ) . Let B ∈ {0 , 1}n×m denote the gene pathway association matrix , where Bij = 1 if gene i is in pathway j . The heterogeneous network H ∈ R ( n + m ) × ( n + m ) is then defined as: H i j = { A i j , i ≤ n , j ≤ n B i - n , j T , i > n , j ≤ n B i , j - n , i ≤ n , j > n 0 , i > n , j > n ( 1 ) PACER adopts diffusion component analysis ( DCA ) , a recently developed network representation algorithm to learn a low-dimensional vector for each node in the network [27] . Because of its ability to handle noisy and missing edges in the biological network , DCA has achieved state-of-the-art results in several computational biology tasks [27 , 28] . DCA takes H as input . It outputs the d-dimensional vectors V ∈ R ( n + m ) × d for each node in H . According to the definition of H , the first n columns of H are the embedding vectors for genes . The remaining columns of H are the embedding vectors for pathways . Since compounds are not nodes in the constructed heterogeneous network , only genes and pathways are projected onto the low-dimensional space . After learning the low-dimensional representations of all nodes ( genes and pathways ) , PACER ranks pathways based on the cosine similarities between the low-dimensional representations of the pathway and a set of genes most correlated with response to a compound . Formally , the PACER score sij between pathway i and compound j is defined as: s i j = ∑ k ∈ RCG ( j ) w k · cos ( V k , V i + n ) , ( 2 ) Here , wk is the weight for gene k . PACER can take input gene weights to weight these cosine similarities . In this paper , we weight the cosine similarities by using the Pearson correlation between the gene expression vector and the chemosensitivity vector . We further calculate an empirical p-value for each compound-pathway association . For a given drug with n response-correlated genes , we use a new , randomly generated set of n genes and compute its pathway association scores using PACER . This is repeated k = 10 , 000 times . With m pathways , we then have a total of km PACER scores . The empirical p-value of each original drug-pathway PACER score is its ( fractional ) rank in this set of PACER scores from random gene sets . LINCS is a data repository of over 1 . 3 million genome-wide expression profiles of human cell lines subjected to a variety of perturbation conditions , which include treatments with more than 20 thousand unique compounds at various concentrations . Each perturbation experiment is represented by a list of differentially expressed genes that are ranked based on z-scores of perturbation expression relative to basal expression . For each gene , we first took the difference between its expression in a perturbation condition and its expression in a control condition ( i . e . , treatment with pure DMSO solvent ) . We then considered the differential expression of the gene in multiple perturbation experiments involving that compound ( i . e . , different concentrations , time points , and cell lines ) . We used the maximum differential expression to represent the compound’s effect on that gene’s expression . All genes were then ranked by their differential expression on treatment with the compound , and the top 250 genes were treated as differentially expressed genes ( DEGs ) of the compound , provided their z-score has an absolute value greater than 2 . We implemented the method of Huang et al . [13] ourselves using the exact same input ( i . e . , chemosensitivity and gene expression data ) as PACER . We first computed a gene’s correlation to a drug by calculating the Pearson correlation coefficient between the gene’s expression values and the drug response values across cell lines . Let the set of genes in pathway p be denoted by Gp , and their correlation values to a drug d by C ( Gp , d ) . Conversely , the set of genes not in pathway p is denoted as G p ¯ , and their correlation values to d as C ( G p ¯ , d ) . We then performed the Kruskal-Wallis H test , following Huang et al . , to test if the medians of C ( Gp , d ) and C ( G p ¯ , d ) were significantly different . We used the resulting p-value to rank pathways for each drug .
Following the work of Rees et al . [7] , we first examined correlations between the compound sensitivity and basal gene expression profiles across hundreds of cell lines . We calculated Pearson correlation coefficients between each gene’s expression and the cellular response to each compound ( measured as AUC , see Methods ) , across different cell lines ( Fig 1A ) . In contrast to IC50 and EC50 scores , AUC simultaneously captures the efficacy and potency of a drug . Of the 8 . 7 million pairs of genes and compounds tested , we found 294 , 789 to be significantly correlated ( p-value < 0 . 0001 after Bonferroni correction , corresponding to a Pearson correlation coefficient of 0 . 215 . ) Since the Rees et al . dataset comprises measurements on 842 cell lines , each correlation was computed over 842 pairs of values ( drug response , gene expression pairs ) . This is why even a modest-looking Pearson correlation of 0 . 215 was deemed highly statistically significant . The key observation from this initial analysis , also noted by Rees et al . , is that basal gene expression levels are highly correlated with cytotoxic response for large numbers of compound-gene pairs . Within these significantly correlated pairs , 26 genes were correlated with over 250 compounds ( Fig 1B , S1 Table ) . We note that these key genes tend to be high-degree nodes in the STRING-based molecular network ( Wilcoxon rank-sum test p-value < 9 . 6e-14 , see Methods ) . We also found that some ( 10 of 481 ) compounds were significantly correlated ( Pearson correlation p-value < 0 . 0001 after Bonferroni correction ) with more than 3 , 200 genes ( Fig 1C ) . Five of these ten compounds are chemotherapeutic agents ( S2 Table ) . In contrast , about 100 compounds were not significantly correlated with any genes; these compounds are mostly probes that either lack FDA approval or are not clinically used . The large disparity among the examined compounds in terms of the number of correlated genes reflects the diversity of these 481 small molecules . While many of them are chemotherapeutic , which can affect the expression of a large number of genes , some compounds may be targeting specific mutations , post-translational modifications , or protein expression . A closer examination revealed that the compounds with the highest AUC had the fewest gene correlations ( i . e . , fewest genes whose expression correlates with cytotoxic response ) ( Fig 1 in S1 Text ) . This suggests that the strategy of identifying compound-associated genes by correlating basal gene expression profiles with cytotoxicity is likely to be more effective for more potent compounds , for which average response is stronger . Note that the gene expression profiles used here are basal and not in response to treatment with compound , hence it was not clear a priori that more effective compounds would have larger numbers of gene correlates . In summary , examination of individual genes’ correlations with chemical response confirmed previous reports [2 , 7 , 29] that basal gene expression is significantly correlated with cytotoxicity across cell lines , especially for effective cytotoxic drugs . For each compound , we refer to the top 250 genes whose expression are most significantly correlated with chemosensitivity ( Pearson correlation p-value < 0 . 0001 after Bonferroni correction ) as “response-correlated genes” ( RCGs ) for this compound . The above evidence for correlations between basal gene expression and chemical response raised the possibility that one might discover important biological pathways associated with the response by a systems-level analysis of gene expression data . To explore this , we considered a collection of 223 cancer-related pathways from the National Cancer Institute ( NCI ) pathway database [26] and used Fisher’s exact test to quantify the overlap between the set of genes in a given pathway and RCGs . A significantly large overlap between the two sets indicates an association between the pathway and the compound . We performed a multiple hypothesis correction on all pathway association tests for each compound , using FDR = 0 . 05 . The results of this baseline method for predicting pathway associations are shown in Fig 1D ( distribution of the number of compounds that are significantly associated with each pathway ) and Fig 1E ( distribution of the number of pathways significantly associated with each compound ) . Both distributions revealed a long tail . For instance , while each pathway was associated with an average of 18 compounds ( of the 481 tested ) , there were 10 pathways that were associated with over 150 compounds ( S3 Table ) . Likewise , while each compound was associated with an average of eight pathways , there were 12 compounds associated with over 25 pathways ( S4 Table ) . We show the details of these long tails in Fig 2 in S1 Text . We observed above that key RCGs ( i . e . , those correlated with many compounds ) tend to be enriched in high degree nodes in the STRING-based molecular network . This suggests that an analysis combining this network with pathway enrichment tests might provide additional insights . We therefore developed a novel network-based method , called PACER , for scoring compound-pathway associations . PACER ( Fig 2A ) first constructs a heterogeneous network consisting of genes and pathways as nodes . In this network , gene-pathway edges denote pathway memberships based on a compendium of pathways and gene-gene edges from the STRING-based molecular network introduced above ( also see Methods ) . PACER then creates a low-dimensional vector representation for each gene and pathway node in the heterogeneous network , reflecting the node’s position in this heterogeneous network . This is done by the Diffusion Component Analysis ( DCA ) approach reported in previous work [27 , 28] . Nodes ( i . e . , pathways or genes ) will have similar vector representations if they are near each other in the network . For instance , two pathway nodes will have similar vector representations if the pathways share genes and/or their genes are related in the STRING-based molecular network . In a similar vein , two genes will have similar representations if they belong to the same pathway ( s ) and/or possess the same network neighbors . A gene and a pathway can also be compared in the low-dimensional space , and will be deemed similar if the gene is in the pathway and/or the gene is related in the network to other genes of the pathway . Using the low-dimensional vectors calculated by DCA , PACER next scores a pathway based on the average cosine similarity between the vector representation of the pathway and those of the RCGs . A pathway can thus be found to be associated with a compound if , in the network , the pathway genes are closely related to the compound’s RCGs; this association can be discovered even if the pathway does not actually include the RCGs . We note that scores assigned by PACER are not statistical significance scores and are meant only to rank pathways for association with a given compound . Also , a negative score assigned to a compound-pathway pair does not imply a negative correlation between expression levels of pathway genes and chemosensitivity . Rather , it only implies a lack of evidence for an association between the compound-pathway pair . Since pathway association analysis is likely to be meaningless for compounds with very few RCGs , we limited the following analysis to the 330 compounds for which more than 5 RCGs were identified . The PACER association scores for all combinations of 330 compounds and 223 NCI signaling pathways are shown in Fig 2B . Since PACER scores are not easily assigned statistical significance levels , we chose to examine , for each compound , the n pathways with the highest PACER scores , where n is the number of statistically significant pathway associations ( FDR < 0 . 05 ) found by the baseline method above for the same compound . ( This choice also allows a fair comparison between the two methods in subsequent sections . ) We found literature support for several of these associations . For example , PACER analysis associates ruxolitinib , a JAK/STAT inhibitor , with integrin-linked kinase signaling pathway . In a previous study , it was shown that beta 4 integrin enhances activation of the transcription factor STAT3 , which is a target of ruxolitinib [30] . Fig 2B reveals that the pathways cluster into many distinct groups , each with different compound association profiles . In some cases , we noted functionally related pathways being grouped together . For example , one group consists of pathways describing various integrin cell surface interactions including “integrin family cell surface interactions” , “alpha E beta 7 integrin cell surface interactions” , “alpha 6 beta 4 integrin-ligand interactions” , and “beta 5 beta 6 beta 7 and beta 8 integrin cell surface interactions” ( marked as blue rectangle in Fig 2B ) . These pathways are known to play crucial roles in communications among cells in response to small molecules [31] . Notably , the integrin-mediated pathways promote invasiveness and oncogenic survival , and contribute to cancer cell survival and resistance to chemotherapy [32 , 33] . Another group consists of different interleukin signaling pathways including “IL4-mediated signaling events” , “IL8- and CXCR1-mediated signaling events” , “IL3-mediated signaling events” , and “IL2 signaling events mediated by PI3K” ( marked as green rectangle in Fig 2B ) . Our analysis found that this group of pathways is associated with decitabine . A recent study shows that decitabine’s effect of PD-1 blockade-based immunotherapy is enhanced in colorectal cancer through upregulation of many immune-related genes [34] . Fig 2B also shows compounds clustered into different groups based on their associations with pathways . We noted examples where many compounds with similar structures were grouped together . For example , teniposide and etoposide had a Tanimoto similarity score of 0 . 94 between their SMILE specifications , which was substantially higher than the average Tanimoto similarity score of 0 . 3716 for all pairs of compounds . They were clustered together in the same group ( marked as black rectangle in Fig 2B ) , which had seven compounds . Among the pathways that are associated with this group , we found a set of similar pathways , including “p53 pathway” , “direct p53 effectors” , “signaling mediated by p38-alpha and p38-beta” , and “signaling mediated by p38-gamma and p38-delta” . We found support in the literature in favor of some of these associations . For example , a previous study reported that etoposide activates p38MAPK and can be used as a combined treatment approach when used with p38MAPK inhibitor SB203580 [35] . As another example , temsirolimus and tacrolimus , which are both epipodophyllotoxins and inhibit topoisomerase II , have a Tanimoto similarity score of 0 . 82 , and are grouped closely in Fig 2B ( marked as pink rectangle in Fig 2B ) . We noted a substantial degree of complementarity between the top predictions of PACER and those of the baseline method that uses Fisher’s exact test between RCGs and pathway genes ( see S5 Table ) . For instance , PACER found that bexarotene is associated with the “IGF1 pathway” . A recent study showed that treating rats with high doses of bexarotene substantially decreased serum IGF1 levels [36] . The baseline approach did not find this association to be significant . Similarly , PACER reported that the “ATM pathway” is associated with simvastatin , while the baseline method did not . Simvastatin has been reported to activate ATM when it is used to treat chronic lymphocytic leukemia patients [37] . For a more systematic comparison between the two methods , we evaluated PACER based on a database of known compound targets . We performed the evaluation under the assumption that a pathway containing at least one known target is an associated pathway . Huang et al . suggested and used this approach [13] . We used it here to evaluate PACER , the baseline method , as well as a third method presented by Huang et al . [13] Although this third method was proposed to detect associations between pathways and drug clades , it can directly detect pathway-compound associations . We implemented the method ourselves ( see Methods ) and included it in our evaluations . We obtained the known targets for our compound set from Rees et al . [7] and STITCH database [38] . We then computed the AUROC of pathway predictions made by PACER for each compound , and plotted this information alongside analogous information for the baseline method and the method of Huang et al . [13] As shown in Fig 2C , PACER identified pathways with higher AUROC compared to the other two methods . For example , PACER identified pathways with an AUROC greater than 0 . 75 for 23 different compounds , while the baseline method achieved this level of AUROC for only 5 compounds . Table 1 shows the 10 compounds for which PACER achieved highest AUROC ( Fig 4-7 in S1 Text ) . We further compared the associations predicted by the three methods to those identified from an external data set . We mined the Library of Integrated Network-Based Cellular Signatures ( LINCS ) L1000 data [21] , which reports genes differentially expressed upon treatment of various cell lines with a compound . For each compound in our analysis that is also included as a perturbagen in the L1000 compendium , we established a LINCS-based benchmark of significantly associated pathways . This was based on a Fisher’s exact test ( p-value < 0 . 05 ) between pathway genes and the most differentially expressed genes from treatments with the same compound ( see Methods ) . We required this criterion to be met in at least one of the cell lines for which data was available from LINCS . We then assessed the concordance between this set of LINCS-based compound-pathway associations and those predicted by either method presented above . We recognize that this is not an ideal benchmark: LINCS data points to genes ( and , indirectly , to pathways ) that are differentially expressed in response to treatment , while PACER and the compared methods base their pathway predictions on genes that have basal expression levels across cell lines that correlate with chemical response . At the same time , we expect the pathways affected by chemical treatment to also be , to an extent , involved in interpersonal variation of chemosensitivity , making this a suitable evaluation procedure . This was inspired by similar observations in cancer biology: genes and pathways disrupted in cancer tissues overlap with genes and pathways whose mutation status in germline non-tumor samples is informative about disease susceptibility and progression . To test whether the significant pathways identified from LINCS data agree with the pathways predicted by one of the methods being evaluated ( based on chemical response variation in CCLE cell lines ) , we counted the compounds for which the two sets of predicted pathways overlapped significantly ( Fisher’s exact test p-value < 0 . 05 ) . As shown in Fig 2D , the PACER approach predicts pathways concordant with the corresponding LINCS-based benchmark for more compounds , compared to the baseline method and that of Huang et al . [13] For instance , when the baseline method used an FDR threshold of 10% to designate significant pathway associations for each compound , and the PACER method predicted the same number of pathways , the latter’s predictions were concordant with the LINCS-based benchmark for 118 , a nearly two-fold improvement over the baseline method’s predictions . Our evaluations actually provide evidence for the above-mentioned possibility that pathways predictive of drug sensitivity overlap with genes that mediate drug response . In fact , we found 113 compounds for which the pathways identified from basal expression correlations and the pathways identified from LINCS signatures overlap with FDR < 5% . After observing the substantial improvement of PACER , we then investigated whether the performance of PACER is stable when using only experimental derived protein-protein interactions as input . We found that this is indeed the case , as per the two evaluation strategies presented above ( Fig 8-9 in S1 Text ) . We further demonstrated that the result of our method is robust to different numbers of top response-correlated genes used in PACER , as shown in Fig 10-11 in S1 Text . We compared different values for ‘k’ in the ‘top k’ genes chosen by PACER . We found that results were comparable when using k = 100 , 150 , 200 , 250 and 300 . This demonstrates the stability of the algorithm’s performance to different but reasonable values of k in its choice of top k response-correlated genes .
We have shown that embedding prior knowledge in a gene network can more accurately identify compound-pathway associations . Our new method , called PACER , identified many compound-pathway associations that are supported by known compound targets as well as literature evidence . Due to its unique ability to incorporate any suitable compendium of gene interactions , our approach may provide complementary insights into drug mechanisms of action . Historically , pathways associated with a particular gene set are identified by using popular statistical methods such as Gene Set Enrichment Analysis [39] , Fisher’s exact test , or the Binomial test ( Reactome [40] ) . These tools test the overlap between differentially expressed genes and pathway members . They may also be applied to the set of drug-response-correlated genes ( RCGs ) analyzed here . Ingenuity Pathway Analysis [41] is another related tool , which utilizes information about causal interactions between pathway members . Our study is similar to the above tools in that PACER also seeks to find pathways implicated by a gene set . However , our approach differs from these existing tools in that known molecular interactions ( e . g . , PPI ) among different genes are taken into consideration . Thus , a gene set , be it the RCGs of a compound or the members of a pathway , is not treated merely as the sum of its parts , but also includes the relationships among those parts . Since the dominant theme in existing approaches is assessment of overlaps between two gene sets ( MSigDB , DAVID , and Reactome adopt variations on this theme ) , our extensive comparisons between PACER and the baseline method of Fisher’s exact test shed light on the relative merits of the new approach . A related line of work aims to identify differentially expressed subnetworks in a given interaction network , e . g . , KeyPathwayMiner [42] , but these studies are only superficially relevant to our work since we aim to prioritize existing pathways instead of finding new pathways . We consider two potential reasons for the strong performance of PACER . First , it is widely appreciated that a chemical compound not only affects individual genes , but also combinations of genes in molecular networks corresponding to core processes , such as cell proliferation and apoptosis . Our method postulates that even if the RCGs and a pathway may only have a few genes in common , they may be close to each other in the network . Although current compound pathway maps are incomplete , much relevant information is available in public databases of human molecular networks . While traditional pathway enrichment analysis methods like Fisher’s exact test identify pathways according to the number of shared genes , PACER prioritizes pathways based on their proximities to RCGs in molecular networks . Second , manually curated pathways may have arbitrary boundaries due to the need to capture knowledge at different levels of detail . Consequently , identifying drug-related pathways might be hindered by pathway boundaries . By leveraging the prior knowledge in molecular networks , PACER is more robust to the noise in pathway boundaries , thus improving the sensitivity of detecting compound-pathway associations . We see many opportunities to improve upon the basic concept of PACER in future work . First , although the current PACER framework was developed in an unsupervised fashion , the scores assigned to each pathway for the given gene set can be used as the feature and plugged into off-the-shelf machine learning classifiers for compound-pathway association identification . Second , although this study focused on chemosensitivity response , the PACER method is broadly applicable to testing the association between two sets of genes according to their proximity in the network . Finally , although we use gene expression data as the molecular profile of each cell line , it might be interesting to test our method based on other molecular data such as somatic mutations and copy number alterations .
|
Gene expression levels have been used to study the cellular response to drug treatments . However , analysis of gene expression without considering gene interactions cannot fully reveal complex genotype-phenotype relationships . Biological pathways reveal the interactions among genes , thus providing a complementary way of understanding the drug response variation among individuals . In this paper , we aim to identify pathways that mediate the chemical response of each drug . We used the recently generated CTRP pharmacogenomics data and CCLE basal expression data to identify these pathways . We showed that using the prior knowledge encoded in molecular networks substantially improves pathway identification . In particular , we integrate genes and pathways into a large heterogeneous network in which links are protein-protein interactions and gene-pathway affiliations . We then project this heterogeneous network onto a low-dimensional space , which enables more precise similarity measurements between pathways and drug-response-correlated genes . Extensive experiments on two benchmarks show that our method substantially improved the pathway identification performance by using the molecular networks . More importantly , our method represents a novel technique of identifying enriched properties of any gene set of interest while also taking into account networks of known gene-gene relationships and interactions .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"cancer",
"genomics",
"medicine",
"and",
"health",
"sciences",
"genetic",
"networks",
"protein",
"interactions",
"protein",
"interaction",
"networks",
"basic",
"cancer",
"research",
"oncology",
"integrins",
"pharmaceutics",
"network",
"analysis",
"cellular",
"structures",
"and",
"organelles",
"computer",
"and",
"information",
"sciences",
"protein-protein",
"interactions",
"proteins",
"cell",
"adhesion",
"extracellular",
"matrix",
"oncogenic",
"signaling",
"gene",
"expression",
"proteomics",
"drug",
"therapy",
"biochemistry",
"signal",
"transduction",
"cell",
"biology",
"gene",
"identification",
"and",
"analysis",
"genetics",
"biology",
"and",
"life",
"sciences",
"genomics",
"cell",
"signaling",
"genomic",
"medicine"
] |
2019
|
Identification of pathways associated with chemosensitivity through network embedding
|
Borrelia burgdorferi , the agent of Lyme disease , has cholesterol and cholesterol-glycolipids that are essential for bacterial fitness , are antigenic , and could be important in mediating interactions with cells of the eukaryotic host . We show that the spirochetes can acquire cholesterol from plasma membranes of epithelial cells . In addition , through fluorescent and confocal microscopy combined with biochemical approaches , we demonstrated that B . burgdorferi labeled with the fluorescent cholesterol analog BODIPY-cholesterol or 3H-labeled cholesterol transfer both cholesterol and cholesterol-glycolipids to HeLa cells . The transfer occurs through two different mechanisms , by direct contact between the bacteria and eukaryotic cell and/or through release of outer membrane vesicles . Thus , two-way lipid exchange between spirochetes and host cells can occur . This lipid exchange could be an important process that contributes to the pathogenesis of Lyme disease .
Borrelia burgdorferi , the causative agent of Lyme disease [1] , [2] , is unusual among prokaryotes in that in addition to phosphatidylcholine , phosphatidylglycerol [3]–[7] and many different lipoproteins [4] , [5] , [8]–[10] , it has free cholesterol and cholesterol-glycolipids in its outer membrane ( OM ) . The glycolipids of B . burgdorferi are mono-α-galactosyl-diacylglycerol ( MGalD ) , which does not contain cholesterol; cholesteryl-β-D-galacto-pyranoside ( CGal ) ; and cholesteryl 6-O-acyl-β-D-galactopyranoside , or cholesteryl 6-O-palmitoyl-β-D-galactopyranoside ( ACGal/Bb-GL-1 ) , which contain cholesterol [3] , [11]–[14] . The cholesterol-glycolipids constitute a significant portion , 45% [11] , of the total lipid content [3] , [5] , [12] , [13] , [15]–[18] . B . burgdorferi does not have the biosynthetic ability to synthesize cholesterol or any long-chain-saturated and unsaturated fatty acids that are required for growth [6] . As a result , the lipid composition of B . burgdorferi reflects that of the culture medium or host animal fluids or tissues [6] . Furthermore , it has been hypothesized that in addition to the activity of galactosyltransferase bb0454 , other uncharacterized spirochetal transferases could be responsible for constructing the cholesterol-glycolipids [18] . Important to the pathogenesis of B . burgdorferi , ACGal , and to a lesser extent MGalD and CGal , are antigenic [13] , [15]–[17] , [19] . These glycolipids induce antibody responses throughout all stages of Lyme disease , being most prominent in the late stages [9] , [11] , [12] , [20] , [21] . Additionally , we demonstrated that antibodies to the cholesterol-glycolipids cross-react with host gangliosides and antibodies to the gangliosides cross-react with the glycolipids [22] , [23] . Borrelia lipid antigens can also be presented in the context of CD1d on NKT cells [24]–[29] . Using ultrastructural , biochemical , and biophysical analysis , we previously determined that the cholesterol-glycolipids in the OM of B . burgdorferi are constituents of eukaryotic-like lipid raft domains [30] . In eukaryotic cell membranes , lipid rafts are microdomains that are rich in sterols , sphingolipids , and phospholipids with saturated acyl tails that allow for tight packing of these lipids into ordered domains [31] , [32] . These cholesterol-rich domains segregate from the disordered membrane domains that contain mostly unsaturated lipids [31] , [33] . In addition to the enrichment of specific lipids , lipid-anchored proteins such as glycosyl phosphatidylinositol ( GPI ) proteins and proteins covalently linked to saturated acyl chains are targeted to lipid rafts [34] . Lipid rafts are important for the segregation of plasma membrane proteins [31]–[33] , [35]–[38] , and contribute to endocytosis , exocytosis , vesicle formation , and budding [39]–[43] . Furthermore , lipid rafts have been identified as important platforms in cell signaling [33] . The presence of free cholesterol and cholesterol-glycolipids with saturated acyl chains in B . burgdorferi creates an opportunity for lipid-lipid interactions with constituents of the lipid rafts in eukaryotic cells . This is of particular interest since B . burgdorferi adheres to many different cell types [44] , [45] . Lipid-lipid interactions could also facilitate the ability of the spirochete to adhere to many different types of cells [46]–[49] and to cellular and matrix proteins [50]–[52] . Furthermore , exchange of lipids between spirochetes and host cells could be important for cholesterol acquisition by the spirochetes , acting as an important nutritional source . Additionally , acquisition of spirochetal antigens by the cells could result in presentation of these antigens in a manner that would be recognized by the immune response leading to a potential mechanism for cellular damage . The requirement for cholesterol is important for other bacteria . The presence of a cholesterol glucoside in spirochetes was first identified in B . hermsii [53] , an agent of relapsing fever . In addition , cholesterol has been documented in the membranes of Helicobacter , Mycoplasma , Ehrlichia , Anaplasma , and Brachyspira [54]–[60] . It is unknown whether raft-like structures similar to that in B . burgdorferi form in these other bacteria . However , acquisition of cholesterol from the plasma membrane of host cells has been documented with H . pylori , another prokaryote that has cholesterol in its OM [61] , and this organism associates with cholesterol rich areas of the eukaryotic cells [61] , [62] . We show here that there is a two-way exchange of lipids between B . burgdorferi and eukaryotic cells and that this exchange is accomplished through direct contact with the spirochete as well as contact with outer membrane vesicles ( OMV ) .
We first investigated whether B . burgdorferi acquires cholesterol through direct contact with HeLa cells using BODIPY-cholesterol . BODIPY-cholesterol is an environment sensitive , lipophilic probe that only fluoresces in hydrophobic , but not aqueous environments [63]–[65] . When B . burgdorferi were incubated with HeLa cells labeled with BODIPY-cholesterol at a multiplicity of infection ( MOI ) of 40∶1 , we observed colocalization ( yellow ) between the BODIPY-cholesterol and B . burgdorferi outer membrane protein OspB on the spirochete and at the point of attachment with the HeLa cell ( Figure 1 , S1 for additional images ) . Colocalization of BODIPY-cholesterol and the spirochetes was demonstrated in single 0 . 5 µm Z-slices and showed the uptake of cholesterol by adherent B . burgdorferi ( Figure 1 , S1 ) . Furthermore , BODIPY-cholesterol labeling extended outward from the point of attachment along the length of the spirochete ( Figure 1 , S1 ) . Acquisition of BODIPY-cholesterol is not detected at the start of the experiment ( Figure 1 , 0 min panels ) . HeLa cells do not release BODIPY-cholesterol into the supernatant over the course of the 1 hr coincubation ( data not shown ) ; therefore , B . burgdorferi most likely acquired BODIPY-cholesterol directly from the labeled HeLa cells and not the supernatant . An important question is whether BODIPY-cholesterol behaves like unlabeled cholesterol in B . burgdorferi . To investigate this , we first determined the incorporation of BODIPY-cholesterol onto the membrane of B . burgdorferi and its toxicity for the spirochete . To grow B . burgdorferi under laboratory conditions , it is necessary to supplement the BSK-II medium with 6% rabbit serum which provides a source of cholesterol ( approximately 0 . 78 mg/L final concentration ) bound to apolipoproteins to form lipoprotein complexes [66] . The free cholesterol ( 0 . 2 mg/L ) , and the most accessible to B . burgdorferi in BSK-II medium , comes from the CMRL-1066 supplement . Removal of cholesterol from the CMRL-1066 eliminated the free cholesterol found in BSK-II . Two of the three concentrations of BODIPY-cholesterol used in the experiments ( 2 . 0 mg/L and 4 . 0 mg/L ) are greater than the 0 . 78 mg/L of cholesterol derived from rabbit serum in BSK-II and are not bound to apolipoproteins . This suggests that the primary and most readily available source of cholesterol to the spirochetes is the fluorescent cholesterol analog . While B . burgdorferi can grow in serum-free formulations [67]–[69] , the use of serum remains the most used supplement . Acquisition of the fluorescent probe by the spirochetes was demonstrated by spectrophotometry , by flow cytometry , and by fluorescent microscopy . The spirochetes incorporated BODIPY-cholesterol in a dose and time dependent manner ( Figure 2A ) . At all concentrations measured , the BODIPY-cholesterol was quickly introduced to the membranes of the spirochetes . At 6 hrs of incubation with the fluorescent cholesterol analog , the variability in the levels of incorporation increased . It is possible that at this time , there may be some self-quenching of the BODIPY fluorophore . For this reason , we determined that the most reproducible fluorescence incorporation level was derived from growing the spirochetes for 4 hrs in the presence of 4 . 0 mg/L BODIPY-cholesterol . It is important to point out that all our incorporation experiments were done at , maximally , 4 hr time periods . Using these optimized conditions , we measured the mean geometric fluorescence of B . burgdorferi labeled with BODIPY-cholesterol compared to unlabeled B . burgdorferi by flow cytometry . B . burgdorferi grown in the presence of BODIPY-cholesterol were significantly more fluorescent when compared to unlabeled bacteria ( Figure 2B ) . Spirochetes grown in DMSO ( which is the diluent for BODIPY-cholesterol ) do not autofluoresce . At the concentrations and time periods used , BODIPY-cholesterol was not cytotoxic as there was no decrease in the numbers of B . burgdorferi during the 6 hr incubation period . Labeled spirochetes did not exhibit any morphological or motility defects in that they maintained their spiral shape and wave-like motion which was demonstrated by several blurry bacteria in the static image ( Figure 2C ) . Furthermore , viability of labeled spirochetes was measured using fluorescent microscopy ( Figure 2C ) . Following incubation of B . burgdorferi with 4 . 0 mg/L BODIPY-cholesterol , spirochetes were reintroduced into BSK-II . The BODIPY-cholesterol did not cause a significant delay in the growth of the spirochetes when compared to controls ( Figure 2D ) . Given that BODIPY-cholesterol can be incorporated by B . burgdorferi , we tested whether the fluorescent cholesterol analog could serve as a substrate for synthesis of cholesterol-glycolipids . The lipids from B . burgdorferi incubated with 4 . 0 mg/L of BODIPY-cholesterol for 4 hrs in BSK-II media lacking free cholesterol were extracted using the Bligh and Dyer solvent extraction method [70] and resolved by chloroform-methanol ( 85/15 ) solvent phase on a High Performance Thin Layer Chromatography ( HPTLC ) plate . Staining of the HPTLC plate with iodine demonstrated that these spirochetes maintain their typical profile ( Figure 2E ) [11] , [12] , [30] . UV excitation of the same chloroform-methanol HPTLC plate showed that the BODIPY-cholesterol is incorporated into the cholesterol glycolipids of B . burgdorferi ( Figure 2E ) . Free cholesterol and cholesterol-glycolipids ( ACGal and CGal ) on the plate are visible , but glycolipids that do not contain cholesterol , MGalD and the phospholipids , are not . Furthermore , incubation of BODIPY-cholesterol with B . burgdorferi at the beginning of the experiment ( no incubation ) indicates that the fluorescent cholesterol probe has the same migration profile as cholesterol in our experimental conditions and is not incorporated into the cholesterol-glycolipids as was observed after 4 hrs of incubation with the probe ( Figure 2E ) . We did observe increased levels of CGal labeled with BODIPY-cholesterol when compared to ACGal labeled with the fluorescent probe . It is possible that CGal could serve as a building block or an intermediate species in the synthesis of ACGal and that over the length of these experiments complete synthesis of ACGal is not achieved . Nonetheless , this shows that B . burgdorferi used the BODIPY-cholesterol as a substrate for cholesterol-glycolipid synthesis , thus making the individual cholesterol-glycolipids fluorescent when excited by UV light . Because B . burgdorferi can incorporate the BODIPY-cholesterol into its cholesterol-glycolipids , we assessed the long term growth and cytotoxicity of the fluorescent cholesterol in BSK-II ( Figure 2F ) . Furthermore , in BSK-II without free cholesterol and without the rabbit serum , the spirochetes with BODIPY-cholesterol do not replicate , but are still alive when viewed using dark field microscopy for up to 96 hrs . This same result is observed when using non-fluorescent cholesterol ( Figure 2F ) . Therefore , BODIPY-cholesterol is similar to non-fluorescent cholesterol in that long-term incubation with the sterols is not cytotoxic to the spirochetes . Furthermore , this experiment indicates that the sterol source does not matter for survival because spirochetes only grow in BSK-II that has to be supplemented with 6% rabbit serum ( Figure 2F ) . We conclude that BODIPY-cholesterol is a useful cholesterol analog in B . burgdorferi , and so its behavior is very likely to reflect that of unlabeled cholesterol . The clustering of cholesterol into lipid rafts is a characteristic that is shared by both the eukaryotic host and B . burgdorferi . We hypothesized that the presence of cholesterol-rich lipid rafts in both the host and pathogen could serve as an ideal platform for lipid-lipid interactions . There is evidence that H . pylori attaches to and partitions in detergent-resistant regions ( lipid rafts ) of the host cell membrane [62] . To determine whether a direct molecular interaction exists between B . burgdorferi and host cells , we performed experiments to observe and quantify if there is any exchange of lipids . HeLa cells were selected for these experiments because B . burgdorferi come into contact with epithelial cells during dissemination and infection , and because they are not phagocytic . HeLa cells were grown on glass coverslips and exposed to B . burgdorferi labeled with BODIPY-cholesterol at a MOI of 20∶1 at four time intervals at 37°C . In addition to the labeled spirochetes , we incubated the HeLa cells with conditioned medium ( after removal of spirochetes labeled with BODIPY-cholesterol ) and cell free wash supernatant ( as negative control ) for the time intervals . Images of HeLa cells incubated with labeled B . burgdorferi and the conditioned medium from BODIPY-cholesterol labeled B . burgdorferi captured by confocal microscopy show that the lowest levels of transfer occur in the shorter incubation times of 15 min ( Figure 3A , E ) and 30 min ( Figure 3B , F ) . There appears to be a time dependent threshold between the 30 min and 1 hr incubation times because the labeling is most robust and highest at the longer incubation times of 1 hr ( Figure 3C , G ) and 2 hrs ( Figure 3D , H ) . In the tested incubation times , we observed little or no transfer in the negative control , the cell free wash supernatant ( Figure 3I , J ) . Time dependency and differences in transfer between labeled spirochetes and conditioned medium were confirmed by RFI analysis ( Figure 3K ) and flow cytometry ( Figure 3L ) . Furthermore at the 2 hr incubation time the images show that B . burgdorferi also come into direct contact with the HeLa cells ( Figure 3M–P ) . This further indicates that the spirochetes make direct contact with the HeLa cells and that this interaction contributes to transfer . These experiments together show that the transfer of B . burgdorferi derived lipids occurs through a combination of contact dependent and contact independent mechanisms . Likewise , transfer is time-dependent because the longer the spirochetes interact with the HeLa cells , the more transfer occurs . To follow the trafficking of BODIPY-cholesterol after incorporation , we used the cis-Golgi marker , GM130 ( Figure 3Q–T ) . Labeled cholesterol trafficked to the Golgi-apparatus within 2 hrs indicating that the cells processed the B . burgdorferi derived cholesterol in the same manner as cholesterol from other sources . To characterize further and understand the process of lipid exchange , we conducted transfer assays using a coincubation time of 2 hrs at different temperatures between 4°C and 37°C . Confocal microscopy demonstrated that at 4°C ( Figure S2 A , D ) there was significantly less transfer of bacterial derived lipids from labeled spirochetes and conditioned medium when compared to images taken at 25°C ( Figure S2 B , E ) and 37°C ( Figure S2 C , F ) . The negative control , the cell free wash supernatant , did not show any visible transfer at any temperature ( Figure S2 G , H ) . For RFI measurements ( Figure S2 I ) and the flow cytometry analysis ( Figure S2 J ) , we observed the same trend when comparing data for two different temperatures where the levels of transferred lipids were significantly higher at the higher temperatures . To confirm that the transfer of B . burgdorferi lipids to HeLa cells was linked to a specific process and not an artifact created by the addition of the BODIPY fluorophore to the cholesterol molecule , we performed lipid transfer experiments using 3H-cholesterol . The increased sensitivity of radiolabeling relative to fluorescence also allowed us to investigate whether cholesterol-glycolipids were transferred to HeLa cells . Lipids from B . burgdorferi labeled with 3H-cholesterol were extracted using the Bligh and Dyer solvent method [70] , and analyzed by HPTLC . The silica from cholesterol lipid spots identified with iodine was scraped from the HPTLC plate and analyzed by liquid scintillation to quantify the amount of 3H-cholesterol incorporated into B . burgdorferi . As an additional control , silica between the observed bands of free cholesterol and ACGal was also scraped to confirm that each band was distinct . When incubated with 3H-cholesterol , B . burgdorferi incorporated the radioactive cholesterol into their membrane fraction ( Figure 4A ) . Similar to the incorporation of BODIPY-cholesterol , we found that the spirochetes also utilized the 3H-cholesterol as a substrate for synthesis of the cholesterol-glycolipids ( Figure 4A ) . The ratio of 3H-cholesterol , 3H-cholesterol labeled ACGal , and 3H-cholesterol labeled CGal in the spirochete was between 53 . 5∶1∶1 and 42 . 6∶1 . 25∶1 ( Figure 4A ) . We did not observe incorporated radioactivity above background in unlabeled controls or in lipids that do not contain cholesterol , MGalD , phosphatidylcholine , or phosphatidylglycerol ( Figure 4A ) . These data support the fluorescence data already presented in Figure 2E in that the cholesterol is incorporated into the bacterial cholesterol glycolipid fractions because we detected significant incorporation of 3H-cholesterol into all cholesterol-glycolipids by HPTLC ( Figure 4B ) . To demonstrate transfer of lipids from spirochetes to HeLa cells , B . burgdorferi labeled with 3H-cholesterol were incubated with the cells at a MOI of 40∶1 , 20∶1 , 10∶1 , and 1∶1 for 2 hrs . The HeLa cells were washed three times and lifted from the tissue culture wells . Each condition was measured by liquid scintillation counting to quantify the levels of transfer ( Figure 4C ) . We observed significant disintegrations per minute ( DPM ) in the higher MOI in a dose-dependent manner ( Figure 4C ) . This agrees with data obtained using BODIPY-cholesterol that there is a transfer of B . burgdorferi lipids to the host eukaryotic cells . Additionally , lipids were extracted from HeLa cells following removal of spirochetes and analyzed for their lipid profile by HPTLC ( Figure 4D ) . Evidence that the cholesterol-glycolipids are transferred to the membrane of the HeLa cells comes from identification of the individual cholesterol-glycolipids in the HeLa cell lipid extracts on the HPTLC plate ( Figure 4D ) . Experimental HeLa cell lipid extracts were run on an HPTLC plate in parallel with a reference control B . burgdorferi lipid extract . The HPTLC plate was stained with iodine and the bands on the HeLa cell extract samples that corresponded to the known glycolipid bands from the B . burgdorferi control lipid extract were scraped for liquid scintillation analysis . The ratio of transferred 3H-cholesterol , 3H-cholesterol labeled ACGal , and 3H-cholesterol labeled CGal was between 56∶4∶1 and 54 . 67∶5 . 33∶1 ( Figure 4D ) . It is noteworthy that the ratio of ACGal to free cholesterol in the HeLa cells was higher than that in the spirochetes . This suggests that it transfers more efficiently than cholesterol , and would not be expected if the HeLa cell-associated radioactivity was an artifact of spirochetes attached to the HeLa cells . As an additional control to account for the contribution of spirochetes that remain attached to the cells after washing to the total DPMs , we utilized a GFP-labeled B . burgdorferi strain [71] to quantify the percentage of bacteria that did not dissociate from the cells . The HeLa cells were analyzed by the SpectraMaxM2 to calculate GFP fluorescence levels . This control disclosed that 2 . 2±0 . 4% ( ∼61% of the total signal of 3 . 5±0 . 9% ) could be derived from spirochetes that remained attached . Despite this , however , ∼39% of the signal of 3H cholesterol and cholesterol-glycolipid was associated with cells . Furthermore , the observation above that the fraction of radioactive ACGal in HeLa cells after transfer ( between 56∶4∶1 and 54 . 67∶5 . 33∶1 [cholesterol∶ACGal∶CGal] ) was 3 to 5-fold times higher than ACGal derived from radioactively labeled B . burgdorferi alone ( between 53 . 5∶1∶1 and 42 . 6∶1 . 25∶1 [cholesterol∶ACGal∶CGal] ) the amount of ACGal bound to the HeLa cells is 6-fold times higher than that expected if it was just derived from B . burgdorferi that had not washed off the HeLa cells . In this study , we have used four different methodologies ( fluorescence confocal microscopy , single cell RFI analyses , flow cytometry , and isotope incorporation ) to demonstrate lipid exchange from B . burgdorferi to cells . Of these four , isotope incorporation was the least robust , but it also provided evidence for lipid exchange through differential incorporation of label and through HPTLC . Thus , using all four experimental approaches , we conclude that there is a transfer or uptake of both cholesterol and the antigenic cholesterol-glycolipids by eukaryotic host cells in vitro , and this could be important implications for the pathogenesis of Lyme disease . Given that HeLa cells acquired BODIPY fluorescence from conditioned medium ( Figure 3 E–H ) , we sought to determine whether free BODIPY-cholesterol was being released nonspecifically or was associated with OMV . To identify released OMV in the supernatants , we used a lipophilic probe , 1 , 6-diphenyl-1 , 3 , 5-hexatriene ( DPH ) . DPH fluoresces in hydrophobic environments such as membranes , but not in aqueous environments and has a linear response with membrane bilayer concentration [72] . By probing the supernatants from BODIPY-cholesterol labeled and unlabeled spirochetes with 1 µg/mL of DPH , we were able to determine if fragments of membrane were being released from the labeled spirochetes and how it compared to release by unlabeled B . burgdorferi . Labeled B . burgdorferi released intact membrane , in amounts similar to unlabeled B . burgdorferi ( Figure 5A ) , suggesting that the membrane release is a natural process and not the result of BODIPY-cholesterol labeling . However , DPH labeling cannot distinguish if membrane release is in the form of an intact OMV . Therefore we analyzed the collected OMV by transmission electron microscopy ( TEM ) . The negative-stain TEM micrographs of isolated vesicles show spherical structures from both labeled and unlabeled B . burgdorferi OMV ( Figure 5B ) . In both OMV preparations , we used double immunogold labeling for OspB ( 18 nm colloidal gold ) and the cholesterol-glycolipids ( 6 nm colloidal gold ) . Both labeled and unlabeled B . burgdorferi release OMV that are similar in morphology and glycolipid content . The release of OMV from the spirochetes ( Figure 5B ) led us to test whether the OMV play a role in the transfer of lipids to HeLa cells . To confirm that the membranes released from labeled B . burgdorferi are OMV and similar in protein content to the unlabeled controls , we purified OMV from labeled and unlabeled B . burgdorferi using an Optiprep density gradient . To compare protein content , we performed SDS-PAGE using 11 purified OMV fractions from the supernatants of labeled and unlabeled B . burgdorferi and observed similar protein content in the fractions where OMV partition ( 20–25% ) . Both preparations had similar distributions of protein across the 11 gradient fractions . Even with the addition of BODIPY-cholesterol to the spirochetes , the OMV are representative of naturally forming vesicles . We further determined by SDS-PAGE and western blot that OspA , OspB , and lp6 . 6 , some of the most abundant lipoproteins in the OMV , are found in both labeled and unlabeled B . burgdorferi ( Figure 6A and 6B ) . These electrophoretic and immunoblot data were also supported by previous mass spectrometry data from our laboratory that showed OM proteins were the most abundant in the OMV [73] . In addition to analyzing the protein content of labeled and unlabeled OMV , we also determined their lipid content . To examine the lipids found in B . burgdorferi OMV , we pooled fractions 15–25% of both the labeled and unlabeled vesicles . We determined BODIPY-cholesterol labeled and unlabeled OMV have both phospholipids and cholesterol , ACGal , and CGal . The cholesterol-glycolipids are significant lipid components of the vesicles ( Figure 6C ) . To identify fluorescent lipids , the same HPTLC plate was exposed to UV light which demonstrated that the vesicles derived from BODIPY-cholesterol labeled B . burgdorferi contain the fluorescent cholesterol analog BODIPY-cholesterol , although in this preparation we were not able to demonstrate BODIPY-cholesterol incorporation into ACGal and CGal ( Figure 6C ) . Detection of fluorescently labeled ACGal and CGal was most likely not observed in the isolated vesicles due to the small amount of material collected and analyzed and the low sensitivity of UV light excitation of the BODIPY-cholesterol probe on the TLC plate . However , the cholesterol-glycolipids were visualized as components of the OMV at the single molecule level using the 6 nm colloidal gold anti-rabbit secondary antibody and TEM ( Figure 5B ) . To test for the possibility that the OMV released by B . burgdorferi could be responsible for transfer of bacterial antigens to the HeLa cells , we isolated vesicles from B . burgdorferi labeled with BODIPY-cholesterol . The HeLa cells were washed with PBS and incubated with Vybrant Cell-Labeling Solution DiI to label their plasma membranes ( Figure 7A ) . We observed that when isolated OMV labeled with BODIPY-cholesterol are incubated with HeLa cells in the transfer assay , there is a colocalization between the BODIPY-cholesterol probe and the Vybrant Cell Labeling Solution DiI probe on the surface of the HeLa cells . We observed this colocalization at the single cell level ( merged cross section , Figure 7B ) and in merged 3D composite image of multiple cells on the cover slip ( Figure 7C ) . Similar to the transfer assays conducted with the whole bacteria , we observed internalization in the form of free BODIPY-cholesterol fluorescence inside of the HeLa cell . This was represented by diffuse staining inside the plasma membrane of the composite images . These confocal micrographs show that there is an exchange of bacterial derived lipids from B . burgdorferi to eukaryotic cells and that this transfer can be executed by OMV .
Through the use of fluorescent and radiolabeled cholesterol , we showed that a lipid exchange between spirochetes and host cells can occur . Two main conclusions can be derived from our studies . First , we show that when B . burgdorferi come in direct contact with epithelial cells the spirochetes can extract cholesterol from epithelial cell membranes . Second , both cholesterol and the antigenic cholesterol-glycolipids of B . burgdorferi are transferred to epithelial cells through direct contact between the spirochete and the plasma membrane and through released OMV . As an extracellular pathogen , nutrient acquisition from the immediate environment is an essential process for B . burgdorferi to be able to persist in the host . Here we provide evidence that B . burgdorferi can extract cholesterol , an essential membrane lipid , from eukaryotic cells . We demonstrated that B . burgdorferi attach to epithelial cells and can incorporate cholesterol directly from the plasma membrane . Using fluorescence microscopy , B . burgdorferi were shown to attach to epithelial cells and associate with BODIPY-cholesterol labeled regions of the plasma membrane . Evidence that the spirochetes extract cholesterol from epithelial cells comes from confocal microscopy images which showed colocalization of the BODIPY-cholesterol from the HeLa cells and the lipoprotein OspB at the point of attachment , and also the presence of BODIPY-cholesterol throughout the spirochete which extended away from the cell . Cholesterol acquisition has been shown to also be an important process for extracellular pathogens . Cholesterol and cholesterol-glycolipids comprise a significant portion of the bacterial membrane of H . pylori . Using similar techniques to ours , H . pylori was shown to attach to cholesterol-rich domains and acquire cholesterol from eukaryotic membranes [61] , [62] . For some obligate intracellular pathogens , such as Coxiella burnetii and the Chlamydia species , lipid acquisition is also essential to establish and maintain an active infection [74] , [75] and an exogenous source of lipids is necessary for their growth [6] , [74] , [75] . In intracellular bacteria , phospholipids and cholesterol along with machinery to synthesize these lipids are trafficked to the inclusion vacuole containing these bacteria [76]–[79] . Understanding how bacteria use or alter these lipids is an area of active research [80] . Together , these findings represent diverse mechanisms for cholesterol acquisition . The host antibody response to B . burgdorferi has been studied extensively , but the understanding of the role of the unique cholesterol-glycolipids in pathogenesis is limited . Antibody responses to these glycolipids have been detected throughout the course of Lyme disease [9] , [11] , [12] , [20] . To observe the role of cholesterol and cholesterol-glycolipids of B . burgdorferi , we developed methods to incorporate labeled cholesterol into the spirochetal membranes and showed that the labeled cholesterol is used as a substrate for synthesis of the cholesterol-glycolipids . This allowed us to track the cholesterol-glycolipids through several experimental approaches . In addition to the ability of B . burgdorferi to tolerate and use labeled cholesterol as a substrate for cholesterol-glycolipid synthesis , we also demonstrated that the cholesterol and cholesterol-glycolipids are transferred to the epithelial cells . After incubation with epithelial cells , evidence from confocal microscopy showed that fluorescently labeled cholesterol and cholesterol-glycolipids were observed in the plasma membrane as well as Golgi apparatus . The presence of the fluorescent cholesterol in the Golgi apparatus indicated that the eukaryotic cells are processing and trafficking all or part of the exogenous BODIPY-cholesterol in a manner similar to that used for cholesterol derived from other sources [64] . To bolster the results from the fluorescence experiments , we also used radioactive cholesterol to demonstrate lipid exchange . With the higher sensitivity of radiolabeled cholesterol , we were able to establish that not just labeled cholesterol , but also labeled ACGal and CGal , the two antigenic cholesterol glycolipids , can be transferred to epithelial cells and are present in the eukaryotic membrane . Because of the increased sensitivity of the experiments using radiolabeled spirochetes , we were able to detect low but significant DPM values in the epithelial lipid extracts , suggesting that there is an exchange of the antigenic cholesterol-glycolipids into the HeLa cells . Because the values obtained from the transfer experiments are generally low , we used several different approaches to demonstrate lipid exchange , including spectrophotometry , flow cytometry , quantifiable fluorescence microscopy ( RFI ) , and liquid scintillation . Further documentation of transfer was obtained at the level of the single cell using fluorescent and confocal microscopy . Regardless of the amount of transfer , together , our results demonstrated that a spirochetal cholesterol based antigen is transferred to and present in the membrane of the epithelial cells . We showed that length of coincubation time and the temperature of the coincubation between the spirochetes and the epithelial cells can change the amount of overall transferred lipid . In addition , we investigated how OMV can contribute to the transfer of lipids between the spirochetes and host cells . Borrelia OMV have been isolated [81]–[83] and extensively studied [84]–[89] . Our method to collect and isolate OMV from Borrelia sought to collect OMV in BSK-II from live organisms , thus replicating the same experimental conditions would be found in as the conditioned medium . We demonstrated that transfer can occur through released OMV from the spirochete , and the vesicles can attach to or be incorporated into the epithelial plasma membrane . Furthermore , we determined that the cholesterol-glycolipids were significant components of the OMV , and similar to other species of Borrelia [81] , the isolated OMV were also rich in OM proteins . We have shown that B . burgdorferi extract cholesterol from the plasma membrane of eukaryotic cells and that cholesterol-glycolipids can be transferred to epithelial cell membranes by a contact dependent mechanism through direct attachment . These two events , cholesterol acquisition and transfer of antigenic lipids , might not be mutually exclusive . One possible explanation for the contact dependent transfer could be that the spirochetes are required to attach to the eukaryotic plasma membrane to acquire cholesterol . Uncharacterized spirochetal transferases [18] potentially associated with the OM could also extract cholesterol from the host cells for synthesis of the cholesterol-glycolipids . During this event , it is possible that the cholesterol-glycolipids are left behind in the plasma membrane of the epithelial cell . There may also be the mechanism in which cells acquire these spirochetal lipids via released OMV that are rich in cholesterol-glycolipids [30] , [73] . The ability of OMV from pathogenic bacteria to participate in host-pathogen interactions as virulence factors has been well documented [90] . Furthermore , there is evidence that OMV from other bacteria can fuse with the cell membrane [91]–[93] . We demonstrated that OMV derived from fluorescently labeled B . burgdorferi can transfer the fluorescent cholesterol to the epithelial cells . Therefore , the OMV of B . burgdorferi could serve as a vehicle to transfer the cholesterol-glycolipids , fuse with the cell membrane , and act as virulence factors that influence and modulate the host immune response . In summary , using fluorescent and radiolabeled cholesterol , we have documented that B . burgdorferi extract cholesterol from the plasma membrane of eukaryotic cells and that prokaryotic cholesterol-glycolipids can be transferred to epithelial cell membranes by two mechanisms ( i ) a contact dependent mechanism through direct attachment and ( ii ) a contact independent method through released OMV . The B . burgdorferi membrane is unique in that it contains lipid rafts , with cholesterol and cholesterol-glycolipids with physical properties that are similar to those of eukaryotic membranes . Transfer of antigenic cholesterol-glycolipids could play a major role in the pathogenesis of the spirochetoses . Given the limited biosynthetic capabilities of B . burgdorferi to make cholesterol and other important lipids , the process of cholesterol extraction from host cells is likely to be more biologically significant for the nutrition of the spirochetes early in infection as nutrient acquisition is crucial for the replication of the bacteria . Once the bacteria have disseminated and an infection has been established , it is likely that even at low levels , the transfer of antigenic lipids from the spirochete to host cells becomes more significant . Whether inserted directly into the plasma membrane of eukaryotic cells , or attached to the surface of these cells , the presence of foreign antigens with similar composition and structural characteristics could have multiple consequences for the host immune response . It is also possible that these transferred lipids could contribute to heightened inflammation and arthritis . Furthermore , if the immune response were to recognize cells with transferred lipid antigens , the cells themselves become targets of immune effectors .
B . burgdorferi strain B31 were grown in microareophilic conditions in BSK-II medium [94] supplemented with 6% rabbit serum ( Sigma ) at 33°C . For the incorporation experiments , a BSK-II without free cholesterol was made by using cholesterol free CMRL-1066 ( Invitrogen ) . Removal of cholesterol from the CMRL-1066 eliminated the free cholesterol found in BSK-II ( 0 . 2 µg/mL . When the BSK-II is supplemented with 6% rabbit serum , the final concentration of cholesterol in the cholesterol-free BSK-II was 0 . 78 µg/mL . The environment sensitive fluorescent cholesterol analog , 23- ( dipyrrometheneboron difluoride ) -24-norcholesterol ( TopFlour Cholesterol or BODIPY-cholesterol , Avanti Polar Lipids ) , was added to BSK-II without free cholesterol medium for 6 hrs at a concentration of 0 . 2 mg/L , 2 . 0 mg/L , and 4 . 0 mg/L of BODIPY-cholesterol . After the 6 hr incubations , the bacteria were washed three times with Hank's balanced salt solution ( HBSS , Gibco ) . Fluorescence readings were calculated using a SpectraMaxM2 ( ex/em - 490/504 ) . The fluorescently labeled B . burgdorferi were used immediately for all assays . B . burgdorferi were also labeled with 3H-cholesterol ( American Radiolabeled Chemicals , Inc . ) . Spirochetes were washed three times with HBSS and the endogenous cholesterol was depleted using 10 mM MβCD . After depletion , the cholesterol was replaced by incubating the spirochetes with 4 . 0 mg/L of cholesterol and 10 . 0 µCi of 3H-cholesterol in HBSS . B . burgdorferi were washed three times with HBSS and were immediately used for all assays . B . burgdorferi were washed three times in large volumes of HBSS , before lipid extraction using the Bligh and Dyer method [70] . The lipid extracts were concentrated under constant nitrogen gas stream . Lipid extracts were separated by thin-layer chromatography on Si250 HPTLC silica plates ( J . T . Baker ) with chloroform/methanol ( 85/15 ) and visualized with iodine vapor staining or exposure to UV light . Lipid extracts from samples containing 3H-cholesterol were separated by HPTLC and stained with iodine vapor . The spots on the plate were scraped and the silica containing the radioactive cholesterol was analyzed by liquid scintillation counting . For direct visualization of the HPTLC plate , the plate was first sprayed with EN3HANCE Spray ( DuPont ) and exposed using BioMax MR Film ( Kodak ) for 4 and 14 days at −80°C and developed using a Medical Film Processor Model SRX-101A ( Konica ) . B . burgdorferi labeled with BODIPY-cholesterol were analyzed by the SpectraMax M2 plate reader ( ex/em - 490/504 ) for incorporation of fluorescence . The viability , morphology , and motility of the BODIPY-cholesterol treated B . burgdorferi were assessed by dark-field enumeration and fluorescence microscopy . The lipids of B . burgdorferi labeled with BODIPY-cholesterol were isolated and resolved on an HPTLC Si250 silica plate with chloroform-methanol ( 85/15 ) and exposed with UV light and stained with iodine vapor for visualization of the lipids . Standards were known Rf values from identical solvent systems [11] , [12] , [30] . B . burgdorferi incubated with BODIPY-cholesterol were also analyzed for their ability to grow or recover following incorporation of the fluorescent label . Labeled B . burgdorferi were washed in HBSS , and resuspended in BSK-II medium , and growth was assessed by dark-field enumeration . HeLa cells were maintained in DMEM medium ( Gibco ) with 10% fetal calf serum ( Pel-Freez ) . The HeLa cells were grown on glass coverslips in T175 tissue culture flasks ( Falcon ) . Cells were infected with B . burgdorferi labeled with BODIPY-cholesterol at a multiplicity of infection ( MOI ) of 20∶1 for 2 hrs . Conditioned medium and negative control cell free wash supernatant were also added to the coverslips for 2 hrs . Conditioned medium supernatants were generated by incubating B . burgdorferi labeled with BODIPY-cholesterol in BSK-II for 2 hrs . The spirochetes were pelleted by centrifugation at high speed for 15 min and the supernatants ( cell-free ) were examined by dark field and fluorescent microscopy to ensure that intact organisms were not present . The supernatants were added directly to the HeLa cells for 2 hrs . The cell free wash supernatant was included as a negative control to ensure that the BODIPY-cholesterol is not loosely associated with the OM or released nonspecifically from B . burgdorferi as there was no observable transfer of label . The negative control or cell free wash supernatant was generated by resuspension of the labeled B . burgdorferi pellet after the final wash in BSK-II , to the HeLa cells for 2 hrs . The coverslips were washed three times in phosphate buffered saline ( PBS , Gibco ) , fixed in 2 . 5% paraformaldehyde and blocked with 1% bovine serum albumin ( Sigma ) in PBS for immunofluorescence staining . The remaining B . burgdorferi that were attached to the HeLa cells were detected with CB2 , a murine monoclonal antibody to OspB [95] followed by an Alexa Fluor 594 goat anti-mouse IgG ( Invitrogen ) . The samples analyzed for BODIPY-cholesterol colocalization with the cis-Golgi complex were probed with a monoclonal rabbit GM130 antibody ( Abcam ) followed by a Texas Red goat anti-rabbit IgG ( Abcam ) . Samples were imaged by confocal laser microscopy using a Zeiss LSM 510 META NLO Two-Photon Laser Scanning Confocal Microscope System . Additional approaches were used to detect lipid exchange between B . burgdorferi and cells . RFI of the HeLa cells from 10 microscope fields of vision were calculated using Zeiss LSM 510 META NLO Two-Photon Laser Scanning Confocal Microscope System Software . The mean geometric fluorescence was calculated from samples that were analyzed by a FACScan/Calibur for BODIPY fluorescence . For experiments using radiolabeled B . burgdorferi , HeLa cells were incubated with B . burgdorferi labeled with 10 . 0 µCi 3H-cholesterol 3H-cholesterol at a MOI of 40∶1 for 2 hrs . The HeLa cells were extensively washed to remove the spirochetes . HeLa cells were lifted from the tissue culture flasks using 0 . 05%Trypsin/EDTA ( Gibco ) . Transferred radioactivity to the HeLa cells was detected by extracting the lipids using the Bligh and Dyer solvent extraction method . Lipids were separated on a HPTLC plate using chloroform-methanol ( 85/15 ) . The HPTLC plate was stained with iodine and spots on the plate were scraped . In addition , incorporation of isotope into the lipids was measured by scraping the silica from the HPTLC plate , analyzed by liquid scintillation using a Beckman LS 6500 Liquid Scintillation Counter and reported as DPM . To control for B . burgdorferi that remained attached to HeLa cells after washing , B . burgdorferi that constitutively expressed GFP [71] was utilized . The GFP expressing B . burgdorferi were incubated with HeLa cells at a MOI of 40∶1 for 2 hrs . The HeLa cells were extensively washed to remove the spirochetes . HeLa cells were lifted from the tissue culture flasks using 0 . 05%Trypsin/EDTA ( Gibco ) . Using the SpectraMaxM2 , the amount of GFP fluorescence ( B . burgdorferi still attached to HeLa cells ) was measured in the HeLa cell pellet . For detection of released membrane material , labeled B and unlabeled B . burgdorferi were incubated in HBSS for 2 hrs . The supernatants were collected at 1 hr and 2 hrs . The supernatants were labeled with 1 µg/ml of the hydrophobic-sensitive , fluorescent , lipophilic probe DPH ( Invitrogen ) for 30 min at 33°C . After 20 min , the supernatants were analyzed for fluorescence in a SpectraMax M2 plate reader using an excitation of 360 nm and emission of 430 nm . B . burgdorferi labeled with BODIPY-cholesterol or unlabeled B . burgdorferi in the late-log phase of growth were pelleted by centrifugation , and resuspended in fresh BSK-II media . To keep similar incubation times as the transfer assay , the labeled and unlabeled spirochetes were incubated for 2 hrs at 37°C to collect released vesicles . Following the removal of spirochetes , the supernatants were filtered twice using 0 . 22 µm-pore-size Steriflip filters ( Millipore ) . Crude membrane and outer membrane vesicles ( OMV ) in the filtered supernatant were concentrated by ultracentrifugation for 1 hr at 100 , 000× g . To purify the OMV , the membrane pellet was resuspended in 60% OptiPrep ( Axis Shield ) . A discontinuous gradient was made following the manufacturer's instructions . The discontinuous gradient was centrifuged for 16 hrs at 100 , 000× g . The OMV were concentrated to form a white band that floated to the interface between the 20% layer and the 25% layer of the gradient . The gradient fractions were collected in 1 mL volumes ( two 1 mL fractions for each OptiPrep gradient percentage ) from the top of the tube with the least dense faction being collected first . The OMV from fraction 15% , 20% , and 25% were pooled based on similar protein contents to maximize the amount of OMV collected [73] . The isolated OMV were concentrated and removed from the OptiPrep solution by centrifugation for 1 hr at 100 , 000× g . The pelleted , purified OMV were resuspended in 20 mM HEPES ( pH 7 . 5 ) . The OMV were immediately used for vesicle transfer assays . HeLa cells were grown on glass coverslips in 24 well tissue culture plates . Cells were incubated with 40 µg ( based on protein content ) of vesicles purified from BODIPY-cholesterol labeled B . burgdorferi for 2 hrs . Protein content was determined by BCA Assay and a Coomassie Plus Assay ( Pierce ) . The HeLa cells were washed with HBSS to remove the vesicles . To label the plasma membrane of the HeLa cells , the coverslips were then incubated with Vybrant Cell Labeling Solution DiI ( Invitrogen ) following the manufacturer's instructions . The coverslips were fixed with 2 . 5% paraformaldehyde and were imaged by confocal laser microscopy using a Zeiss LSM 510 META NLO Two-Photon Laser Scanning Confocal Microscope System . HeLa cells were grown on glass coverslips in 24 well tissue culture plates . The HeLa cells were preloaded with 5 µg/mL of BODIPY-cholesterol for 2 hrs . To remove excess BODIPY-cholesterol , the cells were washed twice with 5 mg/mL of methyl-β-cyclodextrin ( MβCD , Sigma ) and one final time with BSK-II . HeLa cells labeled with BODIPY-cholesterol were incubated in BSK-II for 1 hr , before the supernatants and cells were measured for BODIPY-cholesterol fluorescence using the SpectraMax M2 . To observe the cholesterol extraction from HeLa cell membranes , the cells were infected at an MOI of 40∶1 for 1 hr washed 3 times with PBS and fixed with 2 . 5% paraformaldehyde for 15 min . B . burgdorferi were detected by incubation with CB2 hybridoma supernatants [95] followed by an Alexa Fluor 594 goat anti-mouse IgG ( Invitrogen ) . The cells were imaged by confocal laser microscopy using a Zeiss LSM 510 META NLO Two-Photon Laser Scanning Confocal Microscope System . Statistics were calculated using GraphPad InStat 3 ( GraphPad Software ) .
|
Lyme disease , the most prevalent arthropod-borne disease in North America , is caused by the spirochete Borrelia burgdorferi . Cholesterol is a significant component of the B . burgdorferi membrane lipids , and is processed to make cholesterol-glycolipids . Our interest in the presence of cholesterol in B . burgdorferi recently led to the identification and characterization of eukaryotic-like lipid rafts in the spirochete . The presence of free cholesterol and cholesterol-glycolipids in B . burgdorferi creates an opportunity for lipid-lipid interactions with constituents of the lipid rafts in eukaryotic cells . We present evidence that there is a two-way exchange of lipids between B . burgdorferi and epithelial cells . Spirochetes are unable to synthesize cholesterol , but can acquire it from the plasma membrane of epithelial cells . In addition , free cholesterol and cholesterol-glycolipids from B . burgdorferi are transferred to epithelial cells through direct contact and through outer membrane vesicles . The exchange of cholesterol between spirochete and host could be an important aspect of the pathogenesis of Lyme disease .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"gram",
"negative",
"microbial",
"pathogens",
"biology",
"microbiology",
"host-pathogen",
"interaction",
"bacterial",
"pathogens"
] |
2013
|
Lipid Exchange between Borrelia burgdorferi and Host Cells
|
Type 1 fimbriae are a crucial factor for the virulence of uropathogenic Escherichia coli during the first steps of infection by mediating adhesion to epithelial cells . They are also required for the consequent colonization of the tissues and for invasion of the uroepithelium . Here , we studied the role of the specialized signal transduction system CRP-cAMP in the regulation of type 1 fimbriation . Although initially discovered by regulating carbohydrate metabolism , the CRP-cAMP complex controls a major regulatory network in Gram-negative bacteria , including a broad subset of genes spread into different functional categories of the cell . Our results indicate that CRP-cAMP plays a dual role in type 1 fimbriation , affecting both the phase variation process and fimA promoter activity , with an overall repressive outcome on fimbriation . The dissection of the regulatory pathway let us conclude that CRP-cAMP negatively affects FimB-mediated recombination by an indirect mechanism that requires DNA gyrase activity . Moreover , the underlying studies revealed that CRP-cAMP controls the expression of another global regulator in Gram-negative bacteria , the leucine-responsive protein Lrp . CRP-cAMP-mediated repression is limiting the switch from the non-fimbriated to the fimbriated state . Consistently , a drop in the intracellular concentration of cAMP due to altered physiological conditions ( e . g . growth in presence of glucose ) increases the percentage of fimbriated cells in the bacterial population . We also provide evidence that the repression of type 1 fimbriae by CRP-cAMP occurs during fast growth conditions ( logarithmic phase ) and is alleviated during slow growth ( stationary phase ) , which is consistent with an involvement of type 1 fimbriae in the adaptation to stress conditions by promoting biofilm growth or entry into host cells . Our work suggests that the metabolic sensor CRP-cAMP plays a role in coupling the expression of type 1 fimbriae to environmental conditions , thereby also affecting subsequent attachment and colonization of host tissues .
Bacteria have the ability to rapidly adapt to changes in the environment , a feature that is important for survival and multiplication both during colonization of host organisms and in the environment . An efficient adaptation implies the ability to sense external parameters and to transduce the perceived signals to cellular regulators , which then incite adaptive changes in the physiology of the cell . One form of signal transduction occurs via cytoplasmatic secondary messenger systems , so-called alarmones , which can mediate a rapid response . Alarmones are low molecular mass , non-proteinaceous , enzymatically synthesized compounds . Several modified nucleotides have been described to execute this function in bacteria , among them the 3′ , 5′-cyclic adenosine monophosphate ( cAMP ) . cAMP is a ubiquitous molecule found in both prokaryotes and eukaryotes . In bacteria , the activity of cAMP was initially thought to be restricted to its role in catabolite repression [1] . However , there is evidence for an extended role of cAMP as sensory signal involved in global gene regulation in bacteria [2]–[5] . The level of intracellular cAMP is modulated by several environmental factors [6]–[8] . The cellular target for cAMP-signaling is the cAMP receptor protein ( CRP ) . Dimeric CRP in complex with one molecule of cAMP exhibits DNA-binding activity to sites located near promoter regions [9] . Thereby , CRP-cAMP acts as a global regulator of gene expression by controlling the expression of almost 200 operons in E . coli [10]–[12] . Type 1 fimbriae mediate attachment to both biotic and abiotic surfaces and are involved in the early stages of biofilm formation [13] , [14] . In E . coli , type 1 fimbriae play a crucial role during urinary tract infections by mediating adhesion to mannose-containing receptors on the uroepithelium and promoting the formation of intracellular bacterial communities [15]–[17] . Those adhesins are encoded by the fim determinant composed of two independent transcription units coding for the recombinases FimE and FimB , and a polycistronic operon encoding the structural components ( FimA , FimF , FimG , and FimH ) and a pilus assembly system ( FimC and FimD ) [18] , [19] . Phase variable expression of the fim operon is associated with the inversion of a 314-bp chromosomal region , flanked by two 9-bp inverted repeats , that contains the fimA promoter [20] , [21] . When the invertible element is in the so-called ON orientation , the promoter is directed towards the structural fim genes , thus allowing transcription , whereas transcription is abolished in the inverted OFF orientation . The inversion process is catalyzed by FimB and FimE , two members of the tyrosine site-specific recombinase family [22] , [23] . Several regulators are involved in the fine modulation of the expression of type 1 fimbriae by environmental conditions [24] , [25] . A proper supercoiling state of the DNA and the presence of accessory proteins , such as the DNA binding proteins Lrp and IHF , are essential features that affect the recombination process and determine whether the cell is fimbriated or not [26]–[29] . Other regulators such as RpoS , ppGpp , NanR and NagC modulate type 1 fimbriation mostly by altering the expression of the recombinases that catalyze the recombination event [30]–[32] . Moreover , the global regulator H-NS has been shown to affect type 1 fimbriation both by regulating the expression of the recombinases and by directly interacting with the fim invertible element [33] , [34] . Effects on the expression of type 1 fimbriae in cya derivatives of E . coli K-12 , which are defective in cAMP synthesis , were reported earlier [35] . However , different strains responded divergently upon addition of exogenous cAMP in static cultures and it was not clarified at what level the reported effects were operating . In this work , we describe that CRP-cAMP represses type 1 fimbriation . The dissection of the mechanism underlying the observed phenomenon demonstrated that CRP-cAMP indirectly represses FimB-mediated recombination during the phase variation process . In contrast to many other regulators of the phase variation of type 1 fimbriae described , CRP-cAMP affects phase variation independently of the levels of the recombinases . We propose a novel model by which CRP-cAMP controls the type 1 fimbriation state in the bacterial population by affecting DNA gyrase activity . In addition , our studies led to the new discovery that Lrp expression in E . coli is under the control of the CRP-cAMP complex .
For a successful colonization of hosts by bacteria , it is crucial that the expression of bacterial surface structures , which mediate the interaction with the host tissues , is finely regulated . In E . coli , the CRP-cAMP complex has been shown to regulate the production of several of those surface structures , such as flagella or P-fimbriae [36]–[39] . Using a crp deletion mutant derivative of the extensively studied uropathogenic E . coli ( UPEC ) isolate J96 , we further characterized the role of CRP-cAMP in the modulation of the expression of those colonization factors . Confirming previous data , the CRP-cAMP deficient derivatives were non-motile and had lost the ability to cause mannose-resistant haemagglutination ( MRHA ) ( data not shown ) . Agglutination tests using specific antisera against the Pap and Prs fimbriae , adhesins that mediate MRHA , confirmed that the expression of those fimbriae is strictly dependent on the presence of functional CRP-cAMP in the cell ( data not shown ) . J96 , as most of the UPEC isolates , also expresses type 1 fimbriae , which are essential for the adherence and invasion of the bladder uroepithelium . The expression of type 1 fimbriae can be detected by mannose-sensitive yeast agglutination ( MSYA ) , which attests the ability of type 1 fimbriated bacteria to bind to mannosides-containing receptors on the surface of yeast cells . A clear stimulation in the ability to cause MSYA was observed in the J96crp strain as compared with wt when growing in various culture media ( LB , TBA , CFA , and TSA; data not shown ) . Semi-quantitative MSYA , using serially diluted LB cultures , corroborated these results: agglutination of yeast cells was observed with a higher dilution of the J96crp cell suspension ( 4-fold , i . e . containing 8-times less bacterial cells ) as compared to wt ( Table 1 ) . These results indicate that the deficiency in CRP-cAMP caused a substantial increase in the expression of type 1 fimbriae on the cell surface . In UPEC , a regulatory crosstalk between fimbrial operons occurs , which also affects the expression of type 1 fimbriae . It is known that the UPEC-specific regulators PapB , SfaB , and FocB , which are involved in the regulation of P-related , S-related , and F1C-related fimbriae , respectively , have the ability to repress the expression of type 1 fimbriae [40]–[42] . It has been described that CRP-cAMP is essential for the expression of the operon encoding the P-related fimbriae and the PapB regulator [39] . Therefore , a possible explanation for the observed increase in type 1 fimbriation in the crp derivative could be the lack of expression of UPEC-specific repressors of type 1 fimbriae such as PapB . To test this , MSYA experiments were performed using wt and crp derivatives of an E . coli K-12 strain , which lacks such regulators . The VL751 strain ( mutant in the chromosomal fim gene cluster and consequently deficient in type 1 fimbriation ) , carrying the entire fim determinant of the J96 strain on the pACYC184-based plasmid pSH2 was used . crp derivatives of this commensal strain expressing the fimJ96 determinant had enhanced agglutination ability as compared with the wt strain ( Table 1 ) . As expected , strains carrying pACYC184 did not agglutinate yeast cells . Furthermore , type 1 fimbriae expression was also monitored in the K-12 strain MG1655 ( fim+ ) and its crp derivative , giving the same result ( Table 1 ) . As a control , in all MSYA assays described , the agglutination could be effectively blocked by the addition of mannosides ( data not shown ) . Since the up-regulation of type 1 fimbriation was observed in both UPEC and commensal isolates , our results exclude the possibility that the CRP-cAMP effect strictly requires any UPEC-specific factor or is solely due to regulatory crosstalk between fimbrial operons , although we can not rule out a possible contribution to the regulation in pathogenic isolates . The fimA gene encodes the major subunit of the type 1 fimbriae and its expression is phase variable . Transcriptional studies were performed using two lineages derived from AAEC198A and AAEC374A strains , which carry the same fimA-lacZ operon fusion in the chromosome . The AAEC374A derivatives CBP374 ( wt ) and CBP375 ( Δcrp ) are phase variation deficient due to mutations in the fimB and fimE genes ( encoding the site-specific recombinases ) and have the invertible element locked in the ON orientation . Such strains were used to monitor the transcriptional activity of the fimA promoter . The AAEC198A derivatives CBP198 ( wt ) and CBP199 ( Δcrp ) are phase variation proficient and consequently , the fimA expression monitored using these strains integrates both the percentage of fim-expressing cells ( phase variation ) and the activity of the fimA promoter . Using strains CBP198 and CBP199 , a clear fimA up-regulation was observed in the crp background as compared with wt ( Fig . 1A ) , indicating that CRP-cAMP represses type 1 fimbriation at the transcriptional level , consistent with the agglutination data ( Table 1 ) . On the other hand , when using CBP374 and CBP375 strains ( Fig . 1B ) , a significantly lower fimA promoter activity was detected in the crp mutant , thereby suggesting that CRP-cAMP stimulates fimA promoter activity itself . Having in consideration that CRP activity is strictly dependent on its co-factor cAMP , in-frame cya deletion mutant strains ( cAMP-deficient strains ) were created and compared with isogenic crp and wt strains . The effect of the cya mutation on fimA expression resembled the one observed in the crp strain ( Fig . 1A and 1B ) . Moreover , restoration of CRP-cAMP activity in crp and cya derivatives by using either the low copy number plasmid pCBP68 ( pLG338-crp ) or external addition of cAMP , respectively , restored fimA expression to wt levels ( Fig . 1A and 1B ) . Taken together , we may conclude that CRP-cAMP has a dual role in the transcriptional expression of type 1 fimbriae: i ) to stimulate transcription of the fimA promoter in phase-ON-cells and , ii ) to repress the overall type 1 fimbriae expression . A response in the expression of type 1 fimbriae upon addition of exogenous cAMP was reported in static cultures of different cya derivative strains [35] . However , the response was divergent in different bacterial strains . The physiological heterogeneity in static cultures might be the cause of the strain-dependent variation detected . Our experiments using shaken cultures growing under uniformly aerated conditions gave identical results with all the strains used . Moreover , an increased fimA transcriptional expression in a CRP-cAMP deficient strain could be extracted from a microarray dataset on the effect of the crp mutation on the global pattern of expression in E . coli [10] , consistent with our results . To further probe into the dual role of CRP-cAMP in fimA expression , the effect of rapid restoration of CRP activity by addition of cAMP to cya strains was studied . When using early log phase cultures of the phase variation proficient strain CBP198 and its cya counterpart ( Fig . 1C ) , the exposure to cAMP during one hour period did not significantly alter the expression of fimA when compared to the expression in control cultures ( no addition of cAMP ) . However , a significant rapid alteration in expression ( p = 0 . 036 after 20 minutes ) as a response to the addition of cAMP was detected when using the phase variation deficient strains CBP374 ( wt ) and CMM374 ( Δcya ) ( Fig . 1D ) . The differences in the kinetic of the response suggest that the dual role of CRP-cAMP on fimA expression might be achieved by distinct mechanisms: a direct stimulation of the fimA promoter activity and an indirect role in the overall negative effect of CRP-cAMP on type 1 fimbriae expression . So far , we have described that CRP-cAMP deficiency causes: i ) a higher degree of type 1 fimbriation , ii ) an increase in the expression of type 1 fimbriae in phase variation proficient strains , and iii ) a reduction in fimA promoter activity in phase variation deficient strains . These results suggest that the percentage of fimbriated cells ( ON-cells ) in both crp and cya mutant strains is elevated when compared to wt , causing the overall increase in fimA expression . To test this prediction , the percentage of ON-cells in the population was monitored by plating cultures of the phase variable fimA-lacZ reporter strains on indicator plates containing X-gal . As predicted , a significant increase in the percentage of ON-cells in both the crp and cya mutant strains was observed ( Fig . 2A ) . A quantitative PCR based method was validated ( see Materials and Methods and Fig . S1 ) and used to detect and quantify the subpopulations having the invertible element either in the ON or in the OFF orientation [40] . Consistent with the results obtained from indicator plates , a significantly higher percentage of ON-cells was found for CMM198 ( Δcya ) as compared to CBP198 ( wt ) ( Fig . 2B ) . Moreover , the cya deficiency was chemically complemented by addition of exogenous cAMP in the culture medium . The effect of CRP-cAMP deficiency in the switching between the ON and OFF orientation was further corroborated when in vivo switching frequencies were measured ( see below ) . Similar results were obtained when using the reporterless and type 1 fimbriation-proficient strain MG1655 and its cya mutant derivative ( Fig . 2B ) , thereby excluding the possibility that the lacZYA DNA sequence present in the reporter strains might affect the CRP-mediated effect on phase variation . Moreover , when derivatives of the uropathogenic isolate J96 were used , a significant increase in the percentage of ON-cells was detected in crp derivatives . While most of the cells in the wt population were in the OFF orientation under the culture conditions used , a subpopulation of cells with the invertible element in the ON orientation was clearly detected in the mutant derivative ( Fig . 2C ) . To confirm these results , the level of fimA transcript in cultures of J96 and its derivative J96crp were quantified by Northern blot analysis ( Fig . 2D ) . A 2 . 4-fold increase in fimA transcript was detected in J96crp as compared with wt , corroborating the results obtained when using fimA-lacZ reporter strains ( Fig . 1A ) . The drop in fimA expression in CRP-cAMP deficient strains carrying mutations in fimB fimE ( Fig . 1B ) was assumed to indicate a stimulatory effect of CRP-cAMP on fimA promoter activity . However , it could also be a consequence of an alteration of the percentage of ON-cells by the action of some alternative FimB/FimE-like recombinase , as described for several E . coli strains [43] . As depicted in Fig . 2E , OFF-cells were not observed in cultures of strains CBP374 ( wt ) , CBP375 ( Δcrp ) and CMM374 ( Δcya ) , thereby ruling out the involvement of alternative recombinases , which is consistent with the fact that no genes for such enzymes are detected in the MG1655 genome [43] . Taken together , our results suggest that CRP-cAMP acts on the phase variation process by causing a decrease in the percentage of fimbriated cells in the population . Type 1 fimbriation is growth phase dependent [31] , [32] . The fimA expression profile throughout the growth curve was studied using the strains CBP189 ( wt ) and CPB199 ( Δcrp ) . As previously described , fimA expression in wt cultures was low in the early growth stages , increased in the middle of the logarithmic phase , and stayed constantly high throughout stationary phase ( Fig . 3A ) . In contrast , fimA expression in the crp mutant peaked during early logarithmic phase and then dropped down to almost wt levels during late-logarithmic phase . Consistent with the transcriptional data , a larger difference in the semi-quantitative phenotypic determination of type 1 fimbriae expression ( MSYA ) was observed with mid-log phase cultures of the wt and crp derivatives of MG1655 ( 1/2 versus 1/8 , respectively ) when compared to stationary phase cultures ( 1/4 versus 1/8 , Table 1 ) . The analysis of fimA expression through the growth curve suggests that CRP-cAMP represses type 1 fimbriation in actively growing cells , while during growth arrest , the repression is released and other global regulators such as RpoS and ppGpp assume the control [31] , [32] . This finding is also in agreement with the described growth phase-dependent levels of CRP-cAMP in the cell . As assessed by Northern blot analysis , crp transcriptional expression is high in early exponential phase and significantly reduced in stationary phase [44] . A well-described factor that alters the intracellular levels of CRP-cAMP is carbon source availability , e . g . the presence of glucose causes a significant reduction [6] , [7] . The effect of glucose on the expression of type 1 fimbriae was monitored . A modest but significant increase in the percentage of fimA-expressing cells could be observed when CBP198 ( wt ) cultures were grown in M9-glucose medium compared with cultures grown in M9-glycerol ( Fig . 3B ) . The stimulatory effect of the presence of glucose on transcriptional expression of type 1 fimbriae was also observed by microarray analysis on the effect of glucose in the general expression pattern in E . coli [45] . CRP-cAMP deficient strains have a significant growth defect compared to the wt ( i . e: 89 and 34 minutes generation time in LB for CBP199 and CBP198 , respectively ) , which might raise the question whether the increased type 1 expression in the crp strains is merely due to the growth alterations . However , growth in media that significantly increases the growth rate of the crp strain , i . e . LB medium containing glucose ( 32 and 48 minutes generation time for CBP198 and CBP199 , respectively ) , did not alter the difference in the expression of type 1 fimbriae between the wt and the crp strains ( data not shown ) , suggesting that the CRP specific effect on type 1 fimbriae expression is not coupled to the growth rate . The reported increase in the percentage of ON-cells in the CRP-cAMP deficient strains could be achieved either by stimulating the OFF to ON inversion ( exclusively catalyzed by FimB ) or by causing the opposite effect on the ON to OFF inversion ( mainly catalyzed by FimE ) . To further dissect the role of CRP-cAMP in the recombination event , the percentage of ON-cells in wt and cya derivative strains expressing either FimB ( AAEC370A , fimE ) or FimE ( AAEC261A , fimB ) was determined ( Fig . 4A ) . In the FimB proficient strains ( fimE ) , a significant increase in the percentage of ON-cells was detected in the strain lacking CRP-cAMP ( 16% in cya versus 4% in wt ) . However , in FimE proficient strains ( fimB ) , consistent with published results [46] , all cells were in the OFF orientation independently of the presence or absence of the CRP-cAMP complex . These results suggest that CRP-cAMP is directly or indirectly affecting the FimB-mediated inversion . To corroborate these data , in vitro recombination assays were performed using template plasmids as recombination substrate in bacterial extracts of cya and cya+ strains overexpressing either FimB or FimE . The induction of the synthesis of the recombinases in cultures of the cya and cya+ strains provided apparently identical amounts of the enzymes in the extracts of both strains as determined by Coomassie-stained SDS-PAGE ( Fig . S2 ) . When FimB-mediated OFF to ON inversion was monitored ( Fig . 4B ) , recombination occurred with both cya and wt extracts in the presence of FimB . However , a remarkable 3-fold higher percentage ( p = 0 . 003 ) of invertible fragments in the ON orientation was detected in the extract from the cya strain when compared with wt extracts . On the other hand , FimE-mediated inversion from the ON to the OFF state did not seem to be affected by a mutation in the cya gene ( Fig . 4C ) . The FimB recombinase can also catalyze the switch from ON to OFF . However , no effect of CRP-cAMP on the FimB-mediated ON to OFF inversion was detected when in vitro recombination assays with DNA template in the ON orientation were performed ( data not shown ) . Altogether , our in vitro studies corroborate the results obtained in vivo and suggest that the CRP-cAMP complex specifically affects the FimB-mediated recombination event from the OFF to the ON orientation . Supporting this conclusion , in vivo switching frequency estimations indicated that the OFF to ON switching rate was significantly increased in strain CBP199 ( Δcrp ) as compared with CBP198 ( wt ) ( 1 . 1×10−4 and 1 . 4×10−2 per cell and generation in wt and mutant , respectively ) , while no significant effect was observed in the ON to OFF switching ( 1 . 0×10−6 and 1 . 6×10−6 per cell and generation in wt and mutant , respectively ) . Also supporting our results , it was reported that the FimB-mediated switching frequency from OFF to ON is 3-fold higher in the presence of glucose ( i . e . reduced intracellular levels of CRP-cAMP ) than in the presence of glycerol [24] . Although higher expression of fimB was observed in crp derivatives as compared to wt counterparts in both J96 and MG1655 strains ( Fig . 2D and data not shown ) , the in vitro data , where the recombinases were overexpressed to the same degree in both extracts , suggest that the enhanced OFF to ON switching in absence of CRP-cAMP is not strictly dependent on the levels of the FimB recombinase . To further test this hypothesis , in vivo experiments were performed under conditions of constitutive fimB expression using plasmid pPKL9 , which contains the fimB gene under the control of the tet promoter ( Fig . 4E ) . Control experiments by Northern blot analyses verified that the fimB expression levels from plasmid pPKL9 were essentially identical in CRP-cAMP proficient and deficient genetic backgrounds ( data not shown ) . The percentage of ON-cells in cultures of J96 derivatives constitutively expressing fimB ( carrying plasmid pPKL9 ) was significantly elevated in the CRP-cAMP deficient strain as compared with wt , yielding a 50% higher percentage of ON-cells . Comparable results were obtained when using MG1655 derivative strains ( data not shown ) . Altogether , our results both in vivo and in vitro indicate that the CRP-cAMP complex has a negative effect on the switching process independently of the intracellular concentration of the FimB recombinase . Two possible mechanisms by which CRP-cAMP affects the FimB-mediated switch should be considered: either CRP-cAMP can directly interact with the invertible DNA fragment repressing the OFF to ON switch , or the effect of CRP-cAMP may be indirect . The slow response when adding exogenous cAMP to CMM198 cultures ( Fig . 1C ) suggested that the role of CRP-cAMP in the regulation of the phase variation occurs by an indirect mechanism . Nevertheless , to establish whether CRP-cAMP might also be directly involved in the switching process , in vitro recombination assays were performed using extracts of the cya strain while restoring CRP-cAMP activity by addition of increasing amounts of cAMP ( Fig . 5A ) . No obvious alteration in the FimB-mediated switch was detected , strongly suggesting that CRP-cAMP does not directly interact with the nucleoprotein complex that is the substrate for the FimB recombinase . Accordingly , no effect was observed in the outcome of in vitro recombination assays when purified CRP was added to extracts obtained from a crp strain ( data not shown ) . Simultaneously , the possible binding of CRP-cAMP to various DNA fragments spanning different regions of the fim determinant was tested ( Fig . S3A ) . No strong CRP binding was detected to any of the DNA fragments tested . At most , a low affinity binding was detected in case of the fragment containing the fimA promoter ( PCR7; Fig . S3B ) . However , when DNase I footprinting analysis of this putative CRP binding site was performed , no binding was observed ( data not shown ) . It has been reported that CRP-cAMP might bind to many low affinity binding sites along the E . coli chromosome [47] . Although it is possible that such low affinity CRP binding site ( s ) may exist in the fimA promoter region , our experimental evidence ( Fig . 5A ) suggested that binding is not required for the phase variation control . A possible involvement of the putative CRP binding site ( s ) in the positive control of the fimA promoter activity ( Fig . 1B ) will be further studied . Recently , it has been shown that inhibiting the DNA gyrase promotes the FimB-mediated inversion from OFF to ON and therefore it was concluded that DNA supercoiling determines the directionality of the FimB-mediated recombination [29] , [48] . DNA gyrase is an enzyme that catalyses ATP-dependent DNA breakage , strand passage and rejoining of double-stranded DNA ( for a recent review see Nöllmann et al . [49] ) . DNA gyrase is involved in the regulation of DNA topology , but also in other processes such as replication or illegitimate recombination [50] , [51] . Remarkably , it has been described that CRP-cAMP modulates the expression of the gyrA gene encoding the DNA gyrase . In crp deficient strains , low levels of gyrA expression and DNA gyrase activity , monitored as alterations in the topology of plasmid DNA , were detected [52] . One may hypothesize that the CRP-cAMP mediated effect on the FimB-recombination process could directly result from the low levels of DNA gyrase activity detected in crp deficient strains . To test this hypothesis , the effect of inhibiting the DNA gyrase in vivo was analyzed in both wt ( CBP198 ) and cya ( CMM198 ) strains ( Fig . 5B ) . Addition of increasing amounts of novobiocin ( DNA gyrase inhibitor ) in wt cultures caused a concomitant increase in fimA expression , consistent with previously reported data [29] , [48] . Remarkably , the fimA expression level was essentially unaltered by addition of novobiocin in cultures of the cya mutant strain . In agreement with the hypothesis proposed , the novobiocin mediated inhibition of the DNA gyrase caused an increase in the percentage of ON-cells in the wt strain , but not in the cya derivative ( Fig . 5C , upper panel ) . Results that further corroborated our hypothesis were obtained by inducing overexpression of DNA gyrase from cloned gyrAB genes in the cya strain CMM198 . Both repression of fimA expression and reduction in the percentage of ON-cells were detected ( Fig . S4 ) . Moreover , when fimE mutant derivatives were used , thus only reflecting FimB-mediated inversion , an identical response to novobiocin was observed , indicating that the recombination process that was responsive to gyrase inhibition in vivo is FimB-specific ( Fig . 5C , lower panel ) . To rule out the possibility that the lacZYA sequences present in the fimA-lacZYA fusion might cause alterations in the regional DNA supercoiling and consequently affect the phase variation , similar experiments were performed using reporterless derivatives of strains MG1655 and J96 . Similar results were obtained: i ) an increase in the percentage of ON-cells in the wt strains was observed after addition of increasing novobiocin concentration ( 5-fold and 2-fold increase with the highest concentration of novobiocin tested in MG1655 and J96 strains , respectively ) , ii ) the level of ON-cells was not altered by novobiocin treatment in the CRP-cAMP deficient derivatives , and iii ) the percentage of ON-cells in the wt achieved by novobiocin treatment was similar to the level detected in the CRP-cAMP deficient derivatives ( data no shown ) . It is noteworthy that in all approaches ( Fig . 5B and 5C ) the presence of novobiocin in sub-inhibitory concentrations did not significantly alter the expression of fimA in the cya mutant strains , which is in agreement with a low DNA gyrase activity in the CRP-cAMP deficient background as a result of the low expression of the gyrA gene [52] . To corroborate the in vivo results obtained , in vitro analyses were performed where increasing amounts of novobiocin were added to the wt strain extract . A progressive increase in the OFF to ON switching efficiency in vitro was observed ( Fig . 5D ) , consistent with our in vivo data and with previously reported data [29] , [48] . Moreover , the unaltered fimA expression in the cya mutant strain by addition of novobiocin , together with the in vitro switching data , indicates that the fimA promoter is indifferent to changes in the DNA gyrase activity , in agreement with previous data [48] . The fact that inhibition of the DNA gyrase activity in vitro stimulated the FimB-mediated recombination suggests an active role of the DNA gyrase during the recombination process itself . Altogether , our data provide evidence that the induction of type 1 fimbriation detected in the CRP-cAMP deficient strains is a process mediated by the alteration in DNA gyrase activity and therefore can be mimicked by the specific inhibition of this enzymatic activity by novobiocin . Moreover , our in vitro results comparing OFF to ON switching between wt and cya mutant using the same DNA template for both extracts ( Fig . 4B and Fig . 5D ) demonstrated that the CRP-cAMP effect on phase variation is not merely dependent on the initial supercoiling state of the fim invertible element . Recombination at the fim invertible element requires Lrp , a DNA bending protein that directly binds to specific sites within the invertible element and stimulates DNA inversion [26] . Kelly et al . [29] demonstrated that this binding activity of Lrp is required to promote the FimB-mediated OFF to ON directionality observed when the DNA gyrase was inhibited . The Lrp levels were determined by immunoblot analysis of crp+ and crp strains ( Fig . 6A ) and the Lrp content detected was several fold higher in the crp strains than in the crp+ strains . These results suggest a possible link between the CRP and Lrp regulons . Interestingly , a direct demonstration of CRP dependent regulation of Lrp expression has not been done , although two putative CRP sites have been predicted in the promoter region of the lrp gene [53] , suggesting a possible direct regulation by CRP-cAMP . Additionally , CRP-cAMP could act indirectly by positively regulating GadE , which represses lrp expression [54] , [55] . Transcriptional studies have been performed by Northern blot analysis of RNA from derivatives of MG1655 and J96 ( Fig . 2D ) . An increase in the level of the lrp transcript in the crp derivatives was detected as compared with wt ( 2 . 5 and 1 . 7- fold in MG1655 and J96 derivatives , respectively ) , suggesting a role for CRP-cAMP in the control of lrp transcription , although an additional regulation at the posttranscriptional level can not be ruled out . Further studies would be required to characterize the CRP-cAMP dependent regulation of Lrp expression . Interestingly , when the levels of intracellular Lrp were monitored in the same cultures as in Fig . 5B , again a differential effect of the inhibition of the DNA gyrase in wt and cya strains was observed ( Fig . 6B ) . In a wt strain , the Lrp levels were strongly elevated at the highest concentration of novobiocin , where the amount of Lrp was apparently identical to the amount detected in the cya strain in absence of novobiocin , which again might be explained by the low DNA gyrase activity detected in the CRP-cAMP deficient background [52] . We tested whether increased levels of Lrp by itself would cause an alteration in the FimB-mediated switching process . Our results clearly indicate that Lrp overexpression per se did not result in any significant changes in the percentage of ON-cells when no alteration in DNA gyrase activity was induced ( Fig . 6C ) . The expression of type 1 fimbriae implies the allocation of an important part of the asset of the bacterial cell for the production of those proteinaceous appendages , as considerable amounts of energy and amino acids are needed for their synthesis . A tightly regulated expression of such organelles is therefore expected . Considering the important metabolic effort performed by the bacterial cells committed to be fimbriated , regulation by phase variation can be seen as a selective advantage for the bacterial population , in addition of providing phenotypical heterogeneity in an otherwise genetically homogeneous population . In previous works , we have shown that the expression of type 1 fimbriae is stimulated when intracellular levels of the stringent response alarmone ppGpp are raised [32] , [56] . The level of ppGpp in the cell increases under amino acid starvation and energy stress [57] . In this report , the role of CRP-cAMP , a regulatory complex that is associated with the energy state of the cell , has been included in the extensive list of regulatory networks controlling type 1 fimbriation . Interestingly , many of the global regulators that affect type 1 fimbriae expression , such as H-NS , RpoS , Lrp , and now CRP-cAMP , have been shown to interplay among each other , thereby orchestrating gene regulation cascades in response to the growth conditions [58] . From our studies , we can conclude that CRP-cAMP is a major regulator of fimbriation during the exponential growth phase ( Fig . 3A ) and is required to maintain the growth expression profile of type 1 fimbriae . We dissected the initial observation that the crp derivatives of J96 showed an increased ability to agglutinate yeast cells and conclude that CRP-cAMP represses type 1 fimbriation , as schematically shown in Fig . 7 , by the recently described mechanism of switching directionality established by the activity of the DNA gyrase and the presence of Lrp [29] , thereby affecting FimB-mediated OFF to ON switching . Remarkably , CRP-cAMP inhibited FimB-mediated recombination at a template plasmid isolated from a crp+ background , indicating that the regulatory effect does not merely depend on the supercoiling state of the DNA and thereby suggesting an active role of the DNA gyrase in the OFF to ON recombination event . Interestingly , CRP-cAMP has a dual effect on type 1 fimbriation by repressing phase variation and promoting promoter activity . Further studies will be required for fully understanding the underlying mechanisms by which CRP-cAMP affects both levels of regulation of type 1 fimbriation . In Salmonella , crp cya mutants are avirulent in a mouse model [59] and it has been reported that the crp and the cya genes are strongly repressed during infection of macrophages [60] . Moreover , it has been observed that DNA becomes more relaxed when bacteria are growing in certain intracellular environments and consequently the expression of those genes that are required for intracellular survival is induced [61] . Therefore , CRP-cAMP might be involved in controlling Salmonella virulence in a pathway that includes DNA supercoiling and the sensing of environmental conditions as previously proposed [62] . It is well described that CRP and cAMP levels are affected by environmental conditions such as glucose availability and osmolarity [6] , [7] . Interestingly , such environmental conditions also affect DNA topology in E . coli in a DNA gyrase dependent manner [63] , [64] . The link between CRP-cAMP mediated regulation of gene expression and DNA gyrase activity might represent a specialized signal transduction pathway that senses the metabolic and energetic status of the cell . It can not be ruled out that in this regulatory pathway others factors such as the FIS protein might be involved . FIS has been proposed earlier as a metabolic sensor involved in the homeostatic control of DNA supercoiling . Interestingly , CRP-cAMP modulates fis expression [65] . Supporting this model , a correlation between the sensitivity to catabolite repression and to gyrase inhibitors has been observed earlier for different metabolic operons . Hence , inhibition of DNA gyrase activity represses the expression of several CRP-cAMP sensitive genes ( three maltose operons , the lactose and galactose operons , and the tryptophanase gene ) , whereas CRP-cAMP independent genes such as threonine and tryptophan operons were insensitive to DNA gyrase inhibitors [66] , [67] . We can consider a scenario where disadvantageous environmental conditions , which are sensed by the bacterial cells as low rate of energy flow in the cell , are transduced by different pathways ( CRP-cAMP regulon , ppGpp regulon ) , inducing survival strategies including cell fimbriation . The fact that the highly energy-consuming process of fimbriation is stimulated in conditions of starvation highlights the putative selective advantage represented by the ability of expressing type 1 adhesins , which promotes important virulence properties such as biofilm formation and colonization of host mucosa . The physiological relevance of the CRP-cAMP-mediated signaling pathway controlling type 1 fimbriation needs to be further explored . Nevertheless , the modulation of type 1 fimbriation mediated by glucose availability may provide a possible explanation to the reported observation that diabetic patients commonly are highly susceptible to urinary tract infections [68] . The presence of an unusual high concentration of glucosides in the urine of those patients may cause a reduction in the intracellular level of CRP-cAMP in the bacterial cell . This could result in a concomitant increase in type 1 fimbriation which , in turn , would raise the probability of colonization of the urinary tract .
All strains and plasmids used in this study are listed in Table 2 . Strains were grown to mid-log phase ( corresponding to an OD600nm of around 0 . 5 ) with vigorous shaking ( 200 rpm ) at 37°C in Luria Bertani ( LB ) medium [69] or in M9 medium [70] supplemented with either 0 . 4% glycerol or 0 . 4% glucose , unless otherwise stated . For mannose-resistant haemagglutination ( MRHA ) , bacteria were grown in CFA ( 1% casaminoacids , 0 . 15% yeast extract , 0 . 05% MgSO4 , 0 . 005% MnCl2 ) . For mannose-sensitive yeast agglutination ( MSYA ) , bacteria were grown in different culture media: TBA ( 1% tryptone , 0 . 5% NaCl , 1 . 5% agar ) , TSA ( 1 . 5% trypticase peptone , 0 . 5% phytone peptone , 0 . 5% NaCl , 1 . 5% agar ) and CFA . When necessary , the following antibiotics were used: tetracycline ( 12 . 5 µg ml−1 ) , carbenicillin ( 50 µg ml−1 ) , kanamycin ( 25 µg ml−1 ) and chloramphenicol ( 15 µg ml−1 ) . When indicated , cyclic AMP was added in a final concentration of 5 mM . To study type 1 fimbriae expression , cultures of the different strains were always inoculated using colonies showing an OFF ( non-fimbriated ) phenotype . When fimA-lacZ fusion derivatives were used , OFF-colonies could be identified on X-gal plates ( white colonies ) . In reporterless strains ( MG1655 and J96 derivatives ) , the fimbriation state of the inoculum was estimated from the colony morphology , since ON-colonies are small and convex , while OFF-colonies are large and flat , as described by Blomfield et al . [71] . Standard molecular manipulations were performed according to Sambrook and Russel [70] . The cya deletion mutant ( Δ21–259 ) and Δcya::Kmr deletion mutant ( Δ21–259::Kmr ) were created by allelic exchange as described by Link et al . [72] . The deletion mutant and Kmr deletion mutant were verified by PCR amplification using primers cya-A and cya-D , and cya-up and cya-D , respectively . Gene alleles were introduced by phage P1-mediated transduction [73] using the following donor strains; BEU742 for Δcrp39 ( Tcr ) , SS5357 for Δlrp::Tcr and CMM2 for Δcya::Kmr . Derivatives crp+ and Δcrp39 were initially selected by colony size and confirmed by PCR using primers CRP1 and CRP3 . The plasmid pAAG6 was constructed by cloning a PCR-amplified fragment spanning the lrp gene between the EcoRI-SmaI sites of pBAD30 . The PCR fragment was generated using the primers lrp-1 and 64 and MG1655 as template . All primers used are specified in Table S1 . Bacteria were grown on CFA agar plates . Bacterial cell suspension in PBS containing 3% ( w/v ) mannose ( methyl α-D-mannopyranoside , Sigma ) were prepared and adjusted to an OD600 nm of 5 . MRHA was tested using suspensions from human ( agglutination-positive with P-fimbriae ) and dog ( agglutination-positive with Prs-fimbriae ) erythrocytes ( 8% , v/v ) in PBS , giving identical results . The erythrocytes and bacterial suspensions were mixed in proportion 1:1 ( v/v ) on a glass-slide . Presence of aggregates was considered as agglutination positive . Bacteria were grown on CFA agar plates . Bacterial colonies were mixed with 10 µl of 20× diluted antisera . Two types of antisera were used , pPAP5 and pPAP60 ( originally obtained against P and Prs fimbriae , respectively ) that strongly cross-react and mediate agglutination through both types of fimbriae [74] . Presence of aggregates was considered as agglutination positive . Bacteria were grown on LB plates overnight at 37°C , washed in PBS and resuspended to an OD600nm of 5 . Yeast cells ( Saccharomyces cerevisiae ) were washed and resuspended in PBS to an OD600nm of 5 . The suspensions were mixed in a 1:1 ( v/v ) ratio on a glass-slide placed on ice . After 30 min , the presence of aggregates as sign for agglutination was assessed and scored as + ( weak ) or ++ ( strong ) . Semi-quantitative analysis was performed as described before [32] . Overnight cultures were diluted in LB and adjusted to an OD600nm of 1 . 0 and 4 µl were put on semi-solid LB medium containing 0 . 3% agar and incubated at 37°C . The β-galactosidase assay was performed as described by Miller [75] . Data presented represent average values and standard deviations of duplicate measurements from at least three independent experiments . To monitor the orientation of the fim invertible element in strains carrying a chromosomal fimA-lacZ fusion , indicator plates containing X-Gal ( 5-bromo-4-chloro-3-indolyl-ß-D-galactopyranoside ) were used as previously described [24] . Cells having the fim invertible element in the ON orientation give rise to blue colonies , whereas those cells having it in the OFF orientation exhibit white color . The orientation of the invertible DNA fragment can be determined using a molecular approach described previously [32] , [40] . In brief , a 602-bp DNA fragment containing the fim invertible element was PCR amplified using primers 2535 and 3137 , HinfI restricted and analyzed on TBE-acrylamide-gels . Depending on the orientation of the fim invertible element , this method generates different sized fragments ( 484 and 118 bp when in the ON orientation , 402 and 200 bp when in the OFF orientation ) . Quantification of the percentage of ON-cells in a specific sample was performed as described by Aberg et al . [32] . To prove the reliability of this method , calibration experiments were performed in triplicate using templates from strain CBP374 ( a 100% ON-cells sample ) and CBP198 ( representing the OFF-cells samples ) . Both strains were grown to the same optical density and mixed such that the fraction of ON-cells in the template varied from 0 to 100% . PCR amplification , HinfI digestion and gel electrophoresis were performed as described above . The applicability of this method to quantify the percentage of ON-cells in bacterial populations was evidenced by a linear correlation [regression coefficient ( R2 ) of 0 . 998] between the percentage of ON-cells in the sample and the intensity of the bands ( Fig . S1 ) . Since the invertible element in CBP198 is not completely in the OFF orientation , the intensity derived from ON-cells in the “100%-OFF” sample was subtracted from all intensity values . Total RNA was extracted from mid-log cultures using the hot-phenol method [76] . Contaminating DNA was removed by DNase I ( Roche Diagnostics ) treatment for 1 h at 37°C , followed by RNA cleanup using the RNeasy Mini kit ( Qiagen ) . For Northern analysis , 20 µg of total RNA were separated on a formaldehyde:agarose gel and transferred to Hybond-N membranes ( Amersham ) by capillary blotting as described [77] . The membrane was hybridized with 32P-labeled DNA-fragments corresponding to the coding sequences of fimA , fimB , lrp , and rrnA , which were PCR generated using the primer pairs fimA-RT1&2 , fimB-RT1&2 , lrp-RT1&2 and 16S-RT1&2 , respectively . After hybridization overnight at 52°C , membranes were subsequently washed in 1× SSC-0 . 1% SDS for 15 min at room temperature and in 0 . 1× SSC-0 . 1% SDS for 15 min at 52°C . Autoradiograms were obtained using StoragePhosphor screens ( Molecular Dynamics ) , which were scanned using the Storm Imaging System ( Molecular Dynamics ) . FimB and FimE-promoted switching frequencies were measured as previously described [24] . To perform FimB in vitro recombination assays , bacterial extracts were obtained from cultures of the fim mutant strain NEC026 and its isogenic cya mutant strain harboring the plasmid pIB378 ( fimB gene under the control of an IPTG inducible promoter in pET11 ) . fimB expression was induced with 0 . 4 mM IPTG after the cultures grown in minimal MOPS [78] at 28°C had reached an OD600nm of 0 . 15 . Cells were harvested after 24 h of induction at 28°C and processed as described [40] , [79] . As control , extracts lacking FimB were obtained from cultures of the same strains carrying the pET11 plasmid . To perform FimE in vitro recombination assays , bacterial extracts were obtained from cultures of the strain NEC026 and its isogenic cya mutant strain transformed with the plasmid pIB382 ( fimE gene under the control of an IPTG inducible promoter in pET11 ) . Cultures were manipulated as described above . The in vitro recombination assay was performed as described [40] , [79] . The resulting orientation of the invertible element was analyzed after 3 h incubation at 37°C using the PCR-based method described above . Whole cell extracts from bacterial cultures were separated by SDS-PAGE as described by Laemmli [80] using 15% polyacrylamide gels . Samples were transferred to PVDF membranes using a semidry blotting apparatus . After blocking the membranes overnight in Tris-buffered saline containing 0 . 1% Tween-20 ( TBS-T ) and 5% skimmed milk , membranes were incubated for 1 h at room temperature with 2 , 000× diluted Lrp-specific antiserum or 6 , 000× diluted PapA-specific antiserum [38] in TBS-T containing 5% skimmed milk . After 3× 15 min washes in TBS-T , membranes were incubated for 1 h with 20 , 000× diluted anti-rabbit immunoglobuline-horseradish peroxidase conjugate ( Dianova , Hamburg , Germany ) . After further washing , membranes were developed using the enhanced chemiluminescence ( ECL+ ) kit ( GE Healthcare ) and analyzed on a Chemidoc System ( BioRad ) equipped with the QuantityOne® Software for quantification . Differences between average values were tested for significance by performing an unpaired , two-sided Student's t-test . The levels of significance of the resulting p values are reported by the following symbols: * = p<0 . 05; ** = p<0 . 01; *** = p<0 . 001 and n . s . = non-significant .
|
Attachment of bacteria to the surface of host tissues is a crucial initial step in the establishment of bacterial infections . This process is mediated by adhesins , such as the type 1 fimbriae of Escherichia coli , which play a key role during urinary tract infections by mediating adhesion to the uroepithelium . The expression of type 1 fimbriae is finely regulated attending to environmental signals and is under phase variation control , which determines the percentage of fimbriated cells in the population . In this report , we show that the expression of type 1 fimbriae is repressed by a metabolic sensor of the cell , the global regulatory complex CRP-cAMP . We demonstrate that CRP-cAMP affects the switching outcome by selectively inhibiting the recombination process in one direction only , resulting in a lower percentage of fimbriated cells . Such a switch to the non-fimbriated state after successful adhesion might be advantageous in the urinary tract , where the immune mechanisms of the host favor the removal of bacteria expressing immunogenic surface structures . Understanding the regulatory networks that govern regulation of virulence and colonization factors is both of basic interest and might help to develop novel strategies to treat bacterial infections .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Materials",
"and",
"Methods"
] |
[
"cell",
"biology/microbial",
"physiology",
"and",
"metabolism",
"microbiology/microbial",
"growth",
"and",
"development",
"microbiology/cellular",
"microbiology",
"and",
"pathogenesis"
] |
2009
|
Type 1 Fimbriae, a Colonization Factor of Uropathogenic Escherichia coli, Are Controlled by the Metabolic Sensor CRP-cAMP
|
The mammalian odorant receptor ( OR ) repertoire is an attractive model to study evolution , because ORs have been subjected to rapid evolution between species , presumably caused by changes of the olfactory system to adapt to the environment . However , functional assessment of ORs in related species remains largely untested . Here we investigated the functional properties of primate and rodent ORs to determine how well evolutionary distance predicts functional characteristics . Using human and mouse ORs with previously identified ligands , we cloned 18 OR orthologs from chimpanzee and rhesus macaque and 17 mouse-rat orthologous pairs that are broadly representative of the OR repertoire . We functionally characterized the in vitro responses of ORs to a wide panel of odors and found similar ligand selectivity but dramatic differences in response magnitude . 87% of human-primate orthologs and 94% of mouse-rat orthologs showed differences in receptor potency ( EC50 ) and/or efficacy ( dynamic range ) to an individual ligand . Notably dN/dS ratio , an indication of selective pressure during evolution , does not predict functional similarities between orthologs . Additionally , we found that orthologs responded to a common ligand 82% of the time , while human OR paralogs of the same subfamily responded to the common ligand only 33% of the time . Our results suggest that , while OR orthologs tend to show conserved ligand selectivity , their potency and/or efficacy dynamically change during evolution , even in closely related species . These functional changes in orthologs provide a platform for examining how the evolution of ORs can meet species-specific demands .
Odorant receptors ( ORs ) expressed at the cell-surface of olfactory sensory neurons ( OSNs ) in the main olfactory epithelium detect chemical cues in the proximate environment . The ability to detect these cues is crucial for survival of an individual and a species; odorants can signify favorable or toxic food sources , mating preferences , predators and habitat [1] , [2] . The ecological niche that an animal inhabits is directly associated with the OR repertoire in each species [3] , but we do not know how the functional OR repertoire evolves to maximize an animals' fitness . As a first step to understand the functional evolution of ORs , it is essential to compare OR function in related species , paying attention to the evolutionary relationship of each tested receptor . Gene orthology is a key concept in evolutionary and functional genomics . Here we define orthologs as genes derived from a single ancestral gene that diverged since a speciation event; this is in contrast to paralogs , which are genes related via gene duplication [4] , [5] ( Figure S1 ) . Orthologous genes typically perform equivalent–if not identical—functions , especially when comparing closely related species , while paralogs are thought to be more divergent in function [4]–[7] . Indeed , this is a key assumption in a wide variety of biological research , as it allows research from model organisms to translate into health interventions in humans [7] . Comparisons of sequenced genomes show that many orthologous genes can be identified between divergent species , but the functional equivalency of the vast majority have not been experimentally tested [4] . The OR repertoire suffered extensive gains and losses of genes between species , resulting in a significant decline in the number of putatively functional ORs in primate species when compared to the OR repertoires in other mammals [8]–[12] . Comparisons of high-coverage primate genomes revealed that the size of the functional OR repertoire and percentage of pseudogenized ORs is quite similar between human , chimpanzee ( Great Ape ) and rhesus macaque ( Old World Monkey ) [12] , [13] . However , between humans and chimpanzees , approximately 25% of the OR repertoire exists in only one species [13] , suggesting adaptation to differing environments to meet species-specific demands [3] , [14] , [15] . Importantly , many ORs have clear orthologs in closely related species [13] . Sequence similarity of ORs is often used as a proxy for functional variability [9] , [16]–[19] , but this assumption remains largely untested [9] , [20] due to the paucity of functional data matching ORs with ligands . While full-length sequence comparison provides insight into the evolutionary relationship of ORs , it is thought to have less predictive value about the binding sites of these receptors [9] . Man et al . ( 2004 ) proposed a set of 22 amino acids important for ligand binding under the assumption that orthologs will have more similar odor specificities than paralogs , showing a greater conservation in amino acids residues at odor-binding sites than across the entire coding region [9] , [17] . Several studies have examined changes in ligand selectivity and sensitivity of OR orthologs , but these studies were limited to a single receptor set [17] , [21]–[23] . It is still unclear whether or not this change in sensitivity of orthologs is restricted to a specific family of ORs or is a more general phenomenon across all OR orthologs . To understand how the olfactory system has evolved and how the human OR repertoire was shaped , we must identify the functional changes of orthologous ORs between species . With the development of a high-throughput in vitro assay for OR function , we are now able to directly test how well OR sequence similarity predicts function [24]–[26] . Here we conduct the first multi-receptor comparison of ligand selectivity and sensitivity of OR orthologs in primates and rodents and , further , ask if orthologs respond to a common ligand more often than OR paralogs from the same subfamily .
Starting with a set of human ORs previously matched to at least one ligand ( deorphaned ORs ) [27] , we searched the chimpanzee and rhesus macaque genomes for putatively functional orthologous gene sets where there was no evidence of gene duplication in either species ( one-to-one ortholog ) [13] . We identified 18 ORs that have putatively functional orthologs: 12 orthologous trios ( in all three species ) , five human-chimp duos that lack orthologs in macaque and one human-macaque duo that lacks a chimp ortholog . Additionally , we identified 17 mouse-rat ortholog sets where there is at least one known ligand for the mouse OR ( for sequences of ORs used in functional experiments , see Table S1 ) [27] , [28] . Using the similarity of amino acid properties [29] , we constructed a tree of all 390 putatively functional human ORs; the 18 orthologous primate sets used for analysis are highlighted on the tree ( Figure 1A ) . These ORs represent Class I and II receptors , seven of the 13 families described by Hayden et al . ( 2010 ) [3] and contain human ORs shown to be both broadly and narrowly tuned to odors [27] . Our mouse-rat orthologs also cover both Class I and II ORs and represent 17 of the 228 families described by Zhang and Firestein ( 2002 ) [18] ( Figure 1B ) . This suggests that our set of ORs is not significantly biased towards any particular family of ORs . Using the Jukes-Cantor model of nucleotide substitution rates [30] , comparison between orthologs are consistent with the expected phylogeny between species . Human-chimp orthologs are the most similar , followed by human-macaque and chimp-macaque orthologs , with mouse-rat orthologs being the most divergent ( Figure S2A , Table S2 ) . We used Grantham's distance to compare amino acid similarity of the entire open reading frame ( ORF ) among ortholog sets [29] . Our results show our OR sets are not biased to ORs with highly similar amino acid substitutions ( Figure S2B , Table S2 ) . Additionally , we compared the Grantham's distance of the 22 amino acids predicted to be involved in ligand binding by Man et al . ( 2004 ) [17] . We found that all human-chimp orthologs are identical at these 22 positions , while six human-macaque orthologs and two mouse-rat orthologs differ , but the amino acid substitutions are fairly conservative ( Table S2 , Figure S3A ) . Previous literature suggests there is evidence for both positive and purifying selection in the OR repertoire , so to determine if our OR sets broadly represent genes evolving under different selective pressures , we calculated the ratio of nonsynonymous to synonymous substitutions ( ω or dN/dS ) for our 18 orthologous primate sets and for each of the putatively functional 259 human-chimp or 152 human-macaque 1∶1 orthologs from Go and Niimura ( 2008 ) [13] ( Figure S4 ) . In the absence of selective pressure , the rates of synonymous substitutions per synonymous site ( dS ) are equal to that of nonsynonymous substitutions ( change the resulting amino acid ) per nonsynonymous site ( dN ) , thus , ω = dN/dS = 1; ω<1 suggests evidence of purifying selection and ω>1 indicates evidence of positive selection acting on a gene [31] . The distributions of ω are significantly different between human-chimp and human-macaque ( median human-chimp ω = 0 . 608; median human-macaque ω = 0 . 319; z = −9 . 61 , p<0 . 001 , Wilcoxon Rank Sum ) ( Figure S4 ) . All human-macaque gene pairs have ω<1 , while the human-chimp gene pairs show a wide distribution of ω values ( for branch-test and branch-site tests , see [13] ) . The median ω value of mouse-rat orthologs was 0 . 124 , consistent with previous literature [32] . To determine if gene orthology accurately predicts the functional properties of orthologs , we expressed each OR in a heterologous cell system , using a cyclic adenosine monophosphate ( cAMP ) -mediated luciferase reporter gene to assay the function [26] . We tested each orthologous OR set against a panel of chemically diverse odors to compare their ligand selectivity and responses . We chose a panel of 42 chemically diverse odors to represent most of “odor space” using a method described previously [27] , [33] , and tested these chemicals in triplicate at 100 µM ( Figure S5 , Table S3 ) . Within an OR set , the response of orthologs across the panel of odors was consistent , but with differences in the overall magnitude of the response of an OR ( negative values on the y-axis indicate an odor elicited an inhibitory response ) ( Figure 2 , Figure S6 ) . For example , human OR2W1 responded to 12 ligands , while chimp OR2W1 responded to the same 12 odors but with a diminished magnitude . In some instances the response of the human and mouse ORs could not be used to predict the OR function in other species , as the ligands tested did not activate the orthologous receptors . To address concerns that there is a species-specific interaction between ORs and variants of the Receptor Transporting Protein-1 short form ( RTP1S , the accessory protein necessary for functional expression of ORs at the cell surface ) we tested the functional consequence of swapping human and mouse versions of RTP1S with four human and four mouse ORs [25] , [26] , [34] . With the tested ORs , our data did not support the idea that mouse RTP1S was the most efficient for trafficking only mouse ORs and human RTP1S was the most efficient at trafficking the human ORs ( F ( 7 , 80 ) = 1 . 03 , p = 0 . 416 , 2-way ANOVA ) ( Figure S7 ) . To determine how well sequence similarity among orthologs can predict the function of ORs , we plotted the relationship of Jukes-Cantor distance ( J-C , nucleotide ) , Grantham's distance ( amino acid similarity ) using the entire ORF or 22 amino acid positions predicted to be involved in ligand binding , and ω ( dN/dS ) versus the functional distance of the orthologous sets . Here we define the functional distance as the correlation between OR responses across the 42 odor panel , where a receptor responded to three or more odors . Jukes-Cantor and pairwise ω values do not correlate with functional distance ( J-C , rs = 0 . 14 , p = 0 . 36; ω , rs = 0 . 18 , p = 0 . 24 , Spearman's correlation ) ( Figure 3A , 3C ) . Amino acid similarity using the ORF has a correlation to functional distance ( rs = 0 . 38 , p = 0 . 01 , Spearman's correlation ) and amino acid similarity using predicted binding residues is similar but slightly less significant ( rs = 0 . 32 , p = 0 . 04 ) ( Figure 3B , Figure S3B ) . Additionally , we did not find a correlation between sequence similarity and the number of ligands that activate each OR ( ω , rs = 0 . 02 , p = 0 . 92; J-C , rs = 0 . 10 , p = 0 . 49; Grantham ORF , rs = 0 . 03 , p = 0 . 83; Grantham 22AA , rs = 0 . 02 , p = 0 . 87 , Spearman's correlation , data not shown ) . Finally , the removal of primate OR sets that lack orthologs in one of the three primate species ( five human-chimp duos and one human-macaque duo ) from our analysis did not change the overall conclusions when comparing Jukes-Cantor , Grantham and ω values to the functional distance , suggesting that inclusion of these data does not bias our results . We next wanted to examine the functional changes in sensitivity of orthologs to individual ligands by testing each ortholog across a range of concentrations . We selected a single odor for a given OR to construct a representative dose-response curve , fit the data to a sigmoid curve , and then compared the response of the human allele against the primate orthologs and the mouse allele with the rat ortholog using an extra sum of squares test . Looking at both the potency ( EC50 ) and efficacy ( dynamic range ) of each OR to a particular ligand , an orthologous pair was classified as either indistinguishable , hyper/hypo functional ( one OR had both a lower potency and efficacy ) , or undefined ( orthologs were different but potency and efficacy did not change concordantly ) ( Figure S8 ) . Within each OR set , we saw dramatic differences in the overall potency and efficacy to a particular ligand ( Figure 4 , Figure S9 ) . For example , human and chimp OR8K3 orthologs are indistinguishable in their response to ( + ) -menthol ( Extra sum of squares test , F ( 3 , 36 ) = 0 . 13 , p = 0 . 944 ) , but human and chimp OR8K3 are hypofunctional in comparison to macaque OR8K3 when tested with the same ligand ( Extra sum of squares test , human to macaque F ( 3 , 36 ) = 15 . 16 , p<0 . 001 , chimp to macaque F ( 3 , 36 ) = 17 . 40 , p<0 . 001 ) ( Figure 4C , Table S4 ) . Macaque OR10G7 and human OR10G7 are indistinguishable in response to eugenol ( Extra sum of squares test , F ( 3 , 36 ) = 0 . 97 , p = 0 . 418 ) , but chimp OR10G7 is hypofunctional to human ( Extra sum of squares test , F ( 3 , 36 ) = 54 . 54 , p<0 . 001 ) and macaque ( Extra sum of squares test , F ( 3 , 36 ) = 84 . 82 , p<0 . 001 ) ( Figure 4B , Table S4 ) . Additionally , mouse and rat ORs showed differential responses to a given ligand ( Figure 4D , 4E ) . If we define functional differences as changes in potency and efficacy of a common ligand , comparison of our set of human ORs to primate orthologs revealed functional differences 87% of the time , while mouse-rat orthologs differed 94% of the time ( Figure 5A–5D ) . If our human ORs are randomly compared to other human alleles of the same OR , their dose-response curves are functionally different 25% of the time ( Mainland et al . , unpublished ) . In other words , sequence variation within the human population does not alter receptor function as much as sequence variation between orthologs in closely related species . To further address the question of how well orthology predicts function , we compared the response of orthologous ORs from closely related species to that of orthologs from more distant species and to ORs that are classified in the same subfamily based upon sequence similarity . For our sets of primate orthologs , we identified the putative human-mouse ortholog , as defined by the reciprocal ‘best-hit’ with >80% amino acid identity , and human OR paralogs—members of the same subfamily [11] , [35]—and tested these receptors against a common ligand . Comparison of sequences using Neighbor-Joining phylogenetic analyses showed that our primate orthologs are most similar to the human reference OR , while the mouse best-hit ORs are more distantly related . Human paralogs have a unique relationship for each OR group ( Figure S10 ) , but are generally less related than the primate orthologs . In sum , our ortholog and paralog assignment is congruent with speciation and gene duplication events . Overall , we find that orthologs respond to a common ligand 82% of the time while human OR subfamily members respond to a common ligand 33% of the time . Species-specific comparison of orthologs showed human-chimp orthologs respond to a common ligand 93% ( 14/15 ) of the time , human-macaque 67% ( 8/12 ) , and human-mouse ( 10/12 ) 83% . Using the above criteria to define changes in function to a given ligand , we again find significant differences in the potency and efficacy of each OR within a group ( Figure 6 , Figure S11 , Table S5 ) . For example , human OR5K1 and mouse ortholog mOR184-3 respond to eugenol methyl ether ( human to mouse F ( 3 , 39 ) = 21 . 59 , p<0 . 001 , undefined ) but human OR5K1 is hyperfunctional to both chimp and macaque OR5K1 orthologs . None of the three human 5K family paralogs respond to this ligand ( Figure 6A , Table S5 ) . Human OR8D1 is hyperfunctional to chimp and macaque OR8D1 orthologs , while human paralogs 8D2 and 8D4 do not respond ( Figure 6B , Table S5 ) . Mouse ortholog mOR171-22 is hypofunctional to OR8D1 ( F ( 3 , 42 ) = 873 . 69 , p<0 . 001 ) while mOR171-9 does not respond . From our analysis , orthologs respond to a common ligand more often than OR paralogs of the same subfamily , albeit with differences in sensitivity , suggesting that OR paralogs in the same subfamily may show distinct ligand selectivity . Orthologs were more similar than paralogs when measuring Grantham's amino acid similarity using both the entire ORF and the 22 predicted binding residues ( z = −6 . 61 , p<0 . 0001 ORF; z = −7 . 35 , p<0 . 0001 22AA , Wilcoxon Rank Sum ) ( Figure 7A , Table S6 , Figure S12 ) . Orthologs that responded to the same odor as the human reference OR were not significantly different from orthologs that did not respond when comparing amino acid similarity of both the ORF and the 22 predicted binding residues ( z = 0 . 89 , p = 0 . 37 ORF; z = 1 . 22 , p = 0 . 22 , 22AA , Wilcoxon Rank Sum ) ( Figure 7B ) . This suggests that amino acid similarity did not accurately predict OR function among orthologs . While amino acid similarity of the ORF did not predict the response of paralogs ( z = −1 . 47 , p = 0 . 14 , Wilcoxon Rank Sum ) , the amino acid similarity of the 22 predicted binding residues was significantly different , with responding paralogs being more similar in sequence ( z = −3 . 54 , p<0 . 0004 , Wilcoxon Rank Sum ) ( Figure 7B lower panel , Table S6 ) . Our data suggest that comparing the 22 residues involved in ligand binding is better than the entire ORF when predicting the response of OR paralogs . To determine if individual differences in receptor activity are influenced by the amount of receptor at the cell surface , we assessed cell surface expression of each OR [36] , [37] . Using fluorescent immunocytochemistry in live cells , we measured the Cy3 signal intensity of each ortholog and paralog in our receptor set and compared them against the human counterpart . Within each set of receptors , we found cell-surface signal levels did not predict the potency of the OR to a single ligand ( Figure 8 , Figure S13 ) . For example , human , chimp and macaque OR2W1 are similar in their receptor tuning with differences in response magnitude ( Figure 2A , 2B ) and have differences in EC50 values to a common ligand , allyl phenyl acetate , while human paralogs OR2W3 and OR2W5 do not respond to the common ligand ( Figure 4A , Figure 8A ) . Surface labeling of human OR2W1 was not significantly different from either orthologs or paralogs ( Figure 8B , 8C , Table S7 ) , consistent with the idea that surface expression levels of ORs do not predict sensitivity of ORs . No OR surface expression results in no response to known ligands [34] . However , ORs with very intense surface staining are not necessarily responsive to a common ligand , nor are they the most sensitive to that ligand if they do respond . ORs with very few detectable receptors at the surface still showed functional activity , suggesting receptor amount does not dictate response ( Figure S13 ) .
Here we showed that OR orthologs are similarly tuned within an OR set , but that dramatic differences in efficacy and potency to a common odor are frequent . These functional changes are not specific to the primate lineage where significant gene loss has impacted the size of the OR repertoire and a decline in the relative importance of the olfactory system is commonly assumed [8] , [10] , [13] , [14] , [38] , as we see similar changes in rodent orthologs . Comparison of primate orthologs , more distantly related orthologs ( human-mouse ) and human OR paralogs of the same subfamily suggest that orthologs respond to a common ligand more often than other subfamily members . This idea is consistent with the approach used by Man et al . ( 2004 , 2007 ) comparing residues conserved in orthologs and differing in paralogs to predict 22 amino acid residues involved in ligand binding [9] , [17] . In our example of human OR8D1 and mouse orthologs mOR171-22 and mOR171-9 , we see that mOR171-22 responds to a common ligand while mOR171-9 does not ( Figure 6B ) . Comparison of the predicted binding residues shows that mOR171-22 is identical to human OR8D1 at all 22 sites , while mOR171-9 differs at one position ( Figure S12 , Table S6 ) ; each receptor has an overall amino acid identity of 85% to the human ortholog . While the amino acid similarity using the ORF or the 22 predicted binding residues did not predict OR ortholog response , the amino acid similarity of the 22 predicted binding residues was significantly different for paralogs that responded to a common ligand , ( z = −3 . 54 , p<0 . 0004 , Wilcoxon Rank Sum ) ( Figure 7B lower panel , Table S6 ) . Thus , the identification of a true ‘functional ortholog’ must be supported by both bioinformatics and experimental approaches . A significant portion of mammalian ORs are orphan receptors , although progress has been made in matching individual OR-ligand interactions [9] , [27] . Not all ORs with an intact ORF are necessarily functional . Similarities between coding sequences are used to predict functionality in lieu of real experimental data , thus , ORs grouped into the same subfamily are thought to share similar functional properties based upon sequence homology [10] , [38]–[40] . This has also led to the idea that OR orthologs from different species will maintain the same olfactory capabilities [9] , [16] . We have taken the approach of using an evolutionary analysis to examine the relationship of OR-ligand interactions . Several studies have looked at changes in ligand selectivity and sensitivity of OR orthologs using a single receptor . One functional study identified 18 ligands that activate human paralogs OR1A1 and OR1A2 . Human OR1A1 and mouse ortholog Olfr43 shared 9 common odors; twelve amino acids thought to influence ligand binding properties overlap with the prediction from Man et al . ( 2004 ) [17] , [21] . Krautwurst et al . ( 1998 ) showed that the orthologous mouse I7 and rat I7 receptors show changes in the fine-tuning of ligand selectivity , with mouse I7 preferring heptanal to octanal , and the reverse was shown for the rat ortholog [22] . Zhuang et al . ( 2009 ) showed that OR7D4 orthologs from many primate species differed in potency and efficacy to a common ligand [23] , but the question remained whether these functional differences extended to many ORs or if OR7D4 was a special case . Androstenone and androstadienone , the steroid ligands for OR7D4 , are found in human male sweat , urine and semen , and have extreme perceptual differences in the human population [41] , [42] . Additionally , these odorous steroids have been linked to changes in physiological response in both males and females , making them a unique case [43] , [44] . Our data suggest that many ORs show dynamic functional changes during evolution , thus OR7D4 is not the special case . In the case of many mammalian OR orthologs , we show that amino acid changes have dramatic functional consequences on the OR . While the ligand selectivity across OR orthologs is similar , there are changes in the magnitude of response; there are also frequent functional changes in the potency and efficacy of response to a common odor . We find that orthologs respond to a common odor more often than paralogs ( 82% versus 33% , respectively ) . Species-specific comparison of orthologs shows human-chimp orthologs responding to a common odor 93% ( 14/15 ) of the time , human-macaque orthologs 67% ( 8/12 ) of the time and human-mouse orthologs 83% ( 10/12 ) of the time , again with differences in potency and efficacy . While sequence comparison predicts human-macaque orthologs to be more similar than human-mouse best-hit pairs , it is interesting that human-mouse orthologs respond to a common ligand more often . While our results raises a possibility of accelerated functional changes in the macaque lineage , further investigation with more ORs and additional macaque species will be necessary . However , this result must be interpreted with caution , as our sample size may not extrapolate to a larger data set and our assignment of human-mouse orthologs is based upon the mutual best-hit assignment in BLAST . The use of best-hit to define our human-mouse orthologs may result in an OR pair representing a one-to-many relationship , in contrast to the more closely characterized one-to-one relationship among primate and rodent orthologs . However , investigating the phylogenetic relationship of our orthologs does reflect overall predicted speciation events , supporting the idea that human-macaque comparisons should be more functionally conserved ( Figure S10 ) . One caveat in our study is that we do not examine within-species variation; we are using only one allele from one animal in a particular species and using these data to represent the complexity of OR response . Though we do not know functional variation within non-human species , a study comparing human alleles of the same OR shows much lower frequency of functional differences ( 25% ) when comparing response to a common ligand ( Mainland et al . , unpublished ) , suggesting within species variation would account for minor fraction of functional differences in our analysis . In the future , it would be very interesting to address the within-species allelic variation from primates and rodents to examine how sequence variation within species impacts OR function in comparison with what is seen in the human population . Aside from ORs , there are several examples where the functions of orthologous genes are not equivalent [5]–[7] . First , the FOXP2 transcription factor that plays a role in speech and language in humans , is highly conserved from mice to humans , differing at only 3 amino acid positions; despite the similarity , the genetic substitution of mouse Foxp2 with the human ortholog results in differences in ultrasonic vocalizations and affects dopamine levels , dendrite morphology , gene expression and synaptic plasticity of medium spiny neurons in the basal ganglia [45]–[47] . Second , a functional comparison of rhodopsin genes from 35 vertebrate species and 11 reconstructed ancestral genes revealed that each rhodopsin receptor has a specific wavelength of maximal absorption that can be related to the environmental changes of an organism's habitat [48] . There are also studies elucidating changes in ligand selectivity of nuclear hormone receptors using ancestral reconstruction [49]–[51] , showing changes in ligand preference , sensitivity and general function over time . Human and chimpanzee bitter taste receptors at the TAS2R38 locus show changes in potency to the known ligand PTC ( phenylthiocarbamide ) when tested in a heterologous system [52] , raising the possibility that bitter receptors might also show dynamic functional evolution in closely related species . One concern is that our in vitro system does not mimic the in vivo olfactory sensory system and that the expression of primate receptors is systematically misrepresented . Our data suggest that human RTP1S is capable of trafficking ORs from different species and that these receptors can couple to the canonical signaling pathway , sometimes outperforming the human version of the OR . We can also interchange human and mouse versions of RTP1s and do not see a pattern of species-specific RTP-OR interactions ( Figure S7 ) . While our set of human ORs tends to be hyperfunctional in comparison to primate orthologs , it does not mean that human ORs as a whole are necessarily more functional . Our selection process for OR orthologs began with previously deorphaned human ORs , thus , we would expect all human receptors to show a response while the chimp and macaque orthologs are variable . An additional concern is that codon bias between species may alter the expression levels , leading to variability that would not exist in the in vivo system . Our data include two human-chimp orthologs ( OR5P3 and OR8K3 ) that differ at the nucleotide level but have identical amino acid sequence . For these pairs of OR orthologs , the response is indistinguishable across 42 ligands and within a single odor ( Figure 4C , Figure S6 , Figure S9 ) . This suggests that sequence variation at the nucleotide level does not impact the results in our heterologous expression system . Cell surface expression levels of individual ORs do not appear to dictate the changes in potency of a given receptor in our assay , suggesting the functional changes are an inherent property of the receptor itself ( Figure 8 , Figure S13 ) . While this assay may not provide the resolution for low-levels of OR expression , our data is consistent with the idea that cell-surface expression level does not correlate with OR function , though a minimum level of cell-surface expression , facilitated by the accessory proteins , is required [25] , [26] , [34] , [42] . It is important to note that for ORs that do not have known ligands , a lack of response to a given odor does not mean these ORs are nonfunctional . Rather , it is possible that these ORs have acquired a new functional activation by different odors not used in our assay . In addition to ligand binding to the ORs , it is plausible that differences in G-protein coupling or receptor recycling do exist between these receptors and should be investigated in the future . There are several studies that have shown the reliability of this in vitro system to predict in vivo function and odor perception . Comparison of patch-clamp recordings of olfactory sensory neurons expressing the mouse receptor SR1 to heterologous cells transiently expressing SR1 showed similar patterns of activation to a panel of odors; however , the response from the heterologous system did appear less sensitive than the intact olfactory sensory neurons [53] . In another study , variants of human OR7D4 were shown to respond differently to the ligands androstenone and androstadienone in a heterologous system , and these differences translated to perceptual differences to the odors in the human population [42] . However , the in vitro system is not perfect . Our in vitro assay lacks many components of an in vivo olfactory system , including odorant binding proteins , a mucosal layer , intracellular molecules , and sniffing behaviors . The failure of a specific OR to respond to any of the tested odorants must be interpreted with caution , since it may reflect a failure of the OR to function in our assay rather than a lack of sensitivity to the tested odorant . Taken together , probing the functional differences of OR genes across species using an in vitro system is likely to provide useful information in understanding the evolution of the OR family . Comparisons of high-coverage sequenced genomes show that orthologous relationships of genes between divergent species can be identified for a majority of genes [4] , [7] . The idea that the identified function of a gene is upheld for orthologs of that gene across species is a widely accepted assumption for the progress of bioinformatics , as most sequenced genes may never be subjected to functional experimentation; for many examples in closely related species , this idea of equivalent function is upheld [4] , [54]–[61] . In the multi-gene family of ORs , it appears that even with a clear 1∶1 evolutionary relationship of orthologs between closely related species , that functional equivalency , in terms of efficacy and potency , is limited . To further understand functional changes of ORs , it will be necessary to test the functional properties of individual mammalian ORs from many species to determine if any ORs orthologs have undergone changes in ligand selectivity . Hayden et al . ( 2010 ) showed that the olfactory subgenome in different species is directly associated with the habitat in which the animal exists [3] . Additionally , a comparison of fruit fly Drosophila melanogaster with mosquito Anopheles gambiae OR repertoires suggests ecology has shaped the repertoires and that odorants are differentially encoded in a way consistent with ecological niches of each organisms; the two species show different coverage of a chemically defined odor space , tuned to their food-seeking preferences [62] . In the case of rhodopsin orthologs from vertebrate species , the change in functional wavelength adapted to the environment of the organism [48] . While we do not know if these OR functional changes have an impact on the behavior of an individual or species , we can speculate that the OR repertoire in each species has adapted to meet niche- and species-specific demands .
Starting with a list of human and mouse ORs with previously identified ligands [27] , we identified orthologous genes between human , chimpanzee and rhesus macaque [13] and between mouse and rat [28] . Putative human-mouse OR orthologs were defined as the reciprocal best-hit to the human reference OR with a >80% amino acid identity . Selection of additional human OR subfamily members ( paralogs ) were based upon receptors already cloned and available in our library . ORs were amplified from genomic DNA ( Coriell Cell Repositories ) using Phusion polymerase ( New England Biolabs ) and subcloned into a mammalian expression vector , pCI ( Promega ) , containing the first 20 amino acids of human rhodopsin ( Rho-tag ) . Each receptor was sequenced using a 3100 or 3730 Genetic Analyzer ( ABI Biosystems ) . Analysis of sequence variation was conducted in MATLAB . Evolutionary distance of the nucleotide sequences for each ortholog pair was calculated using the Jukes-Cantor model [30] and the amino acid comparisons were made using Grantham's scale [29] . The 22 amino acid alignment was conducted in Seaview using the ClustalW2 alignment method . The pairwise dN/dS ( ω ) was determined using the Nei-Gojobori 1986 method , based on the Jukes-Cantor model [63] The additional sequence data for ORs used in Figure S4 originated in Go and Niimura [13] , who originally conducted this pairwise analysis using the modified Nei-Gojobori method and additionally assessed selection pressure using branch-test and branch-site test for ORs . Neighbor-Joining trees were built in Seaview . Dual-Glo Luciferase Assay System ( Promega ) was used for the luciferase assay as previously described [26] . Rho-tagged ORs ( 5 ng/well ) were transfected into the Hana3A cell line in 95-well plate format ( Thermo Scientific ) along with the human receptor trafficking protein , RTP1S [25] ( 5 ng/well ) , pRL-SV40 ( 5 ng/well; Promega ) , CRE-luciferase ( 10 ng/well; Stratagene ) and muscarinic acetylcholine receptor ( M3 ) [24] ( 2 . 5 ng/well ) . Luminescence was measured using a Polarstar Optima plate reader ( BMG ) . First , all luminescence values were divided by the Renilla Luciferase activity to control for transfection efficiency and cell viability in a given well . Normalized luciferase activity was calculated by the formula ( LN-Lmin ) / ( Lmax-Lmin ) , where LN is the luminescence of firefly luciferase in response to the odorant , Lmin is the minimum luciferase value on a plate or set of plates , and Lmax is the maximum luciferase value on a plate or set of plates . Data was analyzed using GraphPad Prism 5 . 0 and MATLAB . 42 odorants that quantitatively span chemical space were chosen using a method previously described [27] , [33] . Briefly , we generated 20 physicochemical descriptors that predict 62% of the variance in mammalian OR responses [27] for 2683 commonly used odorants . We then divided the 2683 odorants into 42 clusters using k-means clustering . For each cluster , we selected the odorant closest to the centroid of the cluster among odorants that are previously shown to activate at least one OR . If no such ligand was present in the cluster , we selected the odorant closest to the centroid of the cluster to maximize structural diversity . Each orthologous set and a vector control ( Rho-pCI ) were tested against each odorant at 100 µM ( except androstenone , which was applied at 10 µM ) and compared to a no odor control; each comparison was performed in triplicate and statistical significance was assessed by a t-test with a correction for multiple comparisons ( 2-tailed t-test , α = 0 . 05/42 ) . The human and mouse ORs were used as the reference OR and chimpanzee , rhesus macaque and rat orthologs were the variant ORs . The order of odors is the same within a set of OR orthologs , but is different across ORs . Odors are listed in Table S3 . Dose-response curves were constructed using a single odor at concentrations ranging from 10 nM to 10 mM for the OR-odor pairs for each orthologous set . Each concentration was tested in triplicate and a vector-only control ( Rho-pCI ) was included for each odorant . The odors for dose responses were chosen before we determined the responses to the comprehensive set of 42 odors , thus , we did not always choose the best ligand for dose response curves , although the chosen ligands always robustly activated at least one of the tested orthologs . We tested all the orthologs against this panel of 42 odors and since we did not find changes in ligand selectivity among orthologs , we did not go back to test the best ligands for dose-responses . The dose-response data were fit to a sigmoid curve and the resulting data were fit with a 3-parameter logistic model . An odorant was considered an agonist if the 95% confidence intervals of the top and bottom parameters did not overlap , the standard deviation of the fitted log EC50 was less than 1 log unit , and the extra sums-of-squares test confirmed that the odorant activated the receptor significantly more than the vector-only transfected control . For each pair of ORs , we determined if one model fit the data from both ORs better than two separate models using the extra sums-of-squares test . A pair of ORs is classified as hyper/hypofunctional if one OR in the pair had both a higher EC50 ( lower efficacy ) and a lower potency ( dynamic range , or top-bottom ) . A pair of ORs was undefined if the potency and efficacy showed discordant changes . Dose-response curve images were graphed in GraphPad Prism 5 . 0 and further analyzed in MATLAB . For dose-response curves , data was baslined and normalized to the maximum response across a set of ORS ( Figure 4 , Figure 6 ) . In Figure S9 , the left column was normalized to the human or mouse OR variant response to show the differences in receptor basline activity and the identical data in the right column was baselined and normalized to the maximum response across the set of receptors . Classification of ORs using the above criteria was coducted in MATLAB . Hana3A cells were maintained in minimal essetial medium ( Sigma ) containing 10% fetal bovine serum ( Sigma ) ( M10 ) , 500 µg/ml peniciilin-streptomycin ( Invitrogen ) and 6 µg/ml amphotericin B ( Sigma ) [34] . Live-cell surface staining was done as previously described [25] , [36] , [37] . Briefly , Hana3A cells were seeded on poly-d-lysine coared glass coverslips in 35 mm dishes and transfected with 1000 ng OR , 250 ng RTP1S , and 50 ng of EGFP to control for transfection efficiency . 24-hours post-transfection , primary incubation was carried out at 4°C using mouse monoclonal antibody anti-rhodopsin 4D2 ( provided by R . Molday ) diluted 1∶100 in M10 containing 15 mM NaN3 and 10 mM HEPES ( Invitrogen ) for 45 min . Cells were washed in Hanks' balanced salt solution containing containing 15 mM NaN3 and 10 mM HEPES ( Invitrogen ) , followed by secondary incubation with Cy3-conjugated donkey anti-mouse IgG ( Jackson Immunologicals ) for 30 min at 4°C , fixed in 1% paraformaldehyde and later mounted in Mowiol . Slides were analzyed on a Zeiss Axioskop2 microscope at 40x oil lens and images were captured using QImaging Retiga 2000R camera and QCapture Pro 6 . 0 software . For ORs being compared , staining was performed in parallel and pictures were taken with the same exposure time , brightness and contrast . Images were anaylzed in Adobe Photoshop . Cy3 intensity was measured as integrated density ( grey value mean X area ) and quantifed for background levels and cell-surface expression . Background was subtracted from the cell-surface values and the average and S . E . M . were calculated for each receptor . Cy3 intensity was then compared to the human OR in each OR set using a student's t-test ( p<0 . 05 ) .
|
The mammalian odorant receptor repertoire has been subjected to significant gene duplication and gene loss between species , presumably to adapt to the environment of an organism . However , even in distantly related species , a clear orthologous relationship exists for many genes . While ligands have been identified for several ORs , many of these receptors remain uncharacterized , especially in species other than human and mouse . Due to this paucity of functional data , it is assumed that ORs with similar sequence share functional characteristics . Here we investigate the functional evolution of OR orthologs—genes related via speciation—and OR paralogs—genes related via a duplication event—to provide insight as to how this large gene family has evolved . We show that OR orthologs have similar ligand selectivity to a panel of odors but differ in response magnitude . Additionally , orthologs respond to a common ligand more often than human OR paralogs , but there are vast differences in the potency and efficacy of individual receptors . This result stresses the broad importance of combining evolutionary genomics and molecular biology approaches to study gene function .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"genetics",
"biology",
"evolutionary",
"biology",
"evolutionary",
"genetics",
"genetics",
"and",
"genomics",
"gene",
"function"
] |
2012
|
Functional Evolution of Mammalian Odorant Receptors
|
Hypoxia induces the expression of genes that alter metabolism through the hypoxia-inducible factor ( HIF ) . A theoretical model based on differential equations of the hypoxia response network has been previously proposed in which a sharp response to changes in oxygen concentration was observed but not quantitatively explained . That model consisted of reactions involving 23 molecular species among which the concentrations of HIF and oxygen were linked through a complex set of reactions . In this paper , we analyze this previous model using a combination of mathematical tools to draw out the key components of the network and explain quantitatively how they contribute to the sharp oxygen response . We find that the switch-like behavior is due to pathway-switching wherein HIF degrades rapidly under normoxia in one pathway , while the other pathway accumulates HIF to trigger downstream genes under hypoxia . The analytic technique is potentially useful in studying larger biomedical networks .
Molecular oxygen is the terminal electron acceptor in the mitochondrial electron transport chain . Hypoxia , or oxygen deficiency , induces a number of metabolic changes with rapid and profound consequences on cell physiology . A hypoxia-induced shortage of energy alters gene expression , energy consumption , and cellular metabolism to allow for continued energy generation despite diminished oxygen availability . A molecular interaction map of the hypoxia response network has been proposed [1–3] on the basis of analyzing conserved components between nematodes and mammals . The key element in this network , hypoxia-inducible factor ( HIF ) , is a master regulator of oxygen-sensitive gene expression [4–6] . HIF is a heterodimeric transcription factor which consists of one of the three different members ( HIF-1α , HIF-2α , and HIF-3α ) and a common constitutive ARNT subunit which is also known as HIFβ . The system also includes an enzyme family: prolyl hydroxylases ( PHDs ) , which directly sense the level of oxygen and hydroxylate HIFα by covalently modifying the HIFα subunits . It is very likely that reactive oxidative species ( ROS ) , which are a byproduct of mitochondrial respiration , are also involved in oxygen sensing by neutralizing a necessary cofactor , Fe2+ , for the hydroxylation of HIFα by a PHD [7–10] . There are three members in this enzyme family: PHD1 , PHD2 , and PHD3 . The hydroxylated HIFα is then targeted by the von Hippel-Lindau tumor-suppressor protein ( VHL ) for the ubiquitination-dependent degradation . Hypoxia response element ( HRE ) is the promoter of the hypoxia-regulated genes , and the occupancy of HRE controls the expression levels of these genes . The cascade in Figure 1 ( reproduced from Figure 2 of [1] ) consists of an input ( the concentration of oxygen ) and an output ( the activation of promoters that are under control of HREs ) as the core network . The network is characterized by a switch-like behavior , namely the sharp increase of HIFα when the oxygen decreases below a critical value , followed by a sharp increase of HRE occupancy . It was observed experimentally on many cell lines including Hela cells [11] and Hep 3B cells [12] that HIFα increases exponentially as the oxygen concentration decreases . The past two decades have seen a growing body of work on the use of mathematical modeling to help uncover both general principles behind molecular networks and to provide quantitative explanations of particular network phenomena [13] that may one day have sufficient predictive power to accurately model large subnetworks of the cell . In this sense , Kohn et al . [1] have successfully modeled the switch-like response characteristics of HRE occupancy , by numerically integrating a system of ordinary differential equations ( ODEs ) involving a score of molecular species related to hypoxia . The large model , however , does not identify the smaller components that are actually responsible for the switch-like response and that may occur in other such networks . Furthermore , a numerical solution does not provide the type of insight that mathematical formulas can . At the same time , it is virtually impossible to solve symbolically the type of nonlinear differential equations that model reactions . In this context , methods are desirable that are both tractable , that reduce a system to its key components , and that are not solely reliant on numerical solution . Extreme pathway analysis ( EPA ) is one such recently developed method [14–16] . In this method , the dynamics of interactions between species are formulated as a Boolean network in which the state of a gene is represented as either transcribed or not transcribed . Upregulation and downregulation of genes are captured through an appropriate sign ( plus or minus ) and a scaling constant . The Boolean network is then formulated as a matrix of interaction rules that is then analyzed to help reveal key components and their contributions to the dynamic behavior [16] . The theory of matrices then allows us to look for vectors that characterize the matrix in ways that are helpful for further analysis . The EPA technique , in particular , finds vectors ( extreme pathways ) that correspond to the boundaries of the space of steady-state solutions to the differential equations . We note that similar methods , such as flux balance analysis ( FBA ) and elementary modes analysis , have been developed in other contexts [17–19] . They essentially yield the same results [18] , which have been verified by ExPa [20] and CellNetAnalyzer [21 , 22] . These methods provide a way out of the intractable complexity of sizable molecular networks [23–26] . Our contribution is to go beyond this type of matrix approach and provide a detailed quantitative analysis that explains the observed behavior in the models . This is achieved by combining elementary pathway identification via EPA , which depends solely on the network topology , and the detailed analytical as well as numerical analysis of the governing differential equations in the model , which allows studies of the phase space spanned by the mostly unknown rate constants in the differential equations . Specifically , EPA is first used in our approach to decompose the original network into several underlying pathways . Following this , we make some reasonable approximations to facilitate analytic solution . We show that this analytic solution , in the case of the hypoxia network , explains the switch-like behavior . This explanation is confirmed by comparing the numerical output of the simplified model with the numerical output of the complete ( and complex ) differential equation model . A second contribution of this paper is to highlight a particular mechanism of pathway-switching or pathway branching effect [27] that appears to cause the sharp response to oxygen concentration . In particular , we examine the flux redistribution among the elementary pathways as a function of oxygen concentration . We also identify the key molecular species involved in the subcomponent of the network and show quantitatively how the response of this subcomponent exactly matches the overall response and thus is responsible for it . For hypoxia , our analysis suggests that the cycle of abundant production and efficient degradation of HIFα plays the main role in the sharp response .
For consistency and ease of understanding , we use the notation and nomenclature in [1] and use their 23-species network and differential equation model as the starting point . With this background , the original network shown in Figure 2 of [1] can be further reduced in the following way . Kohn et al . [1] have shown that the feedback of mRNA is not necessary for the switch-like behavior . We therefore eliminate this feedback loop ( reaction k32 ) . Hence , Transcript intermediate 1 , 2 , and 3 ( Species 8 , 9 , and 10 ) can also be dropped , as well as the associated reactions: k7 , k8 , k9 , k10 , and k11 , because they do not affect the dynamics of the network . Species 23 is only the joint name of HIFα:ARNT:HRE ( Species 7 ) and HIFαOH:ARNT:HRE ( Species 22 ) ; therefore , it is dropped . HIFα precursor ( Species 1 ) is a constant and is thus dropped , because its information can be simply encoded in the reaction k1 . The degradation products ( Species 2 ) are also eliminated because they are assumed to leave the network immediately after their production and do not affect the dynamics . Similarly , species 19 , 20 , and 21 do not contribute to the dynamics and are therefore removed . The resulting network is summarized in Tables 1 and 2 , where there are 13 molecular species and 19 reactions in total . The system can be described by the following set of ODEs where [Sn] stands for the concentration of species n as tabulated in Table 1 and [O2] indicates the input cellular oxygen concentration . Table 2 shows the specific reaction each rate constants kn represents . The real values of kn are from [1] . Note that the ODE system below is typical: the terms are based on mass-action principles and , taken together , result in complex behavior not readily discernible by examining the equations . We also dropped the precursor concentration [S1] since it is set to unity . This section assumes some familiarity with linear algebra . The 13 × 19 stoichiometric matrix Φ of the reduced hypoxia response network ( Tables 1 and 2 ) is shown in Figure 2 ( for details , see Materials and Methods ) . The rank of Φ is computed and shown to be 9 , indicating that there are only nine independent molecular species to serve as constraints for the analysis . Therefore , the dimension of the corresponding null space is 10 . The linearly independent basis B vectors are generated by Matlab 6 . 5 and are shown in Figure 3 . According to the constraint that no negative values are allowed in the basis vectors , we can uniquely transform basis B into basis P as shown in Figure 4 . Both b8 and b9 have negative terms . b8 has to be transformed first , otherwise there will be no +1 to cancel out the −1 at the twelfth row of b8 . Each −1 in b8 is canceled out through the operation b8 + b9 . In the second step , one has to use b7 to cancel −1 at the ninth row of b9 . In this way , we obtain the set of basis vectors P . The above analysis indicates that the dimension of this null space is the same as the number of edges for its corresponding convex cone [28] , which is the algebraic basis for extreme pathways [14] and elementary modes [29] . The ten basis vectors of P represent ten underlying pathways of the hypoxia network . They are illustrated in Figures 5 and 6 , from which one finds two distinct patterns: p1 , p7 , and p9 belong to the HIFα degradation pathways ( Figure 5 ) and the others belong to the simple association–dissociation pathways ( Figure 6 ) . More specifically , through p1 , HIFα is directly degraded by reaction k2 , a presumably oxygen-independent degradation pathway; whereas in oxygen-dependent pathways p7 and p9 , the hydroxylated HIFα is recognized by the VHL that channels it through a ubiquitin degradation component that is shown as the dotted box in Figure 5 . Even though p1 , p7 , and p9 are all elementary modes [29] , they can share certain reactions of the network . For example , the total influx for HIFα synthesis from a precursor can thus be decomposed into three parts with the overall rate constant k1 being given by γ1k1 , γ2k1 , and γ3k1 , where γ1 + γ2 + γ3 = 1 . The pathways p7 , and p9 are almost the same . The only difference is that HIFα is associated with ARNT in the middle part of the pathway p9 . Therefore , ARNT must be functionally very important , otherwise it would be hard to explain why the two underlying pathways , which should play significantly different roles , look so similar . Indeed , HIFα degrades differently through the two pathways . The k-sets were selected in [1] on the basis that they produced a switch-like behavior . For all the three k-sets , it was observed that HIFα has high affinity to PHD and low affinity to ARNT . The former is consistent with the usual case of high enzyme-substrate binding affinity . This implies that p9 is not the major degradation pathway because HIFα does not bind with ARNT very well . Moreover , p9 is immediately adjacent to HRE , which suggests its major role is to deliver signals to activate the promoters of hypoxia-regulated genes . As a signal transducer , the rate constant γ3k1 itself need not to be high; what the downstream genes are sensitive to is d ( γ3k1 ) /dt . Therefore , we hypothesize that there is a negligible flux through p9 ( or γ3 ≈ 0 ) . To verify our hypothesis , we calculate the γ1 , γ2 , and γ3 values , as the indication of the relative importance of p1 , p7 , and p9 in HIFα degradation , at different oxygen levels . The results are given in Table 3 . Note that [O2] = 0 . 1 and [O2] = 1 . 0 represent typical low and high oxygen levels according to [1] . One sees that the pathway p9 is always much less important than the other two as far as HIFα degradation is concerned . The majority of HIFα gets degraded via either p1 at low oxygen or p7 at high oxygen . The comparison of hypoxia response network and heat shock response network [30] as in Table 4 shows the similarity between these two networks with respect to the issue of affinity . The huge difference in the affinity can clearly separate the underlying pathways and assign different functions to them . This is also the basis for the Goldbeter-Koshland model [31] . We tested our method to all three parameter sets ( k-sets 1 , 2 , and 3 ) in [1] , and find that the analytical results are almost identical with those of the direct simulations of the entire network , which strongly validates our approximation . For the rest of the paper thereafter , we only report numerical results for k-set 1 . The EPA method gives us a starting point from which to analyze our reaction network in greater and more revealing detail . The verification of our hypothesis implies that the pathways associated with p9 could be neglected in the first place . The following equations describe the combination p1 , p4 , p6 , and p7 , which constitute the oxygen sensing mechanism: Note the differences between Equations 1 , 6 , and 8 , and Equations 14 , 15 , and 17 . The p9 related elements have been omitted due to their smallness . By setting the left-hand sides of the above equations to zero , one obtains the steady state equations: The total amount of PHD is conserved: PHD is either in the form of PHD ( S12 ) or HIFα:PHD ( S13 ) . This implies where is the initial concentration of PHD . By some derivation , the following equation is obtained: where a = k2k12 , b = b1 + b2[O2] , c = c1 +c2[O2] , where Since c < 0 , Equation 25 has one and only one reasonable root Note that none of the species and reactions in the degradation box is present in Equation 26 , which indicates that the components in the degradation box are not responsible for the sharp response curve . Once is determined , the analysis of the remaining network ( p2 , p3 , p5 , p6 , p8 , p9 , and p10 ) becomes straightforward , and the results are given in the section “Additional results . ” In fact , these results can be further simplified ( see Equations 30 and 43 for and ) . Figure 7A and 7B shows the steady-state values of [HIFα] ( ) and [HIFα:ARNT:HRE] ( ) at different oxygen values . The red lines depict the simulation results obtained by the numerical integration of the ODE system ( 1–13 ) until the steady state is reached . The black lines depict the analytical solutions that are obtained by the algebraic Equations 26 and 56 . To better determine the critical point of pathway switching , we calculate ∂ /∂[O2] and ∂ /∂[O2] . The results are shown in Figure 7C and 7D . One sees that both ∂ /∂[O2] and ∂ /∂[O2] change abruptly in a very narrow region of [O2] , with the rest of the values almost zero . One observes that the critical point is about [O2]c = 0 . 65 . We thus show that the sharpness of the response curve can be determined analytically , instead of exhaustively enumerating [O2] values combined with time-consuming numerical integration of a large number of ODEs at each [O2] value .
EPA is a powerful , yet simple tool that can significantly reduce the complexity of the original network and thus make further analytical effort feasible . In this paper , the additional analysis explains precisely the sharp reaction to oxygen of the network as a whole . The clear separation of p7 and p9 indicates their different functions: the pathway p7 and its other associated pathways constitute the sensing of ambient molecular oxygen; in contrast , the pathway p9 and its associated other pathways are responsible for the signal transduction to form the promoters of hypoxia-regulated genes . Most importantly , the simplification allows for a complete explanation of the switch behavior and a clear presentation of the relations between and [O2] , and [O2] , and . The first step below explains the sharp HIFα stabilization . The second step explains the sharp HRE occupancy that is induced by the HIFα stabilization . HIFα stabilization . This involves the dissociation of pathways p1 , p4 , p6 , and p7 from the whole network , due to the fact that the flux through p9 is always small and can be neglected . It reveals a critical value that corresponds to the switching between pathways p1 and p7 . Since an abrupt change often relates to the notion of singularity in mathematics , we proceed to see if a singularity can be found . Under nomoxia , ≈ 0 , and can be neglected in Equation 25 , which yields One immediately finds the singularity For k-set 1 in [1] , one obtains [O2]c = 0 . 64 , which is exactly the critical value found in Figure 7 . When the oxygen level decreases to a value close to [O2]c , the denominator in Equation 27 becomes very small and a can no longer be ignored . Moreover , the term c in Equation 25 can be ignored compared with the large value of . One thus has This explains the linear decrease of versus [O2] increasing in Figure 7A . In summary , one has One can check that k2 can be ignored in the upper branch of Equation 30 due to its smallness , which again demonstrates that the pathway p1 is not important under nomoxia . The very smallness of k2 reflects the importance of p1 under hypoxia , for k2 exists at the denominator of the lower branch of Equation 30 . HRE occupancy . The remaining pathways reveal how HIFα stabilization triggers a sharp increase of HIFα:ARNT:HRE , namely the sigmoid curve of versus [O2] . We conclude that the magnitude of HIFα is crucial for the sharpness of the curve . To show this , we need Equations 47 , 48 , 53 , and 54 . By removing the terms that are negligible , these equations turn into Equation 31 holds because k15 and k16 are far less than k3 and k4 . Equation 33 holds because is far less than and . Equation 34 holds because , , and are far less than , , and . The validity of the above approximations can be easily checked . For example , for [O2] = 0 . 1 , one finds k3 = k4 = 1 . 66 , k15 = k16 = 0 . 005 , = 1 . 19 , = 2 . 69 , = 0 . 11 , = 0 . 89 , = 0 . 23 , = 0 . 0016 , and = 0 . 0005 . From Equations 31–34 , one obtains where α = α1 + α2/ , α1 = + + k6/k5 , and α2 = k4k6/ ( k3k5 ) . has one and only one reasonable solution where x = 2 /α . Taylor expanding , yields No matter what value is , x < 2 /α1 = 0 . 74 , so 1 − x2/2 > 0 . 73 and x4/8 < 0 . 0375 . Therefore ≈1 − x2/2 and Under nomoxia is small , so α ≈ α2/ and 1/α ≈ 0 . From Equation 37 one has which is also small . Under hypoxia , is large , and thus By substituting Equation 39 into Equation 37 , one obtains the important relation between and : where and By substituting Equation 29 into Equation 40 , one obtains where m = /β1 and λ = β2b2/a . It is well-known that Equation 41 represents a sigmoid curve with m controlling the saturation value and λ controlling the sharpness . By ignoring the small term k6/k5 in the expression of α1 , one finds m is a function of and only . One also finds Here The association ( disassociation ) constants k3 , k5 ( k4 , k6 ) exist in the numerator ( denominator ) of the term k3 , k5/ ( k4 , k6 ) , which implies that the higher the affinity , the sharper the response . The third term b2/a is proportional to the HIF level . In summary , our analysis yields One sees that HIFα is the key to triggering the HIFα:ARNT:HRE response . As long as the oxygen level is greater than [O2]c , HIFα is efficiently degraded by the pathway p7 and maintains a very low level , and the HIFα:ARNT:HRE level is also low ( Equation 38 ) . When the oxygen level drops below [O2]c , the system switches to the pathway p1 , and HIFα stabilizes with a large concentration ( b2/a large ) . This triggers the sharp increase of HIFα:ARNT:HRE . The smaller k2 is , the larger b2/a , and the sharper the HRE occupancy response ( see Figure 8 ) . Also , the validity of our analytical approximation is justified by the close resemblance of Figure 8B and 8C . The three major results of Kohn [1] involve HRE occupancy as a function of the oxygen concentration . The dependence of the curve on ARNT , VHL , and PHD are obtained by both simulation ( Figure 9A ) and analysis ( Figure 9B ) and are explained as follows . ARNT dependence . We need only to analyze the pathway p9 . The amount of ARNT does not affect the shape of the response curve or the location of the sharp transition , because p9 is the pathway for HRE expression , while p7 is responsible for the sharpness . HIFα:ARNT would not be generated without ARNT and there would not be expressions of HRE for any level of oxygen . At high oxygen levels , the concentrations are similar because HRE occupancy is low anyway . At low oxygen levels , low levels of ARNT will give low HIFα:ARNT and then low HRE occupancy . VHL dependence . VHL is present in both p7 and p9 . At high oxygen levels , one should analyze p7 because it is the major pathway for HIFα degradation . The VHL source will affect the upstream HIFα . If VHL concentration is low , it cannot degrade HIFα fast , and the system yields high HRE occupancy . At low oxygen levels , p1 is the major pathway for HIFα degradation , which does not depend on VHL . PHD dependence . One interesting property relates to the different locations of the transition . Using the criterion identified by the alternative model without reaction k2 , we can calculate the transition locations as shown in Figure 9B3 for different PHD values . The results are the same as in Figure 9A3 . As a matter of fact , considering the p9 pathway only yields a much simpler analytical solution that is also accurate . This further simplification is due to the fact that the expression of HIFαOH:ARNT:HRE is always negligible . The present model of the hypoxia response network is probably an oversimplified one . Nevertheless , it serves as an important starting point , from both theoretical and experimental perspectives , before a more detailed model can be understood . The present model will be gradually expanded and analyzed , with the input of more quantitative data from future experiments . One advantage of EPA is that the method can easily incorporate mechanistic details as soon as they become available [16] . There are various molecular interactions that can be added to the model . For example , it was demonstrated that HIF influences mitochondrial function by inducing pyruvate dehydrogenase kinase 1 ( PDK1 ) to suppress the tricarboxylic acid ( TCA ) cycle and thus the aerobic respiration . Then the respiration shifts to be anaerobic , whereby the oxygen resource can be preserved to promote cell survival under hypoxic environment [32–34] . Another subject we are interested in is the inclusion of ROS in the network . It has been established that ROS affects HIFα degradation through Fe2+ [7–10] . The direct hydroxylation of HIFα by oxygen requires Fe2+ . Under hypoxia conditions , however , ROS increases dramatically and consistently removes Fe2+ via oxidation to Fe3+ . Together with the shortage of oxygen , this makes the HIFα degradation through the VHL pathway even more difficult . Consequently , the transition would be faster and sharper . Analysis should focus on the explanation of the coexistence of two oxygen sensing components , a matter that does not appear to be settled as yet . To obtain the dynamical response when the oxgen changes continuously in time , Equations 1–13 ( the full model ) are integrated . Figure 10 shows the temporal changes of [HIFα] and [HIFα:ARNT:HRE] as responses to the oxygen decreasing from 1 . 0 to 0 . 1 with different rates . One sees that [S3] and [S7] increase prominently only after [O2] decreases below [O2]c . The faster [O2] decreases , the more rapid the responses are . In particular , when [O2] abruptly jumps from 1 . 0 to 0 . 1 , the responses ensue promptly . However , it is worth noting that one cannot tell practically how fast the responses are because the time scale is unknown . Indeed , the model is dimensionless and no units are given . Nevertheless , the sharp curves illustrate that the responses are very sensitive to the oxygen concentration and imply that the system can provide a timely response under hypoxia . Physiologists have long been puzzled by the ceaseless HIFα cycle , characterized by both abundant generation and efficient degradation , which seems to be a highly wasteful process . Our analysis provides a reasonable explanation . To deal with a sudden environmental change from nomoxia to hypoxia , an organism must respond in time to trigger the genes necessary for adapting to the new environment . To achieve such a sharp response , a high HIFα generation potential is necessary . Since the hypoxia conditions are rare , an efficient degradation pathway has to be designed to maintain a low HIFα under nomoxia . The HIFα cycle is indeed uneconomic , but it appears useful in helping the cell respond to sudden , unpredictable changes in its environment . In summary , we have obtained an accurate analytical solution to the hypoxia response network and have provided a complete explanation of the switch-like behavior first observed and modeled in [1] . The first step of our analysis applied the EPA technique to a reduced , yet complete system that resulted in exposing ten independent pathways , allowing us to focus on analyzing the pathways relevant to HRE occupancy . The analysis showed that the sharp response of HRE occupancy is due to the switch between the pathway p7 ( p1 ) that degrades HIFα under nomoxia ( hypoxia ) .
The remaining network can be solved analytically with HIFα ( ) already determined . The ODE description is as follows: Together with the three constraints of , , and , the steady state equations are expressed as follows: where and have already been determined from the analysis of p7 . By some reasoning , a quartic equation is derived , from which can be obtained . and can then be expressed as functions of : where
Protein accession numbers as listed in Table 1 are from http://www . ncbi . nlm . nih . gov/entrez/query . fcgi ? DB=protein .
|
A complex biomolecular network utilizes different pathways to perform different functions . However , the interactions within the network are typically so complicated that the pathway structure is usually hidden . By some mathematical techniques , the pathways can be identified and possibly decoupled , whereby the insightful details of the network can be exposed . As an example , we study in this paper the hypoxia response network that manifests a dramatic switch-like behavior for certain sets of rate constants: a slight change of the oxygen concentration close to a critical value will lead to distinct reaction patterns . By a technique called extreme pathway analysis , the network is decoupled into three major and some minor pathways . Flux distribution among these pathways can thus be measured by integrating the ordinary differential equations for any given set of rate constants . For the sets of rate constants where the switch-like behavior is observed , we found that such a behavior is due to the switching of flux between two of the three major pathways .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods",
"Supporting",
"Information"
] |
[
"computational",
"biology",
"biophysics",
"animals"
] |
2007
|
Pathway Switching Explains the Sharp Response Characteristic of Hypoxia Response Network
|
Myxococcus xanthus cells self-organize into periodic bands of traveling waves , termed ripples , during multicellular fruiting body development and predation on other bacteria . To investigate the mechanistic basis of rippling behavior and its physiological role during predation by this Gram-negative soil bacterium , we have used an approach that combines mathematical modeling with experimental observations . Specifically , we developed an agent-based model ( ABM ) to simulate rippling behavior that employs a new signaling mechanism to trigger cellular reversals . The ABM has demonstrated that three ingredients are sufficient to generate rippling behavior: ( i ) side-to-side signaling between two cells that causes one of the cells to reverse , ( ii ) a minimal refractory time period after each reversal during which cells cannot reverse again , and ( iii ) physical interactions that cause the cells to locally align . To explain why rippling behavior appears as a consequence of the presence of prey , we postulate that prey-associated macromolecules indirectly induce ripples by stimulating side-to-side contact-mediated signaling . In parallel to the simulations , M . xanthus predatory rippling behavior was experimentally observed and analyzed using time-lapse microscopy . A formalized relationship between the wavelength , reversal time , and cell velocity has been predicted by the simulations and confirmed by the experimental data . Furthermore , the results suggest that the physiological role of rippling behavior during M . xanthus predation is to increase the rate of spreading over prey cells due to increased side-to-side contact-mediated signaling and to allow predatory cells to remain on the prey longer as a result of more periodic cell motility .
Spatial self-organization of developing cells , which results the formation of complex dynamic structures , remains one of the most intriguing phenomena in modern biology [1]–[4] . Analogous developmental behaviors are observed as bacterial cells form biofilms , which are populations of surface-associated cells enclosed in a self-produced matrix [5] , [6] . The dynamic self-organization in biofilms formed by the soil bacterium Myxococcus xanthus is dependent on the ability of the cells to move on solid surfaces [7] , [8] , while sensing , integrating and responding to a variety of intercellular and environmental cues [9]–[12] . M . xanthus is the preeminent model system for bacterial social development . At high density and under nutrient stress M . xanthus cells execute a complex multicellular developmental program by aggregating into multicellular mounds , termed fruiting bodies , and differentiating into dormant , environmentally resistant myxospores [11] . In addition , these bacteria exhibit complex behaviors when they cooperatively prey on other microorganisms by collectively spreading over the prey cells , producing antibiotics and lytic compounds that kill and decompose their prey [13] , [14] . One of the most intriguing forms of collective dynamics exhibited by M . xanthus is their ability to self-organize into ripples – travelling bands of high-density wave crests [15]–[18] . Although the M . xanthus counter-traveling waves appear to pass through each another , they actually reflect off of one another and are termed “accordion waves” [16] , [18]–[21] . These waves are distinct from the waves originating from Turing instability diffusion-reaction patterns , such as those in chemical systems or observed during development of the other well-studied model social microorganism , the amoeba Dictyostelium discoideum [22]–[24] . The initial studies of the mechanisms underlying M . xanthus rippling motility focused on this behavior during starvation-induced multicellular fruiting body development [16]–[20] , [25]–[27] . The application of mathematical modeling to developmental rippling revealed that the wave properties are consistent with contact-induced reversal signaling [18]–[21] , [28] . This signaling was hypothesized to originate from ‘head-to-head’ collisions of cells moving in opposite directions and to result in an exchange of C-signal that accelerates the reversal clock [16] , [19] , [20] . C-signal is an extracellular protein that controls aggregation and sporulation via contact-dependent pole-to-pole transmission [12] . Developmental aggregation and motility coordination are induced through the C-signal-dependent stimulation of the frz chemotaxis-like system , which includes an unconventional soluble cytoplasmic chemoreceptor homologue FrzCD [12] , [29] , [30] . An opportunity to reevaluate and replace the pole-to-pole collision-mediated model was prompted by a new report of FrzCD protein clusters that appear to transiently align and stimulate reversals in cells making side-to-side contact [8] and by the recent discovery that more robust rippling occurs during predation [13] , [15] . In this paper we have investigated predatory rippling behavior with a combination of mathematical modeling and experimentation . We have constructed a mathematical model that faithfully reproduces the travelling wave behavior by adapting the recently proposed reversal-inducing side-to-side contact-mediated signaling model [8] and incorporating the properties of the patterns resulting from these interactions .
To model collective cell behavior we needed a modeling formalism that would allow us to connect the motility of individual cells , intercellular interactions , and the resulting population patterns . To this end , we employed an agent-based model ( ABM ) approach [19] , [31]–[33] . Individual cells are represented as agents that move and interact according to the rules and equations that correspond to experimental observations . Unlike continuous , cell-density-based approaches , the ABM approach allows cell variability and modular implementation of interactions to be easily incorporated . The details and equations describing our ABM are summarized in the Materials and Methods Section . Here we qualitatively describe the main model ingredients that result in predatory rippling behavior . Each agent is simulated as a self-propelled rod on a 2-D surface . The agents move continuously along their long axis and periodically reverse by switching the polarity of their two ends simultaneously . As in the previous models [19]–[21] , we expected the ripples to emerge as a result of intercellular signaling , which leads to synchronized cellular reversals among the cell population . The side-to-side contact-induced signaling mechanism used here is based on the recent observations by Mauriello et al . [8] , which demonstrated that when cells make transient side-to-side contact , their FrzCD clusters align causing one or both of the cells to reverse . The reversals stimulated by this intercellular signaling would be somewhat similar to the reversals induced by pole-to-pole collisions that were hypothesized to occur due to C-signal exchange during M . xanthus development [18]–[21] . Based on this and other experimental observations , our model incorporates four rules to guide the agents' interactions ( see below ) . These rules are converted to mathematical equations that describe rippling motility ( see the Materials and Methods Section ) . The first three rules are sufficient for the model to produce rippling behavior ( Figure 1 , top row; Video S1 ) . Starting from a uniform aligned population of agents ( 0 hrs ) , the model results in their self-organization into periodic traveling bands ( ripples ) within about 3 hrs . As in previous models [18]–[21] , [25]–[27] , the ripples emerge from the synchronized reversals . However , this model , which is based on a side-to-side contact-mediated signaling mechanism , appears to be more robust than the previous models that utilized pole-to-pole collision-mediated signaling ( Figure S2 ) . Rule ( iv ) is not necessary for rippling , but it allows the model to reflect the cell reversal behavior exhibited at low densities when cell contacts are rare , and it does not significantly change the high-density motility patterns studies here . The mean value of the native reversal period is chosen to be about 8 min ( Figure S3 A ) to achieve agreement with experimental observations by us here and others [11] . Within the framework of the proposed model , each rule ( i ) – ( iii ) is necessary to generate rippling behavior . Specifically , rule ( i ) is necessary because eleminating intercellular reversal-generated signaling abolishes rippling motility ( data not shown ) and eliminating the assumptions that signaling occurs only between counter-moving agents has the same effect ( Figure S4 A and B ) . It is noteworthy that rippling motility is robust to the minimal overlap between agents that is required for them to engage in side-to-side signaling ( Figure S5 ) . Hereafter , an arbitrary value of 50% as a minimal overlap threshold is assumed in all simulations . Moreover , Figure S4 C vs D show that rippling motility occurs regardless of whether each signaling event is bidirectional ( when cell #1 signals to cell #2 , cell #2 also signals to cell #1 ) or unidirectional ( cell #1 signaling to cell #2 and cell #2 signaling to cell #1 are independent events ) . In our simulations we use unidirectional signaling assumptions for the reasons explained below . The refractory period ( rule ii ) is also required for ripples , as reducing it to a very short duration leads to the dissapearance of the waves ( Figure S4 E and F ) . In our simulations , the refractory period is a stochastic quantity with a mean value of 2 . 7 min and standard deviation of 0 . 7 min ( Figure S3 B ) . Side-to-side signaling and rippling motility can only occur in a locally aligned cell population , and thus , physical interaction aligning cells , rule ( iii ) , is necesary to maintain the cells' long axes approximately parallel . Since in our simulations the rules ( i ) – ( iii ) induce rippling motility , we addressed the question of which rule is modulated to ensure that rippling motility is observed only when prey cells or the macromolecules associated with their lysis are present . The initiation and maintance of ripples seems to depend on the probability of reversal-inducing signaling events ( Figure S6 ) , which must exceed a threshhold value of ∼5–10% . If the probability is below 5% , then the ripples will not form and the agents will remain uniformly distributed on the 2-D surface . When the signaling probability exceeds the threshold value , the uniform distribution becomes unstable and the agents self-organize into ripples . Therefore , we hypothesize that the presence of prey-associated macromolecules indirectly stimulates rippling by increasing the probability that side-to-side contact generates successive signaling events ( reversals ) . Although the biochemical mechanism of this induction is unknown , various macromolecular substrates , such as peptidoglycan , bovine serum albumin , and salmon testes chromosomal DNA , have been shown to induce rippling motility [13] , [15] . Thus , we predict that the presence of these substrates should increase the probability of reversal-inducing signaling . Although our experimental arrangement does not allow direct testing of this prediction , we can quantitatively compare the emergent properities of the rippling patterns in the model and in the experiments . It should be noted that the experiments demonstrating side-to-side signaling were preformed in the absence of prey cells or prey-associated macromolecules [8] . However , the results reported by Mauriello et al . [8] are consistent with a low probability of side-to-side signaling and the assumption that signaling is unidirectional . This is because in their observations only one of the cells engaged in side-to-side contact signaling reverses its gliding direction [8] ( see also Figure S7 ) . If the probability of signaling is low , it is unlikely that two signaling events will occur simultaneously . Furthermore , once one of the cell reverses , both cells will then be moving in the same direction and as a result , they will no longer be capable of signaling one another . To test the modeling predictions experimentally , we observed cell motility on a solid nutrient agar surface in the presence of prey cells . The ripples were observed with fluorescence and differential interference contrast ( DIC ) time-lapse microscopy , allowing us to track cell density changes and the motility of a small percentage ( 0 . 5% ) of GFP expressing cells in a wild-type population ( see Materials and Methods section and Video S2 ) . These images allowed us to calculate the global properties of the ripples: wavelength ( distance from one wave crest to the next ) and wave-crest width , and at the same time to measure the behavioral properties of individual cells: coordinates , velocity , reversal period , and the time/position of cellular reversals . These data provided crucial input into the model and allowed us to test our modeling predictions . It is clear that the experimental ripple patterns appear very similar to those produced in the simulation ( Figure 1 ) . To compare the timing of wave initiation between the mathematical model and the experimental results , the time point when M . xanthus cells fully cover the prey in the field of view was chosen as the starting time ( 0 hrs in Figure 1; Video S3 ) . Snapshot images at 0 , 1 , 3 and 5 hrs were selected to show the process of ripple formation in both systems . The experimental process of wave initiation appears to follow the same dynamics as the simulations . Initially , the cells homogeneously cover the field of view and the cells align as they cover the prey . During the first 3 hrs the reversals of individual cells become synchronous and result in the formation of ripples . By 5 hrs the ripples are pronounced and are easily discernible . These results indicate that the ABM is capable of qualitatively reproducing the dynamics of rippling motility observed under our experimental conditions . Interestingly , waves generated with the ABM appear somewhat more pronounced than experimentally observed ripples , which have a smaller cell density gradient from crest to trough . This observation suggests that not all the cells in the biofilm participate in rippling behavior . To compare the rippling patterns produced by the ABM to those of the experiments , we quantitatively characterized the ripples and related their patterns to the behavior of individual cells . Previous models of rippling motility [20] , [21] proposed a simple equation , which relates wavelength ( λ ) , individual agent speed ( v ) , and agent reversal period ( τ ) : ( 1 ) This equation indicates that cells in two colliding crests ( relative speed ) reverse their directions every time the crests are superimposed . This prediction was confirmed by both the ABM and experimental results of developing cells [19] . However , our analysis of the measurements by Berleman et al . [13] , showed that wavelengths of their predatory ripples were ∼50% larger than those predicted by Eq . ( 1 ) . Using their experimental values of v = 3 µm/min and τ = 8 minutes , the calculated λ should be 48 µm , however their observed λ was ∼70 µm . To determine if the wavelength relationship , Eq . ( 1 ) , works for our new ABM of rippling motility , two sets of simulations were conducted . First , the agent speed was fixed at 6 µm/min , while the spontaneous reversal period was varied between 5 min and 30 min ( corresponding to the variation between 3 and 12 min of an actual average reversal period , which is smaller due to early reversals triggered by side-to-side contact signaling; Figure 2A , solid line ) . Second , the spontaneous reversal period was fixed at a value corresponding to an average reversal period of approximately 6 . 6 min and the cell speed was varied between 2 µm/min and 12 µm/min ( Figure 2B , solid line ) . These fixed values correspond to the experimental cell motility parameters . As shown in Figure 2A and 2B , the wavelength ( λ ) scales linearly with agent speed ( v ) and average reversal period ( τ ) . However , when no-intercept linear regression was used , regression coefficients of 15 . 2 µm/min for Figure 2A and 16 . 1 min for Figure 2B were obtained . Both values are slightly larger than the predicted coefficients of 2v ( 12 µm/min ) and 2τ ( 13 . 2 min ) , respectively . When we tracked the reversal points of individual agents , we observed that the agent reversals were initiated as soon as the leading edge of each crest came into contact ( Test S1; Figure S6 ) . This indicates that as the agents at the front of each crest reverse , they signal to the other cells in their crests , leading to a “chain-reaction” of signaling and reversals . Given the wave crest width Δ , the cells in each crest only move an average distance of λ−2Δ before reversing again , which results in the average reversal period τ = ( λ−2Δ ) /2v . Thus , we modified our wavelength equation to be: ( 2 ) To test the modified expression in our simulations , we automatically computed the average wave-crest width Δ from the simulation results ( see Text S1 ) and used it to compute the wavelength with Eq . ( 2 ) . The results demonstrate good agreement between the simulated and predicted wavelengths ( Figure 2 A and B , solid vs . dashed line ) . To test the Eq . ( 2 ) prediction experimentally for predatory rippling motility , we tracked 37 GFP-labeled individual cells within ripples for about 2 hr ( or until the cells left the field of view ) . Continuous 1-D wavelet transform of the microscopy images ( see Text S1 ) was used to compute the wavelength and wave-crest width by fitting a Gaussian function to the wave crest calculations . The distributions of average speed and reversal period are shown in Figures 2 C and D; and the ABM-predicted wavelengths are in agreement with the experimentally observed wavelength ( denoted by the stars in Figures 2A and B ) . The prediction of Eq . ( 2 ) is also in good agreement with the data from Berleman et al . [13] . Using their experimentally derived values of v = 3 µm/min , τ = 8 min , Δ∼10–15 µm , the wavelength , λ , is calculated at ∼70–80 µm , which matches their published values . Rippling motility simulated with these parameters is shown in Video S4 . To further test modeling predictions , we attempted to alter rippling wavelengths with changes in agar density and initial prey-cell concentration . We have selected two plates displaying reduced wavelength for detailed analysis and cell tracking . The results show that predictions of Eq . ( 2 ) also hold for these data: a reduced wavelength resulted from a reduction in the cell speed in both movies ( ∼3 µm/min ) and a reduction of the reversal frequency ( ∼4 . 5 min ) in one of the movies . Table S2 summarizes our experimental tests of Eq . ( 2 ) . According to our ABM assumptions and predictions , most of the rippling cells should travel with the wave crest and reverse , essentially as a group , when the leading edges of the two opposing wave crests collide . To test this prediction , we observed reversals of individual cells in the context of wave-crest movement by plotting cell trajectories on the space-time florescence intensity of ripples ( Figure 2E ) . The space-time image illustrates the timing and location of the wave crests ( see the dark gray ridges in Figure 2E ) . By examining trajectories of GFP-labeled cells ( colored lines ) , we observe that the tracked cells travel with the high-density crests and reverse when and where two crests collide . Statistical analysis of the position and timing of cell reversals ( dots ) show that 75 . 0% ( ±2 . 6% ) of all tracked reversals occur during wave crests collisions , matching ABM prediction ( Figure 2F ) . Interestingly , some cells move through a counter-propagating wave crest without reversing and subsequently reverse with the next crest . This “wave-hopping” pattern explains the small peak at ∼12 min ( twice the average reversal time ) in Figure 2D and the more pronounced second peak in the distribution of the average distance travelled per reversal ( Figure S8 E ) .
Myxococcus xanthus cells self-organize into periodic bands of traveling waves , termed ripples , during multicellular fruiting body development and predation on other bacteria . Here we have used an approach that combines mathematical modeling with experimental observations to investigate the mechanistic basis of rippling behavior and its physiological role during predation . The resulting new mathematical model , which is more robust than previous models , is based on the recent observation of Mauriello et al . [8] , that when counter-moving cells come into side-to-side contact , clusters of chemotaxis-like FrzCD receptors within the cells transiently align and thereafter one of the cells reverses . Our model shows that this side-to-side contact-mediated signaling is sufficient to induce rippling self-organization in a locally aligned cell population , assuming that there is a minimal refractory period during which the cells can not reverse again regardless of their signaling state . The existence of the refractory period has also been assumed in our previous model [19] , [20] and this assumption is plausible as reversals are anticipated to require a significant reorganization of the cell-motility machinery [7] , [34] . The existence of a refractory period also naturally follows from the dynamic properties of a negative-feedback oscillator ( Frzilator ) , which was previously hypothesized to regulate cell reversals [35] . Altogether our modeling results suggest that the self-organization of cells into ripples during predation can be explained by the increased efficiency or higher probability of side-to-side signaling induced by the presence of prey macromolecules . This prediction is not tested directly in our experiments , but the emergent properties of simulated waves quantitatively match those in our predation experimental approach . Our model builds on the detailed characterization of M . xanthus predatory rippling behavior by Berleman et al . [15] , which showed that rippling motility occurs during predation on the variety of microorganisms and is induced by the presence of macromolecular substances . However , our model differs from the concept promoted by Berleman et al . [15] that ripples originate solely as an interaction of individual cells with macromolecules without any self-organizing intercellular interactions . In contrast , we propose that ripples result from the self-organization of cells into traveling wave patterns , which result from the intercellular signaling that is stimulated or facilitated by the presence of macromolecules . Indeed , in our experimental approach the macromolecules are likely to be distributed uniformly and their concentration is expected to vary very little during the typical wave period ( ∼10 min ) . Moreover , even if macromolecules induce the periodicity of M . xanthus cell motility as suggested by Berleman et al . [15] , this would not be sufficient to induce ripples because their formation requires temporal and spatial synchronization of cellular behavior that is unattainable without cell-to-cell signaling . Based on the previous modeling of M . xanthus developmental rippling behavior , one is prompted to ask: does the same mechanism control predatory and developmental rippling motility ? Certainly this new model is similar to the previous mathematical models of developmental rippling , as they each consider that self-organization occurs when counter-moving cells interact to induce reversals [19] , [20] . As expected , the new model is in good general agreement with the experimental patterns that were previously observed for developmental rippling motility [16] , [18] . However , our tests reveal that this new side-to-side contact-mediated signaling model is much more robust , in that it can withstand realistic levels of variability in cell speed and reversal times ( Figure S5 left panels ) . Specifically , when the level of randomness in cell motility consistent with the single-cell tracking experiments ( fluctuations of velocity and reversal period over 30% of the mean value ) is used in the pole-to-pole collision-mediated signaling model , the cells do not form ripples ( Figure S5 , bottom right panel ) . It is noteworthy that pole-to-pole signaling can result in more robust waves , if the cells are able to accumulate signals from multiple collisions and if signaling during the refractory period leads to a reduced reversal rate as the Frzilator model predicts [19] . However , for the new side-to-side contact-mediated signaling model , realistic rippling can be observed assuming only that single successful signaling events result in cellular reversals . Furthermore , the experiments of Berleman et al . [13] , [15] provided evidence indicating that developmental rippling occurs as a side effect of cell lysis during aggregation , which suggests that rippling motility is likely to be a response to the released macromolecules . Thus , we propose that our new side-to-side contact-mediated signaling model of rippling describes both predatory and developmental rippling . The new model therefore explains ripples without requiring the pole-to-pole exchange of the starvation-induced C-signal . This may be biologically justified for a number of reasons . First , to date no C-signaling receptor has been identified . Second , localization of CsgA to the cell poles has not been demonstrated directly . Third , the robustness of pole-to-pole signaling-mediated mechanism is questionable as the probability of this type of collision is low . However , as C-signaling mutants fail to display rippling motility [16] , it would be interesting to investigate in future studies how C-signaling affects the FrzCD cluster alignment and whether C-signaling plays a role in predatory rippling . The main hypothesis of this new computational model is that rippling behavior is initiated by side-to-side contact-mediated signaling in the presence of prey cells . This hypothesis cannot be directly tested at this time , since we do not have a complete understanding of the specific biochemical mechanisms involved . However , we can rigorously test the model by comparing the model predictions with experimental data collected by us and others . An important prediction of the model is that M . xanthus cells will reverse more frequently when prey is present . This agrees with our experimental observations ( Table S3 ) and that of Berleman et al . [13] . Moreover , the resulting self-organization of cells into ripples provides various ways to quantitatively and qualitatively compare in silico-generated rippling motility with experimental observations . A second prediction is that if the presence of prey stimulates this side-to-side contact-mediated signaling , then the rippling would only be observed in the regions where signaling is sufficiently probable , i . e . only in the regions covering prey . This is in good agreement with our observations ( Video S6 ) and those of Berleman et al . [13] , [15] . Indeed , our simulations show that the signaling probability can serve as a bifurcation parameter that induces a transition between the homogeneous cell distribution and the formation of ripples ( Figure S6 ) . A third prediction is based on the timescale of rippling self-organization , which can be defined as the time it takes to generate ripples that consist of well-focused wave patterns , starting from an initially homogeneous cell population . Our model predicts that time to be of the order of 3 hrs , which is remarkably consistent with our experimental observations ( Figure 1 ) . The qualitative comparison of the time-lapse dynamics ( Videos S1 vs . S2 , S3 and S6 ) is also in good agreement . Interestingly , the time-scale of rippling origination in the experiments of Berleman et al . [13] is significantly longer ( ∼12 hrs ) . Although it is hard to pinpoint the source of this discrepancy , our model indicates that the cell density and the amount of noise in cell orientation can significantly affect the wave synchronization time . A fourth prediction of the model is based on measuring the rippling wavelengths and correlating them to the parameters of individual cell motility . Our new model predicts a slightly modified relationship ( Eq . ( 2 ) between wavelength , wave-crest width , individual cell speed , and reversal time as compared to the previously established [19] , [20] . This new relationship is confirmed by our simulations and is in excellent agreement with the experimental measurements of wavelength ( Figure 2 A and B , Table S2 ) . The wavelength prediction is also compatible with previously reported measurements [13] and with the observations of Sliusarenko et al . [15] , which show that cells moving in opposite directions tend to inter-penetrate one cell length before a reversal is triggered . Figure S7 shows the sequence of events that occur during two-crest collisions . This cartoon model indicates that once the cells at the front of each crest reverse , they signal to the cells following them , which results in a chain-reaction of signaling and reversal events . This cartoon also illustrates the importance of the refractory period , because once the cells at the front of the crest reverse , it is essential for them to keep signaling to other cells to reverse without reversing themselves . A fifth prediction of the new model is based on tracking the cell reversals and locations of wave-crest collisions in time and space . Just as the model predicts ( Figure 2F ) , the experimental results ( Figure 2E ) indicate that most reversals occur when and where two wave crests collide . Our previous model of developmental rippling motility suggested [28] that periodic travelling waves can ensure a more regular distribution of fruiting-body aggregates at the colony edge , as seen in the submerged culture system of Welch et al . [18] . However , the physiological implications of this observation are unclear as the developmental aggregate distribution can be well organized even without rippling [36] . Furthermore , if rippling motility is predominantly a response to predation , what is its role in these situations ? Berleman et al . [37] proposed two hypotheses . The first , termed the “grinder model” speculates that the movement of the waves of M . xanthus cells during rippling motility causes a physical disruption of the prey colony . The second , termed the “population control model” suggests that waves maximize the prey-predator contact area and push excess predator cells to the edges of the rippling area . Neither of these hypotheses is likely to be correct , based on the biophysics of this environment in which the very-low Reynolds number hydrodynamics will not allow temporary periodic perturbations to affect mixing or transport [38] . Nevertheless , our mathematical model suggests several alternatives for physiological benefits of rippling to predatory cells . These predictions are consistent with the experimental observations reported here and previously . First , the model is in agreement with the observations of Berleman et al . [13] that during the expansion over prey , the presence of side-to-side contact-mediated signaling significantly facilitates the rate of M . xanthus cell spreading ( Figure 3 ) . As a result these cells cover their prey faster . This has obvious physiological benefits in the competitive soil environment . The observation is also consistent with our own experiments . Notably , this result does not require ripples per se , but only reversal-inducing signaling . However , our model indicates that side-to-side contact-mediated signaling is key for rippling self-organization and the other model ingredients can easily be justified by what is known about the biophysics of M . xanthus motility [7] . Furthermore , the increase in spreading also takes advantage of the cell-density gradient of M . xanthus cells that is generated by spreading at the leading edge . It is important to note that the rippling behavior does not require a density gradient of prey cells , as the alternative chemotaxis-based explanation would predict . Second , the model predicts that cells that ripple in the absence of a cell-density gradient ( i . e . when they are behind the leading edge of the swarm or once the prey is fully covered ) , would engage in less noisy and more periodic motion and as a result will have less of a random drift ( Figure 4A ) . This effect would help the predatory cells to remain in the prey area for a longer time and to reduce random movement away from the prey . This prediction was confirmed by the cell-tracking assays ( Figure 4B ) . Notably , this effect requires ripple formation , as the collective interaction of cells in the ripples leads to their synchronization . This effect is analogous to the well-known mathematical phenomena in which a collection of coupled noisy oscillators is less noisy than each oscillator on its own [39] . Third , it is likely that the formation of the ripples increases the cell alignment due to an increase in steric interactions in the denser crests . This prediction agrees with our observations and those of Berleman et al . [13] . However , it is worth noting that the causal relationship between rippling and alignment is not obvious , as ripples also require cell alignment . Therefore , it is likely that there is a positive self-reinforcing feedback loop between the formation of ripples and cell alignment: as cells align , ripples become more pronounced and their crests become more dense leading to further cell alignment . Although the physiological benefit of better alignment is not obvious , it may further enhance the rate of spreading , which contributes to the effects discussed above . Uncovering the mechanistic basis of spatial and temporal multicellular self-organization is a daunting task and a full understanding has not been achieved for even the best-studied model systems . Here , agent-based modeling , time-lapse fluorescence microscopy , and image quantification have been used synergistically to provide new insights into the mechanisms of M . xanthus self-organization into ripples . Our modeling demonstrates that a simple set of ingredients based on experimental observations is sufficient to produce rippling patterns . The subsequent experiments have tested a number of predictions based on the model and have allowed us to refine the model to achieve quantitative agreement with the experimental data . This type of combined approach is essential to further our understanding of self-organization in more complex systems such as development of multicellular organisms .
ABM are widely used to computationally simulate emerging patterns formed by multiple agents . The ABM of M . xanthus rippling presented here is kept simple yet sufficiently flexible to accurately describe the experimentally observed behaviors of M . xanthus cells . The model is an extension of the earlier ABM [19] of M . xanthus self-organization that now incorporates a side-to-side contact-mediated signaling mechanism . In this ABM , each agent represents a cell – a self-propelled rod on a 2-D surface with length of L , width of w , with a center position of ( x ( t ) , y ( t ) ) , and orientation 0≤θ ( t ) ≤2π . Specifically , the agent length and width are constant throughout all simulations , whereas the center position and direction of movement are changed at each time step as the cells move and align . For each simulation , the time is updated by constant increments δt . The simulations are conducted on a fixed 2-D area in which all simulated moving agents are bounded . For most simulations periodic boundary conditions are imposed .
|
Myxococcus xanthus cells collectively move on solid surfaces and reorganize their colonies in response to environmental cues . Under some conditions , cells exhibit an intriguing form of collective motility by self-organizing into bands of travelling alternating-density waves termed ripples . These waves are distinct from the waves originating from Turing instability in diffusion-reaction systems , as these counter-traveling waves do not annihilate but appear to pass through each other . Here we developed a new mathematical model of rippling behavior based on a recently observed contact signaling mechanism – cells that make side-to-side contacts can signal one another to reverse . We hypothesize that this signaling is enhanced by the presence of prey-associated macromolecules and compare modeling predictions with experimentally observed waves generated on E . coli prey cells . The model predicts a modified relationship between the wavelength and individual predatory cell motility parameters and provides a physiological role for rippling during predation . We show that ripples allow predatory cells to increase the rate of their spreading to quickly envelope the prey , and subsequently to decrease their random drift to remain in the prey region for longer . These and other predictions are confirmed by the experimental observations .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"complex",
"systems",
"cell",
"motility",
"mathematics",
"applied",
"mathematics",
"biology",
"computational",
"biology",
"microbiology",
"biophysics",
"microbial",
"growth",
"and",
"development"
] |
2012
|
The Mechanistic Basis of Myxococcus xanthus Rippling Behavior and Its Physiological Role during Predation
|
Chagas' disease is caused by the protozoan parasite Trypanosoma cruzi and affects approximately 10 million people in endemic areas of Mexico and Central and South America . Currently available chemotherapies are limited to two compounds: Nifurtimox and Benznidazole . Both drugs reduce the symptoms of the disease and mortality among infected individuals when used during the acute phase , but their efficacy during the chronic phase ( during which the majority of cases are diagnosed ) remains controversial . Moreover , these drugs have several side effects . The aim of this study was to evaluate the effect of Memantine , an antagonist of the glutamate receptor in the CNS of mammals , on the life cycle of T . cruzi . Memantine exhibited a trypanocidal effect , inhibiting the proliferation of epimastigotes ( IC50 172 . 6 µM ) . Furthermore , this compound interfered with metacyclogenesis ( approximately 30% reduction ) and affected the energy metabolism of the parasite . In addition , Memantine triggered mechanisms that led to the apoptosis-like cell death of epimastigotes , with extracellular exposure of phosphatidylserine , increased production of reactive oxygen species , decreased ATP levels , increased intracellular Ca2+ and morphological changes . Moreover , Memantine interfered with the intracellular cycle of the parasite , specifically the amastigote stage ( IC50 31 µM ) . Interestingly , the stages of the parasite life cycle that require more energy ( epimastigote and amastigote ) were more affected as were the processes of differentiation and cell invasion .
Trypanosoma cruzi is the etiological agent of Chagas' disease , which affects approximately 10 million people living in endemic areas of Mexico and Central and South America , with 28 million people at risk of infection [1] . T . cruzi has a complex life cycle that alternates between a reduviid insect vector and mammalian hosts ( humans among them ) . During its biological cycle , the parasite differentiates several times between infective , non-dividing forms and dividing forms that inefficiently or are unable to infect mammalian cells . Epimastigotes , the replicative form in the insect vector , colonize the digestive tract and differentiate into metacyclic trypomastigotes , the insect-derived infective form , in the terminal midgut . During a blood meal on a mammalian host , the insects defecate and deposit these forms with the feces , which are internalized by the mammalian host and invade cells where they differentiate into the replicative amastigote stage in the cytoplasm . Amastigotes replicate by binary fission until differentiating into mammal-derived trypomastigotes , passing through a transient epimastigote-like stage [2] , [3] . These trypomastigotes induce the lysis of the host cells , bursting into the extracellular milieu where they invade new cells or reach the bloodstream . The parasites disseminate throughout the infected mammal through the blood and can eventually be taken up by a new reduviid insect during a blood meal . In the midgut , the ingested trypomastigotes differentiate into epimastigotes , which replicate , thereby colonizing a new insect vector [3] . The clinical evolution of Chagas' disease in humans can be divided into two phases: acute and chronic . The acute phase is usually asymptomatic with patent parasitemia and non-specific symptoms . The chronic phase is characterized by infrequent tissue parasitism and subpatent parasitemia that persists for the life of the host . Most patients in the chronic phase ( 60–70% ) will never develop clinically apparent disease . However , approximately 30–40% of chronic patients will develop important physiological alterations: the heart is affected , with hypertrophy and dilatation , and furthermore , the digestive tract , mainly the esophagus and large intestine , are affected , with dilatation and the appearance of megaviscera [4]–[6] as reviewed in reference [7] . Chemotherapy relies on two drugs that were discovered approximately 40 years ago: Nifurtimox and Benznidazole . Both drugs are effective for treating the acute phase of the disease . However , their efficacy in treating the chronic phase , when most patients are diagnosed , is controversial [7] . Moreover , drawbacks for both drugs have been reported , such as serious toxic side effects and more recently , the emergence of drug-resistant parasites . These facts underscore the urgent need to intensify the search for new drugs against T . cruzi [7] , [8] . Our group has been exploring drug repositioning strategies , which are being widely employed for the discovery of novel therapeutic strategies to treat tropical diseases [9] , [10] . This strategy seeks new uses for drugs that are already approved for the treatment of diseases in humans . Paveto and colleagues have suggested that T . cruzi epimastigotes have an N-methyl-D-aspartate ( NMDA ) -type L-glutamate receptor that is involved in the control of cytosolic Ca2+ levels , functionally analogous to that reported in neural cells [11] . Moreover , our group characterized a glutamate transporter [12] which is able to bind NMDA , behaving as a glutamate receptor ( unpublished data ) . In addition , analogs of amantadine and Memantine ( 1 , 2 , 3 , 5 , 6 , 7-hexahydro-1 , 5:3 , 7-dimethano-4-benzoxonin-3-yl ) amines with NMDA receptor antagonist activity were also demonstrated to have significant trypanocidal activity against Trypanosoma brucei [13] . These data led us to hypothesize that trypanocidal activities are present in compounds directed against mammalian glutamate receptors . In the present work , we tested the anti-T . cruzi activity of three compounds that have antagonistic effects on NMDA receptors: Amantadine and Memantine , tricyclic amines with low-to-moderate affinity for the NMDA receptor and used for the treatment of Alzheimer's disease [14] , and MK-801 , which is currently being tested in preclinical studies [15] . Memantine , an uncompetitive blocker of continuously overactivated NMDA receptors in neurons , exhibited the highest antiproliferative activity on epimastigotes and a relevant trypanocidal effect against infective forms of T . cruzi . Our experiments show that Memantine mobilizes intracellular Ca2+ and induces apoptosis , which supports the presence of a receptor with similar activity to glutamate NMDA receptors that can be used as drug targets against this parasite .
Memantine was purchased from TOCRIS; MK-801 , MTT ( 3- ( 4 , 5-dimethylthiazol-2-yl ) -2 , 5-diphenyltretazolium bromide ) and a kit for bioluminescent somatic cells were purchased from Sigma-Aldrich ( St . Louis , MO , USA ) . Amplex red , horseradish peroxidase , Fluo-4 AM and annexin V-FITC were purchased from Invitrogen ( Eugene , Oregon , USA ) . Culture medium and fetal calf serum ( FCS ) were purchased from Cultilab ( Campinas , SP , Brazil ) . The Chinese Hamster Ovary cell line ( CHO-K1 ) was cultivated in RPMI medium supplemented with 10% heat-inactivated FCS , 0 . 15% ( w/v ) NaCO3 , 100 units mL−1 penicillin and 100 µg mL−1 streptomycin at 37°C in a humidified atmosphere containing 5% CO2 . T . cruzi CL strain clone 14 epimastigotes [16] were maintained in the exponential growth phase by subculturing every 48 h in Liver Infusion Tryptose ( LIT ) medium supplemented with 10% FCS at 28°C . Trypomastigotes were obtained by infection of CHO-K1 cells with trypomastigotes as described previously [17] . Trypomastigotes were collected from the extracellular medium five or six days after infection . T . cruzi epimastigotes in the exponential growth phase ( 5 . 0–6 . 0×107 cells mL−1 ) were cultured in fresh LIT medium . The cells were treated with different concentrations of drugs or not treated ( negative control ) . A combination of Rotenone ( 60 µM ) and Antimycin ( 0 . 5 µM ) was used as a positive control for inhibition as previously described [5] . The cells ( 2 . 5×106 mL−1 ) were transferred to 96-well culture plates and incubated at 28°C . Cell proliferation was quantified by reading the optical density ( OD ) at 620 nm for eight days . The OD was converted to cell density values ( cells per mL ) using a linear regression equation previously obtained under the same conditions . The concentration of compounds that inhibited 50% of parasite proliferation ( IC50 ) was determined during the exponential growth phase ( five days ) by fitting the data to a typical sigmoidal dose-response curve using OriginPro8 . The compounds were evaluated in quadruplicate in each experiment . Except where otherwise indicated , for experimental purposes epimastigotes ( 1 . 0×106 cells mL−1 ) were cultured in LIT and treated with a concentration corresponding to the IC50 ( 172 . 6 µM ) Memantine or not treated ( control ) . Before conducting the experiments , epimastigotes were washed twice in PBS and resuspended in 50 µl of binding buffer ( 10 mM HEPES , 140 mM NaCl and 2 . 5 mM CaCl2 , pH 7 . 4 ) . The results shown here correspond to three independent experiments . Parasites were treated with Memantine for four days or not treated ( control ) . Annexin V-FITC and propidium iodide were added to the final concentration indicated by the manufacturer . The cells were analyzed by flow cytometry on a Guava cytometer ( General Electric ) . Epimastigote were treated with Memantine for 24 hours , washed and resuspended in PBS ( 5 mM succinate ) . The cells were incubated with 12 µM amplex red and 0 . 05 U mL−1 horseradish peroxidase . Fluorescence was monitored at a λexcitation of 563 nm and a λemission of 587 nm on a Spectra Max M3 fluorometer ( Molecular Devices ) . Calibration was performed using hydrogen peroxide as a standard . Parasites ( 1 . 0×108 cells ) treated Memantine for four days were incubated with 5 µM Fluo-4 AM ( Invitrogen ) for one hour at 28°C . After this period , the cells were washed twice with HEPES-glucose ( 50 mM HEPES , 116 mM NaCl , 5 . 4 mM KCl , 0 . 8 mM MgSO4 , 5 . 5 mM glucose and 2 mM CaCl2 , pH 7 . 4 ) , resuspended in the same buffer and distributed into 96-well plates ( 2 . 5×107 per well ) in triplicate . Readings were performed on a Spectra Max M3 fluorometer using a λexcitation of 490 nm and a λemission of 518 nm [18] . Intracellular ATP levels were measured in treated ( or not ) epimastigote forms . To assess the effect of Memantine on the levels of intracellular ATP , a kit for bioluminescent somatic cells purchased from Sigma-Aldrich was used according to the manufacturer's instructions . Briefly , 50 µl of PBS was added to 100 µl of cellular ATP-releasing reagent and added to 50 µl of parasite suspension containing 5 . 0×106 treated or untreated ( control ) cells mL−1 . The concentration of ATP was determined using a standard curve of different concentrations of ATP . Luminescence was obtained by the reaction between luciferase and the ATP that was released after cell lysis . Light emission levels were measured on a Lumat LB 9507 luminometer at 570 nm . Epimastigotes ( 5 . 0×106 cells mL−1 ) were grown in LIT medium , transferred to Grace's medium [19] and treated or not treated ( control ) with 172 . 6 µM Memantine ( IC50 value ) . On the sixth day , after transfer , the parasites were counted in a Neubauer chamber , and the percentage of metacyclic forms was determined . CHO-K1 cells ( 5 . 0×105 cells mL−1 ) were seeded in 24-well plates in RPMI medium supplemented with FCS ( 10% ) with different concentrations of drugs or not treated ( control ) . Cell viability was evaluated 48 h after the initiation of treatment using the MTT assay [20] . The IC50 was determined by fitting the data to a typical sigmoidal dose-response curve using OriginPro8 . CHO-K1 cells ( 5 . 0×104 per well ) were maintained in 24-well plates in RPMI medium supplemented with 10% FBS and maintained at 37°C . After 24 h , the cells were infected with trypomastigote forms ( 2 . 5×106 per well ) and treated with different concentrations of Memantine ( 50–300 µM ) for four hours . After this period , free parasites and the Memantine were removed . The infected cells were washed twice with PBS . The RPMI medium was replaced , and the plates were incubated at 33°C . Trypomastigotes were collected from the extracellular medium on the fifth day and counted in a Neubauer chamber . CHO-K1 cells ( 5 . 0×104 per well ) were maintained in 24-well plates in RPMI medium supplemented with 10% FBS and maintained at 37°C . After 24 h , the cells were infected with trypomastigote forms ( 2 . 5×106 per well ) for four hours . After this period , free parasites were removed . The infected cells were washed twice with PBS , the RPMI medium was replaced , and the cells were kept in culture in the presence of different concentrations of Memantine ( 5–300 µM ) . The plates were then incubated at 33°C . Trypomastigotes were collected from the extracellular medium on the fifth day and counted in a Neubauer chamber . CHO-K1 cells ( 5 . 0×104 per well ) were maintained in 24-well plates in RPMI medium supplemented with 10% FBS and incubated at 37°C . After 24 h , the cells were infected with trypomastigote forms ( 2 . 5×106 per well ) for four hours . The infected cells were washed twice with PBS , the RPMI medium was replaced , and the cells were treated at different times during invasion , after 24 h ( amastigote stage ) and after 60 h ( epimastigote-like stage ) with 31 µM Memantine ( corresponding to the IC50 value obtained for the treatment of infected cells ) . The plates were incubated at 33°C . Trypomastigotes were collected from the extracellular medium on the fifth day and counted in a Neubauer chamber . One-way ANOVA followed by the Tukey post-test was used for statistical analysis . The T test was used to analyze differences between groups . P<0 . 05 was considered statistically significant .
To investigate the possible presence of targets for mammalian NMDA glutamate receptor inhibitors , leading to a trypanocidal activity , Amantadine , Memantine and MK-801 were evaluated . A preliminary screening for the ability of these compounds to inhibit epimastigote growth was performed . T . cruzi epimastigotes were cultured in LIT medium with different concentrations of the selected drugs or no drug ( control ) . Amantadine , Memantine and MK-801 produced a dose-dependent inhibition of epimastigote growth at 28°C and pH 7 . 5 , the optimal growth conditions for these cells . The observed growth differences between the treated cells and the control were statistically significant ( p<0 . 05 ) , and the IC50 was determined to be 172 . 6 µM for Memantine , 300 µM for MK-801 and 451 . 2 µM for Amantadine ( Figure 1A , 1B and 1C , respectively ) . In spite of being a relatively high IC50 when compared to that obtained herein for Benznidazole , ( which resulted to be 7 µM , see Figure S1 ) , the fact that Memantine is considered a safe drug for humans ( few side effects have been reported ) at relatively high doses ( up to 20 mg/kg day ) , together with the facts that is commercially available and is inexpensive , led us to choice it for further study by investigating its effects on the biological processes of T . cruzi . Programmed cell death is characterized by morphological and biochemical changes . A major change observed in cells undergoing PCD is exposure of phosphatidylserine on the extracellular face of the cytoplasmic membrane . Treated parasites were incubated with annexin V-FITC to assess external exposure of phosphatidylserine ( feature of PCD ) and propidium iodide to assess the possible rupture of the parasite membrane ( feature of necrosis ) , and were further evaluated by flow cytometry . As shown ( Figure 2A ) , untreated parasites ( control ) showed 10% positivity for phosphatidylserine exposure , whereas the parasites treated with Memantine showed 42% positivity ( Figure 2B ) . Another type of necrotic process was excluded because the maintenance of parasite membrane integrity was confirmed by the absence of propidium iodide staining ( Figure 2C ) . To confirm that Memantine induces apoptosis in epimastigotes , hallmarks for this process in trypanosomatids , such as an increase in reactive oxygen species ( ROS ) , decreased ATP levels , increased intracellular Ca2+ levels and cell shrinkage [21]–[23] , were explored . To evaluate the production of H2O2 in parasites treated with Memantine , epimastigote forms were treated with 172 . 6 µM Memantine ( concentration corresponding to the IC50 value ) . After treatment for 24 hours , the parasites were incubated with amplex red and horseradish peroxidase . As observed , treated parasites produced a slightly increased amount of H2O2 than untreated parasites ( Figure 3A ) . To determine intracellular concentrations of Ca2+ , epimastigote forms were incubated with Memantine ( 172 . 6 µM ) or no drug ( control ) for four days . After treatment , the parasites were incubated with Fluo-4 and analyzed by fluorometry . Treated parasites exhibited higher intracellular Ca2+ concentrations compared with untreated parasites ( Figure 3B ) . The levels of intracellular ATP in treated and untreated cells were also determined using a bioluminescence assay . Intracellular ATP levels decreased in the treated parasites compared with the control ( untreated parasites ) ( Figure 3C ) , indicating that the energy metabolism of the parasite is affected by the drug . Finally , we evaluated potential morphological changes in treated parasites compared with the control ( Figure 4 ) . Epimastigotes treated with Memantine exhibit dramatic changes in morphology ( Figure 4C–D ) , presenting a characteristic rounded shape corresponding to shrinkage , a feature that is also described for apoptotic cells including trypanosomatids [23] , [24] . These changes were reflected by changes on the values obtained for the forward and side light scattering ( Table 1 ) . Because Memantine produced apoptotic activity in epimastigotes , we evaluated whether the drug could interfere with parasite differentiation . Metacyclogenesis is a well-characterized process in T . cruzi that involves transient modulation of Ca2+ levels and is dependent upon the parasite's metabolic status [25] , both of which were affected by Memantine . On this basis , we evaluated the effect of Memantine on metacyclogenesis . Memantine-treated parasites sustained a 30% decrease in the number of metacyclic forms compared with the control ( parasites without treatment ) ( Figure 5 ) . To evaluate the effect of treatment on the intracellular forms of the parasites , we first evaluated the toxicity of Memantine for mammalian CHO-K1 cells by MTT assay to avoid cytotoxic doses . Memantine was well tolerated by CHO-K1 cells , with an IC50 of 624 . 5±46 µM ( Figure 6A ) . Based on this result , we evaluated the effect of Memantine on parasite infection using concentrations up to 0 . 4 mM ( below the IC50 for CHO-K1 cells ) . To verify the effect of the drug on trypomastigote infectivity , CHO-K1 cells were infected and treated with different concentrations of Memantine ( ranging from 50 to 400 µM ) or not treated ( control ) . The parasites were treated for four hours ( the interval corresponding to the process of cell invasion ) . At all concentrations , a significant decrease in the number of trypomastigotes released from the lysed treated cells on the 5th day after infection was observed compared with the control , indicating that Memantine interferes with the infection process , and the IC50 under these conditions was determined to be 206 . 3 µM ( Figure 6B ) . We also evaluated the effect of treatment after invasion of the mammalian cells by T . cruzi . All treatments produced a significant reduction in trypomastigote bursting on the 5th day after infection compared with the control ( Figure 6C ) . This result suggests that Memantine also interferes with processes involved in the intracellular cycle . Under these conditions , the IC50 value was 31 µM , less than 20 times the IC50 for CHO-K1 cells ( selectivity index: 20 . 13 ) . Given the effects of treatment of infected cells throughout the entire infection cycle , we determined which stages of the intracellular cycle ( trypomastigote , amastigote or epimastigote-like ) are more susceptible to treatment with Memantine . To explore this question , we took advantage of the fact that the CL14 strain produces a synchronic infection in CHO-K1 cells as previously reported [17] . In this experiment , 31 µM Memantine ( concentration corresponding to the IC50 value when applied throughout the infection ) was added to the infected cultures at different times: period of infection ( four hours ) , between 24 and 60 hours post-infection ( when the parasites are in the host-cells cytoplasm , as amastigotes ) and between 60 and 96 hours post-infection ( when most of the intracellular parasite population is at the epimastigote-like stage and differentiating into trypomastigotes ) . The stage most susceptible to treatment was the amastigote stage ( Figure 6D ) , with a 35% decrease in the number of egressed trypomastigotes compared with the control .
The discovery of novel drugs for neglected diseases is a necessity for the development of more efficient chemotherapies . However , some alternative strategies should be followed in parallel to accelerate the process of optimizing the treatment of these diseases . In this sense , the search for new therapeutic uses ( in this case , against T . cruzi ) of well-known drugs already in use for humans ( such as Memantine ) may help to reduce time- and resource-consuming steps because parameters for their application in humans ( such as pharmacokinetics , toxicity , maximum tolerable doses and interactions with other drugs ) are already well characterized [10] , [26] . Drug repositioning was the main objective of the present work . The uncompetitive NMDA receptor antagonists Amantadine , Memantine , and MK-801 , which are described in the pharmacopeia as antagonists of NMDA glutamate receptors , exhibited trypanocidal activity . These receptors have not been described in T . cruzi at the molecular level , although evidence of their existence in T . cruzi has been reported [11] . All three evaluated drugs produced a dose-dependent inhibition of proliferation and death in T . cruzi epimastigotes . Interestingly , Amantadine and Memantine , which share their basic structure consisting in a tricyclic amine ( Figure 7 ) , were the less and the more effective antagonists , respectively . The presence of two methyl groups in Memantine , which are absent in Amantadine , diminished the IC50 of the first with respect to the second by a factor of 2 . 5 , showing that little modifications on the leader structure can result in an optimized drug . To investigate the mechanism of death , several parameters were evaluated . First , the integrity of the cytoplasmic membrane and the exposure of phosphatidylserine on the extracellular face were evaluated and strongly suggested PCD with the characteristics of apoptosis . This type of PCD has been described for unicellular protists , including T . cruzi , Leishmania and Plasmodium [27]–[30] . Similar to metazoans , apoptosis is triggered by changes in mitochondrial function . The role of mitochondria in different PCD processes including apoptosis is well characterized [21]–[23] . The production of ROS together with diminished intracellular ATP levels suggest this organelle as a main actor in gating this process . Second , increased intracellular Ca2+ levels and morphological changes were consistent with this cell death mechanism . Taken together , these results demonstrate that Memantine triggers PCD with characteristics of apoptosis . Given that Memantine alters epimastigote physiology , we were interested in determining whether in addition to PCD , this drug may also interfere with differentiation into metacyclic trypomastigotes ( metacyclogenesis ) . This process normally occurs in the terminal midgut of the insect vector . It is worth noting that differentiation requires initial metabolic stress conditions and is mainly energetically supported by amino acids present in reduviid urine and feces , such as proline , aspartate and glutamate [31] . These amino acids allow the parasite to reestablish the intracellular ATP levels required to energize metacyclogenesis [32] . Because Memantine reduces parasite ATP levels , we propose that the inhibition of metacyclogenesis occurs as a result of low ATP levels . To evaluate Memantine as a trypanocidal of interest for developing new treatments against T . cruzi infection , its effect throughout the parasite life cycle in mammalian cells was evaluated . Memantine affected the infectivity of trypomastigote forms , which resulted in a reduced number of trypomastigotes bursted from infected host cells . In addition , the amastigote stage was shown to be the most sensitive stage among those infecting the mammalian cells . This is particularly interesting because amastigotes are the forms involved in maintenance of the chronic phase of infection . Taken together , these results reveal promising prospects for a new use for Memantine , a drug that is already approved for use in humans , as an anti-T . cruzi drug . Preclinical studies are underway to support this proposal .
|
Trypanosoma cruzi is a parasite transmitted to mammal hosts by insect vectors known as kissing bugs . This species can result pathogenic for humans , causing Chagas' disease in the Americas . Its treatment relies on two drugs discovered more than 40 years ago . Besides their toxicity , a main drawback of these drugs is the fact that they are highly efficient only during the acute phase of the infection . But due to the lack of specific symptoms , the acute phase of the infection is largely not diagnosed . In fact , most of patients are diagnosed in the chronic phase , where the treatments are not satisfactory . In view of that , it is urgent to look for new drugs with low toxicity and able to kill the parasite in chronic patients . On the basis of previous finding , we looked for drugs against glutamate recognizing surface molecules , keeping special attention on those that are already in use in humans for other purposes ( this strategy is called drug repositioning , and allow to save time and money in clinical trials: several parameters such as toxicity , pharmacokinetics , side effects in humans are already known ) . Here we report that Memantine , a NMDA glutamate receptors antagonist already in use to treat Alzheimer's disease , presents interesting perspectives as a trypanocidal drug .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"parastic",
"protozoans",
"protozoology",
"biology",
"microbiology",
"parasitology"
] |
2014
|
Memantine, an Antagonist of the NMDA Glutamate Receptor, Affects Cell Proliferation, Differentiation and the Intracellular Cycle and Induces Apoptosis in Trypanosoma cruzi
|
It is now widely recognized that robustness is an inherent property of biological systems [1] , [2] , [3] . The contribution of close sequence homologs to genetic robustness against null mutations has been previously demonstrated in simple organisms [4] , [5] . In this paper we investigate in detail the contribution of gene duplicates to back-up against deleterious human mutations . Our analysis demonstrates that the functional compensation by close homologs may play an important role in human genetic disease . Genes with a 90% sequence identity homolog are about 3 times less likely to harbor known disease mutations compared to genes with remote homologs . Moreover , close duplicates affect the phenotypic consequences of deleterious mutations by making a decrease in life expectancy significantly less likely . We also demonstrate that similarity of expression profiles across tissues significantly increases the likelihood of functional compensation by homologs .
The ability of an organism to survive in various environmental conditions indicates robustness to external perturbations . On the other hand , relative insensitivity to harmful genetic mutations represents genetic robustness . Several large scale gene deletion studies demonstrated that organisms exhibit a significant degree of genetic robustness against null mutations [6] . Although these studies have an important caveat that genes without a detectable phenotype may be essential under different growth conditions [7] , [8] , it is clear that genetic robustness is widespread in biological systems [3] . Two distinct mechanisms of genetic robustness have been extensively discussed . Alternative signaling or parallel metabolic pathways illustrate network contributions to genetic robustness [9] . In contrast , a partial functional overlap between sequence paralogs represents the contribution of gene duplicates . The study by Gu et al . [4] demonstrated a significant contribution to functional compensation by duplicate yeast genes . A similar pattern of the functional compensation was also observed in C . elegans [5] . The mechanism of genetic robustness by duplicates was recently investigated by Kafri et al . [10] , who showed that null deletions in yeast are often compensated by over-expression of sequence homologs . The role and magnitude of the paralog contribution to robustness against deleterious human mutations are not currently well understood . While the study by Lopez-Bigas et al . [11] suggested a contribution by highly conserved paralogs , Yue et al . [12] showed recently that disease and all genes have an equal fraction of paralogs . In the present work , we demonstrate the importance of considering the sequence similarity between paralogs for understanding the likelihood and magnitude of functional compensation . We also explore the effects of mRNA co-expression between duplicates on the observed functional back-up . Understanding the mechanisms of genetic robustness will be important for identification and prioritization of medically important human mutations .
We investigated the functional compensation by duplicates using three curated collections of human disease genes . Although we currently do not know the total number of disease genes , more than a thousand genes with known mutations affecting human health have been identified [13] . First , we used the collection of 1003 Swiss-Prot [14] human genes with non-synonymous disease mutations annotated in the OMIM database [13] . Second , we investigated the collection of 1609 human genes from the OMIM Morbid Map annotated to be involved in disease , but not as susceptibility or non-disease . Our third disease gene set , obtained from the study by Jimenez-Sanchez et al . [15] , included a curated collection of 881 human genes and the associated disease phenotypes such as the age of onset and reduction in life expectancy . The considered disease gene sets significantly overlap , i . e . 636 genes are present in all three sets ( see Figure S1 , Supporting Information ) . Without a collection of human genes which are certainly non-disease , we used several large collections of all human genes ( all gene sets ) . We primarily used the comprehensive collection of 20 , 262 human genes from the Ensembl build 35 [16] . As a representative set of well-characterized human genes , we also considered the collection of 12211 human genes from the Swiss-Prot database [14] . To understand the role of gene duplicates in robustness against deleterious human mutations we searched for homologs of the disease and all human genes using protein BLASTP [17] ( see Methods ) . Briefly , for each query sequence its closest human paralog was identified as the non-self hit which can be aligned over more than 80% of the length of both sequences . The sequence hits with an E-value larger than 0 . 001 were not considered ( results are qualitatively insensitive to the gene set used or the cutoffs and parameters applied in the similarity searches , see Table S1–S3 , Supporting Information ) . For the human genes with identified paralogs ( 475 in the disease gene set and 8257 in the all-gene set ) , the distributions of amino acid sequence identities of the closest homologs are significantly different for disease and all-gene sets ( see Figure S2 , Supporting Information ) . The average identity of the closest homolog is 52 . 9% for disease genes and 58 . 3% for all genes ( non-parametric Wilcoxon's test P = 1 . 6*10−7 ) . The observed difference cannot be explained simply by the existence of several large protein families with a small number of known disease genes; after removing sequences with more than one paralog in the human genome , the average identity of the closest homolog is 50 . 0% for disease genes and 54 . 3% for all genes ( P = 2*10−2 ) . Neither can the difference arise due to difficulties in disambiguating allelic variants from close sequence differences in copy number variable genes [18] , [19] . After excluding genes with highly similar paralogs of sequence identity greater than 90% , the average identity of the closest paralog is 51 . 4% for disease genes and 54 . 4% for all genes ( P = 7*10−4 ) . In Figure 1 we show the conditional probability that a human gene will harbor a disease mutation given the amino acid sequence identity of its closest homolog . To calculate the conditional probability ( see Methods ) we assume that , although the total number of human disease genes is not known , the currently available collection of disease genes is unbiased towards sequence identities of the closest homologs . Figure 1 demonstrates that genes with at least 90% sequence identity to their closest homologs are three times less likely to harbor disease mutations compared to genes with remote paralogs . No correlation was observed between the number of disease mutations in a gene ( Spearman's rank correlation rS = −0 . 025 , P = 0 . 6 ) or gene density of disease mutations ( rS = −0 . 036 , P = 0 . 4 ) and the sequence identity of the closest homolog . This suggests that the number of disease mutations identified in genes may be determined primarily by experimental , mutational , or gene history biases [20] , and not affected by the possibility of functional compensation . Similarly , no correlation between deleterious variability and evolutionary distance to murine orthologs was observed in the study by Sunyaev et al . [21] . If close sequence homologs provide functional back-up against medically damaging mutations , it is likely that they also contribute to relaxation of constraints against deleterious human polymorphisms . As was demonstrated by Lynch et al . [22] , most duplicated genes experience a brief period of relaxed selection after duplication . The functional constraints on human genes can be estimated through the normalized ratio of non-synonymous to synonymous single nucleotide polymorphisms ( SNPs ) per site ( Ka/Ks ) [17] , [23] . A small value of the Ka/Ks ratio suggests a higher constraint on a gene , i . e . a smaller fraction of observed non-synonymous polymorphisms . Figure 2 shows the relationship between the average Ka/Ks ratio and sequence identity to the closest homolog ( shown separately for all and validated SNPs from the dbSNP database [24] ) . The Ka/Ks ratio of the validated SNPs is about two times higher for genes with a 90% sequence identity homolog compared to genes with remote homologs . While there are many examples of homologous iso-enzymes providing functional compensation [7] , [25] , this mechanism is less established for other functional classes . To understand the significance of the duplicate compensation among various functional categories we ( applied the approach described in the previous section and ) compared sequence identities of closest paralogs for disease genes and all human genes in the 53 “GO slim” functional classes . Using a false discovery rate of 5% , we found that , in additional to metabolism , the functional category “response to stimulus” showed evidence of statistically significant compensation by duplicates ( see Table 1 and Table S4 , Supporting Information ) ; the “response to stimulus” category contains cytokines , receptors , protein kinases and other proteins involved in signal transduction . Consequently , functional compensation by duplicates is not limited to metabolism and is also significant among other important functional classes . The observed paucity of close homologs for known disease genes could be a consequence of their faster evolution in comparison with all human genes . To investigate this possibility we analyzed Ka and Ka/Ks values calculated using PAML [26] for all 13055 one-to-one human-mouse orthologous pairs from the Ensembl database [27] . Both Ka and Ka/Ks measures for known disease genes are significantly lower than those of all-gene set ( mean/median Ka: disease 0 . 0729/0 . 0833 , all 0 . 0851/0 . 0971 , P = 4*10−2; mean/median Ka/Ks: disease 0 . 119/0 . 105 , all 0 . 137/0 . 113 , P = 1*10−2 . ) . These findings are in agreement with the study by Kondrashov et al . [28] who considered 1273 disease genes and 16580 other human genes . Although the earlier study by Smith and Eyre-Walker [29] reported the opposite pattern ( a higher Ka/Ks ratio for disease genes ) , their results were based on significantly smaller gene sets ( 387 disease and 2024 non-disease genes ) . Consequently , it is unlikely that the elevated sequence similarity between paralogs of non-disease genes is related to their slower rate of evolution . Recently , He et al . demonstrated a lower duplicability of “important” yeast genes ( essential genes and genes with knockout phenotypes ) [30] . To explore the possibility that lower duplicability of disease genes affects our results we followed the approach by He et al . [30] . Based on the Ensembl database [27] we identified singleton human genes ( genes without duplicates in the human genome , see Methods ) with mouse , chicken , and zebrafish orthologs . We then looked at whether the orthologs of singleton human genes have duplicated in the mouse , chicken , and zebrafish genomes ( see Text S1 , Supporting Information ) . The analysis showed that singleton disease genes are as likely to have duplicate orthologs as all human singleton genes ( 9 . 2% of 338 disease singletons and 8 . 5% of 5657 all human singletons , χ2-test P = 0 . 5 . See Figure S3 , Supporting Information ) . Therefore , human disease genes are as likely to retain duplicates in evolution as all human genes . The sequence identity between duplicates influences the phenotypic consequences of gene deletions in yeast [4] . As the sequence identity decreases , null mutations with weak growth phenotypes become less likely and mutations with strong growth phenotypes become more likely . Inspired by this analysis , we decided to investigate if duplicates also affect phenotypic consequences of human disease mutations . For that purpose we used the collection of human disease genes with manually curated phenotypes [15] . While we did not detect a significant correlation between the presence of close duplicates and the age of onset , the population frequency , or the mode of inheritance , we found a significant correlation between the sequence identity to the closest duplicate and the reduction in life expectancy ( Spearman's rank correlation rS = −0 . 21 , P = 2*10−6 , χ2-test , P = 2*10−4 see Figure 3 and Methods ) . Consequently , the functional compensation by close duplicates may protect against “mild” , “moderate” , and “severe” decline in life expectancy . Several known examples illustrate this interesting result . Mutations in red-sensitive opsin gene cause partial colorblindness ( OMIM#303900 ) . Nevertheless , the life expectancy is not seriously affected due to the presence of the green-sensitive opsin gene ( close homolog of the red-sensitive gene ) . Another example involves several homologous iso-enzymes of the human glycogen phosphorylase; the three iso-enzymes are primarily active in muscle , liver , and brain . Although defects in the muscle and liver forms cause glycogen storage disease V ( MIM#232600 ) and VI ( MIM#232700 ) respectively , neither of the defects reduces life expectancy . Because gene duplicates often have different patterns of expression [25] , [31] , [32] , it is likely that the functional compensation depends not only on the sequence similarity , but also on the similarity of their expression profiles across human tissues . We decided to test this hypothesis using the comprehensive expression dataset by Su et al . [33] , which includes expression of 44775 human transcripts in 79 tissues . Initially , we used the absolute values of gene expression in different tissues to calculate the relative expression difference between every gene and its closest sequence homolog . The relative expression difference was defined as ( Exp ( Gene ) −Exp ( Paralog ) / ( 1/2* ( ( Exp ( Gene ) +Exp ( Paralog ) ) . Using this measure we did not find any significant differences between disease and all genes ( P = 0 . 1 ) . It is likely that each gene is expressed primarily in a small number of tissues and the simple averaging of expression values across all tissues will not be informative . Therefore , in order to better reflect the observed expression patterns , we considered a gene to be expressed in a tissue if at least one of the gene transcripts was found to be significantly expressed ( “present call” ) in the tissue by Su et al . [33] . We defined Similarity of Tissue Expression ( STE ) for a gene pair as the ratio of the number of tissues where the two genes are both expressed to the number of tissues where at least one of the genes is expressed; STE is essentially the Jaccard's coefficient of similarity for binary expression patterns . The STE value of one would indicate complete overlap between expression profiles , while values close to zero would indicate poor overlap . Since expression profile similarity and sequence similarity of duplicates tend to be correlated [25] , [31] , [32] , we demonstrated ( see Figure 4 and Table S5 , Supporting Information ) that the STE values are consistently lower for disease gene pairs in different sequence bins; the differences are significant for sequence identity bins from 30% to 80% . We also performed the likelihood ratio test to show that the similarity in tissue expression influences the probability of being a disease gene independently of the sequence identity to the closest homolog ( likelihood ratio test χ2 = 4 . 0 , P<0 . 05 , see Methods ) . Our analysis clearly demonstrates that gene duplicates affect the phenotypic consequences of deleterious human mutations . Several studies suggested possible mechanisms of functional back-up by duplicates [4] , [9] , [10] , [34] . It is likely that similar mechanisms also play a role in human genetic diseases . In some cases duplicates might actively compensate for the mutated homolog , for example by partially carrying the metabolic flux of the mutated gene [25] . In other cases , genes with close duplicates may have smaller functional loads compared to singletons , i . e . genes with duplicates may be essential in a smaller number of environmental conditions [7] . As a result , a disease phenotype is less likely to be observed . We take the view that both of these cases represent functional compensation , although it may be called active compensation in the first case and passive compensation in the second . In our view , the probabilistic approach used in our paper to investigate the likelihood of disease mutations given the sequence identity of the closest homolog can be applied for identification and prioritization of medically relevant mutations . Such prioritization approaches are necessary as large collections of human genetic variation , such as mutations associated with various cancers [35] , [36] and common human polymorphisms [37] , are being generated at an accelerated rate . A probabilistic scheme , similar to the one used in our paper , can be directly applied as a prior in search for causative mutations; the information about homolog expression profiles can be also considered . The development of such probabilistic prioritization schemes is beyond the scope of this paper . Nevertheless , the fact that genes with 70–100% sequence identity homologs are about 2–3 times less likely to harbor disease mutations , and a significant fraction of such genes in the human genome , suggest that duplicate homology information may be important for the prioritization of medically relevant mutations . The collections of disease genes used in our work are incomplete and significantly biased towards Mendelian diseases [15] . When large and reliable datasets of genes responsible for complex diseases become available it will be interesting to investigate whether fundamental differences exist between functional compensation for Mendelian and multi-factorial diseases . In future studies , it will be also important to investigate robustness to deleterious human mutations achieved through various network effects [3] , [9] . Such studies will bring the important biological concept of robustness into the realm of human genetics .
Three sets of human disease genes were used in our study . We obtained a list of 1003 human genes ( 1006 Swiss-Prot entries ) with disease non-synonymous mutations from the Swiss-Prot database [14] ( July 2005; http://expasy . org/cgi-bin/listshumsavar . txt ) . The list of 881 human disease genes ( 923 OMIM entries ) with annotated phenotypes was taken from the study by Jimenez-Sanchez et al . [15] . We also considered another disease set consisting of genes annotated as “disease” , but neither as “susceptibility” nor as “non-disease” in the OMIM Morbid Map [13] . This set included 1609 genes ( 2239 MIM entries ) . Two sets of all human genes were used based on the Ensembl [16] and Swiss-Prot databases . The longest protein isoform of every human gene was obtained from the Ensembl human genome build 35 . We only retained genes annotated as “pep:known” or “pep:CCDS” ( representing genes mapped to human-specific entries of Swiss-Prot , RefSeq , SPTrEMBL or CCDS ) . In total 20 , 262 genes were included . The other all- human gene set consisted of 12 , 211 protein sequences from the Swiss-Prot database . All-against-all BLASTP searches were performed using standard parameters [17] . Sequence homologs were identified as non-self hits with E-value < = 0 . 001 that could be aligned over more than 80% of both the query length and the length of identified sequence . Throughout the manuscript the term “singleton human genes” is used to describe the genes without any sequence homologs which can be identified the BLASTP searches . We obtained H . sapiens to D . rerio , H . sapiens to G . gallus , and H . sapiens to M . musculus orthology information as well as paralogous relationships within D . rerio , G . gallus , and M . musculus from the Ensembl database [27] . Ka and Ka/Ks values of all 1∶1 human-mouse orthologous pairs were calculated using the PAML package and obtained directly from the Ensembl database [27] . The sets of synonymous and non-synonymous human SNPs were obtained from the dbSNP database [24] . These included 87920 SNPs corresponding to 14825 human genes . For each bin of homolog sequence identity , the Ka/Ks ratio was calculated . The proportion of non-synonymous sites ( 0 . 717 ) was calculated from simulation; for each nucleotide in the protein coding region a random transition or transversion mutation was performed at the ratio of 0 . 6/0 . 4 , according to the published estimates in mammals [38] , [39] , [40] , [41] . We used manually curated phenotypes from the study by Jimenez-Sanchez et al . [15] to calculated Spearman's rank correlation between reduction in life expectancy ( ordinal data: none , mild , moderate , and severe ) and sequence identity to the closest homolog . The functional categories of human genes used in our study were based on the annotation by GOA [42]; 53 of GO slims for GOA ( http://www . geneontology . org/GO_slims/goslim_goa . obo ) were considered and Benjamin-Hochberg's algorithm was applied for multiple hypothesis correction . The gene expression profiles in 79 human tissues were obtained from the study by Su et al . [33] . We eliminated probe sets with cross hybridization effects ( as identified by Su et al . ) . In total , we considered expression profiles for 15097 human genes . The expression value of gene G at tissue T was set to 1 if at least one of gene G's transcripts was detected as “Present call” in tissue T based on the Affymetrix detection algorithm ( provided by Su et al . [33] ) . Similarity of Tissue Expression ( STE ) of a gene pair was defined as the Jaccard's coefficient of the binary expression profiles of the two genes , that is , the ratio of the number of tissues where the two genes are both expressed to the number of tissues where at least one of the genes is expressed . We performed the likelihood ratio test to investigate whether the similarity in tissue expression influences the probability of being a disease gene independently of the sequence identity to the closest homolog . The logistic regression was used to model the probability of being a disease gene using the expression and sequence similarities . In the null hypothesis the disease gene probability is determined only by sequence identity of the closest homolog; in the alternative hypothesis the probability is determined by sequence identity and tissue expression similarity of the closest homolog . The probabilities shown in Figure 1 represent conditional probabilities . Specifically , the conditional probability P ( disease|seq_id_homolog ) that a gene is associated with a genetic disease given that it has a closest homolog with a certain sequence identity , was calculated according to the equation:where P ( seq_id_homolog | disease ) is the probability that the closest homolog of a disease gene has a certain sequence identity , P ( seq_id_homolog ) is the probability that a randomly selected human gene ( disease or non-disease ) has a closest homolog with a certain sequence identity , and P ( disease ) is the probability that a random human gene is associated with a genetic disease . Importantly , because P ( disease ) is currently unknown ( as we know only a fraction of all disease genes ) , we estimate P ( disease | seq_id_homolog ) up to a constant by assuming certain P ( disease ) value . For display purposes , we assumed P ( disease ) = 0 . 2 in Figure 1 .
|
Genetic robustness is the ability of an organism to buffer deleterious genetic mutations . It has been previously demonstrated that the functional compensation by duplicates plays an important role in protection against gene deletions in model organisms . Close duplicates often share similar functions , and loss of one paralog may be buffered by others . In the present work we specifically investigate the contribution of gene duplicates to backup against deleterious human mutations . We find that genes with close homologs are significantly less likely to harbor known disease mutations compared to genes with remote homologs . In addition , close duplicates affect the phenotypic consequences of deleterious mutations by making a decrease in life expectancy less likely . Similarity of expression profiles across tissues increases the likelihood of functional compensation by homologs . Taken together , our analysis demonstrates that functional compensation by close duplicates plays an important role in human genetic disease .
|
[
"Abstract",
"Introduction",
"Results/Discussion",
"Methods"
] |
[
"genetics",
"and",
"genomics/bioinformatics",
"computational",
"biology/genomics"
] |
2008
|
Role of Duplicate Genes in Robustness against Deleterious Human Mutations
|
The 24-nucleotides ( nt ) phased secondary small interfering RNA ( phasiRNA ) is a unique class of plant small RNAs abundantly expressed in monocot anthers at early meiosis . Previously , 44 intergenic regions were identified as the loci for longer precursor RNAs of 24-nt phasiRNAs ( 24-PHASs ) in the rice genome . However , the regulatory mechanism that determines spatiotemporal expression of these RNAs has remained elusive . ETERNAL TAPETUM1 ( EAT1 ) is a basic-helix-loop-helix ( bHLH ) transcription factor indispensable for induction of programmed cell death ( PCD ) in postmeiotic anther tapetum , the somatic nursery for pollen production . In this study , EAT1-dependent non-cell-autonomous regulation of male meiosis was evidenced from microscopic observation of the eat1 mutant , in which meiosis with aberrantly decondensed chromosomes was retarded but accomplished somehow , eventually resulting in abortive microspores due to an aberrant tapetal PCD . EAT1 protein accumulated in tapetal-cell nuclei at early meiosis and postmeiotic microspore stages . Meiotic EAT1 promoted transcription of 24-PHAS RNAs at 101 loci , and importantly , also activated DICER-LIKE5 ( DCL5 , previous DCL3b in rice ) mRNA transcription that is required for processing of double-stranded 24-PHASs into 24-nt lengths . From the results of the chromatin-immunoprecipitation and transient expression analyses , another tapetum-expressing bHLH protein , TDR INTERACTING PROTEIN2 ( TIP2 ) , was suggested to be involved in meiotic small-RNA biogenesis . The transient assay also demonstrated that UNDEVELOPED TAPETUM1 ( UDT1 ) /bHLH164 is a potential interacting partner of both EAT1 and TIP2 during early meiosis . This study indicates that EAT1 is one of key regulators triggering meiotic phasiRNA biogenesis in anther tapetum , and that other bHLH proteins , TIP2 and UDT1 , also play some important roles in this process . Spatiotemporal expression control of these bHLH proteins is a clue to orchestrate precise meiosis progression and subsequent pollen production non-cell-autonomously .
Small noncoding RNAs are 20–30 nucleotides ( nt ) long and associate with Argonaute family proteins to serve as guide molecules for RNA silencing in various biological processes , such as cell type specification , cell proliferation , cell death , metabolic control , transposon silencing and antiviral defense [1] . Plant genomes encode precursors of microRNA ( miRNA ) and small interfering RNA ( siRNA ) , as do animal genomes [2] . miRNA is produced from a hairpin structure of a single precursor RNA molecule , and siRNA is derived from a precursor RNA that is either naturally double-stranded or is formed by RNA-dependent RNA polymerases . The third class of animal small RNAs is Piwi-interacting RNA ( piRNA ) . The piRNA is abundantly expressed in the germline and acts in silencing of transposable elements ( TEs ) [3] , massive elimination of paternally derived mRNAs [4] , systemic recognition of self and non-self mRNAs [5 , 6] , and so on . piRNA associates with Piwi family proteins , a distinct subgroup of Argonaute proteins . In contrast , plants have no Piwi family Argonautes [7 , 8] , and consequently lack piRNA species . In place of piRNA , trans-acting siRNA ( tasiRNA ) and phased secondary siRNA ( phasiRNA ) are identified as plant-specific small RNA subgroups . In monocot model plants , rice and maize , phasiRNAs are abundantly expressed in the male reproductive organs , and in this study , the term "phasiRNA" will be used for monocot reproductive phasiRNAs derived from protein-noncoding regions . Both tasiRNA and phasiRNA are produced via miRNA-dependent primary processing , and characterized by phased alignment on both sense and antisense strands in genomic regions . However , they are distinct in several points . First , phasiRNAs are abundantly expressed in developing reproductive organs [9–13] , while 21-nt tasiRNAs are expressed in both vegetative and reproductive phases [14] . Second , phasiRNAs are transcribed from hundreds or thousands of unique , namely nonrepetitive , intergenic regions [9 , 11–13] , while a few tasiRNA-producing ( TAS ) loci are conserved in the plant genome [15–17] . Finally , no phasiRNA targetting a protein-coding gene has been identified , whereas tasiRNAs are complementary to particular genes important for defense and developmental events [14] . In plant reproduction , 24-nt unphased siRNAs or 21-nt epigenetically activated siRNAs ( easiRNAs ) are thought to maintain genome integrity by programmed DNA methylation of TEs [18–20] . The roles of phasiRNAs during plant reproduction largely remain elusive . In rice , a single-stranded PHAS precursor RNA is primarily processed with 22-nt miRNA triggers; miR2118 for 21-PHASs and miR2275 for 24-PHASs [10] . PHAS and TAS RNA members each have one or two conserved complementary sequences to miRNAs , and are cleaved via the one-hit or two-hit processing pathway; the one-hit mode is mediated by the AGO1-miRNA complex for 5'-end cleavage of precursor RNAs [15] to generate the 3' fragment that becomes double-stranded , and the two-hit mode depends on AGO1- or AGO7-miRNA , which potentially associates with both ends and cleaves either end or both [14] . The processed RNA is made double-stranded by RNA DEPENDENT RNA POLYMERASE6 ( RDR6 ) [21] , and chopped into 21- and 24-nt lengths by DICER-LIKE4 ( DCL4 ) and DCL5 ( previous DCL3b in rice ) , respectively [10] . The anther is a four-lobed male reproductive organ in angiosperms . Each anther lobe is composed of central sporogenous cells and four concentric somatic layers; the epidermis , endothecium , middle layer and tapetum , from outward to inward ( Fig 1A ) [22–24] . Sporogenous cells undergo several rounds of mitosis and mature into pollen mother cells ( PMCs ) to prepare for meiosis [22–24] . Maize OUTER CELL LAYER4 ( OCL4 ) , an HD-ZIP IV transcription factor ( TF ) , expressed in the anther epidermis and MALE STERILE23 ( MS23 ) , a basic helix-loop-helix ( bHLH ) TF expressed in the tapetum are required for 21 and 24-nt phasiRNA biogenesis , respectively [12 , 25] . Small RNA-mediated intercellular signaling is proposed in various steps of plant reproduction , for example , between sperm and vegetative cells in the pollen [19 , 26] and between megaspore mother cells and somatic nucellar cells in the ovule [18] . The intercellular movement of reproductive phasiRNAs has been proposed in maize [12 , 13] , while there is yet no decisive evidence . The underlying mechanism to determine the spatiotemporal expression of reproductive phasiRNAs in anthers has largely remained elusive . In this study , we focused on the rice bHLH TFs , because they are key transcriptional regulators for differentiation and development of anther somatic layers . TDR INTERACTING PROTEIN2 ( TIP2 ) /bHLH142 is expressed in several undifferentiated cell layers to form the middle layer and tapetum [27 , 28] . TAPETUM DEGENERATION RETARDATION ( TDR ) /bHLH5 makes a heterodimer with TIP2 to promote tapetal differentiation [29] . ETERNAL TAPETUM1 ( EAT1 ) /bHLH141 , 41% similar to TIP2 , also dimerizes with TDR , and activates transcription of aspartic protease-encoding genes to promote programmed cell death ( PCD ) of postmeiotic tapetal cells [30 , 31] . UNDEVELOPED TAPETUM1 ( UDT1 ) /bHLH164 [32] is expected to function upstream of the regulatory cascade for anther wall development . However , downstream targets of these bHLH TFs are largely unknown . In addition to its role in tapetal PCD , we found that EAT1 is required earlier in tapetal development to support meiosis , while the loss of EAT1 function has little impact on the tapetum morphology . EAT1 shows a bimodal expression at both early meiosis and postmeiosis . Interestingly , EAT1 expressed during early meiosis promoted both transcription and processing of 24-PHAS precursor RNAs to produce 24-nt phasiRNAs in tapetum . This study demonstrates that EAT1 is one of key regulators triggering meiotic phasiRNA biogenesis in anther tapetum , and that other bHLH proteins , TIP2 and UDT1 , also play important roles in this process .
To determine the impact of bHLH proteins in communication between somatic tapetal cells and PMCs in rice anthers , we first performed quantitative reverse-transcription PCR ( qRT-PCR ) of four bHLH genes: UDT1 , TDR , TIP2 and EAT1 , all of which are involved in tapetal cell-fate decision [27–32] . In this study , we separated anther developmental processes into six stages to characterize spatiotemporal expression of these genes ( ST . 1 to ST . 6; Table 1 ) . qRT-PCR of meiotic anthers demonstrated that UDT1 , TDR and TIP2 were expressed as expected from previous reports ( S1 Fig , S1 Data ) . However , EAT1 expression was bimodal , both at early meiosis ( ST . 2 ) and postmeiosis ( ST . 5 ) , whereas it was previously thought to function only in postmeiotic tapetal PCD [30 , 31] . To investigate EAT1 expression during early meiosis , an EAT1pro-EAT1-GFP transcriptional fusion construct ( Fig 1B ) was introduced into male-sterile eat1-4 plants homozygous for a putative null allele with a Tos17-retrotransposon insertion ( S2A–S2F Fig ) . The transgenic plants recovered male fertility ( Fig 1C ) , indicating that the EAT1-GFP protein is functional in planta . EAT1-GFP expression was bimodal at ST . 2 and ST . 5 , as was mRNA expression , and the two expression peaks were clearly separated by the silent ST . 4 ( Fig 1D ) . Transcription of AP25 , an aspartic protease gene required for tapetal PCD initiation [30] , was fully dependent on EAT1 at ST . 5 ( S3 Fig , S1 Data ) , while no AP25 transcript was detected at ST . 2 or ST . 3 . These results confirm that the role of meiotic EAT1 is distinct from its postmeiotic role in tapetal PCD and further suggest that the EAT1 bHLH TF has distinct bHLH partners at these two developmental stages . In wild-type anthers , three concentric layers of somatic-wall cells at ST . 1 become four layered at ST . 2 , and PMCs undergo meiosis at ST . 3 and ST . 4 ( Fig 2A and S4A–S4C Fig ) . During ST . 3-ST . 4 , the middle layer disappears , and during ST . 5-ST . 6 , the tapetal layer degenerates by PCD ( S4D and S4E Fig ) . The eat1-4 mutant phenotype was remarkable in postmeiotic ST . 5 and ST . 6 anthers , in which tapetal cells were unusually degenerated at ST . 5 , concurrent with abortive microspores and male sterility ( S4I and S4J Fig ) . On the other hand , no morphological phenotype was found in earlier stages , ST . 1 to ST . 4 by light microscopy ( Fig 2A and S4F–S4H Fig ) . Degradation of beta-1 , 4-glucan on cell walls of tapetal cells and PMCs occurred normally in eat1-4 anthers at ST . 2-ST . 3 stages ( Fig 2B ) . These observations were largely consistent with previous results [30] . We detected an unreported defect in male meiosis of eat1-4 mutants: PMCs harbor aberrantly decondensed bivalent chromosomes frequently , 74 . 2% at diakinesis ( n = 70 ) and 69 . 4% at metaphase I ( n = 36 ) ( Fig 2C ) . In addition , two out of 27 eat1-4 PMCs at anaphase I harbored lagging chromosomes or chromosomal bridges , which were not found in the wild-type ( n = 42 ) ( Fig 2C ) . Another 5 . 5% eat1-4 PMCs exhibited interphase-like nuclei with fully decondensed chromosomes ( n = 163 ) , in contrast to wild-type PMCs ( n = 192 , Fig 2C ) . In addition , meiotic division timing was retarded in mutant anthers , with asynchronous progression within an anther lobe ( S5 Fig , S1 Data ) . Despite these meiotic defects , male meiosis could complete , but resulting microspores were aborted most likely by the aberrant tapetum , which normally secretes nutrients and exine components required during post-meiotic pollen development ( S4J Fig ) . These results suggest that non-cell-autonomous signaling or some nutrient delivery between somatic tapetal cells and PMCs is mediated by EAT1 during meiosis , in addition to post-meiosis . To identify genes under the control of meiotically expressed EAT1 , we conducted mRNA-seq experiments using whole anther samples and compared the data between wild-type and eat1-4 plants . The data were obtained from three different meiotic stages: premeiosis ( ST . 1 ) , early meiosis ( ST . 2 ) and late meiosis ( ST . 4 ) , each with three biological replicates . 142 , 048 , 793 reads from wild-type and 146 , 928 , 874 reads from eat1-4 anthers ( S1 Table ) in total were mapped to the rice genome . Of all 38 , 311 rice genes , 115 genes were defined to exhibit EAT1-dependent expression , which showed >2-fold greater Fragment per Kilobase per Million ( FPKM ) values in ST . 2 anthers compared to eat1-4 ST . 2 anthers , and also compared to ST . 1 and ST . 4 anthers ( Fig 3A , S2 Table ) . The ontology terms for 7 of 115 genes were enriched in lipid metabolism based on the agriGO algorithm [33] ( S3 Table ) , implying that they function in pollen coat formation [34] . mRNA-seq also identified 6 , 097 regions generating long intergenic noncoding RNAs ( lincRNAs ) , and 248 showed ST . 2-enriched and EAT1-dependent expression ( Fig 3A , S4 Table ) . Next , we conducted small RNA-seq ( sRNA-seq ) to ask whether these lincRNAs are small RNA precursors or not . 52 , 726 , 712 reads of total small RNAs extracted from wild-type and 62 , 364 , 061 from eat1-4 anthers were mapped onto the rice genome ( S1 Table ) . As a result , the 93 lincRNAs were defined as 24-PHAS RNAs , because a large number of 24-nt small RNAs were mapped in a 24-nt phasing manner on the lincRNA loci ( see below for details ) . Of 44 24-PHAS loci previously reported [9 , 10] , 24 were included in the loci identified in this study . Another 8 loci , which were left out of our first selection by their length or overlapping coding genes , generated EAT1-dependent and ST . 2-enriched 24-nt phasiRNAs ( S4 Table ) , while the remaining 12 loci did not . Thus , adding the 8 loci , a total of 101 loci were specified as ST . 2-enriched and EAT1-dependent 24-PHAS loci and analyzed hereafter . Median FPKM values of 24-PHAS transcripts detected at the 101 loci in wild-type ST . 2 anthers were 688-fold and 24-fold higher than those in ST . 1 and ST . 4 anthers , respectively . In addition , the values were 55-fold higher than in eat1-4 anthers at ST . 2 ( Fig 3B and 3C , S4 Table ) . This result reconfirmed the EAT1-dependent and early meiosis-enriched nature of 24-PHAS transcripts . This trend was reproducible in qRT-PCR of five 24-PHASs ( Fig 3D , S1 Data ) . In contrast , most 24-nt RNAs from the corresponding PHAS loci were abundant not only in ST . 2 , but also in ST . 4 anthers ( Fig 3B and 3C , S4 Table ) , implying slower turnover of small RNAs than precursor transcripts . The 101 PHAS loci were unevenly distributed in the genome as reported previously [9] , except for chromosomes 1 and 9 , and many loci formed several clusters on each chromosome ( Fig 4A , S4 Table ) . Sequence comparison by the MEME program [35] demonstrated that 93 out of 101 24-PHAS loci conserved 22-mer sequence complementary to mature miR2275 ( Fig 4B and 4C , S4 Table ) . The miR2275 sites were conserved at the 5'-region in 92 loci ( Fig 4C , S4 Table ) , consistent with previous results that 22-mer miRNA triggers one-hit processing [36 , 37] . The phased pattern tended to start at the 13th position in the 22-mer miR2275 site in most of 24-PHAS loci ( Fig 4D ) . This position corresponded to the cleavage site of the AGO1/miR2275 complex reported previously [10] . Consistent with this , the degradome data from the indica rice variety [38] demonstrated that the cleavage actually occurred at the same position relative to the miR2275 complementarity in 62 of 93 24-PHAS loci ( Fig 4B and 4D , S4 Table ) , and that almost of lincRNAs detected here were the unprocessed , primary 24-PHAS RNAs . Of 24-nt small RNAs mapped on 93 24-PHAS loci , the 77 . 1% reads from wild-type ST . 2 and ST . 4 anthers showed a 24-nt phased pattern which starts from putative AGO1/miR2275 cleavage site ( S6 Fig , S1 Data ) , indicating that 24-nt small RNAs produced from these loci were processed by DCL5 . Most 24-PHAS loci were mapped to unique or low copy regions ( Fig 4C and 4E , S1 Data ) . Only 7 of the so-far reported 15 , 723 TEs showed ST . 2-enriched and EAT1-dependent expression ( Fig 3A right , S2 Table ) . We concluded that meiotic 24-nt phasiRNAs originate from 101 intergenic 24-PHAS loci and that they have a role distinct from TE silencing . Chromatin-immunoprecipitation ( ChIP ) -qPCR analysis was performed to examine EAT1-binding to the upstream cis sequences of two 24-PHAS loci ( chr5-20 and chr6-97 ) using EAT1-GFP-expressing plants . Both sequences included E-box motifs , short CANNTG sequences potentially targeted by bHLH proteins [39] ( Fig 5A ) . The chr5-20-Ebox1 was enriched 5 . 4-fold and the chr6-97-Ebox2 was enriched 6 . 1-fold in ChIP of EAT1-GFP-expressing anthers ( Fig 5B , S1 Data ) , suggesting that EAT1 has a potential to target 24-PHAS loci . The above results prompted the idea that EAT1 activates genes including 24-nt phasiRNA biogenesis-related ( 24-PBR ) genes . Indeed , DCL5 was 2 . 1-fold downregulated in eat1-4 ST . 2 anthers in mRNA-seq analysis ( Fig 3A , S2 Table , S1 Data ) , and this reduction was confirmed by qRT-PCR ( Fig 5C , S1 Data ) . ChIP using EAT1-GFP-expressing anthers and anti-GFP antibody displayed enrichment of the Ebox2 and Ebox3 upstream of DCL5 by 6 . 5- and 2 . 7-fold , respectively ( Fig 5D and 5E , S1 Data ) . In contrast , no EAT1 binding was detectable in two other DCL family genes , DCL3a , responsible for long miRNA production required for cytosine DNA methylation and TE-associating 24-nt siRNA synthesis [40 , 41] , and DCL4 , involved in 21-nt phasiRNA production [10] ( S7A and S7B Fig , S1 Data ) , despite the presence of E-box motifs . A substantial abundance of DCL5 transcripts still in eat1-4 anthers ( Fig 5C ) implies a possibility that other TFs participate in this process . The expression of 24-PBR genes other than DCL5 was examined . DCL1 and RDR6 are respectively required for processing of miR2275 precursors and RNA double-strand formation [10 , 21] . DCL1 and RDR6 transcripts were abundant in ST . 2 anthers; however , both were also abundant in ST . 1 and ST . 4 anthers and were unaffected by the eat1-4 mutation ( S8A Fig , S1 Data ) , indicating that expression of DCL1 and RDR6 is EAT1 independent and not restricted to meiotic stages . Transcripts of pri-miR2275a/b , the precursors of mature miR2275 , were enriched in ST . 2 anthers . In contrast to 24-PHASs and DCL5 , the amount of pri-miR2275 transcripts was elevated in the eat1-4 mutant ( S8A Fig , S1 Data ) . pri-miR2275b promoter sequences were not enriched in ChIP of EAT1-GFP-expressing anthers , despite containing E-box motifs ( S8C and S8D Fig , S1 Data ) . To investigate the EAT1 ability to promote the transcription of 24-PHAS and DCL5 loci , we performed the transient expression assay . The bHLH proteins have homo- and heterodimerization ability [42] . Thus , the effector construct encoding any two of EAT1 , TIP2 , UDT1 and TDR was cotransfected with the 24-PHAS or DCL5 promoter ( pPHAS , pDCL5 ) -Luciferase fusion reporter into rice protoplasts ( S10A Fig ) , and the promoter activity was measured . The activity of two pPHASs was significantly 4 . 46 ( chr5-20 ) and 3 . 99-fold ( chr6-97 ) elevated in EAT1-UDT1 cotransfection , compared to the no effector control ( Fig 6A ) . However , contrary to expectations , the same combination displayed insignificant effects on the pDCL5 ( Fig 6A ) . Little effect on pPHASs nor pDCL5 was observed in the transfection of EAT1 alone and EAT1-TIP2 , while the EAT1-TDR cotransfection slightly affected the activity of pPHASs ( 1 . 85 and 2 . 17 fold ) and pDCL5 ( 1 . 95 fold ) ( Fig 6A ) . Interestingly , EAT1-UDT1 cotransfection induced the pEAT1 activity by greater 7 . 61 fold ( S10B Fig ) , while it was slightly upregulated by the EAT1-TDR cotransfection ( 1 . 58 fold ) . Cotransfection of EAT1 with TIP2 , TDR or UDT1 displayed no significant effect on the pDCL3a ( S10B Fig ) . To examine the protein-protein interaction between EAT1 and UDT1 , we performed the bimolecular fluorescence complementation analysis ( BiFC ) in rice protoplasts . EAT1 fused with the C-terminal split of YFP ( EAT1-cYFP and cYFP-EAT1 ) gave positive BiFC signals when coexpressed with UDT1-nYFP ( Fig 6B , S11A and S11B Fig ) , while they tended to be detectable faintly in the nucleus ( Fig 6B , S11A Fig arrows ) or intensely in the cytoplasm ( Fig 6B , S11A Fig arrowhead ) . In both cases , the positive signals were always more intense compared to negative controls ( Fig 6B , S11A–S11C Fig ) . The above results demonstrate that the meiotic EAT1 TF promotes the transcription of 24-PHAS precursors and the EAT1 gene itself by interacting with UDT1 at the molecular level . EAT1 also promotes the DCL5 transcription , but likely with an unknown bHLH partner . Next , we examined the protein function of TIP2 , an EAT1 paralog [27 , 28 , 37] . The tip2-2 loss-of-function allele newly identified in this study had a T-DNA insertion in the third intron ( S2G–S2L Fig , S1 Data ) . In transverse sections of developing anthers ( S4K–S4X Fig ) , the wild-type tapetal and middle layer cells have dense cytoplasm ( S4M–S4N Fig ) , however , in the mutants the cell layers had sparse cytoplasm at ST . 3 and ST . 4 ( S4T–S4U Fig ) . The central PMCs were eventually collapsed probably due to malformed somatic layers ( S4V Fig ) . These results reconfirmed the previous proposal that TIP2 is essential for differentiation of precursor cells into middle layer and tapetal cells [27 , 28] . When a TIP2pro-YFP-TIP2 transcriptional fusion construct was introduced in the tip2-2 mutant , YFP-TIP2 signals were intensified in tapetal cell nuclei at ST . 2 and ST . 3 , and in addition , weaker signals were observed in the nuclei of middle layer cells ( S9A–S9C Fig ) . TIP2 protein expression was EAT1 independent , while in contrast , EAT1 expression was TIP2 dependent in transgenic plants ( S9D and S9E Fig , S1 Data ) . qRT-PCR indicated that the levels of 24-PHAS , DCL5 , and pri-miR2275a/b transcripts at ST . 2 were severely reduced in tip2-2 anthers ( S8B Fig , S1 Data ) . Using YFP-TIP2-expressing plants , the region upstream of the 24-PHAS locus ( chr5-20-Ebox1 ) was 4 . 3-fold enriched in ChIP of YFP-TIP2 ( Fig 5F , S1 Data ) , and the upstream Ebox2 and Ebox3 sequences of DCL5 also showed 8 . 1 and 3 . 4-fold enrichment , respectively ( Fig 5G , S1 Data ) . In the transient expression assay , TIP2-UDT1 cotransfection resulted in a significant increase of the pPHAS ( 8 . 63 fold on chr5-20 and 2 . 73 fold on chr6-97 ) and pDCL5 activities ( 4 . 91 fold ) ( Fig 6A ) . TIP2-TDR cotransfection also elevated the pPHAS activity ( 2 . 52 and 2 . 67 fold ) ( Fig 6A ) . Both TIP2-UDT1 and TIP2-TDR activated the pEAT1 by 5 . 72 and 2 . 35 fold , respectively ( S10B Fig ) , consistent to TIP2-dependent EAT1-GFP expression in transgenic plants ( S9E and S9F Fig ) and to the previous results [27 , 28] . The BiFC assay clearly indicated that TIP2 has a potential to interact with UDT1 ( Fig 6B and S11A–S11C Fig ) . Collectively , these results suggest that TIP2 has the potential to activate transcription of both 24-PHASs and DCL5 by interacting with UDT1 at the molecular level in early meiosis . Small RNAs are sorted to confer association with specific Argonaute family proteins [43] . MEL1 is a rice Argonaute protein whose function is well characterized in meiosis , and is abundantly expressed in male and female meiocytes , but not in surrounding somatic cells [7] . As supporting this result , the MEL1-GFP expression was limited to premeiotic and meiotic PMCs in transgenic plants ( Fig 7A ) . Here we used MEL1 Argonaute as an indicator for the 24-nt phasiRNA existence or absence in male meiocytes , and performed RNA-immunoprecipitation sequencing using anti-MEL1 antibody ( MEL1-RIPseq ) in flowers at three stages; ST . 1 , ST . 2 and ST . 4 . 1 , 711 , 113 , 1 , 361 , 031 and 2 , 679 , 034 reads of 24-nt small RNAs from three stages were obtained from MEL1-RIPseq of wild-type , eat1-4 and mel1-1 flowers , respectively ( S1 Table ) . After subtraction of mel1-1 mutant results and mapping onto the rice genome , 2 , 110 species ( 98 , 145 reads ) were defined as canonical 24-nt MEL1-associating siRNAs ( masiRNAs ) ( S1 Data ) . Through all three stages , 24-nt masiRNAs originated from repetitive sequences ( 57 . 1 , 55 . 0 and 52 . 7% at ST . 1 , ST . 2 and ST . 4 , respectively ) , intergenic regions other than 24-PHAS loci ( 32 . 6 , 28 . 7 and 27 . 6% ) and protein coding regions ( 10 . 2 , 11 . 1 and 10 . 7% ) ( Fig 7B ) . In contrast , 24-nt masiRNAs from 24-PHAS loci were detected in ST . 2 and ST . 4 ( 5 . 2 and 9 . 0% ) , but hardly detected in ST . 1 anthers ( <0 . 1% ) ( Fig 7B ) . This result corresponds to the temporal expression pattern of EAT1-dependent 24-nt phasiRNAs ( Fig 3B right ) . In eat1-4 mutant , masiRNAs from 24-PHAS loci occupied few portion of masiRNA reads even in ST . 2 ( < 0 . 1% ) and ST . 4 ( <0 . 5% ) in addition to ST . 1 ( <0 . 1% ) ( S12A Fig , S1 Data ) . MEL1 preferentially bound 24-PHAS-derived 24-nt masiRNAs with a 5'-terminal cytosine ( S12B Fig ) , consistent with the 5'-end preference of MEL1 [11] . The mel1 mutant anthers displayed only a few 24-nt RNA reads in MEL1-RIPseq in each stage ( S5 Table , S1 Data ) . The mapping mode of 24-nt masiRNAs was shown in two 24-PHAS loci for example ( chr12-82 and chr12-85 , Fig 7C ) . On the chr12-82 locus , 165 and 207 reads of only a 24-nt masiRNA species ( masiRNA_u_0815 ) were mapped at the third phase of the sense strand in ST . 2 and ST . 4 anthers , respectively ( Fig 7C left ) . A significant reduction of the masiRNA_u_0815 in male-sterile eat1-4 plants ( Fig 7C , S5 Table ) confirmed their origin in anthers , not in pistils , although MEL1 is expressed in both male and female cells [7] . A similar tendency was found in the chr12-85 and masiRNA_u_1708 ( Fig 7C right ) . Collectively , above results indicate that a subset of EAT1-dependent 24-nt phasiRNAs , at least the versions retaining 5'-terminal cytosine , was bound by MEL1 .
Previous studies unveil the complicated interaction of four bHLH proteins , UDT1 , TIP2 , EAT1 and TDR , in aliphatic metabolism and PCD in tapetal cells for rice pollen development . In post-meiosis , the TIP2/TDR heterodimer directly activates the EAT1 transcription , and the EAT1 competes for the TIP2/TDR activity [28] , because EAT1 also dimerizes with TDR [30] . EAT1 activates transcription of AP25 and AP37 , both required for tapetal PCD [30] . This study gave new insights in the relationship of tapetal bHLH proteins during early meiosis . First , the EAT1 expression is bimodal , not only in post-meiosis , but also in early meiosis ( Fig 1 ) . Second , the transient expression assay suggests a possibility that the transcription of EAT1 gene during early meiosis is activated by the TIP2/UDT1 heterodimer , and reinforced by the EAT1/UDT1 ( Fig 6 , S10 Fig ) . Third , both EAT1 and TIP2 can activate transcription of 24-PHAS lincRNAs and the DCL5 gene in tapetum during early meiosis ( Figs 3 , 5 and 6A , S10B Fig ) . The activation by EAT1 is thought to be independent of that by TIP2 , because of no interaction between two proteins as previously reported [27 , 28] and shown in this study ( Fig 6A , S10 Fig ) . In these two pathways , UDT1 is a strong candidate for the dimerization partner of EAT1 and TIP2 ( Fig 6B , S11 Fig ) , while dimerization of unknown bHLH proteins with EAT1 is supposed in the DCL5 transcription ( Fig 6A ) . In the udt1 mutant , the tapetum is aberrantly vacuolated and the tetrads are degenerated during meiosis [32] . This observation is consistent to the idea that UDT1 acts with TIP2 and EAT1 in 24-nt phasiRNA biogenesis in rice anther tapetum during meiosis . The temporal replacement of binding partners from UDT1 to TDR may enable EAT1 and TIP2 to switch downstream targets from meiotic phasiRNA production to postmeiotic tapetal PCD induction . In this study , we performed mRNA-seq and sRNA-seq to estimate 24-nt phasiRNA production only in the eat1-4 ( Fig 3 ) , but not in the tip2-2 . This is because in the tip2 mutant , tapetum is replaced by undifferentiated cell layers [27 , 28] ( S4U–S4X Fig ) , and the absence of 24-PHAS and DCL5 transcripts is possibly a by-product of the missing tapetum . However , the results that at least two 24-PHAS transcripts enriched at ST . 2 were transcribed EAT1-independently ( green spots in Figs 3A and 5C ) , and that non-negligible amounts of 24-PHAS and DCL5 transcripts are expressed still in eat1-4 anthers at ST . 2 ( Fig 3B left , Fig 5C ) . Taken together with the results of ChIP-qPCR and transient expression assay , it is obvious that TIP2 has an indispensable role in 24-nt phasiRNA production . The maize ( Zm ) bHLH122 , the EAT1 ortholog , also shows bimodal expression [25] , and MALE STERILE23 ( MS23 ) , the TIP2 ortholog , promotes the expression of bHLH122/ZmEAT1 , DCL5 , 24-PHAS transcripts and meiotic 24-nt phasiRNAs [12 , 25] . A positive interaction in the yeast two hybrid analysis ( Y2H ) is reported between MS32/ZmUDT1 and bHLH122/ZmEAT1 , consistent to the results of this study ( Fig 6 , S10 and S11 Figs ) . Thus , the bHLH TF-mediated mechanism underlying specification and development of tapetum is well conserved in rice and maize , and commonly coupled with meiotic small RNA production . A contradiction between maize and rice is in the relationship of TIP2 and UDT1 . In maize , a negative Y2H interaction of MS23/ZmTIP2 and MS32/ZmUDT1 is reported [25] , whereas rice TIP2 and UDT1 interact with each other at the molecular level ( Fig 6B ) and promote the activity of pPHASs , pDCL5 and pEAT1 ( Fig 6A , S10B Fig ) . Further analyses will be necessary for conservation and differentiation of tapetal bHLH protein functions in these monocot model plants . The observation that a subset of tapetum-originating phasiRNAs was sorted to MEL1 Argonaute , which is abundantly expressed in PMCs but not in tapetal cells ( Fig 7 ) . Though the possibility that 24-nt phasiRNA functions mainly in tapetum cannot be excluded , the result of this study suggests another possibility that the 24-nt phasiRNA is mobile between somatic and reproductive cells in rice anthers . This idea is attractive and proposed previously [12 , 13] , but should be considered carefully . It is difficult to exclude the possibility that 24-nt phasiRNAs are produced cell-autonomously in PMCs by EAT1 and/or TIP2-independent pathways , for example , DNA double-strand break ( DSB ) -induced small RNAs [44 , 45] . However , we think this unlikely , because mel1 mutant anthers with few meiotic DSBs in male meiocytes [46] produce a robust level of 24-nt phasiRNAs ( S1 and S4 Tables ) . In addition , few amounts of 24-nt phasiRNAs are detectable in eat1-4 anthers ( Fig 3C , S3 Table ) . A recent study unveiled that 24-nt phasiRNA and miR2275 expression is depleted in two rice mutants , multiple sporocytes1 ( msp1 ) and tpd1-like gene in rice1a ( tdl1a ) , in which a subset of inner anther-wall cells turn into PMCs [24 , 47] . In maize , the ms23 anther lacking the tapetum fails to produce 24-nt phasiRNAs , but the ocl4 anther developing the tapetum succeeds [12] . These results suggest that 24-nt phasiRNA production occurs exclusively in tapetum , consistent to the conclusion of this study . An alternative possibility is that precursor PHAS transcripts or their processed intermediates are transferred from tapetum , and processed into mature 24-nt phasiRNAs by 24-PBR components in PMCs . TIP2 and EAT1 are detectable in somatic companions , but hardly in PMCs ( Fig 2D , S9C Fig ) , implicating that most of DCL5-mediated 24-PHAS processing is completed in anther tapetum . However , to answer the above question , further analyses for tissue-specific expression of precursor transcripts and 24-PBR components are required . Another question for the intercellular small-RNA movement is whether the undetectable level of MEL1 proteins accumulates in tapetal cells during meiosis and associates with tapetum-expressing 24-nt phasiRNAs . However , MEL1 mRNA expression is ranked at the top 1 . 7 percentile ( the 629th highest ) of all protein-coding transcripts expressed in ST . 2 anthers ( S13 Fig ) , and as reflecting the higher mRNA level , the MEL1-GFP signal in male meiocytes made a striking contrast to undetectable signals in somatic anther cells in transgenic plants ( Fig 7A ) . Thus , small RNAs immunoprecipitated with somatic MEL1 are , if any , hard to be detected in the RIPseq analysis of anther samples , that is , the MEL1 RIPseq data of this study largely comes from the masiRNA population derived from male meiocytes . In any case , rigorous verification requires some breakthrough technologies for live-imaging of small RNAs or sequestering them into the particular cell type , such as tapetal cells . Molecular transport in plants occur either symplastically through plasmodesmata , or apoplastically across the cell membrane , cell walls and intercellular space [48] . Tapetal cells and PMCs are connected with plasmodesmata and form symplastic continuity by the onset of meiotic leptotene ( ST . 2 in this study ) [49 , 50] , when EAT1-dependent meiotic 24-nt phasiRNAs are produced in tapetal cells ( Fig 3 ) . This interconnection is broken by callose accumulation [49] . Callose is the highly impermeable polysaccharide distinct from cellulose [50] , and can be a barrier for apoplastic molecular movement . However , in ST . 2 anthers , cellulosic components still remain between tapetum and PMCs ( Fig 2B ) , in turn , callose accumulation is absent or less . Thus , both symplastic and apoplastic movements are currently possible mechanisms underlying meiotic phasiRNA movement in anthers . Taking previous findings into consideration , we propose that considerable amounts of 24-nt meiotic phasiRNAs are imported from tapetum to PMCs during early meiosis in rice . If it is true , not only the phasiRNAs with the 5'-teminal cytosine ( C-terminal phasiRNAs ) , but also non-C-terminal ones are supposed to move together in the intercellular movement , because the enrichment of C-terminal phasiRNAs in MEL1-RIPseq in this study is simply due to the selectivity of MEL1 [11] . The analysis of other Argonautes expressed in PMCs will be beneficial to trace tapetum-originating non-C-terminal phasiRNAs . Functions of other Argonaute proteins in plant meiosis still remain to be debated . Rice flowers highly express many Argonaute proteins in addition to MEL1/AGO5 ( AGO1b , AGO1d , AGO2b , AGO4a , AGO9 and AGO18 ) [8 , 51] , whose meiotic roles are largely unknown . Arabidopsis AGO4 plays important roles in chromosome condensation and segregation during the first meiotic division [52] , comparable to rice EAT1 function in male meiosis ( Fig 2C ) . ZmAGO104 , orthologous to Arabidopsis AGO9 , is also required for meiotic chromosome condensation [53] . In either case , the relationship of Argonaute/small RNA complexes to the nuclear RdDM and histone modification will be one of the most important questions regarding epigenetic regulation of plant meiosis . Dukowic-Schulze et al . [13] unveiled that both 21- and 24-PHAS precursor loci showed higher DNA methylation in all cytosine contexts ( CG , CHG , CHH , where H represents A , T or C ) in isolated maize PMCs . The highest context was CHH methylation , implying that reproductive phasiRNAs are involved in RNA-directed DNA methylation ( RdDM ) in PMCs . RdDM includes both de novo DNA methylation and histone H3 lysine-9 ( H3K9 ) methylation in plants [54–59] . Supporting this idea , MEL1 is thought to govern meiosis-specific chromatin remodeling accompanying dynamic alteration in H3K9 dimethylation [46] . Meiosis is a special type of cell division to transmit new haplotypes to the next generation , and additionally , to survey incompatibilities in ploidy levels and chromosomal structures between both parents . This process must be strictly regulated by complicated mechanisms genetically and epigenetically . Recent genome-wide studies have revealed that small RNA-mediated and non-cell-autonomous regulation is likely general in reproduction of eukaryotic species . Further analyses of tapetum-expressing bHLH TFs and meiotic phasiRNAs in anthers will bring new epigenetic insights into plant reproduction systems .
The eat1-4 mutant is a Tos17 insertion line produced from the rice variety , cv . Nipponbare [60] , NF9876 , kindly provided by the Rice Genome Resource Center , Japan . The mel1-1 mutant [7] , another Tos17 line with the Nipponbare background , was kindly provided by the National Bioresource Project ( NBRP ) Rice , conducted by the Japan Agency for Medical Research and Development ( AMED ) . The tip2-2 mutant is a T-DNA tag line with the genetic background of cv . Dongjin [61 , 62] , 1B-24309 , kindly provided by Dr . G . An ( POSTECH , Korea ) . All plants were grown in moist chambers , greenhouses , and/or open paddy fields at the National Institute of Genetics ( NIG ) , Mishima , Japan . Plant genotypes were determined by PCR using GoTaq Green Master Mix ( Promega ) and gene-specific and T-DNA/Tos17-internal primers ( S6 Table ) . Rice spikelets were fixed in PMEG buffer ( 50 mM PIPES , 10 mM EGTA , 5 mM MgSO4 , and 4% glycerol , pH 6 . 8 ) containing 4% paraformaldehyde ( PFA ) for 3 h and washed six times in PMEG buffer for 2 hours . After dehydration using ethanol series , they were embedded in Technovit7100 resin ( Heraeus Kulzer ) , sectioned in 2 μm thick slices using a LM2255 microtome ( Leica Microsystems ) , stained with 0 . 1% toluidine blue O ( Wako Pure Chemicals ) and photographed using a BX50 light microscope ( Olympus ) and a DP50 camera system ( Olympus ) . Cellulosic cell wall staining was conducted according to the method described previously [63] . Fluorescent images were captured using a Fluoview FV300 CLSM system ( Olympus ) , and pseudo-colored and merged using Photoshop CS4 ( Adobe Systems Inc . ) . EAT1pro-EAT1-GFP ( Fig 1D ) was constructed as follows . The 5 . 3 kbp HindIII-XhoI genomic fragment including the upper half of the EAT1 gene and its promoter region was subcloned from a BAC clone , OSJNBa0010K21 , into the pBluescriptII ( pBSII ) -SK ( - ) vector . The 1 . 2 kbp XhoI-EcoRV fragment including the 3’ downstrem region of the EAT1 gene was also subcloned into another pBSII-SK ( - ) , and from this plasmid , the 2 . 0 kbp XhoI-EcoRV fragment harboring a sGFP sequence just in the front of the EAT1 stop codon was constructed using EAT1-specific primers , bHLH141stop-BamHI/bHLH141XhoI-BamHI and bHLH141stop-NotI/M13-Rev , and a CaMV35S-sGFP ( S65T ) -nos3ʹ vector [64] , kindly provided by Dr . Y . Niwa ( Shizuoka U . , Japan ) . The resultant 5 . 3 kbp and 2 . 0 kbp fragments were inserted into a pPZP2H-lac binary vector [65] to assemble EAT1pro-EAT1-GFP . TIP2pro-YFP-TIP2 ( S8A Fig ) was constructed as follows . The 6 . 6 kbp genomic fragment , including the entire TIP2 gene with 4 kbp upstream and 0 . 5 kbp downstream sequences , was cut out from a rice BAC clone OSJNBa0001E17 by digestion with SpeI , and inserted into pBSII-SK ( - ) vector . From the 6 . 6 kbp fragment , the 1 . 8 kbp HindIII-SalI fragment including the translational initiation site ( TIS ) was subcloned into pBSII-SK ( - ) . From this plasmid , the YFP sequence was inserted just in front of TIS by using TIP2-specific primers bHLH142start-NcoI/M13-Rev and bHLH142start-BsrGI/T7-EcoRI and a pEYFP vector ( a cloning vector with EYFP sequence in pUC18 backbone ) . Then , the 1 . 8 kbp fragment with the YFP sequence was inserted back into the original 6 . 6 kbp genomic fragment/pBSII-SK ( - ) plasmid . The resultant 7 . 4-kbp insert was digested , blunt-ended , and reinserted into pGWB601 binary vector [66] , kindly provided by Dr . T . Nakagawa ( Shimane U . , Japan ) . In case of TIP2pro-TIP2-YFP ( S8D Fig ) , the TIP2 stop codon in the above 6 . 6 kbp genomic fragment/pBSII-SK ( - ) was replaced by YFP sequence by using TIP2-specific primers , bHLH142stop-NcoI/M13-20 and bHLH142stop-BsrGI/M13-Rev , and a pEYFP . Finally , the 7 . 4 kbp of TIP2pro-TIP2-YFP insert was assembled in the pPZP2H-lac . In above constructions , KOD-FX DNA polymerase ( TOYOBO ) was used for PCR . In MEL1pro-GFP-MEL1 construction , the GFP sequence was inserted just in front of MEL1 TIS in pKN16 , a binary vector containing the 18 kbp MEL1 genomic fragment [7] . Two DNA fragments , corresponding to 5ʹ upstream and 3ʹ downstream regions of MEL1 TIS , were amplified from pKN16 with primer pairs up_nf/up_nr and up_atgf/up_r , respectively . Linker-attached sGFP coding sequence was amplified from CaMV35S-sGFP ( S65T ) -nos3ʹ with ngfp_f/ngfp_r primers . The PCRs were conducted using a PrimeSTAR Max DNA polymerase ( TaKaRa ) . The three amplified DNA fragments were mixed with the NruI-AscI-digested pKN16 and incubated with an In-Fusion HD enzyme premix ( TaKaRa ) to assemble MEL1pro-GFP-MEL1 , following manufacturer’s instructions . All the primer sequences for the construction were listed in S6 Table . The constructs were transformed into rice calli using agrobacterium-mediated transformation [67] , in which Hygromycin B ( 50 mg/L in media; Wako Pure Chemicals ) or glufosinate-ammonium PESTANAL ( 5 mg/L in media; Sigma-Aldrich ) was used for a positive selection . Anthers embedded in 6% SeaKem GTG agarose ( Lonza ) were sliced into 50 μm thickness by MicroSlicer DTK-ZERO1 ( D . S . K . ) , and mounted on slide grasses with VECTASHIELD ( Vector Laboratories ) containing DAPI . Fluorescent images were captured using Fluoview FV300 CLSM system ( Olympus ) . Spikelet ( lemma ) and anther lengths were measured under SMZ645 stereo microscopy ( Nikon ) . 0 . 8–1 . 2 mm anthers were fixed with 4% PFA/PMEG and provided for chromosome observations as previously described [7] . Fluorescent images of DAPI were taken as described above . Anther or spikelet samples were separated by their lengths as corresponding to ST . 1-ST . 6 stages ( S7 Table ) , immediately frozen with liquid nitrogen in microtubes , and stored at -80°C until use . Total RNAs were extracted from the samples using TRIzol reagent as manufacturer’s recommendation ( Life Technologies ) , and treated with DNase I ( TaKaRa ) . In qRT-PCR , 1 μg of total RNA was reverse-transcribed by oligo ( dT ) 12-18 primer ( Life Technologies ) and SuperscriptIII reverse transcriptase ( Life Technologies ) . The products were 20-fold diluted and supplied for real-time qPCR using gene-specific primers ( S6 Table ) , KAPA SYBR FAST universal qPCR Kit ( KAPA Biosystems ) and Thermal Cycler Dice Real Time System ( TaKaRa ) . Rice Ubiquitine gene was used as an internal standard . Total RNAs were extracted from ST . 1 , ST . 2 and ST . 4 anthers of wild-type and eat1-4 plants , three biological replicates each . For mRNA-seq , 1 μg of total RNA was subjected to library construction using KAPA stranded mRNA-seq Kit Illumina Platforms ( KAPA biosystems ) . Eighteen libraries differentially indexed by FastGene Adapter kit ( Nippon Genetics ) were multiplexed ( 9 per lane ) and sequenced by HiSeq2500 ( Illumina ) with SR50 ( single ended ) . Adapter sequences were removed in silico using R package QuasR [68] . mRNA-seq reads were mapped on the rice genome IRGSP1 . 0 using Tophat ( v2 . 0 . 14 ) [69] . Differential expression analysis of annotated genes were conducted using Cuffdiff2 program [70] . The genes fulfilling all of the following conditions were regarded as EAT1-dependent and ST . 2-enriched genes; ( 1 ) genes showing >2-fold higher FPKM values in wild-type ST . 2 anthers than the values in wild-type ST . 1 and ST . 4 anthers , ( 2 ) genes showing >2-fold higher FPKM values in wild-type ST . 2 anthers than the values in eat1-4 ST . 2 anthers , and ( 3 ) genes with each standard deviation less than a half of the FPKM mean value of three replicates in wild-type ST . 2 anthers . The lincRNAs were determined by Cuffdiff2 ( merged . gtf ) , in which protein-coding genes were removed as referring to MSU7 . 0 annotation , and unannotated but transcribed genomic regions larger than 200 bp were extracted . FPKM values of lincRNAs were calculated by BEDtools [71] . Furthermore , EAT1-dependent and ST . 2-enriched lincRNAs were extracted according to the same conditions described above for coding genes . For sRNA-seq , 1 μg of total RNA was provided for library construction by NEBNext Multiplex Small RNA Library Prep Set for Illumina ( New England BioLabs ) . The libraries were 9-plexed per lane and sequenced by HiSeq2500 ( illumina ) with SR52 , a 2-bp extended version of SR50 , for higher-quality sequencing . After trimming by QuasR , 24-nt long sRNA-seq reads were extracted by ShortRead [72] , and mapped to the rice IRGSP1 . 0 genome using Tophat , in which reads having >50 multi-hits on rice genome or any mismatches were cut off ( -N 0 -g 50 ) . If 24-nt RNAs with >10 FPKM values were mapped on each of EAT1-dependent and ST . 2-enriched lincRNA loci identified above , the loci were defined as 24-PHAS loci . Regional abundance of mRNA-seq and 24-nt sRNA-seq reads mapped on the rice genome ( shown in Fig 4A ) was calculated in a sliding window ( window; 50 kbp , step; 25 kbp ) by BEDtools . Conserved motifs were searched in each 24-PHAS locus , in addition to 200 bp regions both upstream and downstream sequences , by MEME SUITE program [35] . Phased scores were calculated as described by Howell et al . [73] . A degradome-seq dataset from young panicles of indica rice variety , cv . 93–11 , was obtained from Sequence Read Archive of DNA Data Bank of Japan ( DDBJ-SRA ) under the accession code SRR034102 [38] . Adaptor sequence and low-quality reads were removed using FASTX-toolkit ( http://hannonlab . cshl . edu/fastx_toolkit/ ) and the reads retaining 20- or 21-nt length were mapped onto rice IRGSP1 . 0 genome using Bowtie 2 [74] . The frequency of 5’-end of mapped reads were manually examined within and around 24-PHAS loci identified in this study ( S4 Table ) . To determine the TSS of two 24-PHASs and a pri-miR2275 ( chr5-20 , chr6-97 and pri-miR2275b ) , the standard 5ʹ rapid amplification cDNA end ( 5ʹ RACE ) method was applied using a GeneRacer kit ( Thermo Fisher Scientific ) , total RNA from ST . 2 wild-type anther , and gene specific primers ( S6 Table ) . Eight clones from each locus were sequenced using a BigDye Terminator v3 . 1 cycle sequencing kit ( Applied Biosystems ) and a PRISM 3130xl sequencer ( Applied Biosystems ) and the end of the longest read ( s ) was marked as TSS . Rice young panicles from transgenic derivatives were fixed , and the anthers at early meiosis ( around 0 . 5 mm ) were supplied for ChIP as described previously [75] . The anti-GFP antibody No . 598 and the normal rabbit IgG ( both from MBL International ) were used for positive and negative ChIP experiments , respectively . The extracted DNAs were analyzed by real-time qPCR using region-specific primers ( S6 Table ) . The 1/10 volume of chromatin-containing samples without IP treatment was prepared for the input samples . The 2-kbp upstream sequences from the translational start site of 24-PHAS ( chr5-20 , chr6-97 ) , DCL5 , EAT1 and DCL3a genes , all originated from the japonica rice cv . Nipponbare , were inserted in the upstream of the firefly Luciferase CDS and the nopaline synthase ( nos ) terminator . This reporter construct was cloned into pBSII-SK ( - ) plasmid ( S10A Fig ) . For the effector construct , the cauliflower mosaic virus 35S ( CaMV35S ) promoter was fused with the cDNAs of EAT1 , TIP2 , TDR and UDT1 genes , originated from Nipponbare ST . 2 anthers , with the nos terminator . and cloned into pBSII-SK ( - ) ( S10A Fig ) . For normalization of the firefly Luciferase activity , the Luciferase cDNA of Renilla reniformis were fused with the CaMV35S promoter and the nos terminator , inserted into pBSII-SK ( - ) ( S10A Fig ) , and cotransfected with the effector and reporter constructs as an internal control in all experiments . All PCR primers for the above constructions were listed in S6 Table . PrimeSTAR Max DNA polymerase ( TaKaRa ) was used for PCR amplification according to the manufacturer's instruction . Protoplast preparation from rice seedlings , transfection of plasmids , and protein extraction from protoplasts were according to the method previously described [76] . The Luciferase activity was detected using Dual-Luciferase Reporter Assay System ( Promega ) and Filter MAX F5 multi-mode microplate reader ( Molecular Devices ) . A pair of split YFP vectors ( pBS-35S-nYFP and pBS-35S-cYFP ) were kindly provided by Drs . D . Tsugama ( Hokkaido U . , Japan ) and T . Takano ( The U . of Tokyo , Japan ) [77] . Each of EAT1 , TIP2 and UDT1 cDNAs , originated from Nipponbare ST . 2 anthers , was inserted into the either of upstream or downstream of both pBS-35S-nYFP and pBS-35S-cYFP vectors . For the nuclear marker , the rice Histone 2B ( H2B ) cDNA were fused with the maize Ubiquitine promoter and in-frame with the tagRFP gene ( Evrogen ) , cloned into pPZP2H-lac binary vector [65] , and cotransfected with a pair of split YFP constructs in all experiments . All PCR primers used here were listed in S6 Table . PrimeSTAR Max DNA polymerase ( TaKaRa ) was used for PCR . The protoplast preparation and plasmid transfection were same with the method described above . Fluorescent images were captured by Fluoview FV300 CLSM system ( Olympus ) and processed by Photoshop CS4 ( Adobe systems Inc . ) , under the identical conditions and parameters through all experiments . We tried all sixteen combinations of split YFP constructs to assess EAT1-UDT1 and TIP2-UDT1 interactions , and thirteen combinations of negative controls . However , UDT1-cYFP and cYFP-UDT1 gave the intense signal in a single transfection as negative controls , and excluded from the assay . Then , the total eight combinations of EAT1-UDT1 ( EAT1-cYFP/UDT1-nYFP , EAT1-cYFP/nYFP-UDT1 , cYFP-EAT1/UDT1-nYFP , cYFP-EAT1/nYFP-UDT1 ) and TIP2-UDT1 ( TIP2-cYFP/UDT1-nYFP , TIP2-cYFP/nYFP-UDT1 , cYFP-TIP2/UDT1-nYFP , cYFP-TIP2/nYFP-UDT1 ) were assayed . RIP fractions from wild-type , mel1-1 and eat1-4 flowers at ST . 1 , ST . 2 and ST . 4 , each of which included two biological replicates , were obtained using anti-MEL1 antibody as described previously [11] . Library construction , sequencing , adapter trimming , size filtration and mapping to rice genome were done as well as sRNA-seq methods described above . Reads per million ( RPM ) values were calculated in the respective 24-nt RNA sequences and compared among wild-type , mel1-1 and eat1-4 fractions . In this process , 24-nt masiRNAs were defined in 24-nt RNA sequences as having ≥15 RPM detected in wild-type ST . 1 , ST . 2 or ST . 4 flowers , and ≥RPM 4-fold enriched in wild-type compared to mel1-1 . EAT1; Os04g0599300 , TIP2; Os01g0293100 , TDR; Os02g0120500 , UDT1; Os07g0549600 , MEL1 , Os03g0800200 , DCL5; Os10g0485600 , DCL3a; Os01g0909200 , DCL4; Os04g0509300 , DCL1; Os03g0121800 , RDR6; Os01g0527600 , AP25; Os03g0186900 . ( Rice Annotation Project Database ( RAP-DB ) ( http://rapdb . dna . affrc . go . jp ) ) .
|
Meiotic crossover formation shuffles homologous genes between parental genomes , and enables transmission of new gene sets to the offspring . Frequency and positions of crossovers are determined by numerous genetic and epigenetic factors , and low nucleosome-density regions are associated with crossover hot spots in yeasts and Arabidopsis . The epigenetic chromosome landscape is shaped by unevenly distributed modifications of nucleosome components , histones and DNAs . Recently , we found that MEL1 ( ARGONAUTE5 ) promotes large-scale remodeling of meiotic chromosomes with dramatic increases of histone H3 lysine 9 dimethylation , and that loss of MEL1 resulted in early meiotic arrest with few crossovers present . In rice anthers , MEL1-associating small interfering RNAs ( masiRNAs ) were composed of large amounts of premeiotic 21-nt phasiRNAs , plus low levels of both 24-nt repeat-associated siRNA and meiotic 24-nt phasiRNAs . Production of 24-nt phasiRNA during the meiotic stage was largely EAT1-dependent . Collectively , our findings suggest a possibility that unknown small RNA-mediated signaling regulates male meiosis non-cell-autonomously , probably a downstream output involves large-scale chromosome remodeling promoted by Argonaute proteins , while a possibility of EAT1-dependent , but small RNA-independent signaling cannot be excluded . In any cases , the studies on MEL1 and tapetal bHLH proteins will be a clue to reveal small RNA-mediated processes determining meiotic epigenetic landscape .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"biotechnology",
"meiosis",
"plant",
"anatomy",
"gene",
"regulation",
"cell",
"cycle",
"and",
"cell",
"division",
"cell",
"processes",
"plant",
"science",
"rice",
"genetically",
"modified",
"plants",
"experimental",
"organism",
"systems",
"sequence",
"motif",
"analysis",
"plants",
"flower",
"anatomy",
"genetic",
"engineering",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"small",
"interfering",
"rnas",
"grasses",
"genetically",
"modified",
"organisms",
"chromosome",
"biology",
"anthers",
"bioinformatics",
"gene",
"expression",
"genetic",
"loci",
"biochemistry",
"rna",
"eukaryota",
"plant",
"and",
"algal",
"models",
"double",
"stranded",
"rna",
"cell",
"biology",
"nucleic",
"acids",
"database",
"and",
"informatics",
"methods",
"genetics",
"biology",
"and",
"life",
"sciences",
"non-coding",
"rna",
"plant",
"biotechnology",
"organisms"
] |
2018
|
EAT1 transcription factor, a non-cell-autonomous regulator of pollen production, activates meiotic small RNA biogenesis in rice anther tapetum
|
While the contribution of CD8+ cytotoxic T lymphocytes to early containment of HIV-1 spread is well established , a role for NK cells in controlling HIV-1 replication during primary infection has been uncertain . The highly polymorphic family of KIR molecules expressed on NK cells can inhibit or activate these effector cells and might therefore modulate their activity against HIV-1-infected cells . In the present study , we investigated copy number variation in KIR3DH loci encoding the only activating KIR receptor family in rhesus monkeys and its effect on simian immunodeficiency virus ( SIV ) replication during primary infection in rhesus monkeys . We observed an association between copy numbers of KIR3DH genes and control of SIV replication in Mamu-A*01– rhesus monkeys that express restrictive TRIM5 alleles . These findings provide further evidence for an association between NK cells and the early containment of SIV replication , and underscore the potential importance of activating KIRs in stimulating NK cell responses to control SIV spread .
Natural killer ( NK ) cells are the primary effector cells of the innate immune system , representing a first line of defense against viruses through their ability to lyse virally infected cells without prior antigen sensitization [1]–[3] . NK cells express a complicated set of activating and inhibitory receptors on their cell surfaces that recognize specific ligands on target cells [4] . Inhibitory receptors transmit inhibitory signals to NK cells that protect healthy cells from destruction by NK cell-mediated cytotoxicity , whereas activating NK cell receptors transmit activating signals to these effector cells . It is the balance of these opposing signals that determines the activation state of an NK cell and , in so doing , regulates NK cell-mediated killing and cytokine production [5]–[7] . Among these receptor families expressed by NK cells are the inhibitory and activating killer cell immunoglobulin-like receptors ( KIR ) . The highly polymorphic KIRs recognize MHC class I molecules as ligands [8] , [9] , and the coincident expression of certain KIRs and MHC class I molecules in an individual influences the outcome of a number of viral infections [10] , [11] . Recent studies have shown that activating KIRs and their MHC class I ligands can affect AIDS pathogenesis . The expression of KIR3DS1 , an activating KIR receptor , has been shown to delay AIDS progression when its ligand , HLA-B Bw4 alleles with an isoleucine at position 80 ( HLA-B Bw4-80Ile ) , is coexpressed in an individual [12] . Consistent with this finding , an in vitro functional analysis showed that KIR3DS1+ NK cells are able to inhibit HIV-1 replication in HLA-B Bw4-80Ile+ target cells [13] . Further , KIR3DS1+ NK cells selectively expand during acute HIV-1 infection in the presence of HLA-B Bw4-80Ile [14] . In addition to these findings , others have reported an association between the expression of certain inhibitory KIR3DL1 allotypes and protection against HIV-1 disease progression , when the KIR3DL1 ligand , HLA-B Bw4 alleles , is also expressed in an individual [15] . Studies of the contributions of NK cells to HIV-1 control have been limited by the difficulties associated with finding individuals who can be evaluated during the earliest phase of the infection . The SIV-infected rhesus monkey therefore provides a critical model for exploring NK cell biology in the setting of an AIDS virus infection [16] . We have previously shown that there are five KIR receptor families in rhesus monkeys [17] . KIR3DH is the only activating KIR family in this nonhuman primate species , and this family of molecules is highly polymorphic [18]–[21] . An understanding of this KIR gene family of rhesus monkeys provides an important basis for exploring the contributions of KIR receptors and NK cells in early AIDS pathogenesis in the SIV/macaque model . In the present study , we evaluated the copy number variation ( CNV ) of activating KIRs in rhesus monkeys and demonstrated an association between the extent of this CNV and SIV control during primary SIV infection in a cohort of Mamu-A*01– rhesus monkeys that were homozygous for the restrictive TRIM5 alleles .
This study was initiated to explore the copy number variation of activating KIR genes of Indian-origin rhesus monkeys and its contribution to the control of virus replication during the acute phase of SIV infection . To date , only one activating KIR receptor family , KIR3DH , also termed KIR3DS in recent publications to be consistent with the nomenclature used in describing human KIRs [19] , [21] , has been identified in rhesus monkeys . Interestingly , the KIR3DH receptor family in this macaque species comprises genes that display extensive polymorphism . To determine the copy number of activating KIR genes in rhesus monkeys , we developed a quantitative real-time PCR assay ( qPCR ) using a primer/probe set that binds to a conserved region of KIR3DH genes that encodes the transmembrane domain of the KIR3DH proteins . This primer/probe set was designed to amplify 23 previously described KIR3DH alleles ( GenBank accession numbers MmKIR3DH1-4 ( AF334648-AF334651 ) , MmKIR3DH7-21 ( EU702453-EU702473 ) , MmKIR3DH-like1-4 ( AY505479-82 ) ) [17] , [18] . We confirmed that this primer/probe set does not bind inhibitory KIR genes by sequencing the qPCR amplicons ( data not shown ) . Genomic DNA derived from B-lymphoblastoid cell lines ( B-LCLs ) or peripheral blood mononuclear cells ( PBMCs ) of 77 rhesus monkeys was extracted , and KIR3DH copy numbers were determined for each monkey by qPCR using serial dilutions of a plasmid containing the amplicon of the KIR3DH qPCR reaction . STAT6 , since it is present at two copies per diploid genome ( pdg ) , was used as a reference gene in these studies [22] . Intra-experimental reproducibility of the qPCR assay was confirmed by analyzing triplicate samples for each of the 77 monkeys in two separate experiments , determining KIR3DH copy numbers based on the means of the triplicate values ( R2 = 0 . 866 , β = 0 . 816 ) ( Figure 1A ) . To validate the accuracy of the KIR3DH copy number qPCR assay , we developed a multiplex ligation-dependent probe amplification assay ( MLPA ) . Twenty genomic DNA samples were analyzed using one oligonucleotide set for the KIR3DH genes and , for reference genes , two oligonucleotide sets for EP300 encoding the E1A binding protein p300 and one set for CREBBP encoding the CREB-binding protein . EP300 and CREBBP are each present at two copies pdg in the rhesus monkey genome . Because of the extensive polymorphism at the KIR3DH loci , only one MLPA oligonucleotide set could be designed that bound to the same KIR3DH genes amplified in the qPCR assay . Relative MLPA signals for KIR3DH , normalized against the control gene signals , were assessed in two MLPA experiments using the same DNA samples . These signals clustered into groups corresponding to copy numbers 2 , 3 , 4 and 5 ( Figure 1B ) . The KIR3DH copy numbers determined by MLPA were correlated with KIR3DH copy numbers determined by qPCR ( R2 = 0 . 77 , β = 0 . 729 ) ( Figure 1C ) . Using this assay , we determined KIR3DH copy numbers in a cohort of 77 Indian-origin rhesus monkeys ( Figure 2A ) . KIR3DH copy numbers varied extensively in this cohort of monkeys , ranging from 0 to 12 copies pdg ( median = 4 . 11 copies pdg ) . We rejected that this distribution had a Gaussian distribution by the skewness test ( P = 0 . 001; i . e . , the distribution was not symmetric ) as well as by the kurtosis test ( P = 0 . 005; i . e . , the distribution is more sharply peaked than the Gaussian ) , the chi-squared test ( P = 0 . 0006 ) , and the Shapiro-Wilks test ( P = 0 . 003 ) ( data not shown ) . Since the aim of this study was to determine the association of activating KIR copy numbers on early control of SIV replication , we first sought to determine whether activating KIR copy numbers in this species co-stratified with other alleles already implicated in SIV control . Therefore , we assessed whether KIR3DH copy numbers in this cohort of animals were associated with their Mamu-A*01 or Mamu-B*17 status or their expression of particular TRIM5 alleles . The expression of certain MHC class I alleles , particularly Mamu-A*01 and Mamu-B*17 , has been associated with control of virus replication and delayed disease progression following SIV infection of Indian-origin rhesus monkeys [23]–[27] . We have also recently demonstrated the role of rhesus monkey TRIM5 alleles 1–5 in restricting SIV infection and the impact of this restriction on the clinical outcome of SIV infection in vivo [28] . Although MHC class I , TRIM5 and KIR3DH genes are encoded on different chromosomes in rhesus monkeys , contributions to SIV control by one allele could be a surrogate for the effects of another allele , as seen for CCL3L and Mamu-A*01 [29] . We found that KIR3DH copy numbers were not significantly different between Mamu-A*01+ and Mamu-A*01– rhesus monkeys ( Mann-Whitney , P = 0 . 31 ) ( Figure 2B ) . Moreover , KIR3DH copy numbers were also not significantly different between Mamu-B*17– and Mamu-B*17+ monkeys ( Mann-Whitney , P = 0 . 49 ) . There were insufficient Mamu-B*08+ animals ( n = 3 ) in this cohort to assess the associations of this allele with KIR3DH copy numbers . Finally , we found that KIR3DH copy numbers were not significantly different in rhesus monkeys expressing only TRIM5 alleles 1–5 and rhesus monkeys expressing at least one of the permissive TRIM5 alleles 6–11 ( Mann-Whitney , P = 0 . 75 ) , nor were they significantly different when considering the four subsets of Mamu-A*01 and TRIM5 ( Kruskal-Wallis , P = 0 . 19 ) . To explore whether copy number variation of activating KIR3DH alleles in rhesus monkeys might be associated with protection against SIV replication , we evaluated a cohort of 57 rhesus monkeys infected with SIVmac251 . SIV plasma RNA levels were measured in these monkeys at peak and set-point of the infection , days 14 and 70 post-SIV challenge , respectively , since these measures of viremia have been shown to be predictors of SIV disease progression in rhesus monkeys [30] , [31] . No association of KIR3DH copy numbers and viral load at peak ( Figure 3A ) or set-point ( data not shown ) was observed . To explore further a possible relationship between activating KIR copy numbers and control of SIV replication in vivo , we divided this cohort of rhesus monkeys into two groups based on their expression or lack of expression of the MHC class I allele Mamu-A*01 . Since its expression can contribute to SIV control , we reasoned that the expression of Mamu-A*01 might mask effects of other host alleles on controlling viral replication during SIV infection . On the basis of similar reasoning , the Mamu-A*01+ and Mamu-A*01– rhesus monkeys were further divided into monkeys expressing only the TRIM5 alleles 1–5 and monkeys expressing at least one TRIM5 allele from the group 6–11 . Finally , the monkeys were further subdivided into those having KIR3DH copy numbers above the median ( ≥4 . 111 copies pdg ) and those having KIR3DH copy numbers below the median ( <4 . 111 copies pdg ) . As expected , the Mamu-A*01+ monkeys had lower viral load values at peak and set-point than the Mamu-A*01– monkeys ( data not shown ) . In Mamu-A*01+ rhesus monkeys , even when those monkeys were divided according to their TRIM5 allele expression , there was no association between KIR3DH copy numbers and peak ( Figure 3B ) or set-point ( data not shown ) viral load . However , there were too few monkeys with TRIM5 alleles 1–5 to determine whether there was such an association . Strikingly , in the Mamu-A*01– rhesus monkeys , we observed a very strong negative trend toward an association between KIR3DH copy numbers and peak plasma SIV RNA levels ( P = 0 . 08 ) ( Figure 3C ) . The association between KIR3DH copy numbers and peak plasma SIV RNA levels was even more pronounced in Mamu-A*01– rhesus monkeys expressing only the TRIM5 alleles 1–5 . In this group of monkeys , animals having KIR3DH copy numbers above the median had significantly lower peak SIV RNA levels than animals having KIR3DH copy numbers below the median ( Mann-Whitney , P = 0 . 02 ) , with a 0 . 4 log median difference . We also analyzed the association between KIR3DH copy number and peak plasma viral load by fitting the data to a parabola or a linear spline with a fixed knot at a KIR3DH copy number of 5 . The fixed knot of 5 was chosen because of previous reports that an individual NK cell usually expresses 3–5 KIRs that are randomly selected on their surface [32] . We hypothesized that KIR3DH copy numbers above 5 might not be associated with a linear increase in surface expression of KIR3DH molecules . In the Mamu-A*01– , TRIM5 1–5 expressing monkeys , we observed comparable significance for the coefficient of the linear term in the parabola and for the slope of the spline for copy number ≤5 ( P = 0 . 015 and P = 0 . 016 ) , the squared term in the parabola and the slope of the second spline were not significant ( P = 0 . 074 and P = 0 . 53 ) ( Supplementary Figure S2 in Text S1 ) . Interestingly , this association was no longer apparent by the time viral set-point was reached on day 70 post-infection ( data not shown ) . These results suggest that NK cells from this subset of monkeys expressing higher numbers of KIR3DH copies may contribute more to control of SIV replication than NK cells from monkeys expressing lower numbers of activating KIR copies . We then assessed some of the clinical consequences of SIV infection in Mamu-A*01– rhesus monkeys . The loss of peripheral blood CD4+ T cells , and more importantly , central memory ( CM ) CD4+ T cells following SIV infection have been shown to predict survival of the infected rhesus monkeys [31] . Therefore , we evaluated the loss of peripheral blood CD4+ T cells and central memory ( CM ) CD4+ T cells on days 14 ( peak ) and 70 ( set-point ) following SIVmac251 infection in these monkeys . Data were assessed by dividing the Mamu-A*01– rhesus monkeys into those having KIR3DH copy numbers above the median ( ≥4 . 11 copies pdg ) and those having KIR3DH copy numbers below the median ( <4 . 11 copies pdg ) . Neither loss of total CD4+ T cells nor loss of CM CD4+ T cells was significantly associated with KIR3DH copy numbers in these monkeys ( Figure 4A , B ) . We also evaluated survival as a clinical indicator of long-term protection following SIV infection in the Mamu-A*01– rhesus monkeys . Assessing the percentage of monkeys that had died by day 283 following SIV infection ( day 283 was chosen ahead of time; after 283 days some monkeys were euthanized and some monkeys were used in other experiments ) , we observed no association between KIR3DH copy numbers and survival following infection ( data not shown ) . Therefore , peak viral load was the only clinical correlate of KIR3DH copy numbers in this cohort of monkeys . Since some reports by other investigators had suggested that particular KIR alleles were associated with high plasma viral RNA levels in SIV-infected rhesus monkeys [20] , [33] , we investigated whether particular KIR3DH alleles contributed to the association between KIR3DH copy numbers and control of peak SIV RNA levels in these monkeys . Full-length KIR3DH cDNA clones generated from 8 unrelated , Mamu-A*01– rhesus monkeys with different KIR3DH copy numbers ranging from 1–10 copies pdg were acquired by PCR . Two of the isolated KIR cDNA sequences were identical to rhesus monkey KIR alleles that had previously been reported ( Mamu-KIR3DS10-JHB-HQ ( GU112262 ) and Mamu-KIR3DS05-JHB-HH ( GU112301 ) ) [19] . The majority of isolated KIR cDNA clones differed in their sequences from previously reported rhesus monkey KIR sequences . These novel sequences have been assigned the GenBank accession numbers JN613291-JN613300 . The predicted amino acid sequences of all KIR3DH alleles were aligned ( Supplementary Figure S1 in Text S1 ) . While some KIR3DH cDNA sequences were only observed in individual monkeys ( eg . JN613291 ) , other KIR3DH alleles were observed in multiple monkeys ( eg . JN613292 ) ( Figure 5 ) . Importantly , there was no apparent trend toward certain KIR3DH alleles being expressed in rhesus monkeys with low or high KIR3DH copy numbers . Therefore , there was no evidence that a particular KIR3DH allele was responsible for the observed effect on early SIV control . To assess how activating KIR CNV might affect early SIV containment , we first determined the association between the number of KIR3DH copies in a cell and the expression of KIR3DH genes by that cell . We utilized the qPCR assay that we developed for determining KIR3DH CNV to measure KIR3DH mRNA expression levels in peripheral blood CD16+ NK cells of 28 naïve rhesus monkeys . Relative KIR3DH mRNA expression was significantly associated with KIR3DH copy numbers in the evaluated cell populations , as determined by linear regression analysis ( P = <0 . 001 , R2 = 0 . 51 ) ( Figure 6A ) . To determine if these differences in KIR3DH RNA expression levels persisted over time , we sampled 15 of these naïve rhesus monkeys one month later , and KIR3DH RNA expression levels at both sampling dates were measured in the same qRT-PCR run . Relative mRNA expression levels of KIR3DH genes from the first and second sampling associated positively , reaching statistical significance as determined by linear regression analysis ( P = 0 . 03 , R2 = 0 . 33 ) ( Figure 6B ) . These findings suggest that increased KIR3DH copy numbers associate with high , stable KIR3DH mRNA expression levels . We next investigated the relative representation of various subpopulations of NK cells in rhesus monkeys expressing different numbers of KIR3DH copies . Rhesus monkey NK cells were defined as CD3– CD8α+ NKG2A+ , and CD16 and CD56 expression were used to delineate three NK cell subsets: CD16+ , CD56+ and double-negative ( DN ) NK cells ( Figure 7A ) . Since these subsets have been previously shown to mediate different effector functions , an expansion or contraction might only be expected to occur in certain NK cell subsets . The relative representation of NK cells , as a percentage of total circulating lymphocytes , and of NK cell subsets , as a percentage of total NK cells , were determined by flow cytometric analysis of PBMCs of naïve monkeys and of monkeys sampled on day 28 following SIVmac251 infection ( Figure 7B ) . Data were displayed by grouping the monkeys into those having KIR3DH copy numbers below and those having KIR3DH copy numbers above the median . In the naïve rhesus monkeys , the relative representation of circulating NK cells did not associate with differences in KIR3DH copy numbers . We observed a modest increase in the relative representation of NK cells in monkeys harboring a greater number of KIR3DH copies than the median on day 28 post-SIVmac251 infection ( median , 9 . 27%; range , 2 . 15–15 . 82%; n = 9 ) compared to monkeys having KIR3DH copy numbers below the median ( median , 6 . 02%; range , 1 . 34%–9 . 04%; n = 5 ) . However , this difference in circulating NK cells did not reach statistical significance ( Mann-Whitney , P = 0 . 24 ) . Since there is no available antibody for staining KIR3DH molecules , NK cells that express KIR3DH receptors on their surface can't be distinguished from those that do not . We finally assessed some aspects of the functionality of NK cells from rhesus monkeys harboring different copy numbers of KIR3DH loci . We evaluated cytokine secretion by NK cells from naïve and SIVmac251-infected monkeys in response to in vitro stimulation . PBMCs were stimulated with K562 cells and the intracellular expression of two cytokines produced by NK cells – tumor necrosis factor α ( TNFα ) and interferon γ ( IFNγ ) – was measured in the three primary NK cell subsets: CD16+ , CD56+ and DN NK cells . As expected , the CD16+ NK cells secreted little cytokine [34] , while the other NK cell subpopulations did secrete cytokines upon stimulation ( Figure 8 ) . There was , however , no obvious association between TNFα or IFNγ secretion upon stimulation and KIR3DH copy numbers by the CD56+ and DN NK cell subpopulations . This was seen in NK cells sampled from both naïve and recently infected monkeys ( Figure 8A , B and data not shown ) . We also assessed whether the surface expression of the activation-associated molecules CD69 , HLA-DR and NKp46 on NK cell subsets was associated with KIR3DH copy numbers in naïve rhesus monkeys and in the same cohort of monkeys at day 35 and at set-point following SIV infection . The intracellular levels of the proliferation-associated Ki67 molecule in NK cells during primary infection were also assessed . No association was observed between KIR3DH copy number and the expression of these molecules in CD16+ , DN or CD56+ NK cells pre- or post-infection ( Supplementary Figure S3 in Text S1 and data not shown ) .
The protective effects of particular KIRs for HIV-1 infections in humans have , for the most part , been shown in epidemiologic studies [12] , [35]–[37] . Functional studies of NK cells expressing specific KIRs have been difficult to carry out in the early phases of HIV-1 infections because the timing of HIV-1 acquisition cannot be precisely determined from the clinical histories of patients . The SIV-infected rhesus monkey therefore represents a potentially important model for studying KIR receptors expressed on the surface of NK cells and the effects of these cells on the control of viral replication during primary infection . However , the ligands of rhesus monkey KIRs are not well understood . The interaction between a particular KIR and its ligand might , however , be crucial for that KIR to modify disease outcome in HIV-1/SIV infections . In fact , some reports suggest that associations between particular KIR receptors and clinical sequelae of HIV-1 infections are only observed when the ligands of those KIRs are considered [12] , [36] , [38] . For example , coexpression of KIR3DS1 and its ligand HLA-B Bw4-80Ile alleles was associated with a delayed progression to AIDS in HIV-1-infected individuals , whereas the expression of KIR3DS1 in the absence of the HLA-B Bw4-80Ile alleles was not associated with a delayed clinical progression following HIV-1 infection [12] . In contrast to this observation , other studies in humans have demonstrated that the expression of particular KIR molecules was associated with a more favorable HIV-1 disease outcome or decreased risk of HIV-1 acquisition irrespective of the expression of the KIR ligands [37] , [39] , [40] . Also , studies of SIV-infected rhesus monkeys showed that the expression of particular inhibitory KIR3DL and KIR3DH molecules was associated with high levels of SIV replication without consideration of the ligands of those KIR molecules [20] , [33] . Consistent with the findings of these latter studies , we observed an association between KIR3DH copy numbers and peak SIV RNA levels during primary infection in monkeys that did not express Mamu-A*01 and expressed the restrictive TRIM5 alleles 1-5 without considering the contribution of specific KIR ligands . It is possible however , that the observed effect might be more profound if KIR3DH ligands were taken into consideration . The characterization of KIR3DH-expressing NK cell subpopulations could be important for clarifying the role of activating KIRs in modulating SIV replication . There is , however , no monoclonal antibody that recognizes rhesus monkey KIR3DH , and , therefore , surface expression of KIR3DH on monkey NK cells cannot be monitored . Without such an antibody , NK cells expressing high levels of KIR3DH molecules on their surface and high frequencies of KIR3DH+ NK cells cannot be distinguished . In the present studies , we used KIR3DH RNA expression as a surrogate for KIR3DH cell surface expression . The primer/probe set used in the assay to quantify KIR3DH RNA expression binds to a conserved region of the KIR3DH genes that encodes the transmembrane domain of the KIR3DH proteins . Therefore , truncated KIR proteins that have lost their transmembrane domains due to frameshift deletions and are not anchored in the cell membrane , as seen for allotypes of the human KIR2DS4 and KIR2DL4 [41] , [42] , would not be detected . We assume that higher KIR3DH copy numbers , and the resulting higher KIR3DH transcript levels , indicate an increased surface expression of activating KIR receptors on subpopulations of NK cells . The MHC class I allele Mamu-A*01 and TRIM5 alleles are genetic determinants of the control of SIV replication in rhesus monkeys [25] , [29] , [43] . A link was also reported between CCL3L CNV and AIDS progression in SIV-infected rhesus monkeys [22] . We , however , showed that CCL3L CNV was serving as a surrogate for the expression of Mamu-A*01 , and the relatively benign clinical course observed following SIV infection in certain monkeys was actually a consequence of Mamu-A*01 expression by those animals [29] . Because of these findings , it was important in the present studies to show that KIR3DH copy numbers were not acting as a surrogate marker for the Mamu-A*01 and TRIM5 status of the monkeys . In our data , CNV of KIR3DH was not associated with the expression of either Mamu-A*01 , Mamu-B*17 or the restrictive TRIM5 alleles . In addition , when Mamu-A*01 and TRIM5 were both included in models , they did not eliminate the relationship of KIR3DH copy number and peak viral load . Very few studies have evaluated CNV in KIR loci . One study showed that KIR2DS2 copies ranged from 0 to 2 copies in humans [38] . In that study , KIR2DS2 copy numbers were estimated using KIR typing rather than single gene analysis . Pelak et al . documented up to 3 copies of KIR3DS1 and 3 copies of KIR3DL1 in humans using a qPCR-based assay that was similar to the quantitative assay we used in the present studies [Pelak et al . , personal communication] . We , however , observed a wider range of KIR3DH copies in rhesus monkeys with one monkey having 12 KIR3DH copies per cell . We defined copy numbers of the KIR3DH family without evaluating individual KIR3DH genes . These results are consistent with a recent study in rhesus monkeys showing 0-4 KIR3DS genes per haplotype [19] . Both Gaudieri et al . and Pelak et al . investigated the effects of copy number variation of KIR2DS2 and KIR3DS1 on HIV-1 disease outcome . Higher copy numbers of the activating KIR2DS2 were associated with greater CD4+ T cell loss and rapid progression to AIDS [38] . Since these investigators only indirectly determined KIR2DS2 copy numbers and there is a strong linkage disequilibrium between KIR2DS2 and KIR2DL2 , the rapid HIV-1 disease progression in these individuals might not be attributable to the expression of KIR2DS2 . In the study by Pelak et al . , higher numbers of activating KIR3DS1 copies were associated with lower plasma virus RNA levels at set-point in HIV-1 infected individuals that express the KIR3DS1-ligand HLA-B Bw4-80Ile alleles [Pelak et al . , personal communication] . The findings in the present study are in line with these observations . The effect of activating KIR copy numbers on SIV replication may be more modest than the effects on SIV control mediated by the MHC class I molecule Mamu-A*01 and the restrictive TRIM5 alleles [25] , [28] , [43] . Since the KIR3DH copy number effect was only seen in the Mamu-A*01– monkeys that were homozygous for restrictive TRIM5 alleles , it is likely that a stronger Mamu-A*01-associated effect may obscure this NK cell contribution to SIV control . It is not immediately obvious why the KIR3DH copy number effect was only observed in the monkeys expressing the restrictive TRIM5 alleles . In the present studies we showed that higher numbers of KIR3DH copies were associated with lower peak viral load following SIV infection , but we did not observe associations between KIR3DH copy numbers and other clinical sequelae of SIV infection . The contributions of NK cells to the control of SIV replication may be manifested during the early stages of SIV infection . Then , during the course of SIV-infection , virus-specific CD8+ T cells expand that maintain control over viral replication throughout the chronic phase of infection [44]–[48] . The contributions of adaptive CD8+ T cells to viral replication are likely much greater than those mediated by NK cells , and these effects may simply obscure those of KIR3DH-expressing NK cells on SIV control during the later stages of infection . Further , the NK cells may become too dysfunctional later in the course of infections to mediate an antiviral effect [49]–[51] .
All of the animals used in this present study were Indian-origin rhesus macaques . All monkeys were housed in accordance with the guidelines of the NIH Guide for the Care and Use of Laboratory Animals and with the approval of the Institutional Animal Care and Use Committee of Harvard Medical School and the National Institutes of Health . SIV-challenged monkeys were either infected intravenously or intrarectally with an uncloned SIVmac251 inoculum [31] . The expression of Mamu-A*01 , Mamu-B*08 , Mamu-B*17 and the expression of TRIM5 alleles were assessed by PCR [28] , [29] . Plasma viral RNA levels were measured using an ultra-sensitive branched DNA amplification assay ( Bayer Diagnostics , Berkeley , CA ) . Counts of total peripheral blood CD4+ T lymphocytes and central memory CD4+ T lymphocytes were calculated by multiplying the total lymphocyte count by the percentage of CD3+CD4+ T cells , times the percentage of CD95+CD28+ T cells for counts of central memory CD4+ T cells , determined by monoclonal antibody staining and flow cytometric analysis [52] . Total genomic DNA was extracted from peripheral blood mononuclear cells ( PBMCs ) or H . papio-immortalized B-lymphoblastoid cell lines ( B-LCLs ) using the DNeasy Blood and Tissue Kit ( Qiagen , Valencia , CA ) . The purity of the isolated genomic DNA was verified by A260/A280 ratio: average 1 . 88 ( range 1 . 74–2 . 00 ) . The DNA samples were stored at −20°C until use . DNA integrity was verified by gel-electrophoresis of selected samples . Activating KIR copy number determinations were performed by quantitative real-time PCR ( qPCR ) using the 7300 Real-time PCR System ( Applied Biosystems , Foster City , CA ) . With KIR3DH being the only activating KIR receptor family in rhesus monkeys , a KIR3DH primer and TaqMan probe set for qPCR was designed to specifically amplify previously identified Mm-KIR3DH alleles ( GenBank accession numbers MmKIR3DH1-4 ( AF334648-AF334651 ) , MmKIR3DH7-21 ( EU702453-EU702473 ) , and MmKIR3DH-like1-4 ( AY505479-82 ) ) [17] , [18] , thereby avoiding recognition of any inhibitory KIR genes . KIR3DH-specific primer sequences were: Forward: 5′-CACCAGACACCTGCCTATTGTGA-3′; Reverse: 5′-GAGTCTCTTTTTGTCGGAGCACCA-3′; Probe: 5′-FAM-TAGGTACTCGGTGGCCACCATCAT-BHQ-3′ . qPCR products were sequenced to confirm the specificity of the amplification . The STAT6 gene , present in a single copy per haploid rhesus genome [22] , was used as an endogenous reference gene: Forward: 5′-AACCTAAAGAGAATGGGAGTGT-3′; Reverse: 5′-GAATATAGTCACAACCCTGGATC-3′; Probe: 5′-FAM-CTCTGCCCTTCTCCTGCCTCCC-BHQ-3′ . Primers were purchased from Invitrogen and probes were purchased from Biosearch Technologies . Per qPCR reaction , 12 . 5 ng of total genomic DNA were added to TaqMan Universal PCR Master Mix ( Applied Biosystems ) and the specific primers and probe . All samples were run in triplicate . Thermal cycling conditions were as follows: 2 min at 50°C , 10 min at 95°C , followed by 40 cycles of a two-step PCR of 15 s at 95°C and 1 min at 60°C . qPCR results were analyzed using the SDS v1 . 4 . 0 software ( Applied Biosystems ) . To determine absolute KIR3DH copy numbers , plasmid DNA standards for KIR3DH and STAT6 were created . The plasmids contained the specific sequence amplified in the qPCR reaction . KIR3DH standard primers were: Forward 5′-GGAGGAACCTACAGATGCTTCG-3′; Reverse: 5′-TCAGAGTCTCTTTTTGTCGGAGCAC-3′ . STAT6 standard primers were: Forward: 5′-CCTTGTCCAAACTGAGTCCAACTGC-3′; Reverse: 5′-CAGACCCAGGACCTCAGACTTC-3′ . First-strand cDNA was synthesized from RNA isolated from rhesus PBMCs using an oligo ( dT ) 20 primer and the SuperScript III First-Strand Synthesis System for RT-PCR ( Invitrogen , Carlsbad , CA ) , following the manufacturer's protocol . Then , PCR was performed using the Platinum PCR SuperMix High Fidelity ( Invitrogen ) . Amplification conditions were: 5min at 94°C; 35 cycles of 30 s at 94°C , 30 s at 55°C , and 45 s at 68°C; and 20 min at 68°C . PCR products were resolved by gel electrophoresis , excised and purified with the QIAquick Gel Extraction Kit ( Qiagen ) . Purified PCR products were ligated into the pGEM-T Easy Vector ( Promega , Madison , WI ) , resulting in pKIR3DH and pSTAT6 . Clones containing the correct insert were verified by sequencing . Plasmids were isolated and purified using the EndoFree Plasmid Maxi Kit ( Qiagen ) , and the plasmid DNA concentration was measured using a NanoDrop ND-1000 spectrophotometer ( Thermo Scientific , Wilmington , DE ) . Six serial log dilutions ( 108 – 103 copies ) of pKIR3DH and pSTAT6 plasmid DNA were used to generate standard curves by plotting CT values versus log copies of each qPCR plate . An R2 value of a standard curve of less than 99% was considered imprecise and the corresponding qPCR plate ( 96 wells ) of DNA samples was repeated . Absolute copy numbers were calculated by determining the number of KIR3DH copies per sample from the standard curve and then by normalizing against the number of STAT6 copies in the same sample . Values were multiplied by 2 to obtain copy numbers per diploid genome . To validate that the KIR3DH absolute copy numbers calculated from qPCR were accurate , we determined KIR3DH copy numbers using a multiplex ligation-dependent probe amplification assay ( MLPA ) . Two adjacent oligonucleotides per locus – one set for the KIR3DH loci and , as reference loci , two sets for EP300 ( E1A binding protein p300 ) and one set for CREBBP ( CREB-binding protein ) – were designed to contain the same primer binding sequences for later amplification . Only one KIR3DH MLPA oligonucleotide set , recognizing the same KIR3DH alleles as in the qPCR assay , could be designed due to the polymorphic nature of the KIR3DH loci . The oligonucleotide sequences are listed in the Supplementary Table S1 in Text S1 and were purchased from IDT ( IDT , Coralville , IA ) . All MLPA reagents were purchased from MRC Holland ( MRC-Holland , Amsterdam , Netherlands ) . The MLPA reaction was carried out according to the manufacturer's protocol using 100 ng of genomic DNA . For each DNA sample , the oligonucleotide sets for the KIR3DH genes and the reference loci were ligated to the DNA in a multiplex PCR and only ligated oligonucleotides were amplified using the FAM-labeled universal primer pair . Amplified PCR products differed in length and could therefore be distinguished . The amplification products were analyzed on an ABI 3130XL capillary DNA analyzer ( Applied Biosystems ) . The results were analyzed using GeneMapper Software ( Applied Biosystems ) . The final analysis of the MLPA data was carried out using Microsoft Excel software . For each sample , peak signal values for KIR3DH were normalized by the average signal of the reference probes to determine relative MLPA signals [53] . Relative MLPA signals from two different experiments using the same DNA samples formed discrete clusters corresponding to copy numbers . Absolute copy numbers of individual clusters were determined by assuming that the distance between successive clusters corresponded to 1 copy and that the copy number genotype of the first cluster corresponded to the distance from 0 divided by the average distance of successive clusters . PBMCs were sorted for CD14–CD16+ NK cells using magnetic cell sorting ( MACS Microbeads by Miltenyi Biotec ) . Total RNA was isolated from the CD14–CD16+ NK cells using the RNeasy Mini Kit ( Qiagen ) . RNA samples were stored at −80°C until use . KIR3DH RNA expression in CD14–CD16+ NK cells was determined by performing real-time quantitative reverse transcription PCR ( qRT-PCR ) using the same KIR3DH and STAT6 primers and probes used for copy number determination of DNA . Per qRT-PCR reaction , 50 ng of total RNA was added to the Taqman One-Step RT-PCR Master Mix ( Applied Biosystems ) and the KIR3DH- and STAT6-specific primers and probes . All samples were run in triplicates . Thermal cycling conditions were as follows: 30 min at 48°C , 10 min at 95°C , followed by 40 cycles of a two-step PCR of 15 s at 95°C and 1 min at 60°C . qRT-PCR results were analyzed using the SDS v1 . 4 . 0 software ( Applied Biosystems ) . Relative RNA expression was determined using the 2−ΔΔCt method ( Applied Biosystems ) . Briefly , for each sample , CT values of KIR3DH were first normalized against the CT values of STAT6 . Normalized KIR3DH CT values for each monkey were then divided by the normalized KIR3DH CT values of one reference monkey that was evaluated in each qPCR run to determine relative KIR3DH RNA expression . Total RNA , isolated from rhesus CD14–CD16+ NK cells , was used to synthesize first-strand cDNA using an oligo ( dT ) 20 primer and the SuperScript III First-Strand Synthesis System for RT-PCR ( Invitrogen ) , following the manufacturer's protocol . PCR amplification was performed using the Platinum PCR SuperMix High Fidelity ( Invitrogen ) . The primer sequences were as described by Blokhuis et al . [18] . Amplification conditions were as follows: 5 min at 94°C; 35 cycles of 30 s at 94°C , 30 s at 66°C , and 90 s at 68°C; and 20 min at 68°C . PCR products were subjected to gel electrophoresis , excised and purified with a QIAquick Gel Extraction Kit ( Qiagen ) following the manufacturer's protocol . Purified PCR products were ligated into the pGEM-T Easy vector ( Promega ) , according to the manufacturer's instructions . Briefly , ligation reactions were incubated at room temperature for one hour and then transformed into JM109 High Efficiency competent cells ( Promega ) . Between 23 and 28 insert-containing colonies per sample were sequenced and analyzed using the Sequencher Software ( Gene Codes , Ann Arbor , MI ) . Corresponding amino acid sequences were aligned using ClustalW2 Multiple Sequence Alignment [54] . Only KIR sequences that were obtained from more than one clone are reported . The novel rhesus monkey KIR cDNA sequences have been assigned the GenBank accession numbers JN613291-JN613300 . The antibodies used in this study were anti-CD8α-Peridinium Chlorophyll Protein-Cy5 . 5 ( SK1 ) , anti-CD56-Phycoerythrin-Cy7 ( N901 , Beckman Coulter , Brea , CA ) , anti-CD3-Pacific Blue ( SP34 . 2 ) , anti-CD159 ( NKG2A ) -Allophycocyanin ( Z199 , Beckman Coulter ) , anti-CD16-Allophycocyanin-Cy7 ( 3G8 ) , anti-TNFα-Fluorescein Isothiocyanate ( MAb11 ) , anti-IFNγ-Alexa Fluor 700 ( B27 ) anti-CD335 ( NKp46 ) -Phycoerythrin ( BAB281 , Beckman Coulter ) , anti-CD69-energy-coupled dye ( TP1 . 55 . 3 , Beckman Coulter ) , anti-HLA-DR-Phycoerythrin-Cy7 ( L243 ( G46-6 ) , anti-Ki67-Alexa Fluor 488 ( B56 ) , and anti-CD56-Peridinium Chlorophyll Protein-Cy5 . 5 ( B159 ) . All antibodies unless otherwise indicated were purchased from BD Biosciences . The LIVE/DEAD Fixable Aqua Dead Cell stain kit ( Invitrogen ) was used as a viability marker to distinguish live cells from dead cells in all flow cytometric analyses . All acquisitions were made on a LSR II flow cytometer ( BD Biosciences ) and analyzed using FlowJo software ( TreeStar Inc . , Ashland , OR ) . PBMCs were isolated from EDTA-anticoagulated blood by Ficoll-Paque ( GE Healthcare , Piscataway , NJ ) gradient separation and either stained immediately or cryopreserved in the vapor phase of liquid nitrogen . Later , cryopreserved cells were thawed and rested at 37°C in a 5% CO2 environment for 6 hrs . The viability of these cells was > 90% . PBMCs were stained with anti-surface MAbs to delineate NK cells ( CD3 , CD8 , NKG2A , CD56 , and CD16 ) and anti-surface molecules CD69 , HLA-DR and NKp46 . Cells were then fixed and permeabilized with Cytofix/Cytoperm solution ( BD Biosciences ) and stained with antibodies specific for Ki67 . Labeled cells were fixed in 1% formaldehyde-PBS . To determine cytokine production of NK cells in response to K562 cells , PBMC were incubated for 6 hours in the presence of RPMI 1640/10% fetal calf serum alone ( unstimulated ) , with K562 at an effector-to-target ratio of 10∶1 , or with phorbol myristate acetate ( PMA ) as a positive control . All samples contained Monensin ( GolgiStop , BD Biosciences ) and Brefeldin ( Golgi Plug , BD Biosciences ) . Cells were next stained with antibodies specific for cell surface molecules CD3 , CD8 , NKG2A , CD56 , and CD16 . Cells were then fixed and permeabilized with Cytofix/Cytoperm solution ( BD Biosciences ) and stained with antibodies specific for TNFα and IFNγ . Labeled cells were fixed in 1% formaldehyde-PBS . All data are reported after background correction . All statistical analyses and graphic analyses were conducted using GraphPad Prism ( GraphPad Prism Software , La Jolla , CA ) and STATA ( StataCorp LP , College Station , Texas ) . Normality was assessed using the skewness test , the kurtosis test , the chi-squared-test and the Shapiro-Wilks-test . The non-parametric Mann-Whitney-test was used for the comparison of two groups and the Kruskal-Wallis test was used for comparing more than two groups . Linear regression models were done for the relationship of peak viral load and KIR3DH copy number either overall ( and incorporating covariates for Mamu-A*01 and TRIM5 groupings ) or in subgroups . Subsequently , two other regression models ( parabolic , i . e . second degree polynomial , and two linear splines with a fixed knot at 5 ) were used because of previous reports that an individual NK cell usually expressed at most 5 KIRs [32] . All P values are two-sided and none are corrected for multiple comparisons . P values of <0 . 05 were considered significant .
|
NK cells are effector cells of the innate immune system that contribute to protection against virus infections through their ability to lyse virus-infected cells without prior antigen sensitization . Their role in controlling HIV-1 replication during primary infection has been uncertain . NK cell activation is regulated by inhibitory and activating KIRs that recognize MHC class I molecules expressed by target cells . In the present study , we identify an association between the copy number of activating KIR genes in rhesus monkeys and the control of SIV replication during primary infection in Mamu-A*01– rhesus monkeys that express restrictive TRIM5 alleles . This observation underscores the potential importance of activated NK cells in the control of SIV spread during the early stages of infection .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"infectious",
"diseases",
"immune",
"cells",
"nk",
"cells",
"immunity",
"innate",
"immunity",
"hiv",
"immunology",
"biology",
"viral",
"diseases"
] |
2011
|
Association of Activating KIR Copy Number Variation of NK Cells with Containment of SIV Replication in Rhesus Monkeys
|
Our actions take place in space and time , but despite the role of time in decision theory and the growing acknowledgement that the encoding of time is crucial to behaviour , few studies have considered the interactions between neural codes for objects in space and for elapsed time during perceptual decisions . The speed-accuracy trade-off ( SAT ) provides a window into spatiotemporal interactions . Our hypothesis is that temporal coding determines the rate at which spatial evidence is integrated , controlling the SAT by gain modulation . Here , we propose that local cortical circuits are inherently suited to the relevant spatial and temporal coding . In simulations of an interval estimation task , we use a generic local-circuit model to encode time by ‘climbing’ activity , seen in cortex during tasks with a timing requirement . The model is a network of simulated pyramidal cells and inhibitory interneurons , connected by conductance synapses . A simple learning rule enables the network to quickly produce new interval estimates , which show signature characteristics of estimates by experimental subjects . Analysis of network dynamics formally characterizes this generic , local-circuit timing mechanism . In simulations of a perceptual decision task , we couple two such networks . Network function is determined only by spatial selectivity and NMDA receptor conductance strength; all other parameters are identical . To trade speed and accuracy , the timing network simply learns longer or shorter intervals , driving the rate of downstream decision processing by spatially non-selective input , an established form of gain modulation . Like the timing network's interval estimates , decision times show signature characteristics of those by experimental subjects . Overall , we propose , demonstrate and analyse a generic mechanism for timing , a generic mechanism for modulation of decision processing by temporal codes , and we make predictions for experimental verification .
It is likely that the cerebral cortex evolved to provide a model of the world , serving decisions for action . Our actions take place in space and time and both of these dimensions are considered in the dominant hypothesis of decision making , where noisy spatial evidence is averaged over time ( see [1]–[3] ) . The longer we spend averaging , the more accurate our decisions [4] . A trade-off between speed and accuracy is implicit in this framework and is a hallmark of decision tasks [5] , but the mechanism by which we determine how long to spend averaging is an open question [6] . In recent years , there has been increasing acknowledgement that the encoding of time may be as crucial to behaviour as the encoding of space [7] and several studies have considered roles for temporal codes in decision making [8]–[11] . Under this approach , time is not a passive medium for spatial averaging , but is actively encoded during decisions , determining the rate at which they unfold . Accordingly , the speed-accuracy trade-off ( SAT ) can be controlled by the estimation of temporal intervals , determining how long spatial evidence is integrated [11] . Our ability to represent time covers at least twelve orders of magnitude , from the scale of microseconds to circadian rhythms , and different neural mechanisms are believed to support representations of vastly different temporal duration [12] , [13] . Here , we focus on the hundreds of milliseconds range , the relevant order for the most well studied perceptual decision tasks [1] , [3] , [14] . Two fundamental questions in the study of temporal processing are whether the representation of time is centralized or distributed [15] , [16] , and whether the circuitry involved is specialized or generic [17] , [18] . In this paper , we propose that local-circuit cortical processing is inherently suited to the representation of space and time on this order , supporting a distributed , generic processing framework . To this end , we demonstrate that a generic biophysical model of a local cortical circuit can estimate time in the hundreds of milliseconds range , where ‘climbing’ activity resembles that seen in cortex during tasks with a timing requirement and estimates of temporal intervals show signature characteristics of temporal estimates by experimental subjects . The network estimates different intervals by the scaling of a single term controlling local-circuit dynamics by the strength of NMDA receptor ( NMDAR ) conductance . Analysis of network dynamics formally characterizes this timing mechanism and a simple learning rule is sufficient for the network to quickly learn the intervals . In simulations of a decision task , we couple two such generic networks with identical parameters except for the NMDAR scale factor . One network encodes elapsed time relative to a learned interval . The other decides which of two stimuli has more evidence . As climbing activity evolves in the timing network , it governs the rate of downstream decision processing by gain modulation . To trade speed and accuracy , the timing network simply imposes different temporal constraints on the decision network . The model's activity and behaviour are consistent with a large body of electrophysiological and behavioural data from timing and decision tasks , as well as the hypothesis that cortical circuitry is canonical ( see [19] ) . In our opinion , these results should be expected of a generally uniform structure that evolved to provide a model for action in a spatiotemporal world .
The local circuit model is a fully recurrent network of leaky integrate-and-fire neurons [36] , comprised of simulated pyramidal neurons and fast-spiking inhibitory interneurons . For pyramidal-to-pyramidal synapses , is a Gaussian function of the distance between neurons arranged in a ring . The weight between any two pyramidal neurons and is therefore given by ( 1 ) where defines distance in the ring , is a scale factor , and , depicted on the right side of Figure 1A . The biological basis of is the probability of lateral synaptic contact between pyramidal neurons , generally found to be monotonically decreasing over a distance of in layers 2/3 and 5 [37] , [38] . Width parameter therefore corresponds to approximately axially in cortical tissue , consistent with cortical tuning curves [39] , [40] . Like earlier authors ( e . g . [21] , [24] ) , we do not attribute biological significance to the spatial periodicity of the network; rather , this arrangement allows the implementation of with all-to-all connectivity without biases due to asymmetric lateral interactions between pyramidal neurons . Synaptic connectivity from pyramidal neurons to interneurons , from interneurons to pyramidal neurons , and from interneurons to interneurons is unstructured in the network , so for each of these cases , for all and . Each model neuron is described by ( 2 ) where is the membrane capacitance of the neuron , is the leakage conductance , is the membrane potential , is the equilibrium potential , and is the total input current . When reaches a threshold , it is reset to , after which it is unresponsive to its input for an absolute refractory period of . For pyramidal neurons , , , , , and . For interneurons , , , , , and [21] . Excitatory currents from pyramidal neurons were mediated by AMPA receptor ( AMPAR ) and NMDAR conductances , and inhibitory currents from interneurons were mediated by GABA receptor ( GABAR ) conductances ( Figure 1B ) . The total input current to each neuron is given by ( 3 ) where , and are the summed AMPAR , NMDAR and GABAR currents from intrinsic ( recurrent ) synapses , and is background noise , described below . These intrinsic currents are defined by ( 4 ) where , and are the respective strengths of AMPAR , NMDAR and GABAR conductance , is the reversal potential for AMPARs and NMDARs and is the reversal potential for GABARs [21] . AMPAR and GABAR activation ( proportion of open channels ) are described by ( 5 ) where is the Dirac delta function and is the time of firing of a pre-synaptic neuron . For NMDAR activation , has a slower rise and decay and is described by ( 6 ) where controls the saturation of NMDAR channels at high pre-synaptic spike frequencies [21] . The slower opening of NMDAR channels is captured by ( 7 ) where was set to [21] . The voltage-dependence of NMDARs is captured by , where describes the extracellular Magnesium concentration and is measured in millivolts [41] . The scale factor is described below . The time constants and conductance strengths of AMPAR , NMDAR and GABAR synapses onto cortical pyramidal neurons and inhibitory interneurons vary according to tissue preparation , recording method , species , cortical layer , and to some degree , cortical area within a species or layer . En masse , electrophysiological data provide reasonable guidelines for these parameters , but we emphasize that nothing in the model is fine tuned and our results hold for a broad range of parameters . For AMPAR-mediated currents , and at synapses onto pyramidal neurons , and and at synapses onto interneurons , producing fast-decaying monosynaptic AMPAR currents on the order of [42] , [43] that are stronger and faster onto inhibitory interneurons than pyramidal neurons [44]–[46] . For NMDAR-mediated currents , and at synapses onto pyramidal neurons , and and at synapses onto interneurons , producing slow-decaying monosynaptic NMDAR currents on the order of [42] , [47] that are stronger and slower at synapses onto pyramidal neurons than interneurons [46] . Our excitatory synaptic parameters thus emphasize fast inhibitory recruitment in response to slower excitation ( see [48] for discussion ) . For GABAR-mediated currents , and at synapses onto pyramidal neurons and and at synapses onto interneurons [49] , [50] , producing monosynaptic GABAR currents several times stronger than the above excitatory currents , where stronger conductance onto pyramidal cells was meant to reflect the greater prevalence of GABAR synapses onto pyramidal cells than interneurons [51] . See Figure 1B for exemplary synaptic currents in the model . In both simulated tasks , performance was determined by the mean activity of localized populations of neurons . Spike density functions ( SDF , rounded to the nearest millisecond ) were therefore built for these neurons by convolving their spike trains with a rise-and-decay functionwhere t is the time following stimulus onset and and are the time constants of rise and decay respectively [53] . In the interval estimation task , it was necessary to first identify the relevant population ( the bump population ) before averaging its activity . To this end , we built SDFs for all pyramidal neurons in the network and the neuron with the highest mean SDF over the full trial was considered the centre of the bump . We included neurons on either side of this centre in the bump population . In the decision task , the centres of the response fields for the competing stimuli were pre-determined , so SDFs were constructed for these neurons , as well as the neurons on either side . All simulations were run with timestep and the standard implementation of Euler's forward method . To investigate the mechanism by which the timing network produced climbing activity , we simplified the network to an equivalent integral and partial differential system using a Wilson-Cowan type approach [27] , [54] . We then used methods for the study of non-linear dynamics and stochastic processes to analyse the reduced system . Because pyramidal-to-pyramidal synaptic connectivity is structured and all synaptic connections with interneurons are unstructured , the firing rate of pyramidal neurons and interneurons can be modelled as ( 13 ) and ( 14 ) where ( as above ) denotes spatial location and is the activation function ( 15 ) with gain factor and noise factor . The synaptic current at pyramidal neurons at location is and consists of AMPAR- , NMDAR- and GABAR-mediated synaptic currents and background current , i . e . . The first three of these currents can be approximated as ( 16 ) where , and describe the effective synaptic strength . Superscripts ‘r’ denote the correspondence of terms in the ‘reduced’ system with those in the timing network . The synaptic currents onto interneurons are similar . Synaptic activation is described by , and , obeying the dynamics ( 17 ) Because the NMDAR time constant is much longer than the respective time constants of AMPARs , GABARs and neuronal firing rates ( Equations 13 and 14 ) , these last three variables can be given their steady state values , while NMDAR activation dominates the dynamics of the system . Thus , the system can be described by ( 18 ) By further linearizing the activation function of the interneurons , we obtain the integral and partial equation to approximate the timing network: ( 19 ) The chosen parameters were ( ) , , , , , , , , , , , . Scale factors , and were used to tune the model to qualitatively reproduce the steady state firing rates in the timing network , while abstracts over the unstructured interactions between pyramidal neurons and interneurons .
We simulated an interval estimation task with the generic local-circuit model . Estimates of different intervals were produced by scaling the conductance strength of NMDARs by , where different values of supported different rates of buildup of activity by the bump population . On each trial , the time at which the mean SDF of the bump population reached a threshold of was considered the interval estimate . Thus , like earlier authors ( e . g . [56] , [57] ) , we assume that a behaviourally relevant ballistic process is initiated downstream when neural activity encoding the interval estimate reaches a certain firing rate . The task was simulated by running the network for , sufficient time for a bump to develop for all of the above values of on at least of trials . This length of time may seem long for estimates in the hundreds of milliseconds range , but for longer interval estimates , it allowed for the growing variability of estimates with interval duration , commonly seen in interval estimates in experimental tasks ( see Results section The scalar property of interval timing ) . For , bumps did not consistently develop within the allotted time , and for those that did , spiking activity did not consistently reach . For , background spiking was approximately among pyramidal neurons and among interneurons [58] , but climbing activity was not supported by the network . An upper limit of was used because it is consistent with the experimental enhancement of the NMDAR component of cortical excitatory post-synaptic currents by approximately of baseline [59] , [60] and because interval estimates were increasingly indistinguishable above this value . trials were run for each value of . Varying the scale factor furnished a range of rates of buildup activity , where lower values of lead to slower buildup and higher values lead to faster buildup . The lowest value of that consistently supported buildup activity produced a mean interval estimate of ( ) . The highest value of consistent with experimental enhancement of cortical EPSCs [59] , [60] produced a mean estimate of ( ) . The timing network thus supported interval estimates from approximately to , consistent with experimental evidence that temporal coding on this order is supported by a common mechanism [12] , [55] . Example trials for three values of are shown in Figure 2 . Note that the location of the bump differs on each trial , as there is no bias favouring a particular network location . We are unaware of any data to conclusively confirm or refute such trial-to-trial variability , but to produce climbing activity in the same sub-population from trial to trial , we simply need to strengthen excitatory synaptic conductances among a few localized neurons , e . g . by Hebbian long term potentiation among the neurons participating in the bump . In the coupled-circuit decision trials in Section Encoding time constraints for a decision , the location of climbing activity in the timing network does not matter because projections from the timing network to the decision network are spatially non-selective . The mechanism underlying climbing activity in the timing network can be understood by non-linear analysis of the reduced integral and partial differential system . For a given value of the effective synaptic strength , corresponding to NMDAR conductance strength in the timing network , the steady states of the reduced system can be calculated by setting the right hand side of Equation 19 to zero and solving the resulting equations . Our analysis revealed three regimes of the reduced system . 1 ) Sufficiently small values of furnished a flat steady state which is stable and whose eigenvalues are negative . This regime corresponds to the common case in cortex , where background activity is stable and stimulus-selective activity decays to this background state after stimulus offset . This regime in the reduced system corresponds to approximately in the timing network . 2 ) With a moderate increase in , the system enters a bistable regime . The stable flat steady state is retained , but a small unstable bump steady state and a large stable bump steady state emerge . This bistable regime corresponds to the classic persistent storage regime in these networks ( e . g . [21] , [26] ) , in which a stimulus can trigger a bump state , which persists after stimulus offset . This regime in the reduced system corresponds to approximately in the timing network . 3 ) With a further increase in , the stable flat state and the unstable bump steady state coalesce into one unstable flat steady state whose largest eigenvalue is positive , while the stable bump state becomes higher . The magnitudes of the unstable flat steady state and the stable bump state increase with further increase to . This regime in the reduced system corresponds to approximately in the timing network . This third regime is shown in Figure 3 , where panels A and B show NMDAR activation at the stable bump state and the unstable flat steady state respectively with increasing . Panels C and D show the corresponding firing rates . The instantaneous firing rates of the stable bump states in the reduced system were consistent with the steady state spike rates in the timing network , ranging from approximately to as was increased from to , corresponding to an increase in from to in the timing network ( Figure 3C ) . The evolution of the system away from the unstable flat steady state is dominated by the largest positive eigenvalue and a localized activity bump emerges due to the corresponding eigenvector ( Figure 4 ) . On the other hand , the largest eigenvalue of the stable bump steady state is zero and the corresponding eigenvector explains the invariant location of the bump [61] . Climbing activity therefore occurs at arbitrary locations , as shown in Figure 2 above . Not only did the timing network produce interval estimates , but the estimates conformed to the scalar property of interval timing [62] . The scalar property is a strong form of Weber's law where the standard deviation of estimates is proportional to the mean ( Figure 5A ) . Weber's law is widely regarded as a signature characteristic of interval timing across a wide range of temporal orders [13] , [55] , though see [63] and [64] for a systematic description of conformities and violations of the scalar property in humans and non-human animals respectively . The coefficient of variation ( CV , the standard deviation divided by the mean ) of the interval estimates produced by the timing network was approximately constant ( Figure 5B ) and compared favourably to experimental measurements on this order [55] , [65] . The distribution of interval estimates for each value of was roughly normal ( Figure 6A ) , another widely-reported characteristic of interval estimates across a range of temporal orders [13] , [55] . Gaussian fits to the estimates are shown in Figure 6B . For comparisons with experimental data in the hundreds of milliseconds range , see e . g . [66] and [67] . Climbing activity in the timing network can be understood as the evolution of the system from an initial state in the vicinity of the unstable flat steady state to the stable bump state . We linearized the system in the vicinity of the unstable flat steady state as , where denotes for simplicity . To consider the effects of noise , the linear system can be expressed as a Langevin equation , where is the eigenvalue of the matrix , is a vector , and is white noise with standard deviation . According to non-linear dynamics , the system expands along the manifold tangent to the eigenvector with positive eigenvalue . Thus , we focus on the largest positive eigenvalue , which dominates the expansion of the system [68] , and further simplify the system as a 1-dimensional Ornstein-Uhlenbeck ( OU ) process ( 20 ) The parameter of an OU process is typically negative , supporting a stable distribution . Here , is positive because the flat steady state is unstable . Thus , positive implies that the system departs from the flat state starting at initial state . The corresponding Fokker-Plank equation can be written as ( 21 ) with initial state . The distribution of arrival times of at the timing threshold can be calculated as ( 22 ) which shows that the system grows along the curve with standard deviation . Intervals are estimated when the system reaches the threshold , so the interval estimates are the first passage times of the OU process , the distribution of which can be calculated as ( 23 ) with mean and variance given by ( 24 ) and ( 25 ) respectively . Of note , and are the initial value and threshold of NMDAR activation , ( , Equation 19 ) , not the firing rate . To calculate the distribution of first passage times numerically , we need to express the threshold in terms of . For each value of , we therefore calculated by scaling the interval estimation threshold in the timing network by the ratio of the maximum values of and , i . e . . This scaling preserves our use of a fixed firing rate threshold in the timing network . Values for , and the largest positive eigenvalue are given in Table 1 for increasing . We used a constant level of background noise for all simulations ( ) , consistent with our use of constant parameters with the background noise ( point-conductance ) model across the different values of in the timing network . The distribution of first passage times is shown in Figure 7 . These curves are very similar to the distribution of interval estimates by the timing network ( compare Figures 6B and 7A ) . Stronger ( weaker ) NMDAR conductance causes faster ( slower ) ramping and a narrower ( wider ) distribution , while the relationship between the mean and standard deviation is approximately linear ( compare Figures 5A and 7B ) . The previous sections demonstrate that the timing network estimates intervals in the hundreds of milliseconds range as a function of the scale factor and that these estimates share signature characteristics with those of experimental subjects in studies of interval timing . Next , we consider whether the network can learn a given interval in this range , using a simple learning rule [57] . We ran the interval estimation task ( described above in Results section Interval estimates are controlled by NMDAR conductance strength in the timing network ) for desired intervals . For each desired interval , the network began the block of trials in the baseline condition ( ) and was adjusted after each trial according to ( 26 ) where is the estimate of on trial and determines the rate of learning . As shown in Figure 8 , the network learned each interval after a handful of trials [7] , [32] . Estimates of elapsed time occur relative to a start time , so the network requires a start signal to begin each estimate . Such a signal should be able to reset the network to the background state , regardless of its current state . There are a number of plausible mechanisms that could play this role . We demonstrate two such mechanisms . One is a brief pulse of spatially non-selective excitation , generating blanket feedback inhibition and thus shutting the network down . This mechanism was demonstrated in an earlier study simulating persistent mnemonic activity in prefrontal cortex , using a local-circuit model similar to ours [21] . Figure 9A shows this mechanism in the timing network , where the average excitatory conductance of the point conductance model ( ) at pyramidal neurons was increased by a factor of at all pyramidal neurons for to start the estimate , and again at time to stop the estimate . In this case , the start signal potentially corresponds to broad excitation of the timing network by a cue stimulus , while the stop signal potentially corresponds to efference copy at the time of motor initiation [22] . As such , we do not expect these signals to be identical in duration or magnitude , but giving them the same parameters shows robustness of the mechanism ( fine tuning of each signal was not necessary ) . Electrophysiological data showing a similar trajectory can be seen in e . g . [69] , where these data were interpreted as encoding the anticipation of an upcoming stimulus . Another plausible reset mechanism is long-range excitatory targeting of inhibitory interneurons , which in turn inhibit local pyramidal neurons [70] . Such disynaptic inhibition has been suggested to underlie the control of motor initiation in anti-saccade [71] and countermanding tasks [72] and is simulated in the timing network in Figure 9B . In this simulation , the average excitatory conductance of the point conductance model ( ) at pyramidal neurons was increased by a factor of for to start the estimate , and the average excitatory conductance at interneurons was increased by a factor of for at to stop the estimate . Similar electrophysiological data can be seen in e . g . [73] , interpreted as encoding the anticipation of an upcoming stimulus in their study . There are , of course , other mechanisms that could start and stop estimates of elapsed time by climbing activity . Indeed , we do not expect cortical timing circuits to remain indefinitely in a regime with no stable background state . For example , at the onset of a cue stimulus , fast-acting neuromodulation could alter network dynamics in a manner similar to the scaling of , or cortico-thalamo-cortical disinhibition could have a similar effect . To address the hypothesis that the encoding of elapsed time controls the speed and accuracy of decisions by gain modulation [11] , we ran further simulations to determine if the timing network's temporal estimates could control the SAT in a downstream network during a decision task ( Figure 10 ) . As indicated above , the two networks were identical except for the inputs they received and the scale factor . To emphasize the role played by the timing network in these simulations , was given a low value in the decision network ( , one quarter of the baseline NMDAR conductance in the timing network for synapses onto pyramidal neurons and interneurons ) so its intrinsic processing was too weak to make decisions across all task difficulties without spatially non-selective input from the timing network . Note that this low value of did not support climbing activity in the absence of selective input . Down-scaling NMDARs was thus a practical means of limiting the decision network's processing capability . Although we do not assign it a specific biological correlate , we note that the properties of NMDARs can show marked variation between cortical regions [74] and under receptor modulation within a single region [60] , [75] . We simulated a two-choice decision task by providing two noisy stimuli to the decision network for . On each trial , the network's task was to distinguish the higher-rate input ( the target ) from the lower-rate input ( the distractor ) . For each stimulus , independent , homogeneous Poisson spike trains were provided to all pyramidal neurons in the decision network , where spike rates were drawn from a normal distribution with mean corresponding to the centre of a Gaussian response field defined by . Constants and are given above for the pyramidal interaction structure ( Methods section The network model ) . For the target stimulus , we simulated upstream stimulus-selective neurons firing at each by setting and setting extrinsic AMPAR and NMDAR conductance strength to and respectively , trading spatial summation for temporal summation [76] . The distractor stimulus was similarly defined , where task difficulty ( target-distractor similarity ) was determined by multiplying by . On each trial , the decision was considered correct ( incorrect ) when the mean SDF of the target-selective ( distractor-selective ) population reached a threshold of . As with the interval estimation task , the threshold assumes a downstream ballistic process is initiated when neural firing reaches a certain rate , an assumption supported by a large body of experimental and theoretical work from decision tasks ( see [1] , [3] , [77] , [78] . The time of threshold-crossing was considered the decision time . Note that the precise value of the threshold was not crucial . Gain modulation of the decision network by the timing network was implemented by spatially non-selective excitation [24] , [33] , that is , each pyramidal neuron in the decision network received input from all pyramidal neurons in the timing network for the entirety of each trial . Only AMPAR conductances mediated these inter-network inputs , which were set to one fifth the strength of extrinsic AMPARs . The total input current to each neuron in the decision network was therefore ( 27 ) where and mediate stimulus-selective inputs ( set to 0 for interneurons ) , and mediates spatially non-selective inputs from the timing network ( set to 0 for interneurons ) . NMDAR and AMPAR activation at these synapses follows Equations 5 and 6 above . A block of decision trials ( trials for task difficulties ) was run for values of learned by the timing network for a short ( ) and a long ( ) interval ( Section Learning interval estimates above ) . The mean value of over the last trials was used in each case . Tight and loose temporal constraints were thus imposed on the decision task by running the timing network with and respectively on each block of trials , where activity in the timing network served as a spatially non-selective input to the downstream decision network . In both temporal conditions , the model was very accurate on the easiest task ( mean target-distractor similarity ) and performed at chance when the inputs were indistinguishable on average ( mean target-distractor similarity ) . For task difficulties in between , decisions were more accurate with the longer temporal estimate . Decision times were shorter for all task difficulties with the shorter temporal estimate . The coupled-circuit decision model thus traded speed and accuracy as a function of a learned interval ( Figure 11 ) .
Different neural mechanisms are expected to code for widely varying temporal durations , ranging from microseconds to days [12] , [13] . While considerable overlap between mechanisms is expected at timescales in between , it has been proposed that a dedicated mechanism exists for the hundreds of milliseconds range ( see [55] , [67] ) , the relevant order for the most well-studied perceptual decision tasks [1] , [3] , [14] . These proposals are based on the premise that a single mechanism encoding time for different tasks and modalities will show common variability in these different contexts . For example , [55] suggested a dedicated timing mechanism in this range based on pooled data from a variety of tasks and species showing a similar CV from approximately 200 to 1500 ms ( much like Figure 5B ) . Along similar lines , [82] described a constant Weber fraction for 200 to 2000 ms . [83] used the slope analysis method to distinguish timing-based variability from non-timing sources of variability , such as variability due to perceptual and motor processing during timing tasks . Under this approach , the slope of the linear fit to the variance plotted over the square of interval durations reveals the time-dependent process , shown in their study to be similar for intervals from 325 to 550 ms in two timing tasks . [67] showed a common slope under this method for several auditory tasks requiring interval estimates from 350 ms to 1 s , though they also showed significantly different slopes for other auditory tasks , visual tasks , and between auditory and visual implementations of the same task . See their study for a more extensive description of the evidence for and against a common timer in this range . In consideration of the above , it is important to distinguish between a common mechanism for timing and a common timer . A common timer refers to a ‘central clock’ processing time across a set of modalities and tasks . Inherent in this definition is a common mechanism , but a common mechanism does not necessarily imply a common timer . We propose that the capability to code time in the hundreds of milliseconds range is a generic property of local cortical circuits under conditions supporting strong attractor dynamics , but this capability does not imply that any single circuit should code time for all tasks and modalities , nor that all local cortical circuits should code time . Our model fits a distributed processing framework , with local circuits coding time in various cortical regions across different tasks and modalities , supported by inherent properties of local-circuit cortical processing [95] , [99] . There is considerable debate about the strength of evidence supporting distributed vs . central timers for different temporal durations , modalities and tasks ( see [15] , [17] , [90] , [100] ) , but the growing volume of neural data showing climbing activity in the hundreds of milliseconds range in different cortical regions during tasks with a timing requirement provides strong support for a distributed framework . For example , climbing activity in this range has been recorded in several regions of prefrontal cortex , including lateral [69] , [73] , anterior cingulate [84] , [101] and premotor [85] regions in anticipation of upcoming events; as well as in parietal areas 7A [86] and LIP [9] , [87] , [102] . Similar activity has been recorded in anticipation of reward in primary visual cortex [88] , in the timing of movements in the absence of environmental cues in LIP [89] , and in the midbrain superior colliculus during predictable delays [103] . Furthermore , many of these data showed a phasic response at the start of the anticipatory periods , suggesting a reset mechanism ( Figure 9 ) that would allow for the encoding of elapsed time relative to a start time ( e . g . [69] , [73] , [88] , [89] ) . It is important to note , however , that in a hierarchical cortical framework , one or several local circuits could conceivably encode time at the top of the hierarchy for use in more peripheral processing , i . e . despite favouring a distributed framework , we acknowledge that our model could also support a centralized framework . We have focused on the representation of time in the hundreds of milliseconds range by climbing activity , but climbing activity is not limited to the hundreds of milliseconds range , nor is climbing activity the only neural data indicative of temporal coding on this order . Such activity is generally regarded as ‘prospective’ coding [73] , [104] , i . e . neural activity encoding elapsed time in anticipation of an upcoming stimulus or in the timing of an upcoming action . In addition to the range of hundreds of milliseconds , such activity has been recorded in several cortical and subcortical regions in the range of a few seconds , including primary motor and premotor [105] and prefrontal [106] , [107] cortices , as well as the thalamus [104] . Some of these data follow a very similar trajectory to those in the hundreds of milliseconds range and a clear distinction between the neural mechanisms coding for similar ranges is not expected [12] . Under our parameters , mean estimates are limited to approximately , but alternative parameters may furnish estimates of several seconds , potentially unifying some of these data in terms of their underlying mechanism . Descending neural activity is arguably just as common as climbing activity and is generally regarded as ‘retrospective’ coding ( see [7] , [73] , [108] ) , i . e . neural activity encoding the time since some previous occurrence . In the context of interval timing , the relevant interval can be estimated when the descending activity reaches baseline from an elevated firing rate , initiated by a stimulus ( see [32] , [96] ) . Such activity has been recorded in many of the same cortical regions as climbing activity , often in the same experiments ( e . g . [73] , [88] , [101] , [107] ) . Other neural data suggest alternative mechanisms for temporal coding in the hundreds of milliseconds range , for example , [109] recorded phasic neural activity in prefrontal cortex , where the firing rate correlated with the length of a preceding interval . In the range of a few seconds , [110] showed that interval recognition and production were predominantly mediated by different PFC neurons , where the latter showed an increase and subsequent decrease in firing rate before subjects indicated their temporal estimates . A number of neural models have offered mechanistic explanations for temporal coding in recent years . These models can be distinguished along several dimensions , including the temporal range for which they code , the neural and behavioural data for which they provide a mechanistic explanation , the characterization of their dynamics , and whether they function by intracellular or network mechanisms . Most models addressing temporal coding in the range of hundreds of milliseconds to a few seconds are grounded in ascending ( climbing ) or descending ( decaying ) neural activity . Several such models have encoded time by neural decay [92] , [96] , i . e . if a stimulus elicits a neural response , then the time after stimulus offset is implicit in the level of activity remaining . Regenerative mechanisms such as recurrent synaptic processing yield network time constants much longer than the time constants of contributing biophysical processes , such as those of membranes and synapses [11] , [30] , [31] , so these temporal codes are not limited to the tens of milliseconds range . Stronger ( weaker ) intrinsic synapses thus support longer ( shorter ) temporal estimates , which can be learned by synaptic plasticity [96] . With sufficiently strong recurrent processing , neural activity becomes stable after stimulus offset , i . e . attractor states are supported and timing by neural decay is no longer possible . Just outside the attractor regime , however , intervals can be estimated by the time of collapse of quasi-stable activity [93] , including compliance with the scalar property [111] . Like our model , this approach makes use of quasi-bistable dynamics , where time is coded by the time of transition from one state to another . The obvious difference between these collapsing-activity models and our model is the direction of state change , but another important difference is the rate of state transitions by individual neurons , described below . Neural models have simulated ascending activity by a variety of mechanisms , most of which involve attractor dynamics in one form or another . For example , attractor dynamics enable a stimulus-selective population to store the representation of a start cue after its offset , and several models have used such stable , persistent activity as a source of input for the production of climbing activity in a downstream excitatory population . Different models have demonstrated different mechanisms to produce the climbing activity , including slow integration by recurrent synaptic processing [112] , short term facilitation at feedforward and recurrent synapes onto excitatory neurons [113] , and short term depression at feedforward synapses onto inhibitory interneurons that project to the downstream excitatory population , providing gradual disinhibition [94] . The idea that time can be estimated by integrating regular neural activity has long been deployed in clock-counter models ( a . k . a . pacemaker-accumulator models ) , including recent models proposing neural correlates for the required components: an oscillator to provide regular pulses , an integrator to count them , a store to hold a sample interval in memory for comparison with an evolving estimate , and a gate for starting and stopping the timing process ( see [114] for extensive review ) . These models have commonly addressed intervals in the seconds to minutes range and we do not further discuss them here . Suffice to say , persistent mnemonic activity plays a comparable role to the oscillator in models that estimate intervals by the integration of this input to produce climbing activity . The level of background input has been used to similar effect , where stronger ( weaker ) input produces climbing activity with a steeper ( shallower ) slope , reaching the threshold sooner ( later ) and thus estimating shorter ( longer ) intervals [57] , [115] . This mechanism pre-supposes an additional upstream time-sensitive mechanism to govern the strength of input , but differential rates of persistent mnemonic activity in parametric working memory tasks [107] and task-dependent modulation of background cortical spike rates [116] suggest that such a mechanism is plausible . Alternatively , different rates of climbing activity can be produced with a constant mean input if the input variance is integrated [117] or by modulation of recurrent network dynamics , as is the case here . Providing persistent inputs to integrators is not the only role played by attractor dynamics in models that estimate intervals by climbing to threshold . Indeed , the integrators often utilize attractor dynamics . These models can be differentiated by their number of stable or quasi-stable states and the transitions between them . In neural models incorporating cellular bistability , individual neurons switch rapidly from a down-state , characterized by low rate spiking activity or membrane potential , to an up-state ( high rate or membrane potential ) when triggered by sufficiently strong input current . Climbing activity occurs if the probability of switching increases with the number of up-state neurons [57] , where excitatory recurrent processing creates an avalanche effect in network models [115] . As such , these models assume that climbing activity does not reflect a true gradient of spike rates , but rather , reflects the average of a population of binary neuronal states . This characterization of climbing activity is different from that of our model , where individual neurons in the bump population traverse a gradient of firing rates while climbing to threshold . We confirmed this gradient with the interspike interval ( ISI ) analysis in [103] , where graded firing rates are revealed by a decreasing mean ISI over time and a positively-skewed , unimodal ISI distribution . For all values of the scale factor , all of these conditions were satisfied in the model ( not shown , uni-modality determined by Hartigan's dip test ) . See [103] for details of the analysis . Our model thus makes the testable prediction that neurons undergoing climbing activity during tasks with a timing requirement will show gradual increases in their firing rates , rather than abrupt transitions . Like our model , the single-cell model by [56] codes time by climbing to threshold with a true gradient of firing rates . Unlike our model , Durstewitz's model estimates intervals from hundreds of milliseconds to tens of seconds , doing so by an intracellular positive feedback loop between firing rate , voltage-gated calcium influx , and calcium-dependent inward depolarizing current . As described above , fine tuning of this feedback permits firing rate stability along a continuum of rates . Climbing activity is produced when positive feedback provides slightly more current than is required for stability , continually tipping the balance toward higher rates . The amount by which feedback current exceeds the stabilizing current determines the slope of the climbing activity , learned by plasticity at recurrent synapses in the model . The models by Buonomano and colleagues [95] , [99] are based on the premise that the encoding of time in the hundreds of milliseconds range is an inherent property of local cortical circuits , much like our model , but the mechanism by which they do so is very different . Their models do not produce ascending or descending activity , but rather , time is coded implicitly by network state . The long time constants of short term synaptic facilitation and depression in particular ( up to ) allow the network to reflect its recent history by its current state , which can be readout downstream . This approach provides an elegant solution to the challenge of recognizing rapidly-changing temporal patterns , though it is not clear whether this mechanism could be used for production of temporal estimates . Conversely , our model readily estimates intervals , but is presumably ill-suited to rapid temporal pattern recognition , required , for example , for understanding speech and appreciating music . The synfire chain [118] , [119] models by [93] , [120] offer another generic-network approach , consistent with a distributed temporal coding framework . In these models , time is coded by the duration of activity propagating through the network , where output cells learn to fire at expected times by synaptic plasticity . Such a timing mechanism could conceivably recognize and produce interval estimates . The above neural models are considered along these and other dimensions in Table 2 . See the table caption for further explanation . As a final point on generic circuitry , it seems reasonable to expect a generic mechanism to support multiple functions , such as our demonstration of timing and decision tasks , but we do not expect a single mechanism to execute all possible tasks . While our biophysical network model can be described as a basis function network [28] , it is not a universal function approximator [121] , [122] . A number of earlier models have demonstrated the scalar property under different assumptions and mechanisms than our model . In the accumulator network by [123] , noisy firing by linear spiking neurons is precisely balanced by random fan-out connectivity , where the scalar property emerges from the Gamma distribution of spiking activity . [124] investigated a multi-cascade structure with multiple memoryless states , i . e . a Markov Chain , proving that the system achieves optimal reliability if each state is sequentially and irreversibly activated and generates equal information . In the model by [125] , population climbing activity was modelled by an opponent Poisson process , where the scalar property results from the inverse Gaussian distribution of first passage times . The scalar property was also derived in terms of the first passage problem by [111] , where in a fully recurrent network of bistable units , the probability of transition from the ‘on’ state to the ‘off’ state increases until network activity collapses . In the models by [57] , the scalar property results from the exponential distribution of the transition from the ‘off’ state to the ‘on’ state of bistable units , where the probability of transition is inversely proportional to the duration of the interval being estimated . An information-theoretic framework for classifying timing by a stochastic process is provided by [126] , where timing mechanisms that are based on the mean , variance and correlation of the process predict different characteristics of timing errors . In our model , the interval estimation process is the evolution of network activity to the threshold from an initial state close to the unstable flat steady state . Stronger NMDAR conductance leads to a smaller difference between the initial state and the threshold ( see Table 1 ) and to a larger positive eigenvalue of the unstable flat steady state , producing shorter interval estimates . During the evolution from the initial state to the threshold , noise causes the system to fluctuate around a mean trajectory , where higher levels of noise cause a greater deviation . If the level of noise is independent of the synaptic conductance strength , the deviation will be greater with longer evolution times , so longer estimates will be more variable . Our analysis in Section The scalar property of interval timing reveals Weber's law as a necessary property of any timing mechanism that can be expressed as a 1-dimensional OU process with positive drift coefficient ( Equation 20 ) , given three constraints: ( 1 ) the timing threshold is reduced with increasing feedback excitation , ( 2 ) feedback excitation is linear , and ( 3 ) the noise in the system is independent of feedback excitation . The first constraint follows from the dominance of NMDARs in local-circuit processing , ie . the state variable of the reduced system is NMDAR activation , which varies much less with increasing NMDAR conductance than the resulting firing rates . Therefore , with increasing conductance , reducing the timing threshold for NMDAR activation preserves the fixed firing threshold in the timing network . The second constraint is common to many analytic reductions of neural models ( e . g . [127] , [128] ) . The third constraint can be justified by the long time constant of decay of NMDARs , which allows NMDAR activation to filter synaptic noise at intrinsic synapses , i . e . noisy synaptic release is overcome by residual activation . The timing network estimates different intervals by the scaling of NMDAR conductance strength by , controlling the network's dynamics . Because the strength of cortical NMDAR currents are modulated by dopamine ( DA ) [59] , it is possible that could be instantiated by DA . Further evidence in support of this possibility comes from studies of working memory and interval timing . Working memory , or the active retention of information for use in cognitive tasks , is correlated with persistent mnemonic activity in a number of cortical regions and experimental conditions ( [129] , [130] ) . It is widely believed that persistent mnemonic activity is supported by recurrent synaptic processing , where the long time constant of NMDARs is hypothesized to provide an excitatory plateau [131] , [132] and to limit network oscillations [133] , [134] . This hypothesis is supported by studies showing that NMDAR antagonists abolish cortical up-states in vitro and impair working memory performance [135] . DA antagonists also impair working memory performance and DA is elevated in PFC during working memory tasks [60] , [136] . This convergence of evidence has lead to the hypothesis that NMDAR modulation by DA is a crucial factor in controlling attractor dynamics in the service of working memory ( see [135] ) . Our model makes use of the same computational principles , so the possible relationship between and DA is a compelling one , further supported by the common occurrence of climbing activity during delay periods of working memory tasks ( see Discussion section Prospective and retrospective coding ) . DA is also extensively correlated with interval timing in the seconds to minutes range [13] , [137] . In this regard , DA agonists and antagonists are correlated with underestimates and overestimates of intervals respectively ( see [114] ) . Similarly , high and low values of produced short and long interval estimates respectively in our model . Our results therefore suggest that in cortical timing circuitry , DA may strengthen attractor dynamics sufficiently to destablize background states . If so , the slope of climbing activity could be modulated by tonic DA , possibly by the increasing occupancy of D1 receptors due to slow extrasynaptic uptake [138] , [139] , consistent with enhancement of NMDAR currents via D1 activation [59] . Further work is required to address this possibility . For instance , in addition to modulating NMDAR currents , DA modulates cortical GABAR currents [60] , so a more detailed model of DA modulation is required . Furthermore , in the majority of experiments revealing DA involvement in timing in the seconds to minutes range , the effect of DA has been via D2 receptors ( see [13] , [114] ) , which act in a largely antagonistic way to D1 receptors [60] . Overall , it is unclear whether DA agonists ( antagonists ) should be expected to speed up ( slow down ) the representation of time in the hundreds of milliseconds range , but there are notable gaps in the literature that reveal important lines of enquiry . Firstly , the role of D1 and D2 receptors in timing in the seconds to minutes range appears to be task dependent , as a number of recent studies have shown evidence for D1 involvement in timing in this range , using tasks that were not used in earlier studies showing D2 involvement ( e . g . [140]–[143] ) . Additionally , several studies showing D2 involvement in earlier tasks also showed D1 involvement ( e . g . [144] , [145] ) . Secondly , despite the large body of work addressing D1 and D2 involvement in seconds-to-minutes timing in non-human animals , we are unaware of any such studies investigating the hundreds of milliseconds range ( see [146] ) . Thirdly , despite a growing body of work addressing D2 involvement in timing in healthy humans ( see below for studies with clinical populations ) , we are unaware of any studies to address D1 involvement . Studies have also shown timing deficits in the seconds to minutes range among patients with Parkinson's Disease ( [13] , [100] ) , a pathology characterized by a deterioration of dopaminergic activity [147] . Fewer such studies have considered timing in the hundreds of milliseconds range and results have been mixed among those that have [148]–[150] . Finally , we note that studies addressing the role of DA in interval timing have focused on its effect in the striatum ( see [100] ) . Suffice to say , in addition to their striatal projections , dopaminergic neurons in the basal ganglia project extensively and diffusely to cortex ( [151] ) , so these timing hypotheses are by no means incompatible with our model of local-circuit cortical timing . There has long been an appreciation of the role of time in decision making , where it has been viewed as a medium for filtering noise ( see [6] for a historical and mathematical treatment ) . In sequential sampling models , evidence for each option of a decision is integrated until the accumulated evidence for one of the options reaches a threshold level , at which time the decision is made in favour of that option ( see [2] ) . Because the evidence may be ambiguous and neural processing is noisy , temporal integration provides an average of the evidence , so decisions are not made on the basis of momentary fluctuations . The more time spent integrating , the better the average and the greater the probability of an accurate decision ( see [4] ) . Clearly , speed and accuracy impose conflicting demands within this framework , the reconciliation of which defines the SAT . In two-choice decision tasks , integrating the difference between the evidence for each option ( the decision variable ) implements a class of algorithm known as the drift diffusion model ( DDM ) , known to yield the fastest decisions for a given level of accuracy and the most accurate decisions for a given decision time [6] . The DDM thus optimizes speed and accuracy with respect to one another . This approach accounts for a huge volume of experimental data from decision making experiments ( see [3] ) and under reasonable biophysical constraints , is formally equivalent to models in which neural populations selective for each of two decision options compete via mutual inhibition [128] . Intrinsic synapses support temporal integration in these models [30] , [31] , [127] . The SAT can be be achieved within this framework by raising and lowering the decision threshold , an approach that readily accounts for behavioural data from decision making tasks ( see [5] , [6] ) , but conflicts with neural data showing decision-correlated neural activity that is approximately constant at the time of a decision [9] , [97] , [152] . A similar mechanism that is potentially consistent with these data is the adjustment of the initial level of neural activity on which the decision variable builds , a possibility that is supported by recent functional magnetic resonance imaging ( fMRI ) studies [153] , [154] . Such a mechanism requires a means to control the baseline level of activation in decision circuitry , but this requirement could be satisfied by spatially non-selective input , potentially instantiated by the persistent encoding of task requirements in PFC ( see [155] ) . Neural models have demonstrated the SAT under this approach [24] , [128] . Another potential means of trading speed and accuracy with a fixed neural threshold is the adjustment of the strength of synapses onto downstream neurons reading out or implementing the decision [156] . It is not clear , however , that the timescale of plasticity processes is consistent with the rapidity with which experimental subjects trade speed and accuracy from trial to trial [24] . An alternative , compatible mechanism is that decision-makers explicitly encode their temporal constraints , controlling the SAT downstream ( Figures 10 and 11 ) . Several recent studies have considered such an active role for the representation of time in decision making . For example , the DDM has been augmented with time-dependent mechanisms [8] , [10] . The fundamental difference between these and earlier diffusion models is that the representation of elapsed time has an increasing influence on decision processing as each trial progresses , sometimes referred to as an ‘urgency’ signal . In the time-variant DDM by [157] , amplifying the input by a growing temporal signal was shown to earn more reward per unit time than the standard DDM . This approach is functionally equivalent to lowering the decision threshold over the course of each trial , where later evidence is more heavily weighted than earlier evidence at the expense of a decreasing signal-to-noise ratio [8] . Conversely , if the incoming evidence and the evolving decision variable are both amplified by the temporal signal , there is a transition from a heavier weighting of the former to the latter [11] , similar to the transition from extrinsic to intrinsic processing hypothesized to underlie local-circuit cortical processing ( see [29] ) . We hypothesize that climbing activity drives the rate of this transition in downstream decision circuitry , controlling the SAT . This hypothesis is consistent with neural data from experimental tasks with a timing requirement on the relevant order for perceptual decisions ( see the previous section ) and with neural and behavioural data revealing the SAT ( see [5] ) . It is also consistent with neural data showing a fixed decision threshold [9] , [97] , [152] . Indeed , at least one experiment has reported climbing activity that was correlated with the time of decisions in a perceptual task , but not with the evidence [9] . Such activity effectively encodes elapsed time relative to an estimated interval [32] , shown recently to earn more reward per unit time than a persistent , top-down signal in a more abstract network model than the one used here , where the temporal signal was a linear function of time [11] . Despite the longstanding attribution of a prominent role for time in perceptual decisions and the growing appreciation of the role of temporal codes in behaviour more generally , few studies have considered the interactions between spatial and temporal codes in decision making . The SAT provides a potential window into these interactions , but most theories of the SAT have ignored the encoding of time , with other factors limiting the amount of time spent integrating evidence ( see [5] ) . We hypothesize a compatible mechanism: the SAT can be accomplished by estimating one's temporal constraints ( Figures 10 and 11 ) , where climbing activity encodes these estimates ( Figures 2 and 6 ) and controls the SAT by gain modulation ( Figures 10 and 11 ) . This hypothesis is consistent with a growing body of neural data from tasks with a timing requirement ( see above ) and with the notion of urgency in decision tasks [9] , [10] . Our implementation of the same network for timing and decision making is consistent with the network's foundations as a generic , local-circuit cortical model [27]–[29] and with a framework of distributed , generic cortical timing circuitry [12] . In this regard , we have demonstrated a plausible framework for spatiotemporal integration in cortex: the modulation of spatially selective , temporally non-selective processing by temporally selective , spatially non-selective processing . While the present implementation of this framework is uni-directional , the mutual ( bidirectional ) influence of spatial and temporal codes is an interesting future direction . The variability of decision times in perceptual tasks has been an important means of characterizing decision processing ( see [2] , [77] ) . Our study suggests that the encoding of temporal constraints is an important source of this variability . Despite the large body of work characterizing the variability of temporal estimates on the relevant order and the broad range of variability across these different experimental paradigms ( see [55] ) , we are unaware of any studies that have systematically controlled timing variability in a given task , except by varying the length of the intervals being estimated . Our study suggests that such an approach would not only help to characterize the mechanisms underlying temporal coding , but would further characterize decision making and its relationship with time . In recent years , there has been growing interest in the optimality of decisions in terms of reward-maximization [6] , [156]–[158] and a compelling possibility is that decision makers maximize reward rate by estimating their deadlines [11] . It is conceivable that estimates of deadlines imposed by the environment will vary differently than self-imposed deadlines on the same temporal order , a possibility that could be addressed by controlling the time available to respond and the timing of reward in decision tasks . Future work should address this possibility .
|
Studies in neuroscience have characterized how the brain represents objects in space and how these objects are selected for detailed perceptual processing . The selection process entails a decision about which object is favoured by the available evidence over time . This period of time is typically in the range of hundreds of milliseconds and is widely believed to be crucial for decisions , allowing neurons to filter noise in the evidence . Despite the widespread belief that time plays this role in decisions and the growing recognition that the brain estimates elapsed time during perceptual tasks , few studies have considered how the encoding of time effects decision making . We propose that neurons encode time in this range by the same general mechanisms used to select objects for detailed processing , and that these temporal representations determine how long evidence is filtered . To this end , we simulate a perceptual decision by coupling two instances of a neural network widely used to simulate localized regions of the cerebral cortex . One network encodes the passage of time and the other makes decisions based on noisy evidence . The former influences the performance of the latter , reproducing signature characteristics of temporal estimates and perceptual decisions .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"computational",
"neuroscience",
"sensory",
"systems",
"biology",
"computational",
"biology"
] |
2013
|
Trading Speed and Accuracy by Coding Time: A Coupled-circuit Cortical Model
|
Genome wide association studies ( GWAS ) have revealed 11 independent risk loci for polycystic ovary syndrome ( PCOS ) , a common disorder in young women characterized by androgen excess and oligomenorrhea . To put these risk loci and the single nucleotide polymorphisms ( SNPs ) therein into functional context , we measured DNA methylation and gene expression in subcutaneous adipose tissue biopsies to identify PCOS-specific alterations . Two genes from the LHCGR region , STON1-GTF2A1L and LHCGR , were overexpressed in PCOS . In analysis stratified by obesity , LHCGR was overexpressed only in non-obese PCOS women . Although not differentially expressed in the entire PCOS group , INSR was underexpressed in obese PCOS subjects only . Alterations in gene expression in the LHCGR , RAB5B and INSR regions suggest that SNPs in these loci may be functional and could affect gene expression directly or indirectly via epigenetic alterations . We identified reduced methylation in the LHCGR locus and increased methylation in the INSR locus , changes that are concordant with the altered gene expression profiles . Complex patterns of meQTL and eQTL were identified in these loci , suggesting that local genetic variation plays an important role in gene regulation . We propose that non-obese PCOS women possess significant alterations in LH receptor expression , which drives excess androgen secretion from the ovary . Alternatively , obese women with PCOS possess alterations in insulin receptor expression , with underexpression in metabolic tissues and overexpression in the ovary , resulting in peripheral insulin resistance and excess ovarian androgen production . These studies provide a genetic and molecular basis for the reported clinical heterogeneity of PCOS .
Polycystic ovary syndrome ( PCOS ) occurs in 6–10% of reproductive age women by NIH diagnostic criteria , and is characterized by hyperandrogenism and oligo- or amenorrhea [1] . Metabolic risk factors for type 2 diabetes and cardiovascular disease such as insulin resistance and obesity are common in women with PCOS , with increased body weight , insulin resistance , and impaired glucose tolerance most elevated in women with the highest levels of androgens [1 , 2] . PCOS is a complex disorder with both genetic and environmental factors contributing to its pathophysiology . Twin studies have provided heritability estimates for PCOS of 0 . 71 [3] . Two genome wide association studies ( GWAS ) , carried out in Han Chinese populations , identified 15 risk SNPs from 11 loci ( THADA , LHCGR , FSHR , C9orf3 , DENND1A , YAP1 , RAB5B , INSR , TOX3 , SUMO1P1 , and HMGA2 ) [4 , 5] . Six of these risk loci ( THADA , LHCGR , FSHR , DENND1A , YAP1 , INSR ) have been replicated in Caucasian populations [6–10] . In addition , a genetic risk score based on SNPs not individually associated with PCOS was found to be significantly associated with PCOS in Caucasian subjects [10] , suggesting that some or all of the variants identified in Chinese populations are likely also risk variants in Caucasians . GWAS have provided insight into the genetic architecture of many complex diseases , including PCOS . A limited number of functional studies have evaluated the role of several of the newly identified PCOS risk loci , including LHCGR ( Luteinizing hormone/choriogonadotropin receptor ) and DENND1A ( DENN/MADD domain containing 1A ) [11–13] . The LHCGR promoter region was shown to be hypomethylated and mRNA expression level increased in granulosa cells from women with PCOS [12] . An in vitro study reported overexpression of transcriptional variant 2 of DENND1A ( DENND1Av2 ) in the theca cells of PCOS patients and demonstrated its ability to increase androgen and progestin biosynthesis [13] . These functional studies provide early evidence that alterations in methylation and gene expression within the PCOS GWAS susceptibility loci contribute to the pathophysiology of PCOS . In order to gain a greater understanding of the role of the PCOS susceptibility loci identified by GWAS further functional studies in PCOS relevant tissues are urgently needed . The functional characteristics of a locus can include the epigenetic regulation of expression ( for example , DNA methylation or histone modification ) , enhancer binding activity , transcription factor binding profiles , promoter activity , and the gene expression profile . DNA methylation plays an important role in the regulation of gene expression by affecting chromatin state and the ability of transcription factors , enhancers and insulators to bind DNA [14] . DNA methylation profiles are impacted by local SNPs , either directly by the creation/ablation of CpG residues , or indirectly [15] , allowing SNPs in non-coding regions of the genome to have functional impacts on local gene regulation . Tissue specific methylation patterns contribute to gene expression profiles that delineate tissue function; therefore , SNPs may have tissue specific effects on disease pathways . Genetic variants that regulate methylation at CpG residues are known as methylation quantitative trait loci ( meQTL ) . Genetic variation can also impact gene expression in a manner independent of methylation . Identification of genotype effects on gene expression level ( expression quantitative trait loci , or eQTL ) can help to identify the causal transcript in a disease-associated locus . Although each index SNP from the PCOS GWAS loci has been assigned to a gene , this was done following the common practice of selecting the nearest gene , without functional knowledge such as expression profiles of transcripts surrounding the index SNP . In the present study , we measured DNA methylation and gene expression in adipose tissue of PCOS women and normal controls in order to better understand the functional elements surrounding PCOS-associated SNPs . As an endocrine tissue with a clear role in metabolic function and relative ease of collection , adipose tissue is highly suited for functional studies of PCOS genes , particularly those not directly related to androgen excess or ovarian function . Adipose dysfunction in PCOS has been widely reported; studies of subcutaneous adipocytes from PCOS women have demonstrated resistance to insulin stimulated glucose transport and inhibition of lipolysis [16 , 17] . We have generated the first functional maps of PCOS loci , comparing methylation and gene expression patterns between PCOS patients and healthy controls and the interactions between the SNPs in these regions and local methylation ( meQTL ) and local gene expression ( eQTL ) . The aim of our study was to use a systems biology approach to investigate patterns of gene regulation and expression in the genomic regions surrounding the previously identified PCOS susceptibility loci in a PCOS relevant tissue in order to understand the functional context of these loci .
PCOS subjects were not significantly older than controls , and no significant difference in BMI was detected . As expected , PCOS cases had elevated testosterone and hirsutism measured by modified Ferriman-Gallwey ( mFG ) score ( Table 1 ) . In 35 subjects ( 22 cases and 13 controls ) , we examined the gene expression profiles for each transcript that passed normalization and background correction within the 11 PCOS risk loci . A total of 50 transcripts were identified in the genomic windows surrounding the PCOS risk variants and extracted from the genome wide expression dataset . Twenty-eight of these were expressed in the adipose tissue samples . Both LHCGR and STON1-GTF2A1L from the LHCGR locus were overexpressed in PCOS , while WIBG , RAB5B and IKZF4 from the RAB5B locus were underexpressed in PCOS ( Fig 1A ) . After correction for multiple testing with FDR , both LHCGR and WIBG remained significantly differentially expressed . Power estimates indicated we were well powered to detect significant effects at an alpha of 0 . 05 ( power for detection of differences in expression of LHCGR was 0 . 93 ) . In order to investigate the effect of obesity on gene expression in PCOS adipose tissue , we performed secondary analyses in obese and non-obese subjects separately ( Fig 1B ) . In the non-obese subjects , LHCGR was significantly overexpressed in PCOS and WIBG and IKZF4 were significantly underexpressed in PCOS . INSR was underexpressed only in obese PCOS subjects , with no changes in expression in non-obese PCOS women . LHCGR remained significantly overexpressed in non-obese subjects after correction for multiple testing; however , other stratified results were no longer significant ( FDR P value <0 . 05 ) . Mean beta methylation level at a total of 650 CpG sites across the 11 PCOS risk loci windows were analyzed in 13 cases and 11 controls . A total of 17 CpG sites across the 11 windows demonstrated significant differences in methylation levels between PCOS subjects and controls ( empirical P<0 . 05 ) ( Fig 2 ) . Four CpG sites were differentially methylated across the RAB5B window , including two sites located in the intergenic region 5’ to IKZF4 with increased methylation in PCOS subjects ( Fig 2 ) . Within the INSR window a single CpG was hypermethylated in PCOS subjects . Three CpG sites in the LHCGR window , all located near STON1-GTF2A1L , were hypomethylated in PCOS subjects . CpG sites in the C9orf3 , DENND1A , YAP1 , HMGA2 , TOX3 and SUMO1P1 loci were also differentially methylated between PCOS and controls ( Fig 2 ) . We applied FDR correction for multiple testing and did not identify any methylation sites that retained significance; however , due to the highly correlated nature of methylation probes we would consider this approach conservative . Correlation analysis between differentially methylated sites and expression level of genes within the local window did not reveal any significant expression quantitative methylation ( eQTM ) . Of the remaining eight loci , six had PCOS specific changes in methylation ( Fig 2 ) . We also examined each window for changes in gene expression in PCOS , however did not identify any other genes that are over/under-expressed in PCOS . Several windows contained genes that were not expressed in our adipose samples ( S1 Table ) , including the DENND1Av2 transcriptional variant reported to be overexpressed in PCOS theca and urine [13] . We did identify reduced methylation in intron 2 of DENND1A , which may regulate isoform specific expression in a tissue dependent manner . The National Center for Biotechnology Information’s ( NCBI ) GEO database was used to further investigate the differentially expressed genes in our cohort both in subcutaneous adipose and other tissue types ( S2 Table ) . In a small series comparing gene expression in subcutaneous adipose tissue of PCOS and control subjects , WIBG was underexpressed in PCOS patients , similar to our findings . Also consistent with our findings , in cumulus cells LHCGR was overexpressed in PCOS subjects . In the latter series , when the subjects were stratified by obesity , the non-obese PCOS subjects had lower expression of WIBG and LHCGR continued to be overexpressed . In the obese subjects , women with PCOS demonstrated higher expression of INSR in cumulus cells . Lower expression of INSR was seen in PCOS subjects in two different series of skeletal muscle . Relationships between SNPs and methylation and gene expression were further investigated using a systems genetics approach . meQTL were identified in 19 subjects that had both methylation and genotype data available . Within the LHCGR window , SNPs in the 5’ and intron 1 regions of the LHCGR gene , surrounding the PCOS risk SNP rs13405728 , were associated with methylation level of three CpG residues clustered in the STON1-GTF2A1L gene ( S5 Table and Fig 3 ) . Association of one of these methylation sites ( cg01450842 ) with local variants has been previously reported in adipose tissue [14] , suggesting that variants in the 5’ and intron 1 regions of LHCGR may play a role in methylation , and potentially transcriptional regulation of genes at this locus . The minor allele at each of these three meQTL pairings was associated with decreased methylation level at each site , suggesting these variants reduce methylation and may lead to increased expression ( S5 Table and Figs 3 and S1 ) . Five SNPs ( meQTLs ) from across the INSR window were associated with 4 adjacent CpG sites clustered upstream and in intron 1 of ZNF557 ( S5 Table ) . The two most 5’ methylation sites in meQTL pairs ( rs8106126-cg19772356 , rs10401628-cg09022474 ) are upstream of the ZNF557 gene and overlap enhancer , promoter and transcription factor binding sites in the ENCODE data track of Fig 3 ( Panels iii-2 , 3 , 4 ) . Within the gene body of ZNF557 is a cluster of unmethylated probes , two of which were in meQTL pairs with SNPs in intron 11 and 12 of the INSR gene . One of these SNPs , rs8106125 , is in moderate LD ( r2 = 0 . 50 ) with the PCOS GWAS index SNP , shown in Fig 3 by a black triangle in track i , labeled as rs2059807 . mRNA levels of INSR were associated with a number of SNPs across the 5’ region of the window ( P values are shown in S4 Table ) that are in a complex pattern of LD . A 4 SNP ( rs10401628 , rs2352958 , rs7248939 , and rs10418342 ) conditional regression analysis ( shown in green in the regional association plot in S2 Fig ) was required to remove any signal of significant association in the eQTL analysis results and suggests that at least 4 SNPs independently act as eQTLs in this window ( Fig 3 ) . In the RAB5B/SUOX window four SNPs from across the window acted independently as meQTL SNPs with five methylation sites clustered between WIBG and DGKA and at the 5’ and 3’ of RAB5B ( S5 Table ) . ENCODE data indicated that these methylation sites overlap active enhancer and promoter regions . Finally , an eQTL for this locus was also identified with many linked SNPs from across the window and RPS26 ( Fig 3 ) . A conditional analysis with the top SNP ( rs10876864 ) eliminated the significant associations from all other SNPs , indicating that a single association between many linked SNPs accounts for the numerous association signals .
To gain insight into the function of PCOS susceptibility loci we evaluated genotype , methylation and mRNA expression in the regions surrounding these SNPs in PCOS and healthy control adipose tissue . We have generated the first functional maps of PCOS GWAS loci in PCOS tissue , mapping methylation and gene expression surrounding previously identified PCOS risk loci and identifying relationships between genetic variants and these functional elements . These functional maps allowed us to identify PCOS specific changes in gene expression and methylation in several loci in PCOS adipose tissue . We have also identified differences in the gene expression profile of these risk genes in non-obese and obese PCOS subjects . We found LHCGR was overexpressed in the adipose tissue of non-obese women with PCOS , and corresponding decreases in methylation of adjacent CpG residues . This is consistent with prior studies demonstrating increased LHCGR expression in granulosa and theca cells from patients with PCOS compared to normal controls [12] . We found this non-obese specific increase in expression was also present in cumulus cells from women with PCOS in our confirmatory analysis from the GEO database ( S2 Table ) . Women with PCOS , particularly when not obese , have higher levels of LH secreted from the pituitary [18–20] , increased bioactivity of LH [21 , 22] and excessive production of androgens from the ovaries in response to LH [18 , 23 , 24] . It is possible that enhanced sensitivity to LH in the ovary is due to increased receptor number as a result of overexpression of LHCGR , resulting in elevated androgen synthesis from the theca cell . A biological role for LHCGR in adipose is not clear . The Genotype-Tissue Expression Project ( GTEx ) database [25] reports its expression in subcutaneous adipose as well as several other unexpected tissues such as visceral adipose ( omentum ) , tibial nerve , and esophagus . Reduced methylation and overexpression of LHCGR in adipose could represent a conserved gene regulation profile across tissues in non-obese women with PCOS . To confirm that our finding of LHCGR overexpression is a PCOS-specific effect , we identified GEO datasets that could be analyzed with either obesity or insulin sensitivity as a dichotomous trait . We did not find any changes in expression between lean and obese subjects in three adipose GEO datasets where obesity was available to stratify subjects ( S3 Table ) , or in three datasets where insulin sensitivity was available as a dichotomous trait . These findings , together with our own results , suggest that the observed changes in LHCGR expression are private to PCOS , and not a result of metabolic heterogeneity in the cohort . The role of insulin in PCOS has been widely studied [26] . While insulin resistance is a common feature in PCOS women , it is particularly common in obese women with PCOS [27 , 28] . Compensatory increased circulating insulin levels contribute to PCOS by stimulating ovarian androgen production and inhibiting hepatic SHBG production [29 , 30] . In our study of adipose tissue , we found that obese women with PCOS had significantly lower expression of INSR . In keeping with this , INSR was also down regulated in skeletal muscle of PCOS patients in two independent studies ( S2 Table ) . Decreased INSR expression in metabolic tissues is consistent with insulin resistance and provides a potential mechanism for insulin resistance frequently seen in obese women with PCOS . Contrary to decreased INSR expression in metabolic tissues ( adipose and skeletal muscle ) of obese PCOS women , we found INSR to be overexpressed in the cumulus cells of obese PCOS subjects ( S2 Table ) . Studies have demonstrated differences in insulin sensitivity between reproductive and metabolic tissues , where obese mice had a blunted response to insulin in the liver and muscle while the pituitary and ovary maintained insulin sensitivity [31] . Studies in insulin-resistant PCOS women suggest that the ovaries remain sensitive to insulin’s actions on steroidogenesis , even when metabolic tissues demonstrate peripheral insulin resistance by decreased glucose disposal [30] . Our finding of tissue specific underexpression of INSR in metabolic tissues and overexpression in ovarian tissues supports the previously suggested hypothesis of selective insulin resistance in PCOS , where ovarian sensitivity to insulin is maintained despite peripheral insulin resistance , allowing insulin driven androgen synthesis in the ovary to persist . We identified increased methylation of a single CpG site in a largely unmethylated region 5’ to the INSR transcription start site that also overlaps a regulatory motif in the UCSC ENCODE browser that could regulate INSR expression . Future experiments should include mapping the methylation and expression profile of INSR from several PCOS ovarian cell types , potentially supporting the hypothesis of maintained insulin sensitivity in the ovary as a result of alterations in INSR methylation and expression . It is known that obese women with PCOS have significantly more insulin resistance and the LH levels are higher in non-obese women with PCOS [32] . Our findings suggest that the mechanisms underlying hyperandrogenemia in obese and non-obese PCOS may have a different genetic basis . Non-obese women with increased LHCGR expression may have increased LH-dependent androgen production by the ovary due to increased number of LH receptors and increased LH levels . Obese women with increased INSR expression in androgen-synthesizing ovarian cells may have hyperandrogenemia driven by the hyperinsulinemic response to reduced insulin receptor number in metabolic tissues . We also identified a number of changes in gene regulation and expression in the RAB5B window . In PCOS samples WIBG was underexpressed at FDR corrected significance , and reduced expression levels of RAB5B , and IKZF4 were nominally associated with PCOS . Increased methylation was observed at three CpG sites across the locus , but did not meet correction for multiple testing . Our restricted sample size in methylation analysis , and in stratified expression analysis likely reduced our ability to detect smaller effects . While expression and methylation levels were not significantly correlated in an eQTM relationship , we assayed a relatively small number of all potential methylation sites from this locus , and more extensive changes in methylation at unassayed residues may be regulated by eQTM for these genes . We measured methylation in a subset ( 24 of our total 36 samples ) of adipose samples , and while this is the largest study of this type published to date , the relatively small sample size may have reduced our ability to identify eQTM . A publicly available replication data set comparing gene expression between PCOS and controls in subcutaneous adipose tissue also found WIBG to be underexpressed in PCOS ( S2 Table ) . WIBG encodes a cytoplasmic protein that binds to the ribosomal unit and increases translational efficiency of mRNA [33 , 34] . A specific role for WIBG in PCOS is unclear . RAB5B is a small GTPase that plays a role in early endosome formation and is required for the endocytic pathway that mediates the transport of clathrin-coated vesicles from the plasma membrane to the early endosome [35] . RAB5B has also been identified as a susceptibility locus for type 1 diabetes and childhood obesity [36] . Interestingly , DENND1A encodes for a protein , connecdenn 1 , that also facilitates endocytosis and membrane trafficking and is known to interact with Rab family member RAB35 [37] . Functional studies of DENND1A demonstrated increased expression of DENND1Av2 and increased androgen synthesis in the theca cells of PCOS women [13] . This variant was not expressed in our adipose samples . Given RAB5B’s association with type 1 diabetes it is possible that genes in this locus play a regulatory role via that impacts beta cell function or insulin secretion , a process that is impaired in both disorders . A third gene , IKZF4 , was also down regulated in subcutaneous adipose of women with PCOS . IKZF4 is zinc-finger transcription factor that functions as a transcriptional repressor and is known to play a role in immune regulation , specifically in the programming of T regulatory cells [38] . There is evidence suggesting the presence of chronic low-grade inflammation in women with PCOS; studies have found significantly higher levels of C-reactive protein ( CRP ) and other cytokines , independent of BMI [39] . Underexpression of IKZF4 in PCOS adipose tissues may impact the ability of T cells to suppress pro-inflammatory responses , and contribute to the chronic inflammation seen in PCOS . As several markers of inflammation have been correlated with insulin resistance [40–42] , chronic low-grade inflammation may contribute to the etiology of insulin resistance seen in PCOS . In conclusion , PCOS GWAS loci contain extensive alterations in methylation and gene expression profiles between PCOS and controls , which identify genetic and molecular differences between clinical disease subtypes based on presence or absence of obesity . We demonstrated that LHCGR is overexpressed in the subcutaneous adipose tissue of non-obese PCOS women and INSR was underexpressed in obese women with PCOS . This underexpression of INSR in obese women with PCOS was also seen in cumulus cells . Taken together , our findings suggest that the gene expression profiles may be different between obese and non-obese PCOS subjects , with hormonal disturbances playing a more important role in non-obese subjects and metabolic disturbances playing a larger role in obese subjects . Our results suggesting different mechanisms underlying hyperandrogenemia in non-obese versus obese women may one day have clinical implications , as subclassification based on pathophysiology may lead to tailored treatment . While we did not resolve all functional regulatory mechanisms in PCOS loci in adipose tissue , we provide new insight into several of the susceptibility loci discovered in the PCOS GWAS . Given that methylation and expression vary between tissue types , further studies in other tissues relevant to PCOS pathophysiology are needed to further elucidate the function of these PCOS susceptibility loci .
This study was approved by the Cedars-Sinai Institutional Review Board ( IRB ) under approval number 11289 . All subjects gave written informed consent according to the guidelines of the IRB Subcutaneous lower abdominal adipose tissue was obtained from 23 PCOS and 13 control subjects using a previously described protocol for acquiring and processing subcutaneous adipose tissue [43] . PCOS subjects were recruited at a tertiary care academic institution . Cases were premenopausal , nonpregnant , and on no hormonal therapy , including oral contraceptives , for at least 3 months , and met 1990 National Institutes of Health criteria for PCOS [44] . Parameters for defining hirsutism , hyperandrogenemia , ovulatory dysfunction , and exclusion of related disorders were previously reported [45] . Controls were recruited by word of mouth and advertisements to the public calling for healthy women . Controls were healthy women , with regular menstrual cycles and no evidence of hirsutism , acne , alopecia , or endocrine dysfunction and had not taken hormonal therapy ( including oral contraceptives ) for at least 3 months . Clinical characteristics for these subjects are shown in Table 1 . Samples were snap frozen immediately after collection in liquid nitrogen and then stored at -80°C until extraction . DNA and RNA were isolated from subcutaneous fat tissue after rapid thaw at 37°C with the AllPrep DNA/RNA/protein Mini kit ( QIAGEN , Valencia , CA ) . DNA was stored in TE buffer , quantified and checked for quality on a Nanodrop-1000 ( Nanodrop , Wilmington , DE ) and stored at -80°C . RNA samples were quantified and checked for quality using the BioAnalyzer 6000 Pico kit ( Agilent , Santa Clara , CA ) and stored at -80°C . Genotyping of 36 samples was performed at CSMC using the HumanExome chip ( targeting functional ( e . g . , missense and splice junction ) variants ) , the HumanOmniExpress chip ( targeting common variants using a haplotype tagging approach ) and the HumanOmni1S chip ( targeting rare and non-Caucasian SNPs and copy number variants ) following the manufacturer’s protocol ( Illumina , San Diego , CA ) [46 , 47] . Samples were randomized by case/control status and arrayed at a concentration of 50ng/ul prior to genotyping as part of larger experiments . Thirty one samples passed sample based quality control measures for all three chips that included genotyping rate >98% ( five samples had a genotyping rate <98% ) , p10GC ( a sample statistic representing the tenth percentile of the distribution of genotype quality scores across all SNPs genotyped ) and SNP-based gender estimate ( all samples passed gender estimation ) . Genotypes from each chip were exported from Genome Studio ( Illumina , San Diego , CA ) and merged in SVS ( Golden Helix , Bozeman , MT ) . SNPs with MAF>5% and Hardy-Weinberg Equilibrium P Value >1 . 0x10-4 were retained for downstream analysis ( total number of SNPs carried forward was 1 , 180 , 811 ) . Principal components analysis ( PCA ) within SVS was used to generate the top 10 PCs to identify outlier samples , of which none were found . A subset of 24 age and BMI matched subjects were selected for methylation analysis due to restrictions on sample size because samples were run as part of a larger project . DNA methylation levels were measured using the HumanMethylation450 chip ( Illumina , San Diego , CA ) according to the manufacturer’s instructions at CSMC . The HumanMethylation450 chip targets over 485 , 000 CpG residues across 96% of all RefSeq genes ( 21 , 500 gene symbols ) and 95% of CpG islands and flanking regions . Samples were randomized by case/control status across two plates ( 10 chips ) in the context of a larger experiment and arrayed at a concentration of 10ng/ul . Detection P values were calculated to identify failed probes , and beta ( β ) values representing methylation levels were generated from the Genome Studio software for each methylation site , ranging from 0 ( completely unmethylated ) to 100% ( completely methylated ) . The Methylumi package in R was used to background normalize and log transform the beta values [48] . 485 , 577 probes were exported from Methylumi for further analysis . The data was checked for distribution of the mean β value per site , distribution of the mean β value per site in each bead type , mean methylation score across all samples per CpG location , variance of the β levels across individuals , PC analysis and plotting for each sample and the distribution of methylation sites based on location relative to each CpG locus . All 24 samples passed QC measures and were retained for downstream analysis . Methylation level at individual probes was categorized as low-methylated ( beta <0 . 4 ) , semi-methylated ( beta 0 . 4–0 . 6 ) , or highly-methylated ( beta >0 . 6 ) . The HumanHT-12v4 beadchip was used to measure gene expression levels of well-characterized genes , gene candidates , and splice variants with 47 , 000 probes at the UCLA Neuroscience Genomics Core ( UNGC ) . All 36 samples were randomized according to case/control status and RNA was arrayed at 10ng/ul . The TargetAmp-Nano Labeling Kit for Illumina Expression BeadChip ( Epicenter , San Diego , CA ) was used to label samples . Sample probe profile data was exported from Genome Studio after QC metrics ( direct hyb control metrics including hybridization controls , stringency metrics , background and noise of control probes , gene intensity of housekeeping and all genes and labeling and background metrics ) and sample metrics ( number of genes detected , 95th intensity percentile , signal to noise ratio , signal across all samples ) were reviewed . One sample was excluded due to excessive signal to noise ratio . Sample probe profile data was read into the limma package in R version 3 . 1 . 1 [49] . Probes expressed in at least three samples and with a detection P value of <0 . 05 were retained ( 47 , 314 probes were read in and 29 , 081 probes were retained after this step ) . Background normalization , quantile normalization and log transformation of the remaining probes was performed with the neqc ( ) function in limma . Normalized and transformed data was then used to generate PCs using the MDS ( ) function within limma and PCs were plotted with samples labeled with case/control status to identify QC outliers for removal . No samples were outliers or flagged for removal at this step . Normalized and transformed data was exported from limma for analysis . An experimental flowchart describing sample size and data available for DNA genotyping , methylation and gene expression analysis is shown in S3 Fig . Our analysis was focused on genomic windows around each of the 11 loci previously discovered to harbor SNPs associated with PCOS in GWAS . The genomic region 100kb upstream and downstream of each GWAS SNP was evaluated for methylation and mRNA expression . If the window terminated within the coding frame of a gene , the window was extended to 10kb beyond the coding frame ( S1 Table ) . Normalized and log transformed methylation beta levels and gene expression levels were analyzed in case/control analysis using logistic regression in SVS adjusting for age and BMI . Subjects were stratified by obesity status where subjects with BMI ≥ 30kg/m2 were categorized as obese , and subjects with BMI < 30kg/m2 were categorized as non-obese . Obesity stratified analyses were adjusted for age only . Linear regression in SVS adjusting for age , BMI , disease status and PC1 was used for meQTL ( SNP associated with methylation level ) , eQTL ( SNP associated with mRNA level ) and eQTM ( methylation level associated with mRNA level ) analysis . meQTL relationships between methylation probes and multiple SNPs were interrogated for linkage disequilibrium ( LD ) between the SNPs , and conditional analysis using additive model genotype as a covariate in the linear regression was used to identify the variant driving the meQTL association if possible . We applied correction for multiple testing in gene expression and methylation results using the False Discovery Rate [50] , with an FDR P value <0 . 05 held as significant . In light of the relatively small number of independent tests being run within each independent locus , we considered results with an empirical P value of <0 . 05 suggestive of significance . Correction for multiple testing for meQTL , eQTL and eQTM analysis was calculated on a per-window basis to adjust for the number of SNPs analyzed in a modified Bonferroni approach . NCBI’s GEO is a public repository that archives high-throughput functional genomics data . To compare expression levels of the candidate genes discovered in our cohort in additional PCOS tissues we evaluated these genes in other datasets . A search of the GEO database identified 8 datasets comparing gene expression between PCOS patients and controls in various tissue types ( S2 Table ) . The GEO2R interactive web tool was used to perform comparisons between PCOS and control subjects on the original submitter-supplied processed data tables . GEO2R uses GEOquery and limma R packages from the Bioconductor project to perform statistical analysis [51] . All data was log transformed . These analyses were not adjusted for age or BMI , as these traits are not available in the GEO database . ENCODE tracks were displayed in the UCSC Genome Browser using Build37 ( GRCh37/hg19 ) in order to identify poised enhancer ( H3K4Me1 ) , active enhancer ( H3K27Ac ) , active promoter ( H3K4Me3 ) , and transcription activity in the ENCODE reference cell types [52] .
|
Polycystic ovary syndrome ( PCOS ) is the most common hormonal disturbance in reproductive age women and features high levels of male sex hormones , such as testosterone , and infrequent ovulation . Twin studies have demonstrated that inheritance plays a significant role in PCOS , and recent genome wide association studies ( GWAS ) have implicated 11 susceptibility regions . The mechanism by which these genetic loci cause PCOS has yet to be determined . We looked at DNA methylation and gene expression levels in these 11 loci in fat biopsies from women with and without PCOS . We identified differences in the expression of two receptors that bind hormones known to contribute to the pathogenesis of PCOS–the receptors for luteinizing hormone ( LH ) and insulin . We found increased expression of the LH receptor in non-obese PCOS women , while in the obese women with PCOS the insulin receptor was underexpressed . Both excess LH stimulation and elevated insulin levels , due to decreased receptor levels and resulting insulin resistance , can cause increased androgen production from the ovary . Our findings suggest the primary mechanism for elevated androgen levels in PCOS may differ between non-obese and obese women with PCOS and that the clinical heterogeneity seen in PCOS may have genetic underpinnings .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2015
|
Systems Genetics Reveals the Functional Context of PCOS Loci and Identifies Genetic and Molecular Mechanisms of Disease Heterogeneity
|
Anthrax is hyper-endemic in West Africa . Despite the effectiveness of livestock vaccines in controlling anthrax , underreporting , logistics , and limited resources makes implementing vaccination campaigns difficult . To better understand the geographic limits of anthrax , elucidate environmental factors related to its occurrence , and identify human and livestock populations at risk , we developed predictive models of the environmental suitability of anthrax in Ghana . We obtained data on the location and date of livestock anthrax from veterinary and outbreak response records in Ghana during 2005–2016 , as well as livestock vaccination registers and population estimates of characteristically high-risk groups . To predict the environmental suitability of anthrax , we used an ensemble of random forest ( RF ) models built using a combination of climatic and environmental factors . From 2005 through the first six months of 2016 , there were 67 anthrax outbreaks ( 851 cases ) in livestock; outbreaks showed a seasonal peak during February through April and primarily involved cattle . There was a median of 19 , 709 vaccine doses [range: 0–175 thousand] administered annually . Results from the RF model suggest a marked ecological divide separating the broad areas of environmental suitability in northern Ghana from the southern part of the country . Increasing alkaline soil pH was associated with a higher probability of anthrax occurrence . We estimated 2 . 2 ( 95% CI: 2 . 0 , 2 . 5 ) million livestock and 805 ( 95% CI: 519 , 890 ) thousand low income rural livestock keepers were located in anthrax risk areas . Based on our estimates , the current anthrax vaccination efforts in Ghana cover a fraction of the livestock potentially at risk , thus control efforts should be focused on improving vaccine coverage among high risk groups .
Anthrax is a soil-borne , zoonotic disease found on nearly every continent ( except Antarctica ) that primarily infects herbivorous animals while secondarily infecting humans through the handling or ingestion of contaminated meat or animal by-products [1 , 2] . The geographic distribution of the disease appears to be limited by a combination of climatic ( e . g . precipitation and temperature ) and environmental ( e . g . alkaline soil pH ) conditions [3 , 4] . Under the appropriate ecological conditions , which remain poorly understood , the causative agent of anthrax , Bacillus anthracis , can survive for long-periods of time in the environment , perhaps years [1 , 4] . Although it has received much attention as a potential agent of bioterrorism , the World Health Organization ( WHO ) has listed anthrax as a neglected disease [5] . Poor livestock keepers and their animals often experience a disproportionate burden of anthrax in the hyper-endemic regions of Central Asia and West Africa [5 , 6] . Despite the effectiveness of regular animal vaccination and proper outbreak response following recommended guidelines in controlling anthrax in humans , underreporting of the disease often skews its true burden and geographic distribution making it difficult to implement adequate vaccination campaigns [1 , 7] . In Ghana , anthrax outbreaks have been reported annually in humans associated with contact with infected livestock and their contaminated animal by-products ( e . g . meat or hides ) [8] . Anthrax vaccine is manufactured locally by the Central Veterinary Laboratory in Pong-Tamale , Ghana and is fully subsidized by the government . Despite this , animal outbreaks are documented annually , and primarily affect cattle . Although both human and animal cases are reported , few human cases are linked to confirmed animal cases [9] . As a result , surveillance data alone provide limited information to efficiently plan prevention activities . Previous efforts to elucidate the environmental suitability of anthrax in Africa have been focused on southern countries , such as Zimbabwe [10] , or national parks [11] . A recent study from West Africa also used a machine learning algorithm to map and model the distribution of anthrax and B . anthracis in Cameroon , Chad , and Nigeria , however , that effort was based on limited sample size and no comparable efforts have been carried out in Ghana [12] . To support Ghana’s national anthrax control and assessment , we our study had the following objectives: ( 1 ) model the environmental suitability of anthrax; ( 2 ) identify environmental and climatic factors associated with the occurrence of anthrax; ( 3 ) describe seasonal patterns; and ( 4 ) estimate populations at risk .
This work was performed on nationally available data on anthrax outbreaks in livestock from the Ministry of Food and Agriculture in Ghana . We constructed a GIS of livestock anthrax outbreaks using data collected by the Ghana Field Epidemiology and Laboratory Training Program ( GFELTP ) and the Ministry of Food and Agricultural Veterinary Services . ( Fig 1 ) . Outbreaks were mapped using GPS coordinates collected by field personnel responding to outbreaks or the center of the village where the outbreak occurred . We included data on outbreaks from 2005 through the first 6-months of 2016 included information on the geographic coordinates , date , livestock species , and number of individual animals infected ( periodically recording mortality and survival status ) for each outbreak . However , total livestock populations on affected properties was rarely reported . For this study , an outbreak was defined as any location with one or more anthrax cases in animals . We plotted the seasonality of anthrax outbreaks in relation to the average rainfall during 1991–2015 using data obtained from the Climate Change Knowledge Portal ( http://sdwebx . worldbank . org/climateportal/index . cfm ? page=country_historical_climate&ThisCCode=GHA ) . We also obtained livestock anthrax vaccine administration data during 2005–2015 from the World Animal health Information Database Interface ( OIE; http://www . oie . int/animal-health-in-the-world/the-world-animal-health-information-system/data-after-2004-wahis-interface/ ) . Mapping and spatial analysis was performed in Q-GIS version 2 . 14 ( www . qgis . org ) and the R statistical package ( https://www . r-project . org/ ) . Final maps were produced in ArcGIS version 10 . 3 . 1 ( ESRI , Redlands , CA , USA ) . We used a combination of environmental and climatic variables at a spatial resolution of 30-arcseconds ( approximately 1km x 1km ) that followed , in part , recent studies in West Africa [13] and Central Asia [14] ( Table 1 ) . Five “bioclimatic” variables describing measures of temperature and precipitation were obtained from the WorldClim database ( www . worldclim . org ) [15] . WorldClim variables are interpolated monthly measurements recorded at weather stations located worldwide between 1950 and 2000 . WorldClim produces bioclimatic variable grids to describe annual trends , seasonality , and ecological parameters such as temperature of the coldest and warmest quarters . We also used a combination of physical ( sand content ) , chemical ( soil pH ) , and taxonomic classifications of soil characteristics ( cancerous vertisols and humults ) . Soil data were obtained from the SoilGrids1km database http://www . isric . org/explore/soilgrids ) [16] . SoilGrid variables were created using spatial model predictions based on a global database of soil profiles and a combination of environmental covariates . Furthermore , we used two normalized difference vegetation index ( NDVI ) variables describing average conditions and the amplitude of vegetation greenness , which were obtained from the Trypanosomiasis and Land Use in Africa ( TALA ) research group ( Oxford , United Kingdom ) [17] . TALA variables were derived from temporal Fourier analysed ( TFA ) time series data of advanced very-high resolution radiometer ( AVHRR ) satellite measurements taken between 1992 and 1996 [17] . Mapped variables are shown in S1 Fig . Random Forest ( RF ) modeling [18 , 19] was used to identify environmental characteristics associated with the occurrence of anthrax outbreaks using the ‘randomForest’ package for R . Previous studies have used this approach to map and model the distribution of Anopheles spp . mosquito vectors in Africa and Europe [20] and reservoirs of avian influenza [21] . RF modeling has been described and compared to other modeling approaches in detail elsewhere [18 , 22] . Briefly , RF is a non-parametric method derived from classification and regression trees that consists of a combination of trees built using randomly selected bootstrap samples of the training data ( used to build the model ) , with the number of bootstrap samples equal to the number of trees ( ntrees ) selected . Each tree is split by randomly sampling a number of predictor variables to use ( mtry ) at each node and then choosing the best split . Model error estimates are obtained by internal splits of the training data ( 63 . 2% for model building ) and then predicting the data not used to build a tree ( out-of-bag or OOB ) and aggregating these predictions for each ensemble of trees [18] . Since internal validation of the OOB data is performed , no external testing data is required to validate the model , but testing splits ( external data withheld from the model ) of the data are routinely utilized to assess model performance . Partial dependence plots and variable importance of RF models were assessed for covariates in the model . We used an ensemble modeling approach that incorporated information from multiple random splits of our data into training ( 80% ) and testing ( 20% ) sets . Since our data consisted of presence only records of anthrax outbreaks , we generated pseudo-absence data from all available background data . Several studies have either relied on internal derivations of pseudo-absence in species distribution models [23] or user-defined generations such as in the modeling of the global distribution of dengue virus [24] . The required number of user-defined background pseudo-absence draws for every presence location is not standardized . It has been suggested that a 1:1 random draw of pseudo-absence to presence data in machine learning algorithms such as RF produces optimal results [25] , although variations of this ( 2:1 or 3:1 draws ) have been adopted successfully [24] . Similarly , pseudo-absence data creation has been shown to influence results; thus , research has recommended filtering pseudo-absence data from locations that are known to fall within suitable habitat or that occur within a defined proximity threshold [25 , 26] . We first filtered geo-located anthrax presence data in Ghana ( n = 61 ) using a 5km x 5km proximity threshold in order to improve model performance and avoid overfitting [27] . We generated background pseudo-absence data ( n = 200 ) , from all available background [24] , at a ratio of four absence points to every one filtered presence point ( n = 50 ) , restricting pseudo-absence data to exclude landscape within 5km of presence locations . We then generated 10 random draws each of 1:1 , 2:1 , and 3:1 pseudo-absence to presence data ( 30 total draws ) with replacement . Each randomly generated pseudo-absence to presence draw ( n = 30 ) was randomly divided into training and testing data splits to validate model performance . The final RF models were built using a mtry = 4 at each split and ntrees = 1000 with a combination of variables in which the ensemble list contributed to a mean decrease in accuracy >1% . The 30 individual RF models were then combined into an ensemble prediction at a spatial resolution of ~1km x 1km and scaled from 0 ( low suitability ) to 1 ( high suitability ) ; uncertainty in the model prediction was calculated by taking the range in the 95% confidence intervals of the ensemble model scaled from 0 ( low uncertainty ) to 1 ( high uncertainty ) following Deribe et al . [28] . The resulting output of our ensemble RF model represents the environmental suitability of anthrax in Ghana . To estimate the number of livestock and poor rural livestock keepers at risk in anthrax suitable areas , we dichotomized the modeled environmental suitability into a suitable versus not suitable prediction using a probability threshold that maximized sensitivity and specificity . We then overlaid a database of global livestock density at a spatial resolution of ~1km x 1km ( http://www . livestock . geo-wiki . org/ ) [29] with the dichotomized anthrax prediction to estimate the livestock populations ( cattle , sheep , goats , and swine ) at risk . Livestock populations at risk were further stratified to estimate the population at risk within each of the livestock production zones of Ghana using the livestock production systems data version 5 ( http://www . livestock . geo-wiki . org/ ) [29–31] . Furthermore , we estimated the number of low income rural livestock keepers at risk within each livestock production zone by overlaying the dichotomized anthrax suitable areas with estimates of the population of low income rural livestock keepers provided in Robinson et al . [31] and deriving the fraction of cells that were within our model prediction . Uncertainty in the populations at risk and 95% confidence intervals were calculated by using the 2 . 5% ( lower ) and 97 . 5% ( upper ) bounds of the ensemble RF model prediction [28] . Model performance and validation was conducted for each individual RF model and included the internal: OOB error classification , area under the receiver operating characteristics curve ( AUC ) , sensitivity , and specificity . Additionally , we performed accuracy assessments on the external testing data , which consisted of thirty random subsets of 20% of the data sampled with replacement . Mean values and 95% confidence intervals were estimated for each accuracy metric . The AUC has been used extensively in species distribution modeling to measure the discriminatory performance of models [32]; an AUC value of 1 indicates a perfect discrimination while values of >0 . 9 are outstanding , 0 . 8–0 . 9 excellent , 0 . 7–0 . 8 acceptable , and <0 . 7 indicate poor discriminatory performance [28 , 33] .
From 2005 through the first 6 months of 2016 , there were 67 reported anthrax outbreaks in livestock ( 61 that were geo-located ) ( Fig 1 ) . Nationally , there was a mean of 6 ( 95% CI: 4 , 7 ) outbreaks per year with a peak in 2011 ( n = 12 ) and lull in reporting in 2009 ( n = 2 ) ( Fig 2 ) . The geography of outbreaks shows a higher frequency of anthrax in northern Ghana in the Upper East and Northern regions . Of the reported outbreaks , 4 ( 6% ) were comprised of two or more livestock types . Domestic cattle were reported in 53% ( 35 ) of outbreaks , followed by sheep in 32% ( 21 ) , goats in 11% ( 7 ) , and swine in 5% ( 3 ) . During 2005–2016 , cattle anthrax cases were reported every year except in 2009 . Sheep cases were ubiquitous annually and were characterized by a large number of deaths in 2012 , the same year there was also a large number of swine cases ( n = 500 ) ( Table 2 ) . The seasonality of anthrax outbreaks nationally and regionally are illustrated in Fig 3 . Nationally , outbreaks were reported , on average , across seasons and in every month ( except November ) . There was an increase in outbreaks in the late winter and early spring months , with February through April having the highest reported number of outbreaks . On average , there outbreaks appeared to occur in the dry season before the onset of the rains . Trends in livestock anthrax vaccination among livestock type are shown in Fig 4 . From 2000–20015 , there was a median of 17 , 957 doses [0–175 thousand] of anthrax livestock vaccine administered annually livestock vaccination occurred annually with a median number of doses administered of 19 , 709 [range: 0–175 thousand doses] , followed by a decline in vaccine administration during 2008–2015 . No vaccination was administered during the years 2010 , 2012 , and 2013 . During 2008–2015 , there was a median of 542 [range: 0–147 thousand doses] doses administered . In response to ongoing outbreaks , there was a vaccination campaign in 2014 that resulted in nearly an 8-fold increase in the number of doses administered compared to the previous six years . Among livestock types , cattle were most frequently administered vaccine , followed by sheep , goats and swine ( Fig 4 ) . The ensemble RF model suggests a latitudinal gradient in the environmental suitability of anthrax in Ghana ( Fig 5A ) . High environmental suitability was identified in the Northern , Upper East , and Upper West regions of Ghana that encompass seasonal livestock migration routes from Burkina Faso in the north . Conversely , low or no environmental suitability was identified in southern Ghana among the more acidic soils in the Western , Ashanti , Central , and Eastern regions . Uncertainty ( range: 0–0 . 20 ) in the model prediction was scaled from 0 to 1 and showed it was highest in the Upper West and Northern regions ( Fig 5B ) . The internal OOB model validation indicated excellent discrimination with an AUC = 0 . 88 ( 95% CI: 0 . 87 , 0 . 89 ) . The external validation of anthrax outbreak locations withheld from the model ( testing data ) also showed excellent discrimination ( AUC = 0 . 87 [95% CI: 0 . 85 , 0 . 90] ) . The final list of variables used in the ensemble model are shown in Fig 6 . A combination of bioclimatic , environmental , and soil characteristics had the greatest impact on the OOB prediction errors . The most important variables influencing accuracy were: soil pH , bio7 ( annual temperature range ) , and bio14 ( precipitation of the driest month ) ( S2 Fig ) . The probability of the occurrence of anthrax increased in a step like manner in response to soil pH , increasing as the soil became more alkaline , between 5 . 5 and 6 . 5 , and again between 6 . 5 and 7 . 0 . Annual temperature ranges between 16 and 20°C were also related to a greater probability of occurrence . The occurrence of anthrax showed an affinity for low values of precipitation during the driest month ( 0 to 10 mm ) and then dropped off dramatically as precipitation increased from 10 to 40 mm . Furthermore , as average NDVI ( wd0114a0 ) increased from 0 . 3 to 0 . 6 the probability of anthrax occurrence decreased linearly , with a more suitable range of vegetation greenness identified in the lower ranges between 0 . 1 and 0 . 3 ( Fig 6 ) . To estimate livestock and human populations at risk , we dichotomized the environmental suitability prediction ( on a continuous probability scale ) into suitable versus non-suitable environments for anthrax based on the optimal threshold ( 0 . 46 ) that maximized sensitivity ( 0 . 78 ) plus specificity ( 0 . 89 ) ( Fig 7 ) . The dichotomized prediction shows a marked north-south demarcation in the suitability of anthrax , with a majority of northern Ghana predicted as suitable within the accompanying upper ( 97 . 5% ) and lower ( 2 . 5% ) confidence bounds . The national livestock population located in areas environmentally suitable for anthrax was estimated to be ≈ 2 . 2 ( 95% CI: 2 . 0 , 2 . 5 ) million ( Table 3 ) . More than 50% of the livestock populations at risk were sheep and cattle ( 650 [95% CI: 583 , 745] thousand and 480 [95% CI: 434 , 527] thousand , respectively ) . Among livestock production systems , semi-arid rain-fed , mixed crop livestock systems ( MRA ) contained the greatest number of livestock at risk > 1 . 2 ( 95% CI: 1 . 1 , 1 . 3 ) million ( Table 3 ) . Nationally , there are approximately 3 million low income rural livestock keepers in Ghana ( Table 4 ) . Our model suggests that ≈ 805 ( 95% CI: 519 , 890 ) thousand are located in areas suitable for anthrax , with the majority located in a humid and sub-humid , mixed crop livestock system production zone ( MRH ) .
Anthrax is a globally distributed neglected disease that is often underreported , particularly in West Africa where it is hyper-endemic [1 , 2 , 6 , 13] . Given the reliance of control on the vaccination of livestock , understanding the occurrence of anthrax is crucial for identifying populations at risk in order to disseminate limited resources . Here , we used data on the location of livestock outbreaks to identify seasonal patterns and model the environmental suitability of anthrax in Ghana . In keeping with previous studies , our findings indicate a defined outbreak season with a combination of ecological constraints on the potential geographic distribution of anthrax [3 , 34] . Our modeled prediction suggests a marked ecological divide separating the broad areas of environmental suitability in northern Ghana from the southern part of the country . Additionally , we estimated that populations characteristically at high risk for anthrax , which included >3 million combined ruminant livestock and poor rural livestock keepers are situated within the predicted anthrax risk zone . Based on our estimates , current anthrax vaccination efforts cover only a fraction of the livestock potentially at risk . Hence , these findings can be used to better direct public health intervention strategies and inform surveillance . Official reports of livestock anthrax in endemic areas often go undocumented for a number of reasons , including the inability or unwillingness to report , limited surveillance capacity , and a lack of local knowledge about the disease [1] . In Ghana , livestock cases are likely underreported due to the slaughter and consumption of sick or dead animals [8 , 35] , consistent with findings in the Caucasus and elsewhere [1 , 6 , 36 , 37] . This practice is often undertaken as a means of recouping economic losses from livestock mortality as well as providing food and a readily available source of protein [1 , 8 , 35] . The livestock anthrax outbreak data we used in this study were concordant with data reported to OIE during the same time frame suggesting Veterinary Services in Ghana are compliant with international reporting requirements ( http://www . oie . int/wahis_2/public/wahid . php/Wahidhome/Home ) . Despite the close proximity to the equator , we identified marked seasonality in anthrax reporting; outbreaks increased during the onset of the rainy season from February through April . Similar patterns of anthrax outbreaks associated with the rainy-season have also been reported in Namibia [34] . One hypothesis suggests that there is greater soil consumption among ruminants during with the rainy season [34] , although soil exposure during the dry season has also been hypothesized as a cause of anthrax outbreaks [1] . Regardless , these findings suggest vaccination of livestock could be carried out in Ghana ahead of the peak outbreak season ( September–November ) . Livestock anthrax control in Ghana follows a similar trend in many endemic regions of reactively vaccinating in response to anthrax outbreaks [1 , 38] . In Ghana , the livestock population we identified at risk comprises approximately ≈ 25% of the total national livestock population [29] . Based on official vaccination reports ( Fig 4 ) , our estimates of the livestock populations at risk indicates poor vaccine coverage; this finding is consistent with ongoing outbreaks in endemic communities in Ghana where vaccination has not been officially documented for at least a decade [39] . Barriers to vaccine uptake such as practices of livestock keepers my also affect coverage [1 , 40] . However , Ghana faces additional control challenges with the potential presence of B . cereus biovar ( bv ) anthracis and West Africa strains ( D and E Clades , respectively [41] ) . The West African strains have been hypothesized to evade the Sterne vaccine , which is the vaccine used in Ghana and throughout much of the world [13 , 42] . Further research is needed on vaccine efficacy and to understand what proportion of anthrax outbreaks are due to either insufficient application methods or the vaccine itself . Research has suggested that soil pH >6 . 1 in conjunction with high calcium levels are a crucial component of B . anthracis spore survival [1 , 4 , 43] . Alkaline soils were also found to be associated with the persistence of anthrax transmission over several years [43 , 44] . In keeping with these findings , we identified an increasingly higher likelihood of anthrax occurrence in soils as pH increased from 5 . 5 to 7 . 0 and with an increasing level of calcareous vertisols . The association of anthrax suitability with lower levels of precipitation in our model is in line with reports that have documented soil nutrient leaching in regions with high precipitation , which may lead to soil acidification [45] . We predicted an area of environmental suitability for anthrax that encompasses ≈ 36% of Ghana’s total area ( Fig 7 ) ; this is demarcated by a south ( largely unsuitable ) to north ( highly suitable ) divide , which closely mirrors the ecotone transitions from southern tropical and deciduous forests to the northern Sudanian and Guinea Savanna . Our study had several limitations . As with all neglected zoonoses , our data likely represent an underestimation of the true burden of disease due to underreporting and limited resources for surveillance and testing . To better address issues with diagnostic testing and reporting we used a more contemporary dataset of anthrax outbreaks recorded during the last decade . Anthrax can also be transmitted from contaminated feed that is imported , and animal mortality may occur from livestock moved across long distances; however , we had no information on any outbreaks arising in these instances [1 , 46] . The use of machine learning algorithms to model the distribution of environmental pathogens has been well described , but such approaches , by their definition in conjunction with the use of averaged climate data , may over-generalize the landscape that supports the occurrence of anthrax outbreaks . Other factors not included in our models that may influence the occurrence of anthrax include the health and immune status of the livestock [47] . In conclusion , the current anthrax situation in West Africa , and in particular Ghana , remains a public and veterinary health threat due to challenges with reporting , surveillance , and control . Our findings suggest that broad areas of northern Ghana are environmentally suitable for anthrax . Furthermore , based on recent vaccination efforts , our estimates indicate that only a fraction of livestock at risk are being vaccinated . These findings can be used to help improve differential diagnostics , vaccine coverage estimates , and surveillance efforts . Given the reliance on agriculture and the large population of low income rural livestock keepers at risk in the northern part of the country where predicted suitability was highest , future control efforts should focus on improving livestock vaccination coverage and public awareness of the disease , prioritizing communities in the predicted anthrax zone .
|
Anthrax is a soil-borne zoonotic disease found worldwide . In the West African nation of Ghana , anthrax outbreaks occur annually with a high burden to livestock keepers and their animals . To control anthrax in both humans and animals , annual livestock vaccination is recommended in endemic regions . However , in resource poor areas distributing and administering vaccine is difficult , in part , due to underreporting , logistical issues , limited resources , and an under appreciation of the geographic extent of anthrax risk zones . Our objective was to model high spatial resolution anthrax outbreak data , collected in Ghana , using a machine learning algorithm ( random forest ) . To achieve this , we used a combination of climatic and environmental characteristics to predict the potential environmental suitability of anthrax , map its distribution , and identify livestock and human populations at risk . Results indicate a marked ecological divide separating the broad areas of environmental suitability in northern Ghana from the southern part of the country , which closely mirrors the ecotone transitions from southern tropical and deciduous forests to the northern Sudanian and Guinea Savanna . Based on our model prediction , we estimated >3 million combined ruminant livestock and low income livestock keepers are situated in anthrax risk zones . These findings suggest a low level of annual livestock vaccination coverage among high risk groups . Thus , integrating control strategies from both the veterinary and human health sectors are needed to improve surveillance and increase vaccine dissemination and adoption by rural livestock keepers in Ghana and the surrounding region .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"livestock",
"medicine",
"and",
"health",
"sciences",
"immunology",
"geographical",
"locations",
"vaccines",
"preventive",
"medicine",
"bacterial",
"diseases",
"infectious",
"disease",
"control",
"vaccination",
"and",
"immunization",
"africa",
"veterinary",
"science",
"veterinary",
"medicine",
"public",
"and",
"occupational",
"health",
"infectious",
"diseases",
"anthrax",
"veterinary",
"diseases",
"zoonoses",
"livestock",
"care",
"agriculture",
"people",
"and",
"places",
"ghana",
"biology",
"and",
"life",
"sciences"
] |
2017
|
Modeling the environmental suitability of anthrax in Ghana and estimating populations at risk: Implications for vaccination and control
|
Chagas disease ( CD ) is a major public health concern in Latin America and a potentially serious emerging threat in non-endemic countries . Although the association between CD and cardiac abnormalities is widely reported , study design diversity , sample size and quality challenge the information , calling for its update and synthesis , which would be very useful and relevant for physicians in non-endemic countries where health care implications of CD are real and neglected . We performed to systematically review and meta-analyze population-based studies that compared prevalence of overall and specific ECG abnormalities between CD and non-CD participants in the general population . Six databases ( EMBASE , Ovid Medline , Web of Science , Cochrane Central , Google Scholar and Lilacs ) were searched systematically . Observational studies were included . Odds ratios ( OR ) were computed using random-effects model . Forty-nine studies were selected , including 34 , 023 ( 12 , 276 CD and 21 , 747 non-CD ) . Prevalence of overall ECG abnormalities was higher in participants with CD ( 40 . 1%; 95%CIs=39 . 2-41 . 0 ) compared to non-CD ( 24 . 1%; 95%CIs=23 . 5-24 . 7 ) ( OR=2 . 78; 95%CIs=2 . 37-3 . 26 ) . Among specific ECG abnormalities , prevalence of complete right bundle branch block ( RBBB ) ( OR=4 . 60; 95%CIs=2 . 97-7 . 11 ) , left anterior fascicular block ( LAFB ) ( OR=1 . 60; 95%CIs=1 . 21-2 . 13 ) , combination of complete RBBB/LAFB ( OR=3 . 34; 95%CIs=1 . 76-6 . 35 ) , first-degree atrioventricular block ( A-V B ) ( OR=1 . 71; 95%CIs=1 . 25-2 . 33 ) , atrial fibrillation ( AF ) or flutter ( OR=2 . 11; 95%CIs=1 . 40-3 . 19 ) and ventricular extrasystoles ( VE ) ( OR=1 . 62; 95%CIs=1 . 14-2 . 30 ) was higher in CD compared to non-CD participants . This systematic review and meta-analysis provides an update and synthesis in this field . This research of observational studies indicates a significant excess in prevalence of ECG abnormalities ( 40 . 1% ) related to T . cruzi infection in the general population from Chagas endemic regions , being the most common ventricular ( RBBB and LAFB ) , and A-V B ( first-degree ) node conduction abnormalities as well as arrhythmias ( AF or flutter and VE ) . Also , prevalence of ECG alterations in children was similar to that in adults and suggests earlier onset of cardiac disease .
Chagas disease ( CD ) , is caused by the protozoan Trypanosoma cruzi[1] . It affects individuals from more than 21 countries , being a major public health concern in Latin Americas and a potentially serious emerging threat to a number of non-endemic countries[2] . In Latin American countries , there are currently 8-10 million people having CD , with an additional 300 , 000 individuals in the United States and 45 , 000-67 , 000 in Europe[3–6] . Infection with T . cruzi has two phases: acute and chronic , separated by an indeterminate period , in which the patient is relatively asymptomatic[7 , 8] . Chronically infected patients ultimately develop cardiomyopathy , which is the most important and severe manifestation of CD , and is characterized by left ventricular systolic dysfunction , wall motion abnormalities , brady and tachyarrhythmia , heart failure and sudden cardiac death[9–18] . CD alterations are classified in four stages A , B , C , and D . Stage A corresponds to asymptomatic patients with normal ECG , whereas presence of electrocardiographic abnormalities implies progression towards stage B and deterioration of systolic function is observed in stages C/D , associated with heart failure symptoms . Of note , sudden cardiac death may occur at any moment , including early phases[10 , 16] . The association between CD and cardiac abnormalities ( stages B or more ) is widely reported , however , the information is challenged by the diverse design , sample size and quality of the studies[19–24] . This information is largely based on outdated individual studies , and reports vary substantially among the population based studies making the scientific interpretation challenging[8 , 20 , 25–28] . Furthermore , it is known that the prevalence of different types of ECG abnormalities such as intraventricular conduction abnormalities , atrioventricular block , and arrhythmias , is higher in subjects with CD as compared to non-CD subjects , calling for its update and synthesis , which would be very useful and relevant for physicians in non-endemic countries where health care implications of CD are real and neglected[29] . Given the spread of CD , a need exists for an adequate , comprehensive assessment of CD in association with ECG abnormalities[6 , 29 , 30] . We conducted a systematic review and meta-analysis to ( i ) determine the overall prevalence of ECG abnormalities in seropositive and seronegative CD individuals in the general population; ( ii ) quantify the prevalence of subtypes of ECG abnormalities in this population; and ( iii ) characterize these estimates by age , sex and region of origin , and ( iv ) evaluate the discriminatory capacity of ECG abnormalities , both general and specific to classify patients as either CD or non-CD .
This review was conducted in accordance with MOOSE guideline[31] and , we followed the protocol strictly without deviation from it . An extensive literature search of articles published up to March 2017 was conducted with the assistance of a medical librarian in six electronic databases ( EMBASE , Ovid Medline , Web of Science , Cochrane Central , Google Scholar and Lilacs ) without any language restriction . Search combined terms related to CD ( Chagas Disease , Trypanosoma cruzi ) , with terms related to seroepidemiology ( Seroepidemiologic Studies , Seroprevalence , Seroepidemiology ) ( S1 Appendix ) . Reference lists of selected studies and reviews identified on the topic were searched to identify additional publications . Studies were eligible if they: ( i ) selected cases and controls from the general population ( surveys and blood donors ) ; ( ii ) were cross-sectional , case-control and cohort studies; ( iii ) reported CD status based on any techniques for the detection of antibodies; and ( iv ) reported the prevalence of ECG abnormalities in CD participants and non-CD participants , based on electrocardiographic diagnosis . To avoid overestimation of the effect estimates , we did not include studies using cases and controls from a clinical setting . Two independent reviewers screened the titles and abstracts of all initially identified studies according to the selection criteria . Full texts were retrieved from studies that fulfilled all selection criteria . Any disagreements were resolved through consensus or consultation with a third independent reviewer . The data collection form included questions on qualitative aspects of the studies ( such as author , date of publication , country , design , period , setting , area and sample size ) , participant characteristics ( such as age , sex and serological test of CD ) and information on the reported exposure/outcome ( such as measure definition of ECG abnormalities , number of patients seropositive/seronegative CD , number of patients with general and specific ECG abnormalities and confounders adjustment ) . For cohort studies , only the information from the baseline assessment was extracted . Study quality was assessed by two independent reviewers based on the nine-star Newcastle-Ottawa Scale ( NOS ) using three predefined domains , namely: selection of participants ( population representativeness ) , comparability ( adjustment for confounders ) and ascertainment of outcomes of interests[32] . The NOS assigns a maximum of four or five points for selection , one or two points for comparability and three points for outcome , depending on study design ( cross-sectional or cohort ) . We used the NOS scale adapted for cross-sectional studies and the NOS for case-control studies while for cohort studies we used the NOS for a cross-sectional design ( S2 Appendix ) . Studies that received a score of nine to seven stars were judged to be at low risk of bias; studies that scored five or six stars were considered at medium risk , and those that scored four or less were considered at high risk of bias . Narrative synthesis and construction of descriptive summary tables were performed for the studies included . For this meta-analysis , we used odds ratios ( ORs ) and 95% confidence intervals ( CIs ) , to assess the association between presence of CD , and overall and specific ECG abnormalities . Also , we calculated the pooled prevalence ( P ) of ECG abnormalities for CD and non-CD participants . The inverse variance weighted method was used to combine summary measures using random-effects models to minimize the effect of between-study heterogeneity[33] . Heterogeneity was assessed using the Cochrane χ2 statistic and the I2 statistic and was categorized as low ( I2 ≤25% ) , moderate ( I2 >25% and <75% ) , or high ( I2 ≥75% ) [34] . Sensitivity analyses were performed to assess the influence of study-level characteristics including publication year , geographical location ( according to the number of study in each country and by the prevalence of Trypanosoma cruzi genotype ) , design , area , number and age of participants , sex , definition of ECG abnormalities , test for CD diagnoses , level of adjustment and risk of bias , which were pre-specified as characteristics for assessment of heterogeneity and were evaluated using stratified analysis and random-effects meta-regression[35] . Publication bias was evaluated through funnel plots and Egger’s regression symmetry tests[36] . All tests were 2-tailed; p-value ≤0 . 05 was considered statistically significant . Stata release 15 ( StataCorp ) was used for all analyses . Additionally , we evaluated the discriminatory capacity of ECG abnormalities , both general and specific: sensitivity , specificity , positive predictive value ( PPV ) , negative predictive value ( NPV ) to classify patients as either CD or non-CD .
The database searches identified 5 , 396 citations . After screening the titles and abstracts , 252 articles were selected for detailed evaluation of full texts . Of these , 49 articles met our inclusion criteria and were included in the review ( Fig 1 , S3 Appendix and S1–S3 Tables ) . General characteristics of included studies are shows in S1–S3 Tables summarize the key characteristics of the included studies . Of 49 included studies , 43 were cross-sectional and 6 cohort population-based studies and were published between 1964 and 2015 . Studies were based on data from 10 Latin America countries . The majority of studies was conducted in Brazil ( 38 . 7% ) , collected data between 2001 and 2010 ( 34 . 6% ) , and included participants from rural areas ( 44 . 9% ) . In aggregate , 34 , 023 participants ( 12 , 276 CD and 21 , 747 non-CD ) were included in this review . The age range of the participants was 0-97 years and 54 . 4% of participants were women ( reported by 38 articles ) . The majority of studies ( 81 . 6% ) used two or more different diagnostic tests for T . cruzi infection . In the meta-analysis of 34 , 023 participants and 12 , 276 CD participants , overall ECG abnormalities were significantly more prevalent in participants with CD ( Prevalence ( P ) =40 . 1%; 95%CIs=39 . 2-41 . 0 ) compared to those without CD ( P=24 . 1%; 95%CIs=23 . 5-24 . 7 ) ( OR=2 . 78; 95%CIs=2 . 37-3 . 26 ) ( Fig 2 and Table 1 ) . Up to 30 studies ( range 3 to 30 ) , based on 3 , 451 ± 1 , 350 participants , examined prevalence of ventricular conduction defects in CD and non-CD participants ( Table 2 ) . In the pooled analysis , prevalence of complete right bundle branch block ( RBBB ) ( OR=4 . 60; 95%CIs=2 . 97-7 . 11 ) , left anterior fascicular block ( LAFB ) ( OR=1 . 60; 95%CIs=1 . 21-2 . 13 ) and , the combination of complete RBBB and LAFB ( OR=3 . 34 , 95%CIs=1 . 76-6 . 35 ) was higher in CD participants compared to non-CD participants ( Fig 3 ) , while no difference in prevalence of incomplete RBBB ( OR=0 . 85; 95%CIs=0 . 63-1 . 16 ) , incomplete left bundle branch block ( LBBB ) ( OR=0 . 63; 95%CIs=0 . 39-1 . 01 ) , complete LBBB ( OR=0 . 99; 95%CIs=0 . 62-1 . 58 ) and left posterior fascicular block ( LPFB ) ( OR=0 . 96 , 95%CIs=0 . 35-2 . 62 ) was found between CD and non-CD individuals ( Table 2 and S1–S7 Figs ) . Prevalence of AV-B in CD and non-CD participants was examined by 21 studies ( range 5 to 21 ) , including 4 , 438 ± 1 , 847 participants . Pooled prevalence first degree AV-B ( OR=1 . 71; 95%CIs=1 . 25-2 . 33 ) was higher in CD participants compared to non-CD ( Fig 3 ) , while no difference in prevalence of second-degree AV-B ( OR=1 . 15; 95%CIs=0 . 33-4 . 06 ) and third-degree AV-B ( OR=2 . 10; 95% CIs=0 . 60-7 . 34 ) was found between CD and non-CD participants ( Table 2 and S8–S10 Figs ) . Twenty-five studies ( range 11 to 25 studies ) , based on 4 , 848 ± 1 , 591 participants , compared prevalence of arrhythmias between CD and non-CD participants . Pooled prevalence of AF or flutter ( OR=2 . 11 , 95%CIs=1 . 40-3 . 19 ) and ventricular extrasystoles ( VE ) ( OR=1 . 62; 95%CIs=1 . 14-2 . 30 ) was higher in CD participants ( Table 2 , Fig 3 and S11–S14 Figs ) . In a pooled analysis of 10 studies , comprising 5 , 575 individuals , no statistically significant difference in prevalence of other ECG abnormalities ( Low voltage QRS , OR=0 . 84; 95%CIs=0 . 47-1 . 50 ) was found between CD and non-CD ( Table 2 and S15 Fig ) . Regarding of study quality and assessment bias , among the 49 included studies , 10 studies were judged to be at low risk of bias , 28 were at medium risk , and 11 studies were evaluated to be at high risk of bias ( Table 1 , S2 Appendix , S3 Table and S16 Fig ) . In the sensitivity analysis , of 16 meta-analyses , while five showed no heterogeneity ( I2=0 , p>0 . 05 for the Cochrane χ2 statistic ) , four showed low ( I2=<25% ) , six moderate and one high between-study heterogeneity , four of which with an I2 estimate exceeding 50% ( overall ECG abnormalities , complete RBBB , LAFB and VE ) ( Fig 2 , Tables 1 and 2 and S4–S6 Tables ) . No study characteristics could explain the heterogeneity observed in overall ECG abnormalities ( p-value of meta-regression ≥0 . 05 ) ( Table 1 and S17–S27 Figs ) . Additionally , S28 Fig shows meta-analyses estimates of overall ECG abnormalities excluding single studies one by one , with no evidence of change in the pooled . Similarly , the stratified analysis by sex and age , although showed higher magnitudes of the association in children and men , the effects were not significantly different between sex and age strata: higher OR was observed in children ( OR=3 . 88; 95%CIs=1 . 69-8 . 88; I2=63 . 9% ) compared with all ages or ≥10 years categories; however , when a small-sample study was omitted , the OR of children was similar to that of the other categories ( OR=2 . 83; 95%CIs=2 . 26-3 . 53; I2=0 . 0% ) without being statistically significant in the meta-regression analysis ( p-value of meta-regression=0 . 704 ) , and comparison of overall ECG abnormalities by sex , could be quantified only in a limited number of studies ( 12 . 2% ) ( Table 1 , S25 , S26 and S29 Figs ) . For the studies on the association between CD , complete RBBB , LAFB and VE , the identified heterogeneity was largely explained by geographic location , area , number of participants , number of tests used for diagnoses of CD , level of adjustment and risk of bias ( S4–S6 Tables ) . Considering a P value of 0 . 10 , rather than the conventional level of 0 . 05 , is sometimes used to determine statistical significance for heterogeneity , we investigated further the sources of heterogeneity across our meta-analyses using a p-value lower than 0 . 10 as statistical significant . When setting the p-value threshold to 0 . 10 , the findings of meta-regression were similar as when setting the p-value threshold of 0 . 05 , except for the characteristics “location of the study” which on p-value of 0 . 10 , showed to contribute to heterogeneity of the meta-analysis for overall ECG abnormalities . Under visual examination , funnel plots for studies assessing the general and specific prevalence of ECG abnormalities were symmetrical , providing evidence for no publication bias . This was further supported by the results of Egger’s test , showing no evidence of publication bias , either graphically from the funnel plot , or quantitatively ( p≥0 . 05 for Egger’s test asymmetry ) ( S7 Table and S30 Fig ) . Finally , Sensitivity , specificity , PPV and NPV of general ECG abnormalities were 40 . 09% , 75 . 89% , 48 . 42% and 69 . 18% , respectively . PPV of specific ECG abnormalities for diagnosis of CD varied , ranging between 61 . 64% and 83 . 08% , with two ECG abnormalities showing values over 80%: complete RBBB ( 80 . 77%; 95%CIs=78 . 17-83 . 13 ) and the combination of complete RBBB and LAFB ( 83 . 08%; 95%CIs= 77 . 36-87 . 59 ) ( S8 Table ) .
This systematic review and meta-analysis provides an update and synthesis in this field . This research of observational studies indicates a significant excess in prevalence of ECG abnormalities ( 40 . 1% ) related to T . cruzi infection in the general population from Chagas endemic regions , being the most common ventricular ( RBBB and LAFB ) , and A-V B ( first-degree ) node conduction abnormalities as well as arrhythmias ( AF or flutter and VE ) . Also , prevalence of ECG alterations in children was similar to that in adults and suggests earlier onset of cardiac disease .
|
Chagas disease ( CD ) is a major public health concern in Latin America and a potentially serious emerging threat in non-endemic countries . Although the association between CD and cardiac abnormalities is widely reported , study design diversity , sample size and quality challenge the information , calling for its update and synthesis , which would be very useful and relevant for physicians in non-endemic countries where health care implications of CD are real and neglected . We systematically reviewed and meta-analyzed population-based studies that compared prevalence of ECG abnormalities between Chagas disease ( CD ) and non-CD participants . Forty-nine studies were selected , including 34 , 023 unique participants . Our meta-analysis of observational studies indicates CD presented almost a threefold increase prevalence of ECG abnormalities than non-CD participants in the general population from Chagas endemic regions , being the most common ventricular ( RBBB and LAFB ) , and A-V B ( first-degree ) node conduction abnormalities as well as arrhythmias ( AF or flutter and VE ) .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"and",
"health",
"sciences",
"tropical",
"diseases",
"database",
"searching",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"electrocardiography",
"mathematics",
"protozoans",
"statistics",
"(mathematics)",
"bioassays",
"and",
"physiological",
"analysis",
"neglected",
"tropical",
"diseases",
"research",
"and",
"analysis",
"methods",
"cardiology",
"public",
"and",
"occupational",
"health",
"arrhythmia",
"atrial",
"fibrillation",
"mathematical",
"and",
"statistical",
"techniques",
"protozoan",
"infections",
"electrophysiological",
"techniques",
"research",
"assessment",
"trypanosoma",
"cruzi",
"trypanosoma",
"chagas",
"disease",
"eukaryota",
"cardiac",
"electrophysiology",
"meta-analysis",
"systematic",
"reviews",
"database",
"and",
"informatics",
"methods",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"statistical",
"methods",
"organisms"
] |
2018
|
Electrocardiographic abnormalities in Chagas disease in the general population: A systematic review and meta-analysis
|
Antisense transcription is a prevalent feature at mammalian promoters . Previous studies have primarily focused on antisense transcription initiating upstream of genes . Here , we characterize promoter-proximal antisense transcription downstream of gene transcription starts sites in human breast cancer cells , investigating the genomic context of downstream antisense transcription . We find extensive correlations between antisense transcription and features associated with the chromatin environment at gene promoters . Antisense transcription downstream of promoters is widespread , with antisense transcription initiation observed within 2 kb of 28% of gene transcription start sites . Antisense transcription initiates between nucleosomes regularly positioned downstream of these promoters . The nucleosomes between gene and downstream antisense transcription start sites carry histone modifications associated with active promoters , such as H3K4me3 and H3K27ac . This region is bound by chromatin remodeling and histone modifying complexes including SWI/SNF subunits and HDACs , suggesting that antisense transcription or resulting RNA transcripts contribute to the creation and maintenance of a promoter-associated chromatin environment . Downstream antisense transcription overlays additional regulatory features , such as transcription factor binding , DNA accessibility , and the downstream edge of promoter-associated CpG islands . These features suggest an important role for antisense transcription in the regulation of gene expression and the maintenance of a promoter-associated chromatin environment .
The promoter region is intimately tied to the transcription of genes , providing the initial site of transcriptional machinery binding and assembly . Comprised of DNA elements in defined spatial arrangements [1] , promoters also display key genomic features that facilitate gene regulation . Active promoters present a nucleosome-deprived region that allows for the association of RNA polymerase II ( Pol II ) and transcription factors [2] . Active promoters also possess distinct histone marks associated with gene expression [3] . Divergent transcription has emerged as a common feature of mammalian promoters [4–7] . In divergent transcription , an additional transcription event initiates upstream and antisense of a nearby gene promoter . Though divergent transcription at promoters may result in two different protein-coding transcripts in opposite orientations , often a short-lived non-coding RNA is transcribed anti-sense of a gene [8–10] . In divergent transcription at mammalian promoters , antisense transcription initiates at the upstream antisense transcription start site ( uaTSS ) . The position of the uaTSS intersects with distinguishing features associated with promoters , with uaTSSs falling at the border of nucleosome-depleted regions where transcription factor binding may be enriched [11] . Additionally , uaTSSs tend to broadly coincide with the upstream edge of promoter-associated CpG islands [11] . In addition to divergent transcription , convergent transcription has been observed at genes in a variety of systems , ranging from fly and yeast to mammals [12–14] . Convergent transcription is possible in a variety of different gene structures , including non-overlapping genes on opposite strands and genes with internal promoters [13] . Promoter-proximal convergent transcription is common; analysis identified convergent transcription at roughly one quarter of queried genes [15] . Here , we describe genetic and epigenetic features at downstream antisense transcription start sites ( daTSSs ) associated with promoter-proximal convergent transcription in human T47D/A1-2 cells . We find that daTSSs coincide with the downstream edge of promoter-associated genomic features , such as promoter-associated histone marks . Though convergent transcription has been suggested as a repressive feature [15] , we find that genes with observable daTSSs do not display lower gene expression in T47D/A1-2 cells . Despite this , coincidence of daTSSs with a variety of transregulatory factors , such as transcription factors and chromatin remodelers , suggests an intimate connection between antisense transcription and gene regulation .
To characterize transcription initiation in human T47D/A1-2 cells , Start-seq was performed on nascent RNA transcripts [16] . In a Start-seq experiment , RNA is isolated from the nucleus and selected for short size , allowing characterization of nascent RNA transcripts that may be subsequently degraded and that otherwise could not be analyzed in a traditional RNA-seq experiment . Cap-sensitive degradation ensures that the 5’ end of each read in a Start-seq experiment corresponds to a TSS with single nucleotide resolution . Gene TSSs and uaTSSs were identified using previously described methods [11] . daTSSs were called in a method analogous to that used for uaTSS identification . In brief , for each gene TSS called , a search window from the gene TSS to 2 kb downstream was defined . For each search window with reads exceeding an FDR-defined significance threshold ( 5 reads , observing aligned library depth ) , a daTSS was called at the position with greatest read density on the opposite strand relative to the gene TSS . Stringent filtering was used to ensure that an identified daTSS could not be a miscalled TSS of another gene or a uaTSS of an alternative start site . Observable genes with antisense transcription display overlap between identified TSSs and the 5´ end of Start-seq reads on both sense and antisense strands ( in black and in red; Fig 1A and 1B , respectively ) . Identified gene TSSs are consistent with RNA-seq read density , and both gene TSSs and antisense TSSs display overlap with Pol II ChIP-seq reads ( in blue and red , respectively; Fig 1A ) . Over 10 , 391 observed gene TSSs , 5 , 519 uaTSSs ( 53% of gene TSSs ) and 2 , 956 daTSSs ( 28% ) were identified ( Fig 1B ) . Over all observed gene TSSs , both a uaTSS and a daTSS were identified at 1 , 815 genes , indicating a statistically significant association between the two events and implying potential cooperation between upstream and downstream antisense transcription ( two-sided Fisher’s exact test: p-value < 10−6 ) . To ensure reproducibility , a biological replicate was prepared from the same cell-line , and similar read densities were found at the identified TSSs ( S1A Fig ) . The overall rate of identification is consistent with previous approaches using other experimental methods [7 , 11 , 15] . Start-seq read counts are greatest at gene TSSs , with lower counts at uaTSSs and daTSSs ( averages of 776 , 467 , and 276 reads at gene TSSs , uaTSSs , and daTSSs , respectively ) . Given that the same read threshold was used to call all classes of TSS and that stringent filtering was used to limit miscalling of antisense TSSs , we anticipate that our rate of identification underestimates the prevalence of antisense transcription . daTSSs were also identified in mouse macrophage cells considering observed gene TSS positions found in previous work [11] . 4 , 921 daTSSs were identified over 12 , 229 observed gene TSSs , or roughly 40% of genes ( S2 Fig ) . The difference in the identification rate between human T47D/A1-2 cells ( 28% ) and mouse macrophage cells ( 40% ) may be indicative of variability in the landscape of antisense transcription across organisms and cell types . Heatmaps centered on gene TSSs and sorted by the distance from the observed daTSS show that human T47D/A1-2 and mouse macrophage calls are coincident with enriched Pol II ChIP-seq signal ( Fig 1B; S2B Fig ) , supporting calls as bona fide Pol II-dependent transcription events . There is a lack of stranded RNA-seq coverage immediately downstream of uaTSSs and daTSSs ( S3A Fig ) , indicating that transcripts originating from these sites are not present at steady-state in the cytosol and are short-lived . daTSSs identified in T47D/A1-2 cells were investigated in other cell lines using other experimental data types associated with nascent transcription ( MCF-7 , GM12878 , K562 , HeLa S3 , and HEK239T cells; S3B Fig ) . These experiments include a variety of sequencing-based approaches designed to interrogate different biological phenomenon , including direct Pol II-DNA interactions and nascent Pol II-associated transcription . Despite wide ranging differences in technology and cell line , we found an enrichment of nascent RNA- or Pol II-associated signal at daTSS positions . Given the conservation of daTSSs across cell lines , we leveraged a variety of publically available data from different cell lines to annotate daTSSs . Additionally , though we found that a majority of daTSSs are preserved , some do not display signal in other cell lines ( S3C Fig ) . Of the 2 , 956 daTSSs identified in T47D/A1-2 cells , 1 , 985 ( 67% ) and 1 , 966 ( 67% ) display GRO-cap signal in K546 and GM12878 cells , respectively . We examined these genes to see if they displayed cell-specific characteristics . daTSSs with no GRO-cap signal in either K546 or GM12878 cells and that may be considered T47D/A1-2 specific ( 605 genes; 21% ) show an enrichment in categories associated with breast cancer ( “breast or ovarian cancer”: p-value = 1 . 28 x 10−5; “mammary tumor”: p-value = 2 . 10 x 10−5 ) . This enrichment suggests cell-line specificity in downstream antisense transcription . We examined sequence content at all observed TSSs to both further characterize identified TSSs as associated with Pol II-dependent transcription events and to compare across the three observed TSS classes . Generally , we find that sequence content is similar across all three classes . All TSSs show enrichment for GC content ( Fig 2A ) . Consistent with previous observations , we find enriched GC content upstream of T47D/A1-2 uaTSSs [11] . In addition , we see enriched GC content downstream of daTSSs . Taken together , the antisense TSSs coincide with apparent boundaries of GC content enrichment . The positioning of CpG islands is generally consistent with this observation . Observed uaTSSs and daTSSs broadly coincide with the upstream and downstream edges of promoter-associated CpG islands , respectively ( Fig 2B ) [17] . However , the alignment between CpG island boundaries of daTSSs is not absolute . Of those 2 , 392 daTSSs whose associated gene TSS overlaps a CpG island , 1 , 325 ( 55% ) are within 250 bp of the downstream edge of the CpG island . Thus , observation of a daTSS is not dependent on the presence and relative location of a promoter-proximal CpG island . Considering a narrow 100-bp window around each TSS ( Fig 2A ) , we observe sequence patterns that are preserved across each class of TSS . Roughly 25 bps upstream of the TSS , we identify an area enriched for TA content . Centered on the TSS itself , there is a distinct sequence pattern with pyrimidine-purine dinucleotide at its center . De novo motif discovery reveals enriched sequences similar to TATA box and Inr binding motifs at these regions ( Fig 2D ) [20] . Though these motifs are present near each class of TSS , subtle differences in relative enrichment of Inr-like motifs ( found at 24% , 17% , and 41% of gene TSSs , uaTSSs , and daTSSs , respectively; Fig 2D ) may reflect differences in regulation . The occurrences of these specific motifs are consistent with general patterns found in searches performed with Pol II-associated sequence motifs ( Fig 2C ) . Regardless of TSS class , an enrichment of Pol II-associated motifs [18] is observed upstream of TSS positions . GC-box occurrences dominate the enriched motifs ( S3D Fig ) with fewer motifs identified at daTSSs . We additionally investigated sequence conservation at daTSS positions . Over promoters displaying antisense transcription , we examined PhyloP conservation scores calculated from sequence alignments across placental mammals [19] . Positive PhyloP scores indicate enhanced sequence conservation immediately upstream of daTSSs ( Fig 2E ) . This is consistent with observations seen at both gene TSSs and uaTSSs [11] . Positive scores are also evident between gene TSSs and daTSSs , implying sequence conservation in this region ( Fig 2E ) . Evolutionary pressure to maintain sequence at daTSSs suggests functional significance for downstream antisense transcription . We next sought to examine the connection between downstream antisense transcription and the expression of associated genes . Recent studies have proposed that promoter-proximal downstream antisense transcription represses expression of upstream genes [15] . We examined the expression of genes displaying downstream antisense transcription in two distinct ways . We first compared the extent of transcription initiation at those genes with and without observed daTSSs ( categories “w/ daTSSs” and “w/o daTSSs”; Fig 3A ) . We restricted this comparison to those genes without observed uaTSSs to limit the effect of upstream antisense transcription on observed trends ( N = 4 , 872 ) . The difference between genes with and without daTSSs is not statistically significant ( Wilcoxon test: p-value = 0 . 37 ) . We also find that the extent of downstream antisense transcription as measured by Start-seq read counts at daTSS positions shows no correlation with read counts at gene TSSs ( genes with no observed uaTSSs; Spearman test: rho = 0 . 017 , p-value = 0 . 56 ) . We next compared steady-state transcript levels using RNA-seq-derived FPKM values . The results for RNA-seq analysis are comparable to those derived from Start-seq ( Fig 3B ) . Again , comparisons were restricted to genes without observed uaTSSs . The difference in FPKM values between genes with and without daTSSs is not statistically significant ( Wilcoxon test: p-value = 0 . 89 ) . In summation , downstream antisense transcription is not associated with lowly expressed genes in human T47D/A1-2 cells . These results suggest convergent transcription near gene promoters is not inhibitory . These results are distinct from those reported in recent a study describing promoter-proximal downstream antisense transcription as mark of lowly expressed genes [15] . This conclusion was based upon differences in the gene body density of Net-Seq reads between genes displaying only upstream antisense or downstream antisense transcription . When we use a similar categorical separation of genes , we find that genes displaying only uaTSSs show greater steady-state transcription levels than genes displaying only daTSSs ( Wilcoxon and Kolmogorov-Smirnov tests: p-values < 10−6; S4A Fig ) . However , these differences appear attributable not to the presence of downstream antisense transcription but to the absence of upstream antisense transcription . Considering genes without daTSSs ( N = 7 , 435 ) , those genes with uaTSSs display significantly higher steady-state transcription than genes without uaTSSs ( Wilcoxon test: p-value < 10−6; S4B Fig ) . These results suggest that a complete understanding of the effects of downstream antisense transcription requires deconvolution from that of upstream antisense transcription . An examination of promoter proximal pausing suggests that downstream antisense transcription may be coordinated with transcription of associated genes . We find that Ccnt2 ( or CycT2 ) , part of the P-TEFb complex involved in the regulation of Pol II elongation , selectively associates to the area between gene TSSs and daTSSs and shows statistically significant enrichment compared to equivalent regions at genes without daTSSs ( Wilcoxon test , p-value <10−6; Fig 3C ) . This is coincident with enrichment of components of NELF and DSIF ( Wilcoxon tests , p-values < 10−6; Fig 3C ) . PRO-seq experiments measuring strand-specific association of elongating Pol II also show a sense-strand enrichment of Pol II near the daTSS ( Fig 3C ) , further supporting a connection between Pol II pausing and antisense transcription . These results do not necessarily imply that genes displaying downstream antisense transcription are more paused . Pausing itself is a highly regulated event in gene transcription with connections to signal-dependent gene expression [21] . Pol II-dependent transcription antisense of genes may contribute to signal-dependent response of paused genes . Consequently , while downstream antisense transcription does not appear to correlate with steady-state transcription levels , antisense transcription could affect signal-dependent gene expression changes . We find that the daTSS position coincides with the downstream edge of a chromatin environment displaying promoter-associated features . To investigate the interplay between antisense transcription and nucleosome positioning , we performed MNase-seq on T47D/A1-2 cells . As evidenced in the resulting data , nucleosomes are regularly positioned relative to all three classes of identified TSSs . Like gene TSSs , MNase-seq read density is consistent with “+1” nucleosomes placed immediately downstream of daTSSs and uaTSSs , though MNase-seq peaks are less sharp when centered on daTSS positions ( Fig 4A ) . The daTSS location respects the regular positioning of promoter-proximal nucleosomes . A histogram of gene TSS-daTSS distances is anti-correlated with average MNase-seq density ( Fig 4B ) , implying that daTSSs fall between regularly-spaced nucleosomes oriented at gene TSSs . This pattern is reproduced in MNase-seq data from other breast epithelial cells ( MCF-7 cells; S1B Fig ) [22] . The observed MNase-seq density is consistent with Pol II ChIP-seq density at daTSS positions ( Fig 4A ) . Downstream of observed daTSS positions , we find a regular pattern of Pol II ChIP-seq read density that is attributable to Pol II initiating at gene TSSs and that mirrors the MNase-seq data . As previously described [11] , we find that regions between gene TSSs and associated uaTSSs are nucleosome depleted . This is clear in heatmaps of MNase-seq density ( Fig 4C , left ) . In comparison , the region between gene TSSs and daTSSs displays a regular pattern of positioned nucleosomes ( Fig 4C , left ) . When quartiled by TSS-daTSS distance , TSSs with greater distances show less distinct downstream MNase-seq peaks , perhaps indicating less regular or more transient nucleosome associations at these positions ( Fig 4C , right ) . However , the location of MNase-seq peaks downstream of gene TSSs seem to be similar across genes with and without daTSSs ( S4C Fig ) , indicating that the location of the gene TSS predominantly influences nucleosome positioning . Given that nucleosomes are regularly positioned at identified TSSs , we sought to characterize histone modifications in those regions . We find that histone modifications at those nucleosomes positioned between the gene TSS and daTSS are distinct from proximal regions . ChIP-seq data in HMEC cells [23] show an enrichment of histone marks associated with active promoters . When compared to equivalent positions at genes without daTSSs , H3K27ac and H3K3me3 modifications show significant enrichment by Wilcoxon test ( p-values of 1 . 17 x 10−3 and 3 . 94 x 10−2 for H3K27ac and H3K4me3 modifications , respectively; Fig 4D ) . There is tendency for histone modification enrichment to end at the daTSS , as clearly seen in profiles for histone variant H2A . Z ( Wilcoxon test: p-value < 10−6; Fig 4D ) . This same enrichment is not seen for H3K4me1 and H3K36me3 modifications , associated with enhancers and actively transcribed regions , respectively ( Wilcoxon test: p-values of 0 . 53 and 0 . 44 , respectively ) . Consistent with the lack of nucleosomes in this region , we do not see the same enrichment between the gene TSS and uaTSS ( S5 Fig ) . Given the observed histone modification profile , we next sought to characterize the relationship between the association of transregulatory factors and antisense transcription at gene promoters . The association of transcription factors is enriched at open regions of DNA . We characterized accessible regions of DNA in T47D/A1-2 cells using FAIRE-seq . Like gene TSSs and uaTSSs , FAIRE-seq reveals an open genomic region at daTSS positions ( Fig 5A ) [24] . This observed density is consistent with FAIRE-seq results reported in a previous study ( S1C Fig ) [24] and recapitulates observations made using DNase-seq [15] . Following characterization by FAIRE-seq , we performed protein motif searches in similar areas to characterize the potential of these areas to interact with DNA-binding proteins . Analysis of known vertebrate motif occurrences shows a depletion of protein-binding motifs between gene TSSs and daTSSs and an enrichment of motifs immediately upstream of daTSSs ( Fig 5B ) [18] . Consistent with these areas being open and enriched for protein-binding motifs , ChIP-seq data [23] reveal that the daTSS coincides with the binding of trans-regulatory factors . Transcription factors were found to associate with open regions at both the gene TSS and the daTSS ( Fig 5C , top ) . Comparisons with equivalent positions at genes without daTSSs show a significant enrichment of transcription factor-associated signal at daTSS positions ( Wilcoxon test: p-values < 10−6; Fig 5C , top , and S2 Table ) . We present selected transcription factors known to associate at gene promoters and broadly participate in a number of signal-dependent pathways . However , the coincidence of p300 , a known co-activator of numerous additional transcription factors [25] , implies potential interaction with many others at daTSS positions ( Fig 5C , top-right ) . In contrast , chromatin remodelers were found to associate in the area between the gene TSS and the daTSS with a significant enrichment of associated signal at daTSS positions ( Wilcoxon test: p-values < 10−6; Fig 5C , bottom , and S2 Table ) . Though this area displays high GC content , this alone does not explain the enrichment of chromatin remodelers; randomly selected genomic regions matched for GC content show diminished signal relative to these regions ( Wilcoxon test: p-value < 10−6 ) . Like daTSSs , a variety of trans-regulatory factors associate to uaTSS positions ( S6 , S7 and S8 Figs ) . However , differences in regions between gene and antisense TSSs distinguish these two classes of TSS . Unlike daTSSs , transcription factors associate to the nucleosome-depleted areas between gene TSSs and uaTSSs while chromatin remodelers do not . This likely reflects differences in nucleosome occupancy between nucleosome-rich TSS-daTSS regions and nucleosome-deprived TSS-uaTSS regions ( Fig 4 ) . Factors involved in the deposition of histone marks ( CHD1-A and Sap30 ) and in the positioning of nucleosomes ( SWI/SNF-associated factors ) may contribute to the distinct chromatin environment seen at this region , with CTCF potentially contributing to the definition of this region ( Fig 5C , bottom-right ) . This suggests a mechanism similar to that observed in yeast where antisense transcription may contribute to a chromatin environment that ultimately impacts gene expression [26] .
Our analyses indicate that downstream antisense transcription proximal to gene promoters is common in mammals . Its coincidence with a number of different regulatory features suggests that antisense transcription borders the chromatin environment characteristic of promoters and may possess a regulatory role . Previous studies have characterized convergent transcription as a repressive feature of genes [27] . The promoter-proximal convergent transcription described here is a narrow subset of convergent transcription , where downstream antisense transcription initiates at or within 2 kb of a gene TSS . Based on our analysis , we see little evidence for repression of associated genes by downstream antisense transcription . Considered categorically , comparisons between all genes and those displaying observable daTSSs fail to show significant differences in levels of transcription initiation and steady-state expression levels ( Fig 3 ) . Ultimately , the interplay between antisense transcription and gene expression will be complex , as coincidence of daTSSs positions with other promoter-associated features suggests interplay with other regulatory pathways ( S9 Fig ) . Active enhancers are often present and transcribed within intronic regions of gene bodies . Given the position of daTSSs downstream of gene TSSs , there is a question as to whether downstream antisense transcription is predominantly the consequence of canonical enhancer activity within gene introns . The lack of a direct connection to increased expression levels suggests that downstream antisense transcription is not associated with active enhancers regulating the nearby gene . Though a majority of daTSSs are positioned within introns ( 74% ) , a similar proportion remain overlapped with introns when gene models are randomly shuffled ( 75% ) , indicating that daTSSs are not significantly enriched within introns ( S3 Table ) . There is also a conspicuous lack of signal attributable to enhancer-associated H3K4me1 at observed daTSSs , though there is an apparent association of p300 and H3K27ac marks ( Figs 4D and 5C ) . Rather than describing a functionally distinct element , e . g . proximal enhancers , downstream antisense transcription seems to be a feature of promoters themselves . Along with antisense transcription upstream of gene TSSs , downstream antisense transcription may be an intrinsic feature at many mammalian promoters ( Fig 6 ) . There appears to be a connection between antisense transcription and promoter-specific features at genetic and epigenetic levels ( for an overview of features at each class of TSS , see S9 Fig ) . daTSSs respect the positioning of promoter-proximal nucleosomes , with observed daTSSs falling within valleys of MNase-seq read density ( Fig 4B ) . Downstream antisense transcription may affect nucleosome organization at promoters; genes with larger distances between gene TSSs and daTSSs display less prominent nucleosome-associated peaks in MNase-seq data ( Fig 4C ) . daTSSs also coincide with the binding of transregulatory factors . In particular , regions between gene TSSs and daTSSs show association of chromatin remodeling factors ( Fig 5C ) . These factors potentially contribute to the chromatin environment bordered by gene TSSs and daTSSs and distinguished by enrichment of promoter-associated histone marks . It is not clear whether these functions are attributable to generated transcripts or transcription itself , though produced transcripts are not apparently stable ( S3A Fig ) . Recent studies suggest that non-coding RNAs generated near promoters participate in the establishment of nucleosome occupancy [28] . Though both are proximal to promoters , daTSSs and uaTSSs exist in distinct epigenetic environments ( S9 Fig ) . daTSSs and uaTSSs display fundamentally different relationships with nucleosomes in the promoter region . uaTSSs are an apparent boundary of the nucleosome depleted region at promoters [11] while daTSSs initiate from between regularly oriented nucleosomes downstream of gene TSSs ( Fig 4 ) . The different relationships with nucleosomes seem to inform in part the differences observed with other epigenetic features . Nucleosome depletion near uaTSSs allows for transcription factor association between uaTSSs and gene TSSs while the presence of nucleosomes likely prevents transcription factor association between daTSSs and gene TSSs ( Fig 5; S6 Fig ) [11] . Likewise , the presence of nucleosomes gives functional relevance to the association of chromatin remodelers observed between daTSSs and gene TSSs but not between uaTSSs and gene TSSs ( Fig 5; S7 Fig ) . Despite differences in epigenetic features , tendency for association with transregulatory factors , and capacity to produce stable RNA transcripts , all three classes of TSS described in this work display similarities in sequence content , including enrichment for GC content and Pol II-associated sequence motifs ( Fig 2 ) . As such , antisense transcription appears to be encoded in genetic sequence . This connection between sequence content and epigenetic features provides the compelling suggestion that antisense transcription encoded by sequence may direct the positioning of nucleosomes and deposition of histone marks . Antisense transcription may also participate in signal-dependent modulation of epigenetic content where activation of sequence-encoded antisense TSS precedes nearby changes in chromatin structure . In this way , the collection of transcription initiation-associated sequence motifs near promoters may define regulatory potential for a given gene . This connection to sequence also provides a means to interrogate antisense transcription function . Future studies with selective mutation of associated sequence motifs may elucidate the function of antisense transcription and its coincidence with promoter-associated features . Directed mutagenesis could also establish the extent of the effect of antisense transcription on the chromatin environment at promoters . We characterized downstream antisense transcription initiating near gene promoters in human T47D/A1-2 cells . daTSSs fall between regularly positioned nucleosomes downstream of gene TSSs . Histones within this region are enriched for marks closely associated with active promoter regions , such as H3K4me3 and H3K27ac modifications . Chromatin remodeling complexes show enriched binding upstream of observed daTSS positions , suggesting that antisense transcription contributes to the establishment and maintenance of a promoter-specific chromatin environment . Downstream antisense transcription is common to many human promoters , and daTSSs correlate with the downstream edge of promoter-associated chromatin features . Coincidence of daTSSs with these features suggests interplay between antisense transcription and regulatory pathways .
T47D/A1-2 cells were cultured in DMEM containing 10% FBS . Prior to RNA isolation , cells were cultured in medium supplemented with 5% charcoal dextran-treated serum for at least 24 hours . The A1-2 cell line is a previously described derivative of the T47D breast cancer cell line that overexpresses rat glucocorticoid receptor ( GR ) and contains a stably-integrated MMTV luciferase reporter gene [29] . Short capped RNA was isolated from T47D/A1-2 cells as previously described [16] . Libraries were generated using the Illumina TruSeq small RNA kit . Two independent replicates were performed . A primary dataset was generated from combined Illumina HiSeq and MiSeq runs . This data set was used in all TSS calling and downstream analysis . A secondary validation data set with fewer reads was generated on an Illumina MiSeq run . This data set was used to validate reproducibility of read density at called TSSs ( S1A Fig ) . For this and other in-house sequencing experiments , libraries were prepared and sequenced by the NIH Intramural Sequencing Center ( NISC ) . Start-seq reads were first filtered by quality score; reads with an average Sanger score less than 20 were removed from analysis . Following quality filtering , Cutadapt ( version 1 . 2 . 1 ) was used to remove adapter sequences [30] . Alignment of Start-seq reads was performed using Bowtie ( version 0 . 12 . 8 ) to hg19 or mm9 genome assemblies [31] . From each uniquely mapped Start-seq fragment , the 5’ end was taken forward into the TSS calling procedure . Identification of TSSs was performed based on methods described previously [11 , 16] , TSS identification was guided by RefSeq annotation ( retrieved 05/09/2014 ) [32] . From the reference annotation , a list of non-redundant TSSs was taken from all mRNA RefSeq IDs ( “NM_” ) . 2000-nt search windows were created about each RefSeq TSS . If search windows overlapped and had the same common gene name , those search windows were merged . For other overlapping search windows , boundaries were defined as the midpoint between associated TSSs . The intersection of search windows and 5’ Start-seq ends was then determined . TSSs were called within each window in which the strand-specific 5’ end read count at any given nucleotide position met or exceeded a threshold of 5 reads . This threshold was determined in a previously described method [33] . In short , the FDR was estimated based upon the distribution of Start-seq reads across the genome and a background model where the probability of finding a given number of aligned reads by chance is given by a sum of Poisson probabilities . The read threshold was selected to allow less than 1 expected false positive by this measure . In those windows where a single nucleotide position met or exceeded the read threshold , a gene TSS was called . The calling method aims to select as the TSS the position with the highest read counts in window region with the highest read density . To accomplish this , two potential TSSs were first determined . The first was the position with the most aligned 5’ ends across the entire window . For the second , the search window was divided into 200-nt bins at every 10 nt across the search window . At the 200-nt bin with the most overlapped 5’ ends , the second potential TSS was called as the position with the most aligned 5’ ends . Of the two putative sites , the TSS closest to the associated annotated RefSeq TSS was selected . Following gene TSS identification , uaTSSs and daTSSs were found . For uaTSSs and daTSSs , search spaces were defined in antisense orientation as 1 to 1000 nt upstream and 1 to 2000 nt downstream , respectively . uaTSSs and daTSSs were called at the position with the most aligned 5’ ends within the search window if the count at any single nucleotide position met or exceeded a threshold of 5 reads . To ensure that identified daTSSs were not simply mis-called gene TSSs or uaTSSs , additional criteria were used to filter daTSS calls . A daTSS call was filtered if ( 1 ) within 1000 nt upstream of the daTSS there was a genomic position with Start-seq reads greater than or equal to 10% of reads at the associated gene TSS on the same strand as the gene TSS , implicating the daTSS as a potential uaTSS for an uncalled gene TSS or ( 2 ) it was within 1000 nt of an annotated TSS on the same strand , implicating the daTSS as a potential gene TSS . Prior to alignment , Pol II ChIP-seq reads from MCF-7 cells [23] ( see also Supplemental Table 1 ) were filtered based on quality; reads with an average Sanger quality score less than 20 were removed from analysis . Following quality filtering , Cutadapt was used to remove adapter sequences [30] . Alignment was performed using Bowtie [31] . Fragment lengths were estimated using Homer ( version 4 . 6 ) [34] . From each uniquely mapped fragment , the fragment was extended based on the estimated length , and the fragment center was subsequently found . Considering each TSS group separately , the nucleotide composition of each position in a -1000 to +999 window was determined and reported as a percentage . Logo plots were generated using Web Logo 3 considering sequences in a -5 to +5 window about identified TSSs [35] . To identify occurrences of Pol II-associated and known vertebrate motifs , FIMO ( version 4 . 10 . 0 ) was used considering a p < 0 . 0001 significance cutoff and a 0-order Hidden Markov Model from promoter regions as background [36] . Publically available position weight matrices from JASPAR were used in motif identification [18] . The JASPAR POLII database ( 2008 version; 13 motifs ) and vertebrate motifs in the JASPAR CORE database ( 2014 version; 205 motifs ) were used for Pol II-associated and known vertebrate motif identification , respectively . Across all promoter regions , on the order of 105 Pol II-associated and 107 known vertebrate motifs were identified . Motifs and additional information are available at http://jaspar . genereg . net/ . De novo motif discovery was performed using MEME ( version 4 . 10 . 0 ) with default parameters [37] . Sequence windows from -35 to -20 and from -5 to +5 relative to TSS positions were used . Sequences from each TSS class were combined prior to motif analysis . Logo plots were generated using Web Logo 3 after aligning identified motifs [35] . CpG island heatmaps reflect the intersection of annotated CpG islands retrieved from UCSC Genome Browser [17] with TSS-centered windows . Sequence conservation heatmaps were generated using phyloP scores from placental mammal alignments retrieved from UCSC Genome Browser [19] . Each position in the heatmap represents the average score over all positions in a 40-bp bin for which phyloP scores were available . GRO-cap data from K562 and GM12878 cells [7] were compared to daTSSs identified in T47D/A1-2 cells . If a given site did not have GRO-cap signal in either K562 or GM12878 cells , that daTSS was considered T47D/A1-2 specific . The list of genes with T47D/A1-2-specific daTSSs was applied to Ingenuity Pathway Analysis [38] considering experimentally observed associations over mammalian tissues and cell lines . Nuclei were harvested from cultured T47D/A1-2 cells and digested for 5 minutes at 37°C with a range of MNase ( Worthington ) concentrations . Reactions were stopped by the addition of EDTA and then treated with RNase and proteinase K . Digested DNA was isolated by phenol/chloroform extraction and ethanol precipitation . Libraries were prepared using an Illumina TruSeq sample preparation kit and sequenced on an Illumina HiSeq for paired-end 50 base reads . Publically available MNase-seq data [22] and data generated in this work were prepared in the same way . Prior to alignment , MNase-seq reads were filtered based on quality; reads with an average Sanger quality score less than 20 were removed from analysis . Following quality filtering , Cutadapt was used to remove adapter sequences [30] . Alignment of MNase-seq read pairs was performed using Bowtie [31] . From each uniquely mapped fragment , the fragment center was found . Any report of MNase-seq coverage only considers the fragment-center position . FAIRE-seq data were collected as described previously [24] . Publically available FAIRE-seq data [24] and data generated in this work were prepared in the same way . Prior to alignment , FAIRE-seq reads were filtered based on quality; reads with an average Sanger quality score less than 20 were removed from analysis . Following quality filtering , Cutadapt was used to remove adapter sequences [30] . Alignment of FAIRE-seq reads was performed using Bowtie [31] . Aligned FAIRE-seq reads were then de-duplicated using Picard ( version 1 . 118 ) [39] . FAIRE-seq fragment lengths were estimated using Homer [34] . For each uniquely mapped fragment , the fragment was extended based on the estimated length , and the estimated fragment center subsequently found . Reported FAIRE-seq coverage only considers the fragment-center position . Over biological triplicates , total RNA was harvested using an RNeasy Kit ( Qiagen ) with on-column DNase treatment . RNA quality was validated by Bioanalyzer ( Agilent ) . Paired-end strand-specific poly-A enriched libraries were sequenced on an Illumina HiSeq 2500 for 125 base paired-end reads . Prior to alignment , RNA-seq reads were filtered based on quality; reads with an average Sanger quality score less than 20 were removed from analysis . Following quality filtering , Cutadapt was used to remove adapter sequences [30] . Insert lengths were estimated by transcriptome alignment using Bowtie [31] . Sequence alignment was then performed using TopHat ( version 2 . 0 . 4 ) [40] . Following de-duplication by Picard [39] , alignments from individual replicates were merged . FPKM values were calculated using Cufflinks ( version 2 . 2 . 1 ) [41] . For publically available data sets ( with the exceptions of MCF7 Pol II ChIP-seq data and of FAIRE-seq and MNase-seq validation data sets ) , read coverage files were retrieved from public depositories ( S1 Table ) . Given that reported read densities were considered , the data processing of the original authors was effectively observed . deepTools was used to generate matrices describing the intersection of read coverage with TSS-centered genomic windows observing strand specificity when appropriate [42] . These matrices were then used to generate heatmaps . Heatmaps consider 40-bp/40-nt bins over TSS-centered windows . Unless otherwise noted , each position in a heatmap gives the number of reads or other features overlapping with that bin . Heatmap images were generated using Partek ( version 6 . 6 ) [43] . Two-dimensional plots , unless otherwise noted , consider 10-bp/10-nt bins and report average values across all TSSs considered . To test enrichment of ChIP-seq signal at daTSS positions , ChIP-seq coverage was found in a 100-bp window about all identified daTSSs . Equivalent regions were found at genes without daTSSs by selecting a 100-bp window shifted downstream of TSSs by the median observed TSS-daTSS distance ( 507 nts ) . The significance of enrichment was then calculated by Wilcoxon test comparing the two groups . To generate the gene TSS-centered panels in S5 , S6 , S7 and S8 Figs , uaTSS- and daTSS-centered plots were first reflected across uaTSS and daTSS positions , respectively , to orient these plots relative to gene TSSs . These plots were then translated upstream or downstream by the median distances observed between gene TSSs and uaTSSs or between gene TSSs and daTSSs across calls made in T47D/A1-2 cells . The data sets supporting the results of this article are available in the GEO repository , GSE74308 .
|
Gene transcription is regulated by the coordinated interaction of genetic , epigenetic and trans-acting factors . The chromatin environment at gene promoters , including positioned nucleosomes that may display functional histone modifications , is a key regulator of gene expression , contributing to transcriptional activation and repression . In addition to sense-strand transcription of gene sequences , antisense transcription is prevalent at gene promoters . Often resulting in a short-lived non-coding RNA transcript , the function of antisense transcription is poorly understood . Using next-generation sequencing techniques , we characterized transcription in human breast cancer cells and found extensive correlations between antisense transcription and the chromatin environment at promoters . We found that downstream antisense transcription initiates from between regularly positioned nucleosomes and that those nucleosomes between sense and downstream antisense transcription start sites display histone modifications associated with active gene promoters . Chromatin remodelers and other protein complexes responsible for creation and maintenance of the promoter chromatin environment associate with this same region , suggesting an important role of antisense transcription in the regulation of gene expression .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"sequencing",
"techniques",
"gene",
"regulation",
"regulatory",
"proteins",
"dna-binding",
"proteins",
"dna",
"transcription",
"transcription",
"factors",
"sequence",
"motif",
"analysis",
"molecular",
"biology",
"techniques",
"epigenetics",
"chromatin",
"research",
"and",
"analysis",
"methods",
"sequence",
"analysis",
"sequence",
"alignment",
"chromosome",
"biology",
"proteins",
"gene",
"expression",
"molecular",
"biology",
"nucleosomes",
"biochemistry",
"cell",
"biology",
"genetics",
"biology",
"and",
"life",
"sciences"
] |
2016
|
Downstream Antisense Transcription Predicts Genomic Features That Define the Specific Chromatin Environment at Mammalian Promoters
|
Neutrophil abscess formation is critical in innate immunity against many pathogens . Here , the mechanism of neutrophil abscess formation was investigated using a mouse model of Staphylococcus aureus cutaneous infection . Gene expression analysis and in vivo multispectral noninvasive imaging during the S . aureus infection revealed a strong functional and temporal association between neutrophil recruitment and IL-1β/IL-1R activation . Unexpectedly , neutrophils but not monocytes/macrophages or other MHCII-expressing antigen presenting cells were the predominant source of IL-1β at the site of infection . Furthermore , neutrophil-derived IL-1β was essential for host defense since adoptive transfer of IL-1β-expressing neutrophils was sufficient to restore the impaired neutrophil abscess formation in S . aureus-infected IL-1β-deficient mice . S . aureus-induced IL-1β production by neutrophils required TLR2 , NOD2 , FPR1 and the ASC/NLRP3 inflammasome in an α-toxin-dependent mechanism . Taken together , IL-1β and neutrophil abscess formation during an infection are functionally , temporally and spatially linked as a consequence of direct IL-1β production by neutrophils .
Neutrophil abscess formation represents an important component of the innate immune response , which helps control the spread of an invading pathogen into deeper tissues and systemically [1] . At the site of infection , neutrophils primarily function through the phagocytosis of microorganisms and utilize a variety of antimicrobial mechanisms to mediate pathogen killing [2] . To investigate mechanisms that promote neutrophil recruitment and abscess formation , we chose to use S . aureus cutaneous infection as a model [3] . This gram-positive extracellular bacterium is responsible for the vast majority of skin and soft tissue infections in humans and is a common cause of invasive and often life-threatening infections such as bacteremia , abscesses of various organs , septic arthritis , osteomyelitis , endocarditis , pneumonia and sepsis [4] , [5] . S . aureus infection serves as an excellent model system to study neutrophil recruitment since neutrophil abscess formation is required for bacterial clearance in a variety of mouse models of S . aureus infection , including cutaneous infection , bacteremia , septic arthritis and brain abscesses [6]–[8] . The critical role of neutrophils in host defense against S . aureus is also seen in humans , since patients with genetic or acquired conditions with defective neutrophil number or function suffer from recurrent and invasive S . aureus infections in various tissues and organs , including the skin [9] . It is well established that IL-1β plays a central role in initiating the neutrophilic response against S . aureus infections [7] , [10] , [11] . This is mediated by IL-1β activation of IL-1R/MyD88 signaling , which triggers NF-κB and other signaling molecules that induce proinflammatory mediators and chemokines to promote neutrophil trafficking from the circulation into the infected tissue [9] . Given this essential function of IL-1β , there has been intense interest in understanding how its production is triggered during an infection in vivo . On a cellular level , two different signals are required for IL-1β production in response to S . aureus . The first is production of pro-IL-1β , which is in part mediated by activation of pattern recognition receptors ( PRRs ) such as TLR2 , a cell surface PRR that recognizes S . aureus lipopeptides and lipoteichoic acid [12] , [13] , and NOD2 , a cytoplasmic PRR that recognizes muramyl dipeptide , which is a breakdown product of S . aureus peptidoglycan [14] , [15] . The second signal is the triggering of the NLRP3 inflammasome to induce caspase-1 activation and subsequent cleavage of pro-IL-1β into mature IL-1β , the active and secreted cytokine [16]–[18] . The mechanism for inducing IL-1β-dependent neutrophil recruitment in infected tissues in vivo is complex and involves interactions among epithelial cells , stromal cells , resident immune cells , endothelial cells and recruited immune cells . It is known that IL-1β produced at the site of S . aureus skin infection promotes neutrophil recruitment by inducing neutrophil-attracting chemokines and granulopoiesis factors directly via activating IL-1R/MyD88-signaling and indirectly through the production of IL-17 by T cells [3] , [11] , [19] . A key question is which cell types are responsible for IL-1β production during a S . aureus infection in vivo and how these cells utilize PRRs and the inflammasome to induce its production . The precise mechanism is particularly relevant since many different cells , including keratinocytes , mast cells , Langerhans cells , dendritic cells and monocytes/macrophages , can produce IL-1β in various in vivo models of skin inflammation and infection [20]–[23] . In the current study , we used gene expression analysis and noninvasive in vivo imaging to determine the functional and temporal kinetics , cellular sources and mechanisms by which IL-1β induces neutrophil abscess formation during a S . aureus skin infection .
Mice deficient in IL-1β , IL-1R or MyD88 , but not IL-1α , exhibit a severe impairment in neutrophil abscess formation at the site of infection in this mouse model of S . aureus intradermal infection [3] , [11] , indicating that IL-1β is the major cytokine that initiates the IL-1R/MyD88-dependent pathway for neutrophil recruitment . Given these results and evidence from humans and mice that neutrophils are essential for clearance of S . aureus infections [9] , gene expression analysis was performed in an attempt to link IL-1β/IL-1R-dependent gene induction with neutrophil recruitment . To accomplish this , we used a model of S . aureus cutaneous infection in mice , which involves intradermal inoculation of S . aureus ( 2×106 CFUs of strain SH1000 ) in the dorsal back skin of mice [3] . Gene expression analysis was first performed on skin biopsy samples from wt and IL-1R-deficient mice at 4 hrs after S . aureus infection and from uninfected skin . This time point was chosen because we previously observed substantially decreased IL-1β protein levels in S . aureus-infected skin of IL-1R-deficient mice compared with wt mice at 6 hrs after infection [3] and the difference in mRNA levels of IL-1β likely preceded the changes in IL-1β protein levels . By using the criteria that upregulated genes were >1 . 5-fold higher ( p-value<0 . 05 ) than baseline , there were 1 , 288 genes upregulated in wt mice and 606 genes upregulated in IL-1R-deficient mice ( Fig . S1 ) . Comparing S . aureus-infected skin of wt mice versus uninfected skin , the top 4 induced genes were neutrophil-attracting CXC chemokines , including CXCL1 ( KC ) , CXCL2 ( MIP2α ) , CXCL3 ( MIP2β ) and CXCL5 ( LIX ) ( Fig . 1A ) , which bind to CXCR2 on mouse neutrophils to induce chemotaxis [24] . The neutrophil granulopoiesis factors G-CSF and GM-CSF , as well as the proinflammatory cytokines IL-1β , IL-6 and the inflammasome component NLRP3 , were also among the top 20 genes . These data indicate that many of the most highly induced genes in S . aureus-infected wt mice were associated with neutrophil chemotaxis , granulopoiesis and IL-1β production . Pathway analysis ( Ingenuity ) was then used to categorize the genes upregulated in S . aureus-infected wt versus IL-1R-deficient mice into functional groups . In wt mice , the most significantly upregulated functional pathway was the Cellular Movement group , which included 354 genes ( p = 3×10−57 ) ( Fig . 1B ) . In contrast , in IL-1R-deficient mice , the Cellular Movement functional group included only 256 genes and had lower statistical significance ( p = 2×10−25 ) . Categorization of the Cellular Movement functional group by cell type revealed that in wt mice , the Cell Movement of Neutrophils sub-group included 77 genes and was the most statistically significant ( p = 7×10−33 ) . Neutrophils were followed in order of number of genes and significance by macrophages ( 41 genes; p = 2×10−15 ) , T cells ( 41 genes; p = 9×10−15 ) and dendritic cells ( 25 genes; p = 3×10−12 ) . In contrast , in IL-1R-deficient mice , the categorization of the Cell Movement group by cell type was radically different with less numbers of upregulated genes . These differences were most evident for neutrophils , ( 44 genes; p = 1×10−10 ) , followed by macrophages ( 28 genes; p = 4×10−9 ) , T cells ( 29 genes; p = 7×10−9 ) and dendritic cells ( 10 genes; p = 1×10−5 ) . The level of induction of upregulated genes in the Cell Movement of Neutrophils group was then compared between wt and IL-1R-deficient mice using functional network analysis ( Fig . 1D ) . All 77 genes associated with Cell Movement of Neutrophils were significantly upregulated in wt mice . In contrast , over half ( 45 of the 77 genes ) , including CXCL2 , GM-CSF and IL-1β , were not significantly induced in IL-1R-deficient mice . To confirm these results , the expression of a subset of genes of interest in the Cell Movement of Neutrophils sub-group ( Fig . 1D ) were evaluated by quantitative real-time PCR ( Q-PCR ) , which included 2 genes that were similarly-induced ( IL-6 and G-CSF ) and 8 genes that were differentially-induced in wt and IL-1R-deficient mice ( MIP-2β , MIP-2α , IL-1β , GM-CSF , CXCR2 , G-CSF-R , TLR2 and LTB4R ) ( Fig . 1E ) . In agreement with the microarray data ( Fig . 1D ) , we found that the expression of IL-6 and G-CSF between wt and IL-1R deficient mice was similar and the differentially-induced genes had higher expression in wt mice compared with IL-1R-deficient mice . Taken together , these results indicate that genes associated with neutrophil recruitment were upregulated in response to S . aureus skin infection and that the induction of these genes was largely dependent on IL-1β/IL-1R signaling . To differentiate between the specific anti-S . aureus response versus the response to a live infection , the skin of wt and IL-1R-deficient mice was inoculated with either live or heat-killed S . aureus and Q-PCR was performed on skin samples taken at 4 hrs after inoculation and from uninfected mice ( Fig . S2 ) . We found that the top six genes on the microarray were highly expressed in skin samples infected with live S . aureus ( ranging from 242- to 5696-fold ) or heat-killed S . aureus ( ranging from 46- to 1706-fold ) ; however , the level of induction with heat-killed S . aureus was generally a magnitude less ( decreased from 4 . 5-fold to 14 . 1-fold ) than live S . aureus . To determine whether the pattern of gene induction in wt and IL-1R-deficient mice was similar or different in response to live versus heat-killed S . aureus , the same subset of genes of interest in the Cell Movement of Neutrophils sub-group in Fig . 1E was compared ( Fig . S2B , C ) . Live and heat-killed S . aureus had similar expression of IL-6 and G-CSF between wt and IL-1R deficient mice and the differentially-induced genes had higher expression in wt mice compared with IL-1R-deficient mice . However , as with the top 6 induced genes ( Fig . S2A ) , the levels of induction of these genes were lower with heat-killed S . aureus compared with live S . aureus . In summary , the difference in gene expression patterns between wt and IL-1R-deficient mice was consistent between live and heat-killed S . aureus ( albeit the live S . aureus resulted in higher gene expression than heat-killed S . aureus ) , demonstrating that the immune response is more intense in the presence of the live bacterial infection . These results suggest that S . aureus-specific immune responses are significantly impaired in IL-1R-deficient mice . However , we cannot exclude the possibility that the impaired immune responses in the IL-1R-deficient mice were due to a dysregulated immune response in IL-1R-deficient mice in which any stimulus would elicit an aberrant response with a similar pattern of gene expression . Although IL-1β is the major cytokine that induces IL-1R/MyD88-dependent neutrophil recruitment during a S . aureus skin infection [11] , the source and kinetics of IL-1β production during an infection in vivo has remained unknown . Therefore , advanced techniques of in vivo bioluminescence and fluorescence imaging were combined to provide an approximation of the kinetics of IL-1β production and neutrophil recruitment longitudinally over the time course of the S . aureus cutaneous infection . This was accomplished by performing the intradermal inoculation of a bioluminescent S . aureus strain in two fluorescence reporter mouse strains: ( 1 ) pIL1-DsRed transgenic mice , which express the red fluorescent protein DsRed under the control of the mouse IL-1β promoter [21] , and ( 2 ) LysEGFP mice , which possess green fluorescent myeloid cells ( mostly neutrophils ) due to a knock-in of the EGFP gene into the lysozyme M locus [25] . In vivo noninvasive whole animal imaging was then used to track the S . aureus bacterial burden while simultaneously monitoring IL-1β production or neutrophil recruitment in the same anesthetized mice over the 14 day course of infection . Advantages and limitations of using this strategy of in vivo imaging to quantify these endpoints are described in the Discussion . Infection with S . aureus resulted in the development of visible skin lesions , which had a maximum size of 0 . 53±0 . 11 cm2 by day 3 , and healed by day 14 ( Fig . 2A ) . In vivo bioluminescence signals , which closely estimate the bacterial CFUs harvested from the skin lesions during infection [3] , [19] , peaked on day 1 ( up to 1 . 8±0 . 7×106 photons/s ) and slowly decreased to background levels by day 14 ( Fig . 2B ) . IL-1β-DsRed and EGFP-neutrophil fluorescence signals were significantly higher than uninfected control mice at all time points , peaking on days 3 and 1 , respectively ( up to 2 . 7±0 . 5×1010 and 1 . 6±0 . 3×1010 [photons/s]/[µW/cm2] , respectively ) and decreased to background levels by day 14 ( Fig . 2C ) . In summary , IL-1β-DsRed fluorescence and EGFP-neutrophil fluorescence signals had similar temporal kinetics as they both increased rapidly by day 1 and then decreased along with the in vivo bioluminescent signals over the 14 day course of infection . Since many different cell types , including keratinocytes , mast cells , Langerhans cells , dendritic cells and macrophages , have the capacity to produce IL-1β in various in vivo models of skin inflammation and infection [20]–[23] , it is unclear which of these cell types ( or potentially other cell types ) contribute to IL-1β production during a S . aureus skin infection . However , in our previous work using the same S . aureus skin infection model in bone marrow chimeric mice , we found that the source of IL-1β was from bone marrow-derived hematopoietic cells because neutrophil recruitment , host-defense and IL-1β production at the site of infection was restored in IL-1β-deficient mice reconstituted with bone marrow from wt mice but not in wt mice reconstituted with bone marrow from IL-1β-deficient mice [11] . Since in vivo fluorescence imaging demonstrated that IL-1β production was detected shortly after infection , histological evaluation of skin lesions at 4 and 24 hrs after S . aureus skin infection in pIL1-DsRed mice was performed . To identify the number of cells that expressed pro-IL-1β , two-color immunofluorescence labeling and confocal laser microscopy was performed using an antibody against DsRed , which is retained within the cytoplasm of IL-1β-expressing cells [21] in combination with mAbs directed against cell-specific markers . At 4 hrs , the earliest IL-1β-expressing cells were found almost exclusively within the dermis at the site of abscess formation ( Fig . 3A , C ) . To identify these early IL-1β-expressing cells , sections were first co-labeled with anti-DsRed and mAbs directed against CD45 ( pan-leukocyte marker ) or MHCII ( antigen presenting cells ) ( Fig . S3 ) . The vast majority of IL-1β-expressing cells co-localized with CD45 whereas only a few cells co-localized with MHCII , indicating that antigen presenting cells ( e . g . dermal dendritic cells and monocytes/macrophages ) were not the predominant cell type that expressed IL-1β . This was somewhat surprising since monocytes/macrophages produce large amounts of IL-1β in response to S . aureus or S . aureus components in vitro [16]–[18] . Since neutrophils represent the majority of cells recruited at early time points to the site the S . aureus infection in the skin [3] , [11] , [19] , we next evaluated the expression of IL-1β in monocytes/macrophages and neutrophils using mAbs directed against MOMA2 and 7/4 , respectively [26] . We found that only a few IL-1β-expressing cells co-localized with MOMA2 at 4 and 24 hrs after infection ( Fig . 3A and S4A ) . In contrast , the majority of the IL-1β-expressing cells co localized with 7/4 at 4 hrs and especially at 24 hrs after infection ( Fig . 3B and S4C ) . To quantify the degree of co-localization between IL-1β-expressing cells and the cell-specific markers , image analysis was performed using the Manders' coefficient for a value range of 0 ( no pixels co-localize ) to 1 ( all pixels co-localize ) ( Fig . 3B , D ) . The Manders' coefficient between IL-1β-expressing cells and MOMA2+ monocytes/macrophages or MHCII+ antigen presenting cells at 4 hrs was 0 . 27 or 0 . 21 , respectively ( Fig . 3D and Fig . S4 ) , confirming that these cells represented a minority of IL-1β-expressing cells . In contrast , the Manders' coefficient between the IL-1β-expressing cells and 7/4+ neutrophils was 0 . 56 at 4 hrs and 0 . 83 at 24 hrs . Although the co-localization of DsRed with the cellular markers does not provide information about how much IL-1β is made per cell , these data suggest that neutrophils represent the most abundant cell type that expresses IL-1β at the site of infection . Although neutrophils express high levels of 7/4 and monocyte/macrophages express MOMA2 ( which includes recently emigrated monocytes and activated macrophages ) [26] , some subsets of monocytes and macrophages have also been reported to express 7/4 [27] . Therefore , the labeling of 7/4 versus MOMA2 in sections of S . aureus-infected mouse skin was compared ( Fig . S5 ) . At both 4 and 24 hrs after infection , there was only a rare occasional cell that expressed both 7/4 and MOMA2 . Thus , we conclude that the vast majority of the 7/4+ and IL-1β-expressing cells were neutrophils . To evaluate the contribution of neutrophil-derived IL-1β in immunity against S . aureus skin infection , bone marrow-derived neutrophils isolated by Percoll density gradient centrifugation from wt or IL-1β-deficient donor mice were adoptively transferred into IL-1β-deficient recipient mice ( wt PMN→IL-1β−/− mice or IL-1β−/− PMN→IL-1β−/− mice , respectively ) ( Fig . 4 ) . Two hrs after adoptive transfer , these mice along with normal wt and IL-1β-deficient mice , were inoculated intradermally with S . aureus . As expected , IL-1β−/− PMN→IL-1β−/− mice and IL-1β-deficient mice developed larger skin lesions and higher in vivo bioluminescence signals compared with wt mice ( Fig . 4A and B ) . The bioluminescent signals could be seen throughout the areas of the infected lesions , indicating that the total lesion size was a reflection of the degree and extent of the bacterial infection . Furthermore , the defects observed in the IL-1β−/− PMN→IL-1β−/− mice or IL-1β-deficient mice were not likely due to impaired neutrophil function in these mice as in vitro assays for phagocytosis , degranulation , oxidative burst and bacterial killing were not significantly different between neutrophils from IL-1β-deficient mice and wt mice ( Fig . S6 ) . However , wt PMN→IL-1β−/− mice had lesion sizes and in vivo bioluminescence signals that were similar to normal wt mice , indicating that neutrophil-derived IL-1β is sufficient for host defense against the S . aureus skin infection . To further evaluate if the neutrophils and not other cells ( such as the few contaminating monocytes ( Fig . S7A ) in the adoptively transferred cells played a role in promoting neutrophil recruitment and host defense in the in IL-1β-deficient mice , neutrophils or monocytes were specifically depleted from the adoptively transferred cells by positive selection using anti-Ly6G or anti-CD115 MACS bead separation , respectively ( Fig . S7B , C ) . Depletion of neutrophils reduced the absolute number of transferred Ly6G+ neutrophils from 4 . 6×106 to 1 . 8×106 neutrophils/mouse ( 61% depletion efficiency ) and did not decrease the absolute number of transferred CD115+ monocytes ( ∼1 . 5×104 before and after depletion ) . Similarly , depletion of monocytes reduced the absolute number of transferred CD115+ monocytes from 1 . 3×104 to 3 . 0×103 ( 76 . 9% depletion efficiency ) and did not decrease the absolute number of transferred Ly6G+ neutrophils ( ∼4 . 6×106 before and after depletion ) . Although the neutrophil depletion was only 61% complete , the decreased numbers of neutrophils resulted in an inability of adoptively transferred cells from wt mice to rescue the immune impairment in IL-1β-deficient mice . In contrast , monocyte depletion , which decreased the percentage of contaminating monocytes from 0 . 26% to only 0 . 06% of the adoptively transferred cells , had no impact on the ability to rescue the immune impairment in IL-1β-deficient mice . These data provide additional evidence that neutrophils and not monocytes in the adoptively transferred cells played a major role in promoting effective neutrophil recruitment and host defense against the cutaneous S . aureus infection . Histopathological examination of skin biopsies taken one day after infection with S . aureus demonstrated that wt PMN→IL-1β−/− mice and normal wt mice developed large neutrophilic abscesses seen in H&E stained and anti-7/4 labeled sections ( Fig . 4C ) . In contrast , infected skin samples from IL-1β−/− PMN→IL-1β−/− mice and IL-1β-deficient mice had markedly decreased neutrophil recruitment with minimal abscess formation and decreased myeloperoxidase ( MPO ) activity ( which correlates with the degree of neutrophil infiltration ) ( Fig . 4D ) . The levels of IL-1β protein expression from infected skin samples at 4 and 24 hrs were evaluated by ELISA and the amount of IL-1β protein at the site of infection in mice adoptively transferred with wt or IL-1β−/− PMN was below the level of detection ( data not shown ) . In contrast , in wt mice at 4 and 24 hrs the levels of IL-1β protein typically exceeded 20 pg/mg tissue weight ( Fig . 5C and data not shown ) . However , immunohistochemistry with an anti-IL-1β mAb identified scattered IL-1β-expressing cells within the abscess at 24 hrs in wt PMN→IL-1β−/− mice but not in IL-1β−/− PMN→IL-1β−/− mice ( Fig . S8 ) . These data indicate that adoptively transferred wt neutrophils produced IL-1β at the site of infection and that neutrophil-derived IL-1β is sufficient for promoting effective neutrophil abscess formation and host defense against a cutaneous S . aureus infection . Certain PRRs have been shown to recognize S . aureus components and initiate innate immune responses , including TLR2 , a membrane PRR that recognizes S . aureus lipopeptides and lipoteichoic acid [12] , [13] , NOD2 , a cytosolic PRR that recognizes muramyl-dipeptide ( a breakdown product of S . aureus peptidoglycan ) [14] , [15] , and FPRs , which recognize formylated peptides of bacteria [28] . To investigate whether these PRRs contributed to IL-1β production and neutrophil recruitment during a S . aureus skin infection in vivo , we inoculated wt mice and mice deficient in TLR2 , NOD2 or FPR1 with S . aureus ( Fig . 5 ) . TLR2- , NOD2- , and FPR1-deficient mice all developed larger lesions ( up to 4 . 0- , 2 . 8- and 2 . 9-fold , respectively ) ( Fig . 5A ) and higher bioluminescent signals ( up to 5 . 6- , 4 . 9- and 4 . 6-fold , respectively ) than wt mice ( Fig . 5B ) . Taken together , these results demonstrate that TLR2 , NOD2 and FPR1 all significantly contributed to host defense against S . aureus infection in the skin . Furthermore , at 6 hrs after inoculation , S . aureus-infected skin lesions of mice deficient in TLR2 , NOD2 or FPR1 had significant reductions in IL-1β protein ( 80 , 62 , and 91 percent decrease , respectively ) and MPO activity ( 53 , 58 , and 47 percent decrease , respectively ) compared with wt mice ( Fig . 5C , D ) . Therefore , in addition to having higher in vivo bacterial burden , mice deficient in TLR2 , NOD2 and FPR1 also had decreased IL-1β production and neutrophil recruitment during a S . aureus skin infection in vivo . The increased lesion sizes , higher bacterial burden and impaired IL-1β production in TLR2- and NOD2-deficient mice in response to S . aureus skin infection is consistent with previously published studies from our laboratory and others [3] , [15] . To determine whether IL-1β production by neutrophils occurred through direct or indirect mechanisms , neutrophils obtained from bone marrow of wt mice and mice deficient in TLR2 , NOD2 or FPR1 ( purity>99% ) were infected with live S . aureus in vitro ( Fig . 6 ) . This in vitro infection involved incubating the neutrophils with live S . aureus or a community-acquired MRSA strain ( USA300 LAC isolate ) ( at a multiplicity of infection [MOI] of bacteria to neutrophils of 5∶1 ) for a total of 6 hrs and gentamicin was added at 60 min from the start of the infection as previously described [16] . The levels of IL-1β protein produced in these cultures were measured using an ELISA that detects both pro-IL-1β and cleaved IL-1β . During this in vitro infection , we observed increased production of IL-1β protein as the MOI increased from 1∶1 to 5∶1 ( Fig . S9A , B ) . The increased production of IL-1β was not due to increased cell death as there was no decrease in the viability of the neutrophils in the presence of S . aureus or MRSA compared with cultures without any bacterial infection ( Fig . S9C , D ) . Furthermore , the lack of any decrease in viability of the neutrophils in the presence of S . aureus or MRSA suggest that the ELISA likely detected mostly cleaved IL-1β rather than pro-IL-1β released into the supernatants from dying cells . Using this in vitro infection , neutrophils from mice deficient in TLR2 , NOD2 or FPR1 produced significantly less IL-1β protein ( 40 , 43 and 37 percent decrease , respectively ) in response to S . aureus ( Fig . 6A ) , suggesting that activation of TLR2 , NOD2 and FPR1 promoted neutrophil production of IL-1β . To provide further evidence that neutrophils and not other contaminating cells such as monocytes produced IL-1β in these cultures , purified neutrophils from pIL1-DsRed reporter mice were evaluated in this in vitro infection and 43% of the Ly6G+ neutrophils expressed IL-1β-DsRed whereas only 0 . 2% of other cell types ( Ly6G− cells ) expressed IL-1β-DsRed ( Fig . S10 ) . Thus , neutrophils , and not other contaminating cells , were the predominant source of IL-1β in these cultures . An important step in the production of active IL-1β is the enzymatic processing of pro-IL-1β into its active form . Typically , this cleavage is mediated by caspase-1 , which is activated by an intracellular complex of proteins called the inflammasome [29] . However , under certain conditions , cleavage of pro-IL-1β into active IL-1β in neutrophils can be mediated by serine-proteases ( such as proteinase 3 ) or neutrophil elastase rather than inflammasome/caspase-1 activation [30]–[32] . To determine whether IL-1β production by neutrophils in response to S . aureus involved inflammasome activation , we studied neutrophils from mice deficient in ASC , which is required for NLRP3 inflammasome assembly [33] . Neutrophils from wt or ASC-deficient mice were infected with live S . aureus in vitro ( Fig . 6B ) . Neutrophils from ASC-deficient mice had a 58% decrease in IL-1β production in response to S . aureus compared with neutrophils from wt mice , indicating that the majority of IL-1β produced during the S . aureus in vitro infection was dependent on the inflammasome component ASC . Previous studies in mouse and human monocyte/macrophage cultures demonstrated that processing of pro-IL-1β after exposure to S . aureus was dependent upon ASC/NLRP3 inflammasome and caspase-1 activation , which was induced by S . aureus α-toxin and other pore-forming hemolysins [17] , [34] . Therefore , to evaluate whether a similar mechanism of inflammasome activation and IL-1β production was involved in mouse neutrophils , mouse neutrophils were infected with S . aureus or MRSA in vitro in the presence an inhibitor of the NLRP3 inflammasome ( glibenclamide ) [35] , [36] , a specific caspase-1 inhibitor ( Z-YVAD-FMK ) or neutralizing antibodies directed against S . aureus α-toxin [37] ( Fig . 6C , D ) . In mouse neutrophils infected with S . aureus , the amount of IL-1β produced was decreased 62% , 73% and 53% by the NLRP3 inflammasome inhibitor , the caspase-1 inhibitor and the α-toxin neutralizing antibodies , respectively ( Fig . 6C ) . Similarly , in mouse neutrophils infected with MRSA , the amount of IL-1β produced was decreased 44% , 57% and 58% by the NLRP3 inflammasome inhibitor , the caspase-1 inhibitor and the α-toxin neutralizing antibodies , respectively ( Fig . 6D ) . Importantly , addition of inhibitors or neutralizing antibodies did not decrease the viability of neutrophils infected with S . aureus or MRSA ( Fig . S11 ) . To confirm that S . aureus or MRSA infection of mouse neutrophils resulted in the generation of cleaved IL-1β , immunoblotting was performed and cleaved IL-1β was only detected in cultures of S . aureus-or MRSA-infected neutrophils and not in uninfected neutrophil cultures ( Fig . S12 ) . In these mouse neutrophil cultures , there was no decrease in IL-1β produced in control wells containing DMSO ( the vehicle for the NLRP3 inflammasome inhibitor and the caspase-1 inhibitor ) or rabbit IgG ( the control for the α-toxin neutralizing antibodies ) compared with media alone ( data not shown ) . Taken together , these data indicate that the majority of the IL-1β produced was dependent upon activation of caspase-1 via induction of the NLRP3/ASC inflammasome in an α-toxin-dependent mechanism . Furthermore , the dependence of IL-1β production on the NLRP3/ASC inflammasome and caspase-1 provides additional evidence that the IL-1β detected by the ELISA was mostly cleaved IL-1β rather than pro-IL-1β . It should be noted that it was necessary to use methods that yielded purified cultures of mouse neutrophils for the in vitro infection experiments to minimize monocyte contamination . Mouse neutrophils were positively selected from bone marrow cells using anti-Ly6G magnetic bead separation . This method resulted in 99 . 1% purity of mouse neutrophils ( Fig . S13 ) . Although the positive selection with anti-Ly6G magnetic bead separation may have induced some activation of the mouse neutrophils , this degree of activation was unlikely to play a major role in production of IL-1β because there was minimal IL-1β production observed in cultures of uninfected neutrophils ( Fig . 6A–D ) .
Neutrophil abscess formation is an essential component of innate immunity against many pathogens [1] . In this study , using gene expression analysis and advanced techniques of in vivo fluorescence imaging , we found that neutrophil recruitment during a S . aureus cutaneous infection is functionally and temporally linked to IL-1β/IL-1R activation . Based on our prior work [3] , [11] , we hypothesized that this association was a result of IL-1β production by hematopoietic cells such as macrophages or dendritic cells and possibly other cells that reside in the skin , such as keratinocytes or mast cells [20]–[23] . Surprisingly , we found that neutrophils were the most abundant source of IL-1β during infection . Neutrophil-derived IL-1β , in the absence of other cellular sources of IL-1β , was critical for host defense since adoptive transfer of IL-1β-expressing neutrophils was sufficient to restore the impaired neutrophil recruitment and abscess formation in S . aureus-infected IL-1β-deficient mice . In addition , mouse neutrophils produced IL-1β in vitro in response to live S . aureus in a mechanism involving the PRRs , TLR2 , NOD2 and FPR1 , and the ASC/NLRP3 inflammasome . Thus , neutrophil recruitment and IL-1β/IL-1R are functionally , temporally and spatially linked because neutrophils are the predominant source of IL-1β . These findings provide a new paradigm for abscess formation during an infection in which the inflammatory mediators produced by the epithelial , stromal and resident immune cells in the infected tissue may contribute to the recruitment of the very first neutrophils; however , this response is not sufficient for effective abscess formation . Rather , a feed-forward mechanism that involves early recruited neutrophils serving as a source of IL-1β is essential for amplifying and sustaining the neutrophilic response to promote optimal abscess formation and bacterial clearance . Although a link between neutrophil recruitment and IL-1β during S . aureus infections was previously documented [7] , [10] , [11] , our work defines the primary mechanism by which effective neutrophil abscess formation occurs . These findings provide an explanation for a number of puzzling observations in humans and mice and have important implications for neutrophil-derived IL-1β in potentially contributing to other immune responses during infection and inflammation . First , human pediatric patients with deficiency in TLR/IL-1R signaling molecules , MyD88 or IRAK-4 , are predisposed to pyogenic bacterial infections , including S . pneumoniae , S . aureus , and P . aeruginosa , whereas other types of bacterial , fungal and viral infections are exceedingly rare [38] , [39] . The reason for this has remained elusive , especially since these patients do not have impaired neutrophil number or function as seen in other conditions predisposed to pyogenic infections such as severe congenital neutropenia or chronic granulomatous disease [9] . Interestingly , during acute or invasive infections , patients with MyD88 or IRAK-4 deficiency develop neutropenia despite having pus in infected tissues [40] . Our findings suggest a pathway beginning with S . aureus-induced inflammation in the skin tissue that results in an initial early recruitment of neutrophils that produce IL-1β . The neutrophil-derived IL-1β is sufficient to amplify and sustain their recruitment that promotes neutrophilia and effective neutrophil abscess formation . Although patients with MyD88 or IRAK-4 deficiency may recruit neutrophils to the site of infection , they cannot respond to neutrophil-derived IL-1β to amplify the neutrophilic response , providing a potential explanation for their selective predisposition to pyogenic infections . Second , previously published work has demonstrated that neutrophils express pattern recognition receptors , including TLR2 [41] , [42] , NOD2 [43] and FPRs [28] . Thus , we evaluated TLR2- , NOD2- , or FPR1-deficient mice in response to S . aureus skin infection and found that each of these mice had impaired IL-1β production ( Fig . 5C ) and neutrophil recruitment after S . aureus skin infection ( Fig . 5D ) . Based on our in vitro infection experiments with neutrophils from TLR2- , NOD2- , or FPR1-deficient mice , there was decreased IL-1β production compared with neutrophils from wt mice ( Fig . 6A ) , suggesting these each of these PRRs on neutrophils directly contribute to the production of IL-1β in response to S . aureus . Since TLR2 , NOD2 and FPR1 are activated by different bacterial components , they likely provide overlapping and redundant functions to ensure adequate neutrophil IL-1β production and a deficiency in any one of these PRRs would not have a major impact on host defense . However , TLR2 , NOD2 and FPR1 have also been shown to also be involved in other neutrophil functions such as chemotaxis , phagocytosis and oxidative burst [14] , [44]–[47] . Thus , the impaired immune response against S . aureus in TLR2- , NOD2- , or FPR1-deficient mice in vivo may not be solely due to decreased IL-1β production but is likely dependent upon the lack of other functional activities of these PRRs on neutrophils as well as on other cell types in the infected skin in these knockout mice . To determine if TLR2 functioned predominantly on neutrophils or other cell types in vivo , we performed an additional experiment in which wt neutrophils were adoptively transferred into TLR2-deficient mice ( Fig . S14 ) . This adoptive transfer of wt neutrophils did not rescue the immune impairment in TLR2-deficient mice as observed with the adoptive transfer of wt neutrophils into IL-1β-deficient mice ( Fig . 4 ) . Thus , TLR2 activation is needed on other cells to invoke protective mechanisms in vivo . Third , our previous work found that production of IL-1β was likely from bone marrow-derived hematopoietic cells because neutrophil recruitment , host-defense and IL-1β production at the site of infection was restored in IL-1β-deficient mice reconstituted with bone marrow from wt mice but not IL-1β-deficient mice [11] . Here , we found that adoptively transferred wt neutrophils could rescue the immune impairments in IL-1β-deficient mice , indicating neutrophils are the predominant hematopoietic cellular source of IL-1β that was sufficient for effective neutrophil recruitment and abscess formation . These data further argue against an important role for IL-1β produced by non-hematopoietic cells during the S . aureus skin infection . Although keratinocytes produced IL-1β during the infection as detected by immunohistochemistry ( Fig . S15 ) , this amount of IL-1β was not able to promote effective neutrophil recruitment in the absence of IL-1β-expressing hematopoietic cells because wt mice reconstituted with bone marrow from IL-1β-deficient mice show the same impaired neutrophil recruitment response as normal non-irradiated/non-reconstituted IL-1β-deficient mice [11] . Fourth , IL-1 has been shown to play a role in neutrophil recruitment during sterile inflammation . In a mouse model of intraperitoneal injection of necrotic lymphoma cells or acetaminophen-induced liver injury , neutrophil recruitment to the peritoneal cavity or liver was mediated by IL-1α alone or both IL-1α and IL-1β , respectively [48] . Interestingly , in a mouse model of autoimmune inflammatory arthritis , neutrophil recruitment to the inflamed joints was dependent on leukotriene B4 ( LTB4 ) [49] . However , exogenous IL-1β injected into the joints or adoptive transfer of wt neutrophils could restore neutrophil recruitment and arthritis in LTB4-deficient mice , demonstrating that IL-1β-producing neutrophils amplified neutrophil recruitment and arthritis [49] , which was similar to what we observed during a S . aureus skin infection . Thus , although IL-1β-producing neutrophils are sufficient for neutrophil recruitment during a S . aureus skin infection , neutrophil recruitment during sterile inflammation is mediated by IL-1α , IL-1β or both IL-1α and IL-1β , depending on the anatomical site and the type of inflammation . Neutrophil-derived IL-1β may also promote other immune responses at the site of infection . Similar to our findings , a previous report found that NOD2-deficient mice had impaired production of IL-1β during a S . aureus skin infection [15] . This report also found that NOD2-induced IL-1β contributed to production of IL-6 , which enhanced neutrophil killing of S . aureus [15] . In our previous work , we found that IL-1R-mediated neutrophil recruitment ( through production of the neutrophil-attracting chemokines KC and MIP2 ) was dependent upon IL-1R-signaling by resident skin cells rather than bone marrow-derived recruited cells [3] . These findings were based upon data using bone marrow chimeric mice in which the impaired host defense and neutrophil recruitment in IL-1R-deficient mice could not be restored in IL-1R-deficient mice reconstituted with bone marrow from wt mice [3] . In contrast , wt mice reconstituted with bone marrow from IL-1R-deficient had no immune impairment . Additionally , as mentioned above , IL-1β production during the S . aureus skin infection was found to be dependent on bone marrow-derived cells rather than resident skin cells [11] . Combining these previous studies with the present findings , a host defense pathway has been discovered whereby neutrophils represent a source of IL-1β , which subsequently activates IL-1R expressed on non-bone marrow-derived resident skin cells to promote effective neutrophil recruitment in host defense during a S . aureus skin infection . Furthermore , we had previously demonstrated that IL-1R activation was required for inducing IL-17A/F production by γδ T cells in infected mouse skin at early time points after S . aureus infection [19] . In this context , IL-17A/F promoted enhanced neutrophil recruitment via induction of neutrophil-attracting chemokines and granulopoiesis factors . Since IL-1β has also been shown to be important in the generation of Th17 cells [50] , [51] , future studies will be required to determine if neutrophil-derived IL-1β contributes to the IL-6 responses as well as the development of Th17 cells and other IL-17-producing cells following a cutaneous S . aureus infection . It should also be noted that since IL-1β- and IL-1R-deficient mice ultimately clear these infections , compensatory mechanisms exist that eventually promote bacterial clearance . Similar compensatory mechanisms also may play a role in TLR2- , NOD2- and FPR1-deficient mice as these mice also eventually clear the infection and the cellular composition of 7/4+ neutrophils and MOMA2+ monocytes/macrophages on day 10 after infection in TLR2- , NOD2- , FPR1-deficient mice was similar to the cellular composition in wt mice whereas IL-1β-deficient mice had a paucity of 7/4+ neutrophils at this time point ( Fig . S16 ) . These compensatory responses may include activation of other MyD88-dependent receptors such as TLRs , IL-18 or IL-33 because we found that MyD88-deficient mice have a more severe impairment in neutrophil recruitment than IL-1β- or IL-1R-deficient mice [3] , [11] . In addition , IL-17 has also been shown to be critical in promoting neutrophil recruitment and antimicrobial responses against S . aureus in various mouse models of infection ( cutaneous infection , systemic infections , pneumonia and brain abscesses [52]–[56] ) as well as in humans with hyper-IgE syndrome or with a deficiency in IL-17F or IL-17RA [50] , [57]–[60] . Although Th17 development is severely impaired in IL-1R-deficient mice in vivo [61] , [62] and γδ T cell production of IL-17 is enhanced in the presence of IL-1β [19] , [63] , the numbers and activity of Th17 and γδ T cells may increase during the course of the S . aureus skin infection and compensate for absence of IL-1β activity . Consistent with this possibility , humans with deficiency in the IL-1R downstream signaling molecules MyD88 or IRAK-4 do not exhibit impaired development of IL-17-producing cells [64] . The use of in vivo imaging in this study to quantify the bacterial burden , neutrophil recruitment and IL-1β production provides an approximation of these endpoints and there are advantages and limitations that should be considered when interpreting this data . First , in vivo bioluminescence imaging provides only a close estimate of in vivo bacterial burden as several factors such as body temperature , metabolic activity of the bacteria in vivo [65] , [66] and the presence of reactive oxygen mediators produced by neutrophils at the site of infection that could potentially react with the bacterial luciferase as seen with GFP-labeled bacteria [67] , [68] . However , despite these potential confounding factors , we previously demonstrated that in vivo bioluminescent signals directly correlate with ex vivo CFUs harvested at different time points from the S . aureus-infected skin lesions [3] , [19] , [69] . Thus , the bioluminescent signals and actual bacterial burden is not a perfect correlation; however , it is a noninvasive method that approximates the bacterial burden in vivo that does not require euthanasia of numerous animals at every time point to obtain this information . Second , regarding the use of the LysEGFP mice , lysozyme M is expressed in myeloid cells ( including neutrophils and monocytes/macrophages ) and the lysozyme M promoter driven EGFP expression is not specific for neutrophils [25] . However , neutrophils from LysEGFP mice have been shown to have much brighter EGFP fluorescence intensity than monocytes or macrophages [70] , [71] and we found that F4/80+ macrophages constituted less than 10% of the EGFP-expressing cells during the first 5 days after skin wounding of LysEGFP mice [72] , indicating that neutrophils may contribute to more than 90% of the EGFP signals . In addition , the decreasing EGFP signals from days 1 to 10 correlated with the decreasing numbers of 7/4+ neutrophils and not the increasing numbers of MOMA2+ monocytes/macrophages at these time points as detected by immunohistochemistry ( Fig . 3B and S16 ) . Furthermore , EGFP was expressed within the cytoplasm and intracellular vesicles of LysEGFP neutrophils and the intensity of EGFP fluorescent signals was not substantially decreased in culture after neutrophil degranulation induced in response to fMLF or PMA ( Fig . S17 ) . Thus , the EGFP fluorescent signals more closely approximate of the numbers of neutrophils within the infected skin during the course of infection . With respect to using pIL1-DsRed transgenic mice , there are slightly different kinetics between DsRed fluorescence and IL-1β protein expression measured by ELISA , since DsRed fluorescence in extracts of inflamed skin and in cell culture was induced 6–12 hrs slower and persisted ∼24 hrs longer than IL-1β protein levels [21] . However , the difference in kinetics of DsRed fluorescence signals were less than 24 hrs and thus would be unlikely to impact the approximation of IL-1β production in vivo , since we began our measurements 1 day after infection and the infection takes over 14 days to resolve . Furthermore , there was substantial DsRed fluorescence signal on day 1 during our in vivo S . aureus skin infection ( Figs . 2 and 3 ) and at this time point we previously found that IL-1β was detected from the infected skin by ELISA and that both pro-IL-1β and cleaved IL-1β were detected by immunoblotting [11] , [17] . In the present study , we found that processing of pro-IL-1β into active IL-1β by mouse neutrophils was largely dependent upon ASC/NLRP3 inflammasome activation in vitro . We further demonstrated that IL-1β production was dependent upon the activity of α-toxin . These data are consistent with previous work in human or mouse monocyte/macrophage cultures demonstrating that processing of pro-IL-1β during S . aureus cutaneous infections in vivo was dependent upon ASC/NLRP3 inflammasome and caspase-1 activation , which was induced by S . aureus pore-forming toxins ( i . e . α- , β- and γ-hemolysins ) or digestion of peptidoglycan mediated by lysozyme [17] , [18] , [34] . Since there was some IL-1β measured in cultures of neutrophils from ASC-deficient mice as well as in cultures of wt neutrophils in the presence of the NLRP3 or caspase-1 inhibitor , the remainder of IL-1β may have been produced through an inflammasome-independent pathway mediated by serine-proteases ( such as proteinase 3 ) or neutrophil elastase [30]–[32] . In addition , the residual IL-1β detected in these cultures may also reflect transcription of pro-IL-1β since the ELISA detects both pro-IL-1β and cleaved IL-1β . In the adoptive transfer experiments , although the expression of IL-1β protein at the site of infection was extremely low compared with the levels observed in wt mice , scattered IL-1β-producing cells were detected by immunohistochemistry in IL-1β-deficient mice adoptively transferred with wt neutrophils ( Fig . S8 ) . While it is tempting to speculate that these adoptively transferred IL-1β-producing neutrophils directly rescued the neutrophil recruitment response at the site of infection , it is also possible the low levels of IL-1β acted indirectly and/or through another anatomical site such as the blood [73] . Future studies will be necessary to further dissect the mechanism by which neutrophil-derived IL-1β contributes to host defense during a S . aureus skin infection . Nevertheless , we show that IL-1β-producing adoptively transferred neutrophils were sufficient to rescue the impaired immunity in IL-1β-deficient mice . These findings provide evidence that neutrophil-derived IL-1β can promote effective neutrophil abscess formation and host defense against a cutaneous S . aureus infection . IL-1β production in response to live S . aureus cultured with mouse bone marrow-derived macrophages ( BMDMs ) , mouse peritoneal macrophages or human monocytes has previously been described [16]–[18] , [74] . These studies used different S . aureus strains , MOIs and culture conditions and the amount of IL-1β produced was generally 15- to 100-fold greater than the levels we observed with our in vitro infection of mouse neutrophils with S . aureus or MRSA . To evaluate whether IL-1β produced by the few contaminating monocytes/macrophages played any role in the adoptive transfer experiments , monocytes were specifically depleted from the adoptively transferred cells and the lack of monocytes had no impact on the ability of the adoptively transferred wt neutrophils to rescue the immune impairment in IL-1β-deficient mice . Furthermore , our in vitro infection experiments used highly purified mouse neutrophils separated with anti-Ly6G MACS magnetic beads ( 99 . 1% pure with only 0 . 1% monocytes ) to ensure that we were evaluating neutrophil specific production of IL-1β . To provide additional evidence that neutrophils and not other contaminating monocytes produced IL-1β in these cultures , purified neutrophils from pIL1-DsRed reporter mice were evaluated and 43% of the neutrophils expressed IL-1β-DsRed whereas only 0 . 2% of other cell types expressed IL-1β-DsRed . Taken together , these data indicate that neutrophils were the predominant source of IL-1β for both the adoptive transfer experiments and the in vitro infection experiments . Finally , a recent study evaluated sorted mouse bone marrow cells to determine which cell type produced the majority of IL-1β in response to LPS in the presence of the known inflammasome activators ATP or nigericin [75] . They found neutrophils were the predominant source of IL-1β as they produced almost 3-fold more IL-1β than F4/80+ macrophages . They further demonstrated that human neutrophils were responsible for half of all IL-1β secreted by human PBMCs . Lastly , they showed that IL-1β production by mouse and human neutrophils involved activation of inflammasome via NLRP3/ASC/caspase-1 axis . These data are consistent with our findings that neutrophils provide a major source of IL-1β during a S . aureus skin infection that is produced in an NLRP3/ASC/caspase-1-dependent manner In summary , we have identified that neutrophil-derived IL-1β is essential for amplifying the neutrophilic response to promote abscess formation and clearance of a S . aureus skin infection . From a clinical point of view , these findings provide the basis for targeting IL-1β production by neutrophils to improve immunity against pyogenic infections , especially in patients with impaired neutrophilic responses .
All animals were handled in strict accordance with good animal practice as defined in the federal regulations as set forth in the Animal Welfare Act ( AWA ) , the 1996 Guide for the Care and Use of Laboratory Animals , PHS Policy for the Humane Care and Use of Laboratory Animals , as well as UCLA's policies and procedures as set forth in the UCLA Animal Care and Use Training Manual , and all animal work was approved by the UCLA Chancellor's Animal Research Committee ( ARC#: 2008-099 ) . The bioluminescent S . aureus SH1000 strain ALC2906 , which possesses the shuttle plasmid pSK236 with the penicillin-binding protein 2 ( pbp2 ) promoter fused to the modified luxABCDE reporter cassette from Photorhabdus luminescens , was used as a representative S . aureus strain [3] . This strain emits bioluminescence signals from live , actively metabolizing bacteria in all stages of the S . aureus life cycle . In some experiments , a community-acquired MRSA strain was used ( USA300 LAC isolate [76] ) , which was kindly provided by Frank DeLeo ( National Institute of Allergy and Infectious Diseases , Rocky Mountain Laboratories in Hamilton , MT ) . SH1000 cultures were grown in the presence of chloramphenicol ( 10 µg/ml; Sigma-Aldrich , St . Louis , MO ) . S . aureus or MRSA was streaked onto tryptic soy agar ( tryptic soy broth [TSB] plus 1 . 5% bacto agar; BD Biosciences , Sparks , MD ) and single colonies were placed into TSB and grown overnight at 37°C in a shaking incubator . Mid-logarithmic phase bacteria were obtained after a 2 hr subculture of a 1∶50 dilution of the overnight culture . Bacterial cells were pelleted , resuspended , and washed 3 times in PBS . Bacterial concentrations were estimated by measuring the absorbance at 600 nm ( A600 ) ( Biomate 3; Thermo Scientific , Waltham , MA ) . In some experiments , bacteria was heat-killed ( 65°C for 30 minutes ) prior to infection . CFUs were verified by plating dilutions of the inoculum overnight . Male mice on a C57BL/6 genetic background were used in all experiments . pIL1-DsRed-reporter mice [21] , LysEGFP mice [25] , FPR1-deficient mice [77] , IL-1β-deficient mice [78] and ASC-deficient mice [33] were generated as previously described . IL-1R1-deficient mice ( B6 . 129S7-Il1r1tm1Imx/J ) , TLR2-deficient mice ( B6 . 129-TLR2tm1Kir/J ) and NOD2-deficient mice ( B6 . 129S1-Nod2tm1Flv/J ) and wt C57BL/6 mice were obtained from Jackson Laboratories ( Bar Harbor , ME ) . All mouse colonies were maintained in autoclaved cages under specific-pathogen free conditions . The mice were shaved on the back and inoculated intradermally with mid-logarithmic growth phase S . aureus ( 2×106 CFUs ) in 100 µl of sterile saline using a 27-gauge insulin syringe as previously described [3] . Measurements of total lesion size ( cm2 ) were made by analyzing digital photographs of mice using the software program Image J ( http://rsbweb . nih . gov/ij/ ) . Skin punch biopsy ( 8-mm ) specimens from uninfected or lesional skin were taken at 4 hrs after S . aureus intradermal inoculation from wt and IL-1R-deficient mice and homogenized ( Bio-Gen Pro200; Pro Scientific , Oxford , CT ) . RNA was isolated using TRIzol reagent ( Invitrogen , Grand Island , NY ) and purified using the RNeasy Mini kit ( Qiagen , Valencia , CA ) . The UCLA Microarray Core performed probe synthesis and hybridization to the GeneChip Mouse Genome 430 2 . 0 Array ( Affymetrix , Maumee , OH ) according to the manufacturer's protocol . Image files were processed using the invariant set method for probe selection during normalization and the model-based expression method of pooling information across arrays using dCHIP ( DNA-Chip Analyzer ) gene expression software ( www . dchip . org ) [79] . Genes were considered upregulated in S . aureus-infected skin at 4 hrs compared with uninfected skin according to the criteria: fold-change >1 . 5 , p-value<0 . 05 . Functional group and network analysis was performed using Ingenuity Pathway Analysis software ( version 6 . 0; Ingenuity Systems , Redwood City , CA ) as previously described [80] . The raw gene expression data for this study are available through the Gene Expression Omnibus database ( http://www . ncbi . nlm . nih . gov/geo/ ) under accession number GSE36826 . Total RNA from homogenized ( Pro200 Series homogenizer [Pro Scientific] ) 8-mm skin biopsy specimens taken at 4 hrs from skin inoculated with live or heat-killed S . aureus and uninfected skin was extracted by the use of TRIzol reagent ( Invitrogen ) , followed by DNase treatment ( Invitrogen ) according to the manufacturer's recommendations . Real-time quantitative real-time PCR ( Q-PCR ) reactions were performed as previously described [19] . TaqMan Gene Expression Assays primers and probes sets for a subset of genes of interest in the Cell Movement of Neutrophils sub-group , including IL-6 , G-CSF , MIP-2β , MIP-2α , IL-1β , GM-CSF , CXCR2 , G-CSF-R , TLR2 and LTB4R and the normalizer GAPDH were purchased from Applied Biosystems ( Foster City , CA ) . The relative quantities of mRNA per sample were determined using the ΔΔC ( T ) formula as previously described [3] Mice were anesthetized via inhalation of isoflurane and in vivo bioluminescence imaging was performed using the IVIS Lumina II imaging system ( Caliper Life Sciences , a PerkinElmer Company , Alameda , CA ) as previously described [3] . Data are presented on color scale overlaid on a grayscale photograph of mice and quantified as total flux ( photons/s ) within a circular region of interest using Living Image software ( Caliper ) . pIL1-DsRed mice and LysEGFP mice were anesthetized with inhalation isoflurane and in vivo fluorescence imaging was performed ( sequentially after in vivo bioluminescence imaging ) using the IVIS Lumina II imaging system ( Caliper ) . DsRed fluorescence was measured using: excitation ( 535 nm ) , emission ( 575–650 nm ) and exposure time ( 0 . 5 s ) . EGFP fluorescence was measured using: excitation ( 465 nm ) , emission ( 515–575 nm ) and exposure time ( 0 . 5 s ) . Data are presented on color scale overlaid on a grayscale photograph of mice and quantified as total radiant efficiency ( [photons/s]/[µW/cm2] ) within a circular region of interest using Living Image software ( Caliper ) . For histological analysis , lesional 8-mm punch biopsy skin specimens were embedded in Tissue-Tek OTC compound ( Sakura Finetek ) and cut into 4 µm sections by the UCLA Tissue Procurement and Histology Core Laboratory , according to guidelines for clinical samples . Frozen sections were fixed in acetone , air-dried , and rehydrated in PBS . Sections were permeabilized with 0 . 1% saponin in PBS and non-specific binding was blocked with 2% goat serum ( Invitrogen ) and mouse IgG2a ( 10 µg/ml; clone UPC 10 , Sigma-Aldrich ) in PBS . Sections were subsequently labeled with primary antibodies specific for DsRed ( rabbit anti-DsRed antibody; Clontech , Mountain View , CA ) in combination with mAbs specific for neutrophils ( anti-7/4 [Ly-6B . 2]; 5 µg/ml; AbD Serotec , Raleigh , NC ) , monocytes/macrophages ( anti-MOMA2; 5 µg/ml; AbD Serotec ) , antigen presenting cells ( anti-MHC II; 5 µg/ml; clone 2G9; BD Biosciences ) or total leukocytes ( anti-CD45; 5 µg/ml; clone 30-F11; BD Biosciences ) or appropriate isotype controls . Secondary antibodies included goat anti-rabbit IgG-Alexa 568 and goat anti-rat IgG-Alexa 488 ( Invitrogen , Carlsbad , CA ) . All specimens were imaged on a Leica SP2-1P FCS Confocal Microscope ( Leica Microsystems , Heidelberg , Germany ) as previously described [19] . Representative images of isotype controls are shown ( Fig . S4B , D ) . Quantification of co-localization was performed using the Manders' coefficient for a value range of 0 to 1 in which 0 = no pixels co-localize and 1 = all pixels co-localize using Definiens Tissue Studio software ( Definiens , Parsippany , NJ ) . The Manders' coefficient was determined from 4 different mice per group after averaging 2–3 fields of view per specimen . Detection of 7/4+ or MOMA2+ cells on frozen sections of lesional skin was performed with the anti-7/4 mAb or the anti-MOMA2 mAb as described above , followed by the biotinylated goat anti-rat IgG polyclonal antibody ( 5 µg/ml; Vector Labs , Burlingame , CA ) or corresponding isotype control antibodies . To detect IL-1β protein expression , a biotinylated anti-mouse IL-1β ( 20 µg/ml: clone 1400 . 24 . 17; Thermo Scientific ) or corresponding isotype control mAb was employed . All procedures were performed using the immunoperoxidase method as previously described [11] . MPO activity in lesional skin specimens was determined using an established MPO activity assay . Briefly , 8-mm punch biopsies were weighed and homogenized ( Bio-Gen Pro200; Pro Scientific ) in a buffer containing potassium phosphate ( 50 mM , pH 6 . 0 ) and hexadecyltrimethylammonium bromide ( 0 . 5%; Sigma-Aldrich ) . To measure MPO levels , 140 µl assay buffer , containing o-Dianisidine dihydrochloride ( 0 . 168 mg/ml; Sigma-Aldrich ) and hydrogen peroxide ( 0 . 05%; Sigma-Aldrich ) , was added to 10 µl of homogenized supernatant and the change in absorbance ( A490 ) was determined at 40 s intervals for 2 min using the Synergy 2 microplate reader ( BioTek , Winooski , VT ) . Purified MPO ( Sigma-Aldrich ) was used to generate a standard curve and data are presented as MPO activity ( U/mg tissue weight ) . For mouse neutrophils , the expression of the neutrophil-specific marker Ly6G ( FITC-conjugated rat anti-Ly6G mAb; 5 µg/ml; clone 1A8 , IgG1; BD Pharmingen ) and the monocyte-specific marker CD115 ( the M-CSF receptor ) ( PE-conjugated rat anti-CD115 mAb , 2 µg/ml; clone AFS98; eBioscience ) , CD11b ( APC-conjugated rat anti-CD11b mAb; 2 µg/ml; clone M1/70; BD Pharmingen ) and corresponding fluorescently-conjugated isotype control mAbs were used . Using these antibodies , Ly6G+ CD115− or Ly6G+ CD11bhigh represented mouse neutrophils and Ly6G− CD115+ or Ly6G− CD11blow represented mouse monocytes as previously described [81]–[83] . In some experiments , purified neutrophils from pIL1-DsRed reporter mice were used . pIL1-DsRed neutrophils were co-labeled with the anti-Ly6G mAb and prepared for flow cytometry as described above . For adoptive transfer experiments , neutrophils were obtained from the bone marrow of IL-1β-deficient or wt mice using Percoll density gradient centrifugation . Briefly , marrow cavities of the tibias and femurs of 8-week old mice were flushed with complete RPMI 1640 containing 10% FBS . After hypotonic lysis of red blood cells , mature neutrophils were isolated by centrifugation for 30 min at 10°C and 1600 g over a discontinuous Percoll gradient consisting of 50% ( vol/vol ) , 55% ( vol/vol ) , 62% ( vol/vol ) and 81% ( vol/vol ) Percoll ( Sigma-Aldrich , St . Louis , MO ) in PBS . The purity of the adoptively transferred cells was determined by flow cytometry using the neutrophil specific marker Ly6G ( anti-Ly6G mAb , clone 1A8 ) and the monocyte-specific marker ( anti-CD115 mAb , the M-CSF receptor ) . These markers have been shown to distinguish between mouse neutrophils and monocytes by flow cytometry [81] , [82] . The adoptively transferred cells were 90% neutrophils ( Ly6G+ CD115− cells ) , 0 . 26% monocytes ( Ly6G− CD115+ cells ) and 9 . 1% Ly6G− CD115− cells , which likely represented other granulocytes ( eosinophils or basophils ) or residual red blood cells not lysed with the lysis buffer ( Fig . S7A ) . After washing extensively in saline , 5×106 adoptively transferred neutrophils in 100 µl of sterile saline were injected intravenously into IL-1β-deficient mice two hrs prior to intradermal inoculation with S . aureus . In some experiments , neutrophils or monocytes were specifically depleted prior to adoptive transfer using either an anti-Ly6G or an anti-CD115 MicroBead Kit and MACS magnetic bead separation ( 61 . 4% and 76 . 9 percent depletion efficiency , respectively ) according to the manufacturer's protocols ( Miltenyi Biotec , Inc . , Auburn , CA ) ( Fig . S7B , C ) . In another set of adoptive transfer experiments , neutrophils were obtained from the bone marrow of TLR2-deficient or wt mice and adoptively transferred into TLR2-deficient recipient mice and the mice were infected with S . aureus according to the same procedures as described above ( Fig . S14 ) . For all in vitro cultures with mouse neutrophils , neutrophils were obtained from the bone marrow of TLR2- , NOD2 , FPR1- and ASC-deficient mice , pIL1-DsRed reporter mice or wt mice by anti-Ly6G MACs magnetic bead separation according to the manufacturer's protocols ( Miltenyi Biotec , Inc . ) . Purity of the mouse neutrophils was determined by flow cytometry ( see above ) and these cultures contained 99 . 1% Ly6G+ CD11bhigh neutrophils . There were very few Ly6G− CD11b+ monocytes ( 0 . 1% ) and the remaining cells ( 0 . 6% ) were Ly6G− CD11b− cells ( Fig . S13 ) . Murine neutrophils ( from TLR2- , NOD2- , FPR1- , ASC-deficient or wt mice ) were cultured in RPMI 1640 complete media supplemented with 10% heat-inactivated FBS at a density of 1×105 cells per 200 µl/well in a 96-well plate . These neutrophil cultures were infected with live S . aureus ( SH1000 strain ) or MRSA ( USA300 LAC isolate ) at a multiplicity of infection ( MOI ) of bacteria to neutrophils of 5∶1 , 2∶1 or 1∶1 at 37°C and 5% CO2 in a humidified incubator for 6 hrs . Gentamicin ( 20 µg/ml ) was added to the cultures at 60 minutes after infection according to previous methods to study inflammasome activation in response to live S . aureus in vitro [16] . Using these culture conditions , the MOI of 5∶1 for S . aureus or MRSA resulted in the highest production of IL-1β compared with MOI of 2∶1 or 1∶1 ( Fig . S9A , B ) . The levels of IL-1β in wt mouse neutrophils did not differ more than 15% between experiments . There was also no decrease in neutrophil viability in any of the cultures with the different MOI of S . aureus or MRSA compared with neutrophils cultured in the absence of any bacteria ( Fig . S9C , D ) . Therefore , the MOI of 5∶1 was used in all in vitro culture experiments . In some experiments , specific inhibitors were also added to the culture at the same time as S . aureus or MRSA . These include , the NLRP3-inhibitor , glibenclamide ( 100 µM; Imgenex ) [35] , [36] , the caspase-1 inhibitor Z-YVAD-FMK ( 20 µM; Millipore ) , anti-staphylococcal α-toxin antiserum ( 1% vol/vol; Sigma-Aldrich ) [37] , or respective vehicle controls ( DMSO or normal rabbit IgG ) . After in vitro infection of neutrophils with S . aureus or MRSA , cell viability was determined using the CellTiter 96 AQueous One Solution Cell Viability Assay ( Promega Corporation , Madison , WI ) according to the manufacturer's instructions . Protein levels of IL-1β from lesional mouse skin were obtained from tissue homogenates ( Pro200 Series homogenizer [Pro Scientific] ) of 8-mm skin punch biopsy specimens performed at 4 and 24 hrs after S . aureus skin inoculation using a commercially-available ELISA kit ( R&D Systems ) . Levels of mouse IL-1β protein in culture supernatants were determined by using a commercially available ELISA kit ( R&D Systems , Minneapolis , MN ) according to the manufacturer's instructions . For detection of pro-IL-1β ( 35 kDa ) and cleaved IL-1β ( 17 kDa ) by immunoblotting , purified mouse neutrophils from wt mice were cultured in RPMI 1640 media supplemented with 10% heat-inactivated FBS at a density of 1×106 cells per 500 µl/well in a 24-well plate . These neutrophil cultures were infected with live S . aureus ( SH1000 strain ) or MRSA ( USA300 LAC isolate ) at an MOI of bacteria to neutrophils of 5∶1 at 37°C and 5% CO2 in a humidified incubator for 6 hrs and gentamicin ( 20 µg/ml ) was added to the cultures at 60 minutes after infection . Following incubation , cells were lysed using the M-PER Mammalian Protein Extraction Reagent ( Thermo-Fisher ) supplemented with a Protease Inhibitor Cocktail ( Sigma-Aldrich ) . Cell lysates were diluted in SDS-PAGE sample buffer , boiled for 5 min , and proteins were separated by SDS-PAGE . Pro-IL-1β and cleaved IL-1β protein were detected by immunoblotting using a goat anti-mouse IL-1β polyclonal antibody ( 1∶2000 dilution; catalog #: AF-401-NA; R&D Systems , Minneapolis , MN ) followed by HRP-conjugated chicken anti-goat-HRP ( 1∶1000 dilution; catalog #: HAF019; R&D Systems ) . All assays were performed using anti-Ly6G magnetic bead enriched neutrophils obtained from bone marrow cells of wt or IL-1β-deficient mice as described above . Phagocytosis was measured using pHrodo S . aureus BioParticles ( Invitrogen ) , according to the manufacturer's instructions . Briefly , 1×105 neutrophils were incubated with fluorophore-conjugated S . aureus bioparticles at 37°C for 1 hr . Cells were then stained with FITC-conjugated anti-Ly6G and pHrodo-positive neutrophils were quantified by flow cytometry . For neutrophil degranulation , 1×105 neutrophils were stimulated for 30 min . at 37°C with 1 µM fMLF . Lactoferrin release was quantified from the supernatant using a mouse Lactoferrin ELISA kit ( Biotang , Inc . , Waltham , MA ) . Release of neutrophil reactive oxygen species was measured using the Phagoburst kit ( Orpegen Pharma , Heidelberg , Germany ) according to the manufacturer's instructions . Briefly , 5×105 neutrophils were treated with 1 µM fMLF for 10 min at 37°C . The generation of reactive oxygen species was measured by flow cytometry by gating on 10 , 000 neutrophil events and determining the proportion of these cells positive for the conversion of the substrate dihydrorhodamine-123 to fluorescent rhodamine-123 . Finally , neutrophil killing assays were performed by opsonizing S . aureus with 10% serum from C57BL/6 wt mice and adding the opsonized bacteria to purified neutrophils at a 1∶1 ratio ( 2×105 neutrophils∶2×105 CFU bacteria ) for 45 min at 37°C . After incubation , neutrophils were diluted in H2O ( pH 11 ) to lyse the neutrophils and serial dilutions were plated on TSB agar plates to enumerate viable bacterial CFU . As negative a control , bacteria were also incubated in media without neutrophils . Neutrophils were obtained from the bone marrow of LysEGFP mice using Percoll density gradient centrifugation . Neutrophils were washed once and resuspended in 1 ml of RPMI 1640 ( Gibco ) supplemented with 5 µg bisBenzinamide Hoescht 33342 trihydrochloride , a nuclear counterstain , ( Sigma-Aldrich ) for 30 minutes at room temperature . Neutrophils were subsequently attached onto glass slides using a Shandon Cytospin IV ( Thermo Scientific ) and imaged using an Olympus ×61 fluorescence microscope . To determine whether EGFP fluorescent signals were altered after neutrophil degranulation , neutrophils from LysEGFP mice were left unstimulated or stimulated with 1 µM fMLF or 100 ng/ml PMA ( both from Sigma-Aldrich ) for 15 minutes at 37°C . Cells were labeled with a biotinylated anti-mouse Ly6G ( University of California San Francisco Monoclonal Antibody Core ) with streptavidin-PE ( Caltag ) and anti-mouse PE-Cy7 CD11b ( Biolegend ) and analyzed on a Beckman Coulter FC500 flow cytometer . Data were compared using Student's t test ( 2-tailed ) . All data are expressed as mean ± SEM ( standard error of the mean ) where indicated . Values of *p<0 . 05 , †p<0 . 01 , and ‡p<0 . 001 were considered statistically significant .
|
Invasive infections caused by the human pathogen Staphylococcus aureus result in more deaths annually than infections caused by any other single infectious agent in the United States . Although neutrophil recruitment and abscess formation is crucial for effective host defense against this pathogen , how neutrophils sense and mount an inflammatory response are not completely clear . Using gene expression analysis and in vivo bioluminescence and fluorescence imaging , we found that neutrophil recruitment during a S . aureus cutaneous infection is functionally and temporally linked to IL-1β/IL-1R activation . Surprisingly , neutrophils themselves were determined to be the most abundant cell type that produced IL-1β during infection . Further , neutrophil-derived IL-1β , in the absence of other cellular sources of IL-1β , was sufficient for neutrophil recruitment , abscess formation , and bacterial clearance . Finally , mouse neutrophils produced IL-1β in direct response to live S . aureus in vitro . These findings expand our understanding of the acute neutrophil response to infection in which early recruited neutrophils serve as a source of IL-1β that is essential for amplifying and sustaining the neutrophilic response to promote abscess formation and bacterial clearance . Therapies aimed at promoting IL-1β production by neutrophils may be an effective immunotherapeutic strategy to control S . aureus infections .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"medicine",
"dermatology",
"immunity",
"to",
"infections",
"immunology",
"microbiology",
"staphylococcus",
"aureus",
"animal",
"models",
"bacterial",
"diseases",
"model",
"organisms",
"skin",
"infections",
"staphylococci",
"immunologic",
"techniques",
"bacterial",
"pathogens",
"infectious",
"diseases",
"inflammation",
"biology",
"staphylococcal",
"infection",
"mouse",
"immunofluorescence",
"immune",
"response",
"immunity",
"innate",
"immunity"
] |
2012
|
Neutrophil-derived IL-1β Is Sufficient for Abscess Formation in Immunity against Staphylococcus aureus in Mice
|
The concept of dynamical compensation has been recently introduced to describe the ability of a biological system to keep its output dynamics unchanged in the face of varying parameters . However , the original definition of dynamical compensation amounts to lack of structural identifiability . This is relevant if model parameters need to be estimated , as is often the case in biological modelling . Care should we taken when using an unidentifiable model to extract biological insight: the estimated values of structurally unidentifiable parameters are meaningless , and model predictions about unmeasured state variables can be wrong . Taking this into account , we explore alternative definitions of dynamical compensation that do not necessarily imply structural unidentifiability . Accordingly , we show different ways in which a model can be made identifiable while exhibiting dynamical compensation . Our analyses enable the use of the new concept of dynamical compensation in the context of parameter identification , and reconcile it with the desirable property of structural identifiability .
Some biological systems are capable of maintaining an approximatively constant output despite environmental fluctuations . It has long been accepted that negative feedback plays a central role in biological phenomena such as homeostasis . Feedback mechanisms are capable of rendering a system robust to a wide range of external disturbances . The ability to keep a constant steady state has been called exact adaptation , a feature that is known to be achievable with integral feedback [1–5] . The ability of preserving not only the steady state , but also the transient response ( i . e . the dynamic behaviour ) has been less studied and , despite recent contributions [6 , 7] , the mechanisms that make it possible are still less well understood . Recently , Karin et al . [6] addressed the problem of finding mechanisms that allowed to maintain the transient response unchanged in the face of environmental disturbances . To describe this phenomenon they coined the term dynamical compensation with respect to a parameter , which they defined as the property that the output of a system does not depend on the value of that parameter . According to this definition , dynamical compensation amounts to the parameter being structurally unidentifiable . Structural identifiability is a mathematical property originally introduced by Bellman and Åström [8] . If a parameter is structurally unidentifiable , it cannot be determined from experiments because there is an infinite number of values that yield the same model output . For example , symmetric expressions such as A = p1 × p2 or B = p1 + p2 yield the same result if the values of p1 and p2 are exchanged , so it is not possible to infer p1 and p2 by measuring functions of A and/or B . In such case parameters p1 and p2 are called structurally unidentifiable; a model containing structurally unidentifiable parameters is also termed structurally unidentifiable . While in the aforementioned examples structural unidentifiability is apparent , in practice it can be very difficult to detect such situation , even for small models , and many methodologies have been developed for this purpose , as reviewed e . g . in [9–13] . Structurally unidentifiable parameters pose several problems . Their estimated values are biologically meaningless [14] , and the use of a structurally unidentifiable model for predicting the time course of system variables that cannot be directly measured can produce wrong results [15] . This means that the usefulness of a model for obtaining biological insight can be compromised if its structural identifiability is not analysed . In recent years the importance of performing such analyses has been highlighted when modelling e . g . HIV infection [16] , diabetes [14] , infarction [17] , or cancer therapeutics [18] . Therefore , the correspondence between dynamical compensation and structural unidentifiability is relevant in realistic situations , in which the parameters of interest can be unknown . The equivalence between the original definition of dynamical compensation and structural unidentifiability was originally noted in [19 , 20] . In one of those papers [19] , Sontag drew an additional connection between dynamical compensation and system equivalence , and showed that a related property , fold-change detection ( FCD ) or input symmetry invariance , is a particular case of the same phenomenon . In this paper we begin by illustrating the correspondence between structural unidentifiability and the original definition of dynamical compensation , which we refer to as DC1 , using the four case studies presented by Karin et al . [6] . Then , as a new contribution , we suggest a more complete definition ( DC2 ) drawing from ideas implicit in the original publication [6] . Furthermore , given that structural identifiability is a desirable property for system identification , we enquire whether it is possible to reconcile the concept of dynamical compensation with it . We provide a positive answer by suggesting an alternative definition of dynamical compensation ( DC-Id ) which does not necessarily imply lack of structural identifiability , and preserves the intended meaning of the DC concept . Using for illustrative purposes one of the circuits proposed by Karin et al . [6] , we explore different modelling choices and show how they affect the identifiability of the model . We also compare our proposal with a different one suggested in a note by Karin , Alon , and Sontag [21] , which provides an alternative definition of dynamical compensation ( called DC3 in the present paper ) and analyses the aforementioned circuit . Finally , we discuss the implications of structural unidentifiability and show different ways in which it can be avoided , leading to identifiable models which may or may not exhibit the proposed definition of dynamical compensation ( DC-Id ) .
Karin et al . [6] introduced the concept of dynamical compensation to describe a design principle that provides robustness to physiological circuits . The original definition of dynamical compensation , which we refer to as “DC1” , is as follows: As explained above , the DC1 definition explicitly provided by Karin et al . [6] does not mention certain aspects whose omission can lead to confusion , and in fact , it can be considered as a rephrasing of the structural unidentifiability property [19 , 20] . However , the concept of dynamical compensation was not introduced with the aim of describing the same issue as structural unidentifiability . Instead , it was purported to describe a different phenomenon , specifically relevant for the regulation of physiological systems . To clarify the intended meaning of dynamical compensation in the context it was proposed , we use the βIG model of Fig 1D as an example . This model describes a glucose homeostasis mechanism where β stands for the beta-cell functional mass , I for insulin , and G for glucose . The time evolution of its three states in typical scenarios is shown in Fig 2 . The first row describes the behaviour after a series of pulses in glucose , where a pulse corresponds to an external input of glucose resulting from a meal . We consider a typical scenario of three meals , with roughly six hours between them . Both glucose and insulin concentrations reach peaks shortly after the meals , and in a few hours they return to their normal levels ( steady state ) . The second row describes what happens if the value of a parameter , insulin sensitivity ( si ) , is changed . Specifically , the figure represents the case in which s i n e w = 0 . 5 s i o l d . For ease of visualization no external pulses are applied in this simulation , so the plots in this row show the evolution of the system with zero input . There is a slow adaptation of the system’s steady state , which can take months , as seen in the figure . After this period the system has adapted to a new steady state: for glucose concentration it remains the same as the initial one ( exact adaptation ) , while the values of insulin concentration and β-cell mass are doubled . The third row illustrates the phenomenon of DC itself: after the adaptation to a new steady state has occurred , the output of the system ( from the new steady state , and with the new value of insulin sensitivity ) as a response to a pulse in glucose is the same as before the parameter change ( from the old steady state and the old parameter value ) . Note that only the glucose dynamics remains unchanged; for insulin and β-cell mass there is a scaling . In light of this behaviour , the following alternative definition of dynamical compensation ( DC2 ) may be deduced from a detailed reading of the original paper by Karin et al . : In reality , biological models almost always have a number of unknown parameters , whose values must be determined before the model can be used in practical applications . In this context the following question naturally arises: how does the behaviour described by the concept of dynamical compensation relate with structural identifiability of the parameters in the model ? As we have already mentioned , structural unidentifiability was shown to be equivalent to the original explicit definition of dynamical compensation , or DC1 [19 , 20] . The DC2 definition avoids this equivalence by requiring that the parameter is known . Is this , then , the end of the question ? Are unknown parameters with dynamical compensation “doomed” to be structurally unidentifiable , thus potentially limiting the biological insight that can be extracted from the models in which they appear ? We claim here that this is not necessarily the case , provided that we reformulate the definition of dynamical compensation . To show this , let us examine in more detail the structural identifiability of a system with dynamical compensation , the βIG model of Fig 1D . We analysed the structural identifiability of this model in its original formulation earlier in this paper , showing that , when its five parameters ( p , si , γ , c , α ) are considered unknown and plasma glucose concentration ( G ) is the only available measurement , the two parameters that exhibit dynamical compensation ( p , si ) are unidentifiable , while the remaining three are identifiable . Let us now see the results of such analysis when we change key aspects of the model , while preserving its dynamics . The two main choices we can play with are: ( i ) which parameters of the model are considered unknown , and therefore need to be estimated; and ( ii ) which measurements are possible . Regarding the first choice ( i ) , we analyse not only the five-parameter case considered by Karin et al . , but also other representative scenarios: when the unknown parameters are {si , γ , c , α} ( i . e . , all but p ) , when they are {p , si} , and when there is only one unknown , si . The second choice ( ii ) defines the output function of the model . While in general the output can be any function of the states , typically it consists of a subset of the states . In the version of the βIG model used by Karin et al . [6] the only measured variable was glucose concentration ( G ) . Here we consider all the possibilities , to assess the consequences of measuring every possible combination of the three state variables of the model: glucose ( G ) and insulin ( I ) concentrations , and beta-cell mass ( β ) . The set of 28 alternative model configurations and the corresponding results of the structural identifiability analysis are summarized in Table 1 . It can be noticed that there is substantial variability in the identifiability results depending on the modelling choices , despite the fact that the dynamic behaviour of the system is the same in all cases . Let us now see how the different configurations in Table 1 affect dynamical compensation . It should be noted that both the original definition of dynamical compensation ( DC1 ) and the second one ( DC2 ) consider single-output models . Specifically , Karin et al . demonstrated that the βIG model has dynamical compensation in glucose concentration ( G ) with respect to the {p , si} parameters . As can be seen in the first row of Table 1 , both parameters are structurally unidentifiable when G is the only output . To break this correspondence between dynamical compensation and structural unidentifiability we might interpret the “output” in the DC definition to be multi-dimensional . Indeed , if we could measure the three state variables we would make {p , si} identifiable . However , by doing so we would also destroy the dynamical compensation property , because there is no DC for β and I , as seen in Figs 2 and 3 . Thus , additional precisions should be incorporated into our working definition of dynamical compensation in order to make it describe a meaningful systemic property without being equivalent to structural unidentifiability . In light of this , we propose the following definition of dynamical compensation , which we call DC-Id: Shortly after the original DC publication [6] , two preprints noting the equivalence between DC and structural unidentifiability were posted: one by Sontag [25] , which was later published in this journal [19] , and our own [20] . Likewise , a few months later two new preprints appeared independently with the aim of reconciling DC with structural identifiability: the one on which the present paper is based [26] and another one by Karin , Alon , and Sontag [21] , which proposed an alternative definition of dynamical compensation that we will call DC3 . In the present subsection we comment on the latter one and compare it with our own proposal . Briefly , the DC3 definition includes two conditions for DC: ( i ) exact adaptation , and ( ii ) structural unidentifiability of the parameter of interest . Additionally , one of two alternative conditions must hold: either ( iii ) identifiability from perturbations , or ( iv ) identifiability given an additional output function . While more technical details are provided in [21] , the intuition behind DC3 is , in the words of its authors , to “require that while the parameter p of a DC model is unidentifiable from measurements of y at steady-state , it should be identifiable from other experimental measurements—either from measurements of y away from steady-state or from measurements of other system variables . ” . DC3 is clearly a more accurate definition of dynamical compensation than DC1 and DC2 . It is better at describing the biological phenomenon of interest , and discusses the relationship between the new property and structural ( un ) identifiability . It also acknowledges that it is desirable to have an identifiable DC parameter , and suggests ways of making it structurally identifiable . However , there are two main concerns with DC3 and the results provided in [21] . The first one is that the reference to measurements of “other variables” is somewhat confusing because , by definition , the output of a model consists of the quantities that are measured ( which are often states , but may sometimes consist of the sum or other functions of the states and parameters ) and , since the output function is part of the model structure , changing it by measuring additional state variables amounts to having a different model . This makes condition ( iv ) in this definition problematic , strictly speaking . In this regard , it is also worth mentioning that when DC3 is formalized in [21] the state x ( t ) is defined as “an n-dimensional vector of state variables” and the output y ( t ) as “the output variable” . The use of different wording for each of them seems to imply that y is one-dimensional , which would make this definition not valid for the general case of models with multidimensional outputs . The second—and arguably more important—issue is that the reference to measurements “away from steady-state” can be misleading . This is discussed in the following paragraphs , where we analyse the results presented in [21] and show that the approach suggested in said paper can lead to incorrect conclusions . The first result reported in [21] is that conditions ( i ) and ( ii ) hold for the βIG model , and that ( iv ) also holds if insulin and β-cell mass are measured . Therefore the model has dynamical compensation and is structurally identifiable for the output pair {I , β} . This agrees with our own results , as shown in Table 1 . This result is followed by another analysis , which concludes that “given p we can infer s either from either ( i ) measurements of glucose and insulin at steady state , or ( ii ) measurements of glucose off steady state . To infer p we only require some additional measurement such as beta cell mass” ( note that si is written as s in the quoted text ) . The paragraph above contains three claims , of which the first one is correct: given knowledge of p and measurements of glucose and insulin , we can indeed infer si , as reported in Table 1 . However , the remaining two claims are incorrect . Let us examine them in more detail . One claim is that , given p , it is possible to infer si from “measurements of glucose off steady state” . It should be noted that most of the measurements that we would usually collect for the βIG model ( e . g . in a scenario such as the one pictured in the third row of Fig 2 , with inputs of glucose from meals ) are already off steady state ( i . e . dG ( t ) /dt ≠ 0 ) , since the system only returns to steady state a few hours after the external pulse of glucose . However , given the context in which the words “off steady state” are used in [21] , we might interpret that Karin et al . are specifically referring to the particular situation that takes place immediately after the insulin sensitivity parameter is changed as a result of a perturbation , which is illustrated in Figure 1 in [21] and in the second row of Fig 2 in the present paper . As explained before , such perturbation instantaneously modifies the system’s steady state , which then goes back to the initial one after a long adaptation period . During this adaptation period the system is transitioning between two different steady states . Figure 1 in [21] shows that , while the glucose curves are identical for ( I ) si before adaptation and for ( II ) si/2 after adaptation , they do not coincide with ( III ) si/2 during adaptation . For this reason Karin et al . argue that during this period it is possible to identify si from glucose measurements . However , this is not true: as shown in Fig 4 , the time course of glucose ( leftmost plot in the lower row ) is the same for a model with si and for another with si/k , as long as the initial concentrations of insulin and β-cell mass of the second model are multiplied by the same constant k . And this holds even if the value of si is changed during the course of the experiment , triggering the slow adaptation . What is happening in this case is that , although there is only one unknown parameter ( si ) , there are also two unmeasured states ( I , β ) , and it is possible to compensate the variation in model output ( G ) originated from changes in the parameter with coordinated changes in the two unmeasured states . Thus , if only glucose is measured ( left plot in Fig 4 ) , it is impossible to distinguish between si and si/k—even if we know the value of p , which is the same in both cases—and therefore the parameter si is structurally unidentifiable . This is in agreement to the results reported in Table 1 . The remaining claim ( “To infer p we only require some additional measurement such as beta cell mass” ) is also incorrect: if not only si but also p are unknown , and besides glucose we measure also β-cell mass , both parameters are unidentifiable even with measurements off steady state . This can be realised by inspecting the lower row of Fig 3 , which shows that the time courses of glucose and β-cell mass are identical for two different parameter vectors ( {si/2 , 2 ⋅ p} and {si/10 , 10 ⋅ p} ) , both in and off steady state . More generally , any pair of values {si/k , k ⋅ p} will yield the same output as a reference vector {si , p} , as long as insulin is not measured . For this model , inferring both {si , p} always requires measuring at least β-cell mass and insulin , as we have shown in this paper ( see Table 1 ) . The cause of the inaccurate claims in [21] is that often times the cause of unidentifiability is the correlation between parameters , or between parameters and state variables . In that case , as happens with the βIG model , structural identifiability cannot be determined in a step-wise fashion and intuitive reasoning can lead to misjudgements . Such issues can be circumvented by performing a rigorous structural identifiability analysis and adopting a definition of DC like the one proposed in the present paper , DC-Id .
The fact that a model is unidentifiable is important because , after five decades of research , it is now well understood that lack of structural identifiability is the result of choosing an inappropriate model structure for the available measurable variables ( or variables that can be directly observed ) [23 , 27] . When understood in this way , structural unidentifiability can be avoided or surmounted in at least three ways: ( i ) by reducing the number of parameters or changing their definition , ( ii ) by increasing the number of measured variables , if possible , or ( iii ) by determining the unidentifiable parameters in some alternative way , e . g . by direct measurements . Strategy ( i ) entails reformulating the model to remove redundant parameters , for example , by grouping several non-identifiable parameters into a single identifiable one . Perhaps the simplest example would be the merging of two parameters that multiply each other into a single one , i . e . pnew = p1 × p2 . Such relationships can be revealed systematically by performing a structural identifiability analysis . Some techniques such as COMBOS [28] are explicitly designed for finding identifiable combinations of otherwise unidentifiable parameters , and other methods may also be used for this purpose [29–33] . Strategy ( ii ) can be illustrated with the “βIG” model: if it were possible to measure all its three states instead of only glucose , all the parameters in the βIG model would become structurally identifiable . In other words , while the effect on the glucose concentration ( G ) of a change in p can be compensated by changing si , this does not happen for the insulin concentration ( I ) . Since it may not be realistic to measure β-cell mass continuously , another possibility could be to assume it constant ( since it changes very little , as can be seen in Fig 3 ) and use as an estimate of it a single measurement obtained in the past . With this assumption it suffices to monitor the insulin concentration ( I ) to obtain an identifiable model , as seen in Table 1 . Strategy ( iii ) was applied for example by Watson et al . [34] . After determining that two parameters in a homeostatic model were structurally unidentifiable , they decided to measure one of them by means of a tracer experiment and to calculate an estimate of the other using a steady state assumption . Strategies ( ii ) and ( iii ) demonstrate how measurements and data can directly inform modelling decisions . More generally , structural identifiability analysis can inform expectations about how precisely a model can be defined , given measurements and data . For example , if the state of a system changes very little when a parameter varies , the system is sometimes said to be robust or insensitive to variations in that parameter [27] . Speaking in terms of identifiability , this scenario may be connected to poor practical identifiability: although the value of the parameter has some influence on the model output , its effect is too small to allow for its precise determination due to limitations in the information content of the data ( regarding quantity and/or quality ) [23 , 35] . In contrast , when the sensitivity of the model output to a parameter is exactly zero , as implied by DC1 , it corresponds to lack of structural identifiability . In this case , the value of the parameter has no influence at all on the model output . This situation represents an “unreasonable” elasticity which should not be interpreted as a sign of biological robustness , but as an indication that the parameter is not meaningful . Ideally , it should be removed and the model should be modified , as explained above . It should be noted that in realistic situations the values of estimated parameters always have some associated uncertainty . Practical identifiability analysis ( which is sometimes referred to as numerical identifiability , estimability , or a posteriori identifiability ) quantifies the uncertainty that results from limitations in the information content of the data used for calibration [13 , 23 , 27 , 35] . Unlike practical identifiability , dynamical compensation and structural identifiability are both a priori concepts , that is , they can be studied before collecting experimental data . In this regard , the uncertainty in parameter estimates does not play a role in dynamical compensation . As recently stressed by Janzén et al . [36] , the danger of inadvertently using a structurally unidentifiable model is that the biological interpretations of its parameters are not valid , which may lead to wrong conclusions; furthermore , any predictions involving unmeasured states “may be meaningless if the parameters directly or indirectly related to those states are unidentifiable” [36] . This fact can be illustrated with the βIG model , as seen in the second row of Fig 3: if we try to estimate the p , si parameters from glucose ( G ) measurements , we will not be able to recover their true values , because they are structurally unidentifiable: there is an infinite number of combinations of their values that yield the same glucose profile . This , in turn , means that we cannot use the model to predict the time-course of insulin concentration ( I ) , which is an unmeasured state . As seen in the lower plot of the third column , the predictions of insulin can be very different depending on the pair of p , si values used .
Given that deficiencies in identifiability may lead to wrong reconstructions of a system’s behaviour , and that parameter identification is an ubiquitous need in biological modelling , it is necessary to assess the structural identifiability of a model before using it to extract insights about the corresponding biological system . The absence of structural identifiability considerations in the paper that introduced dynamical compensation [6] led to an ambiguous definition of the latter concept , which we have termed DC1 in the present manuscript . The fact that DC1 is essentially equivalent to structural unidentifiability when examined from the viewpoint of model identification , as noted in [19 , 20] , is a source of potential confusion: it opens the door to ( i ) interpreting as dynamical compensation what might be a case of structural unidentifiability , and to ( ii ) inadvertently using structurally unidentifiable models . It is possible to deduce from a detailed reading of the original paper [6] an alternative definition of dynamical compensation ( called DC2 in the present manuscript ) , which removes some ambiguities of DC1 . Another alternative definition , which we have called here DC3 , was suggested in [21] , but its application can be problematic , as shown in this paper . Importantly , neither DC1 nor DC2 are appropriate for realistic modelling scenarios , in which it is necessary to estimate the values of parameters from input-output data . To overcome this limitation we have proposed a modification of the definition of dynamical compensation which can be used in such cases . Our new definition , termed DC-Id , captures the biological meaning of the dynamical compensation phenomenon , which is the invariance of the dynamics of certain state variables of interest with respect to changes in the values of certain parameters . But , additionally , it includes precisions that make it distinct from structural unidentifiability , even in the context of parameter identification—that is , when it is necessary to determine the values of the model parameters . It is thus unambiguous and generally applicable . We see the discussion held in the present paper and the resulting clarification as an example of the gains that can be obtained by exchanging more notes among the different communities working in biological modelling , which we have advocated elsewhere [37] . Such an exchange of notes increases researchers’ awareness of community-specific knowledge and is useful for avoiding potential misconceptions .
We consider state-space models described by ordinary differential equations ( ODEs ) of the following general form: M : { x ˙ ( t ) = f ( x ( t ) , p , u ( t ) ) y ( t ) = h ( x ( t ) , p ) x 0 = x ( p ) ( 2 ) Following the usual convention , we use: x to refer to state variables , u for inputs , y for outputs , and p for parameters . States , inputs , and outputs are in general time-varying , while parameters are constants ( it could also be possible to take into account time-varying parameters , but these are rare in biological models [27]; for an exception , see e . g . the model of glucose turnover by Steele et al . [38] ) . In Eq ( 2 ) , f and h are analytic vector functions of the states and parameters , which are in general nonlinear ( linear models are a particular case ) . For ease of notation we can omit the dependence of f and h on p , and denote initial values of state variables or inputs as x0 = x ( 0 ) and u0 = u ( 0 ) , respectively . We also often drop the time dependence , i . e . we write x instead of x ( t ) , and so on . We remark that by “model structure” we refer not only to the dynamic equations ( x ˙ ) but also to the definition of the observation function , or set of measured model outputs ( y ) , and the known input variables ( u ) . It should also be noted that the model output y ( t ) does not take noise into account , since it does not play a role in the concepts discussed in the present paper . Structural identifiability and dynamical compensation are both a priori properties , which can be analysed before performing any measurements . Of course , in a realistic parameter estimation scenario it is also necessary to take into account limitations introduced by the quantity and quality of the available data . This is the related topic of practical or numerical identifiability , which aims at quantifying the uncertainty in the estimated parameter values that results not only from the model structure but also from data limitations , including noise [13 , 23 , 27 , 35] . Among the existing approaches for structural identifiability ( SI ) analysis , we adopt one that considers SI as a generalization of observability—the property that allows reconstructing the internal state ( x ) of a model from observations of its outputs ( y ) . If a model is observable there is ( at least locally ) a unique mapping from y to x , and two different states will lead to two different outputs . Observability is a classic system-theoretic property introduced by Kalman for linear systems , and extended to the nonlinear case by Hermann and Krener [39] , among others . It can be studied with a differential geometry approach , as described in the remainder of this subsection . A thorough treatment of this matter can be found in the books by Vidyasagar and Sontag [40 , 41] . Observability analysis determines if the mapping from y to x is locally unique by analysing the expression of y = h ( x ) and its derivatives . This is done by constructing an observability matrix that defines this mapping , and then calculating its rank . If the matrix is not full rank , the same output can be produced by an infinite number of state vectors , and the system is unobservable . In the nonlinear case , the observability matrix can be built using Lie derivatives . The extended Lie derivative of h with respect to f is: L f h ( x ) = ∂ h ( x ) ∂ ( x ) f ( x , u ) + ∑ j = 0 j = ∞ ∂ h ( x ) ∂ u ( j ) u ( j + 1 ) ( 3 ) where u ( j ) and u ( j+1 ) denote the ith and ( i + 1 ) th derivatives of the input , respectively . Higher order Lie derivatives can be recursively calculated from lower order ones as: L f i h ( x ) = ∂ L f i - 1 h ( x ) ∂ x f ( x , u ) + ∑ j = 0 j = ∞ ∂ L f i - 1 h ( x ) ∂ u ( j ) u ( j + 1 ) ( 4 ) The nonlinear observability matrix can be written as: O ( x ) = ( ∂ ∂ x h ( x ) ∂ ∂ x ( L f h ( x ) ) ∂ ∂ x ( L f 2 h ( x ) ) ⋮ ∂ ∂ x ( L f n - 1 h ( x ) ) ) ( 5 ) where n is the dimension of the state vector x . We can now formulate the Observability Rank Condition ( ORC ) as follows: if the system given by Eq ( 2 ) satisfies rank ( O ( x 0 ) ) = n , where O is defined by Eq ( 5 ) , then it is ( locally ) observable around x0 [39] . This condition guarantees local observability , which means that the state x0 can be distinguished from any other state in a neighbourhood , but not necessarily from distant states . The distinction between local and global identifiability is usually not relevant in biological applications . By considering the parameters as state variables with zero dynamics ( p ˙ = 0 ) , SI analysis can be recast as observability analysis . To this end , we augment the state vector as x ˜ = [ x , p ] and write the generalized observability-identifiability matrix as: O I ( x ˜ ) = ( ∂ ∂ x ˜ h ( x ˜ ) ∂ ∂ x ˜ ( L f h ( x ˜ ) ) ∂ ∂ x ˜ ( L f 2 h ( x ˜ ) ) ⋮ ∂ ∂ x ˜ ( L f n + q - 1 h ( x ˜ ) ) ) ( 6 ) where n is the dimension of the state vector x and q is the dimension of the parameter vector p . We can now state a generalized Observability-Identifiability Condition ( OIC ) : if a system satisfies rank ( O I ( x ˜ 0 ) ) = n + q , it is ( locally ) observable and identifiable around the state x ˜ 0 . If rank ( O I ( x ˜ 0 ) ) < n + q , the model contains unidentifiable parameters ( and/or unobservable states ) . It is possible to determine the identifiability of individual parameters because each column in OI contains the partial derivatives with respect to one parameter ( or state ) . Thus if the matrix rank does not change after removing the ith column the ith parameter is not identifiable ( if the column corresponds to a state , it is not observable ) . The software used in this paper for analysing structural identifiability is STRIKE-GOLDD ( STRuctural Identifiability taKen as Extended-Generalized Observability with Lie Derivatives and Decomposition ) . It is a methodology and a tool for structural identifiability analysis [24] which can handle nonlinear systems of a very general class , including non-rational ones . At its core is the conception of structural identifiability as a generalization of observability . Since the calculation of rank ( O I ( x ˜ 0 ) ) can be computationally very demanding , even for models of moderate size , STRIKE-GOLDD includes a number of algorithmic modifications to alleviate its cost . One of them is the construction of the observability-identifiability matrix O I with less than n + q − 1 derivatives . In certain cases , this reduced matrix can suffice to establish the identifiability of the whole model; in other cases , it can at least report identifiability of a subset of parameters , even if it cannot decide on the rest . Another possibility is to decompose the model in a number of submodels , which have smaller matrices whose rank is easier to compute . More details about these and other procedures included in the methodology can be found in the STRIKE-GOLDD publication [24] . STRIKE-GOLDD is an open source MATLAB toolbox that can be downloaded from https://sites . google . com/site/strikegolddtoolbox/ . A more complete description of the tool can be found in its user manual , which is available in the website . All the code ( including the STRIKE-GOLDD toolbox ) and instructions required for reproducing the results reported in this paper are provided in S1 File .
|
A robust behaviour is a desirable feature in many biological systems . The study of mechanisms capable of maintaining the transient response unchanged despite environmental disturbances has recently motivated the introduction of a new concept: Dynamical Compensation ( DC ) . However , the original definition of DC with respect to a parameter amounts to structural unidentifiability of that parameter , which means that it cannot be estimated by measuring the model output . Since most biological models have unknown parameters that need to be estimated , DC can be considered a negative property for the purpose of model identification . In this paper we reconcile these two conflicting views by proposing a new definition of DC that captures its intended biological meaning ( i . e . robustness , which should be a systemic property , intrinsic to the dynamics ) while making it distinct from structural unidentifiability ( which is a modelling property that depends on decisions made by the modeller , such as the choice of model outputs or unknown parameters , and on experimental constraints ) . Our definition enables a model to have DC with respect to a structurally identifiable parameter , thus increasing the applicability of the concept .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Conclusions",
"Methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"chemical",
"compounds",
"diabetic",
"endocrinology",
"carbohydrates",
"organic",
"compounds",
"glucose",
"hormones",
"physiological",
"processes",
"systems",
"science",
"mathematics",
"homeostasis",
"evolutionary",
"adaptation",
"insulin",
"computer",
"and",
"information",
"sciences",
"nonlinear",
"systems",
"endocrinology",
"chemistry",
"dynamical",
"systems",
"differential",
"equations",
"nonlinear",
"dynamics",
"biochemistry",
"organic",
"chemistry",
"physiology",
"monosaccharides",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"evolutionary",
"biology",
"evolutionary",
"processes"
] |
2017
|
Dynamical compensation and structural identifiability of biological models: Analysis, implications, and reconciliation
|
Listeria monocytogenes is a bacterial pathogen whose genome encodes many cell wall proteins that bind covalently to peptidoglycan . Some members of this protein family have a key role in virulence , and recent studies show that some of these , such as Lmo0514 , are upregulated in bacteria that colonize eukaryotic cells . The regulatory mechanisms that lead to these changes in cell wall proteins remain poorly characterized . Here we studied the regulation responsible for increased Lmo0514 protein levels in intracellular bacteria . The amount of this protein increased markedly in intracellular bacteria ( >200-fold ) , which greatly exceeded the increase in lmo0514 transcript levels ( ∼6-fold ) . Rapid amplification of 5′-cDNA ends ( RACE ) assays identified two lmo0514 transcripts with 5′-untranslated regions ( 5′-UTR ) of 28 and 234 nucleotides . The transcript containing the long 5′-UTR is upregulated by intracellular bacteria . The 234-nucleotide 5′-UTR is also the target of a small RNA ( sRNA ) denoted Rli27 , which we identified by bioinformatics analysis as having extensive base pairing potential with the long 5′-UTR . The interaction is predicted to increase accessibility of the Shine-Dalgarno sequence occluded in the long 5′-UTR and thus to promote Lmo0514 protein production inside the eukaryotic cell . Real-time quantitative PCR showed that Rli27 is upregulated in intracellular bacteria . In vivo experiments indicated a decrease in Lmo0514 protein levels in intracellular bacteria that lacked Rli27 . Wild-type Lmo0514 levels were restored by expressing the wild-type Rli27 molecule but not a mutated version unable to interact with the lmo0514 long 5′-UTR . These findings emphasize how 5′-UTR length affects regulation by defined sRNA . In addition , they demonstrate how alterations in the relative abundance of two transcripts with distinct 5′-UTR confine the action of an sRNA for a specific target to bacteria that occupy the intracellular eukaryotic niche .
Listeria monocytogenes is a facultative intracellular food-borne bacterium responsible for serious clinical manifestations including febrile gastroenteritis , meningitis , encephalitis and maternofetal infections in humans and livestock , with an estimated fatality rate of 20–30% of infected individuals [1]–[3] . Following ingestion , L . monocytogenes is able to cross the intestinal , blood-brain and placental barriers . The bacterium expresses a number of virulence factors that promote entry into phagocytic and non-phagocytic eukaryotic cells , intracellular survival and proliferation , and spreading to adjacent cells [4] . Genome studies show that all Listeria species sequenced to date have more than 40 genes that encode predicted surface proteins bearing an LPXTG sorting motif [5] . This motif is recognized by sortase enzymes , which anchor these proteins covalently to the cell wall . In pathogenic Listeria , some of these LPXTG proteins direct essential steps throughout the infection process , including bacterial adhesion and uptake by the host cell [6]–[9] . Proteomic analyses indicated that levels of many of these LPXTG surface proteins change on adaptation to different environments . The Listeria cell wall subproteome thus changes substantially in actively growing and resting bacteria . Mutants that lack sortase SrtA and SrtB activity show impaired LPXTG protein anchoring to the peptidoglycan [10] as well as differences in the relative levels of certain LPXTG proteins [11] . Recent studies also showed major changes in the cell wall proteome when L . monocytogenes proliferate inside epithelial cells [12] . Upregulation of defined LPXTG proteins has been observed in intracellular bacteria , including Internalin-A and Lmo0514 [12] . The mechanisms that regulate the coordinated production of such a large number of LPXTG proteins nonetheless remain largely unknown . Bacterial small RNAs ( sRNA ) are a class of bacterial gene expression regulators important in many physiological processes , including virulence and cell envelope homeostasis [13] , [14] . sRNA coordinate target gene expression in response to environmental changes and have regulatory functions that affect protein activity and mRNA stability/translation in many microorganisms , including bacterial pathogens [15] , [16] . More than 100 sRNA have been identified for L . monocytogenes by the use of tiling arrays , global RNA sequencing ( RNA-Seq ) and bioinformatics methods [14] , [17] , [18]; more than 30 of these have been validated by northern blot , but their biological function and mechanisms of action are so far unknown [19] . There is little information on the regulation of sRNA expression in L . monocytogenes . Some reports implicate the alternative sigma factor SigB in regulating expression of the sRNA SbrA ( Rli11 ) and SbrE ( Rli47 ) [17] , [20] , [21] . In addition , 22 sRNA genes are preceded by putative sigma A boxes in the L . monocytogenes genome [14] . Recent studies also show that the sRNAs Rli31 , Rli33-1 , Rli38 and Rli50 modulate virulence in L . monocytogenes [14] , [17] . Despite these studies , there is no model that describes how sRNA expression in L . monocytogenes responds to infection of eukaryotic cells . With the exception of LhrA , which controls expression of the chitinase ChiA post-transcriptionally [22] , and of the multicopy sRNA LhrC , which modulates LapB adhesin expression [23] , the identity of the functions targeted by L . monocytogenes sRNA inside or outside eukaryotic cells , remains unknown . Here we studied the regulatory mechanism responsible for the increase in the LPXTG protein Lmo0514 in the cell wall of intracellular bacteria [12] . Our data demonstrate an sRNA that is a key regulatory element in modulating levels of this cell wall surface protein during intracellular infection . This response to the eukaryotic niche is directed by the activity of two promoters in the target gene that generate transcripts with 5′-untranslated regions ( 5′-UTR ) of distinct length . The relative abundance of these two transcripts differs in extra- and intracellular bacteria . Only the ‘long’ version , enriched in intracellular bacteria , bears the sRNA binding site . This mechanism confines the regulation of lmo0514 by this sRNA to the intracellular eukaryotic niche .
Lmo0514 , a L . monocytogenes LPXTG surface protein of unknown function , is encoded by a gene upregulated by bacteria located within macrophages [24] . Lmo0514 is also more abundant in the cell wall of bacteria that proliferate inside epithelial cells than in bacteria growing in laboratory media [12] . To study the basis of this regulation , we compared lmo0514 expression in extra- and intracellular bacteria . Real-time quantitative PCR ( qPCR ) assays showed enhanced lmo0514 mRNA expression ( ∼6-fold ) in intracellular bacteria after infection of JEG-3 human epithelial cells ( Fig . 1A ) . Consistent with our previous work [12] , the Lmo0514 protein was detected mainly in the cell wall of intracellular bacteria , with very low levels in extracellular bacteria ( Fig . 1B ) . Changes in relative levels of Lmo0514 protein were estimated to be>200-fold ( Fig . 1B ) , much higher than those for lmo0514 mRNA ( ∼6-fold ) . This lack of correlation between induction of lmo0514 transcript and protein levels in intracellular bacteria led us to hypothesize that post-transcriptional regulatory mechanisms act on this gene . To evaluate this possibility , we sought lmo0514 gene expression control mechanisms that operate specifically in intracellular bacteria . Previous in silico predictions by Loh et al . [25] indicated that lmo0514 could be expressed from three promoters at positions −26 , −104 and −163 . Two of these , −26 and −163 , were assigned as tentatively regulated by sigma A ( σA ) and the third , at position −104 , as controlled by sigma B ( σB ) [25] ( Fig . 2A ) . The activity of these putative promoters and the presence of the different transcripts were analyzed by RT-PCR on RNA isolated from L . monocytogenes grown extracellularly and from intracellular bacteria that colonized JEG-3 epithelial cells . lmo0514 transcripts with a long 5′-UTR were detected specifically in intracellular bacteria ( Fig . 2A ) . To confirm these findings , rapid amplification of 5′-cDNA ends ( 5′-RACE ) assays were used to map transcriptional start sites ( TSS ) of lmo0514 in bacteria grown extracellularly and in bacteria isolated from eukaryotic cells . These 5′-RACE assays revealed two distinct TSS at positions −28 and −234 ( Fig . 2B , C ) , and also confirmed expression of the long lmo0514 transcript by intracellular bacteria ( Fig . 2C ) . Putative promoters for these TSS , which we termed P1 and P2 , both bear bona fide −10 TATA boxes ( Fig . 2B , C ) . The existence of two lmo0514 transcripts of different length was verified by northern blot ( Fig . 3A ) , with sizes compatible with cotranscription of lmo0514 with the downstream gene lmo0515 , which encodes a universal stress protein [26] . lmo0514-lm0515 cotranscription was verified by RT-PCR ( Fig . S1 ) . qRT-PCR assays confirmed that expression of the lmo0514 transcript variant with the long 234-nucleotide ( nt ) 5′-UTR was upregulated by ∼12-fold in intracellular bacteria ( Fig . 3B ) . These findings suggested that the specific induction of this mRNA variant with a longer 5′-UTR in intracellular bacteria accounts for or contributes to the 6-fold increase in total lmo0514 mRNA ( Fig . 1A ) . These data supported a model in which intracellular bacteria specifically upregulate expression from the P2 promoter , resulting in an lmo0514 transcript with a long 5′-UTR . This assumption takes into account the different ratios between the two lmo0514 transcripts when L . monocytogenes colonizes the eukaryotic cell . The increased length of the lmo0514 transcript variant that is upregulated in intracellular bacteria prompted us to test whether the distinctive 234-nt 5′-UTR is a target region for sRNA-mediated post-transcriptional regulation . We used in silico analysis to search for putative non-coding RNAs in the L . monocytogenes reference strain EGDe [27] that could bind to this lmo0514 long 5′-UTR . The targetRNA program ( http://cs . wellesley . edu/~btjaden/TargetRNA2/ ) [28] gave a high score to a pairing between defined stretches of the lmo0514 234-nt 5′-UTR and a sequence in the lmo0411-lmo0412 intergenic region . A gene in this region encodes an sRNA termed Rli27 that is upregulated by L . monocytogenes in the intestine of infected mice and in human blood , as shown by transcriptomics [17]; RNA-seq corroborated the expression of this sRNA [18] . Although Rli27 was identified as an sRNA induced in infection conditions [17] , no further characterization of its function or targets was reported . Genomic comparisons of pathogenic and non-pathogenic species are usually carried out to identify virulence genes , including sRNAs [18] , [29] . We analyzed the genomic region of L . monocytogenes containing rli27 and those of the non-pathogenic species L . innocua and L . welshimeri . In L . monocytogenes , rli27 is flanked by lmo0411 and lmo0412 , two genes that map in the opposite DNA strand ( Fig . S2 ) , whereas in the L . welshimeri genome , the same intergenic region has a small ORF ( lwe0373 ) that codes for a predicted protein of unknown function ( Fig . S2 ) . We nonetheless found that Rli27 is highly conserved in L . innocua ( 82% identity , Fig . 4A ) , in contrast with a previous report [17] . The extremely variable rli27 genomic region might thus have been shaped by gain and/or loss of genes during Listeria speciation . Apart from Listeria species , BLAST searches did not identify rli27 orthologs in other bacterial species . Rli27 , identified as a 131-nt sRNA [14] , [18] , is not predicted to encode any protein using the Small Open Reading Frame ( ORF ) tool in the ORF finder program ( http://www . bioinformatics . org/sms2/orf_find . html ) . Although the existence of Rli27 sRNA was inferred based on its detection by genomic and transcriptomic approaches , it has not yet been formally demonstrated . The presence of rli27 and its flanking genes in different strands ruled out the possibility that its detection by tiling arrays and RNA-seq analyses was due to untranslated regions of neighbor genes . rli27 has its own predicted transcription start site and Rho-independent terminator sequence ( Fig . 4A ) , and the respective promoter regions in L . monocytogenes and L . innocua showed no significant divergence ( Fig . 4A ) . Northern blot assays using total RNA isolated from L . monocytogenes wild-type EGD-e and an isogenic Δrli27 mutant strain demonstrated a small transcript consistent with the ascertained size of Rli27 ( ∼130 nt ) ( Fig . 4B ) . Real-time qPCR showed that Rli27 expression is induced ( ∼20-fold ) in intracellular bacteria when compared with extracellular bacteria grown in rich medium to logarithmic or stationary phases ( Fig . 4C ) . These findings indicate that Rli27 is a bona fide sRNA that is upregulated by L . monocytogenes inside eukaryotic cells . Rli27 interaction with the lmo0514 5′-UTR extends to several regions , although it shows a major predicted pairing region involving Rli27 nucleotides 1 to 21 ( Fig . 5A , Fig . S3 ) . We used electrophoretic mobility shift assays ( EMSA ) to assess the validity of this prediction . We generated in vitro wild-type versions of Rli27 and 5′-UTR-lmo0514 , together with variants of both RNA molecules bearing mutations in 3 nt ( mut-1 ) or 14 nt ( mut-3 ) important for pairing ( Fig . 5B ) . Incubation of Rli27 and 5′-UTR-lmo0514 wild-type molecules resulted in a duplex with low electrophoretic mobility ( Fig . 5C ) . Conversely , combination of wild-type 5′-UTR-lmo0514 with mutated Rli27 ( either mut-1 or mut-3 variants ) , reduced duplex formation ( Fig . 5C ) . Duplex formation was partially restored by combining mutations in Rli27 with compensatory mutations in 5′-UTR-lmo0514 ( Fig . 5C ) . Specificity of the Rli27-5′-UTR-lmo0514 interaction was confirmed by lack of duplex formation after incubation of the 5′-UTR-lmo0514 wild-type molecule with SbrA , an unrelated sRNA ( Fig . 5D ) . To determine the biological relevance of the 5′-UTR-lmo0514-Rli27 interaction in vivo , we analyzed the specific contribution of Rli27 binding to Lmo0514 protein upregulation in bacteria that infect eukaryotic cells . We generated a Δrli27 strain and a second isogenic mutant , Δrli27C2T , which bears an artificial strong terminator between the remaining rli27 sequences . This mutant was intended to avoid polar effects on the flanking genes lmo0411 and lmo0412 ( Fig . S4 ) ; we also included mutants in these flanking genes , Δlmo0411 and Δlmo0412 [30] . In addition , we designed a qPCR assay specific for the lmo0514 long 5′-UTR for comparison to the lmo0514 coding region . There were no notable differences among strains in the relative levels of the long 5′-UTR region or the lmo0514 ORF ( Fig . 6A ) . In contrast , Lmo0514 protein levels were ∼2 . 5- to 3-fold lower in the cell wall of the two Rli27-lacking mutant strains isolated from the eukaryotic cell ( Fig . 6B ) . This phenotype was complemented by overproduction of wild-type or mut1 versions of Rli27 from a plasmid ( Fig . 6C , D ) . In contrast , when we tested mut3 , the Rli27 mutant bearing 14 nt changes in the major region predicted to interact with the lmo0514 5′-UTR ( Fig . 5A ) , it did not restore Lmo0514 protein levels in intracellular bacteria ( Fig . 6D ) . Wild-type , mut1 and mut3 Rli27 versions were all produced by the plasmid at similar levels ( Fig . 6C ) . These data showed that Rli27 interaction with the lmo0514 long 5′-UTR was essential for induction of the protein in intracellular bacteria , and that elimination of the Rli27-lmo0514 5′-UTR interaction interfered with the Lmo0514 protein increase while levels for the long transcript isoform remained unchanged . Our findings thus supported the need for Rli27 binding for efficient Lmo0514 translation . Control qPCR experiments in extracellular bacteria showed similar lmo0514 transcript levels in this mutant series ( Fig . 6E ) , whereas there were no marked changes in Lmo0514 protein levels ( Fig . 6F ) . These in vivo experiments based on complementation assays with Rli27 variants supported a mechanism that involves Rli27 binding to the 5′-UTR of the long lmo0514 transcript variant that is upregulated by L . monocytogenes inside eukaryotic cells . Such an interaction could promote translation , which would lead to increased Lmo0514 protein levels ( Fig . 7 ) .
Given the unique architecture of the cell envelope in Gram-positive bacterial pathogens , cell wall-associated proteins have essential functions in the interplay of these microorganisms with the host [31] . Despite the recognized importance of these proteins in infection , relatively few studies address the spatio-temporal regulation of the production of these proteins following host colonization . Obtaining this information is particularly challenging for Gram-positive pathogens such as L . monocytogenes or Staphylococcus aureus , which produce a large arsenal of surface proteins with distinct modes of association to the cell wall [31]–[33] . In this study of the Gram-positive bacterium L . monocytogenes , we identify sRNA-mediated regulation that acts on a cell wall-associated protein , Lmo0514 , during the infection process . During the review process of this work , another report showed regulation of L . monocytogenes adhesin LapB by the multicopy sRNA LhrC , although this regulation was not studied in the context of infection [23] . Our data for lmo0514 also distinguish two transcript isoforms with 5′-UTR of distinct length that are expressed differentially when the pathogen transits between non-host and host environments . These findings are consistent with a regulatory role for the sRNA Rli27 , based on its exclusive binding to the lmo0514 long 5′-UTR variant . This long 5′-UTR is generated from a promoter , here termed P2 , which must respond to environmental cues of the eukaryotic intracellular niche . The regulator itself , Rli27 , is also upregulated by L . monocytogenes following entry into host cells . Transcriptional regulators of L . monocytogenes that operate in intracellular bacteria include the alternative sigma factor SigB and the Listeria-specific virulence regulator PrfA . Transcriptomic analyses in sigB and prfA mutants grown in laboratory media did not indicate lmo0514 as a gene regulated by these factors [34]; our results in intracellular bacteria were also negative ( Fig . S5A ) . A yet undetermined regulator might thus be involved in enhancing transcription from the lmo0514 P2 promoter . Neither SigB nor PrfA appear to upregulate Rli27 in intracellular bacteria , as determined by real-time qPCR in sigB and prfA mutants isolated from infected epithelial cells ( Fig . S5B ) . Comparative transcriptomic studies of L . monocytogenes and L . innocua show that ∼87% of the genes are transcribed with 5′-UTR shorter than 100 nt , whereas there is a subgroup of approximately 100 genes with long 5′-UTR ( >100 nt ) [18]; this subgroup includes virulence-related genes and genes with riboswitches [17] . Similar distribution of 5′-UTR length was also described in the related model organism Bacillus subtilis [35] . About 80 genes shared by L . monocytogenes and L . innocua are produced with different-length 5′-UTR [18] , which might indicate differences in post-transcriptional regulation of these transcripts . Our data imply a third group of genes based on distinct transcript isoforms that differ in 5′-UTR length . lmo0514 is a representative example , as it is expressed as two isoforms with 28- and 234-nt 5′-UTR in extra- and intracellular bacteria , respectively . A close parallel is found in a recent work that analyzed sRNA RydC regulation of the Salmonella enterica cfa gene , which encodes a cyclopropane fatty acid synthase [36] . RydC selectively stabilizes the longer of two cfa transcript isoforms , which is associated to the activity of a distal promoter controlled by σA and a proximal promoter modulated by σB [36] . Unlike lmo0514 , both cfa isoforms are expressed by S . enterica growing extracellularly in laboratory media . These observations indicate that transcript isoforms with distinct 5′-UTR target platforms for sRNA-mediated post-transcriptional regulation could profoundly influence protein production . It is noteworthy that long 5′-UTRs are frequently associated with genes involved in pathogenesis [18] , [37] . An interesting feature predicted by the Mfold program is that Rli27 binding to the lmo0514 long 5′-UTR could expose the Shine-Dalgarno site , in contrast to the occluded configuration predicted when this 5′-UTR folds as single molecule ( Fig . S6 , S7 ) . This led us to propose that Rli27 positively regulates Lmo0514 protein levels by altering the long 5′-UTR conformation . This mechanism resembles that of the translational regulation of the rpoS transcript in Escherichia coli [38] . The Shine-Dalgarno site is blocked by a stem-loop in the rpoS 5′-UTR , which is released by base pairing of three distinct Hfq-binding sRNA to the same region . Other paradigmatic cases in L . monocytogenes include the virulence regulator prfA , actA , and the hemolysin ( hly ) genes [25] , [39]–[41] . Our hypothesis for lmo0514 implies that its 234-nt 5′-UTR has considerable secondary structural complexity in the absence of Rli27 . This assumption is consistent with the study by Wurtzel et al . [18] , in which RNA-seq did not define the lmo0514 transcriptional start site , although 2018 such sites were mapped in the L . monocytogenes genome , which account for 88% of all annotated transcriptional units . Our tentative model ( Fig . 7 ) also considers the lmo0514 transcript as ‘low-efficiency’ in terms of translation; there are marked differences in Lmo0514 protein levels in bacteria isolated from epithelial cells ( >200-fold increase ) that are not reflected at the transcript level . The secondary structure prediction for the short ( 28-nt ) 5′-UTR of the extracellular lmo0514 isoform also suggests probable occlusion of the Shine-Dalgarno site ( Fig . S8 ) . Further work is needed to clarify the extent to which such potential structural changes in the 5′-UTR might explain Rli27-mediated regulation . Our EMSA data infer direct Rli27-5′-UTR-lmo0514 interaction , which was also relevant in vivo , based on data obtained with the Rli27-mut3 variant . This variant did not restore the Lmo0514 protein levels produced by intracellular bacteria ( Fig . 6D ) . We did not obtain perfect complementation with compensatory mutations in the predicted interacting regions , which allows other interpretations . For example , the targetRNA program might have predicted an incorrect pairing site , pairing between the two molecules might require additional factors with a precise stoichiometry , or the lmo0514 transcript could undergo alternative post-transcriptional regulation; future work will address these possibilities . We designed in vivo experiments to assess the lmo0514 long 5′-UTR requirement in Lmo0514 protein production in the cell wall of bacteria located inside eukaryotic cells . We tested strains that bear chromosomal mutations in the lmo0514 long 5′-UTR predicted interaction site or that lack most of the 5′-UTR upstream of the P1 promoter −10 and −35 sites ( Fig . S9A , S9B ) . Lmo0514 protein levels dropped markedly inside the eukaryotic cells for some these mutants , especially in that lacking the lmo0514 5′-UTR ( Fig . S9C , S9D ) . Nonetheless , lmo0514 transcript levels were affected in these mutants in both extra- and intracellular conditions ( Fig . S9C , S9D ) . Due to the clear side effect of the mutations on transcription , these findings remained inconclusive . In summary , our results demonstrate that Rli27 is a regulatory sRNA in L . monocytogenes , with an essential role as a positive regulator of the Lmo0514 surface protein during the intracellular infection cycle . We also provide evidence that the Rli27 regulatory role is directed to a transcript isoform that bears the binding site for this sRNA; in addition , we show that this isoform is specifically upregulated by intracellular bacteria . Further research will be necessary to determine how Rli27 might modify the secondary structure of the 5′-UTR after binding , and whether such a role requires additional factors also probably upregulated in intracellular bacteria . Another challenge will be to identify the host-derived signal that triggers transcription from the P2 promoter in intracellular L . monocytogenes and the bacterial transcriptional factor responsible .
To compare the genome region bearing rli27 in L . monocytogenes EGD-e , L . innocua Clip11262 and L . welshimeri serovar 6b str . SLCC5334 , we used the WEBACT program ( http://www . webact . org/WebACT/home ) . Genome sequences were obtained from the Genbank repository ( http://www . ncbi . nlm . nih . gov/genbank/ ) with entry numbers NC_003210 . 1 , NC_003212 . 1 and NC_008555 . 1 for L . monocytogenes EGD-e , L . innocua Clip11262 and L . welshimeri serovar 6b str . SLCC5334 , respectively . The L . monocytogenes strains of serotype 1/2a used here are isogenic to wild-type strain EGD-e [27] ( listed in Table S1 ) . For sRNA overexpression analyses , the rli27 wild-type allele was cloned in the pP1 plasmid [42] using Lmorli27-pP1-F and Lmorli27-pP1-R primers ( Table S2 ) . Relative expression of cloned sRNA was monitored by semi-quantitative RT-PCR using Lmorli27-F and Lmorli27-R primers ( Table S2 ) . L . monocytogenes strains were grown at 37°C in brain heart infusion ( BHI ) broth . For cloning , E . coli strains were grown in Luria Bertani ( LB ) broth at 37°C . When appropriate , media were supplemented with erythromycin ( 1 . 5 µg/ml ) or ampicillin ( 100 µg/ml ) . Two Rli27 variants , Rli27-mut1 and Rli27-mut3 , were constructed by amplification of the rli27 gene with degenerate primers Lmorli27-pP1-F-mut1 and Lmorli27-pP1-F-mut3 ( Table S2 ) and subsequent cloning in pP1 plasmid [42] . The mut1 mutation introduces 3 nt changes and mut3 , 14 nt changes in the major predicted interaction site ( see Fig . 5B ) . To generate the Δrli27 mutant strain , fragments of ∼500-bp DNA flanking rli27 were amplified by PCR using chromosomal DNA of L . monocytogenes strain EGD-e and cloned into the thermo-sensitive suicide integrative vector pMAD [43] with primers Lmorli27-A , Lmorli27B , Lmorli27-C and Lmorli27-D ( Table S2 ) . Genes were deleted by double recombination as described [43] , and deletion was verified by PCR . To generate the Δrli27 mutant , we left 9 nt in the 5′ end and 50 nt in the 3′ end of the rli27 gene , to avoid interference with the lmo0412 terminator ( shared with rli27 ) and the lmo0411 predicted promoter sequence ( Fig . S4 ) . This Δrli27 mutation affected lmo0411 transcript levels slightly . A new deletion mutant was generated ( Δrli27C2T ) , which retains a 5′ extended region of the predicted lmo0411 promoter , thus maintaining 21 nt in the 5′ end and 50 nt in the 3′ end of rli27 ( Fig . S4 ) . In addition , a strong artificial terminator sequence between the remaining rli27 sequences was introduced in the Δrli27C2T mutant ( Fig . S4 ) . All deletions were confirmed by PCR and sequencing , using primers listed in Table S2 . Three types of mutants were constructed with the following chromosomal mutations: i ) changes in 3 nt of the long 5′-UTR-lmo0514 to compensate the mutation in Rli27-mut1 ( see Fig . S3 , S9 ) , ii ) changes in 14 nt of the long 5′-UTR-lmo0514 to compensate the mutation in Rli27-mut3 ( see Fig . S3 , S9 ) , and iii ) a 174-nt deletion upstream of the −10 and −35 sites of the P1 lmo0514 promoter ( Fig . S9 ) . These changes were generated by double recombination as described [43] and when required , using overlapping SOEing PCR . The oligonucleotide primers for these procedures included Δ0514_P2_A , Δ0514_P2_B , Δ0514_P2_C , Δ0514_P2_D , Mut0514pXG_1-overlap , Mut0514pXG_2-overlap , Mut0514pXG_5-overlap and Mut0514pXG_6-overlap ( Table S2 ) . Intracellular bacteria were collected from the human epithelial cell line JEG-3 at 6 h post-infection , as described [12] . For total RNA isolation , epithelial cells cultured in BioDish-XL plates ( 351040 , BD Biosciences ) at ∼80% confluence ( ∼5 . 6×107 cells ) were infected ( 30 min ) with L . monocytogenes grown in BHI medium ( 37°C , overnight ) in static non-shaking conditions . RNA was purified using the TRIzol reagent method [17] . For cell wall protein analysis , intracellular bacteria were obtained from JEG-3 cells cultured on four BioDish-XL plates and infected for 6 h [12] . Subcellular fractions containing protoplasts and peptidoglycan-associated proteins were obtained by mutanolysin treatment of intact bacteria as described [10] , [12] , except that bacterial pellets were incubated for 5 h in lysis buffer ( 10 mM Tris HCl pH 6 . 9 , 10 mM MgCl2 , 0 . 5 M sucrose , 60 µg/ml mutanolysin , 250 µg/ml RNAse-A , 1× protease inhibitor ) . Subcellular fractions containing protoplasts and cell wall-associated proteins of L . monocytogenes grown at 37°C in BHI media were obtained as described [10] . A volume of protoplasts and the cell wall fraction was analyzed by SDS-PAGE followed by Western blot using B . subtilis RecA-specific rabbit polyclonal antibody ( a gift of Dr . JC Alonso , Centro Nacional de Biotecnología-CSIC ) and rabbit poyclonal sera to the L . monocytogenes LPXTG surface proteins Lmo0263 ( InlH ) , Lmo0433 ( InlA ) and Lmo0514 [12] . RecA ( for the protoplast fraction ) and LPXTG proteins InlA and InlH ( for the cell wall fraction ) were used as loading controls . Goat anti-rabbit antibodies conjugated to horseradish peroxidase ( Bio-Rad ) were used as secondary antibodies . Proteins were visualized by chemoluminescence using luciferin-luminol reagents . Total RNA from extracellular bacteria grown to exponential ( OD600 ∼0 . 2 ) and non-shaking stationary phase ( OD600 ∼1 . 0 ) was prepared as described [11] . Oligonucleotides for RT-PCR assays were designed using Primer Express v3 . 0 ( Applied Biosystems ) ( listed in Table S2 ) . RNA was treated with DNase I ( Turbo DNA-free kit , Ambion/Applied Biosystems ) at 37°C for 30 min . RNA integrity was assessed by agarose-TAE electrophoresis . RT-PCR was performed using the one-step RT-PCR kit ( Qiagen ) . Briefly , RT-PCR were carried out with 10 to 70 ng RNA ( depending on the gene analyzed ) in the following conditions: 50°C for 35 min , 95°C for 15 min , followed by 30 cycles ( 16 cycles for the 16S rRNA gene ) of 94°C for 30 s , 55°C for 30 s , and 72°C for 1 min , and then an additional elongation step at 72°C for 10 min . The gene that encodes 16S rRNA was used as a housekeeping gene for all strains in all experimental conditions [44] . For cDNA library construction , we used 1 µg of total DNA-free RNA and the High-Capacity cDNA Archive kit ( Applied Biosystems ) including a random hexamer mix . Reverse transcription was performed at a one-step run of 25°C for 10 min , 37°C for 2 h and 85°C for 5 min . Primers for qPCR were designed using Primer3 [45] ( listed in Table S2 ) . qPCR was performed in a 10 µl final volume with 1 ng of the cDNA library as template , 500 nM of gene-specific primers and the Power SYBR Green PCR Master Mix ( Applied Biosystems ) . Reactions and data analysis were carried out as described [46] . 5′-RACE was performed as described [47] , with minor modifications . To convert 5′triphosphates to monophosphates , 15 µg DNA-free RNA , isolated from L . monocytogenes growing extracellularly at 37°C to stationary phase or from intracellular bacteria collected at 6 h post-infection of epithelial cells , was treated with 25 U tobacco acid pyrophosphatase ( TAP ) ( Epicentre Technologies ) at 37°C for 60 min in a total reaction volume of 50 µl containing 50 mM sodium acetate ( pH 6 . 0 ) , 1 mM EDTA , 0 . 1% β-mercaptoethanol , 0 . 01% ( v/v ) Triton X-100 and 80 U RNAsin ( Promega ) . TAP-negative ( TAP− ) control RNA was processed in the same conditions in the absence of TAP . Following TAP treatment , RNA was phenol/chloroform-extracted and precipitated with sodium acetate and ethanol . Pellets were rinsed with 70% ethanol in DEPC-dH2O , then resuspended in 65 µl DEPC-dH2O; 29 µl of these TAP+ or TAP-treated RNA were combined with 5 . 5 µl 10× buffer , 120 U RNasin , 10% ( v/v ) dimethylsulfoxide , 70 U RNA ligase , 150 µM ATP and 150 ng RNA oligonucleotide adapter , in a total reaction volume of 55 µl . Samples were denatured ( 95°C , 5 min ) and then chilled on ice . RNA adapter ligation was performed ( 17°C , 12 h ) . Following ligation , RNA was phenol/chloroform-extracted and converted to cDNA with a lmo0514-specific primer ( Lmo0514-Pe-3rv ) and the Thermoscript RT System ( Invitrogen ) . Reverse transcription was performed in three cycles ( 55°C , 60°C and 65°C; 20 min each ) , followed by RNAseH treatment ( 37°C , 20 min ) . lmo0514 cDNA ( 2 µl ) was amplified by PCR with oligonucleotides RaceIN and lmo0514-PE-1rv ( 30 cycles of 95°C for 15 s , 55°C for 30 s , and 72°C for 1 min ) , or with oligonucleotides RaceIN and lmo0514-PE-6rv in the same cycling conditions . PCR products were resolved on 2% agarose gels and bands of interest were excised and subcloned into pCR 2 . 1 TOPO-vector ( Invitrogen ) . Plasmids containing inserts were purified using the QIAprep Spin Miniprep Kit ( QIAgen ) and sequenced . To detect the sRNA Rli27 and the 5S rRNA , 15 µg total RNA were electrophoresed in a 6% polyacrylamide 8 M urea gel ( 1 h , 200 V in 1× TBE ) . RNA was transferred to a Hybond membrane ( Amersham ) for 2 . 5 h at 40 V in 0 . 5× TBE at 4°C and RNA was UV-crosslinked to the membrane . Membranes were pre-hybridized with UltraHyb buffer ( Ambion; 65°C , 2 h ) and hybridized with 106 cpm 32P-labeled specific riboprobes ( 65°C , overnight ) . Membranes were washed with 2× SSC , 0 . 5% SDS and 1× SSC , 0 . 1% SDS and exposed to X-ray film . 5S rRNA was used as control [22] . A nonradioactive digoxigenin ( DIG ) -based RNA detection protocol was used for Northern blot analysis of lmo0514 and the 16S rRNA . Total RNA ( 1 µg for lmo0514 or 200 ng for 16S rRNA ) was separated on a 1 . 5% agarose denaturing gel ( 2% formaldehyde , 1× MOPS ) , overnight capillary transferred to a Hybond membrane in 20× SSC , and UV-crosslinked . The membrane was prehybridized ( 68°C , 1 h ) and then hybridized with DIG-labeled lmo0514 and 16S rRNA probes ( 68°C , overnight ) . Immunological detection of RNA was performed ( DIG Northern starter kit; Roche ) and exposed to X-ray film . Gel mobility shift assays were performed with 1 . 48 pmol in vitro-transcribed RNA corresponding to the lmo0514 5′-UTR ( nucleotides −234 to −14 from the lmo0514 ATG codon ) and increasing concentrations of in vitro-transcribed RNAs for Rli27 wild-type , Rli27-mut1 and Rli27-mut3 . These in vitro-transcribed molecules included Rli27 nucleotides 1 to 131 plus an additional 60 nt at the 3′ end , as designed for optimal amplification . We produced lmo0514 5′-UTR variants with compensatory mutations in 3 nt ( mut-1 ) or 14 nt ( mut-3 ) for those generated in Rli27 . Oligonucleotide primers used are listed in Table S2 . We also generated an amplified molecule corresponding to RNA SbrA ( Rli11 ) encompassed nucleotides 1 to 69 of the total of 71 nucleotides . The reaction was carried out in 10 µl of 1× binding buffer ( 20 mM Tris-acetate pH 7 . 6 , 100 mM sodium acetate , 5 mM magnesium acetate , 20 mM EDTA ) ( 37°C , 1 h ) . The binding reactions were mixed with 2 µl loading dye ( 48% glycerol , 0 . 01% orange G ) and loaded on native 4% polyacrylamide gels , followed by electrophoresis in 0 . 5× TBE buffer ( 200 V , 4°C ) . Gels were stained with Gel Red nucleic acid stain ( Biotium ) and photographed under UV transillumination with the GelDoc 2000 system ( Bio-Rad ) . The bioinformatic program TargetRNA ( http://cs . wellesley . edu/~btjaden/TargetRNA2/ ) [28] was used to predict non-coding RNAs that could bind to the lmo0514 long 5′-UTR ( 234 nt from the ATG codon ) . Predictive folding of the lmo0514 long 5′-UTR alone or with sRNA Rli27 was done using Mfold ( http://mfold . rna . albany . edu/ ? q=mfold ) . Statistical significance was analyzed with GraphPad Prism v5 . 0b software ( GraphPad Inc . ) using Student's t-test . A P value≤0 . 05 was considered significant . For densitometry of bands obtained in western blots , we used ImageJ software ( National Institutes of Health of USA [http://imagej . nih . gov/ij/] ) .
|
Listeria monocytogenes has evolved to adapt to numerous environments , including the intracellular niche of eukaryotic cells . Small RNAs ( sRNA ) play important regulatory roles in changing environments , and are thus predicted to modulate L . monocytogenes adaption to the intracellular lifestyle . This study shows how the regulatory activity of an sRNA on a defined target is restricted to bacteria in the intracellular infection phase . This regulation relies on a long ( 234-nucleotide ) 5′-UTR that bears the sRNA-binding site present in a transcript variant that is upregulated by intracellular L . monocytogenes . The concomitant increase in both the target transcript containing the long 5′-UTR and the sRNA , which is postulated to facilitate opening of the Shine-Dalgarno site , culminates in markedly higher protein levels in intracellular bacteria . The limited amounts of both the target and the regulator in extracellular bacteria ensure that production of this bacterial protein is confined mainly to the host rather than the non-host environment .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"genetics",
"molecular",
"biology",
"biology",
"and",
"life",
"sciences",
"microbiology"
] |
2014
|
The Listeria Small RNA Rli27 Regulates a Cell Wall Protein inside Eukaryotic Cells by Targeting a Long 5′-UTR Variant
|
Pattern separation is a central concept in current theories of episodic memory: this computation is thought to support our ability to avoid confusion between similar memories by transforming similar cortical input patterns of neural activity into dissimilar output patterns before their long-term storage in the hippocampus . Because there are many ways one can define patterns of neuronal activity and the similarity between them , pattern separation could in theory be achieved through multiple coding strategies . Using our recently developed assay that evaluates pattern separation in isolated tissue by controlling and recording the input and output spike trains of single hippocampal neurons , we explored neural codes through which pattern separation is performed by systematic testing of different similarity metrics and various time resolutions . We discovered that granule cells , the projection neurons of the dentate gyrus , can exhibit both pattern separation and its opposite computation , pattern convergence , depending on the neural code considered and the statistical structure of the input patterns . Pattern separation is favored when inputs are highly similar , and is achieved through spike time reorganization at short time scales ( < 100 ms ) as well as through variations in firing rate and burstiness at longer time scales . These multiplexed forms of pattern separation are network phenomena , notably controlled by GABAergic inhibition , that involve many celltypes with input-output transformations that participate in pattern separation to different extents and with complementary neural codes: a rate code for dentate fast-spiking interneurons , a burstiness code for hilar mossy cells and a synchrony code at long time scales for CA3 pyramidal cells . Therefore , the isolated hippocampal circuit itself is capable of performing temporal pattern separation using multiplexed coding strategies that might be essential to optimally disambiguate multimodal mnemonic representations .
Understanding what computations hippocampal neurons perform and how they support episodic memory is a longstanding goal in neuroscience . The hippocampus is notably critical for the discrimination of memories that are similar in content [1] . This cognitive function has long been hypothesized to be supported by a neural process called pattern separation , which is generally thought to be implemented in the dentate gyrus ( DG ) [2–4] . Pattern separation is defined as the transformation of a non-simultaneous set of similar input patterns of neuronal activity into less similar output patterns [5] , and is theorized to happen in the hippocampus before encoding a memory trace in order to avoid confusion between similar memories [6] . Despite a long history of research on the subject , mostly in silico [7] , it is still unclear how the activity of single neurons underlies this computation . Generally speaking , single-neuron computations have been investigated experimentally using two broad strategies: 1 ) by determining how the spiking of individual neurons , recorded in vivo , is tuned to different parameters of the behavior or environment [8] or 2 ) by relating the neuronal infra or suprathreshold membrane potential responses , generally recorded in vitro , to their synaptic inputs . The first strategy has led to the discovery that the principal cells of the hippocampus fire preferentially in specific locations of an environment called place fields [9] . Upon sufficient modification of the environment , the place field ( s ) of a given neuron remap by relocating or changing their firing rate ( in extreme cases , disappearing ) [10 , 11] . These different forms of remapping have often been used as a proxy for pattern separation , yielding conflicting interpretations on the role of DG and CA3 [12–14] . However , neurons remapping between two similar environments is not sufficient evidence that those neurons participate in pattern separation because place field remapping is assessed without knowledge of the direct synaptic inputs: it is possible that separated representations are inherited from upstream networks . Thus , despite extensive recordings of large populations of several hippocampal celltypes during behavior [11 , 14] , it remains unknown how each celltype participates in pattern separation . In contrast to the first strategy , the second one explicitly takes into account the synaptic inputs of recorded neurons . Such an approach has been used to study dendritic integration [15] , short [16 , 17] or long-term plasticity [18 , 19] and the relationship between excitatory inputs and the output spiking probability [19–23] , but most of these studies were limited to simple input activity patterns ( i . e . rhythmic series of spikes or bursts of spikes ) and simple output patterns ( i . e . a single spike ) . Some have used more complex sequences of spatially distributed synaptic inputs [24–26] , but the resulting input patterns were still spikes regularly spaced in time . Yet , neuronal spike trains recorded in vivo are not so regular but instead approximately follow a Poisson distribution [27] or a more bursty distribution [28 , 29] , and what neurons integrate is thus a barrage of irregular synaptic inputs [30 , 31] . To infer the single-neuron suprathreshold input-output relationship in more naturalistic regimes , some have thus recorded the spiking response to intracellular injections of current waveforms resulting from either white noise [32] or from the simulated random discharges of a large number of presynaptic neurons [27 , 30] . But this only gives insight on the operations performed by a neuron isolated from its network . To understand the role of the network , rare studies have directly stimulated afferents with naturalistic spike trains to characterize short-term [33–35] or long-term [35 , 36] synaptic dynamics of hippocampal neurons . An even smaller number of studies have aimed at characterizing the suprathreshold input-output function of a neuron in response to external Poisson spike trains [37 , 38] , and such experiments have determined that this function was different when evaluated with simpler input patterns [39] , confirming the necessity to assess neuronal computations with complex , naturalistic input patterns . Surprisingly , because none of the above studies have quantified and systematically varied the similarity between input patterns , none have directly addressed how the input-output transformation in single hippocampal neurons relates to pattern separation . To answer this question , we developed a new paradigm [40] , in mouse brain slices , where afferent axons of the DG are stimulated with complex input spike trains of varying similarity while the output spike trains from a single neuron are recorded . Thus , in contrast to in vivo studies , we considered input and output temporal patterns instead of spatial ones . With this paradigm , we recently demonstrated that the suprathreshold responses of the principal neurons of the DG and CA3 were strongly decorrelated compared to their inputs , whereas DG interneurons exhibited less temporal decorrelation [40] . Importantly , our previous investigation only considered one type of neural code ( binwise synchrony , as measured by the Pearson’s correlation coefficient ) , even though temporal pattern separation could , in principle , be achieved in a variety of ways depending on the features considered relevant in a spike train . For example , if the only feature carrying information in a spike train was its firing rate , a pattern separator would convert a series of input spike trains with similar rates into a series of output spike trains with dissimilar rates . Alternatively , the firing rate could be irrelevant and two output trains with the same number of spikes could be considered separated if their spike times are desynchronized . Pattern separation is thus a group of potential computations , each computation corresponding to a specific neural code , or , in other words , to a specific definition of "similarity" . By testing a wide range of similarity metrics and time scales , the present work aims to determine through which neural codes temporal pattern separation is performed in the hippocampus .
In Madar et al . ( 2019 ) [40] , we demonstrated for the first time that single GCs exhibit high levels of pattern decorrelation thanks to a novel pattern separation assay in acute brain slices . Our general paradigm has three steps ( Fig 1 , Methods—Experiments ) : 1 ) ensembles of stimulus patterns ( simulating trains of action potentials ) are generated , with known degrees of similarity to each other . These sets of input spike trains are then fed into the DG by stimulating the lateral perforant path . 2 ) The simultaneous response of a single neuron is recorded in whole-cell current-clamp . 3 ) The similarity between the output spike trains is compared to the similarity between the input patterns , revealing the degree of separation or convergence . We used R as a similarity measure between spike trains because it is easy to implement and is commonly used to quantify the similarity between neural activity patterns , both in computational [41] and experimental studies [12 , 13 , 42 , 43] . However , the original Hebb-Marr framework theorized pattern separation as the orthogonalization of the input patterns [5 , 44] . As a result , the terms "decorrelation" ( corresponding to output patterns , viewed as variables , with a lower Pearson's correlation coefficient than their inputs ) and "orthogonalization" ( corresponding to output patterns , viewed as vectors , that are closer to a right angle than inputs ) are often conflated in the literature , even though they are not mathematically equivalent and have a nonlinear relationship to each other ( Figs 2A–2C , S1 . See Methods –Similarity metrics and S1 Appendix- 1 ) . For instance , pairs of spike trains can be uncorrelated ( R = 0 ) without being orthogonal , or can be orthogonal without being uncorrelated ( Fig 2A–2C and S1 ) . To determine whether output spike trains of GCs are truly orthogonalized , we explicitly considered spike trains as vectors of spike-counts and computed the normalized dot product ( NDP , i . e . the cosine of the angle between two vectors ) between pairs of spike trains to assess their similarity ( Fig 2A and 2C . See Methods –Similarity metrics and S1 Appendix—1 ) . For every recording set , NDPoutput was lower than NDPinput , indicating that the angle between output spike trains was closer to a right angle ( i . e . , closer to orthogonal ) than their inputs ( Fig 2D and 2E ) . Vectors can differ by their angle , but also by their norm . In other words , even if neurons fire in the same time bins ( relative to the start of each sweep ) , the number of spikes per bin can be different , as quantified by the ratio between their norms ( scaling factor , SF ) ( Fig 2A and 2C . See Methods –Similarity metrics and S1 Appendix—1 ) . Our results show that for input patterns highly similar in terms of SF , SFoutput is slightly but significantly lower than SFinput ( Fig 2F ) . This indicates that variations in the binwise firing rate of single GCs in response to similar inputs is a potential , but weak , mechanism of pattern separation at the 10 ms time scale . As a whole , these results confirm that input spike trains are decorrelated in the DG at the level of single GCs , and demonstrate that , at the 10 ms time scale , GCs exhibit temporal pattern separation mediated by high levels of orthogonalization and weak levels of scaling . Trains of action potentials can be described in many different ways: their overall firing rate , the timing of their spikes or the number of bursts of spikes , among other statistics ( from now on referred to as spike train features ) . Measuring the similarity between two spike trains is not a trivial problem , if only because it is unclear what spike train features are relevant to the brain and what counts as different or similar . In principle , pattern separation could be achieved by the variation of any spike train feature ( e . g . if the relevant feature is the total number of spikes in two second bins , pattern separation will be achieved if this number varies more in the set of output spike trains than in the input set ) . In other words , many different forms of pattern separation could be performed depending on the neural code considered . In our analyses , the neural code ( i . e . the set of assumptions on which spike train features are considered relevant and what definition of similarity is used ) is determined by the choice of similarity metric and time resolution ( see S1 Appendix– 1 ) . In order to better characterize which neural code ( s ) are used to perform temporal pattern separation in the hippocampus , we thus tested multiple metrics and time scales . This approach also allowed us to ask whether several coding strategies could be multiplexed ( i . e . simultaneously relevant over different time scales ) to achieve pattern separation . The three measures of similarity we have used above ( R , NDP and SF ) all carry different assumptions about the neural code ( Table 1 ) : R and NDP are mostly , but differently , sensitive to the binwise synchrony whereas SF evaluates variations in spike number ( Fig 2A–2C and see S1 Appendix—1 ) . On the other hand , these metrics resemble each other in that they require binning spike trains in time windows of a prespecified duration ( τw ) . The time scales that are meaningful for the brain are uncertain so we assessed the separation of spike trains for different τw . Indeed , different durations of τw assume a different window to read the information contained in a spike train . Our analysis shows that pattern separation , measured through R or NDP , is more pronounced at short time scales ( e . g . 5 ms ) than at longer ones ( e . g . 100 ms ) ( Fig 3A and 3B ) . In contrast , in the case of high input similarity , pattern separation through scaling is rather weak at short time scales but gets stronger at longer ones ( Fig 3C ) . This demonstrates that multiplexed coding allows temporal pattern separation to be carried out by the DG for a large range of time scales . In addition , because a long stream of research suggests that spike trains can carry information directly through the timing of individual spikes [45–47] , we also assessed the similarity between spike trains using SPIKE , a binless metric purely based on spike times [48] , thus assuming a neural code completely different from the binned metrics above ( Table 1 , Methods–Similarity metrics and S1 Appendix ) . Our results show that input spike trains with similar spike times relative to their sweep start ( defined here as spike trains with a high degree of synchrony , see Methods –Similarity metrics ) , are transformed into significantly less synchronous outputs , thereby demonstrating that temporal pattern separation in single GCs can occur through spike timing modifications ( Fig 3D ) . This is consistent with the high levels of decorrelation and orthogonalization at very short time scales ( Fig 3A and 3B ) and confirms that temporal pattern separation can be achieved through multiple coding strategies . Results in Figs 2F and 3C show that the similarity of GCs output spike trains in terms of SF is almost flat , suggesting it might be independent of the input similarity , hence leading to significant pattern convergence in certain conditions of low input similarity and short time scales . However , those experiments were based on responses to ~10 Hz Poisson input spike trains that did not differ much in terms of SF . To further characterize single GC computations , we designed two new input sets spanning a wider range of SF values ( Fig 4A and 4B ) . From its definition , it was apparent that , unlike R and NDP , SF is very sensitive to both differences of overall firing rate and differences in burstiness between spike trains ( S2 Fig ) . Input set A was thus made of highly correlated Poisson spike trains with different firing rates ( 7–31 . 5 Hz ) , and input set B was made of uncorrelated spike trains with constant firing rate but different levels of burstiness ( see Methods - Experiments ) . By recording the spiking output of single GCs in response to these input sets we could thus test how GCs transform input patterns with vastly different statistical structures , and whether they can still perform pattern separation in those conditions . Our results indicate that for both input sets GCs can perform weak levels of pattern separation via scaling for highly similar inputs ( Fig 4C ) , consistent with our experiments using 10 Hz Poisson trains ( see Figs 2F and 3C ) . However , although not independent of input similarity , the similarity of GCs output spike trains ( in terms of SF ) is comprised in a relatively narrow band , yielding high levels of pattern convergence for relatively dissimilar input spike trains ( Fig 4C ) . In other words , when assuming a neural code based on the norm of spike count vectors , GCs outputs are maintained relatively similar to each other , even when their corresponding input patterns are very different . This is true from the millisecond to the second time scale , but pattern convergence decreases and separation increases with longer time scales ( Fig 4C ) , consistent with previous experiments ( see Fig 3C ) . Although SF makes mathematical sense when measuring the similarity between vectors , it is difficult to interpret in terms of basic spike train features , in part because it is sensitive to both firing rate and burstiness ( see S1 Appendix– 1 for details ) . We thus aimed to gain a more intuitive view of how those spike train features vary to result in pattern separation or convergence . For that , we calculated the mean firing rate of each spike train and designed two new indicators providing complementary information on the burstiness of individual spike trains: Compactness and Occupancy ( Fig 5A ) . Briefly , Compactness is anticorrelated to the number of time bins occupied by at least one spike ( the less occupied bins , the more compact a spike train is ) , whereas Occupancy is the average number of spikes per occupied bin ( Methods–Firing rate and burstiness codes and S1 Appendix—2 ) . First , we asked how GCs transform their inputs in terms of firing rate and burstiness . For all time scales , GCs have generally equal or higher Compactness than their inputs but maintain a constant low range of Occupancy ( Fig 5B left ) . This combination suggests that , using S1 Table , GC responses are mostly a sparser version of their inputs , with a decrease in the number of spikes per burst when inputs are bursty . Accordingly , the overall firing rate of GC output spike trains is generally lower than their inputs ( Fig 5B right ) , which is consistent with the known sparsity of GC activity in vivo and fits the view of the DG acting as a filter of cortical activity [49] . We next asked whether temporal pattern separation is performed through variation of the firing rate between GCs output spike trains , or through variations of Compactness or Occupancy , or a combination of those three . Our results indicate that pattern separation , but also pattern convergence , can be exhibited for all three types of codes , depending on certain conditions ( Fig 6 ) . First , separation is favored at longer time scales ( for burstiness codes ) and high input similarity ( for burstiness and rate codes ) ( Fig 6A ) . Second , the direction of the computation ( separation or convergence ) strongly depends on the statistical structure of the input patterns ( Fig 6B ) : 1 ) In response to Poisson inputs with very similar firing rates ( P10Hz ) , GCs exhibit low but significant levels of pattern separation through small variations of Compactness , Occupancy and firing rate , which become more relevant with larger time scales . Pattern separation measured through scaling ( see Figs 2F and 3C ) was thus due to the variability in all three spike train features combined . 2 ) In response to Poisson inputs with varying firing rates ( PΔFR , input set A ) GCs switch , as the time scale is increased , from pattern convergence to pattern separation through Compactness variations , while increasing their levels of convergence in terms of Occupancy . Accordingly , their output spike trains have more similar firing rates than their inputs ( this is because GCs maintain a narrow range of firing rates regardless of the rate of their inputs , as shown in Fig 5B ) . 3 ) In response to inputs with a constant firing rate but varying levels of burstiness ( B 10 . 5 Hz , input set B ) , GCs show low but significant pattern convergence in terms of Compactness and Occupancy . Pattern convergence in terms of Occupancy progressively decreases as the time scale increases . This translates into significant pattern separation through variations of the firing rate ( note that , because spike trains were 2 s long , pattern separation via FR variations is equivalent to pattern separation via Occupancy variations in a single 2 s time scale ) . Interestingly , this analysis demonstrates that GCs can exhibit both separation and convergence simultaneously through different neural codes: for instance , at the 250 ms time scale , the Compactness of GC output spike trains vary more than the Compactness in input set A , whereas GC output spike train Occupancy varies less ( Fig 6B , Table 2 ) . GCs are known to receive strong feedforward and feedback inhibition from a variety of GABAergic interneurons [50 , 51] that impact GCs spiking output [21 , 52 , 53] . Furthermore , theoretical work has long suggested a role for interneuron interactions with GCs in mediating pattern separation [4 , 54 , 55] , but experimental evidence is lacking . To test the influence of fast inhibitory transmission on the complex computations of GCs characterized above , we recorded GCs responses to 10 Hz Poisson trains before and after application of a nonsaturating concentration of Gabazine , a GABAA receptor antagonist ( Fig 7A , see Methods - Experiments ) . Under these conditions , IPSC amplitude was reduced by ~30% , which led to a slight but visible increase in firing rate , in part due to a higher propensity to fire short bursts of action potentials riding a single EPSP ( Fig 7A and 7B ) . This pharmacological manipulation led to lower levels of temporal pattern separation as measured by R , NDP , but also SF ( Fig 7C ) . This was true at all time scales ( Fig 7D ) , with the importance of inhibition for pattern separation even increasing at longer time scales . Interestingly , the impact on pattern separation at larger time scales was not only for SF , whose relevance increases with larger τw ( see Fig 3C ) but also for R and NDP , whose relevance normally decreases with larger τw ( see Fig 3A and 3B ) . The average Routput was well correlated to the average firing rate of a recording set ( linear regression: R2 = 94% , p < 0 . 0001 , n = 14 data points from 7 GCs ) , suggesting that the inhibitory control of sparsity in GC firing helps temporal pattern separation as measured with R . Interestingly , in a previous report we did not detect such a strong relationship between R and the firing rate levels of GCs [40] . Moreover , R is mathematically independent of proportional increases of pairwise firing rate levels and not very sensitive to non-proportional ones ( see S1 Appendix—1 and S1C Fig ) . In the present experiment , the correlation between R and the firing rate is thus likely to arise from physiological reasons that conjointly affect both the sparsity and the binwise synchrony of output spike trains rather than from the mathematical assumptions behind R . In contrast to R and NDP , the mathematical definition of SF makes it sensitive to pairwise firing rate levels ( i . e . SF systematically increases for pairs of spike trains with higher firing rates . See S1 Appendix—1 and S1C Fig ) . As seen before , SF is also designed to be sensitive to differences in firing rate and burstiness of individual spike trains ( e . g . Figs 5 and S2 ) . Therefore , the effect of gabazine on decreasing pattern separation as measured by SF ( Fig 7C and 7D ) could be due: 1 ) to a global increase in firing rates in all the output spike trains , 2 ) to variations of firing rate or burstiness between output spike trains , or 3 ) to a combination of all of the above . The direct assessment of pattern separation in terms of firing rate or burstiness indicates that inhibition actually does not affect temporal pattern separation through rate or Compactness codes and favors pattern convergence through Occupancy codes ( Fig 7E ) . We can thus conclude that the effect of inhibition on pattern separation via scaling ( SF ) is mostly due to global suppression of the firing rate that is relieved under gabazine . Overall , we provide here the first experimental evidence that fast inhibitory transmission in the DG controls levels of temporal pattern separation and convergence through different multiplexed neural codes . GCs are embedded in a network of multiple celltypes , all participating in generating the output of the DG . For instance , computational models of the DG have suggested that different populations of inhibitory and excitatory interneurons could modulate certain forms of pattern separation [43 , 54] . To develop a comprehensive understanding of the computations performed by the DG , and hippocampal networks in general , it is important to characterize the input-output transformation of a diverse set of celltypes . We recorded from dentate fast-spiking GABAergic interneurons ( FSs ) , glutamatergic hilar mossy cells ( HMCs ) and CA3 pyramidal cells ( PCs ) in response to 10 Hz ( for GC , FS , HMC ) or 30 Hz ( for PC and GC ) Poisson input trains with varying levels of correlation ( Fig 8A and Methods–Experiments ) . Previously , we found that temporal pattern separation , measured using R , was not specific to GCs but that GCs exhibit the highest levels of decorrelation among tested dentate neurons , and that CA3 PCs express even more temporal decorrelation than GCs [40] . Here we ask whether this is still the case when considering other neural codes based on different similarity metrics . A pairwise similarity analysis and a comparison of the pattern separation function ( the linear regression model describing Soutput as a function of Sinput , with S representing a given similarity metric ) across celltypes shows that pattern orthogonalization ( NDP ) and scaling ( SF ) levels significantly differ at most time scales ( Fig 8B–8D , Table 3 ) . For NDP , we notice a progression from FSs to HMCs to GCs to CA3 PCs , with GCs performing the highest levels of separation among DG neurons , and CA3 PCs surpassing GCs , especially with their ability to orthogonalize inputs already relatively dissimilar ( Fig 8B ) . The orthogonalization function of GCs , HMCs and CA3 PCs diverge as the time scale is increased ( Fig 8C ) . For SF , scaling functions do not depend much on the input similarity , except for CA3 PCs . Mostly based on the scaling functions intercept , we notice two groups ( Fig 8D ) : FSs and GCs exhibit low levels of pattern separation through scaling , in contrast to HMCs and CA3 PCs ( and their GC controls ) that produce spike trains with lower similarities in terms of SF . Note that among celltypes tested with 10 Hz Poisson trains , HMCs are the best at pattern separation through scaling , not GCs . In contrast to GCs , HMCs and FSs tend to fire bursts of spikes ( although the characteristics of bursts differ between the two celltypes ) , which affects , at least partially , their levels of pattern decorrelation [40] . An analysis of the distribution of inter-spike-intervals in all recorded celltypes confirmed that HMCs and FSs are bursty ( as they diverge from a Poisson process ) but not to the same extent ( Fig 9A , and see Methods - Firing rate and burstiness codes ) . It also revealed that , although neither CA3 PCs nor their GC controls ( recorded under gabazine and 30 Hz inputs ) displayed bursts elicited by a single input pulse [40] , they can be considered bursty in the sense that their output spikes tend to be clustered together in time in a nonrandom fashion ( Fig 9A ) . We thus asked whether these celltypes with diverse bursty behaviors could perform pattern separation through Compactness , Occupancy or firing rate codes . Our analysis shows indeed that these features vary across the output spike trains of hippocampal neurons , but in different ways for each celltype ( Fig 9B , Table 4 ) : for example , at long time scales FSs and HMCs achieve pattern separation mostly through variations of Occupancy and firing rate , whereas CA3 PCs achieve pattern separation mostly through variations of Compactness . Overall , we demonstrate here that not all hippocampal celltypes exhibit the same computations , but that they all can perform temporal pattern separation to different degrees through different neural codes and over different time scales . This suggests that different types of neurons can serve different , complementary , computations that allow the isolated hippocampal network to perform temporal pattern separation using multiplexed coding strategies .
Overall , our study highlights that measuring neuronal computations implies a certain vision of the neural code , and that brain networks or single neurons perform many different computations depending on the assumed coding scheme . The "neural code" is a pervasive but vague concept that refers to the set of features of neural activity that convey information , for example to represent our experiences and memories . The first difficulty in breaking this code comes from the infinitely large number of potential features and combinations of features that could be relevant in the brain . Since the early work of Adrian [60] , the most common view is that neurons transmit information about their stimulus through their firing rate , i . e . the number of output spikes averaged over a certain time window . This rate code is often contrasted to time codes , but this is a false dichotomy [45]: there is an infinite number of possible rate codes , as there is an infinite number of time windows over which the number of spikes could be integrated ( are spikes counted by reader neurons over milliseconds , seconds or minutes ? ) . In addition to spike counts , information can also be carried by the timing of single spikes [45 , 46 , 61] , bursts [62 , 63] or even more complex spike patterns [61 , 64–66] . Even when focusing on time or burstiness codes , many subfeatures could be considered words in the elusive language of the brain: for example , the information could be encoded in the specific spike times [67] or the interspike intervals [64 , 68] , the relative or exact latency or synchrony with regard to other spike trains [67] or the relative timing of individual spikes with regard to a reference network oscillation [69] , the timing or frequency of bursts [62 , 63] , the spike times within a burst or the spike counts per burst [63 , 70] . Pattern separation has long been thought to cover several potential computations depending on the neural code considered [5] . Indeed , past investigations did not always define brain activity patterns the same way , nor did they choose the same similarity metrics and time resolution , thus assuming different forms of pattern separation . When the hypothesis of pattern separation was first formulated , activity patterns were defined as binary vectors representing populations of ON or OFF neurons , without considering a time dimension [4 , 6 , 44] . As a result , modeling work that followed often excluded the dynamics of neural activity , and used the number of common active neurons between two population patterns as their measure of similarity [43 , 54 , 71] . In addition to this purely spatial population code , others have defined activity patterns as maps of firing rates averaged over minutes , or as population matrices of those maps [12 , 14] , measuring pattern separation using R [72] or NDP [73] as their similarity metric . Only two modeling studies have investigated pattern separation by directly considering spike trains at a time scale less than a minute: one considered an input pattern as a static population of ON/OFF neurons , an output pattern as a population vector of firing rates averaged over 500 ms and the similarity between two output patterns was based on the sum of firing rate ratios [71]; the second considered an input pattern as an ensemble of 30 ms spike trains ( 0 or 1 spike per input channel , i . e . akin to a population of ON/OFF neurons but with the time of an input spike carrying some information due to the possibility of temporal summation ) and an output pattern as an ensemble of 200 ms spike trains , using R with a τw of 20 ms to measure the similarity between spatiotemporal patterns [41] . Past computational studies suggest , together , that the DG could perform pattern separation through different codes , but our investigation is the first to systematically explore and compare pattern separation levels based on diverse coding strategies . It is also the first focusing on time codes . Thanks to our slice-based pattern separation assay , we could control input patterns while recording the output of the DG , thus allowing us to experimentally test pattern separation in the DG for the first time [5 , 40] . We considered activity patterns ( for both input and output ) to be two-second long spike trains , and used multiple definitions of spike train similarity at time scales going from submillisecond to second resolution . We first measured spike count correlations with R and NDP , two common metrics with different sets of assumptions on the neural code ( e . g . NDP does not consider common periods of silence as correlated; see S1 Appendix ) , showing that the DG performs decorrelation and orthogonalization of its outputs at subsecond time scales at the level of single GCs ( Fig 3A and 3B ) . Because this phenomenon is stronger at millisecond time scales , and that R and NDP are mostly sensitive to binwise synchrony , we measured spike train similarity with SPIKE , assuming a synchrony code purely based on spike times instead of spike counts [48] . This confirmed that GC outputs are separated through "desynchronization" of their spike times from sweep to sweep ( Fig 3D ) . Moreover , we designed a new simple similarity metric , SF ( which complements NDP in the description of the similarity between two spike count vectors ) , as well as two complementary measures of spike train "burstiness" ( Compactness and Occupancy ) . These revealed that , in addition to desynchronization strategies , GCs can exhibit pattern separation through firing rate or burstiness codes , and that depending on the time scale , the neural code considered and the input statistical structure , different levels of pattern separation or convergence are performed ( Fig 6 ) . In conclusion , our work demonstrates that single neuron pattern separation can be achieved in isolated hippocampal tissue through multiplexed coding , which is a way of simultaneously representing several features of a stimulus or , in our case , simultaneously performing different computations , through different temporal scales and spike train features [59 , 63 , 74] . Our work illustrates the importance of considering multiple neural codes when studying neuronal computations for three reasons: 1 ) The fact that conceptually different metrics lead to converging results bolsters our conclusion that temporal pattern separation occurs within the DG and CA3 at the level of single neurons; 2 ) a single neuron can exhibit either pattern convergence or separation depending on the coding scheme; 3 ) a given neural code can be more relevant for pattern separation/convergence in certain celltypes than others . Our finding that DG can perform pattern convergence in some conditions is particularly intriguing . If pattern separation is conceptualized to support mnemonic discrimination , then pattern convergence would support generalization , another elusive mnemonic phenomenon [75] . In the future , experiments correlating mnemonic discrimination and generalization with the levels of pattern separation or convergence through various potential codes will help pinpoint which computations are actually used in the brain to support episodic memory . Our work , in showing that spiking was difficult to elicit in GCs and that their responses were sparser than their inputs ( Fig 5 ) is in line with past research [20 , 23] . It also resonates with the long-standing idea that the DG acts as the gate of the hippocampus , filtering out incoming cortical activity before it reaches CA3 to prevent overexcitation [49] . Figuring out the computations of the DG is equivalent to reverse-engineering the transfer function defining the DG filter . Some have suggested that the DG is a low-pass [22 , 76] or a ~10 Hz band-pass filter [21] . Our results confirm that the output frequency of GCs is constrained in a narrow band ( Fig 5B ) and go one step further by providing the first demonstration that the filtering properties of the DG allow it to perform different computations including pattern separation in response to complex naturalistic inputs ( Fig 6 ) . The question is then: how are these filtering properties implemented ? The physiological mechanisms underlying DG computations in general are understudied and remain a mystery . In Madar et al . ( 2019 ) [40] , we provided a proof-of-principle that the short-term dynamics of synaptic transmission at the perforant path-GC synapses could explain the levels of temporal pattern decorrelation in the DG . But GCs also receive feedforward and feedback inhibition and excitation from multiple interneurons that also interact between each other [51 , 77 , 78] . Indeed , our finding that partial block of GABAA-mediated neurotransmission impairs temporal pattern separation in the DG ( Fig 7 ) strongly suggests that the computations we investigated are network phenomena . Interactions between multiple celltypes are likely involved , and it is thus critical to characterize the suprathreshold input-output transformation of each celltype to parse out their computational role . In the present work , we shed light on the computations of dentate FSs ( also known as basket cells providing fast feedforward and feedback perisomatic inhibition to GCs [79] ) , HMCs ( glutamatergic interneurons providing feedback monosynaptic excitation and disynaptic inhibition to GCs mostly outside of the lamellar plane [77] ) , and CA3 PCs , the output neurons of the hippocampal subnetwork directly downstream of the DG in the trisynaptic circuit , but which can also provide indirect feedback to the DG via collaterals targeting hilar neurons [77] . We discovered that , in contrast to GCs ( Figs 8 and 9 , and see [40] ) : 1 ) FSs generally exhibit low levels of pattern separation through binwise desynchronization ( R , NDP ) or scaling ( SF , likely due to high firing rates ) , but high levels of separation through simple variation of local firing rates from sweep to sweep , especially at the 100 ms time scale and higher; 2 ) HMCs exhibit intermediate to low levels of binwise desynchronization up to the 500 ms time scale . However , contrarily to GCs and FSs , they are able to achieve high levels of pattern separation through scaling , thanks to a variable bursting response and low overall firing rates; 3 ) CA3 PCs exhibit high levels of pattern separation through all tested neural codes , but significantly more than their GC controls through binwise desynchronization only , especially at long time scales , suggesting that CA3 might complement and amplify the separation inherited from the DG , possibly to make pattern completion more efficient [40] . In the future , computational modelling combining those results with knowledge of the synaptic dynamics between these populations of neurons will help dissect the role played by each element of the network in hippocampal computations . Our study especially calls for more work on the computational role of inhibitory synaptic dynamics as it reveals the first experimental evidence that fast GABAergic transmission favors pattern separation through multiple neural codes ( binwise desynchronization and scaling ) and impairs it through another code ( Occupancy variations at short time scales ) ( Fig 7 ) . Past research has long suspected the importance of inhibition to hippocampal computations . Computational models of the DG implementing perisomatic feedforward inhibition ( provided by FSs ) in the form of a "k-winners take all" rule controlling the number of GCs activated by a given input pattern have stressed the importance of such inhibition on pattern separation through a population code [4 , 54 , 55] . The seminal model of Myers and Scharfman ( 2009 ) [54] also suggested that certain hilar interneurons , providing feedforward and feedback inhibition on the distal dendrites of GCs , could impair population pattern separation through their influence on other GABAergic interneurons ( see also [43] ) . Experimentally , a recent behavioral study on mice showed that normal levels of tonic inhibition in GCs are critical for mnemonic discrimination [80] . Our results now demonstrate that inhibition is also critical for computations that could support this episodic memory function . Future experiments will need to clarify the exact role of different interneuron populations as well as different types of inhibition ( e . g . feedforward vs feedback or somatic vs dendritic ) . Such work will be especially critical to further understand CA3 computations and how they differ from those of DG , because our findings on CA3 are based on recordings with partially reduced inhibition . On one hand , the enhancement of pattern separation we observed in CA3 could be due to abnormal pharmacological conditions , on the other hand , the necessity to lower inhibition to get CA3 PCs to spike could instead indicate that CA3 must be disinhibited during episodic memory formation .
All experiments were performed in accordance with the National Institute of Health guidelines outlined in the National Research Council Guide for the Care and Use of Laboratory Animal ( 2011 ) and regularly monitored and approved by the University of Wisconsin Institutional Animal Care and Use Committee . Briefly , we performed experiments on C57Bl6 male mice ( Harlan/Envigo ) following the general paradigm described in Madar et al . ( 2019 ) [40] . Horizontal acute slices of a mouse brain hemisphere containing the hippocampus were prepared in cutting solution ( CS; two version were used . CS#1/CS#2 , in mM: 83/80 NaCl , 26/24 NaHCO3 , 2 . 5/2 . 5 KCl , 1/1 . 25 NaH2PO4 , 0 . 5/0 . 5 CaCl2 , 3 . 3/4 MgCl2 , 22/25 D-Glucose , 72/75 Sucrose , 0/1 Na-L-Ascorbate , 0/3 Na-Pyruvate , bubbled with 95% O2 and 5% CO2 ) and recorded at 33–35°C in regular artificial cerebrospinal fluid ( aCSF , in mM: 125 NaCl , 25 NaHCO3 , 2 . 5 KCl , 1 . 25 NaH2PO4 , 2 CaCl2 , 1 MgCl2 , and 25 D-Glucose ) . Two second input trains of varying similarity were delivered to neural afferents of DG or CA3 neurons via bipolar electrical stimulation through a glass theta pipette , while recording the membrane potential of a single responsive patched neuron in whole-cell current-clamp mode . For the intracellular solution ( IS ) , we used two different recipes that yielded similar electrophysiological behaviors and results ( IS#1/IS#2 in mM: 140/135 K-gluconate , 0/5 KCl , 10/0 . 1 EGTA , 10/10 HEPES , 20/20 Na-phosphocreatine , 2/2 Mg2ATP , 0 . 3/0 . 3 NaGTP , 0/0 . 25 CaCl2 , adjusted to pH 7 . 3 and 310 mOsm with KOH and H2O ) . The location of the stimulation ( >100um away from dendrites of the recorded neuron ) and its intensity ( 0 . 1–1 mA ) was determined to yield ~50% probability of spiking after an input pulse ( range: ~20–80% ) . A stimulation protocol consisted of five or ten trains ( i . e . a given input set ) , interleaved ( with 5 s of pause between each train ) and repeated ten or five times , respectively , such that fifty output traces were recorded ( Fig 1C ) . The amount of pattern separation or convergence effectuated for a given recording was assessed by pairwise similarity analysis ( defined below ) , displayed as what we refer to as a "pattern separation graph": the pairwise input similarity across all pairs of input trains ( excluding self-comparisons ) was compared to the average similarity of all pairs of output spike trains coming from the corresponding pairs of input trains ( excluding comparisons between output spike trains coming from the same input train ) ( Fig 1D and 1E ) . Figs 1–3 ( GCs ) and Figs 8 and 9 ( all celltypes ) are based on recordings from GCs , fast-spiking interneurons ( FSs ) , hilar mossy cells ( HMCs ) and CA3 pyramidal cells ( CA3 PCs ) reanalyzed from Madar et al . ( 2019 ) [40] . They were performed on slices ( CS#1 for GCs , CS#2 for other celltypes; IS#1 for all ) from young mice ( p15-25 ) , using 11 input sets consisting of five Poisson spike trains with a mean rate of 10 Hz ( GCs , FSs , HMCs ) or 30 Hz ( CA3 PCs and their GC controls ) . GCs ( 10 Hz ) , FSs and HMCs were recorded in regular aCSF . Because it was very difficult to promote firing of CA3 PCs in aCSF , CA3 PCs and their GCs control were recorded in aCSF with 100 nM gabazine ( gzn , SR-95531 ) to decrease inhibitory postsynaptic current ( IPSC ) amplitude by ~30% and allow CA3 PCs to occasionally escape inhibition and fire action potentials . Note that the GCs recorded in gabazine as controls for the CA3 PC experiments could not be directly compared to GC recordings in regular aCSF because input sets with different mean rates were used . To determine the role of inhibition on pattern separation in GCs , we performed additional recordings of single GCs in slices from young mice ( p22-32 , 6 animals ) ( Fig 7 ) . After cutting ( CS#2 ) , slices were moved to a storing chamber containing 100% CS , at 37°C for 30 minutes after dissection . As in all other experiments , the storing chamber was then placed at room temperature and slices left there for at least 30 min . Slices were then transferred to the recording chamber and perfused ( 5 ml/min ) with oxygenated regular aCSF at 33–35°C . Patch-clamp recordings of GCs were performed using IS#1 . An experiment consisted of recording the same GC under the same input set of five 10 Hz Poisson trains ( average Rinput = 0 . 76 at 10 ms , shown in Fig 1B ) before and after addition of 100 nM gabazine to the flow of regular aCSF . We waited at least five minutes after the solution switch , to allow for equilibrium to be reached . The resting membrane potential ( held around -70 mV ) , the location of the stimulation electrode , the stimulus current intensity and all other parameters were unchanged between the two recordings ( e . g . under gabazine , the spiking probability was not re-targeted to ~50% , unlike the gabazine experiments using 30 Hz inputs mentioned above ) . The intrinsic properties of recorded GCs ( n = 7 ) were ( mean ± SEM ) : 1 ) Before gzn: resting membrane potential Vrest = -76 . 5 ± 2 . 5 mV , input resistance Ri = 194 ± 30 MΩ and membrane capacitance Cm = 20 ± 1 . 4 pF; 2 ) After gzn: Vrest = -74 . 4 ± 2 mV , Ri = 184 ± 29 MΩ and Cm = 17 ± 1 . 6 pF . Ri was not significantly affected and Vrest and Cm were only slightly decreased after treatment ( Wilcoxon signed rank test: Vrest p = 0 . 09; Ri p = 0 . 44; Cm p = 0 . 03 ) . The slight drift of Cm was likely an artifact due to the small increase in series resistance occurring over time ( before: 6 ± 0 . 8 MΩ , after: 7 . 3 ± 1 . 4 MΩ; p = 0 . 06 ) . To explore the impact of a variety of input pattern statistical structures on single neuron computations , we designed two new input sets comprised of 10 different two seconds spike trains ( Figs 4–6 ) . Trains in "input set A" follow a Poisson ( P ) distribution and have an average Pearson's correlation of 75% at the 10 ms time scale , but each train has a different mean firing rate ( PΔFR , Fig 4A ) . Trains in "input set B" do not follow a Poisson distribution: they all have 21 spikes , making them all 10 . 5 Hz overall , but are more or less bursty ( B ) ( i . e . for each spike train the same number of spikes was distributed in a different number of time bins . Occupied time bins were selected randomly–following a uniform distribution–among all the possible time bins ) ( B 10 . 5 Hz , Fig 4B ) . For these experiments , we used slices from two adult mice ( p145 and p153 ) injected with saline at p46 as controls for a separate set of experiments in animals with epilepsy ( not reported here ) . Slices were prepared as described above ( CS#2 ) and patch-clamp recordings were performed in regular aCSF using IS#2 . The intrinsic properties of recorded GCs were: 1 ) Input set A ( n = 5 ) : Vrest = -80 . 4 ± 2 . 9 mV , Ri = 155 ± 19 MΩ and Cm = 23 ± 2 pF; 2 ) Input set B ( n = 3 ) : Vrest = -78 . 0 ± 5 . 5 mV , Ri = 178 ± 14 MΩ and Cm = 24 ± 3 pF . All chemicals and drugs were purchased from Sigma-Aldrich ( USA ) . As explained above , three types of input sets were used , defined by the statistical structure ( P for Poisson , B for Bursty ) and the firing rate of their spike trains ( constant or varying across trains ) : 1 ) P10Hz ( Figs 1–3 and 7–9 ) , 2 ) PΔFR ( input set A in Figs 4–6 ) , 3 ) B10 . 5Hz ( input set B in Figs 4–6 ) . Input sets with Poisson spike trains ( P10Hz and PΔFR ) were generated using two different algorithms allowing to specify the number of spike trains in an input set ( 5 or 10 , respectively ) , the firing rate of each spike train , and the average Pearson's correlation coefficient across all spike trains ( computed with τw = 10ms ) . First , through iterative modifications of a random Poisson spike train , we generated five different P10Hz input sets of 5 trains with average correlation Rinput = 0 . 88 , 0 . 84 , 0 . 74 , 0 . 65 , 0 . 56 . Six other P10Hz input sets of 5 trains ( Rinput = 1 . 00 , 0 . 95 , 0 . 76 , 0 . 48 , 0 . 26 , 0 . 11 ) and one PΔFR set of 10 trains ( Rinput = 0 . 75 ) were generated using the Matlab toolbox provided by Macke and colleagues [81] , which uses a more efficient and mathematically principled algorithm that randomly samples from a multivariate Poisson distribution with preset rate ( FR ) and covariance matrix . For both algorithms , although pairwise correlations were constrained around the specified mean Rinput ( e . g . at τw = 10 ms , on average across all P10Hz input sets , the relative standard error of pairwise Rinput values is 4% of the mean Rinput ) , some variability remained ( see Fig 1E right ) which allowed to test a wide array of pairwise input correlations ( Fig 1F ) . Because this variability may be larger when measuring spike train similarity with different metrics than R , or at τw values that were not used to design input sets , we reported pairwise similarity values ( as opposed to the average similarity across all trains ) throughout the article . To design an input set with 10 spike trains spanning a range of burstiness ( B10 . 5Hz ) , we simply constrained the number of spikes to be equal in all trains and specified the number of bins with at least one spike for each train ( bursty trains have a lower number of occupied bins ) . Spikes were assigned to randomly selected bins and to random times within those bins . We assessed the similarity between spike trains in four ways , spanning a range of different assumptions on the neural code: 1 ) with the Pearson's correlation coefficient ( R ) , 2 ) with the normalized dot product ( NDP ) , 3 ) with the scaling factor ( SF ) and 4 ) with a distance metric called SPIKE specifically designed to assess the dissimilarity between two spike trains [48] . In addition , we used measures of differences between spike trains that assume different neural codes , related to the one assumed by SF but easier to interpret in terms of basic spike train features ( see Firing rate and burstiness codes and Dispersion metrics sections below ) . For R , NDP and SF , two spike trains X and Y of the same duration were divided into N time bins of duration τw , with Xi and Yi the respective numbers of spikes in bin i for each spike train . In contrast , the SPIKE metric is binless and based on the spike times of each spike train . For all similarity measures , empty spike trains ( i . e . sweeps during which no spikes were evoked ) were excluded . R ( Pearson's correlation coefficient ) can take values between 1 ( perfectly correlated , i . e . identical ) and -1 ( anticorrelated ) with 0 indicating X and Y are uncorrelated . It was computed with the following equation , where cov is the covariance ( scalar ) , σ is the standard deviation and X¯ and Y¯ are the means of X and Y: R=cov ( X , Y ) σX . σY=∑i=1N ( Xi−X¯ ) ( Yi−Y¯ ) ∑i=1N ( Xi−X¯ ) 2∑i=1N ( Yi−Y¯ ) 2 ( Eq 1 ) NDP ( Normalized Dot Product ) is the cosine of the angle θ between the two vectors: it is 0 when they are perfectly orthogonal , 1 when they are collinear ( see Fig 2A and 2C ) . NDP was computed as the dot product of X and Y divided by the product of their norms , with the following equation ( where Xi and Yi are the coordinates of X and Y ) : NDP=cos ( θ ) =∑i=1NXiYi∑i=1NXi2∑i=1NYi2 ( Eq 2 ) SF ( Scaling Factor ) quantifies the difference of length between the two vectors X and Y . We have defined it as the ratio between the norms of each vector , the smaller norm always divided by the bigger one to have SF values ranging from 0 to 1 . SF = 1 means X and Y are identical in terms of binwise spike number . The closer to 0 SF is , the more dissimilar are X and Y ( see Fig 2A and 2C ) . SF was computed with the following equation ( where 0 < ||X|| ≤ ||Y|| ) : SF=||X||||Y||=∑i=1NXi2∑i=1NYi2 ( Eq 3 ) NDP and SF focus on complementary features of a system of two vectors ( angle vs norms ) , and thus , together they are sufficient to fully describe the similarity between two vectors in a Euclidean space . The binless SPIKE similarity metric was computed from lists of spike times using the Matlab toolbox provided at www . fi . isc . cnr . it/users/thomas . kreuz/sourcecode . html . The provided software computes the SPIKE-distance between two spike trains ( called D ( t ) in our study and S ( t ) in eqn . 19 of the original paper [48] ) and we derived the SPIKE similarity as: SPIKE = 1−1T∫0TD ( t ) dt , where T is the duration of the spike train . Because D ( t ) ranges from 0 to 1 , SPIKE is thus also between 0 and 1 , like NDP and SF . When SPIKE is equal to 1 , spike trains have exactly the same spike times , i . e . they are synchronous ( n . b . in our experiments , spike trains were not simultaneously recorded , but we use "synchronous" in the sense of spike trains aligned to the start of each sweep ) . Note that SPIKE has a large dynamic range ( i . e . sensitivity over large differences of spiketimes ) , and , as a result , realistic spike trains like in our input sets rarely have a SPIKE similarity lower than 0 . 45 [48 , 82] ( see Fig 3D ) . The ranges of possible values for the metrics we used are listed below , where closer to 1 always means more similar ( although what similar means is different for each metric ) : −1≤R≤1 0≤NDP≤1 0<SF≤1 0≤SPIKE−similarity≤1;but≥0 . 45inmostcases Each metric assumes a specific neural code . For instance , R , NDP and SF all consider that the basic informative feature in spike trains is the number of spikes in a time bin ( i . e . a spike count code ) , but R and NDP suppose that spike trains are similar when the spike counts increase or decrease similarly in the same time bins ( with empty bins not counting for NDP ) , thus assuming a binwise synchrony code , whereas SF is independent of the temporal structure ( i . e . the fact that Xi and Yi correspond to the same time bin ) and assumes a code where it is the similarity in overall burstiness or firing rate that is relevant ( see Fig 4A and 4B ) , not the binwise synchrony ( see Fig 2A–2C ) . SPIKE , on the other hand , considers a spike time synchrony code . Details on the neural codes assumed by each metric , how they differ from each other and how they are impacted by global increases in firing rate can be found in the S1 Appendix ( part 1 ) . The main assumptions and properties of each similarity metric on the neural code are summarized in Table 1 . Note that we do not consider one metric to be better than the other , or better than alternative metrics not used in this study . Each metric can be used as a tool to analyze different aspects of the neural code , with a set of pros , cons and assumptions that complement other methods . R , NDP , SF and SPIKE are considering different aspects of spike trains and thus assume different neural codes . However , it can be difficult to interpret those similarity measures in terms of common properties of spike trains like the firing rate or the burstiness . To get a better sense of the neural code assumed by each similarity metric , and to directly assess whether pattern separation was performed through easily interpretable strategies like variation of the firing rate or variation of the burstiness , we measured such properties in each spike train . The firing rate is simply the number of spikes divided by the duration of a sweep ( 2 seconds ) . Measuring burstiness is less straightforward as there is currently no consensus on the definition of a burst of action potentials [83–85] . We used three measures , detailed below . Spike trains are often considered bursty when they diverge from a Poisson process [28 , 85] . To determine the tendency of a neuron to fire bursts , we computed the burstiness of a given output set as the Kullback-Leibler divergence ( DKL ) of its interspike interval ( ISI ) frequency distribution M ( normalized to the median ISI ) from the normalized ISI frequency distribution P of a Poisson process . In practice , output sets from a 10 Hz input set were compared to the ISI distribution from all combined 10 Hz input sets , whereas input sets from 30 Hz input set were compared to the combined distribution of all 30 Hz input sets . Burstiness=DKL ( M||P ) =∑jM ( j ) log ( M ( j ) P ( j ) ) ( Eq 4 ) where M ( j ) or P ( j ) is the value of M or P in the jth bin of the distribution . Normalized ISI frequency distributions were discretized in 250 bins going from 0 to 50 , and DKL was computed using the KLDIV matlab function ( v 1 . 0 . 0 . 0 ) , written by David Fass , available at: https://www . mathworks . com/matlabcentral/fileexchange/13089-kldiv . See S1 Appendix—2 for details on why DKL is a suitable measure of cell burstiness . To assess burstiness in single spike trains , the DKL measure being not well-suited for this ( see S1 Appendix—2 ) , we designed two simple nonparametric measures that we called Compactness and Occupancy . Instead of explicitly defining "burstiness" , these measures are based on the binning of spike trains , offering the advantage of being directly comparable to the neural codes assumed by binned similarity metrics like R , NDP and SF ( S2 Fig ) . A spike train is thus seen as a vector of spike counts , and a time bin ( or vector coordinate ) is considered a window in which spikes can be clustered . Thus , instead of counting the number of detected bursts and the number of spikes in a burst , our two measures provide information on the number of occupied bins ( Compactness ) and the number of spikes in occupied bins ( Occupancy ) . These measures of burstiness should be viewed as complementary to previous definitions , not as replacements . The Compactness of a spike train was computed as follows: Compactness=1−proportionofoccupiedbins=1−numberofbinswithatleast1spiketotalnumberofbins ( Eq 5 ) Compactness thus runs between 0 ( all bins are occupied by at least one spike ) and 1 ( no spikes: this case was excluded so the true maximum , corresponding to any spike train with all its spikes clustered in just one bin , was 1−1totalnumberofbins ) . Values closer to 1 mean very few bins are occupied: either many spikes are clustered in a small number of bins or the measured spike train has very few spikes . The Occupancy of a spike train is the average number of spikes per occupied bin: Occupancy=numberofspikesnumberofoccupiedbins ( Eq 6 ) Because we excluded empty spike trains , the Occupancy can theoretically go from 1 to infinity . Note that the lowest value ( i . e . 1 spike per occupied bin ) means that all spikes are distributed in different bins ( or that there is just 1 spike ) . Note that , in addition to being dependent on the binsize , Compactness and Occupancy are both dependent on the firing rate of a given spike train , but not in the same way ( see S1 Appendix—2 for details ) . They are complementary measures that , together , allow determining whether two spike trains with different firing rates are also different in terms of burstiness ( S1 Table ) . When spike trains have the same firing rate ( e . g . input set B 10 . 5 Hz in Fig 4B , which were all trains with 21 spikes each but were constructed by specifying different numbers of occupied bins ) , Compactness and Occupancy are redundant ( S2C and S2D Fig ) and , for both measures , higher values mean higher burstiness . To assess the amount of pattern separation performed by variation of a given spike train feature like the firing rate , Compactness or Occupancy , we measured the difference in this feature between two spike trains . In the same spirit as for similarity metrics , we could then compute the difference between all pairs of trains in an input set ( excluding self-comparisons ) and compare to the difference of all corresponding pairs of output spike trains , ( excluding comparisons between output spike trains associated with the same input spike train ) . This would yield matrices of pairwise differences organized in the same fashion as the similarity matrix displayed in Fig 1E , which could then be converted to pattern separation graphs showing the average pairwise output absolute difference as a function of the pairwise input difference ( Fig 6A ) . Note that the mean absolute difference ( MAD ) between all spike trains of a given set ( either input or output ) corresponds to the Gini mean difference [86] , a common measure of dispersion of a sample akin to the standard deviation ( Fig 5A ) . The difference in MAD between an input set and its output set indicates whether pattern separation or convergence was achieved through variation of the measured spike train feature ( Fig 6B ) . Data analysis was performed using MATLAB ( 2017a , Mathworks ) . Sample sizes were chosen based on the literature and estimations of the variance and effect size from preliminary data . The one-sample Kolmogorov-Smirnov test was used to verify the normality of data distributions . Parametric or non-parametric statistical tests were appropriately used to assess significance ( p-value < 0 . 05 ) . Assumptions on equal variances between groups were avoided when necessary . All tests on means or medians were two-tailed . To determine whether distributions of similarity values [Sinput , Soutput] ( S standing for any similarity metric ) were significantly different at a given time scale ( Figs 7 and 8 ) , we performed an analysis of the covariance ( ANCOVA ) using separate lines regression models , with Sinput as a continuous predictor and Treatment ( Fig 7 ) or Celltype ( Fig 8 ) as a categorical predictor with 2 or 5 levels respectively ( the aoctool function in Matlab and a custom-written code yielded the same results ) . The 95% confidence interval around the slope and intercept of a given linear model was determined from the regress function ( Fig 8 ) .
|
Pattern separation ( the process of disambiguating incoming patterns of neuronal activity ) is a central concept in all current theories of episodic memory , as it is hypothesized to support our ability to avoid confusion between similar memories . For the last thirty years , pattern separation has been attributed to the dentate gyrus of the hippocampus , but this has been hard to test experimentally . Moreover , because it is unclear how to define activity patterns in the brain , such a computation could be achieved in many different ways . Here , we demonstrate that pattern separation is performed by hippocampal networks ( dentate gyrus and CA3 ) through a variety of neural codes . By systematically testing different definitions of what it means for spike trains to be similar ( using a range of time scales and both standard and innovative metrics that assume different views of the neural code ) , we assessed how the input-output transformation of multiple hippocampal celltypes relate to pattern separation and found that different celltypes favor complementary coding strategies . This might help storing rich but concise and unambiguous representations of complex events . Finally , we provide the first experimental evidence of the importance of inhibitory signals in mediating pattern separation , and identify through which coding strategies .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[] |
2019
|
Temporal pattern separation in hippocampal neurons through multiplexed neural codes
|
Cells must make reliable decisions under fluctuating extracellular conditions , but also be flexible enough to adapt to such changes . How cells reconcile these seemingly contradictory requirements through the dynamics of cellular decision-making is poorly understood . To study this issue we quantitatively measured gene expression and protein localization in single cells of the model organism Bacillus subtilis during the progression to spore formation . We found that sporulation proceeded through noisy and reversible steps towards an irreversible , all-or-none commitment point . Specifically , we observed cell-autonomous and spontaneous bursts of gene expression and transient protein localization events during sporulation . Based on these measurements we developed mathematical population models to investigate how the degree of reversibility affects cellular decision-making . In particular , we evaluated the effect of reversibility on the 1 ) reliability in the progression to sporulation , and 2 ) adaptability under changing extracellular stress conditions . Results show that reversible progression allows cells to remain responsive to long-term environmental fluctuations . In contrast , the irreversible commitment point supports reliable execution of cell fate choice that is robust against short-term reductions in stress . This combination of opposite dynamic behaviors ( reversible and irreversible ) thus maximizes both adaptable and reliable decision-making over a broad range of changes in environmental conditions . These results suggest that decision-making systems might employ a general hybrid strategy to cope with unpredictably fluctuating environmental conditions .
Cellular decision-making underlies many biological processes such as multipotent differentiation , where cells commit to one of several distinct fates . Such cell fate choice must permit individual cells to reach a decision even in fluctuating extracellular environments [1] . At the same time , cells must also be able to adapt their cell fate choice to changes in these conditions . It is unclear how individual cells reconcile these opposing requirements of decisiveness and adaptability during decision-making . Decisive cellular differentiation mechanisms have been proposed to combine ultra-sensitivity and positive feedback to generate an irreversible and all-or-none cell fate choice such as those observed during Xenopus oocyte maturation [2] and yeast mating decision [3] . However , individual cells with irreversible responses can lack the flexibility to respond proportionally to changing environments , since even small changes can trigger irreversible responses . In contrast , progression of cellular differentiation through reversible intermediate states permits flexibility and proportional responses to environmental changes . For example , multipotent differentiation of hematopoietic stem cells is a stepwise process with numerous reversible intermediate states that allows cells to gradually adapt to changes in extracellular signals [4] , [5] , [6] , [7] , [8] , [9] , [10] . Despite these recent insights , how multipotent differentiation systems reach a decisive cell fate choice while maintaining the ability to respond to changes in the environment is largely unknown . To understand cellular decision-making it is critical to determine the single-cell dynamics underlying the progression to cell fate choice . However , these dynamics are poorly characterized in multipotent differentiation systems ranging from bacteria to mammalian stem cells . Simultaneous measurement of multiple components of a differentiation program in the same cell can reveal the dynamics of cellular decision-making underlying multipotent differentiation . The soil bacterium Bacillus subtilis serves as an ideal model system in which the dynamics of multiple genes within a differentiation circuit are simultaneously measurable in single cells [11] , [12] , [13] . In stressful environments the majority of B . subtilis cells form spores that survive environmental extremes [14] , [15] . The sporulation program has been well characterized genetically and multiple stages of sporulation have been described [16] , [17] , [18] , [19] , [20] . However , despite these important insights , how individual cells proceed to spore formation and thus the dynamics of the sporulation program in single cells has not been determined . To uncover cell fate choice dynamics in B . subtilis , we simultaneously measured the activities of multiple sporulation circuit elements with single-cell resolution . We found that individual cells progress gradually to spore formation through reversible activities of early sporulation components . These measurements also allowed us to confirm the previously identified commitment point of cellular decision-making at which cells irreversibly proceed to complete spore formation [21] . More importantly , single cells analysis revealed the precise all-or-none dynamics of this decision-making point that was obscured by variability at the single-cell level in population measurements . Modeling of alternative sporulation dynamics showed that the combination of reversible and irreversible dynamics employed by B . subtilis can represent a general strategy to maximize reliable and yet adaptable cellular decision-making over a broad range of randomly fluctuating environmental conditions .
First , we established the single-cell dynamics of the progression toward spore formation by measuring the temporal profile of four sporulation components ( Fig . 1A ) . Initiation of sporulation is controlled by a multicomponent phosphorelay including two phosphotransferases , Spo0F and Spo0B , and a transcription factor Spo0A [14] , [19] , [22] , [23] , [24] , [25] . Spo0F protein senses and integrates a variety of inputs and as a result becomes phosphorylated [23] , [24] . The phosphate group is subsequently transferred through Spo0B to Spo0A , a master regulator of sporulation , which upon phosphorylation directly controls expression of over 120 genes [20] , [26] , [27] , including spo0F ( in what constitutes a feedback mechanism [28] , [29] , [30] ) . Among these is the gene for SpoIIE , a protein phosphatase that is required for the activation of the forespore specific transcription factor ( σF ) and whose localization to the asymmetric septum is the first morphological marker for forespore formation [31] , [32] , [33] , [34] , [35] . Activation of σF in the forespore switches on the expression of SpoIIR , which in turn leads to the activation of σE , a transcription factor specific to the mother cell that switches on the expression of a large number of late sporulation genes [36] , [37] , [38] . At this point , the cell becomes irreversibly committed to sporulation [21] , [37] , [38] , [39] . Using fluorescent reporter constructs , we quantified the activities of the Spo0A ( Pspo0A ) , Spo0F ( Pspo0F ) and SpoIIR ( PspoIIR ) promoters ( Fig . 1A , top two rows ) . We also visualized SpoIIE protein localization by utilizing a functional translational fusion to SpoIIE ( Fig . 1A , bottom row ) . Each sporulation reporter was simultaneously measured in combination with the PspoIIR reporter . Measurement of these overlapping pair-wise combinations of reporters allowed us to establish a relative temporal profile for multiple sporulation steps with single-cell resolution ( Fig . 1B and Supporting Fig . S1 ) that was consistent with the genetically established hierarchy within the sporulation circuit [14] , [18] , [22] , [40] , [41] , [42] ( Fig . 1C ) . Single cell measurements of sporulation reporters revealed that progression to spore formation is comprised of reversible steps . During the early progression to spore formation , individual cells exhibited bursts of Pspo0A gene expression that did not result in spore formation ( Fig . 2A ) . Activation of Spo0A during B . subtilis sporulation is known to be heterogeneous among single cells [43] , [44] . However , spontaneous bursts of gene expression at single-cell level during sporulation have not been described to date . Bursts in gene expression were not limited to Pspo0A , but were also observed for Pspo0F ( Fig . 2B ) . Given that Pspo0F is , as mentioned above , transcriptionally activated by phosphorylated Spo0A , the dynamics of that promoter is reporting the activity of Spo0A . Consequently , the observed bursts in Pspo0F indicate that not only the expression , but also the activity of the Spo0A master regulator exhibits bursts in single cells . In contrast , bursts of gene expression were not observed for the late stage sporulation reporter PspoIIR ( Fig . 2A–C ) : after the sharp signal increase observed for PspoIIR , a spore is always formed ( and thus the time traces shown in the plots cannot be continued ) . Gene expression bursts can introduce variability and reversibility during the early stages of the progression to spore formation , whereas the later stages do not appear to be subject to such stochasticity . Additionally , we observed reversible protein localization of SpoIIE in approximately 2±1% ( SEM ) of cells ( Fig . 2C ) . Specifically , we observed transient localization events of the SpoIIE-YFP fusion protein to the asymmetric septum that did not give rise to spore formation . In these cells , SpoIIE either switched its localization between opposite poles , or completely delocalized and cells continued with cell division ( Fig . 2C ) . Similar bursts of promoter activity and protein localization in single cells have also been reported in other systems , and have been attributed to the stochastic and reversible nature of the underlying biochemical reactions [43] , [45] , [46] , [47] , [48] , [49] , [50] . The single-cell reversibility observed here is distinct from what has been reported in population-level studies which showed that sporulation can be aborted upon transfer from stress to rich media conditions [21] . In contrast , the single-cell reversibility discussed here appears to be a cell autonomous behavior that occurs randomly without requiring a change in environmental conditions . Such cell autonomous and random sampling of heterogeneous behavior in single cells has also been observed in other systems where it may provide a fitness advantage , such as the generation of antibiotic resistant Escherichia coli persister cells [51] and the survival of Saccharomyces cerevisiae to cellular stress [50] . Together , our data reveal that the early steps of progression towards spore formation are subject to stochastic reversibility at the single-cell level , suggesting an adaptable progression to spore formation . Detailed analysis of single cell dynamics revealed that the irreversible commitment for spore formation is executed within a narrow time window . Specifically , we find that PspoIIR activation ( which as mentioned above , denotes the sporulation commitment point ) is switch-like ( Fig . 2A–C , green line ) . This timing was quantified by using the morphological appearance of an actual spore ( indicated by a phase bright spot in our single-cell movies ) as a reference time point from which to measure commitment time ( Fig . 3A ) . Analysis of single-cell data shows that the temporal distance between PspoIIR activation and the formation of the morphologically visible forespore is tightly distributed ( CV = 0 . 2 , and 15% of median cell cycle duration ) ( Fig . 3B ) . These results show the precise timing of the irreversible decision and the completion of the spore formation process . Therefore , single cell measurements of sporulation dynamics exposed a precise cell intrinsic decision point in time that revealed switch-like dynamics of the commitment to spore formation . The temporal precision of this decision-making point was previously concealed in population measurements ( see Fig . 1B ) by the cell-to-cell variability in the reversible progression to spore formation . Our measurements suggest that the decision to sporulate in B . subtilis is made in an irreversible all-or-none manner , following a reversible and gradual progression toward this decision point . Therefore , sporulation combines reversible and irreversible dynamics , two seemingly opposed decision mechanisms . In order to examine the potential advantages of this hybrid mechanism , we compared the response of three simplified models of cellular decision-making to a variable-stress environment . These models describe the dynamics of a population of cells progressing toward sporulation under stress , in terms of the number of cells existing at any given time in a given state along the progression , beginning with the vegetative state and ending in the spore state . The dynamics of cell populations in all these states is given by a set of coupled ordinary differential equations that are linear , and thus can be solved exactly ( Supporting Text S1 ) . The three models , shown schematically in Fig . 4A and described in detail in the Supporting Information ( Supporting Text S1 and Supporting Fig . S2 ) , involve either 1 ) a purely irreversible/all-or-none , 2 ) a purely reversible/gradual , or 3 ) a “hybrid” process that takes cells from their initial vegetative state to their final spore state . In the irreversible-only scheme , cells decide in a single step whether or not to sporulate , and the decision is irreversible . The reversible-only model takes cells gradually toward sporulation through multiple reversible intermediate states , without any irreversible commitment step taking place along the process . Only the ultimate transition to the spore state , taking place after the decision , is irreversible . Finally , the third “hybrid” model features the actual sporulation dynamics identified here for B . subtilis that combines a gradual progression through reversible intermediate states with an irreversible all-or-none decision prior to the spore transition . We subjected the three models described above to a random variation of the environment , with alternation of a high and a low level of stress ( insets in Fig . 4B and Supporting Fig . S3 ) . For all models , the rates of progression toward/back from sporulation , growth and death rates are all coupled to the level of stress ( Supporting Text S1 ) . We then systematically varied the ratio of high stress to total cycle duration , which was kept constant . The relative fitness of the hybrid model with respect to the irreversible-only and reversible-only models was measured as the ratio of total cells in the population between the models ( Fig . 4B , blue and red lines respectively ) . Results of this analysis show that the survival rate of the different models depends on the stress profile . None of the models dominates over the entire range of environmental stress conditions , but the hybrid model performs best overall . For short phases of high stress , the hybrid outgrows the irreversible-only model ( left half of main plot in Fig . 4B , blue line ) , since the latter is driven irreversibly to sporulation even for short stress periods . Under these conditions , the hybrid and reversible-only models perform similarly ( red line ) since in both models reversible progression delays sporulation . However , in the opposite limit where high stress durations approach the total duration ( right half of main plot in Fig . 4B ) , the irreversible-only model outgrows the hybrid model , since responsiveness is a disadvantage . For such prolonged high stress durations , spores are at an advantage because non-spore cells have a higher death than growth rate under stress . In this limit , the hybrid model is in turn more reliable than the reversible-only model , since the reversible-only model is driven away from sporulation even by short intervals of rich phase . Parameter sensitivity analysis showed that these results hold over a range of parameter values ( Fig . 4B , light blue and light red regions ) . Taken together , these results show that the hybrid model outperforms both the irreversible-only and reversible-only schemes over a broader range of randomly alternating stress profiles , given its responsiveness to the long-term recovery from stress and its reliability during short-term release of stress conditions . This prediction was tested directly by subjecting all three models to random changes in the fraction of time spent in high stress . In these extended simulations cells were exposed to 100 random samplings of environmental stress ratios ( part of an environmental profile is shown in Fig . 4C , inset ) . We then calculated the ratio of the total cell population of the hybrid model to that of the irreversible-only and reversible-only models averaged over the entire simulation , defining the relative fitness of each model ( Fig . 4C ) . Our data demonstrate that the hybrid model has an overall higher fitness over both the irreversible-only and reversible-only models when populations are subjected to a randomly changing range of environmental conditions . Therefore , the combination of gradual progression toward an all-or-none decision in B . subtilis , as represented in the hybrid model and observed experimentally , enables the system to cope with a broader range of unpredictable stress conditions .
Single cell measurements of sporulation components allowed us to precisely determine both the temporal progression of the earliest sporulation events and the consecutive switch-like dynamics of the commitment point . The temporal precision and switch-like nature of the decision point would have been concealed in population measurements that describe average behaviors of cells and cannot account for cell-to-cell variability . Similarly , a recent study of the post-infection decision in bacteriophage lambda to undergo lysis or lysogeny showed how cell-cell variability obscured the precision of cell fate choice [52] . Therefore , stochastic fluctuations and noise observed at the single-cell level can conceal the temporal precision of numerous cellular processes . Our approach based on measuring multiple gene circuit components simultaneously in the same cell can identify cell intrinsic temporal reference points that can accurately establish the relative timing of cellular processes . Multipotent differentiation in B . subtilis appears to combine the opposing dynamic regimes of reversible/gradual and irreversible/all-or-none behavior to reconcile the seemingly contradictory requirements of adaptable and reliable cellular decision-making . The time required for cells to reach spore formation is highly variable , allowing cells to generate a broad distribution of wait times prior to cell fate choice . This extends the period during which B . subtilis cells are known to be responsive to environmental changes . Therefore , variability in sporulation progression can be beneficial biologically . These findings thus extend previous results regarding the advantageous role of stochasticity in enhancing survival under uncertain environments [13] , [50] , [53] , [54] , [55] , [56] . In contrast , the actual decision to sporulate is governed by irreversible and all-or-none dynamics , which provides reliable execution of the sporulation program . Our modeling suggests that this combination of gradual and all-or-none dynamics during cell fate choice allows B . subtilis to successfully survive under a broad range of alternating environmental stress profiles . In mammalian cells , the decision for apoptosis has been described as a slow progression towards a fast decision , suggesting that a hybrid strategy similar to that described here might be employed [57] . Therefore , this strategy of combining opposing dynamic behaviors may be common to various other biological processes , and may represent a general mechanism for decision-making under unpredictably changing environments . Models for decision-making under uncertain conditions have been developed and applied to numerous unrelated complex systems , such as finance [58] . Perhaps a stochastically reversible progression to an irreversible all-or-none switch as observed here in cells , may also serve as a strategy for adaptable and reliable decision-making in other complex systems that are subject to unpredictable conditions .
Bacillus subtilis strains used in the study are isogenic to wild-type B . subtilis PY79 strain and are listed in Supporting Table S1 . Promoter – fluorescent proteins fusions were generated using fusion polymerase chain reaction and cloned into B . subtilis chromosomal integration vectors following standard protocols . The following B . subtilis chromosomal integration vectors were used: pSac-Cm , integrating into the sacA locus ( constructed by R . Middleton and obtained from the Bacillus Genetic Stock Center ) and pLD30 integrating into the amyE locus ( kind gift from Jonathan Dworkin , Columbia University ) . We have also utilized the bifunctional cloning plasmid pHP13 carrying the replication origin of the cryptic B . Subtilis plasmid pTA1060 ( 5 copies per genome ) [59] . Standard one-step transformation protocol was followed to transform B . subtilis with these constructs . For imaging , B . subtilis cells were grown at 37°C in LB with the following concentrations of appropriate antibiotic: 5 µg/ml chloramphenicol , 5 µg/ml neomycin , 5 µg/ml erythromycin or 100 µg/ml spectinomycin . After reaching OD 1 . 8 , the cells were resuspended in 0 . 5 volume of Resuspension Media ( RM; see Supporting Text S2 , Materials and Methods ) supplemented with 0 . 02% glucose . The cells were incubated at 37°C for 1 hour , then diluted 10-fold in RM and applied onto a 1 . 5% low-melting agarose pad placed into a coverslip-bottom Willco dish for imaging . Growth of microcolonies was observed with fluorescence time-lapse microscopy at 37°C with an Olympus IX-81 inverted microscope with a motorized stage ( ASI ) and an incubation chamber . Image sets were acquired every 20 min with a Hamamatsu ORCA-ER camera . The imaging time has been optimized in order to prevent phototoxicity [11] . Custom Visual Basic software in combination with the Image Pro Plus ( Media Cybernetics ) was used to automate image acquisition and microscope control . Time-lapse movie data analysis was performed by custom software developed with MATLAB image processing and statistics toolboxes ( The Mathworks ) described in [60] and [11] . Extended description of the models and methods is available in Supporting Text S1 , Mathematical Modeling .
|
Cells must continuously make decisions in response to changes in their environment . These decisions must be irreversible , to prevent cells from reverting back to unfit cellular states , but also be flexible , to allow cells to go back to their previous state upon environmental changes . Using single-cell time-lapse fluorescence microscopy , we show that these seemingly contradictory properties coexist in Bacillus subtilis cells during their progression to spore formation . We suggest , on the basis of a mathematical population model , that reversible progression towards the irreversible decision to sporulate optimizes respectively adaptability and reliability of decision-making over a broad range of changes in environmental conditions .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"systems",
"biology",
"biology",
"computational",
"biology",
"genetics",
"and",
"genomics"
] |
2011
|
Reversible and Noisy Progression towards a Commitment Point Enables Adaptable and Reliable Cellular Decision-Making
|
In plants , nutrient provision of shoots depends on the uptake and transport of nutrients across the root tissue to the vascular system . Nutrient delivery to the vasculature is mediated via the apoplastic transport pathway ( ATP ) , which uses the free space in the cell walls and is controlled by apoplastic barriers and nutrient transporters at the endodermis , or via the symplastic transport pathway ( STP ) . However , the relative importance of these transport routes remains elusive . Here , we show that the STP , mediated by the epidermal ammonium transporter 1;3 ( AMT1;3 ) , dominates the radial movement of ammonium across the root tissue when external ammonium is low , whereas apoplastic transport controlled by AMT1;2 at the endodermis prevails at high external ammonium . Then , AMT1;2 favors nitrogen ( N ) allocation to the shoot , revealing a major importance of the ATP for nutrient partitioning to shoots . When an endodermal bypass was introduced by abolishing Casparian strip ( CS ) formation , apoplastic ammonium transport decreased . By contrast , symplastic transport was increased , indicating synergism between the STP and the endodermal bypass . We further establish that the formation of apoplastic barriers alters the cell type–specific localization of AMTs and determines STP and ATP contributions . These results show how radial transport pathways vary along the longitudinal gradient of the root axis and contribute to nutrient partitioning between roots and shoots .
A major function of plant roots is the uptake and subsequent translocation of nutrients from soil to above-ground plant organs . To reach the shoot , nutrients need first to be transported radially across the root tissue before entering the xylem for root-to-shoot translocation . Once nutrients cross the plasma membrane of root epidermal cells , they enter the symplastic pathway , on which they move through the cytoplasmic continuum via plasmodesmata from cell to cell until they arrive in the xylem [1] . Nutrients may also enter the free space and cell walls of epidermal and cortical cells and move passively along the apoplastic route , which ultimately becomes blocked by the Casparian strip ( CS ) at the endodermis [2] , where lignin depositions in anticlinal walls form a physical barrier to prevent an endodermal bypass [3] . This barrier prevents further inward movement in the apoplast . To progress further , nutrients must enter endodermal cells via membrane proteins , thereby completing the apoplastic transport pathway ( ATP ) . As both pathways require a membrane transporter–mediated step , we refer here to the ATP and the symplastic transport pathway ( STP ) . In basal root zones , endodermal cells become suberized , i . e . , coated at the inner cell walls with aliphatic polymers , which form another apoplastic barrier , preventing access of nutrients to the plasma membrane [4 , 5] . Endodermal bypass , i . e . , unhindered radial movement through cell walls of the endodermis , is only possible where these apoplastic barriers are not yet formed , such as in the apical root zone . Although the concept of radial nutrient transport , as determined by the ATP and STP , is common to all plant physiology textbooks [6 , 7] , it still remains largely based on coincidences and theoretical considerations , as the significance and quantitative share between these two transport pathways has not yet been characterized for any mineral element . A key methodological aspect required to dissect successfully the contribution of different radial transport pathways is the ability to manipulate the integrity of root apoplastic barriers . Earlier attempts to generate small artificial bypasses by puncturing endodermal cells with microcapillary tubes provoked instable root pressure and wounding responses [8] . However , with the recent isolation of mutants with impaired CS formation , it is possible for the first time to investigate these transport pathways at the molecular level . Among all CS-defective mutants characterized so far , the most suitable mutant is schengen3 ( sgn3 ) , since it exhibits strong and persistent CS defects but no enhanced or precocious accumulation of suberin [9] . SGN3 , also known as GASSHO1 ( GSO1 ) , is a leucine-rich repeat receptor–like kinase that acts as a receptor for 2 tyrosine-sulfated peptides known as CS integrity factors 1 and 2 ( CIF1 and CIF2 ) [10 , 11] . As SGN3 function cannot be replaced by another protein , a large endodermal bypass is created in sgn3 roots , turning this mutant into a valuable tool to assess the contribution and physiological relevance of different radial transport pathways . Ammonium ( NH4+ ) is a major source of soil nitrogen ( N ) for plants and is transported radially through the root tissue . A dedicated set of NH4+ transporters belonging to the ammonium transporter/methylamine permease/Rhesus-type ( AMT/MEP/Rh-type ) protein family is responsible for membrane transport of ammoniacal N in plants . Arabidopsis roots express 5 AMT-type transporters . Short-term influx studies had shown that under N deficiency , which up-regulates transcript levels of all 5 AMT genes in roots , 90%–95% of the high-affinity uptake capacity of NH4+ is mediated by AMT1;1 , AMT1;2 , and AMT1;3 [12 , 13] . In the low-affinity ( millimolar ) range , the contribution of AMT1-type transporters to overall NH4+ uptake shrinks because other low-affinity transporters come into play . For instance , AMT2;1 , the only member belonging to MEP/AMT2 subfamily , contributes 10%–25% to the overall ammonium uptake rate at high external ammonium concentrations [13] . Other low-affinity transporters , such as ammonium facilitator–type transporters or potassium ( K+ ) channels [14 , 15] , may further contribute to low-affinity ammonium transport , but their physiological function in planta still remains unclear . Interestingly , AMT1;1 and AMT1;3 reside primarily in the epidermis and are involved in NH4+ uptake for the early passage into the STP . In contrast , AMT1;2 is located primarily in the endodermis , suggesting that this transporter completes the ATP for NH4+ [12 , 16] . Thus , the distinct cell type–specific expression of individual NH4+ transporters in roots and the near absence of AMT-independent NH4+ uptake in the high-affinity range make NH4+ transport via AMTs ideally suited to study radial transport pathways in plant roots . Here , we quantify the relative contribution of the ATP and STP to radial transport of 15N-labeled ammonium ( 15NH4+ ) . We show that a leaky CS causing an extended endodermal bypass acts synergistically with the AMT1;3-mediated STP while competing with the ATP mediated by AMT1;2 . Together with cell type–specific localization of AMT proteins along the longitudinal axis of the Arabidopsis root , our study provides a novel basis for an improved mechanistic understanding on the contribution of radial transport pathways to nutrient partitioning between roots and shoots .
The sgn3 mutant was used to dissect the apoplastic route into an ATP , which entails an obligatory transporter-mediated step at the plasma membrane of the endodermis ( Fig 1A , blue ) , and an endodermal bypass ( Fig 1A , purple ) , allowing for unhindered diffusion across the endodermal layer . To quantify radial transport , we first exposed N-deficient plants to 15NH4+ and compared 15N accumulation in shoots to that in xylem exudates and found closely related patterns of 15N accumulation ( S1 Fig ) . Since NH4+ uptake rates and the expression of AMTs increase during daytime and decrease in the dark [17] , we assessed shoot accumulation rates during the light period , when transpiration is high and root pressure low , thereby suppressing the effect of root pressure on xylem transport rates . In order to favor transpiration as the major driving force for radial transport of water and nutrients , we calculated radial transport rates based on tracer accumulation in shoots rather than in xylem saps . Referring tracer accumulation in shoots to root dry weight and time ( “normalized shoot accumulation” ) thus integrates the rates of radial transport and of xylem loading in roots . Since AMT2;1 , a critical transporter for root-to-shoot NH4+ translocation [13] , was present in all tested lines , we assumed no significant changes in xylem loading and hence considered “normalized shoot accumulation” as a readout for radial substrate transport rates in roots . We then verified the relevance of an extended endodermal bypass in sgn3 by exposing hydroponically grown wild-type ( WT ) and sgn3 mutant plants to strontium ion ( Sr2+ ) , which , similar to calcium ion ( Ca2+ ) , is transported to shoots mainly via the ATP [18–21] . Indeed , compared to wild type , normalized shoot accumulation of Sr2+ was approximately 3- or 20-fold higher in sgn3 at low or high external Sr2+ , respectively ( Fig 1B and 1C ) . In roots , Sr2+ accumulation increased comparatively little ( S2A and S2B Fig ) , indicating an extraordinary impact of the endodermal bypass generated in the sgn3 mutant to radial transport and delivery of Sr2+ to the shoot . In the case of NH4+ , the impact of the endodermal bypass on shoot or root accumulation of NH4+ was insignificant ( Fig 1D and 1E , S2C and S2D Fig ) , suggesting that the action of the whole set of dedicated NH4+ transporters in WT and sgn3 plants dominated over the contribution of an endodermal bypass . Thus , the relevance of a purely apoplastic pathway in the form of an extended endodermal bypass for radial element transport depends on the substrate and the presence of transporter-mediated radial pathways . In agreement with an earlier study showing that phenotypical changes in sgn3 plants are highly dependent on the prevailing environmental conditions [9] , this mutant had significantly less root and shoot biomass than the wild type ( S3A and S3C Fig ) . Potential pleiotropic effects arising from the sgn3 mutation were assessed by verifying the relation between biomass and normalized shoot accumulation of NH4+ in esb1 and myb36 , 2 other CS-defective mutants [4 , 5] . Both mutants showed a similar decrease in shoot and root dry weight to sgn3 ( S3A and S3C Fig ) , but their normalized shoot accumulation of 15N was 20%–25% lower and went along with increased root accumulation of 15N ( S3B and S3D Fig ) . Thus , we concluded that radial transport rates of NH4+ were not primarily affected by plant biomass but rather by the properties of existing apoplastic barriers . Unlike sgn3 , the CS defects of esb1 and myb36 are partially compensated for by a stronger and earlier suberization of endodermal cells [4 , 5] , which makes endodermal transporters inaccessible for their substrates [21] . Thus , the decreased biomass of sgn3 is not specific to the sgn3 mutation but most likely the consequence of a leaky CS . To compensate for the differences in root dry weight , we normalized all shoot accumulation rates to root dry weight , as practiced in other transport studies [22 , 23] . To quantify the contribution of individual pathways to radial NH4+ transport , we introgressed the sgn3 mutation into the triple amt1;1 amt1;2 amt1;3 ( tko ) knockout line , which has only 5%–10% of the wild type capacity for high-affinity NH4+ uptake [12] . The obtained quadruple knockout ( tko sgn3 ) had an extended endodermal bypass ( Fig 2A ) and smaller rosette leaves than tko but no visible symptoms of nutrient deficiency ( S4 Fig ) . The presence of such endodermal bypass in a root devoid of the 3 major NH4+ uptake transporters increased significantly the normalized shoot accumulation of NH4+-derived N but only at high external supply ( tko versus tko sgn3; Fig 2B and 2C ) . Although alternative uptake pathways for NH3 or NH4+ likely exist in roots , the absence of a significant difference between wild type and sgn3 ( Fig 1D and 1E ) suggests that the increased normalized shoot accumulation of tko sgn3 relative to tko plants at high external ammonium ( Fig 2C ) is most likely due to the extended endodermal bypass rather than the action of low-affinity transport pathways . To compare the contribution of the endodermal bypass with that of the ATP or STP , we generated tko lines with reconstituted expression of either endodermal AMT1;2 ( tko+1;2 ) , thus installing the end point for the ATP at the endodermis , or epidermal AMT1;3 ( tko+1;3 ) , thereby establishing an early entry into the STP ( Fig 2A ) . At low external 15NH4+ , symplastic transport via AMT1;3 alone conferred slightly higher normalized shoot accumulation for the tracer than the AMT1;2-dependent ATP ( Fig 2B and 2D ) . At the same time , 15N accumulation in roots was significantly higher in tko+1;3 than in tko+1;2 ( S5A and S5B Fig ) . However , at elevated substrate levels , the difference between tko+1;3 and tko+1;2 to tko reversed ( Fig 2C ) , and the estimated contribution of the AMT1;2-dependent ATP to shoot 15NH4+ accumulation became twice as high as that through the STP ( Fig 2E ) . Notably , the higher normalized shoot accumulation attributed to the ATP was independent of 15N accumulation in roots ( S5C and S5D Fig ) , indicating that root and shoot accumulation of NH4+-derived N were uncoupled . The radial transport capacity of AMT1;1 , AMT1;2 , and AMT1;3 together , as reflected by the difference of normalized shoot accumulation between wild type and tko ( Figs 1D , 1E , 2B and 2C ) , was somewhat lower than the sum of the 2 individual capacities conferred by AMT1;2 and AMT1;3 ( Fig 2D and 2E ) . This was not unexpected with regard to the fact that AMT1;1 and AMT1;3 can individually compensate for the lacking uptake capacity of each other because the formation of a heterotrimeric complex and concomitant posttranslational down-regulation of interacting monomers was no longer possible [12 , 24] . This was also the reason why the contribution of AMT1;1 could be disregarded here . In conclusion , at low external substrate concentrations , the contribution of the STP to radial transport of NH4+-N prevails over the ATP , whereas at millimolar NH4+ supply , the ATP mediated by AMT1;2 confers an approximately 2-fold higher capacity than that of the symplastic route . These results represent , to our knowledge , the first quantitative comparison of individual radial transport pathways for any nutrient . To investigate the cross-talk between radial transport pathways , we hypothesized that the presence of an extended endodermal bypass will decrease radial 15NH4+ transport via the apoplastic or symplastic pathway because tracer reaching the stele may leak out in the absence of a functional CS . Indeed , shoot 15NH4+ accumulation via the AMT1;2-dependent apoplastic pathway in the presence of an endodermal bypass ( i . e . , tko sgn3+1;2 versus tko sgn3 ) revealed similar or significantly lower 15NH4+ accumulation relative to the contribution of the ATP alone ( Fig 2B and 2C ) . These results indicated that AMT1;2-mediated 15NH4+ transport across the endodermis was compromised by concomitant NH4+ efflux through an endodermal bypass . This observation highlights that an intact CS improves the efficiency of the apoplastic transport route by limiting apoplastic backflow out of the vasculature . Unexpectedly , the interaction between an endodermal bypass and the STP was opposite: normalized shoot accumulation of 15NH4+ via AMT1;3 was significantly higher in the tko sgn3 background ( tko sgn3+1;3 versus tko sgn3 ) than in the tko background ( tko +1;3 versus tko ) ( Fig 2B and 2C ) . Importantly , the estimated contribution of the STP in the presence of an endodermal bypass is higher than either of these transport routes alone , indicating a synergistic interaction between the STP and the endodermal bypass ( Fig 2D and 2E ) . Presumably , part of the NH4+ transported via the symplastic route was exported into the apoplast during the radial move and profited from unhindered apoplastic diffusion at the endodermis to reach the vascular system . The opposite contribution of AMT1;2 and AMT1;3 in the tko sgn3 background was not due to altered gene expression , as the expression of AMTs was not significantly affected by the sgn3 mutation ( S6 Fig ) . Moreover , we supplied sufficient amounts of potassium ( K ) in our hydroponic solution to prevent latent K deficiency in sgn3 [9] . With this measure , sgn3 mutants showed no symptoms of nutrient deficiency , and despite slight variations in the accumulation of other mineral elements , all of these were in a usual physiological range and far away from critical deficiency or toxicity levels [25] ( S4A and S7 Figs , S1 Table ) . Thus , our elemental analysis supported that the results with the sgn3 mutant were largely independent of mineral element disorders or potential interactions between K+ and NH4+ at the level of uptake or xylem loading . To verify the opposite interaction between an extended endodermal bypass and the 2 radial transport pathways in an independent growth system , we carried out a transport assay on horizontally split agar plates , in which 15NH4+ was supplemented only to the lower agar segment ( Fig 3A ) . Roots of N-deficient plants were placed across the trench , separating the upper and lower agar compartments so that only approximately 10 mm of the apical root zone were in contact with the 15N-containing segment . At this developmental stage , the shoot biomass between tko lines and tko sgn3 lines was almost indistinguishable ( Fig 3B ) . As labeled root segments were too small to measure root dry weights , we determined 15N concentrations in shoots as readout for radial transport of 15NH4+-N . Shoot 15N concentrations were higher in tko sgn3+1;3 plants than in tko+1;3 , thus confirming the synergistic interaction between an endodermal bypass and the STP . On the other hand , we observed a slightly decreased contribution of AMT1;2 in the tko sgn3 background ( tko sgn3+1;2 versus tko+1;2; Fig 3C ) , further supporting that an endodermal bypass antagonizes the ATP . To further validate our findings independently of the sgn3 mutation , we treated WT plants with the lignin biosynthesis inhibitor piperonylic acid ( PA ) , which blocks CS formation and hence partially mimics the CS defects obtained by mutating SGN3 [3] . Since the PA effect is confined to newly grown root portions , we carried out the experiment on agar plates and limited the PA treatment to 48 h . Within this period , the zone of unrestricted penetration of the apoplastic tracer propidium iodide ( PI ) greatly expanded , as the endodermal bypass was now open up to >8 mm from the root tip ( Fig 4A and 4B ) . Notably , this short-term disturbance of CS formation in PA-treated roots did not significantly compromise root growth or shoot biomass formation ( Fig 4C and 4D ) . We then exposed the apical 7 mm of the root tip to 15NH4+ in order to determine the contribution of PA-dependent endodermal bypass . Shoot 15N concentrations were strongly suppressed by PA ( Fig 4E and 4F ) , indicating an inhibitory side effect of PA on root physiology . Nevertheless , when compared to tko , the endodermal bypass created by PA decreased shoot 15N accumulation in tko+1;2 further but increased shoot 15N accumulation in tko+1;3 , indicating a synergistic interaction ( Fig 4E and 4F ) . Thus , 3 independent approaches provided evidence for the opposite interaction of the endodermal bypass with either the ATP or STP in roots . The overall dominant contribution of the STP to radial nutrient transport raised the question of the biological significance of the ATP . We thus compared root-to-shoot translocation in our mutant lines by calculating shoot-to-root 15N concentration ratios . At micromolar NH4+ supply , AMT1;2 conferred in the tko background a slightly higher increase in 15N translocation to shoots than AMT1;3 , while in the presence of an endodermal bypass , AMT1;3 delivered the most N to shoots , reflecting the synergistic interaction with the endodermal bypass ( Fig 5A and 5B ) . At millimolar supply , shoot N provision profited most from NH4+ delivered by the AMT1;2-dependent apoplastic pathway or from NH4+ entering the vasculature via the endodermal bypass ( Fig 5C ) . Without the presence of an endodermal bypass , the AMT1;3-dependent STP alone made no contribution to 15N allocation to shoots ( Fig 5D ) . We thus concluded that the AMT1;2-mediated ATP favors partitioning of NH4+-N to the shoots and contributes to shoot N provision particularly at elevated external supply . Since the occurrence of CSs and suberin lamellae show distinct developmental gradients along the root axis [3 , 26] , we compared their localization with that of the 2 investigated AMTs . The root zone below the 13th elongated cell down to the initiation of xylem cells , which is devoid of a functional CS or suberin deposition , represents an endodermal bypass even in WT plants ( Fig 6A and S8 Fig ) . AMT1;2 was absent from the root tip and present at the endodermis from the 11th to the 60th elongated cell ( Fig 6E–6H and S8A–s8D Fig ) . Notably , from the 30th cell onward , this transporter was also detected in cortical cells , while shifting completely to the cortex from the 60th cell onward . This shift in cell type–specific expression coincided with suberin formation and indicated that the presence of suberin in endodermal cells influenced cell type–specific AMT1;2 localization ( Fig 6A–6E and S8A–S8D Fig ) . By contrast , AMT1;3 expression started from the very root tip in epidermal cells and expanded toward the cortex from the 26th cell onward ( Fig 6I–6L and S8E–S8H Fig ) . Regarding the establishment of functional radial transport pathways , the localization pattern of the 2 NH4+ transporters and the apoplastic barriers allowed differentiating 4 zones along the axis of an Arabidopsis root ( Fig 7A ) . We then projected our estimates for the contribution of the individual radial pathways along the longitudinal axis of a WT root to build a model that estimates the relative contribution of different transport pathways to radial NH4+ transport in roots . In the high-affinity range , for which up to 80% of radial NH4+ transport depends on AMTs , the EB does not contribute to radial NH4+ transport along the whole root axis ( Figs 2B and 7B , and S2 Table ) . In “zone 1 , ” radial transport is completely dominated by the AMT1;3-dependent STP , which profits from the synergistic action of the EB . Once AMT1;2 is expressed , its contribution is rather modest ( approximately 16% ) , as it is negatively affected by the presence of an EB . When the CS becomes established ( zone 2 ) , the STP and ATP contribute by 45% and 34% , respectively ( Figs 2B and 7B , and S2 Table ) . With progressing suberization of the endodermis ( zone 3 + zone 4 ) , the estimated contribution of the ATP ceases and is taken over by a combined pathway , which is defined by the expression of both transporters in cortical cells mediating uptake of apoplastically transported NH4+ across the epidermal layer into the symplastic route . In the low-affinity range , the overall contribution of the AMT- and sgn3-dependent pathways is only 45%–65% , as NH4+ is transported additionally by not yet fully characterized low-affinity transporters ( Figs 2C and 7C , and S2 Table ) . In this case , the endodermal bypass contributes by 22% to radial NH4+ transport in “zone 1 . ” AMT1;3-mediated symplastic transport , in turn , contributes by 29% , albeit in a synergistic manner , with the endodermal bypass . In contrast , the contribution of AMT1;2 is strongly decreased by the endodermal bypass ( Fig 2C ) . In “zone 2 , ” AMT1;2-dependent apoplastic transport dominates radial NH4+ transport , with 30% over the 15% of AMT1;3-dependent symplastic transport . In “zone 3 , ” the contribution of the ATP decreases due to suberization of endodermal cells , and in “zone 4 , ” the contribution was taken over by a combined pathway . To validate our model , we exposed apical root tips of different length to a 15NH4+-containing agar plate compartment and correlated root lengths touching the tracer with counted elongated cell numbers ( Fig 8A and 8B ) . Samples were grouped into 4 classes , corresponding to the 4 zones with distinct combinations of transport pathways along the root axis ( Fig 7A ) . With this procedure , radial transport of 15NH4+ was integrated over longitudinal root segments with different sets of radial transport pathways . Longitudinal gradients of CS or suberin were neither affected by the sgn3 mutation nor by AMT1;3 expression ( S9 Fig ) . Shoot 15N accumulation after 4 h of labeling was related to root length and taken as readout for the radial transport rate of each root zone ( Fig 8C ) . In all lines , the highest radial transport rate was detected in the most apical part . The contribution of the STP ( tko+1;3 versus tko ) to radial transport , here represented by shoot accumulation of 15NH4+ , was particularly significant in “zone 1” ( <5 mm ) , where CS are not yet formed ( Figs 7A and 8C ) . The endodermal bypass alone ( tko sgn3 versus tko ) had no significant impact on shoot 15N accumulation . The comparison between tko+1;3 and tko sgn3+1;3 showed that loss of a functional CS increased AMT1;3-mediated symplastic transport capacity , especially in “zone 3” and “zone 4” ( Fig 8C ) , further indicating a synergistic effect between the STP and the EB . While the contribution of the ATP and STP to radial transport of nutrients through the root tissue has been adopted as a general principle in classical textbooks [6 , 7] , the quantitative share between these 2 pathways for radial transport of any nutrient has remained elusive . Here , we took advantage of the differential cell type–specific expression of AMT1–type transporters in combination with the CS-defective mutant sgn3 to dissect radial transport pathways for NH4+ and to determine their quantitative contribution . Thereby , we discovered previously unanticipated interactions between radial transport pathways and differential roles of ATP and STP for nutrient partitioning between roots and shoots . A prerequisite for the dissection of radial transport pathways is the availability of membrane transport proteins mediating either early substrate passage via the epidermis into the symplastic continuum or completing the apoplastic transport by transmembrane passage into endodermal cells . Unlike for most other nutrients , such transporters were known for NH4+ and previously characterized for their contribution to root uptake [12 , 16] . Assessing their contribution to radial NH4+ transport required a longer period of exposure to the tracer and revealed a higher capacity of the AMT1;3-mediated symplastic route at low external NH4+ supply but a higher capacity of the AMT1;2-mediated apoplastic route at elevated supply ( Fig 2B–2E ) . Although AMT1;2 is characterized by lower substrate affinity , i . e . , 234 μM relative to 61 μM for AMT1;3 [12] , better adapted biochemical transport properties to higher apoplastic NH4+ concentrations alone cannot explain its superior contribution to shoot NH4+ accumulation ( Fig 5 ) . AMT1;2-dependent NH4+ transfer to shoots was uncoupled from NH4+ accumulation in roots ( Fig 2E and S5D Fig ) because apoplastically transported NH4+ circumvents retention by root cells in favor of direct movement to the stele . Thus , the ATP may function as a “fast track” for nutrient delivery to the shoot . The reason for its greater importance at elevated external substrate supplies may lie in a lower number of low-affinity NH4+ transporters competing with AMT1;2 for NH4+ transport across the endodermis , whereas low-affinity NH4+ transporters in epidermal cells may be more abundant , masking the contribution of AMT1;3 . At least this holds true for AMT2;1 , which has been shown to localize in the epidermis and cortex of N-deficient roots , where it confers approximately 10%–25% of the low-affinity uptake capacity [13] . A supposedly differential contribution of these poorly characterized low-affinity transport systems for NH4+ in different cell types may restrict direct comparison of AMT1;2- versus AMT1;3-mediated radial transport capacities . In the high-affinity range , >90% of the NH4+ uptake capacity in roots relies on AMT1;1 , AMT1;2 , and AMT1;3 [12] . Respecting the compensatory increase in AMT1;3 capacity in the absence of AMT1;1 [12 , 24] , quantitative estimates of the share between apoplastic and symplastic pathways to radial transport in the high-affinity range are unlikely to be affected by other transport systems . It is important to note that the contribution of an extended endodermal bypass might change according to growth conditions , particularly to substrate concentration , root pressure , or transpiration rates . Therefore , we assessed 15N shoot accumulation during the light phase , when transpiration is high and root pressure is low , thereby suppressing the effect of root pressure on xylem transport rates . Such conditions allowed investigating the contribution of AMT1;2- and AMT1;3-dependent pathways in presence or absence of an extended EB more directly and at a higher resolution . In our experimental conditions , the ATP reached highest capacity when CSs were intact , and this capacity was as high as that mediated by the EB alone ( Fig 2C ) . We therefore conclude that the membrane transport steps required for xylem loading , i . e . , NH4+ transfer to pericycle cells and subsequent export to xylem vessels , are not limiting for root radial transport . Hence , a quantitative comparison between the ATP and STP for NH4+ , as addressed by our approach , appears valid . Apart from determining the contribution of individual pathways to radial NH4+ transport , we also focused on their interplay . This question is of particular biological relevance due to the longitudinal gradients of apoplastic barrier formation , which generate a suite of possible radial pathway interactions . At the primary root axis , an interplay between all 3 pathways is confined to a small zone , in which epidermal AMT1;3 is coexpressed with endodermal AMT1;2 , while CS or suberin are not yet formed ( Figs 6 and 7 ) . In the sgn3 mutant , the largely expanded zone with an endodermal bypass enhanced radial transport , as shown by higher normalized shoot accumulation of the apoplastic tracer Sr2+ ( Fig 1B and 1C ) , whereas EB of NH4+ was outcompeted by AMT-mediated radial transport pathways ( Figs 1D–1E and 2B–2C ) . Surprisingly , we found that the EB has an opposing role on the 2 radial transport pathways , which we confirmed in 3 independent methodological approaches , i . e . , by a ) examining tko and tko sgn3 lines with reconstituted expression of AMT1;2 or AMT1;3 in 5-week-old hydroponically grown plants ( Fig 2 ) , b ) in 8-d-old agar-grown plants ( Fig 3 ) , or c ) by circumventing the use of the sgn3 background with the help of a lignin biosynthesis inhibitor ( Fig 4 ) . A compromising action of the EB on the ATP occurs when NH4+ leaking out through a defective CS must be retrieved by AMT1;2 , i . e . , by the same transporter that is already engaged in the apoplastic transport route . Although the net contribution of an open EB was 0 at low external NH4+ supply , it promoted shoot 15N accumulation via the AMT1;3-dependent symplastic route ( Fig 2D ) . Presumably , part of NH4+ transported via the symplastic route was exported into the apoplast during the radial move and reached the vascular system by profiting from unhindered apoplastic diffusion through the endodermal cell layer . NH4+ efflux has been extensively characterized in physiological studies and shown to be dominated by the export of NH3 [27 , 28] . Currently , it is not clear which membrane transporters mediate NH3 export from root cells , but export across the plasma membrane may be facilitated through NH3-transporting aquaporins , as shown for tonoplast intrinsic proteins [29] or through AMT2;1 , which alters its localization toward inner root cells under NH4+ nutrition [13] and is able to permeate NH3 [30] . Such synergistic interplay between the STP and the EB may be regarded as experimental evidence for the so-called coupled transcellular transport pathway [1] . This third transport pathway has been postulated to couple repeated steps of symplastic and apoplastic transport based on the polarized localization of importers and exporters , as shown for silicon or boron [31–33] . The relative contribution of the individual radial pathways as well as their interplay is subject to root zonation and the formation of apoplastic barriers . As radial NH4+ transport via the AMT1;2-mediated ATP was decreased by an extended endodermal bypass in the most apical root zone and progressively sealed by suberin deposition in a shootward direction , an efficient ATP was restricted to a zone between 13 and approximately 67 cells above the first elongated root cell ( Figs 6 and 7 ) . Thus , the “fast track” for NH4+ delivery to the shoot is already in place as soon as the primary root explores new nutrient-rich soil layers and establishes an almost equal share of NH4+ partitioning between the shoot and the roots , since NH4+ provision to roots depends mainly on the STP ( Fig 5 ) . Since elevated substrate supply can push the relative share of NH4+ in favor of the shoot , apoplastic NH4+ transport may finally contribute to enhanced shoot growth under ample N supply . From a general perspective , this principle appears advantageous during root foraging , when soil nutrients are vertically translocated and accumulate in deeper soil layers [34] . However , it is noteworthy that the apoplastic transport route is highly sensitive to environmental conditions . Abiotic stress factors , including high salt or K deficiency , accelerate the suberization of endodermal cells via abscisic acid ( ABA ) and can restrict the zone of substrate-accessible endodermal transporters to less than 50% [35] . This may also be the reason why the suberin-enriched mutants esb1 and myb36 showed lower normalized shoot accumulation of 15NH4+ ( S3 Fig ) . Endodermal transporters directly impact the nutrient composition of the shoot and , in turn , may be subject to straight control by the nutritional status of the shoot . Thus , characterizing transport processes across the endodermal plasma membrane in both directions and quantifying the contribution of the ATP versus the STP is highly relevant . Such knowledge might help specify breeding targets in plants , enabling more selective nutrient translocation to shoots , e . g . , to meet the increasing N demand of shoots growing under elevated carbon dioxide ( CO2 ) , decreasing shoot accumulation of xenobiotics , or fortifying seeds with essential mineral elements .
tko plants ( amt1;1amt1;2amt1;3 ) were obtained from a selfed F2 population after backcrossing a homozygous amt1;1-1amt1;3-1amt2;1-1amt1;2–1 quadruple insertion line [12] to WT Col-0 . tko and sgn3;3 ( in Col-0 background , SALK_043282 ) [9] were crossed and selfed to obtain an F2 population . Within the F2 population , homozygous lines of tko+AMT1;2 ( tko+1;2 ) , tko+AMT1;3 ( tko+1;3 ) , tko sgn3 , tko sgn3+AMT1;2 ( tko sgn3+1;2 ) , and tko sgn3+AMT1;3 ( tko sgn3+1;3 ) were selected by PCR . For details of primers for genotyping , see S3 Table . Homozygous tko and sgn3;3 from this population served as reference lines . All homozygous mutant lines were selfed , and seeds from F3 generations were used in all experiments . In this work , 1 biological replicate represents a sample consisting of 2 plants from the same line or treatment ( except for xylem sap , for which 4 plants were combined as 1 sample ) . The number of biological replicates indicated in the figure legends represents the number of independent biological replicates ( originating from same line/treatment but different pot or agar plate ) . Plants were grown hydroponically for 5 weeks , as previously described [12 , 16] . Nitrate was used as the sole N source for preculture in all experiments . The climate chamber had a day–night regime of 10 h ( 22 °C ) /14 h ( 18 °C ) , 200 μmol m−2 s−1 light intensity , and 70% humidity . Prior to labeling , plants were precultured for 3 d on an N-free nutrient solution . Just before 15N or Sr2+ labeling , plant roots were rinsed in 1 mM CaSO4 for 1 min to remove nutrients from the apoplast and then transferred to the N-free solution containing different concentrations of ( 15NH4 ) 2SO4 ( 98 atom% 15N ) or SrCl2 . After 1 h , roots were washed again in 1 mM CaSO4 for 1 min to remove tracers from the apoplast . Shoots and roots were harvested , freeze dried , and ground to fine powder . To harvest xylem sap samples , plants were decapitated below the rosette , and xylem sap from 4 plants was combined in 1 sample . 15N concentrations were determined by isotope-ratio mass spectrometry ( DELTAplus XP , Thermo-Finnigan ) . Other mineral elements were determined by high-resolution inductively coupled plasma mass spectrometry ( Element 2 , Thermo Fisher Scientific ) . Sr was determined by high-resolution inductively coupled plasma mass spectrometry ( ICP-MS , Element 2 , Thermo Fisher Scientific ) . Normalized shoot accumulation was calculated by normalizing total Sr2+ or 15N shoot content to root dry weight and labeling time . The formula used is as follows: 15N accumulation rate in shoot ( or xylem exudate ) = ( sample 15N abundance − natural 15N abundance ) × total N concentration × shoot ( or xylem exudate ) dry weight/ ( 15N molecular weight × 15N purity × labeled time × root dry weight ) . Natural 15N abundance was obtained by measuring unlabeled leaf material . For harvest , 2 plants were taken as 1 sample in 4 replicates for each data point . Normalized shoot accumulation in different lines was compared in 3 independent experiments . Construction of promoterAMT1;2:ORF:GFP and promoterAMT1;3:ORF:GFP lines were described previously [12 , 16] . Plants were precultured on a one-half Murashige and Skoog ( MS ) medium containing 1 mM NO3- as the sole N source , 2 . 5 mM 2- ( N-morpholino ) ethanesulfonic acid ( MES ) ( pH 5 . 7 ) , 1% sucrose , and 0 . 8% Difco agar ( Becton Dickinson ) . Five-d-old seedlings were transferred to the N-free one-half MS medium for 3 d . N-deficient plants were used for visualization of GFP and root barriers . For visualizing propidium iodide ( PI ) penetration , plants were stained with PI ( 10 μg mL-1 ) for 10 min in the dark [3 , 36] . PI-dependent red fluorescence and AMT:GFP–dependent green fluorescence in different root zones were observed with a confocal laser scanning microscope Zeiss LSM 780 ( Carl Zeiss ) . Excitation/emission wavelengths of 488 nm/490–540 nm and 561 nm/650–710 nm were used for detection of GFP signals and PI signals , respectively . Cross sections from each zone were reconstructed from z-stacks of longitudinal confocal sections . For visualizing suberin , plants were stained with Fluorol Yellow 088 , as described previously [3] , and observed with a conventional light microscope ( Zeiss Axio Imager M2 , Carl Zeiss ) using a GFP filter . For quantification , “onset of elongation” was defined as the zone where an endodermal cell length was more than twice its width [3] . More than 20 plants were counted . Experiments were repeated at least 3 times . Arabidopsis seeds were surface sterilized and precultured on one-half MS medium containing 1 mM NO3- as the sole N source , 2 . 5 mM MES ( pH 5 . 7 ) , 1% sucrose , and 0 . 8% Difco agar ( Becton Dickinson ) . Five-d-old seedlings were transferred to one-half MS medium without N and supplemented with 2 . 5 mM MES ( pH 5 . 7 ) , 1% sucrose , and 0 . 8% Difco agar . After 3 d , plants with similar root size were transferred to horizontally split agar plates containing the 15N tracer in the bottom segment . To prepare the 15N-labeled agar plates , 50 ml one-half MS medium without N , supplemented with 2 . 5 mM MES ( pH 5 . 7 ) and 1% Difco agar , was spread on the agar in a 12-cm square Petri dish . No sucrose was added in order to slow down the root growth . A trench of 5-mm width was cut in the solidified agar to avoid 15N diffusion between the upper and lower agar segments . ( 15NH4 ) 2SO4 ( 98 atom% 15N ) stock solution was spread on the lower agar segment to reach a final 15NH4+ concentration of 4 mM . Petri dishes were placed in vertical orientation in a growth cabinet at a day–night regime of 10 h ( 22 °C ) /14 h ( 19 °C ) and a light intensity of 120 μmol m-2 s-1 during the day period . In each treatment , 10 plants were placed on the agar plate in a way that approximately 2 mm , 5 mm , 10 mm , or 15 mm of the apical primary root zone touched the 15N-containing bottom agar segment , while the shoot was placed on the upper segment ( -N agar ) . Eight-d-old seedlings were used for this 15N transport assay to avoid interference from lateral roots and to make sure that only a predefined root portion was in contact with the tracer . For each line , 30 plants were placed on 3 independent plates to normalize the variation on different agar plates . After 4 h of incubation , shoots were separated at the hypocotyl and dried separately in tin capsules . To measure the shoot 15N concentration in each plant , whole shoots ( about 40 μg dry weight ) were mixed with BBOT isotope reference standard ( atom percent: 0 . 366 , N percent: 6 . 5 , HEKAtech GmbH ) in order to reach the detection limit for total N . Shoot 15N concentration was determined by isotope-ratio mass spectrometry ( DELTAplus XP , Thermo-Finnigan ) . Labeled root segments were cut off , mounted on slides , and scanned by an Epson Expression 10000XL scanner ( Seiko , Epson ) . Root lengths were determined by Smartroot software V4 . 126 , and the number of cells from the start of cell elongation onward was counted under differential interference contrast using a conventional light microscope ( Zeiss Axio Imager M2 , Carl Zeiss ) . Since the roots grew nearly 2 mm during this 4-h period , final labeled root length was grouped into classes of <5 mm , <10 mm , <15 mm , and <20 mm . For each root length class , 6–9 plants from 3 agar plates were analyzed to yield 1 data point . Col-0 plants were precultured on one-half MS agar with 1 mM NO3- for 5 d and then transferred for 3 d to agar plates without N . During the last 48 h of growth on the N-free medium , part of the plants were transferred to the same medium containing either 0 . 05% DMSO ( control ) or 10 μM of the lignin biosynthesis inhibitor piperonylic acid ( PA ) . To verify the presence of functional CS , roots were stained with PI ( 10 μg mL-1 ) for 10 min in the dark . For shoot 15N analysis , tko , tko+1;2 , and tko+1;3 plants were precultured as described above and then either exposed to 0 . 05% DMSO or 10 μM PA for 48 h before transferring to horizontally split agar plates . Since PA treatment inhibited CS formation up to 9 mm , precisely 7 mm of the apical root tips were exposed to 4 mM 15NH4+ . After 6 h , shoots were collected for 15N analysis , as described above . Total RNA was extracted using the QIAzol Lysis reagent ( Qiagen ) following the manufacturer’s instructions . Prior to cDNA synthesis , samples were treated with DNase ( Thermo Fisher Scientific ) . Reverse transcription was performed using SuperScript II ( Thermo Fisher Scientific ) reverse transcriptase and Oligo ( dT ) 12-18 . Real-time PCR was performed using a Mastercycler ep realplex ( Eppendorf ) and QuantiTect SYBR Green qPCR mix ( Qiagen ) using the primers listed in S3 Table . Relative transcript levels were calculated according to Vandesompele and colleagues ( 2002 ) [37] .
|
Radial transport of nutrients from the soil to the vascular system of plant roots occurs via the symplastic transport pathway ( STP ) and apoplastic transport pathway ( ATP ) . Nutrients move along the STP when crossing the plasma membrane of outer cells and moving to xylem through the cytoplasmic continuum formed by plasmodesmata . Nutrients following the ATP , in turn , initially move passively through the extracellular space but are eventually taken up by endodermal cells , in which Casparian strips ( CSs ) prevent further apoplastic movement . We assessed the contribution of these transport pathways to radial transport in roots and nutrient provision to shoots by expressing cell type–specific ammonium transporters in a CS-defective mutant . Our study reveals that i ) symplastic transport is more efficient at low external ammonium supply; ii ) when endodermal cells become sealed by the deposition of suberin lamellae , the expression of ammonium transporters shifts to cortical cells; and iii ) apoplastic transport depends on a functional apoplastic barrier at the endodermis , favoring nitrogen ( N ) partitioning to shoots at high external ammonium .
|
[
"Abstract",
"Introduction",
"Results",
"Materials",
"and",
"methods"
] |
[
"plant",
"anatomy",
"chemical",
"compounds",
"apoplastic",
"space",
"vascular",
"bundles",
"plant",
"physiology",
"nitrates",
"xylem",
"plant",
"science",
"biological",
"transport",
"cellular",
"structures",
"and",
"organelles",
"proteins",
"endodermis",
"chemistry",
"leaves",
"iodides",
"cell",
"membranes",
"biochemistry",
"cell",
"biology",
"biology",
"and",
"life",
"sciences",
"transmembrane",
"transport",
"proteins",
"physical",
"sciences",
"metabolism"
] |
2018
|
Root zone–specific localization of AMTs determines ammonium transport pathways and nitrogen allocation to shoots
|
The World Health Organization’s 2020 Goals for Chagas disease include access to antiparasitic treatment and care of all infected/ill patients . Policy makers need to know the economic value of identifying and treating patients earlier . However , the economic value of earlier treatment to cure and prevent the Chagas’ spread remains unknown . We expanded our existing Chagas disease transmission model to include identification and treatment of Chagas disease patients . We linked this to a clinical and economic model that translated chronic Chagas disease cases into health and economic outcomes . We evaluated the impact and economic outcomes ( costs , cost-effectiveness , cost-benefit ) of identifying and treating different percentages of patients in the acute and indeterminate disease states in a 2 , 000-person village in Yucatan , Mexico . In the absence of early treatment , 50 acute and 22 new chronic cases occurred over 50 years . Identifying and treating patients in the acute stage averted 0 . 5–5 . 4 acute cases , 0 . 6–5 . 5 chronic cases , and 0 . 6–10 . 8 disability-adjusted life years ( DALYs ) , saving $694-$7 , 419 and $6 , 976-$79 , 950 from the third-party payer and societal perspectives , respectively . Treating in the indeterminate stage averted 2 . 2–4 . 9 acute cases , 6 . 1–12 . 8 chronic cases , and 11 . 7–31 . 1 DALYs , saving $7 , 666-$21 , 938 from the third-party payer perspective and $90 , 530-$243 , 068 from the societal perspective . Treating patients in both stages averted ≤9 acute cases and ≤15 chronic cases . Identifying and treating patients early was always economically dominant compared to no treatment . Identifying and treating patients earlier resulted in a cumulative cost-benefit of $7 , 273-$224 , 981 at the current cost of identification and treatment . Even when identifying and treating as little as 5% of cases annually , treating Chagas cases in the acute and indeterminate stages reduces transmission and provides economic and health benefits . This supports the need for improved diagnostics and access to safe and effective treatment .
While previous studies have evaluated the economic value of earlier treatment of Chagas disease ( caused by the protozoan parasite Trypanosoma cruzi ) in individuals , none have considered the spread or how it may prevent the spread of Chagas disease . Studies have estimated the cost of treating Chagas patients in various stages of disease[1 , 2] and shown that the cost of Chagas disease is lower when treated in the acute stage[1]; however , these studies did not evaluate the cost-effectiveness of treatment nor did they consider the impact on the spread of T . cruzi . Another study has evaluated the cost-effectiveness ( measured in cost per quality-adjusted life year gained ) of vector control strategies plus drug treatment[3] , but misses the broader effects of treatment ( i . e . , reduction in transmission ) . Treating patients in the earlier stages of infection can cure Chagas disease and is associated with a higher efficacy than in the chronic stage[4] , which would lead to a reduction in each of the various forms of T . cruzi transmission ( e . g . , vectorial , congenital , and transfusional ) . Therefore , previous reports on the economic value of earlier treatment may underestimate the benefits of earlier treatment . While the World Health Organization’s ( WHO ) London Declaration 2020 Goals for Chagas disease cite “100% of countries with access to antiparasitic treatment” and “100% of infected/ill patients under care”[5] , it is currently estimated that <1% of those infected with Chagas have access to care and treatment[6] . Given these 2020 goals , policy makers may want to know the economic value of identifying and treating patients earlier in the disease course to inform programs aimed at increasing access to care and treatment . Thus , the question remains , what is the economic value of earlier treatment of Chagas disease when considering transmission ? Therefore , we modified our previously published T . cruzi transmission model[7] ( calibrated to simulate a village in Yucatán , Mexico ) and linked an economic outcomes model to evaluate the economic benefits ( e . g . , cost-effectiveness , cost-benefit ) of treating Chagas cases in the acute and indeterminate stages .
Our compartmental simulation model ( outlined in Fig 1 ) includes vector-borne , congenital , and transfusional , forms of T . cruzi transmission . Briefly , it represents vector and host populations involved in T . cruzi transmission and included triatomines , human hosts , non-human hosts ( i . e . , dogs ) , and dead-end hosts ( i . e . , chickens ) to simulate vector-borne transmission between these populations in both domestic and peridomestic settings , as well as congenital and transfusional transmission . During each time step ( i . e . , t = 1 month or 30 days; chosen as a balance between the shorter acute stage and longer indeterminate and chronic stages and to be consistent with other models ) , fixed fractions and rates determined the number of individuals in each compartment . Each member of the human population could be in any of the following mutually exclusive disease states: susceptible ( not infected with T . cruzi and able to become infected ) , acute Chagas disease ( infected with T . cruzi and able to transmit , exhibiting mild and nonspecific symptoms , but in some cases can show Romaña’s sign or be serious and life-threatening , and having microscopically detectable parasitemia for 1 . 5 to 2 months ) , indeterminate Chagas disease ( asymptomatically infected with T . cruzi , able to transmit ) , and chronic Chagas disease ( infected with T . cruzi , able to transmit , and showing symptoms such as cardiomyopathy and/or megaviscera ) . Those not developing symptomatic chronic disease remained in the intermediate stage . Upon a feeding contact by an infectious triatomine , a susceptible human had a probability of becoming infected with T . cruzi via contamination with bug feces during or immediately after the feeding [represented in the vector-borne force of infection ( FOI ) ] . Pregnant women had a probability of transmitting Chagas to their infants upon birth , with newborns becoming infected based on the congenital FOI . Additionally , a proportion of humans receiving a blood transfusion or organ transplant had a probability of becoming infected with T . cruzi , based on the transfusional FOI . Those in the acute and symptomatic chronic states of disease had probabilities of Chagas-related mortality . Triatomine bugs could be susceptible ( not infected and able to become infected ) or infectious ( infected with T . cruzi and able to transmit to vertebrate hosts ) . Upon feeding on an infectious host , a susceptible bug had a probability of becoming infected with T . cruzi , conditional on the disease state of the host . Dogs served as reservoir hosts for T . cruzi and could be either susceptible or infected , with a susceptible dog becoming infected upon the bite of an infected vector based on the FOI . Dogs were considered competent transmitters and susceptible triatomines could become infected by an infected dog . Chickens served as dead end hosts , unable to become infected or further transmit T . cruzi . [12] Our model included transmission in both domestic and peridomestic habitats , which vary by vector-host contact rates , and allowed for the movement of triatomines between habitats . Vectorial transmission was governed by the vectorial FOI . Consistent with other models of vector-borne diseases[13] , this is a function of: ( 1 ) the triatomine biting rate , ( 2 ) the triatomine feeding proportion for each host type in each habitat , ( 3 ) the probability of transmission from vector to susceptible host , ( 4 ) the probability of transmission from infected host to susceptible bug , and ( 5 ) the proportion of infected hosts in each habitat . Transmission probabilities from vector to host varied with host species , while triatomine biting rates were assumed to be constant , regardless of host species . During each time step , a proportion of the human population were identified and completed treatment with a standard course of benznidazole ( 5mg/kg/day for 60 days ) . Those treated had a probability of moving back to the susceptible state based on the efficacy of benznidazole , which was disease state-specific . Those unsuccessfully treated remain infected with T . cruzi and progress though the model . As antiparasitic treatment is only indicated once , those unsuccessfully treated were not eligible for additional antiparasitic treatment . In the absence of treatment , we assumed that once infected , persons were always infected . Our economic model translated the number of cases identified and treated and number of new chronic cases generated from the transmission model into costs and health effects . The model included the cost associated with identification , diagnosis , treatment , treatment adverse events , and chronic disease costs and measured health effects in disability-adjusted life years ( DALYs ) . Each individual completing treatment accrued the cost of identification , diagnosis , and treatment . Each individual also had a probability of adverse events and those who experienced adverse events accrued the cost of antihistamines , as dermatitis is the most common adverse event . [14] Case identification included the costs for overhead ( e . g . , materials , transport , telecommunications ) , personnel field work , which includes operational costs such as obtaining blood samples and patient data ( estimated from personnel field time and hourly wage ) , and laboratory personnel for specimen processing , diagnostics and confirmation testing . Total diagnostic costs included a two-step process with a recombinant antigen enzyme-linked immunosorbent assay ( ELISA ) and chemiluminescense tests . Total treatment cost was determined from the dosage they received over the 60-day treatment course ( i . e . , cost per dose multiplied by dosage based on the individual’s weight ) . Each chronic case from the transmission model accrued costs and health effects by multiplying each case by the annual cost and DALYs accrued per chronic case for the remainder of their lifetime . This was done by cumulating monthly cases to annual cases . For this cost and DALYs lost per case , we used our previously published Markov model , which provided the annual costs and DALYs accrued , specific to Latin America and discounted to net present value ( NPV ) with a 3% discount rate . [15] This model included costs and probabilities for diagnosis , treatment , and monitoring for chronic cases with and without cardiomyopathy and/or megaviscera , as well as procedures such as pacemaker implant and surgery for megacolon . DALYs were calculated as the years lost due to disability ( YLD ) and the years of life lost ( YLL ) as a result of Chagas-related mortality and included health-related outcomes such as cardiomyopathy , congestive heart failure , and megaviscera . The Markov model converted DALYs into productivity losses , thus representing indirect costs for the duration of disease and for early mortality . Our economic model determined cost from the third-party payer and societal perspectives . The third-party payer perspective included direct costs ( i . e . , treatment and healthcare costs ) and the societal perspective included direct plus indirect ( i . e . , productivity losses ) costs . For each scenario , we calculated both its cost-benefit and incremental cost-effectiveness ratio ( ICER ) , as follows: Cost‑Benefit=Benefit–Cost=DirectCostandProductivityLossesofAvertedChronicCases–CostofTreatment ICER=CostEarlierTreatment–CostNoEarlierTreatmentDALYsNoEarlierTreatment–DALYsEarlierTreatment Results are in net present value ( NPV ) , with all past and future costs and future DALYs discounted to 2018 $US using a 3% discount rate . ICERs were considered highly cost-effective if less than Mexico’s gross domestic product ( GDP ) per capita ( $8 , 709[16] ) , cost-effective if 1 to 3 times the GDP , and not cost-effective if >3 times the GDP . Table 1 lists the model input parameters , values , and sources ( with transmission model inputs adjusted to a monthly time step ) . The model is of a rural village ( 2 , 000 persons ) in Yucatán , Mexico and in the absence of treatment , it was calibrated to assume a median T . cruzi prevalence value of 32 . 5% in T . dimidiata[17–25] , and seroprevalence estimates of 1 . 85% in humans[25–35] , and 14 . 58% in dogs[17 , 28 , 29 , 36–39] . We assumed the impact of treatment in the chronic stage was negligible as it has limited effectiveness and takes years for sero-reversion to occur . As transmission probabilities and T . dimidiata feeding proportions across host species are highly variable and/or not well defined in the literature , these parameters were calibrated to available empirical data for the Yucatán . Efficacy and cost data were Mexico-specific when available and came from the scientific literature . Patient weight came from Mexico’s National Health and Nutrition survey . [40]
In the absence of treatment in the acute and indeterminate stages , Chagas prevalence was maintained at 1 . 8% in humans , with 50 new acute cases ( i . e . , transmission events ) and 22 new chronic cases over 50 years; Fig 2 shows the number of new acute cases over time . Table 2 shows the number of new chronic cases and annual costs and DALYs accrued over time . Overall , chronic cases accrued a total lifetime cost of $44 , 955 ( 95% UI: $27 , 856–65 , 574 ) from the third-party payer perspective , $444 , 483 ( 95% UI: $404 , 803–479 , 798 ) from the societal perspective , and accrued 55 . 8 ( 95% UI: 45 . 2–64 . 5 ) DALYs . This translates to $450 and $4 , 444 per 1 , 000 person-years from the third-party payer and societal perspectives , respectively . Using the cost per case from Ramsey et al . , societal costs totaled $1 . 8 million ( 95% UI: $1 . 6–1 . 9 million ) assuming undiagnosed Chagas disease and $2 . 7 million ( 95% UI: $2 . 5–2 . 9 million ) assuming all chronic cases are diagnosed and treated . Fig 2A shows how treating Chagas in the acute stage impacts transmission over time , with treatment averting 0 . 5 to 5 . 4 new acute cases per 2 , 000 population over 50 years ( 1 . 0–10 . 8% relative reduction compared to no treatment ) , when 5% to 100% of cases detected and completed treatment annually . Table 2 shows the clinical and economic outcomes over time when different percentages of acute stage cases were identified and completed treatment annually . While all scenarios accrued costs in year 1 , cost-savings began to manifest by year 10 . Identifying and treating Chagas in the acute stage resulted in a 2 . 7–25 . 0% relative reduction in new chronic cases ( 0 . 6 to 5 . 5 new cases ) over 50 years ( Table 2 ) . Identifying and treating in the acute stage ( 5% to 100% of cases annually ) resulted in cost-savings totaling $694 to $7 , 419 from the third-party payer perspective , $6 , 976 to $79 , 950 from the societal perspective , and averting 0 . 6 to 10 . 8 DALYs over the lifetime of all chronic cases occurring over the 50-year period . These NPV lifetime costs were $375 to $443 per 1 , 000 person-years from the third-party payer perspective and $3 , 645 to $4 , 374 per 1 , 000 person-years from the societal perspective . Assuming the cost per case from Ramsey et al . , total societal costs ranged from $1 . 5 to $1 . 8 million ( if chronic cases remain undiagnosed ) to $2 . 2 to $2 . 6 million ( assuming all chronic cases are diagnosed and treated ) , varying by the proportion ( 5% to 100% ) of acute cases identified and completing treatment annually . Total cost-savings ranged from $56 , 171 to $1 . 3 million , depending on the cost of the chronic case and the percent treated annually . Identifying and treating Chagas disease in the acute stage was economically dominant ( i . e . , saved costs and provided health benefits ) compared to no treatment from both perspectives . At $1 . 50 per 100mg of benznidazole , treating 5% of cases was cost-effective ( ICER $361/DALY averted ) and dominant when treating ≥10% of cases . Fig 3A shows the incremental cost and effectiveness of increasing the proportion of acute cases treated compared to baseline . Fig 4 shows the cumulative cost-benefit of earlier treatment over time for various treatment costs . At the current cost ( ~$63 , varying by weight ) , identifying and treating Chagas patients in the acute stage would generate a positive return by year 6 , resulting in a cumulative cost-benefit of $7 , 273 to $71 , 705 over 50 years ( treating 5% to 100% of acute patients annually; Fig 4A ) . With a $300 cost , positive returns started in year 10 ( Fig 4C ) . Identifying and treating Chagas disease in the indeterminate stage averted 2 . 2 to 4 . 9 new acute cases ( 4 . 4–9 . 8% relative reduction ) per 2 , 000 population ( 5% to 100% of cases were detected and completed treatment annually ) over 50 years ( Fig 2B ) . Identifying and treating in the indeterminate stage , 41 . 3 to 68 . 1 cases were treated ( treating 5% to 100% annually ) , averting 6 . 1 to 12 . 8 new chronic cases ( 27 . 7–58 . 2% relative reduction ) over 50 years ( Table 3 ) . Over the 50-year period , cost-savings totaled $7 , 666 to $21 , 938 from the third-party payer perspective , $90 , 530 to $243 , 068 from the societal perspective , while 11 . 7 to 31 . 1 DALYs were averted . These NPV lifetime costs were $230 to $372 and $2 , 013 to $3 , 539 per 1 , 000 person-years from the third-party payer and societal perspectives , respectively . Assuming the cost of an undiagnosed chronic case , early treatment in the indeterminate stage cost $0 . 8 to $1 . 4 million ( saving $1 . 3 to $1 . 9 million ) , while it cost $1 . 2 to $2 . 1 million assuming chronic cases are treated ( saving $0 . 6 to $1 . 5 million ) from the societal perspective . Identifying and treating in the indeterminate stage was economically dominant compared to no treatment for all strategies tested from both perspectives , even at a cost of $1 . 50 per 100mg . It was also more cost-effective than treating in the acute stage , so that treating 5% of indeterminate patients annually was dominant compared to identifying and treating 100% of acute patients . Fig 3B shows the incremental gains of increasing the proportion treated . At the current treatment cost ( Fig 4A ) , positive returns were seen by year 3 ( 50%-100% treated ) or year 4 ( 5%-20% treated ) , resulting in a total cost-benefit of $81 , 352 to $224 , 981 over 50 years . At $300 per person ( Fig 4C ) , positive returns started between year 6 ( treating 100% annually ) and year 8 ( treating 5% annually ) . Identifying and treating Chagas patients in the acute and indeterminate stages averted 2 . 7 to 9 . 0 new acute cases ( 5 . 4–18 . 0% relative reduction ) per 2 , 000 persons over 50 years . Table 4 shows the clinical and economic outcomes over time . Scenarios garnered cost-savings by year 5 ( Table 4 ) . Identifying and treating in both stages averted 6 . 3 to 15 . 0 new chronic cases ( 28 . 6–68 . 2% relative reduction ) over the 50-year period , saving a total NPV of $8 , 420 to $25 , 076 from the third-party payer perspective , $94 , 989 to $279 , 379 from the societal perspective , and 12 . 3 to 35 . 7 DALYs over the lifetime of all chronic cases that occurred during the 50-year period . The NPV lifetime costs per 1 , 000 person-years were $199 to $365 from the third-party payer perspective and $1 , 650 to $3 , 494 from the societal perspective . Earlier treatment in both stages garnered societal cost-savings up to $2 . 1 million ( undiagnosed chronic case cost and 100% early treatment annually ) or $1 . 7 million ( 100% chronic diagnosed and treated and 100% early treatment annually ) . Identifying and treating in both stages ( ≥5% annually ) was dominant compared to no treatment and compared to treating 100% of acute patients and a similar percent of indeterminate patients . The incremental gains were largest when treating both acute and indeterminate patients earlier ( Fig 3C ) . At the current cost per person , treatment yielded a positive return by year 3 , except when treating only 5% of cases annually ( positive by year 4 ) , with a total return of $84 , 390 to $253 , 850 by year 50 . At $100 per person , positive returns took 4 to 5 years to manifest , totaling up to $251 , 797 over 50 years; while at $300 per person , positive returns were seen between years 6 and 9 , totaling up to $240 , 581 over 50 years .
All models are simplifications of real life and as such cannot represent every possible event or outcome . Our current model is deterministic in nature and does not include the full heterogeneity possible for Chagas disease transitions between states . Our model inputs were fit to disparate data of varying quality yet can be refined as new data become available . As Chagas disease is underdiagnosed and underreported , our estimates for T . cruzi seroprevalence are subject to limitations; however , we used the best available data for these parameters . Our results only account for those patients completing treatment; thus , our 100% completing treatment scenarios optimistically assume that no one discontinues treatment . We did not account for any costs for those Chagas cases that did not complete treatment as treatment discontinuation increases with age , is more frequent in chronic disease patients , and tends to be better tolerated by younger patients . [14 , 57] Also , the reasons for discontinuing treatment vary , making it difficult to determine the associated costs ( e . g . , rare or more severe side effects would lead to higher costs than treatment non-compliance ) . We assumed acute cases did not seek care and did not include these costs , while some cases are more severe and life-threatening ( estimated to occur in 1–5% of patients ) , the vast majority are minor and unrecognized . [4 , 58] Additionally , we did not include the cost of routine care and monitoring of indeterminate cases which may occur . However , our cost-benefit analysis shows that economic returns are garnered even with a high treatment cost per person , which could factor in these other costs . Treating Chagas cases in the acute and indeterminate stages , reduces transmission and new chronic cases ( up to 18 . 0% and 68 . 2% relative reduction in transmission events and chronic cases in the Yucatán , respectively ) , provides health benefits , and would result in cost-savings within a few years , even when identifying and treating as little as 5% of cases annually . This supports the need for improved diagnostics and access to safe and effective treatment for earlier treatment of Chagas disease .
|
The World Health Organization’s 2020 Goals for Chagas disease include access to antiparasitic treatment and care of all infected/ill patients . Policy makers need to know the economic value of identifying and treating patients earlier and what can be invested . We evaluated the impact and economic outcomes ( costs , cost-effectiveness , cost-benefit ) of identifying and treating different percentages of Chagas patients in the acute and indeterminate disease states using a transmission model linked to a clinical and economic outcomes model . Identifying and treating Chagas cases in the acute and indeterminate stages could result in up to an 18 . 0% relative reduction in transmission events as well as a 68 . 2% relative reduction in new chronic cases over 50 years compared to no treatment and was always economically dominant compared to no treatment . Identifying and treating Chagas disease in its earlier stages would reduce transmission and result in better health outcomes and cost-savings . In fact , the cost-savings would outweigh the cost of treating , meaning that earlier treatment may pay for itself . This supports the need for improved diagnostics and access to effective treatment .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"death",
"rates",
"medicine",
"and",
"health",
"sciences",
"tropical",
"diseases",
"vector-borne",
"diseases",
"social",
"sciences",
"geographical",
"locations",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"health",
"care",
"north",
"america",
"protozoans",
"neglected",
"tropical",
"diseases",
"population",
"biology",
"infectious",
"diseases",
"health",
"economics",
"protozoan",
"infections",
"economics",
"disease",
"vectors",
"people",
"and",
"places",
"trypanosoma",
"cruzi",
"population",
"metrics",
"mexico",
"trypanosoma",
"chagas",
"disease",
"eukaryota",
"biology",
"and",
"life",
"sciences",
"species",
"interactions",
"organisms"
] |
2018
|
The economic value of identifying and treating Chagas disease patients earlier and the impact on Trypanosoma cruzi transmission
|
Novel technologies that include recombinant pathogens and rapid detection methods are contributing to the development of drugs for neglected diseases . Recently , the results from the first high throughput screening ( HTS ) to test compounds for activity against Trypanosoma cruzi trypomastigote infection of host cells were reported . We have selected 23 compounds from the hits of this HTS , which were reported to have high anti-trypanosomal activity and low toxicity to host cells . These compounds were highly purified and their structures confirmed by HPLC/mass spectrometry . The compounds were tested in vitro , where about half of them confirmed the anti-T . cruzi activity reported in the HTS , with IC50 values lower than 5 µM . We have also adapted a rapid assay to test anti-T . cruzi compounds in vivo using mice infected with transgenic T . cruzi expressing luciferase as a model for acute infection . The compounds that were active in vitro were also tested in vivo using this assay , where we found two related compounds with a similar structure and low in vitro IC50 values ( 0 . 11 and 0 . 07 µM ) that reduce T . cruzi infection in the mouse model more than 90% after five days of treatment . Our findings evidence the benefits of novel technologies , such as HTS , for the drug discovery pathway of neglected diseases , but also caution about the need to confirm the results in vitro . We also show how rapid methods of in vivo screening based in luciferase-expressing parasites can be very useful to prioritize compounds early in the chain of development .
It is estimated that around 100 million people live with the risk of infection with T . cruzi in endemic areas in Latin America , with approximately 8 million already infected . The considerable influx of immigrants from Latin American countries to USA , Canada and Europe has also made Chagas disease an important health issue in these countries [1] . Although Chagas disease was discovered more than one hundred of years ago , the medicines available for treatment have serious drawbacks . The two drugs current in use , Benznidazole and Nifurtimox that were released in the 70's , present toxic side effects and low efficacy in some strains [2] . It was believed that both of them were only efficient for the treatment of the acute phase but recent studies suggest that chagasic patients in the chronic phase of the disease treated with Benznidazole show reduced disease progression and increased negative seroconversion than the untreated patients [3] . In an advanced position in the pipeline for future anti-T . cruzi treatments there is only Posaconazole , an oral antifungal that is currently in the market and has been tested successfully in mice [4] and humans [5] infected with T . cruzi . Given the limitations of the current available treatments and the low number of candidates undergoing clinical tests , the development of new anti-T . cruzi compounds combining broad and high efficacy with low toxicity is an urgent need . About ten years ago , the advent of high-throughput screening ( HTS ) technology revolutionized the process of early drug development , enabling researchers to rapidly collect enormous amounts of data and explore compound libraries with unprecedented thoroughness . Even if this technology has not yielded the expected increase in the number of licencesed medicines in the market , it still is considered a fundamental tool in early drug development in the pharmaceutical industry [6] . Additional developments in the field of drug discovery include luminescent reporter gene assays , which appear as the most prominent type of reporter gene assay used in biomolecular and pharmaceutical development laboratories . The success of these techniques is due to the high signal associated with luciferases , which makes them ideal for high throughput screening ( HTS ) in vitro applications , but also for the possibility of adapting these assays for in vivo screening [7] . Major changes are being introduced in the field of Chagas disease drug discovery since the development of recombinant T . cruzi parasites to be used as tools for drug screening . The first example is a transgenic T . cruzi strain expressing the reporter enzyme β-galactosidase [8] that has allowed performing a HTS for compounds active against T . cruzi infection of host cells ( Pubchem AID:1885 ) . Screening of drugs in T . cruzi mouse models has also been made much more rapid and simple with the use of fluorescent [9] or luminescent [10] recombinant parasites . Recombinant parasites expressing luciferase are already available for several species and have been used effectively for drug discovery in Leishmania [11] , [12] . In this work we describe the continuation of a chemical HTS against T . cruzi trypomastigote infection of host cells . Re-testing of some of the HTS hits for in vitro anti-T . cruzi activity revealed that approximately half of them did not confirm the activity . Screening of the active compounds in a mouse model of acute Chagas resulted in the finding of one molecular structure with high anti-trypanosomal activity in mice .
Animal studies were approved by the Institutional Animal Care and Use Committee of New York University School of Medicine ( protocol #81213 ) . This protocol adheres to the guidelines of the Association For Assessment and Accreditation Of Laboratory Animal Care International ( AAALAC ) . The compounds were selected from a HTS campaign performed by the Broad Institute , as part of the MLPCN ( Molecular Libraries Probe Centers Network ) T . cruzi inhibition project . The results from a HTS for T . cruzi trypomastigote infection of host cells were made available at Pubchem ( AID: 1885 ) . This HTS was performed by screening of 303 , 286 molecules ( the NIH collection ) form where 4 , 065 hits were selected by their activity against T . cruzi trypomastigote infection . These compounds were further assayed to determine their IC50 ( Pubchem AID: 2044 ) and their toxicity to host NIH-3T3 cells ( Pubchem AID: 2010 ) . Compounds were selected from the hits of this HTS among the ones with reported IC50<1 . 2 µM and at least 100 fold activity versus toxicity All the compounds selected for this analysis had toxicity activity >60 µM . Chromatographic analyses were performed to determine the degree of purification ( all compounds were >90% pure except for CID-563075 and CID-2234099 that were 87 and 88% pure , respectively ) . Electrospray ionization mass spectrometry was performed to confirm compound identification . Finally , compounds were dissolved in DMSO at 10 mM concentration . LLC-MK2 and NIH/3T3 cells were cultivated in DMEM supplemented with 10% FBS , 100 U/ml penicillin , 0 . 1 mg/ml streptomycin , and 0 . 292 mg/ml glutamine ( Pen-Strep-Glut ) at 37°C and 5% CO2 atmosphere . T . cruzi parasites from the Tulahuen strain stably expressing the β-gal gene ( clone C4 ) [8] and T . cruzi Y strain expressing the firefly Luciferase gene were kept in culture by infection of LLC-MK2 every 5 or 6 days in DMEM with 2% FBS and 1% Pen-Strep-Glut at 37°C and 5% CO2 atmosphere . Trypomastigotes forms were released in the supernatant of infected LLC-Mk2 and harvested between days 5 and 7 . The harvested medium was centrifuged for 7 min at 1 , 237 g and , in order to eliminate the amastigotes , the trypomastigotes forms were allowed to swim out of the pellet for at least 3 h . The parasites were counted in a Neubauer Chamber and 10 million trypomastigotes were used to infect 1 million LLC-MK2 cells plated in a 75 cm2 culture flask . Between 5 to 7 days after the infection , NIH/3T3 cells and T . cruzi Tulahuen expressing β-galactosidase [8] were harvested , centrifuged and washed with DMEM without phenol red supplemented with 2% FBS and Pen-Strep-Glut . The phenol red needed to be eliminated in order to avoid interference with the assay absorbance readings at 590 nM . NIH/3T3 cells ( 50 , 000 per well ) were seeded in 96-well plates 2 h before addition of purified T . cruzi trypomastigotes ( 50 , 000 per well ) and the compounds for testing at the maximum concentration of 50 µM , therefore the DMSO percentage was never higher than 0 . 5% . This concentration of DMSO was tested repeatedly and it does not affect the viability of the parasites . Each determination was performed in duplicate . Amphotericin B ( Sigma-Aldrich ) was used as positive control at a final concentration of 4 µM . Negative and positive controls were carried in every plate . After 4 days , 50 µl of PBS containing 0 . 5% of the detergent NP40 and 100 µM Chlorophenol Red-β-D-galactoside ( CPRG ) ( Sigma ) were added per well . Plates were incubated at 37°C for 4 h and absorbance was read at 590 nm using a Tecan Spectra Mini plate reader . The absorbance obtained was proportional to the viability of the parasite . The value of IC50 was determined using Graph Prism Software . The firefly luciferase gene ( luc ) was used to replace the GFP in the T . cruzi episomal expression vector , pTREX-GFP [13] , a modified version of pRIBOTEX-GFP [14] . T . cruzi ( Y strain ) epimastigotes maintained at 28°C in LIT medium were transfected with 10 µg of pTREX-luc using a nucleofector transfection system ( T-cell protocol; AMAXA ) and selected with 200 µg/ml G418 for 4 weeks . Parallel transfections with pTREX-GFP demonstrated that under similar selection conditions >95% of parasites are strongly positive for GFP after 4 weeks ( not shown ) . Mammalian-infective metacyclic trypomastigotes were harvested from stationary phase epimastigote cultures and enriched following passage over DEAE-cellulose/PBS pH . 8 . 0 as routinely performed [15] . Tissue culture trypomastigotes were harvested from monkey kidney epithelial cells , LLcMK2 , monolayers infected with Y-luc metacyclic trypomastigotes . Relative luciferase activity in Y strain epimastigotes grown in the presence of 200 µg/ml G418 and in mammalian infective trypomastigotes harvested from infected monolayers after the second passage through mammalian cells ( ie . 2 weeks in the absence of drug selection ) was similar ∼2 . 5×106 RLU/106 parasites . Trypomastigotes forms from transgenic T . cruzi Y strain expressing firefly Luciferase were purified , diluted in PBS and injected i . p . in Balb/c mice ( 105 trypomastigotes per mouse ) . Three days after infection the mice were anesthesized by either i . p . injection of 300 mg/kg of Xylazine and 3500 mg/kg of Ketamine or by inhalation of isofluorane ( controlled flow of 1 . 5% isofluorane in air was administered through a nose cone via a gas anesthesia system ) . Mice were injected with 150 mg/kg of D-Luciferin Potassium-salt ( Goldbio ) dissolved in PBS . Mice were imaged 5 to 10 min after injection of luciferin with an IVIS 100 ( Xenogen , Alameda , CA ) and the data acquisition and analysis were performed with the software LivingImage ( Xenogen ) . One day later ( 4 days after infection ) treatment with compounds at 5 mg/kg/day or vehicle control ( DMSO in PBS ) was started by i . p . injection in groups of 5 mice and continued daily for the indicated number of days . On the days indicated , mice were imaged again after anesthesia and injection of luciferin as described above . Parasite index is calculated as the ratio of parasite levels in treated mice compared to the control group and is multiplied by 100 . The ratio of parasite levels is calculated for each animal dividing the luciferase signal one day after the end of the 5 day treatment ( day 9 of infection ) by the luciferase signal one day before the beginning of treatment ( day 3 of infection ) . The compound CID-12402750 was selected for this assay due to its activity against T . cruzi in vivo . NIH-3T3 cells plated on coverslips were infected with T . cruzi Tulahuen expressing β-galactosidase and incubated with or without drug at 5 , 10 , 50 or 100 times the value of the IC50 obtained in the in vitro assay ( IC50 = 0 . 11 µM ) . After 3 days , they were fixed with 4% of paraformaldehyde , rinsed with PBS , permeabilized for 15 min in PBS with 0 . 1% Triton X-100 ( Sigma-Aldrich ) and blocked for 20 min in PBS with 10% goat serum , 1% bovine serum albumin , 100 mM glycine and 0 . 05% sodium azide . The cells were incubated for 1 h at room temperature with a polyclonal rabbit anti-T . cruzi at 1∶2 , 000 dilution . After rinsing , they were incubated for 1 h at a 1∶800 dilution with an Alexa Fluor® 488 goat anti-rabbit IgG secondary antibody ( Molecular Probes , Invitrogen ) . DAPI was used to stain the DNA and the coverslips were mounted on Mowiol . Cells were analyzed using an inverted Olympus IX70 microscope with a 60× oil objective . Data were analyzed using Prism ( v . 4 . 0c , GraphPad ) . t-test was performed . Statistics were considered significant if P<0 . 05 or P<0 . 01 , as indicated .
The results of the first high throughput screening ( HTS ) to identify molecules effective against T . cruzi trypomastigote infection of host cells were used to select 23 compounds with reported IC50<1 . 2 µM and at least 100 fold activity versus toxicity for further analysis . The quality control of these compounds was made by HPLC/MASS . We first tested the activity of the 23 selected compounds against T . cruzi trypomastigote infection of host cells using a similar assay and the same parasite and host cells that were used in the HTS ( see methods ) . Our tests showed a higher value for the IC50 for the majority of the compounds when compared to the ones reported in the HTS ( Pubchem , AID: 2044 ) , with 11 compounds showing no detectable activity against T . cruzi ( Fig . 1 ) . We then selected all compounds that showed activity in our in vitro assay for testing of anti-T . cruzi activity in mice ( except for CID 1473168 , which was not available ) . For this purpose , we adapted a rapid method for drug testing in mouse models using recombinant T . cruzi expressing the firefly luciferase gene in an episomal expression vector . We generated a recombinant T . cruzi expressing luciferase ( Y-luc ) , which presented good infectivity and stability . We find that peak parasitemia was comparable for both WT and Y-luc parasites ( data not shown ) and that Y-luc trypomastigotes harvested from blood on day 7 of infection exhibited comparable levels of luciferase expression as epimastigotes maintained on drug selection or trypomastigotes that were used to inoculate the mice ( Fig . 2 ) . Given that T . cruzi amastigotes divide every 12 h and the intracellular infection cycle is 4–5 days , we estimate that these in vivo passaged Y-luc parasites were free from drug selection for at least 30 generations ( when including time to generate trypomastigotes in culture ) [11] . Stable expression of luciferase from pTREX-luc for a minimum of 7 days in vivo gives us an adequate window of time to assess the effects of small molecule inhibitors on acute T . cruzi infection in vivo . Parasite loads were measured at different days after infection by injecting luciferin , the substrate of luciferase , followed by imaging and quantification of the luminescence signal with an IVIS Lumina imager . Focusing on the area of highest intensity signal ( red and turquoise ) , it is clear that luminescent T . cruzi are concentrated in the intraperitoneal cavity ( site of injection ) ( Fig . 3A ) . Following the signal for 10 days post-inoculation demonstrates that there is a clear indication of parasite migration from the injection site ( peritoneal cavity ) to distal sites , perhaps spleen and liver ( Fig . 3A ) . Using this recombinant parasite , we infected two groups of five Balb/c mice and followed the course of infection over 13 days . To determine whether this method would be useful for testing of drugs , one of the groups was treated with benznidazole , while control group was injected with vehicle control . A reduced signal was obtained in the group treated with benznidazole ( Fig . 3A , B ) . Even if variation between individual animals is high , as expected in this type of in vivo experiments , values between groups are significantly different after only two days of treatment and maintain different levels of infection for the 6 days of treatment . We then used a modification of this protocol with quantification of the parasite loads only at days 3 and 9 after infection , which corresponds to five days of treatment ( Fig . 4A ) to test the activity of the eleven compounds selected from the in vitro assay ( Fig . 1 ) . We found that treatment with some of the compounds had no activity on parasite levels ( index close to 100 ) and others even resulted in increased parasite loads ( index higher than 100 ) , possibly because they interfere with the immune response of the mice . However , two of the compounds tested , CID-24892493 and CID-12402750 , resulted in severe decreases in the levels of T . cruzi in mice that were significantly different from their control group ( Fig . 4B , C ) . No toxic effects were apparent on the mice on visual observation . These two compounds are closely related , they belong to the 1- ( 4-Halogeno-benzyl ) -2 , 4 , 6-triphenyl-pyridinium series and are differing only in the nature of halogen on the para position of the benzyl ( Fluorine for CID-24892493 and Chlorine for CID-12402750 ) . To get a better understanding of the anti-T . cruzi effect observed , we next determined whether compound CID-12402750 could inhibit T . cruzi replication within mammalian host cells . We infected cells for 2 h , rinsed away the remaining free trypomastigotes and , after adding the compound at concentrations between the IC5 and the IC100 , we incubated cells for 3 days to allow for amastigote proliferation . In control cells , amastigotes homogenous in size were distributed throughout the cytoplasm of the host cells ( Fig . 5A ) . Treatment with CID-12402750 resulted in infected cells containing only a few amastigotes of average size ( Fig . 5B , C ) , suggesting that this compound interferes with proliferation of amastigotes .
The use of novel pharmaceutical technologies for neglected diseases is opening new possibilities for drug development in this area . As an example of this , the first HTS performed for Chagas disease ( Pubchem AID: 1885 ) represents a major advance in this field . However , our results illustrate the need for confirmation of the HTS results , since as much as 11 hits from the HTS out of 23 selected did not show activity against T . cruzi in our hands . Since the same parasite and host cells and were used for the HTS and for testing in our laboratory , and the screening assay is also very similar ( see methods ) , it is likely that the reason for the discrepancies resides in the chemical compounds used for analysis . The quality control of the compounds used in our laboratory was checked by HPLC followed by mass spectrometry and therefore we can be confident that the chemical identity and the purity of the compounds was optimal . Another recent advance in the drug development field is the development of new methods for screening of compounds in animal models . Testing of compounds for activity in mice was always considered a labor intensive and expensive step in the chain of pre-clinical drug development and therefore was left at the end of the time line . The availability of sensitive imaging techniques and transgenic parasites , either fluorescent or luminescent [9] , [10] and the Y-luc parasite described here , allow for rapid testing of relatively high number of compounds . Compared to traditional methods that require bleeding of infected mice and counting parasites in a haemocytometer , intraperitoneal injection of the luciferase substrate and imaging requires considerably less time , with the additional advantage that there is no manipulation of infected blood . Detection of parasites expressing luciferase is also more sensitive than conventional counting of parasites in blood samples . In our model of infection , Balb/c mice infected with T . cruzi Y strain , we are not able to detect parasitemia by manual counting in peripheral blood at any time after infection with 105 parasites , but inoculation of the same amount of Y-luc parasites allows to follow the course of infection ( Fig . 4 ) . The stability of the Y-luc parasite and the sensitivity of detection allows for screening of compounds early after in vitro results , accelerating the speed of pre-clinical drug discovery . Probably , this method will also be useful in detection of parasitological cure after drug treatments . This is normally achieved by administering an immunosuppressive treatment to mice after drug treatment , when parasites are no longer detectable . If parasites that were not eliminated during the drug treatment emerge after immunosuppression , it is expected that the luminescence signal will still be detectable . In our project , all the compounds with in vitro activity were tested in mice , where we found only one chemical structure with significant in vivo activity . It is well known that there is a significant attrition rate when compounds are tested in animal models , even if optimization based on pharmacokinetics parameters has been performed . In our case , direct testing without optimization probably reduces the chances of success , but at the same time , this strategy provides a rapid method to find compounds with activity in vivo , which places them in an advanced position in the development chain . Pharmacokinetic analysis can then be performed to optimize compounds accordingly , which combined with testing of anti-parasitic activity in vivo in every round , could lead to an accelerated drug discovery path . We have found a chemical series , 1-benzyl-2 , 4 , 6-triphenylpyridin-1-ium , with strong in vitro and in vivo activity against T . cruzi . The two related compounds , with either a Cl or a F in para position of the benzyl , had same level of anti-trypanosomal activity , confirming the efficacy of this structure . We also found that the anti-T . cruzi effect is mediated by the inhibition of proliferation of amastigotes within host cells . Despite this promising results , the development of this structure as an anti-trypanosomal drug , may be impaired by the quaternary ammonium , that is generally known as having low intestinal permeability by passive diffusion . Additionally , symmetry , planarity and the presence of 4 phenyl rings could contribute to lower solubility , which would also have a negative impact on oral absorption . Further drug development should include optimization of solubility and permeability in vitro before additional tests in vivo could be performed .
|
Chagas is a devastating disease affecting about 100 million people in Latin America . The drugs available for treatment against the causative agent , the parasite Trypanosoma cruzi , have associated toxicity and are not completely effective against the chronic form of the disease , which is the most common presentation in the clinic . There is a great need for new drugs against this disease . Novel technologies in drug development are now being applied for the search of new compounds against Chagas . Taking advantage of a high throughput screening performed recently to identify compounds active against T . cruzi replication in host cells in vitro , we have selected 23 compounds , which have been re-tested to selected active ones . We have also adapted a transgenic T . cruzi expressing luciferase , which allows for direct visualization when mice are infected . These parasites have been used to establish a model for acute Chagas disease useful for drug testing in mice . Using this method , we have tested the activity of the selected compounds and found two compounds with strong anti-T . cruzi activity in mice .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"medicine",
"infectious",
"diseases",
"chagas",
"disease",
"neglected",
"tropical",
"diseases"
] |
2011
|
Activity In Vivo of Anti-Trypanosoma cruzi Compounds Selected from a High Throughput Screening
|
Sensory loss induces cross-modal plasticity , often resulting in altered performance in remaining sensory modalities . Whereas much is known about the macroscopic mechanisms underlying cross-modal plasticity , only scant information exists about its cellular and molecular underpinnings . We found that Caenorhabditis elegans nematodes deprived of a sense of body touch exhibit various changes in behavior , associated with other unimpaired senses . We focused on one such behavioral alteration , enhanced odor sensation , and sought to reveal the neuronal and molecular mechanisms that translate mechanosensory loss into improved olfactory acuity . To this end , we analyzed in mechanosensory mutants food-dependent locomotion patterns that are associated with olfactory responses and found changes that are consistent with enhanced olfaction . The altered locomotion could be reversed in adults by optogenetic stimulation of the touch receptor ( mechanosensory ) neurons . Furthermore , we revealed that the enhanced odor response is related to a strengthening of inhibitory AWC→AIY synaptic transmission in the olfactory circuit . Consistently , inserting in this circuit an engineered electrical synapse that diminishes AWC inhibition of AIY counteracted the locomotion changes in touch-deficient mutants . We found that this cross-modal signaling between the mechanosensory and olfactory circuits is mediated by neuropeptides , one of which we identified as FLP-20 . Our results indicate that under normal function , ongoing touch receptor neuron activation evokes FLP-20 release , suppressing synaptic communication and thus dampening odor sensation . In contrast , in the absence of mechanosensory input , FLP-20 signaling is reduced , synaptic suppression is released , and this enables enhanced olfactory acuity; these changes are long lasting and do not represent ongoing modulation , as revealed by optogenetic experiments . Our work adds to a growing literature on the roles of neuropeptides in cross-modal signaling , by showing how activity-dependent neuropeptide signaling leads to specific cross-modal plastic changes in neural circuit connectivity , enhancing sensory performance .
Sensory loss often elicits cross-modal plasticity , either enhancing or reducing the performance of remaining unimpaired sensory modalities . These effects have been broadly described in humans and other mammals [1 , 2] and exemplify the remarkable plasticity and adaptability of the brain . What drives cross-modal plasticity and how this influences sensory performance has been mainly addressed at the macroscopic level of entire brain structures [3 , 4] . For example , it has been shown that in the blind , the visual cortex is recruited to process various auditory features [5 , 6] , and at the same time the auditory cortex may expand its tonotopic area [7] or exhibit changes in its functional responses [8] . Such system-wide neuroplasticity might stem directly from the silencing of neurons and neural circuits associated with the dysfunctional sensory modality , leading , for example , to reduced competition for neural targets [9] . Additionally , cross-modal plasticity might result also from an increased use-dependent plasticity [10] of the remaining functioning senses , or from an increase in the attention directed towards them [11] . These types of plasticity do not require activity-dependent signaling between modalities . Whereas cross-modal plasticity has been largely studied at the system level of entire brain regions , much less is known about its cellular and molecular underpinnings . Recent work is just beginning to address this . For example , two recent studies have revealed prominent strengthening of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid ( AMPA ) receptor-mediated synaptic transmission in pyramidal neurons of layer 2/3 barrel cortex of visually-deprived rats [11 , 12] . These changes depended on long-distance serotonin signaling , presumably originating from the raphe nuclei [11] . Conversely , sensory deprivation of neonatal mice was shown to down-regulate oxytocin neuropeptide secretion from the hypothalamus , resulting in decreases in synaptic transmission in sensory cortical regions associated with nondeprived sensory modalities [13] . Both mechanisms rely on long-distance signaling . Interestingly , recently , long-distance neuropeptide and hormone signaling from one sensory modality was shown to modulate concurrent sensory responses in another modality in C . elegans [14–16] . We thus asked whether , in C . elegans , long-range signaling also leads to long-lasting plastic changes in sensory acuity . For this purpose , we examined C . elegans mechanosensory ( Mec ) mutants lacking a sense of gentle touch to the body , and observed various changes in behaviors that depend on remaining senses . One notable alteration was enhanced chemosensation , an increased response to low concentrations of certain attractive odors . This suggests that cross-modal sensory compensation following sensory loss is a basic and conserved feature of the nervous system not limited to the complex brains of mammals . In particular , we were able to identify a specific synapse in the chemosensory circuit that is modulated by neuropeptide secretion from the Mec touch receptor neurons ( TRNs ) to tune chemosensory performance and to thus implement a form of cross-modal compensatory plasticity in C . elegans . To our knowledge , such neuropeptide-mediated synaptic plasticity has not been described before in C . elegans .
In order to identify cross-modal plasticity following sensory loss in C . elegans , we focused on the Mec-deficient mutant , mec-4 ( u253 ) , which lacks functional MEC-4 , a DEG/ENaC channel subunit exclusively expressed in the TRNs and necessary for sensing gentle touch to the body ( Fig 1A ) [17–19] . We first examined the mec-4 response to nose touch [20] and found that , even though body and nose touch are mediated by distinct sensory neurons and mechanoreceptors [21] , loss of body touch leads to reduced nose touch ( Fig 1B ) . We also tested mec-4 ( u253 ) chemosensation [22] by performing a chemotaxis assay using the attractive odorant benzaldehyde ( Bz ) . We found that specifically at low Bz concentrations ( 1:10 , 000 ) , the Mec mutants were more proficient in navigating towards the odor source than wild-type worms , as indicated by their higher chemotaxis index ( Fig 1C ) . Bz is sensed by the AWC chemosensory neurons , which are also sensitive to isoamyl alcohol ( IAA ) , another volatile attractant [23] . We found that similarly to Bz , chemotaxis to low concentrations of IAA was enhanced in mec-4 worms ( Fig 1D ) . In contrast , chemotaxis to diacetyl ( DA ) and pyrazine ( Py ) , both of which are sensed by the AWA chemosensory neurons [23] , was attenuated in mec-4 ( Fig 1D ) . These data illustrate two forms of cross-modal plasticity following sensory loss in C . elegans , one enhancing ( Bz and IAA chemotaxis ) and one reducing ( nose touch , DA and Py chemotaxis ) remaining sensory responses . In the present study , we focus on cross-modal sensory enhancement , as exhibited by an augmented AWC-mediated olfactory acuity in touch-deficient worms . C . elegans chemotaxis is based on a biased random walk mechanism , whereby forward motion is interspersed with randomly occurring reorienting reversals and turns , whose frequency varies as a function of odor concentration [24 , 25] . This produces a net movement up chemical gradients towards the source of an attractant . As a consequence , animals removed from their food transiently increase their reversing rate compared to animals on food , since they sense a decrease in odor concentration . Thus , an alternative assay for olfaction consists of measuring the increased reversing rate of animals just removed from their source of food . This response , which like chemotaxis itself has been shown to be controlled by the AWC olfactory circuit [26–28] , thus can serve as a more sensitive measure for analyzing chemotaxis in individuals or small groups of animals [28] . We thus compared the frequency of reversing off and on food between wild-type and Mec mutants . For these experiments we tested two independent mec-4 alleles with full or partial loss of body touch sensitivity , mec-4 ( u253 ) and mec-4 ( e1339 ) , respectively , as well as a mec-10 ( e1515 ) mutant , with dysfunctional MEC-10 , a subunit like MEC-4 of the DEG/ENaC mechanosensory channel complex [29] . The off-food reversing rate of all Mec mutants was significantly higher than that of wild type ( Fig 1E ) and significantly lower than wild type on-food reversing rate ( Fig 1F ) , suggesting that locomotion of Mec mutants is indeed more sharply tuned to the presence or absence of food cues , as sensed by the AWC olfactory circuit [28 , 30] . In addition to chemosensory and olfactory neurons , food is sensed also by a group of dopaminergic mechanosensory neurons , which are distinct from the TRNs , and mediate a slowing in speed in the presence of food [31] . We tested whether the food-dependent changes in reversing rate in Mec mutants ( Fig 1E and 1F ) might be associated with improper mechanical food sensing , but found normal slowing on food in mec-4 mutants ( Fig 1G ) , ruling out this possibility . We asked whether the altered reversing rate of mec-4 mutants is associated with reduced TRN activity . We used an optogenetic approach to address this question . We artificially activated the TRNs of mec-4 and tested whether this manipulation would reduce reversing rate back towards normal . We did this by expressing Channelrhodopsin2 ( ChR2 ) specifically in the TRNs of mec-4 worms using the mec-4 promoter and stimulated the TRNs with random flashes of blue light for a period of 80 min . Since C . elegans must be fed all-trans retinal ( ATR ) in order for ChR2 to be activated by light [32] , we compared the post flashing reversing rate of adult worms that had or had not been fed with ATR and found a significant decrease in reversing rate 2 h , but not immediately , after photo-stimulation in worms exposed to ATR ( Fig 1H ) . This result indicates that the cross-modal change observed in Mec mutants is enduring and depends on the history of TRN activity rather than on ongoing or recent activity . It also demonstrates that this form of cross-modal plasticity does not depend on developmental effects , since it was readily reversible in adults . We next sought to identify the neural mechanisms underlying cross-modal compensatory behavior between touch and smell ( olfaction ) , by comparing the activity patterns of neurons involved in the mechanosensory or AWC-associated olfactory networks in wild type relative to mec-4 ( u253 ) mutants . To this end , we performed calcium-imaging experiments using microfluidic devices [33] . We first considered a subset of premotor interneurons , AVA and AVE , which control , in part , reversing behavior [34–36] , and that have direct synaptic connections with the touch receptor neurons ( Fig 2A ) [17 , 37] . Previous work has shown that spontaneous calcium transients in these neurons correspond to spontaneous reversing behavior [33 , 38] . We thus expected mec-4 ( u253 ) AVA/AVE neurons to exhibit enhanced activity in the absence of food compared to wild type , if they are involved in increasing reversing behavior . We found , however , no differences between AVA/E wild type and mec-4 ( u253 ) averaged spontaneous calcium transient traces , their amplitude , or frequency ( Fig 2B ) . Since the additional reversal-promoting premotor neuron pair , AVD , seems to be mainly involved in orchestrating touch-evoked withdrawals rather than spontaneous reversing [17 , 39 , 34] , we did not attempt to image spontaneous activity in this neuron . Next , we considered neurons involved in Bz chemotaxis ( Fig 2A ) . Bz , IAA , and other odorants are detected by the AWC pair of sensory neurons , which increase their activity as odor concentrations decrease [28] . We expected that if AWC neurons contribute to enhancing off-food reversing rate and chemotaxis following loss of mechanosensation by increasing their sensitivity , then they should show a larger response to reduced odor concentration in the Mec mutants . However , we found no significant difference in the AWC response to Bz ( 1:10 , 000 ) removal between wild type and mec-4 ( u253 ) animals ( Fig 2C ) . This was also true for a considerably lower concentration of Bz ( 1:10 , 000 , 000; Fig 2C , right ) , indicating that the similar AWC responses of wild type and mec-4 ( u253 ) mutants are not likely due to some limit in the AWC or calcium sensor dynamic range . The AWC sensory neurons make excitatory synaptic connections with a pair of interneurons , AIB , which promote reversing when active ( Fig 2A ) [25 , 28] . Thus , for AIB to be considered as a source for enhanced reversing and chemotaxis , it should display an enlarged response to odor removal in Mec-deficient worms compared to wild type . In effect , the initial calcium responses of AIB to Bz removal were similar in wild type and mec-4 ( u253 ) ( Fig 2D ) . Moreover , the AIB response in mec-4 ( u253 ) animals decayed more rapidly ( beginning approximately 5 sec after onset; Fig 2D ) , entailing overall reduced rather than enhanced AIB activity , which , if at all , should reduce and not enhance reversing . This delayed effect might be due to some negative feedback mechanism within the olfactory circuit , perhaps similar to other neuropeptide-dependent feedback loops already shown to act in this circuit [40] . The AWC odor-sensing neurons also make inhibitory synapses with the AIY interneuron pair ( Fig 2A ) [28] . Artificial inhibition of AIY activity has been directly shown to enhance reversing [41] . Thus , if the AIY neurons are involved in enhancing off-food reversing in Mec mutants , then they should respond with a larger inhibition following odor removal . Indeed , mec-4 ( u253 ) worms exhibited more prominent AIY negative calcium responses than wild type ( Fig 2E ) . The enlarged AIY inhibitory response in the Mec mutants ( Fig 2E ) on the one hand , and the similarity in the AWC and initial AIB responses between wild type and mec-4 ( u253 ) worms ( Fig 2C and 2D ) on the other hand , suggest together that the enhanced acuity of Mec-deficient worms to Bz might stem from potentiated AWC→AIY inhibitory transmission . Since there do not seem to be any direct synaptic connections between the TRNs and AWC or AIY ( Fig 2A ) , we hypothesized that the strengthening of AWC→AIY and the ensuing enhanced reversing rate in Mec-deficient worms might depend on neuropeptide signaling , which does not require synaptic contact between neurons . Indeed , TRN-specific RNAi knockdown [43] of EGL-3 , a proprotein convertase necessary for neuropeptide processing [44] , resulted in an increased off-food reversing frequency similar to TRN-specific RNAi knockdown of MEC-4 ( Fig 4A ) , supporting our hypothesis . To rule out the possibility that neuropeptide secretion might be necessary for mechanosensation itself , we compared the response to gentle body touch between wild type , mec-4 ( u253 ) , and egl-3 ( nr2090 ) . Only mec-4 ( u253 ) mutants displayed defective mechanosensation ( Fig 4B ) . Notably , the egl-3 ( nr2090 ) responses were relatively small in magnitude , which might reconcile our results with previous accounts of Mec deficiency in egl-3 mutants [44] . We also tested the effects of TRN-specific RNAi silencing [45] of EGL-21 , a carboxypeptidase required for neuropeptide processing [46 , 47] , on mec-4 ( u253 ) reversing rate 2 h post artificial ( optogenetic ) TRN activation . TRN photo-stimulation had a weaker effect on reversing rate ( relative to baseline: the same strain without stimulation ) upon TRN EGL-21 silencing compared to mec-4 ( u253 ) alone ( Fig 4C ) , indicating that neuropeptide signaling from the TRNs contributes significantly to TRN activity-dependent changes in reversing rate . We note that the residual reduction in reversing may be a result of incomplete knockdown of EGL-21 by the Pmec-4::egl-21 transgene . In order to establish the impact of mechanosensory loss on TRN neuropeptide secretion , we performed a neuropeptide secretion imaging assay [48] . We measured coelomocyte fluorescence in worms expressing an mCherry-tagged insulin-like peptide transgene ( INS-1 ) specifically in their TRNs . Since INS-1::mCherry is loaded together with all other neuropeptides in the cell into the same dense core vesicles , this assay does not segregate between different types of neuropeptides . mec-4 ( u253 ) mutants showed a reduced coelomocyte uptake of mCherry , implying a general decrease in neuropeptide secretion in mec-4 ( u253 ) mutants ( Fig 4D ) . Our results so far suggest that one or several neuropeptides expressed in the TRNs and processed by EGL-3 and EGL-21 convey the cross-modal plasticity observed following sensory loss . Recently , FLP-20 , an FMRFamide-related neuropeptide expressed in the TRNs [49] , has been shown to play a TRN-dependent role in mating behavior [50] and in short-term memory for mechanosensory habituation [51] . We wondered whether FLP-20 might also convey cross-modal plasticity following mechanosensory loss . To this end , we examined the reversing frequency of flp-20 ( pk1596 ) mutants , which harbor a deletion in their flp-20 coding sequence . We found that , similarly to Mec mutants , flp-20 worms show an increased reversing rate off-food ( Fig 5A ) . Moreover , no differences were found in reversing frequency between mec-4 ( u253 ) mutants alone and mec-4 ( u253 ) ; flp-20 ( pk1596 ) double mutants ( Fig 5A ) , suggesting that FLP-20 acts in the same pathway that produces enhanced reversing in mec-4 ( u253 ) . An additional allele , flp-20 ( ok2964 ) , displayed a similar increase in reversing compared to wild type ( Fig 5B ) . Notably , although the reversing rate of flp-20 mutants was higher than wild type , it was still lower than that of mec-4 mutants , suggesting perhaps that additional neuropeptides might be involved in modulating reversing frequency off-food ( Fig 5A ) . Transgenic expression of the FLP-20 transcript specifically in the TRNs reduced the enhanced flp-20 ( ok2964 ) reversing rate off-food ( Fig 5C ) . To test whether this change in reversing rate is food-dependent , we compared the reversing rate of flp-20 mutants and the TRN-specific rescue strain off-food and on-food ( Fig 5D ) . We found a significant interaction between genotype and food ( 2-way ANOVA , F ( 1 , 76 ) = 13 . 30 , p = 0 . 0005 ) , indicating that TRN secretion of FLP-20 is important for modulating reversing rate in a food-dependent manner . Together , these results are consistent with a model whereby FLP-20 released from the TRNs diminishes the tuning of reversing rate to the abundance of food odor concentration . To test whether elimination of FLP-20 specifically in the TRNs might increase reversing rate off-food , we constructed , using the Mos1 single-copy insertion ( MosSCI ) technique [52] , a flp-20 ( pk1596 ) rescue strain carrying single-copy integrated FLP-20 driven by the FLP-20 promoter . We eliminated expression exclusively in the TRNs through cell-specific excision of the FLP-20 rescue sequence , using FLP-recombinase ( see Materials and Methods ) . Reversing in this strain indeed increased following TRN-specific flp-20 excision ( Fig 5E ) , indicating that reduced FLP-20 signaling from the TRNs is sufficient to increase reversing rate . Conversely , overexpression of FLP-20 in the TRNs significantly suppressed the mec-4 ( u253 ) increase in reversing rate ( Fig 5F ) ; further indicating that FLP-20 is functionally released from the TRNs . It is noteworthy that the effectiveness of FLP-20 overexpression in mec-4 ( u253 ) mutants might suggest that the TRNs are still active , thus enabling neuropeptide release , even without mechanosensory input ( e . g . , due to spontaneous activity or input from other neurons ) . Further support for this is reported below ( Fig 5H ) . We also tested whether the recovery of reversing rate in mec-4 ( u253 ) mutants following artificial photo-activation of the TRNs depended on FLP-20 signaling . We indeed found that the reduction in reversing post TRN stimulation was smaller in mec-4 ( u253 ) ; flp-20 ( pk1596 ) double mutants than in mec-4 ( u253 ) single mutants ( Fig 5G ) , providing further evidence for the role of FLP-20 in modulating reversing rate as a function of TRN activity . Above , we showed that reduced TRN activity due to loss of mechanosensation is associated with an overall decrease in neuropeptide secretion from the TRNs ( Fig 4D ) . We wished to examine whether specifically FLP-20 signaling from the TRNs is reduced in Mec mutants . Recently , Laurent et al . introduced a new method for long-term monitoring of neurosecretion . The assay is based on the observation that reduced neurosecretion leads to reduced neuropeptide transcription , presumably via some feedback mechanism [53] . We thus measured ALM ( one of the TRNs ) and PVC ( an interneuron also expressing FLP-20 but not MEC-4 ) fluorescence intensity in worms with an integrated Pflp-20::GFP array [54] . ALM but not PVC fluorescence in mec-4 ( u253 ) mutants was significantly decreased compared to wild type ( Fig 5H ) , suggesting that attenuated TRN activity due to mechanosensory loss specifically reduces FLP-20 secretion from these neurons . We examined whether reduced TRN FLP-20 transcription was directly linked to diminished exocytosis . We did this by expressing tetanus toxin [55] in the TRNs , which cleaves synaptobrevin disrupting both clear and dense core vesicle release . This led to an even more pronounced decrease in Pflp-20::GFP fluorescence in ALM but not in PVC ( Fig 5H ) , implying that indeed reduced neurosecretion in ALM correlates with reduced FLP-20 transcription in ALM , and that in spite of the loss of mechanosensation there might still be residual secretion of FLP-20 from the TRNs , consistent with the effectiveness of FLP-20 TRN overexpression ( Fig 5F ) . To ensure that mechanosensory loss does not confer a general change in transcription , we measured the ALM YFP fluorescence intensity of Pmec-4 driven YC2 . 12 ( a calcium indicator composed of both YFP and CFP ) , in wild type and mec-4 ( u253 ) mutants treated with 0 . 01 M sodium azide ( to eliminate YC2 . 12-sensed calcium fluctuations in these neurons ) . We found no significant difference in YFP expression between them ( S1A Fig ) . To further examine the specificity of the mec-4 effect on FLP-20 release , we measured Pflp-20::GFP fluorescence in ASE neurons , which also express FLP-20 . Whereas mec-4 ( u253 ) - and TRN-expressed tetanus toxin had no effect on PVC fluorescence , surprisingly , both caused a decrease in Pflp-20::GFP fluorescence in the ASE neurons ( Fig 5H ) . This finding , which we leave for future investigation , suggests that the TRNs may modulate also ASE activity , and that in turn; in wild type worms the ASEs might relay and amplify FLP-20 signaling . As we have shown , FLP-20 signaling is correlated with TRN activity and can modify reversing rate . To determine whether it also affects AWC→AIY transmission , we measured calcium responses to Bz removal in wild type and flp-20 ( pk1596 ) mutants . As in mec-4 ( u253 ) mutants ( Fig 2C and 2E ) , flp-20 ( pk1596 ) responses were similar to wild type in AWC ( Fig 6A ) , but significantly enhanced in AIY ( Fig 6B ) . Conversely , artificially reducing AWC→AIY inhibitory transmission by inserting an electrical synapse between AWC and AIY was sufficient to reduce flp-20 ( pk1596 ) reversing rate ( Fig 6C ) , further confirming the link between FLP-20 , reversing rate , and AWC→AIY synaptic transmission . Finally , we compared chemotaxis to low concentrations of Bz between wild type and flp-20 ( pk1596 ) , and found a significant increase in the chemotaxis index of worms lacking functional FLP-20 ( Fig 6D ) , similar to the enhanced chemotaxis of Mec-deficient worms ( Fig 1C and 1D ) . Taken together , these results are consistent with a simple model , whereby in Mec-deficient worms reduced neuropeptide secretion from the TRNs , including FLP-20 , results in increased AWC→AIY inhibitory neurotransmission , leading to enhanced reversing off-food and AWC-dependent chemotaxis . Finally , we sought to identify the FLP-20 neuropeptide receptor , presumably acting in AIY to modulate AWC→AIY transmission . There is currently no known receptor for FLP-20 , but several putative G-Protein Coupled Receptors are prominent in the AIY transcriptome [56] . We screened 15 candidate receptors using an aequorin assay [57] ( see Materials and Methods ) but none of them responded to neither of three FLP-20 derived peptides: AMMRFamide , AVFRMamide and SVFRLamide ( Fig 7 ) . Thus , some other FLP-20 receptor , yet to be determined , might act directly in AIY , or indirectly through another neuron .
Cross-modal compensation for sensory loss is an intriguing adaptive capacity , which has been studied so far in complex mammalian brains . Here we have shown that it also occurs in a considerably simpler animal , C . elegans , and is thus perhaps a fundamental feature of any nervous system . Our results suggest that decreased neuropeptide ( including FLP-20 ) secretion from the TRNs in Mec mutants leads to an increase in the strength of inhibitory synaptic transmission between the AWC and AIY neurons in the olfactory circuit ( Fig 8A ) , resulting in enhanced coupling between reversing frequency and food odor abundance , and in general , in increased olfactory acuity to AWC-sensed odors . Several lines of evidence support a causative link between these effects . First , loss of mechanosensation , and overall elimination of FLP-20 secretion cause a nonadditive increase in reversing rate upon removal from food ( Fig 5A ) . Second , eliminating FLP-20 secretion exclusively from the TRNs , similarly , causes an increase in off-food reversing ( Fig 5E ) . Third , such increased reversing can be counteracted by artificially and specifically attenuating AWC→AIY transmission ( Fig 3D and 3E ) , demonstrating that changes in the strength of AWC→AIY transmission is sufficient to affect reversing rate . Moreover , reducing AWC→AIY transmission suppresses the increased reversal rates of mec-4 and flp-20 mutants ( Fig 3F and Fig 6C ) . Fourth , increased reversing off-food is a key component of chemotaxis towards food , enabling reorientation in search for food when sensing a drop in food odor concentration [26–28] . Finally , AWC→AIY transmission , reversing rate , and chemotaxis are all modulated in a correlated fashion by FLP-20 signaling . Parsimony argues that these effects are likely to be linked , rather than separate actions of FLP-20 released , for example , independently from different neurons . Our findings also hint at further possible components in this pathway . For example , additional TRN-secreted neuropeptides might be involved in signaling ( e . g . , Fig 5A ) ; FLP-20 signaling itself might be amplified , for instance , by the ASE neurons ( Fig 5H ) ; and it is still not clear whether FLP-20 impact on AWC→AIY transmission is direct or indirect . Recent studies have revealed a role for neuropeptide/hormonal signaling in cross-modal interactions in C . elegans during environmental stress [14 , 16] or developmental states of quiescence [58] . Our study indicates that in addition to conjoining simultaneous co-occurring inputs from different sensory modalities , neuropeptides can link between past mechanosensory experience and present chemosensory performance . Our findings also demonstrate that in addition to modulating sensory transduction at the sensory neuron level ( e . g . , tuning receptor strength ) , cross-modal neuropeptide signaling can also act at the circuit level , modifying the strength of synaptic transmission and downstream sensory processing . Furthermore , our study sheds light not only on the mechanisms of adaptation to sensory loss but also on normal sensory function , revealing an innate cross-network suppression mechanism . We speculate that cross-modal suppression by functional sensory neurons might serve to homeostatically limit to a manageable level the overall volume of sensory inputs that the nervous system receives , prioritizing diversity of sensory information over acuity of any one particular sensory modality . However , at the same time this mechanism is also flexible enough to enable the reweighting of sensory inputs in the event of sensory loss . Notably , the increased reversing rate in touch-deficient worms might have an additional advantage , as it effectively restricts their dispersal range , thus avoiding potential hazards that might otherwise be detected by a functional mechanosensory system . In addition to enhanced chemosensation of AWC-sensed odors , cross-modal plasticity following mechanosensory loss appears to impact a range of behaviors , including reduced sensory responses to nose touch and to AWA-sensed attractants ( Fig 1B and 1D ) . Whereas cross-modal enhanced sensory performance has obvious adaptive advantages , as it provides a form of compensation for unavailable sensory information , cross-modal reduced performance appears as maladaptive plasticity , whereby damage to one sensory modality propagates to diminish also other modalities . However , in some cases it might actually be beneficial . For example , worms withdrawing after being touched in the nose probably require body touch information to ensure that they don’t encounter another threat while reversing . When such information is permanently unavailable , it is perhaps more prudent to limit the response to nose touch as mechanosensory animals do ( Fig 1B ) . Uncovering the intricate mechanisms underlying these additional forms of cross-modal plasticity and perhaps also their significance as an adaptive or maladaptive response to sensory deprivation is an appealing direction for future research . In many ways the C . elegans cross-modal plasticity mechanism is analogous to cross-modal plasticity mechanisms underlying enhanced somatosensation in visually deprived rats ( Fig 8B ) [11 , 12] and reduced sensory processing in sensory-deprived neonatal mice ( Fig 8C ) [13] , whereby the loss of one sensory modality leads to long-distance signaling ( serotonin [11] or oxytocin [13] ) affecting a second sensory modality through experience-dependent modification of key glutamatergic synapses in the sensory circuit ( Fig 8 ) . Specifically , our work reveals that the source for cross-modal signaling in touch-deprived C . elegans is the touch sensory neurons themselves ( Fig 8A ) rather than regions downstream of the sensory neurons in sensory signaling pathways ( Fig 8B and 8C ) . Furthermore , it is reduced signaling ( Fig 8A ) rather than enhanced signaling ( Fig 8B and 8C ) that results in increased synaptic strength . Whereas the role of the sensory neurons themselves in this cross modal plasticity might be specific to the simple nervous system of C . elegans , the disinhibition by sensory deprivation , although not previously demonstrated , is likely to be conserved . Importantly , the various parallels that we have revealed between C . elegans and mice and rats support conservation of mechanisms identified in this study and the potential of using C . elegans for further research that will enhance our understanding of the molecular and cellular mechanisms governing cross-modal plasticity . Conversely , our findings generate specific predictions for mammalian systems . Namely , neuropeptides or other neuromodulators , originating from the sensory-deprived brain regions , might have a role in cross-modal plasticity both in normal and in sensory-deprived animals . Optogenetics has been suggested as an approach for treatment of various brain diseases and malfunctions [59 , 60] such as epilepsy [61] , stroke [62] , and blindness [63] . We have shown that random optogenetic activation of the TRNs of Mec-deficient worms ( Fig 8A , grey lightning bolt ) could counteract the changes in locomotion resulting from the loss of mechanosensation . We have also demonstrated that locomotion can be restored by an alternative strategy , synaptic engineering . Inserting an electrical synapse between AWC and AIY that modifies AWC→AIY transmission ( Fig 8A , grey dashed line ) , was sufficient to offset the enhanced reversing of Mec mutants . Thus , manipulating neuronal activity , by optogenetic stimulation , or artificially modifying synaptic transmission , by engineering new synapses into neural circuits , may potentially help recover deviations from normal circuit function .
Strains were grown and maintained under standard conditions at 20°C on nematode growth medium ( NGM ) 2% agar plates seeded with Escherichia coli strain OP50 . All experiments were conducted at 18°C–22°C . We found that higher temperatures considerably altered the results . Wild-type worms were Bristol variety N2 . The other strains used in this study are detailed in Table 1 . A standard assay for gentle body touch was applied [43 , 44] , whereby each worm was alternately touched five times anteriorly or posteriorly with an iris hair . For each worm , the number of anterior withdrawals was recorded . Chemotaxis assays were performed essentially as described [22] . However , we found that washing mec-4 ( u253 ) worms prior to the assay causes them to clump on the test plate . To overcome this , for Figs 1C and 6D , we manually transferred 20 worms to 6 cm unseeded nematode growth medium ( NGM ) test plates after releasing them from food on an empty plate . Prior to the assay , we placed 1 μL drop of 1% Bz diluted in ethanol on one side of the plate and a drop of ethanol alone on the opposite end . To each drop , 1 mM sodium azide was added for trapping the worms . This procedure was very noisy due to the relatively small number ( n = 20 ) of worms in each plate . For low odor dilutions in Fig 1D , we did wash the worms and place them in a 9 cm test plate , but waited 15 min , the time it took mec-4 ( u253 ) to unclump , then applied the odor and the ethanol , waited another 5 min , and applied the sodium azide . In both cases chemotaxis was scored >2 h later by counting the number of paralyzed worms within the field of view of a stereomicroscope centered at the odor spot , N ( odor ) , and the number of paralyzed worms within the field of view centered at the control spot , N ( control ) , and calculating the chemotaxis index ( CI ) [22] equal to: [N ( odor ) -N ( control ) ]/[N ( odor ) +N ( control ) ] . The reversing assay was performed as previously described [27] . A single worm was removed from food , allowed to crawl for a few seconds until no traces of food were visible in its track , and then transferred either to an empty 6 cm NGM plate ( off-food assay ) or to a 6 cm NGM plate seeded the day before with OP50 ( on-food assay ) . After 1 min , reversing events consisting of at least one body bend were counted over a 3 min period . Speed off and on food was measured as in [31] by counting the number of body bends over a period of 20 s in the absence or presence of OP50 bacteria . Worms were grown in the dark , and unless otherwise indicated , were fed OP50 bacteria mixed with ATR at 0 . 5 mM concentration . Just before blue light stimulation , 10 worms were transferred to a 3 cm plate whose lid was removed . For illumination , we used a Royal-Blue ( 447 . 5nm ) LUXEON SR-03-R0500 Rebel LED assembly attached to a Carclo 27° Frosted 20 mm Circular Beam Optic ( Part 10508; www . luxeonstar . com ) . The LED was controlled by an Arduino Uno R3 microcontroller ( www . adafruit . com ) using a Matlab ( Mathworks ) interface , which generated random blue light flashes for approximately 80 min each session . The interval between flashes was drawn from an exponential random distribution with a 10 s mean . The duration of each flash was drawn from a uniform distribution with a 3 s mean . Four to six worms were picked from the 3 cm plates either immediately ( Fig 1H ) , after no flashing at all ( Figs 3C , 4C and 5F ) , or 2 h after the end of the flashing session , and their reversing rate off food was measured . Data in Figs 4C and 5F are presented as the reversing rate after 2 h of TRN stimulation normalized by the average reversing rate without any stimulation for each experiment day . Imaging was performed on a Zeiss Axioscope upright microscope equipped with a Hamamatsu ORCA-ER digital camera . Worms were inserted into a microfluidic PDMS chip [33] , and the neuron of interest was observed through the chip's cover slip , using a 63X oil immersion objective . For imaging of spontaneous AVA/E activity , we used the “locomotion chip” [33] , in which the worm is trapped , but is free to make forward and backward undulating movements similar to those observed during locomotion . Since the AVA and AVE neurons are very close and difficult to separate , we imaged them together as one unit . For imaging calcium responses to Bz removal in AWC , AIB , and AIY , we used the “olfactory chip” [33] , which exposes the worm's nose to a constant flow of buffer ( S-basal without cholesterol ) or odor ( unless otherwise mentioned , 1:10 , 000 Bz diluted in buffer ) . Imaging commenced after 5 min of odor exposure . 10 s after the beginning of each recording , the odor was replaced with buffer and the imaging continued for an additional 30 s . Movies were captured and analyzed using custom written Matlab ( Mathworks ) programs . A rectangular region of interest ( ROI ) was drawn surrounding the cell body ( AVA/E , AWC , and AIB ) or the neurite ( AIY ) [28] , and for every frame the ROI was automatically shifted according to the new position of the center of mass ( in the case of the cell body ) or point of maximum intensity ( AIY neural process ) . The fluorescence intensity , F , was computed as the difference between the sum of pixel intensities and the faintest 10% pixels ( background ) within the ROI . For ratiometric imaging ( all neurons except for AIB ) , only ROIYFP was tracked , whereas ROICFP remained at a fixed offset from ROIYFP . The ratio , R , between FYFP and FCFP was then computed after correcting for bleed through . No correction for bleaching was necessary . Spontaneous calcium transients in AVA/E were detected automatically . All transients in a recording were aligned and averaged , and then all averaged traces were averaged between worms to obtain an overall mean trace . The traces in the olfactory imaging experiments depicting ΔF ( AIB ) or ΔR ( AWC and AIY ) , were computed as ( F − F0 ) / F0 * 100 , whereby F0 equals the average F within the first 3 s of recording . For statistical quantification , ΔF ( and similarly ΔR ) was computed as ( F1 − F0 ) / F0 * 100 , whereby F0 is the average F over T0 , the 10 s preceding odor switching , and F1 is the average F over T1 , the 10 sec following odor switching ( see Fig 2D ) . For AIB , ( F2 − F0 ) / F0 * 100 was computed as well , F2 being the average F over T2 , the last 10 s of the recording . All imaging strains showed normal reversing behavior . We used a strain previously described , AQ2614 [42] , containing an electrical synapse inserted between AWC and AIY . Briefly , the cDNA sequence of Mus musculus gap junction protein , delta 2 ( Gjd2 ) , was codon optimized to produce a synthetic Cx36 gene ( GeneArt ) . We fused to Cx36 an upstream promoter ( either Podr-1 for AWC [28] or Pttx-3 ( int2 ) , the second intron of the ttx-3 gene , for AIY [64] ) and a downstream gene encoding mCherry , and coinjected both plasmids to generate a transgenic worm carrying an extrachromosomal array . TRN-specific RNA interference ( RNAi ) by feeding [43] was performed as follows . Worms from the TU3568 strain , sid-1 ( pk3321 ) V; him-5 ( e1490 ) V; lin-15B ( n744 ) X; uIs71[Pmec-18::sid-1 , Pmyo-2::mCherry] , were fed with bacteria either carrying an empty vector ( control ) or producing double-stranded RNA against mec-4 ( Source BioScienceRNAi library , ID number X-7D15; primer pairs: forward , TACCTGCAACGGAAAGATCC and reverse , ATACAACGGAAAGACGCCAC ) or egl-3 ( Source BioScienceRNAi library , ID number V-7G01; primer pairs: forward , CATGAATGCATTACTCACTTGGA and reverse , CATATCTACTCTGCTTCATGGGG ) [65] . To prepare the bacteria , a single colony of each bacterial strain was shaken overnight at 25°C in LB medium containing 50 μg/ml ampicillin and 12 . 5 μg/ml tetracyclin . Then 25 ml of LB medium containing 50 μg/ml ampicillin were inoculated with 250 μl culture and shaken for 4–5 h at 37°C until a 595 nm absorbance of at least 0 . 8 was reached . Next , an additional 25 ml of LB medium containing 50 μg/ml ampicillin was added to the culture together with 1M IPTG . Culture was shaken at 37°C for an additional 4 h and then spun down . The precipitate was resuspended in 2 ml M9 buffer containing 10 μl 1M IPTG and distributed onto NGM plates supplemented with 25 μg/ml carbenicillin . TRN-specific RNA interference ( RNAi ) by transgenic transformation [45] was performed by coinjecting plasmids containing 738 base pair sense and antisense fragments from the coding region of the egl-21 gene fused to the mec-4 promoter into mec-4 ( u253 ) worms . To gauge overall neuropeptide secretion , we imaged the coelomocytes of wild type and mec-4 ( u253 ) mutants specifically expressing in their TRNs ( using a mec-4 promoter ) INS-1 fused to mCherry . Transgenic animals were picked at the L4 stage onto fresh plates and were grown for an additional 12–16 h before being treated with 0 . 01 M of the anesthetic sodium azide ( NaN3 ) and then imaged . We examined only animals oriented laterally on their right side and accordingly imaged ccDR , ccAR , and ccPR coelomocytes . Fluorescence intensity was determined after background subtraction using Metamorph software . As the difference between wild type and mec-4 coelomocyte fluorescence intensity was similar in anterior and posterior coelomocytes ( S1B Fig ) , we pooled them for analysis . To monitor FLP-20 transcription , we recorded fluorescence intensity in ALM , PVC , and ASE neurons of a Pflp-20::GFP integrated strain , similarly to the coelomocyte fluorescence measurement described above . The Pflp-20::GFP-integrated strain carrying the extrachromosomal array pl5Ex5[Pmec-4::tetX::SL2::mCherry] was imaged with both GFP and RFP channels . We analysed only animals expressing RFP in both ALM and PLM . We controlled mec-4 promoter transcriptional activity by measuring the fluorescence intensity of Pmec-4 driven YC2 . 12 in wild type and mec-4 ( u253 ) mutants treated with 0 . 01 M sodium azide . We found no significant difference in YFP expression between them ( S1A Fig ) . We made a Gateway pENTR221 vector , BJP-I135 , including FRT-flanked FLP-20 genomic DNA fused to SL2 , mCherry , and a terminator , let-858 3’ UTR , by modifying the plasmid pWD178 [66] , a gift from the Jorgensen lab . We used Gateway to recombine Pflp-20 upstream and GFP downstream of the construct into a MosSCI destination vector pCFJ150 [52] , targeting a ttTi5605 Mos1 insertion on chromosome II . We then generated a single copy insertion of this cassette and crossed it into flp-20 ( pk1596 ) to obtain the rescue strain BJH440 . We injected into BJH440 Pmec-4::FLP recombinase and obtained BJH457 . In this strain , FLP recombinase excises the FLP-20::SL2::mCherry sequence specifically in the TRNs . As a result , all neurons expressing FLP-20 have a functional copy of the gene , except for the TRNs . Chinese hamster ovary ( CHO ) cells K1 stably overexpressing the mitochondrial targeted apo-aequorin ( mtAEQ ) and the human Gα16 subunit were cultured and transfected with pcDNA3 . 1D::receptor-cDNA as described fully in [57] . Cells for negative control experiments were transfected with an empty pcDNA3 . 1D vector . The three FLP-20 peptides , ( AMMRFamide , AVFRMamide and SVFRLamide ) were synthesised by Cambridge Research Biochemicals , based on in silico predictions . All peptides were initially tested at a concentration of 10–5 M . In addition , BSA medium containing no peptides was used as a negative control , and ATP , which activates an endogenous CHO receptor , was used as a positive control . BSA and ATP responses were averaged for all transfected constructs . Calcium responses were monitored as previously described [57] for 30 s on a Mithras LB 940 luminometer ( Berthold Technologies ) . The numerical data used in all figures are included in S1 Data .
|
The brain has the remarkable capacity to respond to sensory loss by boosting remaining functioning senses . For example , certain features of hearing are improved in blind people . What are the cellular and molecular mechanisms underlying this effect ? How is a certain sense strengthened ? If it is possible to hear better , why don’t we hear better in the first place ? To simplify these problems , we examined them in an organism with a substantially less complicated nervous system than our own , the roundworm C . elegans . We discovered that C . elegans mutants that cannot sense touch to the body exhibit an improved sense of smell . We were able to pinpoint this change in sensory performance to a change in strength of a specific synapse in the olfactory circuit . We further found that in normal worms , this olfactory synapse is suppressed through a neuropeptide signal transmitted from the touch sensing neurons . In contrast , without any touch input , the touch neurons secrete less neuropeptide , the olfactory synapse becomes stronger , and the sense of smell improves . We were able to reverse these effects by artificially stimulating the touch neurons and by engineering a new synapse into the olfactory circuit .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[] |
2016
|
Neuropeptide-Driven Cross-Modal Plasticity following Sensory Loss in Caenorhabditis elegans
|
Cryptosporidium parvum is a highly prevalent zoonotic and anthroponotic protozoan parasite that causes a diarrheal syndrome in children and neonatal livestock , culminating in growth retardation and mortalities . Despite the high prevalence of C . parvum , there are no fully effective and safe drugs for treating infections , and there is no vaccine . We have previously reported that the bacterial-like C . parvum lactate dehydrogenase ( CpLDH ) enzyme is essential for survival , virulence and growth of C . parvum in vitro and in vivo . In the present study , we screened compound libraries and identified inhibitors against the enzymatic activity of recombinant CpLDH protein in vitro . We tested the inhibitors for anti-Cryptosporidium effect using in vitro infection assays of HCT-8 cells monolayers and identified compounds NSC158011 and NSC10447 that inhibited the proliferation of intracellular C . parvum in vitro , with IC50 values of 14 . 88 and 72 . 65 μM , respectively . At doses tolerable in mice , we found that both NSC158011 and NSC10447 consistently significantly reduced the shedding of C . parvum oocysts in infected immunocompromised mice’s feces , and prevented intestinal villous atrophy as well as mucosal erosion due to C . parvum . Together , our findings have unveiled promising anti-Cryptosporidium drug candidates that can be explored further for the development of the much needed novel therapeutic agents against C . parvum infections .
The zoonotic and anthroponotic protozoan parasite , Cryptosporidium parvum , is a major cause of diarrheal diseases in children under the age of two , resulting in significant morbidity and mortality in poor-resource areas of developing countries [1] . In livestock , particularly in calves , lambs and goat kids , it causes a serious diarrheal syndrome , culminating in growth retardation and high neonatal mortalities [2–4] . C . parvum is highly prevalent because of its enormous capacity to reproduce in infected livestock , resulting in large amounts of infective parasite oocysts being shed in animal feces , and contaminating water sources as well as the general environment . The parasite oocysts in the environment are difficult to eliminate because of their resistance to virtually all kinds of chemical disinfectants , as well as to commonly used water treatments such as chlorination [5] . The efficacy of the only FDA-approved anti-Cryptosporidium drug in humans , nitazoxanide , is modest . Of particular concern , nitazoxanide is ineffective in those individuals most at risk for morbidity and mortality due to Cryptosporidium infections , including malnourished children and immunocompromised individuals [6] . There is currently no vaccine against Cryptosporidium infections . Efforts to develop fully effective drugs against Cryptosporidium have largely been hampered by the lack of genetic tools for functional interrogation and validation of potential molecular drug targets in the parasite . Recently , however , a CRISPR/Cas9 gene manipulation approach [7] , and a morpholino-based targeted gene knockdown approach [8 , 9] in C . parvum have been developed . The completed and annotated genome sequence of Cryptosporidium indicates that , while the parasite lacks genes for conventional molecular drug targets found in other important protozoan parasites , it has several genes encoding plant-like and bacterial-like enzymes that catalyze potentially essential biosynthetic and metabolic pathways in Cryptosporidium [10] . Using a morpholino-based approach for targeted gene knockdown in C . parvum , we have previously validated that the C . parvum lactate dehydrogenase gene ( CpLDH ) that encodes a bacterial-like enzyme , is essential for survival , virulence and reproduction of C . parvum both in vitro and in vivo [8 , 9] . In the present study , we screened compound libraries and identified compounds with inhibitory effect against the enzymatic activity of recombinant CpLDH protein in vitro . Among the identified CpLDH inhibitors , we have demonstrated that two of the inhibitors can effectively block the growth , proliferation and pathogenicity of C . parvum in vivo at tolerable doses , suggesting that they are potential candidates for development of drugs against C . parvum infections .
By sequencing , the cloned open reading frame of CpLDH gene was verified to be 966 bp long , and 99 . 79% identical to that reported in the genome database ( GenBank accession number AF274310 . 1 ) . It coded for a 321 amino acids long protein with amino acid residue substitutions of F-198-L , R-251-K and K-295-E when compared to that in GenBank . The expressed and purified His-tagged CpLDH protein was of the expected molecular size of about 34 kDa ( Fig 1A ) . By analyzing the in vitro catalytic activities of recombinant CpLDH , we found that it depicted more activity in catalyzing the reduction of pyruvate to lactate than the oxidation of lactate to pyruvate . We found that recombinant CpLDH enzymatic catalytic activity was consistent with the Michaelis-Menten kinetics on pyruvate , NADH , lactate and NAD+ ( Fig 1B–1E ) . The Lineweaver–Burk representation of the saturation curves ( insets in Fig 1B–1E ) showed that the Km of recombinant CpLDH for pyruvate was at least 54-fold lower than that for lactate , while its Vmax for pyruvate was 123-fold higher than that for lactate ( Table 1 ) . Our obtained enzymatic kinetic parameters for recombinant CpLDH in comparison to those reported previously for C . parvum CpLDH [11] are summarized in Table 1 . We found that recombinant CpLDH had more catalytic activity for the reduction of pyruvate to lactate than for the oxidation of lactate to pyruvate . Therefore , we used the assay for reduction of pyruvate to lactate to screen chemical compounds for inhibitors of the enzymatic activity of recombinant CpLDH in vitro . Within the group of the 27 diverse chemical compounds ( S1 Table ) [12] , we identified three compounds ( NSC22225 , NSC37031 and NSC158011 ) that significantly ( P < 0 . 05 ) inhibited the catalytic activity of recombinant CpLDH for the reduction of pyruvate to lactate ( Fig 2 ) . On the other hand , among the 800 compounds in the Mechanistic Set IV ( S2 Table ) , we found 20 that had significant ( P < 0 . 05 ) inhibitory effect on the catalytic activity of recombinant CpLDH ( Fig 3 ) . Those 20 compounds included: NSC51148 , NSC1771 , NSC626433 , NSC349438 , NSC10447 , NSC85561 , NSC73413 , NSC657799 , NSC686349 , NSC638352 , NSC34931 , NSC253995 , NSC70929 , NSC70925 , NSC56817 , NSC79688 , NSC18298 , NSC71948 , NSC22842 , NSC33006 , and NSC82116 . The rest of the compounds either had no effect or augmented the catalytic activity of recombinant CpLDH and were thus not pursued further . To decipher the differences in the interactions of NSC150811 and NSC10447 ( S1 Fig ) with CpLDH and human LDH proteins , an in silico molecular docking using Autodock Vina [13] was performed to determine the most energetically favorable poses of the compounds complexed with the rigid structures of both CpLDH and human LDH . Both compounds were found to bind most favorably into the co-factor-binding pocket of the CpLDH and human LDH , where NADH binds to reduce pyruvate to lactate . NSC158011 complexed with CpLDH with an affinity of -6 . 4 kcal/mol . ( Table 2 ) . The ligand was surrounded in hydrophobic and hydrophilic interactions . Hydrophobic interactions occurred with the Ile-100 , Ala-80 , and Ile-15 residues and the nonpolar aromatic rings of the molecule , while polar interactions occurred with the Asn-97 and Gln-14 residues and the highly polar thio-amide group ( S2 Fig ) . Although secondary amines do not possess a strong dipole moment , it is possible that the positively-charged Arg-85 residue interacts with the resonance-stabilized deprotonated thio-amide moiety of NSC158011 . The docked NSC158011 possessed little solvent exposure , likely due to its folded nature and position , tight within the protein pocket ( S2 Fig ) . NSC158011 complexed with human LDH with an affinity of -7 . 2 kcal/mol . ( Table 2 ) . The hydrophobic Phe-119 , Ile-120 , Val-116 , Val-98 , Ala-96 , Val-26 , and Val-28 residues interacted with the non-polar aromatic rings of the NSC158011 ( S3 Fig ) . Remarkably , these same non-polar aromatic rings are also heavily solvent-exposed ( S3 Fig ) . NSC10447 complexed with CpLDH with an affinity of -7 . 6 kcal/mol . ( Table 2 ) . The ligand was involved primarily in polar interactions with the surrounding Ser81 , Thr-79 , Thr-229 , Thr-231 , and positive-charged Arg-85 residues , mediated through interactions with semi-polar carbonyl carbons abundant on one side of the molecule ( S4 Fig ) . The non-polar side of the molecule interacted favorably with the hydrophobic Tyr-233 residue internal to the protein ( S4 Fig ) . There was a weak hydrogen-bonding interaction between the backbone of Asn-97 and an alcohol group on the ligand . Additionally , the hydrophilic carbons had weak but notable solvent exposure ( S4 Fig ) . NSC10447 complexed with human LDH with an affinity of -7 . 1 kcal/mol . ( Table 2 ) . The ligand was involved primarily in hydrophobic interactions with the surrounding Val-98 , Ala-96 , Val-94 , Phe119 , Val-26 , Tyr-83 , and Val-116 residues that stabilized the non-polar moiety of the molecule ( S5 Fig ) . Interestingly , there was a weak , polar , anti-bonding interaction with the Thr-95 residue . The polar alcohol groups and the attached carbons were heavily solvent exposed in the final docking conformation ( S5 Fig ) . The results of the molecular docking simulation showed that due to the high level of interaction between the two lead compounds and the residues within the LDH cofactor-binding pocket , NSC158011 and NSC10447 each bound favorably to both CpLDH and human LDH proteins . It can be proposed that the compounds act as competitive inhibitors for the LDH enzyme , binding favorably to the hydrophobic residues internal to the co-factor-binding pocket and blocking the enzyme from binding NADH , thus preventing the hydride transfer that powers the conversion of pyruvate to lactate . All the compounds that we found to have inhibitory effect against recombinant CpLDH activity were first analyzed for in vitro cytotoxicity in a mammalian cell line , HCT-8 ( American Type Culture Collection Item number: CCL244 ) before testing their anti-Cryptosporidium efficacy . For cytotoxicity screening , varying concentrations of each compound ( from 0 to 700 μM ) were tested in triplicate using the WST-1 cell proliferation assay and the half maximal inhibitory concentration ( cytotoxicity IC50 values ) of the compounds in HCT-8 cells ( S3 Table ) were derived from dose–response curves using GraphPad PRISM software . To test the compounds’ efficacy against C . parvum in vitro , an initial screen was performed using concentrations that were at least 50% lower than the compounds’ respective cytotoxicity IC50 values ( S3 Table ) . NSC10447 and NSC158011 from the diverse group and Mechanistic Set IV group , respectively , were found to significantly ( P < 0 . 05 ) inhibit C . parvum proliferation in vitro at 48 h post-infection . Therefore , these two compounds were selected for secondary analysis of anti-Cryptosporidium efficacy using varying concentrations and durations of culture to derive the IC50 values for the inhibition of parasite proliferation . For each compound , the assays were done in two formats: ( 1 ) by adding the compound to the HCT-8 cells culture shortly before infecting them with C . parvum sporozoites , with the goal to assess whether the compounds would block host cell invasion by sporozoites , and ( 2 ) by adding the compounds to the cells 2 h post-infection to determine the effect of the compounds on intracellular parasites . When the cultures were analyzed at 48 h post-infection , compound NSC158011 was found to have a significant ( P < 0 . 05 ) concentration-dependent effect of inhibiting proliferation of intracellular C . parvum merozoites in HCT-8 cells starting at 10 μM ( with 40 μM blocking parasite growth almost completely ) relative to the control infected cultures without compound treatment ( Fig 4A ) . Treating the cultures with NSC158011 compound 2 h post-infection also resulted in a concentration-dependent reduction in parasite proliferation , but with a slight decrease in compound efficacy relative to treating at the time of infection ( Fig 4A ) . By using GraphPad PRISM software , the half maximal inhibitory concentration ( IC50 ) values of NSC158011 for C . parvum in vitro were derived from the dose–response curves . The NSC158011 IC50 values at 48 h post-infection for inhibition of C . parvum growth when compound was added immediately , or 2 h after infecting the HCT-8 cells were 14 . 88 and 15 . 81 μM , respectively . Analysis of the inhibitory effect of NSC158011 on parasite proliferation at 72 h post-infection , depicted similar concentration-dependent effects ( Fig 4B ) , with IC50 values of 15 . 63 and 16 . 50 μM when compound was added immediately or 2 h post-infection , respectively . After infecting HCT-8 cells , C . parvum is able to proliferate for 3–4 days before becoming growth-arrested . Therefore , during both observation time-points of 48 h and 72 h post-infection , C . parvum if untreated was expected to be in proliferative phase . Consistently , in the untreated infected cells , the relative parasite load at 72 h post-infection was about 2-fold that observed at 48 h post-infection ( Fig 4A and 4B ) . Compound NSC10447 also depicted a concentration-dependent inhibitory effect on parasite growth , both at 48 h ( Fig 5A ) and 72 h ( Fig 5B ) time points of observation . The efficacy of NSC10447 when added immediately or 2 h after infecting the cells was similar ( Fig 5A ) . The NSC10447 IC50 values at 48 h post-infection for inhibition of C . parvum growth when compound was added immediately , and 2 h after infecting the HCT-8 cells were 72 . 65 and 79 . 52 μM , respectively . Consistently , NSC10447 had a concentration-dependent inhibitory effect on parasite growth at 72 h time point of observation ( Fig 5B ) , with IC50 values of 83 . 63 and 95 . 17 μM , when the compound was added immediately , and 2 h after infecting the HCT-8 cells , respectively . Noteworthy , in all instances , NSC158011 depicted significantly higher in vitro efficacy against C . parvum than NSC10447 ( Table 3 ) . The IC50 values of NSC158011 and NSC10447 for the inhibition of the catalytic activity of recombinant CpLDH in vitro were 76 . 59 μM and 46 . 33 μM , respectively ( Table 3 ) . We used paromomycin as the positive control treatment . Using in vitro assays the cytotoxicity of paromomycin in HCT-8 cells has been reported to be negligible even when used at concentrations above 1000 μM [14 , 15] . Therefore , we tested the in vitro efficacy of paromomycin against C . parvum at varying concentrations up to a maximum of 1000 μM , and found it to have a concentration-dependent effect of inhibiting C . parvum growth in vitro , both at 48 h ( Fig 6A ) and 72 h ( Fig 6B ) post-infection , with IC50 values of 450 and 400 μM , respectively . Others have previously reported paromomycin to have an IC50 of 711 μM for inhibition of C . parvum growth in HCT-8 cells [15] . There was no notable significant difference in paromomycin inhibitory effect against C . parvum between starting the treatment immediately or 2 h after infection of the HCT-8 cells . Compound NSC158011 and NSC10447 that were found to inhibit C . parvum growth in vitro were selected for in vivo testing using a mouse infection model . Prior to use in mice , the highest tolerable doses in mice for the two compounds were found to be 400 mg/kg and 1000 mg/kg for NSC158011 and NSC10447 , respectively . These doses consistently did not induce any toxicity signs ( changes from normal physical activity , respiration , body temperature , feeding pattern , body posture , fur condition or occurrence of death ) over 7 days of daily oral gavage in mice . In the case of NSC158011 , the dose of 400 mg/kg intraperitoneal administration in mice has also been previously shown not to be toxic to mice [16] . Thus , the doses of 400 mg/kg and 1000 mg/kg were selected as the highest doses for testing the efficacy of NSC158011 and NSC10447 , respectively , against C . parvum growth and proliferation in mice . Paromomycin at 100 mg/kg daily by oral gavage was used as a positive control . The daily load of C . parvum oocysts shed in mice feces was determined using real time PCR quantification of the C . parvum 18s rRNA gene . As expected , in the feces of untreated infected mice , C . parvum genomic DNA was almost undetectable during the first 2 days post-infection , but was detectable from 3 days post-infection , and increased progressively with increase in the number of days post-infection ( Tables 4–6 ) . The notable lower C . parvum DNA in the feces of the untreated infected mice in Table 6 when compared to those for Tables 4 and 5 is because the infection assays for Tables 3 and 4 were done using freshly purified oocysts , while those for Table 6 were done using oocysts that were purified 3 months earlier , and thus their infectivity could have been lower . We found that NSC158011 at 400 mg/kg significantly ( P < 0 . 05 ) reduced shedding of C . parvum oocysts in mice feces , comparable to the efficacy of paromomycin ( Table 4 ) . As expected , C . parvum DNA was consistently undetectable at all time points sampled in the uninfected mice ( Table 4 ) . By day 7 post-infection , both NSC158011 and paromomycin treatment had reduced the shedding of C . parvum in mice feces by about 3000-fold when compared to the untreated infected mice samples ( Table 4 ) . This suggested that NSC158011 , at 400 mg/kg , had sustained anti-Cryptosporidium efficacy in vivo comparable to that of 100 mg/kg of paromomycin . We titrated the dose of NSC158011 to determine the effect of lower dosages . We observed a dose-dependent reduction in efficacy of NSC158011 , with 200 mg/kg having about 10-fold lower efficacy than paromomycin during the last three days of treatment . Consistently , 100 mg/kg of NSC158011 showed lower efficacy than 200 mg/kg NSC158011 ( Table 5 ) . For testing the in vivo anti-Cryptosporidium efficacy of NSC10447 , in addition to testing the highest tolerable dose of 1000 mg/kg , we also tested lower doses of 250 mg/kg and 500 mg/kg . While the C . parvum DNA was undetectable in the uninfected mice’s feces , the untreated infected mice had readily detectable C . parvum DNA by day 3 post-infection , that then increased progressively with increase in number of days post-infection ( Table 6 ) . From day 3 until day 8 of treatment , compared to the infected untreated , mice treated with NSC10447 at 250 , 500 and 1000 mg/kg showed sustained significantly lower ( by at least 50% ) C . parvum DNA load in their feces ( Table 6 ) . There was a notable dose-dependent effect , with 1000 mg/kg having the highest efficacy ( Table 6 ) . Notably , on day 7 and 8 post-infection , the 1000 mg/kg dose of NSC10447 maintained anti-Cryptosporidium efficacy that was comparable to that of paromomycin , while both 250 and 500 mg/kg doses depicted lower efficacies than paromomycin ( Table 6 ) . During C . parvum infection , usually the distal small intestines are severely affected , characterized by villous atrophy , erosion and ulceration of the intestinal mucosa . Thus , we performed histopathological examination of the distal small intestines of the experimental mice at 9 days post-infection . As expected , while uninfected mice maintained the integrity of the intestinal mucosa ( Fig 7A and Fig 8A ) , infected untreated mice had microscopic lesions characterized by villous atrophy and mucosal erosion ( Figs 7B and 8B ) . Infected mice treated with NSC158011 maintained intact intestinal mucosa and villi ( Fig 7D ) whose integrity was similar to that of mice treated with paromomycin ( Fig 7C ) . Likewise , mice treated with NSC10447 also prevented villous atrophy and maintained the integrity of the intestinal mucosa ( 8D-F ) similar to treatment with paromomycin ( Fig 8C ) . We enumerated the mean percentage of denuded intestinal villi in 4 randomly chosen microscopic fields per sample from representative histopathology images . We observed that in infected mice , just like treatment with paromomycin , treatment with NSC158011 and NSC10447 reduced the percentage of denuded intestinal villi by 7-fold or more compared to infected untreated mice ( Fig 7E and Fig 8G ) . These findings corroborated the observations that treatment with NSC158011 and NSC10447 , just like paromomycin , inhibited C . parvum oocysts shedding in mice’s feces to almost undetectable levels .
Because of the lack of genetic tools for identifying essential molecular components in Cryptosporidium , screening for potential drug lead-compounds against Cryptosporidium has been based on molecular targets identified in other protozoan parasites such as Toxoplasma and Plasmodium . However , the completed and annotated Cryptosporidium genome sequence shows the absence of conventional drug targets being pursued in other protozoan parasites [10] . Nevertheless , the completed genome sequence of Cryptosporidium has unveiled a number of bacterial-like and plant-like classic and novel drug molecular targets that now require functional characterization and validation using genetic tools [10] . Among the identified potential drug molecular targets , is the C . parvum lactate dehydrogenase ( CpLDH ) , which is a bacterial-type lactate dehydrogenase enzyme that the parasite uses to generate metabolic energy ( ATP ) in the glycolytic pathway [11 , 17 , 18] . Importantly , C . parvum lacks both the Krebs cycle and the cytochrome-based respiration chain [10] , suggesting that the glycolysis pathway is the sole energy source in C . parvum [19–21] . Consistently , using morpholino-based targeted knockdown of CpLDH , we recently showed that CpLDH is essential for growth , propagation and viability of C . parvum in vitro and in vivo [8 , 9] . Corroboratively , previous studies have shown that known inhibitors of lactate dehydrogenase enzymes , gossypol and FX11 , are able to inhibit the enzymatic activity of CpLDH [11] . However , both gossypol and FX11 are not specific to CpLDH , but also inhibit mammalian lactate dehydrogenases , implying that they would be toxic to mammalian cells . Regardless , it is noteworthy that CpLDH is unique to C . parvum , and is very significantly different from the lactate dehydrogenase enzymes found in mammals [17] . In the present study , we first established the in vitro enzymatic kinetic parameters of the natively purified recombinant CpLDH protein . Consistent with previous reports by others [11] , we found that recombinant CpLDH preferentially catalyzed the reduction of pyruvate to lactate , and displayed Michaelis-Menten enzymatic kinetics . Using the in vitro enzymatic assay , we identified 29 chemical compounds that inhibited the catalytic activity of recombinant CpLDH for the reduction of pyruvate to lactate . Lactate dehydrogenase is a key enzyme for the anaerobic respiration in which pyruvate is reduced to lactate , with the concomitant oxidation of NADH to NAD+ [22] . Thus , we tested the candidate compounds for toxicity in a mammalian cell line ( HCT-8 ) and selected only those that were tolerable at high micromolar concentrations ( IC50 > 140 μM ) as candidate compounds for further testing . The cytotoxicity IC50 values of the candidate compounds were at least 2-fold higher than the cytotoxicity IC50 values of known mammalian lactate dehydrogenase inhibitors ( gossypol and FX11 ) in HCT-8 cells [11] . We subsequently tested the candidate compounds for anti-Cryptosporidium effect using in vitro infection assays of HCT-8 cells monolayers and identified compounds NSC158011 and NSC10447 that sustainably inhibited the proliferation of intracellular C . parvum . The HCT-8 cells were infected with excysted C . parvum sporozoites that infect host cells and transform into proliferative merozoites . In C . parvum sporozoites and merozoites , CpLDH is expressed and localized in the cytosol [18] , suggesting that it is utilized for energy generation during these parasite stages that are important for host cell invasion and intracellular parasite growth . Interestingly , NSC158011 has been previously shown to inhibit the catalytic activity of the Plasmodium faclciparum phosphoethanolamine methyltransferase enzyme , and to inhibit in vitro intracellular growth of the parasite [23] . However , based on the completed genome sequence of C . parvum , there are no homologs of genes encoding a phosphoethanolamine methyltransferase in C . parvum . Therefore , our findings suggest that the anti-Cryptopsoridium activity of NSC158011 is associated with its ability to inhibit the catalytic activity of CpLDH which is an essential enzyme for survival and growth of C . parvum , both in vitro and in vivo [8 , 9] . At amino acid sequence level , CpLDH is only 25% identical to human LDH , with the active site conformation of CpLDH being significantly different from that of the human LDH [24] . Further , in the 3-dimensional structure model of the two enzymes , the helix-loop portion of CpLDH is more proximal to the active site loop than it is in the human LDH [24] . Additionally , the co-factor binding site of human LDH possesses a network of hydrogen-bonding formed by a serine residue with NAD+ , while the co-factor binding site of CpLDH only forms two hydrogen bonds with NAD+ [24] . This is thought to lower the CpLDH affinity for NAD+/NADH than human LDH [24] . When we modeled NSC158011 and NSC10447 onto the 3-D crystal structure of CpLDH and human LDH , we found that both NSC158011 and NSC10447 bind to the NAD+ co-factor binding site . Interestingly , in the docking simulation , NSC10447 displayed better affinity for CpLDH than human LDH , while NSC158011 displayed better affinity for the human LDH . We had selected NSC10447 and NSC158011 based on their low toxicity in a human cell line , but high inhibitory activity against C . parvum , though these molecules still bind to the human LDH crystal structure . A docking simulation calculates the free energy of the interaction between a protein and a ligand but does not consider the interaction between the ligand and its surrounding solvent ( solvation energy ) . Due to the unfavorable high solvent exposure of the non-polar , aromatic rings in NSC158011 docked to human LDH , it can be inferred that the binding stability is greatly reduced . The docking pose of NSC158011 to CpLDH exposes the non-polar regions of the molecule to less solvent , leading to a much more stable interaction . These ligand binding properties suggest that NSC158011 and NSC10447 would more effectively compete out the binding of NAD+ to CpLDH than to human LDH . This is consistent with our observations that both NSC158011 and NSC10447 effectively inhibit C . parvum growth and replication ( both in vitro and in vivo ) at concentrations that are not toxic to mammalian ( including human ) cells . Typically , solvent-exposed protein pockets like the one in LDH are often not targeted in lead-compound optimization due to their poor binding characteristics , but the results of our in silico docking reveal that solvent-exposed protein pockets may be useful for enhancing lead-compound selectivity . Importantly , these differences in ligand-binding stability between CpLDH and human LDH offer prospects for identifying inhibitors that would specifically target CpLDH , without being toxic to mammalian host cells , and would thus be potential lead-compounds for development of effective anti-Cryptosporidium drugs . We observed that NSC158011’s IC50 for the inhibition of recombinant CpLDH in an in vitro enzymatic assay was higher than its IC50 for inhibition of C . parvum growth . Based on our observation that NSC158011 binds to the co-factor binding site in CpLDH , the likely reason for this discordance is that in the recombinant CpLDH enzymatic assay in vitro , excessive amounts of NADH co-factor ( 1 mM ) were used that in turn required high concentration of NSC158011 to effectively compete out the co-factor and reduce the generation of the product . In comparison , intracellular ( intra-parasite ) levels of co-factor are likely much lower ( μM range ) . For instance , in human cells the absolute concentration of NADH has been reported to be in the range of 97 to 168 μM [25 , 26] . The lower intracellular concentrations of NADH when compared to the higher in vitro concentrations , would translate into lower concentrations of NSC158011 to effectively compete out the co-factor and register a decrease in CpLDH activity , and subsequent reduction in parasite growth . Importantly , the chemical structure of NSC158011 suggests that it possesses promising drug-like properties that render it amenable to drug development [23] . Using doses that were tolerable in mice , we tested the in vivo efficacies of NSC158011 and NSC10447 in Gamma interferon knockout mice ( B6 . 129S7-Ifng ) that when infected with C . parvum , develop debilitating clinical disease , with completion of the parasite life cycle and shedding of oocysts in feces [27] . We found that both NSC158011 and NSC10447 consistently significantly reduced the shedding of C . parvum oocysts during the experimental period of 9 days , and prevented the occurrence of villous atrophy and intestinal mucosal erosion that is associated with C . parvum infection . NSC158011 displayed better efficacy than NSC10447 , both in vitro and in vivo , with lower anti-Cryptosporidium IC50 values . Importantly , compared to the only FDA-approved nitazoxanide that lacks efficacy in immunocompromised individuals , both NCS158011 and NSC10447 were efficacious against C . parvum in the immunocompromised mice we used in the study . In conclusion , we have demonstrated NSC158011 and NSC10477 as specific inhibitors for CpLDH that have efficacy against C . parvum both in vitro and in vivo . Thus , our findings provide promising anti-Cryptosporidium drug candidates that can be explored further for the development of much needed novel cryptosporidiosis therapeutic interventions .
All experiments involving the use of mice and Holstein calves were carried out in accordance with guidelines and protocols number 17024 and 18108 , respectively , approved by the University of Illinois Institutional Animal Care and Use Committee , in compliance with the United States Department of Agriculture Animal Welfare Act and the National Institute of Health Public Health Service Policy on the Humane Care and Use of Animals guidelines . For all experiments , the AUCP-1 isolate of C . parvum was used . The parasites were maintained and propagated in male Holstein calves in accordance with the guidelines of protocol number 18108 approved by the University of Illinois at Urbana-Champaign , USA . Freshly shed C . parvum oocysts in calf feces were extracted and purified by sequential sieve filtration , Sheather's sugar flotation , and discontinuous sucrose density gradient centrifugation , essentially as previously described [28 , 29] . The purified oocysts were rinsed and stored at 4°C in 50 mM Tris–10 mM EDTA , pH 7 . 2 , and used within 3 months . Sporozoites were excysted from C . parvum oocysts following the method described previously [30] . Briefly , to about 1 × 108 purified C . parvum oocysts suspended in 500 μl of PBS , an equal volume of 40% commercial laundry bleach was added and incubated for 10 minutes at 4°C . The oocysts were washed four times in PBS containing 1% ( w/v ) bovine serum albumin and resuspended in Hanks balanced salt solution , incubated for 60 minutes at 37°C , and mixed with an equal volume of warm 1 . 5% sodium taurocholate in Hanks balanced salt solution . The excysted sporozoites were collected by centrifugation and resuspended in supplemented PBS . The sporozoites were purified by passing the suspension through a sterile 5 μM syringe filter ( Millex ) and enumerated with a hemocytometer . The coding sequence of CpLDH was cloned from cDNA prepared from the AUCP-1 isolate of C . parvum , and the His-tagged CpLDH recombinant protein expressed in Escherichia coli , and purified in native form essentially as previously described [8] . Briefly , cDNA was prepared from total RNA extracted from the AUCP-1 isolate of C . parvum , and the coding sequence of CpLDH ( Genebank accession number AF274310 . 1 ) was PCR-amplified from the cDNA using the primer pair 5’-CTCGAGATGATTGAAAGACGCAAGA-3’ ( Forward , with the XhoI restriction site italicized and start codon in bold ) and 5’-GGATCCTTATGCTCCAGCTGGT-3’ ( Reverse , with the BamHI site italicized and stop codon in bold ) . The PCR amplicon was cloned at the XhoI/BamHI site of the pET15b expression vector in-frame with the hexahistidine ( His-tag ) at the N-terminal and sequenced to confirm identity . The recombinant expression vector was transformed into protein expression E . coli BL21-CodonPlus-DE3-RIL ( Stratagene ) . Transformed E . coli was cultured at 37°C in Luria broth medium ( supplemented with 100 μg/ml ampicillin and 34 μg/ml chloramphenicol ) to an A600 of 0 . 8 followed by addition of 1 mM isopropyl-β-d-thiogalactopyranoside to induce protein expression . The expression E . coli was harvested and lysed under native conditions by sonicating in lysis buffer ( 50 mM NaH2PO4 , 300 mM NaCl , 10 mM Imidazole , pH 8 . 0 ) containing a 1x EDTA-free protease inhibitor cocktail , 600 units benzonase and 30 kU lysozyme ( EMD Millipore ) . The lysate was clarified by centrifugation and the His-tagged recombinant protein purified under native conditions by nickel-affinity chromatography according to the manufacturer's instructions ( Novagen ) . The wash buffer used contained 50 mM NaH2PO4 , 300 mM NaCl and 20 mM Imidazole , pH 8 . 0 , while the elution buffer was composed of 50 mM NaH2PO4 , 300 mM NaCl , 250 mM Imidazole , pH 8 . 0 . The eluate was dialyzed using a buffer containing 5 mM Hepes–KOH ( pH 7 . 8 ) and 0 . 5 mM DTT . The purity of the recombinant protein was determined by SDS/PAGE , and the concentration measured using a Qubit 3 . 0 fluorometer ( Life technologies ) . The in vitro enzymatic activity of the recombinant CpLDH protein for catalyzing the reduction of pyruvate to lactate was determined by measuring the change in optical density of a 100 μl reaction mixture containing 10 mM pyruvate , 1 mM NADH , 100 mM Tris , pH 7 . 5 and varying concentrations of CpLDH recombinant protein at 25°C . On the other hand , the catalytic activity of CpLDH recombinant protein for the oxidation of lactate to pyruvate was determined by measuring the change in optical density of a 100 μl reaction mixture containing 100 mM lactate , 1 mM NAD+ , 100 mM Tris , pH 9 . 2 , with varying concentrations of CpLDH recombinant protein at 25°C . For determining the kinetic parameters , a fixed concentration of CpLDH recombinant protein was used in reactions with varying substrate and co-factor concentrations ( pyruvate from 0 . 5–15 mM; NADH from 0 . 25–1 . 5 mM for the reduction reaction , while for the oxidation reaction lactate varied from 25–125 mM; NAD+ from 0 . 05–1 . 5 mM ) . In all assays , reaction mixtures without recombinant CpLDH protein were included as negative controls . All assays were performed in triplicate and repeated at least thrice . The change in optical density was measured every 15 seconds for a total of 2 minutes using a Spectra Max 384 Plus plate reader ( Molecular Devices ) at a wave length of 340 nm . The chemical compounds were obtained from the National Cancer Institute/Developmental Therapeutics Program Open Chemical Repository . They consisted of a diverse set of compounds ( n = 27 ) ( S1 Table ) reported previously [12] , and a Mechanistic Set IV compounds ( n = 800 ) ( S2 Table ) . The compounds were reconstituted in dimethyl sulfoxide ( DMSO ) as stock solutions . Just before use , aliquots of the stock solutions were diluted in sterile distilled water to generate working solutions , such that the final amount of DMSO added to the reaction mixtures was less than 1% ( V/V ) . The compounds were tested for their inhibitory effect against the enzymatic activity of recombinant His-tagged CpLDH for catalyzing the reduction of pyruvate to lactate . The reactions were performed in 100 μl reaction volume containing 10 mM pyruvate , 1 mM NADH , 100 mM Tris , pH 7 . 5 , 15 ng/μl of recombinant CpLDH protein with or without 20 μM of compound . Control reactions without recombinant CpLDH protein were included . Reactions were performed in triplicate and repeated at least thrice . The change in optical density was measured every 15 seconds for a total of 2 minutes using a Spectra Max 384 Plus plate reader ( Molecular Devices ) at a wave length of 340 nm . The mean percent inhibition effect of each compound on recombinant CpLDH activity was derived using the following formula: Mean Percent Inhibition ( MPI ) = ( ΔOD340 of reaction with compound / ΔOD340 of reaction without compound ) X 100 Where: Compounds with inhibitory effect against the enzymatic activity of recombinant CpLDH were tested for cytotoxicity in a human cell line , HCT-8 ( American Type Culture Collection Item number: CCL244 ) , that was used for in vitro culture of C . parvum . A colorimetric assay using the cell proliferation reagent WST-1 ( Roche , USA ) for the quantification of cell viability was performed . HCT-8 cells were cultured in 96-well plates in 200 μl of RPMI 1640 medium without phenol red ( Life Technologies ) , but supplemented with 2 g/L of sodium bicarbonate , 2 . 5 g/L of glucose , 10% FBS ( Gibco , USA ) , 1× antibiotic–antimycotic ( Gibco ) , and 1× sodium pyruvate ( Gibco ) . When the cells were confluent , the old medium was replaced with fresh medium with or without varying concentrations of chemical compound . After 24 h of culture , 10 μl of the cell proliferation reagent WST-1 , ( for quantification of cell viability ) was added to each well , mixed and the plates incubated for 1 h at 37 C with 5% CO2 in the dark . Following incubation , 150 μL of the medium from each well was transferred to a new 96-well plate and quantification of the formazan dye produced by metabolically active cells was read as absorbance at a wavelength of 420 nm using a scanning multi-well spectrophotometer ( Spectra Max 384 Plus; Molecular Devices , USA ) . Three independent assays were performed and the dose–response curves of the means of triplicate assays were generated using GraphPad PRISM software to derive the half maximal inhibitory concentration ( IC50 ) of compounds in HCT-8 cells . HCT-8 cells were cultured in supplemented RPMI-1640 medium in 96-well plates . When the cells were confluent , old medium was replaced with fresh medium . To one set of wells , varying concentrations of recombinant CpLDH inhibitors ( reconstituted in DMSO and diluted in RPMI medium ) were added , while another set was left without inhibitors . Paromomycin was used as a positive control drug reconstituted in distilled sterile water . Then , 4 x 104 freshly excysted sporozoites were added to each well and incubated at 37°C with 5% CO2 . After 2 h incubation , varying concentrations of CpLDH inhibitors were added to the set of infected cells that were not treated initially . Control infected cells without inhibitors , but with added DMSO volumes equivalent to those used in the wells with inhibitors , were included . The cells were maintained in culture for a total of either 48 h or 72 h and processed for immunofluorescence assay as previously described [8 , 16] . Briefly , medium was decanted and the cells fixed with methanol-acetic acid ( 9:1 ) for 2 minute at room temperature . The cells were rehydrated and permeabilized by two successive washes with buffer ( 0 . 1% Triton X-100 , 0 . 35 M NaCl , 0 . 13 M Tris-base , pH 7 . 6 ) and blocked with 5% normal goat serum , followed by staining with antibody to C . parvum ( SporoGlo; Waterborn , Inc . ) overnight at 4°C . The stained cells were washed twice with PBS , followed by water , and then imaged with an inverted fluorescence microscope . Fluorescence quantification was done using ImageJ version 1 . 37v software ( NIH ) . Assays were performed in triplicate and repeated at least thrice . To propose a model for the specific binding of NSC15801 and NSC10447 to lactate dehydrogenase , the crystal structure of CpLDH complexed with substrate pyruvate and cofactor analogue 3-acetylpyridine adenine dinucleotide ( APAD ) was obtained from the RCSB protein database ( 4ND2 ) . The chemical structures of NSC158011 and NSC10447 were obtained from the PubChem library . The protein crystal structure was loaded into the AutoDockTools software suite and a search location box was drawn encompassing the co-factor analogue APAD in one subunit of the homotetrameric protein . APAD was removed from the active site of the crystal structure using the Swiss-Model DeepView software . Using Autodock Vina ( Scripps Institute , USA ) as previously done [13] , polar hydrogen atoms were added to the APAD-deficient LDH structure and its non-polar hydrogen atoms were merged . NSC15801 and NSC10447 were each docked into the empty co-factor binding site of one subunit of the protein with exhaustiveness = 10 . Both compounds were docked using a 40 × 40 × 40 Å grid box , and all single bonds within the ligands were set to allow free rotation . The procedure was subsequently repeated with the crystal structure of human LDH ( 1I0Z ) and a 24 × 14 × 20 Å grid box around the co-factor binding pocket of the new structure . Docking results were visualized using VMD ( University of Illinois at Urbana-Champaign , USA ) as previously done [31] . The most energetically favorable result of each docking was then loaded as a protein-ligand complex into the Schrödinger Maestro software ( Schrodinger LLC , USA ) to investigate the nature of the protein-ligand interactions and propose a mechanism for the lead compounds’ inhibition of LDH . Because the presence or absence of natural LDH substrate , pyruvate , complexed within the active site did not significantly affect binding of the compounds in CpLDH or human LDH , those models were excluded in the final molecular docking simulations . Gamma interferon knockout mice ( B6 . 129S7-Ifng ) , 8 weeks of age , were purchased from Charles River , USA . The care and use of the mice was done following the guidelines of protocol number 17024 approved by the University of Illinois at Urbana-Champaign , USA . The animals were allowed to acclimatize for 1 week before experiments commenced . Stock solutions of recombinant CpLDH inhibitors reconstituted in DMSO were diluted in sterile distilled water to reduce the final amount of DMSO in the solution to less than 1% ( v/v ) before administering them to mice . Prior to testing the anti-Cryptosporidium effect of the inhibitors in mice , the tolerability of each inhibitor was tested by oral gavage using varying dosages ( 100–2000 mg/kg body weight ) of each inhibitor in groups of mice ( three mice for each dose ) daily for 7 days . Mice were observed daily for signs of toxicity including changes from normal physical activity , respiration , body temperature , feeding , posture , fur condition or occurrence of death . The highest dose ( 1000 mg/kg for NSC10447 , and 400 mg/kg for NSC158011 ) of each inhibitor that did not induce any toxicity signs over the 7 days of administration was used as the maximum dose limit for subsequent in vivo experiments . The subsequent dosages of NSC10447 used in the mice infection assays were 250 , 500 and 1000 mg/kg mouse body weight , while the NSC158011 dosages used were 100 , 200 and 400 mg/kg mouse body weight . Mice were allocated to groups as follows: “Infected plus inhibitor treatment”; “Infected minus inhibitor treatment”; “Uninfected minus inhibitor treatment” and “Infected plus paromomycin treatment” . Each group contained at least three mice . Each mouse in the infection groups received 5 , 000 C . parvum oocysts ( resuspended in 50 μl of PBS ) by oral gavage . Mice were housed individually in cages lined with sterile gauze as bedding . One day post-infection ( PI ) , daily oral gavage administration of recombinant CpLDH inhibitor or paromomycin commenced and continued for a total of 7 days . Untreated mice received an equivalent volume of sterile distilled water ( containing DMSO equivalent to the amount administered in the inhibitor-treated group ) by oral gavage . Fecal pellets were collected daily from each cage and placed in individual sterile 15 ml conical tubes . An equivalent volume of PBS containing a cocktail of penicillin ( 100 units/ml ) , streptomycin ( 100 μg/ml ) , chloramphenicol ( 34 μg/ml ) and amphotericin ( 0 . 25 μg/ml ) was added to the fecal samples and stored at 4°C until use for quantification of C . parvum genomic DNA load . Three independent replicate infection assays were performed . At 9 days PI , mice were euthanized and 5 cm of the distal small intestine resected 2 cm anterior to the cecum and immediately submerged in 4% buffered formalin . The intestinal tissues were submitted for histopathology to the Veterinary Diagnostic Laboratory at the University of Illinois at Urbana-Champaign . Briefly , intestinal tissues preserved in 4% buffered formalin were washed in 70% ethanol and embedded in 1% agar and then processed for paraffin embedding . For hematoxylin and eosin staining , five μm transverse and cross sections were cut and processed and stained following standard procedures of the Veterinary Diagnostic Laboratory . Sections were imaged using a Zeiss microscope and images captured with a color camera . Genomic DNA was extracted from individual fecal samples collected from mice at different days as described above . For each sample , 220 mg of homogenized feces were used to extract genomic DNA using the QIAamp PowerFecal DNA Kit ( Qiagen , USA ) following the manufacturer’s instructions . Quantification of the amount of C . parvum 18s rRNA gene ( GenBank accession number AF164102 ) was performed essentially as described previously [9] . Briefly , the primer pair 5′-CTGCGAATGGCTCATTATAACA-3′ ( Forward ) , and 5′-AGGCCAATACCCTACCGTCT-3′ ( Reverse ) was used to generate a 240 bp amplicon from C . parvum genomic DNA by conventional PCR . The PCR product was fractionated on agarose gel , extracted using the QIAquick Gel extraction kit ( Qiagen , USA ) , and the concentration measured by Nanodrop Spectrophotometer ( Fisher , USA ) . Ten-fold serial dilutions of the extracted DNA fragment were made and used as quantification standards for real-time PCR . Each real time PCR mixture contained 2 μl of DNA template , 1 μl of primer mix ( 500 nM each ) , and 10 μl of SsoAdvanced Universal SYBR Green Supermix ( Bio-Rad , USA ) , with the final volume made up to 20 μl with nuclease-free water . The cycling conditions included an initial denaturation for 10 min at 98°C , 40 cycles at 98°C for 15 s and 60°C for 1 min , and a final melting curve step . Cycling was performed using a 7500 Real Time PCR System ( Applied Biosystems , USA ) . DNA quantities were derived by the system software using the generated quantification standard curves . Statistical analyses were performed using two-tailed Student’s t test . P values of 0 . 05 or less were considered significant .
|
Cryptosporidium parvum is a protozoan parasite that can cause a life-threatening gastrointestinal disease in children and in immunocompromised adults . The only approved drug for treatment of Cryptosporidium infections in humans is nitazoxanide , but it is not effective in immunocompromised individuals or in children with malnutrition . C . parvum possesses a unique lactate dehydrogenase ( CpLDH ) enzyme that it uses for generating metabolic energy ( ATP ) via the glycolytic pathway to fuel its growth and proliferation in the host . We have identified novel inhibitors for the enzymatic activity of CpLDH . Further , we have demonstrated that two of the CpLDH inhibitors effectively block the growth , proliferation and pathogenicity of C . parvum at tolerable doses in immunocompromised mice . Together , our findings have unveiled novel CpLDH inhibitors that can be explored for the development of efficacious therapeutic drugs against C . parvum infections .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"parasite",
"groups",
"oocysts",
"chemical",
"compounds",
"respiratory",
"infections",
"ketones",
"enzymology",
"cryptosporidium",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"pyruvate",
"pulmonology",
"parasitology",
"apicomplexa",
"protozoans",
"enzyme",
"inhibitors",
"digestive",
"system",
"proteins",
"acids",
"chemistry",
"recombinant",
"proteins",
"cryptosporidium",
"parvum",
"gastrointestinal",
"tract",
"biochemistry",
"eukaryota",
"anatomy",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"organisms"
] |
2019
|
Novel lactate dehydrogenase inhibitors with in vivo efficacy against Cryptosporidium parvum
|
Filarial parasites ( e . g . , Brugia malayi , Onchocerca volvulus , and Wuchereria bancrofti ) are causative agents of lymphatic filariasis and onchocerciasis , which are among the most disabling of neglected tropical diseases . There is an urgent need to develop macro-filaricidal drugs , as current anti-filarial chemotherapy ( e . g . , diethylcarbamazine [DEC] , ivermectin and albendazole ) can interrupt transmission predominantly by killing microfilariae ( mf ) larvae , but is less effective on adult worms , which can live for decades in the human host . All medically relevant human filarial parasites appear to contain an obligate endosymbiotic bacterium , Wolbachia . This alpha-proteobacterial mutualist has been recognized as a potential target for filarial nematode life cycle intervention , as antibiotic treatments of filarial worms harboring Wolbachia result in the loss of worm fertility and viability upon antibiotic treatments both in vitro and in vivo . Human trials have confirmed this approach , although the length of treatments , high doses required and medical counter-indications for young children and pregnant women warrant the identification of additional anti-Wolbachia drugs . Genome sequence analysis indicated that enzymes involved in heme biosynthesis might constitute a potential anti-Wolbachia target set . We tested different heme biosynthetic pathway inhibitors in ex vivo B . malayi viability assays and report a specific effect of N-methyl mesoporphyrin ( NMMP ) , which targets ferrochelatase ( FC , the last step ) . Our phylogenetic analysis indicates evolutionarily significant divergence between Wolbachia heme genes and their human homologues . We therefore undertook the cloning , overexpression and analysis of several enzymes of this pathway alongside their human homologues , and prepared proteins for drug targeting . In vitro enzyme assays revealed a ∼600-fold difference in drug sensitivities to succinyl acetone ( SA ) between Wolbachia and human 5′-aminolevulinic acid dehydratase ( ALAD , the second step ) . Similarly , Escherichia coli hemH ( FC ) deficient strains transformed with human and Wolbachia FC homologues showed significantly different sensitivities to NMMP . This approach enables functional complementation in E . coli heme deficient mutants as an alternative E . coli-based method for drug screening . Our studies indicate that the heme biosynthetic genes in the Wolbachia of B . malayi ( wBm ) might be essential for the filarial host survival . In addition , the results suggest they are likely candidate drug targets based upon significant differences in phylogenetic distance , biochemical properties and sensitivities to heme biosynthesis inhibitors , as compared to their human homologues .
Human filarial nematodes affect more than 150 million people worldwide with 1 billion people at risk in over 80 countries , and lead to some of the most debilitating tropical diseases , including elephantiasis and African river blindness [1] , [2] . The current anti-filarial treatments e . g . DEC , ivermectin , albendazole ( all suitable for lymphatic filariasis; ivermectin for onchocerciasis ) interrupt the cycle of transmission of the causative filarial parasites Brugia malayi , Onchocerca volvulus and Wuchereria bancrofti , by predominantly killing microfilaria . However , a lower activity against adult worms , which can survive in human hosts for up to decades , is known . DEC and albendazole produce macrofilaricidal activity only after repeated rounds of mass drug administration ( MDA ) [3] . Since the current treatments have to be administered annually on a community-wide basis for many years to break the infection cycle , and drug resistance may be emerging [4] , [5] , there is still an urgent need to develop novel drugs ( particularly macrofilaricidal ) . Numerous lines of evidence , in both laboratory and human trials , show that depletion of Wolbachia in filarial parasites by antibiotics ( e . g . doxycycline , tetracycline ) can kill adult worms in addition to affecting embryogenesis , mf output and worm development [6] , [7] , [8] , [9] , [10] , [11] , [12] , [13] . These studies indicate that these vertically transmitted Wolbachia endosymbionts are indispensible for their filarial hosts and represent a promising therapeutic strategy for filariasis control . Comparative analysis of available genomic sequences for Wolbachia ( wBm , GenBank accession no . AE017321 ) and its B . malayi nematode host ( GenBank accession no . EF588824 to EF588901 ) provides insight into metabolic pathways that might contribute to the mutualistic symbiotic relationship [14] . This approach can be used to aid identification of potential anti-filarial drug targets . One biochemical pathway identified as potentially important in the symbiotic relationship between wBm and its nematode host is heme biosynthesis . Heme , an iron-containing tetrapyrrole , is an essential cofactor for many proteins such as cytochromes , hemoglobins , peroxidases , and catalases , which are involved in a wide range of critical biological processes , including oxidative metabolism and electron transport . All but one of the C4-type heme biosynthetic genes are readily identified from the wBm genome ( Fig . 1 ) . The only missing step , protoporphyrinogen-IX oxidase ( PPO/hemG ) , has not been identified in many heme-producing bacteria [15] . However , all but one heme biosynthetic gene ( FC/hemH , ferrochelatase , the last step in heme biosynthesis ) is absent in the B . malayi genome [16] , implying filarial nematodes are incapable of de novo heme biosynthesis , a condition that seems to be characteristic of all or most nematodes , including Caenorhabditis elegans [17] . Filarial worms presumably salvage heme/intermediates from their surroundings and/or acquire them from their Wolbachia endosymbionts . Heme deprivation may at least partially account for the effects caused by elimination of wBm following antibiotic treatment of filarial worms . For example , it is already known that antibiotic treatment disrupts the L4 to L5 molt in B . pahangi [18] . Furthermore , heme-containing enzymes such as peroxidases have critical functions in the molting of C . elegans and orthologs exist in B . malayi [19] , [20] , [21] . In this report , we indicate that Wolbachia heme biosynthesis likely contributes to filarial worm survival and thus could be a potential anti-filarial drug target pathway .
Human heme gene cDNA clones were purchased from the Invitrogen human cDNA clone collection , except for the 5′-aminolevulinic acid synthetase cDNA clone which was purchased from Open Biosystems . B . malayi worms were purchased from TRS Labs , Athens , GA . B . malayi DNA ( including Wolbachia DNA ) was extracted using DNeasy extraction ( Qiagen ) according to the manufacturer's protocol . Based on available human , Wolbachia and E . coli sequences in the NCBI database , primers were designed with restriction endonuclease sites ( Table S1 ) and used for full-length open reading frame ( ORF ) amplification by PCR with Phusion polymerase ( New England Biolabs , NEB ) . After purification by QIAquick PCR purification ( Qiagen ) and digestion with corresponding restriction endonucleases ( NEB ) , resulting PCR products were cloned into the pET21a+ vector ( Novagen ) for protein expression with a C-terminal 6XHis-tag . Correct clones were first identified by lysed-colony PCR and then verified by DNA sequencing . For improving protein expression and solubility , human 5′-aminolevulinic acid dehydratase ( ALAD ) , Wolbachia porphobilinogen deaminase ( PBGD ) and Wolbachia ferrochelatase ( FC ) genes were codon-optimized by gene re-synthesis using DNAworks oligonucleotide designing software [22] and USER cloning methods [23] . All cloned heme genes were expressed in T7 Express competent E . coli ( NEB ) , either with or without the RIL plasmid ( Stratagene ) which encodes E . coli rare tRNAs for arginine , isoleucine and leucine . Protein expression was induced with starting OD600 0 . 3–0 . 4 , 10–100 µM isopropyl β-D-thiogalactopyranoside ( IPTG , Sigma ) , 18–48 hours at 14–16°C . The 6XHis-tagged proteins were purified under native conditions , using a nickel resin ( Qiagen ) according to a modified manufacturer's protocol . Buffers ( 100 mM Tris-HCl pH 8 . 0 , 300 mM NaCl ) containing different concentrations of imidazole ( 10–20 mM , 40–50 mM and 250 mM ) were used as the lysis , wash and elution buffers , respectively . Purity of the proteins was verified on 4–20% SDS-PAGE gels ( Invitrogen ) and protein concentrations were measured on a Nanodrop ND-1000 ( Thermo Scientific ) . Proteins were stored at −80°C in 10% glycerol for long-term storage or stored at 4°C for no more than 1 month for further analysis . Homologous protein sequences were retrieved from the NCBI database via protein-protein BLAST similarity searches and were aligned using CLUSTAL ×1 . 83 [24] . The sequence alignments were further refined manually after the removal of large gaps and evolutionarily diverse regions . Based on protein sequence alignments , gene phylogenies for the B . malayi Wolbachia heme synthesis genes were derived from both Bayesian inference ( BI ) [25] and Maximum likelihood ( ML ) [26] methods . ML trees were constructed by the PROML programs of PHYLIP package version 3 . 65 [27] with global rearrangements and randomized input order options in conjunction with estimated parameter gamma and the proportion of invariable sites obtained from TREE-PUZZLE 5 . 1 [28] calculation , in which Quartet puzzling maximum likelihood ( QP ) analysis was carried out employing the JTT-f amino acid substitution probability model with a mixed eight category gamma+invariable-sites model of rate heterogeneity and 10 , 000 puzzling steps . ML analyses were performed by subsequent applications of SEQBOOT ( 100 replicates ) , PROML and CONSENSE . BI analyses were conducted with randomly produced starting trees , JTT amino acid substitution frequencies , four category gamma+invariable-sites model , 200 , 000 generations of searches . Posterior possibilities for the best trees were calculated using a 50% majority rule . Fresh live adult male and female B . malayi worms were incubated with different concentrations of succinyl acetone ( SA , Sigma ) or methyl mesoporphyrin ( NMMP , Frontier Scientific ) ( 3 replicates/experiment , 1 adult female or 3 adult males/replicate , experiment repeated three times ) , which target ALAD and FC , respectively . Worms were cultured in RPMI-1640 with 2 mM glutamine , 25 mM HEPES ( Gibco ) with 10% Fetal Calf Serum ( Gibco ) and 100 U/ml streptomycin , 100 µg/ml penicillin , 0 . 25 µg/ml amphotericin B ( Sigma ) . Medium was changed every 2 days . SA was freshly made in water at a concentration of 500 mM before use . Both NMMP and hemin ( Frontier Scientific ) were freshly prepared as 5 mM stock concentrations in 50% ethanol containing 0 . 02 N NaOH . In NMMP tests , control worms were cultured in medium containing 1% ethanol and 0 . 0004 N NaOH ( “solvent only” ) with and without 100 µM hemin . Motility was measured daily ( similar to the method used by Rao et al [11] ) as 0 , no motility; 1 , slight movement clearly observed under microscope; 2 , minor movement readily observed by eye; 3 , non-continual moderate movement; 4 , continual moderate movement; 5 , continual active movement . C . elegans Bristol N2 was cultured and maintained according to standard protocols [29] . Drug testing was performed as follows . Eggs were extracted from gravid hermaphrodites using alkaline hypochlorite treatment followed by extensive washes in M9 buffer ( 22 mM KH2PO4 , 42 mM Na2HPO4 , 86 mM NaCl , 1 mM MgSO4 ) [29] . Eggs were allowed to hatch overnight in S-basal buffer ( 0 . 1 M NaCl , 0 . 05 M KH2PO4 pH 6 , 5 mg/ml cholesterol ) . The concentration of first-stage larvae ( L1 ) was adjusted to obtain an average of 5 to 6 live animals per well . Pre-grown concentrated E . coli OP50 was added as a food source . The compounds to be tested were added at the appropriate concentration . Worms were cultured in 96-well plates ( NUNC , Rochester , NY ) in a 100 µl volume for three days at 20°C . Twenty-four wells were cultured for each compound at a given concentration . Only one generation was followed . The number of parental animals reaching adulthood was scored . Enzyme activities were assayed using purified recombinant C-terminal 6XHis-tagged Wolbachia and human ALAD proteins ( wALAD & hALAD ) at 37°C for 15–30 min . The enzyme reactions were carried out in 100 mM Bis-Tris Propane ( BTP ) buffer ( Sigma , pH range 6 . 5–9 . 5 ) containing 1 µg protein , 5 mM substrate 5′-aminolevulinic acid ( ALA ) ( Sigma ) and 10 mM β-mercaptoethanol ( Sigma ) unless otherwise stated , in a total volume of 100 µl . All assays were initiated by the addition of ALA after enzyme pre-incubation for 20–30 min with various metal ions ( e . g . Zn2+ , Mg2+ ) and/or other reagents ( e . g . the metal ion chelator EDTA , specific enzyme inhibitor SA ) . After determining optimal reaction pH , enzyme assays were further conducted in the presence of different concentrations of substrate ALA ( for determination of Km and Vmax ) or inhibitor SA ( for calculation of EC50 ) . The reaction was stopped by mixing with an equal volume of stop buffer ( 0 . 1 M HgCl2 in 12% Trichloroacetic acid ) followed by the addition of 800 µl modified Ehrlich reagent for 10 min [30] . The product porphobilinogen ( PBG ) was subsequently estimated by measuring the absorbance at OD555 . The molar extinction coefficient for PBG ( 60 , 200 M−1 cm−1 ) was used in calculation of PBG concentration ( µmol PBG/mg of protein/h ) . E . coli hemB ( ALAD ) mutant strain RP523 [31] and HemD ( Uroporphyrinogen III synthase , UROS ) deletion mutation strain SASZ31 were obtained from the E . coli Genetic Stock Center ( http://cgsc . biology . yale . edu/ ) . E . coli hemG ( PPO ) deletion strain SASX38 [32] and hemH ( FC ) deletion strain VS200 [33] were generously provided by Dr . Harry A . Dailey , University of Georgia . The pET21a+ vectors carrying the corresponding human , Wolbachia and E . coli heme gene inserts were transformed into the above-mentioned E . coli mutant strains , both with and without RIL plasmid co-transformation alongside a vector only control , and with appropriate antibiotic selection . Transformants were selected on 20 µM hemin-containing LB plates with appropriate antibiotics and incubated at 37°C overnight . The selected transgenic clones were further tested on LB plates with no hemin addition . E . coli hemH mutants containing human , Wolbachia or E . coli FC genes , were used in growth assays . The fresh transgenic E . coli mutants ( grown to 0 . 6–0 . 8 OD600 ) were diluted to 0 . 01 OD600 before initiating growth assays in the presence or absence of different concentrations of the specific FC inhibitor , NMMP in LB medium . The E . coli cells were grown in a 30°C shaker ( 180 rpm ) for 3 h before estimating the cell growth level by measuring the final OD600 values . The final cell density for the untreated controls varied from 0 . 4 to 1 . 0 OD600 . Relative cell density was taken as a measure of toxicity . The average cell growth ratio ( final OD600/0 . 01 ) for untreated control is set at 1 . 0 .
Based upon genomic DNA sequence analyses , both humans and Wolbachia share the C4-type heme biosynthetic pathway , usually consisting of eight components ( Fig . 1 ) and phylogenies inferred by both ML and BI analyses indicate a deep evolutionary distance existing between homologues in this pathway . Sequences for all heme biosynthetic enzymes ( except for the missing PPO gene ) were obtained by database queries from diverse organisms ( 17–30 species , with exclusion of archaeal sequences due to their extreme divergence ) . They were aligned using ClustalX ( with manual refinement ) . Conserved regions ( 157–312 sites ) were used for phylogenetic reconstruction . The unrooted gene trees ( except for ALAS ) , presented in Fig . 2 and Fig . S1 , show that the gene homologues from both nematode and insect Wolbachia consistently group together and mostly within the Rickettsia subgroup of the alpha-proteobacterial cluster , sharing high amino acid sequence identities/similarities ( 70–87%/80–97% ) . By comparison , B . malayi Wolbachia heme synthesis genes only share 22–34% identities ( 29–53% similarities ) with their human homologues . Among the seven identified components of Wolbachia heme pathway ( Fig . 1 ) , two were of particular interest ( gene trees are presented in Fig . 2 ) owing to their significant divergence and biochemical properties . The ALAD gene tree shows that Wolbachia and human homologues belong to Zn2+-independent and Zn2+-dependent groups , respectively ( Fig . 2A ) , which is confirmed by the absence of the critical cysteine residues in the Zn2+-binding sites in Wolbachia ALADs , while present in human ALAD ( Text S1 ) . Similarly , it is known that human FC contains both an N-terminal extension ( encoding a mitochondrion-targeting signal ) and a C-terminal extension ( involved in formation of homodimers ) , and harbors an [Fe-S] cluster binding site ( formed by 4 cysteine residues ) [34] , [35] . However , sequence analysis of Wolbachia FCs revealed that they do not have any of these features ( Text S1 ) . This is in line with the FC phylogenetic analysis that shows significant evolutionary divergence between human and Wolbachia FCs ( Fig . 2B ) . Two analog inhibitors , SA ( 1–3 mM ) and NMMP ( 10–100 µM ) , were used in B . malayi worm ex vivo motility assays , which specifically target ALAD and FC , respectively . The results are shown in Fig . 3A–D . Motility was measured ( similar to the method used by Rao et al [11] ) . Compared to the untreated controls , both SA and NMMP lead to significantly reduced motilities of adult worms during the nine-day treatments , independent of the addition of hemin to the medium ( Fig . 3A–D ) . The tissue/cell structure of immotile worms ( scaled as 0 ) seemed degenerative after treatment and no recovery was observed even after transferring these worms to fresh medium without inhibitor . Similar results were also observed in B . malayi mf larvae assays ( data not shown ) . The effect of the inhibitors on female adult worms ( Fig . 3B , D ) appears more severe than that on male adult worms ( Fig . 3A , C ) . The free-living nematode C . elegans was used as a worm control as it lacks the heme biosynthetic pathway and does not harbor an obligate endosymbiont , so has to salvage heme from the medium for viability [17] . In the presence of hemin , NMMP ( 10–100 µM ) did not affect the growth of C . elegans larvae into adults ( Fig . 3E ) or worm fertility ( data not shown ) as expected . However , SA ( 1–3 mM ) appeared to have a non-specific inhibitory effect on C . elegans larval development ( Fig . 3E ) and resulted in 100% sterility even at the lowest concentration tested ( data not shown ) . cDNA clones in pET 21a+ of Wolbachia and human ALADs were transformed into E . coli containing the RIL plasmid and expressed as C-terminal 6XHis-tagged proteins . However , expression for hALAD was very poor , thus a codon-optimized version was made for improvement of expression . Examples of purified full-length C-terminal 6XHis-tagged wALAD ( 37 . 7 kDa ) and hALAD ( 37 . 4 kDa ) are presented in Fig . 4 . The pH profiles ( Fig . 5 ) for hALAD and wALAD enzyme activities indicate an overlapped optimal pH range ( pH 6 . 5–7 . 5 vs pH 7 . 0–8 . 5 ) . hALAD is Zn2+-dependent . At optimal pH 7 . 0 , its activity is inhibited by the metal ion chelator EDTA and recovered by addition of Zn2+ ( Fig . 6A ) . wALAD activity ( at its optimal pH 8 . 0 ) is also sensitive to EDTA inhibition , however , its activity is only restored by Mg2+ addition ( Fig 6B ) . This suggests that wALAD is Zn2+-independent , which agrees with the absence of a putative Zn2+ binding site in wALAD , while it is present in hALAD . The maximum activity ( Vmax ) of hALAD ( pH 7 . 0 ) is measured as 57 . 8±2 . 2 µmol porphobilinogen ( PBG ) /mg of protein/h and the Km value for substrate ALA is 0 . 35±0 . 06 mM ( Fig . S2A ) , while for wALAD ( pH 8 . 0 ) , the Vmax and Km values are 22 . 5±1 . 1 µmol PBG/mg of protein/h and 0 . 32±0 . 07 mM , respectively ( Fig . S2B ) . hALAD activity is about 2 . 5 times higher than that of wALAD with similar ALA substrate binding affinity . SA is a specific ALAD inhibitor with different potency depending on the particular ALAD species involved . The sensitivities of wALAD and hALAD to SA are presented in Fig . 7 . Both enzymes could be inhibited by SA , but with strikingly different inhibition profiles - EC50s for wALAD and hALAD were ∼109 µM and ∼0 . 18 µM , respectively , a 600 fold difference . This likely reflects a significant structural variation between these two enzymes . To test whether the cloned Wolbachia heme biosynthetic genes are functional , we performed complementation assays using E . coli heme deficient mutant strains ( hemB , hemD , hemG and hemH ) . These assays complement the in vitro enzyme assays described above . The corresponding Wolbachia and human heme genes , cloned in the pET21a+ vector , were tested for activity , alongside a pET21a+ vector negative control . An E . coli hemB mutant , transformed with the pET21a+ plasmid fails to grow on LB plates , unless hemin is added to the media . Transformations of the E . coli hemB mutant strain with wALAD or hALAD constructs result in strong colony growth on LB plates without hemin addition , similar to wild type E . coli growth , indicating functional expression of these two heme genes in E . coli . Similar complementation was observed for E . coli hemD and hemH mutants with Wolbachia/human UROS and Wolbachia/human FC genes , respectively . As mentioned previously , Wolbachia PPO is still un-recognized . It has been reported that overexpression of E . coli CPO might function as PPO [36] and therefore we tested an E . coli hemG mutant with the wBm CPO construct . No complementation was observed , even under IPTG induction . As a positive control , an E . coli hemG mutant was functionally complemented by transformation with the human PPO construct . Our enzyme assays and E . coli complementation tests verified that Wolbachia ALAD , UROS and FC genes are functional . Since ALAD is the second step and FC is the last step of the heme biosynthesis pathways , our results indicate that Wolbachia has the ability to synthesize endogenous heme . FC is a potential drug target based on its evolutionary divergence and differing protein features compared to its human homologue . Human FC ( hFC ) is an [Fe-S] protein , while Wolbachia FC ( wFC ) lacks [Fe-S] clusters . A potent FC inhibitor - NMMP was used in growth inhibition assays using E . coli hemH mutants transformed with wFC , hFC or E . coli FC ( EcFC ) . The growth of all transformed strains was inhibited by NMMP with significantly different drug sensitivities ( sensitivity level: hFC>EcFC>wFC ) as compared to their non-treated controls ( Fig . 8 ) . The inhibition was readily overcome by the inclusion of hemin in the growth medium ( Fig . 8 ) . This assay further supports the possibility of using a rapid E . coli-based complementation assay to screen for specific inhibitors that will differentially target the Wolbachia heme synthesis enzymes .
Anti-Wolbachia chemotherapeutic treatment is an emerging approach for filariasis control [37] . Comparative genomic and bioinformatic analyses are shedding light onto the mutualistic symbiotic relationship between Wolbachia and its filarial host , revealing essential biochemical pathways in the bacterial endosymbiont that might provide critical metabolites for its worm host survival [14] . Based on our data from ex vivo worm assays , phylogenetic analyses , gene expression and purification profiles , in vitro enzyme assays , and E . coli complementation results , we have evaluated the possibility of the Wolbachia heme pathway as an anti-filarial drug target set . No heme biosynthetic genes except ferrochelatase ( FC , the last step ) have been identified from the worm host B . malayi genome sequence [16] . It appears that the nematode is not capable of synthesizing heme de novo , and thus may have to acquire heme from its Wolbachia endosymbiont or salvage environmental heme or both . The motilities of both B . malayi male and female worms are significantly reduced when exposed to the heme biosynthesis inhibitor SA ( targeting ALAD ) , even in the presence of hemin in the medium ( Fig . 3A–B ) . However , we believe that this effect was non-specific because a similar phenotype was observed when C . elegans , a heme auxotroph , which does not have the biosynthetic pathway at all , was exposed to SA ( Fig . 3E ) , suggesting that SA may have some unspecific effect on B . malayi . In contrast , NMMP appears to be potent and specific in its inhibitory effect on the heme pathway , since it has an in vivo effect on B . malayi ( Fig . 3C–D ) , but not on C . elegans ( Fig . 3E ) . This inhibition can not be rescued by hemin ( Fig . 3C–D ) , implying that B . malayi possibly lacks the capability of salvaging environmental heme , as has been demonstrated for C . elegans [38] . However , it should be noted that the B . malayi genome encodes for FC [16] and the effect of NMMP on B . malayi ferrochelatase ( BmFC ) could contribute to the inhibition of the worm viability ( Wu et al , in preparation ) . We have no experimental evidence for a direct effect of NMMP on Wolbachia . Thus , the survival of B . malayi might be dependent on bacterial derived heme from the Wolbachia heme biosynthetic pathway and/or a functional BmFC which may utilize porphyrin intermediates from the endosymbiont or the environment . Given the observed differences between human and B . malayi FC proteins , specific inhibition of nematode FC in infected humans could be a potential drug target ( Wu et al , in preparation ) . Because mammals also synthesize heme via the C4-type pathway like Wolbachia , caution is needed when considering Wolbachia heme biosynthetic enzymes as anti-filarial drug targets . However , phylogenetic analyses ( Fig . 2 & Fig . S1 ) revealed that significant evolutionary distances exist among the human and Wolbachia heme genes ( except for ALAS ) , as shown by their low sequence similarities/identities ( 22–34%/29–53% ) . We corroborate these in silico studies with biochemical/pharmacological assays by cloning the Wolbachia and human genes and expressing the proteins . Difficulty in protein expression or purification of soluble proteins for hALAD , wPBGD & wFC was addressed by codon optimization . This helped improve expression levels , but still failed to yield soluble proteins , with the exception of synthetic hALAD . Based on the significant difference ( ∼600 fold ) in drug sensitivities between wALAD and hALAD enzymes ( Fig . 7 ) , large amounts of recombinant wALAD and hALAD proteins are currently being prepared for high throughput screening as part of anti-Wolbachia ( A-WOL ) drug discovery and development program ( http://www . a-wol . com ) . Crystal structures are available for ALAD from human , mouse , yeast , Pseudomonas aeruginosa and Chlorobium vibrioforme ( including a structure for yeast ALAD in complex with SA ) from Protein Data Bank ( PDB , http://www . rcsb . org ) . With this information , structural studies of Wolbachia and human ALAD by homology modeling may help explain the observed differences in SA sensitivity and help optimize identification of lead compounds obtained from the on-going drug screening effort . Functional Wolbachia heme synthesis activity for several genes ( ALAD , UROS and FC ) along with their human homologues was demonstrated by complementation tests using the corresponding E . coli heme deficient mutants ( hemB , hemD and hemH ) . As mentioned above , PPO , the penultimate step in heme pathway , is missing in Wolbachia and is unidentifiable from many other bacterial genomes , e . g . Rickettsia [15] . The E . coli hemG mutant was readily complemented by the human PPO gene; however , unlike E . coli CPO , Wolbachia CPO was incapable of rescuing PPO deficiency . Since PPO function is required for heme biosynthesis , it is possible that an unidentified oxidase may function as PPO in Wolbachia . We have attempted to express and purify recombinant wFC ( codon-optimized ) and hFC proteins , however the yield of pure soluble protein was limited due to the formation of inclusion bodies . The availability of an E . coli hemH deficient mutant and functional complementation by wFC and hFC genes permit an alternative E . coli-based inhibition assay and drug-screening strategy . FC , the final step in heme biosynthesis , is responsible for insertion of iron into protoporphyin IX ( PPIX ) to form heme . NMMP is a strong PPIX analog , and competitively binds to the FC active site with Ki values in the nM range [39] . It was reported previously that FC enzymes could have dramatically different sensitivities to NMMP inhibition ( >1000 fold , e . g . Human FC vs chicken FC ) due to the differences existing in their active sites [40] . The E . coli hemH mutant , complemented with wFC , hFC and EcFC , is not sensitive to NMMP when grown in the presence of hemin . Significantly different sensitivities to NMMP were detected between human and Wolbachia FCs in absence of hemin , with human FC being much more sensitive . This may be accounted for by the structural difference in their active sites . Cell-based drug screening may help identify compounds more specifically targeting wFC instead of hFC . FC crystal structures from several species ( human , yeast and Bacillus ) are also available in PDB with a crystal structure of Bacillus subtilis FC complexed with NMMP [41] . In silico comparative studies based on molecular modeling can be conducted for Wolbachia and human FCs . E . coli mutants complemented by other Wolbachia/human heme genes ( ALAD , UROS ) will be used as potential screens as well for identification of compounds specifically inhibiting selected Wolbachia heme biosynthesis enzymes . Our data suggest that the Wolbachia heme biosynthetic pathway is a potential anti-filarial drug target due to its requirement for survival of both Wolbachia and its filarial host . The presumptive transporters , responsible for heme trafficking , could be drug targets as well . However , it still remains unknown how the heme/heme intermediates might transfer from Wolbachia to its filarial host . No enzymes involved in traditional heme catabolism ( e . g . heme oxygenase ) have been identified from the B . malayi genome sequence . It is still an open question how transport , degradation , and regulation of heme occur in filarial parasites .
|
Human filarial nematodes are causative agents of elephantiasis and African river blindness , which are among the most debilitating tropical diseases . Currently used drugs mainly affect microfilariae ( mf ) and have less effect on adult filarial nematodes , which can live in the human host for more than a decade . Filariasis drug control strategy relies on recurrent mass drug administration for many years . Development of novel drugs is also urgently needed due to the threat of drug resistance occurrence . Most filarial worms harbor an obligate endosymbiotic bacterium , Wolbachia , whose presence has been identified as a potential drug target . Comparative genomics had suggested Wolbachia heme biosynthesis as a potential drug target , and we present an analysis of selected enzymes alongside their human homologues from several different aspects—gene phylogenetic analyses , in vitro enzyme kinetic and inhibition assays and heme-deficient E . coli complementation assays . We also conducted ex vivo Brugia malayi viability assays using heme pathway inhibitors . These experiments demonstrate that heme biosynthesis could be critical for filarial worm survival and thus is a potential anti-filarial drug target set .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"Methods",
"Results",
"Discussion"
] |
[
"biochemistry",
"evolutionary",
"biology/human",
"evolution",
"microbiology/microbial",
"evolution",
"and",
"genomics",
"infectious",
"diseases/helminth",
"infections",
"microbiology/parasitology",
"microbiology/microbial",
"physiology",
"and",
"metabolism"
] |
2009
|
The Heme Biosynthetic Pathway of the Obligate Wolbachia Endosymbiont of Brugia malayi as a Potential Anti-filarial Drug Target
|
Protein export is central for the survival and virulence of intracellular P . falciparum blood stage parasites . To reach the host cell , exported proteins cross the parasite plasma membrane ( PPM ) and the parasite-enclosing parasitophorous vacuole membrane ( PVM ) , a process that requires unfolding , suggestive of protein translocation . Components of a proposed translocon at the PVM termed PTEX are essential in this phase of export but translocation activity has not been shown for the complex and questions have been raised about its proposed membrane pore component EXP2 for which no functional data is available in P . falciparum . It is also unclear how PTEX mediates trafficking of both , soluble as well as transmembrane proteins . Taking advantage of conditionally foldable domains , we here dissected the translocation events in the parasite periphery , showing that two successive translocation steps are needed for the export of transmembrane proteins , one at the PPM and one at the PVM . Our data provide evidence that , depending on the length of the C-terminus of the exported substrate , these steps occur by transient interaction of the PPM and PVM translocon , similar to the situation for protein transport across the mitochondrial membranes . Remarkably , we obtained constructs of exported proteins that remained arrested in the process of being translocated across the PVM . This clogged the translocation pore , prevented the export of all types of exported proteins and , as a result , inhibited parasite growth . The substrates stuck in translocation were found in a complex with the proposed PTEX membrane pore component EXP2 , suggesting a role of this protein in translocation . These data for the first time provide evidence for EXP2 to be part of a translocating entity , suggesting that PTEX has translocation activity and provide a mechanistic framework for the transport of soluble as well as transmembrane proteins from the parasite boundary into the host cell .
Five species of Plasmodium parasites cause human malaria . Of these P . falciparum is responsible for the majority of the over 500’000 annually recorded malaria deaths [1] . The pathology of malaria is caused by the continuous propagation of the parasite within red blood cells ( RBCs ) . In this phase P . falciparum parasites modify extensively the host RBC by exporting hundreds of different proteins into the infected cell . These modifications include host cell surface changes that cause the sequestration of infected RBCs ( iRBCs ) in the vasculature , a phenomenon considered to be a major contributor to parasite virulence [2] . Other changes are required for nutrient acquisition , to adjust RBC rigidity and to facilitate protein trafficking in the host cell [3] . Protein export is therefore central for blood stage development and malaria pathology . Two general types of exported proteins have been described in malaria parasites . The first group contains a five amino acid motif termed Plasmodium export element ( PEXEL ) or host targeting signal ( HT ) [4–6] . The second group , termed PEXEL negative exported proteins ( PNEPs ) , is defined by the absence of a PEXEL/HT signal [7 , 8] . Both groups comprise soluble and transmembrane ( TM ) proteins . Despite the distinction into PNEPs and PEXEL proteins , both types of proteins appear to share a similar export domain [9] and at least at one point during their export , the same trafficking factors are involved [10 , 11] . Many aspects of the pathways exported proteins use to reach the host RBC are still unclear . The parasite replicates in a parasitophorous vacuole ( PV ) formed by a PV membrane ( PVM ) [12] . Exported proteins therefore have to cross two membranes , the parasite plasma membrane ( PPM ) and the PVM . A previously postulated protein translocation machine termed ‘Plasmodium translocon of exported proteins’ ( PTEX ) [13] , is involved in this export step [10 , 11] . Of the 5 known PTEX components , heat shock protein 101 ( HSP101 ) , and a parasite-specific protein termed PTEX150 are essential for protein export [10 , 11] . Much less clear is the role of the suspected PTEX membrane pore component EXP2 for which no functional data are available in P . falciparum . Recent work in the apicomplexan Toxoplasma gondii showed that PfEXP2 was able to functionally replace a protein implicated in the solute pore activity at the PVM of this parasite [14] . This raised the possibility that PfEXP2 may have an additional or differing role than in protein export , which would also explain the finding that EXP2 , but not HSP101 , is expressed in liver stage parasites [15 , 16] . Although some of its components are clearly essential for protein export , PTEX is still a translocon in concept , as translocation activity has so far not been shown and a function upstream of membrane translocation would also satisfy the findings so far [3 , 17] . It is also puzzling that it promotes trafficking of both TM and soluble proteins . PTEX is situated on the luminal face of the PVM [13] . While soluble proteins directly reach PTEX after release by transport vesicles into the PV , TM proteins embedded in the transport vesicle membrane will end up integral to the parasite plasma membrane ( PPM ) [18] . There is evidence that these proteins are then extracted out of the PPM in an unfolding dependent step [9] , but the trafficking events at the PPM , PV and PVM remain obscure ( reviewed in [17] ) . Substrates fused to conditionally foldable domains have been used to study translocation processes in various organelles and systems , for instance in mitochondria [19 , 20] . A widely used tool for such studies is mouse dihydrofolate reductase ( mDHFR ) , a protein that can be stabilized in its folding upon addition of a small ligand such as WR [19] . If mDHFR is fused to a translocation substrate , addition of the ligand will render the substrate translocation incompetent . The resulting defect in transport is indicative of membrane translocation and excludes vesicular trafficking for the transport step analyzed . This system has previously been used in malaria parasites to show that soluble truncated PEXEL reporters [21] , TM PNEPs [9] and soluble PNEPs [7] , require unfolding to reach the host cell . However , while clearly indicative of translocation , these experiments did not link this activity with the proposed translocon PTEX . Here we resolve the translocation steps in the parasite periphery , demonstrating that two unfolding events are required for TM proteins to reach the host cell . Further we show that all known types of exported proteins converge at the second translocation step at the PVM and that inducibly jamming this pore arrests general protein export and parasite development . Crucially , we for the first time obtained stable translocation intermediates and use this to provide evidence that links EXP2 with the translocating complex , suggesting that PTEX has translocation activity .
The previously used exported TM protein REX2 fused with mDHFR accumulated at the PPM after addition of WR , leading to an arrest at the first step when exported TM proteins leave the parasite cell [9] . In agreement with previous data for a soluble mDHFR fused PEXEL reporter [21] , this arrest was not reversible ( S1A Fig ) . This precluded the use of this system to study the subsequent transport steps by simply removing WR . We therefore replaced mDHFR with the bovine pancreatic trypsin inhibitor ( BPTI ) , a protein that forms a translocation incompetent folded structure in oxidising but not reducing environments based on 3 disulfide bonds [20] . The rational was that BPTI would only form a folded structure once the construct reached the PV , which is thought to be an oxidising environment [22 , 23] . In contrast , extraction out of the PPM , when BPTI still faces the reducing cytoplasmic side of the PPM , would not be affected ( see S1B Fig for schematic ) . To first test whether the PV indeed is an oxidative environment where BPTI can fold into a translocation incompetent state and to assess whether PVM translocation was sensitive to this domain , we fused BPTI to REX3 , a soluble exported protein that is directly delivered from the secretory pathway into the PV . Export of REX3 was indeed sensitive to BPTI , as evident by a clear , although only partial , accumulation of the construct in the parasite periphery ( Fig 1A ) . In contrast , a REX3 control construct fused with a mutated BPTI [unable to form the stabilising disulfide bridges [24]] , did not show an accumulation in the parasite periphery but was fully exported ( Fig 1B ) , excluding oxidation unrelated trafficking defects . This suggested that the system is suitable to obtain translocation incompetent reporters in the PV and we next fused BPTI to the TM PNEP REX2 . The resulting construct ( REX2-BPTI-GFP ) showed a strong accumulation in the parasite periphery ( Fig 1C ) , suggestive of a translocation-dependent step of TM proteins after passing the PPM . This block was not absolute , as additional fluorescence was detected in the Maurer's clefts ( Fig 1C ) . A control construct with a mutated BPTI was fully exported to the Maurer's clefts ( Fig 1D ) , again excluding oxidation unrelated trafficking defects . Protease protection assays ( see Fig 1E for schematic ) with REX2-BPTI-GFP indicated that the REX2-BPTI-GFP molecules were in the PV in their entireness ( Fig 1F ) and hence had completed the extraction out of the PPM . These results are consistent with an oxidation state-dependent arrest in export in the PV due to fusion with BPTI ( Fig 1G ) and indicated that after PPM extraction exported TM proteins undergo a second unfolding-dependent translocation at the PVM . Unexpectedly , when we analysed two further TM PNEPs fused to BPTI ( SBP1-BPTI-GFP and MAHRP1-BPTI-GFP ) , these constructs showed no accumulation in the parasite periphery but were efficiently exported ( Fig 1H and S1C Fig ) . The most noticeable difference between these PNEPs and REX2 is their larger size . We reasoned that protein length might influence whether BPTI can fold in an intermediate step in the PV . To test this idea , we shortened the N- or C-terminus of SBP1 in SBP1-BPTI-GFP by inserting deletions previously reported not to affect export of this protein [25] . The protein with the shortened N-terminus ( SBP1ΔN-BPTI-GFP ) was not blocked in export ( Fig 1I ) . In contrast , deletion of most of the C-terminus ( SBP1ΔCBPTI-GFP ) resulted in a strong block in the parasite periphery with some left over export to the Maurer's clefts ( Fig 1J ) , comparable to the result with REX2-BPTI-GFP . This was not due to a general export defect introduced by the C-terminal deletion but due to folding of BPTI , as a version of SBP1ΔC-BPTI-GFP with the mutated BPTI was exported ( Fig 1K ) . These results suggested that the length of the C-terminus , specifically the region between the TM and the blocking domain , decides whether BPTI has the chance to fold in the PV and the protein gets blocked in further export . To confirm this and exclude an SBP1-specific effect , we extended the C-terminus in REX2-BPTI-GFP by inserting 3 consecutive REX2 C-termini ( REX2+3C-BPTI-GFP ) . This turned REX2 into a protein unaffected by BPTI as this construct was fully exported ( Fig 1L ) . This lends support for a scenario where translocation substrates with a large distance between the TM and the blocking domain already engage the PVM translocon while still being extracted out of the PPM which would prevent release into the PV and oxidation-state dependent folding of BPTI . In contrast , such a direct ‘hand over’ may not be possible for proteins with a short C-terminus where BPTI would already become exposed to the oxidising milieu of the PV to form the folding stabilising disulfide bonds before the protein can get access to the PVM translocon ( see model S1D and S1E Fig ) . Prompted by the differences seen with different PNEP-BPTI constructs , we fused SBP1 and MAHRP1 with mDHFR to confirm that they at all require unfolding to pass from the parasite into the host cell . Analogous to REX2-mDHFR-GFP [9] , these constructs were efficiently exported to the Maurer's clefts and conditionally blocked in the parasite periphery in the presence of the ligand WR that prevents unfolding of the appended mDHFR domain ( Fig 2A and 2B ) . This indicated translocation as the mode of export for these TM PNEPs , similar to REX2 . However , compared to REX2-mDHFR-GFP ( [9] and Fig 2C ) , there were three notable differences in these constructs after arresting export with WR: firstly , the arrest phenotype was leaky in many cells , i . e . besides the prominent peripheral stain , there was also a detectable signal at the Maurer's clefts ( arrowheads Fig 2A and 2B ) . Secondly , the fluorescence pattern in the parasite periphery was unusual as it included small , mobile , worm-like protrusions reaching into the host cell ( arrows Fig 2A and 2B ) . Thirdly and most remarkably , the internal control ( co-expressed REX2mCherry ) also showed a WR-dependent block in export even though it lacked an mDHFR domain ( Fig 2A and 2B , compare to Fig 2C ) . REX2 fusion with an inverted order of GFP and mDHFR ( REX2-GFP-mDHFR , generated in a failed attempt to obtain a reversible mDHFR-based block ) , showed a similar phenotype to SBP1 and MAHRP1 and differed from REX2-mDHFR-GFP ( Fig 2D ) . Hence , the difference observed was not specific for the exported protein used . These data indicated that in contrast to REX2-mDHFR-GFP , the export-blocked version of these constructs remained arrested in a translocon that also trafficks REX2mCherry and that arresting these constructs in the process of translocation thereby prevented the passage of REX2mCherry . This effect was clearly caused by the mDHFR fusion protein , as in a subpopulation of cells not expressing the mDHFR construct , REX2mCherry was correctly trafficked to the Maurer's clefts in the presence of WR ( S2A Fig ) . In contrast , REX2mCherry was always arrested in parasites harbouring the GFP-tagged mDHFR fusion ( S2A Fig ) . We termed this effect a 'co-block' as it was caused by a protein fused with mDHFR that was arrested as a stable intermediate in the translocon and prevented passage of the fully translocation competent mCherry control . The export-arrested form of REX2-mDHFR-GFP , which does not cause a co-block , was previously found to be located at the PPM [9] . We investigated the site of arrest for the co-blocking SBP1-mDHFR-GFP . Initial protease protection experiments indicated that the protein is neither in the PPM nor PVM but in the PV . This was based on the finding that after permeabilisation of the PVM with saponin , addition of protease K digested SBP1-mDHFR-GFP down to its protease resistant core ( mDHFR-GFP ) which is only possible if this protein is freely accessible in the PV ( S2B Fig ) . However , due to the proportionally small size difference of this core compared to the protected fragment ( if the protein were inserted up to the blocking domain into the PPM ) , we increased the size of the potential protected fragment by appending a protease sensitive domain [a mutated PH domain [26]] to the C-terminus of the construct ( Fig 2E ) . Similar to SBP1-mDHFR-GFP this construct ( SBP1-mDHFR-GFP-PHmut ) was fully exported and conditionally arrested in the parasite periphery after addition of WR ( Fig 2E ) . Protease protection assays showed that in the blocked state the C-terminal PH part was also fully protease accessible in the PV , indicating that the SBP1-mDHFR-GFP-PHmut molecules had entirely passed the PPM ( Fig 2F ) . Hence , the site of block differed from that of REX2-mDHFR-GFP . The co-blocked REX2-mCherry control was also found in the PV ( S2B Fig ) . The presence of the co-blocked molecules in the PV indicated that the site of arrest of the co-blocking construct is the PVM and that PPM extraction is not prevented by clogging the PVM translocons . This further supports a two-step model of translocation for TM proteins . Next we tested whether other kinds of exported proteins besides REX2mCherry were co-blocked by PNEP mDHFR-fusions arrested in translocation . To this end we generated doubly transfected parasites as well as parasites expressing two individual proteins from the same open reading frame using a skip peptide [[27 , 28] , see S3 Fig for demonstration of suitability of this approach] to co-express SBP1-mDHFR-GFP with mCherry tagged members of each of the different known groups of exported proteins . The co-expressed proteins included the soluble PEXEL proteins REX3 and KAHRP , the soluble PNEP MSRP6 , and the TM PEXEL protein STEVOR . In each case the co-expressed mCherry fusion protein was hindered in export if SBP1-mDHFR-GFP trafficking was arrested with WR in the translocon ( Fig 3A–3D ) . Similarly , REX2-GFP-mDHFR ( the domain order that in contrast to REX2-mDHFR-GFP led to a co-block of REX2mCherry ) caused a co-block of the PEXEL protein KAHRP ( S4A Fig ) . These data were also confirmed with endogenous exported proteins detected by IFA using specific antisera: a WR-dependent co-block of the late expressed MSRP6 and KAHRP was seen in SBP1-mDHFR-GFP expressing cells ( S4B Fig ) . In contrast , proteins expressed earlier in the cycle ( before SBP1-mDHFR-GFP under the crt promoter was expressed ) , were unaffected ( S4C Fig ) . Taken together , these data show that TM PNEPs arrested during translocation hinder the passage of all known types of exported proteins , indicating that a single kind of protein conducting pore is used by all exported proteins to cross the PVM . So far it was not tested whether PEXEL TM proteins also require unfolding for export . Our data indirectly indicated that they require translocation , as they were co-blocked by arrested mDHFR fusions , suggesting passage through the same pore . A shared pathway was also indicated by their sensitivity to inactivation of PTEX components [10 , 11] . Indeed , when PTP1 [a PEXEL protein with a single predicted TM [29]] and STEVOR [two predicted TMs [30]] were expressed as mDHFR-fusions , their export was conditionally arrested when WR was added ( Fig 4A ) . Similar results were obtained with the full length soluble PEXEL protein KAHRP ( S5 Fig ) . Together with the previous data [7 , 9 , 21] this indicates that all known types of exported proteins require a membrane translocation step when they pass from the parasite into the host cell . Next we tested whether it is possible to obtain arrested translocation intermediates of PEXEL proteins that induce a co-block . To this end we generated double transfectants expressing either REX3-mDHFR-GFP or PTP1-mDHFR-GFP together with the TM PNEP REX2mCherry . In the case of REX3-mDHFR-GFP , addition of WR caused a co-block of REX2mCherry ( Fig 4B ) . These results show that other types of proteins than TM PNEPs can induce a co-block . As REX3 is directly released into the PV , the PVM translocon has to be the site where the translocation intermediates are arrested and cause the co-block , consistent with the data obtained with SBP1-mDHFR-GFP ( Fig 2 ) . In contrast to the result with REX3-mDHFR-GFP , no co-block was observed with PTP1-mDHFR-GFP ( Fig 4C ) . It is noteworthy that PTP1-mDHFR-GFP in WR treated parasites did not show worm-like protrusions extending from the PVM , similar to REX2-mDHFR-GFP which is also not co-blocking . Of the constructs tested so far , those with the capacity to cause a co-block were also insensitive to fusion with BPTI ( Figs 1 and 2 ) . Prompted by this correlation , we tested whether the capacity to induce a co-block ( as judged by co-expression with REX2mCherry ) also depended on the length of region between the TM and the blocking domain . Indeed , extension of this region in REX2 ( REX2+3C-mDHFR-GFP ) turned this protein into a co-blocker ( Fig 5A ) whereas shortening this region in SBP1 ( SBP1ΔCmDHFR-GFP ) changed it into a non-co-blocking protein ( Fig 5B ) . Next we tested whether this was the reason for the failure of PTP1-mDHFR-GFP to induce a co-block . PTP1 has a short C-terminus of 27 amino acids . Extension of the PTP1 C-terminus in this construct ( PTP1-mDHFR+3C-GFP ) turned this protein into a co-blocker ( Fig 5C ) . Similar to REX2 ( Fig 1 ) , export of PTP1 was sensitive to BPTI ( Fig 5D ) , whereas the version with the extended C-terminus ( PTP1-BPTI+3C-GFP ) again was insensitive ( Fig 5E ) . These findings support the idea that long C-termini enable engagement with the PVM translocon during PPM extraction and that this is responsible for both , failure of BPTI folding and induction of the co-block ( Fig 5F , S1D and S1E Fig ) . This also further emphasises the similarities in the trafficking modalities of PNEPs and PEXEL proteins , as it affects both groups of proteins alike . The jammed translocon prevented export of all so far tested exported proteins and provided the opportunity to apply an inducible pan-export block . To assess the effect of generally blocking protein export on parasite growth , we generated an integration parasite line that expresses SBP1 fused to mDHFR-GFP from the endogenous locus ( S6A and S6B Fig ) . We chose SBP1 , because it is early expressed [31] and has no essential role for in vitro growth [32 , 33] . The resulting protein , SBP1-mDHFR-GFPendo , was correctly trafficked to the Maurer's clefts and arrested after addition of WR ( Fig 6A ) . As this protein was expressed much earlier in the cycle than the mDHFR fusions under episomal crt control , a co-block was now also observed for early expressed endogenous proteins , such as REX1 , REX2 and MAHRP2 , in addition to late expressed proteins like KAHRP ( Fig 6B and S6 Fig ) . Furthermore , the use of an integration cell line ascertained that all cells expressed the co-blocking construct . Growth assays showed that the parasites with the arrested SBP1-mDHFR-GFPendo had a strongly reduced growth rate compared to control ( Fig 6C ) . Giemsa stained smears of export blocked and control cultures revealed a delayed parasite development evident by the accumulation of young trophozoite stage parasites in the export-blocked culture , whereas the controls grew normally ( Fig 6D and S6D Fig ) . Translocation activity has so far not been demonstrated for PTEX and the functional role of the proposed pore component EXP2 is unclear [14] . We took advantage of our constructs stuck in translocation to determine whether this arrest involves the proposed translocation pore EXP2 using immunoprecipitations ( IP ) . To this end we generated a cell line expressing 3xHA tagged EXP-2 from the endogenous locus ( EXP2-3xHAendo ) and further transfected it with SBP1-mDHFR-GFP ( Fig 7A , S7A and S7B Fig ) . As expected , treatment with WR led to an arrest of export of SBP1-mDHFR-GFP in the parasite periphery where it co-localised with EXP2-3xHA by IFA ( Fig 7B ) , as previously shown for the PEXEL leader of GBP fused to mDHFR-GFP , a construct that also co-localised with EXP2 when arrested in export [34] . As to our knowledge PTEX was never defined using IP with EXP2-HA , we first wished to confirm that EXP2-3xHA indeed pulls down other PTEX components . IP using anti HA-beads with EXP2-3xHAendo parasites treated with the protein crosslinker DSP followed by mass spectrometry analysis of the eluates revealed the PTEX components PTEX150 , HSP101 , and PTEX88 as well as several other proteins that may include further PTEX interaction partners ( Fig 7C and S3 Table ) . Hence , EXP2-HA was part of the expected PTEX complex [13] . Next we IPed EXP2-3xHA from parasites that expressed SBP1-mDHFR-GFP in presence of WR ( to arrest SBP1-mDHFR-GFP in the translocon ) and absence of WR ( control ) . While EXP2-3xHA co-IPed SBP1-mDHFR-GFP when this protein was stuck in translocation , this was not the case in the control ( Fig 7D , see Fig 7F for quantification of enrichment ) . SERA5 , a soluble molecule of the PV , to control for non-specific interactions , was not co-IPed . No DSP was used for these experiments but similar results were obtained when DSP treated parasites were used ( S7C Fig ) . This indicated that substrates arrested during translocation are in a complex with EXP2 . To confirm these findings we carried out the reciprocal experiment and tested whether the substrate ( SBP1-mDHFR-GFP ) arrested in the translocon co-IPed the proposed pore component EXP2-3xHA . EXP2-3xHA was indeed enriched after IP of translocon-arrested SBP1-mDHFR-GFP if compared to control parasites without WR ( Fig 7E ) . As SBP1-mDHFR-GFP also showed some enrichment in WR+ over control after the IP ( potentially due to greater stability of the folding stabilised molecule ) , we quantified the band intensities which showed that this was not significant . In contrast the enrichment of the co-IPed EXP2-3xHA was significant ( n = 3 , Fig 7F ) . Taken together these results indicate that SBP1-mDHFR-GFP stuck in translocation is found in a complex with EXP2 . In order to confirm this with a second co-block inducing protein , we repeated the anti-HA IP with EXP2-3xHAendo parasites expressing REX2-GFP-mDHFR . Again this substrate was detected in the WR+ eluate but not in the control , demonstrating co-purification with EXP2-HA when this substrate was arrested during translocation ( S7D Fig ) . In contrast , EXP2-3xHAendo parasites expressing REX2-mDHFR-GFP ( which is not co-blocking and arrests at the PPM extraction step preceding PTEX ) , did not co-IP this protein in WR-treated parasites compared to control ( Fig 7G ) . Hence , only the second of the two consecutive unfolding dependent events involves EXP2 . A further candidate of the PTEX components that might interact with the arrested substrate is HSP101 , the ATPase that may unfold the substrate . To test whether this is the case or not , we generated a cell line expressing 3HA-tagged HSP101 from the endogenous locus ( S7E and S7F Fig ) and co-transfected a construct expressing SBP1-mDHFR-GFP . Unlike EXP2-3xHA , HSP101-3xHA did not pull down the substrate after anti-HA IP ( S7G Fig ) . Whilst this could indicate a disassociation of HSP101 from PTEX when substrates are arrested in the translocon , some HSP101 was still interacting with EXP2 in the substrate-arrested state as judged by its presence in the EXP2-3xHA IPed fraction ( Fig 7D ) . Hence , whilst some HSP101 still appears to be part of the complex containing the arrested substrate , it does not seem to be in direct contact with the substrate .
Here we for the first time obtained intermediates of exported proteins inducibly and stably arrested during translocation into the host cell . Intriguingly , these translocation intermediates prevented the transport of all known types of exported proteins , demonstrating that the actual translocation is a point of convergence for all exported proteins and a single kind of protein-conducting channel mediates export . Our data further support a two-step translocation process for exported TM proteins which are first extracted out of the PPM and then translocated into the host cell in a second unfolding-dependent process at the PVM . Blocking translocation of all exported proteins across the PVM made possible to inhibit general protein export and strongly reduced parasite growth . Similarly to the phenotype observed when HSP101 and PTEX150 were knocked down [10 , 11] , we observed slowed parasite development and an accumulation of young trophozoite stage parasites , although the growth arrest was not absolute . Crucially , our IP data now link the proposed PTEX pore component EXP2 with the entity carrying out the translocation step , fitting with the site of block and the proposed function of PTEX [10 , 11 , 13] . Our substrate was arrested in the process of being translocated by preventing unfolding of its C-terminal fusion part . The fact that other proteins could then not pass the PVM clearly shows that the arrested substrate remained in the translocon , likely partly inserted into the membrane channel . Hence , our data support the idea that EXP2 is part of the translocation pore and that PTEX has translocation activity , although it cannot fully exclude a different role of EXP2 in the complex , especially as the link between substrate and EXP2 may also be indirect . However , if there is an indirect link , it does not seem to be via HSP101 , as we failed to detect an interaction of arrested substrate with this protein . It should also be noted that we cannot formally rule out that EXP2 at the same time also interacts with a complex other than PTEX that in actual fact carries out the translocation in which case PTEX would have a different , translocon-proximal essential function in export . Finally it is noteworthy that in contrast to the co-blocking SBP1-mDHFR-GFP and REX2-GFP-mDHFR no interaction was detected with export blocked REX2-mDHFR-GFP , a construct that is not co-blocking and arrests at the PPM [9] , demonstrating the specificity of the IP results . Recent data highlighted the possibility of an alternative or dual role of EXP2 as part of a solute transporter [14 , 16] both of which would be congruent with recent findings indicating that EXP2 is important for the growth of P . berghei in mice [15] . Our data points to a role in protein export for EXP2 but cannot exclude a dual role . In this respect it is noteworthy that our IP data identified further potential interaction partners of EXP2 and it will be interesting to determine whether these constitute further components of PTEX , interaction partners that ( together with EXP2 ) form a second type of pore conducting solutes or are simply proteins abundant in the PV crosslinked to PTEX by chance . Two findings indicate that fitting with the location for PTEX , the PVM is the site where the co-block inducing translocation intermediates are arrested . Firstly , a soluble PEXEL protein ( a type of protein directly released into the PV that requires translocation at the PVM but not the PPM ) , also induced a co-block . Secondly , exported TM proteins accumulated in the PV when they were co-blocked , indicating that their extraction out of the PPM was not affected . It should be noted that also the co-blocking construct itself will be co-blocked when all translocation sites are clogged and it can be assumed that this population of the construct by far exceeds the population stuck in translocons . Our data provide mechanistic insights into the translocations of TM proteins at the PPM and the PVM . The importance of the distance between the blocking domain and the TM ( henceforth named spacer ) for ( i ) sensitivity to fusion with BPTI and ( ii ) the co-blocking capacity of a construct is particularly intriguing . mDHFR fusions with a short spacer were arrested in the PPM [9] and were not co-blocking whereas mDHFR fusions with a long spacer were arrested in PTEX at the PVM , causing a co-block . While other scenario can also be envisaged , we favor two , non-mutually exclusive , possibilities why only proteins with a long but not a short spacer reach the PVM translocon and prevent the export of other proteins . As not the total length from the N-terminus to the blocking domain but specifically the distance between blocking domain and TM domain was important , the TM domain must play a role in this effect . The first possibility is that the TM domain is part of the recognition signal that has to emerge far enough out of the PPM extractor to become available for engagement with PTEX , leading to a transient interaction and hand-over of the substrate . This idea is supported by the fact that the type of TM region is one critical determinant for a protein to be exported [9 , 25 , 35] . The second possibility is that a short spacer keeps the TM domain in the PPM extractor close to the membrane milieu which could favor lateral release into the membrane if extraction is blocked . The BPTI results also support the hand-over model , as in proteins with long spacers , BPTI is not exposed to the oxidizing environment of the PV and cannot fold , suggesting that the polypeptide chain is passed from the PPM extractor directly through PTEX ( S1D and S1E Fig ) . Taken together the different properties of mDHFR and BPTI constructs indicate that TM proteins with a short C-terminus have a brief intermediate in the PV , similar to soluble proteins , and that TM proteins with a long C-terminus cause a transient interaction of the PPM and PVM protein conducting machines while being translocated into the host cell . This is similar to the situation in mitochondria where the translocators of the outer and inner membrane transiently interact during substrate transport [36] . However , this does not appear to be mandatory in the parasite , as proteins with a short spacer were not directly handed over . From our experiments it can be assumed that a spacer as short as 99 amino acids ( C-terminus of SBP1 ) but longer than 34 amino acids ( C-terminus of REX2 ) enables the TM of the substrate to emerge sufficiently from the PPM to engage PTEX at the PVM and to induce a direct hand over without intermittent release into the PV . Finally , other models for the observed phenotypes and for the translocations at the PPM and PVM may also be possible and it will be interesting to see how future work conforms with the here favored scenario . It is interesting that mDHFR fused export substrates only block the passage of other proteins when arrested in PTEX but not in the PPM extractor . The lack of a co-block at the PPM may be due to the mechanism of how the PPM extractor receives proteins and its resulting architecture . As opposed to PTEX which obtains proteins from the PV , the TM substrates ( arriving from a brefeldin A sensitive secretory pathway [18] ) are already integral to the PPM and the extractor needs the ability to receive substrates laterally . This might be a reversible process , leading to disassociation of the substrate back into the membrane if its export is blocked and hence frees the extractor for other substrates . It should also be noted that blocking translocation generally requires a high saturation with arrested substrate [37] . A small proportion of free extractors may therefore already be sufficient to maintain export of other proteins . The situation for the transport of exported TM proteins at the parasite periphery resembles that in certain plastids where the outer most membrane receives substrates in integral form from the Golgi after which they are translocated across the inner membranes , hence requiring extraction out of the first membrane [38 , 39] . How this extraction is achieved in these plastids is unclear [40] but the ERAD pathway in the ER membrane [41] and the Asi complex in the nuclear inner membrane [42 , 43] clearly demonstrate that a dislocation of membrane embedded proteins is possible . In absence of data on the nature of the PPM extractor , multiple configurations for the translocation set-up at the PPM and PVM can be envisaged ( reviewed in [17] ) . In one possible scenario , the dislocation of TM proteins from the PPM is aided by PTEX , which is especially plausible for the substrates that are handed over to PTEX at the PVM while portions of the molecule are still being extracted out of the PPM . This is also supported by the capacity of HSP101 to disassociate from PTEX [10] . Many exported proteins , including the major virulence factor PfEMP1 , contain TMs , and the identification of the PPM extractor will be crucial to understand how the export of these proteins is achieved . The translocation machineries in the malaria parasite PPM and PVM may have similar capacities to those in other systems such as the mitochondrial membranes that are remarkably versatile in regards to the types of substrates accepted and the destiny of delivery . Stable translocation intermediates will be essential to further unravel these mechanisms in malaria parasites .
Animal handling and immunizations by Eurogentec were done in accordance with good animal practices according to the Belgian national animal welfare regulations for Eurogentec SA , Seraing and was under approval ( CE/Sante/E/001 ) of the ethics committee of the Centre d’Economie Rurale ( CER Groupe , Marloie , Belgium ) . Eurogentec follows the European Union directive 2010/63/EU ( Welfare Legislation for laboratory animals ) . To obtain mDHFR-GFP fusions expressed under the crt-promoter , inserts were PCR amplified with Phusion Polymerase ( NEB ) from 3D7 genomic DNA or cDNA using the primers listed in S1 Table and cloned as detailed in S2 Table . Inserts were digested with XhoI/AvrII and cloned into vector pARL2-DG [21] containing the blasticidine deaminase gene as resistance marker . To obtain SBP1-mDHFR-GFP-PHmut , the last 88 bp of GFP fused to a mutated PH domain [26] was amplified from a pARL1 plasmid containing this domain and cloned into SBP1-mDHFR-GFP using an internal GFP BstBI restriction site and XmaI . To swap the order of GFP and mDHFR in REX2mDHFR-GFP [9] , REX2-GFP was amplified from REX2-GFP [35] with an additional primer-inserted NheI site after the GFP and cloned into pARL2 using XhoI and XmaI . mDHFR was inserted using NheI and XmaI to obtain REX2-GFP-mDHFR . MSRP-6 and REX-3 mCherry constructs were produced in vector pARL1-REX2-mCherry [31] containing human dihydrofolate reductase as resistance marker . Wild type BPTI [20] ( Uniprot accession P00974 ) and mutated BPTI [24] genes were commercially synthesised ( GenScript ) and inserted to replace mDHFR in vector pARL2 containing REX2-mDHFR-GFP using AvrII/KpnI , resulting in pARL2-REX2-BPTI-GFP . REX2+3C , SBP1 , SBP1∆N , SBP1∆C , MAHRP1 , and PTP1and REX3 were inserted into this vector using XhoI/AvrII to obtain the corresponding BPTI fusions . The SBP1∆N deletion insert was generated by overlap PCR using the primers listed in S1 Table . The REX2+3C insert was synthesized ( GenScript ) , fusing three additional codon changed REX2-C termini ( each encoding amino acids 61–94 ) to the construct . An additional SpeI restriction site was inserted before the 3C- termini . PTP1 was PCR amplified and cloned into pARL2 containing REX2-3C-BPTI-GFP and REX2-3C-mDHFR-GFP with Xho and SpeI to generate PTP1-3C-BPTI-GFP and PTP1-3C-mDHFR-GFP . To obtain the skip peptide ( T2A ) vectors , the T2A sequence was inserted between GFP and mCherry in pARL2 GFP-mCherry flanked by XhoI and AvrII sites . MSRP6 , KAHRP , REX-3 and STEVOR inserts were PCR amplified and cloned after T2A using AvRII/KpnI . SBP1mDHFR-GFP was amplified from the corresponding pARL2 construct and cloned before the skip peptide using XhoI/SpeI . For HA tagging of the exp2 locus by single cross-over recombination the last 1000 bp of the exp2 gene were PCR amplified from 3D7 genomic DNA , adding a sequence encoding a 3xHA tag and an additional KpnI restriction site before the 3xHA tag with the reverse primer and cloned into vector pSLI-PfEHD2xFKBP ( Genbank accession KU998257 ) using NotI/SalI to replace PfEHD2xFKBP , leading to an integration plasmid carrying N-terminally truncated exp2 without promoter . For HA tagging of the hsp101 locus , the last 1000 pb of this gene were cloned NotI/KpnI to replace the exp2 fragment inserted into pSLI-PfEHD2xFKBP . To obtain a plasmid to tag the endogenous locus of sbp1 with mdhfr and gfp base pairs 64–1181 of sbp1 were PCR amplified and cloned NotI/AvrII into p-REX2mDHFR-int ( Genbank accession KU998258 ) , leading to an integration plasmid carrying a promoterless N-terminally truncated sbp1 gene fused with the sequence coding for mDHFR-GFP . For GST fusion expression constructs , inserts were amplified with the primers listed in S1 Table and cloned with BamHI and XhoI into pGEX-6-P2 ( GE healthcare ) . Inserts of all plasmids were sequenced to confirm absence of undesired mutations . P . falciparum parasites ( 3D7 ) were cultured in human 0+ erythrocytes ( transfusion blood , Universitätsklinikum Hamburg-Eppendorf ) with RPMI 1640 medium containing 0 . 5% AlbuMAX ( Invitrogen ) according to standard procedures [44] . Transfection of synchronized ring stages was performed with 100 μg of purified plasmid DNA ( Qiagen ) as described [45] . Positive selection was done with 4 nM WR99210 ( Jacobus Pharmaceuticals ) or 2 μg/ml Blasticidin S ( Invitrogen ) . Double transfectant cell lines were produced by transfection of mDHFR-GFP constructs into WR resistant cell lines expressing pARL1-mCherry constructs and selected using Blasticidin S . Once a week these transfected cultures were treated with 4 nM WR to avoid loss of plasmid expressing the mCherry construct . Parasite cell lines expressing mDHFR fusion proteins were synchronized with 5% sorbitol [46] to obtain ring stages before they expressed the transgene . Thereafter the parasites were grown for 24 hours in presence or absence ( control ) of 4 nM WR during which transgene expression occurred . The cells were either directly imaged , processed for immune fluorescence assays ( IFA ) , lysed for parasite extracts , or processed for protease protection assays . Percoll purified [47] late stage SBP1-mDHFR-GFPendo parasites or sorbitol synchronized ring stages of the double transfected cell lines REX-2GFP-mDHFR/REX2-mCherry and REX-2-mDHFR-GFP/REX2-mCherry were washed with RPMI medium and brought back into culture with or without 4 nM WR . Giemsa-stained thin blood smears were collected every day and parasitemia and parasite stages were recorded . Data are representative of three independent experiments . IFAs were performed as described [48] . Briefly , parasites were washed twice with PBS and dried as a thin film on 10-well slides . Cells were fixed in 100% acetone for 30 minutes at room temperature . Antibodies were diluted in PBS/3%BSA and incubated for 1 hour , followed by 5 washes in PBS . Secondary antibodies were applied for 1 hour in PBS/3%BSA containing 1 μg/ml DAPI followed by 5 washes with PBS . Mounting medium ( Dako ) was added and the slide sealed with a coverslip for imaging . Dilutions of primary antibodies were: mouse anti-GFP 1:500 ( Roche ) , rabbit anti-GFP ( Thermo ) 1:500 , rabbit anti-KAHRP 1:500 ( a kind gift of Prof . Brian Cooke ) , mouse anti-MSRP6 1:250 [7] , rabbit anti-myc 1:500 ( Cell Signaling Technologies ) , mouse anti-REX2 1:250 [45] , rabbit anti MAHRP2 1:250 [[49] , a kind gift of Prof . Hans-Peter Beck] , rabbit anti-REX1 1:2000 ( newly raised ) and rat anti HA 1:500 ( Roche ) . Secondary antibodies used were donkey anti-rabbit conjugated with Alexa Fluor-488 , -594 or goat anti-rabbit conjugated with Alexa-647 and goat anti-mouse conjugated with Alexa Fluor-488 , -594 or donkey anti mouse conjugated with Alexa-647 ( Invitrogen ) diluted 1:2000 . IFAs were directly imaged . For live cell imaging , the nuclei of GFP and mCherry-expressing parasites were stained with 1 μg/ml DAPI ( Roche ) for 5 min at 37°C and infected erythrocytes were imaged in medium as described [50] . Microscopy was done with a Zeiss Axio Scope M1 microscope equipped with a 100x/1 , 4 numerical aperture oil immersion lens . Images were collected with a Hamamatsu Orca C4742-95 camera and Zeiss AxioVision software . Images were processed in Corel PHOTO-PAINT X6 . Fragments of SERA-5 [aa 68–184 of PF3D7_0207600 [51]] , Aldolase [aa 9–96 of PF3D7_1444800 , [52]] , REX1 [aa 332–596 of PF3D7_0935900 , [53]] , REX3 [aa 48–326 of PF3D7_0936300 , [45]]; SBP1N [aa 13–208 of PF3D7_0501300 , [54]] were expressed as GST fusion proteins , purified with GST-sepharose ( GenScript ) and antisera were commercially raised by Eurogentec . Single bands of the expected sizes were observed with the antisera in parasite extracts on Western blots ( S8 Fig ) . For total parasites extracts , parasites were released from RBCs using 0 . 03% saponin ( Sigma ) in PBS and washed twice with PBS . Proteins were then extracted with 4% SDS/0 . 5% Triton X-114/0 . 5 x PBS in presence of protease inhibitors ( Roche ) . After centrifugation at 16’000g for 5 min , reducing SDS sample buffer was added to the supernatant which was then separated by SDS-PAGE . Protease protection assays were performed as described [9] . Percoll purified infected RBCs from 10 ml culture ( 5–10% parasitemia ) were washed with RPMI medium and treated with 1 HU tetanolysin ( Sigma ) in 100 μl of Dulbecco PBS ( DPBS ) ( Pan Biotech ) at 37°C for 10 min . The permeabilised parasites were washed with DPBS , equally divided into three tubes that were incubated for 30 min on ice with either 100 μl DPBS alone ( control ) , 100 μl DPBS containing 8 U/ml proteinase K ( NEB ) , or 100 μl DPBS containing 0 . 03% saponin and 8 U/ml proteinase K , respectively . Reactions were quenched and proteins precipitated by adding trichloroacetic acid to 10% final concentration . The sample containing the precipitated proteins was centrifuged at 16'000 g for 20 minutes , washed twice with 100% acetone and resuspended in 50 μl 1x TE buffer and frozen at—20°C . Samples were thawed , SDS sample buffer was added and equal amounts were subjected to Western analysis . Protein samples were resolved by SDS-PAGE and transferred to nitrocellulose membranes ( Protran ) in a tankblot device ( Bio-Rad ) using transfer buffer ( 0 . 192 M Glycine , 0 . 1% SDS , 25 mM Tris ) with 20% methanol . Blocking of membranes and dilutions of antibodies were done in PBS containing 5% skim milk . Washing steps were done with PBS . Primary antibodies were applied in the following dilutions: mouse anti-GFP ( Roche ) 1:1000; rabbit anti-GFP ( Thermo ) 1:2000; rat anti-mCherry , 1:1000 ( Chromotek ) ; rabbit anti-SERA5 , 1:2000 ( newly raised ) ; rabbit anti-REX3 , 1:2000 ( newly raised ) ; mouse anti-SBP1N , 1:2500 ( newly raised ) ; rabbit anti-aldolase , 1:4000 ( newly raised ) ; mouse anti-HSP101 , 1:1000 [7]; rabbit anti-DHFR ( Abcam ) , 1:1000; rat anti-HA ( Roche ) 1:4000 . Horseradish peroxidase-conjugated secondary antibodies used were goat anti-rat ( Dianova ) and goat anti-mouse ( Dianova ) and diluted 1:3000 as well as donkey anti-rabbit ( Dianova ) 1:2500 and applied after three washes . Immunoreactions were detected by enhanced chemiluminiscence ( Bio Rad/ Thermo ) and detected on CEA RP NEW x-ray films ( Agfa ) . For quantification of Western blot signals , band intensities were measured with a Chemi Doc XRS imaging system ( Bio-Rad ) and densitometry analyses were done with Image Lab Software 5 . 2 ( Bio-Rad ) . Data are representative of three independent experiments . The EXP2-3xHAendo cell line expressing SBP1-mDHFR-GFP , REX2-mDHFR-GFP or REX2-GFP-mDHFR , respectively , was sorbitol synchronized and ring stage parasites ( ~10% parasitemia ) cultured with and without WR ( 4 nM ) for 24 hours . The resulting trophozoites were harvested and washed twice with DPBS . For some experiments , cultures were crosslinked with 0 . 5 mM dithiobis ( succinimidylpropionate ) ( DSP , 20 mM stock in DMSO ) ( Pierce ) in DPBS for 30 minutes at room temperature and the reaction was quenched with PBS containing 25 mM Tris-HCl . Infected erythrocytes were purified in a Percoll gradient , washed with DPBS and lysed with RIPA buffer ( 10 mM Tris HCl pH 7 . 5 , 150 mM NaCl , 0 . 1% SDS , 1% Triton ) containing protease inhibitor cocktail ( Roche ) and 1 mM PMSF . After two freeze-thaw cycles , lysates were cleared by centrifugation at 16'000 g for 10 minutes . Supernatants were incubated with 25 μl of mouse monoclonal anti-HA beads ( Pierce ) or anti-GFP beads ( Chromotek ) for 3 hours at 4°C . Samples of input and post binding extracts were saved for SDS-PAGE . Beads were recovered by centrifugation and washed five times with RIPA buffer . Proteins were eluted in 50 μl 4 x SDS sample buffer at 85°C for 5 minutes . Equal volumes of input , post binding extract and bound fractions were subjected to Western analysis . Synchronised trophozoite cultures of EXP2-3XHAendo and 3D7 parasites ( 100 ml each , 5% parasitemia ) , were harvested , washed twice with DPBS and dithiobis ( succinimidylpropionate ) ( DSP , 20mM stock in DMSO ) ( Pierce ) was added to 2 mM in DPBS for 30 minutes at room temperature . The reaction was quenched with 25 mM Tris-HCl pH 7 . 5 in DPBS for 10 minutes . Infected RBCs were purified in a Percoll gradient , washed with DPBS and lysed with RIPA buffer ( 10 mM Tris HCl pH 7 . 5 , 150 mM NaCl , 0 . 1% SDS , 1% Triton ) containing 2X protease cocktail inhibitors ( Roche ) and 1 mM PMSF . After two freeze-thaw cycles at -80°C , lysates were cleared by centrifugation at 16'000 g for 10 minutes . Supernatants were diluted 1:2 with RIPA buffer without detergents ( 10 mM Tris HCl pH 7 . 5 , 150 mM NaCl ) and equal volumes were incubated with 50 μl of anti-HA beads ( Pierce ) for 3 hours at 4°C with end-over-end rotation . Beads were recovered by centrifugation for 10 seconds at 11'000 rpm and washed five times with RIPA buffer . Cross linked interacting partners were released by ReCLIP [55] by incubating beads for 30 minutes at 37°C with RIPA buffer supplemented with 100 mM dithiotreitol followed by centrifugation to obtain the supernatant ( ReCLIPed eluate , designated as Eluate 1 ) . The beads were then incubated shortly with NaOH 50 mM , centrifuged and supernatant was saved ( Eluate 2 ) . Both eluates were then precipitated with trichloroacetic acid ( TCA ) 20% and analysed by mass spectrometry . TCA Protein pellets were solubilized in lysis buffer ( 6 M urea , 2 M thiourea , 10 mM HEPES pH 8 . 0 ) by sonication for 10 min at 4°C . Proteins were reduced with 10 mM DTT for 10 min at room temperature and alkylated with 55 mM iodoacetamide for 20 min in the dark . Proteins were digested with 0 . 5 μg LysC ( Wako ) for 3h at room temperature . Samples were then diluted 1:4 with water and subsequently digested with mass-spectrometry grade trypsin ( Promega ) overnight at 32°C . Tryptic peptides were purified by SPE on a SepPAC-tC18 ( Waters ) according to the manufacturer's instructions , lyophilized and re-dissolved in 0 . 1% formic acid and spiked with 20 fmol/μL of yeast enolase 1 MassPREPTM protein digestion standard ( Waters ) prior to LC-MS analysis . Tryptic peptides were analysed using a nanoscale UPLC system ( nanoAcquityUPLC ) ( Waters ) coupled online to a Synapt G2-S HDMS mass spectrometer ( Waters ) . Peptides were separated on a HSS-T3 1 . 7 μm , 75 μm x 250 mm reversed-phase column ( Waters ) using direct injection mode as described before [56] . Analysis was performed in positive mode ESI-MS using an ion-mobility enhanced data-dependent acquisition workflow ( HD-DDA ) described in detail previously [57] . The data were post-acquisition lock mass corrected using [Glu1]-Fibrinopeptide B . LC-MS data were processed using PEAKS v 7 . 5 ( Bioinformatics Solutions Inc ) searching against a combined database consisting of UniprotKB/Swissprot human database ( UniProtKB release 2015_02 ) and UniProt Plasmodium 3D7 Reference Proteome , supplemented with common contaminant proteins , which was concatenated to a reversed decoy database , using the following search criteria for peptide identification: i ) trypsin as digestion enzyme ii ) up to three missed cleavages allowed iii ) fixed carbamidomethylcysteine and variable methionine oxidation as modifications . Precursor and fragment ion mass tolerances were set to 15 ppm for precursors and 0 . 03 Da for fragment ions . The initial false discovery rate ( FDR ) for peptide identification was set to 1% in PEAKS based on a reversed decoy database search .
|
P . falciparum parasites , the deadliest agent of human malaria , develop within erythrocytes where they are surrounded by a parasitophorous vacuolar membrane ( PVM ) . To ensure intracellular survival , the parasite exports a large repertoire of proteins into the host cell . Exported proteins require unfolding for trafficking across the membrane boundaries separating the parasite from the erythrocyte , typical for transport by protein translocating membrane channels . Here , we dissected the sequence of translocation events at the parasite boundary using substrates that can be conditionally arrested at translocation steps . We for the first time obtained exported proteins arrested in the process of being translocated across the PVM . This jammed the translocons for all other types of exported proteins and inhibited parasite growth . The constructs stuck in translocation were in a complex with EXP2 , a component of a complex known to be essential for protein export that is termed PTEX . Our work links the need for unfolding and the function of this complex in export , giving experimental evidence that PTEX indeed is a translocon . Conditionally unfoldable domains have been instrumental in unravelling transport processes across membranes and here resolve the transport steps the different kinds of exported proteins require to reach the P . falciparum-infected host cell .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"parasite",
"groups",
"medicine",
"and",
"health",
"sciences",
"plasmodium",
"viral",
"transmission",
"and",
"infection",
"enzymes",
"enzymology",
"microbiology",
"parasitic",
"diseases",
"parasitic",
"protozoans",
"parasitology",
"membrane",
"proteins",
"apicomplexa",
"protozoans",
"cellular",
"structures",
"and",
"organelles",
"extraction",
"techniques",
"research",
"and",
"analysis",
"methods",
"protein",
"extraction",
"malarial",
"parasites",
"proteins",
"cell",
"membranes",
"biochemistry",
"host",
"cells",
"cell",
"biology",
"virology",
"biology",
"and",
"life",
"sciences",
"proteases",
"organisms"
] |
2016
|
Stable Translocation Intermediates Jam Global Protein Export in Plasmodium falciparum Parasites and Link the PTEX Component EXP2 with Translocation Activity
|
Zika virus ( ZIKV ) is a mosquito-borne flavivirus first isolated in Uganda in 1947 . Although entomological and virologic surveillance have reported ZIKV enzootic activity in diverse countries of Africa and Asia , few human cases were reported until 2007 , when a Zika fever epidemic took place in Micronesia . In the context of West Africa , the WHO Collaborating Centre for Arboviruses and Hemorrhagic Fever at Institut Pasteur of Dakar ( http://www . pasteur . fr/recherche/banques/CRORA/ ) reports the periodic circulation of ZIKV since 1968 . Despite several reports on ZIKV , the genetic relationships among viral strains from West Africa remain poorly understood . To evaluate the viral spread and its molecular epidemiology , we investigated 37 ZIKV isolates collected from 1968 to 2002 in six localities in Senegal and Côte d'Ivoire . In addition , we included strains from six other countries . Our results suggested that these two countries in West Africa experienced at least two independent introductions of ZIKV during the 20th century , and that apparently these viral lineages were not restricted by mosquito vector species . Moreover , we present evidence that ZIKV has possibly undergone recombination in nature and that a loss of the N154 glycosylation site in the envelope protein was a possible adaptive response to the Aedes dalzieli vector .
Zika virus ( ZIKV ) is a mosquito-borne flavivirus , a member of the Spondweni serocomplex , whose natural transmission cycle involves mainly vectors from the Aedes genus ( A . furcifer , A . taylori , A . luteocephalus and A . africanus ) and monkeys [1] , while humans are occasional hosts . Clinical pictures range from asymptomatic cases to an influenza-like syndrome associated to fever , headache , malaise and cutaneous rash [2] , [3] . Likewise , direct contact is also considered a potential route of transmission among humans , probably during sexual intercourse [4] . The first isolation of ZIKV was in 1947 from the blood of a sentinel Rhesus monkey No . 766 , stationed in the Zika forest , near the Lake Victoria in Uganda , and in 1948 ZIKV was also isolated in the same forest from a pool of A . africanus mosquitoes [5] . Thereafter , serological and entomological data indicated ZIKV infections in the African continent in Nigeria in 1971 and 1975 [6] , Sierra Leone in 1972 [7] , Gabon in 1975 [8] , Uganda in 1969 and 1970 [9] , Central African Republic in 1979 [10] , Senegal from 1988 to 1991 [11] and Côte d'Ivoire in 1999 [12] . Recently , ZIKV was detected in Senegal in 2011 and 2012 ( unpublished data ) . In addition , ZIKV infections in Asia were reported in Pakistan [13] , Malaysia [14] , Indonesia in 1977 and 1978 [15] , Micronesia in 2007 [16] , [17] and Cambodia in 2010 [18] . Although ZIKV was repeatedly isolated , only 14 human cases were reported before April 2007 , when a Zika fever ( ZF ) epidemic occurred in Yap island in Micronesia , where 49 confirmed cases and 73% of the residents older than 3 years provided serologic evidence for recent ZIKV infection [16] . This outbreak showcased the potential of ZF as an emerging disease , which could be misdiagnosed as dengue fever , as happened during the beginning of the Micronesian outbreak [16] , [17] . The ZIKV genome consists of a single-stranded positive sense RNA molecule with 10794 kb of length with 2 flanking non-coding regions ( 5′ and 3′ NCR ) and a single long open reading frame encoding a polyprotein: 5′-C-prM-E-NS1-NS2A-NS2B-NS3-NS4A-NS4B-NS5-3′ , that is cleaved into capsid ( C ) , precursor of membrane ( prM ) , envelope ( E ) and seven non-structural proteins ( NS ) [19] , [20] . The E protein ( ≈53 kDa ) is the major virion surface protein . E is involved in various aspects of the viral cycle , mediating binding and membrane fusion [21] . The NS5 protein ( ≈103 kDa ) is the largest viral protein whose C-terminal portion has RNA-dependent RNA polymerase ( RdRP ) activity and the N-terminus is involved in RNA capping by virtue of its processing due to methyl transferase activity [21] . The 3′NCR of the ZIKV genome contains about 428 nucleotides , including 27 folding patterns [20] that may be involved in the recognition by cellular or viral factors , translation , genome stabilization , RNA packaging , or cyclization [21] . Although diverse studies have contributed greatly to our understanding of the evolutionary biology of flaviviruses in general [22]–[25] , few studies have addressed ZIKV evolutionary biology [17] , [26] . Those studies report three main ZIKV lineages , one from Asia and two from Africa . Aiming to fill this gap and gain better insights ZIKV molecular evolution in the 20th century , we investigated 43 ZIKV strains , sampled from 1947 to 2007 in Africa and Asia , to describe phylogenetic relationships , selective influences , recombination events , phylodynamics , phylogeography , host correlations with viral lineages and glycosylation patterns .
Samples used in this study are part of the Institute Pasteur in Dakar collection ( WHO Collaborating Centre for Arboviruses and/or Hemorrhagic Fever Reference and Research ) . Monkey and human strains ( AnD 30332 and HD 78788 ) were obtained respectively in 1979 and 1991 in Senegal during routine surveillance . None of the data was directly derived from human or animal samples but rather from cell culture supernatant . Therefore all the samples were anonymous and only reference numbers were used during the analysis that originated this study . ZIKV strains were provided by CRORA at the Institute Pasteur of Dakar . The strains were obtained from mosquitoes , humans and other mammals isolated in Burkina Faso , Central African Republic , Côte d'Ivoire and Senegal in West Africa ( Table S1 ) . Viral stocks were prepared by inoculating viral strains into Aedes pseudoscutellaris clone 61 monolayer in Leibovitz 15 ( L-15 ) growth medium ( GibcoBRL , Grand Island , NY , USA ) supplemented with 5% fetal bovine serum ( FBS ) ( GibcoBRL , Grand Island , NY , USA ) , 10% Tryptose Phosphate and antibiotics ( Sigma , Gmbh , Germany ) . Viral infection was confirmed after seven days of propagation by an indirect immunofluorescence assay ( IFA ) using specific hyper-immune mouse ascitic fluid , as described previously [27] . Cultures supernatants were collected for virus RNA isolation . RNA was extracted from ZIKV stocks using the QIAamp RNA Viral Kit ( Qiagen , Hilden , Germany ) according to the manufacturer's recommendations . RNA was eluted in 50 µl of AVE buffer and stored at −80°C until use . For cDNA synthesis , 10 µl of viral RNA was mixed with 1 µl of each of a reverse primer ( 2 pmol ) , 1 µl of deoxynucleotide triphospahte ( dNTP ) ( 10 mM each dNTP and the mixture was heated at 65°C for 5 min . Reverse transcription was performed in 20 µl mixture containing mixed of 2 . 5 U RNasin ( Promega , Madison , USA ) 5 U of Superscript III reverse transcriptase ( Invitrogen , Carlsbad , USA ) and incubated at 55°C for 50 min , followed by 70°C for 15 min . PCR products were generated independently using the primers Unifor/Unirev , FD3/FU1 and VD8/EMFI to amplify partial E , NS5 and NS5/3′NC region respectively [28] . Five microliters of cDNA were mixed with 10× buffer , 5 µl of each primer , 5 µl of dNTPs 10 mM , 3 µl of MgCl2 , and 0 . 5 µl of Taq polymerase ( Promega , Madison , USA ) . PCR products of the expected size were purified from agarose gels with the QiaQuick Gel Extraction Kit ( Qiagen , Hilden , Germany ) as specified by the manufacturer . Both strands of each PCR product were sequenced directly with the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit V3 . 1 on an Applied Biosystems 3100 DNA Analyzer ( Applied Bisoystem , Foster City , CA , USA ) at the Laboratory of Molecular Evolution and Bioinformatics , Biomedical Sciences Institute , University of Sao Paulo , Brazil . We deposited thirty two 753 bp-long sequences from the E gene ( Accession numbers: KF383015-KF383046 ) , thirty one of NS5 ( 708 bp ) ( Accession numbers: KF38304-KF383114 ) , thirty seven of 3′NCR ( 537 bp ) ( Accession numbers: KF383047-KF383083 ) and six genomes ( 10274 bp ) ( Accession numbers: KF383115–KF383120 ) in GenBank ( www . ncbi . nlm . nih . gov/genbank/ ) from thirty eight viral strains ( Table S1 ) . Additional sequences representing strains from Kedougou in Senegal , Nigeria , Malaysia , the Ugandan prototype MR766 , the strain related to Micronesian outbreak in 2007 and the Spondweni virus were obtained from GenBank , with the following accession numbers , respectively: HQ234501 , HQ234500 , HQ234499 , NC_012532 , EU545988 and DQ859064 . 1 ( Table S1 ) . Prior to the analyses , all sequences were aligned with MUSCLE v3 . 7 [29] and manually edited with SeaView v4 . 3 . 3 [30] . To prevent potential biases during phylogenetic inference due to recombination , we first analyzed the sequences of available ZIKV genomes with RDP v4 . 4 . 8 program [31] that incorporates RDP [32] , GENECONV [33] , Chimaera [34] , MaxChi [35] , Bootscan [36] , SiScan [37] and 3Seq [38] methods to uncover evidence for recombination events . Only events with p-values≤0 . 01 that were confirmed by four or more methods were considered , using the Bonferroni correction to prevent false positive results [39] , as implemented in the RDP program [31] . In addition , the occurrence of recombination in genomes was also investigated with the Rec-HMM program that estimates breakpoints based on the Phylo-HMM approach , which models tree topology changes over the columns of a multiple alignment [40] . Moreover , potential intra-gene recombination was also inspected with RDP using individual gene datasets , and the incompatibility among phylogenies inferred from genes ( NS5 and E ) was also investigated with GiRaF v1 . 01 [41] that searches incompatible clades among posterior set of trees ( PST ) obtained from different genomic regions with threshold of 70% for incompatible clades . The PST was obtained during Monte Carlo Markov chain ( MCMC ) stationarity using four chains , one ‘cold’ and three ‘heated’ , after 20 million of generations , sampling every 5000 generations using MrBayes v3 . 2 . 1 [42] . The phylogenetic signal content of the sequence datasets to phylogenetic reconstruction was investigated by Likelihood mapping method [43] , implemented in TREE-PUZZLE v5 . 2 [44] . The concordance among gene ( E and NS5 ) datasets without recombinant sequences was further assessed using a permutation test with significance level ( α ) of 0 . 05 after 10000 permutations , implemented in the Congruence among Distance Matrices ( CADM ) package [45] . Phylogenetic trees were generated by Maximum Likelihood ( ML ) criterion using GARLI v2 . 0 [46] that uses a stochastic algorithm to estimate simultaneously the best topology , branch lengths and substitution model parameters that maximize the log Likelihood ( lnL ) . The confidence of ML trees was assessed by the convergence of lnL scores from ten independent replicates . We used a substitution model based on general time reversible ( GTR ) model with gamma-distributed rate variation ( Γ ) and a proportion of invariant sites ( I ) . Support for the topology was obtained after 1000 non-parametric bootstrap replicates with GARLI . Then , we summarized the bootstrap trees into one consensus tree to access bootstrap values , using Dendropy v3 . 10 . 1 [47] . To infer the selection pressures acting on each gene of ZIKV , we estimated the difference between the non-synonymous ( dN ) and synonymous ( dS ) rates per codon site using the single likelihood ancestor counting ( SLAC ) algorithm available in HyPhy v0 . 99 [48] , assuming a significance level of 1% ( α = 0 . 01 ) . In the HyPhy output , values of ω are expressed as ω = dN - dS . Therefore , ω smaller than zero ( ω<0 ) indicate purifying , negative selection . Potential glycosylation sites that may have adaptive value were previously described in ZIKV proteins [17] , [20] , [26] . Thus , we investigated partial E sequences to detect potential glycosylation sites using NetCGlyc v1 . 0 [49] , NetOGlyc v3 . 1 [50] , YinOYang v1 . 2 [51] and NetNGlyc v1 . 0 [51] , [52] methods that employ algorithms based in neural networks to predict , respectively , C-mannosylated , mucin-type O-linked , N-acetylglucosamine ( GlcNAc ) and N-linked glycosylation sites . To infer the structural position of the predicted glycosylation sites , we modeled the tridimensional structures of E regions of viral polyprotein of the Micronesian strain ( GenBank accession number ACD75819 ) . We used the homologous sequences from Japanese Encephalitis virus ( PDB code 3p54 ) , West Nile virus ( PDB code 2i69 ) and Dengue virus type 3 ( PDB code 1uzg ) . The amino acids sequences were aligned using MUSCLE v3 . 7 [29] , a total of 1000 independent models were generated and optimized using Modeller v9 . 10 [53] , and the best models were validated with PROCHECK v3 . 5 . 4 [54] . Maximum Clade Credibility ( MCC ) trees were inferred using a Markov Chain Monte Carlo ( MCMC ) Bayesian approach implemented on the program BEAST v1 . 6 . 2 [55] under GTR + Γ + I and a relaxed ( uncorrelated lognormal ) molecular clock [56] . MCMC convergence was obtained for four independent runs with 50 million generations , which were sufficient to obtain a proper sample for the posterior at MCMC stationarity , assessed by effective sample sizes ( ESS ) above 200 inspected using Tracer v1 . 5 ( http://tree . bio . ed . ac . uk/software/tracer/ ) . Furthermore , using the concatenated sequences of E and NS5 genes , we employed a discrete model attributing state characters representing isolation locality , animal source , recombination and N- linked glycosylation on E protein of each of the strains with the Bayesian Stochastic Search Variable ( BSSVS ) algorithm [57] , implemented in BEAST . This method estimates the most probable state at each node in the MCC trees , allowing us to reconstruct plausible ancestral states on these nodes . Moreover , we represented the viral migration in Google Earth ( http://www . google . com/earth/ ) , using the SPREAD v1 . 0 . 3 program [58] . We evaluated the correlation among viral states and inferred phylogenies from PST by the parsimony score ( PS ) , association index ( AI ) and monophyletic clade size ( MC ) , with BaTS v1 . 0 [59] after 10000 null replications . In addition , we investigated the occurrence of correlated evolutionary change among ZIKV phenotypes ( glycosylation pattern and vector host ) along PST , employing a ML approach to test the fit of the two evolutionary models , one where the two traits evolve independently on the phylogenetic tree ( independent model ) , and one where they evolve in a correlated way ( dependent model ) [60] , using BayesTraits program ( http://www . evolution . rdg . ac . uk/ ) . To evaluate model suitability to ZIKV data , we estimated the marginal likelihoods for both models after 1000 bootstrap replications and compared Bayes factors ( BF ) between models [61] , using Tracer v1 . 5 .
The primary analyses with RDP suggested 13 recombination events in ZIKV complete genomes ( Table S2 ) , Rec-HMM also detected genomic breakpoints with confidence in the following alignment positions: 1044 to 1095 , 5181 to 5238 , 9007 to 9132 and 9580 to 9631 ( Figure S1 ) . Since the results obtained by both methods revealed breakpoints in the E and NS5 genomic regions , we investigated these evidences with RDP on partial gene sequences . We found a single event in E sequences with estimated breakpoints near to the 414th and 1065th site of E gene reaching nine viral strains: ArA986 , HD78788 , ArA27101 , ArA27290 , ArA27096 , ArA27443 , ArA27407 , ArA27106 and ArA982 . These results were found by Bootscan , Maxchi , Chimaera , SiSscan and 3Seq methods and supported by significant p-values of 1 . 31E-5 , 2 . 85E-3 , 1 . 59E-3 , 1 . 79E-29 and 6 . 85E-19 , respectively . Likewise , only one recombination event was detected in NS5 sequences with estimated breakpoints near sites 1581 and 2152 of the NS5 gene from strains ArD158084 , ArB1362 and ArD157995 . These findings were confirmed by Bootscan , Maxchi , Chimaera , SiSscan and 3Seq methods and supported by significant p-values of 9 . 93E-9 , 3 . 32E-7 , 3 . 32E-7 , 5 . 27E-28 and 7 . 65E-24 , respectively . These potential recombinant sequences were excluded from further analyses to avoid inferential biases [62] , [63] . To perform the phylogenetic analysis we concatenated E and NS5 sequences and replaced inferred recombinant fragments with missing data . This is in line with the use of Maximum Likelihood approaches , which is fairly robust to the introduction of gaps [64] , [65] . In addition , we found incompatibilities between E and NS5 phylogenies using GiRaF . The three discordant strains ( ArD128000 , ArA1465 and ArD142623 ) were excluded , and we used 40 ( 31 from E and 36 from NS5 ) concatenated sequences for phylogenetic analysis . Moreover , we also found that the two remaining datasets for E and NS5 have no conflicting phylogenetic signal , as estimated by a CADM test ( p-value = 9 . 99E-5 and α = 0 . 05 ) . Given the limited sampling that we investigated , these results indicate that ZIKV may be experiencing recombination in the field , which is uncommon among flaviviruses [66] . These findings remain to be properly evaluated and assessed related to their effects on viral spread , zoonotic maintenance and epidemiologic potential . The possibility that our findings could be a consequence of cross contamination among isolates seems highly improbable given the extreme precautions that were taken . RNA extraction and reverse transcription were done separately for each isolate under BSL-II cabinets , sequenced several times leading to identical sequences , even when processed in different laboratories in Sao Paulo , Brazil , and Dakar , Senegal . We first investigated the phylogenetic signal content in our data by reconstructing 50000 quartets for each gene segment using the likelihood mapping method ( see methods section ) . Our results indicated that NS5 and E datasets had high phylogenetic signal content given their lower percentage of unresolved quartets ( 3 . 2% and 3 . 4% , respectively ) , while 3′NCR showed less signal ( 16 . 4% of unresolved quartets ) and was not considered . The ML trees for E ( data not shown ) , NS5 ( data not shown ) and the two concatenated genes ( Figure 1 ) reinforced that ZIKV strains could be classified in three major clusters [17] . Accordingly , the African strains were arranged into two groups: the MR766 prototype strain cluster ( yellow sector on Figure 1 ) and the Nigerian cluster ( green sector on Figure 1 ) ; and the Micronesian and Malaysian strains clustered together forming the Asian clade ( Figure 1 ) , in agreement with [26] . For West Africa , the strains from Côte d'Ivoire and Senegal were found in both African clusters , suggesting that at least two distinct lineages of ZIKV circulated in these countries . Interestingly , we found that the position of the Senegalese cluster , comprising viruses isolated from 1998 to 2001 associated with A . dalzieli , branching as a sister group of HD78788 isolated in Senegal in 1991 , was not simply explained by recombination ( with both Giraf and RDP ) or poor rooting of the tree , since it did not depend on the inclusion ( Figure 1 ) or exclusion ( Figure S2 ) of the Spondweni , which is a bonafide outgroup . It was observed 65% of the time during a highly stringent maximum likelihood ( ML ) analysis with GARLI , not taking into account dates of isolation , but crucially it had a posterior probability of one during Bayesian Inference ( BI ) that do take into account dates of isolation . Although we cannot rule out systematic topological errors , BI was certainly better informed than ML , since RNA viruses evolve fast , making their times of isolation important parameters for phylogenetic inference . Moreover , since we did not find compositional or codon usage biases in those sequences and in agreement with the consistent BI results , we could not rule out that the long branch length observed was not due to a detected increase of almost 10 fold increase in the rate of change along that lineage , which was not caused by detectable positive selection , as evaluated using HyPhy . Selection analyses of E and NS5 genes uncovered several sites under strong negative selection indicated by ω<0 . This suggests frequent purging of deleterious polymorphisms in functionally important genes . Likewise , the lack of positively selected sites , indicated by ω>0 , is typical of highly adapted phenotypes and shows no detectable directional change on the available data . Our findings were expected , as the infection and transmission modes of ZIKV allow the accumulation of synonymous mutations and negatively selected sites [67] . The alternation between arthropod vector and mammal hosts may impose several barriers to non-synonymous mutations in important genes [68] . The μ and the highest posterior densities ( HPD with 95% of confidence interval ) estimated with Beast for E , NS5 and 3′NCR genomic regions were , respectively , 2 . 135E-3 ( 2 . 04E-3 to 2 . 33221E-3 ) , 7 . 1789E-4 ( 6 . 9466E-4 to 7 . 417E-4 ) and 1 . 1285E-3 ( 2 . 708E-4 to 2 . 504E-3 ) substitutions per site per year , which are similar to μ estimated other flaviviruses [69] . As evolutionary rates are the result of spontaneous mutations followed by selection , differences per gene are expected and in accordance with their biological role , given that the NS5 is a polymerase and the E is a surface protein . In addition , the root date estimates and 95% HPDs of the phylogenetic trees for E , NS5 and 3′NCR genomic regions were , respectively , 1900 ( 1851 to 1937 ) , 1927 ( 1887 to 1940 ) and 1923 ( 1874 to 1959 ) . These dates suggest a recent origin for the ZIKV strains ( included in this study ) near to the beginning of the 20th century . Based on our samples we inferred the most likely geographical pathway connecting ZIKV lineages . These results indicated that ZIKV emerged in Uganda around 1920 , most probably between 1892 and 1943 . This inference is in line with the first known ZIKV isolation in Uganda in 1947 [5] . We found two independent ZIKV introductions into West Africa from the Eastern portion of the continent ( Figures 2 and S2A , and kml file in Dataset S1 ) . The first viral introduction into Côte d'Ivoire ( CI ) and Senegal ( SN ) was related to the MR766 cluster ( yellow lines in Figure 2 ) , which possibly moved from Uganda around 1940 into Dezidougou ( CI ) . From there , this lineage probably spread to Kedougou in Senegal ( SN ) around 1985 and to Sokala-Sobara ( CI ) around 1995 . The second introduction was related to a Nigerian cluster ( green lines in Figure 2 ) , when ZIKV probably moved from Uganda to the Central African Republic and Nigeria around 1935 . From Nigeria , the virus probably spread to Saboya ( SN ) around 1950 and from there to Dezidougou ( CI ) and Bandia ( SN ) around 1960 . From Bandia , ZIKV was introduced into Kedougou ( SN ) around 1965 and from there to Burkina Faso around 1980 and to Dakar ( SN ) around 1985 . Moreover , an additional ZIKV lineage from Uganda probably spread to Malaysia around 1945 and from there , the virus reached Micronesia around 1960 , forming the Asian cluster [26] . The correlation between viral location ( coded as character states ) and phylogenies was strongly supported by significant AI and PS values , p-values≤1 . 00 E-4 ( Dataset S2 ) . Thus , assuming an origin of ZIKV in Uganda , our findings revealed at least two independent exits from East Africa in agreement with the two previously proposed African clades [17] and also pointed to a viral migratory flow from Eastern Africa to Asia . Although our sampling was the most comprehensive to this date , our conclusions about ZIKV movement are informed conjectures at best on the most plausible hypotheses on ZIKV spreading patterns , which are limited by the inherent biases of this type of analyses . The association of the animal sources with viral lineages ( Figure S2B ) suggested that ZIKV dispersed widely among distinct animal species without a clear pattern of preference , maybe associated to the enzootic cycle of ZIKV in Africa , whose natural cycle allows a broad range of hosts [70] . Nevertheless , we found significant MC ( p-value≈1 . 00 E-4 , Dataset S2 ) for ZIKV strains isolated from A . dalzieli , suggesting a possible important role of this zoophilic vector [71] in West Africa . This association was found to be robust to the exclusion of vertebrate host from the analysis . The plausibility of the putative recombination events we detected ( Table S2 ) , could in part be explained by mosquitoes taking sequential blood meals from distinct animal species harboring distinct ZIKV lineages , which is in line with ours and others host range findings [70] . Also , when analyzing the increase of ZIKV activity in Kedougou , ( where most of the viruses analyzed herein were collected ) , we noticed that ZIKV activity is much more frequent , with an interval of 1–2 years , compared to the 5 to 8 years cycle of dengue and yellow fever virus . Hence from 1972 to 2002 , ZIKV emerged in 1973 , 1976 , 1979 , 1980 and 1981 . Such frequent activity can also be an opportunity of co-circulation and mixing of multiple genotypes present in the forest and that may favor recombination among them . The occurrence of recombination among ZIKV strains in time-scaled phylogenetic trees suggested that some ZIKV lineages sampled in Dezidougou ( CI ) in 1990 ( ArA27101 , ArA27290 , ArA27096 , ArA27443 , ArA27407 and ArA27106 ) with recombinant E ( Figure S2C ) shared a common ancestor around 1962 ( ranging from 1951 to 1967 HPD with 95% of confidence interval ) . Likewise , the strain ArA982 was also isolated at Dezidougou in 1999 and its sister-group ArA986 , which shared a common ancestor with the former around 1992 ( ranging from 1981 to 1996 HPD with 95% of confidence interval ) , was sampled in the neighbor province Sokala-Sobara ( CI ) in 1999 . Together these results indicated that recombination in envelope protein could be an important trend among the enzootic cycle of ZIKV at this region in Côte d'Ivoire , as ZIKV lineages did not show a clear pattern of host preference and recombination requires the infection of the same host by more than one viral strain . Besides , the other E recombinant strain ( HD78788 ) , isolated from a human case at Dakar ( SN ) in 1991 , shared a common ancestor around 1984 ( ranging from 1976 to 1988 HPD with 95% of confidence interval ) with ZIKV strains from Kedougou ( SN ) . Conversely , the three NS5 recombinants did not cluster along phylogenetic trees ( Figure S2C ) , although two of them were isolated in Kedougou from A . dalzieli mosquitoes in 2001 ( ArD157995 and ArD158084 ) and the other ( ArB1362 ) was isolated in Bouboui , Central African Republic , from A . africanus mosquitoes in 1968 . The preferential distribution of recombinant strains along phylogenies was supported by significant p-values of AI and PS ≤2 . 00E-4 ( Dataset S2 ) and the adjacency patterns of E and NS5 recombinants were also confirmed by MC statistics ( Dataset S2 ) . Our analyses predicted several glycosylation sites in the E protein ( Figure 3 ) . We detected a probable motif ( Asn-X-Thr ) among E sequences from several ZIKV strains ( Figure 3A ) , which suggests a N-linked glycosylation site in the residue Asn-154 , in agreement with [17] , [26] . This residue is located on an α-helix in the E protein structure ( yellow arrow in Figure 3A and yellow bead in Figure 3B ) . Our results also pointed several O-linked glycosylation sites in the E protein ( red arrows in Figure 3A and red beads Figure 3B ) but no C-mannosylated site . We found a probable mucin-type O-linked glycosylated site at residue Thr-170 in E protein from all ZIKV strains , and other mucin sites at residues Thr-245 and Thr-381 in some isolates ( Figure 3A ) . In addition , we also uncovered probable O-GlcNAc attachment sites at residues Ser-142 , Ser-227 , Thr-231 , Ser-304 , Thr-366 and Thr-381 in E from some strains ( Figure 3A ) . Given the importance of the N-linked glycosylation site around position 154 of the E protein for infectivity and assembly of flaviviruses [72]–[74] and the fact that we observed polymorphisms in this motif ( deletions and substitutions 156 Thr/Iso ) , we investigated the correlation between the conservation of this motif ( Asn-X-Thr ) and phylogenies for ZIKV strains . Our results suggested that the acquisition of this glycosylation site is a recurrent event in the history of ZIKV , given the observed changes from Isoleucine to Threonine and vice-versa more than once in the MCC tree ( Figure S2D ) , supported by p-values for AI and PS ≤7 . 00E-4 ( Dataset S2 ) . However , our conclusions are limited due to serial passages of the former ZIKV strains ( Figure S2D ) in mouse brain [26] , which could result in the loss of this glycosylation site , as observed in West Nile virus [75] . Since it was demonstrated that the absence of an N-linked glycosylation site on the E protein enhances viral infectivity for C6/36 mosquito cells [72] , [73] and E protein of ZIKV strains from A . dalzieli , which was the unique vector source with significant MC–showed an absence of this glycosylation site , we investigated the correlation between this mosquito-source and N-linked glycosylation patterns of E protein along PST . Our results indicated the changes in glycosylation patterns ( presence or absence ) and vector ( A . dalzieli or not ) were correlated during ZIKV emergence , which was supported by BF for dependent model ( BF≈47 . 004 ) greater than for them to independent model . These findings could be related to the enzootic cycle of ZIKV in West Africa and the zoophilic behavior of A . dalzieli [71] , whose females take blood meals from a broad range of vertebrates , which provides additional evidence for the absence of host preference ( as described in Animal sources of ZIKV ) . Hence , further studies are necessary to understand the consequences of our results to ZIKV transmission cycle in nature . Our analyses indicated that ZIKV may have experienced several recombination events , which is uncommon among flaviviruses [66] . The recurrent loss and gain of the N-linked glycosylation site in the E protein could be related to mosquito-cell infectivity [73] and the correlated loss of this glycosylation site in ZIKV strains from A . dalzieli also provides indirect evidence for the enzootic cycle , since this vector has a zoophilic behavior [71] that may spread ZIKV among several hosts . Crucially , our results corroborated the notion that at least three distinct ZIKV clusters shared a common ancestor possibly with Ugandan lineages around 1920 , followed by two events of East to West Africa spread ( Figure 2 ) : ( i ) one related to the MR766 cluster introduction to Côte d'Ivoire and posterior spread to Senegal and; ( ii ) other related to the Nigerian cluster introduction in Senegal and posterior dispersion to Côte d'Ivoire and Burkina Faso .
|
Zika fever is a mosquito-borne illness caused by a flavivirus . Human infections with Zika virus ( ZIKV ) could cause fever , malaise and cutaneous rash . Despite several ZIKV reports since 1947 when it was first isolated at Zika forest in Uganda , molecular evolution of ZIKV as an emerging agent remains poorly understood . Moreover , despite several ZIKV reports from Africa and Asia , few human cases were notified until 2007 when an epidemic took place in Micronesia . In West Africa , surveillance programs have reported periodic circulation of the virus since 1968 . To help fill the gap in understanding ZIKV evolution , 43 ZIKV samples were analyzed . We focused on: ( i ) adaptive genetic changes including protein glycosylation patterns , ( ii ) phylogenetic relationship among isolates and their spatiotemporal patterns of spread across Africa and Asia and , ( iii ) dispersion among vertebrate reservoirs and invertebrate vector species . Our results indicated that ZIKV may have experienced recombination in nature and that , after it emerged from Uganda in the early of the 20th century , it moved to West Africa and Asia in the first half of the century , without any clear preference for host and vector species .
|
[
"Abstract",
"Introduction",
"Methods",
"Results/Discussion"
] |
[
"medicine"
] |
2014
|
Molecular Evolution of Zika Virus during Its Emergence in the 20th Century
|
To map the neural substrate of mental function , cognitive neuroimaging relies on controlled psychological manipulations that engage brain systems associated with specific cognitive processes . In order to build comprehensive atlases of cognitive function in the brain , it must assemble maps for many different cognitive processes , which often evoke overlapping patterns of activation . Such data aggregation faces contrasting goals: on the one hand finding correspondences across vastly different cognitive experiments , while on the other hand precisely describing the function of any given brain region . Here we introduce a new analysis framework that tackles these difficulties and thereby enables the generation of brain atlases for cognitive function . The approach leverages ontologies of cognitive concepts and multi-label brain decoding to map the neural substrate of these concepts . We demonstrate the approach by building an atlas of functional brain organization based on 30 diverse functional neuroimaging studies , totaling 196 different experimental conditions . Unlike conventional brain mapping , this functional atlas supports robust reverse inference: predicting the mental processes from brain activity in the regions delineated by the atlas . To establish that this reverse inference is indeed governed by the corresponding concepts , and not idiosyncrasies of experimental designs , we show that it can accurately decode the cognitive concepts recruited in new tasks . These results demonstrate that aggregating independent task-fMRI studies can provide a more precise global atlas of selective associations between brain and cognition .
A major challenge to reaching a global understanding of the functional organization of the human brain is that each neuroimaging experiment only probes a small number of cognitive processes . Cognitive neuroscience is faced with a profusion of findings relating specific psychological functions to brain activity . These are like a collection of anecdotes that the field must assemble into a comprehensive description of the neural basis of mental functions , akin to “playing twenty questions with nature” [1] . However , maps from individual studies are not easily assembled into a functional atlas . On the one hand , the brain recruits similar neural territories to solve very different cognitive problems . For instance , the intra-parietal sulcus is often studied in the context of spatial attention; however , it is also activated in response to mathematical processing [2] , cognitive control [3] , and social cognition and language processing [4] . On the other hand , aggregating brain responses across studies to refine descriptions of the function of brain regions faces two challenges: First , experiments are often quite disparate and each one is crafted to single out a specific psychological mechanism , often suppressing other mechanisms . Second , standard brain-mapping analyses enable conclusions on responses to tasks or stimuli , and not on the function of given brain regions . Cognitive subtraction , via the opposition of carefully-crafted stimuli or tasks , is used to isolate differential responses to a cognitive effect . However , scaling this approach to many studies and cognitive effects leads to neural activity maps with little functional specificity , hard to assemble in an atlas of cognitive function . Indeed , any particular task recruits many mental processes; while it may sometimes be possible to cancel out all but one process across tasks ( e . g . through the use of conjunction analysis [5] ) , it is not feasible to do this on a large scale . Furthermore , it can be difficult to eliminate all possible confounds between tasks and mental processes . An additional challenge to the selectivity of this approach is that , with sufficient statistical power , nearly all regions in the brain will respond in a statistically significant way to an experimental manipulation [6] . The standard approach to the analysis of functional brain images maps the response of brain regions to a known psychological manipulation [7] . However , this is most often not the question that we actually wish to answer . Rather , we want to understand the mapping between brain regions/networks and psychological functions ( i . e . “what function does the fronto-parietal network implement ? ” ) . If we understood these mappings , then in theory we could predict the mental state of an individual based solely on patterns of activation; this is often referred to as reverse inference [8] , because it reverses the usual pattern of inference from mental state to brain activation . Whereas informal reverse inference ( e . g . based on a selective review of the literature ) can be highly biased , it is increasingly common to use meta-analytic tools such as Neurosynth [9] to perform formal reverse inference analyses ( also know as decoding ) . However , these inferences remain challenging to interpret due to the trade-off between breadth and specificity that is necessary to create a sufficiently large database ( e . g . see discussion in [10 , 11] ) . The optimal basis for brain decoding would be a large database of task fMRI datasets spanning a broad range of mental functions . Previous work has demonstrated that it is possible to decode the task being performed by an individual , in a way that generalizes across individuals [12] , but this does not provide insight into the specific cognitive functions being engaged , which is necessary if we wish to infer mental functions associated with novel tasks . The goal of decoding cognitive functions rather than tasks requires that the data are annotated using an ontology of cognitive functions [13–15] , which can then become the target for decoding . Some recent work has used a similar approach in restricted domains , such as pain [16] , and was able to isolate brain networks selective to physical pain . Extending this success to the entire scope of cognition requires modeling a broad range of experiments with sufficient annotations to serve as the basis for decoding . To date , the construction of human functional brain atlases has primarily relied upon the combination of resting-state fMRI and coordinate-based meta-analyses . This approach is attractive because of the widespread availability of resting-state fMRI data ( from which brain functional networks can be inferred through statistical approaches [17] ) , and the ability to link function to structure through the use of annotated coordinate-based data ( such as those in the BrainMap [18] and Neurosynth [9] databases ) . This approach has identified a set of large-scale networks that are consistently related to specific sets of cognitive functions [19 , 20] , and provides decompositions of specific regions [21 , 22] . However , resting-state analysis is limited in the set of functional states that it can identify [23] , and meta-analytic databases are limited in the specificity of their annotation of task data , as well as in the quality of the data , given that it is reconstructed merely from activation coordinates reported in published papers . A comprehensive functional brain atlas should link brain structures and cognitive functions in both forward and reverse inferences [7] . To build such a bilateral mapping , we introduce the concept of “ontology-based decoding , ” , in which the targets of decoding are specific cognitive features annotated according to an ontology . This idea was already present in [9 , 12 , 24]; here we show how an ontology enables scaling it to many cognitive features , to increase breadth . In the present case , we use the Cognitive Paradigm Ontology ( CogPO ) [15] , that provides a common vocabulary of concepts related to psychological tasks and their relationships ( see S1 Text Distribution of terms in our database ) . Forward inference then relies on ontology-defined contrasts across experiments , while reverse inference is performed using an ontology-informed decoder to leverage this specific set of oppositions ( see Fig 1 and methodological details ) . We apply these forward and reverse inferences to the individual activation maps of a large task-fMRI database: 30 studies , 837 subjects , 196 experimental conditions , and almost 7000 activation maps ( see S1 Text Distribution of terms in our database ) . We use studies from different laboratories , that cover various cognitive domains such as language , vision , decision making , and arithmetics . We start from the raw data to produce statistical brain maps , as this enables homogeneous preprocessing and thorough quality control . The results of this approach demonstrate that it is possible to decode specific cognitive functions from brain activity , even if the subject is performing a task not included in the database .
The main challenge to accumulate task fMRI is to account for the disparity in experimental paradigms . One solution is the use of cognitive ontologies that define terms describing the cognitive tasks at hand and enable to relate them . The choice of the ontology must meet two opposite goals: have a good coverage of the cognitive space , and document overlap between studies . In practice , each cognitive term describing mental processes must be expressed in several studies of our database to ensure the generalizability of our inference . Standard forward inference in functional neuroimaging uses the GLM ( general linear model ) , which models brain responses as linear combinations of multiple effects . We use a one-hot-encoding of the concepts , i . e . we represent their presence in the tasks by a binary design matrix . We test for response induced by each concept with a second-level analysis using cross-studies contrasts . To disentangle various experimental factors , brain mapping uses contrasts . Individual studies are crafted to isolate cognitive processes with control conditions , e . g . a face-recognition study would rely on a “face versus place” or a “face versus scrambled picture” contrast . To separate cognitive factors without a strong prior on control conditions , the alternative is to contrast a term against all related terms , e . g . , “face versus place and scrambled picture” . We use the categories of our ontology to define such contrasts in a systematic way for the wide array of cognitive concepts touched in our database . This approach yields groups of terms within the task categories , as described in Table 1: the task categories are used to define the conditions and their controls . Inside each group , we perform a GLM analysis with all the “one versus others” contrasts . We denote these ontology contrasts . Compared to a standard group analysis , the benefit of this GLM is that the control conditions for each effect studied span a much wider range of stimuli than typical studies . For reverse inference , we rely on large-scale decoding [12] . Prior work [12 , 24] tackles this question using a multi-class predictive model , the targets of the classification being separate cognitive labels . Our formulation is different as our goal is to predict the presence or absence of a term , effectively inverting the inference of our forward model based on one-hot-encoding . This implies that each image is associated with more than a single label , which corresponds to multi-label classification in a decoding setting .
Using a database of 30 studies , we demonstrate that our approach captures a rich mapping of the brain , identifying networks with a specific link to cognitive concepts . Prediction of cognitive components in new paradigms validates this claim . We combine forward and reverse inference to construct a one-to-one mapping between brain structures and cognitive concepts . Forward inference across studies requires adapting brain mapping analysis to leverage the ontology . Mapping the brain response to the presence of a concept in tasks selects unspecific regions , as it captures other related effects , e . g . selecting the primary visual cortex for any visual task ( Fig 3 ) . To obtain a more focal mapping , we remove these effects by opposing the concept of interest to related concepts in the ontology . Reverse inference narrows down to regions specific to the term . However , as we use a multivariate procedure , some of its variables may model sources of noise [26] . For instance , when using visual n-back tasks with a motor response to map the visual system , the motor response creates confounding signals . A multivariate procedure could use signal from regions that capture these confounds to subtract them from vision-specific activity , leading to better prediction . As such regions are not directly related to the task , they are well filtered with a standard GLM ( General Linear Model ) used in forward inference . For this reason , our final maps combine statistics from forward and reverse inference: functional regions are composed of voxels that are both recruited by the cognitive process of interest and predictive of this process; see S5 Text Consensus between forward and reverse inference for statistical arguments and [27] for more fundamental motivations regarding causal inference . Fig 3a–3d shows how the neural-activity patterns for the “places” label progressively narrow on the PPA with the different approaches . Thus we link each cognitive concept to a set of focal regions , resulting in a brain-wide functional atlas . To build functional atlases , it is important to clearly identify the regions associated with different cognitive concepts . Fig 3e shows that reverse-inference meta-analysis with Neurosynth also associates the PPA with the “place” term , but the region is not as well delineated as with our approach . Fig 4 shows functional atlases of auditory and visual regions extracted with various mapping strategies . The relative position and overlap of the various maps is clearly visible . Forward-inference mapping of the effect of each term versus baseline on our database gives regions that strongly overlap ( Fig 4a ) . Indeed , the maps are not functionally specific and are dominated by low-level visual mechanisms in the occipital cortex and language in the temporal cortex . Using contrasts helps decreasing this overlap ( Fig 4b ) , and hence reveals some of the functional segregation of the visual system . However , as the stimuli are not perfectly balanced across experiments , contrasts also capture unspecific regions , such as responses in the lateral occipital cortex ( LOC ) for faces or places . Reverse inference with a logistic-regression decoder gives well separated regions , albeit small and scattered ( Fig 4c ) . The ontology-informed approach identifies well-separated regions that are consistent with current knowledge of brain functional organization ( Fig 4d ) . Finally , meta analysis with NeuroSynth separates maps related to the various terms better than forward analysis on our database of studies ( Fig 4e ) . Yet some overlap remains , for instance in the LOC for maps related to visual concepts . In addition , the outline of regions is ragged , as the corresponding maps are noisy ( Fig 3e ) , probably because they are reconstructed from peak coordinates . Note that overlaps across term-specific topographies are ultimately expected to remain , especially in associative cortices . In the following , we first discuss quantitative validation of the reverse-inference atlases , and then study in detail the atlas obtained with the ontology-informed approach . Upon qualitative inspection , the regions extracted by our mapping approach provide a good functional segmentation of the brain . For an objective test of this atlas , we quantify how well these regions support reverse inference . For this , we use the ontology-informed decoder to predict cognitive concepts describing tasks in new paradigms and measure the quality of the prediction . This approach was tested using a cross-validation scheme in which 3 studies were held out of each training fold for subsequent testing . Fig 5 shows the corresponding scores: ontology-informed decoding accurately predicts cognitive concepts in unseen tasks . It predicts these concepts better than other commonly used decoders ( logistic regression and naive Bayes , see also S6 Text Evaluating prediction accuracy: cross-validation ) and NeuroSynth decoding based on meta-analysis . This confirms that the corresponding atlas captures areas specialized in cognitive functions better than conventional approaches . Very general labels such a “visual” are found in most studies , and therefore easy to predict . However , higher-level or more specialized cognitive concepts such as viewing digits or moving the left foot are seldom present ( see S1 Text Distribution of terms in our database ) . For these rare labels , the fraction of prediction errors is not a useful measure . Indeed , simply assigning them to zero images would lead to a small fraction of errors . For this reason , Fig 5 reports the area under the receiver operating characteristic ( ROC ) curve . This is a standard metric that summarizes both false positives and false negatives and is not biased for rare labels . This analysis showed that even for relatively rare concepts , successful decoding was possible . Our approach links different cognitive terms to functionally-specialized brain regions:
Our analysis framework overcomes the loss in specificity typical of data aggregation . As a result , it enables analyzing jointly more cognitive processes . These richer models can map qualitatively different information . Analyzing more diverse databases of brain functional images can bring together two central brain-mapping questions: where is a given cognitive process implemented , and what cognitive processes are represented by a given brain structure . Answers to the “what” question have traditionally been provided by invasive studies or neurological lesion reports . Indeed , in a given fMRI study , brain activity results from the task . Concluding on what processes are implied by the observed activity risks merely capturing this task . Decoding across studies can answer this question , by demonstrating the ability to perform accurate inference from brain activity to cognitive function [36] . Reverse-inference maps are essential to functional brain mapping . A key insight comes from the analysis in NeuroSynth [9]: some brain structures are activated in many tasks . Hence , a standard analysis –forward inference– showing such a structure as activated does not provide much information about what function is being engaged . Reverse inference puts the observed brain activity in a wider context by characterizing the behavior that it implies . The analysis performed in NeuroSynth accounts for the multiple tasks that activate a given structure , performing a Bayesian inversion with the so-called Naive Bayes model; however , it does not account for other activation foci in the brain that characterize the function . Put differently , our approach departs from the model used by NeuroSynth for reverse inference by what it conditions upon: NeuroSynth’s model asserts functional specialization conditional to other terms , while we condition on other brain locations when predicting concept occurrence . This difference should be kept in mind when interpreting differences between the two types of approaches . The Inferior Temporal Gyrus ( ITG ) , for instance , is more active in object-recognition tasks than in other paradigms . However , observing activity in the ITG does not help deciding whether the subject is recognizing faces or other types of objects: the information is in the Fusiform gyrus . An important difference between reverse-inference maps with a Naive Bayes –as in Neurosynth– and using a linear model –as in our approach– is that the Naive Bayes maps do no capture dependencies across voxels . On the opposite , linear models map how brain activity in a voxel relates to behavior conditionally on other voxels . Technically , this is the reason why Neurosynth reverse-inference maps related to object recognition overlap in the IT cortex ( Fig 3e ) while maps produced by our approach separate the representations of the various terms in the ventral mosaic ( Fig 3d ) . Another , more subtle , benefit of the two-layer model over more classical multi-label approaches is that it combines the decisions of classifiers based on subsets of the data , such as the OvO classifiers , which helps learning relevant local discriminative information . In sum , our mapping approach provides a different type of brain maps: They quantify how much observing activity in a given brain location , as opposed to other brain locations , informs on whether the subject was engaged in a cognitive operation . Brain functional atlases are hard to falsify: is a functional atlas specific to the experimental paradigms employed to build it , or is it more generally characteristic of human brain organization ? The success of statistically-grounded reverse inference , which generalizes to new paradigms from unseen studies , suggests that there must be some degree of generality in the present atlas . In demonstrating this generalization , the present work goes beyond previous work that had shown generalization to new subjects under known task conditions [12] , but not to unknown protocols . However , it is worth noting that here too we found that it was easier to predict on held-out subjects ( from one of the training studies ) than on held-out studies ( see S6 Text Evaluating prediction accuracy: cross-validation ) , consistent with a substantial effect of the specific task ( see S2 Text Similarities of activations across the database ) . Despite this , our ontology-enabled approach was able to successfully predict cognitive processes for new tasks . Interestingly , it opens the possibility to perform prospective decoding analyses on novel data , hence makes it easier to grasp the added information of incoming data . To enable this generalization across paradigms , we characterize each task by the multiple cognitive concepts that it recruits , that are specified in the ontology . Departing from the subtractions often used in brain mapping , our framework relies on quantifying full descriptions of the tasks . In the context of decoding , this approach leads to multi-label prediction , predicting multiple terms for an activation map , as opposed to multi-class prediction , used in prior works [12 , 16] , that assigns each new map to a single class . The use of the multi-label approach combined with an ontology capturing the relationships between terms provides a principled way of modeling the multiple components of cognition and thus avoids the need for hand-crafted oppositions that are customarily used in subtraction studies . Defining good ontologies is yet another challenge for the community , but it is not unlikely that brain imaging will become part of that process [36 , 37] . Providing a methodological approach founded on an explicit hierarchy of cognitive concepts would allow to test for different cognitive ontologies , and , provided with a comparison metric , select the best ontology according to the available data . Although the present analysis is limited to a relatively small set of cognitive functions , such an approach will be essential as the field attempts to scale such analyses to the breadth of human cognition . To build brain functional atlases that map many cognitive processes , we have found that reverse inference and an ontology relating these processes were key ingredients . Indeed , because of the experimental devices used in cognitive neuroimaging , some regions –e . g . attentional or sensory regions– tend to be overly represented in forward inferences . An ontology encodes the related cognitive processes that must be studied together to best establish forward or reverse inferences . Using a relatively small number of independent task fMRI datasets , our brain-mapping approach reconciles the conundrum of multiple cognitive processes/labels mapping to often overlapping brain regions in activation studies . More data will enable even more fine-grained process-region mappings . In particular higher-level cognitive processes elude the present work , limited by the amount and the diversity of the studies in our database . Indeed , high-level terms form very rare classes in the datasets employed here ( see S1 Text Distribution of terms in our database ) . With increased data sharing in the neuroimaging community [38] , there is a growing opportunity to perform this kind of analysis on a much larger scale , ultimately providing a comprehensive atlas of neurocognitive organization . A major challenge to such analyses is the need for detailed task annotation; whereas annotation of task features such as the response effector is relatively straightforward , annotation of complex cognitive processes ( e . g . , whether a task involves attentional selection or working memory maintenance ) is challenging and often contentious . The utility of the ontology in the present work suggests that this effort is worthwhile , and that the increased utilization of ontologies in cognitive neuroscience may be an essential component to solving the problem of how cognitive function is organized in the brain .
|
Cognitive neuroscience uses neuroimaging to identify brain systems engaged in specific cognitive tasks . However , linking unequivocally brain systems with cognitive functions is difficult: each task probes only a small number of facets of cognition , while brain systems are often engaged in many tasks . We develop a new approach to generate a functional atlas of cognition , demonstrating brain systems selectively associated with specific cognitive functions . This approach relies upon an ontology that defines specific cognitive functions and the relations between them , along with an analysis scheme tailored to this ontology . Using a database of thirty neuroimaging studies , we show that this approach provides a highly-specific atlas of mental functions , and that it can decode the mental processes engaged in new tasks .
|
[
"Abstract",
"Introduction",
"Materials",
"and",
"methods",
"Results",
"Discussion"
] |
[
"cognitive",
"neurology",
"medicine",
"and",
"health",
"sciences",
"ontologies",
"social",
"sciences",
"neuroscience",
"learning",
"and",
"memory",
"face",
"recognition",
"perception",
"data",
"management",
"cognitive",
"neuroscience",
"cognitive",
"psychology",
"cognition",
"brain",
"mapping",
"memory",
"vision",
"neuroimaging",
"research",
"and",
"analysis",
"methods",
"language",
"computer",
"and",
"information",
"sciences",
"imaging",
"techniques",
"psychology",
"neurology",
"biology",
"and",
"life",
"sciences",
"sensory",
"perception",
"cognitive",
"science"
] |
2018
|
Atlases of cognition with large-scale human brain mapping
|
Although persistent viral diseases are a global health concern , the mechanisms of differential susceptibility to such infections among individuals are unknown . Here , we report that differential interactions between dendritic cells ( DCs ) and virus are critical in determining resistance versus susceptibility in the Theiler murine encephalomyelitis virus–induced demyelinating disease model of multiple sclerosis . This virus induces a chronic demyelinating disease in susceptible mice , whereas the virus is completely cleared in resistant strains of mice . DCs from susceptible mice are more permissive to viral infection , resulting in severe deficiencies in development , expansion , and function , in contrast to DCs from resistant mice . Although protective prior to viral infection , higher levels of type I interferons ( IFNs ) and IFN-γ produced by virus-infected DCs from susceptible mice further contribute to the differential inhibition of DC development and function . An increased DC number and/or acquired resistance of DCs to viral infection render susceptible mice resistant to viral persistence and disease progression . Thus , the differential permissiveness of DCs to infectious agents and its subsequent functional and developmental deficiencies determine the outcome of infection- associated diseases . Therefore , arming DCs against viral infection–induced functional decline may provide a useful intervention for chronic infection-associated diseases .
Susceptibility to immune-mediated diseases induced by microbial infections varies between individuals . Several genes involved in innate and adaptive immune responses are known to be associated with such differential susceptibility [1 , 2] . Consequently , different levels and types of immune responses to causative antigens are often correlated with susceptibility/resistance to diseases . However , the underlying immune mechanisms are largely unknown . Therefore , understanding the mechanisms involved in the determination of resistance versus susceptibility is a crucial step towards better disease control . Recent studies have indicated that dendritic cells ( DCs ) play a critical role in the induction and maintenance of strong immunity to infectious agents [3] . Thus , it is conceivable that the differential responsiveness of DCs to an identical pathogen in resistant and susceptible individuals may lead to disparate levels of protective and/or pathogenic immunity , resulting in differential susceptibility to virus-associated diseases . In this study , we examine this possibility using the Theiler murine encephalomyelitis virus ( TMEV ) –induced demyelinating disease model for multiple sclerosis ( MS ) . TMEV is a picornavirus that causes chronic demyelination in the white matter of the central nervous system ( CNS ) of susceptible mice , such as SJL/J ( SJL ) , following the establishment of persistent infection [4] . This demyelinating disease has immune parameters and histopathology similar to those of chronic progressive MS and thus is extensively studied as an infectious model for MS [4] . However , susceptibility to persistent infection and demyelination varies among inbred strains of mice . For example , in contrast to susceptible SJL mice , resistant strains of mice such as C57BL/6 ( B6 ) are able to completely clear the virus and do not develop late chronic demyelinating disease [5 , 6] . Many previous studies have indicated that TMEV clearance from the CNS is mainly mediated by CD8+ cytotoxic T lymphocytes ( CTLs ) [7–9] , although such T cells may also be pathogenic , particularly during chronic demyelinating disease . The anti-viral CTL response in resistant mice is significantly more rapid and intense compared to that of susceptible mice [10 , 11] , suggesting that this response is critical in preventing the establishment of viral persistence . Similarly , the level of virus capsid–specific CD4+ T cell response following viral infection in resistant mice is significantly higher than that in susceptible mice , despite the higher level of overall CD4+ T cell infiltration in the CNS of susceptible mice [12] . These results demonstrate that lower levels of initial anti-viral T cell response in susceptible mice may consequently allow the virus to establish chronic persistence . However , the underlying cellular mechanisms involved in such differential levels of anti-viral immune responses between susceptible and resistant mice are not fully understood . DCs play a central role in the initiation of both innate and adaptive immune responses to infectious agents , including bacteria and viruses [3] . They are distributed in the skin , mucosa , and in lymphoid organs , where they efficiently take up diverse microbial antigens , then process and present them as major histocompatibility complex ( MHC ) –peptide complexes to T cells . Following the recognition of pathogens , DCs undergo a series of activation processes . Once activated , DCs migrate into the T cell zone of lymphoid organs , where antigen-specific immune cells can be primed . Migration coincides with a maturation process that results in morphological , phenotypic , and functional changes . Moreover , activated DCs have the capacity to recruit and/or activate cells of the innate and adaptive immune systems [3] . However , DC function and its consequent adaptive immune response depend largely upon the interaction between DCs and microbial signals , which can enhance or impair DC function via Toll-like receptors , c-type lectins , CD40 , cytokine receptors , or other pathways [3] . Interestingly , several research groups have recently shown the presence of DCs in the meninges and choroid plexus of healthy rodents [13–16] . Moreover , recent studies have demonstrated that local antigen-presenting cells ( APCs ) , possibly DCs , can activate naïve T cells that enter the inflamed CNS [17] . It has also been noted that the DCs associated with CNS vessels are sufficient to prime myelin-reactive T cells in vivo , resulting in CNS inflammation and clinical disease [18] . Thus , it is possible that neurotropic TMEV evades immune recognition and establishes latency by subverting DC function in susceptible strains of mice , but not in resistant mice . In this study , we address the role of DC responses in viral persistence and its consequent demyelinating disease in prototypically susceptible SJL and resistant B6 mice . We sought to answer the following questions: First , is the level of viral infection/replication in DCs from susceptible mice greater than that of resistant mice ? Second , are DC differentiation and activation/maturation preferentially impaired in susceptible mice ? If this is the case , what are the major mechanisms involved in these differential DC responses following viral infection ? Finally , can an increased DC number and/or acquired resistance to viral infection rescue susceptible mice from chronic viral persistence and from developing demyelinating disease ?
There are significant differences in persistent viral infection and its consequent demyelinating disease between SJL and B6 mice . To examine whether these differences may be associated with differential susceptibility of DCs to TMEV infection , bone marrow ( BM ) DCs were studied for expression of TMEV antigens and virus production . TMEV was able to bind equally well to DCs from either susceptible SJL or resistant B6 mice ( Figure 1A ) . However , ultraviolet ( UV ) -inactivated TMEV failed to bind DCs from SJL and B6 mice , which is consistent with the abrogation of receptor-mediated binding of other picornaviruses by UV inactivation [19] . These findings suggest that the binding of TMEV to DCs is not due to non-specific stickiness . In addition , TMEV+ cells were detected by 6 h post-infection in the SJL DCs , but not at 1 h post-infection , suggesting that new protein synthesis was required for the detection ( Figure 1B ) . In contrast to SJL DC cells , DC cells from B6 mice were far less susceptible to infection , and the virus yield in single-step growth curves from B6 cultures was approximately one log lower than that from SJL cultures ( Figure 1B , right panel ) . Differences in DC viral infection between SJL and B6 mice were evident throughout several days of culturing BM cells ( BMCs ) ( Figure 1C ) , despite comparable levels of CD11c in both SJL and B6 DCs ( unpublished data ) . Even on the peak day , day 6 , the level of viral infection in SJL DCs was significantly greater ( >3-fold ) than that of B6 DCs ( Figure 1C ) . It was clear in these experiments that the cells producing viral proteins were CD11c+ ( Figure 1B ) , suggesting that DCs produce the majority of infectious virus . In fact , assessment of viral production from isolated CD11c+ and CD11c− cells following viral infection confirmed that the great majority of BMCs producing virus proteins ( >95% ) are CD11c+ and the rest ( <5% ) are CD11c− cells ( Figure S1A and S1B ) . Furthermore , the level of infectious virus produced by CD11c+ cells is over 40-fold higher than that produced by CD11c− cells ( Figure S1A and S1C ) . A similar differential virus replication in splenic DCs between these mouse strains was also observed ( Figure S1D ) . Phenotypic analysis of the infected DC cells suggested that virus preferentially infects immature DCs and prevents their maturation . As shown in Figure 1D , the TMEV+ cells in either SJL or B6 cultures were MHC class IIlow , CD86low , and CCR7low , molecules whose expression is enhanced only during DC maturation . Interestingly , when compared to mock-infected cells , lower levels of expression of MHC class II , CD86 , or CCR7 were also observed in infected SJL cultures , even in cells that were not TMEV+ . For example , 64% of mock-infected cells expressed MHC class II while only 22% of the TMEV− cells in the SJL cultures expressed the molecule . This suggests that the maturation of uninfected DCs is altered in the SJL cultures , possibly indirectly via soluble factors produced by virus-infected cells . This potential bystander effect will be discussed further in the next section . In contrast , infected B6 DC cultures did not show dramatic reductions in their maturation-associated markers . These data strongly suggest that TMEV preferentially replicates in immature DCs and that , in the case of SJL DCs , arrests their maturation . To exclude the possibility that resistance to persistent TMEV infection in the B6 strain is attributed to the expression of the H-2Db gene [20] , DCs from B10 ( H-2b ) and B10 . S ( H-2s ) mice , which are relatively resistant to TMEV-induced demyelinating disease compared to susceptible SJL ( H-2s ) mice [21] , were also examined . Interestingly , our results show that BM DCs from resistant B10 ( H-2b ) and B10 . S ( H-2s ) are similarly resistant to viral infection compared to susceptible SJL ( H-2s ) mice ( Figure S2 ) . Thus , these results strongly suggest that differences in viral infection/replication levels in DCs do not reflect the H-2 differences , and that differential susceptibility of DCs correlates closely with susceptibility to TMEV-induced demyelinating disease . To investigate the potential mechanisms for bystander inhibition of DC maturation , cytokines produced in infected DC cultures were examined after 5 d of culture . As shown in Figure 2A , infected cultures produced higher levels of IL-12p40 , IL-6 , and interferon ( IFN ) -α than mock-infected controls . Differences in TNF-α and IL-10 were not observed . Interestingly , infected SJL DCs produced significantly more IL-6 and IFN-α than the B6 counterparts . Infection of SJL DCs induced the expression of mRNAs of several IFNs ( Figure 2B ) with fairly rapid kinetics ( 8 h post-infection ) , suggesting that these are direct consequences of the infection . These included type I IFNs ( IFN-α1 , IFN-α2 , IFN-α4 , IFN-α5 , IFN-α6 , IFN-α7 , IFN-α9 , IFN-α11 , IFN-αb , and IFN-β ) , and a type II IFN ( IFN-γ ) . Figure 2B shows representative examples . In contrast , the levels of other IFNs , such as epsilon , kappa , and lambda , were not significantly different ( unpublished data ) . The finding that high levels of IFNs are produced by SJL-infected DCs is of importance because these cytokines were found to block the activation of DCs ( Figure 2C ) . When SJL DCs were pretreated with IFN-α , IFN-β , or IFN-γ for 18 h , then stimulated with lipopolysaccharide ( LPS ) for the final 6 h of culture , DCs produced dramatically less IL-12p40 , IL-6 , and TNF-α . Interestingly , pretreatment with IFN-γ , but not IFN-α or IFN-β , enhanced IL-10 production , suggesting potential differences among these IFNs in inhibiting DC function ( Figure 2C ) . IL-6 , the major cytokine produced by TMEV-infected DCs , had only marginal effects on DC function , which suggests that type I and II IFNs are primarily responsible for the virus-induced inhibitory effects on DC activation . TMEV was shown to infect immature DCs in Figure 1B , blocking their maturation . In contrast , TMEV does not efficiently infect DCs that have been activated by LPS , a TLR4 ligand ( Figure 3A ) . A variety of TLR ligands , including viral components , are known to induce activation , maturation , and/or cytokine production in DCs [3] . Pretreatment of DCs with LPS greatly reduces the numbers of TMEV+ cells in the culture , but treatment of DCs with LPS after infection has no effect . The addition of LPS at the time of infection had only a partial effect . The reduced infection in cells pretreated with LPS was correlated with restored DC cytokine production ( Figure 3B ) . DCs induced by LPS produce the cytokines IL-12p40 , IL-6 , TNF-α , and IL-10 . TMEV preinfection significantly blocked the ability of these cells to produce IL-12p40 , IL-6 , and TNF-α in response to LPS . In contrast , when DCs were treated first with LPS , cytokine production was not inhibited by the TMEV infection . These results suggest that LPS pretreatment results in inhibition of viral replication and restoration of cytokine production . These patterns were similar in DCs of both SJL and B6 mice , although the levels of viral infection remain higher in SJL DCs compared to B6 DCs . Interestingly , a different pattern was found again with IL-10 , and infected SJL DCs produced significantly higher levels of this cytokine in response to LPS . The findings that virus-infected cells produce IFNs ( Figure 2A ) and that IFNs reduce the ability of DCs to respond to LPS ( Figure 2C ) suggest an IFN-mediated inhibition of DC maturation by viral infection . The IFNs produced by infected cells can reduce DC maturation , even in the uninfected cells in the culture , which explains the apparent bystander effect seen in Figure 1D . In addition , the IFNs not only blocked DC maturation , but they also prevented viral infection in pretreated cells ( Figure 3C ) . Furthermore , antibodies to IFNs were actually able to increase the numbers of TMEV+ cells in DC cultures ( unpublished data ) . These results are consistent with the possibility that the inhibitory effect of LPS on TMEV infection is mediated by IFNs produced by LPS-treated DCs . It is , therefore , important to note that TMEV-infected SJL DCs produce higher levels of IFNs than infected B6 cells . Even though more virus is produced in the initial infection of SJL DCs , the increased levels of IFNs produced would limit the spread of the infection to neighboring cells . This idea was confirmed by experiments in which supernatants from virus-infected cultures were used to pretreat DC cells ( Figure 3D ) . Pretreatment of fresh DC cells with cell-free culture media from SJL-infected cells dramatically reduced the numbers of TMEV+ cells in the new cultures . When DC cells from IFN-α/β receptor knockout ( IFN-α/βR ) animals were infected with TMEV , the culture supernatants did not effectively block a subsequent infection ( Figure 3D , right panel ) , suggesting that the supernatants inhibit viral infection mainly in a type I IFN–dependent manner . In addition , LPS pretreatment failed to inhibit TMEV replication in IFN-α/βR-deficient DCs ( Figure 3D ) , indicating the involvement of type I IFNs in the resistance of normal DCs acquired by pretreatment with LPS . However , IFN-α/βR-deficient DCs remained partially resistant to viral infection following pretreatment with infected culture supernatants , strongly suggesting that there is a cytokine ( s ) blocking in a type I IFN–independent manner . Since an adequate number of DCs is necessary to elicit strong innate and adaptive anti-viral immune responses , a substantial reduction of DCs resulting from viral infection may allow viruses to evade immune recognition [22] . To test the possibility that TMEV infection inhibits DC expansion from hematopoietic progenitors , BMCs that were cultured for 3 d in the presence of granulocyte monocyte colony-stimulating factor ( GM-CSF ) were infected with TMEV , and the rate of DC expansion was subsequently assessed . Only a few small clusters of proliferating cells were observed in virus-infected SJL BMCs at 3 days post-infection ( dpi ) , compared to mock-infected cultures ( Figure 4A ) . In sharp contrast to SJL BMCs , many large aggregates of proliferating cells were evident in virus-infected B6 BMC cultures , similar to that of mock-infected BMCs ( Figure 4A ) . The total number of SJL BMCs infected with TMEV continuously declined during the 4-d culture period , whereas the total number of virus-infected B6 BMCs increased slightly , although the rate of increase was significantly lower than that of mock-infected cultures ( Figure 4B ) . Furthermore , BMCs infected with TMEV failed to differentiate into plasmacytoid DCs or macrophages , as the majority of cells showed CD11c+ CD45RA−CD11b+F4/80− phenotype ( unpublished data ) . It is conceivable that the suppression of DC expansion is in part due to the apoptosis of BMCs induced by TMEV infection . To evaluate this possibility , we examined the levels of BMC apoptosis at 24 , 48 , and 72 h post-infection . The proportions of Annexin V–positive cells undergoing apoptosis were significantly higher in TMEV-infected SJL BMCs compared to those in B6 BMCs during the culture periods ( shown at 24 h post-infection in Figure 4C; unpublished data ) . Thus , SJL DCs are substantially more susceptible not only to viral infection but also to virus-induced apoptosis , and this certainly contributes to the failure of DC expansion from hematopoietic progenitors . The preferential apoptosis of SJL DCs may also contribute to the drastic inhibition of SJL DC expansion . However , not all infected DCs are undergoing apoptosis . As shown in the bottom experiment of Figure 4C , when cells infected for 24 h were isolated into apoptotic ( Annexin V–positive ) and non-apoptotic fractions , a significant number of Annexin V–negative , TMEV+ cells were found to be present ( 22% in SJL versus 5% in B6 ) . It is conceivable that these cells may also be deficient in their expansion . This possibility was confirmed by measuring the loss of carboxyfluorescein diacetate succinimidyl ester stain 3 d after infection ( Figure 4D ) . DCs from SJL mice showed poor proliferation ( 30% in infected cultures versus 90% in control cultures ) . In contrast , DCs from B6 mice ( 73% ) showed significant loss of staining intensity , consistent with ongoing cell divisions . The arrest of DC precursors following infection in SJL cultures is likely to reflect the increased production of IFNs , which is known to block DC expansion [23] . Indeed , when IFN-blocking antibody was included in CFSE experiments , the proliferation of infected SJL DCs was enhanced ( unpublished data ) . It was shown in Figure 3D that culture supernatants from TMEV-infected DCs produced cytokines that blocked further infection , but only when cells from IFNα/βR-sufficient mice were used . Similarly , culture supernatants were able to block the expansion of hematopoietic cell precursors ( Figure 4E ) . The inhibition was dose dependent and the activity of supernatants from infected SJL cells was significantly higher than that of supernatants from infected B6 cells . Type I and II IFNs also similarly reduced cellular expansions . Experiments with IFN-α/βR-deficient DCs indicated that type I IFNs and other inhibitory cytokines , most likely IFN-γ , were responsible for the arresting activity of SJL supernatants ( unpublished data ) . Taken together , these results indicate that BMCs from susceptible SJL mice , but not from resistant B6 mice , are highly permissive to viral infection , leading to virus-induced apoptosis and the production of high levels of regulatory cytokines , including type I IFNs , IFN-γ , and perhaps IL-6 , resulting in growth arrest of BMCs . To examine the effects of TMEV infection on SJL and B6 DCs in vivo , phenotypic and functional alterations in DCs from BM , spleen , and CNS of infected mice were analyzed . Interestingly , levels of CD11c+ DC populations in the BM and spleen from uninfected naïve SJL mice were significantly lower than those in naïve B6 mice ( p < 0 . 01 ) ( Figure S3A ) . After TMEV infection , the number of B6 DCs in BM rapidly increased at 2 dpi ( p < 0 . 05 ) and then decreased to a basal level at 4 dpi , whereas the number of SJL DCs remained low throughout . Similarly , the number of DCs in the spleens of B6 mice gradually increased ( p < 0 . 05 ) , whereas no significant increase was seen in SJL mice . In addition , the levels of DCs undergoing apoptosis in the BM and spleens of virus-infected SJL mice were greater than those of infected B6 mice ( Figure S3B ) . Therefore , lower DC numbers in virus-infected SJL mice are consistent with their differential susceptibility to apoptosis and growth arrest induced by TMEV infection in vitro ( Figure 4 ) . The accumulation of DCs in the CNS has been reported in mice with local injections of heat-killed BCG [16] and EAE [24] . Thus , we examined the properties of DCs in the CNS , the virus target organ following TMEV infection . The number of infiltrating CD45hiCD11c+ DCs in the CNS of infected B6 mice was significantly higher ( >2-fold ) than that of SJL mice ( Figure 5A ) . In particular , B6 mice showed greater levels of myeloid CD11b+ DCs , but decreased levels of lymphoid CD8α+ DCs , compared to SJL mice ( Figure S4A ) . In contrast , splenic CD8α+ DC levels were lower in SJL mice compared to B6 mice , suggesting a differential distribution of this DC population between resistant and susceptible mice , which may have a role in local immunity to viral infection ( Figure S4B ) . Since TNF-α and/or IL-12 are important cytokines in the migration , accumulation , and activation of DCs involved in the elicitation of anti-viral immunity [3] , these cytokines were examined at 5 dpi ( Figure 5B ) . Very low levels of these cytokines were observed in unstimulated DCs from the CNS of TMEV-infected mice . After stimulation with LPS , cytokine production by DCs was enhanced in infected B6 , but poorly in SJL , mice . Similarly , DCs from the spleens of virus-infected SJL mice , but not B6 mice , were refractory to cytokine production in response to LPS stimulation ( Figure 5C ) . This result was confirmed with supernatants from isolated splenic DCs from mock- and TMEV-infected mice to exclude the cytokines produced by non-DC population ( Figure 5D ) . Taken together , these results suggest that TMEV infection severely interferes with the survival , CNS accumulation , and function of DCs in susceptible SJL mice , but not in resistant B6 mice . Mixed lymphocyte responses ( MLRs ) induced by infected DCs were assessed as a measure of DC function . T cells from naïve BALB/c mice were co-cultured with either SJL or B6 BM DCs infected in vitro with virus for 1 d ( unpublished data ) or 3 d ( Figure 6A ) . The allostimulatory function of virus-infected SJL DCs was severely compromised compared to that of either mock-infected or TMEV-infected B6 DCs . DCs isolated from infected SJL mice were similarly impaired in their ability to drive the MLR ( Figure 6B ) . It is unlikely that this differential reduction in the T cell stimulatory function of SJL DCs is due to infectious virus released from DCs during T cell stimulation , since paraformaldehyde ( PFA ) -fixed virus-infected SJL DCs similarly failed to stimulate allogeneic T cells ( Figure 6A ) . Key cytokines produced during allogeneic T cell responses were measured using either BM cells infected in vitro ( Figure 6C ) or splenic DCs from infected mice ( Figure 6D ) . Interestingly , T cells stimulated with either of these sources of infected SJL DCs failed to produce IFN-γ but gained the ability to produce IL-4 , as compared to mock-infected counterparts . In contrast , there was no significant difference in cytokine production when T cells were stimulated with virus-infected or mock-infected B6 DCs . These results strongly suggest that TMEV infection converts SJL DCs from Th1 stimulation to Th2 stimulation , while this infection has a minimal effect on B6 DCs . Since virus-specific Th1 cells are known to be more protective than Th2 cells [7 , 8] , such a switch in CD4+ T cell types may also facilitate the preferential persistence of virus in the CNS of susceptible SJL mice . Consistent with the above altered T cell differentiation , the expression levels of CD69 , an indicator of T cell activation , and TIM-3 , a Th1-associated marker , were reduced when TMEV-infected SJL DCs , but not B6 DCs , were cultured with naïve T cells ( Figure 6E ) . Similar patterns of CD69 and TIM-3 expression on T cells were also observed with splenic DCs from TMEV-infected mice ( Figure 6F ) . These results are consistent with the differential Th1/Th2 stimulation exhibited by SJL and B6 DCs following TMEV infection in vitro as well as in vivo . It is important to note that the expression levels of CD69 on CD8+ T cells are similarly reduced after stimulation with infected SJL DCs or DCs from infected mice compared to those from B6 mice ( Figure 6E and 6F ) , suggesting that CD8+ T cell stimulation is also compromised . The reduction in DC number and function in SJL mice may contribute to their greater susceptibility to demyelinating infections . Thus , interventions to enhance DC function might reduce the occurrence of disease in these animals . To explore this possibility , SJL mice were intracerebrally infected with TMEV alone or along with syngeneic DCs , LPS-pretreated DCs ( LPS-DCs ) , or non-DC splenocytes ( Figure 7 ) . The LPS-DC group was included because LPS-DCs are resistant to TMEV infection and maintain their function ( Figure 3A and 3B ) . At 7 dpi , macrophage levels ( CD45hiCD11b+ ) and virus-specific IFN-γ-producing CD8+ T cells in the CNS were significantly higher in mice infected with virus plus either untreated DCs or LPS-DCs , compared to those in mice infected with virus alone or virus plus non-DC splenocytes ( Figure 7A ) . Control mice receiving LPS-DCs alone failed to accumulate a significant level of leukocytes in the CNS ( Figure 7A ) . The percentage of virus-specific CD8+ T cells in the spleens of these mice was low but correlated with the number of CNS-infiltrating T cells . The group that received virus plus LPS-DCs displayed the highest ( 5 . 4% ) level among all experimental groups ( unpublished data ) . These data indicate that supplying additional DCs , particularly virus-resistant LPS-DCs , facilitates cellular infiltration to the CNS and elicitation of a strong CTL response against the virus . To relate levels of virus-specific CD8+ T cells to viral persistence , the viral load in brains and spinal cords was analyzed by plaque assay ( Figure 7B ) . Levels of virus in the CNS were inversely correlated to the number of virus-specific CD8+ T cell levels; viral persistence was significantly lower in mice that received either LPS-DCs or untreated DCs , which resulted in higher CD8+ T cell responses . The same patterns of viral persistence were maintained at 21 dpi , although the levels of virus were greatly reduced . More importantly , both the incidence and severity of demyelinating disease ( Figure 7C and 7D ) were reduced in animals with high levels of the anti-viral CD8+ T cell response and viral clearance . Mice receiving LPS-DCs , and to a lesser extent , mice receiving untreated DCs , were more resistant ( p < 0 . 001 and p < 0 . 01 , respectively ) to virally induced demyelinating disease than control mice receiving non-DC splenocytes . Abundant inflammatory foci and areas of myelin stripping were prominent within the white matter of spinal cords from mice receiving non-DC splenocytes ( Figure 7D ) . However , only limited levels of inflammatory cells and restricted myelin damage were observed in spinal cords from mice receiving untreated DCs , and these were less apparent in spinal cords from mice receiving LPS-DCs ( Figure 7D ) . These data indicate that the presence of additional DCs and more effective virus-resistant DCs ( LPS-DCs ) enable susceptible SJL mice to generate a strong virus-specific CD8+ T cell response and control viral persistence , resulting in resistance to demyelinating disease . Therefore , the availability of a high level of functionally intact DCs appears to be a critical factor in determining resistance of the host to virus-induced demyelinating disease , and perhaps similarly to other infection-associated chronic diseases .
The ability to generate a strong innate and adaptive immune response is critical for the resistance of a host to infectious agents [25–27] . However , many viruses can elude immune responses by an array of diverse pathways , including alterations in antigen presentation , cell apoptosis , and cytokine-mediated signaling [28] . Initial TMEV infection is successfully cleared in resistant mice , but not in susceptible mice , leading to a chronic , MS-like demyelinating disease . Susceptibility to persistent infection and its consequent demyelination is likely dependent upon the balance between viral persistence and host defense immunity . Since DCs are potent APCs , which are essential for strong anti-viral immunity , it would be advantageous for viruses to evade immunological recognition and establish latency by subverting this particular cell population . However , very few systematic comparative studies have been reported regarding the differential susceptibility of DCs to an identical pathogen between resistant and susceptible individuals . Here , for the first time to our knowledge , we provide evidence that the differential susceptibility of DCs to viral infection between resistant and susceptible mice leads to varying levels of DC function in anti-viral immunity and subsequent resistance/susceptibility to chronic demyelinating disease in these mice ( Figure 8 ) . It has recently been shown that generation of the anti-TMEV CD8+ T cell response in the CNS requires a BM-derived APC population [29] . Similarly , induction of the T cell response is impaired by DCs infected with foot-and-mouth disease virus , a member of the Picornaviridae [30] . In addition , the altered expression of accessory molecules on DCs upon interaction with human rhinovirus leads to down-modulation of adaptive immune responses during viral infection [31] . In this study , we demonstrate that DCs from SJL mice , but not B6 mice , are extremely permissive to TMEV infection/replication ( Figures 1 , S1 , and S2 ) . The differential susceptibility of DCs to TMEV infection between B6 and SJL mice may not be restricted to this cell type; similar differential susceptibilities to viral infection were also observed in macrophages ( unpublished data ) . Since these cell types are the major TMEV reservoir in the CNS [32] , the differential susceptibility of these cells may also contribute to viral persistence in these mice . Nevertheless , resistance of DCs from viral infection appears to be particularly important in facilitating strong anti-viral immunity , leading to viral clearance and protection from developing demyelinating disease . Furthermore , as diagrammed in Figure 8 , our results highlight a series of DC alterations that together may lead to the reduced immune response and viral persistence in susceptible SJL mice , whereas such alterations are not apparent in resistant B6 mice . First , TMEV preferentially infects immature DCs ( Figure 1 ) , blocking their differentiation and activation ( Figure 4 ) . A subset of the infected DCs undergoes apoptosis , but many arrest and fail to expand and mature . These infected DCs may also contribute to disease pathogenesis by altering the profile of cytokines produced and consequently by stimulating T cell subsets that deviate from those stimulated by intact DCs ( Figures 2 and 6 ) . In addition , the infected DCs from SJL mice produce cytokines that lead to bystander effects , inhibiting DC expansion and maturation in response to other stimuli such as LPS ( Figures 3 and 5 ) . TMEV-infected SJL DCs produce significantly higher levels of a variety of type I IFNs and type II IFN-γ compared to B6 DCs ( Figure 2 ) . The presence of type I or II IFNs provides protection against TMEV replication , but only when cells are exposed to IFNs prior to viral infection , indicating a transient nature of protection by IFNs ( Figure 3 ) . The resistance of TMEV to type I IFNs once infection is established is similar to previous observations in simian virus 40 , respiratory syncytial virus , dengue virus type 2 , and bovine viral diarrhea virus [33–37] . Thus , IFNs may not only limit the extent of TMEV infection but also provide the opportunity for these viruses to establish persistent infection . In addition , these cytokines also inhibit the production of IL-12p40 , IL-6 , and TNF-α by DCs in response to LPS stimulation ( Figure 2 ) . These results are consistent with the previous report that type I IFNs can negatively regulate IL-12 expression [38] . It is interesting to note that IFN-γ can also negatively regulate the production of these inflammatory cytokines as well as DC expansion from BM precursors ( Figures 2 and 4 ) . The potent inhibition of DC expansion and function by IFN-γ was unexpected , as this cytokine is essential for strong anti-viral immunity [7 , 8] . As far as we know , this novel inhibitory mechanism of IFN-γ on DC function and differentiation has not been previously described . Notably , the signal pathways involved in the negative regulation by type I and II IFNs may be different since IFN-γ , but not IFN-α/β , displays enhanced IL-10 induction in DCs ( Figure 2C ) . It has been previously reported that IFN-γ can play a negative regulatory role on DC migration and Ag-specific T cell priming in vivo [39] . Since IFN-γ is the major cytokine produced by NK/NKT cells during early innate responses and by CD4+ and CD8+ T cells during adaptive immune responses following viral infection , IFN-γ may play an important regulatory role by switching DCs on and off for further T cell activation . However , many studies have also indicated that type I and II IFNs play an important role in enhancing DC function in adaptive T cell responses [38] . Therefore , it appears paradoxical that type I and II IFNs can confer DCs with viral resistance ( Figure 3 ) , yet these cytokines can also inhibit DC differentiation and function ( Figures 2 and 4 ) , which are critical for strong adaptive T cell responses . Our results here strongly suggest that the level of stimulation , duration of stimulation , and timing of exposure to these cytokines following viral infection affect the fate of DC expansion and function . TMEV infection preferentially abrogates not only maturation and cytokine production by DCs infected with the virus ( producing viral proteins ) , but also bystander DCs ( not producing viral proteins ) from susceptible SJL mice in culture ( Figures 1 and 3 ) . This bystander inhibition is most likely mediated by type I and II IFNs ( Figures 2 and 4 ) . Similar deficiencies in DC function are likewise observed in both CNS-associated and splenic DCs from virus-infected susceptible SJL mice ( Figure 5 ) . In addition , the T cell stimulatory function of either virus-infected BM DCs or splenic DCs from virus-infected mice is severely compromised , particularly in susceptible SJL mice ( Figure 6 ) . Two recent studies using lymphocytic choriomeningitis virus and measles virus have demonstrated that these viral infections prohibit the expansion of DCs in the spleens of mice treated with Fms-like tyrosine kinase 3 ligand [23 , 40] . Therefore , it appears that inhibition of DC expansion or function may be a common tactic used by various viruses to evade anti-viral T cell responses for the establishment of chronic viral persistence . It is also noteworthy that significantly higher levels of IL-10 are induced by LPS stimulation in virus-infected SJL DCs than in B6 DCs ( Figure 3 ) . IL-10 production is not inhibited after TMEV infection , unlike that of other cytokines , such as IL-12 or TNF-α ( Figure 3 ) . High levels of IL-10 may favor Th2 development by TMEV-infected SJL DCs , in contrast to the Th1 development promoted by uninfected DCs or virus-infected B6 DCs ( Figure 6 ) . Interestingly , levels of IFN-γ-producing TMEV-specific CD4+ T cells are significantly lower , and levels of IL-10-producing CD4+ T cells are significantly higher in SJL mice compared to those in B6 mice [12] . Thus , higher IL-10 levels in susceptible SJL mice may promote viral persistence and preferentially elicit regulatory T cells , which further suppress anti-viral responses . Consistent with this notion , HIV-1-infected monocyte-derived DCs do not undergo maturation , but can elicit IL-10 production and T cell regulation [41] . Furthermore , IL-10 may also directly induce immunosuppression , facilitating viral persistence in vivo as recently shown in the lymphocytic choriomeningitis virus model [42 , 43] . Alternatively , it is possible that TMEV-induced type I IFNs in DCs exert a regulatory role in Th1/2 development; one study showed that exposure to IFN-β during DC-mediated stimulation of naïve Th cells inhibits Th1 cell polarization and promotes the generation of an IL-10-secreting T cell subset [44] . Therefore , it appears that the alteration of IL-10 production is an immune evasion mechanism frequently used by many different pathogenic viruses . Susceptibility to this alteration may also be a critical factor for chronic viral persistence leading to various inflammatory diseases . In summary , we have demonstrated that DCs from susceptible SJL mice are considerably more permissive to initial viral infection and replication compared to DCs from resistant B6 mice . In addition , TMEV infection preferentially inhibits maturation , activation , and T cell–stimulating function of DCs from susceptible SJL mice in a bystander fashion . The impaired function and reduced number of DCs are likely to add to the poor anti-viral T responses observed in virus-infected susceptible mice , contributing to establishment of persistent infection . It is quite striking to note that such deficiencies and/or abnormalities of DC function are not apparent after TMEV infection in resistant B6 mice . Furthermore , the adoptive transfer of DCs , particularly pre-activated DCs resistant to viral infection , results in the effective clearance of viral persistence and protection from the development of demyelinating disease in susceptible SJL mice ( Figure 7 ) . Therefore , the differential responses of DCs to infectious agents may determine the outcome of diseases associated with chronic infections . Our results strongly suggest that increasing DC populations in the target organs and arming them against infections may provide a useful intervention for acute or chronic infection-associated diseases .
Female SJL , B6 , and BALB/c mice were purchased from Harlan Sprague Dawley ( http://www . harlan . com/ ) . IFN-α/β receptor knockout mice ( IFN-α/βR ) were kindly provided by Herbert ( Skip ) Virgin ( Washington University , St . Louis , Missouri , United States ) , and control 129S2/SvPas mice ( IFN-α/βR ) were purchased from Charles River Laboratories ( http://www . criver . com/ ) . Mice ( 6–8 weeks old ) were maintained and used according to protocols approved by the Northwestern University Animal Care and Use Committee . The reagents used include IFN-α , IFN-β ( both Invitrogen , http://www . invitrogen . com/ ) , and IFN-γ , IL-6 ( both PeproTech , http://www . peprotech . com/ ) . The BeAn strain of TMEV was propagated and titered in baby hamster kidney ( BHK ) cells grown in DMEM supplemented with 7 . 5% donor calf serum , as previously described [11] . TMEV or virus-infected DC supernatant was inactivated by exposure to a UV light source for 60 min at 4 °C [45] . The complete inactivation of virus after irradiation was confirmed by the absence of viral plaque formation on BHK-21 cells . Mice were infected by intracerebral inoculation of 30 μl ( 1 × 106 pfu ) of the BeAn strain of TMEV , and clinical symptoms of disease were assessed weekly on the following grading scale as previously described [46]: grade 0 = no clinical signs; grade 1 = mild waddling gait; grade 2 = moderate waddling gait and hind limb paresis; grade 3 = severe hind limb paralysis; grade 4 = severe hindlimb paralysis and loss of righting reflex . BM cells were harvested from femurs and tibias of SJL or B6 mice and cultured in RPMI-10 with 20 ng/ml murine rGM-CSF ( PeproTech ) as described [47] . On day 5 , predominantly immature DCs were obtained . For DC expansion experiments , BMCs cultured for 3 d were collected and used as indicated . Mice were perfused with 30 ml HBSS . Brains and spinal cords were forced through steel mesh and digested at 37 °C for 45 min in HBSS containing 250 μg/ml collagenase type 4 ( Worthington Biochemical , http://www . worthington-biochem . com/ ) . CNS infiltrated leukocytes were then enriched on a continuous Percoll ( Amersham Biosciences , http://www . gelifesciences . com/ ) gradient as described [11] . For the binding assay , BM DCs on day 5 of the culture were incubated with TMEV ( MOI 10 ) for 1 h at 4 °C , fixed with 1% PFA , and stained with monoclonal antibody ( mAb ) ( 8C ) to the TMEV VP2 capsid protein followed by FITC-conjugated goat anti-mouse Ig ( Invitrogen ) [6] . For TMEV infection , BMCs were incubated with TMEV ( MOI 10 ) for 1 h at 20 °C with intermittent shaking in RPMI with 0 . 1% bovine serum albumin ( infection medium ) . Cells were washed twice and then resuspended in complete RPMI-10 medium containing 10% FCS , 20 ng/ml GM-CSF , and 5 × 105 cells per well in a 24-well plate . For the infection assay , monensin ( 0 . 7 μl/ml , GolgiStop; BD PharMingen , http://www . bdbiosciences . com/ ) was added for the final 4 h of infection cultures . Cells were fixed , permeabilized ( perm/fix solution , BD PharMingen ) , blocked with 1% normal goat serum ( Invitrogen ) , and stained with 8C mAb followed by FITC-conjugated secondary antibody or PE-conjugated goat anti-mouse Ig ( BD PharMingen ) . Cells were subsequently stained with anti-CD11c-APC ( BD PharMingen ) . Labeled cells were analyzed on a FACSCalibur flow cytometer . For the maturation-associated marker assay , DCs on day 5 of the culture were infected for 24 h . Cells were stained as described in the infection assay and were further labeled with anti-rat RT1B-FITC ( which cross-reacts with mouse I-Ak and I-As alloantigens ) , anti-I-Ab-FITC , anti-CD86-FITC ( all BD PharMingen ) , and anti-CCR7-PE ( eBioscience , http://www . ebioscience . com/ ) . In order to determine the effect of LPS on TMEV replication in DCs , 100 ng/ml LPS ( Escherichia coli 0111:B4 ) ( Sigma-Aldrich , http://www . sigmaaldrich . com/ ) were added to DCs on day 5 of the cultures for the indicated time periods . To determine the anti-viral effect of LPS or cytokines , DCs were pretreated for 6 h with these reagents , washed twice , and then infected with TMEV for 24 h . IL-12p40 , IL-6 , TNF-α , and IL-10 in supernatants of DC cultures at 24 h after infection were measured using BD OptEIA ELISA Set , and IFN-α was determined using an ELISA kit ( PBL Biomedical Laboratories , http://www . interferonsource . com/ ) . Culture supernatants of DCs infected with TMEV for varying time periods were collected for a standard plaque assay on BHK-21 monolayers [11] . In order to analyze viral persistence in vivo , brains and spinal cords were removed from virus-infected mice at 7 or 21 dpi after cardiac perfusion with HBSS . Tissue homogenate was used to perform a standard plaque assay . SJL DCs were pretreated with IFN-α , IFN-β , IFN-γ , or IL-6 for 18 h and then stimulated with PBS or 100 ng/ml LPS for 6 h . Supernatants were collected to determine cytokine production and cells were stained for intracellular cytokine production . CNS-infiltrated leukocytes at 5 dpi and splenocytes at 7 dpi were cultured for 6 h with 20 ng/ml GM-CSF in the absence or presence of LPS . Monensin was added for the final 4 h of culture . DCs were fixed , permeabilized , and then stained with anti-CD11c-APC and anti-IL-12p40/p70-PE or anti-TNF-α-PE ( both BD PharMingen ) . To analyze cytokine production by splenic DCs , DCs were positively selected by using anti-CD11c MicroBead ( Miltenyi Biotec , http://www . miltenyibiotec . com/ ) from the spleens of virus-infected mice at 7 dpi . Cultures were further incubated for 24 h in medium containing 20 ng/ml GM-CSF , with or without LPS ( 100 ng/ml ) during the last 6 h of culture . Supernatants were collected to determine IL-12p40 levels . For the cytokine production assay from T cell cultures , supernatants were collected 96 h later . IFN-γ and IL-4 were measured using BD OptEIA ELISA Set . Total RNA was isolated from SJL DCs or B6 DCs infected with or without TMEV for 8 h , using Trizol ( GIBCO BRL , http://www . invitrogen . com/ ) by the manufacturer's instructions . Reverse transcription was performed using 1–3 μg of total RNA . The sense and antisense primer sequences used are as follows: IFN-α1 ( 5-TAGGCTCTGTGCTTTCCTGATGGT-3 and 5-GGTTGCATTCCAAGCAGCAGAT GA-3 ) , IFN-α7 ( 5-AAGGACTCATCTGCTGCTTGGGAT-3 and 5-ACTCAATCTTGCCAGCAACT TGGC-3 ) , IFN-α11 ( 5-TCAAACACACAGTCCAGAGAGCCA-3 and 5-ATCCAAAGTCCTGC CTGTCCTTCA-3 ) , IFN-β ( 5-GCTCGAGCCCCAGGGAATGAACAACAGG TGG-3 and 5-GCTCGAGTCCCTGGGGGTTTTGGAAGTTTCT-3 ) , IFN-γ ( 5-ACTGGCAA AAGGAT GGTGAC-3 and 5-TGAGCTCATTGAATGCTTGG-3 ) , and GADPH ( 5-AACTTTGGC ATTGTGGAAGG-3 and 5-ACACATTGGGGGTAGGAACA-3 ) . Specific RNA messages were amplified in iCycler SYBR green I mastermix ( Bio-Rad , http://www . bio-rad . com/ ) , using an iCycler ( Bio-Rad ) . Quantification of the genes of interest was normalized to GAPDH in each cDNA and was expressed by fold increase compared with mock-infected DCs . All real-time PCR was performed in triplicate . For the BM DC apoptosis assay , DCs obtained from a 3-d BM culture were infected with TMEV for 24 h and then either stained with FITC-conjugated Annexin V ( Invitrogen ) or labeled with Annexin V MicroBead ( Miltenyi Biotec ) for cell separation . The Annexin V–positive and –negative cells were separated , fixed , permeabilized , and stained with anti-TMEV mAb and anti-CD11c antibody . BMCs on day 3 of the culture were labeled with CFSE ( Molecular Probes , http://probes . invitrogen . com/ ) and then infected with TMEV . After 3 d of incubation , cells were stained with anti-CD11c-APC and the level of proliferation was assessed using FACS . Primary allogeneic MLR was performed with mock-infected or virus-infected BM DCs at 3 dpi that were unfixed or fixed with 0 . 1% PFA as described [48] . DCs from the spleens of mock-infected or TMEV-infected mice were positively selected by using anti-CD11c MicroBead ( Miltenyi Biotec ) at 7 dpi . Responder T cells from naïve BALB/c mice were enriched from spleens by depleting adherent cells at 37 °C for 2 h and then negatively selected by using anti-CD19 MicroBead ( Miltenyi Biotec ) to remove B cells . Varying numbers of DCs were cocultured with purified T cells ( 105/well ) in 200 μl of RPMI-10 . The cultures were then incubated for 96 h , followed by an 18-h pulse with 1 μCi of [3H]-thymidine/well ( Amersham Biosciences ) . For the T cell activation assay , cells were collected 96 h later and stained with anti-CD4-APC , anti-CD8-FITC , and anti-CD69-PE ( all BD PharMingen ) or anti-TIM-3-PE ( eBioscience ) . SJL DCs cultured for 5 d were stimulated with or without 100 ng/ml LPS for 6 h ( LPS-DCs ) and then washed twice . The cell pellets were then resuspended in DMEM and incubated with TMEV ( MOI 10 ) for 30 min at 4 °C . Non-DC splenocytes depleted of DCs by anti-CD11c MicroBead ( Miltenyi Biotec ) were used as an irrelevant cell type control . Mice were infected by intracerebral inoculation of 30 μl containing 1 × 105 cells and/or 1 × 106 pfu TMEV . In addition , mice were injected with LPS-DCs alone as an additional control group . Spinal cords were removed at 50 dpi after perfusion as described above , fixed in 10% buffered PFA overnight , and then embedded in paraffin . Sections were stained with luxol fast blue ( myelin stain ) and hematoxylin-eosin , and then analyzed by microscopy . Ten different sections of the lumbar region of the spinal cord of individual mice were evaluated for demyelination . A representative set of micrographs is shown . Results are expressed as mean ± SD , where applicable . The Student's t test was used to compare two independent groups . The differences in mean severity scores of the two experimental groups were compared between 21 and 98 dpi by using a paired Student's t test . Multiple group comparisons were done by a one-way analysis of variance ( ANOVA ) with Tukey post-hoc analysis . p-Values < 0 . 05 were considered to be significant .
|
Many chronic viral diseases are associated with prolonged viral persistence levels , which vary from one individual to another . However , the mechanisms of differential susceptibility to persistent viral infections are unknown . Theiler murine encephalomyelitis virus ( TMEV ) induces a chronic demyelinating disease similar to multiple sclerosis . In this study , we investigated the potential mechanisms of differential susceptibility to chronic viral persistence in the central nervous system following infection with TMEV . Our results indicate that differential interactions between virus and dendritic cells ( DCs ) , leading to the induction of anti-viral immunity , are critical in determining resistance or susceptibility to virus-induced chronic demyelinating disease . DCs from susceptible mice are much more permissive to viral infection , resulting in severe deficiencies in their development , expansion , and function , whereas DCs from resistant mice are not permissive . Consequently , the DCs in susceptible mice are responsible for poor anti-viral T cell activation , permitting viral persistence and disease development in the host . Interestingly , the administration of additional DCs or pre-activated virus-resistant DCs enables susceptible mice to resist persistent viral infection and disease development . This knowledge may be useful in devising effective means to induce strong anti-viral immune responses , thereby protecting the host from virus-associated chronic diseases caused by persistent viral infection .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"viruses",
"virology",
"in",
"vitro",
"immunology",
"mus",
"(mouse)"
] |
2007
|
Role of Dendritic Cells in Differential Susceptibility to Viral Demyelinating Disease
|
Autosomal recessive nonsyndromic hearing loss is a genetically heterogeneous disorder . Here , we report a severe-to-profound sensorineural hearing loss locus , DFNB100 on chromosome 5q13 . 2-q23 . 2 . Exome enrichment followed by massive parallel sequencing revealed a c . 2510G>A transition variant in PPIP5K2 that segregated with DFNB100-associated hearing loss in two large apparently unrelated Pakistani families . PPIP5Ks enzymes interconvert 5-IP7 and IP8 , two key members of the inositol pyrophosphate ( PP-IP ) cell-signaling family . Their actions at the interface of cell signaling and bioenergetic homeostasis can impact many biological processes . The c . 2510G>A transition variant is predicted to substitute a highly invariant arginine residue with histidine ( p . Arg837His ) in the phosphatase domain of PPIP5K2 . Biochemical studies revealed that the p . Arg837His variant reduces the phosphatase activity of PPIP5K2 and elevates its kinase activity . We found that in mouse inner ear , PPIP5K2 is expressed in the cochlear and vestibular sensory hair cells , supporting cells and spiral ganglion neurons . Mice homozygous for a targeted deletion of the Ppip5k2 phosphatase domain exhibit degeneration of cochlear outer hair cells and elevated hearing thresholds . Our demonstration that PPIP5K2 has a role in hearing in humans indicates that PP-IP signaling is important to hair cell maintenance and function within inner ear .
Hearing loss ( HL ) is a heterogeneous neurosensory deficiency that occurs at all ages with varying severities , affecting 1 in 500 newborns and >360 million people worldwide [1] . Many forms of HL are inherited , and ~400 syndromic forms occur with linked medical comorbidities . Genetically complex non-syndromic recessively inherited HL ( NSRHL ) comprises ~75% of hereditary deafness [2] . RNA profiling studies revealed expression of 18 , 133 genes in the organ of Corti ( OC ) hair and supporting cells [3] , underscoring the complexity of inner ear development and function . Elucidating the roles of each of these genes by conventional methods would require generating mouse or other animal models for each . By contrast , genetic studies of human families segregating HL has significantly helped in overcoming the barriers of sheer scale , functional redundancy , and non-essential roles of some genes . Discovering a variant associated with human deafness rules out functional redundancy at least in the auditory system . Genetic and functional studies of the deafness associated proteins have been pivotal in elucidating various molecular networks essential for hearing [4–6] . Here , we describe two large consanguineous Pakistani families segregating NSRHL that we associate with a missense variant [p . ( Arg837His ) ] in the PPIP5K2 ( E . C . 2 . 7 . 1 . 155 ) . PPIP5Ks are enzymes that interconvert 5-IP7 and IP8 , two key members of the inositol pyrophosphate ( PP-IP ) cell-signaling family [7] . Humans express two PPIP5K enzymes , PPIP5K1 ( 160 kDa ) and PPIP5K2 ( 138 kDa ) [8] . These are large enzymes that contain a kinase domain that phosphorylates 5-IP7 to IP8 , and a separate phosphatase domain that dephosphorylates IP8 back to IP7 . Little is known concerning how these competing reactions are coordinated . Nevertheless , PPIP5Ks regulate the levels of PP-IPs , which impact endocytosis , vesicle trafficking , apoptosis , spermatogenesis , secretion of insulin from pancreatic β cells , and DNA repair [9–11] . IP8 is also a sensor of extracellular inorganic phosphate [8] and regulates bioenergetic homeostasis [12] . Our in vitro biochemical studies indicate that the p . Arg837His deafness-associated variant of PPIP5K2 has reduced phosphatase activity and increased kinase activity . As a consequence , the variant has the capacity to synthesize more of the IP8 signal in vivo . An elevated cellular production of IP8 can be recapitulated in mice harboring a homozygous deletion of the phosphatase domain of PPIP5K2 , thereby unmasking higher kinase activity . These animals exhibited degeneration of cochlear outer hair cells and progressive hearing loss . To the best of our knowledge , this is the first study demonstrating the necessity of PP-IP metabolism for inner ear development and hearing function .
Two large apparently unrelated families PKDF041 and PKDF751 segregating NSRHL were enrolled from Punjab province of Pakistan ( Fig 1A and 1B ) . Affected individuals of both families have prelingual bilateral severe to profound sensorineural hearing loss ( Fig 1C and 1D ) . According to family history no hearing was noted since birth in all of the affected individuals . Tympanometry of three of the affected [V:2 ( 19 yrs ) , V:3 ( 23 yrs ) , V:5 ( 26 yrs ) ] , and one normal hearing individual ( V:7; 22 yrs ) of family PKDF751 revealed no abnormalities of the tympanic membrane or middle ear . However , these affected individuals ( V:2 , V:3 , V:5 ) failed a transient evoked otoacoustic emission test , which is indicative of defective outer hair cell function . Romberg and tandem gait tests did not reveal any overt vestibular dysfunction in affected individuals of both families . Funduscopy examinations of affected individuals of PKDF041 , V:5 ( 42 years ) , VI:3 ( 16 years ) , VI:4 ( 10 years ) and VI:5 ( 20 years ) revealed no evidence of retinitis pigmentosa . In addition , ERG a- and b-waves amplitudes for the affected individuals [V:5 ( 42 yrs ) , V:7 ( 30 yrs ) ] of PKDF041 , were normal [13] , further excluding a retinal degeneration phenotype . Linkage analysis of NSRHL segregating in family PKDF041 with STR markers across the genome , previously revealed linkage to DFNB49 locus on chromosome 5 , with a maximum two-point lod score ( Zmax ) of 4 . 44 ( θ = 0 ) for marker D5S2055 [13] . Subsequently , mutations in TRIC were identified as the cause of DFNB49-linked HL [14 , 15] . Full sequencing of TRIC in the genomic DNA from two affected individuals from the PKDF041 family along with a normal hearing sibling did not reveal any pathogenic variants , suggesting there is an additional gene in the PKDF041 linkage interval in which a pathogenic variant is associated with deafness . Subsequently , genome wide linkage analysis in DNA samples from a second family , PKDF751 , revealed a large region of homozygosity shared among the affected individuals on human chromosome 5q12 . 2-q23 . 3 , which completely overlaps with the linkage interval defined by family PKDF041 ( S1 Fig ) . Similar to family PKDF041 , mutation in the protein coding exons , non-coding exons or in the splice junctions of TRIC were not detected in the affected individuals of family PKDF751 . Therefore , the HUGO nomenclature committee assigned the DFNB100 designation for the locus defined by families PKDF041 and PKDF751 ( S1 Fig ) . Besides TRIC , other candidate genes include OCLN , SHROOM1 , GPR98 , KCNN2 and SLC12A2 ( S1 Fig ) [16–21] . However , sequencing of all these genes did not reveal a potentially pathogenic variant among the affected individuals of families PKDF041 and PKDF751 . Rather than continue hierarchical sequencing of candidate genes based on function or expression in the inner ear , we employed an exome sequencing approach to identify a mutant variant responsible for deafness at the DFNB100 locus . To enrich and capture coding regions , we used genomic DNA samples from one affected individual and one unaffected control from family PKDF751 , and performed massively parallel sequencing of the exome . Copy number variants ( CNVs ) analysis did not reveal any indels ( ≥50bp ) associated with deafness within DFNB100 linkage interval . Our filtering of the exome sequencing data revealed three genes with predicted pathogenic changes within DFNB100 linkage interval ( S1 Table ) . Sanger sequencing revealed a single point mutation ( c . 2510G>A ) in the PPIP5K2 gene ( NCBI RefSeq NM_001276277 ) segregating with HL in family PKDF751 ( Fig 2A ) , which is predicted to replace an evolutionarily-conserved arginine with histidine at residue 837 ( NP_001263206; Fig 2B , S2 Table ) . Sanger sequencing of all coding and non-coding exons of PPIP5K2 in the DNA samples from family PKDF041 confirmed association of the same c . 2510G>A variant with deafness . SNPs linked to PPIP5K2 were genotyped in affected individuals of the PKDF041 and PKDF751 families , and the flanking haplotype was consistent with a founder effect for c . 2510G>A variant ( S3 Table ) . We did not detect this variant in 180 ethnically-matched Pakistani control samples , nor in small publicly-available databases ( 1000 Genome [22] , or NHLBI_EP [23] ) . At present , the c . 2510G>A variant of PPIP5K2 is at a low frequency ( 0 . 000146 ) in the ExAC database [24] , and thus c . 2510G>A is not a common polymorphism . Human PPIP5K2 encodes at least eleven alternatively spliced isoforms , all of which are affected by the p . ( Arg837His ) substitution identified in the two DFNB100 HL families ( Fig 2B ) . PPIP5K2 has three distinct functional regions: a kinase domain , a phosphatase domain and an intrinsically disordered region that may mediate protein-protein interactions ( Fig 2B ) . The Arg837His substitution is located in the phosphatase domain ( Fig 2C ) . Human PPIP5Ks is a mutually competitive kinase and phosphatase that interconverts PP-IPs ( Fig 2C ) . In silico analyses predict that the p . Arg837His residue is likely of structural and/or functional importance ( S2 Table ) . Indeed , Arg837 is conserved throughout the animal kingdom ( Fig 2D ) , and lies nine residues C-terminal to a conserved histidine residue that is catalytically-essential for IP8 phosphatase activity in a yeast orthologue of PPIP5K2 [25] . Thus , p . Arg837 is a candidate for binding the negatively charged PP-IP substrate . The substitution of histidine would be less effective in this role as its side chain is less polar at physiological pH [26] . Another possibility is that p . Arg837 acts in a structurally-stabilizing salt bridge , which could be disrupted by substitution with His [27] . To pursue these ideas , we first assayed directly the impact of the p . Arg837His variant upon the phosphatase activity of PPIP5K2 in vitro , using recombinant , FLAG-tagged WT and PPIP5K2R837H which were expressed in HEK293 cells and then immuno-affinity purified ( Fig 2E ) . The IP8 phosphatase activity of the PPIP5K2R837H protein was reduced by 21% ( *p< 0 . 05 ) as compared to the WT protein ( Fig 2E ) . PPIP5K2 interconverts 5-IP7 to IP8 through the actions of mutually competing phosphatase and kinase domains . Thus , we hypothesized based upon the in vitro kinase assays of 5-IP7 phosphorylation that PPIP5K2R837H protein would accumulate more IP8 than WT PPIP5K2 . That proposal presumes that a potentially higher level of accumulation of IP8 by the PPIP5K2R837H protein is not negated by an increased rate of IP8 dephosphorylation . However , that self-correction did not occur . Assays that used PPIP5K2R837H accumulated 60% higher levels of IP8 compared to assays that contained WT protein ( Fig 2F ) . The proportionately higher effect of the p . Arg837His variant upon the kinase activity may reflect an impact upon conformational coupling between the two domains [8] , in addition to the reduction in phosphatase activity . Note that , in vitro , the phosphatase and kinase domains of PPIP5K2 can also be shown to interconvert IP6 and 1-IP7 [12] . However , these particular reactions proceed relatively slowly [12]; their general biological significance is in doubt , especially as levels of 1-IP7 in mammalian cells are virtually undetectable [11] . Our analysis by real-time PCR showed Ppip5k2 and Ppip5k1 to be widely expressed in mouse tissues , including inner ear ( Fig 3A , S3A Fig ) . Transcriptome analysis indicates that Ppip5k2 is the major Ppip5k gene to be expressed in the mouse cochlear and vestibular sensory hair cells , supporting cells as well as in ganglion neurons , while levels of Ppip5k1 expression in these cells are up to 100-fold lower ( gEAR , SHIELD ) [28] . Similarly , PPIP5K2 immunoreactivity , assessed using an antibody directed against a region within the phosphatase domain ( Fig 2B ) , was observed in all these cell types at various developmental stages ( Fig 3 ) . Moreover , all three layers of the stria vascularis ( marginal , intermediate and basal cells ) exhibited PPIP5K2 immunoreactivity ( Fig 3B and 3C ) . The distribution of PPIP5K1 immunoreactivity within the inner ear , assessed using isoform-specific antibodies ( S2 Fig ) , exhibited an almost identical expression profile to that of PPIP5K2 ( S3B–S3E Fig ) . To determine the role of PPIP5K2 in the inner ear , we obtained a mouse from the Knockout Mouse Phenotyping Program ( KOMP ) that has a gene trap cassette with a LacZ reporter in the intron between exons 13 and 14 of Ppip5k2 ( Fig 4A ) . The gene trap leads to the translation of a protein in which the N-terminal kinase domain is intact , but the adjoining phosphatase domain is truncated to just 84 residues ( Fig 4A ) , which is insufficient to encode phosphatase activity [29 , 30] . To study the impact upon the enzymatic properties of the protein product of the targeted allele , we prepared a recombinant , C-terminally truncated human PPIP5K2 ( PPIP5K21-466 ) . This is equivalent to the murine protein that is expected to be expressed by the Ppip5k2K^ allele . The shorter human protein converted 5-IP7 to IP8 at a 23-fold higher rate than that catalyzed by the kinase domain of the full-length human PPIP5K2 ( Fig 4B ) . We therefore named the murine allele as Ppip5k2K^ . It is impractical to assay the impact of the Ppip5k2K^ allele upon PP-IPs in vivo , due to PP-IPs sub- to low-micromolar levels , and the relative insensitivity of mass assays for these molecules . Instead , we used [3H]-inositol radiolabeled PPIP5K-/- HEK293 cells [12] , as a host for over-expression of either WT human PPIP5K2 or the truncated PPIP5K21-466 mutant . The latter supported the synthesis of 4-fold ( n = 3 ) higher cellular IP8 levels than those sustained by the WT protein ( Fig 4C ) , despite the expression level of truncated PPIP5K21-466 being much lower than that of WT PPIP5K2 ( S4 Fig ) . Based on these findings , we conclude that the Ppip5k2 mutant mice likely expresses a “hyper kinase” activity . Note that the changes in cellular levels of IP7 and IP8 resulting from over-expression of PPIP5K2R837H in PPIP5K-/- HEK293 cells [12] were similar to those observed after over-expression of WT PPIP5K2 ( Fig 4C ) . The catalytic differences between these particular enzymes are less prominent ( Fig 2E and 2F ) , making it harder to detect their differential impact upon PP-IP turnover in intact cells . PP-IP levels will also be influenced by cell-type dependent differences in expression of other PP-IP kinases and phosphatases . In Ppip5k2K^/K^ mice , we observed a significant upregulation of Ppip5k1 message in several tissues at P150 , including a nearly 3-fold elevation in the inner ear ( S3F Fig ) . Any increased expression of Ppip5k1 can be predicted to further enhance the synthesis of IP8 [12] . However , the metabolic consequences of this effect are likely minor in the inner ear , where relative levels of Ppip5k1 are 100-fold less than those of Ppip5k2 ( see above ) . To characterize inner ear function , we measured auditory-evoked brainstem responses ( ABR ) in Ppip5k2K^/ K^ and WT mice at various developmental stages ( S5 Fig ) . No statistically significant difference in hearing thresholds was observed in Ppip5k2K^/ K^ as compare to WT at postnatal day 60 ( P60 ) , when tested for broad band clicks as well as pure tone burst of frequencies 8 , 16 , and 24 kHz ( Fig 4D ) . However , Ppip5k2K^/K^ mice had slightly elevated , but not statistically significant , thresholds at 32 kHz compared to WT mice ( Fig 4D , top left ) . The difference progressively increased with age . At P90 , the threshold at 32 kHz of Ppip5k2K^/K^ mice were statistically significantly ( p<0 . 05 ) elevated as compare to WT ( Fig 4D , top right ) . At P120 and P150 , significantly ( p<0 . 001 ) higher threshold was observed at 24 kHz as well in Ppip5k2K^/K^ ( Fig 4D , bottom ) , which suggest progression of hearing loss from base towards middle cochlear turn . The Ppip5k2+/K^ mice at P150 also exhibited significantly higher threshold ( p < 0 . 05 ) at 24 kHz , which suggest that reduced amount of WT protein is also not sufficient to maintain hearing function in older mice . As anticipated , considering the C57BL/6 genetic background , which is susceptible to age-related HL [31 , 32] , at P120 and P150 , the thresholds for 32 kHz were observed to be elevated in both genotypes . We also compared the ABR wave I amplitudes and latencies in control and Ppip5k2 K^ mutant mice ( Fig 5 ) . Significant reduction in wave I amplitudes at higher frequency ( 24kHz ) was observed in older homozygous mutant mice . Overall , ABR auditory data from mice are consistent with the conclusion that elevated PPIP5K2 kinase activity is a causal factor for inherited auditory defects , although the deficit in the Ppip5k2K^/K^ mice is less severe than that observed in humans homozygous for the p . ( Arg837His ) substitution . To determine whether the auditory deficit in Ppip5k2 K^/K^ mice was secondary to hair cell loss , spiral ganglion degeneration or stria vascularis atrophy , we analyzed whole mounts preparations as well as serial sections of the cochlear tissue at P150 ( Fig 6 ) . The cytoarchitecture and morphology of the cochlea in Ppip5k2 K^/K^ mice appear normal at P150 . No obvious degeneration of inner ( IHCs ) and outer ( OHCs ) hair cells was observed in the apical and middle cochlear turn ( Fig 6 ) . However , consistent with high frequency hearing loss observed at P150 , a greater degree of degeneration of OHCs in the basal coil was evident in Ppip5k2+/K^ and Ppip5k2K^/K^ mice ( Fig 6 ) . Although no difference was observed in the number of IHCs , a statistically significant ( p<0 . 01 ) reduction in OHCs was observed in the basal turn ( Fig 5B ) . No obvious indications of stria vascularis atrophy or spiral ganglion neuron degeneration were found in Ppip5k2K^/ K^ mice at P150 ( S6 Fig ) . Thus , the degeneration of OHCs is likely contributing to the progressive , high frequency hearing deficit in these Ppip5k2 K^/K^ mice . Overall , our data indicate that genetic perturbation of PP-IP turnover has consequences for cellular signaling that promotes an auditory deficit , in both humans and mice .
This study points to the importance of PP-IP signaling and the role of the bifunctional PP-IP kinase/phosphatase , PPIP5K2 , for normal mammalian inner ear function . This conclusion is based upon a novel single variant [c . 2510G>A , p . ( Arg837His ) ] in the PPIP5K2 gene that we associated with autosomal recessive , nonsyndromic , prelingual sensorineural deafness . Biochemical analyses revealed that , compared to WT human PPIP5K2 , the PPIP5K2R837H variant exhibited lower phosphatase activity and higher kinase activity , indicating it promotes increased metabolic flux from 5-IP7 to IP8 in vivo . The availability of a gene-trap murine model for elevated PPIP5K2 kinase activity is consistent with the association of p . ( Arg837His ) with an auditory deficit . Humans who are homozygous for the c . 2510G>A variant failed transient evoked otoacoustic emission tests , indicating an outer hair cell functional defect . Thus , it is significant that , in the Ppip5k2K^/K^ mice , we found that hearing loss is associated with selective degeneration of cochlear outer hair cells . How might the latter phenomenon be promoted by elevated IP8 levels ? One possibility is that hair cell loss is due to apoptosis , since previous work has shown that increased PP-IP synthesis is pro-apoptotic [33] . Another possibility emerges from recent work showing that loss of outer hair cells can result from bioenergetic imbalance [34]; the latter could also result from perturbation to PP-IP turnover and signaling [11 , 12] . Deficits in vesicle trafficking and endocytosis might affect the function of the inner hair cell ribbon synapse [35 , 36] , and thus contribute to hearing loss observed in the Ppip5k2K^ mice . It is intriguing that the hearing deficit in the Ppip5k2K^/K^ mice is less severe than that observed in humans homozygous for the p . ( Arg837His ) variant . Perhaps in mice there is greater functional redundancy of PPIP5K2 with other proteins . The particular susceptibility of humans to a single arginine-to-histidine substitution in PPIP5K2 may also reflect genetic or environmental factors . Elucidation of the causes of this dissimilarity may reveal molecular or cellular pathways for potential interventions to prevent HL in DFNB100 families . Future studies with mice in which the Ppip5k2 gene is targeted on a genetic background resistant to age-related HL , may also help in deciphering the roles of PP-IPs in development , maintenance and function of sensory cells in the inner ear . PPIP5Ks are pivotal enzymes for regulating PP-IP turnover . Their actions at the interface of cell signaling and bioenergetic homeostasis can impact many biological processes [11] . Yet surprisingly , as far as we are aware , our work provides the first description of any amino acid variant in either PPIP5K1 or PPIP5K2 that is both functionally-significant and associates with a human disorder . Two candidate single nucleotide polymorphisms , rs35671301 ( p . Ser419Ala ) and rs17155147 ( p . Thr1267Met ) , in human PPIP5K2 were reported to be enriched in individuals with autism spectrum disorder [37 , 38] . However , in those studies , there is no empirical or in silico evidence that either amino-acid replacement might alter affect protein function . Thus , PPIP5K2 is given new clinical significance by our observations . In conclusion , the identification of the PPIP5K2 variant as a risk factor for deafness raises the possibility that there may be other variants in its coding sequence that exhibit a similar hearing deficit . Moreover , our study reinforces the concept that studies into PP-IP turnover and signaling are relevant to human health and well-being .
This study was conducted under IRB-approved protocols , by IRB committees at University of Maryland School of Medicine ( UMSOM ) , USA ( HP-00059851 ) , National Centre of Excellence in Molecular Biology , Lahore , Pakistan ( FWA00001758 ) , and the National Institutes Health , USA ( Combined Neuroscience Blue Panel IRB; OH-93-N-016 ) , and in accord with the Declaration of Helsinki for the release of clinical information , family history , and blood draw . Written informed consent was obtained from all adult participants and parents of minor subjects . Genomic DNA was extracted from blood samples . Clinical histories were obtained for all the individuals of families PKDF041 and PKDF751 . A general physical examination was performed to evaluate general health . Pure tone audiometry tests for air and bone conduction were performed at frequencies from 250 to 8 , 000 Hz . Tympanometry and otoacoustic emission tests were performed to assess middle ear and cochlear function . Vestibular function was evaluated by testing tandem gait ability and by using the Romberg test . Exome sequencing on the genomic DNA of one affected individual of family PKDF751 was performed and analyzed as described previously [39] . We used the XHMM and CoNIFER methods [40 , 41] to call CNV events in WES data . The segregation of the candidate variants in participating member of family PKDF751 was performed through Sanger sequencing . Primers were designed with Primer3 software , and used to amplify exons as well as flanking introns and untranslated regions ( S4 Table ) . All experiments were approved by the Animal Care and Use Committees at the University of Maryland , School of Medicine in accordance with the National Institutes of Health ( NIH ) Guide for the Care and Use of Laboratory Animals . The B6N ( Cg ) -Ppip5k2tm1a ( EUCOMM ) Wtsi/J strain was generated by KOMP . In this study , we designated this strain as Ppip5k2K^ due to its intact kinase domain . The Ppip5k2K^ mice had a gene-trap cassette downstream of exon 13 , which is expected to cause premature truncation . Lipofectamine 3000 ( Life Technologies ) was used to transiently transfect HEK293 cells with pDest515 plasmids hosting cDNAs encoding Flag-tagged versions of either full-length PPIP5K2 ( isoform 11; accession number NM_001345875 ) , PPIP5K2 p . Arg837His ( PPIP5K2R837H ) or a truncated version of PPIP5K2 containing residues 1 to 466 ( PPIP5K21-466 ) . The mutations were created using a Q5 site-directed mutagenesis kit ( New England Biolabs ) . The primers used are as follows ( mutagenic codons underlined ) : R837H mutation , forward: GTCTATTCTTCACTATGGTGCCTTATG; reverse: AGCAAAGAATGTACATGAC . Truncation ( stop codon insertion ) , forward: TAGGTCATTTTTCTGGAATAAATCG; reverse: TTACTCTAATACAGTCTTAAGTTG . All mutations were confirmed by sequencing . In some experiments , the wild type ( WT ) and mutant PPIP5K2 constructs were expressed in PPIP5K-/- HEK293 cells that had been radiolabeled by growing them in media containing [3H]-inositol . Cellular levels of [3H]IP7 and [3H]IP8 were determined by HPLC [12] . For the purification of WT and mutant PPIP5K2 proteins , the HEK293 cells hosting the various constructs were harvested 16–20 hr after transfection and lysed in ice-cold buffer containing 10 mM HEPES , 130 mM NaCl , 1% Triton X-100 , 10 mM NaF , 10 mM Na2HPO4 , 10 mM Na pyrophosphate and protease inhibitor cocktail ( Roche ) . All subsequent steps were performed on ice in an anaerobic chamber ( Bactron CAT180 ) . The FLAG-tagged PPIP5K2 proteins were immunopurified using FLAG M2 affinity gel ( Sigma ) . Purified FLAG-PPIP5K2 proteins were analyzed by SDS-PAGE and stained with Coomassie Blue . Densitometry analysis ( ImageJ ) was used to calculate the amounts of PPIP5K2 loaded onto the gel , by reference to additional lanes loaded with known amounts of the human recombinant PPIP5K2 kinase domain . All enzyme assays were performed using physiologically relevant substrate concentrations at 37°C; the extent of substrate conversion was up to 30% . IP8 phosphatase activities were determined using approximately 70 ng of purified , WT or mutant PPIP5K2s in 30 min incubations with100 ul of assay buffer containing 1 mM Na2EDTA , 50 mM KCl , 20 mM HEPES pH 7 . 2 , 2 mM MgCl2 , 0 . 5 mg/ml BSA and 1 μM [3H]IP8 . The 5-IP7 kinase activities were determined using approximately 35 ng of these proteins in 90 min incubations with 100 ul of assay buffer containing 1 mM Na2EDTA , 50 mM KCl , 20 mM HEPES pH 7 . 2 , 7 mM MgCl2 , 5 mM ATP , 0 . 5 mg/ml BSA and 1 μM 5-[3H]IP7 . All assays were then acid-quenched , neutralized , and HPLC was used to determine the degree of metabolism [12] . Inner ear tissues were harvested from temporal bones of the WT and Ppip5k2K^ mice , fixed and processed for immunostaining as previously described [42] . The following primary antibodies were used in our studies: anti-Myosin VIIa ( 1:200; Proteus BioSciences , Ramona , CA ) , anti-PPIP5K2 ( 1:200; Stock# ab87054 , Abcam , Cambridge , MA ) and anti-PPIP5K1 ( 1:200; Novus Biologicals , Littleton , CO ) . Samples were mounted using ProLong Gold antifade reagent ( Life Technologies , Carlsbad , CA ) and imaged using an LSM 510 DUO confocal microscope ( Zeiss Microimaging Inc . , Thornwood , NY ) with a ×63 , 1 . 4 N . A . oil immersion objective . Briefly , mice were anesthetized with an IP injection of ketamine/xylazine mix at 80–100 mg/kg and 10–12 . 5 mg/kg , body weight respectively . ABR recordings were performed using a TDT RZ6/BioSigRZ system along with RA4PA preamplifier/digitizer . The system was calibrated between 4–40KHz frequency range from 90 to 5 db using the ACO Pacific type 7016 microphone . For simultaneous recording from both ears , two open filed speakers were placed 10 cm apart from each ear . Two electrodes were placed behind each ear . The reference was placed at the base of the skull , whereas the ground was placed at the base of the tail . Mice were tested at broad band clicks and pure tone frequencies of 8 , 16 , 24 , and 32 kHz for ABRs ( n = 7 per genotype at each time point ) . All the stimuli were tested from 90 dB SPL to10 dB SPL in 5 dB decrements , with a total of 512 responses averaged at each level . Results are displayed as a mean and standard error of the mean ( SEM ) . When no ABR signals were observed at 90db , for calculation purposes , mice were assigned a threshold of 95db . Statistical significance was determined using a 2-way ANOVA . A Bonferroni correction was performed to adjust for multiple statistical tests . The whole inner ear was isolated from P150 WT and Ppip5k2K^/K^ mice using the RiboPure RNA isolation kit ( Life Technologies , Grand Island , NY ) . cDNA was prepared using an oligo-dT primer and SMARTScribe Reverse Transcriptase enzymes ( Clontech , Mountain View , CA ) . To determine the differential expression of Ppip5k1 , SYBRGreen based real-time primers were designed using Integrated DNA Technologies online PrimeTime qPCR assay design tool . The real-time PCR assays were performed in triplicate using an ABI StepOnePlus Real-Time thermal cycler ( ABI , Foster City , CA ) . CT values were normalized using Gapdh as an endogenous control , and fold changes of Ppip5k1 transcripts in different tissues were calculated using 2^-ΔΔCt standard formula . An expression with a 2-fold change and with a P-value less than 0 . 05 based on a Student’s t-test analysis was considered significant . The URLs for data presented herein are as follows: NHLBI Exome Sequencing Project ( Exome Variant Server ) , http://evs . gs . washington . edu/EVS/ Primer3 , http://frodo . wi . mit . edu/primer3 PrimeTime qPCR assay design tool , http://www . idtdna . com/Scitools/Applications/RealTimePCR/ gEAR , http://umgear . org SHIELD ( Shared Harvard Inner-Ear Laboratory Database ) https://shield . hms . harvard . edu/
|
Exome sequencing coupled with homozygosity mapping was used to identify a missense variant [c . 2510G>A; p . ( Arg837His ) ] in PPIP5K2 at the DFNB100 locus that is associated with nonsyndromic , prelingual sensorineural deafness in two large consanguineous Pakistani families . PPIP5Ks are pivotal enzymes for regulating inositol pyrophosphate ( PP-IP ) turnover . Biochemical analyses revealed that , compared to wild type human PPIP5K2 , the PPIP5K2R837H variant exhibited lower phosphatase activity and higher kinase activity , indicating that it promotes increased metabolic flux from 5-IP7 to IP8 in vivo . In rodent inner ears , PPIP5K2 immunoreactivity was observed in the cochlear and vestibular hair cells , supporting cells , and spiral ganglion neurons . Mouse mutants homozygous for the targeted deletion of Ppip5k2 phosphatase domain exhibit degeneration of cochlear outer hair cells and progressive hearing loss . Our work provides the first description of any amino acid variant of PPIP5K2 that is both functionally-significant and associates with a human disorder . The ‘futile cycling’ of the kinase/phosphatase activity of PPIP5K2 makes inner ear function particularly susceptible to even minor changes in the phosphatase activity of PPIP5K2 . We have shown that a pathogenic variant in PPIP5K2 is associated with hearing loss in humans . Thus , PPIP5K2 is given new clinical significance by our observations .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"medicine",
"and",
"health",
"sciences",
"enzymes",
"ears",
"enzymology",
"neuroscience",
"phosphatases",
"animal",
"models",
"outer",
"hair",
"cells",
"otology",
"inner",
"ear",
"model",
"organisms",
"experimental",
"organism",
"systems",
"hearing",
"disorders",
"research",
"and",
"analysis",
"methods",
"animal",
"cells",
"proteins",
"biological",
"tissue",
"mouse",
"models",
"head",
"otorhinolaryngology",
"genetic",
"loci",
"biochemistry",
"cellular",
"neuroscience",
"deafness",
"anatomy",
"cell",
"biology",
"ganglia",
"neurons",
"genetics",
"biology",
"and",
"life",
"sciences",
"cellular",
"types",
"afferent",
"neurons"
] |
2018
|
Mutations in Diphosphoinositol-Pentakisphosphate Kinase PPIP5K2 are associated with hearing loss in human and mouse
|
Travelers who acquire dengue infection are often routes for virus transmission to other regions . Nevertheless , the interplay between infected travelers , climate , vectors , and indigenous dengue incidence remains unclear . The role of foreign-origin cases on local dengue epidemics thus has been largely neglected by research . This study investigated the effect of both imported dengue and local meteorological factors on the occurrence of indigenous dengue in Taiwan . Using logistic and Poisson regression models , we analyzed bi-weekly , laboratory-confirmed dengue cases at their onset dates of illness from 1998 to 2007 to identify correlations between indigenous dengue and imported dengue cases ( in the context of local meteorological factors ) across different time lags . Our results revealed that the occurrence of indigenous dengue was significantly correlated with temporally-lagged cases of imported dengue ( 2–14 weeks ) , higher temperatures ( 6–14 weeks ) , and lower relative humidity ( 6–20 weeks ) . In addition , imported and indigenous dengue cases had a significant quantitative relationship in the onset of local epidemics . However , this relationship became less significant once indigenous epidemics progressed past the initial stage . These findings imply that imported dengue cases are able to initiate indigenous epidemics when appropriate weather conditions are present . Early detection and case management of imported cases through rapid diagnosis may avert large-scale epidemics of dengue/dengue hemorrhagic fever . The deployment of an early-warning surveillance system , with the capacity to integrate meteorological data , will be an invaluable tool for successful prevention and control of dengue , particularly in non-endemic countries .
Dengue outbreaks initiated by international tourists , immigrants , and foreign workers have been reported in numerous developed areas and countries [1] , [2] , [3] . Nevertheless , the interplay between infected travelers , climate , vectors , and indigenous dengue incidence remains unclear [3] , [4] . Historically , the link between imported cases and indigenous cases has been established through phylogenetical analysis and viral sequence comparisons [5] , [6] . However , these retrospective studies are not capable of providing timely , relevant information about transmission dynamics , nor do they provide quantitative insight for disease control strategies in a broader context . For example , epidemiological data has indicated that imported dengue cases enter Taiwan almost every month from other countries ( Figure 1B ) but have not always resulted in local dengue epidemics [6] , [7] . This suggests that the timing of imported dengue's entrance may have considerable effect on domestic dengue epidemics [3] , [8] . However , the role of these foreign-origin cases in local dengue epidemics has not yet been quantitatively assessed [9] . The aims of this study were to clarify the relationship between imported dengue , local weather , and domestic epidemics of dengue , and to further identify the role of imported cases ( in different phases ) during a dengue epidemic in non-endemic areas such as Taiwan . The study used data of all imported and indigenous dengue cases nationwide that had been confirmed by the Centers for Disease Control in Taiwan ( Taiwan-CDC ) [10] , [11] , Republic of China ( R . O . C . ) to investigate the relationship between imported and indigenous dengue , and all concurrent meteorological characteristics with potential for facilitating disease transmission .
Dengue , including dengue fever ( DF ) and dengue hemorrhagic fever ( DHF ) , are notifiable infectious diseases to be reported within 24 hours in Taiwan . Information on these confirmed cases of dengue fever ( DF ) and dengue hemorrhagic fever ( DHF ) were obtained from Taiwan-CDC from 1998 to 2007 through dengue surveillance in Taiwan . This surveillance system is made up of three parts: passive , active and semi-active surveillance . In passive surveillance , dengue-like illness reports by health care workers to local health authorities account for most confirmed dengue cases . Active surveillance , including volunteer reporting and fever screenings at international airports ( identifying fever cases by infrared thermal scanner , which has been routinely operated by the government since 2003 ) [5] , [6] . In semi-active surveillance , fever cases are investigated in residential areas , schools , and work places with epidemiological linkage , and specimens are taken once confirmed dengue cases are identified . These active and semi-active components , serve to complement and reinforce in support of comprehensive virus detection . Among active strategies , fever screening detects imported dengue cases efficiently [6] . All febrile patients identified through fever screening are required to submit blood samples for testing . In addition , public health professionals at local health authorities monitor suspected cases for the development of dengue-like symptoms/signs until dengue virus infection is excluded [10] . These strategies identify and manage potential dengue cases before they enter into the community . The current definitions for dengue , including DF , DHF and dengue shock syndrome ( DSS ) in Taiwan have been applied since the 1980s . Cases of “probable DF” are patients with body temperatures ≥38°C and two or more of the following clinical manifestations: headache , retro-orbital pain , myalgia , arthralgia , rash , hemorrhagic manifestations and leucopenia . Cases of “probable DHF” and DSS are further identified based on criteria established by the World Health Organization [12] . Identified probable dengue cases must provide blood specimens for laboratory confirmation tests . These laboratory tests include molecular identification of dengue virus by reverse-transcriptase polymerase chain reaction ( RT-PCR ) [13] , single or paired serum samples testing for dengue-specific IgM seropositives , 4-fold dengue-specific IgG serotiter rises ( with the exclusion of Japanese encephalitis virus infection ) [12] , or virus isolation [14] , [15] . Date of onset of dengue illness , age , gender , clinical manifestations , reporting hospital , and laboratory results were all thoroughly documented for each dengue case . Epidemiological questions such as travel history , incubation period , and first day of illness were evaluated to identify the possible origin of dengue infection . “Imported dengue cases into Taiwan” were defined as laboratory-confirmed dengue cases with travel history to endemic countries within 14 days before the date of onset of dengue ( based on Taiwan-CDC's definition ) [10] . Confirmed indigenous dengue cases in three epidemic areas in Southern Taiwan [Tainan ( TN ) , Kaohsiung ( KH ) , and Pingtung ( PT ) ] were investigated . All three areas had identified both Aedes aegypti and Ae . albopictus mosquitoes as vectors for transmitting dengue virus . KH , including both metropolitan Kaohsiung and Kaohsiung County , had served as the location for the majority of Taiwan's dengue epidemics involving all four serotypes of dengue viruses . Smaller scale epidemics of dengue also occurred in both TN and PT , located adjacent to Kaohsiung . For this study , TN included Tainan City and County , while PT referred to Pingtung City and County . The subtropical climate of southern Taiwan presents an annual hot and rainy summer season lasting from June to August and daily mean temperatures ranging from 18° to 32°C year round . Information on the predominant serotype of isolated dengue viruses in TN , KH and PT ( Figure 1C ) was obtained from Taiwan-CDC [10] . Taiwan's dominant serotypes/genotypes of epidemic DENV varied by year and area [6] . However , in 2002 , a DENV-2 epidemic attacked all three areas of our study . During our study period , KH had the most frequent occurrence of dengue epidemics , with epidemics of DENV-2 in 1998 and 2001–2003; DENV-1 in 2004; DENV-3 in 2006 , and DENV-1 in 2007 . TN had four major epidemics , including DENV-3 in 1998 [16] , DENV-4 in 2000 , DENV-2 in 2002 , and DENV-1 in 2007 . PT had two major epidemics , including DENV-2 in 2001–2003 and DENV-1/DENV-4 in 2004 . We found that local dengue epidemics , with geographical variations in these three areas , only had higher numbers of indigenous cases during certain years . We systematically collected daily weather data for Taiwan that was publicly available through the 26 branch stations of the Central Weather Bureau ( http://www . cwb . gov . tw/ ) . The meteorological variables analyzed in this study were selected after comprehensive evaluation of all available data with biological relevance to vectors or cases of dengue , including daily mean temperature , daily maximum temperature , daily minimum temperature , daily mean relative humidity , daily mean wind speed , daily accumulative rainfall , daily accumulative rainy hours , daily sunshine accumulation hours , daily mean sunshine rate ( from sunrise to sunset ) , and daily sunshine total flux . Unlike weather stations in Tainan and Kaohsiung , Pingtung County's station is located a far distance from Pingtung City , where the majority of Pingtung's dengue cases occurred . We therefore used weather data collected by the Environment Protecting Agency ( EPA ) at their station in Pingtung City . This EPA weather station was only able to provide data regarding daily mean temperature , daily maximum temperature , daily minimum temperature , daily mean wind speed , and daily accumulative rainfall . We then substituted Kaohsiung's data for Pingtung's meteorological variables not provided by the EPA because of Pingtung City's close proximity to Kaohsiung City . All laboratory-confirmed daily dengue cases , according to the date of onset of dengue illness , were summed into total case numbers in bi-weekly intervals for data analysis . The mean value of each meteorological variable was also calculated for each biweekly interval . Abbreviations of all variables analyzed are listed in the Table 1 . As the effects of imported dengue and meteorological factors on indigenous dengue logically had a time lag , we thus tested different time lags for each variable from lag 1 up to lag 12 ( lag 1 representing two weeks , lag 2 representing four weeks , and so on ) . Logistic regression was used to analyze the correlation between the occurrence/increase of indigenous dengue and the number of imported cases , as well as the correlation between the occurrence/increase of indigenous dengue and each meteorological variable across various time lags ( from 2 weeks to 24 weeks ) . Poisson regression was used to analyze the correlation between the number of indigenous dengue cases and the number of imported cases , as well as the correlation between the number of indigenous dengue cases and quantitative data of each meteorological variable across time lags from 2 weeks to 24 weeks . Regression with the negative binomial model [17] was used for over-dispersed data . All models were adjusted by area ( two dummy variables , area_KH and area_TN ) , popd ( area-specific population density ) , and sine24 plus cosine24 ( the oscillatory sine and cosine functions were used to model seasonal variations of dengue cases [18] ) . Because the quantitative relationship between indigenous and imported dengue cases may exist only at the onset of local dengue epidemics , we further divided all bi-week intervals into three categories for further analysis: 1 ) Period of “low intensity transmission”: From March to May during our study period , 94 . 44% ( 170/180 ) of bi-week intervals during these three months had no indigenous dengue cases in these studied areas . 2 ) Period of “early phase of outbreaks”: Those bi-week intervals presenting <10 indigenous dengue cases for months excluding March to May . 3 ) Period of “late phase of outbreaks”: Those bi-week intervals presenting ≧10 indigenous dengue cases . Further information on these regression models are listed in the Text S1 . Two-tailed p<0 . 05 was regarded as statistically significant . The statistical analysis was conducted using S-PLUS Enterprise Developer Version 8 . 0 . 4 ( TIBCO Software Inc . , Palo Alto , CA , USA ) and SAS 9 . 1 . 3 Service Pack 4 ( SAS Institute Inc . , Cary , NC , USA ) .
Among the 9 , 910 laboratory-confirmed indigenous dengue cases ( mean age ± standard deviation = 45 . 63±18 . 67 years ) from 1998 to 2007 , 9 , 195 ( 92 . 79% ) were adults ( >15 years old ) . Cases in study areas accounted for 98 . 45% ( 9 , 756/9 , 910 ) of total confirmed indigenous cases in Taiwan during this period ( Figure 1A ) . Figure 1B indicates that the number of biweekly imported dengue cases in Taiwan significantly increased over time ( β = 0 . 026±0 . 002 , p<0 . 0001 ) . Comparing local dengue case numbers of the three study areas ( TN , KH and PT ) in Figure 1D with all of Taiwan's dengue cases in Figure 1B , Figure 1D serves to illustrate that imported dengue cases entered southern Taiwan almost every month ( within one year ) without a clear pattern . Indigenous dengue cases , in contrast , exhibited a strong seasonal regularity ( Figure 1C ) . Figure 2 displays estimates of regression coefficients of independent variables ( Xs ) in the logistical regression model for the “occurrence” of indigenous dengue cases . We found that the variables of the number of imported cases ( imported , p = 0 . 0023∼0 . 0315 ) and daily maximum/mean/minimum temperatures ( tmax/tmean/tmin , p = 0 . 0002∼0 . 0495 ) were positively correlated . On the contrary , relative humidity ( rh ) was negatively correlated with indigenous case occurrences ( p<0 . 0001∼p = 0 . 0433 ) . These findings indicate that an increase in imported cases , in conjunction with warmer and drier weather , is favorable for the occurrence of indigenous dengue . Among other meteorological variables , one sunshine related variable and wind speed did not exhibit consistently significant relationships with indigenous dengue ( data not shown ) . However , Figure 3 reveals that , the influence of both imported cases and weather conditions on the “increase” of indigenous dengue was less significant . In addition , when binary outcomes were replaced with indigenous case counts ( Figure 4 ) , the quantitative relationships between imported and indigenous dengue cases became insignificant . In Figure 5 , we observed variation in the impact of imported dengue at different epidemic phases ( please see definitions in Methods ) . Using Poisson models , we found that the imported dengue cases were significantly correlated with indigenous dengue with lag 4 ( i . e . 8 weeks ) only in periods of “low intensity transmission” ( Figure 5A ) . However , this relationship was more statistically significant in the “early phase of outbreaks” ( Figure 5B ) . Imported dengue had their greatest impact on epidemics during this phase . When local epidemics entered a period of “late phase of outbreaks” , these correlations disappeared ( Figure 5C ) , suggesting that imported cases were unlikely to have influence on indigenous cases during this period . These findings may indicate that imported dengue cases initiate local dengue cases almost exclusively during the onset of an epidemic .
This study examined all laboratory-confirmed dengue cases detected through a combination of active , semi-active , and passive surveillance , and found that imported dengue are able to serve as an initial facilitator , or spark , for domestic epidemics . Nevertheless , imported dengue cases do not have a noteworthy effect from March to May , during the low transmission period of dengue in Taiwan . When these sparks meet suitable weather conditions , the tinder , local dengue epidemics result . Eventually , this relationship disappears once biweekly indigenous case numbers rise over ten , indicating that a local epidemic has occurred . Our findings thus provide evidence that a significant quantitative relationship between Taiwan's imported and indigenous dengue case numbers exists solely at the onset of an epidemic and in the context of appropriate meteorological conditions . Because the numbers of imported dengue cases that initiate indigenous cases have been increasing in non-endemic areas such as Taiwan [4] , [5] , [6] , [8] ( further supported by the high nucleotide identities of dengue viruses isolated from travelers with travel history to endemic countries [5] , [6] , [13] , [16] , [19] ) , this study ventures to provide epidemiological evidence of the combined impact of both imported dengue and weather conditions on local outbreaks . Climate factors have provided helpful clues for monitoring dengue's transmission in affected areas [20] , [21] , [22] , [23] . Higher temperature has the effect of shortening the time intervals of extrinsic incubation in the mosquito life cycle [23] , [24] and is positively correlated with more occurrences of indigenous dengue in our study . This is consistent with previous findings that demonstrate the suitability of warm or hot weather for the survivorship of adult mosquitoes and , thus , dengue transmission [22] , [25] . In addition , although increased rainfall has been shown to increase the number and quality of mosquito breeding sites ( as well as the density of resting sites ) [21] , lower rainfall and relative humidity ( RH ) were significantly related to indigenous dengue in this study . The correlation between lower RH and indigenous dengue with time lags was also observed in previous studies in Thailand [26] , [27] . We explain this phenomenon as follows . Drier conditions may facilitate dengue transmission through the increase of water storage behavior , which result in an increase of breeding sites for Aedes mosquitoes , particularly in areas without reliable water supplies [28] , [29] , [30] . Although piped water supply is available in 90% of Taiwan ( http://www . water . gov . tw/eng/08statistics/sta_a_main . asp ? bull_id=4341 ) , water storage for gardening or agricultural use is popular during water restriction period in the dry season , October to April , in southern Taiwan . In addition , a previous field survey identified water buckets as the most common breeding sites of Ae . aegypti in southern Taiwan [31] . Entomologically , lower RH ( 50% vs . 90% ) aids higher flight speed of female adult Ae . aegypti at temperatures higher than 21 degrees of Celsius [32] thus facilitating dengue transmission . This may explain why both RH and rainfall showed a negative correlation with the number of indigenous dengue ( Figure 4 ) and , that while higher temperatures occurred during July to September in the summer of Taiwan , the number of indigenous dengue cases usually peak in October–November . On the other hand , although the correlation between drier conditions and increased transmission is unlikely to be caused by higher temperatures , we acknowledge that the effects of meteorological factors have a complex relationship . Unlike the consistent negative correlation across lags 3–8 ( rain ) and lags 4–10 ( rainhr ) in Figure 4 , the positive correlation of “rain” and “rainhr” in Figure 3 occurred only in lag 9 , and was therefore most likely a random statistical anomaly rather than a conclusive finding . We believe that weather-based mechanisms that support the proliferation of indigenous dengue therefore need further region-specific investigation and more international collaboration . To the best of our knowledge , this is the first study to simultaneously identify the relationship between indigenous and imported dengue cases in the context of meteorological factors . Our findings provide a highly accurate epidemiological portrait of dengue in Taiwan because of the following components of the research: First , a better surveillance system was instituted to actively rather than passively detect dengue cases . This system was also laboratory-based to minimize confounding infection and manifestations [5] , [13] , [33] , [34] . Second , we avoided a potential bias as a result of delays in dengue notification by analyzing all confirmed dengue cases in accordance to their onset dates of illness rather than their reporting dates . We consider that vector control efforts on dengue cases do not affect outbreak initiation , but rather the size and magnitude of an outbreak . A dengue notification delay of over one month allows for two transmission cycles , and increases the potential for a large outbreak [35] . Vector control operations in Taiwan are unlikely to influence imported cases to initiate local dengue epidemics because they are implemented after case notification [10] . By the time indigenous dengue cases increase , the relationship with imported cases disappear ( Figure 5C ) . Hence , the focus of this study was to verify the correlation between imported dengue and the onset of local dengue epidemics under appropriate weather conditions . In order to construct the best possible regression models to reflect meteorological conditions , we built alternative statistical models to demonstrate the role of imported cases in the onset of dengue epidemics . Previous modeling studies using ARIMA ( Autoregressive Integrated Moving Average ) found that the number of imported dengue cases was not associated with the incidence of local dengue [9] , [36] . ARIMA examined a linear relationship between case numbers of imported dengue cases and incidence of indigenous dengue cases over several time lags . However , the quantitative relationship between imported and indigenous dengue was likely limited to the onset ( i . e . early phase ) of outbreaks , and was therefore not subject to linear modeling . We believe these conclusions by logistic and Poisson regression models are not only demonstrable in countries with distinct seasonality , but also applicable in non-endemic areas of dengue . However , meteorological conditions may need to be modified for countries in higher altitudes . Under suitable weather conditions , dengue viruses introduced via travelers are likely to result in further domestic spread and subsequent occurrence of epidemics . In addition , the introduction of more virulent genotypes of dengue viruses has been documented as a potential factor for driving new epidemics [37] , [38] , [39] . For example , Thai strains belonging to the 1980–1994 clade within the genotype I of dengue virus serotype 1 ( DENV-1 ) were replaced by a 1990–2002 clade [6] . Additionally , an old clade in genotype I of DENV-3 during 1976–1978 was also replaced by a new 1991–2002 clade in genotype II [5] , [6] . Furthermore , cosmopolitan genotypes of DENV-2 , the causing agent of Taiwan's largest-scale epidemic of dengue/DHF in last thirty years , had been gradually and effectively replacing Asian genotype 2 in the Philippines since 1998 and entered Taiwan in 2001 [40] . This cosmopolitan genotype of DENV-2 is different from the Asian 1 and Asian 2 genotypes of Taiwan's DENV-2 isolates from 1981 to 1998 and the American/Asian genotype of Taiwan's isolates in 2005 , when the majority of dengue cases were dengue fever [41] . In other words , the more virulent genotypes/strains of the same serotype that have emerged during later years have resulted in more severe and/or larger-scale epidemics of dengue/DHF in many Asian countries [37] , [39] . Based on phylogenetic analyses of dengue viruses isolated from imported cases [6] at the micro-level , we find that local dengue epidemics in Taiwan typically originate in South East Asia . It is therefore imperative to establish a stable surveillance system to detect the spread of different genotypes of DENV . Currently , Taiwan's comprehensive dengue surveillance system is evolving and , hopefully , it may continuously monitor the possible evolution of DENV in SE Asian countries through international collaboration . We believe that global warming may have further impact on the incidence of imported dengue cases and future dengue/DHF epidemics [42] . Advanced research integrating virus displacement and meteorology will be necessary to provide a fuller understanding of both the macro and micro changes contributing to the increasing severity of dengue/DHF epidemics . This study had notable limitations . First of all , meteorology alone does not initiate an epidemic . Herd immunity also plays a decisive role in the spread of disease . Once a new or more virulent genotype/strain of dengue virus is introduced , public health officials should alert the public and implement prevention efforts regardless of meteorological conditions . Second , local entomological data from Taiwan's entomology surveillance was not included . Non-government scholars do not have access to such data prior to 2002 . Furthermore , the data was divided by village or “Li” – the basic administration unit in Taiwan , and was not systematically collected with a standardized process . Therefore , we did not use entomological data because of its lack of consistency and inability to adequately represent the locations covered in our study . Lastly , although socioeconomic status may influence vector habitat [43] , it was assumed to be relatively stable during the studied ten years . As an increase in viremic international travelers has caused global DHF case numbers to surge in the past several decades [44] , efficient measures have to be instituted to prevent imported dengue cases from igniting local dengue/DHF epidemics . Additionally , it has been previously found that DHF cases with higher viral load [45] appeared when the number of dengue fever cases increased rapidly , particularly in areas with higher dengue clusters [46] . All these findings suggest that the entrance of imported cases , in conjunction with suitable meteorological conditions , may have the potential to precipitate severe epidemics involving more DHF cases . Careful tracking and clinical management of imported dengue cases , along with relevant meteorological information , are able to provide earlier warning signals for emerging indigenous dengue epidemics than current surveillance systems [11] , [20] . These early alerts allow for the proper implementation of targeted public health interventions and valuable buffer time for preventing subsequent large-scale epidemics of dengue/DHF locally and in affected countries . Advanced investigation and integration of immunological , virological , meteorological , and entomological findings with prevention/control strategies will support a more comprehensive understanding of the mechanisms that initiate dengue epidemics , and will help guide realistic public health interventions in the era of rapid globalization and climate change [47] .
|
Dengue/dengue hemorrhagic fever is the world's most widely spread mosquito-borne arboviral disease and threatens more than two-thirds of the world's population . Cases are mainly distributed in tropical and subtropical areas in accordance with vector habitats for Aedes aegypti and Ae . albopictus . However , the role of imported cases and favorable meteorological conditions has not yet been quantitatively assessed . This study verified the correlation between the occurrence of indigenous dengue and imported cases in the context of weather variables ( temperature , rainfall , relative humidity , etc . ) for different time lags in southern Taiwan . Our findings imply that imported cases have a role in igniting indigenous outbreaks , in non-endemics areas , when favorable weather conditions are present . This relationship becomes insignificant in the late phase of local dengue epidemics . Therefore , early detection and case management of imported cases through timely surveillance and rapid laboratory-diagnosis may avert large scale epidemics of dengue/dengue hemorrhagic fever . An early-warning surveillance system integrating meteorological data will be an invaluable tool for successful prevention and control of dengue , particularly in non-endemic countries .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"public",
"health",
"and",
"epidemiology/epidemiology",
"infectious",
"diseases/viral",
"infections",
"public",
"health",
"and",
"epidemiology/infectious",
"diseases",
"infectious",
"diseases/tropical",
"and",
"travel-associated",
"diseases",
"infectious",
"diseases/epidemiology",
"and",
"control",
"of",
"infectious",
"diseases"
] |
2010
|
The Role of Imported Cases and Favorable Meteorological Conditions in the Onset of Dengue Epidemics
|
Phenotypic plasticity is the ability of a genotype to produce contrasting phenotypes in different environments . Although many examples have been described , the responsible mechanisms are poorly understood . In particular , it is not clear how phenotypic plasticity is related to buffering , the maintenance of a constant phenotype against genetic or environmental variation . We investigate here the genetic basis of a particularly well described plastic phenotype: the abdominal pigmentation in female Drosophila melanogaster . Cold temperature induces a dark pigmentation , in particular in posterior segments , while higher temperature has the opposite effect . We show that the homeotic gene Abdominal-B ( Abd-B ) has a major role in the plasticity of pigmentation in the abdomen . Abd-B plays opposite roles on melanin production through the regulation of several pigmentation enzymes . This makes the control of pigmentation very unstable in the posterior abdomen , and we show that the relative spatio-temporal expression of limiting pigmentation enzymes in this region of the body is thermosensitive . Temperature acts on melanin production by modulating a chromatin regulator network , interacting genetically with the transcription factor bric-à-brac ( bab ) , a target of Abd-B and Hsp83 , encoding the chaperone Hsp90 . Genetic disruption of this chromatin regulator network increases the effect of temperature and the instability of the pigmentation pattern in the posterior abdomen . Colocalizations on polytene chromosomes suggest that BAB and these chromatin regulators cooperate in the regulation of many targets , including several pigmentation enzymes . We show that they are also involved in sex comb development in males and that genetic destabilization of this network is also strongly modulated by temperature for this phenotype . Thus , we propose that phenotypic plasticity of pigmentation is a side effect reflecting a global impact of temperature on epigenetic mechanisms . Furthermore , the thermosensitivity of this network may be related to the high evolvability of several secondary sexual characters in the genus Drosophila .
Phenotypic plasticity and buffering are concepts describing the phenotypic outcome of genotype-environment interactions . Phenotypic plasticity is the ability of a given genotype to produce different phenotypes in different environments [1] . It has been the subject of increasing interest as it is involved in adaptation and evolution [1–7] . Buffering , or canalization , is the ability of an organism to maintain a stable phenotype despite genetic variation or environmental fluctuations [8] . Buffering can be challenged by environmental stress , such as extreme temperatures . Thus , the question arises whether the plasticity of a particular phenotype is a specifically targeted reaction of the organism to changes in the environment or whether it is a side effect , reflecting a global process at the level of the transcriptome/proteome , but visible for weakly buffered phenotypes . To answer this question , we investigated the genetic basis of a particularly well described trait subject to phenotypic plasticity: the abdominal pigmentation of female Drosophila melanogaster , which strongly depends on the temperature conditions during development [9 , 10] . In the posterior abdomen , the differences of pigmentation between females grown at 20 °C and 29 °C are comparable to the phenotypic effect of mutations in major structural or developmental regulatory genes . The extreme plasticity of this phenotype makes it a particularly suitable model to dissect the responsible mechanisms . Within the last ten years , key studies have identified structural and developmental regulatory genes playing major roles in abdominal pigmentation patterning [11–16] . Because these studies focused on genetic factors , they were performed under standard temperature conditions [11 , 13–17] . Following a classical developmental genetics approach , we use mutations in key regulatory or structural genes to destabilize the underlying genetic networks and analyze how they interact with temperature .
Pigmentation is sexually dimorphic in D . melanogaster . In males , the abdominal tergites 5 and 6 are black and maintain this pigmentation at all temperatures . In contrast , the posterior abdominal pigmentation in females is highly polymorphic and plastic [18 , 19] . Figure 1A shows the abdominal pigmentation phenotypes of females from three different wild-type genotypes grown at different temperatures . At a given temperature , the extent of the dark region of the segments in females can differ dramatically between Drosophila lines , showing strong genetic basis [18 , 19] . NO1 and Samarkand are outliers and most lines have a pattern similar to that of BV1 , comparable to the patterns described previously through pigmentation score [10] . Differences in plasticity are observed within each segment along the antero-posterior axis [10] and along the dorso-ventral axis ( Figure 1A ) . This is extreme for A7 , which can shift from completely black at 20 °C to completely yellowish at 29 °C . In addition , the transition border between the yellowish and the dark region of the tergites is not smooth but variegated ( Figure 1B ) , implying that the control of pigmentation is not robust . The spatial restriction of the phenotypic plasticity of pigmentation suggests the involvement of developmental regulatory genes . The morphology of abdominal tergites A5 , A6 , and A7 is specified by the posterior homeotic gene Abdominal-B ( Abd-B ) [20] . In tergites , Abd-B expression is low in A5 , intermediate in A6 , and high in A7 [21] . Thus , the increasing plasticity observed in the abdomen along the antero-posterior axis [10] perfectly correlates with the expression level of Abd-B . We used the Transabdominal mutation [17] to test whether ectopic expression of this gene in another body region is sufficient to generate a plastic pigmentation pattern . This mutation is a chromosomal rearrangement that fused the regulatory region of the stripe gene to the Abd-B locus [17 , 22] . It induces an ectopic expression of Abd-B on the thorax at the flight muscle attachment sites . This phenotype was previously described as sexually dimorphic , inducing melanin production in the whole sites of ectopic expression in males and in only restricted areas in females [17] . The pattern is indeed sexually dimorphic , but it is also extremely plastic ( Figure 2 ) . Remarkably , in females , the regions that are not brown at the sites of ectopic Abd-B expression show a very strong reduction in the production of yellowish pigments ( Figure 2G–2L ) . This indicates that Abd-B plays opposite roles in melanin production . It either increases melanin production or represses the production of all pigments . Furthermore , these two roles are extremely thermosensitive . The increase of melanin production is much higher at low temperature , whereas the decrease in pigment production is much stronger at high temperature . These two roles of Abd-B are concomitantly observed within the same spot of ectopic expression , which suggests that they are influenced by other developmental pathways . In order to quantify the effects of Abd-B and temperature on pigmentation , we tested how modifications of Abd-B expression level interact with temperature in the development of the pigmentation pattern . We varied the copy number of Abd-B using a deficiency of Abd-B ( Df ( 3R ) -RS-1–98 ) and a duplication of Abd-B ( Dp-P5 ) . Both mutations are carried in the same stock , which reduces background effects as much as possible . We found that high temperature decreases melanin production in all genotypes , but the effects of Abd-B level differed in A6 and A7 and along the dorso-ventral axis within A6 ( Figure S1A–S1I ) . Thus , we quantified the melanin production along the antero-posterior axis in the lateral , median , and dorsal region of A6 , in the lateral region of A7 , and along the dorso-ventral axis in A7 ( Figure S2 ) . Temperature , Abd-B as well as the Abd-Bxtemperature interaction , strongly influenced pigmentation in each of these regions ( p < 0 . 001 for all ) , except the dorsal midline ( Table S1 ) . These effects explained a large proportion of the total variation in abdomen pigmentation . The Abd-Bxtemperature interactions are particularly striking in the lateral regions of A6 and A7 and in the median region of A6 ( Figure 3 ) . In line with previous studies [13] , Abd-B strongly increases melanin production in the lateral region of A6 under all temperature regimes ( Figure 3A ) . The Abd-Bxtemperature interaction in the lateral region of A6 was mainly attributable to the pronounced reduction in pigmentation at high temperatures when the expression of Abd-B is low ( Figure 3A ) . There was also a significant interaction between Abd-B and temperature in the lateral region of A7 ( Figure 3C ) . However , in contrast to the lateral region of A6 , Abd-B represses melanin production in the lateral region of A7 . The opposite roles of Abd-B on melanin production are best illustrated by the very pronounced Abd-Bxtemperature interaction in the median region of A6 . At low temperature , high Abd-B levels increase melanin production , whereas at high temperature they reduce melanin production ( Figure 3B ) . Based on these data , we conclude that Abd-B has two opposite roles on melanin production in females and can either increase or decrease melanin production . This makes the balance between melanin production and repression very unstable in the posterior abdomen , generating phenotypic plasticity in pigmentation . Indeed , this balance is very sensitive to temperature and is most pronounced in A7 showing the highest Abd-B level . Abd-B is a developmental regulatory gene encoding a homeodomain transcription factor [23] . Its opposite roles on melanin production must be ultimately mediated by pigmentation enzymes . Indeed , pigment precursors move only a few cell diameters; thus , the spatial restriction of some of the enzymes synthesizing them is directly responsible for the pigmentation pattern observed in the adult [14 , 24 , 25] . A consensus model of pigment synthesis pathway is discussed in Text S1 . Two classes of enzymes can be distinguished . Enzymes of the first class such as the tyrosine hydroxylase ( TH ) or the dopa decarboxylase ( DDC ) are required for the production of pigment precursors involved in the synthesis of all pigments . Enzymes of the second class , such as Ebony , Yellow , or Tan , are involved in the switch between the production of yellowish ( NBAD sclerotin ) or black-brown ( dopa-melanin and dopamine-melanin ) pigments [14 , 26] . The strong reduction in the production of all pigment observed in regions expressing Abd-B on the thorax of Tab/+ females suggests that Abd-B represses melanin production through the downregulation of one or several enzymes of the first class . The strong production of melanin observed at low temperature in the posterior abdomen and in some of the regions expressing Abd-B on the thorax of Tab/+ females suggests that Abd-B also regulates one or several enzymes of the second class . Mutations in genes encoding enzymes of the first class are homozygously lethal but loss-of-function mutations for enzymes of the second class are homozygously viable and can be used to identify the target ( s ) of Abd-B involved in plasticity . We postulated that if phenotypic plasticity of pigmentation is caused by temperature-dependent activity or regulation of a particular pigmentation enzyme , loss-of-function mutations in the gene encoding this enzyme should generate a nonplastic pigmentation phenotype . We used mutations in yellow ( y1 ) , tan ( t1 ) , and ebony ( e1 ) ( Figure S3 ) . We observed that the plasticity of abdominal pigmentation is still visible in females mutant for y ( Figure S3D–S3F ) , which lack black melanin but still have brown melanin . In t mutants , melanin is produced in males and females in the posterior abdomen at 20 °C , but is strongly reduced at higher temperatures in both sexes ( Figure S3G–S3I and S3U ) . In contrast , e flies remain very dark at high temperature and show very limited plasticity of pigmentation ( Figure S3J–S3L ) . Thus , a functional e gene is required for the plasticity of pigmentation . In flies mutant for t , which antagonizes e [15] , the role of e is magnified . This suggests that the system responsible for plasticity in A5 or A6 also exists partly in males , but that it is normally hidden by the activity of Tan . The production of melanin in t mutants requires the repression or the strong downregulation of e , as even the gain-of-function of y cannot induce melanin production in the presence of Ebony [14] . Thus , the pigmentation pattern of t mutants at different temperatures implies that e is differentially regulated at different temperatures , relative to the production of melanin precursors , and that a major temperature-induced regulatory switch occurs between 20 °C and 25 °C . We thus investigated the expression of the genes encoding Ebony and the two enzymes of the first class , TH and ddc . In order to visualize the expression of these enzymes , we stained pharate adults dissected out of their pupal case . We first investigated the expression pattern of e , ddc , and TH at 25 °C using e-LacZ , ddc-LacZ , and UAS-LacZ; TH-Gal4 flies , respectively . We observed that e , ddc , and TH are highly expressed in the pattern of the thoracic trident ( Figure S4 ) . The trident is a cryptic pattern fully visible in e mutants , y; e double-mutants ( Figure S4B and S4C ) [14] , or when flies are grown at extreme temperatures [27] ( Figure 2A ) . A similar pattern was described with an antibody against Ebony [14] . This suggests that the coexpression of e with TH and ddc assures that most of the locally produced pigment precursors are normally converted into yellowish NBAD sclerotin by Ebony . In the absence of Ebony , the excess of dopamine is converted into dopamine- melanin [14] . Thus , the melanin pattern in the absence of e is completely correlated to the spatial expression of TH and DDC in the epidermis . We then looked at the expression of these enzymes in abdominal segments to see how their combined spatio-temporal expression could explain the pigmentation pattern . We observed with the e-LacZ transgene an expression similar to that previously reported using an antibody against Ebony [14] . It starts at the base of the bristles around 90 h after puparium formation ( Figure 4A ) and then becomes progressively uniform in the epidermal cells of the tergites ( Figure 4B and 4C ) . We observed that the epidermal expression starts in the anterior region of the segment , as a weak antero-posterior gradient is first visible ( Figure 4B ) . e-LacZ is stronger in the anterior region ( Figure 4B , arrowhead ) than in the posterior region of the segment ( Figure 4B , arrow ) . We observed that the expression of ddc is also very dynamic in the posterior abdomen ( unpublished data ) , but it is even more pronounced for TH , which encodes the first and limiting enzyme in the pigment synthesis pathway . Thus , we focused on TH . In the abdomen , TH expression started at the base of the large bristles on the posterior border of the segments before complete maturation of the bristles ( Figure 4D ) . Expression at the base of more anterior bristles begins later ( Figure 4E ) . Finally , TH is later expressed in epidermis of the whole tergites ( Figure 4F ) . In the abdomen , it is expressed along an antero-posterior gradient as the expression starts much later in the more posterior segments ( Figure 4D–4F ) . In particular , no strong expression is visible before hatching in A7 ( Figure 4F ) . Thus , the expression of TH is lower in the posterior abdomen where the Abd-B level is the highest . We then looked at the expression of TH on the thorax of Tab/+ pharate females . We observed that it is also very dynamic . The expression is first visible at the base of the bristles located in the region of the trident ( Figure 4G ) ; then , a uniform expression is visible in the epidermal cells of the trident ( Figure 4H ) . In the regions of ectopic Abd-B expression , expression starts at the base of the bristle located close to the teeth of the trident ( Figure 4H , arrowhead ) . Later on , it is visible in the epidermal cells of these regions , but the regions that are devoid of pigments in Tab/+ flies show a much reduced staining ( Figure 4I , arrows ) . Thus , the most plastic regions , i . e . , the posterior abdomen and the regions of Abd-B ectopic expression in Tab/+ females , express TH very late . Furthermore , regions of ectopic expression of Abd-B devoid of pigments correspond to strongly reduced expression of TH . The delayed pigmentation in the posterior abdomen and the loss of TH expression in the regions of ectopic expression of Abd-B in Tab/+ females suggest that Abd-B represses TH , at least indirectly . There is an obvious difference between e and TH expression at 25 °C in the posterior abdomen . In particular , at 25 °C , e is already expressed in A7 before hatching , whereas TH is not yet expressed . We assume that TH is expressed in A7 , which is pigmented , but this expression probably occurs after hatching . This is likely , as the activity of TH was reported to peak 50 min after hatching [28] . This means that when TH starts to be expressed , e being already expressed , DOPA and dopamine can be used to produce NBAD , the precursor of the yellowish pigment . Because the phenotype of tan mutants suggested that a major regulatory switch occurs between 20 °C and 25 °C , we analyzed the expression of e and TH in females grown at 20 °C . When flies are grown at 20 °C , the expression of e just before hatching is much weaker in the posterior abdomen than at 25 °C ( Figure 4J ) . Furthermore , at 20 °C , TH expression can be observed very clearly before hatching in A7 , but is mainly seen in association with bristles in the inner region of the tergite ( Figure 4K ) . Thus , Abd-B plays opposite roles in melanin production by repressing at least two genes encoding pigmentation enzymes with different roles in melanin production: TH required for the production of all pigments and Ebony required for the production of the yellowish pigment . It makes the expression of these enzymes particularly sensitive to temperature in the posterior abdomen . At low temperature , the stronger repression of e and the reduced repression of TH correlate with the increased melanin production observed in the posterior abdomen and on the thorax of Tab+/ females . In contrast , at higher temperature , the strong repression of TH and its delayed expression correlate with the reduced pigment production observed in the posterior abdomen and on the thorax of Tab/+ females . The effect on expression timing is visible on the pigmentation phenotype of the A7 tergite in limiting conditions , for example , in females with three doses of Abd-B grown at 25 °C: the melanin remaining is clearly associated with bristles , which mark the first sites of TH expression ( Figure 4L ) . How could temperature influence these opposite roles of Abd-B on pigmentation ? Abd-B was previously shown to induce melanin production both via the repression of the transcription factor bric-à-brac ( bab ) and independently of bab [13] . bab , which was shown to repress melanin production , is strongly repressed in males by Abd-B in A5 and A6 [13] . This leads to the melanic pigmentation observed in the posterior abdomen [13] . bab is activated by the female-specific isoform of doublesex ( dsxF ) , which compensates for the repression of bab by Abd-B , and thus reduces the amount of melanin produced in this part of the abdomen compared to males [13] . In female A7 , bab is not repressed by Abd-B [13] . In order to analyze potential interactions between bab and environmental temperature , we used the babAR07 mutation that completely abolishes the expression of the two paralogs bab1 and bab2 and induces a well characterized haplo-insufficient melanic phenotype [13 , 18 , 29] . We observed that this phenotype is fully visible at 25 °C in A6 , but is less obvious at other temperatures compared to wild-type ( Figure S1J–S1L ) . Multivariate analysis of the effect of bab and temperature on melanin production ( Table S2 ) revealed a very strong effect of bab and babxtemperature interaction in the lateral and median region of A6 and along the dorso-ventral axis of A7 ( p < 0 . 001 for all , Figure 5A and 5B; Table S2 ) . No significant effect was observed in the lateral region of A7 ( Figure 5C ) . Thus , BAB level is less limiting in wild-type flies in A7 , where bab is not repressed by Abd-B , than in A6 . The role of bab on melanin production has been described previously [13 , 29] , but these experiments did not reveal whether bab acts mainly by regulating pigmentation enzymes of the first or the second class . To identify the main targets of bab , we overexpressed bab1 in the dorsal domain using the pannier driver as previously described [13] , but in an e or in a y background ( Figure 6A and 6B ) . We observed that the production of both melanin and yellowish NBAD sclerotin is strongly repressed by the overexpression of bab1 ( Figure S5A and S5B , arrows ) . It suggests that bab represses an enzyme of the first class . BAB has been reported to physically interact with products of the Broad-Complex [30] , a direct regulator of ddc in pharate adults [31] . We used a ddc-lacZ transgene and observed that ddc-lacZ is downregulated in the dorsal domain of flies overexpressing bab1 ( Figure S5C , compared to Figure S5D , arrows ) . Thus , the effect of bab on melanin production is mediated at least through the repression of ddc . Interestingly , temperature was shown previously to modulate the effect of bab loss-of-function on another phenotype: the presence of an ectopic sex comb observed in males on the second tarsal segment of the first leg [32] . The sex comb is a structure made with modified bristles present on the first tarsal segment in Drosophila melanogaster males . Ectopic sex comb are extremely informative phenotypes frequently used to identify and quantify particular genetic interactions . The sex comb phenotype of bab mutants ( distal sex comb ) is observed also in some chromatin regulator mutants of the Polycomb group ( PcG ) and Enhancer of Trithorax and Polycomb group ( ETP ) [33–36] . The Polycomb group ( PcG ) and the antagonizing Trithorax-group ( TrxG ) proteins were identified through their role in the regulation of homeotic genes ( Hox ) [37] , but it is now clear that they regulate hundreds of genes [38 , 39] . The PcG are involved in silencing of Hox genes , whereas the TrxG are involved in their activation . A third class of chromatin regulators has been described , the Enhancers of trithorax and Polycomb ( ETP ) , required for both TrxG and PcG normal functions [40] . Most PcG mutants induce the formation of ectopic sex combs on the first tarsal segment of the second and third legs caused by ectopic expression of the homeotic gene Sex-comb reduced ( Scr ) [37 , 41] . However , ectopic distal sex combs are observed in mutants of only a few PcG or ETP genes [34–36] . This suggests that these different ectopic sex comb phenotypes correspond to the disruption of two distinct processes , and that bab and a subset of chromatin regulators are required for the repression of the distal sex comb . Other data suggest that these genetic interactions probably correspond to physical interactions between BAB and chromatin regulators . The bab locus encodes two closely related transcription factors with a BTB/POZ domain [29] . This interaction domain is present in many chromatin regulators [42] or transcription factors recruiting chromatin regulators [43] . In particular , BAB has been reported to bind to Batman/LolaL , which is part of PcG and TrxG complexes [42] . Interestingly , the activity of chromatin regulators such as members of the Trithorax/Polycomb system are known to be temperature-sensitive [38 , 44] . Silencing by PcG through characterized regulatory sequences known as Polycomb Response Elements was shown to be stronger at high temperature [38 , 44 , 45] . Thus , we hypothesized that the modulation of bab activity by temperature could take place via an effect of temperature on a network of chromatin regulators interacting with BAB . We used the sex comb phenotype to test for genetic interactions between bab , genes encoding ETP or PcG , and temperature . We observed strong genetic interactions between bab and corto , cramped ( crm ) , batman/lolal , and Trithorax-like ( Trl ) that encodes GAGA ( Figure 6 ) . Temperature strongly enhances the sex comb phenotype of crm7/Y; babAR07/+ males . At 29 °C , they die in their pupal case with sex comb teeth also visible on the third tarsal segment of the first leg ( Figure 6B ) in a large fraction of the individuals ( 6/18 observed legs ) . This phenotype is not observed at lower temperature or in single mutants . In addition , the second tarsal segment is inflated and shortened , which reduces the size of the ectopic sex comb and makes quantification impossible . Therefore , we quantified the crm-bab interaction only at 25 °C using the number of teeth in the ectopic sex comb on the second tarsal segment of the first leg ( Figure 6C ) . Wild-type flies have no sex comb teeth on the second tarsal segment ( 0 ± 0 , n = 16 ) . The crm7/Y; babAR07/+ males have many more teeth ( 6 . 18 ± 0 . 25 , n = 22 ) than crm7/Y hemizygotes ( 2 . 75 ± 0 . 18 , n = 12 ) and babAR07 heterozygotes ( 0 . 18 ± 0 . 05 , n = 36 ) . This strong genetic interaction is shown in Figure 6C . We also quantified the effect of temperature on the interactions between bab and other chromatin regulators ( Figure 6A and 6D , Tables S3 and S4 ) . The single heterozygote mutants for corto , ban , or Trl do not show ectopic sex combs . We analyzed how these mutations modify the babAR07/+ phenotype . The genotype and the temperature accounted for 71% of the variance ( Table S4 ) . Temperature had a strong effect and increased sex comb teeth number across all genotypes . All heterozygote double mutants were significantly different from babAR07/+ ( Tukey post hoc test , p < 0 . 001 ) , which shows that the effect of bab mutation is strongly enhanced by mutations in corto , ban/lolal , or Trl . We tested three corto alleles . All alleles showed a similar trend , but the effects were stronger for corto420 and cortoL1 . In addition , there was a significant genotype/temperature interaction ( Table S4 ) visible in the curves corresponding to babAR07 single mutants and double heterozygotes ( Figure 6D ) . Given these strong effects , we analyzed the effect of mutations in chromatin regulators on the abdominal pigmentation of babAR07 heterozygote females . All the mutants we looked at had been induced in different backgrounds , so it was not possible to clearly differentiate the effect of the mutation itself from the background . Balancers are frequently used as control when the mutant stock is out-crossed to a wild-type line . However , most of the balancer chromosomes from the mutant stocks carry mutations in pigmentation genes , which is particularly inadequate in our case . Thus , we focused on genes for which we observed very strong phenotypes and interactions , and/or for which we could test different alleles . We observed very strong effects for the three corto alleles . They dominantly induce a reduced pigmentation in A7 at 25 °C . This is extreme for corto420 and cortoL1 ( Figures S6 and S7 ) and weaker for corto07128 ( Figure S7E ) . We tested how they would modify the haplo-insufficient pigmentation seen in babAR07 females . We observed a strong temperature-sensitive effect on pigmentation in babAR07/corto420 in A6 with a completely black phenotype at 20 °C , a strong variegation at 25 °C , and a completely white pigmentation at 29 °C ( except for the dorsal midline ) ( Figure S7G–S7I ) . In A7 , the pigmentation was very weak at 25 °C . A similar effect was observed with cortoL1 ( Figure S8 ) and a visible but weaker effect with corto7128 ( Figure S8 ) . Quantification of melanin production revealed very strong effect for corto420 and strong interactions between corto and temperature and between bab , corto , and temperature ( Figure 5A–5C , Table S1 ) . In particular , whereas reducing bab level by half has no significant effect on melanin production in the lateral region of A7 , it interacts very strongly with corto for this phenotype ( Figure 5C , Table S1 ) . In addition , in the median region of A6 , babAR07/corto420 females produce less melanin than wild-type or single heterozygous females , whereas reducing bab level alone has the opposite effect ( Figure 5B ) . This corresponds to the variegated phenotype observed at 25 °C ( Figure S7H ) and shows that bab and corto work together to increase melanin production in this region of A6 . The females homozygous for the crm7 allele of the PcG gene crm show an absence of melanin in A7 and a very reduced and variegated pigmentation in A6 ( Figure S8A ) . This is not observed in their heterozygote siblings crm7/FM7c ( Figure S8B ) or when out-crossed to a wild-type stock ( Figure S8C ) . We also quantified how heterozygosity for crm7 would modify the pigmentation phenotype of babAR07/+ females . Except for the dorsal midline , we observed strong effect of crm and interaction between crm and bab , between crm and temperature , and between crm , bab , and temperature ( Table S1 ) . The genetic interactions between bab and chromatin regulators for abdominal pigmentation and sex comb development , and previously reported physical interaction between BAB and the ETP Batman/Lolal [42] , suggest that BAB and chromatin regulators cooperate in the regulation of particular targets . In order to test this hypothesis , we used antibodies against BAB , CRM , and Corto to localize their products on salivary gland polytene chromosomes . We observed many colocalizations on polytene chromosomes of BAB with Corto and BAB with CRM ( unpublished data ) . In particular , we observed clear staining for BAB , Corto , and CRM in the cytological region corresponding to the locus of TH ( 65C ) ( Figure 7 ) . BAB and CRM colocalized in the cytological region of the ddc ( 37C ) ( Figure 7G and 7H ) . We detected BAB alone in the cytological region of e ( 93C ) ( Figure 7E , 7F , 7K , and 7L ) . Chaperones , in particular Hsp90 , have been shown to buffer against the effect of cryptic genetic variation and environmental stress , in particular against high temperature [46 , 47] . Recent studies revealed a link between the chaperones and chromatin regulators [48–50] , which suggested that the effect of temperature on chromatin regulators might be partly mediated by chaperones . We tested this hypothesis using two different alleles of Hsp83 , the gene encoding Hsp90 . In females , the allele Hsp83e6D dominantly induced a very low pigmentation in A7 at 25 °C ( Figure S6Q ) , not observed with the allele Hsp83e6A ( unpublished data ) , which had a weaker TrxG-like effect than Hsp83e6A [48] . We tested the effect on abdominal pigmentation of the two Hsp83 alleles in babAR07 heterozygous females . We observed that Hsp83e6D ( Figure S6T and S6U ) , but not Hsp83e6A ( unpublished data ) , strongly reduced the pigmentation phenotype of babAR07/+ at 25 and 29 °C . Quantification of melanin production revealed strong effect of Hsp83e6D and a strong interaction with temperature ( Table S1 ) . In contrast , interactions between Hsp83e6D and bab were only significant in the median region of A6 . Significant interactions were observed between bab , Hsp83e6D , and temperature in the median region of A6 and along the dorso-ventral axis of A7 ( Table S1 ) . We also quantified the effect of Hsp83e6D on babAR07 heterozygote sex comb phenotype and found that it increased the number of teeth in the ectopic sex comb at 20 °C and 25 °C , but slightly decreases it at 29 °C ( Figure 6D , Table S4 ) . Because of the similarity of effects of Hsp83 and corto on the pigmentation phenotypes , we tested for potential genetic interactions between these two genes . We observed extragenic noncomplementation ( lethality ) between Hsp83e6D and cortoL1 ( observed when crossed in both directions ) , whereas cortoL1 is viable with Hsp83e6A . The Hsp83e6D/corto420 and Hsp83e6D/corto07128 genotypes were viable , but strongly enhanced the loss of pigmentation observed in Hsp83e6D/+ , corto420/+ , or corto7128/+ females at 25 °C ( Figures S6V–S6X and S7P–S7R ) . Quantification of melanin production revealed very strong interactions between Hsp83 and corto and between Hsp83 , corto , and temperature ( Figure 5D–5F; Table S1 ) . In particular , the double heterozygote Hsp83e6D/corto420 females have an extremely reduced melanin production ( Figure 5D–5F ) and are the only genotype we analyzed where the pigmentation in A6 at the dorsal midline is affected at 25 and 29 °C . Furthermore , the Hsp83e6D allele ( Figure S9 ) also dominantly induced pigmentation defects in males in the A5 segment observed at 25 °C and 29 °C , but not at 20 °C ( Figure S9B and S9C ) . Most importantly , corto alleles strongly increased the pigmentation defects observed in Hsp83e6D/+ males at 25 °C and 29 °C ( Figure S9D–S9I ) . Loss of pigmentation was visible in A6 with both alleles . In contrast , corto420/+ and corto07128/+ males had a normal pigmentation in A5 and A6 at all tested temperatures ( unpublished data ) . We observed a similar phenotype in Hsp83e6D/babAR07 males ( Figure S9J–S9L ) . This suggests that Hsp83 , bab , and corto work tightly together to control abdominal pigmentation in both males and females and are much more required at 25 °C and 29 °C than at 20 °C .
Buffering or canalization describes the ability of individuals of a given species to show a constant phenotype despite genetic variations or environmental fluctuations . Phenotypic plasticity could therefore be interpreted as a weaker buffering of some phenotypes . Chaperones , and in particular Hsp90 , have been identified as components of this buffering system and are thought to become limiting under stressful conditions such as high temperature [46 , 48] . The chaperone Hsp90 was proposed to act as a general evolutionary capacitor by releasing the effect of cryptic genetic variation under stressful environment [46] . However , more recent studies have revealed that the influence of Hsp90 on the variation of particular traits was very limited [51] . This suggests that the ability of chaperones , and Hsp90 in particular , to buffer phenotypic variation is not so general , but might rely on very specific interactions more tightly involved in particular phenotypes . Recent studies revealed a link between chaperones and chromatin regulators [48–50] . In particular , Hsp90 and several TrxG chromatin regulators were shown to buffer the same phenotype [48] . These studies on buffering in flies were based on the penetrance of deleterious phenotypes caused by cryptic genetic variation or known introduced mutations [46 , 48] . We found that mild modulation of a similar system by environmental temperature is involved in phenotypic plasticity of abdominal pigmentation . The chromatin regulator network we found to be sensitive to environmental temperature interacts genetically with the chaperone Hsp90 and the transcription factor BAB . It contains the PcG gene crm and the ETP gene corto . We observed very strong genetic interactions between corto and Hsp83 , the gene encoding Hsp90 . In particular , we observed extragenic noncomplementation between Hsp83e6D and cortoL1 , whereas the viable trans-heterozygote combinations with the two other corto alleles induce strong reduction in melanin production in both sexes . This suggests that Hsp90 and Corto are involved in a common process . Hsp90 has been shown to physically interact with histones , to induce chromatin condensation , and to interact with topoisomerase II , which plays a crucial role in chromosome condensation [52 , 53] . Interestingly , corto is also required for the normal condensation of chromosomes during mitosis [33] . Therefore , the interactions between corto and hsp83 in gene regulation might be linked to a general role and common involvement of these genes in chromatin condensation . The pigmentation in the posterior abdomen is particularly sensitive to environmental temperature because it is sensitized by the input of the homeotic gene Abd-B . Abd-B plays opposite roles in melanin production . The positive role of Abd-B in melanin production in females has already been described [13] . It is linked to the establishment of the sexually dimorphic pattern of pigmentation , and , for this role , Abd-B works antagonistically with bab by repressing it in A6 and A5 [13] . The role of Abd-B in repressing melanin production has not been described previously . It is very strong in A7 and is probably linked to the very peculiar morphology of this segment in females . In A7 , bab is not repressed by Abd-B , and both genes work together not only to repress melanin production , but also to control some aspects of the particular development of this segment such as the absence of fusion of the tergites at the dorsal midline [29 , 54] . Abd-B plays these opposite roles in melanin production by repressing several pigmentation enzymes such as TH and Ebony . These enzymes start to be expressed at the end of pupal development [14 , 28] . Modulation of the regulation of these enzymes by temperature induces a local difference in their relative timing of expression in the abdominal epidermis . The effect is particularly visible in A7 , which exhibits the highest Abd-B level . Studies in Drosophila wing have shown that pigment precursors can also be provided by the hemolymph [55] . Hence , it is possible that a change in the general level of pigment precursors in the hemolymph might contribute partly to the phenotypic plasticity of abdominal pigmentation . However , the diffusion of pigment precursors from the hemolymph does not seem to play an important role in the pigmentation of the body: recent studies in lepidopterans showed that the local production of DOPA and dopamine by TH and DDC in the epidermis are major components of the pigmentation pattern [56] . In Drosophila abdomen , epidermal clones of cells mutant for TH or DDC are albino [55] , which shows that pigment precursors potentially available in the hemolymph cannot contribute significantly to pigment production in the body epidermis . In addition , in the thorax , the pigmentation patterns visible in e and y , e mutants perfectly correlate with the epidermal expression of TH and DDC , the enzymes providing pigment precursors ( Figure S4 ) . Furthermore , we clearly show that the spatial restriction of plasticity is strongly conditioned by Abdominal-B expression and the repression of pigment precursor production in the thoracic epidermis in Transabdominal mutants ( Figure 2 ) . Thus , we conclude that the modulation of the relative temporal expression of TH and ebony by temperature in the epidermis of the posterior abdomen is responsible for the phenotypic plasticity of female abdominal pigmentation . Mutations in corto , crm , hsp83 , and bab enhance the effect of temperature on melanin production in the posterior abdomen . The colocalization of bab , corto , and crm at the locus containing TH in polytene chromosomes suggest that they might all cooperate in the direct regulation of this pigmentation enzyme , and that they might counteract the effect of temperature and Abd-B on TH expression . Their mutations enhance the repression of TH by Abd-B and high temperature , which explains why it has a particularly strong effect in A6 and A7 , at 25 °C and 29 °C . We therefore propose the model presented in Figure 8 to explain some aspects of the pigmentation pattern plasticity in the posterior abdomen . It does not exclude that temperature also modifies the expression of other genes . This is likely as the PcG/TrxG have hundreds of targets [38 , 39] , and the thermosensitivity of the PcG/TrxG system is a general phenomenon observed with PRE from several different genes [38 , 44] . It is possible that other genes ( developmental regulators or structural genes ) are also modulated by temperature and contribute to the phenotypic plasticity of pigmentation . We observed , indeed , many colocalizations of Corto and BAB , and Corto and CRM , suggesting that this particular network of chromatin regulators regulate many targets . We demonstrate that this network is involved in at least two different phenotypes , abdominal pigmentation in females and sex comb development in males , both showing high temperature sensitivity . Thus , we propose that the plasticity of Drosophila pigmentation is a visible side effect , at particularly sensitive loci , of a process affecting the whole genome through alteration of epigenetic mechanisms . Interestingly , abdominal pigmentation and morphology of the sex comb along the proximo-distal axis of the first leg evolve very rapidly in the Drosophila genus [57–59] . Remarkably , we found that these two morphological traits are under the control of a common thermosensitive network including the transcription factor bab , chromatin regulators , and the chaperone Hsp90 . This suggests that the thermosensitivity of this particular regulatory network might be linked to the high evolvability of several secondary sexual characters in the genus Drosophila . Our results corroborate other studies , which have shown that the plasticity of specific traits is correlated to their evolvability [60] .
Most of the fly stocks used in this study were provided by the Bloomington Drosophila Stock Center ( http://flystocks . bio . indiana . edu ) . The following ones were kindly sent to us by various researchers: babAR07and UAS-bab1 ( Jean-Louis Couderc ) , crm7 ( Neel Randsholt ) , Df ( 3R ) RS-1–98/Dp-P5 ( Artyom Kopp ) , ebony-LacZ ( Bernard Hovemann ) , and Pc3 Tab ( Ian Duncan ) . The Tab mutation is associated with the Pc3 inversion in the stock we used , but the pigmentation phenotype visible on the thorax has been shown to be caused by the ectopic expression of Abd-B [17] . Flies were grown on standard agar-corn medium . Standard balancer chromosomes were used . For the effect of temperature , crosses were kept at 25 °C and tubes transferred after 2 d to the desired temperature . Oregon-R was used as a wild-type stock to outcross mutant stocks . All fly stocks are described in Flybase ( http://flybase . bio . indiana . edu ) . The interaction between bab , chromatin regulators , and chaperones was analyzed using sex comb phenotypes ( except for the bab/crm interaction ) and female abdominal pigmentation in the progeny of crosses between females Oregon-R or babAR07/+ and males carrying the mutation to be tested , or wild-type Oregon-R males . The interactions between crm ( located on the X chromosome ) and bab on sex comb was analyzed in the male progeny of crosses between crm7/FM7c females and males babAR07/TM6b , and compared to the effect of these mutations alone when crossed to Oregon-R . The effect and interaction with corto and bab of the hsp83e6D allele on male abdominal pigmentation was observed and analyzed in the male progeny of crosses made with females hsp83e6D/TM6b and wild-type males or carrying the mutation to test . Flies were fixed in 75% ethanol 3 d after hatching to allow proper pigmentation of the cuticle to develop . Abdominal cuticles were cut on one side of the dorsal midline , cleaned , and mounted in Euparal ( Roth ) . At least 15 flies were observed for each genotype/temperature condition , except the crm7 homozygote female escapers ( Figure S3A ) for which we had only five individuals . Thoraces were dissected in 75% ethanol , fixed 5 min in 100% ethanol , and mounted in Euparal . Flies were dissected out of their pupal case , fixed , and stained without further dissection with X-Gal according to previously described protocols for pharate abdomen [61 , 62] . Because developmental time is conditioned by temperature [63] , we staged pharate adults according to progression of eye color , bristle and wing melanization , meconium appearance , and ability of the fly to walk prematurely when dissected out of the pupal case [64] . Stainings were performed overnight at 37 °C for e-LacZ and ddc-LacZ , or 2 h for TH-GAL4; UAS-LacZ genotypes . Thoraces or abdomen were dissected after staining in PBS . This allowed us to make sure that absence of staining was not caused by tissue disruption during dissection prior to staining . Tissues were then dehydrated 5 min in 75% ethanol , 5 min in 100% ethanol , and mounted in Euparal . Immunostaining of polytene chromosomes was performed as described by Cavalli ( http://www . igh . cnrs . fr/equip/cavalli/link . labgoodies . html ) on larvae of the w1118 genotype . The rabbit anti-CRM [35] , rat anti-BAB2 [30] , and rabbit anti-Corto [65] antibodies were used respectively , at 1:50 , 1:200 , and 1:25 dilutions . All statistical analyses were performed using the software SPSS . 10 . 0 or SPSS . 13 [66] . We scored sex comb teeth number in the ectopic sex comb on the second tarsal segment . The number of teeth on the left and right legs of individuals was highly correlated for the bab-crm interaction ( rs = 0 . 892 , p < 0 . 001 ) , as well as for bab and other chromatin regulator interactions across temperatures ( rs = 0 . 667 , p < 0 . 001 ) . Therefore , we averaged the number of teeth of the left and right leg for all subsequent analysis . We analyzed the effect of mutations in bab and chromatin regulators on sex comb teeth number at three temperatures using a parametric two-way ANOVA . Log transformed data ( ln + 1 ) were not exactly normally distributed; however , the residuals of the analysis did not deviate from normality ( Kolmogorov-Smirnov test , p > 0 . 05 ) . We observed that the different genetic combinations affected melanin production differently in abdominal segments A6 and A7 . Furthermore , within segments the effects differed along the dorso-ventral axis . Thus , we scored the proportion of melanin visible along the antero-posterio axis at the dorsal midline ( A6D ) , on both sides in the lateral region of A6 ( A6L1 , A6L2 ) , and in the median region of A6 ( A6M1 , A6M2 ) ( Figure S2 ) . In A7 we scored the proportion of melanin on both sides along the dorso-ventral axis ( A7DV1 , A7DV2 ) and the lateral region along the antero-posterior axis ( A7L1 , A7L2 ) ( Figure S2 ) . Ten individuals were scored for each genotype/temperature combination . Pigmentation scores varied between 0 ( no melanin ) and 4 ( fully black ) ( Figure S2 ) . Pigmentation scores between the left and right side were highly correlated within all regions ( all: rs ≥ 0 . 925 , p < 0 . 001 ) . They were averaged in each individual ( A6L , A6M , A7L , A7DV ) . We analyzed these pigmentation data using a multivariate analysis of variance . A6L , A6M , A6D , A7L , and A7DV were used as dependent variables; temperature and genotypes at Abd-B ( one , two , or three doses ) , bab ( one or two doses ) , corto ( wild-type or corto420/+ ) , Hsp83 ( wild-type or Hsp83e6D/+ ) , and crm ( wild-type or crm7/+ ) as fixed factors . We included all main effects as well as possible interaction terms in the model ( Table S1 ) . The model includes genes interacting with bab ( encoding putative cofactors ) . Thus , in this model , although the effect of bab and the interaction between bab and temperature are highly significant ( Table S1 ) , they are also allocated to the interactions between bab and other genes , and between bab , other genes , and temperature . In order to test more generally the effect of bab and its interaction with temperature before dissecting the network ( see Results ) , we also performed a multivariate analysis using the same dependant variables: temperature and genotype at bab as fixed factors in a reduced dataset with only wild-type and babAR07/+ females ( Table S2 ) .
|
The phenotype of an individual is not fully controlled by its genes . Environmental conditions ( food , light , temperature , pathogens , etc . ) can also contribute to phenotypic variation . This phenomenon is called phenotypic plasticity . We investigate here the genetic basis of the phenotypic plasticity of pigmentation in the fruit fly Drosophila melanogaster . Drosophila pigmentation is strongly modulated by temperature , in particular in the posterior abdominal segments of females . The development of these segments is controlled by the homeotic gene Abdominal-B ( Abd-B ) . Abd-B sensitizes pigmentation patterning in this region of the body by repressing several crucial pigmentation enzymes . It makes the regulation of their spatio-temporal expression in the posterior abdomen particularly sensitive to temperature variation . We show that temperature modulates the mechanisms regulating the dynamic structure of the chromosomes . Chromosomal domains can be compacted and transcriptionally silent , or opened and transcriptionally active . Temperature interacts with a network of chromatin regulators and affects not only the regulation of pigmentation enzymes but several traits under the control of this network . Thus , we conclude that the phenotypic plasticity of female abdominal pigmentation in Drosophila is a visible consequence for a particularly sensitive phenotype , of a general effect of temperature on the regulation of chromosome architecture .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"evolutionary",
"biology",
"drosophila",
"genetics",
"and",
"genomics"
] |
2007
|
Phenotypic Plasticity in Drosophila Pigmentation Caused by Temperature Sensitivity of a Chromatin Regulator Network
|
Hypoxia inducible factors ( HIFs ) are transcription factors belonging to the basic helix−loop−helix PER-ARNT-SIM ( bHLH-PAS ) protein family with a role in sensing oxygen levels in the cell . Under hypoxia , the HIF-α degradation pathway is blocked and dimerization with the aryl hydrocarbon receptor nuclear translocator ( ARNT ) makes HIF-α transcriptionally active . Due to the common hypoxic environment of tumors , inhibition of this mechanism by destabilization of HIF-α:ARNT dimerization has been proposed as a promising therapeutic strategy . Following the discovery of a druggable cavity within the PAS-B domain of HIF-2α , research efforts have been directed to identify artificial ligands that can impair heterodimerization . Although the crystallographic structures of the HIF-2α:ARNT complex have elucidated the dimer architecture and the 0X3-inhibitor placement within the HIF-2α PAS-B , unveiling the inhibition mechanism requires investigation of how ligand-induced perturbations could dynamically propagate through the structure and affect dimerization . To this end , we compared evolutionary features , intrinsic dynamics and energetic properties of the dimerization interfaces of HIF-2α:ARNT in both the apo and holo forms . Residue conservation analysis highlighted inter-domain connecting elements that have a role in dimerization . Analysis of domain contributions to the dimerization energy demonstrated the importance of bHLH and PAS-A of both partners and of HIF-2α PAS-B domain in dimer stabilization . Among quaternary structure oscillations revealed by Molecular Dynamics simulations , the hinge-bending motion of the ARNT PAS-B domain around the flexible PAS-A/PAS-B linker supports a general model for ARNT dimerization in different heterodimers . Comparison of the HIF-2α:ARNT dynamics in the apo and 0X3-bound forms indicated a model of inhibition where the HIF-2α-PAS-B interfaces are destabilised as a result of water-bridged ligand-protein interactions and these local effects allosterically propagate to perturb the correlated motions of the domains and inter-domain communication . These findings will guide the design of improved inhibitors to contrast cell survival in tumor masses .
Hypoxia inducible factors ( HIFs ) are obligate heterodimers belonging to the basic helix−loop−helix ( bHLH ) superfamily of transcription factors that mediate the physiological responses to hypoxia . This extensive protein family is characterized by a 4–6 basic amino acids next to a HLH dimerization domain , both required to properly bind DNA targets . Within the bHLH superfamily HIFs belong to the subfamily containing the PER/aryl hydrocarbon receptor nuclear translocator ( ARNT ) /single minded ( SIM ) ( PAS ) homology domain ( bHLH-PAS ) [1–3] . Based on their heterodimerization behavior , bHLH-PAS proteins can be further divided into two classes: class I members only form heterodimers with a member of class II , which , by contrast , can promiscuously homo- and heterodimerize . Class I includes aryl hydrocarbon receptor ( AhR ) , aryl hydrocarbon receptor repressor ( AhRR ) , neuronal PAS proteins ( NPAS1-4 ) , single minded proteins ( SIM1-2 ) , clock circadian regulator ( CLOCK ) and three HIF-α subunits isoforms , HIF-1α , HIF-2α , and HIF-3α , each targeting both shared and distinct genes [4] . When transcriptionally active , HIF-α subunit dimerizes with the constitutive ARNT ( also known as HIF-β ) subunit , the best characterized class II protein; other members of this class include the tissue restricted ARNT2 , and the circadian rhythm proteins BMAL1 and BMAL2 [2 , 3] . The poorly conserved C-terminal region of bHLH-PAS proteins hosts the transactivation domains ( TAD ) where the transcriptional coactivators are recruited to initiate the transcription [1]; the N-terminal part contains three well-defined domains: bHLH , PAS-A , and PAS-B . The bHLH domain offers the primary dimerization interfaces and , together with the protein partner , determines the target gene recognition [5] . Despite low sequence identity , the PAS domains show conserved three-dimensional structures in a wide range of prokaryotic and eukaryotic proteins [2] . They contribute to the dimerization and increase the specificity of partner choice [6 , 7] . PAS-A , in particular , prevents dimerization with non- PAS-containing bHLH proteins and participate to the binding of DNA sequences that differ from the prototypical E-box motif [5] . The PAS-B domain commonly functions as a signalling domain and hosts hydrophobic cavities for small molecules and/or cofactors that relay environmental or metabolic signals [7]; consequently to the binding , allosteric changes occur that affect the affinity for partner molecules [8] . HIF-1α and HIF-2α contain an additional N-terminal TAD and an oxygen dependent degradation domain ( ODD ) that enable HIFs to monitor oxygen concentration . Under normoxia ( 20% O2 ) , two prolines in the ODD are hydroxylated by the HIF prolyl hydroxylases , PHD1–3 , and recognized by the von Hippel–Lindau ( VHL ) tumor suppressor protein , the substrate-binding subunit of the E3 ubiquitin ligase complex . The binding causes polyubiquitination and targets HIF-α to the proteasome [1] . In hypoxic conditions , HIF-α escapes degradation and , after translocation to the nucleus , heterodimerizes with ARNT , binds hypoxia response elements ( HREs ) in the enhancer regions of target genes , interacts with CBP-p300 complex and initiates transcription [1] . Activated genes are involved in glycolysis , erythropoiesis and angiogenesis; the gene products include erythropoietin , that stimulates the production of red blood cells , and vascular endothelial growth factor ( VEGF ) , a regulator of blood vessel growth [1] . In tumor masses , the abnormal vasculature creates hypoxic regions that activate HIFs to promote angiogenesis and to switch to anaerobic metabolism , sustaining cell viability under hypoxic conditions [9] . HIFs are commonly upregulated in a broad range of cancers [10–12] where they contribute also to resistance to oxidative stress , epithelial–mesenchymal transition ( EMT ) , and tumor invasiveness . HIF-1α and HIF-2α accumulation can also be caused by reduced degradation , as in VHL syndrome , an inherited familial cancer syndrome where mutation of VHL causes its inactivation [13] . Internal hydrophobic cavities are observed in all available structures of bHLH-PAS family within both their PAS-A and PAS-B domains [14] . AhR uses its PAS-B internal cavity for binding a diverse set of small molecules thus activating nuclear translocation , dimerization with ARNT and DNA binding [15 , 16] . In other PAS domains , ligand binding in the pockets induces long distance conformational changes that affect protein-protein interactions [17] , suggesting that PAS cavities can contain allosteric sites [7 , 18] . In addition to the PAS domain cavities , HIF-1α:ARNT and HIF-2α:ARNT structures show a pocket at HIF-α PAS-B:ARNT PAS-A interface , which has been targeted by acriflavin [19] , a mix of trypaflavin and proflavin that acts as a potent inhibitor of both HIF-1α and HIF-2α dimerization with ARNT in cells [20] . As HIF-α:ARNT dimerization is essential to bind DNA and initiate transcription , destabilizing protein–protein interactions in this system represents an optimal therapeutic approach for tumor treatment . However , while direct antagonizing of the interfaces with small molecules is pharmacologically demanding and often unsuccessful due to the troublesome identification of the key residues to target [21] , exploiting PAS internal cavities offers potential advantages , especially in terms of selectivity . The need of developing isoform-specific drugs emerges from the observations that HIF-1α and HIF-2α target distinct genes [4] , in some cases affecting tumor progression in opposite ways [11] . HIF-2α PAS-B domain contains a relatively large ( 290 Å3 ) cavity that can be occupied by either water or small molecules with sub-micromolar affinities . These small binders have been shown to impair heterodimerization of isolated PAS-B domains in vitro [22 , 23] . In the framework of extensive efforts directed to identify inhibitors of HIF-2α:ARNT dimerization [24 , 25] , a molecule has been recently developed , 0X3 ( N- ( 3-chloro-5-fluorophenyl ) -4-nitro-2 , 1 , 3-benzoxadiazol-5-amine ) , that is able to disrupt heterodimerization also in living cells [26] . The compound fails to bind to HIF-1α , which has a smaller cavity in PAS-B domain . Albeit the molecular details of 0X3 interaction with HIF-2α PAS-B have been unveiled , how the ligand binding destabilizes the HIF-2α:ARNT complex remains unexplained . The recently determined structures of the entire bHLH-PAS region of the HIF-2α:ARNT dimer in the unbound , DNA-bound , and inhibitor-bound ( 0X3 and proflavin ligands ) forms [19] provide a sound basis for assessing the inhibition mechanism . These dimer structures show that , while the two bHLH domains become linked in a pseudo-symmetric arrangement , the PAS domains interact asymmetrically . Besides the interfaces between corresponding PAS domains , there are also interfaces formed by HIF-2α PAS-B-ARNT PAS-A and HIF-2α intramolecular interactions between PAS-A and PAS-B and between PAS-A and bHLH domains . The lack of physical interaction between ARNT domains facilitates flexibility for arrangements with different partners . Indeed in the NPAS1: and NPAS3:ARNT heterodimers , ARNT PAS-B domain is slightly displaced in comparison with HIF-α:ARNT complex [14] . PAS-B cavity residues facing 0X3 in the HIF-2α:ARNT-0X3 complex are not significantly perturbed , while the PAS domains slightly shifts one respect to the others , with major rearrangements occurring at the interface between the HIF-2α and ARNT PAS-B domains [19] . Here we hypothesized that the ligand-induced local perturbation at the HIF-2α PAS-B domain dynamically propagates through the HIF-2α:ARNT dimerization interfaces by an allosteric inhibition mechanism . To study the functional dynamics of the complex and shed light into the mechanism of regulation of dimer stability , we compared the evolutionary , dynamical and energetic properties of HIF-2α:ARNT dimer structure in its unbound and 0X3-bound form . We identified both the molecular features of the ligand-induced perturbation and the key residues involved in inter-domain communication paths . This novel insight in HIF-2α regulation will guide the development of new specific inhibitors of aberrant HIF-2α activity .
Mapping of residue evolutionary conservation on protein structure can provide reliable prediction of functionally relevant elements and their role in multi domain organisation [27] . Residue conservation on protein surfaces was analysed with ConSurf [28 , 29] . PAS-domain sequences were detected with a PSI-BLAST [30] search ( 3 iterations; E-value cutoff 0 . 0001 ) of the PDB sequence of HIF-2α:ARNT ( PDB ID: 4ZP4 ) against the UniProt database [31] . Orthologous sequences were manually selected for each protein independently: HIF-1α , HIF-2α and HIF-3α for HIF-2α , and ARNT1 and ARNT2 for ARNT , for a total of 110 and 170 sequences ( S1 Table ) . Input multiple sequence alignments were generated with Muscle [32] . Crystal structures for HIF-2α:ARNT dimer in its apo ( PDB ID: 4ZP4 ) and holo ( PDB ID: 4ZQD ) forms [19] were obtained from the Protein Data Bank ( PDB ) [33] . The structures have five unresolved segments on each partner: two inter-domain ( bHLH/PAS-A and PAS-A/PAS-B ) linkers and three intra-domain PAS-A loops . Among these unresolved segments , the GH loop and the PAS-A/PAS-B linker of HIF-2α are available in at least another crystal structure . The other eight unresolved segments in the apo deposition were modelled using the Rosetta all-atom de novo loop modelling method with the Next Generation Kinematic closure ( NGK ) procedure , a variant of the Kinematic Closure ( KIC ) approach [34–36] . A starting set of 1000 loop models was generated with the parameters proposed in [37] , enabling the Taboo sampling feature and using Monte-Carlo simulated annealing for rotamer-based side-chain optimization in a neighborhood of 10 Å around the loop structures . The ensemble of models was then clustered by backbone structural similarity using the Self Organizing Map ( SOM ) approach previously described in [38–40] . The best conformation for each loop was selected from the most populated cluster by Rosetta energy score . Missing regions in the holo structure were completed by grafting the atomic coordinates of the loops modelled for the apo form and refined using Modeller 9v8 [41] . The completed structures were then pre-processed for simulation with the Schrodinger's Protein Preparation Wizard tool [42]: hydrogen atoms were added , all water molecules were removed , C and N terminal capping were added , disulfide bonds were assigned and residue protonation states were determined by PROPKA [43] at pH = 7 . 0 . Each system was then solvated in an octahedral box with TIP3P water molecules , and neutralized with Na+ ions using the GROMACS [44] preparation tools . The minimal distance between the protein and the box boundaries was set to 12 Å . Simulations were run using GROMACS 5 . 1 [44] with Amber ff99sb*-ILDNP force-field [45] . 0X3 inhibitor in the holo structure was parameterised using GAFF [46] . 0X3 charges were calculated with the restricted electrostatic potential ( RESP ) method [47] at HF/6-31G* after ab-initio optimization of the ligand . A multistage equilibration protocol ( modified from [48] ) was applied to all simulations to remove unfavourable contacts and provide a reliable starting point for the production runs: the system was first subjected to 1000 step of steepest descent energy minimization , followed by 1000 step of conjugate gradient with positional restraints ( 2000 kJ mol-1 nM-2 ) on all resolved atoms . This minimization process was then repeated with weaker ( 1000 kJ mol-1 nM-2 ) restraints on the backbone of resolved regions . Subsequently a 200 ps NVT MD simulation was used to heat the system from 0 to 100 K with restraints lowered to 400 kJ mol-1 nM-2 and then the system was heated up to 300 K in 400 ps during a NPT simulation with further lowered restraint ( 200 kJ mol-1 nM-2 ) . Finally , the system was equilibrated during a NPT simulation for 1 ns with backbone restraints lowered to 50 kJ mol-1 nM-2 . All the restraints were removed for the production runs at 300 K . A set of three production replicas of 300 ns each were performed for the apo and holo forms . In the NVT simulations temperature was controlled by the Berendsen thermostat [49] with coupling constant of 0 . 2 ps , while in the NPT simulations the V-rescale thermostat [50] ( coupling constant of 0 . 1 ps ) was used and pressure was set to 1 bar with the Parrinello-Rahman barostat [51] ( coupling constant of 2 ps ) . A time step of 2 . 0 fs was used , together with the LINCS [52] algorithm to constrain all the bonds . The particle mesh Ewald method [53] was used to treat the long-range electrostatic interactions with the cutoff distances set at 12 Å . The dynamics of the HIF-2α:ARNT dimer both in the unbound and 0X3-bound forms was investigated to characterise the flexibility of the inter-domain interfaces . Global structural changes during the simulations were monitored by Root Mean Square Deviation ( RMSD ) from the initial structure . RMSD values were calculated after best fit superposition on the protein Cα atoms . Average per-residue flexibility was measured by RMSF of the atomic positions . RMSD and RMSF values were calculated for the protein Cα atoms using the R [54] Bio3D package [55] , both for complete dimer and core domains excluding linkers and loops . All RMSF values were computed on a trajectory obtained concatenating the three replicas , excluding the first 50 ns of each simulation . Secondary structure attribution was done with DSSP [56] . Cluster analysis of the inhibitor geometries in the binding pocket was performed using the GROMOS nearest neighbour algorithm [57] implemented in GROMACS analysis tools , after fitting on the Cα atoms of HIF-2α PAS-B domain . The occurrence during the simulations of water-mediated interactions between HIF-2α residue S304 and 0X3 ligand nitro group was evaluated using GROMACS analysis tools . H-bond detection was done with two sets of thresholds , in agreement with GROMACS ( donor-acceptor distance < 3 . 5 Å and hydrogen-donor-acceptor angle < 30° ) and HBplus ( donor-acceptor distance < 3 . 9 Å and hydrogen-donor-acceptor angle < 90° ) [58] criteria . MD simulations of the unbound and 0X3-bound HIF-2α:ARNT dimers were also analysed to identify the role of domains and secondary structures in the dimer stability . The binding free energy of dimer formation was estimated using the Molecular Mechanics Generalized Born Surface Area ( MM-GBSA ) method [59 , 60] , implemented in the AMBER software package [61 , 62] . In this method , the ΔGbinding is obtained as the sum of energy and configurational entropy contributions associated with complex formation in the gas-phase and the difference in solvation free energies between the complex and the unbound monomers . The configurational entropy of the solute can be estimated using various approximations , but its determination remains a challenging task and usually is neglected [63] . In this work , ΔGbinding is determined omitting the entropic term and therefore it is referred to as dimerization energy . The method includes an implicit solvent model . The polar solvation term was approximated with the Generalized Born ( GB ) model [64] using OBC re-scaling of the effective Born radii [65] . The non-polar solvation term was calculated as the product of the surface tension parameter and the solvent accessible surface area ( SA ) evaluated using the Linear Combination of Pairwise Overlap ( LCPO ) algorithm [66] . The single-trajectory approach [63] was used . In this approach both monomer conformations for the calculation were obtained from the dimerized state MD simulation instead of performing distinct simulations of the three different states ( monomeric ARNT , monomeric HIF-2α , and bound state ) . MM-GBSA calculations were performed on a subset of conformations from the equilibrated part of the MD simulations . For this purpose , the domain contributions were calculated at 100ps frequency on the stable interval ( 100–300 ns ) of each replica and each energy component was determined by averaging over the contributions from all the conformers . Single interfaces were analysed using a per-residue energy decomposition . For this purpose , a common ensemble of conformations sampled in all three replicas was identified in the principal component subspace of inter-domain motions , calculated on the subset of Cα at domain interface ( S1 Fig ) . In this analysis , a residue of a domain A was considered within the A-B interface if at least one of its atoms was found within 3 . 5 Å from an atom of the B domain in at least 10% of the simulation . To identify protein segments with correlated atomic motions , a correlation network analysis [67] was performed using Bio3D [55 , 68] . Pairwise residue cross correlation coefficients were calculated from the displacement of the Cα atom pairs [69] . A weighted graph was generated from the cross correlation matrix , in which each residue represents a node and an edge is drawn when the absolute correlation between two residues is greater than 0 . 4 . Edges with positive weights connect residues with correlated motion , while negative weights describe anti-correlated motions . Shortest and suboptimal path analysis [68] , conducted on the 50 shortest detectable paths , was used to highlight differences in interdomain communication in the apo and holo states . The obtained network contains substructures of nodes , called communities , that are more densely interconnected to each other than to other nodes in the network . This community structure was detected using the random walker algorithm [68] on the edges with absolute correlation values greater than 0 . 5 . The role of residues identified as important for dimer stabilisation or inhibition by computational analysis was validated by comparison with point mutations reported in the literature . Two sets of mutations were considered: 1 ) selected mutations with experimentally verified effects; 2 ) missense mutations from whole-exome sequencing analysis of different tumor tissues annotated in the COSMIC v83 database [70] for HIF-2α and ARNT . The impact of the mutations in the latter set was predicted using two online webservers: Polyphen2 [71] and SIFT [72] . These webservers estimate the impact of amino acid substitutions based on degree of conservation , physical and evolutionary properties . Mutations were considered ‘benign’ if indicated as such by either one of the two webservers , ‘probably damaging’ if Polyphen2 score was greater than 0 . 95 and SIFT score was between 0–0 . 01 with a Median Information Content ≥ 2 . 6; in all other cases mutations were considered ‘possibly damaging’ . Sequence alignment of ARNT and HIF-2α was obtained by Clustal Omega [73] and visualised using the ESPript server [74] . Protein structure representations were generated with PyMol [75] .
Domains and secondary structures for the two protomers of the HIF-2α:ARNT dimer [19] are presented in Fig 1A . Each protomer includes three domains ( bHLH , PAS-A and PAS-B ) . An overall view of the complex structure , including the modelled loops , is shown in Fig 1B . The dimer has a compact core region formed by ARNT-PAS-A , HIF-2α-PAS-A and HIF-2α-PAS-B domains; this is connected to the bHLH region of both the dimerization partners on one side and the ARNT-PAS-B domain on the opposite side . ConSurf conservation profiles highlight highly conserved patches on the bHLH and PAS core domains , especially for the residues lying at the dimerization interfaces ( Fig 2 and annotated sequences in S2 Fig ) . As expected , the most conserved region is the bHLH domain responsible for DNA binding , while loops are generally poorly conserved . It should be noted that most of the modelled loops belonging to the PAS-A domains resemble structural embellishment with a typical Ω-loop shape and no expected functional role . On the contrary , the medium to high conservation observed for many residues belonging to the ARNT-PAS-A FG loop , the HIF-2α-PAS-A GH loop and the HIF-2α-PAS-A/PAS-B linker ( S2 Fig ) suggests that these elements may have a functional role . Also the C-terminal linker of the HIF-2α PAS-B , including a loop and a short α-helix and inserted into the PAS-B:PAS-B interface , shows some conserved residues ( S2 Fig ) , suggesting this connecting element may give a contribution to dimerization . The RMSD plots of the core domains and complete system ( S3 Fig ) show well equilibrated trajectories after 50 ns . High flexibility in the PAS-A loops is evident from the root mean square fluctuation ( RMSF ) plot of the concatenated trajectories for the complete system ( S4 Fig ) , while the bHLH regions show enhanced flexibility due to their terminal position and lack of DNA interactions . As shown in the previous subsection ( see The domain structure of HIF-2α:ARNT dimer ) ARNT PAS-B domain is involved in fewer interactions with the rest of the complex . In all three 300 ns simulation replicas of the unbound dimer we detected a reorientation of ARNT PAS-B and observed that the domain hinge-bending motion around the PAS-A/PAS-B linker spans the same conformational space sampled by different crystallographic structures of ARNT in complex with HIF-2α and other partners ( S5 Fig ) [14] . A’ helices of both PAS-A domains show high flexibility ( Fig 3 ) , especially in ARNT , while other secondary structure elements are generally rigid , with relatively high RMSF values only in correspondence of connective loops . The only exception is in the HIF-2α PAS-B G-strand , which shows a high degree of flexibility in its central residues . This is specific of the HIF-2α PAS-B domain and is not found in the other PAS domains . Moreover , this β-strand is partially unstable and its central region alternates between unstructured ( ≈ 70% ) and folded ( ≈ 30% ) conformations during the simulations ( S6 Fig ) . This instability of the Gβ strand is consistent with the structures in the NMR ensemble of the isolated HIF-2α PAS-B ( PDB ID: 1P97 ) which contains a completely structured G-strand only in 7 out of 20 states ( S7 Fig ) . Interface interaction energies were estimated by MM-GBSA . A summarized view of the relative contribution to the dimerization energy provided by each domain is reported in S2 Table . Values were derived as sum of per-residue contributions averaged over the three replicas of 300 ns . The bHLH domains of the two units equally contribute to the stabilization of the dimer . Due to its central position within the quaternary assembly of the dimer ( see Fig 1B ) , the HIF-2α PAS-B domain highly contributes to the dimerization energy , while the ARNT PAS-B domain only interacts with the HIF-2α PAS-B domain , and seems less important in the dimerization . Notably , the HIF-2α C-terminal linker shows a high contribution to the dimerization energy ( 4 . 1% ) compared to that of the entire HIF-2α PAS-B domain ( 14 . 0% ) . As expected , the dominant role in the dimer association is adopted by the ARNT PAS-A domain ( 21 . 0% ) , that interacts with the bHLH region and with both the HIF-2α PAS domains . A relevant insight arising from the MM-GBSA analysis concerns the importance of ARNT PAS-A FG loop . The 30-residue long loop stands out from the other PAS-A loops for its contribution to dimerization which is at least four times greater than the others ( S2 Table ) . This is due to strong interactions with both the C terminal region of the HIF-2α PAS-A-PAS-B linker and two HIF-2α PAS-B elements , A-strand and C-helix ( both highly conserved ) . Similar to the apo state , the 0X3-bound form has limited flexibility , mostly located in the loop regions . The holo and apo simulations show a quite similar intra-domain RMSF profile for all the domains ( S8 Fig ) . Some differences emerge in the Eα helix of ARNT PAS-B , that is outside the dimerization interfaces , and in the Bβ-Cα region of HIF-2α PAS-A , lying at the PAS-A:PAS-A interface . This latter change in flexibility may indicate a perturbation of protein-protein interactions at this interface . Moreover , significant differences appear on HIF-2α PAS-B domain in the F-helix and G-strand elements , that are in strong contact with the ligand , and thus probably subjected to a local perturbation . Even the HIF-2α PAS-B C-terminal is subjected to rigidification , probably correlated with that of the interacting G-strand . A comparative analysis of the dimerization energy at the interfaces was done using MM-GBSA . The per-residue decomposition of the dimerization energy highlights a weakening of the interactions at the PAS-B:PAS-B interface in presence of the ligand ( S3 Table ) . The most perturbed region involves key residues for the interface stabilization at both the HIF-2α ( from Y278 to T290 ) and ARNT ( from R366 to Y456 ) side ( Fig 4 ) . In details , the perturbation affects the HIF-2α:ARNT electrostatic interactions between E279 and R366 , K253 and E455 , D251 and N448 , as well as the T-stacking between Y278 and F446 . The perturbed region of HIF-2α PAS-B includes residues lying on the E and F helices . As discussed before ( S8 Fig ) , the arrangement of the F-helix appears to be affected by the presence of the ligand , so it is conceivable that this perturbation propagates through the helical bundle , destabilizing the whole PAS-B:PAS-B interface . The central part of the HIF-2α PAS-B G-strand is highly flexible and partially unstructured in the apo simulation , while in presence of the 0X3 ligand it is more rigid and fully structured in a β-strand during the entire simulation ( S6 Fig ) . This region of the G-strand includes residues S304 , G305 and Q306 . S304 is a highly-conserved residue whose side-chain lies within the HIF-2α PAS-B cavity in contact with the ligand . Monitoring of the presence of interactions between S304 and the nitro group of the ligand during the simulations highlighted a stable H-bond network mediated by one or two water molecules ( S9 Fig ) . Overall , interactions occur for about 36 . 4% of simulation time ( 21 . 0% with one water molecule and 16 . 4% with two water molecules ) according to the GROMACS H-bond definition , and 60% of simulation time ( 28 . 4% with one water molecule and 31 . 6% with two water molecules ) according to the HBplus H-bond definition . A set of representative ligand conformations ( Fig 5 ) was extracted by cluster analysis and shows different arrangements of these water bridged interactions . The presence of these interactions looks essential for the complete folding of the G-strand . Analysis of residue correlated motions by means of the distance cross correlation matrix [67 , 69] ( DCCM ) show long-distance effects of this local perturbation ( Fig 6 ) consistent with a lower stability of the dimer in presence of 0X3 . Each domain of the system holds strong internal positive correlation ( except for the long PAS-A loops ) in both apo and holo simulations , confirming the rigidity of all the PAS domains during the simulations . Major differences are evident in the region of ARNT residues 130–180 . In particular , ARNT bHLH-PAS-A linker ( magenta squares in Fig 6 ) has anti-correlated motions towards the ARNT PAS-A domain only in the holo simulation , while in the apo simulation the ARNT PAS-A A’ helix ( green squares in Fig 6 ) is correlated with the HIF-2α PAS-A domain . This suggests that the motion of ARNT PAS-A A’ helix , lying at the PAS-A:PAS-A interface , becomes decoupled from the PAS-A domain after inhibitor binding , probably indicating lower inter-domain interaction . To correlate the altered flexibility of HIF-2α PAS-B G-strand with the perturbed dynamics and correlated motions of ARNT PAS-A A’ helix , we calculated the optimal and suboptimal paths [76] between these two regions from the DCCM networks of the apo and holo simulations ( Fig 7 ) . In the case of the apo network , the shortest paths connect the HIF-2α PAS-B G-strand with the HIF-2α PAS-A strands and then with the ARNT PAS-A A’ helix ( Fig 7 ) . In the case of the holo network , the shortest path is altered . For this system , it links the ligand-perturbed HIF-2α PAS-B G-strand to the ARNT PAS-A , implying a longer connection to the A’ helix ( Fig 7 ) . The change observed in the communication paths of the apo and holo dimers can also be appreciated by comparing the frequency of each residue occurrence in the best fifty suboptimal paths of the two systems ( S10 Fig ) . Several residues in the HIF-2α PAS-A that participate to these paths with high frequency in the apo system simulation ( F168 , F169 , L193-T196 , M225-E227 ) no longer occur in the holo paths . A modification in the relative motion of PAS domains after inhibitor binding is also highlighted by a different community structure in the residue correlation network analysis ( S11 Fig ) . Holo simulations are characterised by the independent motion of each single PAS domain . First , a notable change is visible at the PAS-B:PAS-B interface , where ARNT and HIF-2α residues belong to the same community in the apo simulations , while they are separated in domain-specific communities in the holo simulations . Second , residues of HIF-2α-PAS-A:ARNT-PAS-A interface and the HIF-2α PAS-B G-strand are in the same community in the apo simulations , while there is a clear separation by domain in the holo simulations .
Since the discovery of a large cavity within the PAS-B domain of HIF-2α and the identification of compounds that bind this cavity and dissociate HIF-2α from ARNT [22 , 23 , 26] , several structure-based research programs have been started to find selective and potent antagonists of the HIF-2α transcriptional activity [24 , 25 , 77 , 78] . However , mechanistic understanding of ligand effects on the dimer association had remained elusive until recently , because the available X-ray structures of HIF-2α in complex with artificial ligands encompassed only the isolated HIF-2α and ARNT PAS-B domains . It was first suggested that ligands can induce conformational changes at the PAS-B β-sheet of HIF-2α that weaken the interactions with the ARNT PAS-B β-sheet [24 , 26] , but no evidence in the context of the full dimer was available . Only recently the determination of the crystallographic structures of the entire bHLH-PAS region of the dimer in the apo and 0X3-bound forms [19] has opened the way to a better understanding of the inhibition mechanism . The proximity of 0X3 to the α-helices region of the HIF-2α PAS-B domain , as well as the small perturbation observed in the X-ray structure of the 0X3-bound dimer at this interface , supported a model in which ligand binding could influence the heterodimer stability through a perturbation of its PAS-B:PAS-B interface [19] . However , deeper insight in the atomistic details of this perturbation is limited by the static view provided by the crystallographic structure . Indeed , it is conceivable that 0X3 perturbation could propagate through the structure and affect other interfaces thanks to the dimer intrinsic dynamics and in agreement with a previously suggested allosteric inhibition mechanism [19 , 26] . To investigate this hypothesis , we characterised the evolutionary , dynamical and energetic properties of the dimerization interfaces in the apo and 0X3-bound form . The results shed light on the atomistic details of 0X3 inhibition mechanism , on the residues involved in dimer stabilisation and on pharmacophoric features required for future development of analogues of 0X3 . Our modelling predictions were tested against a wide set of relevant mutagenesis data ( reported in Fig 8 and S4 Table ) . Several experimental mutations that were proved to destabilize the HIF-2α:ARNT dimer and homologous systems were reported in the recent literature [19][14][79][80][81] . Furthermore , it is known that HIFs function can be affected by mutations that have been observed in different carcinomas , brain gliomas , and skin melanomas [82]; in-depth analysis within the COSMIC database , performed in this work and by other Authors [19] indicated that many cancer-related missense mutations are located in the PAS-A-PAS-B regions at the HIF-2α:ARNT interface , thus offering a valid reference point to validate our hypotheses . Our residue conservation analysis ( Fig 2 ) detected high scoring patches on all inter-domain interfaces confirming homomeric and heteromeric interactions in agreement with the crystallographic structure [19] . In addition , our results highlighted strong conservation in some connecting elements , thus suggesting their functional role: the HIF-2α-PAS-A GH loop , which involvement in DNA binding was previously demonstrated [19]; the HIF-2α-PAS-A/PAS-B linker , which participation to the ARNT-PAS-A:HIF-2α-PAS-B interface indicates its role in the dimer stabilization; and the ARNT-PAS-A FG loop that we suggest may be involved in the ARNT flexible arrangement around HIF-2α and different partners . Past mutagenesis and co-immunoprecipitation ( co-IP ) studies indicated that the bHLH:bHLH , PAS-A:PAS-A and HIF-2α PAS-B:ARNT PAS-A interfaces are critical for dimer stability [19] . We assessed this by calculation of the contributions provided by each domain and secondary structure element to the dimerization energy . We confirmed the importance of the bHLH and PAS-A domains of both partners and of the HIF-2α PAS-B domain in the dimer stabilization ( S2 Table ) and we showed that the ARNT-PAS-A domain gives the major contribution to binding . In addition , we highlighted that the dynamic behaviour of the ARNT-PAS-A FG loop enhances the domain intermolecular interactions by wrapping around the HIF-2α PAS-B domain and that the C-terminus of the HIF-2α PAS-B contributes to the PAS-B:PAS-B interface stabilization . Evidences of the above predictions were found in cancer-related mutations located in the ARNT PAS-A FG loop ( in particular , D238N ) and in the HIF-2α faced strands ( R247H , D258N , Fig 8B and S4 Table ) , as well as in the short α-helical region of the HIF-2α C-terminal linker ( S355F , T359A , Fig 8A and S4 Table ) . From MD simulations we detected a general internal rigidity in the PAS domains of both partners , except for the PAS-A loops . This suggests that the dynamics of the system mainly involves quaternary structure oscillations . These motions are particularly evident for the ARNT PAS-B domain , that gives few interactions with the rest of the dimer , and shows characteristic hinge-bending motions around the flexible PAS-A/PAS-B linker . The dynamics of this domain , along with its arrangement ( S5 Fig ) in the crystallographic structures of ARNT in complex with different bHLH-PAS class-I partners ( HIF-1α , HIF-2α , NPAS1 and NPAS3 ) [2 , 3 , 14 , 19] , supports a general model for ARNT dimerization in different heterodimers: strong interactions at the dimerization interfaces in the bHLH/PAS-A region stabilize the dimerization , while the domain bending motion of ARNT PAS-B provides adaptation to different partners through different dimerization geometries in the PAS-B:PAS-B region . We compared the dimer dynamics in the apo and 0X3-bound forms and identified the dimerization interfaces that are mainly affected by ligand binding as well as the ligand-induced perturbations on intra-domain correlated motions and on inter-domain communication paths . The holo dimer has reduced flexibility in the Eα-Fα region of the HIF-2α PAS-B domain and weakened residue interactions in the PAS-B:PAS-B interface ( Fig 4 and S3 Table ) . This behaviour is in agreement with the hypothesis of Wu and co-workers about the involvement of this interface in dimer inhibition , supported by mutagenesis and co-IP experiments on the HIF-2α:ARNT dimer ( ARNT R366A , N448A , Y456D , Fig 8A and S4 Table ) [19] as well as on ARNT dimers with different bHLH-PAS class I partners [19][14] . We found additional confirmation by several cancer-related mutations that lie at both sides of the PAS-B:PAS-B interface ( in particular the HIF-2α mutations S276L and E279V , Fig 8A and S4 Table ) . However , in the holo simulations we also detected a previously undescribed perturbation on the opposite side of the HIF-2α PAS-B domain: the G-strand , which is flexible and partially unstructured in the apo simulation , becomes more rigid and fully structured in a β-strand in the presence of the ligand . This regularisation of the strand is triggered by water-bridged interactions of the 0X3 nitro group with S304 sidechain ( Fig 5 ) . Previous studies on the isolated HIF-2α PAS-B domain have demonstrated that , among a number of artificial ligands , the ones with a nitrobenzoxadiazole group connected to aromatic/heterocyclic rings by a amine linker , like 0X3 , show the highest binding affinities and inhibition potency [24] . While the heterocycle and the nitro group in this molecular moiety were suggested to contribute to high affinity through favourable electrostatic interactions with a few side-chains in the PAS-B binding cavity [24] , no clear explanation was provided for their role in the inhibition mechanism . On the other hand , a critical role of water molecules in the stabilisation of the apo cavity was previously reported [83] , but our insight on the dynamics of the bound form explains for the first time the atomistic details of 0X3 perturbation mediated by water . Here we propose that the local effect of the inhibitor propagates through HIF-2α-PAS-B interfaces with other domains toward the core dimerization region . This is evident in the perturbed flexibility of the Bβ-Cα region of HIF-2α PAS-A ( S8 Fig ) and in the change of correlated atomic motions between HIF-2α and ARNT domains ( Fig 6 ) . The DCCM network analysis showed that a communication path connecting HIF-2α-PAS-B—HIF-2α-PAS-A—ARNT-PAS-A is present in the apo but lost in the holo simulations ( Fig 7 ) . The motion decoupling induced by the loss of this communication is consistent with a weakening of the HIF-2α-PAS-A:ARNT-PAS-A interaction in the A’ helix key region . This is also confirmed by the change in the community structure of the residue correlation networks ( S11 Fig ) from the apo to holo form , where in the ligand-bound state the domains segregate in different communities . Our prediction of the inhibition mechanism can be validated on the basis of several experimental evidences ( Fig 8C and S4 Table ) . The ligand-induced local perturbation of the HIF-2α G-strand directly affects S304 , whose importance was attested by previous mutagenesis experiments reporting that the S304M mutant is unable to bind 0X3 and other similar ligands [24] . Moreover , another residue in the G-strand , Q306 , is known to interact with the proflavin inhibitor that , in a reported crystal structure ( PDBID: 4ZPH ) [19] , is shown to bind outside the PAS-B ligand binding cavity known for the bHLH-PAS proteins [18] . The Q306L tumor-associated mutation , along with others in the adjacent HIF-2α H-strand ( i . e . T321I and G323E , Fig 8C ) and in the faced ARNT PAS-A elements , confirm this region as a key point of ligand perturbation . Noticeably , our MD analysis showed that the H-bonds between S304 and T321 in the holo dimer facilitate the structuring of the G-strand ( Fig 5 ) . Moreover , in our DCCM network analysis a set of residues in the region 168–227 of the HIF-2α-PAS-A domain ( S10 Fig ) were only present in the shortest path of the apo simulations and are expected to be critical to sustain the dimer interaction . Indeed three of them have been already shown to be essential by previous mutagenesis and co-IP studies ( F169D , V192D , H194A , Fig 8C and S4 Table ) [19] . Finally , our prediction of the allosteric destabilization of the PAS-A:PAS-A interaction may be supported both by HIF-2α missense mutations at the PAS-A:PAS-A interface ( in particular , I223M ) and by three point mutations on the ARNT A’ helix that were demonstrated to strongly affect the stability of ARNT dimers not only with HIF-2α ( L167E , I168D , A171D , Fig 8C and S4 Table ) but also with HIF-1α , NPAS1 and NPAS3 partners [19][14] . In conclusion , the results here presented support a model of inhibition by 0X3 where both HIF-2α-PAS-B interfaces are destabilised: the helical bundle side interacting with the ARNT-PAS-B β-sheet , and the β-sheet side interacting with both the PAS-A domains . This latter perturbation allosterically propagates to the PAS-A:PAS-A interaction interface thus destabilizing one of the most important region for dimer association . A critical role in the initial induction of these effects is played by water-bridged ligand-protein interactions . This suggests that , in addition to previously identified features of successful inhibition of 0X3 [24 , 26] , future drug design may be targeted to insert functional groups to stabilise water-bridged interactions with the key residues in the HIF-2α-PAS-B G-strand .
|
A low oxygen condition , called hypoxia , often occurs in tumor masses and generally correlates with worse prognosis . Cells in a tumor react to low oxygen levels with a metabolism modification induced by the activation of hypoxia inducible factors ( HIFs ) through dimerization with a partner protein and binding to a DNA target . Disrupting this protein-protein interaction could be a potential therapeutic strategy , but directly interfering with dimer formation can be troublesome because of the difficulty to design drugs that bind to protein interfaces . However , ligands that bind internal protein cavities can indirectly perturb the interfaces reducing dimers stability . Albeit protein crystallography had offered a detailed static picture of a HIF dimer bound to candidate inhibitors , it is not able to describe either the perturbation caused by binding or the molecular mechanism of dimer destabilization . Here we exploit molecular dynamics to identify the crucial interfaces in the HIF dimer stabilization and , by comparing the results obtained in the bound and unbound forms , we reveal the mechanism of ligand inhibition at atomic detail . All these findings will guide toward the design of improved dimerization inhibitors , to contrast cell survival in tumor masses .
|
[
"Abstract",
"Introduction",
"Methods",
"Results",
"Discussion"
] |
[
"dimers",
"(chemical",
"physics)",
"crystal",
"structure",
"condensed",
"matter",
"physics",
"dna-binding",
"proteins",
"mutation",
"mutation",
"databases",
"crystallography",
"basic",
"helix-loop-helix",
"domains",
"physical",
"chemistry",
"chemical",
"properties",
"research",
"and",
"analysis",
"methods",
"solid",
"state",
"physics",
"dimerization",
"proteins",
"biological",
"databases",
"chemistry",
"physics",
"biochemistry",
"point",
"mutation",
"database",
"and",
"informatics",
"methods",
"protein",
"domains",
"genetics",
"biology",
"and",
"life",
"sciences",
"physical",
"sciences",
"chemical",
"physics"
] |
2018
|
Ligand-induced perturbation of the HIF-2α:ARNT dimer dynamics
|
Many microbes exhibit quorum sensing ( QS ) to cooperate , share and perform a social task in unison . Recent studies have shown the emergence of reversible phenotypic heterogeneity in the QS-responding pathogenic microbial population under laboratory conditions as a possible bet-hedging survival strategy . However , very little is known about the dynamics of QS-response and the nature of phenotypic heterogeneity in an actual host-pathogen interaction environment . Here , we investigated the dynamics of QS-response of a Gram-negative phytopathogen Xanthomonas pv . campestris ( Xcc ) inside its natural host cabbage , that communicate through a fatty acid signal molecule called DSF ( diffusible signal factor ) for coordination of several social traits including virulence functions . In this study , we engineered a novel DSF responsive whole-cell QS dual-bioreporter to measure the DSF mediated QS-response in Xcc at the single cell level inside its natural host plant in vivo . Employing the dual-bioreporter strain of Xcc , we show that QS non-responsive cells coexist with responsive cells in microcolonies at the early stage of the disease; whereas in the late stages , the QS-response is more homogeneous as the QS non-responders exhibit reduced fitness and are out competed by the wild-type . Furthermore , using the wild-type Xcc and its QS mutants in single and mixed infection studies , we show that QS mutants get benefit to some extend at the early stage of disease and contribute to localized colonization . However , the QS-responding cells contribute to spread along xylem vessel . These results contrast with the earlier studies describing that expected cross-induction and cooperative sharing at high cell density in vivo may lead to synchronize QS-response . Our findings suggest that the transition from heterogeneity to homogeneity in QS-response within a bacterial population contributes to its overall virulence efficiency to cause disease in the host plant under natural environment .
Pathogenic bacteria coordinate several social behaviors via production , secretion and perception of diverse diffusible cell-cell signaling molecules by a process called quorum sensing ( QS ) . QS synchronizes the bacteria to perform social task in unison by coordinating production of exo-products as ‘public goods’ that are beneficial to the population as a whole . Such social tasks include the production of virulence associated function-components involved in biofilm formation , extracellular enzymes , extracellular polysaccharide and surfactants that promote motility and spread [1 , 2 , 3 , 4] . The simplified model is that QS coordinates the collective bacterial behavior in unison to maximize the inclusive fitness of individual cells in the community at high cell density , thus avoiding costly production of public goods at low cell density [5 , 6 , 7] . Although QS has been associated with cooperation at high cell density , recent experimental and theoretical modelling studies in pathogenic bacteria such as Pseudomonas , Vibrio and Xanthomonas have demonstrated that QS-response is complex; as bacteria exhibit reversible non-genetic phenotypic heterogeneity in QS-response generating two distinct sub-populations of QS-responsive and non-responsive cells under artificial laboratory conditions [8 , 9 , 10 , 11] . It has been argued that generation of non-genetic phenotypic heterogeneity in bacterial QS-response by stochastic fate determination in an isogenic population may be a bet-hedging survival strategy that enables the population adaptation to fluctuating environmental condition [9] . In other words , heterogeneity in performing social task may have adaptive functions , such as division of labour and sharing of environmental resources . It has been proposed that heterogeneity in QS may arise due to highly sensitive QS-response , which may result in intrinsic stochasticity of QS-response at lower concentrations of auto-inducer [12]; production of sub-optimum level of QS signals even at high cell density [8]; or inherent heterogeneity in QS-response even in the presence of saturating concentration of QS signals [9] . Recent theoretical modelling study suggested that coupling mediated by quorum sensing between ecological and population dynamics can induce phenotypic heterogeneity in a QS experiencing microbial population [11] . Although heterogeneity in QS-response has been studied under artificial laboratory conditions in bacterial pathogens , very little is known about the QS-response dynamics in natural environment particularly inside the host . This raises interesting and significant question regarding the bacterial behavior towards QS heterogeneity for better adaptation in natural host , under fluctuating environmental condition as well as change in cell density . To address the above question , here we investigated whether pathogenic bacteria exhibit QS heterogeneity and its possible role towards social-cooperation and adaptation within its natural plant host , using plant pathogen Xanthomonas campestris pv . campestris ( Xcc ) as a model system that causes black rot disease of cabbage and several other cruciferous plants [13] . In Xanthomonas group of phytopathogens , QS is mediated by the synthesis and perception of fatty acid signaling molecules called DSF ( Diffusible Signal Factor; cis-11-methyl-2-dodecenoic acid ) and its derivatives , which are involved in regulation of expression of several virulence-associated factors such as motility , biofilm formation , production of extracellular polysaccharides ( EPS ) and extracellular enzymes [14 , 15 , 16 , 17] . The phytopathogen Xcc is able to detect its population density through QS via production ( via RpfF; DSF synthase ) and perception ( via RpfC; DSF sensor ) of DSF as quorum signal; where the bacteria significantly depend on QS regulation to coordinate its colonization and infection of plant hosts [2 , 18] . In this study , we visualized the spatial and temporal dynamics of DSF dependent QS-response at the single cell level in the wild-type Xcc during the early and late stage of the disease progression in the host cabbage plant . We have shown that at early stage of disease , the QS non-responsive cells coexist with QS-responsive cells in the wild-type population . However , at the late stage of the disease , the QS-response was more homogeneous . Using single and mixed infection studies in planta with various bioreporter strains of the wild-type Xcc 8004 ( QS performers/responders; that are able to synthesize and sense the DSF to perform QS ) and its QS deficient mutants ( QS non-performers/non-responders/cheaters; that are defective in either synthesis or sensing of DSF and hence unable to perform QS , including QS null ΔrpfF and QS blind ΔrpfC ) , we have shown that inspite of in planta survival of cheaters in the QS-induced Xcc population at the early stages of disease , the QS non-responders ( i . e . QS cheaters ) are outcompeted by the wild-type QS responders at the later stage of disease , as there is significant decline in growth , migration and spread of non-responders , resulting in a more homogeneous QS-response within the quorum induced bacterial population . These results contrast with the earlier studies; those describe that expected cross-induction and cooperative sharing of public goods at high cell density in vivo may lead to synchronize QS-response [4 , 7 , 10 , 18 , 19 , 20] . Here , we argue that under natural condition during bacterial colonization of host plants , the interplay between heterogeneity and homogeneity towards QS-response may provide a stage specific adaptive advantage to the bacterial populations .
Previously we have demonstrated that the wild-type Xcc exhibits heterogeneity in the DSF mediated QS-response in vitro which is a reversible stochastic phenomenon [9] . However , to investigate the detailed dynamics of DSF mediated QS-response in Xcc in vitro as well as more importantly inside the host plant at the single cell level , we engineered a DSF responsive whole-cell QS dual-bioreporter that harbours a gene encoding monomeric red fluorescent protein fused to the DSF responsive promoter eng ( Peng:rfp ) to monitor the QS-response and a constitutive Pkan:gfp marker gene to enable an accounting for all the bioreporter cells in the isogenic bacterial population both in vitro and in vivo . The above Xcc dual-biosensor expressed red fluorescence in response to DSF , and green fluorescence constitutively whose intensity was independent of the amount of DSF produced by it ( S1 Fig ) . To analyze the detailed QS induction dynamics under artificial laboratory conditions , we performed both in vitro confocal laser-scanning microscopy ( CLSM ) and colony forming unit ( CFU ) studies using the DSF responsive whole-cell dual-bioreporter of the wild-type Xcc 8004 in the nutrient rich PS medium , along with its DSF deficient ΔrpfF mutant harbouring the dual-bioreporter construct ( pPkan:gfp-Peng:rfp ) as a QS negative control alone and in the presence of external DSF at optimal level ( i . e . 4 . 84 μM ) separately ( see Materials and Methods ) . At each time period , all the gfp expressing cells were considered for bacterial cell density calculation , but the cells expressing both gfp and rfp were considered for QS induction calculation . The quorum induced average RFP pixel intensity per bacterial cell was represented in arbitrary units ( A . U . ) . The average cell-normalized RFP fluorescence of “Xcc 8004 ( pPkan:gfp-Peng:rfp ) ” as well as “Xcc ΔrpfF ( pPkan:gfp-Peng:rfp ) supplemented with external DSF” increased in a typical density-dependent fashion 12 hr onwards from a initial culture density of ~ 6 × 104 cells ml-1 , with maximum induction ( average RFP pixel intensity ~ 44 A . U . ) occurring between 20–28 hrs of inoculation with approximately 108 to 109 cells ml-1 culture . Analysis of the fractions of induced ( average RFP pixel intensity > 9 A . U . ) and uninduced ( average RFP pixel intensity < 7 A . U . ) cells of Xcc 8004 ( pPkan:gfp-Peng:rfp ) and “Xcc ΔrpfF ( pPkan:gfp-Peng:rfp ) supplemented with external DSF” revealed that the percentage of QS-induced ( RFP+ ) cells increased with time , where only ~ 80–85% cells in the population exhibited QS-induced state even at high cell density ( ~ 109 to 1010 cells ml-1 ) between 24 to 44 hr of growth ( S2 Fig , S3 Fig ) . However , at any sampling point on 20 hrs onward only , the Xcc ΔrpfF ( pPkan:gfp-Peng:rfp ) cells were able to exhibit little RFP fluorescence with average pixel intensity of ~ 3 to 6 A . U . towards minimal promoter activity within population ( S4 Fig ) . These results revealed that Xcc experiences a temporal QS heterogeneity in response to DSF at high cell density in vitro . In parallel , the analysis of distribution of constitutive GFP and DSF responsive RFP fluorescence intensity of at least 100 individual cells for each strain at mentioned sampling times by confocal microscopy revealed the co-existence of both QS-induced and QS uninduced sub-populations with bimodal QS-distribution in the quorum induced populations of Xcc 8004 ( pPkan:gfp-Peng:rfp ) at 24 hr and 36 hr of growth ( S5A Fig ) . With no substantial QS-response in the DSF deficient Xcc ΔrpfF ( pPkan:gfp-Peng:rfp ) population even at sufficiently high cell density ( S5B Fig ) , the bimodal QS-distribution pattern could be restored under similar conditions of growth in the population of Xcc ΔrpfF ( pPkan:gfp-Peng:rfp ) upon initial supplementation with 4 . 84 μM external DSF into the culture ( S5C Fig ) . However , the maximum bimodal gene expression was observed with strongest QS-response ( average RFP pixel intensity ~ 47 A . U . ) within the QS-induced bacterial population of “wild-type Xcc” as well as “its DSF deficient Xcc ΔrpfF supplemented with DSF” at 24 hrs of growth , but not in the QS null Xcc ΔrpfF population alone ( S5D Fig ) . Xcc causes black rot disease of cruciferous plants such as cabbage , cauliflower [13]; where it gains entry inside water conducting xylem vessels of the host plant through natural openings at the tip of the leaf known as hydathodes or through leaf wounds [21 , 22] . During in planta colonization , Xcc group of phytopathogens primarily localize and grow within the vascular regions , and subsequently can escape to surrounding mesophyll regions at the late stage of the infection [23 , 24] . To analyze the detailed in planta localization and growth of Xcc during disease progression , we performed wound infection assays in cabbage leaves with the wild-type Xcc harbouring the DSF responsive dual-bioreporter construct and visualized the in planta distribution patterns for the bacterial population by confocal microscopy . The growth of the wild-type Xcc 8004 harbouring the dual-bioreporter construct ( pPkan:gfp-Peng:rfp ) was similar to the wild-type Xcc 8004 strain alone ( S6A Fig ) . Although , the bacterial localization was found to be comparatively higher within the proximal vascular regions compared to their surrounding mesophyll regions up to 12 dpi , the Xcc 8004 ( pPkan:gfp-Peng:rfp ) population size increased significantly faster in the proximal xylem vessel lumens from the initial population size of 104 cells to 105 cells per leaf within 3 days of incubation as compared to their surrounding mesophyll regions ( S6B Fig ) . We analyzed the DSF dependent QS-response of Xcc population at single cell level or in cell-aggregates inside the host plant by visualizing the initiation of QS-induced rfp expression patterns for wild-type Xcc dual-biosensor cells spanning both vascular and mesophyll regions of the wound inoculated cabbage leaves by confocal microscopy . In planta , both the bacterial populations exhibited no detectable red fluorescence on leaves initially after inoculation ( 0 day post inoculation; 0 dpi ) , but could be readily detected because of their bright green fluorescence . However , heterogeneously QS-induced populations with sufficient amount of QS dependant red fluorophore expression were observed for only wild-type Xcc 8004 dual-bioreporter within both proximal vascular and mesophyll leaf regions on 6 dpi ( i . e . the last dpi before the appearance of characteristics diseased phenotype on the inoculated leaves , considered as “early stage of disease establishment” ) , but not for DSF synthesis mutant Xcc ΔrpfF dual-bioreporter under similar conditions except the background plant autofluorescence of uninfected control cabbage leaves ( Fig 1 ) . These results revealed that Xcc experience the QS heterogeneity in response to DSF at single cell level during early stage of disease establishment within its natural host plant . The detailed confocal microscopic analysis revealed the spatio-temporally regulated QS initiation and distribution within the wild-type Xcc 8004 dual-bioreporter population in planta within the infected cabbage leaves upto dpi 12; spanning both proximal vascular and their surrounding mesophyll area respectively ( Fig 2A ) . The QS dependant expression of red fluorophore was only visible in the wild-type Xcc 8004 dual-bioreporter population from day 3 onwards post inoculation only , where almost half of the population ( ~ 50 . 75% ) was found to be QS-induced producing sufficient amount of red fluorophore ( with average fluorescence pixel intensity of ~ 47 A . U . ) to be detected other than background red fluorescence due to plant chloroplasts and other plant debris . By the 6 days post inoculation , approximately 80% of the Xcc population was found to be QS-induced with higher quorum intensity per cell ( average red fluorescence pixel intensity ~ 59 A . U . ) . However , interestingly at the later stage at dpi 12 , almost all cells of the Xcc population were found to be QS-induced ( ≥ 98% ) with much stronger QS induction per cell ( average red fluorescence pixel intensity ~ 68 A . U . ) compared to the in vitro QS-response in culture ( Fig 2B ) . Further study to understand the spatial distribution of Quorum response along with Xcc localization in both proximal vascular and its surrounding mesophyll regions indicated a much earlier Quorum response within xylem vessels with approximately 65% induced population on 3 dpi as compared to a late Quorum response within surrounding mesophyll extracellular regions with approximately 45% induced population 4 dpi . Xcc maintained a significantly larger quorum size within vascular regions compared to mesophyll regions with maximum heterogeneity in QS-response during early stage of disease progression upto 6 days post inoculation . However , more homogeneously QS-induced populations with more than 98% quorum induced cells were observed during late stage of disease progression from 6 dpi onwards in both the regions ( Fig 2C and 2D ) . In parallel , the confocal microscopy analysis towards distribution of constitutive GFP and DSF responsive RFP fluorescence intensity for at least 100 representative individual cells from each strain on mentioned sampling days {i . e . dpi ( s ) 1 , 6 and 12} revealed the co-existence of both QS-induced and QS uninduced sub-populations with a typical and strong bimodal QS-distribution ( in ~ 81:19 ratio ) in the quorum induced heterogeneous population of wild-type Xcc 8004 ( pPkan:gfp-Peng:rfp ) on dpi 6 at early stage of disease establishment , but a comparatively weaker bimodal QS-distribution within the strongly quorum induced population dominated largely with QS responders over QS non-responders ( in ~ 96:04 ratio ) on dpi 12 at late stage of disease establishment ( Fig 3A and 3B ) . During maximum bimodal gene expression in the heterogeneously QS-induced Xcc population on dpi 6 , the bacteria exhibited a strong QS-response ( average RFP pixel intensity ~ 49 A . U . ) similar to in vitro . However , the average QS intensity was comparatively stronger ( average RFP pixel intensity ~ 64 A . U . ) with a weaker bimodal gene expression in the wild-type Xcc population on dpi 12 , unlike in vitro . As a QS negative control , the DSF deficient QS null Xcc ΔrpfF population was unable to exhibit any QS-response and bimodal gene expression alone in the plant host upto dpi 12 ( Fig 3C ) . In addition to the QS induction and localization studies with the wild-type Xcc dual-bioreporter strain , we also observed the role of DSF dependent QS-response towards spatio-temporal bacterial localization in planta using two other previously reported biosensor strains; β-glucuronidase ( GUS ) assay with DSF dependent QS-responsive GUS reporter strain of wild-type Xcc 8004 {Xcc 8004 ( pLAFR/Peng:gusA ) } and confocal microscopy with the wild-type Xcc harbouring a DSF responsive gfp reporter {Xcc 8004 ( pKLN55/Peng:gfp ) } . Analysis of GUS and GFP expression pattern in planta further corroborated the heterogeneity in the DSF dependent QS-response and the pattern of bacterial colonization; wherein , the Xcc cells colonizing the xylem vessels exhibited initiation of QS induction followed by the escape of QS-induced Xcc cells to the surrounding mesophyll region ( S7 Fig ) . Our in planta GUS study gives a clear idea that QS induction happens to be initiate within vascular region followed by the distribution of the QS-induced bacterial population spanning both vascular and their surrounding mesophyll regions at different stages of infection . To understand the distribution of QS-induced and uninduced cells within the bacterial aggregates or microcolonies of different size inside the area of colonization , we analyzed the percent of QS induction within different bacterial aggregates spanning both vascular as well as the surrounding mesophyll region by taking 24 different representative Xcc dual-bioreporter aggregates at different time points post infection . Initially , even cell aggregates as large as 100 cells did not exhibit QS induction till 2 dpi within vascular regions and upto 3 dpi within mesophyll regions , as measured by red fluorescence indicative of DSF dependent QS induction . The inductions of QS-response in individual cells in the aggregates were more evident in the vascular and mesophyll region after 2 and 4 dpi , respectively . The percent of QS induction was much higher in similar sized bacterial aggregates within vascular region compared to the surrounding mesophyll area . Relatively , smaller aggregates were able to get auto-induced within vascular regions from 3 dpi onwards . The cell aggregate size at which QS induction was observed was progressively smaller . A more heterogeneously distributed quorum response was observed within almost similar aggregate sizes from 3 to 8 dpi in the vascular region and from 4 to 10 dpi in the mesophyll region , which thereafter became more homogeneous gradually upto 12 dpi with almost 100% induction in larger aggregates at a given dpi ( Fig 4 ) . Despite several earlier reports describing high fitness of Gram-negative QS non-responders in animal pathogens [25 , 26 , 26] , some recent reports argue that the spatial structure , occurrence of well separated microcolonies of wild-type and QS non-responders in early-stage infections in vivo may limit sharing of public goods by QS null mutants which may limit mutant fitness [19 , 27] . However , in our QS induction and localization studies in the wild-type Xcc indicated no spatial structures that could limit sharing of public goods , as both QS-induced ( responsive ) and uninduced ( non-responsive ) cells were localized together in similar size cell aggregates or microcolonies . We therefore wanted to address whether the QS non-responders have fitness disadvantage at late stage of infection or there is excess sharing of QS signal within the population which could results in a homogeneous QS-responding population at the late stage of disease establishment in host plant . We performed in planta competition assays with individual and mixed ( in 1:1 ratios ) inoculums separately using different constitutive reporter cells of wild-type Xcc 8004 ( i . e . QS responders; able to produce and sense DSF ) , Xcc ΔrpfF ( i . e . QS null mutant; defective in DSF synthesis but able to sense DSF ) and Xcc ΔrpfC ( i . e . QS blind mutant; hyper-producer of DSF but defective in DSF sensing ) harbouring either a constitutive gfp ( or ) mCherry marker gene , to elucidate the QS-response benefits towards survival fitness among QS+ versus QS- cells of Xcc population at different cell densities within host plant ( see Materials and Methods ) ; where the bioreporter cells were observed under a CLSM along with in planta CFU assay to analyse the growth/localization , migration , survibility and cell-aggregate formation patterns for each bacterial population within the host plant during different stages of disease establishment . The preliminary CLSM analysis for in planta competition assay with single and co-cultures of wild-type Xcc 8004 bioreporters expressing either constitutive Pkan:gfp or Plac:mCherry ruled out any significant fitness difference due to different constitutive marker genes under differential promoter activities upto dpi 12 , indicating the similar survival fitness of both the reporter strains within 1 cm2 proximal green regions of infected cabbage leaves inspite of harbouring different reporter constructs ( S8 Fig ) . Analysis of single and mixed infections ( from CLSM studies ) with the constitutive reporter strains of QS proficient wild-type Xcc 8004 , its QS deficient ΔrpfF and ΔrpfC mutants indicated the reduced fitness in case of both the QS deficient ΔrpfF and ΔrpfC populations compared to QS proficient wild-type Xcc 8004 population during disease establishment in planta . However , in the mixed infection studies , both the QS null ΔrpfF and QS blind ΔrpfC mutants exhibited a significant reduction in fitness at later stage of disease establishment ( 12 dpi ) ( Fig 5A and 5B , S9A and S9B Fig ) . In the mixed infections , the wild-type QS responders outcompeted both the QS null ΔrpfF as well as QS blind ΔrpfC mutants towards in planta growth/localization separately , to exhibit a typically heterogeneous population with approximately 20% QS non-responders on 6 dpi and a more homogeneous population with only approximately 2% QS non-responders on 12 dpi for each combination ( Fig 5C , S9C Fig ) . Furthermore , our in planta competition assay by CFU analysis for the above bacterial populations isolated from surface sterilized infected cabbage leaves on dpi ( s ) 1 , 6 and 12 further corroborates with our in planta competition assays by CLSM analysis , indicating the significant QS-response benefits towards the survival fitness in case of QS proficient wild-type Xcc 8004 population as compared to its QS deficient ΔrpfF and ΔrpfC mutant populations on dpi ( s ) 6 and 12 ( S10 Fig ) . The analysis of frequency distribution of QS+ and QS- bacterial populations in the in planta competition assay indicated sufficiently higher population size in case of QS+ wild-type Xcc 8004 ( i . e . ~ 5 to 10 fold higher ) compared its QS null ΔrpfF and QS blind ΔrpfC mutants on dpi ( s ) 6 and 12 . The average population size of wild-type Xcc 8004 bioreporter cells reached quickly upto ~ 3 × 106 cells per cm2 leaf region to develop the disease symptoms within the inoculated leaves on dpi 6 , and also attended maximum population size of ~ 107 cells per cm2 leaf region within infected cabbage leaves as compared to the populations of its QS null ΔrpfF ( i . e . ~ 2 × 106 cells per cm2 leaf region ) and QS blind ΔrpfC mutants ( i . e . ~ 1 . 5 × 106 cells per cm2 leaf region ) on dpi 12 . However , the population sizes for QS null ΔrpfF and QS blind ΔrpfC mutants were found to be significantly reduced in the presence of QS+ wild-type Xcc 8004 cells at the late stages of disease establishment ( i . e . from dpi 6 to 12 ) in the co-infected cabbage leaves ( Fig 6A , S11A Fig ) . Further , analysis of frequency distribution of QS+ and QS- bacterial populations was carried out to understand the QS benefits towards in planta migration and spatio-temporal regulation of population size in Xcc . On specified sampling dpi ( s ) , analysis of population distribution patterns within proximal , middle and distal regions of inoculated leaves with single cultures revealed significantly higher population size per cm2 infected leaf region in case of wild-type Xcc compared to the QS null ΔrpfF and QS blind ΔrpfC mutant populations from dpi ( s ) 6 ( ~ 5–7 folds higher than ΔrpfF , ~ 7–11 folds higher than ΔrpfC ) to 12 ( ~ 4–5 folds higher than ΔrpfF , ~ 5–7 folds higher than ΔrpfC ) . However in the mixed infections , there was a drastic reduction in the population size for both the QS null and QS blind mutants in the presence of QS-responding wild-type Xcc ( ~ 23–33 folds lower for ΔrpfF , ~ 43–62 folds lower for ΔrpfC ) on 12 dpi . Detailed analysis indicated a drastic reduction in bacterial population size in case of QS blind ΔrpfC mutant population compared to both QS-responsive wild-type Xcc 8004 as well as QS null Xcc ΔrpfF mutant populations spanning all the proximal , middle and distal regions upto dpi 12 , and the bacterial population size was found to be maximum and minimum within proximal and distal regions respectively ( Fig 6B , S11B Fig ) . To understand the QS-regulated cell aggregate formation in planta during disease establishment , we have also analyzed the frequency distribution of aggregate numbers as a function of time in the single and mixed infections . Confocal microscopy of infected cabbage leaves at late stage of disease establishment indicated the presence of significantly higher no . of larger bacterial aggregates in the bioreporter populations of wild-type Xcc 8004 as compared to its QS null ΔrpfF and QS blind ΔrpfC mutants within proximal vascular regions on dpi 12 ( Fig 7A ) . On specified sampling dpi ( s ) , analysis of aggregate formation patterns within proximal , middle and distal regions of inoculated leaves with single cultures revealed significantly higher no . of aggregate formation per cm2 infected leaf region in case of wild-type Xcc compared to the QS null ΔrpfF and QS blind ΔrpfC mutant populations from dpi ( s ) 6 ( ~ 5 folds higher than ΔrpfF , ~ 24–38 folds higher than ΔrpfC ) to 12 ( ~ 2 . 5 folds higher than ΔrpfF , ~ 18–21 folds higher than ΔrpfC ) . However in the mixed infections , there was a drastic reduction in the number of bacterial aggregates for both the QS null and QS blind mutants in the presence of QS-responding wild-type Xcc ( ~ 10–14 folds lower for ΔrpfF , ~ 53–67 folds lower for ΔrpfC ) on 12 dpi ( Fig 7B , S12A Fig ) . Detailed analysis indicated a drastic reduction in the no . of bacterial aggregates in case of QS blind ΔrpfC mutant population compared to both QS-responsive wild-type Xcc 8004 as well as QS null Xcc ΔrpfF mutant populations spanning all the proximal , middle and distal regions upto dpi 12 , and the frequency towards larger aggregate formation for each population was found to be maximum and minimum within proximal and distal regions respectively ( Fig 7C , S12B Fig ) . Furthermore , we analyzed the frequency of distribution of bacterial aggregate size as a function of time in the single and mixed infection studies . Larger aggregates were observed for the wild-type Xcc population rather than the QS null and QS blind mutants from 6 dpi onwards . Between 6 to 12 dpi , significantly higher number of large size bacterial aggregates was observed for wild-type Xcc population within the proximal regions , however , on 12 dpi , more number of solitary and comparatively smaller size bacterial aggregates was observed within the middle regions as compared to proximal regions . Both , QS null ΔrpfF as well as QS blind ΔrpfC mutant populations exhibited a significant reduction in the number as well as in size of bacterial aggregates and migration in the presence of QS proficient wild-type Xcc 8004 population on 12 dpi within the proximal , middle and distal region from the point of inoculation in the cabbage leaves ( Fig 8 , S13 Fig ) .
Quorum sensing plays an important role in the virulence of several plant and animal pathogenic bacteria by coordinating the production of different sets of virulence associated factors via synchronizing gene expression , in a density dependent fashion [1 , 2 , 3 , 4] . However , an increasing body of research suggests that bacteria exhibit non-genetic phenotypic heterogeneity in the QS-response within the isogenic bacterial population under homogeneous laboratory culture conditions [8 , 9 , 10 , 11] . However , little is known about the nature of phenotypic heterogeneity in the QS-response and its role in cooperative behavior in natural environment such as inside the host . Pathogenic bacteria depend quite significantly on QS regulation to coordinate their colonization and infection of plant hosts [2] . DSF family mediated QS-response regulates diverse virulence factors towards Xanthomonas virulence in natural host plant [15 , 16] . Within host plant , initially the pathogen at a low cell density escapes the host immune system by not performing QS . However , upon achieving a certain cell density the bacterial population activates the QS circuit to maximize its in planta fitness via exhibiting stochastic phenotypic heterogeneity within the host plant [14 , 15 , 19] . Recent in vitro studies of QS-response at the single cell level in Xanthomonas campestris pv . campestris and Pseudomonas syringae have indicated that bacteria maintain QS-responsive and non-responsive sub-populations in a ~ 80:20 ratio independent of their origin , even at high cell density and in the presence of excess of exogenously supplemented QS signal . The mixed motility assay indicated that the presence of both responding and non-responding cells could serve as a bet hedging strategy , thus promoting QS-responsive cells for more spread inside the vessel and non-responsive cells to utilize local host resources [9] . However , it was unclear whether the inherent stochastic heterogeneity in the QS-response exhibited under laboratory condition is influenced by change in environmental conditions , and whether there is selection pressure to cooperate under natural conditions particularly in host-pathogen interaction . In the present study , we have now added a detail statement about the lifestyle of the pathogen in which QS-regulated virulence associated functions are involved in adaptation of different stages of infection in its host plant . Our in planta results indicated that the plant pathogen Xcc exhibits heterogeneity in the QS-response with bimodal QS distribution in its population at early stage of disease establishment , with the occurrence of both responding and non-responding cells . In contrast to earlier studies , the studies presented here argue that heterogeneity in QS-response is not due to the lack of cross-induction which may arise due to spatial structures that could limit sharing of public goods such as the QS signal among the members of the community , as QS-responsive and non-responsive cells coexist together in similar size aggregates or microcolonies inside the host plant [10 , 19] . However , during the later stages of the infection , the wild-type Xcc exhibited a synchronized homogeneous QS-response with almost all viable cells to be QS-induced state . In this study , we have shown that the QS-response benefits towards in planta survival fitness of QS responders over QS non-responders as a potential regulator to interplay between heterogeneity and homogeneity towards QS-response within Xcc population at sufficiently high cell density under nutrient scarce conditions inside the host plants . Based on our recent results , here we argue that this interplay between heterogeneity and homogeneity towards QS-response inside the host plants could provide a stage specific adaptive advantage to the bacterial populations towards successful utilization of environmental resources , which in turn helps them to adapt to changing environmental condition . Our current QS induction response dynamics studies indicated towards the existence of bimodal QS distribution with the heterogeneously QS-induced Xcc populations in vitro ( S2 and S5 Figs ) as well as in planta ( Figs 1 , 2 and 3 ) . The quorum size within Xcc aggregate was highly influenced by the aggregate size spatio-temporally within host plant leaves . The existence of QS heterogeneity even within larger bacterial aggregates ( ~ 103 cells per aggregate ) during 3rd to 10th dpi indicates that , the QS non-responders are unable to share the QS benefits in presence of QS responders without any spatial restriction for QS distribution in planta ( Fig 4 ) . In our dual-bioreporter based in planta studies ( Figs 2 , 3 and 4 ) , the absence of QS non-responders in wild-type Xcc dual-bioreporter population at the late stage of infection ( on dpi 12 ) indicated towards the non-sharing of QS benefits towards social co-operation under unfavorable conditions , such as nutrient limitation in natural host plant . The inability of QS non-responders of Xcc to exhibit the QS-response even in presence of excess signal under artificial laboratory conditions [9] discards the possibility of QS-response by those non-responders in the homogeneously QS-induced population at late stage of infection at high concentration of quorum signal in planta . Hence , we hypothesized that the non-sharing of QS benefit could be driving force towards the selective fitness of QS responders over QS non-responders under unfavorable plant host environment . To further prove this hypothesis , an in planta competition assay was performed using single constitutive reporter strains ( Fig 5 ) , where we have used the QS-responsive wild-type Xcc along with its ΔrpfF ( DSF null ) and ΔrpfC ( DSF blind ) mutants as QS negative controls . It is known that QS mutants ( ΔrpfF and ΔrpfC ) in Xcc are growth deficient with compromised fitness as compared to the QS performing wild-type cells in planta [28] . Through our in planta competition assay , we wanted to find out whether the QS benefits can be shared by QS mutants in the presence of QS performers ( without any spatial restriction towards QS distribution as mentioned in earlier studies [11 , 19] to rescue their in planta fitness . In earlier studies , the QS mutants of Xylella fastidiosa and Ralstonia solanacearum exhibit significant fitness defects in associating with their insect and plant hosts respectively [29 , 30] . However , our recent in planta competition assays with the QS-responding wild-type , its QS null ( DSF synthase ) and QS blind ( DSF sensor ) mutant strains indicated that although QS mutants and wild-type cells co-exist together sharing common micro-niche inside the host plant , both the QS null and blind mutants exhibited significant retardation in growth ( Fig 5 ) , migration and survival ( Fig 6 ) and cell-aggregate formation ( Figs 7 and 8 ) in the presence of wild-type , particularly at the late stage of the disease . This suggests that the declined in fitness of QS non-responders in the presence of QS responders may be the reason for a homogeneously QS-induced population at high cell density during late stages of disease establishment ) in host plant . These results contrast with earlier report where it has been shown that the QS cheats or non-responders have fitness cost in the presence of wild-type due to spatial constrain , as QS non-responders and wild-type cells form well separated and discrete microcolonies inside the host which results in non sharing of public goods [19] . Here , we propose that the pathogen interplays between non-genetic heterogeneity and homogeneity towards QS-response spatio-temporally for their better survival and successful disease establishment in host plant . The idea is that , at the early stage of disease , presumably under nutrient sufficient condition , QS-responsive cells contribute to spread and establishment of systemic infection . The QS non-responsive cells contribute more towards colonization and utilization of resources locally . However during the later stage of disease , presumably under condition of nutrient limitation due to the large increase in bacterial load , bet-hedging may be disadvantageous as the free-loaders share the limited resources . At this stage , QS-responsive cells have growth advantage probably by the production of ‘private goods’ [18] required for survival under these condition ( Fig 9 ) . In other words , we assume that during in planta proliferation , the part of the over-saturated bacterial population lagging behind in migration experiences a severe nutrient scarcity ( referred as “nutrient limitation” ) locally . Under such nutrient limitation , iron [31] , nitrogen and phosphorus sources also get depleted along with total carbon ( as a major nutrient ) locally [32] , and there is a decrease in the availability of such resources for QS non-responders as compared to QS responders within this part of the obove population . As a result , the QS non-responders experience comparatively higher nutrient limitation over a time period and gradually get eliminated out from the population locally . Previously , we have shown that under in vitro laboratory conditions , Xcc exhibit stochastic heterogeneity in QS-response with the distribution of both QS-responsive and non responsive cell approximately in an 80:20 ratio at high cell density . Exogenous addition of excess QS signal DSF did not alter the distribution of QS-responsive and non-responsive sub-populations . The fact that an E . coli strain harbouring the QS-responsive signalling components exhibited unison response in the presence of exogenous QS signal molecule , indicating that the QS-responding bacteria in general exhibit inherent stochastic phenotypic heterogeneity in QS-response [9] . It is likely that under natural conditions; such as later stages of disease progression in planta , there is selective advantage of QS-responsive sub-population as evident in our in planta competition assays , co-inoculated with wild-type Xcc 8004 and either its DSF null ( or ) DSF blind mutants separately . Interestingly , it has been shown that Salmonella typhimurium , a human pathogen exhibited phenotypic heterogeneity in production of virulence factors which are required for host colonization , are expressed in a bistable fashion , leading to sub-populations of virulent and avirulent cells in the population [33] . It has been proposed that the heterogeneity in production of virulence factor functions gives stability to the population as whole , as the non-producers have growth advantage that could limit spontaneous occurrence of cheaters in the population , which could be more deleterious [33] . It is pertinent to note that in Xanthomonas , it has been reported that during stationary phase , extracellular polysaccharide deficient mutants arise spontaneously in the wild-type population due to transposition of the endogenous transposon in the EPS biosynthetic genes [34] . In Xanthomonas , DSF is also involved in the regulation of production of EPS . Therefore , it is possible that maintaining phenotypic heterogeneity in QS-responding population could possibly also limit the spontaneous occurrence of EPS deficient mutants in planta which could possibly affect virulence by the sharing or utilization of recourse produced by the QS-responding population . Taken together , our results indicate that interplays between QS heterogeneity and homogeneity at specific stages of infection maximize the phytopathogenic bacterial population fitness under changing environmental conditions in host plant and hence to cause successful disease establishment .
Xcc 8004 and its derived strains were maintained on Peptone Sucrose Agar ( PSA ) and grown in PS broth at 28°C with 200 rpm , as described previously [35 , 36] . For in vitro QS induction experiment , the exponential phase 1o cultures were sub-cultured into fresh PS broth and grown upto 44 hrs . For all plant infection experiments , the exponential phase 1o cultures were sub-cultured and grown upto 107 cells per ml at 28°C . The Escherichia coli DH5α and its derived strains used for routine cloning were maintained on Luria-Bertani Agar ( LBA ) and grown in LB broth [35] at 37°C with 200 rpm . The concentrations of the appropriate antibiotics used were rifampicin ( Rif; 50 μg/ml ) , ampicillin ( Amp; 400 μg/ml or 100 μg/ml ) , nalidixic acid ( Nal; 50 μg/ml ) and 5-bromo-4-chloro-3-indolyl-D-galactoside ( X-Gal; 25 μg/ml ) . Standard molecular biology and microbiology techniques were employed for generating different Xcc derived reporter strains , as mentioned earlier [31]; where , different transcriptional fusions were constructed by fusing the promoter regions ( Pkan , Peng or Plac ) upstream of a gene of interest , to a fluorescent protein gene ( gfp , rfp , mCherry or gusA ) , and cloned into either pBBR4 ( Pkan:gfp , Peng:rfp , Plac:mCherry ) or pLAFR6 ( Peng:gusA ) plasmids . The DSF responsive dual-reporter strains of Xcc {i . e . Xcc 8004 ( pBBR4/Pkan:gfp-Peng:rfp ) and Xcc ΔrpfF ( pBBR4/Pkan:gfp-Peng:rfp ) } were generated , where pBBR4 [37] harboured both a constitutive gfp marker gene and a DSF regulated rfp reporter gene . Briefly , the Kanamycin promoter ( i . e . Pkan ) sequence ( 137 bp fragment ) was amplified from EZ-Tn5 <KAN-2> Insertion Kit ( Cat . No . EZI982K ) with the forward primer ( with EcoRI ) ; 5’-GCGAATTCCTGTCTCTTATACACATC-3’ and reverse primer ( with SalI ) ; 5’-GCGTCGACAACACCCCTTGTATTAC-3’ . The gfp coding sequence ORF ( 716 bp fragment ) was amplified from pPROBE-GT plasmid with the forward primer ( with SalI and universal rbs sequence before the START codon ) ; 5’-GCGTCGACAGGAGGACAGCTATGAGTAAAGGAGAAGAA-3’ and reverse primer ( with BamHI and STOP codon ) ; 5’-GCGGATCCTCATTTGTATAGTTCATCCATG-3’ . Ligation of 3’ end of Pkan with 5’ end of gfp ORF with the SalI restriction site followed by double digestion at the two ends of the ligated product formed the first constitutive gfp reporter cassette ( a 853 bp EcoRI–BamHI fragment ) , which then was cloned into pBBR4 plasmid creating pBBR4/Pkan:gfp . For the second DSF responsive rfp reporter cassette , the predicted Endoglucanase ( XC_0639 ) promoter ( i . e . Peng ) sequence ( 372 bp fragment ) was amplified from Xcc 8004 genomic DNA with the forward primer ( with XhoI ) ; 5’-GCCTCGAGTCACAAACGACGCGAACA-3’ and reverse primer ( with EcoRI ) ; 5’-GCGAATTCCATGGTGATCTCCCTAG-3’ . The rfp coding sequence ORF ( 675 bp fragment ) was amplified from pDsRed-monomer vector ( Cat . No . 632467 ) with the forward primer ( with EcoRI ) ; 5’-GCGAATTCGACAACACCGAGGACGTCATC-3’ and reverse primer ( with KpnI and STOP codon ) ; 5’-GCGGTACCCTACTGGGAGCCGGAGTG-3’ . Ligation of 3’ end of Peng with 5’ end of rfp ORF with the EcoRI restriction site followed by double digestion at the two ends formed the second DSF responsive rfp reporter cassette ( a 1053 bp XhoI–KpnI fragment ) which then was cloned into pBBR4/Pkan:gfp plasmid construct creating a dual construct ( pBBR4/Pkan:gfp-Peng:rfp ) , wherein the Pkan:gfp and Peng:rfp cassettes were divergent . After further confirming each cassette’s orientation by sequence analysis , the dual reporter construct ( pBBR4/Pkan:gfp-Peng:rfp ) was then introduced into Xcc 8004 and its DSF deficient ΔrpfF strain by electroporation resulting Xcc 8004 ( pBBR4/Pkan:gfp-Peng:rfp ) and Xcc ΔrpfF ( pBBR4/Pkan:gfp-Peng:rfp ) separately . Initial screening for Xcc 8004 ( pBBR4/Pkan:gfp-Peng:rfp ) and Xcc ΔrpfF ( pBBR4/Pkan:gfp-Peng:rfp ) strains were performed by visualizing their optimally grown cultures in nutrient rich PS media ( supplemented with required antibiotics ) for their GFP and RFP fluorescence using confocal laser-scanning microscopy ( CLSM ) ; where GFP was excited at 488 nm and the fluorescence was collected in the range of 505–550 nm ( filter set 38 HE eGFP , Zeiss ) , and RFP was excited at 555 nm and the fluorescence was collected in the range of 582–800 nm ( filter set 20 Rhodamin , Zeiss ) . In addition , the DSF responsive GUS reporter strain Xcc 8004 ( pLAFR6/Peng:gusA ) were also constructed , harbouring DSF responsive gusA marker gene within a stable plasmid pLAFR6 [38] . Construction of the DSF responsive GFP reporter strain Xcc 8004 ( pKLN55/Peng:gfp ) used in this study has been previously described [6] . Other reporter strains Xcc 8004 ( pBBR4/Pkan:gfp ) , Xcc ΔrpfF ( pBBR4/Pkan:gfp ) , Xcc ΔrpfC ( pBBR4/Pkan:gfp ) , Xcc 8004 ( pBBR4/Plac:mCherry ) , Xcc ΔrpfF ( pBBR4/Plac:mCherry ) and Xcc ΔrpfC ( pBBR4/Plac:mCherry ) were generated; harbouring either constitutive gfp or mCherry marker genes in stable plasmid pBBR4 . All the plasmids are low copy number , and were stably maintained during infection . Confocal microscopy as well as GUS assay was performed to screen and demonstrate the reporter expression the newly generated reporter strains in broth cultures . All the newly constructed Xcc bioreporter strains were checked for in vitro and in planta growth . The in planta and/or in vitro growth assay with different bioreporter constructs revelled similar growth pattern with either the GFP , RFP and m-Cherry based reporters as compared to the respective strains of Xcc without those reporter constructs . In vitro GUS assay was performed to screen the GUS reporter strains of Xcc . Briefly , the wild-type GUS reporter strain Xcc 8004 ( pLAFR6/Peng:gusA ) was grown along with Xcc ΔrpfF ( pLAFR6/Peng:gusA ) as its QS negative control strain separately in nutrient rich PS media with the appropriate antibiotics at 28°C and 200 rpm overnight . After appropriate OD normalization , 0 . 2% of primary inoculum for each culture was transferred into nutrient rich fresh PS media and incubated at 28°C and 200 rpm upto 44 hrs . The absorbance at 600 nm and GUS expression were measured at regular time intervals of 12 hrs . GUS expression assays were performed as described previously [39] with a few modifications . Briefly , cells were harvested from 1 ml of culture aliquot by centrifugation at 5000 rpm for 6 min ( New Brunswick Scientific , Innova 43 , Edison , NJ , USA ) for specified time period , followed 0 . 2% NaCl wash of the cell pellet for twice . Pellets were resuspended in 250 μl extraction buffer [50 mM sodium di-hydrogen phosphate ( pH 7 . 0 ) , 10 mM ethylene di-amine tetra acetic acid ( EDTA ) , 10 mM β-mercaptoethanol , 0 . 1% Triton X-100 and 0 . 1% sodium lauryl sarcosine] with added 1 mM MUG ( 4-methylumbelliferyl β-D-glucuronide ) and incubated at 37°C . After a definite time interval ( i . e . 30 mins of incubation ) , reactions were terminated by the addition of 675 μl of 0 . 2 M Na2CO3 into 75 μl of reaction mixture . Fluorescence was measured with 4-methylumbelliferone ( 4-MU; Sigma ) as standard at an excitation wavelength of 365 nm and emission wavelength of 455 nm . GUS activity was presented as nano moles of 4-MU produced per minute . The Xcc 8004 dual-bioreporter strains Xcc 8004 ( pBBR4/Pkan:gfp-Peng:rfp ) and Xcc ΔrpfF ( pBBR4/Pkan:gfp-Peng:rfp ) were grown in the liquid PS broth with the respective antibiotics upto a cell concentration ( i . e . 107 cells per ml ) at which QS induction has yet to occur . 0 . 2% ( v/v ) inoculum of this primary culture was used for the 2o cultures to attain an initial culture density of ~ 6 × 104 cells ml-1 , followed by its incubation at 28°C with 200 rpm upto 44 hrs . For the DSF supplementation , the extracted Xoo DSF as well as commercial DSF ( dissolved in ethyl acetate ) were placed in glass culture tubes , air-dried , and resuspended with fresh PS broth to final concentration of 4 . 84μM ( i . e . optimal concentration; that is the threshold amount of DSF required to phenocopy the wild-type Xcc towards QS induction in its ΔrpfF culture ) separately , followed by addition of the 2o inocula of the Xcc ΔrpfF ( pBBR4/Pkan:gfp-Peng:rfp ) strain . The bacterial cells from 1ml of 2o culture aliquotes were harvested in triplicates at specific time intervals upto 44 hrs , by centrifugation at 5000 rpm for 6 min ( New Brunswick Scientific , Innova 43 , Edison , NJ , USA ) followed 0 . 2% NaCl wash of the cell pellet for twice and re-constitution in sterile PBS ( 1X , pH 7 . 4 ) buffer . Approximately , 8 μl of each sample was mounted on glass slide ( Rohem Industries pvt . Ltd; IS-3099 ) at each at each sampling time and observed using a confocal laser-scanning microscope under 100x/1 . 4 oil DIC M27 objective ( LSM700; Carl Zeiss , Germany ) for the expression of both gfp ( excitation: 488 nm and emissions: 505 to 550 nm band pass , with 518 nm emission maximum ) and rfp ( excitation: 555 nm and emissions: 582 to 800 nm band pass , with 585 nm emission maximum ) reporter genes in wild-type Xcc 8004 along with its DSF synthesis mutant , Xcc ΔrpfF ( with and without initial supplementation with external DSF to the culture ) . Multiple images were acquired using green and red fluorescence and bright field ( DIC ) for each slide . The actual QS-induced RFP fluorescence intensities for both the wild-type Xcc 8004 as well as its ΔrpfF ( supplemented with 4 . 84μM external DSF ) were calculated by subtracting the background RFP fluorescence intensities of Xcc ΔrpfF for basal level promoter expression at mentioned time periods with 4 hrs intervals upto 44 hr of growth . Confocal images for GFP ( green ) , RFP ( red ) and Differential Interference Contrast ( DIC ) were constructed simultaneously using a multitrack mode via Pigtail-coupled solid-state lasers . Outlines of the individual bacterial cells were recognized form the DIC images for each time point . Approximately 400 to 600 cells per sample were analyzed for both GFP and RFP fluorescence patterns ( approximately 70 to 100 cells per field were observed for 5 different fields ) with experimental repeats for at least thrice . Simultaneously , appropriate concentrations of 100 μl sample from the 1 ml culture aliquotes for each strain was dilution plated on the nutrient rich solid PSA medium supplemented with suitable antibiotics to determine the bacterial cell density in terms of CFUs per ml for each culture at each sampling time . At each time point , the samples were observed under CLSM for the constitutive gfp and DSF responsive QS-regulated rfp expression by acquiring multiple images using green and red fluorescence for each strain at different cell densities ( CFU/ml ) throughout their growth in vitro . The 2o cultures of Xcc bioreporter strains were grown to a cell density of 106 cells ml-1 and the bacterial cells in 1 ml of culture aliquot were harvested by centrifugation at 5000 rpm for 6 min ( New Brunswick Scientific , Innova 43 , Edison , NJ , USA ) , and reconstituted in sterile PBS ( 1X , pH 7 . 4 ) buffer . The appropriate bacterial suspensions ( approximately 20 μl per leaf ) of QS uninduced cells were then clip inoculated with the help of sterilized scissors into 40 days old healthy cabbage ( Brassica oleracea ) cultivar ( Super Ball; Indian F1 Hybrid variety ) by gently incising at the apex area of the healthy leaves ( 5–6 leafs per plant , total 6 plants ) . Cabbage plant inoculated with sterile PBS ( 1X , pH 7 . 4 ) buffer was used as a negative control . In order to facilitate the initial survival and growth of Xcc on leaves , the inoculated cabbage plants were placed in plant growth chamber ( Adaptis by Conviron; CMP 6010 ) at 28°C with ambient R . H . ( 65% R . H . ) , where artificial light was maintained for 10 hr periods within the 24 hr post inoculation , and then removed from the chamber and kept under natural condition throughout the experiment . Briefly , 40 days old healthy cabbage plants were infected by clip inoculating the leaves with co-cultures ( in 1:1 ratios , from ~ 106 cells ml-1 2o culture ) of wild-type Xcc 8004 ( expressing constitutive gfp ) in combinations with , either wild-type Xcc 8004 ( or ) its ΔrpfF ( or ) its ΔrpfC ( each one expressing constitutive mCherry ) along with their single cultures separately . In parallel , the obove experimental repeats were also performed with the reciprocal reporter constructs for each reporter strain to rule out the possible differential survival fitness effects due to different reporter constructs . On specified sampling dpi upto 12 days post inoculation , the proximal green regions spanning the mid-rib ( upto immediate 1 cm distance from the clipped site excluding diseased part ) were observed under a CLSM with 100x objective along with in planta CFU assay to analyse the bacterial colonization within the specific host tissue . Inoculated leaves from different cabbage plants ( at least five leaves per plant ) were examined upto 12 days after inoculation to visualize both localization and QS induction ( in case of DSF responsive single and dual-bioreporter strains of Xcc , along with their negative control strains ) , as well as both localization and migration ( in case of constitutive single bioreporter strains of Xcc ) . On each sampling dpi , the leaves were collected immediately prior to sectioning and sample preparation and the transverse sections of leaf slices were observed under a CLSM with 100x objective for its vascular and mesophyll regions . For each inoculated leaf , after excising the diseased part , if present on specific day , the green regions from the clipped end were cut in transverse orientation as proximal , middle and distal region maintaining 1cm width for each region along the mid-rib . From each region , multiple thin transverse sections ( including both vascular and surrounding mesophyll regions ) were hand-prepared with a razor blade with each sections approximately 100 to 150 μm thickness . Multiple sections from different parts of the infected leaves were then mounted on separate glass slides ( Rohem Industries pvt . Ltd; IS-3099 ) and directly scanned under CLSM ( LSM700; Carl Zeiss , Germany ) for the bacterial cells with green and red fluorescence separately indicating the presence of the gfp ( excitation: 488 nm and emissions: 505 to 550 nm band pass , with 518 nm emission maximum ) and rfp ( excitation: 555 nm and emissions: 582 to 800 nm band pass , with 585 nm emission maximum ) marker genes . Multiple Z section scans were acquired at 0 . 5 μm increments for large aggregates in each field . The aggregate size was determined by dividing the area of an aggregate within a single Z section by the area of a single cell , followed by addition of results of all the Z sections spanning the size of the entire aggregate for wild-type bioreporter strains of Xcc . At least three sections were sampled from proximal , middle and distal regions of each leaf with the experimental repeat for thrice independently . The cabbage leaves inoculated with wild-type GUS reporter strain Xcc 8004 ( pLAFR6/Peng:gusA ) were harvested on dpi 1 , 3 , 6 , 9 and 12 and were stained with 1 mM of chromogenic substrate X-Gluc ( 5-bromo-4-chloro-3-indolyl-β-D-glucuronide ) in GUS assay buffer [50 mM sodium di-hydrogen phosphate ( pH 7 . 0 ) , 10 mM EDTA , 0 . 1% sodium lauryl sarcosine , 0 . 1% Triton X-100 and 10 mM β-mercaptoethanol] to determine in planta β-glucuronidase activity . Briefly , each leaf was subjected to vacuum ( 15 psi ) application for 1 hr to facilitate X-Gluc penetration into the infiltrated leaves and then incubated at 37°C for 2 hr [32 , 35] . Subsequently , chlorophyll was completely removed from the stained leaves by incubating in absolute ethanol for 72 hr at 37°C followed by observation under white light by using a bright field stereomicroscope ( SteREO Lumar . V12; Carl Zeiss ) for different blue coloured stained regions for the in planta GUS activity . The experiment was performed with a minimum of five infected leaves per plant for total 5 plants and repeated thrice . CFU assay was performed on the nutrient rich solid PSA plates supplemented with suitable antibiotics to calculate the in planta growth of different bioreporter strains of Xcc 8004 . On specific dpi ( s ) , bacterial CFUs were obtained for 1 cm2 green leaf region proximal to the clip inoculation site from surface sterilized cabbage leaves . The leaves were surface sterilized by dipping in 1% ( vol/vol ) sodium hypochlorite for 2 min followed by three washes with sterile MQ water and then crushed with 1 ml of autoclaved MQ water using sterile mortar and pestle , and further dilution plated at concentrations . After sufficient incubation of the inoculated plates at 28°C , the no . of optimally developed bacterial colonies for each combination were observed for their constitutive gfp and rfp fluorescence under stereomicroscope ( SteREO Lumar . V12; Carl Zeiss ) and finally normalized to CFUs per cm2 proximal leaf region . All the in vitro and in vivo CLSM raw images were analyzed using ZEN lite 2012 ( Carl Zeiss ) software for fluorescence pixel intensity calculation minimizing the background intensity , and FIJI ( Image J ) software for co-localization and final picture brightness correction respectively . The GFP and RFP fluorescence pixel intensities for bacterial cells/populations were represented in Arbitrary Units ( A . U . ) . Statistical comparisons were computed using the Student’s test ( non-parametric , paired , two-tailed test ) as denoted in figure legends ( Prism 5 , GraphPad Software ) . A “p value” of less than 0 . 05 was considered significant .
|
Pathogenic bacteria synchronize and coordinate the production of virulence associated function-components in a density dependent fashion via quorum sensing . In general , QS-response and regulation has been studied under laboratory conditions in vitro , where the QS-responding bacterial population exhibits heterogeneous QS-response with the emergence of both QS responders and non-responders irrespective of their parental kind , as a possible bet hedging strategy . However , very little is known about the dynamics of QS-response inside the host . Using Xanthomonas campestris pv . campestris ( Xcc ) and cabbage as a model plant pathogen-host , we show that there is stage specific interplay of heterogeneous and homogeneous QS-response in the wild-type Xcc population inside the host plant . We show that at the initial stage of the disease , Xcc maintains a stochastically heterogeneous population wherein , the QS non-responders are localized locally and QS-responders contribute to the migration and spread . However at the later stage of disease , the non-responders are outcompeted by the responders , thus minimizing QS signal benefit and in turn maximizing the utilization and optimizing limited recourses in the host . Our findings suggest that the interplay of heterogeneity and homogeneity in QS-response gives a stage specific adaptive advantage in a host-pathogen natural environment .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"fluorescence",
"imaging",
"plant",
"anatomy",
"plant",
"cell",
"biology",
"brassica",
"light",
"microscopy",
"plant",
"science",
"confocal",
"laser",
"microscopy",
"microscopy",
"confocal",
"microscopy",
"plant",
"pathology",
"plants",
"research",
"and",
"analysis",
"methods",
"imaging",
"techniques",
"plant",
"bacterial",
"pathogens",
"leaves",
"eukaryota",
"mesophyll",
"cell",
"biology",
"plant",
"pathogens",
"biology",
"and",
"life",
"sciences",
"organisms"
] |
2019
|
New insight into bacterial social communication in natural host: Evidence for interplay of heterogeneous and unison quorum response
|
HIV-1 set-point viral load—the approximately stable value of viraemia in the first years of chronic infection—is a strong predictor of clinical outcome and is highly variable across infected individuals . To better understand HIV-1 pathogenesis and the evolution of the viral population , we must quantify the heritability of set-point viral load , which is the fraction of variation in this phenotype attributable to viral genetic variation . However , current estimates of heritability vary widely , from 6% to 59% . Here we used a dataset of 2 , 028 seroconverters infected between 1985 and 2013 from 5 European countries ( Belgium , Switzerland , France , the Netherlands and the United Kingdom ) and estimated the heritability of set-point viral load at 31% ( CI 15%–43% ) . Specifically , heritability was measured using models of character evolution describing how viral load evolves on the phylogeny of whole-genome viral sequences . In contrast to previous studies , ( i ) we measured viral loads using standardized assays on a sample collected in a strict time window of 6 to 24 months after infection , from which the viral genome was also sequenced; ( ii ) we compared 2 models of character evolution , the classical “Brownian motion” model and another model ( “Ornstein–Uhlenbeck” ) that includes stabilising selection on viral load; ( iii ) we controlled for covariates , including age and sex , which may inflate estimates of heritability; and ( iv ) we developed a goodness of fit test based on the correlation of viral loads in cherries of the phylogenetic tree , showing that both models of character evolution fit the data well . An overall heritability of 31% ( CI 15%–43% ) is consistent with other studies based on regression of viral load in donor–recipient pairs . Thus , about a third of variation in HIV-1 virulence is attributable to viral genetic variation .
The outcome of infection by the human immunodeficiency virus-1 ( HIV-1 , henceforth “HIV” for simplicity ) is highly variable across individuals , with time to AIDS ranging from 2 years to more than 20 years [1–4] . Quantifying the fraction of this variability explained by genetic variability in the virus is important to our understanding of the mechanisms of pathogenesis and of the evolution of virulence [5] . Most studies have focused on set-point viral load ( SPVL ) , which is a robust predictor of time to AIDS [6] . Following HIV infection , viraemia rapidly increases and reaches a peak before dropping precipitously once HIV-specific cytotoxic T-cells are produced by the immune system [7 , 8] . After this transient peak has passed , about a month after infection , the subsequent viraemia is much more stable [8] ( although it slowly increases over the course of untreated infection [9] ) . This relatively stable value defines the SPVL . The extent to which SPVL is determined by the viral genotype—the heritability of SPVL—is defined as the fraction of phenotypic variance in SPVL attributable to variability in viral genotypes [10–12] . The total variance in SPVL in the population may emerge from genetic variation in the host , variation in the immune response , variation because of measurement error , and variation in viral genotypes . Heritability estimates the contribution of the latter . Heritability will determine the statistical power and necessary sample size to find viral molecular determinants of virulence ( for example , in genome-wide association studies [13] ) . Heritability also determines the rate at which viral populations can evolve in response to selective pressures . For example , it has been suggested that an intermediate value of SPVL maximizes viral fitness , because higher SPVL translates into higher transmission rates but shorter disease duration [14] . The rate at which HIV evolves to this optimal value depends on the heritability of SPVL [15 , 16] . Alternatively , SPVL may change over time not because it is directly under selection but because viral mutations that indirectly affect SPVL are under selection—for example , selection for immune escape or drug resistance . These ideas are not just theoretical possibilities: recent meta-analyses have shown that the population mean SPVL varies over time in many settings [17 , 18] . For example , in Europe , the mean SPVL of individuals who seroconverted at the beginning of the 1980s was 10 , 000 copies/mL , whereas it is 30 , 000 copies/mL for those who seroconverted at the beginning of the 2000s [18] . In contrast , in Botswana , it has been hypothesised that a decline in set-point viral loads was caused by the evolution of costly cytotoxic T lymphocyte ( CTL ) escape mutations [19]; in Uganda , a decline in set-point viral loads was explained by viral adaptation to a low optimal value under a transmission–virulence trade-off [16] . As well as being a predictor of disease progression , viral load is also a strong predictor of transmission [20]: as a result , spatiotemporal variation in population SPVL will affect epidemic trends . Thus , it is of both biological and public health interest to assess if trends in SPVL can be explained by viral evolution . Despite many studies into the determinants of virulence , estimates of the fraction of variability in SPVL attributable to genetic variability in HIV vary greatly . Heritability can be estimated in 2 ways . Firstly , as the regression coefficient of recipient SPVL onto donor SPVL in a set of transmission pairs . This is exactly analogous to the classical parent–offspring regression used in quantitative genetics [10 , 11] . With this method , heritability of SPVL has been estimated to be 33% ( 95% confidence interval 20% to 46% ) in a meta-analysis of studies from sub-Saharan Africa including 433 couples [21–23] ( reviewed in [5] ) . Secondly , heritability can be estimated from SPVL measurements and phylogenetic relationships between viruses ( the “phylogenetic mixed model” [24] ) . This comparative approach is based on the assumption that the covariance between the SPVL of 2 individuals is proportional to their shared ancestry on the viral phylogeny [24 , 25] . This assumption holds if SPVL evolution can be described by Brownian motion ( BM ) ; the phylogenetic mixed model , and methods based on summary statistics such as Pagel’s lambda [26] or Blomberg’s K [27] , were all developed under this assumption . Estimation of heritability of SPVL using such comparative methods have yielded more variable estimates of heritability: from 0% to around 60% in Switzerland depending on the subset of the data [12 , 28]; no significant heritability in Uganda and the Netherlands [28]; and 6% ( CI 3%–9% ) in the United Kingdom [29] . Thus , no consistent picture has emerged from these various reports . Inconsistency between estimates may be caused by genuine differences in heritability across populations , limited sample size , uncontrolled variability in SPVL , or limitations in the methods . Heritability may vary across populations and over time because it depends on the genetic variability present in the viral population . For example , one may expect heritability of SPVL to be larger in sub-Saharan Africa , where the viral population is very genetically diverse and multiple subtypes are present [30 , 31] , than in Europe or North America , where the viral population is dominated by subtype B [32] . Moreover , any uncontrolled source of environmental variation—such as variability in the assays used to quantify viraemia , host variability , or intermittent coinfections increasing HIV replication—decreases heritability . This may explain why heritability was highest in the most homogeneous and strictly defined subset of individuals in the Swiss HIV cohort study [12] . The methods used to measure heritability also have several limitations . Methods based on donor–recipient regression do not rely on assuming a specific model of evolution of SPVL , but they can only be applied to datasets that consist of transmission pairs and thus are not suited to estimating heritability for large populations or in a wide range of settings . In addition , genetic confirmation of these transmission pairs is important to avoid downward-biased estimates of heritability [21] . Phylogenetic methods typically include more data and can be applied to more settings , because they use the SPVL measures of a whole cohort ( not only transmission pairs ) , together with the inferred phylogenetic relationships between viral samples [24 , 25] . This comes at the cost of making specific assumptions on the model of SPVL evolution ( for example , the BM model is often assumed [12 , 29] ) . The BM model can be interpreted as random unconstrained neutral evolution of virulence , defined as an intrinsic but unknown property of the virus that influences SPVL . In this model , when a patient is infected , the patient’s SPVL is determined both by the virulence of the virus and by other external factors , such as host genotype , dynamics of the host immune response , environmental factors , random effects , etc . The change in virulence over a time step Δt is drawn from a normal distribution with mean 0 and stochastic variance proportional to Δt . Thus , changes in virulence are random and independent from one time step to the next , resulting in no sustained directional trend in virulence and a constant increase in the genetic variance of virulence ( and so SPVL ) over time . These features may be considered unrealistic , as multiple selective pressures may act upon HIV virulence [14 , 33] , and directional trends in SPVL are observed in some cohorts [17 , 18] . When virulence evolves under selection , inference of heritability under BM is biased , usually downwards [34 , 35] . Similarly , phylogenetic methods have little power to detect heritability when the evolution of virulence does not follow a BM model [28] . We hypothesised that inconsistency between estimates in phylogenetic studies of SPVL heritability may have been caused by 5 factors: different viral genetic variance between settings , limited sample size , heterogeneous SPVL measures and definitions , additional noise because of imperfectly estimated phylogenies , and inappropriate use of the Brownian motion model of character evolution . To overcome these limitations and understand what factors caused inconsistencies , we measured heritability of SPVL in a large European cohort collaboration ( N = 2 , 028 ) . We obtained blood samples of seroconverters from Belgium , France , the Netherlands , Switzerland , and the UK . Using a cohort of seroconverters allowed us to control for the time since infection , which is associated with viral load because of time trends within patients . We inferred heritability using a classical definition of SPVL , namely the mean ( after a log10 transformation ) of all viral loads measured between 6 and 24 months after infection . Averaging multiple viral load measurements from different time points reduces error variance but may also average out biologically important fluctuations in viral genotype or phenotype within the patient . We also remeasured viraemia using a standardized choice of assay on a single sample taken between 6 and 24 months after infection and before antiretroviral therapy was started ( “gold standard viral load” , GSVL ) , and we reconstructed the viral genome from the same sample . The controlled time window of 6 to 24 months limits variability because of the stage of the infection for both the SPVL and GSVL measures . Additionally , the GSVL measure was defined to limit spurious variation because of variability in assays , to more closely link the viral genotype and phenotype by measuring them from the same sample and to expose variation because of within-host fluctuations in genotype and phenotype that is averaged out in SPVL measures . Henceforth , we use the term ‘viral load’ to mean either SPVL ( as classically defined ) or GSVL . To improve the resolution of our inferred phylogenies , we generated whole viral genomes rather than genotypes based on the polymerase ( pol ) gene as in previous studies . To avoid the limitations associated with an inappropriate model of character evolution , we compared the classical phylogenetic mixed model based on BM to the Ornstein–Uhlenbeck ( OU ) model , which includes stabilising selection on viral load . The OU model has been widely used in comparative biology since its introduction in the 1990s [36] . Recently , the OU model was argued to be more appropriate than the BM model to quantify heritability of viral traits [34] and was applied to several datasets ( [35] and the present study ) . Using a large dataset , carefully measured viral load , whole-genome HIV sequences , and new methods , we showed that the heritability of GSVL is 31% ( CI 15%–43% ) , and the heritability of SPVL is 21% ( CI 10%–36% ) .
We fitted the following 3 stochastic models , presented in order of increasing complexity . ( i ) A null model with only a random component uncorrelated across the tips of the phylogeny , implying zero heritability . ( ii ) The BM model in which viral load evolves randomly along the branches of the tree , leading to unconstrained increases in genetic variance over time—the model most commonly used in comparative studies [24 , 25] . ( iii ) The OU model , which includes a random component similar to BM as well as stabilising selection that brings viral load towards an optimal value and maintains variance at a stochastic equilibrium level [36] ( Fig 1A , Materials and methods ) . Both BM and OU , like all common models of character evolution on phylogenies , assume that evolution occurs continuously on the branches of the phylogenetic tree and make no distinction between within-host and between-host evolution . The consequences of these models of character evolution on viral load can be described as a linear model with a fixed effect representing the expectation of viral load at the tips and a random effect with a covariance structure depending on the phylogenetic tree and the model of character evolution ( Materials and methods ) . We also adjusted for the confounding covariates sex , transmission mode , age , ethnicity , and viral load assay by including them as additional fixed effects in the linear model . If these factors affected the viral load , they would have contributed to the environmental variance in viral load . Failing to adjust for these covariates would have then inflated estimates of heritability if these factors were clustered in the phylogeny . We did not adjust for the covariates “country” and “viral subtype” in the main analysis as they are correlated with viral genotype . Viral load was either GSVL , a measure of viraemia remeasured on the same sample as used for viral genetics using a standardized choice of assay ( Materials and methods ) , or the traditional SPVL measure , which is classically defined as the mean of log10 viral load for samples from multiple longitudinal samples during a defined period after the first HIV-positive test . The phylogenetic tree was computed from whole-genome sequences reconstructed from short-read next-generation sequence data [37] and stripped of positions associated with previously identified drug-resistant mutations [38 , 39] and CTL escape mutations [40] ( Materials and methods ) ( S1 Fig ) . Mutations at these positions are under strong selection and may independently evolve in different branches of the tree , thus leading to incorrect phylogenetic inference . Results were similar when including these mutations ( S3 Table ) . Models in which viral load is partly determined by viral genetic factors evolving along the phylogenetic tree had a significantly better fit than the null model , but there was no strong support for the OU over the BM model ( Table 1 , model comparison based on Akaike Information Criterion ( AIC ) , ΔAIC = –19 . 6 for BM , ΔAIC = –21 . 4 for OU compared to the null model ) . The OU model with stabilising selection was more strongly supported in the set of all subtypes . The maximum likelihood ( ML ) models of character evolution implied heritability for GSVL of 17% [8%–26%] under the BM model and 31% [15%–43%] under the OU model ( Table 1 , Fig 2 ) . Heritability is defined as the fraction of phenotypic variance explained by genetic variance in a given population . More precisely , we estimated broad-sense heritability: the contribution to phenotypic variance of all genetic variation ( including variation generated by epistatic effects between loci ) . This contrasts with narrow-sense heritability , which quantifies only the contribution of additive effects of individual genetic variants [35] . We estimated heritability by resimulating the ML model of viral load evolution on the tree . A simulation attributed a value of viral load to each tip of the tree as the sum of a viral genetic component and an environmental component ( the latter including epidemiological covariates ) . Heritability was the variance of the genetic component ( across tips ) divided by the total variance in the simulations . Because the models of evolution are stochastic , each run gave a different value of heritability; therefore , we reported the mean heritability across 1 , 000 stochastic simulations . We also developed an analytical expression for the expectation of heritability as a function of model parameters that allows faster calculations and proved very accurate ( Materials and methods ) . In accordance with theoretical expectations [35] , assuming a BM model of character evolution led to lower heritability than assuming an OU model . The BM model theoretically results in steady expansion of the genetic variance over phylogenetic time . A SPVL variance similar at tips closer to the root and at tips further away from the root goes against this prediction and leads to a downward bias in the estimated genetic variance and heritability . In contrast , the OU model allows high heritability even with a stable variance ( Fig 1A ) . In our data , epidemiological covariates ( sex , transmission mode , age , ethnicity , assay ) would have inflated heritability by 4% were they not accounted for ( heritability of GSVL without adjustment for these covariates was at 21% and 35% under BM and OU ) , with an effect of sex in particular . This means some of these covariates affected viral load and were clustered in the phylogenetic tree . Specifically , in the subset of subtype B , for GSVL , under the null model , males had +0 . 3 log10 copies/mL higher GSVL than females ( CI 0 . 15–0 . 45 ) ( type II analysis of variance , p = 0 . 0002 ) . Mode of transmission , age , and ethnicity did not have a significant effect in the subset of subtype B ( S4 Table ) , although mode of transmission had a significant effect when considering all participants ( S5 Table ) , with men having sex with men ( MSM ) transmission associated with +0 . 14 [CI 0 . 046–0 . 24] log10 copies/mL higher GSVL compared to heterosexual transmission . The type of assay had an effect on SPVL ( type II analysis of variance , p = 0 . 0019 ) but no effect on GSVL ( p = 0 . 24 ) . This confirms the better standardisation of the GSVL measure . Effects were similar under the BM and OU model . Lastly , including “country” as a covariate in the phylogenetic regression lowered heritability by 5% to 6% ( S3 Table ) . This difference partly represents genuine heritability because the viral genotype is expected to differ by country . We quantified uncertainty on the parameters and on heritability using parametric bootstrapping on the bootstrapped trees . This method combines uncertainty because of finite sample size and uncertainty in the phylogenetic tree inference . For each of the 100 bootstrap trees , we simulated a stochastic outcome of the ML model , reinferred ML parameters from these simulations , and calculated the confidence intervals on these reinferred parameters . Uncertainty on heritability was large for the OU model , ranging from 15% to 43% , in spite of the large dataset ( N = 1 , 581 ) . Heritability of GSVL was higher than heritability of SPVL . This difference , however , was not due to higher environmental variance for SPVL but rather was due to higher genetic variance for GSVL ( environmental variance VE was similar for the 2 measures , but the stochastic variance σ2 describing evolution of the genetic component is higher for GSVL , Table 1 ) . The structuring of the viral population into several subtypes did not contribute much to heritability , as heritability was only slightly higher across all subtypes ( Table 1 ) , and we did not detect any effect of subtype on viral load ( type II analysis of variance , p = 0 . 65 , S5 Table ) . The interpretation of heritability across all subtypes is difficult , as many of the nonsubtype B sequences were recombinants . This means that in the phylogenetic tree of all subtypes , the topology , and branch lengths between subtypes cannot be interpreted in terms of the amount of evolution on a line of vertical transmission , hindering phylogenetic interpretation . Because of the diversity of non-B viruses in our sample , we did not have sufficient data to individually estimate heritability for specific non-B subtypes . We next stratified the analysis by country , sex , and mode of transmission . To first estimate the power to detect heritability in smaller subsets of data , we systematically subsampled the main dataset at random and measured maximum likelihood heritability and confidence intervals as a function of sample size ( S2 Fig ) . We found that accurate and precise estimation of heritability required samples of at least 500 individuals , especially when using the OU model of character evolution . Accordingly , we found no significant heritability in most stratifications of the data . Heritability measured separately within each country was significant only in the Netherlands , at 26% ( 12%–60% ) ( S2 Table , N = 434 , ΔAIC = –12 . 2 for OU compared to the null model ) . In Switzerland , the largest cohort in this study , GSVL was not significantly heritable ( S2 Table , N = 742 , ΔAIC = +2 . 1 for OU compared to the null model ) . This could reflect the limited genetic diversity of our Swiss samples; in other countries , the lack of detected heritability is most likely due to limited power to detect a phylogenetic signal . In males infected by subtype B viruses ( N = 1 , 446 ) , GSVL heritability was 16% under BM and 32% under OU , but heritability was not significant in females ( N = 135 ) . In MSM infected by subtype B viruses ( N = 1 , 196 ) , GSVL heritability was 17% under BM and 30% under OU , but heritability was not significant in injecting drug users ( IDUs ) ( N = 110 ) and heterosexuals ( N = 211 ) . Both BM and OU models fitted the data well . One major difference in the prediction of the 2 models is that in BM , the genetic variance keeps increasing with genetic distance as neutral genetic variation accumulates , whereas in OU genetic variance initially increases and then eventually reaches equilibrium between generation of variation and stabilising selection ( Fig 1A ) . The maximum likelihood BM and OU models both predicted an increasing genetic variance at a rate of +0 . 0024 log10 copies2/mL2/year for BM and +0 . 0011 log10 copies2/mL2/year for OU ( S3 Fig ) . The intermediate increase in variance predicted by the OU model means that the distribution of viral loads has not yet converged to its steady state . In our dataset , the average viral load was constant over time , and phenotypic variance increased over time in GSVL at +0 . 01 log10 copies2/mL2/year but did not significantly increase in SPVL ( S1 Table ) . As well as predicting the variance , the models predict the covariance structure of the viral loads at the tips of the phylogeny . We assessed goodness of fit on the subset of 511 phylogenetic cherries ( pairs of adjacent tips that are each mutually closest to each other ) on the tree of subtype B viruses . In addition to testing goodness of fit , focusing on cherries allows phylogenetic approaches to be compared to the donor–recipient regression that are based on classical quantitative genetics [28 , 35] . Indeed , phylogenetic cherries that are genetically very similar ( separated by a small patristic distance ) on a phylogenetic tree are more likely to be donor–recipient pairs than genetically distant pairs [41] . We computed the Pearson correlation coefficient between viral load values across cherries , stratified by the patristic distance ( the distance between the 2 tips ) ( Fig 1C ) . In the limit in which patristic distance is 0 , the expected value of both correlation coefficients is equal to heritability ( S2 Text , see also [35] ) . For OU , the correlation decreases with patristic distance . For BM , all else being equal , the correlation should in theory be independent of patristic distance ( Materials and methods ) . However , here we see that for the BM model the correlation decreased with patristic distance , because those pairs of tips separated by a large patristic distance tend to also have less shared ancestry . In accordance with the predictions of both models , the correlation coefficient for GSVL of cherries separated by a small patristic distance was around 30% , not far from the predicted heritability of 17% ( under BM ) and 31% ( under OU ) ( Fig 1C ) . The observed negative relationship between the correlation coefficient and the patristic distance resembled closely the prediction from both models ( Fig 1C ) . When using only the phylogenetic information contained in a single gene instead of the whole genome , we found heritability for the OU model was 27% ( 14%–36% ) in the gag gene , 28% ( 16%–39% ) in pol , and 21% ( 11%–39% ) in env ( Table 2 , Fig 3 ) . The lower heritability for each gene independently compared with the whole genome ( h2 = 31% ) was expected . In the limit of no recombination , the phylogenetic history of all genes is the same: thus , with perfect phylogenetic resolution , heritability is the same across genes and equal to total heritability for the whole genome . At the other limit where the 3 genes evolve independently ( linkage disequilibrium between them is 0 ) , the heritability for each gene reflects the contribution of molecular variation at that gene on total variation , and whole-genome heritability is the sum of heritability across genes . Indeed , if the viral load can be written v = ggag + gpol + genv + e where the terms in g denotes the additive genetic contributions of each gene and e denotes the effect of the environment , then fitting a model of character evolution on the gag tree estimates ggag , while gpol + genv is subsumed in e because this quantity is randomly distributed on the gag tree by the assumption of linkage equilibrium . At a finer scale , when inferring the heritability across the genome in 17 overlapping windows that are 1 , 000 base pairs in length separated by 500 base pairs , we found accordingly that heritability was almost always lower than for the whole-genome inference . Heritability was highest around the region where the gag and pol genes overlap and lowest in the region between the pol and env genes , including the vif , vpr , and vpu genes ( Fig 3 ) . Linkage disequilibrium dropped rapidly with genetic distance and was small when the distance was greater than 100 bp ( S4 Fig ) . Confidence intervals reflecting phylogenetic uncertainty ( Materials and methods ) do not overlap between the low- and high-heritability regions . These observations suggest 2 distinct island contributions to heritability , 1 from the low-diversity region coding for the replication machinery ( gag–pol ) and the second from the high-diversity region coding for the envelope gene . Lastly , we investigated the heritability of another viral phenotype , the CD4 cell count slope ( the rate at which the patient’s CD4 cell count declines ) . CD4 slope was not found to be heritable in a previous analysis [12] . We computed the CD4 slope for N = 1 , 476 patients with at least 5 CD4 measures after the date of the first positive HIV test and before antiretroviral therapy . Models in which CD4 slope was partly determined by viral genotype evolving along the phylogeny were favoured over the null model ( Table 1 , ΔAIC = –6 . 2 for BM , ΔAIC = –5 . 7 for OU compared to the null model ) . Heritability was weak , at 11% [0%–19%] in the favoured BM model ( the ML OU model has weak stabilising force , and the inferred heritability was similar to that of the BM model ) . The coefficient of determination of the relationship between CD4 slope and viral load was R2 = 5 . 2% for GSVL , R2 = 7 . 4% for SPVL . This weak relationship is similar to that found previously [42 , 43] and may partly be explained by the noisiness of the CD4 slope measure and the difference between viral load and CD4 count in the blood versus in the whole body .
Using a large dataset of whole-genome HIV sequences ( N = 2 , 028 ) from patients with a known seroconversion date with carefully measured viral load , we established that viral genetic factors account for 20% to 30% of variation in viral load in Europe . This estimate of heritability is consistent with those obtained with donor–recipient regression ( around 30% [5] ) , unlike results of previous phylogenetic studies [12 , 29] . It agrees with a recent analysis performed on 8 , 483 patients with pol sequences from the UK , reporting a lower bound for heritability of SPVL at 25% [35] . We hypothesized that the large variation across previously published phylogenetic estimates could be due to genuine biological differences across cohorts , limited sample sizes , less rigorous selection criteria for patient inclusion leading to heterogeneous viral load measures , less well resolved phylogenies , and/or use of methods based on inappropriate models of character evolution . We found an effect of limited sample sizes and the model of character evolution . However , there was little evidence for genuine biological differences across cohorts included in this data . This study provides several methodological recommendations for the estimation of heritability . Firstly , precise and accurate estimation of heritability required a sample of at least 500 individuals in our cohort ( S2 Fig ) . Limited sample sizes can generate substantial heterogeneities across estimates , in accordance with previous results showing limited power to detect heritability in existing datasets , and large confidence intervals [28] . Secondly , we compared 2 models of character evolution: BM and OU ( the latter including stabilising selection ) . The BM model of evolution was used in most previous studies of HIV viral load heritability , while the OU model was only recently applied to HIV data ( [35] and the present study ) . In this dataset , we found the 2 models were almost equally supported for subtype B , and the OU model was more supported for all subtypes combined . OU implied higher heritability than BM . Support for the OU model was stronger in the set of all subtypes ( Table 1 ) . Because the tree of all subtypes spans a larger phylogenetic distance than the tree of subtype B ( the root-to-tip distance is almost twice as large ) , the BM model fits less well the covariation of viral load between tips and the observed phenotypic variance . In general , we suggest the OU model may better describe viral load evolution for 3 reasons . ( i ) The OU model , with stabilising selection , is a better fit on previously analysed data from the UK [29] and Switzerland [12] , leading to consistent estimates of heritability at about 30% in both cohorts [35] . ( ii ) Data on 56 donor–recipient pairs show that the correlation coefficient between viral loads of the donor and that of the recipient decreases as the recipient viral load is measured later in infection [44] . This is in accordance with the OU model , in which stabilising selection progressively erases the correlation between donor and recipient viral loads . But it is not in accordance with the BM model , in which the correlation coefficient does not depend on the time elapsed between donor and recipient but only on the amount of evolutionary ancestry that they share . ( iii ) The OU model , unlike the BM model , is compatible with the hypothesis that viral load is under a transmission–virulence trade-off favouring those viruses giving intermediate viral loads [14] . Under this trade-off , viral load is under stabilising selection as in the OU model , because very low viral loads ( resulting in very low transmission rates ) and very high viral loads ( resulting in very short duration of infection ) do not allow much onward transmission of the virus , such that viral load cannot drift unconstrained to very low or very high values . Consistent with this hypothesis , we found that the optimal GSVL was 4 . 4 log10 copies/mL in subtype B viruses ( CI [3 . 5–5 . 2] ) , an estimate close to the optimum inferred using epidemiological data [14 , 16] . Unfortunately , although the OU model is probably more biologically realistic , one must keep in mind that with this model , any temporal trend in viral load will be interpreted as caused by evolution of viral genetic factors . For example , in our dataset ( in which SPVL did not significantly change over time ) , adding a temporal trend of +0 . 02 log10 copies/mL/year ( comparable to what has been observed in various European and North American cohorts [17 , 18] ) resulted in the OU model being much more favoured ( ΔAIC = 12 compared to the BM model ) . The BM model is unlikely to result in a sustained temporal trend , especially when the sample size is large . Yet , a temporal trend in viral load may be due to uncontrolled environmental factors—for example , a changing prevalence of coinfections [45] or variability in the viraemia assays—and not necessarily to viral evolution . In a dataset presenting a significant temporal trend in which OU is the best model , the dataset may be analysed with the temporal trend removed to check that the correlation structure of viral load also corresponds best to the OU model . Thirdly , our results reconcile estimation of heritability based on donor–recipient pairs ( which consistently estimate heritability at around 30% ) and those based on phylogenetic analysis . If we assume the phylogenetic cherries separated by a small patristic distance are indeed sampled from donor–recipient pairs , the heritability estimated here by donor–recipient regression was similar to the prediction based on the phylogenetic analysis ( Fig 1C ) . We also derive formal links between donor–recipient regressions and phylogenetic models ( S2 Text ) . In this large European cohort , there was no evidence that heritability differed between countries . Given the small sample sizes for each country , we had limited power to detect such differences . However , the similar phenotypic variance across countries and the fact that sequences from different countries are interspersed in the tree also suggested genetic variance in viral load was similar across countries . We found significant heritability only in the Netherlands—26% ( N = 434 , S2 Table ) . We found no significant heritability in the 742 subtype B sequences from Switzerland ( S2 Table ) . Previous studies have estimated heritability of SPVL in Switzerland at around 60% but only when focusing on the subset of MSM patients with at least 3 viral load measures with little fluctuations across measures [12 , 28] , a subset accounting for 20% of all the data; no significant heritability was found when using all of the data . The limited number of samples from the UK ( N = 87 , S2 Table ) made it difficult to compare our results with a previous study finding a heritability of 6% in the UK [29] . We estimated the heritability of 3 viral phenotypes: the single viral load GSVL measured in a standardised way , SPVL , and the rate of CD4 decline . GSVL had higher genetic variance than SPVL , resulting in heritability of 31% for GSVL compared to 21% for SPVL , and we suggest this is due to the closer link between the viral sequence and the phenotype , both obtained from the same sample ( S1 Text ) . However , SPVL was a better predictor of CD4 decline , as it explained R2 = 7 . 4% of the variance in CD4 decline , while GSVL explained R2 = 5 . 2% . This is perhaps not surprising: CD4 decline and SPVL are both temporal averages summarising virulence over a period of time , whereas GSVL is a single measurement typically taken at the beginning of infection ( median time between date first positive and sample = 266 days ) . Note though that assay variability significantly contributed to variation in SPVL ( S4 Table ) , showing the importance of adjusting for assay type when estimating heritability of SPVL . In our definition of heritability , assay error and variability in assays contributed to the denominator ( total phenotypic variance ) . Yet , this variation does not correspond to biological variation in the true trait but to variation caused by imperfect measurement . In principle , we should remove this source of variance from the denominator to obtain the phenotypic variance in the true trait , and this will result in higher heritability . This effect would be negligible for assay variability , which represents only a small fraction of the total variance ( S4 Table ) . Similarly , a plausible value for assay error at 0 . 04 log10 copies2/mL2 ( Materials and methods ) would result ( with a phenotypic variance of 0 . 54 log10 copies2/mL2 and heritability at 31% ) in a slightly higher heritability of 0 . 31 x 0 . 54 / ( 0 . 54–0 . 04 ) = 33% . Lastly , 11% of variation in the rate of CD4 cell count decline was explained by viral genetic variation . The small correlation between CD4 decline and viral load ( around 5% ) , the CD4 decline heritability of 11% , and the viral load heritability of 31% can be explained by the existence of several classes of viral variants: ( i ) variants causing more intense exploitation of CD4 cells resulting in faster CD4 depletion and greater viral load , contributing to the heritability of CD4 decline and of viral load; ( ii ) variants increasing viral load without intensifying the exploitation of CD4 cells , contributing to the heritability of viral load only; and ( iii ) variants intensifying the exploitation of CD4 cells without increasing viral load , contributing to the heritability of CD4 decline only . Both models of character evolution assume an additive effect of the viral genotype and an “environment” effect ( including host factors ) . A limitation of this study is that host genotype data was not available , in particular for the host class I HLA alleles , and therefore we could not control for this important factor [46] in the regression . Moreover , variability in viral load may depend on interactions between viral and host genetic factors [13 , 47] , in particular between host class I HLA alleles and viral CTL epitopes . Our model assumes additive contributions of the host and the virus and would force the ( host genotype ) x ( viral genotype ) interaction variance into the 2 additive components . This calls for the development of new methods to measure the fraction of variance determined by virus–host interactions [47] . The confirmation that a significant fraction of variability in viral load is determined by viral genetic factors motivates searching for individual viral genetic variants responsible for variation in viral load . The finding of 2 distinct islands’ contributions to heritability , 1 from the replication machinery ( gag and pol genes ) and the second from the envelope gene , will guide this search . A previous viral genome-wide association study did not find any significant viral genetic variation associated with viral load [13] but was only powered to detect individual effects accounting for 4% or more heritability . This suggests that the effects of individual variants are small , requiring large sample sizes for their discovery . Viral load may also be determined by epistatic effects between mutations , the detection of which would require even larger sample sizes . Knowledge of these molecular variants would allow us to relate changes in frequency of these variants to the observed temporal trends in viral load [17–19] , and most importantly , would provide new insights into HIV pathogenesis , a fascinating but challenging task [5] . In conclusion , around 30% of variation in viral load was explained by viral genetic variation in these European cohorts . This study highlights the need for large datasets , comparison between different models of character evolution , and goodness of fit tests to consistently estimate heritability of viral load in HIV-1 infection and so explains and resolves the inconsistency in previously published estimates .
We selected HIV-positive seroconverters from the Antwerp cohort in Belgium ( BE ) , the Swiss HIV Cohort Study in Switzerland ( CH ) , the ANRS PRIMO Cohort in France ( FR ) , the ATHENA cohort in the Netherlands ( NL ) , and the UK register of seroconverters in the UK—all part of the BEEHIVE ( “Bridging the Evolution and Epidemiology of HIV in Europe” ) collaboration . These were selected as the set of all patients meeting the study’s inclusion criteria , for which we had complete clinical and virus genetic data at the time of analysis . The inclusion criteria were the following . ( i ) Participants were seroconverters ( i . e . , the first positive test was less than 1 year after the last negative test ) , or the participant presented with evidence of recent infection ( laboratory evidence or seroconversion illness ) , ensuring the date of infection was known precisely . ( ii ) No antiretroviral therapy was taken in the first 6 months following the first positive test . ( iii ) At least 1 viral load or 1 sample from which viral load can be determined was taken between 6 and 24 months following the first positive test . ( iv ) At least 1 sample of at least 500 μL of frozen EDTA plasma or serum was taken between 0 and 24 months following the first positive test while antiretroviral therapy ( ART ) -naive . We also collected information on age , sex , and mode of transmission . All patients consented to this study . All studies within the cohorts were approved by in-country institutional review boards , and the overall BEEHIVE study , which only accessed anonymised data , was approved by the ethics panel of the European Research Council . The stable value of viraemia after acute infection and before the onset of AIDS ( our phenotype of interest ) was calculated in 2 ways . First , the viral load was remeasured in a standardized way , on a single blood sample taken more than 6 months and less than 24 months after the first positive HIV test and before the start of ART . If viral load had been previously measured with 1 of 3 assays ( COBAS AmpliPrep/COBAS TaqMan HIV-1 Test , v2 . 0 from Roche; Abbott RealTiMe HIV-1 Assay from Abbott; Quantiplex HIV-1 RNA Assay , version 3 . 0 from Chiron Diagnostics , Emeryville , CA ) , on the same visit when the sample used to determine the viral sequence was taken , we did not repeat the assay . Otherwise , viral loads were repeated with COBAS AmpliPrep/COBAS TaqMan HIV-1 Test , v2 . 0 on the same sample used to determine the viral sequence . We defined the GSVL as log10 of the single viral load ( in copies per mL ) measured in this way . In several cases , the GSVL was below detection limit , which may be due to genuinely low viraemia or assay failure , but the assay could not be repeated because of material shortage . We eliminated from the analysis the undetectable GSVL values with SPVL greater than 3 log10 copies/mL ( N = 1 value eliminated ) . The error variance of the assay used for GSVL is between 0 . 01 log10 copies2/mL2 ( standard deviation of 0 . 1 copies/mL estimated in a standardised way using aliquots of the same sample [48] ) and 0 . 25 log10 copies2/mL2 ( standard deviation of <0 . 5 log10 copies/mL estimated from the replicated viral load measures for problematic samples in this study ) . An intermediate value of 0 . 04 log10 copies2/mL2 is plausible . Second , we used the series of viral loads previously measured on the same patients to define the set-point viral load ( SPVL ) , calculated as the average log10 ( viral load , in copies per mL ) for all viral load measurements available between 6 and 24 months after the first positive HIV test ( using different assays ) . For both the GSVL and the SPVL measures , we adjusted for the potential impact of assay type on viral load within the phylogenetic regression . Additionally , we quantified virulence of the virus using the rate of CD4 decline in patients . We selected patients who had at least 5 CD4 count measures between the date of the first positive HIV test and the date that ART was first prescribed . We fitted a linear model describing the decline in CD4 count over time within this patient . We recorded the slope of this relationship . The characteristics of the cohort are summarized in Table 3 . Full genome HIV sequences were obtained from blood samples of seroconverters taken between 6 and 24 months after seroconversion and before ART was initiated . In order to measure heritability of SPVL , we fitted a series of stochastic models that describe how viral load evolves along the branch of the tree . Under all these models , the distribution of SPVL at the tip of the trees is a random drawing in a multivariate normal distribution whose mean vector and variance–covariance matrix , denoted ( μ , Σ ) , depend on the model and its parameters and on the structure of the tree .
|
The severity of the outcome of infection by a pathogen depends on many distinct factors . These include the environment and the genetic sequences of both the host and the pathogen , among others . The fraction of variability in disease outcome explained by pathogen genetic factors is termed “heritability” , because these factors are “inherited” by the new host upon infection . Quantifying heritability is key to understanding the development of the disease and the evolution of the virus . Here , we determined heritability of set-point viral load ( SPVL ) in HIV-1 . SPVL is the stable value of viraemia in asymptomatic infection and it is a strong predictor of disease severity . While heritability of SPVL has been estimated using comparisons of viral genome sequences , this has resulted in widely variable estimates of heritability . Using a large dataset of patients living in Europe , standardised viral loads measures , and new methods , we obtain a more definitive estimate of HIV-1 SPVL heritability in Europe at about 30% . Thus , a significant amount of the variation in disease outcome is explained by the genetics of the virus .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"methods"
] |
[
"taxonomy",
"organismal",
"evolution",
"medicine",
"and",
"health",
"sciences",
"pathology",
"and",
"laboratory",
"medicine",
"viral",
"transmission",
"and",
"infection",
"pathogens",
"microbiology",
"retroviruses",
"viruses",
"immunodeficiency",
"viruses",
"phylogenetics",
"data",
"management",
"rna",
"viruses",
"phylogenetic",
"analysis",
"microbial",
"evolution",
"microbial",
"genetics",
"viral",
"load",
"genetic",
"epidemiology",
"computer",
"and",
"information",
"sciences",
"lentivirus",
"medical",
"microbiology",
"epidemiology",
"hiv",
"microbial",
"pathogens",
"evolutionary",
"systematics",
"evolutionary",
"genetics",
"viral",
"evolution",
"viral",
"genetics",
"virology",
"viral",
"pathogens",
"genetics",
"biology",
"and",
"life",
"sciences",
"evolutionary",
"biology",
"organisms"
] |
2017
|
Viral genetic variation accounts for a third of variability in HIV-1 set-point viral load in Europe
|
Heterochromatin preferentially assembles at repetitive DNA elements , playing roles in transcriptional silencing , recombination suppression , and chromosome segregation . The RNAi machinery is required for heterochromatin assembly in a diverse range of organisms . In fission yeast , RNA splicing factors are also required for pericentric heterochromatin assembly , and a prevailing model is that splicing factors provide a platform for siRNA generation independently of their splicing activity . Here , by screening the fission yeast deletion library , we discovered four novel splicing factors that are required for pericentric heterochromatin assembly . Sequencing total cellular RNAs from the strongest of these mutants , cwf14Δ , showed intron retention in mRNAs of several RNAi factors . Moreover , introducing cDNA versions of RNAi factors significantly restored pericentric heterochromatin in splicing mutants . We also found that mutations of splicing factors resulted in defective telomeric heterochromatin assembly and mis-splicing the mRNA of shelterin component Tpz1 , and that replacement of tpz1+ with its cDNA partially rescued heterochromatin defects at telomeres in splicing mutants . Thus , proper splicing of RNAi and shelterin factors contributes to heterochromatin assembly at pericentric regions and telomeres .
Eukaryotic genomic DNA associates with histone and non-histone proteins to form chromatin , which is necessary for the spatial and temporal organization of chromosomes . A distinction is commonly drawn between two types of chromatin: euchromatin and heterochromatin . Euchromatin is less condensed and often associated with genes that are actively transcribed . Heterochromatin is highly condensed and often forms over repetitive DNA elements such as transposons . The formation of heterochromatin prevents expression of transposons , improper recombination of repetitive genomic loci , and missegregation of chromosomes during mitosis and meiosis , thus maintaining genome stability [1] , [2] . Histones within heterochromatin regions are usually hypo-acetylated and methylated at histone H3 lysine 9 , which serves as a binding site for heterochromatin protein 1 ( HP1 ) [3]–[5] . HP1 subsequently recruits diverse proteins to regulate cellular processes such as transcriptional silencing , recombination suppression , and chromosome segregation [1] . The mechanism that attracts histone methyltransferases and deacetylases to repetitive DNA elements is under intensive study . In certain cases , sequence-specific DNA binding proteins are directly involved in recruitment of these enzymes [6]–[9] . Alternatively , the repetitive nature may itself be sufficient to trigger heterochromatin assembly [10] , [11] . Recent work has shown that DNA repeats are transiently transcribed , and the transcripts are processed by the RNA interference ( RNAi ) machinery into small interfering RNAs ( siRNAs ) , which help target histone-modifying enzymes to repeat regions . However , the mechanistic details of RNAi-mediated heterochromatin assembly is not yet well understood [12]–[14] . The mechanisms by which heterochromatin is assembled and regulated have been extensively studied in the fission yeast S . pombe because it shares basic pathways of heterochromatin assembly with mammals , yet has the key advantages of facile genetics and single representative genes for most key families of mammalian chromatin-modifying factors . In this organism , heterochromatin is present mainly at pericentric regions , subtelomeres , and the silent mating-type region , all of which contain similar repeat sequences composed of dg and dh repeats [1] . These repeats are transcribed during S-phase of the cell cycle by RNA polymerase II . The transcripts are sliced by Ago1 and then reverse-transcribed by the RNA-directed RNA polymerase complex ( RDRC: Rdp1 , Cid12 , and Hrr1 ) into double stranded RNAs , which are processed by Dcr1 into small interfering RNAs ( siRNAs ) . These siRNAs are loaded onto Argonaute siRNA chaperone complex ( ARC: Arb1 , Arb2 , and Ago1 ) and then onto the RNA-induced transcriptional silencing complex ( RITS: Ago1 , Chp1 , and Tas3 ) . RITS is targeted to the repeat regions through base pairing between siRNAs and the nascent transcripts and recruits the H3K9 methyltransferase complex CLRC , which contains SET domain protein Clr4 as its catalytic subunit . H3K9 methylation recruits HP1 family proteins Swi6 and Chp2 , which in turn recruit histone deacetylases ( HDACs ) such as SHREC to further compact chromatin ( see review [12]–[14] ) . Surprisingly , in addition to these complexes , splicing factors are required for heterochromatin assembly at pericentric regions [15]–[17] . In fission yeast , 43% of genes contain introns [18] , indicating the prevalence of splicing in this organism . The spliceosome and the splicing reactions of fission yeast are also highly conserved with those of higher eukaryotes , which utilize snRNAs U1 , U2 , U4 , U5 , and U6 , as well as over one hundred protein components ( see review [19] ) . The process of splicing starts when U1 and U2 snRNPs bind to the 5′ splice site and branch point on a pre-mRNA , respectively . The U4/U6 . U5 snRNP and the Prp19 complex ( also known as NineTeen Complex , or NTC ) are then recruited to form the precatalytic spliceosome . After the release of U1 and U4 , the spliceosome is activated and the 5′ splice site is cleaved and fused to the branch point to form a lariat structure . Next , 3′ intron cleavage is coupled to exon ligation in a post-spliceosomal complex of U2 , U5 , and U6 . The mature mRNA is then released and the snRNPs recycled . Most notably , temperature-sensitive mutants prp10-1 ( component of U2 ) and cwf10-1 ( component of U5 ) exhibit silencing defects at pericentric regions [15] . Like dcr1Δ , these mutants lose most siRNAs derived from pericentric repeats [15] . Additionally , the spliceosome associates with RDRC component Cid12 [15] , [20] , indicating a possible direct role of splicing factors in connecting the nascent transcripts and the RNAi machinery during heterochromatin formation . It was hypothesized that splicing factors act in the RNAi pathway independently of RNA splicing because silencing defects are obvious when the splicing of a control tbp1 mRNA is intact and introducing cDNAs of ago1+ or hrr1+ was unable to rescue pericentric heterochromatin silencing defects of prp10-1 cells [15] , [17] . However , the possibility that splicing factors regulate the proper processing of RNAi factors has not been rigorously tested . Notably , a number of recently identified factors involved in RNAi ( arb1+ , arb2+ , ers1+ , and dsh1+ ) contain introns and might require splicing factors for their proper expression [21]–[24] . In this study , we performed a screen of the S . pombe nonessential gene deletion strain library and discovered four new putative splicing factors involved in pericentric heterochromatin assembly . We demonstrated that the phenotype of the strongest of these , cwf14Δ , is similar to those of RNAi mutants in regulating pericentric heterochromatin assembly . RNA-seq analyses further found that cwf14Δ resulted in mis-splicing of a subgroup of genes , including a number of RNAi factors . Moreover , we showed that introducing the cDNAs of three RNAi factors , ago1+ , arb2+ , and ers1+ , significantly alleviated silencing defects associated with cwf14Δ . Furthermore , we found that the mRNA of telomere shelterin protein Tpz1 , which is involved in telomeric silencing , was also mis-spliced in splicing mutants and that introducing tpz1+ cDNA partially rescues telomeric silencing defects of splicing factor mutants . Thus splicing factors are involved in heterochromatin assembly mainly through regulating the proper splicing of heterochromatin assembly factors .
To comprehensively identify factors required for pericentric heterochromatin assembly , we performed a screen of the fission yeast haploid deletion library for mutants that affect silencing of a reporter inserted into pericentric heterochromatin , otr::ade6+ ( Figure 1A–C ) [25] . In wild-type cells , otr::ade6+ is silenced , causing red colony color on low adenine medium . However , when heterochromatic silencing is lost at the pericentric region , otr::ade6+ is expressed , and colonies are white . Strains with intermediate silencing defects show variable degrees of pink/red color , allowing rough phenotypic quantification ( Figure 1C ) . In order to eliminate strains that have inherent metabolic defects causing lighter-than-red color , we performed a control screen with an ade6-M210 query strain lacking the otr::ade6+ reporter ( Figure 1B ) , which allowed us to filter false positives out of the screen . Our finalized list of hits is shown in Figure 1C . Each hit colony was assigned a score between 1 and 4 , with 4 indicating the strongest silencing defects . Among the mutants identified , 25 were previously known to be required for pericentric heterochromatin assembly , validating the effectiveness of our screen . These were mutants in the complexes of CLRC , ARC , RITS , RDRC , CUL4-DDB1 , HIRA , Clr6C , TRAMP , and SHREC , and in individual factors such as HP1 homolog Swi6 , NAD+ histone deacetylase Sir2 , CENP-B homolog Cbp1 , and CHD1 remodeler Hrp1 ( Figure 1C ) . There were also a number of previously reported heterochromatin mutants that are listed in the library but were missed in our screen , such as ago1Δ , rdp1Δ , and dcr1Δ . We confirmed by PCR that the null mutation was missing in these strains , indicating that false negatives are more likely the result of incorrect deletions present in the library than due to methodological bias . Most interestingly , there were eight novel mutants identified in this screen , of which four are uncharacterized genes implicated in various steps of mRNA splicing: cwf14 ( SPBC24C6 . 11 ) , dre4 ( SPAC13C5 . 02 ) , cwf12 ( SPBC32F12 . 05c ) , and the human SRRM1 homolog we named srm1 ( SPCC825 . 05c ) . Each of these strains showed elevated levels of pericentric dh transcripts , a common phenotype of heterochromatin mutants , suggesting that these mutants indeed affected pericentric heterochromatin assembly ( Figure 1D ) . As cwf14Δ showed the strongest phenotype among these four splicing mutants , we chose it as the focus of our subsequent experiments . We first confirmed by serial dilution analysis that cwf14Δ cells containing otr::ade6+ formed white colonies similar to those of dcr1Δ ( Figure 2A ) . We also examined the effect of cwf14Δ on silencing of another reporter inserted at the same location , otr::ura4+ [26] . The silencing of this reporter gene in wild-type cells allows them to grow on counterselective medium containing 5-fluoroorotic acid ( FOA ) . Serial dilution analyses showed that cwf14Δ has comparable silencing defects to dcr1Δ , as indicated by attenuated growth on FOA medium ( Figure 2A ) . Moreover , chromatin immunoprecipitation ( ChIP ) analyses showed increased enrichment of RNA Polymerase II at pericentric regions in cwf14Δ cells , indicating that the effect on pericentric transcript levels was at least in part due to increased transcription ( Figure 2B ) . Further ChIP analyses showed strong reduction in levels of heterochromatin hallmarks such as H3K9me2 and Swi6 at otr::ura4+ and to a lesser extent at the endogenous dh repeats in cwf14Δ cells ( Figure 2C ) . This pattern is similar to RNAi mutants such as dcr1Δ , but less pronounced than that of clr4Δ , suggesting that Cwf14 might regulate RNAi-mediated heterochromatin assembly . Consistent with this idea , siRNAs derived from pericentric repeats were eliminated in cwf14Δ cells , similar to dcr1Δ cells ( Figure 2D ) . It was shown that a 1 . 6 kb fragment of pericentric dg repeats ( termed L5 ) induces heterochromatin assembly and silencing of adjacent genes when inserted into an ectopic site in an RNAi-dependent manner [27] , [28] . Silencing is lost in cwf14Δ cells as well , consistent with the idea that Cwf14 is involved in RNAi-mediated heterochromatin assembly ( Figure 2E ) . Since pericentric heterochromatin promotes proper loading of cohesin near centromeres to promote chromosome bi-orientation [29]–[31] , most heterochromatin mutants show defects in chromosome segregation [26] and sensitivity to the microtubule-destabilizing drug thiabendazole ( TBZ ) [32] . As expected , cwf14Δ was also sensitive to TBZ ( Figure 2A ) . Moreover , both the silencing and TBZ-sensitivity phenotypes were rescued by complementation of cwf14Δ with a plasmid containing an intact copy of cwf14+ under the control of its endogenous regulatory elements ( Figure 2F ) . Since cwf14Δ cells had some growth defects , we also performed PCR-mediated random mutagenesis and isolated a mutant ( cwf14-F26L ) that resulted in silencing defects without significantly affecting growth ( Figure 2G ) . To test whether cwf14Δ affects H3K9 methyltransferase activity of CLRC , we used a system in which Clr4 is tethered via a fused Gal4 DNA binding domain ( GBD ) to an ectopically integrated ade6+ reporter adjacent to three copies of the Gal4 binding site ( 3xgbs-ade6+ ) [33] ( Figure 3A ) . This tethering induces heterochromatin formation over a 6 kb locus , silencing transcription of the ade6+ reporter . RNAi factors are not required for this silencing , consistent with the idea that RNAi is required for the targeting of CLRC to DNA repeats [33] . Dilution analysis showed that cwf14Δ had no effect on Clr4-tethered silencing of 3xgbs-ade6+ , and ChIP analyses showed that H3K9me2 was enriched at nearby loci 2 and 3 kilobases away in cwf14Δ cells , similar to wild-type cells ( Figure 3B ) . Collectively , these data suggest that Cwf14 is involved in heterochromatin formation at pericentric repeats through the RNAi pathway . Cwf14 is highly conserved across species . Its budding yeast homolog , Bud31p , co-purifies with the spliceosome [34] , and BUD31Δ causes mis-splicing of the mRNAs of ARP2 and SRC1 , two factors required for proper budding [35] . Cwf14 was initially identified in a purification of splicing factor Cdc5 , together with a number of spliceosome components [36] . However , whether Cwf14 is a stable component of the spliceosome and involved in splicing has not been tested directly . In order to further specify the mechanism of Cwf14 action , we constructed C-terminally tagged versions of cwf14 at its endogenous locus . Cwf14-GFP and Cwf14-myc were fully functional , as they did not show silencing defects of otr::ade6+ ( Figure 4A ) . Imaging of Cwf14-GFP showed that Cwf14 localizes predominantly in the nucleus ( Figure 4B ) . Western blot analysis showed that Cwf14-myc is a 40 kD protein . Moreover , the F26L mutation resulted in reduced Cwf14 protein levels , indicating that it is a partial loss-of-function allele ( Figure 4C ) . We also performed immunoprecipitation of Cwf14-myc followed by MudPIT mass spectrometry to identify its interacting proteins . Most factors that co-immunoprecipitated with Cwf14 , but not in a control purification , were components of the spliceosome , especially from subcomplexes NTC and U5 ( Figure 4D and Table S1 ) . This result corroborates data that Cwf14 co-precipitates with Prp17 , Prp19 , and Cwf2 , all members of U5-associated NTC [19] , [37] , as well as a component of U5 snRNP , Cwf10 [38] . Previous work suggests that the spliceosome is involved in heterochromatin assembly through tethering RDRC to pericentric transcripts [15] . This is because RDRC component Cid12 associates with the spliceosome [15] , [20] , and no splicing defects were observed in the well-characterized splicing substrate tbp1 in splicing factor mutants when silencing defects were apparent [15] , [17] . However , purifications of spliceosome components , including our purification of Cwf14 , have not identified any other heterochromatin components [36] , [37] , [39] ( Table S1 ) . These results indicate either that the physical connection between the spliceosome and RDRC is very weak , or that Cid12 is present in two separate complexes , RDRC and spliceosome . It remains a possibility that splicing factors regulate the correct processing of mRNAs involved in RNAi-mediated heterochromatin assembly . In order to test whether cwf14Δ has a general splicing defect , we performed RNA-seq analyses of total cellular RNAs from wild-type , dcr1Δ , and cwf14Δ cells . The gene expression profiles of cwf14Δ and dcr1Δ showed strong overlap of significantly up-regulated genes ( Figure 5A and Tables S2 and S3 ) , consistent with the idea that Cwf14 functions in the RNAi pathway . We found that cwf14Δ indeed resulted in intron retention of a portion of genes ( Figure 5B and Table S4 ) , consistent with the finding that it is associated with the spliceosome . The majority of introns were properly processed , which indicates that Cwf14 only moderately affected the activity of the spliceosome . Interestingly , many RNAi factors contain introns . Although unbiased ranking of exon-exon junction ratios between wild-type and cwf14Δ RNAs did not show preferential enrichment of RNAi or heterochromatin assembly factors ( Table S4 ) , significant intron retention was observed within mRNAs from ago1 , arb2 , ers1 , and dsh1 in cwf14Δ cells as compared to those in wild-type cells , despite similar sequencing depths genome-wide and similar levels of each gene transcript in both samples ( Figure 5C ) . RT-PCR analyses confirmed that these mRNAs were indeed spliced inefficiently in cwf14Δ as well as in cwf10-1 and prp10-1 cells ( Figure 5D ) . Moreover , Western blot analysis showed strong reduction of protein levels of Flag-Ago1 , the RNAi factor most severely affected by cwf14Δ , in both cwf14Δ and cwf10-1 strains ( Figure S1 ) , indicating that the mis-splicing of ago1 mRNA , and possibly other RNAi factors as well , resulted in altered protein levels . Previously , it was shown that introducing cDNA versions of ago1+ or hrr1+ failed to rescue silencing defects of prp10-1 [15] . We also generated a cDNA version of ago1+ at its endogenous chromosomal location and under the control of its endogenous regulatory elements ( ago1+::cDNA ) . This cDNA construct showed no defects in silencing of otr::ura4+ , indicating that this replacement created a functional Ago1 protein ( Figure S2 ) . However , ago1+::cDNA was unable to rescue otr::ura4+ silencing defects and TBZ sensitivity of cwf14Δ ( Figure 6A ) , even though it restored Flag-Ago1 protein levels ( Figure S1 ) . We reasoned that the inability of ago1+::cDNA to rescue cwf14Δ defects is because other RNAi factors are also mis-spliced . We thus generated cDNA versions of arb2+ and ers1+ at their endogenous chromosomal loci , which were both functional ( Figure S2 ) . Neither arb2+::cDNA nor ers1+::cDNA alone showed any effect on silencing of otr::ura4+ in cwf14Δ cells ( Figure 6A ) . However , when combinations of two cDNAs were introduced together into cwf14Δ cells , there was a detectable rescue of silencing defects and TBZ sensitivity , and the effect was stronger when all three cDNAs were introduced ( Figure 6A ) . Further ChIP analysis showed that both H3K9me and Swi6 protein levels were significantly increased at both otr::ura4+ and pericentric dh repeats in cwf14Δ cells supplemented with ago1+ , arb2+ , and ers1+ cDNAs ( 3cDNAs ) ( Figure 6B ) . We also found that introducing ago1+::cDNA was sufficient to rescue the silencing defects of otr::ade6+ associated with cwf10-1 ( Figure S3A ) . Moreover , there is a significant increase in both H3K9me and Swi6 levels at both otr::ade6+ and dh repeats in cwf10-1 ago1+::cDNA cells ( Figure S3B ) . Altogether , these results suggest that mutations in different splicing factors affect the splicing of diverse RNAi factors to regulate heterochromatin assembly at pericentric regions . Thus our results clearly demonstrated that splicing factors mainly exert their effects on pericentric heterochromatin assembly by promoting the proper splicing of RNAi factors . However , we caution that the rescue of heterochromatin silencing defects of cwf14Δ cells was incomplete even with ago1+ , arb2+ , and ers1+ cDNAs . This probably reflects the requirement of Cwf14 for the proper splicing of other RNAi factors such as dsh1+ , rdp1+ , arb1+ , hrr1+ , or some unidentified factors involved in heterochromatin assembly . Such an idea is supported by the fact that the rescue of cwf14Δ silencing defects correlated with the number of cDNA constructs that were introduced . It is also possible that splicing factors have a direct contribution in pericentric heterochromatin assembly . However , the strong rescue of pericentric silencing in cwf14Δ cells with cDNA constructs suggests that the regulation of RNAi factor splicing is a major role of splicing factors in this process . We also found that telomere shelterin component tpz1+ showed a strong reduction in exon-exon junction sequencing reads in cwf14Δ cells relative to wild-type cells , and this phenotype was confirmed by RT-PCR analyses in cwf10-1 and prp10-1 as well ( Figure 7A ) . A C-terminally Flag-tagged version of Tpz1 [40] affected silencing of a reporter gene inserted near telomere repeats ( Figure S4 ) , suggesting that Tpz1 is required for telomere silencing . We then analyzed the effect of splicing factor mutations on silencing of a reporter inserted near telomere repeats of chromosome two ( Tel2::ura4+ ) [41] . Indeed , cwf14Δ resulted in silencing defects at this reporter gene ( Figure 7B ) , accompanied by a reduction of H3K9me levels at tlh1 , which is embedded at telomeric heterochromatin , as well as the accumulation of tlh1 transcripts ( Figure 7C and 7D ) . Both cwf10-1 and prp10-1 cells showed increased tlh1 transcript levels , indicating that loss of telomeric silencing is a general feature of splicing mutants ( Figure 7C and 7D ) . Because multiple DNA sequences contribute to heterochromatin assembly at telomeres , including tlh1+ , telomere associated sequences ( TAS ) , and terminal TEL repeats [8] , we further tested silencing at TEL::ade6+ , which is inserted on the mini-chromosome Ch16 adjacent to telomere repeats [42] . We found that cwf14Δ , cwf14-F26L , and cwf10-1 resulted in loss of silencing of this reporter ( Figure 7E ) . Interestingly , replacement of tpz1+ with its cDNA partially rescued telomeric silencing phenotypes of the cwf14-F26L and cwf10-1 mutants ( Figure 7E and 7F ) . We could only marginally rescue the telomere silencing defects of cwf14Δ cells with tpz1+::cDNA ( Figure 7F and data not shown ) . Thus it is possible that additional factors involved in telomere silencing might also contain introns and depend on the spliceosome to properly regulate their splicing . Alternatively , splicing factors might affect telomere silencing through additional mechanisms . Nonetheless , these results demonstrate that inefficient splicing of tpz1 contributes to telomeric silencing defects in splicing mutants .
The formation of heterochromatin requires RNAi-mediated processing of repeat-derived transcripts and the targeting of histone modifying activities to repeat regions , leading to H3K9me and the recruitment of HP1 proteins . It has been shown that splicing factors are required for RNAi-mediated heterochromatin assembly in fission yeast , although the mechanism by which they are involved is not well characterized . The previously prevailing model was that the spliceosome physically associates with RNAi factors to regulate heterochromatin assembly , rather than acting through its splicing activity [15] , [16] . One of the main evidences for this idea is that splicing mutants show severe silencing defects even though the splicing of tbp1 mRNA was not affected [15] , [17] . However , whether these splicing factor mutants selectively affect the splicing of RNAi factors has not been rigorously tested . Our RNA-seq analyses showed prominent intron retention of a subgroup of mRNAs in cwf14Δ cells , even though the majority of introns ( including those of tbp1 ) are still properly processed ( Figure 5B and Table S4 ) . Interestingly , a number of key RNAi factors were among the list of strongly mis-spliced introns , a result that is further corroborated by RT-PCR analyses of a selective set of RNAi factor mRNAs in other spliceosome mutants such as cwf10-1 and prp10-1 ( Figure 5D ) . Most importantly , we found that introducing a combination of cDNAs of RNAi factors significantly alleviated pericentric heterochromatin defects associated with cwf14Δ and cwf10-1 ( Figure 6 and S3 ) . Thus splicing factors regulate the proper splicing of RNAi factors , which is a major , if not sole , contributor to heterochromatin assembly defects in splicing mutants . That introducing tpz1+::cDNA was able to partially rescue telomere silencing defects associated with splicing factors further supports the idea that mis-splicing of heterochromatin factors is the reason splicing factor mutants show heterochromatin assembly defects ( Figure 7E and 7F ) . It is noteworthy that cwf14Δ has only moderate splicing defects , with some introns show very strong sensitivity , whereas most others show little to no defects ( Figure 5B and Table S4 ) . This raises the question of whether specific intron sensitivity is a result of introns that are inherently difficult to splice . Consistent with this idea , our RT-PCR analyses showed prominent unspliced precursor mRNAs of RNAi factors even in wild-type cells ( Figure 5D ) . It is also a striking pattern that heterochromatin factors in the same complexes tend to either have or not have introns . For example , many members of RNAi , such as ago1+ , arb1+ , arb2+ , ers1+ , dsh1+ , rdp1+ , and hrr1+ , have introns , but none of CLRC ( clr4+ , rik1+ , raf1+ , raf2+ , cul4+ , stc1+ ) , SHREC ( clr1+ , clr2+ , clr3+ , mit1+ ) , or swi6+ have any introns , raising the possibility of selective regulation of the RNAi pathway through general or specific changes in splicing efficiency . In fact , an analysis of splicing changes during meiosis showed that one intron of arb1+ is induced to be spliced , while a different intron exhibits splicing repression [43] . Since splicing factors have been identified in screens that affect RNAi-based processes in worms , flies , and plants [44]–[48] , it seems a conserved mechanism that proper splicing of the mRNAs of RNAi factors regulates the efficiency of RNAi inside the cell .
The Bioneer library is composed of strains of mixed endogenous ade6 alleles: ade6-M216 , which forms pink colonies on low adenine medium , and ade6-M210 , which give rise to red colonies , similar to ade6Δ . To avoid the complication of ade6-M216 alleles in our screen , we included in the query strain an ade6-M210-mCherry allele linked to a hphMX6 cassette , which confers resistance to the antibiotic hygromycin B , allowing us to generate a uniform ade6-M210 background . Screens were carried out according to previous protocols [49] , with slight modifications . Yeast strains arrayed in 384 strains/plate format were pinned on YES agar medium containing 100 µg/ml G418 , and otr::ade6+-natMX6 ade6-210-hphMX6 and ade6-210-hphMX6 strains were pinned on YES+100 µg/ml hygromycin B . After two days growth , strains were mated on SPA agar medium . Plates were incubated at 25°C for 3 days , then 42°C for 3 more days to kill vegetative cells . Strains were then germinated and correct genotype selected by pinning to YES+GNH ( G418 , Nat , and Hygromycin ) or YES+GH ( G418 and Hygromycin ) medium , and then pinned to YE ( low ade ) medium for color readout . Cwf14-GFP and Cwf14-13myc strains were constructed by a PCR-based module method . cwf14Δ , cwf12Δ , dre4Δ , and srm1Δ were derived from the Bioneer fission yeast deletion library , verified via PCR , and backcrossed . The p-cwf14+ plasmid was constructed by insertion of a PCR product containing cwf14+ promoter and coding region into SphI and XmaI sites of pREP41 . The cwf14-F26L-myc mutant strain was constructed by integrating a cwf14+-myc-kanMX6 cassette amplified by error-prone PCR ( high dNTP concentration ) into the endogenous cwf14+ locus . ago1+::cDNA , arb2+::cDNA , and ers1+::cDNA strains were constructed by replacing the endogenous genes with intronless cDNA versions . All were sequenced to confirm full replacement and lack of mutation . cwf10-1 and prp10-1 strains were a kind gift from Robin Allshire . Genetic crosses were used to construct all other strains . The genotype of strains is provided in Table S5 . For serial dilution plating assays , ten-fold dilutions of mid-log-phase culture were plated on the indicated medium and grown for 3 days at 30°C unless otherwise indicated . Total cellular RNA was isolated from log-phase cells using MasterPure yeast RNA purification kit ( Epicentre ) according to the manufacturer's protocol . Quantification with qRT-PCR was performed with Power SYBR Green RNA-to-CT one-step Kit ( Applied Biosystems ) . RNA serial dilutions were used as templates to generate the standard curve of amplification for each pair of primers , and the relative concentration of target sequence was calculated accordingly . An act1 fragment served as reference to normalize the concentration of samples . Sequence of DNA oligos is provided in Table S6 . For RNA-seq , purified RNA was prepared by TruSeq Stranded Total RNA Kit ( Illumina ) , which includes rRNA depletion and chemical fragmentation . Index adapters were added to allow for multiplexing . Paired-end sequencing with 100 bp read lengths was performed on Illumina HiSeq . Mapping was performed with the Tuxedo Suite consisting of Bowtie , TopHat , and Cufflinks . For cwf14Δ , 57 , 521 , 513 reads were obtained , and 83% mapped to the genome via Bowtie . For WT , 47 , 636 , 518 reads were obtained , and 82% mapped . For dcr1Δ , 64 , 972 , 007 reads were obtained , and 83% mapped . RNA-seq data have been deposited to the Sequence Read Archive ( http://www . ncbi . nlm . nih . gov/sra/ ) with accession number SRP040479 . For the dot plot , exon-exon junction ratios were filtered to remove several types: 1 ) junctions which mapped zero times in any sample ( possible mapping noise ) , 2 ) junctions whose sum in WT , cwf14Δ , and dcr1Δ was less than 30 ( to avoid randomness due to small sample sizes ) , and 3 ) ratios whose values were greater than 150 ( to focus the diagram on splicing reduction ) . Northern blot of siRNAs was performed as described previously [50] . ChIP analyses were performed as described previously [51] . Antibodies used were H3K9me2 ( Abcam 1220 ) , Swi6 [52] , and RNA Pol II ( Covance 8WG16 ) . Quantification with qPCR was performed with Maxima SYBR Green/ROX qPCR Master Mix ( Thermo ) . Enrichment was calculated as: relative levels in ChIP/relative levels in total DNA . An act1 promoter fragment was used as a control for normalization unless otherwise indicated . Sequence of DNA oligos was provided in Table S6 . Exponentially growing yeast cells were harvested , washed with 2xHC buffer ( 300 mM HEPES-KOH pH 7 . 6 , 2 mM EDTA , 100 mM KCl , 20% glycerol , 0 . 1% NP-40 , 2 mM DTT , and protease inhibitor cocktail ( Roche ) ) and frozen in liquid nitrogen . Crude cell extracts were prepared by vigorously blending frozen yeast cells with dry ice using a household blender , followed by sonication and incubation with 30 ml 1xHC buffer containing 250 mM KCl for 30 min . The lysate was cleared by centrifugation at 82 , 700×g for 3 hours . The supernatants were incubated with 50 µl of C-myc antibody ( Sigma C3956 ) overnight , for three hours the next day with 50 µl protein G agarose beads , washed eight times with 1xHC containing 250 mM KCl , then two times with the same buffer without NP-40 . For mass spectrometry analysis , bound proteins were eluted with 2×100 µl of 50 mM Tris pH 7 . 5 , 5% SDS , 5% glycerol , 50 mM DTT . MudPIT mass spectrometry analysis was performed as described previously [53] .
|
Heterochromatin formation at specific genomic regions is critical for processes as diverse as gene expression and chromosome segregation . The formation of silent heterochromatin at repetitive DNA elements requires processing of transcripts by the RNA interference machinery . Curiously , factors involved in proper RNA splicing are required for heterochromatin assembly , and it was proposed that splicing factors provide a platform for the recruitment of RNAi complexes independently of their role in regulating splicing . In this study , we found several novel splicing factors involved in heterochromatin assembly . Our analysis of genome-wide splicing patterns by RNA sequencing showed that the mRNAs of RNAi factors are very sensitive to perturbations of RNA splicing machinery . Moreover , we found that splicing factors are critical for the production of a telomere shelterin component and proper telomeric heterochromatin assembly . Most importantly , we showed that introducing the cDNA versions of RNAi and shelterin components alleviates heterochromatin defects associated with splicing factor mutations . Thus splicing factors are required for heterochromatin assembly mainly by regulating the proper splicing of heterochromatin assembly factors .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Materials",
"and",
"Methods"
] |
[
"biotechnology",
"rna",
"interference",
"chromosome",
"structure",
"and",
"function",
"gene",
"function",
"histone",
"modification",
"fungi",
"model",
"organisms",
"epigenetics",
"chromatin",
"schizosaccharomyces",
"genetic",
"engineering",
"research",
"and",
"analysis",
"methods",
"chromosome",
"biology",
"gene",
"expression",
"schizosaccharomyces",
"pombe",
"yeast",
"biochemistry",
"rna",
"rna",
"processing",
"cell",
"biology",
"genetic",
"screens",
"gene",
"identification",
"and",
"analysis",
"genetics",
"biology",
"and",
"life",
"sciences",
"yeast",
"and",
"fungal",
"models",
"genomics",
"organisms",
"chromosomes"
] |
2014
|
The Proper Splicing of RNAi Factors Is Critical for Pericentric Heterochromatin Assembly in Fission Yeast
|
Operant learning requires that reinforcement signals interact with action representations at a suitable neural interface . Much evidence suggests that this occurs when phasic dopamine , acting as a reinforcement prediction error , gates plasticity at cortico-striatal synapses , and thereby changes the future likelihood of selecting the action ( s ) coded by striatal neurons . But this hypothesis faces serious challenges . First , cortico-striatal plasticity is inexplicably complex , depending on spike timing , dopamine level , and dopamine receptor type . Second , there is a credit assignment problem—action selection signals occur long before the consequent dopamine reinforcement signal . Third , the two types of striatal output neuron have apparently opposite effects on action selection . Whether these factors rule out the interface hypothesis and how they interact to produce reinforcement learning is unknown . We present a computational framework that addresses these challenges . We first predict the expected activity changes over an operant task for both types of action-coding striatal neuron , and show they co-operate to promote action selection in learning and compete to promote action suppression in extinction . Separately , we derive a complete model of dopamine and spike-timing dependent cortico-striatal plasticity from in vitro data . We then show this model produces the predicted activity changes necessary for learning and extinction in an operant task , a remarkable convergence of a bottom-up data-driven plasticity model with the top-down behavioural requirements of learning theory . Moreover , we show the complex dependencies of cortico-striatal plasticity are not only sufficient but necessary for learning and extinction . Validating the model , we show it can account for behavioural data describing extinction , renewal , and reacquisition , and replicate in vitro experimental data on cortico-striatal plasticity . By bridging the levels between the single synapse and behaviour , our model shows how striatum acts as the action-reinforcement interface .
Learning from reinforcement requires a neural interface between reinforcement signals and action representations . Since the tentative identification of the ventral striatum as this “limbic-motor” interface by Mogenson and colleagues [1] , separate strands of work have elaborated four key elements centred on the striatum . First , that phasic activity of midbrain dopamine neurons signals a prediction error between expected and received reinforcement , or the stimuli that predict reinforcement [2]–[5] . Second , that in the primary target for these signals , the striatum , the plasticity of cortical inputs to striatal medium spiny neurons ( MSNs ) is modulated by dopamine [6]–[8] . Third , that intact regions of striatum are necessary for the expression and likely acquisition of goal-directed and habitual actions [9]–[11] . Fourth , that the basal ganglia , for which the striatum is the input station , collectively implement a system for action selection via selective disinhibition of targets in motor thalamus and brainstem [12]–[14] . Consequently , a plausible hypothesis for the reinforcement-action interface is the interaction between cortico-striatal weights and phasic dopamine . Thus , the adjustment of cortico-striatal weights by value-conditioned environmental feedback , in the form of the phasic dopamine signal , changes which actions are prioritised in future [15] . Despite the extent of work on each of these elements , to our knowledge no model has integrated them all to test this widely held hypothesis . Such a model is required to tackle three critical challenges to this hypothesis . First , theories of reinforcement learning by the basal ganglia are based on simple dichotomies for cortical-striatal plasticity: that low and high dopamine respectively promote long-term depression ( LTD ) and long-term potentiation ( LTP ) at cortico-striatal synapses [15]; or in a more nuanced version that high dopamine promotes LTP at cortical synapses on D1-receptor expressing MSNs and low dopamine levels promote LTP at cortical synapses on D2-receptor expressing MSNs [16] . However , a recent study by Shen and colleagues [17] showed that whether these synapses express LTP or LTD is dependent on a three-way interaction between pre- and postsynaptic spike timing , postsynaptic dopamine receptor type ( D1 versus D2 expressing MSNs ) and dopamine level . Moreover , no combination of these factors maps onto a simple dichotomy . It is thus an open question whether this complex combination of plasticity rules can be reconciled with the reinforcement learning hypothesis . Second , the D1 and D2 MSN populations project through separate pathways that converge in the output nuclei of basal ganglia . A broad class of hypotheses propose that these “direct” and “indirect” pathways respectively permit and prevent the selection of specific actions [16] , [18]–[20] . It is unclear whether the just-described different plasticity rules operating on the cortical inputs to these pathways can be reconciled with this functional hypothesis . Third , the timing of the relevant signals spans many scales . At short time scales ( ∼10–100 ms ) cortical synapses onto the MSNs have spike-timing dependent plasticity ( STDP ) [21] , [22] . At longer time scales ( hundreds of milliseconds to greater than 1 s ) , there is the well-known credit assignment problem [23] , [24]: that cortical-striatal signals for action selection appear transiently , and long before the phasic dopamine signal carrying feedback from the environment arrives in the striatum [4] . How the short-term STDP and long-term feedback interact is unknown . We present here a model that provides the basis for integrating these strands of work on reinforcement learning and answering these challenges . It bridges the gap between the intricate subtleties of cortico-striatal plasticity at the synaptic level and the behaviour of the whole animal , thereby providing strong evidence that the striatum is indeed the locus of the action-reinforcement interface .
To ground this exercise we imagine a stylised instrumental conditioning experiment with reinforcement learning of an action , such as a rat lever pressing for food pellet ( in the Discussion we consider how our model of this task and of the inputs to striatum relate to the well-known distinction between goal-directed and habitual behaviour in instrumental tasks ) . We separate the experiment into epochs , and divide each epoch into notional trials corresponding to one action and its outcome . The timeline for the experiment is shown at the top of Figure 1 . Initially , there is a “baseline” epoch of free action choice . Following this , there is a “learning” epoch in which a key action–such as a lever press–is reliably paired with reinforcement , and consequently repeated . In the subsequent “intermission” epoch , the rat is removed from the arena and again has free action choice . This is followed by an “extinction” epoch , where the rat is reintroduced into the arena , but reinforcement is no longer paired with the previously reinforced action . We assume there ensues a period of repeated ( but unsuccessful ) attempts to obtain reinforcement . At some point the animal extinguishes its reinforced action and engages in a final bout of free-choice action in the “post-extinction” epoch . The baseline and intermission epochs will serve as controls for the models , testing that the absence of reinforcement does not lead to aberrant learning through noise ( in baseline ) and that the execution of other actions does not interfere with the learnt representation of the reinforced action ( in intermission ) . There is considerable in vivo evidence that striatal activity evolves during the course of operant learning , with both increases and decreases in activity observed , consistent with the hypothesis of cortico-striatal plasticity driving changes in activity over learning [26]–[31] . However , detailed interpretation of these data is difficult as there is no distinction made between D1- and D2-type MSNs . By contrast there are good recent data on the opposing roles of D1 and D2 MSNs in controlling behaviour , from which we can establish predictions for the start and end-points of learning and extinction . Cui and colleagues [25] showed that the execution of a specific action was immediately preceded by coincident activation of both D1 and D2 MSNs , showing that both direct and indirect pathways are active when selecting an action . Selective optogenetic stimulation has shown that activating D1 MSNs initiates or increases locomotion whereas activating D2 MSNs ceases or prevents locomotion [19] , [20] , [32] . Together , these data support the broad hypothesis for the competing influence of the two pathways on action selection , that D1 MSN activity is permissive for action and D2 MSN activity is preventative for action [18] . In the context of learning , this hypothesis has been interpreted as the D1 and D2 MSNs , respectively , learning the go and no-go contexts for a given action [16] . Optogenetic stimulation during learning suggests this interpretation is correct [33] . We here hypothesise that this extends beyond active suppression of an action in a specific context ( no-go learning ) to also include active suppression of a learnt action in extinction—we later show this hypothesis is consistent with renewal and reacquisition phenomena . Currently missing are data or hypotheses for how the representation of the same action in corresponding D1 and D2 MSN populations changes over learning and over extinction . A straightforward extension of the competing pathways hypothesis is that after learning D1 MSN activity will be high and corresponding D2 MSN activity will be low or zero , thus favouring the selection of the action; and conversely that after extinction D1 MSN activity will be low or zero and D2 MSN activity high , thus favouring the suppression of the action . We used our prior model of action selection in the basal ganglia [34] , [35] to test this hypothesis and predict the relative responsiveness of D1 and D2 MSNs that optimises selection performance within a trial after learning or after subsequent extinction . Our model of the basal ganglia simulates how their internal circuitry can resolve competition between salient inputs from cortex ( Figure 2 ) —see Methods for a full description . Under the interpretation that basal ganglia mediate action selection [12]–[14] , cortical signals afferent to striatum associated with a single potential action comprise an “action request” [36] . The neural populations throughout basal ganglia that process this request comprise an action “channel . ” In general , an action request is a complex pattern of signals encoding the action whose overall level of activity represents the “salience” or urgency of the request . Selection of an action is then signalled by a sufficient fall in the level of inhibition ( relative to tonic ) in the channel encoding the action in the basal ganglia’s output nuclei . Our model simulates the mean firing rate of each neural population within the basal ganglia in response to a given set of action requests . Figure 2B shows the model’s response to a single phasic input from cortex . Consistent with the labelled-recording study of [25] , a single action is represented by coincident activity in a small population of D1 and D2 MSNs . Consistent with the optogenetic stimulation studies of [19] and [20] , activity in the two pathways is antagonistic: greater activity of the D1 MSN population drives inhibition of the corresponding basal ganglia output population , whereas greater activity of the D2 MSN population drives excitation of the corresponding basal ganglia output population . The model therefore shows that key to whether an action is selected or suppressed is the relative weighting of cortical input to the D1 and D2 MSN populations representing that action . We thus used our model to find the relative weights of cortical input to the D1 and D2 MSN populations that optimised selection of an action ( emulating the target situation at the end of the learning epoch ) and , separately , that optimised the suppression of an action ( emulating the target situation at the end of the extinction epoch ) . The ability to select a particular action can only be tested with reference to at least one other possible alternative action , so we considered two competing signals , one signal representing a fixed “control” action , available for selection throughout , and another signal representing the key action learnt and extinguished over the course of the experiment . We input this pair of salient signals to two channels in the model . For a given pair of inputs , we read out the outcome of the competition from the output of the basal ganglia ( SNr/GPi in Figure 2 ) : a sufficient decrease in inhibition from the output population signalled selection of the corresponding action . Thus three outcomes were possible: no action selected , one action selected , or both actions selected . Given these possible outcomes for each input pair , we defined ideal outcomes for a range of pairs of salience values , shown at the top left of Figure 3A and 3B for selection and suppression , respectively . We expect low salience signals to give no selection as the unresponsiveness of MSNs to low inputs ensures that these signals do not change basal ganglia output [34] . Otherwise , for selection we expect the input with the highest salience to win and thus a single action to be selected; and for suppression we expect no selection of the suppressed action , and only selection of the control action when it is sufficiently salient . Figure 3A shows that selection of an action was best achieved when its coding D1 MSN population was more responsive than its coding D2 MSN population . But , importantly , our results show that the best selection was achieved with some activity in the action’s coding D2 MSN population ( Figure 3A , bottom right ) , suggesting the novel prediction that D2 MSN activity must also be present to achieve optimal selection , and so does not only block selection ( in Figure S1 and Text S1 , we explain why the model makes this prediction ) . Figure 3B shows that suppression of an action was best achieved when its coding D2 MSN population was more responsive than its coding D1 MSN population . Importantly , our results showed that the action-coding D1 MSN population could remain highly active , with an lower limit of about 1∶1 for its input to output ratio . These results show that , rather than requiring that the D1 MSN input weight falls close to zero , the suppression of an action is robust to a large range of such weights . Our model thus shows that the competing-pathways hypothesis is broadly true for the D1 and D2 populations coding a single action , but more nuanced: there is a non-intuitive contribution of D2 MSN activity to optimal selection; and successful suppression can tolerate high levels of D1 MSN activity . We capture these non-intuitive predictions as the hypothesized target activity at end-points of learning and extinction during the stylised experiment in Figure 1 ( respectively , symbols 2 and 5 ) . There , we extend these end-points to their changes over the entire experiment with mild assumptions for MSN activity outside periods of learning . In the baseline epoch we assume a small , but non-zero response in both D1- and D2-MSNs , which is sufficient to initiate learning . In addition we demand that this baseline response is relatively stable during this period , such that randomly occurring pre- and postsynaptic spike pairings in this baseline activity do not cause either LTP or LTD . For similar reasons , we require stable responses in the intermission and post-extinction epochs . These profiles form the predicted targets for changes in MSN activity over learning for the rest of the paper . The key hypothesis is that these changes in MSN activity are driven by feedback from changes in the environment that are carried by dopamine signalling in the striatum . The bottom panel of Figure 1 plots the corresponding trial-by-trial change in striatal dopamine during the behavioural task . Throughout the baseline , intermission , and post-extinction epochs , the absence of any reinforcing stimuli is reflected in the constant tonic dopamine level on every trial . At the onset of the learning epoch , the initial reinforcement , being unexpected , is assumed to elicit a phasic dopamine burst [2]–[4] , [37] , [38] . As the reinforcement becomes predictable , the amplitude of elicited phasic dopamine declines [39] . During the extinction epoch , the omission of the expected reinforcement is assumed to elicit phasic dopamine “dips” [2] , [37] , [38] , [40] , whose magnitude gradually declines , as the omission too becomes predictable [41] . With these target trial-by-trial changes in MSN activity and corresponding striatal dopamine profile in hand we turn to the central question of how that dopamine signal drives the required MSN activity changes . The long-standing answer has been that dopamine modulates cortico-striatal plasticity [15] , but recent data have shown a partially complete picture of how nuanced that modulation is . On the one hand , Pawlak and Kerr [22] showed that cortico-striatal synapses have STDP , but not how that depends on postsynaptic neuron type ( D1 or D2 ) . On the other hand , Shen and colleagues [17] showed that the direction of modulation is dependent on the three factors of postsynaptic neuron type ( D1 or D2 ) , dopamine concentration ( high or low ) , and the sign of pre- and postsynaptic event timing ( positive or negative ) , but not how it depends on the delay itself . We therefore used these data as the starting-point for a new framework for cortico-striatal plasticity . This framework extrapolates naturally from the data in three ways . First , it extrapolates from the Shen data to the STDP functions described by Pawlak and Kerr . Second , it establishes a simple way of defining plasticity rules over a continuum of dopamine levels , proposing dopamine-dependent STDP . Third , it incorporates an eligibility trace to solve the temporal credit assignment problem—that the change in dopamine level is locked to environmental feedback , and so occurs long after the signals for action are input at cortico-striatal synapses . Figure 4 shows how we interpret the data of Shen and colleagues [17] in terms of STDP functions , generalising from the data of [22] by assuming that each combination of MSN type and sign of pre- and postsynaptic event timing has a standard exponential function of time [42] . The dopamine level in the experiment is assigned one of two values—“high” or “low” ( depleted ) —where the term “high” is simply used as a contrast with “low” and no implication is made that this is a biologically high level . To deal with spike timing , let be a pair of presynaptic and postsynaptic spike times , respectively . Letting , we refer to the conditions , as “positive” and “negative” spike-pair timing , respectively . For a given pair of pre- and postsynaptic events separated by , we model the exponential dependency of plasticity on timing by , where sets the time scale of the exponential decay , and coefficient sets the scale of contribution to plasticity: high values of indicate a larger contribution . The consequent change in weight is , where is a learning rate . We define separate functions for each combination of receptor type ( D1 , D2 ) , dopamine level ( low , high ) , and sign of pre-post event timing ( + , − ) in the Shen and colleagues′ [17] data . As an example consider the case of low dopamine with D1-MSNs shown in the top right panel of Figure 4 . For positive spike timing , the data show clear LTD and so we assign a negative function describing the relation between plasticity change and , with amplitude to capture the LTD in the data ( note the “+” superscript refers to the positivity of , not the sign of the function value; “lo” indicates “low dopamine” ) . Duplicating this whole procedure for all other combinations results in a set of four plasticity coefficients for each of D1 and D2 type MSNs: . Even at this qualitative stage of the model , our distillation of the complex dataset of Shen and colleagues [17] shows that their data imply “standard” STDP ( LTP and LTD in positive and negative timing , respectively ) applies only for D2 MSNs under high dopamine levels; all other combinations of MSN type and dopamine level imply non-standard combinations of LTP and LTD with pre- and postsynaptic spike timing . In order to extend these results to arbitrary levels of dopamine , we define functions for any by smoothly mixing or “blending” the functions at the extremes of the range , and , according to : Figure 5D plots the particular mixing functions used here ( see Methods ) . For a given level of dopamine , the mixing function determines the consequent amplitude of the STDP functions , thus setting the change in weight—we plot these “plasticity factors” for each spike-timing ( + , − ) and receptor type ( D1 , D2 ) in Figure 5C ( D1 ) and 5D ( D2 ) . Figure 5A and 5B plots the resultant two-dimensional STDP functions over the full range of dopamine level for D1 ( Figure 5A ) and D2 ( Figure 5B ) MSNs , showing that various combinations of LTP and LTD emerge naturally from the mixing scheme . In particular , the smooth morphing of the STDP functions predicts that , at some intermediate levels of dopamine , both D1 and D2 MSNs would express “standard” STDP; this case is highlighted by the dark blue lines in Figure 5A and 5B . The parameters of the mixing function were chosen so that this standard STDP in both MSN types occurred at our nominal level of tonic dopamine . We expect such tonic dopamine to be present outside of the learning and extinction epochs ( Figure 1 ) , yet for there to be no change in synaptic strength despite the ongoing pre- and postsynaptic spike-pairings in background spiking activity . We show below that using these standard STDP functions at tonic dopamine levels indeed results in no overall change in synaptic strength outside learning and extinction . In operant conditioning experiments schematised in Figure 1 , at some time during or immediately after the action request , the action is executed , and any environmental consequences made apparent . If unpredicted , these will cause a phasic dopamine signal . The delay between action request and consequence is largely regulated by the physics of the world and can be as much as 1–2 s , or even longer , while still allowing action discovery [43] . There is therefore a temporal credit assignment problem [23] , [24]: for if cortico-striatal plasticity is the proposed locus of reinforcement learning and is dopamine-dependent , how can the transient cortico-striatal action request lead to correct changes in cortico-striatal weights by dopamine signals arriving long afterwards ? Solutions often involve some kind of “eligibility trace” in which pre- and postsynaptic activity at a neuron establishes the potential for plasticity , which is later converted into permanent change with dopamine . Here we adopt the dopamine and STDP-dependent eligibility trace model introduced by Izhikevich [44] , and extend by incorporating the non-standard forms of STDP and the plasticity-function mixing framework described above ( see Methods for a formal description ) . In this model , plasticity is not governed directly by the STDP functions; rather , these are used to establish an eligibility trace , which subsequently decays over time in the order of seconds . It is this trace , together with its interaction with dopamine , that governs synaptic weight change . We therefore refer to this plasticity framework as “spike timing dependent eligibility” ( STDE ) . The process is illustrated for positive spike timing in Figure 6 , which also shows our model of an action request—see below . Each pre- and postsynaptic spike pair for which creates a step-change contribution to an eligibility trace , where is the time dependent STDP function used previously . The eligibility decays exponentially with time constant , where , so the eligibility , due to a single spike pair , is therefore . In contrast to learning under STDP , STDE introduces time-dependence within a single trial of both dopamine level —describing the phasic dopamine response to environmental events ( Figure 6 , green trace ) —and the eligibility trace . Thus each synaptic weight is updated continuously in STDE , with the change at time proportional to both the current state of the eligibility trace and the current dopamine level , as shown in Figure 6 . The magnitude of the change is still given by the dopamine-dependent plasticity factor , but now depends on time . Put together , the change in weight for positive spike-timing is thus proportional to . The plasticity rule may be extended to spike pairs with negative timing by introducing an eligibility . Overall plastic change at a single synapse is then the sum of contributions from both and . Multiple spike pairs are accommodated by assuming their contributions combine linearly . The learning rule was chosen so that , under constant dopamine , STDE reduces to STDP; that is , the overall change in synaptic strength for a spike pair is the same as that in STDP . Later , we show that this STDE model of cortico-striatal plasticity is able to account for the original experimental data of Shen and colleagues [17] . Here , we continue with our programme relating plasticity to operant learning . We now have on the one hand predicted D1 and D2 MSN activity changes over trials of an operant learning task , and on the other an in vitro-derived model for cortico-striatal synaptic plasticity as a function of given pre- and postsynaptic spike timing , MSN type , and dopamine level . Together these allowed us to test the basic hypothesis of reinforcement learning: that adjustment of cortico-striatal weights by value-conditioned environmental feedback , in the form of the phasic dopamine signal , changes which actions are prioritised in future . To do so , we simulated the stylised experiment described above ( Figure 1; see Methods for a formal description ) using our previously developed spiking models of the D1 and D2-type MSNs [45] as representatives of the action-coding populations of D1 and D2 MSNs . The spiking model simulates background synaptic input from cortical ( via AMPA and NMDA receptors ) and intra-striatal ( via GABA receptors ) sources , and incorporates tonic dopamine modulation of the MSN’s excitability . The top panel of Figure 6 shows the model of spiking input and dopamine feedback signals occurring around a single MSN during a single trial of the simulated experiment , comprising a single action and its possible reinforcement . Within each trial we simulate a phasic action request by a subset , , of cortical afferents to the MSN that generate a short burst of spikes with a higher firing rate than background levels , with the remaining afferent subset at background rate . Random action choice in the baseline and intermission epochs are modelled by randomly choosing the active subset of cortical signals , , on each trial . During learning and extinction epochs , the same set of cortical signals representing the reinforced action is transiently active in each trial of the epoch . Where reinforcement was presented ( in learning ) or expected ( in extinction ) the phasic dopamine signal on that trial was delayed by 150 ms . Across trials the magnitude of the dopamine signal changed according to the envelope shown in the bottom panel of Figure 1 . Each AMPA synapse of the model was updated using the STDE rules . Our only free parameters were thus the key plasticity coefficients , but these were constrained to have the correct sign for LTP or for LTD as shown in Figure 4 ( that is , for D1 MSNs , and for D2 MSNs ) . Within these constraints , we easily found coefficients that produced the target changes in activity for both D1 MSNs and D2 MSNs across all epochs of the simulated operant experiment . Figure 7A and 7D shows the resulting change in D1 and D2 MSN activity over the simulated experiment for an example well-performing set of coefficients . Thus , we see that dopamine-modulated STDE synapses can indeed drive the required activity changes in D1 and D2 MSNs despite reinforcement or its omission being delayed beyond the end of the STDP time-window . We particularly note that the two unintuitive properties of the MSN responses derived from the network model arise naturally from the in vitro-derived STDE rules: first , that the reduction in D1-MSN activity over extinction need not drive this activity to zero , or even to the average activity of the preceding intermission epoch; second , that D2-MSN activity does increase during the learning epoch as a consequence of the STDE rules . In Figure S2 and Text S2 we further show that the resultant cortical input weights to the D1 and D2 MSN models from each epoch of the operant task do , in turn , produce the required action selection performance for the whole basal ganglia network model . In both D1 and D2 MSN profiles , we also note there was no change in activity across trials in the baseline , intermission , or post-extinction epochs , showing that our choice of using the “standard” STDP functions at tonic dopamine levels ( Figure 5 ) is indeed sufficient to suppress plastic change overall despite many pairs of pre- and postsynaptic spikes and the presence of dopamine . These activity changes over the course of the experiment were driven by the dopamine-dependent changes in cortical input weights . We plot the evolution of the mean synaptic strengths ( AMPA conductances ) in the fixed afferent set for D1-MSNs and D2-MSNs in Figure 7C and 7F , respectively; illustrative snapshots at trials 1 and 55 of the full synaptic sets are shown alongside in Figure 7B and 7E . There is clear evidence of the development of matching between the patterns of cortical signals and synaptic conductances in the fixed afferent set . Note how , in both MSN types , conductances increase during the learning phase ( compare outcome at key trials 1 and 55 ) , and are preserved during free action choice of the intermission epoch ( compare trials 55 and 85 ) . For D1-MSNs the conductances in decrease during extinction , while for D2-MSNs they increase ( compare across trials 55 and 125 ) . In constructing our target changes in MSN activity over learning we advanced the hypothesis that increased D2 MSN activity in extinction causes active suppression of a previously reinforced action . That this increased activity in extinction emerged from our STDE plasticity model ( Figure 7D ) is partial evidence in support of the hypothesis . To further test this hypothesis , we sought to determine whether the active suppression hypothesis could be reconciled with the post-extinction behavioural phenomenon of renewal ( context-switch evoking immediate display of the previously acquired behaviour ) and reacquisition of the key action ( after a subsequent bout of reinforcement ) [46] . Given that the action-representing weights for D1 MSNs returned to baseline after extinction ( Figure 7C ) , while those for D2 MSNs reached their highest value ( Figure 7F ) , it was not clear that the plasticity model could account for these post-extinction phenomena . In renewal and reacquisition protocols , learning and extinction are carried out in two environments with differing contextual cues that may be visual , structural , or olfactory [47] . Typically an operant task is learned in a context , extinguished in context , or another , and behaviour then tested for renewal or reacquisition in a context different from that used during extinction . This leads to protocols , but results are also sometimes reported for control sequences , in which , unsurprisingly , the “renewal” performance is close to that observed at the end of extinction [48] . Our goal was to test whether synaptic changes due to the STDE plasticity model could both allow renewal and cause reacquisition . To do so , we simulated these protocols using the spiking MSN model with STDE to find the changes in the cortico-striatal synaptic weights; to assess performance at the different stages of the protocols , we took the weights found at these stages and constructed equivalent rate-coded D1 and D2 MSNs , tested the resultant basal ganglia network model’s response behaviour , and compared it to experimental results . We did this for sequences ( test for renewal and reacquisition ) , ( control for the same context in learning and renewal/reacquisition ) , and ( control for the same context in extinction and renewal/reacquisition ) . Figure 8A shows a summary of relevant data from experiments by Nakajima and colleagues [49] ( from their Figure 3 ) on extinction and renewal . We plot there the results of testing response behaviour in the context used for renewal both before extinction ( point labelled ‘acquis . ’—acquisition ) as a control for the effect of changing the context alone , and after extinction ( point labelled ‘renewal’ ) . Figure 8B is a summary of relevant data from experiments in [50] on extinction and reacquisition ( see Figure 2 therein ) —see Methods for details of our interpretation . In order to simulate the use of different contexts with the STDE-equipped MSN spiking model we manipulated the strongly active afferent synapse set . We assumed that 50% of the original set , used to obtain the previous results , is responsible for sensory components common across contexts and , as well as any pre-motor components of the action request for the key action . We then established a new set , which included this 50% of , with the remaining half of its synapses drawn randomly from the set complement . The cortical input under context or then takes the salient input value ( see Figure 6 ) at synapses in and , respectively . Using these input sets , we simulated the three sequences for the renewal protocol , and then tested for reacquisition in context or ( reinstating the phasic dopamine signal in each trial to simulate the reintroduction of reinforcement ) . The behavioural performance at each stage of the simulated sequences was determined by testing the response of the spiking D1 and D2 MSN models to cortical input at that stage ( given their learnt weights ) , and using their responses to parameterise an equivalent rate-coded neuron that captures their learnt responsiveness at that stage of the sequence ( see Methods ) . Embedding these in one channel representing the key action , the resultant basal ganglia network model was then tested with the paired-input protocol used to assess selection ( Figure 3 ) ; the performance metric was the number of selections of the key action channel ( channel 1 ) , corresponding to the numbers of responses in the in vivo experiments . Figure 8C shows that the model’s behavioural performance both before and after extinction is consistent with the data in Figure 8A: there is reduced selection of the key action under context after initial acquisition , selection under renewal is always diminished with respect to corresponding acquisition performance , and selection under renewal in the protocol is greater than that in the and protocols . Figure 8C also shows that the model’s behavioural performance following the subsequent reintroduction of reinforcement is consistent with the data in Figure 8B: requisition allows increased selection , and the ordering under both contexts is preserved . The relative cortico-striatal weight changes in contexts and underpinned these performance changes . Figure 8D shows the trajectory of the mean AMPA conductance of each of the synaptic sets , , under learning with the protocols described above . As we might expect , at the start of extinction ( Trial 1 ) , , since learning has been carried out with respect to . This accounts for the “acquisition” selection results in Figure 8C . In all cases , extinction causes a reduction/increase in mean conductance for D1/D2-MSNs , with both features promoting diminution of selection under “renewal . ” However , the changes with extinction under context for synaptic set are most marked , which explains the correspondingly larger decrease in renewal selection under extinction with . New learning under reacquisition causes increased/reduced conductances for D1/D2-MSNs resulting in the increased selection observed . We thus found that active suppression of the key action by D2 MSNs during extinction could nonetheless give rise to its renewal and reacquisition . Thus far we have shown that in vitro data-derived dopamine-modulated STDP functions are sufficient to generate putative D1 and D2 MSN responses over the course of an operant-learning task . We now ask to what extent this complex set of non-standard STDP functions ( Figure 4 ) are necessary to generate such responses: that is , could the complexity of the three-factor dependency ( on receptor type , dopamine concentration , and spike-timing ) be explained by the need to generate a particular set of MSN responses ? To address this , we performed an exhaustive , “brute-force” search in the 4D parameter space of plasticity coefficients for each MSN type . Full details are supplied in the Methods but , briefly , each search was divided into two stages: a first stage with an extensive parameter range , followed by a more focused search around the best-fitting responses . For each set of plasticity coefficients encountered , we ran a set of the simulated learning experiments to obtain spike count profiles . We then used a feature-based method to define a score to determine how well the profiles matched the targets in Figure 1 . Figure 9 illustrates the search process , and the diversity of activity profiles encountered for D1 MSNs . Figure 10 shows the range of satisfactory plasticity coefficients discovered by the search for both MSN types . Figure 11 shows the range of STDP functions resulting from the distribution of values for each plasticity coefficient that gave good matches to the MSN response profiles . Across the three factors of spike-timing ( negative , positive ) , MSN type ( D1 , D2 ) , and dopamine level ( low , high ) , six of the eight functions were always restricted to the same sign ( LTP or LTD ) as the data of Shen and colleagues [17] . Thus , our model predicts that the dependencies on timing , dopamine-level , and dopamine-receptor for these STDP functions are necessary for the putative MSN response profiles under operant conditioning . However , we also predict some diversity in the necessary learning rules for two functions with negative spike-timing ( ) . For D1 MSNs at high dopamine levels ( Figure 11A , top left panel ) our model predicts the possibility of either LTP or LTD for . The overall sign of plasticity , averaged over randomly chosen pre-post spike timings , is determined by the sum , shown in the plot inset . For D1 MSNs at high dopamine , we therefore predict an overall LTP-like outcome . For D2 MSNs at low dopamine levels ( Figure 11A , lower right ) , our model also predicts the possibility of either LTP or LTD for . However , once again , the overall direction of plasticity is almost always ( with one outlier ) LTP-like with . We derived our cortico-striatal plasticity model by extrapolating and combining Pawlak and Kerr’s [22] report of STDP at cortico-striatal synapses and Shen and colleagues′ [17] data on that plasticity’s dependence on dopamine receptor type , concentration , and the sign of spike-timing , and extending to include arbitrary levels of dopamine and an eligibility trace . Here we answer the question of whether this extrapolated and extended model can capture these underlying data . In Figure 11B we plot the range of STDP kernels predicted by the sets of successful plasticity coefficients from our exhaustive search if , as in the study of Pawlak and Kerr [22] , D1 and D2 MSNs were indistinguishable . We find that the mean kernels give the classic STDP profile and some evidence of LTP at negative spike timings , exactly replicating Pawlak and Kerr’s [22] result . To check that our models could replicate the results of Shen and colleagues [17]—shown in the insets in Figure 4—we simulated their plasticity induction protocols at a single AMPA synapse of the spiking MSN model using the full STDE model . Each condition of D1 or D2-type MSN , “high” or “low” dopamine , and positive or negative spike-pair timing was simulated; details are given in the Methods . The outcomes of the experiment were a set of EPSP-ratios , one per condition , comparing the EPSPs before and after the period of plasticity induction . We simulated such a complete experiment using different sets of successful plasticity coefficients found by the exhaustive search . Figure 12 plots the EPSP-ratios for the data against those obtained using a typical set of coefficients , showing that the sign of plasticity is preserved in all cases and several of the rank-order relations between pairs of experimental conditions are preserved . Thus , the plasticity model parameters necessary for successful action selection and suppression in an operant task are consistent with in vitro data on plasticity at a single cortico-striatal synapse .
In going from in vitro data to learning rules , some interpretation of that data was clearly necessary . For example , we adopted the naturally occurring level of dopamine in the in vitro experiments as the nominally “high” value in setting function parameters . The precise levels of dopamine here may not correspond with the highest values accessible in vivo but this is not critical . Rather , we assume that the trend in parameters is monotonic with dopamine level so that the data determine these trends rather than the values per se . The monotonicity assumption is a key aspect of our framework and more experimental work is required to establish if this is the case . While the data of Shen and colleagues [17] form the most complete picture of the factors controlling cortico-striatal plasticity , our extrapolation to the set of STDP kernels ( Figure 4 ) is based on a particular interpretation of their experimental protocol . They used an asymmetric stimulation protocol with three postsynaptic spikes preceding each pre-synaptic spike in the negative timing condition , but three pairs of pre- then postsynaptic spikes in the positive timing condition , each pair spaced by 15 ms . Thus their positive-timing protocol contains both positive and negative delays , implying that it contains contributions from both positive and negative STDP kernels . In our interpretation , we simplified this by assuming the positive-timing protocol was predominantly receiving contributions from the positive STDP kernel ( Figure 4 ) . Nonetheless , it was encouraging that our unconstrained search returned kernel coefficients with the signs we extrapolated from the Shen and colleagues’ data , and recovered the generic MSN STDP kernel reported by [22] . A further common limitation for any extrapolation from in vitro work to in vivo application is that many of the in vivo-like conditions are intentionally removed during in vitro studies to provide close control over the experimental question at hand . For the Shen and colleagues’ [17] data , these include the injection of current to hold the membrane potential close to −70 mV , thus minimising the impact of NMDA receptors , and the use of GABAa antagonists to prevent any effect of inhibition ( which may play a key role in STDP [53] ) . Despite these limitations , we showed that the single spiking MSN models with our plasticity rules could produce the required activity profiles over an operant task even though they incorporated input to both NMDA and GABAa synapses . Also missing in vitro are the dynamics of the intra-striatal signals in vivo that may directly or indirectly affect plasticity at cortical synapses on MSNs , particularly those originating from the interneurons . As well as GABAergic signals from the fast-spiking interneurons , cholinergic interneurons may play a dual role through both postsynaptic modulation of plasticity [54] and the shaping of dopamine release in the striatum [55] . Thus , a complete systems model of cortico-striatal plasticity will require the integration of synaptic and network level contributions . Finally , STDP is a phenomenological description at the level of spikes of a set of intra-cellular signalling processes , and more detailed modelling of those processes ( e . g . , [56]–[59] ) will be essential to shed light on the effects of spiking history , of dopamine’s triggering of intra-cellular signalling cascades , and particularly on the discontinuity at . The plasticity rules developed here are consistent with a range of interpretations of the origin of the phasic dopamine signal . They are consistent with the dominant hypothesis that phasic firing of dopamine neurons encodes a reward prediction error [2] , [3] , [5] , [37] , [38] . However , we note that they are also consistent with our recent proposal that phasic dopamine is , in part , associated with a sensory prediction error that can enable intrinsically motivated action discovery [4] , [60] . Here , serendipitous interaction with the environment to effect some predictable outcome therein , can cause learning of the contingency between action and outcome . Recently [61] we have tested the ability of the plasticity rules developed here to effect action discovery by embedding a model of the basal ganglia , equipped with these rules , in a simulated behaving agent that can learn simple action outcome associations . The agent was able to successfully learn the associations and , moreover , the specific plasticity rules described here demonstrated superior performance to a range of plausible alternatives . There have been numerous attempts to model the learning taking place in basal ganglia and that identify the locus of plasticity as the cortico-striatal connections . Many of these models use a temporal difference ( TD ) learning rule or variants therein; for a recent review see [62] . The learning signal in TD algorithms is an “error” or discrepancy between a predicted reward and the actual value received . The error is derived from algorithms grounded in machine learning [63] , but , in biological terms , it is often identified with phasic dopamine [2] , [64] . In contrast , we have no algorithmic origin for phasic dopamine because our account does not address this level of description ( the dynamics of dopamine are described phenomenologically ) . Nevertheless , we might , in principle , attempt to map components of the TD “rule” onto mechanisms we have described here . This exercise would probably fail however , as the the TD rule is not inherently of the three-factor kind in which our framework sits; that is , it does not explicitly include pre- and postsynaptic firing , and an error/dopamine modulatory term . The difficulties encountered with mapping TD in this way have been discussed at length by Worgotter and Porr [65] . However , this does not preclude our plasticity framework from supporting operant learning in which phasic dopamine is obtained algorithmically from internal models of prediction . Indeed , we have recently demonstrated such a model in complete cortico-basal ganglia-thalamic loops , embodied in a behaving agent [61] . This model showed how our plasticity rules have rate-coded ( non-spiking ) equivalents that are part of the well-known BCM family of learning rules [66] , [67] . This was made possible because of the intimate relation between BCM rules and STDP [68] . A key distinction in instrumental learning tasks is made between goal-directed and habitual behaviour . An animal expressing goal-directed behaviour modifies that behaviour in response to a change in the value of its outcome or in the contingency between the action and the outcome; one expressing habit behaviour does not [9] , [69] , [70] . The inference is then drawn that goal-directed animals have access to explicit representations of outcomes linked to actions to guide behavioural choice , which are updated after changes to the outcome irrespective of performing the action . By contrast , habitual animals make behavioural choices on the basis of stimulus-response pairings and can only update this association after repeatedly performing the action cued by the stimulus [69] , [71] . Habitual and goal-directed behaviour have been respectively linked to the dorsolateral and dorsomedial striatum [9]–[11] , [72] . Lesioning the dorsolateral striatum [73]–[75] or disrupting dopamine signalling within it [76] prevent habit formation . Correspondingly , there is a re-organisation of single neuron activity in the dorsolateral striatum during habit formation [26] , [27] , [29] , [75] . Lesioning the dorsomedial striatum [74] , [75] , [77] prevents sensitivity to devaluation or contingency changes . Recent studies of comparative plasticity have shown that only the dorsomedial striatum has evidence of synaptic plasticity unique to goal-directed learning [78] , [79] . Together , these data raise the key question of what differs between circuits containing the dorsomedial striatum and dorsolateral striatum that ultimately results in goal-directed and habitual behaviour [71] . Our model framework here has three separate components: ( 1 ) models of the signals from cortex and of dopamine release , both per trial and their changes over trials; ( 2 ) a synaptic-level plasticity model ( dopamine-dependent STDP ) ; and ( 3 ) a circuit-level action selection model . Any or all of these could be a source of difference between dorsomedial and dorsolateral striatum , and hence candidates for the difference between goal-directed and habitual behaviour . We consider the first two here , as basal ganglia circuitry is well-conserved between regions [80] ( but see [81] ) and it is not immediately clear how differences in the action selection mechanism could differentiate between outcome-driven and stimulus-driven behaviour . Together , model components 1 and 2 reinforce an action by increasing the probability of its selection on a subsequent trial , and do this by increasing the influence of a fixed salience signal from cortex over the basal ganglia selection process . In this respect , the model mechanisms are neutral as to whether the action request from cortex is primed by a representation of the outcome to follow ( goal-directed ) or a representation of the preceding stimulus ( habitual ) . However , for simplicity we assumed throughout that the input from cortex had the same salience on every trial whether the outcome was delivered or not , and so did not reflect changes in value . Thus , our model of inputs is currently consistent only with stimulus-response behaviour , and therefore our model framework as a whole is most consistent with the dorsolateral striatum . Nevertheless , within this framework , component 2 ( the synaptic-level plasticity model ) remains neutral to the goal/habit distinction . Extending our model framework to account for goal-directed behaviour would require identifying where information about value or contingency become encoded . Dorsolateral and dorsomedial striatum receive inputs from different cortical regions [82] and so one possibility is that only the action-request inputs to dorsomedial striatum encode value and contingency information . One candidate here is orbitofrontal cortex: it projects to the dorsomedial striatum [83] , its neurons’ activity represents the expected value of an action [84] , [85] , and optogenetic stimulation of its projection neurons promotes the maintenance of action during extinction [75] consistent with their encoding of value . In this view , changes to value or contingency update their representations in cortex and are reflected in the changed salience of the action request to striatum , allowing for more rapid changes to behaviour than could occur solely via synaptic plasticity . A particular challenge for this view are non-contingent reinstatement phenomena where an action is immediately re-energised after extinction by a single non-contingent presentation of its pre-extinction outcome [86] . For if goal-directed behaviour is driven by the rapidly diminishing salience of an action during extinction , then reinstatement forces us to assume that a single outcome presentation is sufficient to restore that salience . Another possibility is that the dopamine signal is not the same in dorsomedial and dorsolateral striatum , as we have assumed here . Separate midbrain dopamine systems project to these regions [81] , [87] , [88] . Reflecting this , intact dopamine signalling in dorsolateral striatum is necessary for the formation of habitual behaviour [76] , and blunting dopamine signalling prevents the formation of habitual behaviour but does not prevent goal-directed behaviour [89] . In this view , changes to value and contingency would be reflected by the evoked dopamine signal in dorsomedial striatum and not in dorsolateral striatum , and thus appropriately modulate cortico-striatal plasticity only in dorsomedial striatum . Particular challenges for this view are that dopamine signals to the striatum seem to encode the same information everywhere [90] ( but see [91] ) and the speed of change—if behavioural change depends solely on synaptic plasticity , then behaviour is likely altered slowly but the goal-directed system seems to rapidly adapt [71] . A further possibility ( which challenges our synaptic-level neutrality ) is that dopamine-dependent STDP is different between the dorsolateral and dorsomedial striatum , so that even with the same input signals ( cortical and dopaminergic ) , the cortico-striatal weights are updated differently between the two regions . There is good evidence that synaptic weight change differs between the two regions in both skill-learning [92] and goal-directed learning [79] , though these data cannot distinguish between whether the inputs differed , thus differentially recruiting the same plasticity mechanism , or the mechanism of plasticity itself differed . Consistent with the latter , in vitro work has suggested differences in high-frequency stimulation induced LTP between medial and lateral striatum [93] . In this view , for the synaptic plasticity rules themselves to reflect changes to outcome in dorsomedial and not dorsolateral striatum , it follows that the outcome-related signals ( cortical and/or dopaminergic ) must be input to both areas , but that the plasticity mechanisms are sensitive to changes in these inputs only in dorsomedial and not dorsolateral striatum . Again a particular challenge for this view is the speed of behavioural change for goal-directed behaviours if they are solely dependent on synaptic plasticity and not on computations performed elsewhere [71] . The above ideas are naturally speculative , reflecting the current lack of data on the precise relationship between different forms of behaviour and the details of cortico-striatal plasticity in different striatal regions [70] . A contribution of our model framework is that by bridging the levels from a single synapse to overt behaviour it provides a basis for framing the alternative hypothesises and their implications . Our search for the necessary plasticity coefficients to generate the D1 and D2 MSN activity profiles predicts that two of the eight coefficients could be positive or negative ( Figure 11 ) . Thus , for D1-MSNs at high levels of dopamine and for D2-MSNs at low dopamine levels , there is a possibility of LTD or LTP for negative spike-pair timing . This apparent ambiguity may be resolved in two ways: ( i ) that there is a corresponding variation of plasticity rules across individual MSNs ( or even individual synapses ) in an individual animal brain; or ( ii ) that these rules are subject to constraints that lie outside our framework , and thus in vivo all combinations of LTP and LTD are those we inferred from the Shen and colleagues′ [17] data ( Figure 4 ) . Such constraints could include that the specific dopamine-activated intracellular signaling pathways that ultimately give rise to changes in plasticity can allow only a single direction of change for a given combination of dopamine receptor and level , and consequently can only express one of LTD or LTP at a single synapse for that combination . We hypothesised that extinction in operant learning involves active suppression of the action by D2 MSNs , not ( solely ) unlearning of the action at cortico-striatal synapses onto D1 MSNs . While this is compatible with modern theories of behaviour that posit that extinction is not a simple unlearning of previous competence [46] , it leaves open the question of how post-extinction phenomena of spontaneous recovery of action can occur if the action is actively suppressed . We showed our model nonetheless could account for both phenomena of contextual renewal ( immediate recovery of extinguished action in new context ) and reacquisition ( rapid re-learning of extinguished action ) . This occurred because , in extinction , we predict that D1-MSN synaptic conductances would regress to their original untrained state only when extinction and post-extinction testing were in the same context , and so a change of context allows rapid recovery of action . Thus in our model spontaneous post-extinction recovery arises solely from the plasticity rules without recourse to additional hypotheses such as state-space splitting proposed by the model of Redish and colleagues [94] . The complexities of cortico-striatal plasticity’s dependence on dopamine receptor-type , dopamine level and spike-timing mean that inferring the effect of changes in these factors is fraught with difficulty , and models are necessary to guide us . Simplifying such models in turn provides us with useful heuristic guides . On the basis of the data available at the time , Reynolds and Wickens [15] sketched a widely used and valuable heuristic guide to the overall direction of weight change at cortico-striatal synapses as a function of dopamine concentration ( see Figure 4 in [15] ) . Our data-derived cortico-striatal plasticity model predicts a smooth morphing of STDP kernels with changing levels of dopamine , switching gradually from LTP to LTD . We can thus use our model to update the heuristic guide to the dopamine-dependence of plastic change , and importantly separate the effects on D1 and D2 MSNs . In Figure 13 we plot the sum of the STDP kernel amplitudes as a function of dopamine concentration , which approximates the expected overall weight change for random trains of input and output spikes , for every successful coefficient set from the exhaustive search . The range of weight changes shown are hence consistent with successful action selection and suppression of the key action . We see that , if we plot the equivalent curve to that in [15] by not distinguishing D1 and D2 MSNs , then our model predicts that the average total measured weight change approximates the curve in [15] . However , the range of total weight change we observed , consistent with successful selection of the key action , covers both LTD and LTP at many dopamine levels . This is accounted for in the model by its prediction that increasing dopamine switches D1 MSN synapses from LTD to LTP and D2 MSN synapses from LTP to LTD . Our results thus suggest that the dependence on both dopamine receptor and dopamine concentration forms the minimal model of cortico-striatal plasticity .
Figure 2A shows the basal ganglia network implemented by the model ( see [34] , [80] , [95] for a detailed discussion of assumptions behind this architecture ) . Each action is encoded in a discrete “channel” throughout the model . Within each nucleus , each channel is represented by a single , rate-coded leaky-integrator unit whose output stands for the mean activity of a population of neurons that might instantiate the channel in vivo . The assumption of a channel architecture is based on the long-standing concept of parallel anatomical loops running throughout the basal ganglia nuclei [96] , [97] . Both anatomical and electrophysiological evidence points to the existence of channels representing discrete actions . For example , the somatotopic map found within the striatal motor territory is maintained throughout the basal ganglia circuit , such that there are separate channels for arm , leg , and face representations [18] , [98] . Similar topographic maps have been proposed for the other macroscopic channels [18] . Moreover , within these limb representations , there are discrete channels corresponding to particular movements , demonstrated in striatum by microstimulation [99] and markers for metabolic activity during behavior [100] . Recently , Fan and colleagues [101] provided a compelling demonstration that basal ganglia output neurons coding for selection of the same action are physically clustered , just as predicted by the channel architecture . Cortical input to each channel represents the “salience” of that action . In general , the salience of an action at any given moment will depend on the integration of diverse information on current motor commands , sensory information , and context by convergent inputs to individual MSNs [13] , [80] , [102] . For the rate-coding model of the basal ganglia network , we collapse this into a single scalar value for the salience of the represented action , as we are interested in the ability of the network model to perform selection or suppression on the basis of this salience signal , not in how that signal is computed . Consistent with this assumption , a recent optogenetic study has shown that selecting an action is controlled by the activity of cortico-striatal neurons in sensory cortex [103] . For the spiking MSN model , we explicitly represent changes in context by altering the sub-set of active cortical inputs ( detailed below ) , and thus simulate how salience is dependent on context . Competition between channels for behavioural expression is provided in a “selection pathway” comprising D1-MSNs , STN , and the output nuclei that form a feedforward , off-centre , on-surround network . The circuit with STN , D2-MSNs , and GPe acts to moderate the overall levels of excitation and inhibition in the selection pathway and also perform action suppression for individual channels ( Figure 2B ) . The average activity of all neurons comprising a channel’s population changes according to ( 1 ) where is a time constant and is summed , weighted input . We used ms throughout . The normalised firing rate of the unit is given by a piecewise linear output function ( 2 ) with threshold . Negative thresholds thus ensure spontaneous output , which we use to ensure STN , GPe , and GPi/SNr have tonic output ( see below ) . The following describes net input and output for the channel of each structure , with channels in total . The full model is given by [35]: Striatum D1: , Striatum D2: , Subthalamic nucleus: , Globus pallidus external segment: GPi/SNr: , Each cortical signal simulating an action request was input to channel in the D1-MSN , D2-MSN , and STN populations . The network model included opposite effects of activating D1 and D2 receptors on MSN activity: D1 activation facilitated cortical efficacy at the input , while D2 activation attenuated this efficacy [45] , [104] , [105] . Thus , if the relative activation of D1 and D2 receptors by tonic dopamine are , then the increase in efficacy due to D1 receptor activation was given by ; the decrease in efficacy due to D2 receptor activation was given by . In the implementation used here , the model had six channels but only two were actively driven by cortical input . The other channels are required , however , as they have quiescent firing rates in STN and GPe that contribute to overall activity . We used this model to predict the relative responsiveness of D1 and D2 MSNs that optimised selection of an action ( emulating the target situation at the end of the learning epoch ) and , separately , that optimised the suppression of an action ( emulating the target situation at the end of the extinction epoch ) . The ability to select a particular action can only by tested with reference to at least one other possible alternative action , so we considered two competing signals , one signal representing a fixed “control” action , available for selection throughout , and another signal representing the key action learnt and extinguished over the course of the experiment . We input this pair of salient signals ( ) to two channels in the model , respectively termed the control ( subscript ) and experimental channel ( subscript ) . For a given pair of inputs , we read out the outcome of the competition from the output of the basal ganglia ( SNr/GPi in Figure 2 ) : signalled a sufficient fall in GPi’s tonic inhibition for selection of the corresponding action on channel . Each input pair thus had four possible outcomes: no selection , control channel selected , experimental channel selected , or dual selection . The ideal selector outcomes were then defined as follows . For both learning and extinction we demanded that no action be selected if both inputs ( ) were less than the MSN output threshold . After action learning we required that , if , then the experimental channel is selected , and if , the control channel is selected; if , then no selection is required . After extinction of a previously learned action represented by the experimental channel , we required that that channel is never selected no matter what the value of —representing suppression of that action—and that the control channel is selected if . The salience pairs were constructed by allowing each of to range over a set of ten discrete values in the interval . The set of ideal outcomes ( for each of learning and extinction ) over all 100 salience pairings constitutes an ideal selector template for model comparison , and these are plotted in Figure 3 for learning ( Figure 3A ) and extinction ( Figure 3B ) , with experimental and control channels being identified with channels 1 and 2 , respectively . For each of the 100 input pairs , the input on the experimental and control channels occurred at t = 1 s , and t = 2 s , respectively . The GPi output was read out at equilibrium , and the simulation time-step was 0 . 01 s . Over all 100 input pairs , the model performance was then compared to the template , and summarised as a percentage match . The ability of the network model to match these two templates was tested by varying the relative “responsiveness” to input of the D1 and D2 MSN populations of the experimental channel . Responsivess is defined here as the ratio of the input to output value for the population . As both the cortico-striatal input weights and the level of tonic dopamine affect responsiveness , for this channel alone we set and varied the D1 ( ) and D2 ( ) MSN input weights independently over the range . To allow us to investigate a full range of MSN behaviour , we dropped the saturation requirement on the output ( condition ( iii ) in Equation 2 ) . For the control channel , we set and the input weights to = = 1 , following our prior models [35] . Here , we give details of the plasticity framework that incorporates the three factors of postsynaptic neuron type , dopamine concentration , and spike-timing at the scale of STDP . All parameters are collected together in Table 1 . We start by assuming constant dopamine and STDP ( no eligibility ) . Let be a pair of postsynaptic and presynaptic spike times respectively , and put . For each of the two classes , D1- , D2-MSNs we define STDP functions ( kernels ) for the following four cases: ( 3 ) We define functions for any , by “mixing” the functions at the extremes of the range , and ( see Figure 5 ) . We use a simple linear blending scheme ( 4 ) where the mixing functions for each of D1- and D2-MSNS are shown in Figure 5D . It is conveniently expressed by a Naka-Rushton equation ( 5 ) but no special significance is assigned to this form; all that is required is a rapidly increasing , then saturating , monotonic function of with no point of inflexion . The parameters were chosen to ensure: ( i ) over the range of dopamine level used; ( ii ) that , for each of D1- and D2-MSNs , with typical plasticity coefficients consistent with the data in [17] , there is little or no overall plastic change at tonic levels of dopamine . In extending the formalism further to incorporate eligibility ( next section ) , it is useful to rewrite ( 4 ) in an alternative form ( 6 ) We refer to the as “plasticity factors , ” and plot them in Figure 5C and 5E . For STDP , the resulting change in synaptic weight due to a single pre-post spike pair is given by ( 7 ) where is a learning rate . We base our eligibility trace model on that of Izhikevich [44] , extending to incorporate arbitrary levels of dopamine , and testing its application across all forms of non-standard STDP we observe for cortico-striatal synapses . The basic idea is that each spike pair creates a step-change contribution to a corresponding eligibility trace , where are the normalised STDP functions defined in ( 3 ) , and the positive/negative sign applies according to whether or . The step change for either can be positive or negative , corresponding to a potential increase ( LTP ) or decrease ( LTD ) in synaptic weight . The eligibility decays exponentially with time constant , so the eligibility , due to a single spike pair , is . The process is illustrated for positive spike timing in Figure 6 . Synaptic weights are updated according to ( 8 ) where are functions of the ( possibly changing ) dopamine level , and is a learning rate . We now put , where are the plasticity factors given by ( 6 ) , but allowing time-dependent dopamine . Then , using the first relation in ( 6 ) , the learning rule for a single spike pair becomes ( 9 ) Here , the factor is given by the same functional form as ( 4 ) but now has a time-dependence with dynamically changing dopamine . The effects of multiple spike pairs are assumed to add linearly . The complete STDE learning rule for a single synapse is thus given by Equation 9 , which uses the STDP kernel from Equation 4 defined by mixing the extreme STDP kernels in Equation 3 with the mixing function in Equation 5 . The dynamic dopamine level is specified by the modeller: for our simulated operant conditioning experiment we specify the within- and between-trial changes in dopamine below . The choice of learning rule for STDE was dictated by the constraint that STDE reduces to STDP for constant levels of dopamine . Thus , integrating ( 9 ) gives the total change in weight due to the spike pair and , for constant dopamine , this is equal to the change for STDP in Equation ( 7 ) ( up to the time constant , which may be absorbed into ) . The spiking model MSN is based on that in [45] . Essentially , this is an Izhikevich model [106] of a MSN , with the addition of direct dopaminergic modulation of both synaptically induced and intrinsic membrane currents . In the biophysical form of the Izhikevich model neuron [107] , is the membrane potential and the “recovery variable” is the contribution of the neuron class’s dominant ion channel: ( 10 ) ( 11 ) with reset condition if then , where , in the equation for the membrane potential ( 10 ) , is capacitance , and are the resting and threshold potentials , is the current due to synaptic input , and is the reset potential . Parameter is a time constant governing the time scale of the dominant ion channel . Parameters and are arbitrary scaling constants , with the sign of controlling whether the neuron is an integrator ( ) or a resonator ( ) . Parameter describes the after spike reset of recovery variable , and can be tuned to modify the rate of spiking output . The MSN model’s parameter values and their sources are given in Table 2 . In [45] we showed how this model can capture key dynamical phenomena of the MSN the slow-rise to first spike following current injection; paired-pulse facilitation lasting hundreds of milliseconds; and bimodal membrane behaviour emulating up- and down-state activity under anaesthesia and in stimulated slice preparations . Synaptic input comprises the source of current in Equation 10: ( 12 ) where , , are current input from AMPA , GABA , and NMDA receptors , respectively , and is a term that models the voltage-dependent magnesium plug in the NMDA receptors . Each synaptic input type ( where is one of ampa , nmda , gaba ) is modelled by ( 13 ) where is the maximum conductance and is the reversal potential . We use the standard single-exponential model of postsynaptic currents ( 14 ) where is the appropriate synaptic time constant , and is the number of pre-synaptic spikes arriving at all the neuron’s receptors of type at time . The term in Equation ( 12 ) is given by [108] ( 15 ) where is the equilibrium concentration of magnesium ions . Synaptic conductances were initialised with Gaussian noise so that they have a coefficient of variation of 0 . 1 . Any synapses with negative conductance as a result of this initialisation was set to . There was a ceiling on the synaptic conductance of . The following models of dopamine modulation are detailed in [45] . Let and be the proportion of activated D1 and D2 receptors . For activation of D1 receptors we used the linear mappings: ( 16 ) and ( 17 ) which respectively model the D1-receptor mediated enhancement of the inward-rectifying potassium current ( KIR ) ( 16 ) and enhancement of the L-type Ca2+ current ( 17 ) . For activation of D2 receptors we used the linear mapping: ( 18 ) which models the the small inhibitory effect on the slow A-type potassium current , increasing the neuron’s rheobase current [105] . We add D1 receptor modulation of NMDA receptor evoked EPSPs by ( 19 ) and we add D2 receptor modulation of AMPA receptor evoked EPSPs by ( 20 ) where and are scaling coefficients determining the relationship between dopamine receptor occupancy and the effect magnitude . The dopamine dependent factors used in the dopamine-modulated neuron model are related to dopamine level by , where . This ensured that , for most of the phasic dopamine signal , are both almost 1 . The neuron incorporated excitatory and inhibitory ( GABAergic ) synapses , with . Each excitatory synapse contained a model of NMDA and AMPA receptors , as described above . Every synapse received a Poisson train of spikes at some specified firing rate . For the main experiments with operant learning , the GABAergic synapses received background input at three spikes/s; for the replication of the STDP protocols , they received no input . The firing rates of the excitatory synapses are detailed below . Details are given here of the search for plasticity coefficients that give rise to MSN response profiles of the form in Figure 1 . The 4D space of coefficients was divided into a regular rectangular lattice defined by the intersection of five regularly spaced points along each of the axes ( giving 625 points ) . This was augmented by a point corresponding to the coefficients used in the data-constrained experiments reported in Figure 7 . At each lattice point , three experiments were run using the experiment defined by Figure 1 , but the numbers of trials in some epochs were reduced to expedite computation . Thus , for D1-MSNs , the number of trials in each epoch ( baseline , learning , intermission , extinction , post-extinction ) was reduced to 15 , 30 , 30 , 20 , 15 , respectively , and for D2-MSNs , to 15 , 40 , 30 , 20 , 15 . Initially , the lattice was rather coarse grained with a liberal range of values; we were keen not to exclude any non-intuitive combinations of coefficient values . For D1 MSNs , the lattice was defined by drawing the coefficients from the five equi-spaced values across the following intervals: , . For D2 MSNs the intervals were . However , a second search was then conducted using a smaller lattice , whose domain was restricted by the more successful experiments from the first pass . For D1 MSNs this was given by , and for D2 MSNs by For each group of three experiments at each lattice point , the spike counts at each trial were averaged over this group , and across a window of three trials . These smoothed , ensemble-mean spike counts were then characterised with a feature-based metric in terms of their match to the target profiles in Figure 1 . This metric was used as a guide for selecting MSNs with well-matched profiles , and fit to the target was ultimately corroborated by visual inspection ( any feature-based method is only as good as the quality of the features it uses ) . We simulated the cortico-striatal plasticity induction protocols described in Shen and colleagues [17] using the spiking MSN model with a single AMPA synapse . They used a theta-burst protocol , with an asymmetric design for the positive ( pre-post ) and negative ( post-pre ) spike timing tests . For the pre-post test , each burst was three pre-synaptically induced EPSPs spaced by 20 ms , each EPSP followed by a fictive postsynaptic spike after 5 ms . For the post-pre test , each burst was three fictive postsynaptic spikes spaced by 20 ms , the last spike followed by a pre-synaptically induced EPSP after 10 ms . For both tests , the bursts were presented in blocks of 5 at 5 Hz ( that is , the first event of a burst occured every 200 ms ) , and ten blocks were presented at 0 . 1 Hz ( i . e . , every 10 s ) . To simulate this protocol we used a single synaptic input obeying the STDE rules to which we applied afferent spikes , and generated artificial postsynaptic spikes with the correct timing relations . The only difference was the extended period of time between blocks of stimuli was reduced to 2 s to avoid unnecessarily large simulation times ( the neural membrane had returned to rest over this time , and all time constants in the model are substantially shorter than 2 s ) . Ten blocks of stimuli with potential plasticity were used , sandwiched between blocks with no plasticity ( learning rate of zero ) , which served to allow measurement of mean EPSPs before and after learning . In line with the protocol of Shen and colleagues [17] , the membrane potential was set to an initial holding value of −70 mV ( by current injection ) . At no time were any spontaneous action potentials generated so that all spike pairs were synthetically created by the spike-pair timing protocol .
|
A key component of survival is the ability to learn which actions , in what contexts , yield useful and rewarding outcomes . Actions are encoded in the brain in the cortex but , as many actions are possible at any one time , there needs to be a mechanism to select which one is to be performed . This problem of action selection is mediated by a set of nuclei known as the basal ganglia , which receive convergent “action requests” from all over the cortex and select the one that is currently most important . Working out which is most important is determined by the strength of the input from each action request: the stronger the connection , the more important that action . Understanding learning thus requires understanding how that strength is changed by the outcome of each action . We built a computational model that demonstrates how the brain's internal signal for outcome ( carried by the neurotransmitter dopamine ) changes the strength of these cortical connections to learn the selection of rewarded actions , and the suppression of unrewarded ones . Our model shows how several known signals in the brain work together to shape the influence of cortical inputs to the basal ganglia at the interface between our actions and their outcomes .
|
[
"Abstract",
"Introduction",
"Results",
"Discussion",
"Methods"
] |
[
"developmental",
"neuroscience",
"cellular",
"neuroscience",
"synaptic",
"plasticity",
"behavioral",
"neuroscience",
"computational",
"neuroscience",
"single",
"neuron",
"function",
"biology",
"and",
"life",
"sciences",
"computational",
"biology",
"neuroscience",
"neuronal",
"plasticity"
] |
2015
|
A New Framework for Cortico-Striatal Plasticity: Behavioural Theory Meets In Vitro Data at the Reinforcement-Action Interface
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.