text
stringlengths 87
880k
| pmid
stringlengths 1
8
| accession_id
stringlengths 9
10
| license
stringclasses 2
values | last_updated
stringlengths 19
19
| retracted
stringclasses 2
values | citation
stringlengths 22
94
| decoded_as
stringclasses 2
values | journal
stringlengths 3
48
| year
int32 1.95k
2.02k
| doi
stringlengths 3
61
| oa_subset
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1635525310.1371/journal.pcbi.001006905-PLCB-RA-0153R2plcb-01-07-02Research ArticleBioinformatics - Computational BiologyDevelopmentEvolutionGenetics/GenomicsGenetics/Comparative GenomicsGenetics/EvolutionGenetics/Gene ExpressionSystems BiologyDrosophilaNematodesRevealing Posttranscriptional Regulatory Elements Through Network-Level Conservation Posttranscriptional NetworksChan Chang S Elemento Olivier Tavazoie Saeed *Department of Molecular Biology and The Lewis-Sigler Institute for Integrative Genomics, Princeton University, Princeton, New Jersey, United States of AmericaMattick John EditorUniversity of Queensland, Australia* To whom correspondence should be addressed. E-mail: [email protected] 2005 9 12 2005 2 11 2005 1 7 e698 7 2005 2 11 2005 Copyright: © 2005 Chan et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.We used network-level conservation between pairs of fly (Drosophila melanogaster/D. pseudoobscura) and worm (Caenorhabditis elegans/C. briggsae) genomes to detect highly conserved mRNA motifs in 3′ untranslated regions. Many of these elements are complementary to the 5′ extremity of known microRNAs (miRNAs), and likely correspond to their target sites. We also identify known targets of RNA-binding proteins, and many novel sites not yet known to be functional. Coherent sets of genes with similar function often bear the same conserved elements, providing new insights into their cellular functions. We also show that target sites for distinct miRNAs are often simultaneously conserved, suggesting combinatorial regulation by multiple miRNAs. A genome-wide search for conserved stem-loops, containing complementary sequences to the novel sites, revealed many new candidate miRNAs that likely target them. We also provide evidence that posttranscriptional networks have undergone extensive rewiring across distant phyla, despite strong conservation of regulatory elements themselves.
Synopsis
Organisms have evolved extensive regulatory mechanisms for the appropriate expression of genes within precise spatiotemporal contexts. Until recently most of this regulation was thought to be implemented by processes that operate at the “transcriptional” level, that is, by modifying the rate at which mRNA is synthesized. The discovery of short RNAs, termed microRNAs (miRNAs), which can affect gene expression either by degradation of target mRNAs or by inhibiting their translation, has focused much recent effort on determining their specific functional roles and the extent to which they contribute to establishing protein repertoires within individual cells. Chan and colleagues have applied a computational comparative genomic approach for identifying the targets of these miRNAs within 3′ untranslated regions of mRNAs in closely related flies and worms. Their approach identifies a large number of target genes for most of the known miRNAs in these species, providing evidence that these regulators have a much more extensive role than previously thought. The sets of genes targeted by each miRNA are enriched in various known functional classes, providing strong clues for their role in physiology and development. The authors went on to identify many novel miRNAs based on the sequence of highly conserved target sites. They also found a large number of targets that do not correspond to miRNAs, some of which match the targets of known RNA-binding proteins. By comparing the large catalog of putative regulatory elements between flies and worms, they show that, although a large fraction of these elements are conserved, they are targeting, by and large, different sets of genes.
Citation:Chan CS, Elemento O, Tavazoie S (2005) Revealing posttranscriptional regulatory elements through network-level conservation. PLoS Comput Biol 1(7): e69.
==== Body
Introduction
Complex cellular and developmental processes depend on precise spatiotemporal regulation of mRNA and protein levels and activities. Such regulation arises essentially at the transcriptional, posttranscriptional, and posttranslational levels. While tremendous progress has been made in understanding transcriptional regulation and in mapping transcriptional regulatory networks, posttranscriptional regulatory networks are only beginning to be uncovered. Posttranscriptional regulation has been shown to arise through both protein-RNA and RNA-RNA interactions. RNA-binding proteins have been implicated in many aspects of posttranscriptional regulation, e.g., RNA processing, export, localization, degradation, and translational efficiency. Posttranscriptional regulation through RNA-RNA interactions has recently received much attention, in large part due to the discovery of microRNAs (miRNAs) [1].
miRNAs are 21- to 23-nucleotide (nt) single-stranded RNAs, derived from stem-loop precursors. It has been demonstrated that miRNAs regulate mRNA expression either by inducing degradation of the targeted transcript or by decreasing translational efficiency [2]. Recent studies suggest that only the degree of complementarity between a miRNA and its target determines the nature of regulation [2]. Targets with strong complementarity to the miRNA are cleaved by the RNA-induced silencing complex [3]. Such targets appear to be common in plants but rare in animals [4,5]. In some cases, targets with weaker complementarity appear to have decreased translational efficiency, although the molecular mechanism for this repression is currently unknown. It was also recently shown that some miRNAs might be involved in mRNA degradation in animals [6]. Indeed, decreased mRNA levels (in human HeLa cells) were observed for dozens of genes upon transfection of two distinct miRNAs, miR-1 and miR-124; it was also shown that the 3′ untranslated regions (UTRs) of these down-regulated mRNAs have significant complementarity to the 5′ extremity of the transfected miRNAs.
While hundreds of animal miRNAs have been discovered [7], it is likely that many more have not, probably because the conditions under which they are expressed are not known, or because they may be expressed at very low levels. Moreover, very few targets of known miRNAs have been fully characterized experimentally [8]. However, these studies, along with computational ones, indicate that complementarity between miRNAs and their targets is stronger in the 5′ extremity of the miRNAs. Several computational and experimental efforts, based on features of the few verified miRNA/target duplexes, have been directed at finding targets for known miRNAs in the Drosophila melanogaster transcriptome [9–11]. In a recent study, human miRNA targets were predicted on the basis of conservation of the seed (the 6-nt sequence at the 5′ extremity of the miRNA) in multiple alignments of five vertebrates [12]. Based on their results, the authors suggest that up to one-third of human genes may be regulated by miRNAs. In another recent study [13], 3′UTR alignments from four mammalian genomes were used to identify highly conserved targets of known miRNAs. Other highly conserved short sequences within their alignments were subsequently used to discover novel miRNAs.
In this paper, we use network-level conservation [14,15] to show that many motifs are highly conserved in the 3′UTRs of orthologous genes from pairs of fly and worm genomes. We show that many of these highly conserved short sequences are complementary to the 5′ extremity of known miRNAs. We show that our approach naturally defines sets of putative target genes for each of these miRNAs, and that some of the target sets are enriched for genes within specific functional categories, shedding new light on miRNA involvement in these processes. Our approach also discovers known sites for RNA-binding proteins, motifs known to be involved in mRNA decay in other species, and many novel sites that are strongly associated with specific functional enrichments. We show that some of the highly conserved sites are often simultaneously conserved within the same 3′UTRs, suggesting combinatorial regulation of these transcripts. Since our approach uncovers many sites that are not known to be targeted by miRNAs or RNA-binding proteins, we describe a simple approach for discovering new miRNAs in the worm and fly genomes, and show that the candidate novel miRNAs have all the features of known miRNAs.
Results
Scoring Exhaustive Motif Lists for Network-Level Conservation
We modified FastCompare [15], for processing mRNA sequences (i.e., we performed single-strand analyses), to calculate a conservation score for all 7-, 8-, and 9-mers from the 3′UTRs of worm and fly genes. Briefly, a k-mer is given a high conservation score if there is a significant overlap between the sets of orthologous genes having at least one copy of the k-mer anywhere in their 3′UTRs. The hypergeometric distribution is used to evaluate the significance of the overlap. Conservation scores are defined as the negative logarithm of the cumulative hypergeometric p-values (see Figure 1A and Materials and Methods). However, the hypergeometric p-values are only treated as relative measures of conservation, and are not used in the traditional null hypothesis rejection scheme. Further details can be found in Materials and Methods and in [15].
Figure 1 Schematic Representation of the Approach
(A) In the first stage of our approach, we scored exhaustive lists of k-mers for network level conservation. Schematic examples for a nonconserved k-mer (AAAAAAA) and a highly conserved one (UGUGAUA) are given in the left and right graphics, respectively.
(B) In the miRNA discovery stage, seed k-mers are used to search the genome for conserved and stable stem-loops.
As a control, we applied FastCompare to sets of randomized 3′UTRs with the same length and same level of divergence as the original sequences [15]. Figure 2 shows the distribution of conservation scores for all 7-mers in worms, obtained for real data and a single randomized control; it clearly shows that the extremely high conservation scores obtained on real data are very unlikely to be obtained by chance. The same pattern was observed for flies (Figure S1). We retained the top 500 7-mers (3.5% of all 16,384 7-mers) for further analysis (see below for a justification of this cutoff, in terms of number of captured miRNAs). We determined that the conservation score threshold induced by retaining the top 500 7-mers is, on average, greater than 99.9% of the scores obtained from randomized data. Using the procedure described in Materials and Methods, we mapped the 500 7-mers into 442 k-mers for worms and 497 for flies (with k = 7, 8, or 9). We observed that, in both cases, the list of highest-scoring k-mers often contains several overlapping, slightly distinct variants of the same sites, as shown for some of the highest-scoring worm k-mers in Table S1.
Figure 2 Distribution of Conservation Scores for the C. elegans/C. briggsae Analysis on 3′UTR Sequences
Distributions of actual (red) and randomized (black) sequences are shown. Scores corresponding to some of the known miRNA target sites and RNA-binding protein sites in worms are indicated by arrows. The top portion of both distributions are not shown, for the purpose of presentation.
We provide two lines of evidence that the high-scoring k-mers obtained in this study are not DNA regulatory elements (transcription factor binding sites [TFBSs]). First we determined that the overlap between the highest-scoring k-mers obtained in the present study and the ~400 highest-scoring k-mers found in the analysis of 2-kb upstream regions in the same worm and fly genomes [15] is very small (15 for worms, seven for flies). Since TFBSs are most abundant in upstream regions (see [16] for a review), this provides supportive evidence for the limited presence of TFBSs among our highest-scoring k-mers; although we cannot rule out the presence of unique transcription factor binding sites that function exclusively within 3′ downstream regions. Second, we reasoned that, if our highest-scoring k-mers were TFBSs, they would generally not have any strand bias, i.e., a k-mer and its reverse complement should be roughly equally conserved. However, we found that 97% (worm) and 90% (fly) of our highest-scoring k-mers have a higher conservation score than their reverse complement (which are themselves generally not among our highest-scoring k-mers). As an example of highly significant strand bias, the highest-scoring worm k-mer, CUGUGAU, is conserved in 187 genes, while its reverse complement, AUGACAG, is conserved in only 16 genes.
High-Scoring k-Mers Are Complementary to the 5′ Ends of Many miRNAs
We observed that the highest-scoring fly k-mer, UGUGAUA, corresponds to the K box, a short sequence that has been found in the 3′UTRs of many genes of the E(spl) complex in D. melanogaster [17]; this sequence has been shown to reduce mRNA transcript levels in vivo, and to a lesser extent, to also reduce protein levels [17]. Another short sequence with a verified posttranscriptional role [18,19], the Bearded (Brd) box (AGCUUUA, rank 23) was also identified as a highly conserved motif. These two motifs were shown to be complementary to the 5′ extremity of several fly miRNAs [20]. We then systematically matched our highest-scoring worm and fly k-mers to the 117 Caenorhabditis elegans and 79 D. melanogaster known (and experimentally verified) miRNAs, from the miRNA registry [7].
We found that 87 and 73 of our 442 and 497 highest-scoring worm and fly k-mers (respectively) had perfect complementarity to at least one known miRNA, and that, conversely, 77 and 57 different miRNAs were complementary to at least one high-scoring k-mer. The expected numbers of miRNAs matched by chance are significantly lower (approximately 38 and 24 for worm and fly, respectively; see Figures 3A for worms and S2A for flies; see Materials and Methods for explanations). However, we found that the vast majority of k-mers matched miRNAs within their 5′ extremity (see Figures 3A and 4 for worms, and Figures S2A and S3 for flies). Note that here, and in the rest of this study, we define complementarity to the 5′ extremity of a miRNA as complementarity starting within 1 nt or less of the actual miRNA 5′ extremity (e.g., positions 1 or 2 of the miRNA). Of the k-mer/miRNA pairings, 76% and 67% occur within the miRNA 5′ extremity, and the number of distinct miRNAs that are complementary to at least one k-mer within their 5′ extremities is 73 for worms and 49 for flies; this represents 62.4% and 62.0% of all known and experimentally verified miRNAs in C. elegans and D. melanogaster, respectively. For both worms and flies, the expected number of miRNAs whose 5′ extremity is complementary to the same number of k-mers selected at random is small: 5 for worms and 3.5 for flies (see Figures 3A and S2A), signifying that only a small proportion (6.8% and 7.1%) of the captured miRNAs is expected to be due to chance. Figure 3A shows that significantly increasing the initial number of retained worm 7-mers (we currently retain the 3.5% highest-scoring 7-mers) would yield very few additional complementary miRNAs; however, it would significantly increase the number of complementary miRNAs expected by chance. The same holds for flies (see Figure S2A).
Figure 3 High-Scoring k-Mers Are Complementary to the 5′ Ends of Many miRNAs
(A) Number of complementary worm miRNAs as a function of initial number of retained 7-mers. Solid lines correspond to complementarity anywhere within the miRNAs. Dashed lines correspond to complementarity to the 5′ extremity of miRNAs only. Complementarity to the 5′ extremity of a miRNA is defined as starting within 1 nt of the actual miRNA 5′ extremity.
(B) Proportion of 7-mers complementary to the 5′ extremity of at least one miRNA, as a function of the conservation rank (using a sliding window [w] of size 50).
Figure 4 Distribution of Distances from the First Nucleotide of the k-Mer to the 5′ Extremity of the miRNA
Distances are given for all pairs of high-scoring k-mers/complementary miRNAs. The distribution clearly shows that complementarity between high-scoring worm k-mers and miRNAs occurs primarily at the 5′ extremity of the miRNAs.
Interestingly, almost all the k-mers that are complementary to miRNAs are 7-mers (i.e., they were not extended into 8-mers). We observed that, for most worm miRNAs in this study (58/73), 7-mers that are complementary to positions 2–8 of miRNAs are more conserved than 7-mers complementary to positions 1–7. Intriguingly, the situation was almost opposite in flies, with only 18/49 miRNAs having a more conserved complementary 7-mer in positions 2–8.
We also investigated whether highly conserved k-mers that are not exactly complementary to the 5′ extremity of any miRNA can still pair with certain miRNAs, if we tolerate a single non-Watson-Crick GU pairing. We found that 41 highly conserved k-mers, complementary to 53 distinct miRNAs, fit that scenario in worms. This number of k-mers is much larger than the average (approximately 11) obtained when we start from the same number of randomly selected k-mers (repeated 100 times). Interestingly, out of these 53 complementary miRNAs, 45 (85%) are also exactly complementary (in their 5′ extremity) to one of our high-scoring k-mers. This suggests that at least certain miRNAs in worms can bind their targets either through exact complementarity or through inexact complementarity involving a small number of GU pairs. The high network-level conservation of certain k-mers with imperfect complementary to miRNAs may indicate that miRNA targets involving imperfect pairing through GU pairing constitute a functionally distinct class of targets, similar to observations made for transcription factors in bacteria [21]. The same analysis in flies yielded 17 k-mers, complementary (through one GU pairing) to 19 miRNAs. This number of k-mers was closer to the expected number (approximately nine) obtained from randomly selected k-mers, than in the worm analysis. This suggests that targets involving GU pairing may be less common in fly than in worms.
Since the signal-to-noise ratio appears to be much higher when considering only exact complementarity to miRNAs (in worms, 57 k-mers are exactly complementary to the 5′ extremity of known miRNAs, with only 4.3 expected by chance), we restricted the rest of our analyses to such exact complementarity. Nonetheless, the list of k-mers/complementary miRNAs with one GU pairing is available from our Web site (http://tavazoielab.princeton.edu/mirnas/).
In Figure 3B we show the proportion of 7-mers (within a sliding window of 50 7-mers) that are (exactly) complementary to the 5′ extremity of at least one miRNA as a function of the conservation score rank in worms. As can be seen in the figure, complementarity to the 5′ extremity of a miRNA correlates very strongly with conservation at the network level. A similar correlation was observed for flies (see Figure S2B).
The known C. elegans miRNAs with 5′ complementarity to at least one k-mer are shown in Table 1. The same information for flies is shown in Table S2. It is interesting to note that many of the highest-scoring worm k-mers that are complementary to known miRNAs are also highly conserved in flies, and vice versa. Moreover, worm k-mers that are highly conserved in flies are almost always complementary to the 5′ extremity of at least one fly miRNA (see Table 1). Table S3 shows the few miRNAs with complementarity to a highly conserved k-mer not occurring at the 5′ extremity. These cases may be due to chance; alternatively, they may be due to slightly erroneous annotation of the mature miRNA boundaries, or they may signify that some miRNAs are not restricted to recognizing their targets through their 5′ extremity.
Table 1 Worm k-Mers Complementary to 5′ Extremity of Known Worm miRNAs for 73 Distinct miRNAs
Table 1 Continued
Prediction and Analysis of miRNA Targets
The observations above suggest that the presence of a conserved k-mer within the 3′UTR of a given gene indicates targeting and regulation by the miRNA whose 5′ extremity is complementary to the k-mer. Our approach thus conveniently defines sets of putative targets for each miRNA: A gene is predicted to be a target of a given miRNA if its 3′UTR and that of its ortholog contain a high-scoring k-mer that is complementary to the 5′ extremity of the miRNA. Note that a similar approach for defining miRNA targets has been described [12,13,22,23], in which targets were defined as short sequences conserved within multiple alignments of several 3′UTR sequences from closely related species (vertebrates and flies). For flies, we observed that many of the largest 3′UTRs correspond to genes involved in development (for example, the 200 genes with largest 3′UTRs are strongly associated with the organ development Gene Ontology [GO] category, p < 10−19). To avoid systematically biasing our predicted targets toward these genes, we used real-length 3′UTRs when the length is less than 500 nt, but truncate larger 3′UTRs to 500 nt. Although many real targets are likely to be located beyond the 500-nt cutoff, we expect most of them to be retained (80% of annotated fly 3′UTRs are less than 500 nt). For worms, we used the real-length 3′UTRs. Although few miRNA targets have been experimentally verified, our predicted targets include some for which experimental evidence is available. For example, the predicted targets for worm let-7 includes hbl-1, a gene that is likely regulated by let-7 [24]. As another example, recent in vitro and in vivo experiments suggest that members of the fly miR-2 family (miR-2a/2b/2c) regulate the proapoptotic genes reaper, grim, and sickle in D. melanogaster [9]. Indeed, our predicted target set for miR-2a/2b/2c (259 genes having a conserved CUGUGAU or UGUGAUA in their 3′UTRs) contains the reaper and sickle genes (but not grim), as well as several other genes known to be involved in apoptosis: CG10345, CG11593, tartan, croquemort, and Ice. These genes are not yet known to be miRNA targets, and therefore constitute strong candidates for experimental verification of regulation by members of the miR-2 family.
We found that many miRNAs are also associated with significant functional enrichment(s) (see Table 2 for worms and flies). For example, we found that the predicted target set of C. elegans miR-1 (197 genes having a conserved ACAUUCC or CAUUCCA in their 3′UTRs) contains many genes involved in proton transport (p < 10−10) and ATPase activity (p < 10−9). In fact, most of the mRNAs encoding cytosolic sector subunits (A, B, D, E, F, and H) of a C. elegans vacuolar H+-ATPase contain a conserved target site for miR-1, suggesting a miRNA-mediated regulation of this proton-pumping complex.
Table 2 Functional Enrichments for Some of the Known C. elegans and D. melanogaster miRNA Target Sets
In flies, the predicted target set for miR-2a/b/c is enriched with genes annotated in GO as involved in the Notch signaling pathway (p < 10−6). In previous studies, the K box was found in the 3′UTR of many members of the E(spl) and Brd gene complexes, which are targets of the Notch signaling pathway [17,20,25]. Indeed, the target set for miR-2a/b/c contains the Brd genes m2, m4, and mα and the E(spl) genes m3, m5, mδ, and E(spl). It also contains the fringe and serrano genes, which are other known components of the Notch pathway (note that these two genes were also predicted as targets in [9,10]). The target sets for the miRNAs targeting the Brd box, miR-4 and miR-79, are also enriched with genes involved in the Notch signaling pathway (p < 10−5 in both cases).
The predicted target set for worm miR-277 (the union of conserved sets for GCAUUUA, UGCAUUU) is highly enriched with fatty acid metabolism (p < 10−15), carboxylic acid metabolism (p < 10−10), and branched chain family amino acid metabolism (p < 10−8). In a recent computational study, several enzymes of the branched chain amino acid degradation pathway were proposed to be targets for miR-277 [9]. The functional enrichment of its target set suggests a much broader role for miR-277, perhaps acting as a general metabolic switch, slowing down metabolic activity by repressing translation of these genes.
We used conserved sets obtained from randomized sequences to show that the number of targets we predict is much larger than the number expected by chance (see Figure 5A for an example with bantam, a D. melanogaster miRNA). The expected number of targets provided us with an estimate of the number of false positives in our sets of targets. Once sets of targets have been corrected for false positives, our approach could then provide insights into the topology of miRNA regulatory networks in metazoan genomes. For example, Figure 5B shows that some worm miRNAs potentially regulate hundreds of genes (e.g., miR-2/miR-43), while others may regulate fewer than ten genes (e.g., miR-273). Most miRNAs appear to regulate between 50 and 100 genes both in worms and flies (see Figures 5B and S4), a number that agrees with other recent estimates [8].
Figure 5 Number of miRNA Targets
(A) Example showing that the number of predicted targets for D. melanogaster bantam is much larger than expected by chance. The number of predicted targets is the number of genes whose 3′UTR contains at least one conserved k-mer complementary to the 5′ extremity of the corresponding miRNA. The distribution of numbers of targets expected by chance was obtained by running the same analysis using 100 pairs of randomized genomes with the same level of divergence as the original ones (see Materials and Methods for details).
(B) Estimated numbers of targets for C. elegans miRNAs (only for miRNAs that are complementary to at least one of our high-scoring k-mers). Each number corresponds to the number of predicted targets (as defined above) minus the average number of targets expected by chance over the 100 randomizations. The error bars correspond to two standard deviations.
To further validate our sets of predicted targets, we investigated whether coexpressed genes are regulated by the same miRNAs. When using the C. elegans early embryonic microarray time-course [26], we found that 3,131 pairs of highly coexpressed genes (Pearson correlation ≥ 0.8) contain at least one predicted target for the same miRNA. Using randomizations, we calculated that this number is significantly higher than expected by chance (p < 0.044), thus providing statistical evidence that the same miRNAs tend to regulate mRNAs that are coexpressed (at least during C. elegans early embryogenesis).
Finally, we generated the subsets of target genes for which high-scoring k-mers are also conserved within global alignments of the 3′UTRs (we used CLUSTALW with default parameters to generate the alignments). The list of these target genes is available on our Web site (http://tavazoielab.princeton.edu/mirnas/). On average, 49% and 75% of our initial target predictions correspond to k-mers at the same position in these alignments, in worms and flies, respectively. These subsets thus contain predicted target sites that are further constrained. Although nonaligned predicted targets may contain many false positives, we suspect that small scale DNA rearrangements, fast evolution of noncoding sequences, or imprecise definition of 3′UTR boundaries may, in many cases, make alignment-based methods unreliable. Our global list of predicted targets may therefore contain many functional targets that will not be found using traditional alignment-based approaches.
Biological Significance of Other High-Scoring k-Mers
Many of our high-scoring k-mers are not complementary to any known miRNA. For example, we found several AU-rich motifs that are highly conserved both in worms and flies, e.g., UAAUUUAU (ranks 4 and 11 in worms and flies, respectively) and UAUUUAUU (ranks 6 and 2). These motifs are similar to the AU-rich element (ARE), defined as UUAUUUAUU [27]; the ARE was found in the 3′UTRs of cytokines and proto-oncogenes in human [28]. It was shown to destabilize these mRNAs at least in part by triggering rapid deadenylation [27]. AREs have not been shown to be functional in worms or flies; however, a chimeric mRNA consisting of the rabbit β-globin gene fused to the 3′UTR of the human TNF-α (which contains several AREs) was rapidly degraded in Drosophila S2 cells [29]. Moreover, this degradation involves homologs of human genes known to be involved in ARE-mediated mRNA decay [29]. Interestingly, it was shown in the same study that the human miR-16 miRNA is required for ARE-mediated mRNA turnover. In worms, the genes whose 3′UTR contains conserved UAAUUUAU appear to be enriched for genes whose products localize to the endoplasmic reticulum (p < 10−11) and to the proteasome complex (p < 10−10); for example, eight (out of 14) genes encoding products that localize to the core proteasome complex contain a conserved UAAUUUAU in their 3′UTRs.
We also found that UGUAAAUA, a sequence bound by some members of the PUF family, was highly conserved both in worms and flies (ranks 9 and 4, respectively). Interestingly, as we will see below, the PUF binding site appears to be better represented by a gapped motif in worms. D. melanogaster possesses a single PUF protein named Pumilio. Early in embryogenesis, Pumilio controls anterior/posterior body patterning by binding to the 3′UTR of hunchback mRNA and repressing its translation (via interaction with Nanos) [30]. There is strong evidence that Pumilio also inhibits pole-cell division in early embryogenesis by repressing the translation of cyclin B [31]. Finally, Pumilio has also been shown to be involved in neuronal excitability [32], long-term memory [33], and dendrite neurogenesis [34]. However, no additional targets have been identified yet. Altogether, these studies suggest that Pumilio targets many mRNAs, with potentially a small fraction of them having been identified. The fly conserved set for UGUAAAUA contains 314 genes, thus providing a large number of potential targets awaiting experimental verification; it would be particularly interesting to focus experiments on genes that are expressed early in embryogenesis (maternal and zygotic) and genes expressed in the brain.
Finally, we found many highly conserved k-mers that are not yet known to be bound by any RNA-binding proteins (and are not complementary to any known miRNAs), but which are associated with strong functional enrichment. In worms, the sequences UUGUUGA, UGUUGUU, and UUGUUAU appear to be highly conserved in the 3′UTRs of many genes involved in cell growth (p < 10−17, p < 10−23, and p < 10−24, respectively). Indeed, out of the 497 genes containing a 3′UTR with a conserved UUGUUGA, UGUUGUU, or UUGUUAU, 192 are annotated as being involved in growth (p < 10−49, 64 expected). Moreover, the protein products of 60 of these 497 genes are known to localize to the ribosome (p < 10−49, six expected by chance). The same elements are also conserved downstream of many genes involved in larval development (p < 10−38) and gametogenesis (p < 10−15). This motif may be involved in slowing down cell growth by repressing translation or degrading large numbers of mRNAs at a certain developmental stage, or under stressful environmental conditions.
In flies, CAG-repeats (CAGCAGC, rank 71; and GCAGCAG, rank 104) are strongly associated with genes involved in transcriptional regulation (p < 10−14 for both k-mers). For example, among the 14 genes that have a conserved motif consisting of four tandem copies of CAG, six are known transcription factors (fork head, ventral veins lacking, SoxNeuro, cropped, spineless, and ypsilon schachtel), and two are transcriptional co-activators or co-repressors (big brother, smrter). CA repeats are also highly enriched with genes involved in organ development (p < 10−11 for CACACAC, rank 32). There is growing evidence that CA repeats are involved in mRNA transcript stability, at least in human. For example, in one recent study it was shown that CA repeats in the 3′UTR of the bcl-2 mRNA are responsible for destabilization of the transcript [35]. In another study, CA repeats within intron 13 of the human endothelial nitric oxide synthase gene were found to mediate cleavage of the pre-mRNA, unless bound by heterogeneous nuclear ribonucleoprotein L [36]. Although experiments are needed to validate our observations, these results provide strong support for the functionality of certain classes of repeats in posttranscriptional regulation.
Motifs Represented by Gapped k-Mers Often Show Stronger Conservation than Ungapped k-Mers
The binding sites for the PUF RNA-binding proteins in yeast are UGUA..UA, UGUA...UA, and UGUA....UA for Puf3p, Puf4p, and Puf5p, respectively [37]. This means that UGUA and UA are required for binding to the PUF proteins, but also that the nucleotides between the two half-sites are less important (although nucleotide preferences still exist within the gaps [37]. In yeast, the length of the gap between UGUA and UA determines which PUF protein binds the site with highest affinity. In what follows, we refer to such elements as “gapped” k-mers. To search for these regulatory elements within the 3′UTRs of worm and fly mRNAs, we calculated a conservation score for all sequences of the form s1-gap-s2, where the lengths of s1, gap, and s2 vary between 2 and 4 nt. We found that 157 (worm) and 215 (fly) gapped k-mers were more conserved (in terms of network-level conservation score) than any of the ungapped k-mers they matched. Although some of these gapped k-mers are complementary to some miRNAs, the position at which complementarity begins is less biased than for ungapped k-mers: Out of 52 gapped k-mer/miRNA pairs in worms, only 17 (33%) occur within 1 nt or less from the 5′ extremity of the miRNAs. The same holds for fly, in which out of 26 gapped k-mer/miRNA pairs, only ten (38%) occur within 1 nt or less from the 5′ extremity of the miRNAs. Some of the highest-scoring worm gapped k-mers are shown in Table 3.
Table 3 Selection of Highest-Scoring Gapped k-Mers in Worms
The highest-scoring gapped k-mer in worms, UGUA..UA, matches the experimentally defined binding sites for Puf3p in yeast [37]) and mouse PUM-2 (consensus UGUA.AUA [38]). Although in yeast, Puf3p targets the mRNA of many genes encoding proteins that are localized to the mitochondrion, in worms we do not find any particular functional enrichment using the GO annotations. The C. elegans genome contains eight distinct PUF proteins, although some of them duplicated recently and might be redundant [39]. FBF-1 and FBF-2 (93% identical) regulate the germ line switch from spermatogenesis to oogenesis by posttranscriptionally repressing fem-3 [40]. Although the nature of this repression is unknown, the required physical interaction between FBF and NANOS-3, a homolog of Drosophila Nanos, suggests that FBF represses the translation of fem-3 [41], as Pumilio does for hunchback.
Table 3 contains several motifs that also resemble the PUF binding site (e.g., UUGU..AUA and UGUA..AUA); these motifs could be bound by other members of the PUF family in C. elegans.
Several of the highest-scoring gapped k-mers in worms are variants of the growth-related motifs described above (e.g., UU..UGUUG, UU..UGUUA; see Table 3) and have similar functional enrichments. Several other highly conserved gapped k-mers appear to be involved in embryonic development, e.g., UUU..CCC and CCC..UUU (p < 10−9 and p < 10−10, respectively).
Coregulation by Multiple Target Sites
Simultaneous conservation of two distinct high-scoring k-mers (termed conserved co-occurrence) provides a simple way to discover putative coregulation by pairs of regulatory elements (or, more precisely, between the molecules that bind them). Pair members of high-scoring k-mers that differ in at least 3 nt were scored and sorted according to network-level conservation, as described above for single elements. Only pairs of k-mers conserved in at least ten genes were retained for further analysis. For the most conserved co-occurrences, we also calculated the statistical significance of the overlap between the conserved sets corresponding to each element taken separately (unlike for network-level conservation, we assumed that these conserved sets were approximately independent, provided that the k-mers were different enough in sequence). In both worms (Table 4) and flies (Table S4), we found that many distinct miRNA target sites are strongly co-conserved. For example, in worms, the target sites for miR-75/79 and miR-86 are simultaneously conserved in the 3′UTRs of 16 genes (p < 10−9). Similarly, the target sites for miR-2/43 and miR-1 are coconserved in 12 genes (p < 10−6). Consistent with previous observations [17], target sites for miR-2a/2b/2c/6/13a/13b (UGUGAUA, K box) and miR-277 (AGCUUUA, Brd box) are significantly coconserved within fly 3′UTRs (p < 10−17). As shown in Tables 4 and S4 for both worms and flies, miRNA target sites are very often coconserved with AU-rich elements, potentially linking miRNA-based regulation and regulation through AU-rich elements. Interestingly, a recent study provided evidence for involvement of miRNAs in AU-rich element-mediated mRNA decay [29]).
Table 4 Top 20 Most Conserved k-Mer Co-occurrences in Worms
We also looked at whether coconserved sites are significantly clustered within their 3′UTRs. To do so, we calculated the median distance between the conserved occurrences within the 3′UTRs of the reference species (C. elegans for worms and D. melanogaster for flies). To calculate a statistical significance, we randomly selected the same number of pairs of positions within the same 3′UTRs, 10,000 times, and from this we calculated the null distribution of median distances. We found that most interactions were not associated with any statistically significant clustering. This may be due to the fact that pairs of k-mers are coconserved in relatively few genes, preventing strongly significant statistical assertions. We found that, in worms, the target sites for miR-86 (UUCACUU) and miR-87/miR-233/miR-356 (UUGCUCA) are significantly clustered (p < 10−4), with a median distance of 35 nt between coconserved occurrences in C. elegans. In another case, also in worms, the target sites for two coconserved and distinct miRNAs (or miRNA sets) miR-2/miR-43 (CUGUGAU) and miR-80/miR-81/miR-82 (UGAUCUC), were found to overlap more often than expected by chance (p < 10−5) (note that we limited the extent of the overlap to 4 nt in our coconservation analysis). This may be related to the as-yet unexplained observation that, in many cases, several distinct miRNAs appear to target the same site (see Tables 1 and S2).
Discovery of Novel miRNAs
Starting from our list of highly conserved known and putative mRNA regulatory elements (k-mers), we systematically searched the C. elegans and D. melanogaster genomes for novel miRNAs that might target them. Our goal was to find candidate novel miRNAs that have not been found by previous approaches. We relaxed some assumptions about the structure of the miRNA precursor stem-loops and their pattern of conservation (e.g., see [42]), while introducing a new one, i.e., one having a 5′ extremity with perfect complementarity to at least one of our high-scoring k-mers. Briefly (see Materials and Methods for more details), we searched only for conserved stem-loop–forming sequences with a folding strength above a selected threshold, in which both orthologous stem-loops are required to yield a mature miRNA with a 5′ extremity complementary to the same high-scoring k-mer. We use this stringent condition when we refer to conservation of candidate miRNAs.
In worms, we obtained 80 candidate miRNAs that meet the (stringent) requirements described in Materials and Methods, with 30 of them being known C. elegans miRNAs. Note that the number of known miRNAs that are conserved (using only the conservation of the 80-nt stem-loop precursor sequence at the same threshold as described in Materials and Methods) between C. elegans and C. briggsae is 49; therefore, our miRNA discovery procedure for discovering highly conserved miRNAs has a 61% sensitivity in worms.
The top 30 worm miRNAs, ranked by decreasing folding strength (increasing ΔG), are listed in Table 5. They include 17 previously known miRNAs, the rest being candidate novel miRNAs. Table 5 shows that the candidate novel miRNAs are very similar to the known ones, in terms of ΔG, conservation in C. briggsae, and location within the genome (intergenic region or introns). Note that, although they were derived from k-mers lying in 3′UTRs of genes, none of the candidate novel miRNAs lie across from 3′UTRs. The 30 highest-scoring fly miRNAs, of which ten are known, are listed in Table S5. While all our candidate miRNAs should be systematically verified by experiments, our study suggests that the total number of miRNAs in metazoans may be much higher than previously thought, a prediction that agrees with a recent one made in mammalian genomes [13]. A conservative estimate based on the top 30 miRNAs discovered by our procedure would predict that the total number of miRNAs in worms is almost twice the current number in the miRNA registry. A less conservative estimate based on our 80 predicted miRNAs would indicate between 300 and 400 miRNAs in worms.
Table 5 Top 30 Predicted C. elegans miRNAs, Sorted by ΔG
Comparison of 3′UTR Regulation Between Worm and Fly
The intersection between the 442 and 497 ungapped highest-scoring worm and fly k-mers consists of 79 k-mers (19 of them complementary to the 5′ extremity of miRNAs in both organisms). This overlap is much higher than expected by chance (p < 10−37), indicating significant conservation of posttranscriptional regulatory sequences between these two phylogenetic groups. However, we found no significant overlap between the gene sets bearing the same k-mers in C. elegans and D. melanogaster. These results strongly resemble the ones we obtained for transcription factor binding sites [15], indicating that the regulators (miRNAs and RNA-binding proteins) are highly conserved, hence the sequences they bind are conserved; however, the sets of genes regulated by those regulators appear to differ significantly, indicating large-scale rewiring of the posttranscriptional regulatory network across large phylogenetic distances. This scenario is not surprising, since modifying the RNA-binding affinities of the regulators would cause drastic changes to the regulatory network topology, while appearance/deletion of single regulatory elements would presumably have much less drastic consequences for the cell. These results also indicate that motif discovery using the network-level conservation principle (at least as presented here) would fail if the compared species were very distantly related.
Discussion
We have described an integrated approach for discovering conserved elements involved in posttranscriptional regulation and for predicting the miRNA regulators that may target these elements. Our approach is based on comparative genomics, but does not require the orthologous 3′UTRs to be aligned, and it requires only two genomes. Many of the regulatory elements we discovered were complementary to the 5′ extremity of known miRNAs, both in worms (in which we captured 62.4% of the known miRNAs) and in flies (62.0%). There may be several reasons why we failed to detect complementary k-mers for the remaining known miRNAs. It is possible that the sets of genes regulated by these miRNAs are small, decreasing the statistical power for detecting them. Alternatively, the corresponding posttranscriptional networks may have undergone extensive rewiring, rendering them undetectable by network-level conservation.
We have shown that the high-scoring k-mers are unlikely to be TFBSs (although we expect some level of contamination, due to transcription factors that bind downstream regions). Note, however, that some miRNAs may target mRNA sequences that match known transcription factor binding sites. For example, the target site of C. elegans miR-248 matches the E-box, a site known to be bound by several transcription factors of the basic helix-loop-helix family; similarly, the target site of D. melanogaster miR-184* matches the binding site for GATA factors.
We have also shown that our approach conveniently defines sets of targets for miRNAs; a gene is predicted to be a target of a miRNA if the 3′UTR of the gene and its ortholog contain a globally conserved k-mer that is complementary to the 5′ extremity of the miRNA. While this simple approach appears to group together coherent sets of genes (as defined by functional enrichments and coexpression, for example), it has several limitations. For example, it does not predict lin-41 as being a target of let-7 in C. elegans, because this interaction involves a non-Watson-Crick pairing [43] and relatively extensive complementarity across the entire length of the miRNA/mRNA duplex. As recently shown in [8], inexact complementarity between the 5′ extremity of the miRNA and its target sequence can be rescued by a more extensive pairing across the length of the miRNA. Computational approaches for recovering these targets have been described elsewhere [10,44]. Also, our approach predicts only conserved targets; it is unclear what proportion of functional targets is not conserved between the species under consideration.
Since many of the most highly conserved k-mers do not match known miRNAs, we searched the C. elegans and D. melanogaster genomes for candidate novel miRNAs. We found many such putative miRNAs and showed that these predictions have all the features of known miRNAs. While experiments are now required to validate our predictions, a conservative estimate based on the highest-scoring C. elegans miRNAs indicates that there may be more than twice as many functional miRNAs in worms as is currently thought. A total of 101 (worms) and 110 (flies) of our high-scoring k-mers are complementary to the 5′ extremity of at least one known or novel miRNA. As illustrated in Table S1, we found that many of the remaining high-scoring k-mers extensively overlap with these sites (e.g., 120 and 147 of these remaining k-mers have at least 6-nt overlap with at least one k-mer that is complementary to the 5′ extremity of a known or novel miRNA). A small fraction (17 in worms, five in flies) of the remaining k-mers were complementary to miRNA 5′ extremities, if a single GU-pairing is allowed. Therefore, 204 k-mers in worms and 235 in flies are unaccounted for. As we have shown above, several of these k-mers are known to be protein-binding sites (e.g., by members of PUF family of RNA-binding proteins), and it is likely that many of the remaining k-mers are bound by RNA-binding proteins that have not been characterized yet. Finally, it is also possible that many of the remaining sites are targeted by miRNAs that are less conserved and/or those that interact weakly with their targets.
We have also shown that posttranscriptional regulatory networks have undergone extensive rewiring between worms and flies. The binding sites for miRNAs and known RNA-binding proteins are present in both phyla, suggesting that the regulators are still largely the same. However, these elements seem to be regulating entirely different sets of genes.
We envision several directions for further research. For example, our current approach does not make any assumptions about how miRNA or RNA-binding protein target sites evolve. We believe that, once the evolution of RNA regulatory elements is better understood, our approach may be refined to take RNA-specific modes of evolution into account, similarly to what was done with transcription factor binding sites [45]. Allowing more degeneracy within our RNA regulatory element representation also represents an interesting direction for further research, especially for RNA-binding protein target sites.
Materials and Methods
The approach is outlined in Figure 1. It consisted of two main stages, the motif discovery stage (Figure 1A), in which FastCompare [15] was used to score exhaustive k-mer lists for network-level conservation, and the miRNA discovery stage (Figure 1B), in which novel miRNA candidates with 5′ complementarity to our high-scoring k-mers were predicted based on a combination of stem-loop structure and conservation.
Data.
Entire genome sequences were downloaded from ENSEMBL [46]. The D. pseudoobscura genome sequence was obtained from [47]. We used real-length 3′UTRs, calculated according to ENSEMBL gene boundary coordinates for C. elegans and D. melanogaster. When several 3′UTRs were present for a single gene, only the longest one was retained. When no 3′UTR was present, we used 300 or 500 nt downstream of the stop codon, which correspond approximately to the 80th percentile of worm and fly 3′UTR lengths, respectively. For C. briggsae and D. pseudoobscura genes, we used 3′UTRs of the same length as those of their C. elegans and D. melanogaster orthologs, respectively. C. elegans/C. briggsae gene orthology relationships were obtained from Stein et al. [48]. D. melanogaster/D. pseudoobscura gene orthology relationships were mapped using reciprocal best BLAST hits. Functional annotations and GO classifications were downloaded from the GO web site (http://www.geneontology.org/). Functional enrichment p-values were calculated as described in [15,49]. Only functional enrichments with p-values lower than 10−5 are presented here.
Motif finding using network-level conservation.
We applied FastCompare [15] to motif discovery using network-level conservation. FastCompare was modified for processing mRNA sequences (single-strand analysis). Briefly, each possible k-mer was considered as a candidate regulatory element. For each k-mer, we found the set of open reading frames in the first species that had at least one exact occurrence of the k-mer in their 3′UTR. We then found the set of open reading frames in the second species that had at least one occurrence of the same k-mer in their 3′UTR. The matches could be anywhere in the 3′UTRs: They do not have to be at the same positions in two orthologous 3′UTRs (as with multiple alignment). Since both functional and nonfunctional elements are expected to be conserved between two closely related species, the two sets are expected to overlap. However, under the network-level conservation principle, the extent of the overlap will be much greater for k-mers that represent functional mRNA regulatory elements. The strength of the overlap was measured using the hypergeometric distribution, which defines the probability of drawing two sets of size s
1 and s
2, having i or more elements in common, from a set of N elements, and this probability is given by:
It is important to note that, due to basal conservation (that is, conservation arising from common ancestry), the hypergeometric p-values will generally be very small for most k-mers. Therefore, we use only these p-values as relative measures of network-level conservation and focus on k-mers with the greatest conservation. For simplicity, we define the “conservation score” as the negative logarithm (base e) of the hypergeometric p-value obtained for a given k-mer. Conservation scores were normalized for unequal lengths among 3′UTRs by weighing the contribution of each 3′UTR by 1/length, where length represents the length (in nt) of the 3′UTR. The variables s
1, s
2, and i were obtained by multiplying the corresponding weighted counts by 300 (for worms) and 500 (for flies), then rounding to the nearest integer. Further details about the motif discovery method are described in [15].
We calculated a conservation score for all possible k-mers, with k ranging from 7 to 9. We used randomized sequences to show that the high conservation scores obtained in this study were unlikely to be obtained by chance. To generate randomized pairs of orthologous 3′UTRs with the same level of divergence as in the actual data, we aligned each pair of orthologous 3′UTRs using ClustalW [50], and used the alignments to obtain an estimate of the substitution rates between the orthologous sequences. Starting from one of the orthologs, we created a randomized ortholog by mutating the initial sequence according to the estimated substitution frequencies. We then repeated FastCompare on the randomized sequences.
In the present study, we retained the 500 most conserved 7-mers (obtained from the actual sequences) and processed them as described in [15]. Briefly, we first extended 7-mers into 8-mers if, for a given 7-mer, there existed an 8-mer with a higher conservation score such that the 8-mer contained the 7-mer. We extended 8-mers into 9-mers in the same way. We also retained high-scoring 8-mers and 9-mers that did not have any substrings among 7-mers. Then we systematically removed k-mers that had higher scoring substrings. We define the conserved set of a given k-mer as the set of genes whose 3′UTRs contained at least one conserved occurrence of the k-mer.
To estimate the expected number of miRNAs complementary to a set of m high-scoring k-mers, we chose m k-mers at random, with the same numbers of 7-, 8-, and 9-mers as in the original set. We then evaluated the number of known miRNAs complementary to at least one of these randomly selected k-mers. We repeated the same procedure 100 times, and calculated the average number of complementary miRNAs.
Defining miRNA targets.
We defined the target set of a given miRNA as the union of all conserved sets corresponding to the highly conserved k-mers complementary to its 5′ extremity (complementarity had to begin within 1 nt of the 5′ extremity). To estimate the number of targets expected by chance for each miRNA, we generated pairs of randomized orthologous sequences retaining the same level of divergence as the original pairs of sequences, as described above. Conserved sets for all k-mers complementary to miRNAs were then determined for these randomly generated sequences and subsequently used to create pseudo-target sets for each miRNA. The randomization procedure was repeated 100 times. Then, for each miRNA, the average size and standard deviation of the 100 corresponding pseudo-target sets were calculated.
miRNA discovery.
For each conserved k-mer, we searched through the entire C. elegans or D. melanogaster genome for occurrences of the reverse-complementary k-mer on both strands of DNA. For each occurrence, we took two windows of length 80 nt (potentially corresponding to the two possible candidate miRNAs lying on the two arms of the 80-nt stem-loop sequence). We folded each window using RNAfold from the Vienna Package [51] to give the fold with minimal folding energy (MFE). If the fold formed a single stem-loop and the MFE was less than −30 kcal/mol at 25 °C, we retained the sequence as a potential miRNA precursor. We then tested for conservation of the potentially novel miRNA by searching for its homolog. We used BLAST (blastn) to search the second genome (C. briggsae or D. pseudoobscura) for regions homologous to the 80-nt stem-loop sequence, requiring the best matching sequence to have an E-value below a cutoff corresponding to conservation of 40% of the known miRNAs for that species (10−17 for worms and 10−29 for flies). We also required that the best matching sequence contains the exact conserved k-mer above, and folds into a single stem-loop with MFE less than −30 kcal/mol at 25 °C. We removed miRNA candidates located in exons or on the opposite strand of exons. Candidate mature miRNAs were defined as 23-nt sequences, such that the conserved reverse-complementary k-mer began at the second nucleotide from the 5′ extremity. A candidate mature miRNA matched a known mature miRNA if the positions of their 5′ ends were located within 2 nt on the genome (because miRNAs are generally 21–23 nt long). For the miRNA precursor stem-loops with candidate miRNAs from both arms of the stem-loop, we chose the one that matched the more conserved k-mer as the more likely candidate, except in the case where one matched a known miRNA. We ranked the candidate miRNAs by the MFE of their precursor stem-loop. To minimize false-positives, we present only candidate miRNAs with MFEs smaller than −34 kcal/mol at 25 °C as our final list of high-confidence predictions (80 for C. elegans and 92 for D. melanogaster).
We named our predicted miRNAs (pmi) by a number followed by a letter (e.g., cel-pmi-74a). The number corresponds to the rank of the conserved k-mer matched by our predicted miRNA. The letter corresponds to the ordinal value from all pmi matching that k-mer as ranked by MFE. When our pmi corresponds to a known miRNA, both names are shown.
Data and Web site.
The sequences, programs, and detailed results described in this paper are available at http://tavazoielab.princeton.edu/mirnas/.
Supporting Information
Figure S1 Distribution of Conservation Scores for the D. melanogaster/D. pseudoobscura Analysis, on Actual and Randomized 3′UTR Sequences
Actual sequences are depicted in red and randomized sequences in black. Scores corresponding to some of the known miRNA target sites and RNA-binding protein sites are indicated by arrows. The top portions of both distributions are not shown, for the purpose of presentation.
(72 KB PDF)
Click here for additional data file.
Figure S2 High-Scoring k-Mers Are Complementary to the 5′ Ends of Many miRNAs
Number of complementary fly miRNAs as a function of initial number of retained 7-mers (A), and proportion of fly 7-mers complementary to the 5′ extremity of at least one fly miRNA (B), as a function of the conservation rank (using a sliding window of size 50).
(66 KB PDF)
Click here for additional data file.
Figure S3 Distribution of Distances (in Nucleotides) from the First Nucleotide of the k-Mer to the 5′ Extremity of the miRNA, for All Pairs of High-Scoring k-Mers/Complementary miRNAs
The distribution clearly shows that complementarity between high-scoring fly k-mers and miRNAs occurs primarily at the 5′ extremity of the miRNAs.
(66 KB PDF)
Click here for additional data file.
Figure S4 Estimated Number of Targets for D. melanogaster miRNAs That Are Complementary to One or More of Our High-Scoring k-Mers
These numbers correspond to the number of genes with a 3′UTR containing at least one conserved k-mer complementary to the 5′ extremity of the corresponding miRNA (i.e., number of predicted targets), minus the expected number of targets by chance. Expected numbers were obtained by running the same analysis using 100 pairs of randomized fly genomes with the same level of divergence as the original ones, and averaging the obtained number of targets over the 100 runs. The error bars correspond to two standard deviations.
(80 KB PDF)
Click here for additional data file.
Table S1 Illustrating the Redundancy Among Some of the Highest-Scoring Worm k-Mers
The top box shows k-mers (and their ranks) that have a 1-nt difference with four of the highest-scoring k-mers. The bottom box shows k-mers (and their ranks) that have a 6-nt overlap with the same four k-mers.
(69 KB PDF)
Click here for additional data file.
Table S2 Fly k-Mers Complementary to 5′ Extremity of Known Fly miRNAs
k-Mers are grouped by sequence similarity and overlap. Each k-mer within a group is complementary to (i.e., matches) at least one miRNA, indicated by a number. If the k-mer is also found within the list of highest conserved worm k-mers, its rank is given, and * indicates that the k-mer is also complementary to the 5′ extremity of a worm miRNA
(79 KB PDF)
Click here for additional data file.
Table S3
k-Mers Complementary to Known miRNAs, But Not Within the 5′ Extremity in Worms and Flies
Complementary k-mers in worm (A) and fly (B) are listed.
(81 KB PDF)
Click here for additional data file.
Table S4 Top 20 Most Conserved k-Mer Co-occurrences in Flies
Pairs of k-mers were considered (scored) only if the pair members differed in at least 3 nt and if they are coconserved in at least ten genes. The number of genes for which the pairs of k-mers were conserved within the 3′UTRs is indicated in the table. The p-value represents the statistical significance of the intersection between the conserved sets of k-mer 1 and k-mer 2.
(79 KB PDF)
Click here for additional data file.
Table S5 Top 30 Predicted D. melanogaster miRNAs, Sorted by ΔG
Red miRNAs are known, black are novel. Pos is the position of the k-mer in the chromosome. ΔG is the MFE of the precursor stem-loop. E-value measures the conservation of the stem-loop sequence in D. pseudoobscura. L is the annotation of the location in the genome in which the predicted miRNA lies (IG, intergenic; IN, intron).
(82 KB PDF)
Click here for additional data file.
We are grateful to Jean-Baptiste Boulé, Morten Krogh, Virginie Orgogozo, and the two anonymous reviewers for critical reading of the manuscript. We are also grateful to members of the Tavazoie group for insightful discussions. ST is supported by grants from National Science Foundation, Defense Advanced Research Projects Agency, and National Institutes of Health. CSC is supported by a grant from the National Institutes of Health.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. CSC, OE, and ST conceived and designed the experiments. CSC and OE analyzed the data. CSC, OE, and ST wrote the paper.
A previous version of this article appeared as an Early Online Release on November 2, 2005 (DOI: 10.1371/journal.pcbi.0010069.eor).
Abbreviations
AREAU-rich element
GOGene Ontology
MFEminimal folding energy
miRNAmicroRNA
ntnucleotide
TFBStranscription factor binding site
UTRuntranslated region
==== Refs
References
Lee R Feinbaum R Ambros V 1993 The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14 Cell 75 843 854 8252621
Hutvagner G Zamore P 2002 A microRNA in a multiple-turnover RNAi enzyme complex Science 297 2056 2060 12154197
Vaucheret H Vazquez F Crete P Bartel D 2004 The action of ARGONAUTE1 in the miRNA pathway and its regulation by the miRNA pathway are crucial for plant development Genes Dev 18 1187 1197 15131082
Rhoades M Reinhart B Lim L Burge C Bartel B 2002 Prediction of plant microRNA targets Cell 110 513 520 12202040
Yekta S Shih IH Bartel DP 2004 MicroRNA-directed cleavage of HOXB8 mRNA Science 304 594 596 15105502
Lim L Lau N Garrett-Engele P Grimson A Schelter J 2005 Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs Nature 433 769 773 15685193
Griffiths-Jones S 2004 The microRNA Registry Nucleic Acids Res 32 (Database issue) D109–111
Brennecke J Stark A Russell R Cohen S 2005 Principles of microRNA-target recognition PLoS Biol 3 e85 15723116
Stark A Brennecke J Russell R Cohen S 2003 Identification of Drosophila microRNA targets PLoS Biol 1 e60 14691535
Enright A John B Gaul U Tuschl T Sander C 2003 MicroRNA targets in Drosophila
Genome Biol 5 R1 14709173
Rajewsky N Socci N 2004 Computational identification of microRNA targets Dev Biol 267 529 535 15013811
Lewis B Burge C Bartel D 2005 Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets Cell 120 15 20 15652477
Xie X Lu J Kulbokas E Golub T Mootha V 2005 Systematic discovery of regulatory motifs in human promoters and 3′ UTRs by comparison of several mammals Nature 434 338 345 15735639
Pritsker M Liu Y Beer M Tavazoie S 2004 Whole-genome discovery of transcription factor binding sites by network-level conservation Genome Res 14 99 108 14672978
Elemento O Tavazoie S 2005 Fast and systematic genome-wide discovery of conserved regulatory elements using a non-alignment based approach Genome Biol 6 R18 15693947
Wray GA Hahn MW Abouheif E Balhoff JP Pizer M 2003 The evolution of transcriptional regulation in eukaryotes Mol Biol Evol 20 1377 1419 12777501
Lai E Burks C Posakony J 1998 The K box, a conserved 3′ UTR sequence motif, negatively regulates accumulation of enhancer of split complex transcripts Development 125 4077 4088 9735368
Lai E Posakony J 1997 The Bearded box, a novel 3′ UTR sequence motif, mediates negative post-transcriptional regulation of Bearded and Enhancer of split complex gene expression Development 124 4847 4856 9428421
Leviten M Lai E Posakony J 1997 The Drosophila gene Bearded encodes a novel small protein and shares 3′ UTR sequence motifs with multiple Enhancer of split complex genes Development 124 4039 4051 9374401
Lai E 2002 MicroRNAs are complementary to 3′ UTR sequence motifs that mediate negative post-transcriptional regulation Nat Genet 30 363 364 11896390
Kalir S Alon U 2004 Using a quantitative blueprint to reprogram the dynamics of the flagella gene network Cell 117 713 720 15186773
Grun D Wang Y-L Langenberger D Gunsalus KC Rajewsky N 2005 MicroRNA target predictions across seven Drosophila species and comparison to mammalian targets PLoS Comput Biol 1 e13 16103902
Krek A Grun D Poy MN Wolf R Rosenberg L 2005 Combinatorial microRNA target predictions Nat Genet 37 495 500 15806104
Lin S Johnson S Abraham M Vella M Pasquinelli A 2003 The C. elegans hunchback homolog, hbl-1, controls temporal patterning and is a probable microRNA target Dev Cell 4 639 650 12737800
Lai E Bodner R Kavaler J Freschi G Posakony J 2000 Antagonism of notch signaling activity by members of a novel protein family encoded by the bearded and enhancer of split gene complexes Development 127 291 306 10603347
Baugh L Hill A Slonim D Brown E Hunter C 2003 Composition and dynamics of the Caenorhabditis elegans early embryonic transcriptome Development 130 889 900 12538516
Zubiaga A Belasco J Greenberg M 1995 The nonamer UUAUUUAUU is the key AU-rich sequence motif that mediates mRNA degradation Mol Cell Biol 15 2219 2230 7891716
Shaw G Kamen R 1986 A conserved AU sequence from the 3′ untranslated region of GM-CSF mRNA mediates selective mRNA degradation Cell 46 659 667 3488815
Jing Q Huang S Guth S Zarubin T Motoyama A 2005 Involvement of microRNA in AU-rich element-mediated mRNA instability Cell 120 623 634 15766526
Murata Y Wharton R 1995 Binding of pumilio to maternal hunchback mRNA is required for posterior patterning in Drosophila embryos Cell 80 747 756 7889568
Asaoka-Taguchi M Yamada M Nakamura A Hanyu K Kobayashi S 1999 Maternal Pumilio acts together with Nanos in germline development in Drosophila embryos Nat Cell Biol 1 431 437 10559987
Schweers B Walters K Stern M 2002 The Drosophila melanogaster translational repressor pumilio regulates neuronal excitability Genetics 161 1177 1185 12136020
Dubnau J Chiang A Grady L Barditch J Gossweiler S 2003 The staufen /pumilio pathway is involved in Drosophila long-term memory Curr Biol 13 286 296 12593794
Ye B Petritsch C Clark I Gavis E Jan L 2004
Nanos and Pumilio are essential for dendrite morphogenesis in Drosophila peripheral neurons Curr Biol 14 314 321 14972682
Lee J Jeon M Seo Y Lee Y Ko J 2004 CA repeats in the 3′-untranslated region of bcl-2 mRNA mediate constitutive decay of bcl-2 mRNA J Biol Chem 279 42758 42764 15294893
Hui J Reither G Bindereif A 2003 Novel functional role of CA repeats and hnRNP L in RNA stability RNA 9 931 936 12869704
Gerber A Herschlag D Brown P 2004 Extensive association of functionally and cytotopically related mRNAs with Puf family RNA-binding proteins in yeast PLoS Biol 2 e79 15024427
White E Moore-Jarrett T Ruley H 2001 PUM2, a novel murine puf protein, and its consensus RNA-binding site RNA 7 1855 1866 11780640
Wickens M Bernstein D Kimble J Parker R 2002 A PUF family portrait: 3′UTR regulation as a way of life Trends Genet 18 150 157 11858839
Zhang B Gallegos M Puoti A Durkin E Fields S 1997 A conserved RNA-binding protein that regulates sexual fates in the C. elegans hermaphrodite germ line Nature 390 477 484 9393998
Kraemer B Crittenden S Gallegos M Moulder G Barstead R 1999 NANOS-3 and FBF proteins physically interact to control the sperm-oocyte switch in Caenorhabditis elegans
Curr Biol 9 1009 1018 10508609
Lim L Lau N Weinstein E Abdelhakim A Yekta S 2003 The microRNAs of Caenorhabditis elegans
Genes Dev 17 991 1008 12672692
Vella M Choi E Lin S Reinert K Slack F 2004 The C. elegans microRNA let-7 binds to imperfect let-7 complementary sites from the lin-41 3′UTR Genes Dev 18 132 137 14729570
John B Enright AJ Aravin A Tuschl T Sander C 2004 Human microRNA targets PLoS Biol 2 e363 15502875
Moses AM Chiang DY Pollard DA Iyer VN Eisen MB 2004 MONKEY: Identifying conserved transcription-factor binding sites in multiple alignments using a binding site-specific evolutionary model Genome Biol 5 R98 15575972
Birney E Andrews TD Bevan P Caccamo M Chen Y 2004 An overview of Ensembl Genome Res 14 925 928 15078858
Richards S Liu Y Bettencourt BR Hradecky P Letovsky S 2005 Comparative genome sequencing of Drosophila pseudoobscura : Chromosomal, gene, and cis-element evolution Genome Res 15 1 18 15632085
Stein L Bao Z Blasiar D Blumenthal T Brent M 2003 The genome sequence of Caenorhabditis briggsae: a platform for comparative genomics PLoS Biol 1 e45 14624247
Tavazoie S Hughes J Campbell M Cho R Church G 1999 Systematic determination of genetic network architecture Nat Genet 22 281 285 10391217
Thompson JD Higgins DG Gibson TJ 1994 CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 22 4673 4680 7984417
Hofacker I 2003 Vienna RNA secondary structure server Nucleic Acids Res 31 3429 3431 12824340
|
16355253
|
PMC1309705
|
CC BY
|
2021-01-05 09:18:23
|
no
|
PLoS Comput Biol. 2005 Dec 9; 1(7):e69
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010069
|
oa_comm
|
==== Front
PLoS Comput BiolPLoS Comput. BiolpcbiplcbploscompPLoS Computational Biology1553-734X1553-7358Public Library of Science San Francisco, USA 1635525410.1371/journal.pcbi.001007205-PLCB-RA-0220R2plcb-01-07-01Research ArticleBioinformatics - Computational BiologyCell BiologyMolecular Biology - Structural BiologyStatisticsYeast and FungiSaccharomycesFolding Free Energies of 5′-UTRs Impact Post-Transcriptional Regulation on a Genomic Scale in Yeast Regulatory Effects of 5′-UTR FoldingRingnér Markus *Krogh Morten Complex Systems Division, Department of Theoretical Physics, Lund University, Lund, SwedenStormo Gary EditorWashington University in St. Louis, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 9 12 2005 9 11 2005 1 7 e7229 8 2005 9 11 2005 Copyright: © 2005 Ringnér and Krogh.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Using high-throughput technologies, abundances and other features of genes and proteins have been measured on a genome-wide scale in Saccharomyces cerevisiae. In contrast, secondary structure in 5′–untranslated regions (UTRs) of mRNA has only been investigated for a limited number of genes. Here, the aim is to study genome-wide regulatory effects of mRNA 5′-UTR folding free energies. We performed computations of secondary structures in 5′-UTRs and their folding free energies for all verified genes in S. cerevisiae. We found significant correlations between folding free energies of 5′-UTRs and various transcript features measured in genome-wide studies of yeast. In particular, mRNAs with weakly folded 5′-UTRs have higher translation rates, higher abundances of the corresponding proteins, longer half-lives, and higher numbers of transcripts, and are upregulated after heat shock. Furthermore, 5′-UTRs have significantly higher folding free energies than other genomic regions and randomized sequences. We also found a positive correlation between transcript half-life and ribosome occupancy that is more pronounced for short-lived transcripts, which supports a picture of competition between translation and degradation. Among the genes with strongly folded 5′-UTRs, there is a huge overrepresentation of uncharacterized open reading frames. Based on our analysis, we conclude that (i) there is a widespread bias for 5′-UTRs to be weakly folded, (ii) folding free energies of 5′-UTRs are correlated with mRNA translation and turnover on a genomic scale, and (iii) transcripts with strongly folded 5′-UTRs are often rare and hard to find experimentally.
Synopsis
In cells, proteins are made from messenger RNA copied from genes in the DNA. The amount of each protein needs to be controlled by cells. For this purpose, cells use a strategy that includes decomposing RNA and varying the number of proteins made from each RNA. One part of the RNA molecule is called the 5′–untranslated region (UTR), and it is known that this region can fold into a three-dimensional structure. For some genes, such structures are important for protein production. In this article, structures in 5′-UTRs are calculated for all genes in the yeast Saccharomyces cerevisiae. The authors show that structures in 5′-UTRs likely play a role in RNA decomposition and protein production for many genes in the genome: RNA molecules with weakly folded 5′-UTRs live relatively longer and produce more proteins. This study provides an example of how genome-wide computational analysis complements experimental results.
Citation:Ringnér M, Krogh M (2005) Folding free energies of 5′-UTRs impact post-transcriptional regulation on a genomic scale in yeast. PLoS Comput Biol 1(7): e72.
==== Body
Introduction
Regulation of gene expression is important for many cellular processes. Numerous studies have focused on the transcriptional level to investigate under what conditions a gene is transcribed and to what extent. These investigations have led to descriptions of system architectures, in which the activity of specific transcription factors regulates the activity of downstream target genes in such a way that the combined activity results in large developmental or physiological programs. In recent years, such descriptions have benefited from DNA microarray technology, which has provided overall mRNA levels for many systems. However, much less is known about the system architecture of regulation of gene expression at the post-transcriptional level, including regulation of mRNA subcellular localization, stability, and translation rate.
mRNA consists of three parts: a 5′–untranslated region (UTR) beginning with a 7-methyl-guanosine cap, a coding region, and a 3′-UTR ending in a poly(A) tail (Figure 1A). UTRs of mRNAs are known to be a crucial part of post-transcriptional regulation [1]. In yeast, the exact lengths of 5′- and 3′-UTRs are unknown for most genes. Mignone et al. [1] estimated the average lengths for yeast as 134 nucleotides (nt) for 5′-UTRs and 237 nt for 3′-UTRs. Later, Hurowitz and Brown [2] performed genome-wide measurements of total transcript lengths and calculated the average combined 5′- and 3′-UTR length to be 260 nt. Cis-acting sequence motifs in 3′-UTRs can interact with specific RNA-binding proteins (RBPs) to direct subcellular localization [3] and stability [4] of mRNAs. DNA microarrays have also enabled a growing body of work on global analysis of RBPs that supports the importance of RBPs in many cellular processes through post-transcriptional regulation of mRNAs [5–7]. Translation of the majority of mRNAs depends on cap-dependent ribosomal scanning of 5′-UTRs [8], and this process is influenced by features of 5′-UTRs. For example, ribosomal scanning is severely hampered by 5′-UTRs containing start codons or secondary structure [9–15].
Figure 1 Structure of Yeast mRNA
(A) The mRNA has a tripartite structure consisting of a 5′-UTR, a coding region, and a 3′-UTR. CRE, cis-acting regulatory element; m7G, 7-methyl-guanosine cap; SS, secondary structure.
(B) The computed minimum free energy secondary structure for the 5′-UTR of the gene YBR296C-A.
The purpose of mRNA degradation is 2-fold: to regulate transcript abundance and to destroy faulty transcripts. Degradation of mRNA in yeast occurs via 5′ to 3′ exonucleotic, 3′ to 5′ exonucleotic, and endonucleotic pathways [16–19]. Regulation of transcript abundance via the exonucleotic pathways occurs by first shortening the poly(A) tail followed either by removal of the 5′ cap, resulting in rapid 5′ to 3′ degradation, or by degradation from the 3′ end without prior decapping [18,20]. The dual importance of the cap structure, for translation initiation and 5′ to 3′ mRNA decay, has led to the hypothesis that there is a competition between translation and decay for access to the cap [13,20,21]. Transcripts with 5′-UTRs that hamper their translation often encode for proteins that need to be strongly and finely regulated, such as growth factors, transcription factors, and proto-oncogenes [14], suggesting that 5′-UTRs are sometimes structured in a way to prevent harmful overproduction of regulatory proteins. Indeed, some diseases are caused by mutations in 5′-UTRs [22,23]. In agreement with this picture, proteins involved in the regulation of dynamic cellular processes such as transcription, signal transduction, cell cycle control, and metabolism have long UTRs [2].
To our knowledge no one has found genome-wide associations between secondary structure in 5′-UTRs and mRNA half-life, translation rates, or other transcript features. For example, Bernstein et al. [24] performed a genome-wide experiment of mRNA decay in Escherichia coli and found no association to secondary structure in UTRs. To study the regulatory effects of secondary structure in UTRs, we performed genome-wide computations of secondary structures and their folding free energies in 5′-UTRs for 5,888 verified genes in Saccharomyces cerevisiae. The folding free energy is the difference in free energy between the unfolded and folded state. For a given mRNA length, a lower folding free energy corresponds to a more stable secondary structure. We analyzed associations between folding free energies and various transcript features including translation and decay rates. One result of our analysis is that low folding free energy of 5′-UTRs is, on average, associated with low translation rates and high transcript turnover, in concordance with previous results for single genes (e.g., [13]). We also found that 5′-UTRs on average are more weakly folded than random sequences with the same dinucleotide frequencies, and than intergenic, coding, and 3′-UTR sequences. Strikingly, genes with unknown function were enriched among genes with strongly folded 5′-UTRs.
Results
Folding Free Energies of 5′-UTRs
To investigate secondary structure in 5′-UTRs, we used the Vienna RNA package [25] to compute secondary structures and the corresponding free energy changes for folding (ΔG). The lower ΔG is, the more strongly folded is the secondary structure. Using 5′-UTRs of length 50 nt, the average ΔG was −4.3 kcal/mol (standard deviation [SD] = 2.9 kcal/mol) for the 5,888 open reading frames (ORFs) investigated. The range of ΔG was from −18.1 kcal/mol to 0 kcal/mol. The lowest value of ΔG, −18.1 kcal/mol, was obtained for the gene YBR296C-A, whose computed 5′-UTR secondary structure is illustrated in Figure 1B. There were 231 5′-UTRs with folding free energies below −10 kcal/mol. These thermodynamically most stable structures had on average 12.9 base pairs (SD = 2.2), i.e., more than half of the bases were typically paired. Their average GC-content was 47% (SD = 7%). The structures were mostly hairpins similar to Figure 1B with unpaired bases in internal or bulge loops or at the ends of the sequences, but also structures containing two hairpins were found. There were 727 5′-UTRs with folding energies above −1 kcal/mol. These 5′-UTRs formed minimum free energy structures having on average 2.6 base pairs (SD = 3.0) and their average GC-content was 29% (SD = 7%).
Folding Free Energies of Other Genomic Regions
Folding free energies were computed for three control groups, all containing 5,888 sequences of length 50 nt. The first group consisted of randomly chosen sequences from intergenic regions and had an average ΔG of −5.4 kcal/mol (SD = 3.4 kcal/mol). The second group consisted of the first 50 nt of the 3′-UTR of each ORF and had an average ΔG of −4.5 kcal/mol (SD = 3.1 kcal/mol). The third group consisted of the 50 nt located after the start codon of each ORF and had an average ΔG of −6.3 kcal/mol (SD = 3.2 kcal/mol). The free energies of the 5′-UTRs were significantly higher than those of the three other groups (3′-UTR: p < 3 × 10−4, intergenic: p < 2 × 10−70, coding: p < 3 × 10−253; Mann–Whitney U test). Figure 2A shows cumulative distributions of all free energies for the four groups.
Figure 2 Folding Free Energies of 5′-UTRs
(A) Cumulative distributions of folding free energies, ΔG, are shown for 5,888 ORFs for 5′-UTRs (50 nt upstream of the ORF; solid line), 3′-UTRs (50 nt downstream of the ORF; dashed-dotted line), coding sequences (50-nt sequences following downstream of the start codon of each ORF; dotted line), and 5,888 sequences of length 50 nt selected randomly from intergenic regions (dashed line).
(B) Distribution of Z-scores for 5′-UTRs of 5,888 ORFs. Each 5′-UTR sequence was shuffled 100 times and a Z-score was calculated for each to compare the folding free energy of the native sequence to the shuffled sequences. A histogram of these Z-scores is shown together with a standard normal distribution (dashed line).
Folding Free Energies of Randomized Sequences
The free energy of secondary structure in RNA is highly dependent on nucleotide composition. A base pair stacking term that depends on dinucleotides contributes to the free energy. We obtained the free energy contributions for each of the 16 possible dinucleotides from Xia et al. [26]. We calculated the dinucleotide frequencies in the four groups of sequences and used the dinucleotide energy contributions as weights in a weighted average of the dinucleotide frequencies. This calculation gave us a rudimentary measure for the contribution to the free energies coming from dinucleotide composition without actually folding the structures. The weighted dinucleotide composition for the four groups was −1.74 kcal/mol for 3′-UTRs, −1.81 kcal/mol for 5′-UTRs, −1.81 kcal/mol for intergenic sequences, and −1.95 kcal/mol for coding sequences. The dinucleotides with lowest free energy are GC, CC, GG and CG, so GC-content is an even simpler measure for the relative contribution of nucleotide composition to the free energy. The GC-content of the four groups was 31% for 3′-UTRs, 34% for 5′-UTRs, 34% for intergenic sequences, and 40% for coding sequences. Interestingly, the two measures are in perfect agreement.
We checked whether the folding free energies of 5′-UTRs were not only higher than for the other groups of sequences, but also different from what was expected from 5′-UTR dinucleotide composition [27]. For this purpose, we used a dinucleotide shuffling algorithm [28,29]. Native 5′-UTR sequences were shuffled 100 times each, and minimum free energies were calculated for all randomized sequences. The mean free energy of the randomized sequences was −4.4 kcal/mol as compared to −4.3 kcal/mol for the native sequences. Z-scores were calculated to compare the folding free energy of each 5′-UTR with the free energies of its randomized sequences. 5′-UTRs with positive Z-scores had higher folding free energies than the average of their randomized sequences and are therefore thought to have less secondary structure. We found an overabundance of 5′-UTRs with positive Z-scores (Figure 2B). The mean value of the Z-scores was 0.050 (standard error of the mean [SEM] = 0.013), which is significantly different from zero (p < 10−4; t-test). Also 58% of the 5,888 ORFs had a positive Z-score, which is significantly more than expected by chance (p < 3 × 10−35).
Folding Free Energies of 5′-UTRs and Transcript Features
We investigated the correlation between ΔG and the ribosome density measured by Arava et al. [30]. We observed a small but significant correlation (Figure 3). The Pearson correlation was 0.12, with an associated p-value of 3 × 10−16. Beyer et al. [31] argue that it is preferable to define ribosome density as the number of ribosomes divided by transcript length instead of ORF length. They provide a processed dataset of such ribosome densities, and these densities had a Pearson correlation of 0.09 (p < 10−10) with ΔG. Likewise, using mRNA half-lives measured by Wang et al. [32], we observed a small but significant correlation between ΔG and mRNA half-lives (Figure 4). The Pearson correlation was 0.10 (p < 3 × 10−10). We also found significant correlations between ΔG on the one hand and ribosome occupancy, the number of ribosomes bound on the transcript, the mRNA copy number, and protein abundance on the other hand (Table 1). To avoid potential pitfalls in the assumptions used to calculate p-values for Pearson correlations, we also calculated Spearman rank correlations. We observed similar results for both correlation measures (Table 1). In contrast to our results for 5′-UTRs, we found no significant correlations between folding free energies of 3′-UTRs and transcript features.
Figure 3 Comparison between Ribosome Densities and Folding Free Energies of 5′-UTRs
(A) Scatter plot of mRNA ribosome density and folding free energy of the 5′-UTR (ΔG) for 5,888 ORFs.
(B) ORFs were grouped based on the change in free energy (ΔG). For each energy group, the average ribosome density (±SEM) is shown. From left to right, the number of ORFs in each energy group used to calculate the average density was 573, 796, 1,214, 1,438, and 1,187.
Figure 4 Comparison between mRNA Half-Lives and Folding Free Energies of 5′-UTRs
(A) Scatter plot of mRNA half-life and folding free energy of the 5′-UTR (ΔG) for 5,888 ORFs.
(B) ORFs were grouped based on the folding free energy (ΔG). For each energy group, the average mRNA half-life (±SEM) is shown. From left to right, the number of ORFs in each energy group used to calculate the average density was 467, 657, 982, 1,169, and 983.
Table 1 Correlations between Secondary Structure in 5′-UTRs and Transcript Features
As expected, we observed a large correlation between ΔG and GC-content for the 5′-UTRs. The Pearson correlation was 0.48 (p < 3 × 10−16). To rule out that our observed correlations between ΔG and transcript features were merely a consequence of GC-content, we investigated whether ΔG was correlated with the transcript features independently of GC-content. We regressed the transcript features as a function of GC-content and free energy in a multivariate model. First, significance was calculated for the correlation between GC-content and a transcript feature. Second, significance was calculated for free energy being correlated to the transcript features after subtraction of the GC-content effect. For ribosome density, we obtained p = 5 × 10−4 for GC-content and p < 5 × 10−14 for free energy. For mRNA half-life, we obtained p < 10−15 for GC-content and p < 0.004 for free energy. For the combined protein abundance dataset [31], we obtained p < 2 × 10−12 for GC-content and p < 0.0002 for free energy. Similar results were obtained when correcting for weighted dinucleotide composition instead of for GC-content.
Fast and Slowly Decaying Genes
In order to check whether the relations between various transcript features depended on the half-life of the mRNA, we designated the 1,013 genes with a half-life below 13 min as fast decaying, and the 1,058 genes with a half-life above 33 min as slowly decaying. These cutoffs were chosen to get closest to, and above, 1,000 genes. The only correlations between ΔG and any of the other nine transcript features in Table 1 that changed significantly (p < 0.001) were with half-life and heat shock: in the fast decaying group of genes, ΔG and half-life had a correlation of −0.06, which is significantly different from their correlation of 0.10 among all genes (p < 8 × 10−7). Similarly in the fast decaying group of genes, ΔG and heat shock had a correlation of −0.01, which is significantly different from their correlation of 0.10 among all genes (p < 6 × 10−4).
Correlation between Decay and Translation
It has been argued that translational efficiency of a transcript is a determinant of mRNA half-life: decreased translation leads to decreased half-life. Evidence for this model has come from yeast strains either mutated in translation initiation factors [33] or with translation of individual mRNAs inhibited [13]. To see whether such an effect is present globally in yeast without such modifications, we calculated the correlations between half-life on the one hand and ribosome density and ribosome occupancy on the other hand. We found a small, but significant, correlation among all genes. However, for the fast decaying genes the correlations were much stronger, especially between half-life and ribosome occupancy, for which the correlation was 0.24 (Figure 5).
Figure 5 Correlations between Decay and Translation Rates
Pearson correlations together with corresponding p-values are shown for mRNA half-life versus (A) ribosome density and (B) ribosome occupancy. ORFs were, depending on mRNA half-life, grouped into all ORFs, 1,058 slowly decaying ORFs with t
1/2 ≥ 33 min, and 1,013 fast decaying ORFs with t
1/2 ≤ 13 min.
Gene Ontology Analysis
To see whether folding free energies of 5′-UTRs were associated with functional annotations, we mapped the 5,888 genes to 3,678 Gene Ontology (GO) categories [34]. The genes were ranked according to ΔG in both increasing and decreasing order, and a Wilcoxon rank sum test was employed for each GO category [35]. The significant categories, using a very stringent p-value cutoff of 10−8, corresponding to a Bonferroni corrected cutoff of 4 × 10−5, are listed in Table 2. Among genes with strongly folded 5′-UTRs, three categories were significant and no other categories were close to being this significant. Remarkably, these three categories were “molecular function unknown,” “biological process unknown,” and “cellular component unknown.” Among genes with weakly folded 5′-UTRs, 12 categories were significantly overrepresented. Chief among these were categories related to retrotransposons.
Table 2 GO Terms Overrepresented among Genes with Strongly and Weakly Folded 5′-UTRs
RNA-Binding Proteins
Affinity tagging of RBPs followed by microarray hybridizations has been used to obtain genome-wide lists of bound transcripts. We obtained the lists of bound transcripts for the RBPs Yra1, Mex67 [6], and for five members of the Puf family [7]. Furthermore, we identified all the genes whose 3′-UTR contained the consensus motifs for Puf3p, Puf4p, and Puf5p. For each of these ten gene sets, we examined whether there was a significant difference in the number of fast decaying genes relative to slowly decaying genes, and whether there was a significant difference in the number of genes having strongly folded 5′-UTRs relative to genes with weakly folded 5′-UTRs (Table 3). The most significant associations were that transcripts bound by Puf3p, Puf4p, and Puf5p were fast decaying. The Puf3p, Puf4p, and Puf5p motifs confirm this picture. The most significant associations with folding free energy were that Mex67 and Yra1 preferentially bind transcripts with weakly folded 5′-UTRs.
Table 3 Number of RBP Targets and Sequence Motifs Found in All mRNAs, mRNAs with Fast and Slow Decay Rates, and mRNAs with Strongly and Weakly Folded 5′-UTRs
Comparison with Longer Upstream Regions
The 5′-UTRs of yeast genes vary in length. In this study, we used the 50 nt upstream of the start codon as a representation of the 5′-UTR. Since 50 nt is shorter than many 5′-UTRs, we also used 100- and 200-nt 5′-UTRs for comparison. The correlation between ΔG and ribosome density decreases for longer regions, but is still significant for 100 nt (Table 4). Similar behavior was observed for other transcript features.
Table 4 Pearson Correlations between Ribosome Density and ΔG in 5′-UTRs of Length 50, 100, and 200 nt
Discussion
We carried out genome-wide computations of secondary structures in 5′-UTRs of mRNA in yeast, and correlated 5′-UTR folding free energy with various other transcript features. We chose somewhat arbitrarily to fold sequences of length 50 nt upstream of the coding start, because these sequences are almost certainly inside the 5′-UTR. We also folded 100- and 200-nt sequences, and had similar but weaker results (Table 4). Folding of RNA is somewhat local in sequence: when folding 100- or 200-nt upstream sequences, the last 50 nt were typically computed to fold into the same structure as when the 50-nt upstream sequences were folded. Translation has been shown to be most sensitive to secondary structure close to the 5′ end of mRNA [12]. Hence, we think that the weaker results obtained for longer upstream sequences reflect an increase of sequence spanning genomic DNA not being transcribed, and not that secondary structure close to the translation start is most important for the transcript features we have investigated. We used 5′-UTRs of fixed length to avoid comparing free energies for sequences of different lengths. Bernstein et al. [24] used predicted UTRs for each gene in E. coli and found no association between secondary structure in UTRs and mRNA half-life. Our different findings may be due to differences between pro- and eukaryotes, or difficulties in comparing UTRs of different length.
To compare 5′-UTRs with other genomic regions, 50-nt sequences from intergenic regions, coding regions, and 3′-UTRs were also folded. These three sets of sequences had significantly lower free energies than the 5′-UTR sequences (see Figure 2A). The folding free energy of RNA depends on both nucleotide composition and the order of the nucleotides. The nucleotide composition, quantified both by GC-content and weighted dinucleotide composition, was similar in 5′-UTRs and intergenic regions, indicating that the difference in free energies between these groups is due to nucleotide order. Indeed, the 5′-UTRs had higher folding free energies than random sequences with the same dinucleotide composition (Figure 2B). In contrast, yeast coding regions have lower folding energies than randomized sequences preserving the encoded protein, the codon usage, and the dinucleotide composition [36]. This opposite behavior is in agreement with the huge difference in folding free energies between coding regions and 5′-UTRs (Figure 2A), even though GC-content probably is more important for this difference. Our results indicate that there has been evolutionary selection for 5′-UTRs to be weakly folded and suggest that folding free energy might be used as one probabilistic component of a gene prediction program.
In line with our observation that 5′-UTRs tend to be weakly folded is our finding that uncharacterized ORFs are overrepresented among the genes with strongly folded 5′-UTRs. Assuming that uncharacterized genes typically are expressed at low levels or under rare conditions, or even are pseudogenes, this finding hints at a larger selective pressure for absence of secondary structure for commonly or highly expressed genes. Confirming this picture is our finding that 5′-UTR folding free energy is significantly positively correlated with mRNA copy number and protein abundance (see Table 1). Since we only investigated verified genes, we could look into the source of the verification of the genes with strongly folded 5′-UTRs. The 5′-UTR of the gene YBR296C-A (see Figure 1B) has the secondary structure with the lowest free energy of all genes, and is annotated as unknown in GO. Remarkably, this gene has only one literature reference, in which Kumar et al. [37] describe an approach for finding overlooked genes in yeast. Of the 137 new genes reported by Kumar et al., 41 are annotated as verified in the Saccharomyces Genome Database (SGD). Ten of these 41 genes have a free energy below −10 kcal/mol, which is significantly more genes than expected by chance (p = 4 × 10−6, Fisher's exact test).
The three most significant GO categories among the genes with weakly folded 5′-UTRs were related to Ty element retrotransposons (see Table 2). Ty element retrotransposons are stretches of DNA that replicate and move in the genome through RNA intermediates [38]. The Ty elements contain various genes in their sequences, e.g., proteases, integrases, and reverse transcriptases. The fact that they have weakly folded 5′-UTRs suggests that folding of their RNA is detrimental to their function or integration in the genome. Interestingly, Ty elements showed up in a study of RNA half-life where different methods of transcriptional inhibition were compared [39]. The RNA transcripts whose stability differed most between rpb1–1 inhibition on the one hand and Thiolutin, 1,10-phenanthroline, and 6-azauracil on the other hand were predominantly Ty elements. It may be worth investigating whether there is a connection between this difference in transcript stability and the lack of 5′-UTR secondary structure.
We found that 5′-UTR folding free energy was significantly positively correlated with both translational activity and mRNA half-life (see Table 1). These correlations were still significant after correction for GC-content, indicating that the correlations are not simply a secondary effect caused by nucleotide frequencies. Parker and colleagues showed that the insertion of secondary structures into the 5′-UTR of PGK1 yeast mRNA inhibited translation and stimulated decay of PGK1 [13]. Together, these findings suggest a widespread use of 5′-UTR secondary structure in post-transcriptional regulation. Our correlations may not be caused by any biochemical mechanism, e.g., transcripts of one evolutionary origin could have both strongly folded 5′-UTRs and low translation rates, whereas transcripts of another evolutionary origin could have weakly folded 5′-UTRs and high translation rates. Nevertheless, we believe that the correlations do reflect more direct connections. Our findings may be explained by an inhibitory effect of 5′-UTR secondary structure on translation initiation combined with competition between translation and decay. However, more direct biochemical pathways preferentially degrading mRNA with 5′-UTR secondary structure might also exist. Early support for the inhibitory effect of 5′-UTR secondary structure on translation came from insertion of hairpin loops into 5′-UTRs [9,10]. Later studies have shown connections between mRNA 5′ secondary structure and proteins important for translation such as eIF4A [15]. Competition between translation and decay has been proposed because both may require cap access [20,33]. Moreover, during translation the mRNA is circularized through interactions between cap-binding translation initiation factors and the poly(A)-binding protein (PABP). This conformation presumably protects mRNA from degradation by preventing access to both the cap and the poly(A) tail, suggesting that also the poly(A) tail is important for competition [17]. We expected that such competition would be more easily seen for short-lived transcripts because degradation takes up a larger part of their lives. Indeed, our global analysis revealed that transcript half-life is positively correlated with both ribosome density and ribosome occupancy, in particular for short-lived transcripts (see Figure 5).
A major mediator of heat shock response is mRNA decay [40], and the mRNA decay profile is similar to the heat response [39]. In line with these observations, we found a positive correlation between 5′-UTR free energy and mRNA response to heat shock (Table 1), i.e., transcripts with weakly folded 5′-UTRs are, in addition to being relatively long-lived, relatively upregulated after a heat shock. Given that transcripts that are upregulated by heat shock have weakly folded 5′-UTRs, it is expected that they would be translated at relatively high rates. Indeed, the correlation between ribosome occupancy and relative upregulation 10 min after heat shock was 0.23 (p < 2 × 10−60; similar for 5 min). Of interest, the heat shock mRNA Hsp90 in Drosophila has extensive secondary structure in its 5′-UTR. Hsp90 translation is inefficient at normal growth temperature, and is activated by heat shock, perhaps by thermal destabilization of the secondary structure in the 5′-UTR [41]. It may be worthwhile to perform genome-wide protein abundance experiments of heat shock response to investigate whether preferential heat shock translation is a common mechanism.
We assessed whether transcripts associated with RBPs, or with sequence motifs associated with these RBPs in their 3′-UTRs, were over- or underrepresented among fast decaying transcripts or among transcripts with strongly folded 5′-UTRs. Puf proteins are known to enhance mRNA turnover or repress translation [42]. We found targets of Puf3p, Puf4p, and Puf5p proteins to be significantly associated with fast decay, extending an earlier study [43]. Perhaps of interest, we note that the three Puf proteins for which Gerber et al. identified sequence motifs [7] were associated with fast decaying transcripts, while the remaining two Puf proteins, as well as Mex67 and Yra1, instead tended to be associated with weakly folded 5′-UTRs.
To summarize, we found that (i) 5′-UTRs have higher folding free energies than other genomic regions and than expected from their nucleotide composition, (ii) secondary structures in 5′-UTRs likely play a role in mRNA translation and turnover on a genomic scale, and (iii) genes with strongly folded 5′-UTRs are generally rarer, harder to find experimentally, and less annotated. It is important to keep in mind that the highly significant correlations we have found are small, showing that folding of 5′-UTRs is, as expected, only one aspect of post-transcriptional regulation. However, the correlations may be larger in subgroups of mRNAs, such as mRNAs targeted by individual decay pathways [44] and specific RBPs [45]. An example of a larger correlation in a subgroup is our observation that translational activity and mRNA decay are highly correlated for mRNAs with short half-lives.
Materials and Methods
Untranslated regions.
The exact 5′- and 3′-UTR lengths are unknown for most yeast genes. Mignone et al. [1] estimated the average lengths for yeast as 134 nt for 5′-UTRs and 237 nt for 3′-UTRs. With these numbers in mind, we retrieved 50, 100, and 200 nt of predicted 5′-UTRs and 237 nt of predicted 3′-UTRs from SGD for the 5,888 ORFs annotated as verified in SGD. As three additional control groups of 50-nt sequences, we retrieved nt 4–53 downstream (the first 50 nt following the start codon) for each of the 5,888 ORFs, the first 50 nt of the 3′-UTR region for each of the 5,888 ORFs, and 5,888 randomly chosen 50-nt sequences from intergenic regions from SGD.
Folding of RNA secondary structures.
We used the RNAfold program in the Vienna RNA package [25] with default values for parameters (T = 37 °C) to compute secondary structures from RNA sequences. For each sequence, we used the free energy of the minimum free energy structure (the most negative ΔG) as a measure for secondary structure formation. For a given sequence, there may be other structures with similar ΔG, but we are interested in the possible free energy change in folding and not the secondary structure itself. A low ΔG corresponds to a strongly folded UTR, while a high ΔG corresponds to a weakly folded UTR. To avoid any pitfalls with using the free energy of the most strongly folded structure for each sequence, we also performed our analysis using the ensemble free energies [25], and none of the conclusions presented in this study changed. In fact, the correlations were typically somewhat more significant using ensemble averages. The free energies for all 5,888 genes are available in Dataset S1.
Transcript feature datasets.
Translation profiles measured by Arava et al. [30] were downloaded from http://genome-www.stanford.edu/yeast_translation/. From this dataset, the number of bound ribosomes, the ribosome density (number of ribosomes per unit ORF length), and the ribosome occupancy (the fraction of the transcripts engaged in translation) were extracted for 5,700 genes, together with the mRNA copy number for 5,643 genes. Half-lives for 4,687 genes measured by Wang et al. [32] were downloaded from http://genome-www.stanford.edu/turnover/. A second dataset of mRNA half-lives for 6,092 genes [39] was obtained from http://hugheslab.med.utoronto.ca/Grigull/. For this dataset, the decay ratios 5 min after temperature shift of the rpb1–1 strain were used. Changes in transcript abundance in cells responding to heat shock for 5,259 genes measured by Gasch et al. [46] were downloaded from http://www-genome.stanford.edu/yeast_stress/. A dataset containing protein abundance information for 2,044 genes, constructed by Greenbaum et al. [47] by merging publicly available two-dimensional electrophoresis and MudPit data, was downloaded from http://bioinfo.mbb.yale.edu/expression/prot-v-mrna. Another dataset containing protein abundance information for 1,669 genes was obtained from the experiment by Ghaemmaghami et al. [48]. A merger of these two protein abundance datasets was obtained from Beyer et al. [31], along with a set of ribosome densities normalized by transcript length instead of ORF length.
Dinucleotide shuffling.
Each native 5′-UTR sequence was shuffled 100 times keeping the dinucleotide frequencies constant using a publicly available implementation [28] of an algorithm developed by Altschul and Erickson [29]. For each 5′-UTR, the mean and the SD of the free energies of its randomized sequences were calculated. A Z-score was defined for each 5′-UTR as the free energy of the native sequence minus the mean of its randomized sequences divided by the SD of its randomized sequences [49].
Statistical analysis.
Pearson correlations, Spearman rank correlations, Fisher's exact tests on 2 × 2 contingency tables, Mann–Whitney U tests, t-tests, and corresponding p-values were calculated using the statistics package R [50]. For Pearson correlations, p-values were calculated as the probability of obtaining a better correlation by chance if the two vectors were drawn independently from a Gaussian distribution. The multivariate linear model was done in R as well, and p-values were obtained with the ANOVA test of a linear model. All p-values were two-sided.
GO analysis.
The 5,888 genes were mapped to 3,678 GO categories [34] using annotations from SGD. The genes were ranked according to ΔG in both increasing and decreasing order, separately, and a Wilcoxon rank sum test was employed for each GO category using Catmap [35]. Catmap outputs p-values calculated as the probability that a random ordering of the genes produces a lower, or equally low, Wilcoxon rank sum as the ordering investigated. The p-values were multiplied with the number of categories (3,678) to obtain Bonferroni corrected p-values.
Supporting Information
Dataset S1 Calculated Folding Free Energies for All 5,888 Genes
(487 KB TXT)
Click here for additional data file.
We thank Peter Johansson and Kasper Astrup Eriksen for valuable discussions, three anonymous reviewers for helpful suggestions, and Peter Schuster for giving a talk that inspired us to enter the world of RNA folding. This work was in part supported by the Swedish Research Council and the Knut and Alice Wallenberg Foundation through the Swegene consortium.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. MR and MK conceived and designed the experiments, performed the experiments, analyzed the data, and wrote the paper.
A previous version of this article appeared as an Early Online Release on November 9, 2005 (DOI: 10.1371/journal.pcbi.0010072.eor).
Abbreviations
GOGene Ontology
ntnucleotide(s)
ORFopen reading frame
RBPRNA-binding protein
SDstandard deviation
SEMstandard error of the mean
SGD
Saccharomyces Genome Database
UTRuntranslated region
==== Refs
References
Mignone F Gissi C Liuni S Pesole G 2002 Untranslated regions of mRNAs Genome Biol 3 REVIEWS0004 11897027
Hurowitz EH Brown PO 2003 Genome-wide analysis of mRNA lengths in Saccharomyces cerevisiae
Genome Biol 5 R2 14709174
Jansen RP 2001 mRNA localization: Message on the move Nat Rev Mol Cell Biol 2 247 256 11283722
Bashirullah A Cooperstock RL Lipshitz HD 2001 Spatial and temporal control of RNA stability Proc Natl Acad Sci U S A 98 7025 7028 11416182
Keene JD Tenenbaum SA 2002 Eukaryotic mRNPs may represent posttranscriptional operons Mol Cell 9 1161 1167 12086614
Hieronymus H Silver PA 2003 Genome-wide analysis of RNA-protein interactions illustrates specificity of the mRNA export machinery Nat Genet 33 155 161 12524544
Gerber AP Herschlag D Brown PO 2004 Extensive association of functionally and cytotopically related mRNAs with Puf family RNA-binding proteins in yeast PLoS Biol 2 e79. DOI: 10.1371/journal.pbio.0020079 15024427
Kozak M 1989 The scanning model for translation: An update J Cell Biol 108 229 241 2645293
Pelletier J Sonenberg N 1985 Insertion mutagenesis to increase secondary structure within the 5′ noncoding region of a eukaryotic mRNA reduces translational efficiency Cell 40 515 526 2982496
Baim SB Sherman F 1988 mRNA structures influencing translation in the yeast Saccharomyces cerevisiae
Mol Cell Biol 8 1591 1601 2837649
Kozak M 1987 An analysis of 5′-noncoding sequences from 699 vertebrate messenger RNAs Nucleic Acids Res 15 8125 8148 3313277
Kozak M 1989 Circumstances and mechanisms of inhibition of translation by secondary structure in eucaryotic mRNAs Mol Cell Biol 9 5134 5142 2601712
Muhlrad D Decker CJ Parker R 1995 Turnover mechanisms of the stable yeast PGK1 mRNA Mol Cell Biol 15 2145 2156 7891709
van der Velden AW Thomas AAM 1999 The role of the 5′ untranslated region of an mRNA in translation regulation during development Int J Biochem Cell Biol 31 87 106 10216946
Svitkin YV Pause A Haghighat A Pyronnet S Witherell G 2001 The requirement for eukaryotic initiation factor 4A (elF4A) in translation is in direct proportion to the degree of mRNA 5′ secondary structure RNA 7 382 394 11333019
Muhlrad D Decker CJ Parker R 1994 Deadenylation of the unstable mRNA encoded by the yeast MFA2 gene leads to decapping followed by 5′ to 3′ digestion of the transcript Genes Dev 8 855 866 7926773
Wilusz CJ Wormington M Peltz SW 2001 The cap-to-tail guide to mRNA turnover Nat Rev Mol Cell Biol 2 237 246 11283721
Mata J Marguerat S Bahler J 2005 Post-transcriptional control of gene expression: A genome-wide perspective Trends Biochem Sci 30 506 514 16054366
Parker R Song H 2004 The enzymes and control of eukaryotic mRNA turnover Nat Struct Mol Biol 11 121 127 14749774
Tucker M Parker R 2000 Mechanisms and control of mRNA decapping in Saccharomyces cerevisiae
Annu Rev Biochem 69 571 595 10966469
de la Cruz BJ Prieto S Scheffler IE 2002 The role of the 5′ untranslated region (UTR) in glucose-dependent mRNA decay Yeast 19 887 902 12112242
Kozak M 2002 Emerging links between initiation of translation and human diseases Mamm Genome 13 401 410 12226704
Pickering BM Willis AE 2005 The implications of structured 5′ untranslated regions on translation and disease Semin Cell Dev Biol 16 39 47 15659338
Bernstein JA Khodursky AB Lin PH Lin-Chao S Cohen SN 2002 Global analysis of mRNA decay and abundance in Escherichia coli at single-gene resolution using two-color fluorescent DNA microarrays Proc Natl Acad Sci U S A 99 9697 9702 12119387
Hofacker IL Fontana W Stadler PF Bonhoeffer S Tacker M 1994 Fast folding and comparison of RNA secondary structures Monatsh Chem 125 167 188
Xia T SantaLucia JJ Burkard ME Kierzek R Schroeder SJ 1998 Thermodynamic parameters for an expanded nearest-neighbor model for formation of RNA duplexes with Watson-Crick base pairs Biochemistry 37 14719 14735 9778347
Workman C Krogh A 1999 No evidence that mRNAs have lower folding free energies than random sequences with the same dinucleotide distribution Nucleic Acids Res 27 4816 4822 10572183
Clote P Ferre F Kranakis E Krizanc D 2005 Structural RNA has lower folding energy than random RNA of the same dinucleotide frequency RNA 11 578 591 15840812
Altschul SF Erickson BW 1985 Significance of nucleotide sequence alignments: A method for random sequence permutation that preserves dinucleotide and codon usage Mol Biol Evol 2 526 538 3870875
Arava Y Wang Y Storey JD Liu CL Brown PO 2003 Genome-wide analysis of mRNA translation profiles in Saccharomyces cerevisiae
Proc Natl Acad Sci U S A 100 3889 3894 12660367
Beyer A Hollunder J Nasheuer HP Wilhelm T 2004 Post-transcriptional expression regulation in the yeast Saccharomyces cerevisiae on a genomic scale Mol Cell Proteomics 3 1083 1092 15326222
Wang Y Liu CL Storey JD Tibshirani RJ Herschlag D 2002 Precision and functional specificity in mRNA decay Proc Natl Acad Sci U S A 99 5860 5865 11972065
Schwartz DC Parker R 1999 Mutations in translation initiation factors lead to increased rates of deadenylation and decapping of mRNAs in Saccharomyces cerevisiae
Mol Cell Biol 19 5247 5256 10409716
Harris MA Clark J Ireland A Lomax J Ashburner M 2004 The Gene Ontology (GO) database and informatics resource Nucleic Acids Res 32 258 261
Breslin T Edén P Krogh M 2004 Comparing functional annotation analyses with Catmap BMC Bioinformatics 5 193 15588298
Katz L Burge CB 2003 Widespread selection for local RNA secondary structure in coding regions of bacterial genes Genome Res 13 2042 2051 12952875
Kumar A Harrison PM Cheung KH Lan N Echols N 2002 An integrated approach for finding overlooked genes in yeast Nat Biotechnol 20 58 63 11753363
Kim JM Vanguri S Boeke JD Gabriel A Voytas DF 1998 Transposable elements and genome organization: A comprehensive survey of retrotransposons revealed by the complete Saccharomyces cerevisiae genome sequence Genome Res 8 464 478 9582191
Grigull J Mnaimneh S Pootoolal J Robinson MD Hughes TR 2004 Genome-wide analysis of mRNA stability using transcription inhibitors and microarrays reveals posttranscriptional control of ribosome biogenesis factors Mol Cell Biol 24 5534 5547 15169913
Lindquist S 1981 Regulation of protein synthesis during heat shock Nature 293 311 314 6792546
Ahmed R Duncan RF 2004 Translational regulation of Hsp90 mRNA. AUG-proximal 5′-untranslated region elements essential for preferential heat shock translation J Biol Chem 279 49919 49930 15347681
Wickens M Bernstein DS Kimble J Parker R 2002 A PUF family portrait: 3′UTR regulation as a way of life Trends Genet 18 150 157 11858839
Graber JH 2003 Variations in yeast 3′-processing cis-elements correlate with transcript stability Trends Genet 19 473 476 12957538
He F Li X Spatrick P Casillo R Dong S 2003 Genome-wide analysis of mRNAs regulated by the nonsense-mediated and 5′ to 3′ mRNA decay pathways in yeast Mol Cell 12 1439 1452 14690598
Tenenbaum SA Carson CC Atasoy U Keene JD 2003 Genome-wide regulatory analysis using en masse nuclear run-ons and ribonomic profiling with autoimmune sera Gene 317 79 87 14604794
Gasch AP Spellman PT Kao CM Carmel-Harel O Eisen MB 2000 Genomic expression programs in the response of yeast cells to environmental changes Mol Biol Cell 11 4241 4257 11102521
Greenbaum D Colangelo C Williams K Gerstein M 2003 Comparing protein abundance and mRNA expression levels on a genomic scale Genome Biol 4 117 12952525
Ghaemmaghami S Huh WK Bower K Howson RW Belle A 2003 Global analysis of protein expression in yeast Nature 425 737 741 14562106
Seffens W Digby D 1999 mRNAs have greater negative folding free energies than shuffled or codon choice randomized sequences Nucleic Acids Res 27 1578 1584 10075987
Ihaka R Gentleman R 1996 R: A language for data analysis and graphics J Comput Grap Stat 5 299 314
|
16355254
|
PMC1309706
|
CC BY
|
2021-01-05 09:18:23
|
no
|
PLoS Comput Biol. 2005 Dec 9; 1(7):e72
|
utf-8
|
PLoS Comput Biol
| 2,005 |
10.1371/journal.pcbi.0010072
|
oa_comm
|
==== Front
Mol PainMolecular Pain1744-8069BioMed Central London 1744-8069-1-331629723810.1186/1744-8069-1-33ResearchFormalin injection causes a coordinated spinal cord CO/NO-cGMP signaling system response Shi Xiaoyou [email protected] Xiangqi [email protected] J David [email protected] Stanford University Department of Anesthesiology, Stanford, CA, USA2 Veterans Affairs Palo Alto Healthcare System, Palo Alto, CA, USA2005 18 11 2005 1 33 33 15 7 2005 18 11 2005 Copyright © 2005 Shi et al; licensee BioMed Central Ltd.2005Shi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The CO/NO-cGMP signalling system participates in the regulation of many physiological processes. The roles this system plays in spinal cord nociceptive signalling are particularly important. While individual components have been examined in isolation, little study has been dedicated to understanding the regulation and functioning of the system as a whole.
Results
In these studies we examined the time course of expression of 13 genes coding for components of this system including isoforms of the heme oxygenase (HO), nitric oxide synthase (NOS), soluble guanylate cyclase (sGC), cGMP dependent protein kinase (PKG) and phosphodiesterase (PDE) enzyme systems. Of the 13 genes studied, 11 had spinal cord mRNA levels elevated at one or more time points up to 48 hours after hindpaw formalin injection. Of the 11 with elevated mRNA, 8 had elevated protein levels 48 hours after formalin injection when mechanical allodynia was maximal. No component had an increased protein level which did not have an increased mRNA level at one or more time points. Injection of morphine 10 mg/kg prior to formalin completely abolished the acute nociceptive behaviours, but did not alter the degree of sensitivity which developed in the formalin treated hind paws during the subsequent 48 hours. Morphine treatment did, however, eliminate formalin induced increases in enzyme protein levels.
Conclusion
Our results indicate that the expression of the components of the CO/NO-cGMP signalling system seems to be coordinated in such a way that a generalized multi-level enhancement rather than a tightly limited step specific response occurs with noxious stimulation. Furthermore, the analgesic morphine administered prior to noxious stimulation can prevent long-term changes in gene expression though not necessarily nociceptive sensitisation.
==== Body
Background
Traditional approaches to the study of nociceptive signaling mechanisms often focus on single gene products. The functions of such molecules are commonly examined using pharmacological, electrophysiological, genetic and behavioural techniques. While there is little doubt that these investigations have been of substantial utility in understanding nociceptive signaling, it is widely appreciated that nociceptive signaling involves systems of interacting component molecules rather than individual molecules functioning independently. Furthermore, experimental paradigms are often optimised to highlight the roles of the molecules of interest making it difficult to integrate the existing literature towards a better understanding of how signaling system components function together under a single well defined set of parameters.
The analysis of changes in gene expression in models of pain is one type of paradigm often used to infer participation of a gene product in nociceptive signaling. Unfortunately, examination of the existing literature reveals the use of different species, strains, types of noxious stimulation, tissues analysed, time courses, mRNA versus protein measurement and other methods of analysis. Few studies have carefully examined changes in gene expression for large sets of genes coding for the many members of nociception-related signaling pathways.
The spinal cord CO/NO-cGMP signaling system is one of the best studied nociceptive signaling systems, and the available reports demonstrate some of the issues introduced above. The monoxides CO and NO are produced by heme oxygenase (HO) and nitric oxide synthase (NOS) respectively. The roles of HO in many pain models have been examined with changes in the expression of the spinal cord HO-2 isoform reported for both neuropathic and incisional pain models, though always studied apart from other CO/NO-cGMP signaling system components [1-4]. Likewise for NOS, increases in the spinal cord expression of nNOS I, iNOS and eNOS have been reported after noxious stimulation of various types. In 2 reports 2 NOS species were studied simultaneously after a noxious inflammatory stimulus with Wu et al. reporting increased spinal cord expression of nNOS and iNOS after capsaicin injection [5], but Tao et al. reporting increased expression of nNOS but not iNOS after carrageenan injection [6]. Once produced, these monoxides converge on guanylate cyclase to stimulate the production of cGMP. For soluble guanylate cyclase, type 1α (sGC1α), up-regulation in a model of inflammatory pain has been described, though other isoforms are not well studied [7]. The presumed target of cGMP is cGMP dependent protein kinase (PKG). PKG1α undergoes up-regulation in spinal cord tissue though PKG1β and PKG II expression do not appear to have been studied [8,9]. Finally, various phosphodiesterases (PDE's) can metabolise cGMP thus terminating its signaling functions, but we could identify no studies following the spinal cord expression of these enzymes in any model of pain. Basal expression of PDE2,3 and 5 has been demonstrated in spinal cord tissue [10,11].
Thus while the data pertaining to gene expression for some CO/NO-cGMP signaling system components and the large number of pharmacological studies not cited above support roles for this system in nociceptive signal transmission, few studies have attempted to follow the expression of the various components under standardized conditions. Such data would be useful in understanding how the system may be regulated as a whole in the setting of a specific type of pain. When 10 components of the spinal cord CO/NO-cGMP signaling system were studied together during the chronic exposure of mice to morphine, it was observed that 7 of the components were up-regulated in a coordinated fashion [12], thus coordination of expression in a nociceptive system seems plausible. The studies outlined below follow the expression of a set of 13 genes coding for the various components of the spinal cord CO/NO-cGMP signaling system over time in the formalin model of inflammatory pain.
Results
Formalin testing
To characterize the spontaneous pain behaviors and persistent mechanical allodynia associated with the hindpaw injection of formalin in the C57Bl/6J mice used in subsequent experiments, we first measured the total times spent licking the injected hind paws after formalin injection. The phase I (0–5 min) average licking behavior was 65 +/- 10 sec, and the total phase II (10–40 min) average was 179 +/- 20 sec similar to the observations reported earlier by our laboratory [2] (Figure 1A). The administration of 10 mg/kg morphine prior to formalin injection nearly completely eliminated formalin-induced licking behaviors during phase I and phase II.
Figure 1 Nociceptive behaviours displayed after formalin injection. In panel A the nociceptive behaviour of mice is presented as the total number of seconds spent licking in 5 minute intervals beginning with the injection of formalin. Some mice were pre-treated with a 10 mg/kg injection of morphine. In panel B, the time course of formalin-induced mechanical allodynia with and without morphine pre-treatment is displayed. Mechanical withdrawal thresholds were followed using von Frey fibers. Data are presented as mean licking times or thresholds +/- SEM, **p < 0.01.
In Figure 1B data from experiments in which mechanical von Frey withdrawal thresholds were measured in mice up to 48 hours after formalin injection are presented. This mechanical nociceptive sensitization is a more chronic consequence of formalin injection than the phase I and II licking behavior. The mechanical thresholds in the formalin injected mice were significantly lower than those observed in saline-injected animals at all time points after formalin injection (P < 0.01), and the allodynia was relatively stable over this 48 hour time course. The administration of morphine did not significantly alter the measured allodynia over this time course.
Expression of components of the CO/NO-cGMP signaling system – mRNA
To examine possible changes in expression in the various components of the CO/NO-cGMP signaling system, we harvested lumbar spinal cord tissue at time points up to 48 hours after formalin injection. In Figure 2, 3, 4, 5, 6 the time courses of spinal cord expression of these genes are displayed. The components of the CO/NO-cGMP signaling system were grouped into HO isoforms (Figure 2), NOS isoforms (Figure 3), sGC isoforms (Figure 4), PKG isoforms (Figure 5), and PDE isoforms (Figure 6). Each of these groups had at least one member showing increased expression during the period of observation. In fact, 11/13 components had an increased mRNA level measured at one or more time points. There was no consistent temporal pattern for these increases in expression, though the final 48 hour time point was the most common one (8/13 genes) for increased expression to be observed while the 8 hour time point was the point least likely (0/13 genes) to be associated with increased expression. For only one gene was there observed to be a significantly decreased mRNA level at any time point (iNOS), and this was observed only at 8 hours post formalin injection. For no gene did saline injection significantly change expression at any time point analyzed (data not shown).
Figure 2 Alterations in expression of HO isoforms after the hind paw injection of formalin. Panel A provides a time course analysis of changes in levels of HO-1 mRNA up to 48 hours after formalin injection relative to controls. Panel B summarizes measurements for HO-2 mRNA. Data are presented as mean values +/- S.E.M, *p < 0.05, **p < 0.01, ***p < 0.001.
Figure 3 Alterations in expression of NOS isoforms after the hind paw injection of formalin. Panel A provides a time course analysis of changes in levels of nNOS mRNA up to 48 hours after formalin injection relative to controls. Panels B and C summarize measurements for iNOS and eNOS mRNA respectively. Data are presented as mean values +/- S.E.M, *p < 0.05, **p < 0.01, ***p < 0.001.
Figure 4 Alterations in expression of sGC isoforms after the hind paw injection of formalin. Panel A provides a time course analysis of changes in levels of sGCα1 mRNA up to 48 hours after formalin injection relative to controls. Panel B summarizes these measurements for sGCα2 mRNA. Data are presented as mean values +/- S.E.M, *p < 0.05, **p < 0.01, ***p < 0.001.
Figure 5 Alterations in expression of PKG isoforms after the hind paw injection of formalin. Panel A provides a time course analysis of changes in levels of PKG1α mRNA up to 48 hours after formalin injection relative to controls. Panels B and C summarize measurements for PKG 1β and PKG II mRNA respectively. Data are presented as mean values +/- S.E.M, *p < 0.05, **p < 0.01, ***p < 0.001.
Figure 6 Alterations in expression of PDE isoforms after the hind paw injection of formalin. Panel A provides a time course analysis of changes in levels of PDE 2 mRNA up to 48 hours after formalin injection relative to controls. Panels B and C summarize measurements for PDE 3 and PDE 5 mRNA respectively. Data are presented as mean values +/- S.E.M, *p < 0.05, **p < 0.01, ***p < 0.001.
Expression of components of the CO/NO-cGMP signaling system – protein
With the mRNA measurements in hand we turned to the issue of whether the changes in message level corresponded to changes in actual protein content in spinal cord tissue. Table 2 provides a summary of the protein measurements, and Figure 7 shows the appearance of the bands observed on immunoblots. We chose for these studies the 2 hour time point because we desired to measure possible protein level changes in the acute period following nociceptive behaviors. We hypothesized that no protein level changes would be observed this close to the time of formalin injection. The 48 hour time point was chosen because our pervious studies had shown that spinal cord protein accumulation lagged behind mRNA level changes after hindpaw formalin stimulation [13], and our primary interest was in the changes in protein content that might support the persistent mechanical allodynia characteristic of this pain model. Of the 11 genes displaying an increase in mRNA at one or more time points, none displayed a greater protein level 2 hours after formalin injection. However, 8 were observed to have an increased spinal content of protein at 48 hours after formalin injection. For one protein (PDE3) we were not able to optimize our assay sufficiently to allow reliable detection despite the use of multiple antibodies and assay conditions. The 2 genes not having an increased mRNA level at any time point studied also had no increase in protein at 48 hours post formalin injection. The analysis of samples from morphine treated mice revealed that morphine treatment prior to formalin injection eliminated the changes in protein expression seen at 48 hours (Table 2).
Table 2 Summary of measured changes in spinal cord protein levels. The data presented represent the fold changes in spinal cord protein levels +/- SEM relative to control mice 2 and 48 hours after formalin injection. Some mice received 10 mg/kg morphine one time 30 min prior to formalin injection. *p < 0.05, **p < 0.01, ***p < 0.001. NRD, Not readily detected.
Gene Formalin (2 hr) Formalin (48 hr) Formalin (48 hr)+ Morphine
HO-1 1.00 +/- 0.03 1.03 +/- 0.02 1.02 +/- 0.08
HO-2 1.01 +/- 0.08 1.14 +/- 0.02* 0.92 +/- 0.09
NOS-1 1.05 +/- 0.04 2.29 +/- 0.07*** 0.84 +/- 0.13
NOS-2 1.06 +/- 0.06 2.00 +/- 0.11*** 1.00 +/- 0.13
NOS-3 1.01 +/- 0.03 1.16 +/- 0.01** 1.05 +/- 0.07
sGCα1 1.00 +/- 0.04 1.24 +/- 0.06** 1.02 +/- 0.05
sGCα2 0.99 +/- 0.03 0.96 +/- 0.01 1.09 +/- 0.03
PKG Iα 0.99 +/- 0.02 1.00 +/- 0.01 1.04 +/- 0.02
PKG Iβ 1.00 +/- 0.05 1.03 +/- 0.01 0.99 +/- 0.02
PKG II 1.01 +/- 0.05 1.17 +/- 0.01* 0.99 +/- 0.08
PDE 2 0.99 +/- 0.04 1.16 +/- 0.01* 1.06 +/- 0.06
PDE 3 NRD NRD NRD
PDE 5 0.99 +/- 0.05 1.25 +/- 0.02** 1.00 +/- 0.08
Figure 7 Immunoblots for selected CO/NO-cGMP signaling system proteins. Spinal cord protein samples from control animals (left hand lanes) and animals treated 48 hours previously with formalin (right hand lanes) are displayed for the 8 proteins found to display elevated expression at this time point. The molecular weights of the proteins are displayed next to the bands. After processing for CO/NO-cGMP signaling system proteins, blots were stripped and incubated with anti-actin antibodies in order to normalize for protein loading differences. The results of one such actin immunoblot is presented in the lower part of the figure.
Discussion
The spinal cord CO/NO-cGMP system has received much attention, and is thought to regulate both nociceptive and analgesic pathways. The existing published studies are significantly limited, however. These limitations include, 1) the study of single molecules to infer roles for the much more complex enzymatic pathway, 2) the use of multiple species and multiple strains of individual species making the results of separate studies difficult to interpret together, and 3) the arbitrary lack of examination of many of this pathway's enzyme isoforms.
In the present studies we attempted to include the majority of CO/NO-cGMP signaling system components in a time course study following the levels of spinal cord expression of corresponding genes in a commonly used model of inflammatory pain. Overall, at each CO/NO-cGMP system enzymatic level, we identified at least one gene having correspondingly elevated mRNA and protein levels after formalin injection. The principal observations of our studies were, 1) C57Bl/6J mice display not only robust phase I/II licking behaviour after formalin injection, but also a long lasting mechanical allodynia, 2) the majority of the CO/NO-cGMP genes selected for study had correspondingly elevated mRNA levels at some time point within 48 hours of formalin injection, 3) of the genes showing increased mRNA levels at some time point, most of those had correspondingly elevated protein levels as measured in spinal cord homogenates collected 48 hours after formalin injection, and 4) pre-treatment with the analgesic morphine eliminated the acute nociceptive response to formalin injection as well as the delayed changes in spinal cord protein expression, but not chronic nociceptive sensitisation. Figure 8 provides a diagram of the CO/NO-cGMP signaling system highlighting the principal enzymatic steps.
Figure 8 Diagrammatic representation of the CO/NO-cGMP signaling system.
These observations of changes in expression do not by themselves constitute comprehensive evidence of participation in inflammatory pain for each gene. However, our observations do support the hypothesis that within the spinal cord multi-level changes in CO/NO-cGMP system gene expression follow noxious stimulation. Previously provided pharmacological evidence demonstrates the participation of HO [2,3], NOS [14,15], sGC [7] and PKG [8,9] in at least the acute phases of formalin stimulation. PDE inhibitors did not change acute formalin-induced pain behaviours in one study, though those investigations did not look at effects on persistent allodynia or changes in expression [10]. Also, HO-2 null mutants have reduced acute phase licking behaviour and long term allodynia after formalin injection [13], nNOS null mutants have normal formalin-induced licking but lack sensitivity to a NOS inhibitor, and PKG I null mutants have reduced formalin-induced licking behaviour [16]. Previous pharmacological and genetic studies provide valuable evidence concerning the functioning of the various individual components in support of nociceptive signaling. The present data, however, force us to consider the CO/NO-cGMP system as a functional unit with multiple components responding with some degree of coordination.
In the Background section several papers from the existing literature were cited which in various rodent pain models documented alterations in the expression of some of these genes. Though our results are largely consistent with the existing literature, there were differences between the present data and those provided in previous reports using similar paradigms. For example, 2 other groups have reported changes in PKG1α expression after hindpaw formalin injection where we found none [8,9]. Rather, the PKG II isoform seemed to have more robust changes in both mRNA and protein levels in our hands. The use of different species, rats versus mice, provides one possible explanation for the differences in observations. None-the-less, such differences highlight the importance of signaling system components being studied simultaneously if possible to eliminate model, strain and laboratory specific factors from confounding interpretation of the results as may occur when comparing results from the disparate existing literature.
Our morphine data were particularly helpful in understanding the mechanisms governing the control of expression of the CO/NO-cGMP system genes and the consequences of the enhanced expression. It was observed that the injection of a dose of morphine prior to formalin injection providing no more than 2 hours of analgesia in these mice [17] eliminated the wide-spread enhancement of expression of CO/NO-cGMP system genes. Thus it appears both that 1) the enhanced expression of CO/NO-cGMP system genes requires the intense acute nociception characteristic of formalin injection, and 2) that long term hind paw sensitisation does not require enhanced expression of these genes. This is not to say that the enzymatic activity of the existing proteins was not altered. Likewise, abundant pharmacological data indicates that some degree of activity of the CO/NO-cGMP system is required to support nociceptive sensitisation caused by formalin administration. Yet, it is clear that the simple observation of enhanced gene expression occurring in the context of increased pain sensitivity does not prove the enhanced expression is required for that increased sensitivity. In the formalin model the changes in gene expression might be considered indicators of the gene product's involvement in nociceptive signaling, not evidence for the necessity of an increase in gene expression in order for nociceptive sensitisation to occur. Also, though several of the genes studied are most prominently expressed in sensory processing areas of spinal cord dorsal horn tissue, e.g. HO-2 [3], more focused investigations looking at the precise spinal cord areas of altered gene expression may help us to more precisely define the function of those changes. The changes in gene expression we observed might be related to influencing processes other than nociception.
We should consider what new questions our observations raise. One set of issues is: By what mechanism is the response of the CO/NO-cGMP signaling system coordinated? For many years investigators have reported acute changes in spinal cord transcription factor levels after noxious formalin stimulation. Some of the identified factors include c-fos, Fos B, c-jun, jun B, jun D and CREB [18-20]. These could potentially distribute the nociceptive signal to multiple genes sharing particular transcription factor consensus sequences. It need to be recognized, however, that the time courses for mRNA expression increases were not identical for all genes we studies. Thus we would not necessarily expect a single mechanism or transcription factor to govern all the responses measured. In addition to transcription factors and CO/NO-cGMP signaling system components, many other types of molecules including ion channels, nociceptive neurotransmitters, enzymes and other molecules increase in spinal cord abundance after formalin injection [21-25]. We would hypothesize that ongoing investigation into the mechanistic basis for these changes will reveal the involvement of a relatively small group of signaling molecules and transcription factors which influence the expression of a large number of genes via interaction with common promoter region elements in the involved genes. Given the level of our knowledge of the murine genome and the available computational tools, comparison of promoter region sequences for genes responding in a similar manner to noxious stimuli like formalin is becoming feasible.
Conclusion
Noxious inflammatory stimulation causes the increased expression of enzymes participating at each level in the CO/NO-cGMP signalling system. Changes in gene expression corresponding to components of this enzyme system are blocked by the co-administration of morphine though chronic nociceptive sensitisation can still occur. Future studies may be directed at understanding the coordinated CO/NO-cGMP signalling system response is achieved, and whether other pain-related enzyme systems share this type of multi-level response.
Methods
Animal use
All experimental protocols were reviewed and approved by the VAPAHCS institutional animal care and use committee prior to the initiation of work. Male mice 12–14 weeks old of the C57Bl/6J strain were kept under standard conditions with a 12 h light/dark cycle and allowed food and water ad libitum. Mice were obtained from Jackson Laboratories (Bar Harbor, MA) and were kept in our animal facility a minimum of 1 week prior to use in experiments.
Drug administration
5% Formalin solution in 0.9% NaCl was made fresh on the day of experimentation. To induce inflammation, 20 μl of this solution was injected subcutaneously using a 27gauge needle and a microsyringe on the dorsal surface of both hind paws [19]. Control animals received injections of an equal volume of 0.9% NaCl. Animals also received subcutaneous injections of 0.9% NaCl or this vehicle containing 10 mg/kg morphine sulphate (Sigma Chemical, St. Louis, Mo) 30 min prior to formalin injection. Previous experiments demonstrated analgesia to be maximal within 30 minutes of morphine injection [17].
Pain models
The formalin assay was carried out as we have previously described [2,13]. After the administration of formalin (see Drug Administration) the mouse was placed in a clear circular enclosure 25 cm in diameter with a glass floor, and the time spent licking the injected hind paw was measured. Phase I (0–5 min) hind paw licking was tabulated along with phase II licking (10–40 min) by recording total licking times in sequential 5 minute intervals.
Mechanical allodynia was assayed using nylon von Frey filaments according to the "up-down" algorithm described by Chaplan et al. [26] as we have used previously to detect allodynia after formalin injection [13]. In these experiments mice were placed on wire mesh platforms in clear cylindrical plastic enclosures of 10 cm diameter. After 20 minutes of acclimation, fibers of sequentially increasing stiffness (0.2–2 grams, 7 fibers) were applied to the centre of the plantar surface of the right hind paw just distal to the first set of foot pads and left in place 5 sec. Withdrawal of the hind paw from the fiber was scored as a response. When no response was obtained the next stiffest fiber in the series was applied to the same paw; if a response was obtained a less stiff fiber was next applied. Testing proceeded in this manner until 4 fibers had been applied after the first one causing a withdrawal response allowing the estimation of the mechanical withdrawal threshold [27].
Tissue harvest for expression studies
Animals were sacrificed at specific time points by CO2 asphyxiation. Spinal cord lumbar segments (L3-S1) were harvested by extrusion and rapid dissection at the indicated times after formalin injection on a pre-chilled surface. Tissue was then quick frozen in liquid nitrogen and stored at -80°C until use.
Total RNA isolation, reverse transcription and real-time PCR
The isolation of RNA and quantification using real time PCR were performed as described previously for spinal cord samples [12,13,28]. The isolation of total RNA was performed using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to manufacturer's instructions. The purity and concentration was determined spectrophotometrically. Subsequently, cDNA was synthesized from this total RNA using random hexamer priming and a First Strand cDNA Synthesis Kit (Invitrogen, Carlsbad, CA). Briefly, 1 μg of total RNA was mixed with 4 μl of 10 × RT buffer, 8 μl of 25 mM MgCl2, 4 μl 0.1 M DTT, 1 μl RNasin, 2 μl SSII (50 u/μl), 5 μl hexomers and RNase-free water to 40 μl. Incubation was then carried out at 42°C for 60 minutes followed by heat inactivation at 70°C. Finally 1 μl RNase H was added to each reaction and incubated at 37°C for 20 minutes to degrade the RNA.
For real-time PCR, reactions were conducted in a volume of 4 μl using the Sybr Green I master kit (PE applied Biosystems, Foster City, CA). Briefly, 2 μl of a mixture of 2 × sybr green and primers (see Table 1) was loaded with 2 μl diluted cDNA template in each well. 8 μl mineral oil was loaded in each well to prevent loss of solution. PCR parameters were 95°C, 5 min then [95°C, 30 s→ 60°C, 30 s→ 72°C, 60 s] for 40 cycles. Melting curves were performed to document single product formation, and agarose electrophoresis confirmed product size. 18 s RNA was used as an internal control. The 18 s primers were purchased from Ambion (Austin, TX). Amplification kinetics for these products were found to be similar. The data from real-time PCR experiments were analysed by the comparative CT method. For these calculations average Ct values from triplicate PCR reactions for CaMKIIα were normalized to average Ct values for ones from 18 s from the same cDNA preparations. The ratio of comparative expression of each gene between the treated and untreated samples was calculated as 2-(ΔΔCt). Ct represents threshold cycle of PCR amplification. ΔCt represents the difference in threshold cycle between target and reference. ΔΔCt represents the difference between ΔCt (treatedsample) and ΔCt (untreated sample) for same gene.
Table 1 Summary of primer sequences and antibodies used. The sequence of forward (F) and reverse (R) primers used in PCR experiments are provided along with product sizes. Also provided are the sources for antibodies used and the dilution ranges.
Gene Primer Sequences Product Size Antibody Source, Cat# Dilution Range
HO-1 F:ACGCATATACCCGCTACCTG 227 Oncogene, PC340 1:100–1:500
R:GAAGGCGGTCTTAGCCTCTT
HO-2 F:ACTGAAGAAGGTTGCCCAGA 179 Santa Cruz, SC17786 1:100–1:500
R:CTTTATTGGCCTCCTCCACA
nNOS F:TCAGTCTCCCAGGCTAATGG 200 Santa Cruz, SC5302 1:200–1:1000
R:CTGTCCACCTGGATTCCTGT
iNOS F:CTCACTGGGACAGCACAGAA 199 Santa Cruz, SC650 1:200–1:500
R:TGGTCAAACTCTTGGGGTTC
eNOS F:CTCACTGGGACAGCACAGAA 199 Santa Cruz, SC8311 1:200–1:1000
R:TGGTCAAACTCTTGGGGTTC
sGCα1 F:AGCGACTGAACCTTGCACTT 119 Sigma, G4280 1:1000–1:2000
R:ACCTGCTGCAATTGCTTCTT
sGCα2 F:CGAAAGCAACTTCGATGTGA 120 Santa Cruz, SC20954 1:100–1:400
R:AAATGGGGTGGACAATCGTA
PKG Iα F:AAGCATGATGGGAAAACAGG 184 Santa Cruz SC10335 1:100–1:400
R:GTGACTGCTGGCTTGTGGTA
PKG Iβ F:GACAGCTGCATCATCAAGGA 198 Stressgen KAP-PK002 1:500–1:1500
R:GATGGCCCAGAGTTTCACAT
PKG II F:TGAACCGTGACGATGAAAAA 186 Santa Cruz SC25430 1:200–1:500
R:CAAGCTCCACTCTTCCGAAC
PDE 2 F:TTCAAGCTGCTGCAAGAAGA 224 Santa Cruz SC17227 1:200–1:1000
R:TTCCTGAGGACCTGGATACG
PDE 3 F:TTGCATAATTCAATGCCAAG 143 Multiple Multiple
R:TAGGTCCCGATCTTTTGCTG
PDE 5 F:AAATGGTGGGACCTTCACTG 201 Cell Signaling 4072 1:200–1:1000
R:GTGGCCGCTATCTTCTTCAG
Western blot analysis
We performed Western blot analysis for spinal cord samples as we have described previously [12,13,28,29]. Lumbar spinal cord tissue was homogenized in 56.8 mol/l Tris buffer, pH 6.8 with 1.8% (V/V) β-mercaptoethanol, 9.1% glycerol. The homogenate was centrifuged at 13,000 × g for 15 min at 4°C. The supernatant was decanted from the pellet and used for Western blot analyses. The concentration of protein in the homogenate was measured using DC Protein Assay kit (Bio-Rad, Hercules, CA). Equal amounts of protein (50 μg) were size fractionated by SDS-PAGE and electrotransferred onto a polyvinylidene difluoride membrane. The blots were blocked overnight with 5% non-fat dry milk in tris-buffered saline with 0.5% Tween-20 (TBST), incubated with primary antibody on a rocking platform at 4°C for 72 hrs. Primary polyclonal antibodies were obtained from the suppliers listed in Table 1 and used at the indicated dilutions. After washing in TBST, the blot was incubated 2 hrs at room temperature in horseradish peroxidase conjugated anti-rabbit or anti-goat antibody (diluted 1:2,000) (Chemicon, Temecula, CA), washed again, incubated in ECL Plus chemoluminescence reagents and exposed to Kodak XAY-2 film. Bands were quantified using scanning densitometry. Each blot was then stripped and re-probed with anti-β actin antibodies thus allowing normalization of expression between samples.
Sample sizes
Experiments measuring acute and persistent pain related behaviors used 6 mice in each of the control, formalin and formalin plus morphine injected groups. For the quantification of mRNA levels, each time point represents the results from 12 mice with each resulting mRNA sample analyzed in triplicate or quadruplicate in each of at least 2 independent real time PCR experiments. The measurement of protein levels involved the use of 8 mice in each of the control, formalin and formalin plus morphine treated groups with homogenate samples analyzed in triplicate or quadruplicate on at least 2 Western blots.
Statistical analysis
Analysis of repeated parametric measures was accomplished using an ANOVA analysis of variance for repeated measures followed by post-hoc t-testing. For simple comparisons of two means, two-tailed t-testing was performed. A value of p < 0.05 was taken to be significant. All data are presented as means +/- S.E.M. unless otherwise noted.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
XS performed the majority of the expression assays. This author was also responsible for design of the primer pairs used and the optimisation of PCR conditions.
XL performed several of the Western blotting experiments.
JDC was the senior investigator responsible for overall design, coordination and presentation of the experiments.
Acknowledgements
This work was supported by NIH grant GM61260.
==== Refs
Li X Clark JD The role of heme oxygenase in neuropathic and incisional pain Anesth Analg 2000 90 677 682 10702456 10.1097/00000539-200003000-00031
Li X Clark JD Heme oxygenase type 2 plays a role in formalin-induced nociception Pain 2000 86 75 80 10779663 10.1016/S0304-3959(00)00238-4
Li X Clark JD Heme oxygenase inhibitors reduce formalin-induced Fos expression in mouse spinal cord tissue Neuroscience 2001 105 949 956 11530233 10.1016/S0306-4522(01)00207-X
Li X Clark JD Spinal cord heme oxygenase participates in glutamate-induced pain-related behaviors Eur J Pharmacol 2002 450 43 48 12176107 10.1016/S0014-2999(02)02051-4
Wu J Fang L Lin Q Willis WD Nitric oxide synthase in spinal cord central sensitization following intradermal injection of capsaicin Pain 2001 94 47 58 11576744 10.1016/S0304-3959(01)00340-2
Tao F Tao YX Mao P Zhao C Li D Liaw WJ Raja SN Johns RA Intact carrageenan-induced thermal hyperalgesia in mice lacking inducible nitric oxide synthase Neuroscience 2003 120 847 854 12895524 10.1016/S0306-4522(03)00362-2
Tao YX Johns RA Activation and up-regulation of spinal cord nitric oxide receptor, soluble guanylate cyclase, after formalin injection into the rat hind paw Neuroscience 2002 112 439 446 12044461 10.1016/S0306-4522(02)00075-1
Schmidtko A Ruth P Geisslinger G Tegeder I Inhibition of cyclic guanosine 5'-monophosphate-dependent protein kinase I (PKG-I) in lumbar spinal cord reduces formalin-induced hyperalgesia and PKG upregulation Nitric Oxide 2003 8 89 94 12620371 10.1016/S1089-8603(02)00165-9
Tao YX Hassan A Haddad E Johns RA Expression and action of cyclic GMP-dependent protein kinase Ialpha in inflammatory hyperalgesia in rat spinal cord Neuroscience 2000 95 525 533 10658633 10.1016/S0306-4522(99)00438-8
Tegeder I Schmidtko A Niederberger E Ruth P Geisslinger G Dual effects of spinally delivered 8-bromo-cyclic guanosine mono-phosphate (8-bromo-cGMP) in formalin-induced nociception in rats Neurosci Lett 2002 332 146 150 12384231 10.1016/S0304-3940(02)00938-2
Nakamizo T Kawamata J Yoshida K Kawai Y Kanki R Sawada H Kihara T Yamashita H Shibasaki H Akaike A Shimohama S Phosphodiesterase inhibitors are neuroprotective to cultured spinal motor neurons J Neurosci Res 2003 71 485 495 12548704 10.1002/jnr.10483
Liang DY Clark JD Modulation of the NO/CO-cGMP signaling cascade during chronic morphine exposure in mice Neurosci Lett 2004 365 73 77 15234476 10.1016/j.neulet.2004.04.054
Liang DY Li X Clark JD Formalin-induced spinal cord calcium/calmodulin-dependent protein kinase II alpha expression is modulated by heme oxygenase in mice Neurosci Lett 2004 360 61 64 15082179 10.1016/j.neulet.2004.02.050
Yamamoto T Shimoyama N Mizuguchi T Nitric oxide synthase inhibitor blocks spinal sensitization induced by formalin injection into the rat paw Anesth Analg 1993 77 886 890 7692770
Malmberg AB Yaksh TL Spinal nitric oxide synthesis inhibition blocks NMDA-induced thermal hyperalgesia and produces antinociception in the formalin test in rats Pain 1993 54 291 300 8233543 10.1016/0304-3959(93)90028-N
Tegeder I Del Turco D Schmidtko A Sausbier M Feil R Hofmann F Deller T Ruth P Geisslinger G Reduced inflammatory hyperalgesia with preservation of acute thermal nociception in mice lacking cGMP-dependent protein kinase I Proc Natl Acad Sci U S A 2004 101 3253 3257 14973199 10.1073/pnas.0304076101
Davies MF Haimor F Lighthall G Clark JD Dexmedetomidine fails to cause hyperalgesia after cessation of chronic administration Anesth Analg 2003 96 195 200, table of contents 12505952 10.1097/00000539-200301000-00041
Herdegen T Tolle TR Bravo R Zieglgansberger W Zimmermann M Sequential expression of JUN B, JUN D and FOS B proteins in rat spinal neurons: cascade of transcriptional operations during nociception Neurosci Lett 1991 129 221 224 1745403 10.1016/0304-3940(91)90466-7
Li X Lighthall G Liang DY Clark JD Alterations in spinal cord gene expression after hindpaw formalin injection J Neurosci Res 2004 78 533 541 15389827 10.1002/jnr.20274
Anderson LE Seybold VS Phosphorylated cAMP response element binding protein increases in neurokinin-1 receptor-immunoreactive neurons in rat spinal cord in response to formalin-induced nociception Neurosci Lett 2000 283 29 32 10729626 10.1016/S0304-3940(00)00908-3
Noguchi K Morita Y Kiyama H Sato M Ono K Tohyama M Preproenkephalin gene expression in the rat spinal cord after noxious stimuli Brain Res Mol Brain Res 1989 5 227 234 2725196 10.1016/0169-328X(89)90039-9
McCarson KE Enna SJ Nociceptive regulation of GABA(B) receptor gene expression in rat spinal cord Neuropharmacology 1999 38 1767 1773 10587092 10.1016/S0028-3908(99)00121-5
McCarson KE Krause JE NK-1 and NK-3 type tachykinin receptor mRNA expression in the rat spinal cord dorsal horn is increased during adjuvant or formalin-induced nociception J Neurosci 1994 14 712 720 8301359
Gaunitz C Schuttler A Gillen C Allgaier C Formalin-induced changes of NMDA receptor subunit expression in the spinal cord of the rat Amino Acids 2002 23 177 182 12373534 10.1007/s00726-001-0125-3
Sands SA McCarson KE Enna SJ Differential regulation of GABA B receptor subunit expression and function J Pharmacol Exp Ther 2003 305 191 196 12649368 10.1124/jpet.102.046342
Chaplan SR Bach FW Pogrel JW Chung JM Yaksh TL Quantitative assessment of tactile allodynia in the rat paw J Neurosci Methods 1994 53 55 63 7990513 10.1016/0165-0270(94)90144-9
Poree LR Guo TZ Kingery WS Maze M The analgesic potency of dexmedetomidine is enhanced after nerve injury: a possible role for peripheral alpha2-adrenoceptors Anesth Analg 1998 87 941 948 9768799 10.1097/00000539-199810000-00037
Liang D Li X Clark JD Increased expression of Ca2+/calmodulin-dependent protein kinase II alpha during chronic morphine exposure Neuroscience 2004 123 769 775 14706789 10.1016/j.neuroscience.2003.10.007
Li X Clark JD Morphine tolerance and transcription factor expression in mouse spinal cord tissue Neurosci Lett 1999 272 79 82 10507546 10.1016/S0304-3940(99)00559-5
|
16297238
|
PMC1310513
|
CC BY
|
2021-01-04 16:40:08
|
no
|
Mol Pain. 2005 Nov 18; 1:33
|
utf-8
|
Mol Pain
| 2,005 |
10.1186/1744-8069-1-33
|
oa_comm
|
==== Front
Virol JVirology Journal1743-422XBioMed Central London 1743-422X-2-851630067510.1186/1743-422X-2-85ResearchFunctional inaccessibility of quiescent herpes simplex virus genomes Minaker Rebecca L [email protected] Karen L [email protected] James R [email protected] Department of Medical Microbiology & Immunology, University of Alberta, Edmonton, Alberta, T6G 2S2, Canada2 Center for Gene Therapeutics, Department of Pathology and Molecular Medicine, McMaster University, Hamilton, Ontario, L8N 3Z5, Canada 2005 21 11 2005 2 85 85 21 9 2005 21 11 2005 Copyright © 2005 Minaker et al; licensee BioMed Central Ltd.2005Minaker et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Newly delivered herpes simplex virus genomes are subject to repression during the early stages of infection of human fibroblasts. This host defence strategy can limit virus replication and lead to long-term persistence of quiescent viral genomes. The viral immediate-early protein ICP0 acts to negate this negative regulation, thereby facilitating the onset of the viral replication cycle. Although few mechanistic details are available, the host repression machinery has been proposed to assemble the viral genome into a globally inaccessible configuration analogous to heterochromatin, blocking access to most or all trans-acting factors. The strongest evidence for this hypothesis is that ICP0-deficient virus is unable to reactivate quiescent viral genomes, despite its ability to undergo productive infection given a sufficiently high multiplicity of infection. However, recent studies have shown that quiescent infection induces a potent antiviral state, and that ICP0 plays a key role in disarming such host antiviral responses. These findings raise the possibility that cells containing quiescent viral genomes may be refractory to superinfection by ICP0-deficient virus, potentially providing an alternative explanation for the inability of such viruses to trigger reactivation. We therefore asked if ICP0-deficient virus is capable of replicating in cells that contain quiescent viral genomes.
Results
We found that ICP0-deficient herpes simplex virus is able to infect quiescently infected cells, leading to expression and replication of the superinfecting viral genome. Despite this productive infection, the resident quiescent viral genome was neither expressed nor replicated, unless ICP0 was provided in trans.
Conclusion
These data document that quiescent HSV genomes fail to respond to the virally modified host transcriptional apparatus or viral DNA replication machinery provided in trans by productive HSV infection in the absence of ICP0. These results point to global repression as the basis for HSV genome quiescence, and indicate that ICP0 induces reactivation by overcoming this global barrier to the access of trans-acting factors.
==== Body
Background
Herpes simplex virus (HSV) is a significant human pathogen and the prototypical member of the herpesviridae, a large family of enveloped nuclear DNA viruses. HSV displays two modes of interaction with its human host: lytic and latent (reviewed in [1]). Primary infection of epithelial cells produces the lytic response – productive virus replication followed by cell death. Progeny virions then infect adjacent sensory neurons, establishing a life-long latent interaction. Productive infection is characterized by the sequential expression of three classes of viral genes, immediate-early (IE), early (E) and late (L). This regulatory cascade is initiated by VP16, an abundant tegument protein that activates transcription of the IE genes. Four of the IE proteins (ICP0, ICP4, ICP22 and ICP27) then serve to drive further progression into the lytic program. Three of these, ICP4, ICP22 and ICP27, contribute in various ways to the activation of the E and/or L genes [1]. The role of ICP0 appears to be distinct, in that it is also required for efficient IE gene expression [2-4]. Thus, ICP0 mutant viruses display reduced levels of IE gene expression during infection [3-5], and ICP0 activates expression of IE, E and L genes in transient transfection assays [6-10]. Moreover, expression of ICP0 in trans substantially complements the defect of VP16 mutants [11,12], which are otherwise arrested prior to the IE phase following low multiplicity infection. The function of ICP0 therefore seems to lie upstream of those of the other IE gene products in the HSV regulatory cascade.
ICP0 has been described as a promiscuous activator capable of stimulating the expression of a wide range of viral and cellular promoters in transient co-transfection assays (reviewed in [13]). It acts to enhance mRNA accumulation, at least in part by stimulating transcription [14,15]. However it does not bind DNA and there is no evidence that it acts directly on the transcriptional apparatus. Rather, ICP0 appears to stimulate HSV gene expression at least in part by counteracting one or more cellular repression mechanisms that otherwise silence newly delivered viral genomes (reviewed in [16]). This hypothesis emerged from the finding that viral genomes unable to express ICP0 often fail to engage the viral lytic program of gene expression and instead persist for extended periods in the nucleus in an extrachromosomal non-linear configuration without giving rise to appreciable levels of viral gene products [17-21]. Such quiescent genomes however remain potentially functional, as they can be efficiently reactivated by superinfecting the cultures with HSV or human cytomegalovirus (HCMV, another herpesvirus) or by providing ICP0 or HCMV pp71 in trans [17-19,22-24]. The IE promoters residing in quiescent HSV genomes appear to be repressed rather than simply inactive, as they fail to respond to VP16 and several other stimuli that otherwise augment their activity [18]; however, they remain susceptible to activation by ICP0 or pp71 [18,23]. Repression of genomes entering quiescence occurs gradually: newly delivered IE promoters are initially responsive to VP16 and other stimuli and are only later rendered refractory to stimuli other than ICP0 [18]. Perhaps unexpectedly, the otherwise constitutively active HCMV IE promoter is also repressed as recombinant HSV genomes enter quiescence [18-20]. Taken in combination, these data suggest that newly delivered HSV and HCMV IE promoters are targeted by a cellular repression mechanism that is inactivated by ICP0. HSV E and L promoters are also inactive during quiescence; however it is not yet clear if they are actively repressed like the IE promoters or simply inactive due to the absence of the IE proteins.
The mechanisms underlying repression and reactivation of quiescent HSV genomes remain unclear. ICP0 interacts with numerous cellular proteins (reviewed in [25]) including some that could plausibly contribute to gene silencing (for example, type II histone deacetylases [26] and the coREST/REST repressor complex [27]). In addition, ICP0 bears a RING-finger E3 ubiquitin ligase domain [28-30] that is essential for reactivation [24], suggesting that it may act at least in part by targeting key components of the cellular repression machinery for ubiquitination and degradation. Consistent with this view, reactivation is blocked by proteasome inhibitors [24]. However, the crucial target(s) of ICP0 relevant to reactivation have yet to be defined. It may be significant that infecting HSV genomes initially localize to the periphery of nuclear ND10 domains [31-33], and that ICP0 disrupts ND10 [34-36] by targeting several components, including PML, for destruction [37-39]. However, the intranuclear location of quiescent genomes has yet to be determined, and current evidence suggests that transcriptional activity is required for the association of viral genomes with ND10 [33]. Thus, it is not clear what, if any, role ND10 play in quiescence.
A remarkable feature of quiescent HSV genomes is that they fail to detectably respond to superinfection with ICP0-deficient HSV [22,40-42]. The result is striking because ICP0-deficient HSV is itself capable of productively infecting many cell types including those used to establish quiescence, giving rise to infectious progeny. One interpretation of these data is that quiescent HSV genomes are inaccessible to the virally modified transcriptional apparatus and HSV DNA replication machinery provided in trans by the superinfecting virus in the absence of ICP0 [16,41]. If this interpretation is correct, then it follows that: (1) quiescence involves a global restriction in the accessibility of the viral genome to trans-acting factors perhaps akin to that associated with the heterochromatinization of silent host chromosomal loci, and (2) ICP0 induces reactivation by overcoming this generalized barrier to genome activity. However, another hypothesis to explain the inability of ICP0-deficient viruses to induce reactivation is that ICP0 may be required for productive infection of cells harboring quiescent HSV. Under this alternative scenario, ICP0-deficient HSV is effectively excluded from the cells harboring the resident virus, thereby precluding genome reactivation. This "superinfection-immunity" model has not been examined in previous studies; however several considerations suggest that it should be carefully evaluated. First, the severity of the phenotype of ICP0-deficient mutants varies markedly between cell types [43] and during cell cycle progression [44], raising the possibility that such mutants may be unusually sensitive to any perturbations of cellular physiology induced by quiescent HSV infection. Second, the data of Hobbs et al [22] indicate that the replication of ICP0-deficient HSV is severely compromised under the conditions used by those authors in their reactivation assays. Third, HSV virions trigger the induction of a potent antiviral state associated with activation of a subset of IFN-inducible genes in human fibroblasts under conditions where viral gene expression is prevented [45-48], as in quiescence. Moreover, ICP0 serves to block this cellular antiviral response [48], by preventing the activation of IRF3 through unknown mechanisms [49]. Consistent with this particular mechanism of superinfection immunity, ICP0 mutants are hypersensitive to the antiviral effects of type I IFN [50-52] and thus might also be expected to be unusually sensitive to the IFN-independent antiviral state provoked by HSV virions. Fourth, it is possible that quiescent HSV itself gives rise to one or more gene products that interfere with replication of superinfecting ICP0-deficient HSV in a fashion analogous to the repressors produced by temperate bacteriophages.
Considering the foregoing, we examined the susceptibility of human embryonic lung (HEL) fibroblasts harboring quiescent HSV-1 genomes to productive superinfection by ICP0-deficient HSV. We found that such cells are capable of supporting expression and replication of superinfecting ICP0-deficient genomes, given a sufficiently high input multiplicity of infection (MOI). However, the resident quiescent viral genomes were not detectably expressed or replicated in these superinfected cultures. Our results therefore rule out the superinfection-immunity model for the inability of ICP0-deficient HSV to reactivate quiescent HSV, and document that quiescent HSV genomes fail to respond to the virally modified host transcriptional apparatus or viral DNA replication machinery during productive HSV infection in the absence of ICP0. These results point to global repression as the basis for HSV genome quiescence, and indicate that ICP0 induces reactivation by overcoming this global barrier to trans-acting factors.
Results and Discussion
ICP0 is specifically required for reactivation of gene expression from quiescent HSV-1 KM110-R genomes
We first confirmed that ICP0 is required for reactivation in a model of HSV genome quiescence previously developed in our laboratory. The HSV-1 KOS isolate KM110 bears mutations that inactivate the transactivation functions of VP16 and ICP0, severely inhibiting IE gene expression [53]. KM110 fails to enter the lytic cycle following high multiplicity infection of human embryonic lung (HEL) fibroblasts; instead, the infected cell monolayer survives and the KM110 genome persists in a quiescent and reactivation-competent state for at least 10 days [53]. In the present study we used a marked derivative of KM110 (KM110-R) bearing a transgene consisting of red fluorescent protein coding sequences (DsRed2) driven from the human cytomegalovirus IE promoter inserted at the thymidine kinase locus (Methods) in order to facilitate detection of reactivation of KM110 in individual cells. Monolayers of HEL cells were infected with 2 PFU/cell KM110-R to establish quiescence. Four days later the cultures were mock infected or infected with 10 PFU/cell of wild-type HSV-1 KOS or viral mutants bearing lesions in various IE genes. Cells were harvested 24 hours later, then scored for reactivation of the RFP transgene carried by KM110-R by flow cytometry (figure 1). Only 1% of mock-superinfected cells detectably expressed the RFP transgene; in contrast, ca. 28% of cells expressed RFP following superinfection with wild-type HSV-1 KOS. These data indicate that at least a subset of cells in the culture contained reactivation-competent KM110-R genomes and confirm previous reports that expression driven from the HCMV promoter is inhibited during HSV quiescence [18-20]. KM110-R cannot spread to neighboring cells following reactivation with wild-type KOS under the conditions used in this experiment because all cells in the monolayer were productively infected with a high multiplicity of KOS at the outset of the reactivation process therefore excluding superinfecting HSV ([54-56] and data not shown). The data presented in figure 1 therefore indicate that a minimum of ca. 27% of the cells in the monolayer harbored silent but reactivation-competent KM110-R. This value may underestimate the true proportion of quiescently infected cells as dsRed2 folds into the mature fluorescent form quite slowly (CLONTECHniques XVI:3, 2001; Clontech, Palo Alto, Calif.), raising the possibility that some reactivation events may be missed. Reactivation was also observed following super-infection with HSV mutants lacking functional ICP4 (d120), ICP22 (d22-lacZ), and ICP27 (d27-1), confirming that none of these proteins plays an essential role in the reactivation process. We consistently found that the proportion of cells expressing RFP was significantly higher following superinfection with d120 than with any of the other virus isolates tested. Inasmuch as d120 is less effective at excluding super-infecting HSV than any of the other viruses examined (data not shown), it is possible that some or all of this increase stems from spread of reactivated KM110-R to neighboring cells over the course of the reactivation assay. Alternatively, overproduction of ICP0 and other IE proteins during d120 infection may lead to a greater reactivation frequency. In striking contrast to the other viral isolates, the ICP0 mutant n212 failed to detectably induce RFP expression from quiescent KM110-R. These data confirm that ICP0 is required for reactivation of gene expression driven from the HCMV IE promoter located in quiescent HSV-1 genomes, in accordance with previous reports.
Figure 1 ICP0 is required for reactivation of the HCMV IE promoter in quiescent HSV genomes. Confluent monolayers of HEL cells were infected with 2 PFU/cell of KM110-R in order to establish a quiescent infection. Four days later the cells were mock infected or superinfected with wild-type HSV-1 KOS or the indicated IE mutant at an MOI of 10. Samples were harvested 24 hours later and analyzed by flow cytometry. The results are presented as a scatter plot in which the fluorescence in the red and green channels are plotted for each cell analyzed. Values in each panel report the fraction of cells that were scored as positive for RFP expression (indicated as red dots).
We next asked if ICP0 is also required to reactivate expression of the E/L HSV gene encoding VP16. VP16 arising from the KM110-R genome can be readily distinguished from that produced by the superinfecting viruses because KM110-R bears a linker insertion mutation (V422) that truncates VP16 after amino acid residue 422, altering its electrophoretic mobility [53]. To test the requirements for reactivation of VP16 expression, monolayers containing or lacking quiescent KM110-R (input MOI of 6) were superinfected with the same panel of HSV-1 isolates as before, then analyzed by Western blot using a VP16 monoclonal antibody (figure 2A). As expected on the basis of previous work [53], wild-type HSV-1 KOS efficiently reactivated VP16 expression from the resident KM110-R genome, as did d120 (ICP4-), d22-lacZ (ICP22-) and d27-1 (ICP27-). In contrast, the ICP0-deficient mutant n212 failed to detectably reactivate VP16 expression. The phenotype displayed by n212 was distinct from that exhibited by the other IE mutants in that the VP16 gene residing in the superinfecting n212 genome was efficiently expressed while the corresponding gene of KM110-R remained silent. By contrast, both versions of VP16 were efficiently expressed following superinfection with all of the other viruses, including the ICP4-deficient mutant d120, despite the fact that ICP4 is stringently required for expression of VP16 and other HSV E and L genes [57,58]. Presumably, the requisite ICP4 is provided in trans by the reactivated KM110-R. Indeed, as expected, d120 failed to express VP16 following infection of control HEL cells lacking KM110-R. Taken in combination, these data demonstrate that ICP0 is required for reactivation of VP16 expression from the quiescent genome.
Figure 2 ICP0 is required for reactivation of VP16 gene expression and viral DNA replication. Confluent monolayers of HEL cells were infected with 6 PFU/cell of KM110-R to establish quiescence. Four days later the cells were mock infected or superinfected with the indicated HSV strains (MOI of 10). Samples harvested 18 hours later were then analyzed for VP16 expression by Western Blot (panel A) or viral DNA replication by Southern blot (panel B). (A) Samples were scored for VP16 and cellular β-actin by Western blot. (B) Total cellular DNA was cleaved with Bam HI and Nhe I, then analyzed by Southern blot hybridization using an HSV-1 VP16 probe. U2OS cells: samples extracted from permissive U2OS 24 hours after infection infected with KM110-R (MOI of 10). wt: wild-type VP16 protein or gene; mt: mutant VP16 protein or gene.
ICP0 is required for reactivation of viral DNA replication
Previous work has implied that ICP0 is required for replication of the resident viral genome following superinfection of cells harboring quiescent HSV [40,41]. To determine if this is the case in our system, we used Southern blot hybridization to monitor replication of the resident KM110-R genome following superinfection with wild-type and mutant virus (figure 2B). The genome of KM110-R can be readily distinguished from that of wild-type HSV-1 because it bears an NheI linker at the VP16 locus that marks the V422 mutation [59]. As a result, the 8.1 kb BamHI fragment that spans the VP16 locus is cleaved by NheI in KM110-R, yielding fragments of 4.9 and 3.2 kb (figure 2B). Quiescent KM110-R genomes were not detectable prior to reactivation under the conditions used in our Southern blot assay; however the expected KM110-R signal was readily detected following genome amplification induced by super-infection with wild-type KOS (figure 2B). Mutants lacking ICP4, ICP22, and ICP27 (d120, d22-lacZ and d27-1 respectively) triggered replication of the KM110-R genome as effectively as wild-type KOS; in contrast, no amplified KM110-R signal was observed following infection with the ICP0-deficient mutant n212. As expected [1], the ICP4 and ICP27 null mutants each displayed a severe DNA replication defect in cells lacking KM1110-R. These defects were however complemented in cells harboring KM110-R, presumably due to provision of the missing gene products in trans from the reactivated KM110-R genome.
Taken in combination, the data presented above clearly document that ICP0 is essential for the reactivation of expression from the HCMV IE and HSV VP16 promoters and replication of quiescent HSV-1 genomes in superinfected cultures, confirming and extending the results of previous studies.
ICP0-deficient HSV-1 is able to infect cells that harbor quiescent KM110-R
We next sought to determine if ICP0-deficient HSV is able to productively infect cells that contain quiescent KM110-R. Previous work has documented that the magnitude of the defect exhibited by ICP0-deficient HSV varies with cell type and is particularly pronounced on HEL cells [43]. Consistent with these findings, preliminary experiments indicated that only ca. 45% of HEL cells proceeded to the stage of viral DNA replication following infection with 10 PFU/cell n212, compared to >95% following infection with 10 PFU/cell of KOS (data not shown). However, this proportion could be increased to ca. 90% when the MOI of n212 was raised to 30 PFU/cell (figure 6 and additional data not shown). Therefore, in order to maximize the proportion of cells productively infected by ICP0-deficient HSV, all of the remaining reactivation experiments described in this report employed an MOI of 30 for n212 and 10 for wild-type KOS. Importantly, n212 was unable to reactivate quiescent KM110-R in any of our assays following infection at this higher MOI (see below).
Figure 6 Efficiency of replication compartment formation. Cells containing or lacking quiescent KM110-R were superinfected with n212 or KOS as described in the legend to figure 5, then examined for viral replication compartments by visualizing the intranuclear distribution of ICP4 (figure 5). Cells exhibiting large ICP4 structures that filled an appreciable fraction of the nucleus were scored as positive while cells displaying diffuse or small punctate ICP4 structures were scored as negative. 100–650 cells were scored for each treatment group in each experiment. The data presented are the average of three independent experiments. Bars represent the standard deviation.
To determine if ICP0-deficient HSV is able to initiate gene expression in HEL cells containing quiescent virus, cultures were superinfected with a marked n212 derivative (n212-G) bearing an eGFP transgene driven from the HCMV IE promoter inserted at the thymidine kinase locus; an analogous derivative of KOS (KOS-G) served as a control. Replicate monolayers harboring quiescent KM110-R (input MOI, 6 PFU/cell) were superinfected with these indicator viruses on day four, then analyzed for transgene expression by flow cytometery 18 hours later (figure 3). For technical reasons RFP expression from reactivated KM110-R cannot be reliably assessed by flow cytometry in HEL cells that also express GFP (see Methods). Therefore, the proportion of cells harboring reactivation-competent KM110-R was estimated by superinfecting parallel cultures with unmarked KOS and n212 (MOIs of 10 and 30 respectively). As before, mock-superinfected cultures displayed a relatively low proportion of cells expressing RFP (average of 6.5% in four experiments using an MOI of KM110-R of 6), while superinfection with wild-type HSV-1 KOS increased this value to ca. 45% (figure 3). Thus, at minimum, ca. 40% of the cells harbored quiescent but reactivation-competent KM110-R at the time of superinfection. Consistent with previous experiments, n212 at an MOI of 30 did not increase the proportion of RFP-positive cells beyond that observed in mock-infected cultures (figure 3), or reactivate VP16 expression from KM110-R (figure 4). However, n212-G was able to infect >95% of the cells in the cultures as judged by eGFP expression, a value similar to that obtained with KOS-G (figure 3). These data therefore indicate that ICP0-deficient HSV-1 is able to initiate viral gene expression in essentially every cell that harbors quiescent KM110-R.
Figure 3 ICP0-deficient HSV is able to initiate infection in cells harbouring quiescent KM110-R. Confluent monolayers of HEL cells were infected with 6 PFU/cell KM110-R. 4 days later the cells were mock treated or superinfected with n212 or n212-G (MOI 30), or KOS or KOS-G (MOI 10). Samples were harvested 18 hours later and analyzed by Flow cytometry. The intensity of red fluorescence is shown on the y-axis, and green fluorescence on the x-axis. The red and green dots indicate cells expressing RFP and GFP. The purple dots indicate cells that clearly expressed both proteins (however note that GFP expression interferes with the detection of RFP in most cells, see Methods). The proportion of cells scored as positive for expression of RFP (mock, n212, KOS) or GFP (KOS-G, n212-G) are indicated; values represent the average of four independent experiments. Standard deviations were mock: 4%, n212: 3%, KOS: 4%, n212-G: 2%, KOS-G: 0%.
Figure 4 ICP0-deficient HSV fails to reactivate VP16 expression or viral DNA replication following high MOI infection. Confluent monolayers of HEL cells were mock infected or infected with 6 PFU/cell of KM110-R to establish quiescence. Four days later the cells were mock infected or superinfected with 30 PFU/cell of n212 or 10 PFU/cell of KOS. Samples harvested 18 hours later were then analyzed for VP16 expression by Western Blot (panel A) or viral DNA replication by Southern blot (panel B). (A) Samples were scored for VP16 and cellular β-actin by Western blot. (B) Total cellular DNA was cleaved with Bam HI and Nhe I, then analyzed by Southern blot hybridization using an HSV-1 VP16 probe. Lane U2OS cells: samples extracted from U2OS cells 24 hours after infection with 10 PFU/cell KM110-R; Lanes KOS 1 hr and n212 1 hr: samples harvested from HEL cells one hour postinfection with KOS or n212 (MOIs of 10 and 30 respectively), documenting that the input virus does not interfere with detection of newly synthesized VP16 or viral DNA. wt: wild-type; mt: mutant
Previous studies have shown that ICP0-deficient HSV-1 often stalls at varied points in the viral gene expression program subsequent to the immediate-early phase [3,43]. Therefore, expression of eGFP from the HCMV IE promoter does not necessarily imply productive infection with n212-G. As one measure of the ability of ICP0-deficient HSV to progress to later stages of infection, we scored the superinfected cells for the presence of viral DNA replication compartments in parallel experiments. Quiescently infected HEL cells grown on coverslips were mock-treated or superinfected with KOS or n212 (MOIs of 10 and 30 respectively) in the presence or absence of 400 μg/mL phosphonoacetic acid (PAA) to block viral DNA replication; replication compartments were then visualized 9.5 hours later by examining the intranuclear distribution of the immediate-early protein ICP4 by indirect immunofluorescence (figure 5). Previous work has shown that ICP4 is initially recruited to small nuclear foci termed pre-replicative sites at early times post-infection; pre-replicative sites then develop into much larger ICP4-positive DNA replication compartments that fill much of the nucleus at late times post-infection in a process requiring viral DNA replication [60]. As expected, only a very small fraction (2%) of cells in KM110-R infected monolayers expressed ICP4 in the absence of superinfection. The ICP4 staining in these "background" positive cells illuminated large replication compartments that filled most of the nuclear volume (figure 5). Presumably, this signal marks cells that are undergoing productive infection by KM110-R. In contrast, the great majority of cells (>90%) in the KM110-R infected cultures expressed ICP4 following superinfection with KOS or n212, irrespective of the presence or absence of PAA (see figure 5 for data obtained with n212). These results document that both KOS and n212 are able to initiate HSV IE gene expression in the majority of cells in the quiescently infected cultures. As expected, the ICP4 staining pattern in superinfected cells was highly dependent on the presence or absence of PAA (figure 5). The signal in the presence of PAA was diffuse with many small discrete foci of staining in some cells; in contrast, large ICP4-positive structures (replication compartments) that filled much of the nucleus were observed in most cells maintained in the absence of PAA (figure 5). The proportion of cells displaying replication compartments was quantified in three experiments and the data obtained are summarized in figure 6. Only 2% of the cells in cultures harboring quiescent KM110-R displayed viral DNA replication compartments in the absence of superinfection (991 cells scored in total over three experiments). In contrast, replication compartments formed in 97% +/- 2% of these cells following superinfection with KOS (866 cells scored), a value that did not differ from that observed following KOS superinfection of cultures lacking quiescent KM110-R (97% +/- 2%, 1069 cells scored). At the MOI of 30 used in these experiments, n212 formed replication compartments in 92% +/- 2% and 88% +/- 2% of cells in mock-infected and KM110-R infected cultures (samples sizes of 1578 and 1661 cells respectively). Inasmuch as a minimum of ca. 40% of the cells in the KM110-infected cultures harbored silent but reactivation-competent KM110-R at the time of superinfection (figure 3), these data demonstrate that n212 is able to form DNA replication compartments in the majority of cells that contain quiescent KM110-R. However, n212 (30 PFU/cell) did not detectably reactivate replication of the quiescent KM110-R genome in parallel cultures (figure 4).
Figure 5 ICP0-deficient HSV is able to form replication compartments in cells harbouring quiescent KM110-R. Confluent monolayers of HEL growing on coverslips were mock infected (not shown) or infected with 6 PFU/cell of KM110-R to establish quiescence. Four days later the cells were either mock infected or superinfected with 30 PFU/cell of n212 or 10 PFU/cell KOS (not shown) in the presence or absence of 400 μg/mL PAA. 9.5 hours later the cells were fixed and processed for visualization of ICP4 by indirect immunofluorescence. Nuclei were counter-stained with Hoescht 33342. Representative fields of cells harbouring KM110-R are shown following mock-infection or infection with n212 in the presence and absence of PAA.
As another measure of the ability of the KM110-R genome to replicate following superinfection, we asked if KM110-R was recovered in the progeny virus thus produced. To this end, progeny virus harvested 18 hours after superinfection was subjected to plaque assay in permissive U2OS cells, and the titres of RFP-positive (KM110-R) and RFP-negative virus were determined (table 1). KOS and n212 gave rise to approximately equivalent numbers of infectious progeny, and the yields of both viruses were reduced by approximately 50% on cultures harboring KM110-R relative to mock-infected HEL cells. Thus replication of n212 was not greatly impaired relative to wild-type HSV under the conditions of this experiment. Approximately 25% of the progeny recovered following KOS superinfection expressed the RFP marker characteristic of KM110-R. In contrast only ca. 0.06% of the progeny of the n212 infection bore the RFP marker, a reduction of ca. 3 orders of magnitude relative to wild-type HSV. These data document that ICP0 is required for efficient recovery of a genetic marker carried by the quiescent genome into progeny virus. Similar results have been reported previously [22,40,41], however it was not clear from the data presented in those reports if the ICP0-deficient superinfecting virus was competent to replicate in those cells that harbored quiescent HSV.
Table 1 Viral progeny recovered from superinfected cells. HEL cells containing or lacking quiescent KM110-R (MOI 6) were superinfected on day 4 with either KOS or n212 (MOIs of 10 and 30 respectively). Progeny virus harvested 18 hours later was then titrated on U2OS cells in the presence of HMBA (Methods).
Superinfecting virus KM110-R RFP- titre (PFU/mL) RFP+ titre (PFU/mL)
none - 0 N/A
+ 0 1.25 × 103 ± 1.5 × 103
n212 - 3.66 × 107 ± 1.3 × 107 N/A
+ 1.79 × 107 ± 8.6 × 106 1.00 × 104 ± 6.8 × 103
KOS - 5.52 × 107 ± 2.2 × 107 N/A
+ 2.63 × 107 ± 3.9 × 106 8.30 × 106 ± 7.4 × 105
Summary and implications
Our results document that ICP0-deficient HSV is capable of productively infecting cells that harbor quiescent HSV genomes: given a sufficiently high multiplicity of infection the superinfecting virus initiates gene expression and progresses to at least the stage of viral DNA replication in the majority of such cells. Remarkably, this productive infection does not provoke reactivation of the resident viral genomes. These data exclude the superinfection-immunity model for the failure of ICP0-deficient HSV to trigger reactivation and provide strong support for the suggestion that quiescent HSV genomes are functionally inaccessible to the modified transcription apparatus and viral DNA replication factors provided by the superinfecting virus [16,41]. As pointed out by Preston [16], the refractory state of quiescent HSV genomes appears to be distinct from that adopted by the viral genome during latent infection of sensory neurons, as latent HSV genomes can be reactivated in response to external signals or by expression of any of HSV VP16, ICP4 or ICP0 [12]; in contrast, the only known means of reactivating quiescent genomes is via expression of ICP0 or its HCMV functional counterpart pp71. The implication is that quiescent genomes are more effectively shielded from trans-acting factors than latent genomes.
The mechanisms that prevent quiescent HSV genomes from responding to trans-acting factors are of great interest, as is the mode of action of ICP0 in overcoming this barrier to gene expression and DNA replication. Sequence-specific repression seems unlikely, for two reasons. First, the results of this and previous [18,41] reports indicate that genes driven from at least three distinct categories of viral promoters (HCMV IE, HSV IE, and HSV VP16) remain silent in cells superinfected with ICP0-deficient HSV, despite the activity of the corresponding genes located in the superinfecting viral genome. Similarly, the quiescent genome fails to respond to the viral DNA replication and recombination machinery provided by the superinfecting virus. These data suggest that the inhibitory mechanism renders many (if not all) of the cis-acting elements (eg. promoters and origins of DNA replication) located in the quiescent genome non-operative. Second, the quiescent genome is not activated by replication of the superinfecting viral genome within the same nucleus, a condition that would likely titrate classical sequence-specific DNA-binding repressors. These data suggest that quiescent genomes may be stably associated with repressive material that does not readily equilibrate between viral genomes, or located at one or more inaccessible intranuclear sites.
The functional inaccessibility of quiescent HSV genomes documented here is reminiscent of that displayed by genes located in cellular heterochromatin [61]; however it is worth emphasizing that previous work has shown that quiescent HSV genomes lack regularly spaced nucleosomes at the tk locus [20], a feature that distinguishes them both from classical heterochromatin and the latent HSV genomes present in sensory neurons [62]. Moreover, HSV infection (and ICP0) does not activate the heterochromatinized endogenous cellular β-globin gene in present fibroblasts, although transfected (and presumably euchromatic) copies of this gene are susceptible to activation by HSV infection [63,64]. These considerations raise the possibility that HSV genome quiescence involves novel mechanisms, perhaps related to those that inhibit HSV transcription in response to type I IFN [51,65]. Indeed, ICP0 is able to overcome the IFN-induced barrier to HSV transcription [51], in addition to triggering reactivation of quiescent HSV genomes. It therefore seems likely that further studies of the mode of action of ICP0 may illuminate one or more intranuclear mechanisms of antiviral defense.
Conclusion
Our results provide strong support for the hypothesis that quiescent HSV genomes are silenced by a cellular mechanism that renders them globally inaccessible to most trans-acting factors. The implication is that ICP0 triggers reactivation from quiescence by overcoming this generalized barrier to gene expression and DNA replication. Further studies designed to identify the components of this repression mechanism will clarify how the balance between host intranuclear repression mechanisms and viral countermeasures regulates the onset of the HSV lytic program of gene expression.
Methods
Cells and Virus
Human U2OS osteosarcoma cells, Human Embryonic Lung (HEL) fibroblasts and African green monkey kidney (Vero) cells were obtained from the American Type Culture Collection. E5 [66] and V27 [67] cells were gifts from N. A. DeLuca and S. Rice respectively. Cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) (Gibco) supplemented with 10% (U2OS and HEL) or 5% (Vero) fetal bovine serum (FBS), 50 units/ml penicillin (P) and 5 μg/ml streptomycin (S). E5 and V27 cells were additionally supplemented with 100 μg/ml G418 (Geneticin®, GIBCO).
KOS 1.1 (a wild-type strain of HSV-1), KOS-G (see below) and d22lacZ ([68] ICP22-) were grown and titered on Vero cells. n212 ([6] ICP0-), n212-G, KM110 ([53] VP16/ICP0- double mutant) and KM110-R were grown and titered on U20S cells (in the presence of 3 mM HMBA for KM110). d120 ([58] ICP4-) and d27-1 ([67] ICP27-) were grown and titered on complementing E5 and V27 cells respectively.
In experiments where the progeny of superinfected cultures were examined for recovery of the dsRED gene (table 1), the superinfected cells were treated with an acid glycine wash to remove any input superinfecting virus that had not penetrated the host cells, as follows. 2 hrs post-superinfection, the growth medium from monolayers grown in 12 well plates was aspirated. The cells were then incubated with 1 ml Acid Glycine wash (8 g/L NaCl, 1.8 g/L KCl, 0.1 g/L MgCl2·6H2O, 0.1 g/L CaCl2·6H2O, 7.5 g/L glycine, pH 3) for 30 seconds. After two washes with 1 ml Phosphate Buffered Saline (PBS: 10 mg/ml NaCl, 0.25 mg/ml KCl, 1.8 mg/ml Na2HPO4, 0.3 mg/ml KH2PO4, pH 7.5), regular growth medium was added.
Construction of recombinant viruses
We modified KOS1.1, n212, and KM110 by inserting transgenes encoding eGFP (KOS-G, n212-G) or dsRed2 (KM110-R) driven from the human cytomegalovirus immediate-early promoter into the viral thymidine kinase (tk) locus in the tk sense orientation. To this end, 1.6 kbp Ase I-Mlu I fragments bearing the HCMV promoter, DsRed2 or eGFP coding sequence, and SV40 early polyadenylation signal were excised from pDsRed2-C1(Clontech) or pEGFP-C1 (Clontech) and inserted into SstI site in the tk coding sequences carried by pTK173 after making all ends blunt, generating pTK-Red and pTK-Green. The resulting tk-deficient insertion mutations were then transferred into the intact viral genomes of KOS1.1, n212, and KM110 via DNA-mediated marker rescue using standard methods. Briefly, 350 ng of pTK-Red or pTK-Green (cleaved with Afl III) was combined with 1–2 μg of total cellular DNA extracted from cells infected with the target virus, and the resulting mixture was transfected into U2OS cells using Fugene (Roche). Recombinants were then isolated from the progeny of the co-transfection by picking red or green fluorescent plaques. After several rounds of plaque purification the identity and purity of the recombinants was confirmed by Southern blot analysis of the viral tk, VP16, and ICP0 loci.
Western blot
Samples were subject to electrophoresis through 12% SDS polyacrylamide gels along with 10 μl pre-stained molecular weight standards, Low Range (BIO-RAD), then transferred to a nitrocellulose membrane (Hybond ECL, Ambersham Pharmacia) using a wet protein transfer apparatus (Bio-Rad Trans-blot cell). Following the transfer, the membrane was incubated in 10% skim milk TBS-Tween (25 mM Tris, pH 8, 150 mM NaCl, 0.1% Tween-20) overnight at 4°C. Monoclonal antibodies to VP16 (LP1, [69] a generous gift from A. Minson) and β-actin (Sigma Aldrich) were used at dilutions of 1:16,000 and 1:5,000 respectively. The membrane was incubated with the primary antibody diluted in TBS-Tween/5% skim milk for 30 min at room temperature then washed three times for 10 min in TBS-Tween. The membrane was then incubated with secondary antibody, goat anti-mouse IgG-HRP (BioRad) diluted 1:3,000 in TBS-Tween/5% skim milk, for 30 min at room temperature. After washing three times as before, the membrane was developed using ECL+plus system (Amersham Biosciences) according to the manufacture's instructions. The signal was detected by exposure to Fuji Super RX X-Ray film.
Southern blot
Total cellular DNA extracted as previously described was cleaved with a mixture of Bam HI and Nhe I, then subjected to electrophoresis through a 1% agarose gel in Tris-acetate EDTA (TAE) for 2 hrs at 80 V in TAE buffer. The gel was then stained with SYBR Gold Nucleic Acid Gel Stain (Molecular Probes) according the manufacturer's instructions and quantified by phospho-imager analysis on a Storm 860 (Molecular Dynamics). The gel was washed sequentially in the following solutions for 15 min each: 0.25 M HCl, 0.5 M NaOH, 1 M Tris/1.5 M NaCl, and 10 × SSC. DNA was transferred to a GeneScreen Plus nylon membrane (NEN Life Sciences Products) in 10 × SSC. The membrane was UV-cross linked using Stratalinker 2400 (Stratagene) before being hybridized to a 32P-labelled 1537 nt probe VP16 probe generated by random priming. The probe fragment was obtained by polymerase chain reaction using pVP16 KOS [70] as the template and the primers 5' CGCCGTCGGGCGTCCCACAC 3' and 5' CGGGGGATGCGGATCCGGTCGCGC 3'. The 32P signal was detected by exposure to Kodak BioMax MS film at -80°C.
Flow cytometry
Cells were detached from the growth surface with trypsin, resuspended in DMEM and transferred to a 5 ml Falcon tube. Red and green fluorescence was quantified by passing the cells through a Becton Dickson FACScan and analyzed using CellQuest Software. HEL cells exhibit substantial levels of autofluorescence, potentially interfering with the analysis. However, we found that the intensities of the red and green autofluorerescent signals emitted by individual HEL cells are highly correlated (see for example figure 1) such that cells expressing neither dsRED2 nor eGFP fall on the diagonal of plots of green versus red signal intensity. This correlation allows cells expressing even low levels of dsRED2 to be readily detected as signals above the control diagonal (shown as the red dots figure 1). Note that this procedure uses the green autofluorescent signal emmited by each cell to estimate its autofluorence in the read channel. However, this procedure cannot be used if the cells also express eGFP (see figure 3), because the green autofluorescence is masked by the eGFP fluorescence. Hence, the only a minority of the RFP+ cells can be detected when the cells also express GFP (indicated by the purple dots in figure 3).
Detection of viral DNA replication compartments via indirect immunofluorescence of ICP4
Monolayers of HEL cells grown on 18 mm coverslips (Fisher Scientific) in a 12 well plate were fixed by washing twice with 1 ml PBS and incubating in 400 μl PBS containing 5% formaldehyde and 2% sucrose for 10 min. This and subsequent manipulations were at room temperature. The cells were then permeabilized by washing twice with 1 ml PBS and incubating in 400 μl PBS containing 0.6% NonidetP-40 and 10% sucrose for 10 min. After washing twice more with 1 ml PBS/1% FBS, the cells were incubated with 100 μl primary anti-ICP4 monoclonal antibody (#1114, Goodwin Institute) diluted 1:1000 in PBS/1%FBS for 1 hr, and washed six times with PBS/1% FBS over 15 min. The cells were then incubated in 100 μl Alexa Fluor® 488 labeled goat anti-mouse IgG (Molecular Probes) diluted 1:1000 in PBS/1%FBS for 1 hr and washed six times as before. The cell nuclei were stained by incubating in 100 μl of 500 ng/ml Hoescht 33342 (Molecular Probes) in PBS solution for 10 min, protected from the light. After washing three times in PBS/1% FBS, the coverslips were dipped in H2O, and allowed to dry for 15 min, protected from the light. The coverslips were mounted on slides using 20 μl Vectashield mounting medium, and secured with clear nail polish. Slides were examined using a Zeiss LSM 510, 2 photon Laser Scanning Microscope system with two lasers giving excitation lines at 488 nm (for Alexa Fluor 488) and 780 nm (for Hoescht stain), and using a 40× oil immersion objective lens.
List of abbreviations
IFN: interferon
ICP: infected cell protein
eGFP: enhanced green fluorescent protein
HCMV: human cytomegalovirus
HEL: human embryonic lung fibroblasts
HMBA: hexamethylene bis-acetamide
HSV: herpes simplex virus
MOI: multiplicity of infection
mt: mutant
ND10: nuclear domain 10
PAA: phosphonoacteic acid
RFP: red fluorescent protein
VP16: viral protein 16
wt: wild-type
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RLM conducted all of the experiments reported in this manuscript. KLM conducted preliminary experiments that lead to the initiation of this work. JRS conceived of the study, and JRS and RLM wrote the manuscript.
Acknowledgements
We thank Holly Saffran and Rob Maranchuk for technical support, and Jennifer Corcoran for comments on the manuscript. This work was supported by an operating grant from the Canadian Institutes for Health Research. JRS holds a Canada Research Chair in Molecular Virology.
==== Refs
Roizman B Knipe DM Knipe DM, Howley PH Herpes simplex viruses and their replication. Fields Virology 2001 2 4th Philadelphia , Lippincott Williams and Wilkins 2399 2459
Cai W Astor TL Lipak LM Cho C Coen DM Schaffer PA The herpes simplex virus type 1 regulatory protein ICP0 enhances virus replication during acute infection and reactivation from latency J Virol 1993 67 7501 7512 8230470
Cai W Schaffer PA Herpes simplex virus type 1 ICP0 regulates expression of immediate-early, early, and late genes in productively infected cells J Virol 1992 66 2904 2915 1313909
Stow ND Stow EC Isolation and characterization of a herpes simplex virus type 1 mutant containing a deletion within the gene encoding the immediate early polypeptide Vmw110 J Gen Virol 1986 67 2571 2585 3025339
Everett RD Construction and characterization of herpes simplex virus type 1 mutants with defined lesions in immediate-early gene 1 J Gen Virol 1989 70 1185 1202 2543774
Cai W Schaffer PA Herpes simplex virus type 1 ICPO plays a critical role in the de novo synthesis of infectious virus following transfection of viral DNA J Virol 1989 63 4579 4589 2552142
Everett RD Transactivation of transcription by herpes virus products: requirement for two HSV-1 immediate-early polypeptides for maximum activity. EMBO J 1984 3 3135 3141 6098466
Gelman IH Silverstein S Identification of immediate early genes from herpes simplex virus that transactivates the virus thymidine kinase gene Proc Natl Sci USA 1985 82 5265 5269 2991915
O'Hare P Hayward GS Three trans-acting regulatory proteins of herpes simplex virus modulate immediate-early gene expression in a pathway involving positive and negative feedback regulation J Virol 1985 56 723 733 2999428
Quinlan MP Knipe DM Stimulation of expression of a herpes simplex virus DNA-binding protein by two viral functions Mol Cell Biol 1985 5 957 963 2987684
Ace CI McKee TA Ryan JM Cameron JM Preston CM Construction and characterization of a herpes simplex virus type 1 mutant unable to transinduce immediate-early gene expression Journal of Virology 1989 63 2260 2269 2539517
Halford WP Kemp CD Isler JA Davido DJ Schaffer PA ICP0, ICP4, or VP16 expressed from adenovirus vectors induces reactivation of latent herpes simplex virus type 1 in primary cultures of latently infected trigeminal ganglion cells J Virol 2001 75 6143 6153 11390616 10.1128/JVI.75.13.6143-6153.2001
Everett RD ICP0, a regulator of herpes simplex virus during lytic and latent infection Bioessays 2000 22 761 770 10918307 10.1002/1521-1878(200008)22:8<761::AID-BIES10>3.0.CO;2-A
Jordan R Schaffer P Activation of gene expression by herpes simplex virus type 1 ICP0 occurs at the level of mRNA synthesis J Virol 1997 71 6850 6862 9261410
Samaniego LA Wu N DeLuca NA The herpes simplex virus immediate-early protein ICP0 affects transcription from the viral genome and infected-cell survival in the absence of ICP4 and ICP27 J Virol 1997 71 4614 4625 9151855
Preston CM Repression of viral transcription during herpes simplex virus latency J Gen Virol 2000 81 1 19 10640537
Stow EC Stow ND Complementation of a herpes simplex virus type 1 Vmw110 deletion mutant by human cytomegalovirus. J Gen Virol 1989 70 695 704 2543754
Preston CM Nicholl MJ Repression of gene expression upon infection of cells with herpes simplex virus type 1 mutants impaired for immediate-early protein synthesis J Virol 1997 71 7807 7813 9311867
Samaniego LA Neiderhiser L DeLuca NA Persistence and expression of the herpes simplex virus genome in the absence of immediate-early proteins J Virol 1998 72 3307 3320 9525658
Jamieson DR Robinson LH Daksis JI M.J. N Preston CM Quiescent viral genomes in human fibroblasts after infection with herpes simplex virus type 1 Vmw65 mutants. J Gen Virol 1995 76 1417 1431 7782770
Jackson SA DeLuca NA Relationship of herpes simplex virus genome configuration to productive and persistent infections. Proc Natl Acad Sci U S A 2003 100 7871 7876 12796511 10.1073/pnas.1230643100
Hobbs WE Brough DE Kovesdi I DeLuca NA Efficient activation of viral genomes by levels of herpes simplex virus ICP0 insufficient to affect cellular gene expression or cell survival J Virol 2001 75 3391 3403 11238865 10.1128/JVI.75.7.3391-3403.2001
Preston CM Nicholl MJ Human cytomegalovirus tegument protein pp71 directs long-term gene expression from quiescent herpes simplex virus genomes J Virol 2005 79 525 535 15596845 10.1128/JVI.79.1.525-535.2005
Everett RD Orr A Preston CM A viral activator of gene expression functions via the ubiquitin- proteasome pathway Embo J 1998 17 7161 7169 9857173 10.1093/emboj/17.24.7161
Hagglund R Roizman B Role of ICP0 in the strategy of conquest of the host cell by herpes simplex virus 1 J Virol 2004 78 2169 2178 14963113 10.1128/JVI.78.5.2169-2178.2004
Lomonte P Thomas J Texier P Caron C Khochbin S Epstein AL Functional interaction between class II histone deacetylases and ICP0 of herpes simplex virus type 1 J Virol 2004 78 6744 6757 15194749 10.1128/JVI.78.13.6744-6757.2004
Gu H Liang Y Mandel G Roizman B Components of the REST/CoREST/histone deacetylase repressor complex are disrupted, modified, and translocated in HSV-1-infected cells Proc Natl Acad Sci U S A 2005 102 7571 7576 15897453 10.1073/pnas.0502658102
Boutell C Sadis S Everett RD Herpes simplex virus type 1 immediate-early protein ICP0 and its isolated RING finger domain act as ubiquitin E3 ligases in vitro J Virol 2002 76 841 850 11752173 10.1128/JVI.76.2.841-850.2002
Everett RD O'Hare P O'Rouke D Barlow P Orr A Point mutations in the herpes simplex virus type 1 Vmw110 ring finger helix affect activation of gene expression, viral growth, and interaction with PML-containing structures. J Virol 1995 69 7339 7344 7474166
Hagglund R Van Sant C Lopez P Roizman B Herpes simplex virus 1-infected cell protein 0 contains two E3 ubiquitin ligase sites specific for different E2 ubiquitin-conjugating enzymes Proc Natl Acad Sci U S A 2002 99 631 636 11805320 10.1073/pnas.022531599
Ishov AM Maul GG The periphery of nuclear domain 10 (ND10) as site of DNA virus deposition J Cell Biol 1996 134 815 826 8769408 10.1083/jcb.134.4.815
Maul GG Ishov AM Everett RD Nuclear domain 10 as preexisting potential replication start sites of herpes simplex virus type-1 Virology 1996 217 67 75 8599237 10.1006/viro.1996.0094
Everett RD Sourvinos G Leiper C Clements JB Orr A Formation of nuclear foci of the herpes simplex virus type 1 regulatory protein ICP4 at early times of infection: localization, dynamics, recruitment of ICP27, and evidence for the de novo induction of ND10-like complexes J Virol 2004 78 1903 1917 14747555 10.1128/JVI.78.4.1903-1917.2004
Everett RD Maul GG HSV-1 IE protein Vmw110 causes redistribution of PML. EMBO J 1994 13 5062 5069 7957072
Maul GG Everett RD The nuclear location of PML, a cellular member of the C3HC4 zinc-binding domain protein family, is rearranged during herpes simplex virus infection by the C3HC4 viral protein ICP0 J Gen Virol 1994 75 1223 1233 8207389
Maul GG Guldner HH Spivack JG Modification of discrete nuclear domains induced by herpes simplex virus type 1 immediate early gene 1 product (ICP0) J Gen Virol 1993 74 2679 2690 8277273
Everett RD Freemont P Saitoh H Dasso M Orr A Kathoria M Parkinson J The disruption of ND10 during herpes simplex virus infection correlates with the Vmw110- and proteasome-dependent loss of several PML isoforms J Virol 1998 72 6581 6591 9658103
Chelbi-Alix MK de The H Herpes virus induced proteasome-dependent degradation of the nuclear bodies-associated PML and Sp100 proteins. Oncogene 1999 18 935 941 10023669 10.1038/sj.onc.1202366
Parkinson J Everett RD Alphaherpesvirus proteins related to herpes simplex virus type 1 ICP0 affect cellular structures and proteins. J Virol 2000 74 10006 10017 11024129 10.1128/JVI.74.21.10006-10017.2000
Russel J Stow ND Stow EC Preston CM Herpes simplex virus genes involved in latency in vitro J Gen Virol 1987 68 3009 3018 2826646
Harris RA Everett RD Zhu XX Silverstein S Preston CM Herpes simplex virus type 1 immediate-early protein Vmw110 reactivates latent herpes simplex virus type 2 in an in vitro latency system. Journal of Virology 1989 63 3513 3515 2545921
Harris RA Preston CM Establishment of latency in vitro by the herpes simplex virus type 1 mutant in1814 J Gen Virol 1991 72 907 913 1849973
Everett RD Boutell C Orr A Phenotype of a Herpes Simplex Virus Type 1 Mutant That Fails To Express Immediate-Early Regulatory Protein ICP0 J Virol 2004 78 1763 1774 14747541 10.1128/JVI.78.4.1763-1774.2004
Cai W Schaffer PA A cellular function can enhance gene expression and plating efficiency of a mutant defective in the gene for ICPO, a transactivating protein of herpes simplex virus type 1 J Virol 1991 65 4078 4090 1649316
Nicholl MJ Robinson LH Preston CM Activation of cellular interferon-responsive genes after infection of human cells with herpes simplex virus type 1 J Gen Virol 2000 81 2215 2218 10950979
Preston CM Harman AN Nicholl MJ Activation of interferon response factor-3 in human cells infected with herpes simplex virus type 1 or human cytomegalovirus Journal of Virology 2001 75 8909 8916 11533154 10.1128/JVI.75.19.8909-8916.2001
Mossman KL Macgregor PF Rozmus JJ Goryachev AB Edwards AM Smiley JR Herpes simplex virus triggers and then disarms a host antiviral response J Virol 2001 75 750 758 11134288 10.1128/JVI.75.2.750-758.2001
Eidson KM Hobbs WE Manning BJ Carlson P DeLuca NA Expression of herpes simplex virus ICP0 inhibits the induction of interferon-stimulated genes by viral infection J Virol 2002 76 2180 2191 11836395 10.1128/jvi.76.5.2180-2191.2002
Lin R Noyce RS Collins SE Everett RD Mossman KL
The herpes simplex virus ICP0 RING finger domain inhibits IRF3- and IRF7-mediated activation of interferon-stimulated genes Journal of Virology 2004 78 1675 1684 14747533 10.1128/JVI.78.4.1675-1684.2004
Mossman KL Saffran HA Smiley JR Herpes simplex virus ICP0 mutants are hypersensitive to interferon J Virol 2000 74 2052 2056 10644380 10.1128/JVI.74.4.2052-2056.2000
Mossman KL Smiley JR Herpes simplex virus ICP0 and ICP34.5 counteract distinct interferon-induced barriers to virus replication J Virol 2002 76 1995 1998 11799195 10.1128/JVI.76.4.1995-1998.2002
Harle P Sainz BJ Carr DJ Halford WP The immediate-early protein, ICP0, is essential for the resistance of herpes simplex virus to interferon-alpha/beta Virology 2002 293 295 304 11886249 10.1006/viro.2001.1280
Mossman KL Smiley JR Truncation of the C-terminal acidic transcriptional activation domain of herpes simplex virus VP16 renders expression of the immediate-early genes almost entirely dependent on ICP0 J Virol 1999 73 9726 9733 10559282
Meurens F Schynts F Keil GM Muylkens B Vanderplasschen A Gallego P Thiry E Superinfection prevents recombination of the alphaherpesvirus bovine herpesvirus 1 J Virol 2004 78 3872 3879 15047803 10.1128/JVI.78.8.3872-3879.2004
Johnson RM Spear PG Herpes simplex virus glycoprotein D mediates interference with herpes simplex virus infection J Virol 1989 63 819 827 2536105
Banfield BW Kaufman JD Randall JA Pickard GE Development of pseudorabies virus strains expressing red fluorescent proteins: new tools for multisynaptic labeling applications J Virol 2003 77 10106 10112 12941921 10.1128/JVI.77.18.10106-10112.2003
Preston CM Control of herpes simplex virus type 1 mRNA synthesis in cells infected with wild-type virus or the temperature-sensitive mutant tsK J Virol 1979 29 275 284 219222
DeLuca NA McCarthy AM Schaffer PA Isolation and characterization of deletion mutants of herpes simplex virus type 1 in the gene encoding immediate-early regulatory protein ICP4 J Virol 1985 56 558 570 2997476
Smiley JR Duncan J Truncation of the C-terminal acidic transcriptional activation domain of herpes simplex virus VP16 produces a phenotype similar to that of the in1814 linker insertion mutation J Virol 1997 71 6191 6193 9223515
Knipe DM Senechek D Rice SA Smith JL Stages in the nuclear association of the herpes simplex virus transcriptional activator protein ICP4 J Virol 1987 61 276 284 3027360
Craig JM Heterochromatin--many flavours, common themes Bioessays 2005 27 17 28 15612037 10.1002/bies.20145
Deshmane SL Fraser NW During latency, herpes simplex virus type 1 DNA is associated with nucleosomes in a chromatin structure Journal of Virology 1989 63 943 947 2536115
Everett RD Activation of cellular promoters during herpes virus infection of biochemically transformed cells Embo J 1985 4 1973 1980 2998778
Cheung P Panning B Smiley JR Herpes simplex virus immediate-early proteins ICP0 and ICP4 activate the endogenous human alpha-globin gene in nonerythroid cells J Virol 1997 71 1784 1793 9032307
Nicholl MJ Preston CM Inhibition of herpes simplex virus type 1 immediate-early gene expression by alpha interferon is not VP16 specific J Virol 1996 70 6336 6339 8709261
DeLuca NA Schaffer PA Activities of herpes simplex virus type 1 (HSV-1) ICP4 genes specifying nonsense peptides Nucleic Acids Res 1987 15 4491 4511 3035496
Rice SA Knipe DM Genetic evidence for two distinct trans-activation functions of the herpes simplex virus a protein ICP27 J Virol 1990 64 1704 1715 2157053
Long MC Leong V Schaffer PA Spencer CA Rice SA ICP22 and the UL13 protein kinase are both required for herpes simplex virus-induced modification of the large subunit of RNA polymerase II J Virol 1999 73 5593 5604 10364308
McLean C Buckmaster A Hancock D Buchan A Fuller A Minson AC Monoclonal antibodies to three nonglycosylated antigens of herpes simplex virus type 2 J Gen Virol 1982 63 297 305 6296279
Weinheimer SP Boyd BA Durham SK Resnick JL O'Boyle DR Deletion of the VP16 open reading frame of herpes simplex virus type 1 J Virol 1992 66 258 269 1309245
|
16300675
|
PMC1310514
|
CC BY
|
2021-01-04 16:38:58
|
no
|
Virol J. 2005 Nov 21; 2:85
|
utf-8
|
Virol J
| 2,005 |
10.1186/1743-422X-2-85
|
oa_comm
|
==== Front
Thromb JThrombosis Journal1477-9560BioMed Central London 1477-9560-3-181629724410.1186/1477-9560-3-18Original Clinical InvestigationIntermittent pneumatic compression for prevention of pulmonary thromboembolism after gynecologic surgery Suzuki Nao [email protected] Fumio [email protected] Atsushi [email protected] Takeshi [email protected] Sachiko [email protected] Hiroyuki [email protected] Akiyo [email protected] Nobuyuki [email protected] Daisuke [email protected] Department of Obstetrics and Gynecology, St. Marianna University School of Medicine, 2-16-1 Sugao, Miyamae-ku, Kawasaki-City, Kanagawa 216-8511, Japan2 Department of Obstetrics and Gynecology, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan2005 19 11 2005 3 18 18 9 9 2005 19 11 2005 Copyright © 2005 Suzuki et al; licensee BioMed Central Ltd.2005Suzuki et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
To investigate the incidence of pulmonary embolism and risk factors for this condition after obstetric and gynecologic surgery, as well as the efficacy of intermittent pneumatic compression.
Methods
A total of 6,218 patients operated at Keio University Hospital excluding obstetric or infertility-related surgery and uterine cervical conization were evaluated retrospectively to determine the preventive effect of intermittent pneumatic compression on postoperative pulmonary embolism.
Results
Pulmonary embolism occurred in 42 patients (0.68%). Multivariate analysis showed that malignancy, blood transfusion, and a body mass index ≥25 kg/m2 or ≥28 kg/m2 were independent risk factors for postoperative pulmonary embolism. A significantly lower incidence of pulmonary embolism occurred in patients receiving pneumatic compression postoperatively versus those without it. Among gynecologic malignancies, endometrial cancer was a significant risk factor for pulmonary embolism.
Conclusion
Preventive measures, including intermittent pneumatic compression, should be taken to avoid postoperative pulmonary thromboembolism in the gynecology field.
==== Body
Background
Previously, postoperative venous thromboembolism (VTE) did not attract much attention in Japan because its incidence was lower than in the USA and Europe [1]. However, an increasing number of patients have recently been diagnosed with VTE in Japan along with improved detection thanks to progress in imaging technologies and increasing medical interest in VTE. VTE is associated with pulmonary thromboembolism (PTE), which causes death in nearly 50% of patients if untreated and which appears to be caused by embolism arising from deep venous thrombosis. According to the statistics compiled by the Japanese Ministry of Health, Labor and Welfare, the number of deaths due to PTE increased more than 10-fold from 1951 to 2000 [2]. However, a recent report estimated the annual number of PTE patients in Japan at 3,492, which is approximately 1/25 of the number in the USA [3]. Prevention of VTE has been studied intensively in the USA and Europe since the American College of Chest Physicians (ACCP) Consensus Conference was held in 1985. Development of guidelines for the prevention of VTE has been discussed based on high-level evidence, and the seventh ACCP Consensus Statement [4] and the International Consensus Statement [5] were published in 2001. Although the incidence of VTE has been rapidly increasing in Japan, there is still a large difference from the incidence in the USA and Europe. Many deaths due to VTE with PTE may be regarded as sudden deaths because clinical diagnosis is difficult; therefore suitable Japanese guidelines for the prevention of VTE were compiled in 2004 [6].
The Japanese Society of Anesthesiologist (JSA) conducted a survey of perioperative PTE in 2003 and reported that there were 4.41 events per 10,000 operations; the number of PTE patients was third highest in the gynecologic field after the orthopedic and gastrointestinal surgery fields [7]. The incidence of VTE is higher in pregnant women than in non-pregnant women because of hypercoagulability, hypofibrinolysis, platelet activation, venous smooth muscle relaxation by female hormones, and venous compression by the enlarged uterus.
There have been few reports on postoperative PTE in the gynecologic field, apart from that by Nicolaids et al. [8], which indicated that the incidence of PTE was 40–80% higher after extended surgery for malignancy. Therefore, more data about VTE after gynecologic surgery, including that for malignant tumors, are needed. In the present study, we retrospectively investigated the incidence and risk factors for PTE after obstetric and gynecologic surgery performed at Keio University Hospital, and also evaluated the usefulness of intermittent pneumatic compression (IPC) for prevention of postoperative PTE.
Methods
A total of 6,218 patients who underwent operations at the Department of Obstetrics and Gynecology, Keio University School of Medicine between January 1995 and December 2003 were analyzed in this study; operations excluded obstetric, infertility-related and uterine cervical conization surgeries. 42 patients were found to have developed postoperative PTE. As a control group, a total of 929 patients who underwent obstetric, infertility-related surgery and uterine cervical conization operations who did not develop PTE in either 1995 (471 patients) or 2002 (458 patients) were selected.
Univariate analysis was used to assess the relationship between PTE and risk factors, such as the age, body mass index (BMI), smoking habits, presence/absence of complications (hypertension, abnormal glucose tolerance, heart disease, and collagen disease), operating time, perioperative bleeding, surgical indication (benign or malignant disease), presence/absence of retroperitoneal lymph node dissection, and perioperative blood transfusion. We also investigated the relationship between PTE and risk factors, including the use of IPC, by multivariate logistic regression analysis. The patient pools were analyzed according to the following risk subcategories: age was divided into less than 40 years of age, between 40 and 50 years of age, and greater than 50 years of age; BMI was divided into less than 25 kg/m2, between 25 and 28 kg/m2, and greater than 28 kg/m2; operating time was divided into less than 4 hours, between 4 to 6 hours, and greater than 6 hours; and perioperative blood loss was divided into less than 1000 mL, between 1000 and 2000 mL, and greater than 2000 mL. At our hospital, IPC has been used in patients undergoing gynecologic surgery to prevent postoperative PTE since 1999, before which elastic bandages or stockings were used (both methods were employed during the transition year of 1998).
Symptomatic PTE patients were further evaluated by electrocardiography, arterial blood gas analysis, chest X-ray examination, or echocardiography. A confirmative diagnosis was made by chest helical computed tomography scan and pulmonary ventilation-perfusion scintigraphy. Furthermore, asymptomatic patients presenting with only a decreased transcutaneous oxygen saturation (SpO2) were similarly evaluated for confirmative diagnosis.
Results
Incidence of PTE
The overall incidence of PTE was 0.68% (42/6,218). In patients undergoing surgery for benign diseases, excluding obstetrics and infertility-related surgery, the incidence was 0.32% (10/3,158), while the incidence was 2.21% (32/1,451) in patients undergoing surgery for malignancy apart from uterine cervical conization.
Symptoms
Symptoms were present in 27 of the 42 patients who had postoperative PTE, with chest pain and dyspnea occurring in approximately half of them (Table 1). No symptoms were observed in 15 patients who were diagnosed by a decreased SpO2. PTE was diagnosed at a mean of 2.69 days after surgery.
Table 1 Symptoms of PTE
Symptom No. of patients
Chest pain 16
Dyspnea 15
Tachycardia 3
Chest discomfort 2
Palpitations 1
Hypotension 1
Cyanosis 1
Chills 1
Vomiting 1
No symptoms 15
Note: There were 42 patients, but some had multiple symptoms.
PTE: pulmonary thromboembolism
Risk factors (Figure 1)
Figure 1 Prevalence of risk factors in patients with postoperative PTE. Six factors (malignancy, retroperitoneal lymph node dissection, blood transfusion, hypertension, abnormal glucose tolerance, and heart disease) were significantly associated with PTE.
We investigated the association between postoperative PTE and various risk factors, including the surgical indication (benign or malignant disease), perioperative blood transfusion, presence/absence of retroperitoneal lymph node dissection, smoking habits, and presence/absence of complications (hypertension, abnormal glucose tolerance, heart disease, and collagen disease). As a result, postoperative PTE was significantly associated with malignancy, perioperative blood transfusion, retroperitoneal lymph node dissection, hypertension, abnormal glucose tolerance, and heart disease.
Influence of age and BMI
Among the 42 patients with postoperative PTE, 93% were aged 40 years or older and 69% were aged 50 years or older (Figure 2-A). There was no such bias in the age distribution of the control group, with approximately 30% of the subjects in each age group. Patients with a BMI ≥25 kg/m2 accounted for 40% of the PTE group versus 15% of the control group (Figure 2-B).
Figure 2 Distribution of risk factors in the postoperative PTE and control groups. A, age; B, BMI; C, perioperative bleeding; D, operating time.
Operating time and perioperative bleeding
Sixty-two percent of patients with postoperative PTE had an operating time of four hours or longer and 36% had a time of six hours or longer (Figure 2-C). In the control group, 21% of patients had an operating time of four hours or longer and 9% underwent operations lasting six hours or longer. In addition, perioperative bleeding was ≥1,000 mL in 45% of PTE patients and ≥2,000 mL in 21%. In the control group, however, perioperative bleeding was ≥1,000 mL in 18% of patients and ≥2,000 mL in 6% (Figure 2-D).
Univariate analysis (Table 2)
Table 2 Results of univariate analysis
Factor Odds ratio p value (chi-square)
Age ≥40 years old 6.637 <0.01
Age ≥50 years old 4.362 <0.01
BMI ≥25 kg/m2 3.718 <0.01
BMI ≥28 kg/m2 3.911 <0.05
Smoking 0.809 n.s.
Hypertension 3.018 <0.01
Abnormal glucose tolerance 6.285 <0.05
Heart disease 2.951 <0.05
Collagen disease 2.161 n.s.
Malignant tumor surgery 8.327 <0.01
Retroperitoneal lymph node dissection 7.577 <0.01
Perioperative bleeding ≥1,000 mL 3.797 <0.01
Perioperative bleeding ≥2,000 mL 4.252 <0.01
Blood transfusion 7.152 <0.01
Operating time ≥4 hours 5.999 <0.01
Operating time ≥6 hours 5.516 <0.01
BMI: body mass index
n.s.: not significant
Analysis by the χ2 test showed a significant association between PTE and seven background factors: age ≥40 years, age ≥50 years, BMI ≥25 km/m2, BMI ≥28 kg/m2, hypertension, abnormal glucose tolerance, and heart disease. A significant association was also observed with the following seven surgical factors: operation for malignancy, operating time ≥4 hours, operating time ≥6 hours, perioperative bleeding ≥1,000 mL, perioperative bleeding ≥2,000 mL, retroperitoneal lymph node dissection, and perioperative blood transfusion. These variables were also selected as risk factors.
Multivariate analysis (Table 3)
Table 3 Results of multivariate analysis
Factor Risk ratio 95% CI p value (chi-square)
Lower Upper
Age ≥40 years old 2.645 0.703 9.946 n.s.
Age ≥50 years old 3.418 0.926 12.615 n.s.
BMI ≥25 kg/m2 2.718 1.149 6.427 <0.05
BMI ≥28 kg/m2 3.922 1.297 11.863 <0.05
Smoking 1.029 0.364 2.913 n.s.
Hypertension 0.838 0.311 2.259 n.s.
Abnormal glucose tolerance 2.176 0.675 7.016 n.s.
Heart disease 1.149 0.304 4.345 n.s.
Collagen disease 1.181 0.199 7.018 n.s.
Malignant tumor surgery 2.860 1.083 7.522 <0.05
Blood transfusion 3.834 1.683 8.737 <0.01
IPC 0.396 0.193 0.814 <0.05
BMI: body mass index
CI: confidence interval
IPC: intermittent pneumatic compression
n.s.: not significant
Before performing multivariate analysis, the correlations between variables were investigated by calculating Spearman's correlation coefficients. There were strong correlations between malignancy, retroperitoneal lymph node dissection, operating time, and the amount of perioperative blood loss (data not shown), therefore we selected two of these factors related to surgery (malignancy and blood transfusion) for analysis.
As shown in Table 3, multivariate analysis was done for a total of 12 variables, including the use of IPC. Significant associations with PTE were observed in the case of surgery for malignant tumors, blood transfusion, BMI ≥25 km/m2, and BMI ≥28 kg/m2, with their risk ratios being 2.860, 3.834, 2.718, and 3.922, respectively. A significantly lower rate of PTE was observed in the patients treated with IPC (risk ratio of 0.396).
Postoperative PTE and gynecologic diseases
Thirty-two patients among 42 postoperative patients who developed PTE had malignant tumors (76%) compared to an overall rate of PTE development of 2.21% (32/1,451) among all patients undergoing surgery for gynecologic malignancies.
There were 16 patients who had endometrial cancer among the 32 PTE patients with malignancy (Table 4). BMI was ≥25 kg/m2 in 10/16 patients (62.5%) with endometrial cancer, 2/9 patients (22.2%) with ovarian cancer, and 2/6 patients (33.3%) with uterine cervical cancer (Table 5). There was also a significant difference of postoperative PTE between patients undergoing pelvic lymph node dissection alone and those undergoing combined pelvic and para-aortic lymph node dissection, with the incidence being 2.4% (15/636) and 6.2% (13/211), respectively (p < 0.01) (Table 6).
Table 4 Postoperative PTE and gynecologic diseases
Malignant tumors No. of patients Benign tumors No. of patients
Endometrial cancer 16 Uterine myoma 3
Ovarian cancer 9 Uterine adenomyosis 3
Cervical cancer 6 Ovarian tumor 1
Others 1 Cystocele 1
Uterine myoma/ovarian tumor 1
Others 1
Total 32 Total 10
PTE: pulmonary thromboembolism
Table 5 Malignancy and BMI in patients with postoperative PTE
Endometrial cancer Ovarian cancer Cervical cancer
BMI <25 kg/m2 6 7 4
25–27 kg/m2 7 0 2
≥25 kg/m2 3 2 0
Total 16 9 6
BMI: body mass index
PTE: pulmonary thromboembolism
Table 6 Postoperative PTE and lymph node dissection
Site Incidence of postoperative PTE
Pelvic lymph nodes 2.4% (15/636) p < 0.01
+ Para-aortic lymph nodes 6.2% (13/211)
PTE: pulmonary thromboembolism
On the other hand, 10 of the 42 patients (24%) who developed PTE had benign disease compared to an overall rate of PTE development of 0.32% (10/3,158) among all patients undergoing surgery for benign gynecologic tumors.
Patients with uterine myoma and uterine adenomyosis accounted for almost half of these 10 patients (Table 4).
Effect of IPC
There was a significant difference between the incidence of postoperative PTE before and after the introduction of IPC, being 1.19% (23/1,928) versus 0.40% (14/3,525), respectively (p < 0.01) (Table 7).
Table 7 Prophylaxis for postoperative PTE
1995–1997 1998 1999–2003
No prevention 9 0 0
Elastic bandages 4 1 0
Elastic stockings 10 4 0
IPC 0 0 14
Before vs. after introduction of IPC: 1.19% (23/1,928) vs. 0.40% (14/3,525) (p < 0.01)
IPC: intermittent pneumatic compression
PTE: pulmonary thromboembolism
Discussion
Based upon a survey conducted by The Japan Society of Obstetrics, Gynecology, and Neonatal Hematology between 1991 and 2000 at 92 medical institutions in Japan, the incidence of postoperative PTE was 0.08% (168/203,058) in all patients undergoing gynecologic surgery, with a breakdown of 0.03% (51/175,448) in patients undergoing surgery for benign disease compared to 0.42% (117/27,610) in patients undergoing surgery for malignant disease.
The incidence was approximately 14 times higher in patients undergoing surgery for malignancy than in patients undergoing surgery for benign disease [9]. The present study retrospectively investigated the incidence of PTE after gynecologic surgery excluding obstetric surgery, infertility-related surgery, and uterine cervical conization, and we found that the incidence of postoperative PTE was 0.68% (42/6,218), which was higher than the average incidence in Japan. This result appears to be attributable to the high percentage of patients with malignancy among those undergoing gynecologic surgery in our hospital, since patients who had malignant disease surgery accounted for 76.5% of the 42 patients with postoperative PTE. The relationship between VTE and malignancy has long been known as Trousseau's syndrome. It has been reported that VTE is caused by release of procoagulant factors from cancer cells and direct damage to venous endothelial cells and that the incidence of PTE is 3–5 times higher in patients with malignancy [10]. In addition, cancer patients tend to be older and more often have complications, such as hypertension, abnormal glucose tolerance, and heart disease. Furthermore, surgery for malignancy requires a longer operating time, causes more bleeding, and often requires blood transfusion. These were all significant risk factors according to univariate analysis (χ2 test) in the present study. In addition, multivariate analysis selected malignancy as an independent risk factor for postoperative PTE along with blood transfusion and BMI (≥25 kg/m2 or ≥28 kg/m2). There have been few reports on VTE after gynecologic surgery; Horowitz found that obesity, a long period of immobilization, extensive cancer surgery, trauma, radiotherapy, a past history of VTE, severe varices, diabetes, and heart failure were risk factors of postoperative VTE [11]. This report is comparable with results of the present study. Our results are also comparable with the ACCP guidelines, which categorize patients ≥40 years old with extensive surgery and malignant tumors as the highest-risk group [4]. The JSA has also reported that obesity, long-term immobilization, and malignant tumors are risk factors for perioperative PTE, especially in female patients [7]. Of the 32 patients with postoperative PTE in the present study, 50% had endometrial cancer, which is often associated with obesity, hypertension, and abnormal glucose tolerance.
Endometrial cancer is increasing relative to cervical cancer in Japan as well as the USA and Europe. Therefore, endometrial cancer appears to be one of the strongest risk factors for postoperative PTE among malignant gynecologic tumors. Thrombosis occurs due to Virchow's triad, namely 1) hypercoagulability, 2) stagnation of blood, and 3) vascular endothelial cell damage. Retroperitoneal lymph node dissection was not identified as an independent risk factor for PTE according to multivariate analysis in the present study. Lymph node dissection may be closely related to the occurrence of VTE since this procedure causes vascular damage and also accumulations of lymph may compress the veins after surgery and cause stagnation of blood. As shown in Table 6, the incidence of postoperative PTE was increased in patients who had both pelvic and para-aortic lymph node dissection.
Retroperitoneal lymph node dissection is important for the treatment of endometrial cancer, but these findings suggest that it is necessary to carefully consider the performance of lymph node dissection in patients with a number of risk factors as well as the use of postoperative radiotherapy.
There are also patients with benign gynecologic disease in whom attention should be paid to the risk of postoperative PTE. Of the 10 patients with benign gynecologic disease in the present series, six were had relatively large uterine tumors, including myoma and adenomyosis. Although age is specified as a risk factor in the ACCP guidelines [4], neither an age ≥40 years nor an age ≥50 years was identified as a risk factor in the present study. However, multivariate analysis by the χ2 test showed a significant difference for patients ≥50 years old (p = 0.0651) in this study, indicating that elderly patients are at increased risk of developing postoperative PTE.
As well as minimizing risk factors, it is important to employ adequate preventive measures for VTE in high-risk patients according to the guidelines. Multivariate analysis showed that the risk ratio of patients treated with IPC was 0.396 (p < 0.05), so this technique was an independent preventive factor for postoperative PTE. Our department introduced IPC in 1999, as did other institutions in Japan [6]. Comparison of the incidence of postoperative PTE before and after the introduction of IPC showed a significant decrease from 1.19% to 0.40%. In addition, the five patients experiencing postoperative PTE in 1998 (transitional period for introduction of IPC) were all managed with other preventive methods. The ACCP guidelines state that IPC reduces the risk of postoperative VTE by 88%, which is superior to the risk reduction rate for low-dose unfractionated heparin (68%) or low-molecular-weight heparin (76%), suggesting that IPC is one of the most useful preventive methods for VTE and PTE [5].
The JSA reported that the incidence of perioperative PTE is 4.41/10,000, with the mortality rate being 18% [7]. It also reported that 57.7% of PTE was likely to be preventable. In Japan, there is a need to accumulate more evidence-based clinical data, in order to better define the risk factors for VTE and PTE and allow the selection of appropriate preventive methods.
Conclusion
Preventive measures, including intermittent pneumatic compression, should be taken to avoid postoperative pulmonary thromboembolism in the gynecology field.
List of abbreviations
ACCP American College of Chest Physicians
BMI body mass index
IPC intermittent pneumatic compression
JSA Japanese Society of Anesthesiologist
PTE pulmonary thromboembolism
SpO2 oxygen saturation
VTE venous thromboembolism
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
NS (Suzuki) and SE were involved in the sequence alignment and drafted the manuscript.
HN and NS (Suzuki) were involved in writing of method.
NS (Suzuki) and FK were involved in writing manuscript.
FK, AH, SE, and HN were involved in analysis of data.
*** was involved in the sequence alignment.
NS (Suzuki), FK, and DA were involved in the design of the study.
NH, AT and NS (Susumu) performed the statistical analysis.
AH and HT helped to draft the manuscript.
NS (Suzuki), NS (Susumu) and DA were involved in planning, experimental setup.
All authors read and approved the final manuscript.
Acknowledgements
The authors thank Keiko Abe for her secretarial work.
==== Refs
Gore I Hirst AE Tanaka K Myocardial infarction and thromboembolism Arch Int Med 1964 113 323 330 14096382
Sakuma M Konno Y Shirato K Increasing mortality from pulmonary embolism in Japan, 1951–2000 Jpn Circ J 2002 66 1144 1149
Kumasaka N Sakuma M Shirato K Incidence of pulmonary thromboembolism in Japan Jpn Circ J 1999 63 439 441 10406582 10.1253/jcj.63.439
Geerts WH Pineo GF Heit JA Bergqvist D Lassen MR Colwell CW Ray JG Prevention of venous thromboembolism. The Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy Chest 2004 126 338S 400S 15383478 10.1378/chest.126.3_suppl.338S
Nicolaids AN Breddin HK Fareed J Goldhaber S Haas S Hull R Kalodiki E Myers K Samama M Sasahara A Prevention of venous thromboembolism. International Consensus Statement. Guidelines complied in accordance with the scientific evidence Int Angiol 2001 20 1 37 11342993
Kobayashi T Nakabayashi M Ishikawa M Ikenoue T Adachi T Final reports of venous thrombosis in obstetrics and gynecology and new guideline for thromboprophylaxsis J J Obstet Gynecol Neonat Hematol 2004 14 5 6
Kuroiwa M Furuya H Seo N Morita K Sha M Iwao Y Sasaki J Ito M Incidence and characteristics of perioperative pulmonary thromboembolism in Japan J J Anesthesiology 2004 53 454 463
Nicolaids AN Bergqvist D Hull RD Prevention of venous thromboembolism. International Consensus Statement Int Angiol 1997 16 3 38 9165356
Kobayashi T Interdisciplinary practice. Pulmonary thromboembolism/deep vein thrombosis Acta Obst Gynaec Jpn 2004 56 382 391
Lee AYY Levine MN Venous thromboembokism and cancer: Risks and outcomes Circulationt 2003 107 I17 I21
Horowitz IR Postanesthesia and postoperative care Te Linde's Oprative Gynecology 1997 8 Philadelphia: Lippincott Williams & Wilkins 127 140
|
16297244
|
PMC1310515
|
CC BY
|
2021-01-04 16:36:24
|
no
|
Thromb J. 2005 Nov 19; 3:18
|
utf-8
|
Thromb J
| 2,005 |
10.1186/1477-9560-3-18
|
oa_comm
|
==== Front
Popul Health MetrPopulation Health Metrics1478-7954BioMed Central London 1478-7954-3-101628198210.1186/1478-7954-3-10ResearchTrading people versus trading time: What is the difference? Damschroder Laura J [email protected] Todd R [email protected] Christine C [email protected] Molly E [email protected] Peter A [email protected] VA Health Service Research & Development Center of Excellence, VA Ann Arbor Healthcare System, Ann Arbor, MI. USA2 Division of General Internal Medicine, University of Michigan Medical School, Ann Arbor, Michigan, USA3 The Center for Behavioral and Decision Sciences in Medicine, University of Michigan and VA Ann Arbor Healthcare System, Ann Arbor, MI USA4 Department of Psychology, University of Michigan, Ann Arbor, MI USA2005 10 11 2005 3 10 10 28 4 2005 10 11 2005 Copyright © 2005 Damschroder et al; licensee BioMed Central Ltd.2005Damschroder et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Person trade-off (PTO) elicitations yield different values than standard utility measures, such as time trade-off (TTO) elicitations. Some people believe this difference arises because the PTO captures the importance of distributive principles other than maximizing treatment benefits. We conducted a qualitative study to determine whether people mention considerations related to distributive principles other than QALY-maximization more often in PTO elicitations than in TTO elicitations and whether this could account for the empirical differences.
Methods
64 members of the general public were randomized to one of three different face-to-face interviews, thinking aloud as they responded to TTO and PTO elicitations. Participants responded to a TTO followed by a PTO elicitation within contexts that compared either: 1) two life-saving treatments; 2) two cure treatments; or 3) a life-saving treatment versus a cure treatment.
Results
When people were asked to choose between life-saving treatments, non-maximizing principles were more common with the PTO than the TTO task. Only 5% of participants considered non-maximizing principles as they responded to the TTO elicitation compared to 68% of participants who did so when responding to the PTO elicitation. Non-maximizing principles that emerged included importance of equality of life and a desire to avoid discrimination. However, these principles were less common in the other two contexts. Regardless of context, though, participants were significantly more likely to respond from a societal perspective with the PTO compared to the TTO elicitation.
Conclusion
When lives are at stake, within the context of a PTO elicitation, people are more likely to consider non-maximizing principles, including the importance of equal access to a life-saving treatment, avoiding prejudice or discrimination, and in rare cases giving treatment priority based purely on the position of being worse-off.
==== Body
Background
Preliminary results from this study were presented in a poster session at the 2002 AcademyHealth Conference on June 23, 2002, and final results were presented at the 25th Annual Meeting of the Society Medical Decision Making on October 2003.
Cost-effectiveness analyses (CEAs) show how to maximize the number of quality adjusted life-years (QALYs) that can be obtained within a given budget – an approach we refer to as "QALY-maximization." The United States Public Health Service Panel on Cost Effectiveness recommended three methods for measuring preferences in CEAs: the rating scale, the standard gamble, and the time trade-off elicitation technique [1]. Some people argue that none of the three recommended preference measurement techniques are particularly suited for capturing public preferences in allocation or rationing contexts because they support a distributive principle based on maximizing treatment benefits, measured as QALYs, without regard for who receives the benefits. People have described other principles they believe are normatively important to ensure a just distribution of healthcare treatments, including the importance of giving: 1) more weight to patients with a more severe health condition [2-4]; 2) equal weight on saving the lives of people with or without disabilities [4,5]; and 3) sufficiently high priority on treating patients in clear need of beneficial treatments [6-10].
Nord has argued that, "the obvious way to make sure that the QALY procedure captures social preferences for person trade-offs is, of course, to use person trade-off exercises as the basis for scoring health states in the first place" [[3], page 201]. page 201 The person trade-off (PTO) preference elicitation method was proposed to do just that [3,11-16]. In traditional preference measurement methods, such as the time trade-off (TTO) and standard gamble (SG), people are asked, typically from a personal perspective, to state their preference for a health state by imagining they are in that health state but could be returned to perfect health if they lived fewer years (TTO) or won a gamble (SG). In PTO elicitations, people are asked to make tradeoffs between treating different groups of patients who differ by a constellation of attributes. For example, respondents might be asked how many patients would need to be cured of moderate leg pain to be equally good as curing 100 patients of severe shortness of breath.
The debate about the relative merits of traditional utility measures versus the PTO measure is not merely a theoretical one. Numerous studies have shown that preferences elicited by traditional utility measurement methods can differ significantly from those elicited by the PTO method [3,14-21]. Differences are especially striking when lives are at stake. For example, all else being equal, saving the life of someone with a medical disability (e.g. paraplegia) produces fewer QALYs than saving the life of someone who can be returned to perfect health. However, public preferences, as measured by PTO elicitations, place nearly equal value on saving the lives of healthy people and people with paraplegia [22-24].
Why do PTO elicitations yield preference values that are different than values obtained using traditional utility measures? Traditional measures elicit subjective expected utilities for a health condition which are used to set priorities based on principles of QALY-maximization (see for example, Nord [16], Ubel [12], and Olsen [15]). Many people believe that PTO elicitations, in addition to capturing the underlying subjective expected utility for the condition being evaluated, incorporate other distributive principles. Indeed, Salomon and Murray demonstrated how these "distributional concerns" could be derived as weights separate from the "core strength of preference" that is common across many elicitation methods [25]. We conducted this study to gain insight into the factors people consider when responding to PTO elicitations compared to one traditional preference measurement method, the TTO method. We used a think-aloud protocol in face-to-face interviews in which we asked participants to state their reasoning out loud as they responded to both types of elicitations. Specifically, our goal was to determine whether people mention considerations related to distributive principles other than QALY-maximization more often in PTO elicitations than in TTO elicitations and whether this could account for the empirical differences.
Methods
We recruited a convenience sample of participants from the general public in and around a small Midwestern city in the U.S. Venues included a nearby major metropolitan airport, a local Laundromat, a Veterans Affairs Medical Center, and we recruited some over the phone who traveled on-site for the interview. Willing respondents participated in a semi-structured interview that consisted of one TTO elicitation, followed by multiple PTO elicitations. We obtained participants' strength of preferences for paraplegia, severe shortness of breath, and moderate leg pain using both methods.
Elicitations
We started the interviews with a TTO elicitation, asking participants to consider two friends, Mr. (Mrs.) Adams and Mr. (Mrs.) Brown who were both 30 years old and who would live with their health condition (e.g. perfect health or paraplegia) for 50 more years, and then die in their sleep. We matched the gender of the imaginary friends to the gender of the participant. Participants responded to one of two TTO elicitations – either making tradeoffs between paraplegia and perfect health or between severe shortness of breath and moderate leg pain. We started off by asking which friend was better off and then, assuming the respondent chose the friend with the less severe condition, asked them to choose between one friend living in the worse-off condition for 50 more years and the other friend living for only a "few more days" with the better-off condition. We chose a lengthy timeframe (50 years) because we believed more participants would willingly trade years over a longer time horizon compared to a shorter time horizon [26,27]. Most TTO elicitations use a personal perspective in which respondents are asked to imagine themselves in the worse-off condition with a chance for a cure if they are willing to live in perfect health for a shorter period of time [28]. However, we used an impersonal perspective (depicting the two friends) because one study found that people were more willing to trade years to improve quality of life with this perspective while only marginally affecting overall values [29] and another study did not find any differences in values when they compared personal versus non-personal perspectives [21].
After the TTO elicitation, participants responded to PTO elicitations in which we asked them to imagine being "on a panel of experts trying to decide between two different medical treatments." We described two alternative treatment programs (see the Appendix) and asked the participants, "Which would you choose or are the choices equally good?" Participants responded to at least two PTO scenarios (as shown in Table 1) in which they were asked to choose between: 1) curing a life-threatening infection in previously health people versus curing a life-threatening infection in patients with paraplegia; 2) curing a life-threatening infection in previously healthy people versus curing patients with a spinal cord injury to prevent paraplegia; or 3) curing patients of severe shortness of breath versus curing patients of moderate leg pain. Our intent was to compare responses from the TTO and PTO elicitations.
Table 1 Interview Structure for the Three Experimental Groups
Elicitation Save-Save Group Cure-Save Life Group Cure-Cure Group
TTO Paraplegia v. Perfect Health Paraplegia v. Perfect Health Leg Pain v. Severe Shortness of Breath
PTO 2 Distractor PTOs 1 Distractor PTO 1 Distractor PTO
Cure life-threatening infection in previously healthy patients v. patients with paraplegia Cure life-threatening infection v. Cure SCI1 Leg Pain v. Severe Shortness of Breath
Comparison & Reflection Questions You have answered two different types of questions in this interview. Health policy experts use both types of questions to research people's opinions about treating illness and disability. Now I want to show you how your answers might be interpreted. I especially want to know what you think about these interpretations
In the first part of our interview, we compared Mr./Mrs. Adams and Mr./Mrs. Brown in a type of question called the Time Trade-Off. You said that if Mr./Mrs. Brown lived <indifference point> years in perfect health, that he/she would be no better or worse off than Mr./Mrs. Adams, who lived 50 years with <condition> A policy expert would interpret that you think one year of perfect health is about the same as <computed ratio> years with paraplegia. In other words, you think paraplegia is <computed ratio> times worse than perfect health. What do you think about this interpretation?
Later in the interview, you made a choice between curing some people of <condition1> or curing other people of <condition2> in a type of question called the Person Trade-Off. Based on your answer to the Time Trade-Off question, a policy expert might assume you would say that <predicted PTO> people would have to be cured of <condition2> to make that choice as good as curing 100 people of <condition1>. Your actual answer was <PTO indifference point>.
Which number do you think is a better reflection of your thoughts: the number you actually gave for the person tradeoff or the one that was predicted based on your time tradeoff answer?
Do the time tradeoff and person tradeoff questions make you think about different issues?
1SCI = Spinal Cord Injury
We assigned participants to one of three versions of a structured interview. Table 1 highlights the TTO and PTO comparisons that were the focus of each of the three experimental groups. In the Save-Save group, our focus of comparison was: 1) a TTO elicitation where participants were asked to trade off years of perfect health versus living 50 years with paraplegia; and 2) a PTO elicitation that asked participants to choose between saving the lives of previously healthy patients or patients with paraplegia. In the Cure-Save group, we compared: 1) the same TTO elicitation as the Save-Save group; and 2) a PTO elicitation that asked participants to choose between saving the lives of previously healthy patients and curing spinal cord injury to prevent paraplegia. In the Cure-Cure group, we compared: 1) a TTO elicitation where participants were asked to trade off years with moderate leg pain versus living 50 years with severe shortness of breath; and 2) a PTO elicitation that asked participants to choose between curing patients with severe shortness of breath and curing patients with moderate leg pain. We did not want responses to the PTO elicitation to be influenced by responses to the TTO elicitation which was presented first. To help mitigate this possibility, we first asked the participant to respond to PTO "distractor" tasks that presented two unrelated health conditions. We started interviews with the Save-Save version of the interview first. We presented 2 distractor PTO elicitations to participants in this group. Participants in the Save-Save group did not refer back to the earlier TTO elicitation when responding to any of the PTO tasks. We made a decision to reduce to a single distractor task when interviewing the other two groups. Again, in the other two groups, no one made a reference to their earlier TTO response in any of the PTO tasks so we felt that the single distractor task was sufficient.
Indifference point calculations
In the TTO elicitation, we used a "ping-pong" method [30] to converge on an indifference point; the point at which it was difficult to choose which friend was better off. For example, a participant may have said that Mrs. Adams, who was living with paraplegia for 50 years, was just as well off as Mrs. Brown, who was living in perfect health for 40 years. The utility for e.g. paraplegia can be computed as:
Where U(h2) = utility for the health state being evaluated (e.g. paraplegia); th1 = time spent in the less severe health state (e.g. perfect health). In the example, we compute U(h2) as 0.8 for paraplegia. A utility cannot be computed for severe shortness of breath because participants traded-off relative to moderate leg pain instead of perfect health or death. In this context, the value is better regarded as a "relative utility." Based on the TTO utility, we predicted a PTO indifference point so we could compare this value to their actual PTO indifference point:
In the example above, the predicted PTO indifference point would be 125. This means that based on the participant's TTO valuation, we expect the participant to be indifferent between saving the lives of 125 people with paraplegia and saving the lives of 100 non-disabled people. We modified the predicted PTO indifference point calculation for participants in the Cure-Save group because the PTO elicitation compares curing a spinal cord injury (to prevent paraplegia) versus saving lives. Therefore, the denominator in Equation (2a) is substituted with the utility for curing paraplegia rather than the utility for the condition itself, as follows:
Continuing with the example above, the utility for curing paraplegia would be 0.2 and the predicted PTO indifference point for curing spinal cord injury (to prevent paraplegia) would be 500. That is, we would expect the participant to be indifferent between curing 500 patients of spinal cord injury versus saving the lives of 100 non-disabled people.
In the PTO elicitations, we asked participants how many patients would need to be cured in the comparison group (e.g. patients with moderate leg pain) to be equally good as curing 100 patients in the baseline group (e.g. patients with shortness of breath). The indifference point is the number of patients that the participant said was needed in the comparison group for both choices to be equally good. The higher the indifference point, the lower the value placed on curing the comparison group of patients relative to curing the baseline group of patients. We used an open-ended approach to obtain a PTO indifference point rather than a search procedure because the response range was open-ended. People could give as high a number as they wanted. One option, in this situation, might have been to use a titration approach but this approach yields responses that are different depending on the direction values are presented [31,32] and are different than those obtained using a ping-pong procedure [30]. Some participants were unable to state a number without prompting, however. In this situation, we did use a ping-pong approach using 6 billion (roughly equal to the population of the world) as the starting point for the high end and then used a procedure comparable to that used in the TTO elicitation to narrow down to an indifference point.
Qualitative data collection
Three trained interviewers used a verbal report method to explore the thought process of participants as they responded to the TTO and PTO elicitations. We used a concurrent think aloud protocol accompanied by verbal probing,[33] and asked participants to literally think aloud while responding to the elicitations, verbalizing any and all thoughts [33-36]. We directed participants to "just say out loud whatever is going through your mind as you answer my questions, even if it seems obvious. There is no right or wrong answer; we just want to hear how you think about these issues." After the interviewer posed each question, she reminded participants to think aloud as they answered the question. We complemented this approach with verbal probing, in which the interviewer prompted participants to expand upon their "think aloud" statements and/or to provide retrospective reports of their thoughts to elicit more complete verbalization [37]. We accomplished this by saying, for example, "Can you tell me a little more about what you were thinking as you came up with that answer?" We used these two methods to produce a combination of concurrent and retrospective reports which, when used together, provided a comprehensive description of participants' thought processes [34].
After the participant responded to the TTO and PTO elicitations, the interviewer used the TTO response to predict what the PTO indifference point would be for the corresponding PTO elicitation. The interviewer presented this prediction and the participant's actual indifference point and asked the participant to reflect on the two values by asking the questions listed at the bottom of Table 1. We audio-recorded and created verbatim transcripts of every interview.
Quantitative analysis
We compared demographic characteristics across the experimental groups using analysis of variance for continuous variables (i.e. age), and χ2 tests for categorical variables (Fisher's Exact Test for 2 × 2 comparisons) using SPSS Release 10. We coded African American, Hispanic, Native American, and Alaska Native participants as racial minorities. We compared indifference points predicted from TTO responses to actual PTO indifference points using the Wilcoxon signed rank test for paired observations. We also conducted a concordance analysis of qualitative codes for the TTO and PTO elicitations using McNemar's paired comparison tests.
Qualitative analysis
We conducted a descriptive analysis [38] of the verbatim transcribed interviews. The authors read one-third of the interviews (randomly chosen from each of the three experimental groups) and independently listed themes that arose from the readings. They consolidated themes in a step-wise process (reading and coding in two batches of transcripts) to create a final coding scheme. Once themes and definitions were established, three judges (LJD, TRR, CCG) continued reading through all of the transcripts using the coding scheme to identify themes. We used a consensus approach to resolve differences in coding before proceeding to the next batch of transcripts.
Results
A total of 75 participants participated in the study. We excluded a total of 11 (15%) participants: 2 interviews were not recorded properly and could not be transcribed; 5 interviews were not completed because participants had to catch their plane or go to their clinic appointment; 2 participants were confused about the questions and their answers were uninterpretable; and 2 participants protested against the questions. One of the "protesters" said the "ultimate answer is that there is no answer" in response to the TTO questions and that the PTO questions were "disgusting...it's like, which child do you love more, with big ears or the one with the small ears." The other protester said that the comparisons presented in both the TTO and PTO elicitations could never be equal. Of the 64 participants included in the analysis, 50% were female and participants had an average of 15 years of education. Table 2 shows participant demographics. Participants in the Save-Save group were marginally younger than participants in the Cure-Save and Cure-Cure groups. The Cure-Save group had a 9% racial minority of participants, while the Save-Save and Cure-Cure groups had 27% and 20%, respectively. We were unable to test for statistical differences in minority representation because of low minority counts in the Cure-Save group.
Table 2 Participant Demographic Characteristics
Save-Save Group (n = 22) Cure-Save Group (n = 22) Cure-Cure Group (n = 20) Overall p
Minority Status N/A
Yes 27% 9% 20% 19%
Age – Range 19 – 53 19 – 85 20 – 80 19 – 85
Mean (SD) 33 (12.9) 45 (16.3) 42 (19.2) 41 (17.2) 0.100
Gender
%Female 46% 55% 50% 50% 0.834
Education – Range 9 – 21 11 – 19 12 – 21 9 – 21
Avg Years (SD) 16 (3.3) 15 (2.2) 16 (2.6) 15 (2.6) 0.411
Location N/A
Metro Airport 50% 0% 0% 17%
Laundromat 41% 46% 35% 41%
VAMC 9% 23% 30% 20%
On-site 0% 32% 35% 22%
Seven themes emerged through our analysis of the transcribed interviews. Table 4 lists the themes along with illustrative quotes. We organized six of the themes into whether they were consistent with the QALY-maximization objective or not. The seventh theme differentiated participants who responded from a personalized perspective, either in terms of their own experience or mentioning other people they knew well who had experienced the condition. These participants used this prior experience to project what it would be like to live with the condition under consideration themselves. The next three sections present findings for each of the three groups.
Table 4 Qualitative Themes and Illustrative Quotes
Themes consistent with QALY-maximization considerations:
Quality of life References to the condition's impact on quality of life. Could be a positive or negative comment on a societal or individual level. TTO – "...it sounds to me like Mrs. Adams has ... the more severe symptoms, and I would ... think that...it might impact her quality of life more."
PTO – "I would choose to cure the people who are severely short of breath because ... I think it so affects the quality of life, every move they make. ...People don't have to walk up hill, they can avoid doing that, they can drive, they can find some other means. ...I don't see that as being something that's completely disabling, whereas I see the severe shortness of breath as completely disabling."
Years of life Expresses the importance of choosing the option that maximizes the length of life. TTO – "Probably the one that lives longer is in better shape..."
Non-health benefits Consideration of the choice that would result in the most benefit in terms of non-health dimensions such as contribution to society, economic contribution, employment, etc. TTO – "Even though he might have less years to live, he's really capable of doing a lot more."
PTO – "I guess if you wanted to look at it really, really coldly you could say this group is consuming fewer resources ..."
Themes consistent with equity considerations:
Fair Chance Consideration of giving someone who is worse-off a fair chance even though the benefit may be less than the alternative. ... if ... the non-paraplegic versus the paraplegic, have same length of life span ... probably the quality of life would be a little better for the non-paraplegic. So therefore...I would cure...the paraplegics.
Equality of life Consideration of equal value of lives as a basis of moral judgement or to be fair. If they're both living, it doesn't mean one's life is more valuable over the other.
Prejudice Specific mention of wanting to avoid prejudice or discrimination. Now I feel like I'm going to be prejudiced by picking this individual over that one.
Self-Perspective Comments about self (or close-others) who have experienced the condition or projecting self into having the condition. TTO – "...I would rather have pain on the leg than not be able to breathe properly."
PTO – "I think I would have to try to choose the ones with the shortness of breath. ... my son had asthma pretty bad, ... I could live with the leg pain but I don't think I could live with the shortness of breath."
Save-save group
The TTO elicitation reduced the number of years one friend would have to live in perfect health to be equally good as another friend who would live 50 years with paraplegia. We were interested in comparing responses to a PTO elicitation that asked how many people with paraplegia would need to be cured of a life-threatening infection to make them indifferent between curing that group versus curing 100 healthy people who had the life-threatening infection. Table 3 shows the median indifference point predicted from TTO responses and the median actual indifference point obtained through the PTO elicitation. The indifference point predicted from TTO responses was significantly higher than that obtained directly through the PTO elicitation. In the PTO elicitation, the median participant thought that curing the life-threatening infection in 100 people with paraplegia was equally good as curing the life-threatening infection in 100 healthy people. However, based on participants' responses to the TTO elicitation, we predicted that it would take saving 130 lives of people with paraplegia to be just as good as saving 100 healthy lives. In fact 91% of participants placed equal value on saving the lives of both groups when asked directly in the PTO elicitation but only 25% of TTO responses implied this value.
Table 3 Comparison of Predicted TTO and Actual PTO Indifference Points
Median (Interquartile Range) Indifference Points:
Group Predicted from TTO responses Actual Response to PTO elicitation p6
Save – Save1 (n = 22) 130 (100–339) 100 (100–100) 0.001
Cure – Save2 (n = 22) 176 (100–583) 135 (100–3250) 0.320
Cure – Cure3 (n = 20) 323 (131-infin4) 650 (250-infin5) 0.736
1. The number of people with paraplegia whose lives would need to be saved to be equally good as saving the lives of 100 healthy people.
2. The number of people who would need to be cured of spinal cord injury to prevent paraplegia to be equally good as saving the lives of 100 healthy people.
3. The number of people who would need to be cured of moderate leg pain to be equally good as curing 100 people of severe shortness of breath.
4. Participants in the 75th percentile believed that living 10 years with severe shortness of breath was equivalent to living less than a day in perfect health.
5. Participants in the 75th percentile believed that all the people in the world would need to be cured of severe shortness of breath to be equally good as curing 100 people of moderate leg pain.
6. Based on the paired comparisons using the Wilcoxon signed rank test.
Why were the indifference points predicted from TTO responses larger than those obtained directly through the PTO elicitation? During the TTO elicitation, all participants voiced considerations that were consistent with QALY-maximization principles and only one participant mentioned a non-maximizing principle. In contrast, during the PTO elicitation, less than half (45%) mentioned QALY-maximization considerations, while two-thirds (68%) mentioned non-maximizing principles. The dominant consideration that emerged during the PTO elicitation was the belief that the lives of the people being traded-off were equal, regardless of pre-existing paraplegia. One man said, "I don't see leaning towards somebody because they can walk, valuing their life more than someone that would be in a wheelchair...Based on this description there isn't anything that says that I should lean towards one group or the other." Another man seemed to intertwine consideration of equality on the basis of saving a life and on the premise that the people with paraplegia were worse-off and should get higher weight, "They're already living life rough, you know what I mean, and then...not be able to save them?" Two participants wanted to avoid any taint of prejudice that may arise by choosing to cure the non-disabled group. One woman said, "I feel like I'm going to be prejudiced by picking this individual over that one." Another woman expressed strong emotion when asked to choose between curing the two groups, "Oh, this is horrible. This makes me feel like a bad person or something. I think I would choose the non-paraplegia... but...it doesn't make me feel good about myself...because these are people too and as many people may love them, as love these people. They may even have as many people dependent on them as these people." This participant later said, "...it's not different enough from saying 'Oh, well you're black or you're a woman or ... that kind of impermissible difference.' I guess [the two groups are] the same." Yet, this same participant was willing to trade-off years of life in exchange for perfect health when she responded to the TTO elicitation.
During the TTO elicitation, nearly half of participants (45%) took a personal perspective, imagining what it would be like to live in the condition themselves. Only two participants did so in the PTO elicitation. One man said, as he was responding to the TTO elicitation, "To have something like that happen to me, me personally ... if you only lived a day more. I wouldn't want to live like that" as he thought about living with paraplegia. Yet, many of the participants who personalized the scenario while responding to the TTO elicitation took a decidedly societal perspective in the PTO elicitation. The same man, as he responded to the PTO elicitation later said, "...They're already suffering...it's not fair to knock them off, you know what I mean? It wouldn't be right, it wouldn't be moral. If these people want to live like that, then you know what I mean? More power to 'em [sic]." One woman, after saying she'd prefer to live in perfect health even less than a day rather than live with paraplegia for 50 years in response to the TTO elicitation said, "When I think about having all your faculties up until the end, no matter what that point it is, I just prefer [living less than a day in perfect health]." However, later in the interview during the PTO elicitation, she said, "it's...like choosing to say who's life is more important – somebody who is fully functional or a person who is not. So, to me it's equal."
Cure-save group
We presented the same TTO scenario to this group of participants as presented to participants in the Save-Save group. But the PTO scenario was different. The baseline of comparison in the PTO elicitation was curing 100 healthy people of a life-threatening infection and we asked participants how many people would need to be treated for a spinal cord injury to prevent the onset of paraplegia to make them indifferent about which group to treat. We found no significant differences between actual PTO indifference points and those predicted from TTO responses. During the PTO elicitation, the median participant believed that treating 135 people with a spinal cord injury to prevent paraplegia was equally good as saving the lives of 100 healthy people. The median indifference point predicted from TTO responses implied 176 people with spinal cord injury would need to be treated to be equivalent. Based on actual PTO indifference points, 45% of participants believed that curing spinal cord injury was at least as valuable as saving the life of a healthy person. This stance was also implied by 41% of the indifference points predicted from TTO responses.
When responding to the TTO elicitation, all participants voiced considerations related to QALY-maximization principles and most (68%) took a personalized perspective, as seen in Table 5. Likewise, when responding to the PTO elicitation, most participants (82%) voiced considerations consistent with QALY-maximization principles. One man weighed length of life versus quality of life as he responded to the PTO elicitation saying, "With the other ones ... they're not in as good condition because ... they're confined to a wheel chair, they're limited in what their activities are...I think I would go with [curing] the infection, just because, the fact is, that could result in death. These people [with paraplegia]: it would [be] nice to have them cured, so they can get around leading a normal life but without the treatment, they still are here ...They can still lead some sort of a life." Many participants focused on living with paraplegia compared to dying from an infection as one man who said, "...You'd like to live as long as you can...being paraplegic at least you're still on the planet where you can laugh and you can enjoy life and things like that" and a woman said, "So, at least they're alive, where if we hadn't helped these people, in 48 hours with this infection they would die."
Table 5 Percentage of Participants Coded for Each Theme
Percentage of Participants:
Group: Save-Save (n = 22) Cure-Save (n = 22) Cure-Cure (n = 20)
Topic TTO PTO p* TTO PTO p* TTO PTO p*
Topics consistent with QALY-maximization principles
Quality of life 91 36 95 55 100 95
Length of life 55 5 55 55 50 5
Non-health benefits 18 14 36 9 15 5
Overall: 100 45 <0.001 100 82 0.13 100 95 1.00
Topics not related to QALY-maximization principles
Concern for the worse-off 0 5 0 0 0 0
Equality of life 5 59 5 0 0 10
Prejudice 0 9 0 5 0 5
Overall: 5 68 <0.001 5 5 1.00 0 15 0.25
Personalized the Scenario 45 9 0.004 68 27 0.01 45 25 0.10
% who said the TTO and PTO questions were different questions
64 59 35
• Using McNemar's paired comparison test.
Many participants in this group implied that living with paraplegia was at least as bad as death by saying that living less than one day with perfect health was equally good as living 50 years with paraplegia as they responded to the TTO elicitation. Some participants continued to reflect this belief as they responded to the PTO elicitation by saying that treating people with a spinal cord injury to prevent paraplegia would be equally good as saving the lives of healthy people. One woman said, "...Both result in something totally life changing...because it results in death, but this is just really bad too...They just seem...equally as important..." This woman went on to say, "...What if I were an emergency room doctor or something, and somebody came in with this infection [and] somebody came in with this [spinal cord injury], which one would I take first and help...Both result in something totally life changing... [One] results in death, but this is just really bad too, so I don't know. They just seem...equally as important. Even though one results in death."
Cure-cure group
The baseline of comparison for participants in the Cure-Cure group was curing 100 people of severe shortness of breath and we asked participants how many people would need to be cured of moderate leg pain to be equally good when responding to the PTO elicitation. The median participant believed that 650 people would need to be cured of moderate leg pain to be indifferent between treating the two groups when responding to the PTO elicitation. The median indifference point predicted from TTO responses implied that 323 people would need to be cured of moderate leg pain to be equivalent. Despite the seemingly large difference between the two values, it was not statistically significant. The non-significance is likely due to the exceedingly wide range of responses to both the TTO and PTO elicitations. Participants in the 75th percentile, when responding to the TTO elicitation, would rather live less than a day in perfect health than live 50 years with severe shortness of breath. Similarly, participants in the 75th percentile, when responding to the PTO elicitation, believed that every possible person should be cured of severe shortness of breath before anyone was ever cured of moderate leg pain. On the other end of the spectrum, two participants refused to trade any time in the TTO elicitation and two participants refused to choose a group to treat in the PTO elicitation; both sets of cases implied that curing either condition had equal value.
Our qualitative analysis revealed similar types of considerations across the two types of elicitations. All participants voiced considerations consistent with QALY-maximization when they responded to the TTO elicitation. One woman said, "I think that the shortness of breath would be...more suffering even though she's going to live longer and die peacefully." Another woman said, "I'd rather have pain in the leg than not be able to breathe properly." Naturally, because the TTO elicitation asked participants to trade-off years of life versus quality of life, directly, many more participants mentioned the importance of length of life as they responded to the TTO compared to the PTO elicitation saying, for example, "I think that a 30-year-old would value having more years of life, even if it involved having pain – daily pain."
All but one participant also voiced QALY-maximization considerations as they responded to the PTO elicitation too. One man said he would "still choose the shortness of breath because those people cannot function and the people with the leg pain can function; basically enjoy life with a little pain." Qualitatively, several participants said they wanted to be sure everyone was cured in the worse off group before anyone in the group with the less severe condition was cured. One man said, "The shortness of breath is...more severe...Moderate leg pain,... just about everybody has aches and pains. But that shortness of breath,...that's scary. So, I'd rather see everyone cured of the...shortness of breath." Another said, "I think that it would be more important to me to cure shortness of breath than to cure some moderate leg pain they may be having. I don't know, maybe I'm wrong to think that even six billion people, every person in the world had leg pain, I still think that it'd be more important to help the people with shortness of breath." A woman said, "To me it seems comparable to eliminating the common cold versus like getting rid of AIDS what would I choose? I would choose getting rid of AIDS...People die from that, they suffer terribly whereas this leg pain, it's a drag but it doesn't affect your whole life. Again, I would still cure one hundred and let the six billion deal with it."
Three participants (15%) voiced non-maximization principles while responding to the PTO elicitation, while none did so when responding to the TTO elicitation. Two participants specifically mentioned the importance of equality, but from different perspectives. One woman believed that everyone should have equal access to treatment, regardless of severity, "...Everybody is worthy of being alleviated from that pain." But another participant felt that everyone should have equal opportunity for a decent quality of life, "I feel like no matter what number I come up with for moderate leg pain, I'm gonna [sic] be left feeling guilty about ... all these people I know still have shortness of breath and can't make it from the bedroom to the bathroom". Another participant was concerned about inappropriately discriminating against treating a patient on the basis of their condition, "...They're both the same age, so you have to look at them equally...It's...like discriminatory if you put more weight towards the person that's going to die sooner,...if her death has nothing to do with the condition."
Nearly half of participants took a personalized perspective when responding to the TTO elicitation while only 25% did so as they responded to the PTO elicitation. When responding to the TTO elicitation, one man said, "I had leg pain ... I know some people that have shortness of breath and they can do a lot less than I can. They...might not have problems with their legs or anything but they can't...do anything. Walk a few steps, or else some of them have to carry their tanks around, their air tanks. So really, they're more hindered than...the person with the leg pain." He brought out considerations from a societal perspective as he responded to the PTO elicitation saying, "Well, it comes to a point, where would society be better off ... where you can cure more of the leg pain people, and then it would be more beneficial to society. I mean it sounds a little cruel but, part of those people could help take care of the people with severe shortness of breath."
Perceived differences between TTO and PTO responses
When we asked participants to reflect on the indifference point we predicted from their TTO response compared the actual indifference point they gave in response to the PTO elicitation, we heard a wide spectrum of responses across the groups. In the Cure-Cure group, only 35% of participants felt the two elicitations were asking different questions, as shown in Table 5. However, most participants in the Save-Save and Cure-Save groups thought the elicitations were asking different questions (59% and 64%, respectively). One man said, "I think you still are worse off with paraplegia, right? You are definitely worse off, 'cause like I said, there's [sic] things you can't enjoy... If you're ... a politician or ... a public health person, I think you should value life either way ... even though you're worse-off with paraplegia." A woman said, "They're not the same question...You're asking me in the first one what is my perception of the illness...In that first question I took it as how would I feel if I had that illness or how would it affect me. Whereas this situation is asking me is how I value the two subsets as people. And as people, I value them equally. So one is a quality of life issue and the other one is who is more worthy of saving. They're not related at all." Some simply echoed one man who said, "I know the numbers don't match, but even looking at them now, I still would say the same thing." Some expanded a bit more as one woman did, saying that her TTO response was based on "Me personally...everybody doesn't think like I do," but as she talked about the PTO elicitation she said, "...One of the reasons is that both, without the treatment they'll both die. That's why I said they were 50/50. I couldn't decide to save one or the other... I had to pick both."
Discussion
Based on our findings, non-maximizing principles explain part of the empirical differences seen in previous studies when values obtained from TTO responses were compared to those obtained from PTO elicitations. However, it is clear that context drives the extent to which these other principles come into play. Participants were clear in their belief that, within the context of choosing between life-saving treatments, having pre-existing paraplegia should not be a consideration. Though many participants took a decidedly personalized perspective when responding to the TTO elicitation, most did not carry this into the PTO elicitation and instead took a decidedly societal perspective. Even participants who preferred living less than a day in perfect health rather than live with paraplegia when responding to the TTO elicitation said that it was equally important to save the lives of both groups in the PTO elicitation. People were consistent in taking an egalitarian approach in the sense that everyone has a right to choose to live, regardless of their health condition. Our findings align with Williams's fair innings argument based on the belief that everyone deserves to live some normal length of life [39] and confirm Nord's [14] theory that people may feel free to make decisions about their own life but be reluctant to make life decisions for others. These findings are also empirically consistent with other studies where the median participant placed equal (or nearly equal) value on saving the lives of people with paraplegia or perfect health in PTO elicitations [22-24] and the importance of giving patients equal access to care [40].
Indifference points predicted from TTO responses were not significantly different from those obtained directly through the PTO elicitation when participants traded off curing spinal cord injury to prevent paraplegia versus saving a healthy life. Participants voiced considerations related to QALY-maximization principles while responding to both the TTO and PTO elicitation and they rarely mentioned other principles during either elicitation. Most previous studies comparing the PTO method to other traditional preference measures framed choices in terms of curing a less severe chronic condition than ones described in our study, versus saving the lives of previously healthy patients. In nearly all cases, contrary to our results, indifference points were higher when obtained through a PTO elicitation compared to those computed from a traditional utility elicitation method; this was true for the rating scale [18], visual analog scale [17], TTO [14,19,25], and standard gamble [14,17]. One reason for the lack of differentiation in our study may be because of the way we framed the PTO scenario. Rather than curing a pre-existing condition, we presented an opportunity to prevent new onset of paraplegia. Our findings do confirm a study done by the European Disability Weights group in which a substantial number (43%) of participants placed equal or higher value on preventing onset of quadriplegia compared to saving the lives of healthy people [41]. These findings may be related to the fact that the condition evaluated in the two studies (paraplegia or quadriplegia) involves severely impaired mobility. In two other studies, people also placed significantly less value on saving the lives of patients who would suffer new onset paraplegia compared to saving the lives of people with pre-existing paraplegia [22,23]. In a follow-up study, when participants were encouraged to consider their own ability to adapt to difficult situations, the relative valuation for a life-saving treatment with new onset paraplegia increased significantly [24]. This study provides qualitative confirmation that many people are focused on the initial trauma of having paraplegia in the context of this kind of elicitation. If people perceive the condition to be cured as highly traumatic, they will focus on the "awfulness" of the initial trauma such that, in the words of one of our participants, "both result in something totally life changing" when compared to the alternative of saving lives. There may be a theoretical threshold where as the condition to be cured becomes less severe, people focus on the imperative of saving lives instead. In this context, people's choices may become lexicographical when lives were at stake: lives typically trump cures [14,17-19].
Nearly all participants voiced only considerations that were consistent with QALY-maximization principles when asked to trade off years (TTO) to live in a less severe condition versus living longer with a more severe condition or when choosing which of two groups to cure (PTO) of a non-life-threatening condition. Indifference points predicted from TTO responses were comparable to indifference points obtained directly through the PTO elicitation. Some people think that more severely ill patients should be given higher priority even if they gain less benefit than treating less severely ill patients [4,18,19,22]. One study found that PTO valuations for curing patients who were worse-off were higher than those obtained from the rating scale [18]. Only one small-scale study compared the PTO to TTO preferences within the context of curing two non-life-threatening conditions. Dolan and Green found, qualitatively, that participants most often based their responses on what the difference in treatment benefit was likely to be [42]. Many of our participants expressed the belief that everyone should be cured of severe shortness of breath before anyone moderate leg pain was cured. As one participant stated, "if you've got a pain in the leg you can take an aspirin. If you can't breathe you can't breathe." Severe shortness of breath, the way we described it, was exceedingly more severe than the moderate leg pain we described. Taurek asserted that "the discomfort of each of a large number of individuals experiencing a minor headache does not add up to anyone's experiencing a migraine" [[43], page 308]. In the same vein, curing ever more people with moderate leg pain doesn't take away the suffering of a single person with severe shortness of breath. Several participants expressed this sentiment and this may help explain our findings. There may again be a threshold (working in the opposite direction) beyond which, if the difference in severity is pronounced enough, the PTO will result in higher indifference points than would be predicted by the TTO measure.
Nord [16] assumes that the PTO provides a social context within which to obtain preference weights and Olsen [15] maintains that the PTO implies social weights while the TTO implies private weights. Another study found qualitative evidence that some participants took a societal perspective in a PTO elicitation [42]. Our results confirm these assertions. Regardless of context, participants were more likely to take a personal perspective when responding to the TTO compared to the PTO elicitation. This is in spite of the fact that we framed the TTO elicitation from a non-personal perspective; we asked participants to evaluate two friends rather than imagine what it might be like to live with that condition personally. On the whole, participants in our study were still more likely to personalize the TTO compared to the PTO elicitation.
This study has several weaknesses. The sample size was small, with trends toward being unbalanced between the groups with respect to age and minority status. The actual indifference points are not generalizable because of the small convenience sample. In addition, our sample was limited to a Midwestern location in the U.S. The European Disability Weights study found that preferences elicited by a PTO scenario, similar to the one we presented to participants in the Save-Save group, varied widely across five European countries [41]. Additionally, we used a long time horizon (50 years) in the TTO elicitation to encourage more people to trade time for improved quality of life. However, indifference points may have been exceptionally high because people were more averse to living such a long time in the health condition being evaluated. This phenomenon may have contributed the lack of differentiation between TTO and PTO indifference points in two of three of the treatment contexts in our study though others have found differences in similar contexts. Despite differences in recruiting location, demographics, and structure of the interview, participants were consistent in their tendency toward taking a more personalized perspective when responding to the TTO compared to the PTO elicitation, regardless of context.
Conclusion
When trading off groups of patients within the context of PTO elicitations, respondents are more likely to take a societal perspective than when trading time within the context of TTO elicitations – even when the TTO elicitation is framed in a non-personal way. Furthermore, when lives are at stake within the context of a PTO elicitation, people are more likely to consider principles other than simply maximizing QALYs, including the importance of equal access to a life-saving treatment, avoiding prejudice or discrimination, and in rare cases giving treatment priority based purely on the position of being worse-off. However, the extent to which these non-maximizing principles are expressed depends on context.
Competing interests
Financial Disclosure: Financial support for this study was provided by the VA Ann Arbor Healthcare System, Ann Arbor, MI and NIH Grant #R01 HD40789. The funding agreement ensured the authors' independence in designing the study, interpreting the data, writing and publishing the report. The following authors are employed by the sponsor: Laura J. Damschroder and Peter A. Ubel. All funding agreements ensured the authors' independence in designing the study, interpreting the data, writing and publishing the report.
Authors' contributions
LJD participated in design, implementation, led the qualitative and quantitative analyses, and drafted the manuscript. TRR participated in qualitative analyses of the data. CCG carried out the study and participated in the qualitative analyses. MEM participated in design and carrying out the study. PAU participated in design of the study, participated in qualitative analyses, and consulted on the analyses. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Appendix 1 PDF file with detailed elicitation text that was presented to participants for the TTO and PTO elicitations.
Click here for file
Acknowledgements
We could not have completed this study without all the people who willingly participated. We thank Brian Zikmund-Fisher for helpful comments on an earlier draft, Susan Metosky for contributing her interviewing skills, and for the very helpful comments received from David Schwappach, Michael Schwarzinger, and Joshua Salomon.
==== Refs
Gold MR Siegel JE Russell LB Weinstein M Cost-Effectiveness in Health and Medicine 1996 New York, Oxford University Press
Hadorn DC The problem of discrimination in health care priority setting Journal of the American Medical Association 1992 268 1454 1459 1387422 10.1001/jama.268.11.1454
Nord E The person trade-off approach to valuing health care programs Medical Decision Making 1995 15 201 208 7564933
Nord E The trade-off between severity of illness and treatment effect in cost-value analysis of health care Health Policy 1993 24 227 238 10128127 10.1016/0168-8510(93)90042-N
Harris J QALYfying the value of life Journal of Medical Ethics 1987 13 117 123 3669036
Hadorn DC Setting health care priorities in Oregon. Cost-effectiveness meets the rule of rescue Journal of the American Medical Association 1991 265 2218 2225 1901610 10.1001/jama.265.17.2218
Jonsen AR Bentham in a box: Technology assessment and health care allocation Law, Medicine & Health Care 1986 14 172 174
Rawls J A Theory of Justice 1971 Cambridge, Harvard University Press 607
Winslow GR Triage and Justice 1982 Berkeley, University of California Press 228
Miller D Social Justice 1976 Oxford, Clarendon Press
Nord E Pinto-Prades JL Richardson J Menzel P Ubel PA Incorporating societal concerns for fairness in numerical valuations of health programmes Health Economics 1999 8 25 39 10082141 10.1002/(SICI)1099-1050(199902)8:1<25::AID-HEC398>3.3.CO;2-8
Ubel PA Nord E Gold M Menzel P Prades JL Richardson J Improving value measurement in cost-effectiveness analysis Medical Care 2000 38 892 901 10982111 10.1097/00005650-200009000-00003
Nord E The QALY--a measure of social value rather than individual utility? Health Economics 1994 3 89 93 8044215
Nord E Methods for quality adjustment of life years Social Science & Medicine 1992 34 559 569 1604362 10.1016/0277-9536(92)90211-8
Olsen JA Persons vs years: Two ways of eliciting implicit weights Health Economics 1994 3 39 46 8167798
Nord E Richardson J Macarounas-Kirchmann K Social evaluation of health care versus personal evaluation of health states. Evidence on the validity of four health-state scaling instruments using Norwegian and Australian surveys International Journal of Technology Assessment in Health Care 1993 9 463 478 8288424
Pinto-Prades JL Is the person trade-off a valid method for allocating health care resources? Health Economics 1997 6 71 81 9142532 10.1002/(SICI)1099-1050(199701)6:1<71::AID-HEC239>3.0.CO;2-Z
Ubel PA Loewenstein G Scanlon D Kamlet M Value measurement in cost-utility analysis: Explaining the discrepancy between rating scale and person trade-off elicitations Health Policy 1998 43 33 44 10178799 10.1016/S0168-8510(97)00077-8
Ubel PA Loewenstein G Scanlon D Kamlet M Individual utilities are inconsistent with rationing choices: A partial explanation of why Oregon's cost-effectiveness list failed Medical Decision Making 1996 16 108 116 8778528
Lindholm L Rosen M Emmelin M An epidemiological approach towards measuring the trade-off between equity and efficiency in health policy Health Policy 1996 35 205 216 10157398 10.1016/0168-8510(95)00783-0
Richardson J Nord E The importance of perspective in the measurement of quality-adjusted life years Medical Decision Making 1997 17 33 41 8994149
Ubel PA Richardson J Pinto-Prades JL Life-saving treatments and disabilities. Are all QALYs created equal? International Journal of Technology Assessment in Health Care 1999 15 738 748 10645115
Ubel PA Richardson J Baron J Exploring the role of order effects in person trade-off elicitations Health Policy 2002 61 189 199 12088891 10.1016/S0168-8510(01)00238-X
Damschroder LJ Zikmund-fisher BJ Ubel PA The impact of considering adaptation in health state valuation Social Science & Medicine 2005 61 267 277 15893044 10.1016/j.socscimed.2004.11.060
Salomon JA Murray CJ A multi-method approach to measuring health-state valuations Health Econ 2004 13 281 290 14981652 10.1002/hec.834
Martin AJ Glasziou PP Simes RJ Lumley T A comparison of standard gamble, time trade-off, and adjusted time trade-off scores International Journal of Technology Assessment in Health Care 2000 16 137 147 10815360 10.1017/S0266462300161124
Bleichrodt H Luis Pinto JL Abellan-Perpinan JM A consistency test of the time trade-off Journal of Health Economics 2003 22 1037 1052 14604559 10.1016/S0167-6296(03)00046-8
Torrance GW Utility approach to measuring health-related quality of life Journal of Chronic Diseases 1987 40 593 603 3298297 10.1016/0021-9681(87)90019-1
Chapman G Elstein A Kuzel T Sharifi R Nadler R Andrews A Bennett C Prostate cancer patients' utilities for health states: How it looks depends on where you stand Medical Decision Making 1998 18 278 286 9679992
Lenert LA Cher DJ Goldstein MK Bergen MR Garber A The effect of search procedures on utility elicitations Medical Decision Making 1998 18 76 83 9456212
Melsop KA Boothroyd DB Hlatky MA Quality of life and time trade-off utility measures in patients with coronary artery disease American Heart Journal 2003 145 36 41 12514652 10.1067/mhj.2003.37
Percy ME Llewellyn-Thomas H Assessing preferences about the DNR order: does it depend on how you ask? Med Decis Making 1995 15 209 216 7564934
Willis GB Royston P Bercini D The Use of Verbal Report Methods in the Development and Testing of Survey Questionnaires Applied Cognitive Psychology 1991 5 251 267
Ericsson KA Simon HA Protocol Analysis: Verbal Reports as Data 1993 Revised Cambridge, MIT Press 443
Payne JW Braunstein ML Carroll JS Exploring Predecisional Behavior: An Alternative Approach to Decision Research Organizational Behavior and Human Performance 1978 22 17 44 10.1016/0030-5073(78)90003-X
Stone DN Schkade DA Numeric and linguistic information representation in multiattribute choice Organizational Behavior and Human Decision Processes 1991 49 42 59 10.1016/0749-5978(91)90041-Q
Russo JE Johnson EJ Stephens DL The validity of verbal protocols Memory & Cognition 1989 17 759 769 2811673
Sandelowski M What ever happened to qualitative description? Research in Nursing & Health 2000 23 334 340 10940958 10.1002/1098-240X(200008)23:4<334::AID-NUR9>3.0.CO;2-G
Williams A Intergenerational equity: an exploration of the 'fair innings' argument Health Economics 1997 6 117 132 9158965 10.1002/(SICI)1099-1050(199703)6:2<117::AID-HEC256>3.0.CO;2-B
Dolan P Cookson R A qualitative study of the extent to which health gain matters when choosing between groups of patients Health Policy 2000 51 19 30 11010223 10.1016/S0168-8510(99)00079-2
Schwarzinger M Stouthard ME Burstrom K Nord E Cross-national agreement on disability weights: the European Disability Weights Project Popul Health Metr 2003 1 9 14633276 10.1186/1478-7954-1-9
Dolan P Green C Using the person trade-off approach to examine differences between individual and social values Health Economics 1998 7 307 312 9683091 10.1002/(SICI)1099-1050(199806)7:4<307::AID-HEC345>3.3.CO;2-E
Taurek JM Should the numbers count? Philosophy & Public Affairs 1977 6 293 316 11662404
|
16281982
|
PMC1310516
|
CC BY
|
2021-01-04 16:37:42
|
no
|
Popul Health Metr. 2005 Nov 10; 3:10
|
utf-8
|
Popul Health Metr
| 2,005 |
10.1186/1478-7954-3-10
|
oa_comm
|
==== Front
J NeuroinflammationJournal of Neuroinflammation1742-2094BioMed Central London 1742-2094-2-261628751110.1186/1742-2094-2-26ResearchCytokine responses during chronic denervation Ruohonen Saku [email protected] Mohsen [email protected] Maja [email protected] Hanna-Stiina [email protected] Tomas [email protected]öyttä Matias [email protected] Department of Pathology, University of Turku, Kiinanmyllynkatu 10, 20520 Turku, Finland2 Department of Neuroscience, Karolinska Institute, 17176 Stockholm, Sweden3 Department of Handsurgery, Turku University hospital, Kiinanmyllynkatu 10, 20520, Turku, Finland2005 18 11 2005 2 26 26 29 8 2005 18 11 2005 Copyright © 2005 Ruohonen et al; licensee BioMed Central Ltd.2005Ruohonen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The aim of the present study was to examine inflammatory responses during Wallerian degeneration in rat peripheral nerve when the regrowth of axons was prevented by suturing.
Methods
Transected rat sciatic nerve was sutured and ligated to prevent reinnervation. The samples were collected from the left sciatic nerve distally and proximally from the point of transection. The endoneurium was separated from the surrounding epi- and perineurium to examine the expression of cytokines in both of these compartments. Macrophage invasion into endoneurium was investigated and Schwann cell proliferation was followed as well as the expression of cytokines IL-1β, IL-10, IFN-γ and TNF-α mRNA. The samples were collected from 1 day up to 5 weeks after the primary operation.
Results
At days 1 to 3 after injury in the epi-/perineurium of the proximal and distal stump, a marked expression of the pro-inflammatory cytokines TNF-α and IL-1β and of the anti-inflammatory cytokine IL-10 was observed. Concurrently, numerous macrophages started to gather into the epineurium of both proximal and distal stumps. At day 7 the number of macrophages decreased in the perineurium and increased markedly in the endoneurium of both stumps. At this time point marked expression of TNF-α and IFN-γ mRNA was observed in the endo- and epi-/perineurium of the proximal stump. At day 14 a marked increase in the expression of IL-1β could be noted in the proximal stump epi-/perineurium and in the distal stump endoneurium. At that time point many macrophages were observed in the longitudinally sectioned epineurium of the proximal 2 area as well as in the cross-section slides from the distal stump. At day 35 TNF-α, IL-1β and IL-10 mRNA appeared abundantly in the proximal epi-/perineurium together with macrophages.
Conclusion
The present studies show that even during chronic denervation there is a cyclic expression pattern for the studied cytokines. Contrary to the previous findings on reinnervating nerves the studied cytokines show increased expression up to 35 days. The high expressions of pro-inflammatory and anti-inflammatory cytokines in the proximal epi-/perineurial area at day 35 may be involved in the formation of fibrosis due to irreversible nerve injury and thus may have relevance to the formation of traumatic neuroma.
==== Body
Background
During Wallerian degeneration, macrophages enter the peripheral nervous system [1-3] when permeability of the BNB is increased. The increased permeability of BNB leads to the increased infiltration of macrophages into the endoneurium [4,5]. Injury-activated Schwann cells and resident macrophages are probably partly responsible for the recruitment of hematogenous macrophages by secreting MCP-1 [6] and the pro-inflammatory cytokines IL-1, IL-6, IL-12; and especially TNF-α, in the early phase of the inflammatory reaction [7-13]. The main population of macrophages in the peripheral nerve is blood-derived after injury. These gather in the endoneurium to start phagocytosis of axons and myelin stumps [14,15]. Macrophages and Schwann cells mediate and promote inflammation after injury by production of pro-inflammatory cytokines, but they have also positive effects on neurotrophic factors, which can induce axonal sprouting [16]. Macrophages as well as Schwann cells express also anti-inflammatory cytokines such as IL-10 and TGF-β1 to inhibit inflammation [17-21]. The role of TGF-β1 during nerve injury seems to be controversial. It increases neuronal regeneration [19,22] but decreases NGF production [23-25] and kills Schwann cells together with TNF-α [26].
In the present study permanent nerve damage was induced by suturing the distal and proximal ends of the transected sciatic nerve in order to create a model of chronic denervation. Since relatively small amounts of cytokine mRNAs are present in the nerve stumps, real-time polymerase chain reaction was used to facilitate detection of the studied cytokines. Additionally, morphological differences were followed to investigate possible correlations with produced cytokines.
Materials and methods
Experimental animals
Young adult male Sprague-Dawley rats (n = 70) were used in the present study. The animals were kept in the Turku University Animal Centre. (Circadian 12-h rhythm, T = 21 ± 1°C, humidity 50 ± 5%)
Normal daily care was provided with nutrition (Chow Lactamin R36, Södertälje, Sweden) and water ad libitum. The present study was approved by the Committee for Ethical Animal Experiments (permission no. 1080/01).
Operative procedures
The left sciatic nerves were exposed and transected at the level of hip joint under pentobarbital (Mebunat®) anesthesia. Regeneration at the left sciatic nerve was prevented with suturing of distal and proximal stumps beside the point of transection [27]. The right sciatic nerve was left intact. Additionally, normal control nerve samples were collected from normal, unoperated rats of the same age. Samples were collected 1 day, 3 days, 5 days, 7 days; and 2, 3, 4 and 5 weeks after the primary operation. For the subsequent biochemical studies six rats were sacrificed at each time point, and two at each time point for the immunohistochemical studies. The sciatic nerves of three normal rats were studied as negative controls. The rats were perfused intracardially with sterile 0.9% saline or with 4% phosphate-buffered formalin. The nerves were cut in 4-mm sections in freezing conditions (on a covered Petri dish filled with ice).
Real-time PCR and histological samples
The samples were taken both proximally and distally from the point of transection. To avoid local damage and/or sutures beside the point of transection, the initial 1-mm section beside the point of transection was discarded. For real-time PCR and immunohistochemical analysis two 4-mm sections (P1, P2) were cut starting 1 mm proximally from the point of transection. Also, distally two sections were cut, one (D1) starting immediately 1 mm from the point of transection and the other (D2) starting 5 mm distally from the point of transection (Fig. 1). Furthermore, for the real-time PCR studies, the endoneurium was separated from the surrounding peri- and epineurium (Fig. 2) [28]. All real-time PCR samples were immediately immersed in a GITS solution, frozen to -196°C with liquid nitrogen, and stored at -70°C. The samples were pooled from six different rats for each time point to obtain more homogenous material for the PCR studies.
Figure 1 Transection point. Left sciatic nerves were exposed and transected at the level of the hip joint under pentobarbital anaesthesia. Regeneration of the left sciatic nerve was prevented by suturing of distal and proximal stumps beside the point of transection. The right sciatic nerve was left intact. For biochemical and morphological analysis two 4-mm areas (P1, P2) were cut starting 1 mm proximally to the point of transection. Also, distally two sections were cut, one (D1) starting immediately 1 mm from the point of transection and the other (D2) starting 5 mm distal to the site of transection.
Figure 2 Separation of endo-and epi-/perineurium. For biochemical studies, endoneurium was separated from the surrounding peri- and epineurium. This was done on the icy cover of a Petri dish filled with ice. The endoneurium was pulled out with fine tip forceps.
Immunohistochemistry
The animals for morphological samples were perfusion-fixed with 4% phosphate-buffered formalin and whole peripheral nerve samples were exposed and fixed overnight in 4% phosphate-buffered formalin. The specimens were embedded in paraffin and 4-μm sections were cut for immunohistochemical analyses. The paraffin sections were treated with xylene and decreasing alcohol solutions to remove paraffin, and then hydrated and digested with 0.4% pepsin in 0.01 M HCl for 60 min. Endogenous peroxidase activity was blocked using 0.3% H2O2 in methanol, after which sections were incubated for 60 min with horse serum to prevent non-specific staining. The sections were then treated with monoclonal antibody ED-1 (Serotec 0591) for monocytes/macrophages or S-100 (Dakopatts) for Schwann cells overnight at 4°C. Bound antibodies were demonstrated using an avidin-biotin method with a Vectastain ABC kit according to the manufacturer's instructions. Additionally, normal rat serum was used with the second antibody to diminish non-specific staining.
The recruitment of macrophages was evaluated microscopically. Semi-quantification of changes was performed as in our previous studies [1,29]. In the endoneurium the visual evaluation of macrophages was done with a scale of four: 0 = only occasional macrophages similar to that seen in the control nerve; + = a few macrophages, no focal accumulation of macrophages; ++ = several macrophages with focal accumulation, phagocytotic activity present; +++ = numerous macrophages with phagocytotic activity. Similar principles were used in the epineurium. However, this evaluation was done with a scale of three: 0 = only occasional macrophages similar to that seen in the control nerve; + = a few macrophages per cross section in the epineurium; ++ = over 10 macrophages per cross-section in the epineurium.
Schwann cells were assessed using similar techniques. However, the proliferation pattern of Schwann cells was similar to that found in previous studies [29]. Therefore, the data from Schwann cells is not shown.
Immunofluorochemisty
For immunofluorochemical studies veins of animals were stained with FITC-Dextran (Sigma, FD2000S) iv-injection. The amount of FITC-Dextran was titrated to 100 mg/kg to show the veins of the sciatic nerve in a whole-mount experiment. Thirty minutes after injection the animals were perfused with saline and 4% phosphate-buffered formalin. Samples were taken distally and proximally from the point of transection. The nerve stumps were fixed overnight in 4% phosphate-buffered formalin and then transferred to 70% EtOH. The nerve stumps were cut into four longitudinal sections before washing with TTBS and Triton-X permeabilization. BSA (1%) was used to prevent non-specific staining. The sections were then treated with monoclonal antibody ED-1 (1:200; Serotec 0591) 2 h at 37°C. Bound antibodies were demonstrated using goat-anti-mouse-Alexa 555 antibody (1:200; Molecular Probes). The numbers of ED-1 positive cells were compared to that observed in control nerve from unoperated rats, using cross-sectional samples from the same area. The samples were studied by confocal microscope (LSM 510, Zeiss Axiovert 200).
Determination of cytokine mRNA from endo- and epi/perineurium
Pooled and frozen nerve stumps were first homogenized with Ultra-Turrax. The RNA preparation was modified from Chomczynski and Sacchi [30]. RNA was extracted using Trizol (Gibco BRL 15596-018) according to the acid phenol-quanidium thiocyanate-chloroform extraction method. mRNA was purified from total RNA, and then reverse-transcribed to cDNA. The reverse transcription was performed using 10 μg of mRNA, and Superscript reverse transcriptase (200 U; Life Technologies). The samples were then measured with real-time PCR.
cDNA amplification was performed with the ABI PRISM 7700 Sequence Detector (Applied Biosystems) with a two-step PCR protocol (preincubation 10 min at 95°C followed by 40 cycles at 95°C for 15 sec and at 60°C for 1 min). All cytokine primers and probes were designed using Primer Express software (Applied Biosystems) avoiding contaminating genomic DNA amplification by positioning one of the primers or a probe over the exon/intron boundary. The probes were labeled with FAM at the 5' end as reporter dye and TAMRA at the 3' end as quencher dye. The GAPDH gene was used as a stable endogenous control.
Absolute quantification was performed using a standard curve method. Standard curves were constructed using three 100-fold dilutions of standard sample (for IL-1β the standards were plasmids containing the corresponding cytokine gene; for GAPDH, TNF-α and IL-10 the standards were cDNA from concanavalin-A stimulated rat lymph node cells) and corresponding cycles of threshold (Ct) value. The samples were run in triplicate for target cytokine and endogenous GAPDH controls. After computing the relative amounts of target cytokine and endogenous control for one sample, the final amount of cytokine in that sample was presented as a ratio between and the amount of cytokine and amount of endogenous control, GAPDH. For real-time PCR, quantitative results are available directly after PCR without additional purification or analysis steps. Protein levels of cytokines were not studied.
The results were imported into Microsoft Excel and legends were made with GraphPad Prism. SEM was calculated with one-way ANOVA and t-test, which were made with Microcal Origin. The source SEM does not reflect differentiation between individual animals but methodological accuracy.
Results
Results of real-time PCR studies
IL-1β (Figure 3.)
Figure 3 Cytokine expression. Cytokine expressions of IL-1β, IL-10, IFN-γ and TNF-α after transection. Statistical analyses were made with one-way ANOVA and t-test. Only those expressions that had t-values > 0 (relative to control expression) were taken into account. Significance values are marked as follows: * p = 0.05; ** p = 0.01; *** p = 0.001. The expression changes of different cytokines mRNAs could be observed at multiple time points after peripheral nerve injury. Expressions in control samples could only be observed in the epi-/perineurium. The proximal stump expressed IL-1β, TNF-α and IL-10 in a cyclic manner. The main time points for the cyclic expression pattern were 1-5, 14-21 and 35 days. The distal stump showed marked expression of these cytokines during days 1-3 after injury, but for cytokine IL-1β also at day 14. IL-10 was markedly expressed at 21 and 35 days after injury. IFN-γ was not expressed in either compartment before day 5. The proximal stump showed marked expression of this cytokine at 14 days after injury and the distal stump at 21 days. IL-4 was not markedly expressed in this study.
Controls
Very low expression of IL-1β mRNA was observed only in the epi/perineurium of the control samples.
Proximal areas
In the endoneurium the strongest expressions of IL-1β mRNA were observed in both proximal areas (P1 and P2) at 24 hours and at day 21 after injury. At other time points the expression remained low.
The epi/perineurium expressed IL-1β mRNA in a cyclic manner in proximal areas. Three expression peaks were observed: the first one during days 3 and 5, the second one at day 14 (proximal 2) and the third and strongest one at day 35. Proximal 2 areas showed lower expression of cytokines than the proximal 1 area.
Distal areas
The endoneurium showed a slight increase in the expression of IL-1β starting from day 1 until day 7 after injury. The expression rose again dramatically (110-fold) at day 14. Otherwise the expression of IL-1β mRNA was low.
In the epi/perineurium marked expressions were observed during days 1 and 3 in both distal areas. At the other time points the expression stayed at the control level.
TNF-α (Figure 3.)
Controls
Very low TNF-α mRNA expression was observed in the epi/perineurium but not in the endoneurium.
Proximal areas
The endoneurium showed strong TNF-α mRNA expression in the first proximal area at 24 hours. After that only small peaks of expression were noted at days 7, 14, 21 and 35.
The epi/perineurium expressed TNF-α mRNA in a fashion similar to that of the endoneurium. The first proximal areas epi/perineurium showed marked expression of TNF-α at 24 hours. However, the most pronounced expression was noted at day 35 in the proximal 2 area.
Distal areas
In the endoneurium the strongest expression of TNF-α mRNA was noted during the first days after injury but was at the control level at day five. The expression rose slightly at days 7, 14 and 21 in both distal areas, after which only low expression was observed.
In the epi/perineurium the expression pattern of TNF-α mRNA followed the pattern observed in the endoneurium but with increased expressions. From day 5, the expression stayed at the control level.
IFN-γ (Figure 3.)
Controls
Very low expression of IFN-γ mRNA was observed only in the epi/perineurial control samples but not in the endoneurium.
Proximal areas
The endoneurium did not show IFN-γ mRNA until day 5 after which it declined. The most marked expression of IFN-γ mRNA was noted at day 14 after injury.
In the epi-/perineurium only the first proximal area showed some peaked expression of IFN-γ mRNA and only at days 7 and 28 after injury.
Distal areas
The endoneurium showed marked expression of IFN-γ mRNA in the distal 1 area at day 21, but some expression was also noted at day 28 in both distal stumps. Also some expression was noted at days 5 and 14.
The epi/perineurium showed increased expression of IFN-γ only at days 5 and 21 after injury.
IL-10 (Figure 3.)
Controls
Only slight IL-10 mRNA expression was noted in the epi-/perineurium of the operated sciatic nerves but not in the endoneurium.
Proximal areas
The endoneurium showed increased expression of IL-10 at 24 hours and at days 7 and 14 after injury.
In the epi/perineurium IL-10 mRNA was expressed already at 24 hours but the most marked expression was seen at day 35 (900-fold) compared to controls after injury.
Distal areas
The endoneurium showed a marked increase in expression only at 24 hours. Slight expression was noted at days 14, 21 and 35.
The epi/perineurium showed marked peaks of expression at days 1, 21 and 35; otherwise the expression remained at the control level.
Results of morphological studies
Macrophages (Figures 4A,B,C,D,E,F)
Figure 4 Macrophages. (A-C) The areas marked with asterisk (*) are shown with a higher magnification (×420). Black arrows indicate macrophages. (A) At day 3 macrophages are present in the epineurium (EP) of distal 1 area (×210). A few ED-1 positive cells are also seen in the endoneurium (EN). (B) In distal 1 area at day 14 there are numerous macrophages in the endoneurium (×210). (C) Some macrophages are still present in both epi- and endoneurium in the proximal 2 area at 35 days (× 210). (D-F) Longitudinal sections studied by confocal microscope. Macrophages are visualized with red color, endoneurial vessels with green, and yellow color indicates macrophages inside blood vessels. White arrows indicate the epineurial area. The pictures are focused to the endoneurial level in the middle. (D) At day 14 several macrophages can be observed in the epineurial and endoneurial area of proximal 2 area (× 120). (E) At day 35 several macrophages are still present in the epineurium of proximal 2 area, but start to decrease in the endoneurium (× 120). (F) Control from sciatic nerve (non-operated control animal) (× 120).
In the distal and proximal stump the epineurial macrophages could be observed at day 3, (++) (Fig. 4A). Invasion of macrophages into the endoneurial space took place at days 5 and 7 in the proximal and distal stumps, (++) [1]. At days 14 (Fig. 4B) and 21 several macrophages were observed in the distal endo- and epineurium, (endoneurium +++; epineurium ++). At this time point the proximal stump seemed to have only a few MHC II-expressing cells in cross-section samples (+). However, confocal-microscopically studied longitudinal sections of proximal 2 area revealed focal accumulation of macrophages in the epineurial area (Fig. 4D) compared to normal control nerves (Fig. 4F). After this time point only a few macrophages were observed in the distal stump. However, occasional macrophages could still be observed in the proximal and distal endo- (++) and epineurium (+) at day 35 in cross-sections (Fig. 4C), as well as in the longitudinal samples studied by confocal microscope (Fig. 4E).
Discussion
To achieve a model for a chronic nerve injury, and in an attempt to provoke marked and long-lasting immunological changes, transected sciatic nerve was sutured and ligated to prevent reinnervation. In order to better understand the possible dynamics in different compartments within the peripheral nerve, the endoneurium was separated from the surrounding epi/perineurium (Fig. 2) [28] and these compartments were analyzed separately. We have previously shown that after peripheral nerve transection a marked number of macrophages appear first in the epineurium, after which they enter the endoneurium [1]. Hence, the expression of cytokines may differ in these peripheral nerve compartments due to expression of cytokines such as IL-1β and TNF-α, which affect vascular permeability [31] and lead to increased infiltration of macrophages into the endoneurial space during Wallerian degeneration. Our previous studies on peripheral nerve degeneration and regeneration [29,32,33] indicate that when nerve stumps are sutured to prevent regeneration, the most prominent source for endoneurial later-phase expression of cytokines could be Schwann cells and possibly fibroblasts. These cell types are significantly present during the later time points when monocytes/macrophages are already almost absent in the endoneurium. The main source for marked pro-inflammatory cytokine expression in the epi-/perineurium would be macrophages. But, mast cells may also present a potent source for cytokines in the peripheral nervous system [34].
The main sources of IL-1β after peripheral nerve injury are macrophages and Schwann cells [35,36]. In the present study macrophages were numerous in the proximal perineurium showing simultaneous high expression of IL-1β mRNA in the epi-/perineurium at days 1-5. The expression of IL-1β rose to a peak at days 14 and 35 after injury in the epi-/perineurium. At day 14 in the epineurial cross-section samples macrophages were almost absent. However, in the longitudinal samples studied by confocal microscope, focal accumulation of macrophages was seen in this particular area. The distal stump showed marked expression of IL-1β in the endoneurium at day 14 and at this time point numerous macrophages as well as Schwann cells were present in the distal endoneurium.
We have previously observed that, after classical Wallerian degeneration and regeneration, TNF-α mRNA is upregulated in the distal areas from 14 hours to day 5 and at day 14. In the proximal areas TNF-α is expressed from 14 hours to day 1 and again at day 5 [32]. The present results with sutured nerves showed a similar pattern of expression. TNF-α mRNA expression increased rapidly at 24 hours both in the endo- and epi/perineurium of distal and proximal areas. TNF-α has several inflammation-mediating capabilities during peripheral nerve injury [37-40]. Thus, marked rapid activation is needed for stimulating inflammatory reactions involving macrophages [41,42]. This includes the infiltration of macrophages into the endoneurium when BNB is broken down [43] at the beginning of Wallerian degeneration [44] where TNF-α shows pathogenic actions [8]. We found a mild increase in TNF-α mRNA expression at day 14 in the distal area. This augmentation of expression could be related to increased vascular permeability at two weeks [5,31]. Also, the number of Schwann cells starts to decrease at this time point [29] and thus an explanation for this distal TNF-α mRNA expression at 14 days could be the ability of TNF-α to induce Schwann cell apoptosis [45]. In the present study, however, the expression of TNF-α in the distal stump was relatively much lower at day 14 compared to our previous studies on freely regenerative nerves. This indicates that the arrival of axons to the distal stump would also be a possible starter of apoptosis of Schwann cells and hence, an inducer of expression of TNF-α in the distal area. The most remarkable TNF-α mRNA expressions were noted at day 35 in the epi-/perineurium of the proximal area 2. This peak of expression has not been observed in freely regenerative nerve models [32,46]. The reason for this is not known.
In the present study the expression of IL-10 mRNA increased rapidly in proximal and distal sections at 24 hours after injury both in the endo- and epi/perineurium, most likely as an inhibitory response to inflammatory reactions that TNF-α and IL-1β had provoked [17,47,48]. The source for IL-10 expression at the very beginning after injury cannot be infiltrating macrophages [49] because these cells are not usually observed before day 3 in the endoneurium. The most probable source at this time would be Schwann cells [20]. However, the resident activated macrophages that are normally present in the endoneurium are also activated during the inflammation [7,9] and may be responsible for the observed expression. Infiltrating hematogenous macrophages are probably partially responsible for the high IL-10 expression noted at day 21 in the distal epi-/perineurium, and in the endoneurium this expression would be augmented by Schwann cells [20,36]. An interesting observation was that from day 1 onwards the expression of IL-10 mRNA was mainly found in both distal and proximal epi/perineurium. At day 28 the expression of IL-10 was minimal in the distal epi-/perineurium. However, despite the low expression of TNF-α, IL-1β and IL-10 at this time point, the distal stump presents several morphological findings [27,50]. Endoneurial fibroblasts, which are markedly present in this area, form minifascicle-like structures, which possibly support basal lamina tubes from collapsing [27]. Interestingly the expression of IL-10 mRNA was markedly increased at day 35 in distal and proximal epi/perineurium. These changes could be related to the noted simultaneous high expressions of the two pro-inflammatory cytokines (IL-1β and TNF-α). IL-10 is known to inhibit the production of these inflammation-related cytokines [17,49,51]. IL-10 diminishes apoptosis provoked by TNF-α [45] and down-regulates MHC class II expression in macrophages/monocytes [49]. The producers of IL-10 during this later phase are at least partially macrophages [20,49], which were simultaneously present with IL-10 expression in the epineurium in the present study. However, mast cells [52] are also capable of producing IL-10. Why this expression of cytokines in the epi/perineurium takes place at day 35 is unknown. The present experimental design, with prevention of axonal reinnervation, may provoke these changes. Thus one cannot exclude the possibility that the observed expression at day 35 may be related to structural reorganization of peripheral nerve compartments.
In the present study IFN-γ mRNA was present after day 5. This observation can be temporally linked to macrophages, which enter the endoneurium [1,32]. In neuroinflammatory processes IFN-γ, which is primarily released by T-lymphocytes, has been noted to have an important role as an upregulator of MHC class II antigen expression [53]. IFN-γ increases influx of T cells and macrophages into the peripheral nervous system [54], and increases TNF-α and IL-1 production in macrophages [55] and fibroblasts [56]. By this action IFN-γ would induce a massive infiltration of macrophages into endoneurium 3 to 5 days after injury. The marked expression of IFN-γ mRNA at day 14 in the proximal endoneurium is also the time point at which a marked invasion of endoneurial blood-derived macrophages occurs in the proximal stump (Figure 4B.)
At day 35 several cytokines showed augmented expressions in the proximal 2 area in the present study. When axonal regeneration is prevented by suturing the peripheral nerve, a traumatic neuroma starts to form. A component of this neuroma is epineurial fibrosis, which is increased by TGF-β1 [57]. The expression of TGF-β1 increases dramatically in the proximal 2 area epineurium at day 35 during neuroma formation [58]. In the present study, a similar increase in expression of cytokines IL-1β, IL-10 and TNF-α was observed. This finding has not been reported previously.
When axonal sprouts start to grow in the proximal stump, some of them could accidentally start to grow backwards in the epineurium. However, these axons do not extend this growth very far, and sprouting is rapidly disabled. The disabled growth of axons in the epineurium has not been clarified but TGF-β1 is of special interest in this case. It has a function as a mitogen for Schwann cells [59] but it is also capable of killing them by activating c-Jun [60]. Activation of c-Jun leads also to increased expression of TNF-α, which is also capable of inducing Schwann cell apoptosis [26] as well as neuronal death [45,61].
Interestingly, non-injured contralateral sciatic nerve shows marked endoneurial expression of cytokines IL-1β, IL-10 and TNF-α mRNAs at day 35 after injury [62]. However, at this same time marked expressions of these cytokines at the site of injury were now observed only in the epi-/perineurium. What is the mechanism behind these paradoxical findings? This observation could indicate that the observed changes in expression of cytokines at the site of injury can spread through humoral stimulus via circulation. Activated macrophages have also been observed in the endoneurium of the contralateral side after injury [9]. Could the observed endoneurial changes in the contralateral nerve be a consequence of macrophage stimulation from the ganglions, brain, medulla or hypothalamus? After chronic constriction injury, blocking of NMDA-receptor was found to eliminate mRNA cytokine expression in contralateral sciatic nerve [63]. The biological significance of this finding is unknown.
Conclusion
Our results show that prolonged inflammatory mediator changes take place in peripheral nerve in a chronic denervation injury model. The expression is cyclic and correlates partly with those noted in the non-injured contralateral sciatic nerve. This study also supported previous suggestions that Schwann cells may have a central role in the expression of cytokines in nerve [36], which could be partly responsible for pain behavior as well as for support of neural growth during neuroma formation. This denervation model offers new and interesting ways to study the pathogenesis of traumatic neuroma and neuroinflammatory changes, some of which may be related to pain.
List of abbreviations
IL Interleukin
IFN-γ Interferon-γ
BNB Blood-nerve-barrier
BSA Bovine serum albumin
GAPDH Glyseraldehyde-3-phosphate-dehydrogenase
GITS Guanidine thiocyanate 6-9277 Sigma, N-lauroylsarcosine L-5125 Sigma, tri-Na-citrate dihydrate 1.12005 Merck, 0.7% h-mercaptoethanol
MHC Major histocompatibility complex
NGF Nerve growth factor
TGF-β1 Transforming growth factor-β1
NMDA N-methyl-D-aspartate
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SR carried out the real-time PCR studies, the histological studies, statistics, performed the animal operations together with MR and HST, and prepared the manuscript. MK, MJ and TO performed the real-time PCR analysis, as well as the preparation of mRNA samples. HST participated in the design and coordination of the study. MR designed the study, carried the responsibility for the study protocol and aided in the preparation of the manuscript.
Table 1 List of primers and probes used in the present study
Cytokine 5' primer 3' primer Probe
GAPDH TCAACTACATGGTCTACATGTTCCAG TCCCATTCTCAGCCTTGACTG TGACTCTACCCACGGCAAGTTCAACG
IFN-γ TCGAATCGCACCTGATCACTA GGGTTGTTCACCTCGAACTTG CATCCTTTTTTGCTTTACTGTTGCTGAGAAG
IL-1β GAAAGACGGCACACCCACC AAACCGCTTTTCCATCTTCTTCT TGCAGCTGGAGAGTGTGGATCCCAAAC
IL-10 CCCTCTGGATACAGCTGCG GCTCCACTGCCTTGCTTTTATT CGCTGTCATCGATTTCTCCCCTGTGA
TNF-α GACCCTCACACTCAGATCATCTTCT ACGCTGGCTCAGCCACTC TAGCCCACGTCGTAGCAAACCACCAA
The probes were labeled with FAM in the 5' end and TAMRA in the 3' end.
Acknowledgements
We are thankful to Pirkko Huuskonen, MA for reviewing the language. Financial support was provided by Turku University Hospital (EVO-grant).
==== Refs
Taskinen HS Röyttä M The dynamics of macrophage recruitment after nerve transection Acta Neuropathol (Berl) 1997 93 252 259 9083556 10.1007/s004010050611
Stoll G Griffin JW Li CY Trapp BD Wallerian degeneration in the peripheral nervous system: participation of both Schwann cells and macrophages in myelin degradation J Neurocytol 1989 18 671 683 2614485 10.1007/BF01187086
Perry VH Brown MC Gordon S The macrophage response to central and peripheral nerve injury. A possible role for macrophages in regeneration J Exp Med 1987 165 1218 1223 3559478 10.1084/jem.165.4.1218
Mellick RS Cavanagh JB Changes in blood vessel permeability during degeneration and regeneration in peripheral nerves Brain 1968 91 141 160 4296122
Olsson Y Studies on vascular permeability in peripheral nerves. I. Distribution of circulating fluorescent serum albumin in normal, crushed and sectioned rat sciatic nerve Acta Neuropathol (Berl) 1966 7 1 15 5971099 10.1007/BF00686605
Toews AD Barrett C Morell P Monocyte chemoattractant protein 1 is responsible for macrophage recruitment following injury to sciatic nerve J Neurosci Res 1998 53 260 267 9671983 10.1002/(SICI)1097-4547(19980715)53:2<260::AID-JNR15>3.0.CO;2-A
Oldfors A Macrophages in peripheral nerves. An ultrastructural and enzyme histochemical study on rats Acta Neuropathol (Berl) 1980 49 43 49 7355672 10.1007/BF00692218
Stoll G Jung S Jander S van der Meide P Hartung HP Tumor necrosis factor-alpha in immune-mediated demyelination and Wallerian degeneration of the rat peripheral nervous system J Neuroimmunol 1993 45 175 182 8331160 10.1016/0165-5728(93)90178-2
Mueller M Leonhard C Wacker K Ringelstein EB Okabe M Hickey WF Kiefer R eds Macrophage response to peripheral nerve injury: the quantitative contribution of resident and hematogenous macrophages Lab Invest 2003 83 175 185 12594233
Shamash S Reichert F Rotshenker S The cytokine network of Wallerian degeneration: tumor necrosis factor-alpha, interleukin-1alpha, and interleukin-1beta J Neurosci 2002 22 3052 3060 11943808
Wagner R Myers RR Schwann cells produce tumor necrosis factor alpha: expression in injured and non-injured nerves Neuroscience 1996 73 625 629 8809782 10.1016/0306-4522(96)00127-3
Sommer C Schafers M Painful mononeuropathy in C57BL/Wld mice with delayed wallerian degeneration: differential effects of cytokine production and nerve regeneration on thermal and mechanical hypersensitivity Brain Res 1998 784 154 162 9518588 10.1016/S0006-8993(97)01327-9
George A Buehl A Sommer C Wallerian degeneration after crush injury of rat sciatic nerve increases endo- and epineurial tumor necrosis factor-alpha protein Neurosci Lett 2004 372 215 219 15542243 10.1016/j.neulet.2004.09.075
Beuche W Friede RL The role of non-resident cells in Wallerian degeneration J Neurocytol 1984 13 767 796 6512566 10.1007/BF01148493
Griffin JW George R Lobato C Tyor WR Yan LC Glass JD Macrophage responses and myelin clearance during Wallerian degeneration: relevance to immune-mediated demyelination J Neuroimmunol 1992 40 153 165 1430148 10.1016/0165-5728(92)90129-9
Asakura K Rodriguez M A unique population of circulating autoantibodies promotes central nervous system remyelination Mult Scler 1998 4 217 221 9762677 10.1191/135245898678909222
Fiorentino DF Zlotnik A Mosmann TR Howard M O'Garra A IL-10 inhibits cytokine production by activated macrophages J Immunol 1991 147 3815 3822 1940369
Kiefer R Streit WJ Toyka KV Kreutzberg GW Hartung HP Transforming growth factor-beta 1: a lesion-associated cytokine of the nervous system Int J Dev Neurosci 1995 13 331 339 7572285 10.1016/0736-5748(94)00074-D
Scherer SS Kamholz J Jakowlew SB Axons modulate the expression of transforming growth factor-betas in Schwann cells Glia 1993 8 265 276 8406683 10.1002/glia.440080407
Jander S Pohl J Gillen C Stoll G Differential expression of interleukin-10 mRNA in Wallerian degeneration and immune-mediated inflammation of the rat peripheral nervous system J Neurosci Res 1996 43 254 259 8820973 10.1002/(SICI)1097-4547(19960115)43:2<254::AID-JNR13>3.0.CO;2-6
O'Keefe GM Nguyen VT Benveniste EN Class II transactivator and class II MHC gene expression in microglia: modulation by the cytokines TGF-beta, IL-4, IL-13 and IL-10 Eur J Immunol 1999 29 1275 1285 10229095 10.1002/(SICI)1521-4141(199904)29:04<1275::AID-IMMU1275>3.0.CO;2-T
Sulaiman OA Gordon T Transforming growth factor-beta and forskolin attenuate the adverse effects of long-term Schwann cell denervation on peripheral nerve regeneration in vivo Glia 2002 37 206 218 11857679 10.1002/glia.10022
Rogister B Delree P Leprince P Martin D Sadzot C Malgrange B Munaut C Rigo JM Lefebvre PP Octave JN Transforming growth factor beta as a neuronoglial signal during peripheral nervous system response to injury J Neurosci Res 1993 34 32 43 8423635 10.1002/jnr.490340105
Einheber S Hannocks MJ Metz CN Rifkin DB Salzer JL Transforming growth factor-beta 1 regulates axon/Schwann cell interactions J Cell Biol 1995 129 443 458 7536747 10.1083/jcb.129.2.443
Hattori A Iwasaki S Murase K Tsujimoto M Sato M Hayashi K Kohno M Tumor necrosis factor is markedly synergistic with interleukin 1 and interferon-gamma in stimulating the production of nerve growth factor in fibroblasts FEBS Lett 1994 340 177 180 8131840 10.1016/0014-5793(94)80132-0
Skoff AM Lisak RP Bealmear B Benjamins JA TNF-alpha and TGF-beta act synergistically to kill Schwann cells J Neurosci Res 1998 53 747 756 9753202 10.1002/(SICI)1097-4547(19980915)53:6<747::AID-JNR12>3.0.CO;2-V
Röyttä M Salonen V Peltonen J Reversible endoneurial changes after nerve injury Acta Neuropathol (Berl) 1987 73 323 329 3618125 10.1007/BF00688254
Siironen J Sandberg M Vuorinen V Roytta M Laminin B1 and collagen type IV gene expression in transected peripheral nerve: reinnervation compared to denervation J Neurochem 1992 59 2184 2192 1279123
Siironen J Collan Y Roytta M Axonal reinnervation does not influence Schwann cell proliferation after rat sciatic nerve transection Brain Res 1994 654 303 311 7987679 10.1016/0006-8993(94)90492-8
Chomczynski P Sacchi N Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction Anal Biochem 1987 162 156 159 2440339 10.1016/0003-2697(87)90021-2
Schnell L Fearn S Klassen H Schwab ME Perry VH Acute inflammatory responses to mechanical lesions in the CNS: differences between brain and spinal cord Eur J Neurosci 1999 11 3648 3658 10564372 10.1046/j.1460-9568.1999.00792.x
Taskinen HS Olsson T Bucht A Khademi M Svelander L Röyttä M Peripheral nerve injury induces endoneurial expression of IFN-gamma, IL-10 and TNF-alpha mRNA J Neuroimmunol 2000 102 17 25 10626662 10.1016/S0165-5728(99)00154-X
Taskinen HS Röyttä M Increased expression of chemokines (MCP-1, MIP-1alpha, RANTES) after peripheral nerve transection J Peripher Nerv Syst 2000 5 75 81 10905466 10.1046/j.1529-8027.2000.00009.x
Burd PR Thompson WC Max EE Mills FC Activated mast cells produce interleukin 13 J Exp Med 1995 181 1373 1380 7535336 10.1084/jem.181.4.1373
Bergsteinsdottir K Kingston A Mirsky R Jessen KR Rat Schwann cells produce interleukin-1 J Neuroimmunol 1991 34 15 23 1894731 10.1016/0165-5728(91)90094-N
Al-Shatti T Barr AE Safadi FF Amin M Barbe MF Increase in inflammatory cytokines in median nerves in a rat model of repetitive motion injury J Neuroimmunol 2005 167 13 22 16026858 10.1016/j.jneuroim.2005.06.013
Liefner M Siebert H Sachse T Michel U Kollias G Bruck W The role of TNF-alpha during Wallerian degeneration J Neuroimmunol 2000 108 147 152 10900348 10.1016/S0165-5728(00)00262-9
Sommer C Schmidt C George A Toyka KV A metalloprotease-inhibitor reduces pain associated behavior in mice with experimental neuropathy Neurosci Lett 1997 237 45 48 9406876 10.1016/S0304-3940(97)00813-6
Wagner R Myers RR Endoneurial injection of TNF-alpha produces neuropathic pain behaviors Neuroreport 1996 7 2897 2901 9116205
Schafers M Geis C Brors D Yaksh TL Sommer C Anterograde transport of tumor necrosis factor-alpha in the intact and injured rat sciatic nerve J Neurosci 2002 22 536 545 11784800
Nawroth PP Stern DM Modulation of endothelial cell hemostatic properties by tumor necrosis factor J Exp Med 1986 163 740 745 3753996 10.1084/jem.163.3.740
Misawa R Kawagishi C Watanabe N Kobayashi Y Infiltration of neutrophils following injection of apoptotic cells into the peritoneal cavity Apoptosis 2001 6 411 417 11595830 10.1023/A:1012406121993
Tracey KJ Cerami A Tumor necrosis factor: a pleiotropic cytokine and therapeutic target Annu Rev Med 1994 45 491 503 8198398 10.1146/annurev.med.45.1.491
Redford EJ Hall SM Smith KJ Vascular changes and demyelination induced by the intraneural injection of tumour necrosis factor Brain 1995 118 869 878 7655885
Viviani B Corsini E Galli CL Marinovich M Glia increase degeneration of hippocampal neurons through release of tumor necrosis factor-alpha Toxicol Appl Pharmacol 1998 150 271 276 9653057 10.1006/taap.1998.8406
Okamoto K Martin DP Schmelzer JD Mitsui Y Low PA Pro- and anti-inflammatory cytokine gene expression in rat sciatic nerve chronic constriction injury model of neuropathic pain Exp Neurol 2001 169 386 391 11358451 10.1006/exnr.2001.7677
Bethea JR Nagashima H Acosta MC Briceno C Gomez F Marcillo AE Loor K Green J Dietrich WD Systemically administered interleukin-10 reduces tumor necrosis factor-alpha production and significantly improves functional recovery following traumatic spinal cord injury in rats J Neurotrauma 1999 16 851 863 10547095
Perrin FE Lacroix S Aviles-Trigueros M David S Involvement of monocyte chemoattractant protein-1, macrophage inflammatory protein-1{alpha} and interleukin-1{beta} in Wallerian degeneration Brain 2005 15689362
Moore KW O'Garra A de Waal Malefyt R Vieira P Mosmann TR Interleukin-10 Annu Rev Immunol 1993 11 165 190 8386517 10.1146/annurev.iy.11.040193.001121
Salonen V Aho H Röyttä M Peltonen J Quantitation of Schwann cells and endoneurial fibroblast-like cells after experimental nerve trauma Acta Neuropathol (Berl) 1988 75 331 336 3364158 10.1007/BF00687785
Fiorentino DF Zlotnik A Vieira P Mosmann TR Howard M Moore KW O'Garra A IL-10 acts on the antigen-presenting cell to inhibit cytokine production by Th1 cells J Immunol 1991 146 3444 3451 1827484
Bullock ED Johnson EM Jr Nerve growth factor induces the expression of certain cytokine genes and bcl-2 in mast cells. Potential role in survival promotion J Biol Chem 1996 271 27500 27508 8910334 10.1074/jbc.271.44.27500
Halloran PF Autenried P Ramassar V Urmson J Cockfield S Local T cell responses induce widespread MHC expression. Evidence that IFN-gamma induces its own expression in remote sites J Immunol 1992 148 3837 3846 1602132
Heininger K Schafer B Hartung HP Fierz W Linington C Toyka KV The role of macrophages in experimental autoimmune neuritis induced by a P2-specific T-cell line Ann Neurol 1988 23 326 331 3260088 10.1002/ana.410230403
Collart MA Belin D Vassalli JD de Kossodo S Vassalli P Gamma interferon enhances macrophage transcription of the tumor necrosis factor/cachectin, interleukin 1, and urokinase genes, which are controlled by short-lived repressors J Exp Med 1986 164 2113 2118 3097240 10.1084/jem.164.6.2113
Murwani R Armati P Peripheral nerve fibroblasts as a source of IL-6, TNFalpha and IL-1 and their modulation by IFNgamma J Neurol Sci 1998 161 99 109 9879689 10.1016/S0022-510X(98)00271-8
Davison SP McCaffrey TV Porter MN Manders E Improved nerve regeneration with neutralization of transforming growth factor-beta1 Laryngoscope 1999 109 631 635 10201754 10.1097/00005537-199904000-00021
Taskinen HS Ruohonen S Jagodic M Khademi M Olsson T Roytta M Distinct expression of TGF-beta1 mRNA in the endo- and epineurium after nerve injury J Neurotrauma 2004 21 969 975 15307908 10.1089/0897715041526131
Ridley AJ Davis JB Stroobant P Land H Transforming growth factors-beta 1 and beta 2 are mitogens for rat Schwann cells J Cell Biol 1989 109 3419 3424 2557356 10.1083/jcb.109.6.3419
Parkinson DB Dong Z Bunting H Whitfield J Meier C Marie H Mirsky R Jessen KR Transforming growth factor beta (TGFbeta) mediates Schwann cell death in vitro and in vivo: examination of c-Jun activation, interactions with survival signals, and the relationship of TGFbeta-mediated death to Schwann cell differentiation J Neurosci 2001 21 8572 8585 11606645
Raivich G Liu ZQ Kloss CU Labow M Bluethmann H Bohatschek M Cytotoxic potential of proinflammatory cytokines: combined deletion of TNF receptors TNFR1 and TNFR2 prevents motoneuron cell death after facial axotomy in adult mouse Exp Neurol 2002 178 186 193 12504878 10.1006/exnr.2002.8024
Ruohonen S Jagodi M Khademi M Taskinen HS Ojala P Olsson T Roytta M Contralateral non-operated nerve to transected rat sciatic nerve shows increased expression of IL-1beta, TGF-beta1, TNF-alpha, and IL-10 J Neuroimmunol 2002 132 11 17 12417428 10.1016/S0165-5728(02)00281-3
Kleinschnitz C Brinkhoff J Sommer C Stoll G Contralateral cytokine gene induction after peripheral nerve lesions: dependence on the mode of injury and NMDA receptor signaling Brain Res Mol Brain Res 2005 136 23 28 15893583 10.1016/j.molbrainres.2004.12.015
|
16287511
|
PMC1310517
|
CC BY
|
2021-01-04 16:38:21
|
no
|
J Neuroinflammation. 2005 Nov 18; 2:26
|
utf-8
|
J Neuroinflammation
| 2,005 |
10.1186/1742-2094-2-26
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-741630574310.1186/1477-7525-3-74ResearchComparative assessment of three different indices of multimorbidity for studies on health-related quality of life Fortin Martin [email protected] Catherine [email protected] Marie-France [email protected] José [email protected] Lise [email protected] Hassan [email protected] Department of Family Medicine, Sherbrooke University, Sherbrooke, Que, Canada2 Centre de Santé et de Services Sociaux de Chicoutimi, Que, Canada3 Department of Community Health Sciences, Sherbrooke University, Sherbrooke, Que, Canada4 Research Center on Aging, Sherbrooke University Geriatric Institute, Sherbrooke, Que, Canada2005 23 11 2005 3 74 74 29 9 2005 23 11 2005 Copyright © 2005 Fortin et al; licensee BioMed Central Ltd.2005Fortin et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Measures of multimorbidity are often applied to source data, populations or outcomes outside the scope of their original developmental work. As the development of a multimorbidity measure is influenced by the population and outcome used, these influences should be taken into account when selecting a multimorbidity index. The aim of this study was to compare the strength of the association of health-related quality of life (HRQOL) with three multimorbidity indices: the Cumulative Illness Rating Scale (CIRS), the Charlson index (Charlson) and the Functional Comorbidity Index (FCI). The first two indices were not developed in light of HRQOL.
Methods
We used data on chronic diseases and on the SF-36 questionnaire assessing HRQOL of 238 adult primary care patients who participated in a previous study. We extracted all the diagnoses for every patient from chart review to score the CIRS, the FCI and the Charlson. Data for potential confounders (age, sex, self-perceived economic status and self-perceived social support) were also collected. We calculated the Pearson correlation coefficients (r) of the SF-36 scores with the three measures of multimorbidity, as well as the coefficient of determination, R2, while controlling for confounders.
Results
The r values for the CIRS (range: -0.55 to -0.18) were always higher than those for the FCI (-0.47 to -0.10) and Charlson (-0.31 to -0.04) indices. The CIRS explained the highest percent of variation in all scores of the SF-36, except for the Mental Component Summary Score where the variation was not significant. Variations explained by the FCI were significant in all scores of SF-36 measuring physical health and in two scales evaluating mental health. Variations explained by the Charlson were significant in only three scores measuring physical health.
Conclusion
The CIRS is a better choice as a measure of multimorbidity than the FCI and the Charlson when HRQOL is the outcome of interest. However, the FCI may provide a good option to evaluate the physical aspect of HRQOL for the ease in its administration and scoring. The Charlson index may not be recommended as a measure of multimorbidity in studies related to either physical or mental aspects of HRQOL.
==== Body
Background
The coexistence of multiple chronic diseases in the same individual or multimorbidity has led to increasing interest in its measure in research studies as a potential confounder or as a predictor of study outcome [1,2].
Health-related quality of life (HRQOL) is an outcome measure that is adversely affected by the presence of multimorbidity. This association can be demonstrated using the simple count of chronic conditions as a measure of multimorbidity [3-8]. However, we found in a recent study that the use of a multimorbidity index, the Cumulative Illness Rating Scale (CIRS), revealed a stronger association of HRQOL with multimorbidity than a simple count of chronic diseases [8]. Measures of multimorbidity are often applied to source data, populations or outcomes outside the scope of the original developmental work [9]. However, as the development of a multimorbidity measure is influenced by the population and outcome used, these influences should be taken into account when selecting a multimorbidity index [10]. Although the CIRS is a comprehensive evaluation of medical problems by organ system, it was not developed in light of HRQOL. Therefore, it can be argued that another measure of multimorbidity (or comorbidity if an index disease is the object of study) specifically designed for HRQOL could bear a stronger relationship with HRQOL than the CIRS, and would be a better measure of multimorbidity when the outcome of interest is HRQOL.
Several indices have been described to measure multimorbidity or comorbidity [1,2,11]. However, some problems related to many of these indices have been reported such as insufficient data on their clinimetric properties and moderate inter-rater reliability [2,12]. Two indices stand out as potential alternatives to the CIRS, the Charlson Index and the Functional Comorbidity Index (FCI). The Charlson index [13] is, with the CIRS [14], among the most valid and reliable measures of multimorbidity [2]. The Charlson index is the most extensively studied comorbidity index and, although the weights originally used to develop it were based on the relative risk of dying, it has been found to significantly predict the number of ambulatory visits, the probability of an inpatient admission, the length of stay, and hospital costs [9,15]. However, the association between the Charlson index and HRQOL has been assessed only in patients of age 65 or older [16]. Recently developed, the Functional Comorbidity Index (FCI) [11] was specifically developed with physical functioning, an aspect of HRQOL, as the validity criterion. The index was developed using two databases totalizing 37,772 Canadian and US adults seeking treatment for spine ailments. It is possible that the association of this index with physical aspects of HRQOL could outperform the CIRS, but this hypothesis has not been tested yet.
Using these three indices (CIRS, FCI and Charlson) on the same target population would allow a better comparison of their performance when the outcome of interest is HRQOL, but we could not find any study with such comparison. Thus, the primary purpose of this study was to compare the strength of the association of the CIRS, the Charlson index and the FCI measures of multimorbidity, with HRQOL.
Methods
We used data collected on the diagnoses of chronic diseases in a group of 238 adult primary care patients (age 18 or older) who participated in a study on HRQOL [8]. Patients were recruited from the clientele of 21 family physicians in the Saguenay region, Canada. Details of the sampling are described elsewhere [17]. In brief, we randomly selected patients from 980 patients who had also been selected at random for a prevalence study on multimorbidity [17]. Our goal was to recruit 60 patients for each CIRS quintile to have enough representation of different levels of multimorbidity. Of the 419 patients we tried to contact by phone, 66 could not be reached, despite repeated attempts. Of the remaining 353 patients, 238 agreed to participate (Table 1). Patients completed the self-administered 36-item short form of the Medical Outcomes Study questionnaire (SF-36) [18] to assess HRQOL. The SF-36 comprises 8 multi-item scales divided into 2 main groups: physical and mental aspects of quality of life. Two summary scores for each group are obtained through a weighted sum of these scales. To compute the Physical Component Summary scale, high positive weights are given to the scales of the physical aspects of quality of life and low negative weights to those of the mental health. To calculate the Mental Component Summary scale, low negative weights are given to the scores of the physical aspects of quality of life and high positive weights are given to those of the mental health. For all scales and both summary scales, lower scores indicate lower HRQOL.
Table 1 Characteristics of the Sample
Characteristic Refusals (n = 115) Participants (n = 238) P value
Mean (SD) age, y 56.5 (17.4) 59.0 (14.3) 0.17*
Mean (SD) diagnoses, n 5.5 (3.2) 5.3 (2.8) 0.49*
Male, % 33.9 29.0 0.39†
*t-test.
†Chi-square test.
From an exhaustive chart review, we extracted a comprehensive list of diagnoses of all chronic conditions for every patient after informed consent. We then used the list to score the CIRS [19], the FCI [11] and the Charlson index [13] (Table 2). To obtain the most reliable measures for analysis, the three indices were scored by two investigators independently in a group of patients (the number of patients varied from 49 to 73 for the 3 indices), and inter-rater reliability was calculated. During a standardization period, the scoring process was discussed to reach a consensus and repeated until the inter-rater reliability was judged acceptable [20].
Table 2 Main characteristics of CIRS, FCI and Charlson†
Comorbidity indices CIRS FCI Charlson
Items 1. Cardiac
2. Vascular
3. Hematological
4. Respiratory
5. Ophthalmological and ORL
6. Upper gastrointestinal
7. Lower gastrointestinal
8. Hepatic and pancreatic
9. Renal
10. Genitourinary
11. Musculoskeletal and tegumental
12. Neurological
13. Endocrine, metabolic, breast
14. Psychiatric 1. Arthritis (rheumatoid and osteoarthritis)
2. Osteoporosis
3. Asthma
4. COPD, ARDS*
5. Angina
6. Congestive heart failure or heart disease
7. Heart attack
8. Neurological disease
9. Stroke or transient ischemic attack
10. Diabetes types I and II
11. Peripheral vascular disease
12. Upper gastrointestinal disease
13. Depression
14. Anxiety or panic disorders
15. Visual impairment
16. Hearing impairment
17. Degenerative disk disease
18. Obesity and/or BMI > 30 kg/m2 1. Myocardial infarct
2. Congestive heart failure
3. Peripheral vascular disease
4. Cerebrovascular disease
5. Dementia
6. Chronic pulmonary disease
7. Connective tissue disease
8. Ulcer disease
9. Stroke or transient ischemic attack
10. Diabetes
11. Hemiplegia
12. Moderate or severe renal disease
13. Diabetes with end organ damage
14. Any tumor
15. Leukemia
16. Lymphoma
17. Moderate or severe liver disease
18. Metastatic solid tumor
19. AIDS
Weights All systems weighted from 0 to 4:
0 No problem
1 Mild
2 Moderate
3 Severe
4 Extremely severe Presence (yes) or absence (no) of diagnoses Conditions from 1 to 10, weight = 1
Conditions from 11 to 16, weight = 2
Condition 17, weight = 3
Conditions 18 and 19, weight = 6
Final score Sum of weights assigned to each system Sum of "yes" answers Sum of weights assigned to each condition that a patient has
† CIRS = Cumulative Illness Rating Scale; FCI = Functional Comorbidity Index; Charlson = Charlson index.
* COPD = chronic obstructive pulmonary disease; ARDS = acquired respiratory distress syndrome
Data for potential confounders (age, sex, self-perceived economic status and self-perceived social support) were also collected. Self-perceived social support was measured with the Social Provisions Scale [21]. The research ethics board of the Centre de santé et de services sociaux de Chicoutimi approved this study.
Statistical analysis
To investigate the relationship between HRQOL and the multimorbidity indices as well as the direction of the relationships (positive or negative), we first calculated the Pearson correlation coefficients of the SF-36 scores with the three measures of multimorbidity. We also compared CIRS correlation coefficients with those of the FCI and the Charlson index [22]. Next, the coefficient of determination, R2, was calculated to measure the percentage of variation in the dependent variables (all SF-36 scales and two SF-36 summary scores) explained by each measure of multimorbidity over and above that explained by age, gender, self-perceived social support and self-perceived economical status. We obtained these estimates through multiple regression analysis for which underlying assumptions were judged satisfactory. All analyses were done using the SAS system for Windows (version 8.02, SAS Institute, Inc, Cary, NC, USA).
Results
After standardization of the scoring process, the intraclass correlation coefficients for the inter-rater reliability were 0.96, 0.92 and 0.90 for the CIRS, the FCI and the Charlson respectively.
Figure 1 shows the distribution of each multimorbidity score. The CIRS had the widest variation, with a range of 1 to 27 with a mode of 9 (mean = 10.3). The FCI had a range of 0 to 8, with a mode of 3 (mean = 2.4). The Charlson index had a similar range (0–7) but a different distribution from that of the FCI, with 120 patients (50.4%) having a score of zero (mean = 0.9).
Figure 1 Distribution of scores on multimorbidity measures. CIRS = Cumulative Illness Rating Scale; FCI = Functional Comorbidity Index; Charlson = Charlson index.
Pearson correlation coefficients of SF-36 with the three measures of multimorbidity are shown in Table 3. The CIRS was negatively correlated with all scales of SF-36 except the Mental Component Summary; i.e. higher morbidity or multimorbidity level was associated with lower HRQOL. The FCI was negatively correlated with all SF-36 scales measuring the physical aspect of HRQOL; it was also negatively correlated with two scales measuring the mental aspect of HRQOL. The Charlson index was negatively correlated with all scales of SF-36 evaluating the physical aspect of HRQOL; it was not correlated with any of the scales evaluating the mental aspect. There was an unexpected positive correlation of the Charlson index with the Mental Component Summary that did not have any meaningful interpretation. The CIRS correlation coefficients were significantly different from those of the FCI for the SF-36 scales of Physical Functioning, Role Physical and Social Functioning as well as for the Physical Component Summary; whereas the Charlson correlation coefficients were significantly different from those of CIRS for all SF-36 scales.
Table 3 Pearson correlation coefficients of the SF-36† scores with the measures of multimorbidity
HRQOL (SF-36) Correlation coefficients (r)
Physical Health CIRS‡ FCI Charlson
Physical Functioning -0.55** -0.47** -0.31**
Role physical -0.41** -0.32** -0.14*
Bodily Pain -0.38** -0.33** -0.16*
General Health -0.40** -0.34** -0.21**
Mental Health
Vitality -0.30** -0.23** -0.08
Social Functioning -0.29** -0.21** -0.04
Role Emotional -0.18** -0.10 +0.03
Mental Health -0.18** -0.14** +0.07
Physical Component Summary -0.54** -0.47** -0.31**
Mental Component Summary -0.06 -0.001 +0.16*
† SF-36 = Self-administered 36-item short form of the Medical Outcomes Study questionnaire.
‡ CIRS = Cumulative Illness Rating Scale; FCI = Functional Comorbidity Index; Charlson = Charlson index. * p < 0.05; ** p < 0.01
Table 4 shows the percentage of variation in HRQOL explained by each measure of multimorbidity over and above that explained by age, gender, self-perceived social support and economic status. The CIRS explained the highest percent of variation in all scores, except for the Mental Component Summary score where the explained variation was not significant.
Table 4 Percentage of variation of the SF-36 ◇ scores explained by each measure of multimorbidity
HRQOL (SF-36) Percentage of variation explained by the control variables† Partial R2(%)‡
Physical Health CIRS§ FCI Charlson
Physical Functioning 21.08** 15.59** 9.53** 4.52**
Role physical 7.84** 11.14** 5.21** 0.56
Bodily Pain 10.02** 9.91** 6.80** 1.04
General Health 11.63** 14.07** 7.96** 2.99**
Mental Health
Vitality 10.68** 6.78** 2.56* 0.10
Social Functioning 10.77** 6.72** 2.94** 0.002
Role Emotional 6.71** 2.64* 0.43 0.52
Mental Health 17.07*** 2.26* 1.02 1.18
Physical Component Summary 13.18** 17.75** 11.81** 5.46**
Mental Component Summary 12.60** 0.75 0.02 2.80
◇ SF-36 = Self-administered 36-item short form of the Medical Outcomes Study questionnaire.
† control variables: age, gender, self-perceived social support and self-perceived economical status
‡ controlling for age, gender, self-perceived social support and self-perceived economical status.
§CIRS = Cumulative Illness Rating Scale; FCI = Functional Comorbidity Index; Charlson = Charlson index.
* p < 0.05 ; ** p < 0.01
Discussion
We compared the strength of association of three multimorbidity indices (CIRS, FCI and Charlson index) with HRQOL as the outcome of interest in a primary care context. In terms of percent of explained variation in HRQOL, the CIRS performed as well as and often better than the FCI and the Charlson index in all scores of the SF-36. Correlation coefficients of the SF-36 scores with the measures of multimorbidity were always higher for the CIRS, followed by the FCI (Table 3); the correlations of the SF-36 scores with the Charlson index were always the weakest. We also found an unexpected positive correlation of the Charlson index with the SF-36 Mental Component Summary.
Among the three indices, the CIRS was the one that explained the highest percent of variation in all scores of the SF-36. Despite the fact that the FCI was developed with physical function as the outcome of interest, it did not perform better than the CIRS in any of the scales of the SF-36 evaluating the physical aspect of HRQOL. This result may be due in part to the wider range of possible scores on the CIRS. Indeed, an index ranging from 0 to 27 can better predict variations in an outcome than one that ranges from 0 to 7 or 8 with more than half the patients being classified in the first 2 or 3 levels of the scale. It may also be due to the fact that the CIRS evaluates the number and severity of all chronic diseases whereas the FCI evaluates a limited number of diagnoses and does not take into account disease severity. However, R2 values for the FCI related to physical health scores, although lower than those of the CIRS, remained highly significant after controlling for confounders. Given that the FCI is very easy to administer and score, researchers may consider, depending on the characteristics of the study, to trade off a lower explained variation for simplicity to evaluate the physical aspect of HRQOL. In the case of the Charlson index, the percent of explained variation was significant only in the Physical Functioning, the General Health, and the Physical Component Summary scales. In the mental aspect of HRQOL, the percent of variation explained by the Charlson index was not significant in any of the scales of the SF-36. Given these results, the Charlson index may not be recommended as a measure of multimorbidity in HRQOL studies in adults.
The FCI was the only index of multimorbidity that we were aware of that was developed using a component of HRQOL (Physical Functioning) as outcome. However, two other articles reporting multimorbidity measures related with HRQOL were published upon completion of the present study. One of the articles describes a new self-reported assessment of comorbidity, or self-reported disease burden [16]; the other article describes five indices or approaches to scoring multimorbidity derived from a self-administered multimorbidity questionnaire [23].
In the article on the self-reported disease burden [16], the index was validated using two scales of the SF-36 evaluating the physical aspect of HRQOL (Physical Functioning and one item of General Health) as well as the outcomes of depression and self-efficacy. The authors studied these outcomes using the Charlson index and the findings were similar to ours. They found a negative correlation between the Charlson index and the Physical Functioning and General Health outcomes [16]. However, our study expanded the analysis of the Charlson index to all scales of the SF-36 evaluating both physical and mental aspects of the HRQOL. Moreover, we included adults aged 18 and over, whereas age was restricted to 65 years or older in the study on the self-reported disease burden [16]. In the second paper by Byles et al [23], the study was a comparison of the performance of five indices derived from a self-administered multimorbidity questionnaire. None of the indices was compared to other indices previously published. Unfortunately, it was not possible to include these five indices in our comparative study because of the chart review method that we used. However, future research comparing CIRS with these five indices as well as with the self-reported disease burden index is warranted.
In our analysis of the relationship between mental aspects of HRQOL and multimorbidity, we found some contradictory results that may reflect a limitation in our instruments. All scales of the SF-36 used to measure the mental aspect of HRQOL were related to the CIRS, whereas the Mental Component Summary was not (Tables 3 and 4). This summary score was created by the developers of the SF-36 with the hope to reduce the number of statistical comparisons involved in analyzing the SF-36 without substantial loss of information [24]. The lowest possible score of the Mental Component Summary indicates frequent psychological distress, social disability due to emotional problems, and a poorly self-rated health [24]. However, the lack of relationship we found between the CIRS and the Mental Component Summary contradicts the relationship we found between the CIRS and all mental scales of the SF-36 of which the Mental Component Summary is a composite. One possible explanation may be that the calculation of the Mental Component Summary takes into account not only the four scales measuring mental health, but also the four scales measuring physical health which are weighted negatively [25]. As a result, the positive weights of the mental health scales may be canceled out by the negative weights of the physical health scales which have a stronger relationship with the CIRS in our study. This problem was evident in the relationship between the CIRS and the Mental Component Summary, but it also affected the relationships between this summary score and the other measures of multimorbidity. These results suggest that the Mental Component Summary produced a substantial loss of information in the context of our study.
Conclusion
In summary, our study suggests that the CIRS is a better choice as a measure of multimorbidity than the FCI and the Charlson index in a primary care context when HRQOL is the outcome of interest. However, if researchers were interested only in the physical aspect of HRQOL, then the FCI, despite its lower explained variation in HRQOL, may provide a good option for the ease in its administration and scoring. Finally, based on our results, the Charlson index may not be recommended as a measure of multimorbidity in studies related to either physical or mental aspects of HRQOL.
Authors' contributions
MF participated in the conception and design of the study, supervised data collection and analysis and drafted the manuscript. CH participated in the conception and design of the study and data analysis and helped draft the manuscript. M-FD participated in the design of the study, performed the statistical analysis and helped draft the manuscript. JA participated in the data analysis and helped draft the manuscript. LL participated in the data analysis and helped draft the manuscript. HS participated in data analysis and critically reviewed the manuscript. All authors gave their final approval of the version of the manuscript submitted for publication.
Acknowledgements
Sources of support: Fonds de la Recherche en Santé du Québec (Grant number: 24300-2028) and Pfizer Canada (Independent Research Grant).
==== Refs
Harboun M Ankri J [Comorbidity indexes: review of the literature and application to studies of elderly population][French] Rev Epidemiol Sante Publique 2001 49 287 298 11427831
de Groot V Beckerman H Lankhorst GJ Bouter LM How to measure comorbidity. a critical review of available methods. J Clin Epidemiol 2003 56 221 229 12725876 10.1016/S0895-4356(02)00585-1
Cheng L Cumber S Dumas C Winter R Nguyen KM Nieman LZ Health related quality of life in pregeriatric patients with chronic diseases at urban, public supported clinics Health and Quality of Life Outcomes 2003 1 63 14613559 10.1186/1477-7525-1-63
Wensing M Vingerhoets E Grol R Functional status, health problems, age and comorbidity in primary care patients Qual Life Res 2001 10 141 148 11642684 10.1023/A:1016705615207
Michelson H Bolund C Brandberg Y Multiple chronic health problems are negatively asociated with health related quality of life (HRQoL) irrespective of age. Qual Life Res 2000 9 1093 1104 11401042 10.1023/A:1016654621784
Cuijpers P van Lammeren P Duzijn B Relation between quality of life and chronic illnesses in elderly living in residential homes: a prospective study Int Psychogeriatr 1999 11 445 454 10631590 10.1017/S1041610299006067
Grimby A Svanborg A Morbidity and health-related quality of life among ambulant elderly citizens Aging Clin Exp Res 1997 9 356 364
Fortin M Bravo G Hudon C Lapointe L Almirall J Dubois MF Vanasse A Relationship between multimorbidity and health-related quality of life of patients in primary care Qual Life Res 2006
Perkins AJ Kroenke K Unutzer J Katon W Williams JW Hope C Callahan CM Common comorbidity scales were similar in their ability to predict health care costs and mortality J Clin Epidemiol 2004 57 1040 1048 15528055 10.1016/j.jclinepi.2004.03.002
Extermann M Measurement and impact of comorbidity in older cancer patients Crit Rev Oncol Hematol 2000 35 181 200 10960800
Groll DL To T Bombardier C Wright JG The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 2005 58 595 602 15878473 10.1016/j.jclinepi.2004.10.018
Imamura K McKinnon M Middleton R Black N Reliability of a comorbidity measure: the Index of Co-Existent Disease (ICED) J Clin Epidemiol 1997 50 1011 1016 9363035 10.1016/S0895-4356(97)00128-5
Charlson ME Pompei P Ales KL MacKenzie CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chronic Dis 1987 40 373 383 3558716 10.1016/0021-9681(87)90171-8
Linn BS Linn MW Gurel L Cumulative illness rating scale J Am Geriatr Soc 1968 16 622 626 5646906
Beddhu S Bruns FJ Saul M Seddon P Zeidel ML A simple comorbidity scale predicts clinical outcomes and costs in dialysis patients. Am J Med 2000 108 609 613 10856407 10.1016/S0002-9343(00)00371-5
Bayliss EA Ellis JL Steiner JF Sujective assessments of comorbidity correlate with quality of life health outcomes: Initial validation of a comorbidity assessment instrument. Health and Quality of life Outcomes 2005 3 51 16137329 10.1186/1477-7525-3-51
Fortin M Bravo G Hudon C Vanasse A Lapointe L Prevalence of multimorbidity among adults seen in family practice Ann Fam Med 2005 3 223 228 15928225 10.1370/afm.272
Ware JE Sherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Miller MD Paradis CF Houck PR Mazumdar S Stack JA Rifai AH Mulsant B Reynolds CF Rating chronic medical illness burden in geropsychiatric practice and research: application of the Cumulative Illness Rating Scale Psychiatry Res 1992 41 237 248 1594710 10.1016/0165-1781(92)90005-N
Hudon C Fortin M Vanasse A Cumulative Illness Rating Scale was a reliable and valid index in a family practice context J Clin Epidemiol 2005 58 603 608 15878474 10.1016/j.jclinepi.2004.10.017
Cutrona CE Russell DW The provisions of social support and adaptation to stress Advances in Personal Relationships 1987 1 37 67
Meng XL Rosenthal R Rubin DB Comparing correlated correlation coefficients. Psychol Bull 1992 111 172 175 10.1037/0033-2909.111.1.172
Byles JE D'Este C Parkinson L O'Connell R Treloar C Single index of multimorbidity did not predict multiple outcomes. J Clin Epidemiol 2005 58 997 1005 16168345 10.1016/j.jclinepi.2005.02.025
Ware JEJ SF-36® Health Survey Update. Available from: http://www.sf-36.org/tools/sf36.shtml. Last access August 2005
Leplège A Ecosse E Pouchot J Coste J Perneger T Le questionnaire MOS SF-36. Manuel de l'utilisateur et guide d'interprétation des scores. 2001 Paris, Estem
|
16305743
|
PMC1310518
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Nov 23; 3:74
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-74
|
oa_comm
|
==== Front
Clin Mol AllergyClinical and molecular allergy : CMA1476-7961BioMed Central London 1476-7961-3-151631368110.1186/1476-7961-3-15ResearchInterleukin-4 (IL4) and Interleukin-4 receptor (IL4RA) polymorphisms in asthma: a case control study Isidoro-García María [email protected]ávila Ignacio [email protected] Elena [email protected] Esther [email protected] Félix [email protected]ález-Sarmiento Rogelio [email protected] Molecular Medicine Unit, Department of Medicine, Faculty of Medicine, University of Salamanca, Campus Miguel de Unamuno, Salamanca 37008, Spain2 Department of Allergy, University Hospital of Salamanca, Paseo de San Vicente 58, Salamanca 37007, Spain2005 29 11 2005 3 15 15 29 7 2005 29 11 2005 Copyright © 2005 Isidoro-García et al; licensee BioMed Central Ltd.2005Isidoro-García et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
IL4/IL4RA pathway plays an important role in atopy and asthma. Different polymorphisms in IL4 and IL4RA genes have been described. Particularly, -33C>TIL4 and 576Q>RIL4RA SNPs have been independently associated to atopy and asthma. The purpose of this study was to analyse these polymorphisms in a population of patients with a well-characterized asthma phenotype.
Methods
A total of 212 unrelated Caucasian individuals, 133 patients with asthma and 79 healthy subjects without symptoms or history of asthma or atopy and with negative skin prick tests were recruited. Lung function was measured by spirometry and asthma was specialist physician-diagnosed according to the ATS (American Thoracic Society) criteria and classified following the GINA (Global Initiative for Asthma) guidelines. Skin prick tests were performed according to EAACI recommendations. -33C>TIL4 was studied with TaqMan assay and 576Q>RIL4RA by PCR-RFLP technique. Hardy-Weinberg equilibrium was analysed in all groups. Dichotomous variables were analysed using χ2, Fisher exact test, Monte Carlo simulation test and odds ratio test. To model the effects of multiple covariates logistic regression was used.
Results
No statistically significant differences between the group of patients with asthma and the controls were found when the allele and genotype distribution of -33C>TIL4 and 576Q>RIL4RA polymorphisms were compared. However, the T allele of the -33C>TIL4 SNP was more frequent in patients with persistent asthma. Multivariate analysis adjusted for age and sex confirmed that carriers of allele T had an increased risk of persistent asthma (OR:2.77, 95%CI:1.18–6.49; p = 0.019). Analysis of combination of polymorphisms showed that patients carrying both the T allele of -33C>TIL4 and the A allele of 576Q>RIL4RA had an increased risk of asthma. This association was particularly observed in persistent asthma [Fisher's p value = 0.0021, Monte Carlo p value (after 104 simulations) = 0.0016, OR:3.39; 95% CI:1.50–7.66].
Conclusion
Our results show a trend of association between the genetic combination of the T allele of -33C>TIL4 and the A allele of 576Q>RIL4RA with asthma. This genetic variant was more frequently observed in patients with persistent asthma. As long as this study was performed in a small population, further studies in other populations are needed to confirm these results.
==== Body
Background
IL-4 is a Th2 cytokine that plays an essential role in IgE regulation. It triggers isotype switching from IgM to IgE, induces differentiation to Th2 phenotype on T cells and plays a critical role in the induction and maintenance of allergy.IL4 gene has been mapped to chromosome 5q31 where asthma and atopy have also been linked [1-3]. Evidence that IL4 polymorphisms are associated with total IgE levels and potentially with asthma and other allergy related phenotypes has been provided, although ethnical differences have been reported [4]. Specifically, the promoter region of IL4 has been associated with asthma phenotype [5] and a -33C>T polymorphism has been reported in this region [6]. An association between this polymorphism and asthma or atopy has been found, although this relation is still controversial [4,7-10]. IL-4 acts through the IL-4 receptor (IL-4R) that consists of two subunits, the α chain (IL-4Rα) and the γ chain (γc) [11,12]. IL-4Rα is a component of both the IL-4 and the IL-13 receptor complexes [13]. The IL4RA gene is located on chromosome 16p (16p12.1) [14], a region reported in linkage with atopy in different populations [15,16]. Several single-nucleotide polymorphisms (SNPs) have been identified in the coding region of the IL4RA gene, many of them resulting in aminoacid substitutions [17,18]. One of these polymorphisms, 576Q>R, consists of an A-to-G transition at nucleotide 1902, causing a change from glutamine to arginine at codon 576 (Q576R) in the cytoplasmic domain of the IL-4Rα. It has been reported that B-lymphocytes isolated from allergic patients bearing the 576Q>R mutation have an enhanced CD23 induction in response to IL-4 [19]. However, this result has not been confirmed by other authors [20]. Association of the 576Q>R polymorphism with the atopic phenotype has been described, but this relationship is still controversial [19,21-31]. Due to the central role of the IL-4/IL-4RA pathway in atopy and the scarce information about combinations of both genes in South European populations, we have analysed the -33C>T polymorphism of IL4 gene and the 576Q>R polymorphism of IL4RA gene in a Spanish population of patients with a well-characterized phenotype of asthma.
Methods
Subjects
We studied 212 unrelated Caucasian individuals, 133 patients and 79 controls, recruited from the outpatient Allergy Department of the University Hospital of Salamanca. The study was performed following the recommendations of the Ethical Committee of the University Hospital of Salamanca and informed written consent was obtained from each patient. Individuals who met all the following criteria were selected as controls: (i) no symptoms or history of asthma or other pulmonary diseases; (ii) no symptoms or history of atopy; (iii) negative skin prick tests to a battery of common aeroallergens (<1 mm wheal greater than saline) and (iv) absence of first-degree relatives with a history of asthma or atopy. Asthmatic patients were recruited if they had specialist physician-diagnosed asthma with the following characteristics: (i) at least two symptoms consistent with asthma (cough, wheeze and dyspnoea); (ii) either a positive bronchial hyperresponsiveness or a positive bronchodilator test defined as a ≥ 15% increase in baseline FEV1 after bronchodilator use; (iii) absence of other pulmonary disorders. Lung function was measured by spirometry according to ATS (American Thoracic Society) standards and severity of asthma was classified following GINA (Global Initiative for Asthma) guidelines. Asthma patients were grouped into intermittent and persistent by the clinical severity and into allergic and non allergic asthma by the clinical etiology.
Skin prick tests were performed according to EAACI recommendations with a battery of common aeroallergens that included D pteronisynuss, D farinae, L destructor, T putrescentiae, A siro, G domesticus, E maynei, mix of grasses, mix of trees, P judaica, C album, A vulgaris, P lanceolata, O europaea, A alternata, C herbarum, P notatum, A fumigatus, dog, cat, hamster, horse and rabbit dander and cockroach (ALK-Abelló, Madrid, Spain). Saline was used as negative control and histamine 10 mg/ml was used as positive control. Antihistamines were discontinued before skin testing according to published guidelines. Skin tests were considerer positive if at least one allergen elicited a wheal reaction of more than 3 mm of diameter after subtraction of the negative control. Patients were considered atopic if at least they had one positive skin test result. Total serum IgE was measured by a fluoroenzymeimmunoassay (Pharmacia Cap System®; Pharmacia, Uppsala, Sweden), according to the manufacturer's instructions.
Genotyping analysis
After purification from peripheral blood leukocytes, DNA was amplified by polymerase chain reaction (PCR). Genotyping of -33C>TIL4 SNP was performed using a TaqMan assay in the ABI 7700 sequence detector and the allelic discrimination software Sequence Detector v1.7 according to the manufacturer's recommendations (Applied Biosystems). Primers and probes were obtained by means of the Assays-by-Demand SNP genotyping service of Applied Biosystems, Assay ID: C 16176215. Genotyping of the 576Q>RIL4RA polymorphism was performed according to a previously published assay [28]. Two oligonucleotides were used to amplify the polymorphic region of IL4RA: 5'-CCCCCACCACCAGTGGCTACC-3' and 5'-CCAGGAATGAGGTCTTGGAA-3' [24]. PCR reactions were carried out in a total volume of 25 μL, containing 50 ng of DNA and 12.5 μL of PCR Master Mix 2 × (Promega, Madison, Wisconsin). Amplification was performed with an initial denaturation step at 94°C for 5 min followed by 30 cycles of denaturation at 94°C for 1 min, primer annealing at 55°C for 1 min, and extension at 72° for 1 min. A final extension was carried out at 72° for 10 min. A blank amplification tube was always run to check for the presence of contamination. Strict rules were taken to avoid contamination. PCR reactions were prepared on a laminar flow hood and PCR products were examined in a different room. PCR products were digested for 4 h at 37°C with 1 U of MspI (New England Biolab, Boston, Massachusetts) restriction enzyme. After enzymatic digestion of the amplified fragments, the samples were analyzed by electrophoresis in 3% nusieve agarose gel. Control and patients were not genotyped in separated batched and the analysis was performed blindly with respect to case-control status.
Statistical analysis
For case-control studies, the allele and genotype frequencies in patients with asthma were compared to a control non-asthmatic population. All the groups were tested for Hardy-Weinberg equilibrium using χ2 analyses. The dichotomous variables were analysed using χ2, Fisher exact test, Monte Carlo simulation test (after 104 iterations) and odds ratio test. IgE levels were transformed to log10 values to produce a normal distribution for statistical analysis and analysed by ANOVA. To model the effects of multiple covariates on the dichotomous and continuous variables, logistic regression was used. In multivariate analysis, sex and age were included as potential covariates. A p-value less than 0.05 was considered statistically significant. Bonferroni correction was applied when appropriate. Case-control studies were also undertaken using combination of polymorphisms. Frequencies of combinations were estimated individually in controls and in samples to give the results of both single combinations and global data. For management of data, SHEsis software platform [32] and SPSS version 11 (SPSS Inc, Chicago, IL, USA) were used.
Results
-33C>TIL4 SNP
Characteristics of patients and controls are shown in Table 1. Genotype and allele frequencies are shown in Table 2. The -33TIL4 allele was found at a frequency of 0.15 in patients with asthma versus 0.09 in controls. No statistically significant differences between patients with asthma and controls were found. However, we observed an increase of the -33T IL4 allele in patients with persistent asthma compared to controls [Fisher's p value = 0.014, Monte Carlo p value (after 104 simulations) = 0.019]. Multivariate analysis of the genotypes adjusted for age and sex confirmed a trend of association of -33C>TIL4 polymorphism with an increased risk of persistent asthma (OR: 2.77, 95% CI: 1.18–6.49; p = 0.019). No differences were found in the group of subjects suffering allergic asthma compared to controls. Analysis of the total IgE levels failed to reveal any significant difference (p = 0.22), even when separate analysis for each gender was performed (data not shown).
Table 1 Demographic characteristics of patients
Characteristic Controls Patients P value
No. of subjects 79 133
Age ± SD (y) 40 ± 18 32 ± 17 0.001
Sex (No.)
Male 28 56 0.38
Female 51 77
Log IgE ± SD 1.36 ± 0.67 2.18 ± 0.71 <0.001
SD: standard deviation
y: years
Table 2 Genotype and allele frequencies of -33C>TIL4 and 576Q>RIL4RA SNPs
Phenotype Genotype Allele HWE p value
-33C>T IL4 N Log IgE ± SD CC TC TT C T
Controls 79 1.36 ± 0.67 0.81 0.19 0 0.91 0.09 0.35
Asthma 133 2.18 ± 0.71 0.70 0.29 0.01 0.85 0.15 0.15
Allergic Asthma 99 2.42 ± 0.56 0.69 0.30 0.01 0.84 0.16 0.24
Non-allergic Asthma 34 1.59 ± 0.76 0.74 0.26 0 0.87 0.13 0.37
Intermittent Asthma 54 2.39 ± 0.60 0.79 0.19 0.02 0.89 0.11 0.65
Persistent Asthma 79 2.09 ± 0.76 0.63 0.37* 0 0.82 0.18† 0.05
576Q>R IL4RA AA AG GG A G
Controls 79 1.36 ± 0.67 0.62 0.33 0.05 0.79 0.21 0.82
Asthma 133 2.18 ± 0.71 0.68 0.31 0.01 0.84 0.16 0.10
Allergic Asthma 99 2.42 ± 0.56 0.71 0.29 0 0.85 0.15 0.09
Non Allergic Asthma 34 1.59 ± 0.76 0.59 0.38 0.03 0.78 0.22 0.51
Intermittent asthma 54 2.39 ± 0.60 0.67 0.33 0 0.83 0.17 0.14
Persistent asthma 79 2.09 ± 0.76 0.68 0.30 0.01 0.84 0.16 0.35
HWE: Hardy-Weinberg Equilibrium
* Fisher's p value = 0.013, Monte Carlo p value (after 104 simulations) = 0.019
†Fisher's p value = 0.023, Monte Carlo p value (after 104 simulations) = 0.037
576Q>RIL4RA SNP
576RIL4RA arginine allele (G) was found at a frequency of 0.16 in patients with asthma versus 0.21 in healthy subjects (Table 2). We did not observe differences between patients and controls in allele or genotype frequencies. No association was detected with asthma phenotype or with asthma severity (Table 2). 576Q>RIL4RA polymorphism was not related to total serum IgE levels in our population (p = 0.35).
Gene-gene interaction analysis
We did not detect differences in the global distribution of -33C>TIL4 / 576Q>RIL4RA combinations between the group of patients with asthma and the group of controls (Monte Carlo p value = 0.074) (Table 3). However, patients who were carriers of both the T allele of -33C>TIL4 and the A allele of IL4RA had an increased risk of asthma [Fisher's p value = 0.017, Monte Carlo p value (after 104 simulations = 0.012, odds ratio, 2.58; 95 % CI, 1.18–5.66].
Table 3 Distribution of combinations of -33C>TIL4 and 576Q>RIL4RA SNPs
Genetic Variants CTR Asthma Atopy AA PA Monte Carlo p value (after 104 simulations)
CTR vs Asthma CTR vs Atopy CTR vs AA CTR vs PA†
CQ 0.73 0.71 0.72 0.73 0.68 0.584 0.813 0.900 0.281
CR 0.18 0.14 0.12 0.11 0.14 0.334 0.184 0.092 0.425
TQ 0.05 0.12 0.13 0.13 0.16 0.013* 0.016** 0.017*** 0.002‡
TR 0.04 0.03 0.03 0.03 0.02 0.395 0.392 0.786 0.492
CTR, Controls; AA, Allergic Asthma; PA, Persistent Asthma
* Odds ratio: 2.59 95% CI: 1.18–5.66
** Odds ratio: 2.65 95% CI: 1.19–5.87
*** Odds ratio: 2.62 95% CI: 1.17–5.90
†Global distribution of genetic variants: Fisher's p value = 0.018, Monte Carlo p value (after 104 simulations) = 0.017
‡ Odds ratio: 3.39 95% CI: 1.50–7.66
Slight differences in the global distribution of genetic variants were detected between the group of patients with atopy and the group of controls [Fisher's p value = 0.05, Monte Carlo p value (after 104 simulations) = 0.045]. Patients who were carriers of both the T allele of -33C>TIL4 and the A allele of IL4RA had an increased risk of atopy [Fisher's p value = 0.013, Monte Carlo p value (after 104 simulations = 0.016, odds ratio, 2.64; 95 % CI, 1.93–5.87] (Table 3).
Differences in genetic variants distribution were also observed in the group of patients with allergic asthma compared to controls, although these differences did not reach statistical signification globally considered (Fisher p value = 0.053, Monte Carlo p value = 0.053). Again the combination of T and A alleles showed a trend of association [Fisher's p value = 0.016, Monte Carlo p value = 0.017, odds ratio, 2.62; 95 % CI, 1.17–5.90] (Table 3).
When we compared patients with persistent asthma to controls, significant differences in the global combination distribution were observed (Fisher's p value = 0.018, Monte Carlo p value = 0.017). Patients who carried both the T allele of -33C>TIL4 and the A allele of IL4RA had an increased risk of persistent asthma [Fisher's p value = 0.0021, Monte Carlo p value = 0.0016, odds ratio, 3.39; 95 % CI, 1.50–7.66] (Table 3).
Discussion
-33C>TIL4 SNP
We studied -33C>TIL4 and 576Q>RIL4RA polymorphisms in a well-characterized Spanish population of patients with asthma and in a healthy control population. Controls were older than patients allowing a longer period for asthma diagnosis to be made. When we analysed the -33C>T polymorphism independently, we did not detect significant differences in allele or genotype frequencies between the group of patients and the group of controls. Nevertheless, we observed a higher incidence of the T allele of -33IL4 in the group of patients with asthma (Table 2).
An association between -33C>TIL4 polymorphism and asthma or atopy has been previously reported, although this association is controversial [4,7-9]. In previous studies, no statistically significant association between -33C>TIL4 polymorphism and atopic dermatitis, bronchial hyperresponsiveness, atopic rhinitis and skin prick test reactivity was found [7,9]. However, a significant trend for an association between serum IgE levels and this SNP has been detected in children with positive skin prick tests, independent of asthma status [7].
As shown in table 2, we detected that the allele -33TIL4 is more frequent in patients with allergic asthma, although, we did not detect an association between this polymorphism and IgE levels.
It has been reported that polymorphisms within the promoter region of IL4 gene seems to correlate with enhanced IL4 activity [5,33], secondary to modification of IL4 gene transcription [34]. In this sense, it has been hypothesized that the T allele may be associated with severity of asthma [23,34,35]. In our study, a trend of association of -33TIL4 allele with persistent asthma was observed.
576Q>RIL4RA SNP
Analysis of 576Q>RIL4RA polymorphism did not reveal any association with asthma phenotype. It has been reported a relationship of the 576Q>RIL4RA SNP with the atopic phenotype, however this relationship is still controversial [19,21-31,36-38]. In a previous study, we did not find any association of this polymorphism with atopy or IgE levels, except in a specific of group of patients with family history of atopy [28].
Kruse et al [21] associated the R allele with lower total IgE values, but Hersey et al [19]found an association with high levels of total IgE. We did not detect any association with IgE levels. Genetic association studies are often difficult to interpret due to poor reproducibility in different populations [39,40]. This may well result, among other reasons, from the fact that these studies are focused only on one SNP [18,26,27] and that penetrance of the alleles may be influence by other factors [19].
Gene-gene interaction analysis
When we analysed both polymorphisms simultaneously, differences between the group of patients with asthma and controls were detected. Particularly, patients who were carriers of both the T allele of -33C>TIL4 and the A allele of 576Q>RIL4RA had an increased risk of asthma.
Significant differences were observed in the group of patients with allergic asthma compared to controls. In addition, patients who carried both the T allele of -33C>TIL4 and the A allele of 576Q>RIL4RA had an increased risk of persistent asthma, in our population.
It has been previously suggested that both SNPs may modify the susceptibility to atopy or atopic asthma, independently. The importance of analysis of genetic variants has been previously illustrated, because the functional significance of a given polymorphism may only be evident in a specific setting of additional SNPs in the same or different genes [40]. It has also been pointed out that genetic association studies need careful classification of phenotypes, application of quality control in the performance of laboratory procedures and very stringent significant levels to assure reproducibility [20,39]., although it also may indicate true heterogeneity in gene-disease associations. We describe for the first time a specific genetic combination of IL4/IL4RA polymorphisms that shows a trend of association to persistent asthma in a South European population. As long as this study was performed in a small population, further studies in other populations are needed to confirm these results.
Conclusion
We show a trend of association between -33C>TIL4 and 576Q>RIL4RA polymorphisms and asthma phenotype in a Spanish population. Patients who carried both the T allele of -33C>TIL4 and the A allele of 576Q>RIL4RA showed an increased risk of allergic asthma. In the population included in our study this combination was observed more frequently in patients with persistent asthma.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MIG participated in the design of the study, carried out the molecular genetic studies, performed the statistical analysis and drafted the manuscript.
IDG participated in the design of the study, coordinated the clinical aspects of the study, helped to perform the statistical aspects and to draft the manuscript.
EL participated in the clinical aspects of the study.
EM participated in the clinical aspects of the study.
FL participated in the design and coordination of the study.
RGS conceived the study, participated in its design and coordination and helped to draft the manuscript.
All authors have read and approved the final manuscript.
Acknowledgements
This work was partially supported by a grant of the Fundación de la Sociedad Española de Alergología e Inmunología Clínica and a grant from the Junta de Castilla y Leon.
The authors would like thank Mrs. Nieves Mateos for her technical support.
==== Refs
Marsh DG Neely JD Breazeale DR Ghosh B Freidhoff LR Ehrlich-Kautzky E Linkage analysis of IL4 and other chromosome 5q31.1 markers and total serum immunoglobulin E concentrations Science 1994 264 1152 1156 8178175
Doull IJ Lawrence S Watson M Begishvili T Beasley RW Lampe F Allelic association of gene markers on chromosomes 5q and 11q with atopy and bronchial hyperresponsiveness Am J Respir Crit Care Med 1996 153 1280 1284 8616554
Palmer LJ Daniels SE Rye PJ Gibson NA Tay GK Cookson WO Linkage of chromosome 5q and 11q gene markers to asthma-associated quantitative traits in Australian children Am J Respir Crit Care Med 1998 158 1825 1830 9847274
Basehore MJ Howard TD Lange LA Moore WC Hawkins GA Marshik PL A comprehensive evaluation of IL4 variants in ethnically diverse populations: association of total serum IgE levels and asthma in white subjects J Allergy Clin Immunol 2004 114 80 87 15241348 10.1016/j.jaci.2004.05.035
Rosenwasser LJ Borish L Genetics of atopy and asthma: the rationale behind promoter-based candidate gene studies (IL-4 and IL-10) Am J Respir Crit Care Med 1997 156 S152 S155 9351597
Suzuki I Yamaguchi E Hizawa N Itoh A Kawakami Y A new polymorphism in the 5' flanking region of the human interleukin (IL)-4 gene Immunogenetics 1999 49 738 739 10369939 10.1007/s002510050677
Kabesch M Tzotcheva I Carr D Hofler C Weiland SK Fritzsch C A complete screening of the IL4 gene: novel polymorphisms and their association with asthma and IgE in childhood J Allergy Clin Immunol 2003 112 893 898 14610476 10.1016/j.jaci.2003.08.033
Suzuki I Hizawa N Yamaguchi E Kawakami Y Association between a C+33T polymorphism in the IL-4 promoter region and total serum IgE levels Clin Exp Allergy 2000 30 1746 1749 11122213 10.1046/j.1365-2222.2000.00983.x
Elliott K Fitzpatrick E Hill D Brown J Adams S Chee P The -590C/T and -34C/T interleukin-4 promoter polymorphisms are not associated with atopic eczema in childhood J Allergy Clin Immunol 2001 108 285 287 11496248 10.1067/mai.2001.117180
Noguchi E Nukaga-Nishio Y Jian Z Yokouchi Y Kamioka M Yamakawa-Kobayashi K Haplotypes of the 5' region of the IL-4 gene and SNPs in the intergene sequence between the IL-4 and IL-13 genes are associated with atopic asthma Hum Immunol 2001 62 1251 1257 11704288 10.1016/S0198-8859(01)00338-X
Nelms K Keegan AD Zamorano J Ryan JJ Paul WE The IL-4 receptor: signaling mechanisms and biologic functions Annu Rev Immunol 1999 17 701 738 10358772 10.1146/annurev.immunol.17.1.701
Miloux B Laurent P Bonnin O Lupker J Caput D Vita N Cloning of the human IL-13R alpha1 chain and reconstitution with the IL4R alpha of a functional IL-4/IL-13 receptor complex FEBS Lett 1997 401 163 166 9013879 10.1016/S0014-5793(96)01462-7
Mak T Simard J Handbook of immune response genes 1998 Plenum Press Ref Type: Generic
Pritchard MA Baker E Whitmore SA Sutherland GR Idzerda RL Park LS The interleukin-4 receptor gene (IL4R) maps to 16p11.2-16p12.1 in human and to the distal region of mouse chromosome 7 Genomics 1991 10 801 806 1679753 10.1016/0888-7543(91)90466-R
Ober C Tsalenko A Parry R Cox NJ A second-generation genomewide screen for asthma-susceptibility alleles in a founder population Am J Hum Genet 2000 67 1154 1162 11022011
Deichmann KA Heinzmann A Forster J Dischinger S Mehl C Brueggenolte E Linkage and allelic association of atopy and markers flanking the IL4-receptor gene Clin Exp Allergy 1998 28 151 155 9515586 10.1046/j.1365-2222.1998.00159.x
Deichmann K Bardutzky J Forster J Heinzmann A Kuehr J Common polymorphisms in the coding part of the IL4-receptor gene Biochem Biophys Res Commun 1997 231 696 697 9070874 10.1006/bbrc.1997.6115
Lozano F Places L Vila JM Padilla O Arman M Gimferrer I Identification of a novel single-nucleotide polymorphism (Val554Ile) and definition of eight common alleles for human IL4RA exon 11 Tissue Antigens 2001 57 216 220 11285129 10.1034/j.1399-0039.2001.057003216.x
Hershey GK Friedrich MF Esswein LA Thomas ML Chatila TA The association of atopy with a gain-of-function mutation in the alpha subunit of the interleukin-4 receptor N Engl J Med 1997 337 1720 1725 9392697 10.1056/NEJM199712113372403
Franjkovic I Gessner A Konig I Kissel K Bohnert A Hartung A Effects of common atopy-associated amino acid substitutions in the IL-4 receptor alpha chain on IL-4 induced phenotypes Immunogenetics 2005 56 808 817 15712015 10.1007/s00251-004-0763-1
Kruse S Japha T Tedner M Sparholt SH Forster J Kuehr J The polymorphisms S503P and Q576R in the interleukin-4 receptor alpha gene are associated with atopy and influence the signal transduction Immunology 1999 96 365 371 10233717 10.1046/j.1365-2567.1999.00705.x
Shirakawa I Deichmann KA Izuhara I Mao I Adra CN Hopkin JM Atopy and asthma: genetic variants of IL-4 and IL-13 signalling Immunol Today 2000 21 60 64 10652462 10.1016/S0167-5699(99)01492-9
Sandford AJ Chagani T Zhu S Weir TD Bai TR Spinelli JJ Polymorphisms in the IL4, IL4RA, and FCERIB genes and asthma severity J Allergy Clin Immunol 2000 106 135 140 10887316 10.1067/mai.2000.107926
Kauppi P Lindblad-Toh K Sevon P Toivonen HT Rioux JD Villapakkam A A second-generation association study of the 5q31 cytokine gene cluster and the interleukin-4 receptor in asthma Genomics 2001 77 35 42 11543630 10.1006/geno.2001.6613
Wjst M Kruse S Illig T Deichmann K Asthma and IL-4 receptor alpha gene variants Eur J Immunogenet 2002 29 263 268 12047364 10.1046/j.1365-2370.2002.00300.x
Bottini N Borgiani P Otsu A Saccucci P Stefanini L Greco E IL-4 receptor alpha chain genetic polymorphism and total IgE levels in the English population: two-locus haplotypes are more informative than individual SNPs Clin Genet 2002 61 288 292 12030894 10.1034/j.1399-0004.2002.610408.x
Cui T Wang L Wu J Hu L Xie J Polymorphisms of IL-4, IL-4R alpha, and AICDA genes in adult allergic asthma J Huazhong Univ Sci Technolog Med Sci 2003 23 134 137 12973929
Isidoro-Garcia M Davila I Moreno E Laffond E Lorente F Gonzalez-Sarmiento R [IL4RA gene polymorphism (Q576R) is associated with higher total IgE levels in Spanish patients with family history of atopy] Med Clin (Barc) 2005 124 211 212 15737301 10.1157/13071760
Hytonen AM Lowhagen O Arvidsson M Balder B Bjork AL Lindgren S Haplotypes of the interleukin-4 receptor alpha chain gene associate with susceptibility to and severity of atopic asthma Clin Exp Allergy 2004 34 1570 1575 15479272 10.1111/j.1365-2222.2004.02069.x
Lee SG Kim BS Kim JH Lee SY Choi SO Shim JY Gene-gene interaction between interleukin-4 and interleukin-4 receptor alpha in Korean children with asthma Clin Exp Allergy 2004 34 1202 1208 15298559 10.1111/j.1365-2222.2004.02015.x
Migliaccio C Patuzzo C Malerba G Trabetti E Galavotti R Pescollderung L No linkage or association of five polymorphisms in the interleukin-4 receptor alpha gene with atopic asthma in Italian families Eur J Immunogenet 2003 30 349 353 14641543 10.1046/j.1365-2370.2003.00415.x
Shi YY He L SHEsis, a powerful software platform for analyses of linkage disequilibrium, haplotype construction, and genetic association at polymorphism loci Cell Res 2005 15 97 98 15740637
Rosenwasser LJ Promoter polymorphism in the candidate genes, IL-4, IL-9, TGF-beta1, for atopy and asthma Int Arch Allergy Immunol 1999 118 268 270 10224407 10.1159/000024096
Burchard EG Silverman EK Rosenwasser LJ Borish L Yandava C Pillari A Association between a sequence variant in the IL-4 gene promoter and FEV(1) in asthma Am J Respir Crit Care Med 1999 160 919 922 10471619
Beghe B Barton S Rorke S Peng Q Sayers I Gaunt T Polymorphisms in the interleukin-4 and interleukin-4 receptor alpha chain genes confer susceptibility to asthma and atopy in a Caucasian population Clin Exp Allergy 2003 33 1111 1117 12911786 10.1046/j.1365-2222.2003.01731.x
Noguchi E Shibasaki M Arinami T Takeda K Yokouchi Y Kobayashi K Lack of association of atopy/asthma and the interleukin-4 receptor alpha gene in Japanese Clin Exp Allergy 1999 29 228 233 10051727 10.1046/j.1365-2222.1999.00458.x
Malerba G Trabetti E Patuzzo C Lauciello MC Galavotti R Pescollderungg L Candidate genes and a genome-wide search in Italian families with atopic asthmatic children Clin Exp Allergy 1999 29 27 30 10641562
Patuzzo C Trabetti E Malerba G Martinati LC Boner AL Pescollderungg L No linkage or association of the IL-4Ralpha gene Q576R mutation with atopic asthma in Italian families J Med Genet 2000 37 382 384 10905893 10.1136/jmg.37.5.382
Colhoun HM McKeigue PM Davey SG Problems of reporting genetic associations with complex outcomes Lancet 2003 361 865 872 12642066 10.1016/S0140-6736(03)12715-8
Risma KA Wang N Andrews RP Cunningham CM Ericksen MB Bernstein JA V75R576 IL-4 receptor alpha is associated with allergic asthma and enhanced IL-4 receptor function J Immunol 2002 169 1604 1610 12133990
|
16313681
|
PMC1310519
|
CC BY
|
2021-01-04 16:36:25
|
no
|
Clin Mol Allergy. 2005 Nov 29; 3:15
|
utf-8
|
Clin Mol Allergy
| 2,005 |
10.1186/1476-7961-3-15
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-731630573910.1186/1742-4690-2-73ResearchCharacterization of two candidate genes, NCoA3 and IRF8, potentially involved in the control of HIV-1 latency Munier Sandie [email protected]ête Delphine [email protected]éna Laëtitia [email protected] Audrey [email protected] Uriel [email protected] Département des Maladies Infectieuses, Institut Cochin, INSERM U567/CNRS UMR-S 8104/Université Paris 5-René Descartes, 22 rue Méchain, 75014 Paris, France2 UFR de Biochimie, Université Paris 7-Denis Diderot, 2 Place Jussieu, 75251 Paris, France2005 23 11 2005 2 73 73 28 7 2005 23 11 2005 Copyright © 2005 Munier et al; licensee BioMed Central Ltd.2005Munier et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The persistence of latent HIV-1 reservoirs is the principal barrier preventing the eradication of HIV-1 infection in patients by current antiretroviral therapy. It is thus crucial to understand the molecular mechanisms involved in the establishment, maintenance and reactivation of HIV-1 latency. Since chromatin remodeling has been implicated in the transcriptional reactivation of the HIV-1 promoter, we assessed the role of the histone deacetylase inhibitor sodium butyrate (NaB) on two HIV-1 latently infected cell lines (U1 and ACH-2) gene expression.
Results
Analysis of microarrays data led us to select two candidate genes: NCoA3 (Nuclear Receptor Coactivator 3), a nuclear receptor coactivator and IRF8 (Interferon Regulatory Factor 8), an interferon regulatory factor. NCoA3 gene expression is upregulated following NaB treatment of latently infected cells whereas IRF8 gene expression is strongly downregulated in the promonocytic cell line following NaB treatment. Their differential expressions were confirmed at the transcriptional and translational levels. Moreover, NCoA3 gene expression was also upregulated after treatment of U1 and ACH-2 cells with phorbol myristyl acetate (PMA) but not trichostatin A (TSA) and after treatment with NaB of two others HIV-1 latently infected cell lines (OM10.1 and J1.1). IRF8 gene is only expressed in U1 cells and was also downregulated after treatment with PMA or TSA. Functional analyses confirmed that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription and that IRF8 represses the IRF1-mediated activation through the HIV-1 promoter Interferon-stimulated response element (ISRE).
Conclusion
These results led us to postulate that NCoA3 could be involved in the transcriptional reactivation of the HIV-1 promoter from latency and that IRF8 may contribute to the maintenance of the latent state in the promonocytic cell line. Implication of these factors in the maintenance or reactivation of the viral latency may provide potential new targets to control HIV-1 replication in latent viral reservoirs.
==== Body
Background
The use of highly active antiretroviral therapy (HAART) in HIV-1 infected individuals has led to a significant decrease of plasma viremia to undetectable levels and has considerably improved the survival and quality of life of infected individuals (reviewed in [1]). However, the presence of cellular reservoirs that contain latent viruses capable of producing infectious particles after cellular activation lead to a rebound of the viral load after interruption of HAART (reviewed in [2]). The persistence of these latently infected viral reservoirs, despite prolonged HAART treatments, represents a major obstacle to the eradication of HIV-1 in infected patients [3-5]. Therefore, a greater understanding of the molecular mechanisms involved in establishment, maintenance and reactivation of viral latency is essential to expect the reduction of latent HIV-1 reservoirs in infected patients.
Latent HIV-1 infection can exist in many reservoirs, such as macrophages and resting memory CD4+ T cells (reviewed in [6]). At the cellular level, two major forms of HIV-1 latency have been described: pre- and post-integration latency [7]. CD4+ T cells in the post-integration state of latency represent the most stable reservoir for HIV-1 (half-life of 43 months) [8]. Several mechanisms have been proposed to account for the low level of transcription observed during post-integration latency (reviewed in [9]): the inaccessibility of the integrated provirus to the transcriptional machinery, the absence in resting cells of transcription factors involved in HIV-1 gene expression, the presence of transcriptional repressors, and the premature termination of HIV-1 transcription elongation due to the absence of the viral protein Tat and its associated cofactors. Moreover, the chromatin structure appears to be involved in the regulation of HIV-1 gene expression (reviewed in [10]). Indeed, a repressive nucleosome (nuc-1), located immediately downstream of the HIV-1 transcription start site under latency conditions, is disrupted upon transcriptional activation of the HIV-1 promoter in response to Tat, phorbol esters and histone deacetylase (HDAC) inhibitors [11]. Transcriptional activation of the HIV-1 promoter in response to PMA involves the recruitment of SWI/SNF chromatin remodeling complex [12] and cellular proteins with histone acetyltransferase (HAT) activity [13]. Therefore, chromatin remodeling plays a significant role in the transcriptional reactivation of the HIV-1 promoter from latency. Identification of host transcription factors that may regulate chromatin structure is thus critical to understand the molecular mechanisms involved in HIV-1 reactivation.
Gene expression analysis using high-density microarrays have provided a greater understanding of host-pathogen interactions (reviewed in [14]). Previous microarray studies on HIV-1 have described changes in cellular genes transcription in response to HIV-1 protein expression (Nef [15,16], Tat [17,18], gp120 [19] or Vpr [20]) or following acute infection of cell lines [21-24] or Peripheral Blood Mononuclear Cells (PBMC) [25]. DNA microarrays have also been used to characterize gene expression in latently infected resting CD4+ T cells in viremic versus aviremic HIV-1 infected individuals [26]. Recently, global gene expression changes in cell lines latently infected with HIV-1 and induced by PMA for completion of viral replication was described by Krishnan et al. [27].
To complete the results obtained by Krishnan et al., we used the same strategy to assess the role of the HDAC inhibitor NaB on HIV-1 latently infected cells gene expression. We performed microarray experiments on two HIV-1 latently infected cell lines (U1 and ACH-2) treated or not with NaB to induce viral reactivation. Analysis of microarrays data led us to select two candidate genes encoding transcription factors: NCoA3 (reviewed in [28]), which expression is upregulated following treatment of latently infected cells with NaB, and IRF8 (reviewed in [29]), which expression is downregulated in treated cells. Differential expression of these genes was confirmed at the transcriptional and translational levels. Moreover, NCoA3 gene expression was also upregulated after treatment of U1 and ACH-2 cells with PMA but not TSA and after treatment with NaB of two others latently infected cell lines (OM10.1 and J1.1). IRF8 gene is only expressed in U1 cells and was also downregulated after treatment with PMA or TSA. Functional analyses confirmed that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription and that IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element. Implication of IRF8 in the maintenance and NCoA3 in the reactivation of the viral latency may thus provide new insights into the control of HIV-1 replication in latent viral reservoirs.
Results
Microarray analysis
In order to understand the molecular mechanisms regulating HIV-1 latency, we studied the modifications of cellular transcription using microarrays in the promonocytic U1 and T CD4+ lymphocytic ACH-2 chronically HIV-1 infected cell lines after reactivation of latency. The two cell lines were treated with 10 mM of the histone deacetylase inhibitor NaB. Viral reactivation was monitored by coculture with P4 indicating cells (Figure 1A) and measuring gag viral mRNA expression (Figure 1B). Increase in both β-galactosidase activity and gag mRNA expression showed that the viral reactivation after NaB treatment was efficient. Total RNAs were extracted after 24 h and sent to the Affymetrix Microarray Facilities for subsequent hybridization on U-133A microarrays.
Figure 1 Analysis of viral reactivation after treatment of U1 and ACH-2 cells with NaB. U1 and ACH-2 cells were treated or not (NT) with 10 mM of NaB for 24 h and cocultured with P4 indicating cells. β-galactosidase activity was determined after 24 h coculture (A). RNA from U1 and ACH-2 cells treated or not with NaB were extracted after 24 h and gag viral mRNA expression was measured by real-time RT-PCR (B). Results are representative of three independent experiments.
The pattern of cellular mRNA from chronically infected cells treated with NaB was compared to that from non-treated cells. We used as specific criteria a log2 ratio change ≥ 1 with a change p-value ≤ 0.0001 for increased genes and a log2 ratio change ≤ -1 with a 1-change p-value ≥ 0.9999 for decreased genes. Hybridization experiments were performed once. We identified 740 genes that were upregulated by twofold or higher in NaB treated U1 cells and 896 genes that were downregulated, 482 genes in NaB treated ACH-2 cells that had a level increased greater than twofold and 634 genes that had a level decreased greater than twofold (data not shown). Moreover, 260 genes were commonly increased and 337 genes were decreased in both U1 and ACH-2 NaB-treated cells (data not shown). Pathways involved in regulation of transcription, signal transduction, immune response, protein transport, metabolism, apoptosis and RNAs modifications showed altered expression following treatment with NaB. Some of the genes involved in these pathways are shown in Additional Files 1, 2, 3, 4, 5 and 6. Our analysis identified genes that have previously been associated with HIV-1 replication or latency, such as CDK9 [16], Jun [16,23], PSMB10 [27], MAPK1 [26] or OAS1 [30]. This supported the accuracy of our approach, even though, as the hybridization experiments had been performed once, the statistical relevance of the results could not be estimated.
Among the differentially expressed genes, we chose to focus on two candidate genes encoding transcription factors: NCoA3 and IRF8 (Tables 1 and 2). We selected these two genes based on their biological properties, their described effects on viral replication [31,32] and their differential expression observed by microarray experiments. Indeed, NCoA3 and IRF8 gene expression are respectively upregulated and downregulated following treatment with NaB of latently infected cells (Tables 1 and 2). Therefore, NCoA3 and IRF8 could be implicated respectively in the reactivation and maintenance of HIV-1 latency.
Table 1 Differential gene expression obtained for NCoA3 and IRF8 mRNAs in U1 cells treated or not with NaB.
Gene Probe set Name a U1 Signal b U1 Detection p-value c U1NaB Signal U1NaB Detection p-value U1NaBvsU1 Signal log2 ratio d U1NaBvsU1 Change p-value e
NCoA3 207700_s_at 17.7 0.01416 98.9 0.000244 2.5 0.000035
209060_x_at 16.9 0.171387 77.2 0.000244 2.3 0.000023
209061_at 48.4 0.037598 166.4 0.000732 2.3 0.00002
209062_x_at 6.3 0.72583 91.8 0.010742 4.5 0.000147
211352_s_at 7.2 0.303711 68.6 0.00293 3.2 0.000101
IRF8 204057_at 707.9 0.000244 47 0.010742 -4 0.99998
a Affymetrix U133-A reference probe set.
b Signal intensity of hybridization.
c Signal detection p-value < 0.048 for specific hybridization.
d Signal log2 ratio > 1 for increased genes and < -1 for decreased genes.
e Change p-value < 0.0001 for significant increased genes and 1-change p-value > 0.9999 for significant decreased genes.
Table 2 Differential gene expression obtained for NCoA3 mRNA in ACH-2 cells treated or not with NaB.
Gene Probe set Name a ACH-2 Signal b ACH-2 Detection p-value c ACH2NaB Signal ACH2NaB Detection p-value ACH2NaBvsACH2 Signal log2 ratio d ACH2NaBvsACH2 Change p-value e
NCoA3 207700_s_at 43.3 0.001953 99.6 0.001221 1.2 0.000241
209060_x_at 34.5 0.01416 72.9 0.001953 1 0.000273
209061_at 65.8 0.000732 82.6 0.000732 1.6 0.005409
209062_x_at 20 0.466064 76.7 0.095215 2 0.000114
211352_s_at 2.7 0.5 37 0.030273 3.8 0.004481
a Affymetrix U133-A reference probe set.
b Signal intensity of hybridization.
c Signal detection p-value < 0.048 for specific hybridization.
d Signal log2 ratio > 1 for increased genes and < -1 for decreased genes.
e Change p-value < 0.0001 for significant increased genes and 1-change p-value > 0.9999 for significant decreased genes.
NCoA3 gene expression is upregulated following treatment with NaB of both U1 and ACH-2 latently infected cells (Tables 1 and 2). NCoA3 is a nuclear receptor coactivator of the Steroid Receptor Coactivator (SRC) family that interacts with nuclear receptors in a ligand-dependent manner and enhances transcriptional activation via histone acetylation and recruitment of general transcription factors and additional cofactors (reviewed in [28]). NCoA3 (Unigene Hs. 382168) gene expression in U1 cells is significantly upregulated by 4.9 to 22.6 fold (U1NaBvsU1 signal log2 ratio ranging from 2.3 to 4.5 with a change p-value < 0.00015) following treatment with NaB (Table 1). Similarly, NCoA3 gene expression is upregulated in NaB-treated compared to non-treated ACH-2 cells by 2 to 13.9 fold but with a lower significance (ACH2NaBvsACH2 signal log2 ratio ranging from 1 to 3.8 with a change p-value < 0.0055) (Table 2).
IRF8 gene expression is downregulated following treatment of U1 cells with NaB (Table 1). IRF8 is a transcription factor of the Interferon (IFN) Regulatory Factor (IRF) family that binds to IFN-stimulated response element and regulates expression of genes stimulated by IFNs (reviewed in [29]). IRF8 (Unigene Hs. 137427) is expressed in the promonocytic cell line U1 (detection signal of 707.9 with a p-value of 0.000244) (Table 1) but is not expressed in the T CD4+ lymphocytic cell line ACH-2 (data not shown). Following NaB treatment, IRF8 gene expression in U1 cells is downregulated by 16 fold (U1NaBvsU1 signal log2 ratio of -4 with a 1-change p-value of 0.99998) (Table 1).
Validation of NCoA3 and IRF8 differential transcriptional expression
Real-time RT-PCR quantifications were performed to confirm that NCoA3 and IRF8 genes were differentially expressed in the NaB-treated chronically infected cells compared to the non-treated cells. We performed quantification on RNA samples obtained from five independent NaB treatments of U1 and ACH-2 cells and real-time RT-PCR experiments were run in duplicate. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The results show in Figure 2 represent the NCoA3 expression increase fold (Figure 2A) obtained from U1 and ACH-2 cells and the IRF8 expression decrease fold (Figure 2B) obtained from U1 cells treated with NaB for 24 h and 48 h compared to non-treated cells. Concerning NCoA3, real-time RT-PCR showed an upregulation consistent with microarray data in 24 h NaB-treated U1 cells of 8.34 ± 2.42 fold compared to non-treated cells (Figure 2A). NCoA3 gene expression is also increased with a 48 h NaB treatment (upregulation of 8.40 ± 2.33 fold) (Figure 2A). Similarly, an increase of NCoA3 gene expression can be observed on ACH-2 cells following treatment with NaB (upregulation of 4.56 ± 1.28 fold in 24 h and 6.80 ± 2.34 fold in 48 h NaB-treated ACH-2 cells) (Figure 2A). Concerning IRF8, real-time RT-PCR showed a 14.96 ± 4.85 fold decrease in 24 h NaB-treated U1 cells (Figure 2B) in correlation with the microarray ratio previously obtained. Downregulation of IRF8 gene expression is also observed following 48 h NaB-treatment of U1 cells (22.06 ± 11.29 fold decrease) (Figure 2B). Taken together, results from real-time RT-PCR performed on NCoA3 and IRF8 genes corroborate with those obtained using microarray hybridizations.
Figure 2 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in NaB-treated U1 and ACH-2 cells. Total RNAs were isolated from U1 or ACH-2 cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U1 (solid bars) or ACH-2 (white bars) cells and the IRF8 decrease fold (B) in U1 cells treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of five independent experiments performed in duplicate.
We next determined whether NCoA3 and IRF8 gene expression were similarly modified in the uninfected parental cell lines. U937 and CEM cells were subjected to identical treatment and RT-PCR quantifications were performed (Figure 3). NCoA3 is upregulated both in U937 and CEM cells following treatment with NaB (upregulation of 7.32 ± 1.74 fold in 24 h and 11.45 ± 2.95 fold in 48 h NaB-treated U937 cells, upregulation of 1.93 ± 1.04 fold in 24 h and 5.59 ± 0.06 fold in 48 h NaB-treated CEM cells) (Figure 3A). IRF8 is only expressed in the promonocytic cell line U937 and, as in U1 cells, its expression was downregulated after NaB treatment (downregulation of 17.95 ± 4.15 fold in 24 h and 22.32 ± 10.82 fold in 48 h NaB-treated U937 cells) (Figure 3B). Thus, NaB treatment modify NCoA3 and IRF8 gene expression in uninfected parental cell lines U937 and CEM at a similar level than in chronically infected cells.
Figure 3 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in NaB-treated U937 and CEM cells. Total RNAs were isolated from U937 or CEM cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U937 (solid bars) or CEM (white bars) cells and the IRF8 decrease fold (B) in U937 cells treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of five independent experiments performed in duplicate.
We then performed additional experiments to determine whether the gene expression variations observed could also be mediated by treatments with the phorbol ester PMA and another HDAC inhibitor, TSA. We thus assessed the differential regulation of NCoA3 and IRF8 gene expression in U1 and ACH-2 cells treated with PMA or TSA (Figure 4). Results indicated that NCoA3 expression is upregulated by 24 h and 48 h PMA treatment of U1 and ACH-2 cells (upregulation of 5.70 ± 1.37 fold in 24 h and 9.85 ± 0.90 fold in 48 h PMA-treated U1 cells, upregulation of 3.12 ± 1.05 fold in 24 h and 7.12 ± 1.20 fold in 48 h PMA-treated ACH-2 cells (Figure 4A). However, TSA treatment had no significant effect on NCoA3 expression in U1 and ACH-2 cells, although TSA increased viral expression (data not shown). Concerning IRF8 expression in U1 cells, PMA and TSA treatments for 24 h induced a decrease of 3.22 ± 0.45 fold and 5.32 ± 1.09 fold, respectively (Figure 4B). These results show that NCoA3 expression is upregulated following phorbol ester but not with other HDAC inhibitor treatments in U1 and ACH-2 cells. Moreover, IRF8 gene expression in U1 cells is downregulated with PMA or TSA treatments but at a lower extent than with NaB.
Figure 4 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in PMA- or TSA-treated U1 and ACH-2 cells. Total RNAs were isolated from U1 or ACH-2 cells treated or not with PMA for 24 h and 48 h or TSA for 24 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U1 (solid bars) or ACH-2 (white bars) cells treated with PMA for 24 h and 48 h and the IRF8 decrease fold (B) in U1 cells treated with PMA or TSA for 24 h compared to non-treated (NT) cells were determined. Results represent the means of three independent experiments performed in duplicate.
We also assessed the differential regulation of NCoA3 and IRF8 gene expression in others chronically HIV-1 infected cell lines. The chronically infected promonocytic OM10.1 and T CD4+ lymphocytic J1.1 cell lines were treated with NaB for 24 h and 48 h and real-time RT-PCR were performed to measure NCoA3 and IRF8 gene expression. As shown in Figure 5, NCoA3 expression is upregulated by 4.94 ± 0.78 fold in OM10.1 and by 2.56 ± 0.64 fold in J1.1 after 24 h NaB treatment. NCoA3 expression increased with time of NaB treatment in both cell lines (upregulation of 12.89 ± 3.10 fold in OM10.1 and 3.51 ± 0.69 fold in J1.1 cells) (Figure 5). Like ACH-2 and unlike U1 cells, the T CD4+ lymphocytic J1.1 and the promonocytic OM10.1 cell lines did not express IRF8 (data not shown). Thus, the differential regulation of NCoA3 but not IRF8 gene expression is similar in two other related latently HIV-1 infected cell line models.
Figure 5 Real-time RT-PCR analysis of NCoA3 mRNAs expression in OM10.1 and J1.1 cells. Total RNAs were isolated from OM10.1 or J1.1 cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3 or Cyclophilin A. NCoA3 expression was normalized to the expression of Cyclophilin A. The NCoA3 increase fold in OM10.1 (solid bars) or J1.1 cells (white bars) treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of two independent experiments performed in duplicate.
gag mRNA activation is correlated with NCoA3 mRNA increase and IRF8 mRNA decrease
We performed reactivation experiments at different times, sooner than 24 h and until 48 h. Quantitative RT-PCR experiments were carried out on total RNAs. This was done using U1 cells to analyze both NCoA3 mRNA increase (Figure 6A) and IRF8 mRNA decrease (Figure 6B) relative to HIV gag mRNA along with ACH-2 cells (Figure 6C) to analyze NCoA3 mRNA increase relative to HIV gag mRNA.
Figure 6 Analysis of HIV gag, NCoA3, and IRF8 mRNA expression after NaB stimulation on U1 and ACH-2 cells. U1 (A and B) and ACH-2 (C) cells were stimulated with 10 mM NaB and 5.106 cells were taken at t = 0, 4, 8, 16, 24, 48 h for RNA extraction to perform qRT-PCR. NCoA3 (A and C), IRF8 (B) and gag (A, B and C) mRNA contents were measured. Cylophilin A was used as internal standard. Results represent a representative experiment performed in duplicate.
As observed on Figure 6C, the obtained results, both on ACH-2 and U1 cells, clearly show that gag mRNA activation occurs after NCoA3 mRNA increase and accumulation. Moreover, in U1 cells, gag mRNA activation occurs after IRF8 mRNA decrease. Shorter kinetics (0 to 8 h) correlated with these results (data not shown).
Validation of NCoA3 and IRF8 differential translational expression
To confirm that the changes seen at the RNA level correlated with protein levels, we performed Western blot experiments on nuclear extract of U1, ACH-2, OM10.1 and J1.1 cells treated or not with NaB for 24 h (Figure 7). Results indicated that NaB increased the expression level of NCoA3 protein in U1, ACH-2, OM10.1 and not in J1.1 cells (Figure 7). Moreover, IRF8 protein expression is strongly downregulated in U1 cells following NaB treatment (Figure 7). These results correlate with the differential expression of NCoA3 and IRF8 genes observed with both microarray and real-time RT-PCR experiments.
Figure 7 Western blot analysis of NCoA3 and IRF8 proteins expression. Nuclear extract (100 μg) from U1, ACH-2, J1.1 and OM10.1 treated (+) or not (-) with NaB for 24 h were resolved by SDS-PAGE and immunoblotted with anti-NCoA3 or anti-IRF8 antibody, as indicated. The amount of protein was normalized using anti-actin antibody. Figures below NCoA3 immunoblot indicated the results of the quantification using Image Tool (Syngene) software of the ratio NCoA3/actin upon NaB-treatment (+) versus NCoA3/actin non-treated (-). Results are representative of three independent experiments.
Transcriptional activation of the HIV-1 promoter by NCoA3
We analyzed the functional role of NCoA3 on viral transcription by transfection assays. HEK293 cells were cotransfected with pLTRX-luc reporter plasmid containing the luciferase gene under the control of the HIV-1 U3-R promoter region (nt -640 to +78) with or without Tat and/or NCoA3 expression vectors. As shown in Figure 8, NCoA3 increased Tat-stimulated HIV-1 LTR activity by 2.8 ± 1.4 fold. The presence of NCoA3 had synergistic effect on the HIV-1 LTR activity induced by suboptimal expression of Tat. When HEK293 cells were transfected with pLTRΔTAR-luc reporter plasmid lacking the Tat-transactivation response element TAR, Tat was not able to activate the LTR transcription, as expected, and NCoA3 had no effect on the LTR activity (Figure 8). Thus, functional analyses confirm that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription, as expected [31], and that this effect is dependent on the presence of the TAR region.
Figure 8 NCoA3 increases the Tat-stimulated HIV-1 LTR activity. HEK293 cells were cotransfected with pLTRX-luc (10 ng, grey bars) or pLTRΔTAR-luc (10 ng, white bars) with (+) or without (-) suboptimal amounts of pCMV-Tat (5 ng) and/or pNCoA3 (1 μg) expression vectors. NLI (normalized luciferase index) were measured after 24 h and the activation folds compared to the basal activity of the corresponding pLTR-luc were determined. Results represent the means of five independent experiments.
Transcriptional repression of the HIV-1 ISRE element by IRF8
We analyzed the functional role of IRF8 on viral transcription by transfection assays. HEK293 cells were cotransfected with pISRE-TK-luc reporter plasmid corresponding to the HIV-1 IFN-stimulated response element, located downstream transcription start site (nt +194 to +223) [33], with or without IRF1 and/or IRF8 expression vectors. As shown in Figure 9, the basal activity of the ISRE-TK was increased by 7.4 ± 1.0 fold in the presence of IRF1 as expected [32], whereas a decrease was detected in the presence of IRF8 (21.9 ± 10.6 to 41.4 ± 9.5 %). The expression of IRF8 inhibited by 43.5 ± 10.6 to 74.7 ± 2.5 % the IRF1-mediated activation of the ISRE-TK in a dose dependent fashion (Figure 9). The expression of the dominant negative IRF8 DNA-binding domain (IRF8-DBD) inhibited by 76.4 ± 6.5 % the IRF1-mediated activation of the ISRE-TK, as expected [34] (Figure 9). The inhibitory effects of IRF8 and IRF8-DBD expression and activation effect of IRF1 expression was abolished when the ISRE sequence was mutated (pISREmut-TK-luc, Figure 9). These results show that IRF8 represses the ISRE-TK promoter transcription through the ISRE element from the HIV-1 promoter, as expected [32].
Figure 9 IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element. HEK293 cells were cotransfected with pISRE-TK-luc (250 ng, solid bars) or pISREmut-TK-luc (250 ng, white bars) with (+) or without (-) pIRF1 (250 ng), pIRF8 (1–2.5 μg), or pIRF8-DBD (1 μg) expression vectors. NLI (normalized luciferase index) were measured after 24 h and the activation folds compared to the basal activity of the pISRE-TK-luc or pISREmut-TK-luc were determined. Results represent the means of five independent experiments.
Discussion
The existence of long-lasting HIV-1 reservoirs is the principal barrier preventing the eradication of HIV-1 infection in patients by current antiretroviral therapy. It is thus crucial to understand the molecular mechanisms involved in establishment, maintenance and reactivation of HIV-1 latency. In this study, the role of the HDAC inhibitor NaB on HIV-1 latently infected cells gene expression was explored using microarrays. Since chromatin remodeling is involved in the regulation of HIV-1 gene expression (reviewed in [10]), differential expression of cellular genes in latently infected cells following treatment with NaB might be related to the maintenance and reactivation of latency.
Recently, Krishnan et al. [27] described the global gene expression changes in HIV-1 latently infected cell lines treated or not with PMA to induce viral reactivation compared to the uninfected parental cell lines treated under the same conditions. Here, we compared gene expression profiles of two HIV-1 latently infected cell lines (U1 and ACH-2) treated with NaB to that of non-treated corresponding cell lines. We thus avoided identification of genes which differential expression could result from the establishment and cloning of the chronically infected cell lines. Based on our specific criteria, we identified few hundreds of genes affected by NaB treatment implicated in biological pathways previously shown to be modulated by HIV-1 replication. For example, reactivation of latency induced an upregulation of CDK9, the catalytic component of transcription elongation factor b (P-TEFb), which acts in concert with Tat to direct the processivity of HIV-1 transcription. It was shown that CDK9 mRNA and protein levels are induced following T cell activation and Nef expression, and that this correlates with kinase activity, thus enhancing HIV-1 transcription [16,35].
After NaB treatment of latently infected cell lines, we observed an upregulation of genes involved in vesicular transport of protein like syntaxin and nexin. It was found by Chun et al. that numerous genes involved in protein/vesicle transport are upregulated in resting T CD4+ cells of viremic patients, strongly suggesting that enhanced activities in secretory pathways may help in the assembly and release of viral particles [26]. Recently, it was shown that multiple genes involved in cholesterol synthesis are induced by Nef [36]. NaB treatment also induced some of these genes (INSIG1, HMGCS1, IDI1, LSS or SREBF1) and could thus enhanced virion infectivity and viral replication.
Krishnan et al. have described an increase in expression of several proteasome subunits in ACH-2 cells prior induction of lytic replication by PMA and proposed that the higher expression of proteasomes may lead to increased degradation of HIV-1 mRNA [27]. After induction of lytic replication by NaB, proteasome subunits PSMB10 and PSMB8 were downregulated in ACH-2 and U1 cells, suggesting a role in the maintenance of the latent state. Indeed, reactivation of latency was achieved with proteasome inhibitors [27]. Among the downregulated genes after NaB treatment, we identified genes involved in RNA modifications. Krishnan et al. have shown alterations in the expression of DEAD-box and other RNA binding proteins during HIV-1 replication [37]. Especially, DDX18 and DDX39 are upregulated in latently infected cells [37]. After NaB treatment of latently infected cells, we observed a decrease in the expression of these two proteins, thus providing more support for their role in maintaining HIV-1 latency.
The only purpose of our microarray analysis was to identify candidate genes potentially involved in the control of the HIV latency. For this reason, we decided to focus on two candidate genes previously described to influence viral expression and that may be involved in reactivation and maintenance of latency: NCoA3 and IRF8, respectively. Hybridization experiments were performed once. Consequently, we did not further analyze the statistical relevance of the results and performed complementary approaches to confirm the mRNA variations of the selected candidate genes.
NCoA3 is a nuclear receptor coactivator that enhances ligand-induced transcriptional activation of nuclear receptors (reviewed in [28]). We show that NCoA3 (Unigene Hs. 382168) gene expression is upregulated following treatment with NaB of U1 and ACH-2 latently infected cells. This differential transcriptional expression was confirmed by real-time RT-PCR and is also mediated by PMA but not TSA. Upregulation of NCoA3 is thus achieved following phorbol ester but not other HDAC inhibitor treatment. However, NaB and TSA act on different pathways and at different concentrations and target different genes [38]. Transcriptional increase of NCoA3 was observed in parental uninfected corresponding cell lines U937 and CEM and in two others latently HIV-1 infected cell lines, OM10.1 and J1.1. NCoA3 protein level is also upregulated following treatment with NaB in the U1, ACH-2 and OM10.1 cell lines. Moreover, NCoA3 increases the Tat-induced HIV-1 LTR promoter transcriptional activity through the TAR region, in accordance with other data [31]. The differential expression of NCoA3 observed led us to postulate that NCoA3 could be involved in the transcriptional reactivation of the HIV-1 promoter from latency, at low concentrations of Tat.
This hypothesis is supported by several findings. Previous microarray studies on latently infected resting CD4+ T cells in infected individuals have shown an upregulation of NCoA3 gene expression in viremic versus aviremic patients [26]. Moreover, Kino et al. showed that NCoA factors improve Tat transactivation of HIV-1 LTR promoter activity and interact with Tat [31]. Tat transactivation activity is mediated by its interaction with components of the basal transcription machinery (including TBP, TAFII250, RNA polymerase II), with kinase complexes able to phosphorylate the C-terminal domain of RNA polymerase II (in particular with the P-TEFb complex composed of cyclin T1/CDK9) and with cellular proteins possessing HAT activity (p300/CBP, P/CAF and GCN5) (reviewed in [39]). Kino et al. showed that one member of the family, NCoA2, functions as a Tat coactivator on the HIV-1 LTR by bridging promoter-bound proteins with the Tat-P-TEFb complex through its interaction with Tat and Cyclin T1 [31]. Stimulation of Tat transactivation activity by NCoA3 could involve similar mechanisms.
Furthemore, it has been recently demonstrated that recruitment of HATs to the LTR is an early event in HIV-1 transcriptional activation [13] and that a consequence of histone acetylation is the recruitment of the ATP-dependent chromatin remodeling complex hSWI/SNF to the LTR [12]. NCoA3 could mediate chromatin remodeling by recruitment of additional cofactors with HAT activity (such as p300/CBP and P/CAF) and by an intrinsic HAT activity [40] and may thus contribute to the transcriptional reactivation of the HIV-1 promoter from latency.
IRF8 is a transcription factor that binds to ISRE and regulates expression of genes stimulated by IFNs (reviewed in [29]). IRF8 is able to both activate and repress gene transcription depending on the target gene. We show that IRF8 (Unigene Hs. 137427) gene is only expressed in the promonocytic cell line U1 and its expression is strongly downregulated following NaB treatment of these cells. This differential transcriptional expression was confirmed by real-time RT-PCR and is also observed, albeit at lower extent, after PMA and TSA treatments of U1 cells. IRF8 protein level is similarly downregulated following treatment with NaB. Moreover, IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element of the LTR, in accordance with other data [32]. The decreased expression of IRF8 following reactivation of latency using different molecules suggest that IRF8 may contribute in the maintenance of the latent state in the promonocytic cell line.
It has been shown that binding of specific transcription factors downstream of the HIV-1 transcription start site is crucial to control HIV-1 transcription [33,41]. Among these sites is an ISRE element that recruits IRF1 and IRF2 in vivo [33]. Previous studies have investigated the role of IRFs on the modulation of HIV-1 replication (reviewed in [42,43]) and showed that IRF1 activates HIV-1 LTR transcription, interacts with Tat [32] and increases HIV-1 replication [44]. However, IRF8 represses IRF1-Tat-mediated transactivation of the LTR by interfering with IRF1-Tat association [32]. Moreover, it has been shown that IRF8 inhibits HIV-1 replication in T CD4+ lymphocytic and promonocytic cell lines [32,34]. These data support the hypothesis that repression of HIV-1 transcription by IRF8 could be implicated in the maintenance of proviral quiescence in latently infected cells.
Moreover, the result obtained after measurement of gag, NCoA3 and IRF8 mRNA after different times of NaB stimulation clearly showed a correlation between gag mRNA increase and NCoA3 mRNA increase or IRF8 mRNA decrease, respectively. These correlations support the hypothesis that IRF8 and NCoA3 factors may be involved in the control of the HIV latency.
Chronically HIV-1 infected cell lines used in this study provide useful models for studying HIV-1 latency but are not in a quiescent state as cellular reservoirs in vivo. Moreover, it has been shown that mutations in the tat gene and in the TAR sequence are responsible for the latency observed in U1 and ACH-2 cells, respectively [45,46]. We thus confirmed the differential expression of NCoA3 but not IRF8 genes in two others chronically HIV-1 infected cell lines, OM10.1 and J1.1. We will now investigate the involvement of NCoA3 and IRF8 to regulate viral expression in primary cells such as resting T CD4+ lymphocytes or macrophages.
Conclusion
Additional experiments are currently underway to validate the biological relevance of the differential expression of IRF8 and NCoA3 genes in latency maintenance and reactivation. Since the persistence of integrated HIV-1 genomes despite potent suppression of viral replication is a major obstacle for current antiretroviral therapy, selective disruption of the HIV-1 proviral latency may provide good strategies to decrease latent HIV-1 reservoirs. Thus, identification of cellular genes that are differentially expressed during HIV-1 reactivation of latency is crucial to understand the molecular mechanisms involved in the control of HIV-1 latency.
Methods
Cell cultures and treatments
The chronically HIV-1 infected T CD4+ lymphocytic cell lines ACH-2 [47] and J1.1 [48] derived from CEM and Jurkat cells respectively, and the chronically HIV-1 infected promonocytic cell lines U1 [49] and OM10.1 [50] derived from U937 and HL-60 cells respectively, were obtained through the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program. Suspension cell lines were grown in RPMI 1640 (Invitrogen) with 10% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen). Cells were treated with 10 mM of sodium butyrate (NaB; Sigma), or with 10 ng/mL of PMA (Sigma), or with 300 nM of TSA (Sigma). Cells were harvested generally 24 h and 48 h after treatment and cell viability was estimated before subsequent RNA extraction or nuclear extract preparation. P4 indicator cells are HeLa CD4+ cells carrying the lacZ gene under the control of the HIV-1 LTR. P4 and HEK293 cells were grown in DMEM (Invitrogen) containing 5% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen).
Plasmids
The pLTRX-luc construct contains the luciferase (luc) gene downstream of the HIV-1 BRU U3-R promoter region (nt -640 to +78) [51]. The pLTRΔTAR-luc construct corresponds to the pLTRX-luc plasmid in which the TAR region (nt +38 to +78) was deleted [51]. The pCMV-Tat expression vector was kindly provided by S. Emiliani (Institut Cochin, Paris, France). The pIRF8 expression vector (pcDNAmycHis-ICSBP) and dominant negative construct pIRF8-DBD, which contains the DNA binding domain of IRF8, were a kind gift of B.Z. Levi (Technion-Israel Institute of Technology, Haifa, Israel). The pNCoA3 expression vector (pcDNA3.1-AIB1) was a kind gift of P.S. Meltzer (NIH, Bethesda, USA) [52]. The pIRF1 construct was generated by cloning the fragment excised from pHuIRF-3-1 (a kind gift of T. Taniguchi, University of Tokyo, Tokyo, Japan) by HindIII/NotI digestion in the pcDNA3.1 plasmid (Invitrogen). The pISRE-TK-luc and pISREmut-TK-luc constructs were generated by cloning a wild-type (AGGGACTTGAAAGCGAAAGGGAAACCAGAG) or mutated (AGGGACTTGCCCGCGCCCGGGAAACCAGAG) synthetic oligonucleotide corresponding to the HIV-1 BRU ISRE sequence (nt +194 to +223) [33,53] in the pTK-luc plasmid in which the luciferase gene is under the control of the truncated HSV-1 thymidine kinase promoter minimum region [51]. The pCMV-LacZ was kindly provided by M. Alizon (Institut Cochin, Paris, France).
Total RNA extraction
Total RNAs were extracted using the RNeasy Mini Kit (Qiagen). The procedure included an "on-column" DNase I digestion step according to the manufacturer's instructions. RNA quality was assessed using the Agilent Bioanalyzer 2100 and spectrophotometric analysis prior to cDNA synthesis.
Microarray experiments
Microarray experiments were performed using the U133-A microarrays (Affymetrix) containing 22283 oligonucleotides spots. Total RNAs obtained from chronically infected U1 and ACH-2 stimulated or not with NaB for 24 h were sent to Dr. C. Thibault (Affymetrix Microarray Facilities, IGBMC, Strasbourg, France) for amplification, labeling and hybridization. Hybridization experiments were performed once. Results were then analyzed with the Mas5.0 Software (Affymetrix) and interpreted using the Data Mining Tool (Affymetrix) and Microsoft Excel softwares. For individual analyses, the p-value cut off was 0.048 as suggested by Affymetrix. For comparative analyses, a log2 ratio change ≥ 1 for increased genes and ≤ -1 for decreased genes were defined. Gene expression changes were considered to be significant when the change p-value was ≤ 0.0001 for increased genes and 1-change p-value ≥ 0.9999 for decreased genes.
Real-time RT-PCR
Quantifications of cellular RNAs were performed using a Light Cycler instrument (Roche Diagnostics). Briefly, cDNAs were synthesized from 1 μg of total RNA with MoMLV reverse transcriptase (Superscript II, Invitrogen) and 1/10th aliquots of the corresponding samples were used for real-time PCR in a 20 μL reaction mixture containing 1X LightCycler FastStart DNA Master SYBR Green I (Roche Diagnostics), 4 mM MgCl2, and 500 nM of each primer. The reactions were carried out in duplicate and the results were normalized to the expression of Cyclophilin A. Primers for quantitative PCR were designed using Oligo 6 software. All primer pairs produced single amplification product as determined by melting curve analyses. The sequences of the primers used were (5' to 3'): NCoA3 forward CTTTGGGCATTCCTGAACTTGTC, NCoA3 reverse GCCTCATCACCGCAGCAC, IRF8 forward GGAGTGCGGTCGCTCTGAAA, IRF8 reverse GTCGTAGGTGGTGTACCCCGTCA, Cyclophilin A forward AGTGGTTGGATGGCAAGC, Cyclophilin A reverse GATTCTAGGATACTGCGAGCAAA. PCR reactions were carried out with a denaturation step of 10 min at 95°C followed by forty-five cycles of 10 s at 95°C, 5 s at annealing temperature (55°C for NCoA3 and Cyclophilin A, 59°C for IRF8) and 20 s amplification at 72°C. Quantifications of cDNAs were determined in reference to a standard curve prepared by amplification of serial dilutions of PCR product containing matching sequences. Analyses were performed using the second-derivative-maximum method provided by the Light Cycler quantification software, version 3.5 (Roche Diagnostics).
Quantification of gag viral mRNA was performed by real-time RT-PCR as described in [54].
Nuclear extracts preparation
For nuclear extract preparation, 10.106 cells were harvested, washed and nuclei were isolated by addition of 150 μL of buffer I (50 mM Tris pH 7.9, 10 mM KCl, 10% glycerol, 1 mM EDTA, 0.2% NP40) followed by a centrifugation at 3000 g for 3 min. Nuclear extracts were prepared by addition of 15 μL of buffer II (20 mM Hepes pH 7.9, 400 mM NaCl, 10 mM KCl, 20% glycerol, 1 mM EDTA) for 20 min at 4°C followed by a centrifugation at 15000 g for 10 min. Protein concentrations were determined by the Bio-Rad protein assay.
Western blot analysis
Nuclear extracts (100 μg) were loaded on 8% SDS-polyacrylamide gel and the proteins were transferred to nitrocellulose membrane (Hybond-C, Amersham) that was subsequently blocked for 1 h with 5% non-fat dry milk in PBS-T (PBS, 0.05% Tween20) and incubated with antibodies directed against NCoA3 (goat polyclonal anti-ACTR C-20, Santa Cruz Biotechnology, Inc.), IRF8 (goat polyclonal anti-ICSBP C-19, Santa Cruz Biotechnology, Inc.) or actin (mouse monoclonal anti-actin, Calbiochem) for 2 h. The membranes were then washed and incubated with secondary antibodies conjugated to horseradish peroxidase (HRP conjugated rabbit anti-goat (DakoCytomation) or goat anti-mouse (Calbiochem) immunoglobulins). Hybridizations were revealed using an ECL enhanced chemiluminescence kit (ECL, Amersham). The quantification was done using the Image Tools (Syngene) software.
Transient transfection and enzymatic assays
HEK293 cells were transfected using calcium phosphate co-precipitation method. Cells were lysed 24 h after transfection with a buffer containing 60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 10 mM MgSO4, 2.5 mM EDTA, 50 mM β-mercaptoethanol and 0.125% Nonidet P-40. Luciferase activities were measured as previously described [55]. Cotransfection with pCMV-LacZ plasmid was performed to normalize transfection efficiency and β-galactosidase activities were determined using Chlorophenol red β-D-galactopyranoside (CPRG, Roche Diagnostics) assay as previously described [55]. The normalized luciferase index (NLI) was defined as the ratio of luciferase to β-galactosidase activities.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SM performed the microarray analyses, real-time RT-PCR and drafted the manuscript. DD and LC carried out real-time RT-PCR, Western blot and transfection experiments. AG participated in transfection experiments. UH conceived the study, participated in its design and coordination and helped to draft and finalize the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Genes upregulated in U1 and ACH-2 cells.
Click here for file
Additional File 2
Genes specifically upregulated in U1 cells.
Click here for file
Additional File 3
Genes specifically upregulated in ACH-2 cells.
Click here for file
Additional File 4
Genes downregulated in U1 and ACH-2 cells.
Click here for file
Additional File 5
Genes specifically downregulated in U1 cells.
Click here for file
Additional File 6
Genes specifically downregulated in ACH-2 cells.
Click here for file
Acknowledgements
We thank Drs. S. Emiliani, B.Z. Levi, P.S. Meltzer, T. Taniguchi, M. Alizon for providing plasmids, and the National Institutes of Health AIDS Research and Reference Reagent Program for the kind gift of reagents. We are grateful to Dr. C. Thibault for her precious help in microarray hybridization data mining. L.C. holds a fellowship from the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche. S.M. is supported by a grant from Sidaction. Sidaction (AO15) supported this work. We thank S. Nisole for helpful discussion.
==== Refs
Pomerantz RJ Horn DL Twenty years of therapy for HIV-1 infection Nat Med 2003 9 867 873 12835707 10.1038/nm0703-867
Pierson T McArthur J Siliciano RF Reservoirs for HIV-1: mechanisms for viral persistence in the presence of antiviral immune responses and antiretroviral therapy Annu Rev Immunol 2000 18 665 708 10837072 10.1146/annurev.immunol.18.1.665
Wong JK Hezareh M Gunthard HF Havlir DV Ignacio CC Spina CA Richman DD Recovery of replication-competent HIV despite prolonged suppression of plasma viremia Science 1997 278 1291 1295 9360926 10.1126/science.278.5341.1291
Finzi D Hermankova M Pierson T Carruth LM Buck C Chaisson RE Quinn TC Chadwick K Margolick J Brookmeyer R Gallant J Markowitz M Ho DD Richman DD Siliciano RF Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy Science 1997 278 1295 1300 9360927 10.1126/science.278.5341.1295
Chun TW Stuyver L Mizell SB Ehler LA Mican JA Baseler M Lloyd AL Nowak MA Fauci AS Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy Proc Natl Acad Sci U S A 1997 94 13193 13197 9371822 10.1073/pnas.94.24.13193
Blankson JN Persaud D Siliciano RF The challenge of viral reservoirs in HIV-1 infection Annu Rev Med 2002 53 557 593 11818490 10.1146/annurev.med.53.082901.104024
Lassen K Han Y Zhou Y Siliciano J Siliciano RF The multifactorial nature of HIV-1 latency Trends Mol Med 2004 10 525 531 15519278 10.1016/j.molmed.2004.09.006
Finzi D Blankson J Siliciano JD Margolick JB Chadwick K Pierson T Smith K Lisziewicz J Lori F Flexner C Quinn TC Chaisson RE Rosenberg E Walker B Gange S Gallant J Siliciano RF Latent infection of CD4(+) T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy Nat Med 1999 5 512 517 10229227 10.1038/8394
Williams SA Greene WC Host factors regulating post-integration latency of HIV Trends Microbiol 2005 13 137 139 15817380 10.1016/j.tim.2005.02.006
He G Ylisastigui L Margolis DM The regulation of HIV-1 gene expression: the emerging role of chromatin DNA Cell Biol 2002 21 697 705 12443539 10.1089/104454902760599672
Verdin E Paras PJ Van Lint C Chromatin disruption in the promoter of human immunodeficiency virus type 1 during transcriptional activation Embo J 1993 12 3249 3259 8344262
Henderson A Holloway A Reeves R Tremethick DJ Recruitment of SWI/SNF to the human immunodeficiency virus type 1 promoter Mol Cell Biol 2004 24 389 397 14673171 10.1128/MCB.24.1.389-397.2004
Lusic M Marcello A Cereseto A Giacca M Regulation of HIV-1 gene expression by histone acetylation and factor recruitment at the LTR promoter Embo J 2003 22 6550 6561 14657027 10.1093/emboj/cdg631
Kato-Maeda M Gao Q Small PM Microarray analysis of pathogens and their interaction with hosts Cell Microbiol 2001 3 713 719 11696031 10.1046/j.1462-5822.2001.00152.x
Shaheduzzaman S Krishnan V Petrovic A Bittner M Meltzer P Trent J Venkatesan S Zeichner S Effects of HIV-1 Nef on cellular gene expression profiles J Biomed Sci 2002 9 82 96 11810028
Simmons A Aluvihare V McMichael A Nef triggers a transcriptional program in T cells imitating single-signal T cell activation and inducing HIV virulence mediators Immunity 2001 14 763 777 11420046 10.1016/S1074-7613(01)00158-3
De la Fuente C Santiago F Deng L Eadie C Zilberman I Kehn K Maddukuri A Baylor S Wu K Lee CG Pumfery A Kashanchi F Gene expression profile of HIV-1 Tat expressing cells: a close interplay between proliferative and differentiation signals BMC Biochem 2002 3 14 12069692 10.1186/1471-2091-3-14
Izmailova E Bertley FM Huang Q Makori N Miller CJ Young RA Aldovini A HIV-1 Tat reprograms immature dendritic cells to express chemoattractants for activated T cells and macrophages Nat Med 2003 9 191 197 12539042 10.1038/nm822
Cicala C Arthos J Selig SM Dennis GJ Hosack DA Van Ryk D Spangler ML Steenbeke TD Khazanie P Gupta N Yang J Daucher M Lempicki RA Fauci AS HIV envelope induces a cascade of cell signals in non-proliferating target cells that favor virus replication Proc Natl Acad Sci U S A 2002 99 9380 9385 12089333 10.1073/pnas.142287999
Janket ML Manickam P Majumder B Thotala D Wagner M Schafer EA Collman RG Srinivasan A Ayyavoo V Differential regulation of host cellular genes by HIV-1 viral protein R (Vpr): cDNA microarray analysis using isogenic virus Biochem Biophys Res Commun 2004 314 1126 1132 14751250 10.1016/j.bbrc.2004.01.008
Corbeil J Sheeter D Genini D Rought S Leoni L Du P Ferguson M Masys DR Welsh JB Fink JL Sasik R Huang D Drenkow J Richman DD Gingeras T Temporal gene regulation during HIV-1 infection of human CD4+ T cells Genome Res 2001 11 1198 1204 11435401 10.1101/gr.GR-1802R
Geiss GK Bumgarner RE An MC Agy MB van 't Wout AB Hammersmark E Carter VS Upchurch D Mullins JI Katze MG Large-scale monitoring of host cell gene expression during HIV-1 infection using cDNA microarrays Virology 2000 266 8 16 10612655 10.1006/viro.1999.0044
Van 't Wout AB Lehrman GK Mikheeva SA O'Keeffe GC Katze MG Bumgarner RE Geiss GK Mullins JI Cellular gene expression upon human immunodeficiency virus type 1 infection of CD4(+)-T-cell lines J Virol 2003 77 1392 1402 12502855 10.1128/JVI.77.2.1392-1402.2003
Wen W Chen S Cao Y Zhu Y Yamamoto Y HIV-1 infection initiates changes in the expression of a wide array of genes in U937 promonocytes and HUT78 T cells Virus Res 2005 15885842
Vahey MT Nau ME Jagodzinski LL Yalley-Ogunro J Taubman M Michael NL Lewis MG Impact of viral infection on the gene expression profiles of proliferating normal human peripheral blood mononuclear cells infected with HIV type 1 RF AIDS Res Hum Retroviruses 2002 18 179 192 11839152 10.1089/08892220252781239
Chun TW Justement JS Lempicki RA Yang J Dennis GJ Hallahan CW Sanford C Pandya P Liu S McLaughlin M Ehler LA Moir S Fauci AS Gene expression and viral prodution in latently infected, resting CD4+ T cells in viremic versus aviremic HIV-infected individuals Proc Natl Acad Sci U S A 2003 100 1908 1913 12552096 10.1073/pnas.0437640100
Krishnan V Zeichner SL Host cell gene expression during human immunodeficiency virus type 1 latency and reactivation and effects of targeting genes that are differentially expressed in viral latency J Virol 2004 78 9458 9473 15308739 10.1128/JVI.78.17.9458-9473.2004
Xu J Li Q Review of the in vivo Functions of the p160 Steroid Receptor Coactivator Family Mol Endocrinol 2003 12 12
Levi BZ Hashmueli S Gleit-Kielmanowicz M Azriel A Meraro D ICSBP/IRF-8 transactivation: a tale of protein-protein interaction J Interferon Cytokine Res 2002 22 153 160 11846986 10.1089/107999002753452764
Bosinger SE Hosiawa KA Cameron MJ Persad D Ran L Xu L Boulassel MR Parenteau M Fournier J Rud EW Kelvin DJ Gene expression profiling of host response in models of acute HIV infection J Immunol 2004 173 6858 6863 15557180
Kino T Slobodskaya O Pavlakis GN Chrousos GP Nuclear receptor coactivator p160 proteins enhance the HIV-1 long terminal repeat promoter by bridging promoter-bound factors and the Tat-P-TEFb complex J Biol Chem 2002 277 2396 2405 11704662 10.1074/jbc.M106312200
Sgarbanti M Borsetti A Moscufo N Bellocchi MC Ridolfi B Nappi F Marsili G Marziali G Coccia EM Ensoli B Battistini A Modulation of human immunodeficiency virus 1 replication by interferon regulatory factors J Exp Med 2002 195 1359 1370 12021315 10.1084/jem.20010753
Van Lint C Amella CA Emiliani S John M Jie T Verdin E Transcription factor binding sites downstream of the human immunodeficiency virus type 1 transcription start site are important for virus infectivity J Virol 1997 71 6113 6127 9223506
Thornton AM Ogryzko VV Dent A Sharf R Levi BZ Kanno Y Staudt LM Howard BH Ozato K A dominant negative mutant of an IFN regulatory factor family protein inhibits both type I and type II IFN-stimulated gene expression and antiproliferative activity of IFNs J Immunol 1996 157 5145 5154 8943426
Herrmann CH Carroll RG Wei P Jones KA Rice AP Tat-associated kinase, TAK, activity is regulated by distinct mechanisms in peripheral blood lymphocytes and promonocytic cell lines J Virol 1998 72 9881 9888 9811724
Van 't Wout AB Swain JV Schindler M Rao U Pathmajeyan MS Mullins JI Kirchhoff F Nef induces multiple genes involved in cholesterol synthesis and uptake in human immunodeficiency virus type 1-infected T cells J Virol 2005 79 10053 10058 16014965 10.1128/JVI.79.15.10053-10058.2005
Krishnan V Zeichner SL Alterations in the expression of DEAD-box and other RNA binding proteins during HIV-1 replication Retrovirology 2004 1 42 15588285 10.1186/1742-4690-1-42
Marks P Rifkind RA Richon VM Breslow R Miller T Kelly WK Histone deacetylases and cancer: causes and therapies Nat Rev Cancer 2001 1 194 202 11902574 10.1038/35106079
Brigati C Giacca M Noonan DM Albini A HIV Tat, its TARgets and the control of viral gene expression FEMS Microbiol Lett 2003 220 57 65 12644228 10.1016/S0378-1097(03)00067-3
Chen H Lin RJ Schiltz RL Chakravarti D Nash A Nagy L Privalsky ML Nakatani Y Evans RM Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300 Cell 1997 90 569 580 9267036 10.1016/S0092-8674(00)80516-4
El Kharroubi A Martin MA cis-acting sequences located downstream of the human immunodeficiency virus type 1 promoter affect its chromatin structure and transcriptional activity Mol Cell Biol 1996 16 2958 2966 8649407
Battistini A Marsili G Sgarbanti M Ensoli B Hiscott J IRF regulation of HIV-1 long terminal repeat activity J Interferon Cytokine Res 2002 22 27 37 11846973 10.1089/107999002753452638
Marsili G Borsetti A Sgarbanti M Remoli AL Ridolfi B Stellacci E Ensoli B Battistini A On the role of interferon regulatory factors in HIV-1 replication Ann N Y Acad Sci 2003 1010 29 42 15033691 10.1196/annals.1299.005
Sgarbanti M Marsili G Remoli AL Ridolfi B Stellacci E Borsetti A Ensoli B Battistini A Analysis of the signal transduction pathway leading to human immunodeficiency virus-1-induced interferon regulatory factor-1 upregulation Ann N Y Acad Sci 2004 1030 187 195 15659797 10.1196/annals.1329.024
Emiliani S Van Lint C Fischle W Paras PJ Ott M Brady J Verdin E A point mutation in the HIV-1 Tat responsive element is associated with postintegration latency Proc Natl Acad Sci U S A 1996 93 6377 6381 8692823 10.1073/pnas.93.13.6377
Emiliani S Fischle W Ott M Van Lint C Amella CA Verdin E Mutations in the tat gene are responsible for human immunodeficiency virus type 1 postintegration latency in the U1 cell line J Virol 1998 72 1666 1670 9445075
Clouse KA Powell D Washington I Poli G Strebel K Farrar W Barstad P Kovacs J Fauci AS Folks TM Monokine regulation of human immunodeficiency virus-1 expression in a chronically infected human T cell clone J Immunol 1989 142 431 438 2463307
Perez VL Rowe T Justement JS Butera ST June CH Folks TM An HIV-1-infected T cell clone defective in IL-2 production and Ca2+ mobilization after CD3 stimulation J Immunol 1991 147 3145 3148 1833465
Folks TM Justement J Kinter A Schnittman S Orenstein J Poli G Fauci AS Characterization of a promonocyte clone chronically infected with HIV and inducible by 13-phorbol-12-myristate acetate J Immunol 1988 140 1117 1122 2449497
Butera ST Perez VL Wu BY Nabel GJ Folks TM Oscillation of the human immunodeficiency virus surface receptor is regulated by the state of viral activation in a CD4+ cell model of chronic infection J Virol 1991 65 4645 4653 1678437
Hazan U Thomas D Alcami J Bachelerie F Israël N Yssel H Virelizier JL Arenzana-Seisdedos F Stimulation of human T-cell clone with anti-CD3 or tumor necrosis factor induces NF-kappa B translocation but not human immunodeficiency virus 1 enhancer-dependent transcription. Proc Natl Acad Sci USA 1990 87 7861 7865 2146676
Anzick SL Kononen J Walker RL Azorsa DO Tanner MM Guan XY Sauter G Kallioniemi OP Trent JM Meltzer PS AIB1, a steroid receptor coactivator amplified in breast and ovarian cancer Science 1997 277 965 968 9252329 10.1126/science.277.5328.965
El Kharroubi A Verdin E Protein-DNA interactions within DNase I-hypersensitive sites located downstream of the HIV-1 promoter J Biol Chem 1994 269 19916 19924 8051074
Brussel A Sonigo P Evidence for gene expression by unintegrated human immunodeficiency virus type 1 DNA species J Virol 2004 78 11263 11271 15452245 10.1128/JVI.78.20.11263-11271.2004
Dumonceaux J Nisole S Chanel C Quivet L Amara A Baleux F Briand P Hazan U Spontaneous mutations in the env gene of the human immunodeficiency virus type 1 NDK isolate are associated with a CD4-independent entry phenotype J Virol 1998 72 512 519 9420253
|
16305739
|
PMC1310520
|
CC BY
|
2021-01-04 16:36:40
|
no
|
Retrovirology. 2005 Nov 23; 2:73
|
utf-8
|
Retrovirology
| 2,005 |
10.1186/1742-4690-2-73
|
oa_comm
|
==== Front
RetrovirologyRetrovirology1742-4690BioMed Central London 1742-4690-2-731630573910.1186/1742-4690-2-73ResearchCharacterization of two candidate genes, NCoA3 and IRF8, potentially involved in the control of HIV-1 latency Munier Sandie [email protected]ête Delphine [email protected]éna Laëtitia [email protected] Audrey [email protected] Uriel [email protected] Département des Maladies Infectieuses, Institut Cochin, INSERM U567/CNRS UMR-S 8104/Université Paris 5-René Descartes, 22 rue Méchain, 75014 Paris, France2 UFR de Biochimie, Université Paris 7-Denis Diderot, 2 Place Jussieu, 75251 Paris, France2005 23 11 2005 2 73 73 28 7 2005 23 11 2005 Copyright © 2005 Munier et al; licensee BioMed Central Ltd.2005Munier et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The persistence of latent HIV-1 reservoirs is the principal barrier preventing the eradication of HIV-1 infection in patients by current antiretroviral therapy. It is thus crucial to understand the molecular mechanisms involved in the establishment, maintenance and reactivation of HIV-1 latency. Since chromatin remodeling has been implicated in the transcriptional reactivation of the HIV-1 promoter, we assessed the role of the histone deacetylase inhibitor sodium butyrate (NaB) on two HIV-1 latently infected cell lines (U1 and ACH-2) gene expression.
Results
Analysis of microarrays data led us to select two candidate genes: NCoA3 (Nuclear Receptor Coactivator 3), a nuclear receptor coactivator and IRF8 (Interferon Regulatory Factor 8), an interferon regulatory factor. NCoA3 gene expression is upregulated following NaB treatment of latently infected cells whereas IRF8 gene expression is strongly downregulated in the promonocytic cell line following NaB treatment. Their differential expressions were confirmed at the transcriptional and translational levels. Moreover, NCoA3 gene expression was also upregulated after treatment of U1 and ACH-2 cells with phorbol myristyl acetate (PMA) but not trichostatin A (TSA) and after treatment with NaB of two others HIV-1 latently infected cell lines (OM10.1 and J1.1). IRF8 gene is only expressed in U1 cells and was also downregulated after treatment with PMA or TSA. Functional analyses confirmed that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription and that IRF8 represses the IRF1-mediated activation through the HIV-1 promoter Interferon-stimulated response element (ISRE).
Conclusion
These results led us to postulate that NCoA3 could be involved in the transcriptional reactivation of the HIV-1 promoter from latency and that IRF8 may contribute to the maintenance of the latent state in the promonocytic cell line. Implication of these factors in the maintenance or reactivation of the viral latency may provide potential new targets to control HIV-1 replication in latent viral reservoirs.
==== Body
Background
The use of highly active antiretroviral therapy (HAART) in HIV-1 infected individuals has led to a significant decrease of plasma viremia to undetectable levels and has considerably improved the survival and quality of life of infected individuals (reviewed in [1]). However, the presence of cellular reservoirs that contain latent viruses capable of producing infectious particles after cellular activation lead to a rebound of the viral load after interruption of HAART (reviewed in [2]). The persistence of these latently infected viral reservoirs, despite prolonged HAART treatments, represents a major obstacle to the eradication of HIV-1 in infected patients [3-5]. Therefore, a greater understanding of the molecular mechanisms involved in establishment, maintenance and reactivation of viral latency is essential to expect the reduction of latent HIV-1 reservoirs in infected patients.
Latent HIV-1 infection can exist in many reservoirs, such as macrophages and resting memory CD4+ T cells (reviewed in [6]). At the cellular level, two major forms of HIV-1 latency have been described: pre- and post-integration latency [7]. CD4+ T cells in the post-integration state of latency represent the most stable reservoir for HIV-1 (half-life of 43 months) [8]. Several mechanisms have been proposed to account for the low level of transcription observed during post-integration latency (reviewed in [9]): the inaccessibility of the integrated provirus to the transcriptional machinery, the absence in resting cells of transcription factors involved in HIV-1 gene expression, the presence of transcriptional repressors, and the premature termination of HIV-1 transcription elongation due to the absence of the viral protein Tat and its associated cofactors. Moreover, the chromatin structure appears to be involved in the regulation of HIV-1 gene expression (reviewed in [10]). Indeed, a repressive nucleosome (nuc-1), located immediately downstream of the HIV-1 transcription start site under latency conditions, is disrupted upon transcriptional activation of the HIV-1 promoter in response to Tat, phorbol esters and histone deacetylase (HDAC) inhibitors [11]. Transcriptional activation of the HIV-1 promoter in response to PMA involves the recruitment of SWI/SNF chromatin remodeling complex [12] and cellular proteins with histone acetyltransferase (HAT) activity [13]. Therefore, chromatin remodeling plays a significant role in the transcriptional reactivation of the HIV-1 promoter from latency. Identification of host transcription factors that may regulate chromatin structure is thus critical to understand the molecular mechanisms involved in HIV-1 reactivation.
Gene expression analysis using high-density microarrays have provided a greater understanding of host-pathogen interactions (reviewed in [14]). Previous microarray studies on HIV-1 have described changes in cellular genes transcription in response to HIV-1 protein expression (Nef [15,16], Tat [17,18], gp120 [19] or Vpr [20]) or following acute infection of cell lines [21-24] or Peripheral Blood Mononuclear Cells (PBMC) [25]. DNA microarrays have also been used to characterize gene expression in latently infected resting CD4+ T cells in viremic versus aviremic HIV-1 infected individuals [26]. Recently, global gene expression changes in cell lines latently infected with HIV-1 and induced by PMA for completion of viral replication was described by Krishnan et al. [27].
To complete the results obtained by Krishnan et al., we used the same strategy to assess the role of the HDAC inhibitor NaB on HIV-1 latently infected cells gene expression. We performed microarray experiments on two HIV-1 latently infected cell lines (U1 and ACH-2) treated or not with NaB to induce viral reactivation. Analysis of microarrays data led us to select two candidate genes encoding transcription factors: NCoA3 (reviewed in [28]), which expression is upregulated following treatment of latently infected cells with NaB, and IRF8 (reviewed in [29]), which expression is downregulated in treated cells. Differential expression of these genes was confirmed at the transcriptional and translational levels. Moreover, NCoA3 gene expression was also upregulated after treatment of U1 and ACH-2 cells with PMA but not TSA and after treatment with NaB of two others latently infected cell lines (OM10.1 and J1.1). IRF8 gene is only expressed in U1 cells and was also downregulated after treatment with PMA or TSA. Functional analyses confirmed that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription and that IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element. Implication of IRF8 in the maintenance and NCoA3 in the reactivation of the viral latency may thus provide new insights into the control of HIV-1 replication in latent viral reservoirs.
Results
Microarray analysis
In order to understand the molecular mechanisms regulating HIV-1 latency, we studied the modifications of cellular transcription using microarrays in the promonocytic U1 and T CD4+ lymphocytic ACH-2 chronically HIV-1 infected cell lines after reactivation of latency. The two cell lines were treated with 10 mM of the histone deacetylase inhibitor NaB. Viral reactivation was monitored by coculture with P4 indicating cells (Figure 1A) and measuring gag viral mRNA expression (Figure 1B). Increase in both β-galactosidase activity and gag mRNA expression showed that the viral reactivation after NaB treatment was efficient. Total RNAs were extracted after 24 h and sent to the Affymetrix Microarray Facilities for subsequent hybridization on U-133A microarrays.
Figure 1 Analysis of viral reactivation after treatment of U1 and ACH-2 cells with NaB. U1 and ACH-2 cells were treated or not (NT) with 10 mM of NaB for 24 h and cocultured with P4 indicating cells. β-galactosidase activity was determined after 24 h coculture (A). RNA from U1 and ACH-2 cells treated or not with NaB were extracted after 24 h and gag viral mRNA expression was measured by real-time RT-PCR (B). Results are representative of three independent experiments.
The pattern of cellular mRNA from chronically infected cells treated with NaB was compared to that from non-treated cells. We used as specific criteria a log2 ratio change ≥ 1 with a change p-value ≤ 0.0001 for increased genes and a log2 ratio change ≤ -1 with a 1-change p-value ≥ 0.9999 for decreased genes. Hybridization experiments were performed once. We identified 740 genes that were upregulated by twofold or higher in NaB treated U1 cells and 896 genes that were downregulated, 482 genes in NaB treated ACH-2 cells that had a level increased greater than twofold and 634 genes that had a level decreased greater than twofold (data not shown). Moreover, 260 genes were commonly increased and 337 genes were decreased in both U1 and ACH-2 NaB-treated cells (data not shown). Pathways involved in regulation of transcription, signal transduction, immune response, protein transport, metabolism, apoptosis and RNAs modifications showed altered expression following treatment with NaB. Some of the genes involved in these pathways are shown in Additional Files 1, 2, 3, 4, 5 and 6. Our analysis identified genes that have previously been associated with HIV-1 replication or latency, such as CDK9 [16], Jun [16,23], PSMB10 [27], MAPK1 [26] or OAS1 [30]. This supported the accuracy of our approach, even though, as the hybridization experiments had been performed once, the statistical relevance of the results could not be estimated.
Among the differentially expressed genes, we chose to focus on two candidate genes encoding transcription factors: NCoA3 and IRF8 (Tables 1 and 2). We selected these two genes based on their biological properties, their described effects on viral replication [31,32] and their differential expression observed by microarray experiments. Indeed, NCoA3 and IRF8 gene expression are respectively upregulated and downregulated following treatment with NaB of latently infected cells (Tables 1 and 2). Therefore, NCoA3 and IRF8 could be implicated respectively in the reactivation and maintenance of HIV-1 latency.
Table 1 Differential gene expression obtained for NCoA3 and IRF8 mRNAs in U1 cells treated or not with NaB.
Gene Probe set Name a U1 Signal b U1 Detection p-value c U1NaB Signal U1NaB Detection p-value U1NaBvsU1 Signal log2 ratio d U1NaBvsU1 Change p-value e
NCoA3 207700_s_at 17.7 0.01416 98.9 0.000244 2.5 0.000035
209060_x_at 16.9 0.171387 77.2 0.000244 2.3 0.000023
209061_at 48.4 0.037598 166.4 0.000732 2.3 0.00002
209062_x_at 6.3 0.72583 91.8 0.010742 4.5 0.000147
211352_s_at 7.2 0.303711 68.6 0.00293 3.2 0.000101
IRF8 204057_at 707.9 0.000244 47 0.010742 -4 0.99998
a Affymetrix U133-A reference probe set.
b Signal intensity of hybridization.
c Signal detection p-value < 0.048 for specific hybridization.
d Signal log2 ratio > 1 for increased genes and < -1 for decreased genes.
e Change p-value < 0.0001 for significant increased genes and 1-change p-value > 0.9999 for significant decreased genes.
Table 2 Differential gene expression obtained for NCoA3 mRNA in ACH-2 cells treated or not with NaB.
Gene Probe set Name a ACH-2 Signal b ACH-2 Detection p-value c ACH2NaB Signal ACH2NaB Detection p-value ACH2NaBvsACH2 Signal log2 ratio d ACH2NaBvsACH2 Change p-value e
NCoA3 207700_s_at 43.3 0.001953 99.6 0.001221 1.2 0.000241
209060_x_at 34.5 0.01416 72.9 0.001953 1 0.000273
209061_at 65.8 0.000732 82.6 0.000732 1.6 0.005409
209062_x_at 20 0.466064 76.7 0.095215 2 0.000114
211352_s_at 2.7 0.5 37 0.030273 3.8 0.004481
a Affymetrix U133-A reference probe set.
b Signal intensity of hybridization.
c Signal detection p-value < 0.048 for specific hybridization.
d Signal log2 ratio > 1 for increased genes and < -1 for decreased genes.
e Change p-value < 0.0001 for significant increased genes and 1-change p-value > 0.9999 for significant decreased genes.
NCoA3 gene expression is upregulated following treatment with NaB of both U1 and ACH-2 latently infected cells (Tables 1 and 2). NCoA3 is a nuclear receptor coactivator of the Steroid Receptor Coactivator (SRC) family that interacts with nuclear receptors in a ligand-dependent manner and enhances transcriptional activation via histone acetylation and recruitment of general transcription factors and additional cofactors (reviewed in [28]). NCoA3 (Unigene Hs. 382168) gene expression in U1 cells is significantly upregulated by 4.9 to 22.6 fold (U1NaBvsU1 signal log2 ratio ranging from 2.3 to 4.5 with a change p-value < 0.00015) following treatment with NaB (Table 1). Similarly, NCoA3 gene expression is upregulated in NaB-treated compared to non-treated ACH-2 cells by 2 to 13.9 fold but with a lower significance (ACH2NaBvsACH2 signal log2 ratio ranging from 1 to 3.8 with a change p-value < 0.0055) (Table 2).
IRF8 gene expression is downregulated following treatment of U1 cells with NaB (Table 1). IRF8 is a transcription factor of the Interferon (IFN) Regulatory Factor (IRF) family that binds to IFN-stimulated response element and regulates expression of genes stimulated by IFNs (reviewed in [29]). IRF8 (Unigene Hs. 137427) is expressed in the promonocytic cell line U1 (detection signal of 707.9 with a p-value of 0.000244) (Table 1) but is not expressed in the T CD4+ lymphocytic cell line ACH-2 (data not shown). Following NaB treatment, IRF8 gene expression in U1 cells is downregulated by 16 fold (U1NaBvsU1 signal log2 ratio of -4 with a 1-change p-value of 0.99998) (Table 1).
Validation of NCoA3 and IRF8 differential transcriptional expression
Real-time RT-PCR quantifications were performed to confirm that NCoA3 and IRF8 genes were differentially expressed in the NaB-treated chronically infected cells compared to the non-treated cells. We performed quantification on RNA samples obtained from five independent NaB treatments of U1 and ACH-2 cells and real-time RT-PCR experiments were run in duplicate. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The results show in Figure 2 represent the NCoA3 expression increase fold (Figure 2A) obtained from U1 and ACH-2 cells and the IRF8 expression decrease fold (Figure 2B) obtained from U1 cells treated with NaB for 24 h and 48 h compared to non-treated cells. Concerning NCoA3, real-time RT-PCR showed an upregulation consistent with microarray data in 24 h NaB-treated U1 cells of 8.34 ± 2.42 fold compared to non-treated cells (Figure 2A). NCoA3 gene expression is also increased with a 48 h NaB treatment (upregulation of 8.40 ± 2.33 fold) (Figure 2A). Similarly, an increase of NCoA3 gene expression can be observed on ACH-2 cells following treatment with NaB (upregulation of 4.56 ± 1.28 fold in 24 h and 6.80 ± 2.34 fold in 48 h NaB-treated ACH-2 cells) (Figure 2A). Concerning IRF8, real-time RT-PCR showed a 14.96 ± 4.85 fold decrease in 24 h NaB-treated U1 cells (Figure 2B) in correlation with the microarray ratio previously obtained. Downregulation of IRF8 gene expression is also observed following 48 h NaB-treatment of U1 cells (22.06 ± 11.29 fold decrease) (Figure 2B). Taken together, results from real-time RT-PCR performed on NCoA3 and IRF8 genes corroborate with those obtained using microarray hybridizations.
Figure 2 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in NaB-treated U1 and ACH-2 cells. Total RNAs were isolated from U1 or ACH-2 cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U1 (solid bars) or ACH-2 (white bars) cells and the IRF8 decrease fold (B) in U1 cells treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of five independent experiments performed in duplicate.
We next determined whether NCoA3 and IRF8 gene expression were similarly modified in the uninfected parental cell lines. U937 and CEM cells were subjected to identical treatment and RT-PCR quantifications were performed (Figure 3). NCoA3 is upregulated both in U937 and CEM cells following treatment with NaB (upregulation of 7.32 ± 1.74 fold in 24 h and 11.45 ± 2.95 fold in 48 h NaB-treated U937 cells, upregulation of 1.93 ± 1.04 fold in 24 h and 5.59 ± 0.06 fold in 48 h NaB-treated CEM cells) (Figure 3A). IRF8 is only expressed in the promonocytic cell line U937 and, as in U1 cells, its expression was downregulated after NaB treatment (downregulation of 17.95 ± 4.15 fold in 24 h and 22.32 ± 10.82 fold in 48 h NaB-treated U937 cells) (Figure 3B). Thus, NaB treatment modify NCoA3 and IRF8 gene expression in uninfected parental cell lines U937 and CEM at a similar level than in chronically infected cells.
Figure 3 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in NaB-treated U937 and CEM cells. Total RNAs were isolated from U937 or CEM cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U937 (solid bars) or CEM (white bars) cells and the IRF8 decrease fold (B) in U937 cells treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of five independent experiments performed in duplicate.
We then performed additional experiments to determine whether the gene expression variations observed could also be mediated by treatments with the phorbol ester PMA and another HDAC inhibitor, TSA. We thus assessed the differential regulation of NCoA3 and IRF8 gene expression in U1 and ACH-2 cells treated with PMA or TSA (Figure 4). Results indicated that NCoA3 expression is upregulated by 24 h and 48 h PMA treatment of U1 and ACH-2 cells (upregulation of 5.70 ± 1.37 fold in 24 h and 9.85 ± 0.90 fold in 48 h PMA-treated U1 cells, upregulation of 3.12 ± 1.05 fold in 24 h and 7.12 ± 1.20 fold in 48 h PMA-treated ACH-2 cells (Figure 4A). However, TSA treatment had no significant effect on NCoA3 expression in U1 and ACH-2 cells, although TSA increased viral expression (data not shown). Concerning IRF8 expression in U1 cells, PMA and TSA treatments for 24 h induced a decrease of 3.22 ± 0.45 fold and 5.32 ± 1.09 fold, respectively (Figure 4B). These results show that NCoA3 expression is upregulated following phorbol ester but not with other HDAC inhibitor treatments in U1 and ACH-2 cells. Moreover, IRF8 gene expression in U1 cells is downregulated with PMA or TSA treatments but at a lower extent than with NaB.
Figure 4 Real-time RT-PCR analysis of NCoA3 and IRF8 mRNAs expression in PMA- or TSA-treated U1 and ACH-2 cells. Total RNAs were isolated from U1 or ACH-2 cells treated or not with PMA for 24 h and 48 h or TSA for 24 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3, IRF8 or Cyclophilin A. NCoA3 and IRF8 expressions were normalized to the expression of Cyclophilin A. The NCoA3 increase fold (A) in U1 (solid bars) or ACH-2 (white bars) cells treated with PMA for 24 h and 48 h and the IRF8 decrease fold (B) in U1 cells treated with PMA or TSA for 24 h compared to non-treated (NT) cells were determined. Results represent the means of three independent experiments performed in duplicate.
We also assessed the differential regulation of NCoA3 and IRF8 gene expression in others chronically HIV-1 infected cell lines. The chronically infected promonocytic OM10.1 and T CD4+ lymphocytic J1.1 cell lines were treated with NaB for 24 h and 48 h and real-time RT-PCR were performed to measure NCoA3 and IRF8 gene expression. As shown in Figure 5, NCoA3 expression is upregulated by 4.94 ± 0.78 fold in OM10.1 and by 2.56 ± 0.64 fold in J1.1 after 24 h NaB treatment. NCoA3 expression increased with time of NaB treatment in both cell lines (upregulation of 12.89 ± 3.10 fold in OM10.1 and 3.51 ± 0.69 fold in J1.1 cells) (Figure 5). Like ACH-2 and unlike U1 cells, the T CD4+ lymphocytic J1.1 and the promonocytic OM10.1 cell lines did not express IRF8 (data not shown). Thus, the differential regulation of NCoA3 but not IRF8 gene expression is similar in two other related latently HIV-1 infected cell line models.
Figure 5 Real-time RT-PCR analysis of NCoA3 mRNAs expression in OM10.1 and J1.1 cells. Total RNAs were isolated from OM10.1 or J1.1 cells treated or not with NaB for 24 h and 48 h and real-time PCR were performed on cDNAs using gene specific primers for NCoA3 or Cyclophilin A. NCoA3 expression was normalized to the expression of Cyclophilin A. The NCoA3 increase fold in OM10.1 (solid bars) or J1.1 cells (white bars) treated with NaB for 24 h and 48 h compared to non-treated (NT) cells were determined. Results represent the means of two independent experiments performed in duplicate.
gag mRNA activation is correlated with NCoA3 mRNA increase and IRF8 mRNA decrease
We performed reactivation experiments at different times, sooner than 24 h and until 48 h. Quantitative RT-PCR experiments were carried out on total RNAs. This was done using U1 cells to analyze both NCoA3 mRNA increase (Figure 6A) and IRF8 mRNA decrease (Figure 6B) relative to HIV gag mRNA along with ACH-2 cells (Figure 6C) to analyze NCoA3 mRNA increase relative to HIV gag mRNA.
Figure 6 Analysis of HIV gag, NCoA3, and IRF8 mRNA expression after NaB stimulation on U1 and ACH-2 cells. U1 (A and B) and ACH-2 (C) cells were stimulated with 10 mM NaB and 5.106 cells were taken at t = 0, 4, 8, 16, 24, 48 h for RNA extraction to perform qRT-PCR. NCoA3 (A and C), IRF8 (B) and gag (A, B and C) mRNA contents were measured. Cylophilin A was used as internal standard. Results represent a representative experiment performed in duplicate.
As observed on Figure 6C, the obtained results, both on ACH-2 and U1 cells, clearly show that gag mRNA activation occurs after NCoA3 mRNA increase and accumulation. Moreover, in U1 cells, gag mRNA activation occurs after IRF8 mRNA decrease. Shorter kinetics (0 to 8 h) correlated with these results (data not shown).
Validation of NCoA3 and IRF8 differential translational expression
To confirm that the changes seen at the RNA level correlated with protein levels, we performed Western blot experiments on nuclear extract of U1, ACH-2, OM10.1 and J1.1 cells treated or not with NaB for 24 h (Figure 7). Results indicated that NaB increased the expression level of NCoA3 protein in U1, ACH-2, OM10.1 and not in J1.1 cells (Figure 7). Moreover, IRF8 protein expression is strongly downregulated in U1 cells following NaB treatment (Figure 7). These results correlate with the differential expression of NCoA3 and IRF8 genes observed with both microarray and real-time RT-PCR experiments.
Figure 7 Western blot analysis of NCoA3 and IRF8 proteins expression. Nuclear extract (100 μg) from U1, ACH-2, J1.1 and OM10.1 treated (+) or not (-) with NaB for 24 h were resolved by SDS-PAGE and immunoblotted with anti-NCoA3 or anti-IRF8 antibody, as indicated. The amount of protein was normalized using anti-actin antibody. Figures below NCoA3 immunoblot indicated the results of the quantification using Image Tool (Syngene) software of the ratio NCoA3/actin upon NaB-treatment (+) versus NCoA3/actin non-treated (-). Results are representative of three independent experiments.
Transcriptional activation of the HIV-1 promoter by NCoA3
We analyzed the functional role of NCoA3 on viral transcription by transfection assays. HEK293 cells were cotransfected with pLTRX-luc reporter plasmid containing the luciferase gene under the control of the HIV-1 U3-R promoter region (nt -640 to +78) with or without Tat and/or NCoA3 expression vectors. As shown in Figure 8, NCoA3 increased Tat-stimulated HIV-1 LTR activity by 2.8 ± 1.4 fold. The presence of NCoA3 had synergistic effect on the HIV-1 LTR activity induced by suboptimal expression of Tat. When HEK293 cells were transfected with pLTRΔTAR-luc reporter plasmid lacking the Tat-transactivation response element TAR, Tat was not able to activate the LTR transcription, as expected, and NCoA3 had no effect on the LTR activity (Figure 8). Thus, functional analyses confirm that NCoA3 synergizes with Tat to enhance HIV-1 promoter transcription, as expected [31], and that this effect is dependent on the presence of the TAR region.
Figure 8 NCoA3 increases the Tat-stimulated HIV-1 LTR activity. HEK293 cells were cotransfected with pLTRX-luc (10 ng, grey bars) or pLTRΔTAR-luc (10 ng, white bars) with (+) or without (-) suboptimal amounts of pCMV-Tat (5 ng) and/or pNCoA3 (1 μg) expression vectors. NLI (normalized luciferase index) were measured after 24 h and the activation folds compared to the basal activity of the corresponding pLTR-luc were determined. Results represent the means of five independent experiments.
Transcriptional repression of the HIV-1 ISRE element by IRF8
We analyzed the functional role of IRF8 on viral transcription by transfection assays. HEK293 cells were cotransfected with pISRE-TK-luc reporter plasmid corresponding to the HIV-1 IFN-stimulated response element, located downstream transcription start site (nt +194 to +223) [33], with or without IRF1 and/or IRF8 expression vectors. As shown in Figure 9, the basal activity of the ISRE-TK was increased by 7.4 ± 1.0 fold in the presence of IRF1 as expected [32], whereas a decrease was detected in the presence of IRF8 (21.9 ± 10.6 to 41.4 ± 9.5 %). The expression of IRF8 inhibited by 43.5 ± 10.6 to 74.7 ± 2.5 % the IRF1-mediated activation of the ISRE-TK in a dose dependent fashion (Figure 9). The expression of the dominant negative IRF8 DNA-binding domain (IRF8-DBD) inhibited by 76.4 ± 6.5 % the IRF1-mediated activation of the ISRE-TK, as expected [34] (Figure 9). The inhibitory effects of IRF8 and IRF8-DBD expression and activation effect of IRF1 expression was abolished when the ISRE sequence was mutated (pISREmut-TK-luc, Figure 9). These results show that IRF8 represses the ISRE-TK promoter transcription through the ISRE element from the HIV-1 promoter, as expected [32].
Figure 9 IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element. HEK293 cells were cotransfected with pISRE-TK-luc (250 ng, solid bars) or pISREmut-TK-luc (250 ng, white bars) with (+) or without (-) pIRF1 (250 ng), pIRF8 (1–2.5 μg), or pIRF8-DBD (1 μg) expression vectors. NLI (normalized luciferase index) were measured after 24 h and the activation folds compared to the basal activity of the pISRE-TK-luc or pISREmut-TK-luc were determined. Results represent the means of five independent experiments.
Discussion
The existence of long-lasting HIV-1 reservoirs is the principal barrier preventing the eradication of HIV-1 infection in patients by current antiretroviral therapy. It is thus crucial to understand the molecular mechanisms involved in establishment, maintenance and reactivation of HIV-1 latency. In this study, the role of the HDAC inhibitor NaB on HIV-1 latently infected cells gene expression was explored using microarrays. Since chromatin remodeling is involved in the regulation of HIV-1 gene expression (reviewed in [10]), differential expression of cellular genes in latently infected cells following treatment with NaB might be related to the maintenance and reactivation of latency.
Recently, Krishnan et al. [27] described the global gene expression changes in HIV-1 latently infected cell lines treated or not with PMA to induce viral reactivation compared to the uninfected parental cell lines treated under the same conditions. Here, we compared gene expression profiles of two HIV-1 latently infected cell lines (U1 and ACH-2) treated with NaB to that of non-treated corresponding cell lines. We thus avoided identification of genes which differential expression could result from the establishment and cloning of the chronically infected cell lines. Based on our specific criteria, we identified few hundreds of genes affected by NaB treatment implicated in biological pathways previously shown to be modulated by HIV-1 replication. For example, reactivation of latency induced an upregulation of CDK9, the catalytic component of transcription elongation factor b (P-TEFb), which acts in concert with Tat to direct the processivity of HIV-1 transcription. It was shown that CDK9 mRNA and protein levels are induced following T cell activation and Nef expression, and that this correlates with kinase activity, thus enhancing HIV-1 transcription [16,35].
After NaB treatment of latently infected cell lines, we observed an upregulation of genes involved in vesicular transport of protein like syntaxin and nexin. It was found by Chun et al. that numerous genes involved in protein/vesicle transport are upregulated in resting T CD4+ cells of viremic patients, strongly suggesting that enhanced activities in secretory pathways may help in the assembly and release of viral particles [26]. Recently, it was shown that multiple genes involved in cholesterol synthesis are induced by Nef [36]. NaB treatment also induced some of these genes (INSIG1, HMGCS1, IDI1, LSS or SREBF1) and could thus enhanced virion infectivity and viral replication.
Krishnan et al. have described an increase in expression of several proteasome subunits in ACH-2 cells prior induction of lytic replication by PMA and proposed that the higher expression of proteasomes may lead to increased degradation of HIV-1 mRNA [27]. After induction of lytic replication by NaB, proteasome subunits PSMB10 and PSMB8 were downregulated in ACH-2 and U1 cells, suggesting a role in the maintenance of the latent state. Indeed, reactivation of latency was achieved with proteasome inhibitors [27]. Among the downregulated genes after NaB treatment, we identified genes involved in RNA modifications. Krishnan et al. have shown alterations in the expression of DEAD-box and other RNA binding proteins during HIV-1 replication [37]. Especially, DDX18 and DDX39 are upregulated in latently infected cells [37]. After NaB treatment of latently infected cells, we observed a decrease in the expression of these two proteins, thus providing more support for their role in maintaining HIV-1 latency.
The only purpose of our microarray analysis was to identify candidate genes potentially involved in the control of the HIV latency. For this reason, we decided to focus on two candidate genes previously described to influence viral expression and that may be involved in reactivation and maintenance of latency: NCoA3 and IRF8, respectively. Hybridization experiments were performed once. Consequently, we did not further analyze the statistical relevance of the results and performed complementary approaches to confirm the mRNA variations of the selected candidate genes.
NCoA3 is a nuclear receptor coactivator that enhances ligand-induced transcriptional activation of nuclear receptors (reviewed in [28]). We show that NCoA3 (Unigene Hs. 382168) gene expression is upregulated following treatment with NaB of U1 and ACH-2 latently infected cells. This differential transcriptional expression was confirmed by real-time RT-PCR and is also mediated by PMA but not TSA. Upregulation of NCoA3 is thus achieved following phorbol ester but not other HDAC inhibitor treatment. However, NaB and TSA act on different pathways and at different concentrations and target different genes [38]. Transcriptional increase of NCoA3 was observed in parental uninfected corresponding cell lines U937 and CEM and in two others latently HIV-1 infected cell lines, OM10.1 and J1.1. NCoA3 protein level is also upregulated following treatment with NaB in the U1, ACH-2 and OM10.1 cell lines. Moreover, NCoA3 increases the Tat-induced HIV-1 LTR promoter transcriptional activity through the TAR region, in accordance with other data [31]. The differential expression of NCoA3 observed led us to postulate that NCoA3 could be involved in the transcriptional reactivation of the HIV-1 promoter from latency, at low concentrations of Tat.
This hypothesis is supported by several findings. Previous microarray studies on latently infected resting CD4+ T cells in infected individuals have shown an upregulation of NCoA3 gene expression in viremic versus aviremic patients [26]. Moreover, Kino et al. showed that NCoA factors improve Tat transactivation of HIV-1 LTR promoter activity and interact with Tat [31]. Tat transactivation activity is mediated by its interaction with components of the basal transcription machinery (including TBP, TAFII250, RNA polymerase II), with kinase complexes able to phosphorylate the C-terminal domain of RNA polymerase II (in particular with the P-TEFb complex composed of cyclin T1/CDK9) and with cellular proteins possessing HAT activity (p300/CBP, P/CAF and GCN5) (reviewed in [39]). Kino et al. showed that one member of the family, NCoA2, functions as a Tat coactivator on the HIV-1 LTR by bridging promoter-bound proteins with the Tat-P-TEFb complex through its interaction with Tat and Cyclin T1 [31]. Stimulation of Tat transactivation activity by NCoA3 could involve similar mechanisms.
Furthemore, it has been recently demonstrated that recruitment of HATs to the LTR is an early event in HIV-1 transcriptional activation [13] and that a consequence of histone acetylation is the recruitment of the ATP-dependent chromatin remodeling complex hSWI/SNF to the LTR [12]. NCoA3 could mediate chromatin remodeling by recruitment of additional cofactors with HAT activity (such as p300/CBP and P/CAF) and by an intrinsic HAT activity [40] and may thus contribute to the transcriptional reactivation of the HIV-1 promoter from latency.
IRF8 is a transcription factor that binds to ISRE and regulates expression of genes stimulated by IFNs (reviewed in [29]). IRF8 is able to both activate and repress gene transcription depending on the target gene. We show that IRF8 (Unigene Hs. 137427) gene is only expressed in the promonocytic cell line U1 and its expression is strongly downregulated following NaB treatment of these cells. This differential transcriptional expression was confirmed by real-time RT-PCR and is also observed, albeit at lower extent, after PMA and TSA treatments of U1 cells. IRF8 protein level is similarly downregulated following treatment with NaB. Moreover, IRF8 represses the IRF1-mediated activation of the HIV-1 ISRE element of the LTR, in accordance with other data [32]. The decreased expression of IRF8 following reactivation of latency using different molecules suggest that IRF8 may contribute in the maintenance of the latent state in the promonocytic cell line.
It has been shown that binding of specific transcription factors downstream of the HIV-1 transcription start site is crucial to control HIV-1 transcription [33,41]. Among these sites is an ISRE element that recruits IRF1 and IRF2 in vivo [33]. Previous studies have investigated the role of IRFs on the modulation of HIV-1 replication (reviewed in [42,43]) and showed that IRF1 activates HIV-1 LTR transcription, interacts with Tat [32] and increases HIV-1 replication [44]. However, IRF8 represses IRF1-Tat-mediated transactivation of the LTR by interfering with IRF1-Tat association [32]. Moreover, it has been shown that IRF8 inhibits HIV-1 replication in T CD4+ lymphocytic and promonocytic cell lines [32,34]. These data support the hypothesis that repression of HIV-1 transcription by IRF8 could be implicated in the maintenance of proviral quiescence in latently infected cells.
Moreover, the result obtained after measurement of gag, NCoA3 and IRF8 mRNA after different times of NaB stimulation clearly showed a correlation between gag mRNA increase and NCoA3 mRNA increase or IRF8 mRNA decrease, respectively. These correlations support the hypothesis that IRF8 and NCoA3 factors may be involved in the control of the HIV latency.
Chronically HIV-1 infected cell lines used in this study provide useful models for studying HIV-1 latency but are not in a quiescent state as cellular reservoirs in vivo. Moreover, it has been shown that mutations in the tat gene and in the TAR sequence are responsible for the latency observed in U1 and ACH-2 cells, respectively [45,46]. We thus confirmed the differential expression of NCoA3 but not IRF8 genes in two others chronically HIV-1 infected cell lines, OM10.1 and J1.1. We will now investigate the involvement of NCoA3 and IRF8 to regulate viral expression in primary cells such as resting T CD4+ lymphocytes or macrophages.
Conclusion
Additional experiments are currently underway to validate the biological relevance of the differential expression of IRF8 and NCoA3 genes in latency maintenance and reactivation. Since the persistence of integrated HIV-1 genomes despite potent suppression of viral replication is a major obstacle for current antiretroviral therapy, selective disruption of the HIV-1 proviral latency may provide good strategies to decrease latent HIV-1 reservoirs. Thus, identification of cellular genes that are differentially expressed during HIV-1 reactivation of latency is crucial to understand the molecular mechanisms involved in the control of HIV-1 latency.
Methods
Cell cultures and treatments
The chronically HIV-1 infected T CD4+ lymphocytic cell lines ACH-2 [47] and J1.1 [48] derived from CEM and Jurkat cells respectively, and the chronically HIV-1 infected promonocytic cell lines U1 [49] and OM10.1 [50] derived from U937 and HL-60 cells respectively, were obtained through the National Institutes of Health (NIH) AIDS Research and Reference Reagent Program. Suspension cell lines were grown in RPMI 1640 (Invitrogen) with 10% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen). Cells were treated with 10 mM of sodium butyrate (NaB; Sigma), or with 10 ng/mL of PMA (Sigma), or with 300 nM of TSA (Sigma). Cells were harvested generally 24 h and 48 h after treatment and cell viability was estimated before subsequent RNA extraction or nuclear extract preparation. P4 indicator cells are HeLa CD4+ cells carrying the lacZ gene under the control of the HIV-1 LTR. P4 and HEK293 cells were grown in DMEM (Invitrogen) containing 5% fetal bovine serum (Invitrogen), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen) and 2 mM glutamine (Invitrogen).
Plasmids
The pLTRX-luc construct contains the luciferase (luc) gene downstream of the HIV-1 BRU U3-R promoter region (nt -640 to +78) [51]. The pLTRΔTAR-luc construct corresponds to the pLTRX-luc plasmid in which the TAR region (nt +38 to +78) was deleted [51]. The pCMV-Tat expression vector was kindly provided by S. Emiliani (Institut Cochin, Paris, France). The pIRF8 expression vector (pcDNAmycHis-ICSBP) and dominant negative construct pIRF8-DBD, which contains the DNA binding domain of IRF8, were a kind gift of B.Z. Levi (Technion-Israel Institute of Technology, Haifa, Israel). The pNCoA3 expression vector (pcDNA3.1-AIB1) was a kind gift of P.S. Meltzer (NIH, Bethesda, USA) [52]. The pIRF1 construct was generated by cloning the fragment excised from pHuIRF-3-1 (a kind gift of T. Taniguchi, University of Tokyo, Tokyo, Japan) by HindIII/NotI digestion in the pcDNA3.1 plasmid (Invitrogen). The pISRE-TK-luc and pISREmut-TK-luc constructs were generated by cloning a wild-type (AGGGACTTGAAAGCGAAAGGGAAACCAGAG) or mutated (AGGGACTTGCCCGCGCCCGGGAAACCAGAG) synthetic oligonucleotide corresponding to the HIV-1 BRU ISRE sequence (nt +194 to +223) [33,53] in the pTK-luc plasmid in which the luciferase gene is under the control of the truncated HSV-1 thymidine kinase promoter minimum region [51]. The pCMV-LacZ was kindly provided by M. Alizon (Institut Cochin, Paris, France).
Total RNA extraction
Total RNAs were extracted using the RNeasy Mini Kit (Qiagen). The procedure included an "on-column" DNase I digestion step according to the manufacturer's instructions. RNA quality was assessed using the Agilent Bioanalyzer 2100 and spectrophotometric analysis prior to cDNA synthesis.
Microarray experiments
Microarray experiments were performed using the U133-A microarrays (Affymetrix) containing 22283 oligonucleotides spots. Total RNAs obtained from chronically infected U1 and ACH-2 stimulated or not with NaB for 24 h were sent to Dr. C. Thibault (Affymetrix Microarray Facilities, IGBMC, Strasbourg, France) for amplification, labeling and hybridization. Hybridization experiments were performed once. Results were then analyzed with the Mas5.0 Software (Affymetrix) and interpreted using the Data Mining Tool (Affymetrix) and Microsoft Excel softwares. For individual analyses, the p-value cut off was 0.048 as suggested by Affymetrix. For comparative analyses, a log2 ratio change ≥ 1 for increased genes and ≤ -1 for decreased genes were defined. Gene expression changes were considered to be significant when the change p-value was ≤ 0.0001 for increased genes and 1-change p-value ≥ 0.9999 for decreased genes.
Real-time RT-PCR
Quantifications of cellular RNAs were performed using a Light Cycler instrument (Roche Diagnostics). Briefly, cDNAs were synthesized from 1 μg of total RNA with MoMLV reverse transcriptase (Superscript II, Invitrogen) and 1/10th aliquots of the corresponding samples were used for real-time PCR in a 20 μL reaction mixture containing 1X LightCycler FastStart DNA Master SYBR Green I (Roche Diagnostics), 4 mM MgCl2, and 500 nM of each primer. The reactions were carried out in duplicate and the results were normalized to the expression of Cyclophilin A. Primers for quantitative PCR were designed using Oligo 6 software. All primer pairs produced single amplification product as determined by melting curve analyses. The sequences of the primers used were (5' to 3'): NCoA3 forward CTTTGGGCATTCCTGAACTTGTC, NCoA3 reverse GCCTCATCACCGCAGCAC, IRF8 forward GGAGTGCGGTCGCTCTGAAA, IRF8 reverse GTCGTAGGTGGTGTACCCCGTCA, Cyclophilin A forward AGTGGTTGGATGGCAAGC, Cyclophilin A reverse GATTCTAGGATACTGCGAGCAAA. PCR reactions were carried out with a denaturation step of 10 min at 95°C followed by forty-five cycles of 10 s at 95°C, 5 s at annealing temperature (55°C for NCoA3 and Cyclophilin A, 59°C for IRF8) and 20 s amplification at 72°C. Quantifications of cDNAs were determined in reference to a standard curve prepared by amplification of serial dilutions of PCR product containing matching sequences. Analyses were performed using the second-derivative-maximum method provided by the Light Cycler quantification software, version 3.5 (Roche Diagnostics).
Quantification of gag viral mRNA was performed by real-time RT-PCR as described in [54].
Nuclear extracts preparation
For nuclear extract preparation, 10.106 cells were harvested, washed and nuclei were isolated by addition of 150 μL of buffer I (50 mM Tris pH 7.9, 10 mM KCl, 10% glycerol, 1 mM EDTA, 0.2% NP40) followed by a centrifugation at 3000 g for 3 min. Nuclear extracts were prepared by addition of 15 μL of buffer II (20 mM Hepes pH 7.9, 400 mM NaCl, 10 mM KCl, 20% glycerol, 1 mM EDTA) for 20 min at 4°C followed by a centrifugation at 15000 g for 10 min. Protein concentrations were determined by the Bio-Rad protein assay.
Western blot analysis
Nuclear extracts (100 μg) were loaded on 8% SDS-polyacrylamide gel and the proteins were transferred to nitrocellulose membrane (Hybond-C, Amersham) that was subsequently blocked for 1 h with 5% non-fat dry milk in PBS-T (PBS, 0.05% Tween20) and incubated with antibodies directed against NCoA3 (goat polyclonal anti-ACTR C-20, Santa Cruz Biotechnology, Inc.), IRF8 (goat polyclonal anti-ICSBP C-19, Santa Cruz Biotechnology, Inc.) or actin (mouse monoclonal anti-actin, Calbiochem) for 2 h. The membranes were then washed and incubated with secondary antibodies conjugated to horseradish peroxidase (HRP conjugated rabbit anti-goat (DakoCytomation) or goat anti-mouse (Calbiochem) immunoglobulins). Hybridizations were revealed using an ECL enhanced chemiluminescence kit (ECL, Amersham). The quantification was done using the Image Tools (Syngene) software.
Transient transfection and enzymatic assays
HEK293 cells were transfected using calcium phosphate co-precipitation method. Cells were lysed 24 h after transfection with a buffer containing 60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 10 mM MgSO4, 2.5 mM EDTA, 50 mM β-mercaptoethanol and 0.125% Nonidet P-40. Luciferase activities were measured as previously described [55]. Cotransfection with pCMV-LacZ plasmid was performed to normalize transfection efficiency and β-galactosidase activities were determined using Chlorophenol red β-D-galactopyranoside (CPRG, Roche Diagnostics) assay as previously described [55]. The normalized luciferase index (NLI) was defined as the ratio of luciferase to β-galactosidase activities.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SM performed the microarray analyses, real-time RT-PCR and drafted the manuscript. DD and LC carried out real-time RT-PCR, Western blot and transfection experiments. AG participated in transfection experiments. UH conceived the study, participated in its design and coordination and helped to draft and finalize the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Genes upregulated in U1 and ACH-2 cells.
Click here for file
Additional File 2
Genes specifically upregulated in U1 cells.
Click here for file
Additional File 3
Genes specifically upregulated in ACH-2 cells.
Click here for file
Additional File 4
Genes downregulated in U1 and ACH-2 cells.
Click here for file
Additional File 5
Genes specifically downregulated in U1 cells.
Click here for file
Additional File 6
Genes specifically downregulated in ACH-2 cells.
Click here for file
Acknowledgements
We thank Drs. S. Emiliani, B.Z. Levi, P.S. Meltzer, T. Taniguchi, M. Alizon for providing plasmids, and the National Institutes of Health AIDS Research and Reference Reagent Program for the kind gift of reagents. We are grateful to Dr. C. Thibault for her precious help in microarray hybridization data mining. L.C. holds a fellowship from the Ministère de l'Education Nationale, de l'Enseignement Supérieur et de la Recherche. S.M. is supported by a grant from Sidaction. Sidaction (AO15) supported this work. We thank S. Nisole for helpful discussion.
==== Refs
Pomerantz RJ Horn DL Twenty years of therapy for HIV-1 infection Nat Med 2003 9 867 873 12835707 10.1038/nm0703-867
Pierson T McArthur J Siliciano RF Reservoirs for HIV-1: mechanisms for viral persistence in the presence of antiviral immune responses and antiretroviral therapy Annu Rev Immunol 2000 18 665 708 10837072 10.1146/annurev.immunol.18.1.665
Wong JK Hezareh M Gunthard HF Havlir DV Ignacio CC Spina CA Richman DD Recovery of replication-competent HIV despite prolonged suppression of plasma viremia Science 1997 278 1291 1295 9360926 10.1126/science.278.5341.1291
Finzi D Hermankova M Pierson T Carruth LM Buck C Chaisson RE Quinn TC Chadwick K Margolick J Brookmeyer R Gallant J Markowitz M Ho DD Richman DD Siliciano RF Identification of a reservoir for HIV-1 in patients on highly active antiretroviral therapy Science 1997 278 1295 1300 9360927 10.1126/science.278.5341.1295
Chun TW Stuyver L Mizell SB Ehler LA Mican JA Baseler M Lloyd AL Nowak MA Fauci AS Presence of an inducible HIV-1 latent reservoir during highly active antiretroviral therapy Proc Natl Acad Sci U S A 1997 94 13193 13197 9371822 10.1073/pnas.94.24.13193
Blankson JN Persaud D Siliciano RF The challenge of viral reservoirs in HIV-1 infection Annu Rev Med 2002 53 557 593 11818490 10.1146/annurev.med.53.082901.104024
Lassen K Han Y Zhou Y Siliciano J Siliciano RF The multifactorial nature of HIV-1 latency Trends Mol Med 2004 10 525 531 15519278 10.1016/j.molmed.2004.09.006
Finzi D Blankson J Siliciano JD Margolick JB Chadwick K Pierson T Smith K Lisziewicz J Lori F Flexner C Quinn TC Chaisson RE Rosenberg E Walker B Gange S Gallant J Siliciano RF Latent infection of CD4(+) T cells provides a mechanism for lifelong persistence of HIV-1, even in patients on effective combination therapy Nat Med 1999 5 512 517 10229227 10.1038/8394
Williams SA Greene WC Host factors regulating post-integration latency of HIV Trends Microbiol 2005 13 137 139 15817380 10.1016/j.tim.2005.02.006
He G Ylisastigui L Margolis DM The regulation of HIV-1 gene expression: the emerging role of chromatin DNA Cell Biol 2002 21 697 705 12443539 10.1089/104454902760599672
Verdin E Paras PJ Van Lint C Chromatin disruption in the promoter of human immunodeficiency virus type 1 during transcriptional activation Embo J 1993 12 3249 3259 8344262
Henderson A Holloway A Reeves R Tremethick DJ Recruitment of SWI/SNF to the human immunodeficiency virus type 1 promoter Mol Cell Biol 2004 24 389 397 14673171 10.1128/MCB.24.1.389-397.2004
Lusic M Marcello A Cereseto A Giacca M Regulation of HIV-1 gene expression by histone acetylation and factor recruitment at the LTR promoter Embo J 2003 22 6550 6561 14657027 10.1093/emboj/cdg631
Kato-Maeda M Gao Q Small PM Microarray analysis of pathogens and their interaction with hosts Cell Microbiol 2001 3 713 719 11696031 10.1046/j.1462-5822.2001.00152.x
Shaheduzzaman S Krishnan V Petrovic A Bittner M Meltzer P Trent J Venkatesan S Zeichner S Effects of HIV-1 Nef on cellular gene expression profiles J Biomed Sci 2002 9 82 96 11810028
Simmons A Aluvihare V McMichael A Nef triggers a transcriptional program in T cells imitating single-signal T cell activation and inducing HIV virulence mediators Immunity 2001 14 763 777 11420046 10.1016/S1074-7613(01)00158-3
De la Fuente C Santiago F Deng L Eadie C Zilberman I Kehn K Maddukuri A Baylor S Wu K Lee CG Pumfery A Kashanchi F Gene expression profile of HIV-1 Tat expressing cells: a close interplay between proliferative and differentiation signals BMC Biochem 2002 3 14 12069692 10.1186/1471-2091-3-14
Izmailova E Bertley FM Huang Q Makori N Miller CJ Young RA Aldovini A HIV-1 Tat reprograms immature dendritic cells to express chemoattractants for activated T cells and macrophages Nat Med 2003 9 191 197 12539042 10.1038/nm822
Cicala C Arthos J Selig SM Dennis GJ Hosack DA Van Ryk D Spangler ML Steenbeke TD Khazanie P Gupta N Yang J Daucher M Lempicki RA Fauci AS HIV envelope induces a cascade of cell signals in non-proliferating target cells that favor virus replication Proc Natl Acad Sci U S A 2002 99 9380 9385 12089333 10.1073/pnas.142287999
Janket ML Manickam P Majumder B Thotala D Wagner M Schafer EA Collman RG Srinivasan A Ayyavoo V Differential regulation of host cellular genes by HIV-1 viral protein R (Vpr): cDNA microarray analysis using isogenic virus Biochem Biophys Res Commun 2004 314 1126 1132 14751250 10.1016/j.bbrc.2004.01.008
Corbeil J Sheeter D Genini D Rought S Leoni L Du P Ferguson M Masys DR Welsh JB Fink JL Sasik R Huang D Drenkow J Richman DD Gingeras T Temporal gene regulation during HIV-1 infection of human CD4+ T cells Genome Res 2001 11 1198 1204 11435401 10.1101/gr.GR-1802R
Geiss GK Bumgarner RE An MC Agy MB van 't Wout AB Hammersmark E Carter VS Upchurch D Mullins JI Katze MG Large-scale monitoring of host cell gene expression during HIV-1 infection using cDNA microarrays Virology 2000 266 8 16 10612655 10.1006/viro.1999.0044
Van 't Wout AB Lehrman GK Mikheeva SA O'Keeffe GC Katze MG Bumgarner RE Geiss GK Mullins JI Cellular gene expression upon human immunodeficiency virus type 1 infection of CD4(+)-T-cell lines J Virol 2003 77 1392 1402 12502855 10.1128/JVI.77.2.1392-1402.2003
Wen W Chen S Cao Y Zhu Y Yamamoto Y HIV-1 infection initiates changes in the expression of a wide array of genes in U937 promonocytes and HUT78 T cells Virus Res 2005 15885842
Vahey MT Nau ME Jagodzinski LL Yalley-Ogunro J Taubman M Michael NL Lewis MG Impact of viral infection on the gene expression profiles of proliferating normal human peripheral blood mononuclear cells infected with HIV type 1 RF AIDS Res Hum Retroviruses 2002 18 179 192 11839152 10.1089/08892220252781239
Chun TW Justement JS Lempicki RA Yang J Dennis GJ Hallahan CW Sanford C Pandya P Liu S McLaughlin M Ehler LA Moir S Fauci AS Gene expression and viral prodution in latently infected, resting CD4+ T cells in viremic versus aviremic HIV-infected individuals Proc Natl Acad Sci U S A 2003 100 1908 1913 12552096 10.1073/pnas.0437640100
Krishnan V Zeichner SL Host cell gene expression during human immunodeficiency virus type 1 latency and reactivation and effects of targeting genes that are differentially expressed in viral latency J Virol 2004 78 9458 9473 15308739 10.1128/JVI.78.17.9458-9473.2004
Xu J Li Q Review of the in vivo Functions of the p160 Steroid Receptor Coactivator Family Mol Endocrinol 2003 12 12
Levi BZ Hashmueli S Gleit-Kielmanowicz M Azriel A Meraro D ICSBP/IRF-8 transactivation: a tale of protein-protein interaction J Interferon Cytokine Res 2002 22 153 160 11846986 10.1089/107999002753452764
Bosinger SE Hosiawa KA Cameron MJ Persad D Ran L Xu L Boulassel MR Parenteau M Fournier J Rud EW Kelvin DJ Gene expression profiling of host response in models of acute HIV infection J Immunol 2004 173 6858 6863 15557180
Kino T Slobodskaya O Pavlakis GN Chrousos GP Nuclear receptor coactivator p160 proteins enhance the HIV-1 long terminal repeat promoter by bridging promoter-bound factors and the Tat-P-TEFb complex J Biol Chem 2002 277 2396 2405 11704662 10.1074/jbc.M106312200
Sgarbanti M Borsetti A Moscufo N Bellocchi MC Ridolfi B Nappi F Marsili G Marziali G Coccia EM Ensoli B Battistini A Modulation of human immunodeficiency virus 1 replication by interferon regulatory factors J Exp Med 2002 195 1359 1370 12021315 10.1084/jem.20010753
Van Lint C Amella CA Emiliani S John M Jie T Verdin E Transcription factor binding sites downstream of the human immunodeficiency virus type 1 transcription start site are important for virus infectivity J Virol 1997 71 6113 6127 9223506
Thornton AM Ogryzko VV Dent A Sharf R Levi BZ Kanno Y Staudt LM Howard BH Ozato K A dominant negative mutant of an IFN regulatory factor family protein inhibits both type I and type II IFN-stimulated gene expression and antiproliferative activity of IFNs J Immunol 1996 157 5145 5154 8943426
Herrmann CH Carroll RG Wei P Jones KA Rice AP Tat-associated kinase, TAK, activity is regulated by distinct mechanisms in peripheral blood lymphocytes and promonocytic cell lines J Virol 1998 72 9881 9888 9811724
Van 't Wout AB Swain JV Schindler M Rao U Pathmajeyan MS Mullins JI Kirchhoff F Nef induces multiple genes involved in cholesterol synthesis and uptake in human immunodeficiency virus type 1-infected T cells J Virol 2005 79 10053 10058 16014965 10.1128/JVI.79.15.10053-10058.2005
Krishnan V Zeichner SL Alterations in the expression of DEAD-box and other RNA binding proteins during HIV-1 replication Retrovirology 2004 1 42 15588285 10.1186/1742-4690-1-42
Marks P Rifkind RA Richon VM Breslow R Miller T Kelly WK Histone deacetylases and cancer: causes and therapies Nat Rev Cancer 2001 1 194 202 11902574 10.1038/35106079
Brigati C Giacca M Noonan DM Albini A HIV Tat, its TARgets and the control of viral gene expression FEMS Microbiol Lett 2003 220 57 65 12644228 10.1016/S0378-1097(03)00067-3
Chen H Lin RJ Schiltz RL Chakravarti D Nash A Nagy L Privalsky ML Nakatani Y Evans RM Nuclear receptor coactivator ACTR is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300 Cell 1997 90 569 580 9267036 10.1016/S0092-8674(00)80516-4
El Kharroubi A Martin MA cis-acting sequences located downstream of the human immunodeficiency virus type 1 promoter affect its chromatin structure and transcriptional activity Mol Cell Biol 1996 16 2958 2966 8649407
Battistini A Marsili G Sgarbanti M Ensoli B Hiscott J IRF regulation of HIV-1 long terminal repeat activity J Interferon Cytokine Res 2002 22 27 37 11846973 10.1089/107999002753452638
Marsili G Borsetti A Sgarbanti M Remoli AL Ridolfi B Stellacci E Ensoli B Battistini A On the role of interferon regulatory factors in HIV-1 replication Ann N Y Acad Sci 2003 1010 29 42 15033691 10.1196/annals.1299.005
Sgarbanti M Marsili G Remoli AL Ridolfi B Stellacci E Borsetti A Ensoli B Battistini A Analysis of the signal transduction pathway leading to human immunodeficiency virus-1-induced interferon regulatory factor-1 upregulation Ann N Y Acad Sci 2004 1030 187 195 15659797 10.1196/annals.1329.024
Emiliani S Van Lint C Fischle W Paras PJ Ott M Brady J Verdin E A point mutation in the HIV-1 Tat responsive element is associated with postintegration latency Proc Natl Acad Sci U S A 1996 93 6377 6381 8692823 10.1073/pnas.93.13.6377
Emiliani S Fischle W Ott M Van Lint C Amella CA Verdin E Mutations in the tat gene are responsible for human immunodeficiency virus type 1 postintegration latency in the U1 cell line J Virol 1998 72 1666 1670 9445075
Clouse KA Powell D Washington I Poli G Strebel K Farrar W Barstad P Kovacs J Fauci AS Folks TM Monokine regulation of human immunodeficiency virus-1 expression in a chronically infected human T cell clone J Immunol 1989 142 431 438 2463307
Perez VL Rowe T Justement JS Butera ST June CH Folks TM An HIV-1-infected T cell clone defective in IL-2 production and Ca2+ mobilization after CD3 stimulation J Immunol 1991 147 3145 3148 1833465
Folks TM Justement J Kinter A Schnittman S Orenstein J Poli G Fauci AS Characterization of a promonocyte clone chronically infected with HIV and inducible by 13-phorbol-12-myristate acetate J Immunol 1988 140 1117 1122 2449497
Butera ST Perez VL Wu BY Nabel GJ Folks TM Oscillation of the human immunodeficiency virus surface receptor is regulated by the state of viral activation in a CD4+ cell model of chronic infection J Virol 1991 65 4645 4653 1678437
Hazan U Thomas D Alcami J Bachelerie F Israël N Yssel H Virelizier JL Arenzana-Seisdedos F Stimulation of human T-cell clone with anti-CD3 or tumor necrosis factor induces NF-kappa B translocation but not human immunodeficiency virus 1 enhancer-dependent transcription. Proc Natl Acad Sci USA 1990 87 7861 7865 2146676
Anzick SL Kononen J Walker RL Azorsa DO Tanner MM Guan XY Sauter G Kallioniemi OP Trent JM Meltzer PS AIB1, a steroid receptor coactivator amplified in breast and ovarian cancer Science 1997 277 965 968 9252329 10.1126/science.277.5328.965
El Kharroubi A Verdin E Protein-DNA interactions within DNase I-hypersensitive sites located downstream of the HIV-1 promoter J Biol Chem 1994 269 19916 19924 8051074
Brussel A Sonigo P Evidence for gene expression by unintegrated human immunodeficiency virus type 1 DNA species J Virol 2004 78 11263 11271 15452245 10.1128/JVI.78.20.11263-11271.2004
Dumonceaux J Nisole S Chanel C Quivet L Amara A Baleux F Briand P Hazan U Spontaneous mutations in the env gene of the human immunodeficiency virus type 1 NDK isolate are associated with a CD4-independent entry phenotype J Virol 1998 72 512 519 9420253
|
16269085
|
PMC1310521
|
CC BY
|
2021-01-04 16:38:20
|
no
|
Ann Clin Microbiol Antimicrob. 2005 Nov 3; 4:18
|
latin-1
|
Ann Clin Microbiol Antimicrob
| 2,005 |
10.1186/1476-0711-4-18
|
oa_comm
|
==== Front
Microb Cell FactMicrobial Cell Factories1475-2859BioMed Central London 1475-2859-4-321628393610.1186/1475-2859-4-32ReviewInteins and affinity resin substitutes for protein purification and scale up Banki Mahmoud Reza [email protected] David W [email protected] Department of Chemical Engineering, A213 E-QUAD, Princeton University, Princeton, NJ 08544, USA2005 11 11 2005 4 32 32 4 10 2005 11 11 2005 Copyright © 2005 Banki and Wood; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The development of self-cleaving fusion-tag technology has greatly simplified the purification of recombinant proteins at laboratory scale. The self-cleaving capability of these tags has recently been combined with additional purification tags to generate novel and convenient protein purification methods at a variety of scales. In this review, we describe some of these methods, and provide a rudimentary economic analysis of hypothetical large-scale applications. This work is expected to provide a rough outline for the evaluation of these methods for large-scale bioprocessing of a variety of products.
==== Body
Introduction
An important development in the area of recombinant protein purification has been the incorporation of self-cleaving protein elements into a variety of fusion-based purification systems [1-3]. These elements are derived from naturally occurring self-splicing inteins through various protein engineering strategies, and have been combined with conventional affinity tags in a variety of configurations to yield highly effective separations methods. Very recently, these elements have also been combined with non-conventional purification tags to yield "self-purifying" proteins, which can deliver highly purified native products using simple mechanical means without chromatographic methods [4,5].
This review will compare conventional affinity-tag methods, with and without proteolytic tag removal, to three newer methods based on self-cleaving purification tags. The three newer methods include a conventional affinity tag separation with a self-cleaving chitin-binding tag (the IMPACT system), a more recent method where the expression host produces a granular affinity matrix during fermentation (the PHB system), and a third in which the target protein is tagged with a reversibly-precipitating self-cleaving polypeptide (the ELP system). In particular, the advantages and disadvantages of each method will be compared, and the large-scale economics of each of these systems will be examined from a simple raw-materials cost standpoint. This simple analysis is intended to describe the relative merits of these methods, and to provide an initial benchmark for evaluating their potential future role in the large-scale manufacture of recombinant products.
Conventional Affinity-Tag Methods
Affinity fusion-based protein purification is a simple and now widely used method which takes advantage of the selective binding property of a genetically fused binding protein (tag) to purify a given target protein [6,7]. In place of physicochemical properties of the target protein, this technique relies on the specific binding of the affinity tag to an immobilized ligand. By exploiting this highly specific interaction, a single purification step can effectively isolate and purify a given target protein with ease. The development of numerous tags has further demonstrated the flexibility and potential of this method. Despite these strengths, however, the use of conventional gene-fusion affinity tags suffers from two main drawbacks.
The first limitation arises from the requirement that the tag be removed in order to recover a native target protein. This is generally accomplished by enzymatically removing the tag from the purified target by the addition of an appropriate protease. To facilitate this procedure, the target sequence of the selected protease is genetically included between the tag and the target protein when the fusion is constructed, allowing specific cleaving to take place. Although this procedure is generally effective at laboratory scales, the cost of protease enzymes is prohibitive at manufacturing scale. In addition, yield losses can arise from incomplete cleaving or unexpected cleaving within the target, and the affinity tag and protease must also be separated from the cleaved target protein in a separate purification step. Both of these aspects increase the cost and complexity of the purification, while decreasing the yield.
A second limitation arises from the equipment and consumable resin costs associated with these procedures. Conventional affinity resins typically consist of various cross-linked polymers, derivitized with appropriate ligands at the end of optimized spacer arms. Manufacturing costs for these resins are typically much higher than for ion-exchange and other chromatography resins, which can offset the appeal of the simpler affinity-based separation. A notable exception has been the widespread use of Protein A affinity columns in the purification of antibody therapeutics. However, this separation is limited to native antibodies, without the addition of a fusion tag. This suggests that conventional affinity tag methods may be attractive if tag removal can be simplified [8].
For these reasons, new methods which eliminate the need for protease treatment and expensive affinity resins are likely to make a significant impact on large-scale protein purification processes or high-throughput screening of protein libraries. The next two sections address the two drawbacks mentioned above and offer recently developed solutions.
Self-cleaving Affinity Tags
Inteins (INTervening protEINS) are naturally occurring protein sequences capable of post-translational self-excision from a host-intein precursor protein through a process known as "protein splicing" [9,10]. Several intein examples have been identified where the intein is capable of functioning outside of its native context, allowing these inteins to be developed for a variety of biotechnological applications. One of the most significant of these is the creation of self-cleaving protein elements that can be combined with conventional affinity tags to generate effective self-cleaving affinity tags [1-3]. A critical feature of these tags is their ability to release a target protein, fused either C or N-terminally to the tag, in response to a simple chemical or physical stimulus. The highly specific cleaving reaction thus allows the affinity tag to be removed without the addition of expensive protease, while at the same time preventing unwanted cleaving. An additional important advantage is that the cleaving reaction can be induced while the tagged target is bound to the affinity column, thus eliminating the need for subsequent removal of the cleaved tag.
The first commercially available intein purification system was developed by New England Biolabs (NEB), and is based on a modified Saccharomyces cerevisiae vacuolar ATPase subunit A intein (Sce VMA intein) [1]. This intein possesses a particular mutation, leading it to exhibit N-terminal cleaving in the presence of 30 mM 1,4-dithiothreitol (DTT) or β-mercaptoethanol over a wide pH range (5.5 – 9.0). This intein was combined with a chitin binding domain and appropriate resin to form the IMPACT system, and has been effective in the purification of several proteins and restriction enzymes at laboratory scale. The original IMPACT system has now been enhanced to form the IMPACT-CN system, where C-terminal cleaving can also be induced through the addition of DTT to an appropriate fusion precursor. Thus both N and C-terminally fused target proteins can be purified, allowing the design of the fusion to be optimized for expression and folding of the target. NEB has more recently expanded their line of intein-based separation systems to include three modified mini-inteins derived from the Mycobacterium xenopi GyrA enzyme, the Synechocystis spp. strain PCC6803 DnaB helicase, and the Methanothermobacter thermautotrophicus Ribonucleoside-diphosphate reductase enzyme [2,11,12]. These inteins are included in the pTWIN system, and provide the capability to induce cleaving by thiol addition (such as DTT or β-mercaptoethanol as above), small shifts in pH or increases in temperature.
In addition to these commercially available inteins, an engineered mini-intein derived from the Mycobacterium tuberculosis (Mtu) RecA intein has been developed independently [3,13]. This 18 kDa intein was developed through a deletion of the endonuclease domain from the native Mtu RecA intein, followed by mutagenesis and selection for rapid and controllable cleaving. This intein, referred to as the ΔI-CM mini-intein, can be controlled by pH and temperature to yield isolated C-terminal cleavage, and has been successful in delivering native, purified target proteins from various E. coli expression systems. It is this intein that has been combined with the novel purification tags described above to generate convenient "self-purifying" expression systems. The potential of these systems in the purification of recombinant proteins at large scale will be examined in the following sections.
The PHB System
Polyhydroxybuterates (PHBs) are a subclass of biodegradable polymers produced in various organisms and are generally thought to be a means for storing excess carbon in the absence of oxygen, nitrogen or phosphorus [14]. Intracellular PHB takes the form of small granules when expressed, which can vary in morphology based on the expressing organism, the carbon source, and the expression level of accompanying proteins called phasins [15]. The specific affinity of phasin proteins for PHB has been exploited in the development of a self-contained affinity purification system [5]. In this case, the expressing cells harbor two plasmids; one expresses the PHB-synthesis genes, while the other expresses a target protein in fusion to a self-cleaving phasin tag. The large molecular weight of PHB granules allows the simple recovery and cleaning of the granules and bound fusion protein, while the self-cleaving intein allows the purified native target protein to be released from the granule surface once it is purified. Because PHB granules can be readily synthesized from cheap carbon sources such as glucose or lactate, this system provides a low-cost alternative to manufactured and processed affinity beads. Utility of PHB granules in purification has been described, where a phasin tag has been combined with the ΔI-CM mini-intein and used for the purification of several active proteins with competitive yields [5].
The ELP System
Another alternative to conventional resins is a recombinantly produced elastin-like polypeptide (ELP), generally comprised of repeating units of the five amino acids VPGXG (X = any amino acid) [16,17]. Because of the unique salt and temperature-sensitive solubility of ELP, it can be easily purified by salt addition and mild temperature shifts. By combining an ELP tag with the ΔI-CM mini-intein, a method has been created which allows the rapid and simple purification of arbitrary tagged target proteins [4]. In this case, the tagged target is separated from the insoluble components of the cell debris by centrifugation at low temperature, where the ELP is soluble. Addition of salt and an increase in temperature to 30°C causes the ELP portion of the fusion to self-assemble into an insoluble precipitate, allowing it to be easily separated from the remaining soluble components of the cell lysate. Because the precipitation is limited to the ELP portion of the protein, which is separated from the intein and target by a flexible linker peptide, the activity of the intein and target are not affected. Intein cleaving then releases the native target from the ELP tag, which can then be easily separated by an additional precipitation step. This method is compatible with both centrifugation and filtration for recovering and separating the ELP-fusion precipitate. Initial reports indicate that this technique is highly effective in purifying active and native protein products expressed in E. coli, although it is anticipated it will be compatible with a large number of expression systems.
Large-Scale Economics
Despite the clear potential of intein-based separations in industry, little has been done to adapt these methods to large-scale processes. Recent work describes the activity of inteins expressed in high cell-density fermentation, and one recent paper examines the use of inteins with vortex-flow affinity-resin loading [18,19]. Here, we present a comparison of several conventional and intein-based affinity processes from a materials standpoint. Two conventional affinity-based purification methods were used as benchmarks: the maltose-binding protein fusion with proteolytic tag removal (pMAL) and His-tagged purification without tag removal. The pMAL system is available from New England Biolabs (Beverly, MA, USA) and we have chosen the Novagen (Madison, WI, USA) His-bind Purification System from a number of available His-tag purification systems. These two techniques are frequently used for small, lab-scale processes. Despite the high purities attainable with these two systems, they have not yet been adopted for large-scale enzyme production primarily due to the high cost of proteases and affinity resins. The IMPACT system, which circumvents the protease problem through DTT-induced intein cleaving, is the third system we have analyzed. Finally, two recently-developed PHB and ELP methods, which allow pH-induced intein cleaving by the ΔI-CM mini intein and virtually eliminate affinity resin costs, are also included for comparison (Figure 1). Comparison of triggers of target protein cleavage and recovery from the precursor protein fusions are also shown for all five methods (Figure 2).
A materials-based cost comparison can be a decisive prelude to the adoption of the PHB-intein and ELP-intein methods and also in predicting the significant cost savings possible with these two new technologies. Our economic analysis (Table 1) is based on published technical manuals from New England Biolabs (for the Maltose-binding and IMPACT methods), Novagen (for the His-tag method) and published work (for the PHB and ELP methods), and is limited to the consumable costs associated with each process. For this analysis, prices were calculated using supplier list prices for the largest quantities available (best rates) and are at times extrapolations on small-scale amounts. Although bulk order of chemicals and materials could further reduce the cost, the comparison presented here uses the same cost basis for all five methods and hence does not favor a method over another in individual sub-categories. Furthermore, costs associated with pH adjustments, ultrapure water, centrifugation, cooling, heating and plant operation are not considered due to their commonality in all of the processes. Finally, the induction (IPTG) cost is listed separately because it can potentially be eliminated in all five methods by using self-inducing expression strains, some of which are now commercially available (Novagen).
The pMAL and His-tag methods have been commercialized and have thus matured and been optimized for buffer consumption. Therefore, these two methods are most sensitive to the resin cost; a cost that can not be reduced or compromised. However, in the newly developed intein-dependent methods, the growth media and buffers are a larger fraction of the total cost. These components can potentially be replaced by cheaper alternatives, adding to the economic attractiveness of the PHB and ELP methods.
Even without exhaustive buffer and growth medium optimization, this comparison shows up to a 125 fold decrease in materials cost for the PHB and ELP methods in comparison to the pMAL affinity based purification. Likewise, these two technologies reduce the materials cost up to 11 fold in comparison with the His-bind purification procedure. This is a dramatic improvement over previously existing technologies and could thus have a significant impact on the future of the biotechnology industry.
Conclusion
The cost analysis presented here shows the dramatic improvements possible for large-scale protein purification processes through the use of non-chromatographic self-cleaving purification tags. These methods are immediately attractive for large-scale industrial products, where small levels of impurities are tolerable. In pharmaceutical and other applications where high purity is required, these methods can act as a first-capture step, delivering substantially purified material for downstream polishing. In addition, these methods are ideal for high throughput applications, where the simplicity and generality of each method can be applied to large libraries of targets in a highly parallel configuration. As the biochemistry associated with self-cleaving tags is further optimized, it is clear that this platform will be adaptable to many additional separation processes.
Acknowledgements
This work was partially supported by the National Science Foundation Graduate Student Fellowship and U.S. Army Research Office grant W911NF-04-1-0056.
Figures and Tables
Figure 1 Protein purification schematic for (a) conventional affinity-base purification (pMAL), (b) intein-based affinity purification where the linker between the affinity tag and the target protein is replaced by a self-cleaving intein, (c) PHB-intein mediated protein purification where in vivo PHB granules substitute for the affinity resin, and (d) ELP-intein mediated protein purification where an in vivo temperature sensitive, self-cleaving tag replaces both the affinity resin and the affinity tag. † Scanning electron micrograph image of PHB granules expression in E. coli was taken from Reference [20].
Figure 2 Comparison of triggers of cleavage as well as final products for the five methods. Amino acid sequences at fusion junctions are noted in one-letter amino acid code. X denotes that specific amino acids are preferred for this position.
Table 1 Material cost comparison for five different protein purification methods based on 1 kg product protein yield†
pMAL fusion w/tag removal His-tag w/o tag removal IMPACT-CN PHB System ELP System
Supplier NEB Novagen NEB WOOD LAB WOOD LAB
Typical yield 40 mg/L 74 mg/L 4.5 mg/L 36 mg/L 85 mg/L
Resin capacity 3 mg/ml 8 mg/ml 2 mg/ml - -
Growth medium $ 44,198.38 $ 23,101.67 $ 379,894.17 $ 126,257.54 $ 58,699.27
Induction cost $ 35,373.35 $ 19,101.61 $ 314,115.30 $ 39,304.03 -
Buffer cost $ 51,900.11 $ 5,921.45 $ 103,596.75 $ 27,212.63 $ 15,810.47
Resin cost $ 624,375.00 $ 790,000.00 $ 192,000.00 - -
Denaturation cost $ 428,835.99 - - - -
Protease cost $ 8,480,000.00 - - - -
Total cost per kg of product $ 9,664,682.82 $ 838,124.73 $ 989,606.23 $ 153,470.17 $ 74,509.75
† Cost calculations are based on a simple scaling-up of published protocols for laboratory-scale processes. Bulk prices of chemicals and growth media components were obtained from the best available rates per unit mass from Fisher Scientific. Items specific to individual methods were priced by the supplier (i.e. the protease cost was obtained from NEB, His-tag resin cost was obtained from Novagen, etc). Required quantities of all components (i.e. growth media, buffers, resin) were calculated based on the typical yields of each process for a final product yield of 1 kg. In addition, the following assumptions are included for each method: pMAL method – Yield estimate and material usage requirements were based on the supplier, NEB, recommendations in the pMAL Protein Fusion and Purification System manual [21]. In place of the additional DEAE sepharose ion exchange chromatography step for separating the protease and maltose tag, it was assumed that the same amylose resin would be used twice with regeneration. Therefore, one round of purification uses the amylose resin twice. Therefore with an assumed regeneration of four times, the same bed can be used for two separate purifications. Amylose resin binding capacity is 3 mg of fusion protein per ml of resin. This translates to 1 to 2 mg of target protein per ml of resin depending on the molecular weight of the target protein. Even so, it was assumed that 3 mg of maltose tag binds to each ml of resin, hence, underestimating the amylose resin cost. The recommended amount of protease (Factor Xa) is 1% (w/w) of the fusion protein, hence, if an average product protein 1/3 the size of the maltose binding domain tag is used, 4000 g of fusion is needed to produce 1000 g of target protein assuming perfect recovery. Despite this recommendation, the quantity of protease used was based on 1000 kg of fusion protein as opposed to target protein, hence, underestimating the protease cost. His-tag method – Typical yield was based on the published yield using Novagen Standard HisBind for purification of HisTag GST expressed in E. coli [22]. Material requirements were based on supplier (Novagen) recommendations for the HisBind kit [23]. In these calculations the protease step is not factored into the cost and the target protein retains the HisTag after purification. The regeneration is not taken into account because Novagen recommends using a different resin for each different protein. However, a routine regeneration single-step procedure or complete resin regeneration (16 steps) are available even though not considered here. IMPACT-CN method – The yield estimate was based on the average yield published by the supplier, NEB, in the IMPACT-CN manual [24]. Purified protein examples included in this estimate were [24]: Maltose-binding protein, McrB, T4 DNA ligase, Bst DNA polymerase large fragment, BamH I, Bgl II, CDK2, CamK II, T4 Gene 32 product, FseI GFP, CamK II, Invertase, and T4 Endo VII. It was assumed that the chitin beads can be regenerated 5 times as recommended by the manual. PHB method – Excess DTT in buffers for this method is only necessary for specific target proteins and has not been factored into the cost. In addition, cheaper phosphate substitute buffer is used in place of Bis-Tris in the calculation. IPTG induction does not significantly change the yield and has not been used in the calculation. Cost of purification for this method prior to these three modifications (DTT and IPTG elimination as well as Phosphate for Bis-Tris substitution) is $391,265.84 as opposed to the listed $153,470.17. ELP method – This method does not involve sonication and cell lysis can be achieved with 0.2 mg/ml lysozyme content in the lysis buffer (unpublished data) as opposed to the 1 mg/ml published [4]. Furthermore, glycerol was not included in the TB media as noted in the publication. In addition cheaper phosphate buffer is substituted for the published Bis-Tris buffer in this calculation. Cost of purification for this method without this modification (Phosphate for Bis-Tris substitution) is $122,754.12 as opposed to the listed $74,509.75.
==== Refs
Chong S Mersha FB Comb DG Scott ME Landry D Vence LM Perler FB Benner J Kucera RB Hirvonen CA Pelletier JJ Paulus H Xu MQ Single-column purification of free recombinant proteins using a self-cleavable affinity tag derived from a protein splicing element Gene 1997 192 271 281 9224900 10.1016/S0378-1119(97)00105-4
Southworth MW Amaya K Evans TC Xu MQ Perler FB Purification of proteins fused to either the amino or carboxy terminus of the Mycobacterium xenopi gyrase A intein Biotechniques 1999 27 110 4, 116, 118-20 10407673
Wood DW Wu W Belfort G Derbyshire V Belfort M A genetic system yields self-cleaving inteins for bioseparations Nat Biotechnol 1999 17 889 892 10471931 10.1038/12879
Banki MR Feng L Wood DW Simple bioseparations using self-cleaving elastin-like polypeptide tags Nat Methods 2005 2 659 662 16074986 10.1038/nmeth787
Banki MR Gerngross TU Wood DW Novel and economical purification of recombinant proteins: intein-mediated protein purification using in vivo polyhydroxybutyrate (PHB) matrix association Protein Sci 2005 14 1387 1395 15883185 10.1110/ps.041296305
Hearn MT Acosta D Applications of novel affinity cassette methods: use of peptide fusion handles for the purification of recombinant proteins J Mol Recognit 2001 14 323 369 11757069 10.1002/jmr.555
LaVallie ER McCoy JM Gene fusion expression systems in Escherichia coli Curr Opin Biotechnol 1995 6 501 506 7579661 10.1016/0958-1669(95)80083-2
Chapman T Protein purification: pure but not simple Nature 2005 434 795 798 15815637 10.1038/434795a
Chong S Shao Y Paulus H Benner J Perler FB Xu MQ Protein splicing involving the Saccharomyces cerevisiae VMA intein. The steps in the splicing pathway, side reactions leading to protein cleavage, and establishment of an in vitro splicing system J Biol Chem 1996 271 22159 22168 8703028 10.1074/jbc.271.36.22159
Perler FB Davis EO Dean GE Gimble FS Jack WE Neff N Noren CJ Thorner J Belfort M Protein splicing elements: inteins and exteins--a definition of terms and recommended nomenclature Nucleic Acids Res 1994 22 1125 1127 8165123
Evans TCJ Martin D Kolly R Panne D Sun L Ghosh I Chen L Benner J Liu XQ Xu MQ Protein trans-splicing and cyclization by a naturally split intein from the dnaE gene of Synechocystis species PCC6803 J Biol Chem 2000 275 9091 9094 10734038 10.1074/jbc.275.13.9091
Evans TCJ Benner J Xu MQ The in vitro ligation of bacterially expressed proteins using an intein from Methanobacterium thermoautotrophicum J Biol Chem 1999 274 3923 3926 9933578 10.1074/jbc.274.7.3923
Wood DW Derbyshire V Wu W Chartrain M Belfort M Belfort G Optimized single-step affinity purification with a self-cleaving intein applied to human acidic fibroblast growth factor Biotechnol Prog 2000 16 1055 1063 11101334 10.1021/bp0000858
Slater S Gallaher T Dennis D Production of poly-(3-hydroxybutyrate-co-3-hydroxyvalerate) in a recombinant Escherichia coli strain Appl Environ Microbiol 1992 58 1089 1094 1599234
Wieczorek R Pries A Steinbuchel A Mayer F Analysis of a 24-kilodalton protein associated with the polyhydroxyalkanoic acid granules in Alcaligenes eutrophus J Bacteriol 1995 177 2425 2435 7730274
McPherson DT Morrow C Minehan DS Wu J Hunter E Urry DW Production and purification of a recombinant elastomeric polypeptide, G-(VPGVG)19-VPGV, from Escherichia coli Biotechnol Prog 1992 8 347 352 1368456 10.1021/bp00016a012
Meyer DE Chilkoti A Purification of recombinant proteins by fusion with thermally-responsive polypeptides Nat Biotechnol 1999 17 1112 1115 10545920 10.1038/15100
Ma J Cooney CL Application of vortex flow adsorption technology to intein-mediated recovery of recombinant human alpha1-antitrypsin Biotechnol Prog 2004 20 269 276 14763852 10.1021/bp0341803
Sharma S Zhang A Wang H Harcum SW Chong S Study of protein splicing and intein-mediated peptide bond cleavage under high-cell-density conditions Biotechnol Prog 2003 19 1085 1090 12790686 10.1021/bp034009r
Maehara A Ueda S Nakano H Yamane T Analyses of a polyhydroxyalkanoic acid granule-associated 16-kilodalton protein and its putative regulator in the pha locus of Paracoccus denitrificans Journal of Bacteriology 1999 181 2914 2921 10217786
pMAL Protein Fusion and Purification System: Instructional manual #E8000S. New England Biolabs, Beverly, MA, 2004.
Purification of His*Tag GST Expressed in E. coli. Novagen 2002-2003 Catalog: page 197. Novagen Inc., Madison, WI.
His*Bind Kits: User Protocol TB054 Rev. E 0405. Novagen Inc., Madison, WI, 2005.
IMPACT-CN System: Instructional manual #E6900S. New England Biolabs, Beverly, MA, 2004.
|
16283936
|
PMC1310522
|
CC BY
|
2021-01-04 16:05:47
|
no
|
Microb Cell Fact. 2005 Nov 11; 4:32
|
utf-8
|
Microb Cell Fact
| 2,005 |
10.1186/1475-2859-4-32
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-731628866410.1186/1477-7525-3-73ResearchThe International Classification of Functioning as an explanatory model of health after distal radius fracture: A cohort study Harris Jocelyn E [email protected] Joy C [email protected] James [email protected] School of Rehabilitation Sciences, University of British Columbia, T325-2211 Wesbrook Mall, Vancouver, British Columbia, V6T 2B5, Canada 2 Rehabilitation Research Lab, GF Strong Rehab Centre, 4255 Laurel Street, Vancouver, British Columbia, V5Z 2G9, Canada 3 School of Rehabilitation Sciences, McMaster University, Institute of Applied Health Science, 1400 Main Street West, 4th Floor, Hamilton, Ontario, L8S 1C7, Canada 4 Hand and Upper Limb Centre, St. Joseph's Health Centre, PO Box 5777, London, Ontario, N6A 4L6, Canada2005 16 11 2005 3 73 73 19 7 2005 16 11 2005 Copyright © 2005 Harris et al; licensee BioMed Central Ltd.2005Harris et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Distal radius fractures are common injuries that have an increasing impact on health across the lifespan. The purpose of this study was to identify health impacts in body structure/function, activity, and participation at baseline and follow-up, to determine whether they support the ICF model of health.
Methods
This is a prospective cohort study of 790 individuals who were assessed at 1 week, 3 months, and 1 year post injury. The Patient Rated Wrist Evaluation (PRWE), The Wrist Outcome Measure (WOM), and the Medical Outcome Survey Short-Form (SF-36) were used to measure impairment, activity, participation, and health. Multiple regression was used to develop explanatory models of health outcome.
Results
Regression analysis showed that the PRWE explained between 13% (one week) and 33% (three months) of the SF-36 Physical Component Summary Scores with pain, activities and participation subscales showing dominant effects at different stages of recovery. PRWE scores were less related to Mental Component Summary Scores, 10% (three months) and 8% (one year). Wrist impairment scores were less powerful predictors of health status than the PRWE.
Conclusion
The ICF is an informative model for examining distal radius fracture. Difficulty in the domains of activity and participation were able to explain a significant portion of physical health. Post-fracture rehabilitation and outcome assessments should extend beyond physical impairment to insure comprehensive treatment to individuals with distal radius fracture.
==== Body
Background
In 1980 the WHO [1] published a framework for classifying the consequences of disease. This classification system included the domains of impairment, disability, and handicap where a linear relationship was thought to exist between domains. This framework emphasized the multifaceted nature of health and led to changes in the measurement of health outcomes, specifically, the evaluation of disability, and handicap [2]. With increased application of the model it became apparent that the relationship between the domains was not linear and other relevant contributions to health (e.g., environmental, socio-demographic, and psychological has been ignored).
The WHO updated the framework to reflect emerging understanding of health. In 2001 the International Classification of Functioning, Disability, and Health (ICF) was published [3,4]. It has three main domains, Body Structure/Function, Activity, and Participation, that can be used to classify the impact of health. In this framework the domains interact with each other (not necessarily in a linear manner) and are influenced by both environmental and personal factors [3]. Problem areas within the domains are called impairment, activity limitation, and participation restriction. These terms decrease the negative connotations associated with earlier terminology, i.e., disability and handicap [1]. Recently, studies have linked outcome measures to the ICF domains to better reflect all aspects of health, body function, activity, and participation in musculoskeletal conditions [5-13]. With the emergence of this broader model of health, clinical research has started to focus on how ICF might explain health outcomes across a spectrum of health conditions [5-12].
Distal radius fractures are the most common fracture [14]. A 17% increase in incidence rate has been noted over the past few decades [15]. In the United Kingdom 71, 000 persons will sustain a distal radius fracture each year with an incidence rate of 36.8/10,000 for women and 8.9/10,000 for men [16]. Though distal radius fractures are found throughout the life span, women demonstrate an increase in incidence rate from age 50–70 (while men do not) which has been attributed to decreased bone mineral density [16-18].
Usually the majority of recovery from a distal radius fracture occurs within six months post fracture [19,20]. Until recently, descriptions of the clinical outcomes of distal radius fracture have focused on impairment, e.g., radiographic findings, range of motion, and strength. What are missing from these studies are outcome measures that evaluate an individual's ability to perform day-to-day tasks and engage in meaningful activities and roles. Recently studies have included broader outcome measures that reflect performance in self-care, household, work, recreational, and social activities. These studies show that despite the fact that the majority of individuals receive rehabilitation services, residual difficulty in work, sport, and leisure activities are reported [13,19,20].
Studies addressing quality of life in individuals following distal radius fracture are few. Two cross-sectional studies examined the relationship between radiographic findings and the Medical Outcomes Survey Short Form SF-36 [21] and SF-12 [22] (in long-term follow-up). Both studies found that radiographic findings did not correlate with either the SF-36 or SF-12, and that patients' post-rehabilitation scores were similar to those of the general population. However, in the study by Fernandez and colleagues [21], men between the ages of 35–44 (physical component score only) demonstrated a significant difference in SF-36 scores from their age-matched general population norm. It was suggested that this group represents a segment of the population that has greater functional demand from both work and family life and thus the health impact of mild residual physical impairment was greater.
In longitudinal studies that evaluated recovery from a distal radius fracture it is clear that health is affected in the early post-fracture period and that there is substantial recovery. MacDermid and colleagues [20] reported SF-36 scores that improved from the early post-fracture evaluation to a one-year evaluation for the Physical Component Summary Score (PCSS) (from 37–48) but found that the Mental Health Summary Component Score (MCSS) remained within normal range throughout recovery (from 51–53).
One study evaluated the adjustment to distal radius fracture over a three month time frame [23]. This study used scales that measure physical, emotional, social role function, and meaning of injury, i.e., the SF-36, the Enforced Social Dependency Scale, and the Meaning of Illness scale. Findings suggested that as time from fracture increases, scores in physical, emotional, and social role function reflect adjustment to injury [23]. The authors suggested that during the early stages of recovery significant issues in roles, physical function, and adjustment to injury are evident and should not be neglected during rehabilitation. Overall studies suggest that the impact of distal radius fracture on physical and/or mental health abates by three months post injury and occurs to a greater extent within the physical health domain as compared to mental health.
Although previous work has suggested that distal radius fracture has an impact on overall health, these studies have not focused on the extent to which health effects fit the ICH health model. An understanding of how the model applies to this common injury would assist those involved in planning or providing health services to clients with these injuries. The purposes of this study were 1) to determine whether the ICF framework serves as an explanatory model for distal radius fracture and 2) to determine the impact of impairment, activity limitation, and participation restriction on physical and mental health after distal radius fracture.
Methods
This study used a prospective cohort design. Patients with distal radius fracture attending the Hand and Upper Limb Centre for primary care were identified by clinic lists and attending physicians. All identified patients were enrolled in the outcome evaluation process, unless they were unable to participate because of incompetence. Patients who failed or were unable to comply with their scheduled appointments were contacted by phone to determine whether they could reschedule their appointments. The university ethics review board approved the use of this clinical outcomes database for this study.
Patients completed standardized testing at one week, and at three and twelve months post fracture. Demographic data was collected at the initial one-week post injury visit. The ICF was used as a conceptual model to frame the health outcome of distal radius fracture. We have outlined the model, adapted from the WHO, in Figure 1.
Figure 1 International Classification of Functioning (ICF) model applied to distal radius fracture. (Adapted from the World Health Organization, International Classification of Functioning, Disability, and Health training materials, Geneva, 2002). .
Outcome measures
All patients completed the Patient Rated Wrist Evaluation (PRWE) [24-26], and the SF-36 [27-29] at all 3 time points and the Wrist Outcome Measure (WOM) [30] at 3 and 12 months. A research assistant verbally administered the questionnaires (PRWE and SF-36) to patients who were unable to read or write. When patients were unable to understand English sufficiently to answer, the questionnaires were translated with the assistance of a bilingual family member or friend. All questionnaires were administered and scored according to the author's instructions. An independent research assistant administered the Wrist Outcome Measure.
The Wrist Outcome Measure is a composite impairment scale with components that reflect range of motion (ROM), grip strength, and dexterity [30]. Range of motion measures were measured on the N-K computerized hand evaluation system. A total score out of 30, scored by extent of attainment of normative values was given. Six wrist motions (extension/flexion [31], radial/ulnar deviation, pronation/supination [32]) and a gross finger flexion measure was summated. Grip strength was performed using the NK Digit-grip device. The standard protocol recommended by the American Society of Hand Therapists was followed [33]. High reliability has been demonstrated for this protocol and test instrument [34]. A grip strength (score out of 40) was determined as a ratio of the uninjured hand with the injured hand and adjusting for dominance. Dexterity was measured using the checkers subtest of the Jebson's Hand Function Test (score/15) [35]. A total Wrist Outcome Measure score out of 85 was devised from these scales. Further background and discussion on the development of an impairment rating score can be found elsewhere [30].
Patient Rated Wrist Evaluation is a 15-item questionnaire that equally rates wrist-related pain and disability in functional activities (see BMC reference for complete form) [19,24,25,36]. Scoring is done on an eleven-point scale (0–10) with zero being no issues or pain and 10 being unable to do or severe pain. There are five questions that require the individual to rate their pain doing activities such as at rest, repeated motion, and lifting. Functional items are divided into two categories, specific and usual activities. There are six specific tasks such as turning a doorknob, cutting meat, fastening a button, and four usual activity categories, self-care, work, household duties, and recreation. The PRWE can be divided into three sub-scales, pain, specific activities, and usual activities. The total of the combined scales is 100 (50 from pain, 60 from specific, and 40 from usual). The psychometric properties of this scale are excellent [19,24,25,36] and the patterns of recovery following a fracture have been described using this scale [19,20].
The SF-36 is a widely used health outcome measure. It is comprised of eight scales and two summary scores [27-29,37]. There is a large database of normative data available through the Medical Outcomes Trust. The scale has eight sub-scales that portray various domains of health: physical function, physical role, bodily pain, vitality, general health, emotional role, mental health, and social function. These sub-scales are scored out of a maximum score of 100 (higher is better). The physical and mental health component summary scores represent the two main dimensions of health. These scores are calculated in a three-step process which involves weighting, transforming and aggregating the subscale scores to compute summary scores scaled to a US population which represent these two distinct domains of health (US population mean = 50). While the ICF model portrays health as a single concept with multiple, interacting contributors, the SF-36 separates physical and mental health. Since we expected largely a physical effect of wrist fracture and because the SF-36 has been shown to be preferable to other general health measures for musculoskeletal disorders [38,39] we choose it to represent health status. While lesser effects were expected in mental health we decided to include both the Mental and Physical Component Summary Scores as outcomes to determine the relative effects on both domains of health providing a more complete picture of overall health.
Data analysis
Descriptive statistics were calculated for the dependent variables (SF-36 physical and mental health component summary scores) and independent variables (WOM total score, PRWE sub-scales pain, specific activities, and usual activities). All data was inspected for assumption violation by using histograms, box, and scatter-plots. Missing values were replaced using linear extrapolation. Missing values accounted for less than 5% of data points.
Univariate analysis was completed to determine the relationship between variables of interest and outcome variables. Pearson's Correlation Coefficient was used to determine the relationship between SF-36 physical and mental health summary scores and the PRWE pain, specific, and usual activity scales, and the WOM total score. Correlation was determined at time one (one week post injury, time two (three months post injury), and time three (twelve months post injury).
Multivariate analysis was used to determine the explanatory model for distal radius fracture health outcome at time one (one week post), two (three months post), and three (twelve months post). Multiple regression equations were calculated using the SF-36 physical and mental health summary scores as the dependent variables, and patient characteristics, PRWE sub-scales and the WOM as the independent variables as described in Figure 1. The WOM was measured at three and twelve months but not at one week. It was felt that the variables of sex and age are known to be related to health, so we controlled for age and sex by blocked entry of these variables and then continued with stepwise entry of the independent variables. Six stepwise regression models were built. Data was inspected for assumption violation by examining box and scatter-plots of residuals against explanatory variables from each model. Influential data points were examined using Cook's distance. The F to enter was 0.05 and the F to remove was 0.10. Statistical significance was set at 0.05 for all outcomes. All statistics was performed using SPSS 13.
Results
Sample characteristics
There was a total of 790 persons, mean age of 51.4 (SD = 17.6, age range 18–91) in this study. The majority of the people in the study were female (68%). Descriptive characteristics of the sample can be found in Table 1. Summary scores for the outcome measures at each time frame can be found in Table 2. The mean score of the outcome measures improve at each follow up time period. At one week post injury scores demonstrate moderate to severe activity limitation and participation restriction and the one-year measures demonstrate little activity limitation or participation restriction. The PRWE specific activity sub-scale showed the most change over time, from 51.3 (severe limitation) to 6.3 (minimal limitation).
Table 1 Sample characteristics
Variable Description
Sex Male = 251
Female = 539
Dominant Hand Right = 90%
Left = 10%
Wrist Injured Right = 45%
Left = 49%
Mechanism of Fracture Fall on ice = 18%
Other fall = 66%
Other = 10%
Energy of Fracture* Low = 69%
Medium = 19%
High = 6%
Highest Level of Education Finished high school = 26%
Finished college = 18%
Finished university = 8%
Finished graduate school = 4%
Occupation at Injury† Retired = 27%
Service = 13%
Professional = 12%
Occupational Demand§/P > Low = 57%
Moderate = 24%
High = 19%
Had Physiotherapy 83%
* Low = fall from a standing position, Medium = fall from a height, High = trauma
† Top three occupations
§Self-report of how much they used their hand at work; Low = low force, low repetition, Moderate = frequent repetition, intermittent force, High = high force, constant repetition
Table 2 Descriptive statistics for outcome measures at one week, three months, and one year post injury.
Variable Time 1 Mean (SD) Time 2 Mean (SD) Time 3 Mean (SD)
Wrist Outcome Measure (/85; 85 = best) N/A 59.6(8.9) 73.9(7.4)
PRWE pain scale (/50; 50 = worst) 30.2(11.6) 17.0(10.4) 8.1(9.5)
PRWE specific scale (/60; 60 = worst) 51.3(14.1) 19.4(15.0) 6.3(10.3)
PRWE usual scale (/40; 40 = worst) 26.3(11.9) 11.4(12.5) 5.6(12.6)
SF-36 physical health (US norm 50) 37.2(8.7) 43.7(8.9) 49.0(8.7)
SF-36 mental health (US norm 50) 49.8(11.2) 51.5(9.8) 54.8(7.5)
Univariate analysis
Results from the univariate analysis can found in Table 3. All PRWE sub-scales were correlated with SF-36 physical health at one week, three, and twelve months post injury. At one week the sub-scale of usual activity demonstrated the highest correlation (r = -0.31, p = 0.01), at three and twelve months it was specific activity (r = -0.53, p = 0.01, r = -0.52, p = 0.01). Only usual activity was correlated with mental health (r = 0.09, p = 0.05) at one-week post injury. However, at three months all PRWE sub-scales were significantly correlated to physical health and at twelve months all independent variables (PRWE and WOM) were significantly correlated.
Table 3 Correlation results between outcome measures at one week, three months, and one year post injury.
Variable PRWE pain PRWE specific PRWE usual SF-36 physical health SF-36 mental health
Wrist Outcome Measure
Time 2 -0.27** -0.35** -0.20** 0.21** 0.07
Time 3 -0.43** -0.46** -0.44** 0.32** 0.14**
PRWE pain
Time 1 0.46** 0.44** -0.27** 0.05
Time 2 0.75** 0.53** -0.50** -0.27**
Time 3 0.79** 0.34** -0.50** -0.34**
PRWE specific
Time 1 0.48** -0.29** 0.01
Time 2 0.57** -0.42** -0.23**
Time 3 0.34** -0.52** -0.30**
PRWE usual
Time 1 -0.31** 0.09*
Time 2 -0.42** -0.26**
Time 3 -0.16** -0.10**
SF-36 physical health
Time 1 -0.002
Time 2 0.10**
Time 3 0.13**
* p = 0.05, **p = 0.01
† Outcome variables are correlated with each variable at their respective time periods, i.e., time 1 with time 1, time 2 with time 2, etc.
Multivariate analysis
All regression results can be found in Tables 4, 5, and 6. The result from the forward stepwise regression model for SF-36 physical health at one-week post injury yielded a weakly predictive model where all PRWE sub-scales were retained with an R2 = 0.13, p < 0.0001 for the full model (Table 4). Usual activity was most predictive and accounted for R2 = 0.10 of the model. For SF-36 mental health, minimal effects were observed with only PRWE usual activity retained within the model R2 = 0.01, p = 0.04.
Table 4 Multiple regression results for SF-36 Physical Health one week post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.13, p < 0.0001
PRWE usual* 0.10 -0.19 0.0001
PRWE specific 0.12 -0.15 0.0001
PRWE pain 0.13 -0.12 0.003
* Independent variable scores are from one-week post injury
Table 5 Multiple regression results for SF-36 Physical and Mental Health three months post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.33, p < 0.0001*
PRWE specific† 0.28 -0.27 0.0001
PRWE pain 0.31 -0.23 0.0001
PRWE usual 0.33 -0.15 0.0001
SF-36 Mental Health Summary Scale, Total Model R2 = 0.10, p < 0.0001§
PRWE pain 0.08 -0.18 0.0001
PRWE usual 0.10 -0.17 0.0001
* Excluded variable Wrist Outcome Measure
† Independent variable scores are from three months post injury
§Excluded variables Wrist Outcome Measure, PRWE specific
Table 6 Multiple regression results for SF-36 Physical and Mental Health one year post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.28, p < 0.0001*
PRWE specific† 0.25 -0.29 0.0001
PRWE usual 0.28 0.28 0.0001
* Excluded variables PRWE pain and Wrist Outcome Measure
† Independent variable score are from one-year post injury
The regression results for three-month post injury models yielded greater prediction form included variables (Table 5). All PRWE sub-scales were retained in the SF-36 physical health model with a total R2 = 0.33, p < 0.0001. PRWE specific activity was most predictive and accounted for R2 = 0.28 of the total model. Again at the three-month model, SF-36 mental health was less explained by wrist scores, but the independent variables PRWE pain and usual activity were retained, R2 = 0.10, p < 0.0001. PRWE pain accounted for the major effect, R2 = 0.08 of the total model.
The regression results for one-year post injury can be found in Table 6. At one year post injury the regression model for SF-36 physical health showed that PRWE specific and usual activity accounted for a total R2 = 0.28, p < 0.0001, with specific activity accounting for the majority of the model, R2 = 0.25. The regression model for SF-36 mental health one year post injury showed that only PRWE pain was retained and accounted for a total R2 = 0.08, p < 0.001.
Discussion
This study determined that the ICF framework is supported when evaluating the impact of distal radius fracture on health because impairment, activity limitations and participation restrictions, individually and in combinations, were related to self-reported physical health status as measured on the SF-36. Statistically significant models for SF-36 physical and mental health were found. The models for SF-36 physical health were strong with the PRWE accounting for 13% (one week), 33% (three months), and 28% (one year) of the variance. Models for SF-36 mental health demonstrated weak relationships with the PRWE accounting for 1% (one week), 10% (three months), and 8% (one year) of the variance. This study confirms previous work that a distal radius fracture mainly affects the physical domains of health, although it does suggest that pain levels and mental health are also related.
The ICF framework is advantageous as the inclusion of aspects relating to the injury, the individual and the environment provide a broader view of how health interventions might be undertaken. In the past the focus of outcome for distal radius fracture has been impairment based (e.g., radiographic data, strength, and range of motion). However, studies have shown that impairment is not necessarily the best method to measure outcome as it does not always reflect activity and or participation restrictions [13,22,40]. We measured health by self-report allowing us to capture early and late health effects. The PRWE sub-scales were able to explain a significant portion of the SF-36 physical health score at all time periods. When examining the areas measured by the PRWE sub-scales this indicates that problems in areas such as pain, dexterity, lifting, work, household duties, recreation, and self-care after fracture do contribute to overall physical health. However, there were differences in the magnitude of the models and the most prevalent sub-scale between the time periods suggesting that impairment, activity, and participation have different health impacts at different time points in recovery from distal radius fracture.
The regression equations at one-week explained relatively little of the health impact of the fracture. This may have been because PRWE scores are consistently very poor at this point across all patients [19], and variations in health status were not well reflected at this point. For example, many patients would be immobilized providing a common restriction on specific activities like lifting or getting up from a chair. The PRWE usual activity focused on ability to do usual activities and was significant, although accounting for only 10% of the variance of the physical health model (R2 = 0.13) at one-week post injury. This sub-scale evaluates perform their usual self-care, work, household duties, and recreational activities (i.e., participation). This suggests that there is more variability in ability to participate in usual activity, than is observed on the specific activities subscale and these variations impacted on health status. This is in agreement with qualitative studies where clients emphasized the impact of distal radius fracture on work and household activities [41,42]. It is worth noting that issues with pain and inability to do specific activities were not unimportant at this time as they were rated as being at high levels of limitation. Given that participation in usual activity was related to early health status it might be worthwhile to provide some focus on methods to adapt to limitations in the early stage of fracture treatment. Individuals might benefit from educational materials that outline expected activity limitations and possible adaptations to maximize their ability to perform common tasks of daily life.
SF-36 physical health scores at three months and one-year post injury were explained to a greater extent by variations in the PRWE. The PRWE sub-scale of specific activity explained between 25% (model R2 = 0.28) and 28% (model R2 = 0.33) of the variance, replacing usual activity as the prominent variable. At three months post injury the cast has been removed and rehabilitation is underway and patients may be variable in their inability to do specific tasks that require pain-free wrist motion or strength. Self-reported wrist/hand activity limitations and participation restrictions explained a significant portion of overall physical health scores indicating that the wrist injury has a substantial impact on overall health in a manner that is consistent with the view of health portrayed in ICF. Composite wrist impairment rating had minimal additional influence on the models accounting for between 1% and 3% additional variance in health. This is in agreement with the studies that found that impairment was not a good indicator of function [13,21,22,30,40]. Physical impairment contributes to activity limitation, however, there is not a direct relationship.
Though the PRWE sub-scales accounted for a significant portion of the SF-36 physical health score (13%-33%), there are obviously other influencing factors that remain unexplained. We were unable to address a broad spectrum of potentially useful variables given database limitations. More complex models are needed to explore the remainder of the variance and additional concepts such as workplace environment; rehabilitation, surgery, socio-economic status, etc., should be included.
Physical and mental health domains have been seen as distinct domains of health according to SF-36 development and validation. This was also true in this study where the correlation between the physical and mental component summary scores was very low (r<0.10). Neither physical impairments, activities limitations nor participation restrictions were strongly predictive of SF-36 mental health scores. This is consistent with the view that distal radius fracture primarily affects physical health. The PRWE pain sub-scale accounted for slight variation in SF-36 mental health scores. Pain however, has not been shown to significantly impact recovery of distal radius fracture [43,44]. In fact in a study by Bialocerkowski [42], clients with wrist disorders were asked to explore difficulties post injury and pain was not mentioned. Instead issues with household duties, work demands, recreation, and fine motor skills were identified. Because regression reveals associations, it is not clear whether higher pain lowers mental health, or if those with poorer mental health experience more pain.
In describing rehabilitation following distal radius fracture, the importance of a staged approach has been suggested [45-47]. This study would support a staged approach as the mediators of physical health vary over time. In the early phases where immobilization and fracture healing limit motion and activity, the ability to perform usual activities is important. In addition to the customary attention to pain management, rehabilitation should also include compensatory strategies or aids as required to assist individuals complete their usual activities. In the rehabilitative phase where ability to perform specific tasks plays a larger role, remediation of impairment, incorporation of client driven goals and activity-based rehabilitation [48] are needed. Finally, the importance of participation is highlighted in these models and suggests that participation in usual self-care, household, work, and recreation, must be maximized to restore physical health. It has been demonstrated that even after adjusting for age and comorbidity patients more than 65 years of age who sustain a wrist fracture have a 57% 7-year survival rate, as compared to 71% for the comparative US population [49]. One possible contributor to this problem can be reduced participation in an active lifestyle that increases risk for additional health problems. Addressing participation during fracture rehabilitation may have short-term and longer-term health benefits.
Conclusion
The ICF model is useful in framing the health effects of a distal radius fracture that has implications for optimal management of distal radius fractures. Self-reported health measures should provide insight into the impairment, activity limitations and participation restrictions that result from a distal radius fracture. These aspects of health can be addressed at different phases of fracture management and rehabilitation to provide optimal physical health recovery.
Authors' contributions
JM formed the original study design and clinical database, obtained the ethics approval and obtained database funding. JH and JM developed the research question and planned statistical analyses. JH conducted statistical analyses and drafted the manuscript. JR enrolled, treated and evaluated patients in this study and contributed to the design of evaluation procedures, study equipment and personnel funding. All authors approved the final study protocol, contributed to interpretation of the study results and participated in revisions of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This project was supported by the New Investigator Award, Canadian Institute of Health Research (JCM), and Doctoral Research Award, Canadian Institute of Health Research, and a Canadian Institute of Health Research Training Fellowship in Quality of Life Research in Rehabilitation, sponsored by the Institute of Musculoskeletal Health and Arthritis (JEH).
==== Refs
Organization WH International Classification of Impairments, Disabilities and Handicaps. A Manual of Classification Relating to the Consequences of Disease. 1980 Geneva, World Health Organization
Garratt A Schmidt L Mackintosh A Fitzpatrick R Quality of life measurement: bibliographic study of patient assessed health outcome measures BMJ 2002 324 1417 12065262 10.1136/bmj.324.7351.1417
International Classification of Functioning, Disability and Health 2003 World Health Organization
Grimby G Smedby B ICF approved as the successor of ICIDH J Rehabil Med 2001 33 193 194 11585148 10.1080/165019701750419545
Cieza A Brockow T Ewert T Amman E Kollerits B Chatterji S Ustun TB Stucki G Linking health-status measurements to the international classification of functioning, disability and health J Rehabil Med 2002 34 205 210 12392234 10.1080/165019702760279189
Tremayne A Taylor N McBurney H Baskus K Correlation of impairment and activity limitation after wrist fracture Physiother Res Int 2002 7 90 99 12109238
MacDermid JC Measurement of health outcomes following tendon and nerve repair J Hand Ther 2005 18 297 312 15891987
Kjeken I Dagfinrud H Slatkowsky-Christensen B Mowinckel P Uhlig T Kvien TK Finset A Activity limitations and participation restrictions in women with hand osteoarthritis: Patients descriptions, and associations between dimensions of functioning Ann Rheum Dis 2005 15829571
Pap G Angst F Herren D Schwyzer HK Simmen BR Evaluation of wrist and hand handicap and postoperative outcome in rheumatoid arthritis Hand Clin 2003 19 471 481 12945645 10.1016/S0749-0712(03)00004-0
Stucki G Cieza A Geyh S Battistella L Lloyd J Symmons D Kostanjsek N Schouten J ICF Core Sets for rheumatoid arthritis J Rehabil Med 2004 87 93 15370754 10.1080/16501960410015470
Stucki G Cieza A The International Classification of Functioning, Disability and Health (ICF) Core Sets for rheumatoid arthritis: a way to specify functioning Ann Rheum Dis 2004 63 Suppl 2 ii40 ii45 15479870 10.1136/ard.2004.028233
Weigl M Cieza A Harder M Geyh S Amann E Kostanjsek N Stucki G Linking osteoarthritis-specific health-status measures to the International Classification of Functioning, Disability, and Health (ICF) Osteoarthritis Cartilage 2003 11 519 523 12814615 10.1016/S1063-4584(03)00086-4
Dekkers M Soballe K Activities and impairments in the early stage of rehabilitation after Colles' fracture Disabil Rehabil 2004 26 662 668 15204505 10.1080/09638280410001683173
Larsen CF Lauritsen J Epidemiology of acute wrist trauma Int J Epidemiol 1993 22 911 916 8282472
Melton LJIII Amadio PC Crowson CS O'Fallon WM Long-term trends in the incidence of distal forearm fractures Osteoporos Int 1998 8 341 348 10024904 10.1007/s001980050073
O'Neill TW Cooper C Finn JD Lunt M Purdie D Reid DM Rowe R Woolf AD Wallace WA Incidence of distal forearm fracture in British men and women Osteoporos Int 2001 12 555 558 11527052 10.1007/s001980170076
Singer BR Mclauchlan GJ Robinson CM Christie J Epidemiology of fractures in 15000 adults J Bone Joint Surg Br 1998 80B 243 248 9546453 10.1302/0301-620X.80B2.7762
Thompson PW Taylor J Dawson A The annual incidence and seasonal variation of fractures of the distal radius in men and women over 25 years in Dorset, UK Injury 2004 35 462 466 15081322 10.1016/S0020-1383(03)00117-7
MacDermid JC Roth JH Richards RS Pain and disability reported in the year following a distal radius fracture: A cohort study BMC Musculoskelet Disord 2003 4 24 14588078 10.1186/1471-2474-4-24
MacDermid JC Richards RS Roth JH Distal radius fracture: a prospective outcome study of 275 patients J Hand Ther 2001 14 154 169 11382255
Fernandez JJ Gruen GS Herndon JH Outcome of distal radius fractures using the short form 36 health survey Clin Orthop 1997 36 41 9269152 10.1097/00003086-199708000-00007
Anzarut A Johnson JA Rowe BH Lambert RG Blitz S Majumdar SR Radiologic and patient-reported functional outcomes in an elderly cohort with conservatively treated distal radius fractures J Hand Surg [Am] 2004 29 1121 1127 15576226 10.1016/j.jhsa.2004.07.002
Schappler Morris N Distal radius frcatures in adults: Self-reported physical functioning, role functioning and meaning of injury Orthopedic Nursing 2005 19 37 48
MacDermid JC Turgeon T Richards RS Beadle M Roth JH Patient rating of wrist pain and disability: a reliable and valid measurement tool J Orthop Trauma 1998 12 577 586 9840793 10.1097/00005131-199811000-00009
MacDermid JC Development of a scale for patient rating of wrist pain and disability J Hand Ther 1996 9 178 183 8784681
MacDermid JC Tottenham V Responsiveness of the Disability of the Arm, Shoulder, and Hand (DASH) and Patient-Rated Wrist/Hand Evaluation (PRWHE) in evaluating change after hand therapy J Hand Ther 2004 17 18 23 14770134
Ware JEJ Sherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Ware JE Snow KS Kosinski M Gandek B SF-36 Health Survey Manual and Interpretation Guide 1993 Boston, The Health Institute, New England Medical Center
Ware JE Kosinski M Keller SD SF-36 Physical and Mental Health Summary Scales: A User's Manual 1994 3rd Boston, Mass., The Health Institute, New England Medical Center 1.1 10.12
MacDermid JC Donner A Richards RS Roth JH Patient versus injury factors as predictors of pain and disability six months after a distal radius fracture J Clin Epidemiol 2002 55 849 854 12393071 10.1016/S0895-4356(02)00445-6
LaStayo PC Wheeler DL Reliability of passive wrist flexion and extension goniometric measurements: a multicenter study Phys Ther 1994 74 162 174 8290621
Armstrong AD MacDermid JC Chinchalkar S Stevens RS King GJ Reliability of range-of-motion measurement in the elbow and forearm J Shoulder Elbow Surg 1998 7 573 580 9883416 10.1016/S1058-2746(98)90003-9
Fess EE Casanova JS Grip strength Clinical Assessment Recommendations 1992 5 2nd Chicago, American Society of Hand Therapists 41 46
MacDermid JC Alyafi T Richards RS Test-retest reliability of static and endurance grip strength tests performed on the Jamar and NK Physiother Can 2001 53 48 54
Jebson RH Taylor N Trieschmann RB Trotter MJ Howard LA An objective and standardized test of hand function. Arch Phys Med Rehabil 1969 50 311 319 5788487
MacDermid JC Richards RS Donner A Bellamy N Roth JH Responsiveness of the short form-36, disability of the arm, shoulder, and hand questionnaire, patient-rated wrist evaluation, and physical impairment measurements in evaluating recovery after a distal radius fracture J Hand Surg [Am ] 2000 25A 330 340 10.1053/jhsu.2000.jhsu25a0330
Ware JEJ Conceptualization and measurement of health-related quality of life: comments on an evolving field Arch Phys Med Rehabil 2003 84 S43 S51 12692771 10.1053/apmr.2003.50246
Beaton DE Bombardier C Hogg-Johnson SA Measuring health in injured workers: a cross-sectional comparison of five generic health status instruments in workers with musculoskeletal injuries Am J Ind Med 1996 29 618 631 8773722 10.1002/(SICI)1097-0274(199606)29:6<618::AID-AJIM6>3.0.CO;2-F
Stucki G Liang MH Phillips C Katz JN The Short Form-36 is preferable to the SIP as a generic health status measure in patients undergoing elective total hip arthroplasty Arth Care Res 1995 8 174 181 7654802
Karnezis IA Fragkiadakis EG Association between objective clinical variables and patient-rated disability of the wrist J Bone Joint Surg Br 2002 84 967 970 12358387 10.1302/0301-620X.84B7.12673
Bialocerkowski AE Grimmer KA Compensatory mechanism use during the first six months following distal radius fracture International Journal of Therapy and Rehabilitation 2004 11
Bialocerkowski AE Difficulties associated with wrist disorders--a qualitative study Clin Rehabil 2002 16 429 440 12061478 10.1191/0269215502cr516oa
Young BT Rayan GM Outcome following nonoperative treatment of displaced distal radius fractures in low-demand patients older than 60 years J Hand Surg [Am] 2000 25 19 28 10642469 10.1053/jhsu.2000.jhsu025a0019
Fujii K Henmi T Kanematsu Y Mishiro T Sakai T Terai T Fractures of the distal end of radius in elderly patients: a comparative study of anatomical and functional results J Orthop Surg (Hong Kong ) 2002 10 9 15 12401915
Distal Radius Fracture: Practice Guidelines 1998 Chicago, Il, American Society of Hand Therapists
Laseter GF Carter PR Management of distal radius fractures J Hand Ther 1996 9 114 128 8784675
MacDermid JC Hand therapy management of intra-artiocular fractures with open reduction and pi plate fixation: A therapist's perspective Techniques in Hand and Upper Extremity Surgery 2004 8 219 223 16518095
Maciel JS Taylor NF McIlveen C A randomised clinical trial of activity-focussed physiotherapy on patients with distal radius fractures Arch Orthop Trauma Surg 2005 125 515 520 16136342 10.1007/s00402-005-0037-x
Rozental TD Branas CC Bozentka DJ Beredjiklian PK Survival among elderly patients after fractures of the distal radius J Hand Surg [Am ] 2002 27 948 952 12457343 10.1053/jhsu.2002.36995
|
16288664
|
PMC1310523
|
CC BY
|
2021-01-04 16:38:13
|
no
|
Health Qual Life Outcomes. 2005 Nov 16; 3:73
|
utf-8
|
Health Qual Life Outcomes
| 2,005 |
10.1186/1477-7525-3-73
|
oa_comm
|
==== Front
Health Qual Life OutcomesHealth and Quality of Life Outcomes1477-7525BioMed Central London 1477-7525-3-731628866410.1186/1477-7525-3-73ResearchThe International Classification of Functioning as an explanatory model of health after distal radius fracture: A cohort study Harris Jocelyn E [email protected] Joy C [email protected] James [email protected] School of Rehabilitation Sciences, University of British Columbia, T325-2211 Wesbrook Mall, Vancouver, British Columbia, V6T 2B5, Canada 2 Rehabilitation Research Lab, GF Strong Rehab Centre, 4255 Laurel Street, Vancouver, British Columbia, V5Z 2G9, Canada 3 School of Rehabilitation Sciences, McMaster University, Institute of Applied Health Science, 1400 Main Street West, 4th Floor, Hamilton, Ontario, L8S 1C7, Canada 4 Hand and Upper Limb Centre, St. Joseph's Health Centre, PO Box 5777, London, Ontario, N6A 4L6, Canada2005 16 11 2005 3 73 73 19 7 2005 16 11 2005 Copyright © 2005 Harris et al; licensee BioMed Central Ltd.2005Harris et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Distal radius fractures are common injuries that have an increasing impact on health across the lifespan. The purpose of this study was to identify health impacts in body structure/function, activity, and participation at baseline and follow-up, to determine whether they support the ICF model of health.
Methods
This is a prospective cohort study of 790 individuals who were assessed at 1 week, 3 months, and 1 year post injury. The Patient Rated Wrist Evaluation (PRWE), The Wrist Outcome Measure (WOM), and the Medical Outcome Survey Short-Form (SF-36) were used to measure impairment, activity, participation, and health. Multiple regression was used to develop explanatory models of health outcome.
Results
Regression analysis showed that the PRWE explained between 13% (one week) and 33% (three months) of the SF-36 Physical Component Summary Scores with pain, activities and participation subscales showing dominant effects at different stages of recovery. PRWE scores were less related to Mental Component Summary Scores, 10% (three months) and 8% (one year). Wrist impairment scores were less powerful predictors of health status than the PRWE.
Conclusion
The ICF is an informative model for examining distal radius fracture. Difficulty in the domains of activity and participation were able to explain a significant portion of physical health. Post-fracture rehabilitation and outcome assessments should extend beyond physical impairment to insure comprehensive treatment to individuals with distal radius fracture.
==== Body
Background
In 1980 the WHO [1] published a framework for classifying the consequences of disease. This classification system included the domains of impairment, disability, and handicap where a linear relationship was thought to exist between domains. This framework emphasized the multifaceted nature of health and led to changes in the measurement of health outcomes, specifically, the evaluation of disability, and handicap [2]. With increased application of the model it became apparent that the relationship between the domains was not linear and other relevant contributions to health (e.g., environmental, socio-demographic, and psychological has been ignored).
The WHO updated the framework to reflect emerging understanding of health. In 2001 the International Classification of Functioning, Disability, and Health (ICF) was published [3,4]. It has three main domains, Body Structure/Function, Activity, and Participation, that can be used to classify the impact of health. In this framework the domains interact with each other (not necessarily in a linear manner) and are influenced by both environmental and personal factors [3]. Problem areas within the domains are called impairment, activity limitation, and participation restriction. These terms decrease the negative connotations associated with earlier terminology, i.e., disability and handicap [1]. Recently, studies have linked outcome measures to the ICF domains to better reflect all aspects of health, body function, activity, and participation in musculoskeletal conditions [5-13]. With the emergence of this broader model of health, clinical research has started to focus on how ICF might explain health outcomes across a spectrum of health conditions [5-12].
Distal radius fractures are the most common fracture [14]. A 17% increase in incidence rate has been noted over the past few decades [15]. In the United Kingdom 71, 000 persons will sustain a distal radius fracture each year with an incidence rate of 36.8/10,000 for women and 8.9/10,000 for men [16]. Though distal radius fractures are found throughout the life span, women demonstrate an increase in incidence rate from age 50–70 (while men do not) which has been attributed to decreased bone mineral density [16-18].
Usually the majority of recovery from a distal radius fracture occurs within six months post fracture [19,20]. Until recently, descriptions of the clinical outcomes of distal radius fracture have focused on impairment, e.g., radiographic findings, range of motion, and strength. What are missing from these studies are outcome measures that evaluate an individual's ability to perform day-to-day tasks and engage in meaningful activities and roles. Recently studies have included broader outcome measures that reflect performance in self-care, household, work, recreational, and social activities. These studies show that despite the fact that the majority of individuals receive rehabilitation services, residual difficulty in work, sport, and leisure activities are reported [13,19,20].
Studies addressing quality of life in individuals following distal radius fracture are few. Two cross-sectional studies examined the relationship between radiographic findings and the Medical Outcomes Survey Short Form SF-36 [21] and SF-12 [22] (in long-term follow-up). Both studies found that radiographic findings did not correlate with either the SF-36 or SF-12, and that patients' post-rehabilitation scores were similar to those of the general population. However, in the study by Fernandez and colleagues [21], men between the ages of 35–44 (physical component score only) demonstrated a significant difference in SF-36 scores from their age-matched general population norm. It was suggested that this group represents a segment of the population that has greater functional demand from both work and family life and thus the health impact of mild residual physical impairment was greater.
In longitudinal studies that evaluated recovery from a distal radius fracture it is clear that health is affected in the early post-fracture period and that there is substantial recovery. MacDermid and colleagues [20] reported SF-36 scores that improved from the early post-fracture evaluation to a one-year evaluation for the Physical Component Summary Score (PCSS) (from 37–48) but found that the Mental Health Summary Component Score (MCSS) remained within normal range throughout recovery (from 51–53).
One study evaluated the adjustment to distal radius fracture over a three month time frame [23]. This study used scales that measure physical, emotional, social role function, and meaning of injury, i.e., the SF-36, the Enforced Social Dependency Scale, and the Meaning of Illness scale. Findings suggested that as time from fracture increases, scores in physical, emotional, and social role function reflect adjustment to injury [23]. The authors suggested that during the early stages of recovery significant issues in roles, physical function, and adjustment to injury are evident and should not be neglected during rehabilitation. Overall studies suggest that the impact of distal radius fracture on physical and/or mental health abates by three months post injury and occurs to a greater extent within the physical health domain as compared to mental health.
Although previous work has suggested that distal radius fracture has an impact on overall health, these studies have not focused on the extent to which health effects fit the ICH health model. An understanding of how the model applies to this common injury would assist those involved in planning or providing health services to clients with these injuries. The purposes of this study were 1) to determine whether the ICF framework serves as an explanatory model for distal radius fracture and 2) to determine the impact of impairment, activity limitation, and participation restriction on physical and mental health after distal radius fracture.
Methods
This study used a prospective cohort design. Patients with distal radius fracture attending the Hand and Upper Limb Centre for primary care were identified by clinic lists and attending physicians. All identified patients were enrolled in the outcome evaluation process, unless they were unable to participate because of incompetence. Patients who failed or were unable to comply with their scheduled appointments were contacted by phone to determine whether they could reschedule their appointments. The university ethics review board approved the use of this clinical outcomes database for this study.
Patients completed standardized testing at one week, and at three and twelve months post fracture. Demographic data was collected at the initial one-week post injury visit. The ICF was used as a conceptual model to frame the health outcome of distal radius fracture. We have outlined the model, adapted from the WHO, in Figure 1.
Figure 1 International Classification of Functioning (ICF) model applied to distal radius fracture. (Adapted from the World Health Organization, International Classification of Functioning, Disability, and Health training materials, Geneva, 2002). .
Outcome measures
All patients completed the Patient Rated Wrist Evaluation (PRWE) [24-26], and the SF-36 [27-29] at all 3 time points and the Wrist Outcome Measure (WOM) [30] at 3 and 12 months. A research assistant verbally administered the questionnaires (PRWE and SF-36) to patients who were unable to read or write. When patients were unable to understand English sufficiently to answer, the questionnaires were translated with the assistance of a bilingual family member or friend. All questionnaires were administered and scored according to the author's instructions. An independent research assistant administered the Wrist Outcome Measure.
The Wrist Outcome Measure is a composite impairment scale with components that reflect range of motion (ROM), grip strength, and dexterity [30]. Range of motion measures were measured on the N-K computerized hand evaluation system. A total score out of 30, scored by extent of attainment of normative values was given. Six wrist motions (extension/flexion [31], radial/ulnar deviation, pronation/supination [32]) and a gross finger flexion measure was summated. Grip strength was performed using the NK Digit-grip device. The standard protocol recommended by the American Society of Hand Therapists was followed [33]. High reliability has been demonstrated for this protocol and test instrument [34]. A grip strength (score out of 40) was determined as a ratio of the uninjured hand with the injured hand and adjusting for dominance. Dexterity was measured using the checkers subtest of the Jebson's Hand Function Test (score/15) [35]. A total Wrist Outcome Measure score out of 85 was devised from these scales. Further background and discussion on the development of an impairment rating score can be found elsewhere [30].
Patient Rated Wrist Evaluation is a 15-item questionnaire that equally rates wrist-related pain and disability in functional activities (see BMC reference for complete form) [19,24,25,36]. Scoring is done on an eleven-point scale (0–10) with zero being no issues or pain and 10 being unable to do or severe pain. There are five questions that require the individual to rate their pain doing activities such as at rest, repeated motion, and lifting. Functional items are divided into two categories, specific and usual activities. There are six specific tasks such as turning a doorknob, cutting meat, fastening a button, and four usual activity categories, self-care, work, household duties, and recreation. The PRWE can be divided into three sub-scales, pain, specific activities, and usual activities. The total of the combined scales is 100 (50 from pain, 60 from specific, and 40 from usual). The psychometric properties of this scale are excellent [19,24,25,36] and the patterns of recovery following a fracture have been described using this scale [19,20].
The SF-36 is a widely used health outcome measure. It is comprised of eight scales and two summary scores [27-29,37]. There is a large database of normative data available through the Medical Outcomes Trust. The scale has eight sub-scales that portray various domains of health: physical function, physical role, bodily pain, vitality, general health, emotional role, mental health, and social function. These sub-scales are scored out of a maximum score of 100 (higher is better). The physical and mental health component summary scores represent the two main dimensions of health. These scores are calculated in a three-step process which involves weighting, transforming and aggregating the subscale scores to compute summary scores scaled to a US population which represent these two distinct domains of health (US population mean = 50). While the ICF model portrays health as a single concept with multiple, interacting contributors, the SF-36 separates physical and mental health. Since we expected largely a physical effect of wrist fracture and because the SF-36 has been shown to be preferable to other general health measures for musculoskeletal disorders [38,39] we choose it to represent health status. While lesser effects were expected in mental health we decided to include both the Mental and Physical Component Summary Scores as outcomes to determine the relative effects on both domains of health providing a more complete picture of overall health.
Data analysis
Descriptive statistics were calculated for the dependent variables (SF-36 physical and mental health component summary scores) and independent variables (WOM total score, PRWE sub-scales pain, specific activities, and usual activities). All data was inspected for assumption violation by using histograms, box, and scatter-plots. Missing values were replaced using linear extrapolation. Missing values accounted for less than 5% of data points.
Univariate analysis was completed to determine the relationship between variables of interest and outcome variables. Pearson's Correlation Coefficient was used to determine the relationship between SF-36 physical and mental health summary scores and the PRWE pain, specific, and usual activity scales, and the WOM total score. Correlation was determined at time one (one week post injury, time two (three months post injury), and time three (twelve months post injury).
Multivariate analysis was used to determine the explanatory model for distal radius fracture health outcome at time one (one week post), two (three months post), and three (twelve months post). Multiple regression equations were calculated using the SF-36 physical and mental health summary scores as the dependent variables, and patient characteristics, PRWE sub-scales and the WOM as the independent variables as described in Figure 1. The WOM was measured at three and twelve months but not at one week. It was felt that the variables of sex and age are known to be related to health, so we controlled for age and sex by blocked entry of these variables and then continued with stepwise entry of the independent variables. Six stepwise regression models were built. Data was inspected for assumption violation by examining box and scatter-plots of residuals against explanatory variables from each model. Influential data points were examined using Cook's distance. The F to enter was 0.05 and the F to remove was 0.10. Statistical significance was set at 0.05 for all outcomes. All statistics was performed using SPSS 13.
Results
Sample characteristics
There was a total of 790 persons, mean age of 51.4 (SD = 17.6, age range 18–91) in this study. The majority of the people in the study were female (68%). Descriptive characteristics of the sample can be found in Table 1. Summary scores for the outcome measures at each time frame can be found in Table 2. The mean score of the outcome measures improve at each follow up time period. At one week post injury scores demonstrate moderate to severe activity limitation and participation restriction and the one-year measures demonstrate little activity limitation or participation restriction. The PRWE specific activity sub-scale showed the most change over time, from 51.3 (severe limitation) to 6.3 (minimal limitation).
Table 1 Sample characteristics
Variable Description
Sex Male = 251
Female = 539
Dominant Hand Right = 90%
Left = 10%
Wrist Injured Right = 45%
Left = 49%
Mechanism of Fracture Fall on ice = 18%
Other fall = 66%
Other = 10%
Energy of Fracture* Low = 69%
Medium = 19%
High = 6%
Highest Level of Education Finished high school = 26%
Finished college = 18%
Finished university = 8%
Finished graduate school = 4%
Occupation at Injury† Retired = 27%
Service = 13%
Professional = 12%
Occupational Demand§/P > Low = 57%
Moderate = 24%
High = 19%
Had Physiotherapy 83%
* Low = fall from a standing position, Medium = fall from a height, High = trauma
† Top three occupations
§Self-report of how much they used their hand at work; Low = low force, low repetition, Moderate = frequent repetition, intermittent force, High = high force, constant repetition
Table 2 Descriptive statistics for outcome measures at one week, three months, and one year post injury.
Variable Time 1 Mean (SD) Time 2 Mean (SD) Time 3 Mean (SD)
Wrist Outcome Measure (/85; 85 = best) N/A 59.6(8.9) 73.9(7.4)
PRWE pain scale (/50; 50 = worst) 30.2(11.6) 17.0(10.4) 8.1(9.5)
PRWE specific scale (/60; 60 = worst) 51.3(14.1) 19.4(15.0) 6.3(10.3)
PRWE usual scale (/40; 40 = worst) 26.3(11.9) 11.4(12.5) 5.6(12.6)
SF-36 physical health (US norm 50) 37.2(8.7) 43.7(8.9) 49.0(8.7)
SF-36 mental health (US norm 50) 49.8(11.2) 51.5(9.8) 54.8(7.5)
Univariate analysis
Results from the univariate analysis can found in Table 3. All PRWE sub-scales were correlated with SF-36 physical health at one week, three, and twelve months post injury. At one week the sub-scale of usual activity demonstrated the highest correlation (r = -0.31, p = 0.01), at three and twelve months it was specific activity (r = -0.53, p = 0.01, r = -0.52, p = 0.01). Only usual activity was correlated with mental health (r = 0.09, p = 0.05) at one-week post injury. However, at three months all PRWE sub-scales were significantly correlated to physical health and at twelve months all independent variables (PRWE and WOM) were significantly correlated.
Table 3 Correlation results between outcome measures at one week, three months, and one year post injury.
Variable PRWE pain PRWE specific PRWE usual SF-36 physical health SF-36 mental health
Wrist Outcome Measure
Time 2 -0.27** -0.35** -0.20** 0.21** 0.07
Time 3 -0.43** -0.46** -0.44** 0.32** 0.14**
PRWE pain
Time 1 0.46** 0.44** -0.27** 0.05
Time 2 0.75** 0.53** -0.50** -0.27**
Time 3 0.79** 0.34** -0.50** -0.34**
PRWE specific
Time 1 0.48** -0.29** 0.01
Time 2 0.57** -0.42** -0.23**
Time 3 0.34** -0.52** -0.30**
PRWE usual
Time 1 -0.31** 0.09*
Time 2 -0.42** -0.26**
Time 3 -0.16** -0.10**
SF-36 physical health
Time 1 -0.002
Time 2 0.10**
Time 3 0.13**
* p = 0.05, **p = 0.01
† Outcome variables are correlated with each variable at their respective time periods, i.e., time 1 with time 1, time 2 with time 2, etc.
Multivariate analysis
All regression results can be found in Tables 4, 5, and 6. The result from the forward stepwise regression model for SF-36 physical health at one-week post injury yielded a weakly predictive model where all PRWE sub-scales were retained with an R2 = 0.13, p < 0.0001 for the full model (Table 4). Usual activity was most predictive and accounted for R2 = 0.10 of the model. For SF-36 mental health, minimal effects were observed with only PRWE usual activity retained within the model R2 = 0.01, p = 0.04.
Table 4 Multiple regression results for SF-36 Physical Health one week post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.13, p < 0.0001
PRWE usual* 0.10 -0.19 0.0001
PRWE specific 0.12 -0.15 0.0001
PRWE pain 0.13 -0.12 0.003
* Independent variable scores are from one-week post injury
Table 5 Multiple regression results for SF-36 Physical and Mental Health three months post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.33, p < 0.0001*
PRWE specific† 0.28 -0.27 0.0001
PRWE pain 0.31 -0.23 0.0001
PRWE usual 0.33 -0.15 0.0001
SF-36 Mental Health Summary Scale, Total Model R2 = 0.10, p < 0.0001§
PRWE pain 0.08 -0.18 0.0001
PRWE usual 0.10 -0.17 0.0001
* Excluded variable Wrist Outcome Measure
† Independent variable scores are from three months post injury
§Excluded variables Wrist Outcome Measure, PRWE specific
Table 6 Multiple regression results for SF-36 Physical and Mental Health one year post injury
Variable R2 Standardized Beta P value
SF-36 Physical Health Summary Scale, Total Model R2 = 0.28, p < 0.0001*
PRWE specific† 0.25 -0.29 0.0001
PRWE usual 0.28 0.28 0.0001
* Excluded variables PRWE pain and Wrist Outcome Measure
† Independent variable score are from one-year post injury
The regression results for three-month post injury models yielded greater prediction form included variables (Table 5). All PRWE sub-scales were retained in the SF-36 physical health model with a total R2 = 0.33, p < 0.0001. PRWE specific activity was most predictive and accounted for R2 = 0.28 of the total model. Again at the three-month model, SF-36 mental health was less explained by wrist scores, but the independent variables PRWE pain and usual activity were retained, R2 = 0.10, p < 0.0001. PRWE pain accounted for the major effect, R2 = 0.08 of the total model.
The regression results for one-year post injury can be found in Table 6. At one year post injury the regression model for SF-36 physical health showed that PRWE specific and usual activity accounted for a total R2 = 0.28, p < 0.0001, with specific activity accounting for the majority of the model, R2 = 0.25. The regression model for SF-36 mental health one year post injury showed that only PRWE pain was retained and accounted for a total R2 = 0.08, p < 0.001.
Discussion
This study determined that the ICF framework is supported when evaluating the impact of distal radius fracture on health because impairment, activity limitations and participation restrictions, individually and in combinations, were related to self-reported physical health status as measured on the SF-36. Statistically significant models for SF-36 physical and mental health were found. The models for SF-36 physical health were strong with the PRWE accounting for 13% (one week), 33% (three months), and 28% (one year) of the variance. Models for SF-36 mental health demonstrated weak relationships with the PRWE accounting for 1% (one week), 10% (three months), and 8% (one year) of the variance. This study confirms previous work that a distal radius fracture mainly affects the physical domains of health, although it does suggest that pain levels and mental health are also related.
The ICF framework is advantageous as the inclusion of aspects relating to the injury, the individual and the environment provide a broader view of how health interventions might be undertaken. In the past the focus of outcome for distal radius fracture has been impairment based (e.g., radiographic data, strength, and range of motion). However, studies have shown that impairment is not necessarily the best method to measure outcome as it does not always reflect activity and or participation restrictions [13,22,40]. We measured health by self-report allowing us to capture early and late health effects. The PRWE sub-scales were able to explain a significant portion of the SF-36 physical health score at all time periods. When examining the areas measured by the PRWE sub-scales this indicates that problems in areas such as pain, dexterity, lifting, work, household duties, recreation, and self-care after fracture do contribute to overall physical health. However, there were differences in the magnitude of the models and the most prevalent sub-scale between the time periods suggesting that impairment, activity, and participation have different health impacts at different time points in recovery from distal radius fracture.
The regression equations at one-week explained relatively little of the health impact of the fracture. This may have been because PRWE scores are consistently very poor at this point across all patients [19], and variations in health status were not well reflected at this point. For example, many patients would be immobilized providing a common restriction on specific activities like lifting or getting up from a chair. The PRWE usual activity focused on ability to do usual activities and was significant, although accounting for only 10% of the variance of the physical health model (R2 = 0.13) at one-week post injury. This sub-scale evaluates perform their usual self-care, work, household duties, and recreational activities (i.e., participation). This suggests that there is more variability in ability to participate in usual activity, than is observed on the specific activities subscale and these variations impacted on health status. This is in agreement with qualitative studies where clients emphasized the impact of distal radius fracture on work and household activities [41,42]. It is worth noting that issues with pain and inability to do specific activities were not unimportant at this time as they were rated as being at high levels of limitation. Given that participation in usual activity was related to early health status it might be worthwhile to provide some focus on methods to adapt to limitations in the early stage of fracture treatment. Individuals might benefit from educational materials that outline expected activity limitations and possible adaptations to maximize their ability to perform common tasks of daily life.
SF-36 physical health scores at three months and one-year post injury were explained to a greater extent by variations in the PRWE. The PRWE sub-scale of specific activity explained between 25% (model R2 = 0.28) and 28% (model R2 = 0.33) of the variance, replacing usual activity as the prominent variable. At three months post injury the cast has been removed and rehabilitation is underway and patients may be variable in their inability to do specific tasks that require pain-free wrist motion or strength. Self-reported wrist/hand activity limitations and participation restrictions explained a significant portion of overall physical health scores indicating that the wrist injury has a substantial impact on overall health in a manner that is consistent with the view of health portrayed in ICF. Composite wrist impairment rating had minimal additional influence on the models accounting for between 1% and 3% additional variance in health. This is in agreement with the studies that found that impairment was not a good indicator of function [13,21,22,30,40]. Physical impairment contributes to activity limitation, however, there is not a direct relationship.
Though the PRWE sub-scales accounted for a significant portion of the SF-36 physical health score (13%-33%), there are obviously other influencing factors that remain unexplained. We were unable to address a broad spectrum of potentially useful variables given database limitations. More complex models are needed to explore the remainder of the variance and additional concepts such as workplace environment; rehabilitation, surgery, socio-economic status, etc., should be included.
Physical and mental health domains have been seen as distinct domains of health according to SF-36 development and validation. This was also true in this study where the correlation between the physical and mental component summary scores was very low (r<0.10). Neither physical impairments, activities limitations nor participation restrictions were strongly predictive of SF-36 mental health scores. This is consistent with the view that distal radius fracture primarily affects physical health. The PRWE pain sub-scale accounted for slight variation in SF-36 mental health scores. Pain however, has not been shown to significantly impact recovery of distal radius fracture [43,44]. In fact in a study by Bialocerkowski [42], clients with wrist disorders were asked to explore difficulties post injury and pain was not mentioned. Instead issues with household duties, work demands, recreation, and fine motor skills were identified. Because regression reveals associations, it is not clear whether higher pain lowers mental health, or if those with poorer mental health experience more pain.
In describing rehabilitation following distal radius fracture, the importance of a staged approach has been suggested [45-47]. This study would support a staged approach as the mediators of physical health vary over time. In the early phases where immobilization and fracture healing limit motion and activity, the ability to perform usual activities is important. In addition to the customary attention to pain management, rehabilitation should also include compensatory strategies or aids as required to assist individuals complete their usual activities. In the rehabilitative phase where ability to perform specific tasks plays a larger role, remediation of impairment, incorporation of client driven goals and activity-based rehabilitation [48] are needed. Finally, the importance of participation is highlighted in these models and suggests that participation in usual self-care, household, work, and recreation, must be maximized to restore physical health. It has been demonstrated that even after adjusting for age and comorbidity patients more than 65 years of age who sustain a wrist fracture have a 57% 7-year survival rate, as compared to 71% for the comparative US population [49]. One possible contributor to this problem can be reduced participation in an active lifestyle that increases risk for additional health problems. Addressing participation during fracture rehabilitation may have short-term and longer-term health benefits.
Conclusion
The ICF model is useful in framing the health effects of a distal radius fracture that has implications for optimal management of distal radius fractures. Self-reported health measures should provide insight into the impairment, activity limitations and participation restrictions that result from a distal radius fracture. These aspects of health can be addressed at different phases of fracture management and rehabilitation to provide optimal physical health recovery.
Authors' contributions
JM formed the original study design and clinical database, obtained the ethics approval and obtained database funding. JH and JM developed the research question and planned statistical analyses. JH conducted statistical analyses and drafted the manuscript. JR enrolled, treated and evaluated patients in this study and contributed to the design of evaluation procedures, study equipment and personnel funding. All authors approved the final study protocol, contributed to interpretation of the study results and participated in revisions of the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This project was supported by the New Investigator Award, Canadian Institute of Health Research (JCM), and Doctoral Research Award, Canadian Institute of Health Research, and a Canadian Institute of Health Research Training Fellowship in Quality of Life Research in Rehabilitation, sponsored by the Institute of Musculoskeletal Health and Arthritis (JEH).
==== Refs
Organization WH International Classification of Impairments, Disabilities and Handicaps. A Manual of Classification Relating to the Consequences of Disease. 1980 Geneva, World Health Organization
Garratt A Schmidt L Mackintosh A Fitzpatrick R Quality of life measurement: bibliographic study of patient assessed health outcome measures BMJ 2002 324 1417 12065262 10.1136/bmj.324.7351.1417
International Classification of Functioning, Disability and Health 2003 World Health Organization
Grimby G Smedby B ICF approved as the successor of ICIDH J Rehabil Med 2001 33 193 194 11585148 10.1080/165019701750419545
Cieza A Brockow T Ewert T Amman E Kollerits B Chatterji S Ustun TB Stucki G Linking health-status measurements to the international classification of functioning, disability and health J Rehabil Med 2002 34 205 210 12392234 10.1080/165019702760279189
Tremayne A Taylor N McBurney H Baskus K Correlation of impairment and activity limitation after wrist fracture Physiother Res Int 2002 7 90 99 12109238
MacDermid JC Measurement of health outcomes following tendon and nerve repair J Hand Ther 2005 18 297 312 15891987
Kjeken I Dagfinrud H Slatkowsky-Christensen B Mowinckel P Uhlig T Kvien TK Finset A Activity limitations and participation restrictions in women with hand osteoarthritis: Patients descriptions, and associations between dimensions of functioning Ann Rheum Dis 2005 15829571
Pap G Angst F Herren D Schwyzer HK Simmen BR Evaluation of wrist and hand handicap and postoperative outcome in rheumatoid arthritis Hand Clin 2003 19 471 481 12945645 10.1016/S0749-0712(03)00004-0
Stucki G Cieza A Geyh S Battistella L Lloyd J Symmons D Kostanjsek N Schouten J ICF Core Sets for rheumatoid arthritis J Rehabil Med 2004 87 93 15370754 10.1080/16501960410015470
Stucki G Cieza A The International Classification of Functioning, Disability and Health (ICF) Core Sets for rheumatoid arthritis: a way to specify functioning Ann Rheum Dis 2004 63 Suppl 2 ii40 ii45 15479870 10.1136/ard.2004.028233
Weigl M Cieza A Harder M Geyh S Amann E Kostanjsek N Stucki G Linking osteoarthritis-specific health-status measures to the International Classification of Functioning, Disability, and Health (ICF) Osteoarthritis Cartilage 2003 11 519 523 12814615 10.1016/S1063-4584(03)00086-4
Dekkers M Soballe K Activities and impairments in the early stage of rehabilitation after Colles' fracture Disabil Rehabil 2004 26 662 668 15204505 10.1080/09638280410001683173
Larsen CF Lauritsen J Epidemiology of acute wrist trauma Int J Epidemiol 1993 22 911 916 8282472
Melton LJIII Amadio PC Crowson CS O'Fallon WM Long-term trends in the incidence of distal forearm fractures Osteoporos Int 1998 8 341 348 10024904 10.1007/s001980050073
O'Neill TW Cooper C Finn JD Lunt M Purdie D Reid DM Rowe R Woolf AD Wallace WA Incidence of distal forearm fracture in British men and women Osteoporos Int 2001 12 555 558 11527052 10.1007/s001980170076
Singer BR Mclauchlan GJ Robinson CM Christie J Epidemiology of fractures in 15000 adults J Bone Joint Surg Br 1998 80B 243 248 9546453 10.1302/0301-620X.80B2.7762
Thompson PW Taylor J Dawson A The annual incidence and seasonal variation of fractures of the distal radius in men and women over 25 years in Dorset, UK Injury 2004 35 462 466 15081322 10.1016/S0020-1383(03)00117-7
MacDermid JC Roth JH Richards RS Pain and disability reported in the year following a distal radius fracture: A cohort study BMC Musculoskelet Disord 2003 4 24 14588078 10.1186/1471-2474-4-24
MacDermid JC Richards RS Roth JH Distal radius fracture: a prospective outcome study of 275 patients J Hand Ther 2001 14 154 169 11382255
Fernandez JJ Gruen GS Herndon JH Outcome of distal radius fractures using the short form 36 health survey Clin Orthop 1997 36 41 9269152 10.1097/00003086-199708000-00007
Anzarut A Johnson JA Rowe BH Lambert RG Blitz S Majumdar SR Radiologic and patient-reported functional outcomes in an elderly cohort with conservatively treated distal radius fractures J Hand Surg [Am] 2004 29 1121 1127 15576226 10.1016/j.jhsa.2004.07.002
Schappler Morris N Distal radius frcatures in adults: Self-reported physical functioning, role functioning and meaning of injury Orthopedic Nursing 2005 19 37 48
MacDermid JC Turgeon T Richards RS Beadle M Roth JH Patient rating of wrist pain and disability: a reliable and valid measurement tool J Orthop Trauma 1998 12 577 586 9840793 10.1097/00005131-199811000-00009
MacDermid JC Development of a scale for patient rating of wrist pain and disability J Hand Ther 1996 9 178 183 8784681
MacDermid JC Tottenham V Responsiveness of the Disability of the Arm, Shoulder, and Hand (DASH) and Patient-Rated Wrist/Hand Evaluation (PRWHE) in evaluating change after hand therapy J Hand Ther 2004 17 18 23 14770134
Ware JEJ Sherbourne CD The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection Med Care 1992 30 473 483 1593914
Ware JE Snow KS Kosinski M Gandek B SF-36 Health Survey Manual and Interpretation Guide 1993 Boston, The Health Institute, New England Medical Center
Ware JE Kosinski M Keller SD SF-36 Physical and Mental Health Summary Scales: A User's Manual 1994 3rd Boston, Mass., The Health Institute, New England Medical Center 1.1 10.12
MacDermid JC Donner A Richards RS Roth JH Patient versus injury factors as predictors of pain and disability six months after a distal radius fracture J Clin Epidemiol 2002 55 849 854 12393071 10.1016/S0895-4356(02)00445-6
LaStayo PC Wheeler DL Reliability of passive wrist flexion and extension goniometric measurements: a multicenter study Phys Ther 1994 74 162 174 8290621
Armstrong AD MacDermid JC Chinchalkar S Stevens RS King GJ Reliability of range-of-motion measurement in the elbow and forearm J Shoulder Elbow Surg 1998 7 573 580 9883416 10.1016/S1058-2746(98)90003-9
Fess EE Casanova JS Grip strength Clinical Assessment Recommendations 1992 5 2nd Chicago, American Society of Hand Therapists 41 46
MacDermid JC Alyafi T Richards RS Test-retest reliability of static and endurance grip strength tests performed on the Jamar and NK Physiother Can 2001 53 48 54
Jebson RH Taylor N Trieschmann RB Trotter MJ Howard LA An objective and standardized test of hand function. Arch Phys Med Rehabil 1969 50 311 319 5788487
MacDermid JC Richards RS Donner A Bellamy N Roth JH Responsiveness of the short form-36, disability of the arm, shoulder, and hand questionnaire, patient-rated wrist evaluation, and physical impairment measurements in evaluating recovery after a distal radius fracture J Hand Surg [Am ] 2000 25A 330 340 10.1053/jhsu.2000.jhsu25a0330
Ware JEJ Conceptualization and measurement of health-related quality of life: comments on an evolving field Arch Phys Med Rehabil 2003 84 S43 S51 12692771 10.1053/apmr.2003.50246
Beaton DE Bombardier C Hogg-Johnson SA Measuring health in injured workers: a cross-sectional comparison of five generic health status instruments in workers with musculoskeletal injuries Am J Ind Med 1996 29 618 631 8773722 10.1002/(SICI)1097-0274(199606)29:6<618::AID-AJIM6>3.0.CO;2-F
Stucki G Liang MH Phillips C Katz JN The Short Form-36 is preferable to the SIP as a generic health status measure in patients undergoing elective total hip arthroplasty Arth Care Res 1995 8 174 181 7654802
Karnezis IA Fragkiadakis EG Association between objective clinical variables and patient-rated disability of the wrist J Bone Joint Surg Br 2002 84 967 970 12358387 10.1302/0301-620X.84B7.12673
Bialocerkowski AE Grimmer KA Compensatory mechanism use during the first six months following distal radius fracture International Journal of Therapy and Rehabilitation 2004 11
Bialocerkowski AE Difficulties associated with wrist disorders--a qualitative study Clin Rehabil 2002 16 429 440 12061478 10.1191/0269215502cr516oa
Young BT Rayan GM Outcome following nonoperative treatment of displaced distal radius fractures in low-demand patients older than 60 years J Hand Surg [Am] 2000 25 19 28 10642469 10.1053/jhsu.2000.jhsu025a0019
Fujii K Henmi T Kanematsu Y Mishiro T Sakai T Terai T Fractures of the distal end of radius in elderly patients: a comparative study of anatomical and functional results J Orthop Surg (Hong Kong ) 2002 10 9 15 12401915
Distal Radius Fracture: Practice Guidelines 1998 Chicago, Il, American Society of Hand Therapists
Laseter GF Carter PR Management of distal radius fractures J Hand Ther 1996 9 114 128 8784675
MacDermid JC Hand therapy management of intra-artiocular fractures with open reduction and pi plate fixation: A therapist's perspective Techniques in Hand and Upper Extremity Surgery 2004 8 219 223 16518095
Maciel JS Taylor NF McIlveen C A randomised clinical trial of activity-focussed physiotherapy on patients with distal radius fractures Arch Orthop Trauma Surg 2005 125 515 520 16136342 10.1007/s00402-005-0037-x
Rozental TD Branas CC Bozentka DJ Beredjiklian PK Survival among elderly patients after fractures of the distal radius J Hand Surg [Am ] 2002 27 948 952 12457343 10.1053/jhsu.2002.36995
|
16269089
|
PMC1310524
|
CC BY
|
2021-01-04 16:36:23
|
no
|
Cytojournal. 2005 Nov 4; 2:18
|
latin-1
|
Cytojournal
| 2,005 |
10.1186/1742-6413-2-18
|
oa_comm
|
==== Front
BMC Dev BiolBMC Developmental Biology1471-213XBioMed Central London 1471-213X-5-251628198610.1186/1471-213X-5-25Research ArticleDistinct types of glial cells populate the Drosophila antenna Sen Anindya [email protected] Chetak [email protected] Dhanisha [email protected] Veronica [email protected] Department of Biological Sciences, Tata Institute of Fundamental Research, Homi Bhabha Rd., Mumbai 400005, India2 National Centre for Biological Sciences, TIFR, GKVK PO, Bellary Rd., Bangalore 560065, India3 Dept. of Physiology and Cellular Biophysics, Columbia University, New York. USA4 Queensland Brain Institute, University of Queensland, Brisbane, Australia2005 11 11 2005 5 25 25 27 7 2005 11 11 2005 Copyright © 2005 Sen et al; licensee BioMed Central Ltd.2005Sen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The development of nervous systems involves reciprocal interactions between neurons and glia. In the Drosophila olfactory system, peripheral glial cells arise from sensory lineages specified by the basic helix-loop-helix transcription factor, Atonal. These glia wrap around the developing olfactory axons early during development and pattern the three distinct fascicles as they exit the antenna. In the moth Manduca sexta, an additional set of central glia migrate to the base of the antennal nerve where axons sort to their glomerular targets. In this work, we have investigated whether similar types of cells exist in the Drosophila antenna.
Results
We have used different P(Gal4) lines to drive Green Fluorescent Protein (GFP) in distinct populations of cells within the Drosophila antenna. Mz317::GFP, a marker for cell body and perineural glia, labels the majority of peripheral glia. An additional ~30 glial cells detected by GH146::GFP do not derive from any of the sensory lineages and appear to migrate into the antenna from the brain. Their appearance in the third antennal segment is regulated by normal function of the Epidermal Growth Factor receptor and small GTPases. We denote these distinct populations of cells as Mz317-glia and GH146-glia respectively. In the adult, processes of GH146-glial cells ensheath the olfactory receptor neurons directly, while those of the Mz317-glia form a peripheral layer. Ablation of GH146-glia does not result in any significant effects on the patterning of the olfactory receptor axons.
Conclusion
We have demonstrated the presence of at least two distinct populations of glial cells within the Drosophila antenna. GH146-glial cells originate in the brain and migrate to the antenna along the newly formed olfactory axons. The number of cells populating the third segment of the antenna is regulated by signaling through the Epidermal Growth Factor receptor. These glia share several features of the sorting zone cells described in Manduca.
==== Body
Background
Odor information in animals is represented as a spatial pattern of activity among glomeruli in the olfactory lobe [1]. This odotopic map is generated by the projection of olfactory receptor neurons (ORNs) each expressing a single odorant receptor (Or) gene to a defined glomerulus(i). How is this wiring pattern achieved? Compelling evidence exists in vertebrates for a role of the Ors themselves in providing cues for connectivity [2,3]. Such a mechanism seems unlikely in insect olfactory development where a carefully orchestrated interaction between ORNs, glial cells and lobe interneurons pattern the structural units underlying odor coding [reviewed in [4]].
The cellular events occurring during development of the Drosophila olfactory system have been reviewed recently [5]. Adult ORNs are specified within the antennal disc and project to the brain during early pupation. The neurons travel over the lobe anlage in the outer nerve layer occupying positions specified by interaction of Roundabout receptors with a gradient of the ligand Slit [6]. Axon terminals invade the lobe and project to specific glomeruli where they synapse with local lobe and projection interneurons. An attractive hypothesis is that targeting of ORNs and subsequent synapse formation is regulated by transcripts of the Down Syndrome Cell Adhesion Molecule (Dscam) gene [5,7].
There are a number of studies that demonstrate interdependence between neurons and glia during development [8,9]. In the Drosophila olfactory system, peripheral glial cells have been shown to arise from sensory lineages and play a role in patterning ORNs within distinct fascicles as they exit the antenna [10]. A set of central glia associated with the developing olfactory lobe elaborate projections into the neuropil to ensheath the newly formed protoglomeruli [10]. Ablation of the equivalent cells in the moth Manduca sexta, results in a failure in glomerular maturation and stabilization [11,12]. Here the migration of central glia to the glomerular borders is triggered by the earliest arriving ORNs which signal via nitric oxide [13]. In Drosophila a group of about 200 neurons, determined by the basic helix-loop-helix (bHLH) transcription factor Atonal (Ato) are the first to enter the lobe and have been proposed to act as pioneers [14]. In their absence the remaining ~1000 neurons fail to make appropriate targets in the lobe. The fate of central glia has not been investigated in these mutants. In Manduca an additional group of central glia, termed sorting glia, migrate to the base of the antennal nerve where ingrowing axons sort according to glomerular targets [15]. In vivo and culture studies have shown that these glia induce morphological changes in growth cones of the ORNs suggestive of alterations in adhesion and cytoskeletal dynamics [16,17]. This argues for reciprocal signaling between neurons and glia which needs to be investigated further.
In this paper, we demonstrate the presence of a set of cells in Drosophila, which share similarities with the sorting zone glia in the moth. The cells which are labeled with GH146::GFP become associated with the antennal nerve when it reaches the brain. The cells migrate into the third segment of the antenna where they ensheath axons of the ORNs as they project from the antenna to the brain. Although evidence that these glia influence the development or the function of the ORNs is lacking, their possible function is discussed with respect to findings in other insects.
Results
Cellular markers define two distinct subsets of glial cells within the Drosophila antenna
We stained pupal antenna with antibodies against the transcription factor Reversed Polarity (Repo) which serves as a marker for differentiated glial cells. Repo-positive cells are first detected at about 16 hours (hrs) after pupa formation (APF) when they closely associate with developing sensory axons [10,14]. At 36 hrs APF sensory neurons have differentiated (blue in Fig. 1A) and the third segment is invested with 100 ± 2 glial cells (red in Fig. 1A). The use of P(Gal4) lines to drive GFP in subsets of cells allows us to distinguish two classes of antennal glia. We constructed stocks carrying recombinant chromosomes bearing the UAS-GFP transgene with either Mz317-Gal4 (Mz317::GFP) or GH146-Gal4 (GH146::GFP). Mz317::GFP is known to mark peripheral, perineural and cell body glia [18] and labels about 70% of glial cells which we denote Mz317-glia (Fig. 1B). GH146::GFP, is a widely used marker for projection interneurons which connect the olfactory lobes to higher centres in the calyx of the mushroom bodies and the lateral horn [[19]; see Fig. 3A,B]. This marker labels about 30% of Repo-positive cells in the third antennal segment which we refer to as GH146-glia (Fig. 1C). In order to demonstrate that these reporters mark non-overlapping glial populations, we stained antennae from pupae carrying both Mz317::GFP and GH146::GFP transgenes with anti-GFP and anti-Repo (Fig. 1D). The co-localization of Repo positive cells with GFP expressed was evaluated carefully in 1 μm confocal sections through the entire antenna (Fig. 1G). All the Repo-positive cells barring a small set of 12 ± 3 (n = 6) (arrowheads in Fig. 1D) were labeled by the GFP reporters. These data together with that presented in Figures 1B and 1C suggest that antennal glial cells are either of the Mz317 or the GH146 type. The cells unlabeled by either reporter could represent yet another class of cells or could reflect lower levels of expression of GFP of one of the Gal4 drivers. This needs to be examined further.
Figure 1 Two populations of glial cells are present in the third segment of the antenna. 36 hr pupal antenna stained with the neuron-specific antibody mAb22C10 (blue in A) and anti-Repo (red). (B) Mz317::GFP stained with anti-GFP (green) and anti-Repo. Only a few 1 μm confocal sections have been stacked to show the double labeled cells (small arrows). Repo-positive cells that do not express Mz317-GFP are indicated with arrowheads. (C) GH146::GFP stained with anti-GFP (green) and anti-Repo. Numbers in B (N = 10) and C (N = 3) indicate the GFP expressing Repo-positive cells. (D) 36 hr antenna from GH146::GFP/Mz317-Gal4 pupae stained with anti-GFP (green) and anti-Repo (red). Most of the cells stain with both antibodies except for a few cells indicated with arrowheads. (E) In the late pupa the projections of the GH146-glia are similar to that at 36 hrs APF while those of Mz317 (F) are markedly different. Scale bar = 40 μm for panels A-F; Ar-arista. (G) Magnified region from single confocal sections demonstrating glial cells stained with expressing both Repo and GFP (G1) and those expressing Repo alone (G2). (H.I) Enlarged view of an axon fascicle in late pupal antennae from Mz317::GFP (H) and GH146::GFP (I) (scale bar = 24 μm). Dotted lines in (I) mark the boundaries of the GH146-glial processes. This region is devoid of Mz317-glial processes. (J) Diagrammatic representation of glial patterning in the adult antenna. GH146-glia (green) tightly wrap around the axon bundles (red) while those of Mz317-glia (blue) are peripheral to this layer.
Figure 3 Developmental profile of GH146-glia. (A-C) Dissected brains of GH146::GFP pupa. The brightfield image in (C) shows the antenna (A) connected to the brain via the antennal nerve (AN). The I, II and III segments of the antenna are indicated. Fluorescence images of 20 hr and 30 hr preparations are shown in (A) and (B) respectively. GH146::GFP labels interneurons (arrows) that connect the olfactory lobes (AL-dotted circles) to the calyx of the mushroom bodies (Ca) and dorsal horn (arrowhead). A few GFP labeled cells can be seen in the antennal nerve (small arrows) at 20 hrs APF which increase by 30 hrs (small arrows in B). (D-G) Antenna from GH146::GFP pupae stained with anti-Repo (red) and anti-GFP (green). At 20 hrs APF (D,E), a large number of glial cells stained by anti-Repo are present in the second segment (II in D) and only few of these express GFP. Only one or two GH146-glia appears in the third segment (arrow in E). This number increases at 24 hrs APF (F) and 36 hrs APF (G). (H) MARCM clone generated using ey-FLP show presence of only two marked GH146-positive cells in the third segment of the antenna (small arrow). (I,J) Antennal nerve of 36 hr APF GH146::GFP pupae stained with anti-GFP (green), anti-Repo (red) and mAb22C10 (blue). The nerve is enlarged in to show the relative positioning of neurons and glia. (I) Only a subset of the nerve associated glia are of the GH146 subtype (arrow). A significant number of Repo-positive cells do not express GFP (arrowhead). (J) Processes of GH146-glia segregate axons into distinct bundles (dotted lines). (K) 36 hr APF antenna and attached antennal nerve to show the processes of the GH146-glia (green) with respect to the neurons stained with mAb22C10 (blue) and glial cell bodies marked with anti-Repo (red). The positions of the I II and III segment of the antenna are marked.
At 36 hr APF, all the glial processes tightly ensheath fascicles of sensory axons as they exit the antenna (green in Fig. 1D). The GH146-glia remain in the same position up to adulthood (Fig. 1E), while projections of Mz317-glia appear to 'loosen' from the fascicles to ensheath the cell bodies of the peripheral sense organs (Fig. 1F). A careful comparison of GFP labeled processes in Mz317::GFP (Fig. 1H) and GH146::GFP (Fig. 1I) suggest that the latter ensheath the neurons directly while Mz317-glia form an outer layer (schematized in Fig. 1J).
What is the origin of the antennal glia?
There are three morphologically distinct types of sense organs on the antennal surface- the coeloconica, trichoidea and basiconica [20]. Progenitors of the approximately 70 coleoconic sensilla which are specified by Ato, each give rise to a glial cell [10,20]. A second b-HLH protein Amos, in combination with the Runt family transcription factor Lozenge (Lz), specifies the trichoidea and basiconica [21,22]. Amos-dependent lineages are also gliogenic but nascent glial cells undergo programmed cell death [23]. We stained 36 hr APF antennae from animals null for ato (ato1/Df(3R)p13) with anti-Repo (Fig. 2B). Only 30% of the ~100 cells were formed (Fig. 2A,B). GH146::GFP was crossed into this genetic background to demonstrate that the extant cells were all of the GH146-glial subtype (not shown). In late pupae, projections of GH146-glia are more extensive than in the wildtype (compare Fig. 2C with 1E). We believe that the aberrant morphology of projections could be explained by a lack of Mz317-glia which could restrict the extension of GH146-glial projections in normal animals.
Figure 2 GH146-glia does not originate from the sensory lineages. (A) Third antennal segment from 36 hr APF wildtype pupae stained with anti-Repo showing the position of glial cells. (B) 36 hr APF ato1/Df(3R)p13 antenna shows a significant reduction of glial cells, compared to the wildtype. (C) GH146::GFP was crossed into ato1/Df(3R)p13 to visualize GH146-glia in the late pupa. (D) 36 hrs APF lz3 antenna stained with anti-Repo. (E). lz3; GH146::GFP antenna stained with anti-GFP. (F) Antenna in (E) was also stained with mAb22C10. Sensory neurons are markedly reduced in number. Ar-arista. Numbers in panels A,B and D represent the mean and standard deviation of glial cell number in at least 8 antenna. For E only 5 antennae were counted.
We next examined whether the GH146-glia could be a subset of Amos-dependent glial cells which escape apoptosis. The majority of Amos-dependent sensilla fail to form in strong lz alleles. We stained 36 hr antenna from lz3 pupae with anti-Repo and found that the number of cells observed was not significantly different from that of wildtype controls (p < 0.05; Fig. 2D). In animals where GH146::GFP was crossed into this background these glia were observed (Fig. 2E). The somewhat reduced number of cells as compared to the wildtype can be explained by the strong decrease in ORN number [21], assuming GH146-glial cells require axons to navigate into the antenna (discussed below).
The observation that GH146-glia appears in both ato and lz mutants leads us to suggest that the cells do not arise from sensory lineages but probably 'home' into the antenna from elsewhere.
The time course of appearance of GH146-glia suggests migration into the third antennal segment
Since the GH146-glial cells do not have a peripheral origin, we decided to examine the possibility that they migrate into the antenna from the Central Nervous System (CNS). Pupal dissections exposing both antenna and brain (Fig. 3C) were stained with anti-GFP and anti-Repo to observe the appearance of glial cells at different pupal ages (Fig. 3A,B). The first cells are seen on the antennal nerve at 20 hrs APF (small arrows in Fig. 3A), a time corresponding to the entry of olfactory neurons into the brain [10]. One or two GFP positive cells are also detected within the third segment of the antenna at this time (arrow in Fig. 1E). The number of cells associated with the nerve increases steadily with pupal age (not shown) up to about 30 hrs APF (compare small arrows in Fig. 3B with 3A). GH146::GFP positive cells also increase within the third antennal segment to reach a maximum of ~30 at 36 hrs APF (Fig. 3E–G). The GFP labeled cells are only a small fraction of total glial cells detected by anti-Repo staining within the second (Fig. 3D) and third antennal segment (Fig. 3E–G). A few cells can also be detected within the arista (arrow in Fig. 3G).
While these observations support the idea that GH146-glia migrate into the third segment along the olfactory neurons, it is possible that cells already located within the antenna turn on the GH146 enhancer in a dynamic fashion. We exploited the MARCM method [24] to further investigate the origin of GH146-glia. Flipase driven by the eyeless (ey) promoter (ey-FLP) generates a high frequency of large clones covering at least half of the antennal tissue [25,26]. In a control experiment using a marker for ORNs (Or83b::GFP), we obtained >90% clonal frequency (45/49; Pinky Kain personal communication). We reasoned that if, like the ORNs, the GH146-glial cells originate within the antenna, we would obtain a similar frequency of clones. We however obtained only a 15% (3/20) clonal frequency and these were very small covering only two or three cells in each case (arrow in Fig. 3H). These animals had large clones of marked cells within the brain. A possible interpretation is that FLP-mediated recombination generated GH146::GFP cells in the brain and some of these cells migrated peripherally into the antenna.
Data presented in this section, leads us to suggest that the GH146-glial cells arise in the brain and migrate along the nascent olfactory axons to line the fascicles within the third antennal segment. These cells are however, only a small subset of the nerve associated cells and both GFP (arrow in Fig. 3I) and non-GFP (arrowhead in Fig. 3I) Repo-positive cells are detected. Staining of the mature antennal nerve with mAb22C10 and anti-GFP revealed that projections of GH146-glia segregate groups of ORNs as they project to the brain (Fig. 3J,K).
What are the mechanisms that control GH146-glial number in the antenna?
Several studies have demonstrated that glial cell number is regulated by a combination of cell migration, proliferation and programmed cell death. Aigouy and colleagues [27] have followed glia cells as they migrate along the axons in the wing blade and observed dramatic cytoskeleton changes and proliferation during migration. The signals for migration as well as glial cell survival arise from the epidermal growth factor receptor (DER) pathway [28] which impinges upon the small GTPases Ras, Rho and Rac [29-31]. We crossed a DER-lacZ transgene into the GH146::GFP strain and stained 36hr APF antenna with antibodies against β-galactosidase and GFP (Fig. 4A,B). The presence of the DER reporter in GH146-glial cells demonstrates that these cells expressed the receptor at some time in their developmental history. Ectopic expression of a dominant negative form of DER (DN-DER) in these glia (GH146::GFP/UAS-DNDER resulted in a significant reduction in the numbers of GH146-glial cells (Fig. 4C; P < 0.01). Since DER signals through the Ras/MAPK kinase pathway, we confirmed the requirement for this pathway by ectopically expressing a dominant negative form of Ras- RasN17 (Fig. 4D). The phenotype observed would be expected if abrogation of DER signaling affects survival, proliferation or migration of GH146-glial cells. Programmed cell death can be inhibited by expression of the Baculovirus Inhibitor of Apoptosis -P35. When P35 was co-expressed in GH146-glia together with DN-DER there was no rescue of glial cells as compared to expression of DN-DER alone (Fig. 4E; P < 0.0001). We ascertained that expression of P35 alone did not affect GH146-glial cells. The cell number was not distinguishable from that in the wildtype (P < 0.001; Fig. 4J). These data together indicate that failure of DER signaling in GH146-glial cells does not affect cell survival, thus indicating defects in proliferation and/or cell migration.
Figure 4 Factors regulating the numbers of GH146-glia within the third segment of the antenna. (A,B) GH146::GFP /DER-lacZ stained with anti-β-galactosidase (red in A,B) and anti-GFP (green in B). Merged image in (B) shows that GH146-glia express the DER. (C-G). 36 hr APF antenna stained with anti-GFP (green) and anti-Repo (red). Cell numbers shown in each panel are from at least 10 antenna. (C) GH146::GFP/UAS-DN-DER. (D) UASRasN17/+; GH146::GFP/+ (E) GH146::GFP/UAS-DN-DER; UAS-P35/+ (F). GH146::GFP/UAS-Cdc42v12 (G) UAS-p21/+; GH146::GFP/+. (H,I) 36 hour APF antennae of GH146::GFP/UAS-DN-DER (H) and GH146::GFP/ UAS-Cdc42v12 stained with mAb22C10 (blue). ORNs exit the third segment of the antenna in three fascicles denoted I, II and III. The slightly distorted fasciculation pattern seen in (I) is within normal variation. The arista (Ar) is to the right in each figure. (J) Numbers of GH146-glia in antenna expressing different transgenes. Cells in 10 antennae were counted in each case and the mean and standard deviation is represented.
Cell migration is regulated by signaling to the cytoskeleton by small GTPases, Rac, RhoA and Cdc42 [31]. Ectopic expression of dominant negative and constitutively active forms of Cdc42 leads to a dramatic reduction in GH146-glia (Fig. 4J). The effect was particularly striking with Cdc42v12; no GH146-glia was detected within the third segment of the antenna (Fig. 4F). These data lend some support to the hypothesis that GH146-glia migrate into the third antennal segment probably from the CNS. The morphology and arrangement of GH146-glia suggests that these cells undergo division during migration along the antennal nerve. We blocked cell division by ectopic expression of the human cyclin-dependent kinase inhibitor p21CIP1/WAF1 in the GH146 lineage. We noticed a significant reduction (Fig. 4G; P < 0.01) in the number of these cells within the antenna. These data do not allow us to distinguish whether the cells undergo proliferation during migration or upon reaching their final destination.
What is the consequence of a lack of GH146-glia on the ORNs?
ORNs exit the antenna towards the brain in three well defined fascicles [[10], Fig. 4H). Previous work had shown that constitutive activation of small GTPases in Mz317-glia compromises the patterning of sensory axons [10]. In order to test whether GH146-glial cells play a similar role, we stained 36 hr antennae from animals in which these cells were either reduced (Fig. 4H) or absent (Fig. 4I), with the neuron-specific antibody mAb22C10. The pattern of ORNs was not significantly altered in either case; the small irregularities seen in Figure 4I are within normal variation.
In Manduca, glial cells that migrate to the base of the antennal nerve act to sort ORNs to different glomeruli [15]. GH146-glial cells in Drosophila move from the CNS to third segment of the antenna. If this position marks a 'sorting zone', one would expect sensory axons to be patterned within the nerve according to their glomerular targets. Processes of GH146-glia have been shown to segregate sensory axons into groups within the antennal nerve (Fig. 3J). ORNs expressing a given Or project to the same glomerulus in the olfactory lobe [32], although these axons use different fascicles within the third antennal segment to enter the nerve [33]. Do functionally similar axons group together within the nerve? We examined 1 μm confocal sections through the antennal nerve of Or22a::GFP and Or47b::GFP animals stained with antibodies against GFP. Axons of ORNs expressing a given Or are positioned differently within the antennal nerve as they travel the brain (small arrows in Fig. 5A–C). Staining with mAb22C10 showed the presence of several other neurons intermingled between those of Or22a and Or47b (not shown). In the case of Or22a, the axons project to the DM2 glomerulus in two distinct fascicles (arrowheads in Fig. 5A).
Figure 5 Positioning of functionally similar olfactory neurons within the antennal nerve. Dissected brains of Or22a-Gal4;UAS-N-SybGFP (A) and Or47b-Gal4;UAS- N-SybGFP (B,C) were stained with antibodies against GFP. The antennal nerves (AN) are marked with white dots and the boundaries of the olfactory lobe (OL) by yellow dotted lines. Or22a-neurons travel in different regions of the antennal nerve (thin arrows) to enter the DM2 glomerulus in two bundles (arrowheads). Similarly Or47b axons traverse the nerve to reach VA1 glomerulus and also cross the midline (marked in *) to the contralateral lobe. The image in C is magnified (scale bar = 30 μm) and shows that neurons expressing the same Or gene do not group together within the lobe. Scale bar = 40 μm in A, B.
Hence while bundles of axons are segregated within the antennal nerve by projections of GH146-glial cells (Fig. 3I–K), this fasciculation does not appear to be based on Or identity. The functional significance of this axonal grouping needs to be investigated further.
Discussion
Studies in Manduca, have described three types of glia associated with the developing olfactory neuropil and nerve: i) peripheral glia that arises from the antenna and ensheath olfactory axons [16]; ii) lobe neuropil glia that surround and stabilize olfactory glomeruli [11,12]; iii) sorting zone glia which are central in origin and migrate to the base of the antennal nerve where they serve to segregate axons that target independent glomeruli [15,17]. In Drosophila, previous work has described the direct equivalents of peripheral antennal and the neuropil glia [10]. In this paper, we identify an addition set of glia marked by GH146::GFP that share common features with sorting zone glia. Our model, depicted in Figure 6, suggests that these glia arise in the brain and migrate along the ORNs which have newly connected to the brain. Cells proliferate as they migrate along the nerve and enter the third segment of the antenna. Coincident with GH146-glia arrival at the periphery, the Ato-derived glia marked by Mz317::GFP which are already present on the olfactory axons move to wrap around the cell bodies of the sensory organs. The GH146-glial processes tightly ensheath the olfactory fascicles and form a layer below that of the Mz317-glial projections. In all 36 hrs APF preparations examined, we detected ~30 GH146-glia suggesting there must exist mechanisms to regulate the number of cells terminating in the antenna.
Figure 6 Model for the migration of the GH146-glia into the antenna. Olfactory axons arrive at the brain at 20 hrs APF. At this time the Mz317-glia (blue) are already present within the third segment and are associated with developing axons. GH146-glial cells (green) appear on the antennal nerve (AN) when it arrives in the brain. By 24 hrs additional glia are observed on the nerve; the morphology of the cells suggest that they undergo mitosis as they travel along the nerve. By 30 hrs APF, large numbers of cells are present on the nerve and these some of these enter the third segment of the antenna. The cell bodies of Mz317-glial cells appear to move towards the cell bodies of the ORNs at this time. In late pupae, GH146-glia remain closely associated with the neuronal fascicles while Mz317-glial processes appear to move away and wrap around the cells bodies of the sense organs. ORNs expressing Or22a do not appear to fasciculate together in the Drosophila antennal nerve.
What is the function of GH146-glia during olfactory development?
Although these cells share many of the properties of sorting zone glia described in the moth, we were unable to implicate them in any guidance function. In Manduca, Fasciclin II (FasII), the insect ortholog of N-CAM, is expressed in a subset of ORNs which are scattered throughout the antennal nerve. Upon reaching the glial rich sorting zone, these projections segregate from the non-expressing axons and terminate in distinct glomeruli [34]. In Drosophila, we were unable to detect the presence of FasII protein on the ORNs during axon projection using the available antibodies (unpublished observations). When the GH146-glia were completely absent within the third antennal segment, we could not detect significant abnormalities in ORN fasciculation within the antenna. The processes of the GH146-glia extend into the antennal nerve to ensheath axon bundles. This segregation is not based on Or gene type and its functional significance is obscure. A similar migration of glial cells from the centre into the periphery has been described during development of the Drosophila compound eye [35].
What are the factors that induce migration of glia from the brain towards the periphery?
Cell migration in insect glia is well known to be mediated by signaling through the Fibroblast Growth Factor Receptors (FGFRs) [36]. Our preliminary observations (unpublished) appear to rule out a requirement for FGFR signaling during GH146-glial migration. The downstream effector of FGF (Dof), is not expressed in these cells at any time during their development. Further, expression of dominant negative forms of Drosophila FGF receptors do not affect glial cell number in the antenna (our unpublished data).
We suggest that GH146-glial migration is mediated by EGF signaling [37]. The ligand for DER, Vein has been shown to be known to be expressed in the antennal epidermis at a time consistent with the arrival of cells from the centre. The mechanisms involved in triggering this homing of cells into the antenna require extensive investigation.
Conclusion
We have identified a new population of glial cells in the antenna of Drosophila. The cells originate centrally and use the newly targeted olfactory axons to migrate to the third segment of the antenna. The signal for migration appears to involve EGF signaling. The data in this paper support the idea that in Drosophila there is no discrete sorting region equivalent either to the sorting zone in Manduca or to the inner nerve layer in the vertebrate olfactory bulb.
Methods
Fly strains
Mz137-Gal4 was kindly provided by Kei Ito and GH146-Gal4 by Reinhard Stocker. These lines were used to drive UAS-GFP (1010T2) thus labeling peripheral glia. The ato strains ato1/TM3 and Df(3R)p13/TM3 were obtained from Andrew Jarman [14]. The Epidermal growth factor receptor (EGFR)-lacZ strain, UAS-DN DER, UAS-RasN17, UAS-Cdc42v12, UAS-Cdc42N17, UAS-P35, UAS-p21, yw;P [ry+t7.2-neoFRT}19A, Tub-Gal80 P[ry+t7.2-neoFRT}19A and ey-FLP were obtained from the Bloomington Stock Centre at Indiana, USA. Or22a-Gal4, Or47b-Gal4 and Or83b-Gal4 lines were kindly provided by Leslie Vosshall [32].
All flies were reared at 25°C on standard cornmeal media containing yeast. For staging, white prepupae (0 h after puparium formation, APF) were collected and allowed to develop on moist filter paper. This stage lasts for an hour; hence the error in staging is 30 minutes. Ages of pupae grown at other temperatures were normalized with respect to growth at 25°C. Wild type pupae take about 100 hrs to eclose when grown at 25°C in our laboratory.
Immunohistochemistry
Pupal tissues were dissected in phosphate-buffered saline (PBS) and treated as described in [5]. The primary antibodies used were mAb22C10 (1:50, DSHB donated by Seymour Benzer), Rabbit-anti-Repo (1:1000, Susinder Sundaram), Rat-anti-Repo (1:500, Susinder Sundaram), Rabbit anti-GFP (1:10,000, Molecular Probes). Secondary antibodies used were: Alexa 488 goat anti-rabbit, Alexa 568 goat anti-mouse, Alexa 568 goat anti-rat (Molecular Probes) at 1:400; Cy5 conjugated goat anti-mouse (Amersham); at 1:300. The fluorescently labeled preparations were mounted in Vectashield (Vector Labs) and viewed on a BioRad MRC1024 or Radiance 2000 confocal microscope. Two-dimensional projections were generated by stacking appropriate sections for each channel using Confocal Assistant software (distributed by Bio-Rad). Image processing, including pseudo-coloring and labeling, were done using Adobe Photoshop 7.0.
Imaging of late pupal antennae
Antibody staining of the late pupal stages cannot be carried out because of the hardening of the cuticle. To visualize GFP fluorescence, antennae were removed, mounted in 70% glycerol and scanned on a Radiance 2000 confocal microscope immediately.
Generation of clones for lineage analysis
Clones were generated using the mosaic analysis with repressible cell marker (MARCM) method described in [25]. Pupae of genotype FRT19A/Tub-Gal80 FRT19A; GH146-Gal4 UAS-GFP/+; ey-FLP/+ and FRT19A/Tub-Gal80 FRT19A; Or83b-Gal4 UAS-GFP/+; ey-FLP/+ were dissected and antennae and brains were stained with antibodies against GFP. Preparations were examined by confocal microscopy for GFP positive cells.
Authors' contributions
AKS carried out the characterization of the Mz317 and GH146 glia in the antenna and studied the effect of DER and small GTpases signaling. CS established the migration of glia by following expression in dissected preparations and by clonal analysis. DJ carried out the initial observations on the GH146. VR conceived and designed the study, participated in drafting the manuscript. All authors have read and approved the final manuscript.
Acknowledgements
We are grateful to Reinhard Stocker, Kei Ito and Leslie Vosshall for generous gifts of fly stocks. We thank Susinder Sundaram for generous supply of rabbit and rat antibodies against Repo, and K. VijayRaghavan for useful discussions and comments on the manuscript. We acknowledge the referees of the previous version of the manuscript for many constructive criticisms. AKS and DJ acknowledge the Sarojini Damodharan fellowship for funding.
==== Refs
Korsching SI Odor maps in the brain: spatial aspects of odor representation in sensory surface and olfactory bulb Cell Mol Life Sci 2001 58 520 530 11361087
Wang F Nemes A Mendelsohn M Axel R Odorant receptors govern the formation of a precise topographic map Cell 1998 93 47 60 9546391 10.1016/S0092-8674(00)81145-9
Mombaerts P How smell develops? Nature Neuroscience 2001 4 192 1198 10.1038/nn751
Oland LA Tolbert LP Key interactions between neurons and glial cells during neural development in insects Annu Rev Entomol 2003 48 89 110 12194908 10.1146/annurev.ento.48.091801.112654
Jefferis GSXE Luo L Gilbert LI, Iatrou K, Gill SS The Development of the Olfactory System Comprehensive Molecular Insect Science 2005 1 Elsevier Pergamon 421 463
Jhaveri D Saharan S Sen A Rodrigues V Positioning glomeruli in the olfactory lobe of Drosophila by Robo signaling Development 2004 131 1903 1912 15056612 10.1242/dev.01083
Hummel T Vasconceles ML Clemens JC Fishilevich Y Vosshall LB Zipursky SL Axonal targeting of olfactory receptor neurons in Drosophila is controlled by Dscam Neuron 2003 37 221 231 12546818 10.1016/S0896-6273(02)01183-2
Chotard C Salecker I Neurons and glia: team players in axon guidance Trends in Neurosci 2004 27 6551 6561
Oland LA Tolbert LP Glial patterns during early development of antennal lobes of Manduca sexta:acomparison between normal lobes and lobes deprived of antennal axons J Comp Neurol 1987 255 196 207 3819014 10.1002/cne.902550204
Jhaveri D Sen A Rodrigues V Mechanisms underlying olfactory neuronal connectivity in Drosophila-the Atonal lineage organizes the periphery while sensory neurons and glia pattern the olfactory lobe Dev Biol 2000 226 73 87 10993675 10.1006/dbio.2000.9855
Baumann PM Oland LA Tolbert LP Glial cells stabilize axonal protoglomeruli in the developing olfactory lobe of the moth Manduca sexta J Comp Neurol 1996 373 118 28 8876467 10.1002/(SICI)1096-9861(19960909)373:1<118::AID-CNE10>3.0.CO;2-G
Oland LA Tolbert LP Mossman KL Radiation-induced reduction of the glial population during development disrupts the formation of olfactory glomeruli in an insect J Neurosci 1988 8 353 67 3339417
Gibson NJ Rössler W Nighorn AJ Oland LA Hildebrand JG Tolbert LP Neuron-glia communication via nitric oxide is essential in establishing antennal lobe structure in Manduca sexta Dev Biol 2001 240 326 339 11784067 10.1006/dbio.2001.0463
Jhaveri D Rodrigues V Sensory neurons of the Atonal lineage pioneer the formation of glomeruli within the adult Drosophila olfactory lobe Development 2002 129 1251 1260 11874920
Rössler W Oland LA Higgins MR Hildebrand JG Tolbert LP Development of a glia-rich axon-sorting zone in the olfactory pathway of the moth Manduca sexta J Neurosci 1999 19 9865 9877 10559396
Tucker ES Tolbert LP Reciprocal interactions between olfactory receptor axons and olfactory nerve glia cultured from the developing moth Manduca sexta Dev Biol 2003 260 9 30 12885552 10.1016/S0012-1606(03)00207-0
Tucker ES Oland LA Tolbert LP In vitro analyses of interactions between olfactory receptor growth cones and glial cells that mediate axon sorting and glomerulus formation J Comp Neurol 2004 472 478 495 15065121 10.1002/cne.20058
Ito K Urban J Technau GM Distribution, classification, and development of Drosophila glial cells in the late embryonic and early larval ventral nerve cord Roux Arch Dev Biol 1995 204 284 307 10.1007/BF02179499
Stocker RF Heimbeck G Gendre N de Belle JS Neuroblast ablation in Drosophila P[GAL4] lines reveals origins of olfactory interneurons J Neurobiol 1997 132 443 456 9110257 10.1002/(SICI)1097-4695(199705)32:5<443::AID-NEU1>3.0.CO;2-5
Sen A Reddy GV Rodrigues V Combinatorial expression of Prospero, Seven-up and Elav, identify progenitor cell types during sense-organ differentiation in the Drosophila antenna Dev Biol 2003 254 79 92 12606283 10.1016/S0012-1606(02)00021-0
Gupta BP Flores GV Banerjee U Rodrigues V Patterning an epidermal field, Drosophila lozenge, a member of the AML-1/Runt family of transcription factors, specifies olfactory sense organ type in a dose-dependent manner Dev Biol 1998 203 400 411 9808789 10.1006/dbio.1998.9064
Goulding SE zur Lage P Jarman AP amos, a proneural gene for Drosophila olfactory sense organs that is regulated by lozenge Neuron 2000 25 69 78 10707973 10.1016/S0896-6273(00)80872-7
Sen A Kuruvilla D Pinto L Sarin A Rodrigues V Programmed cell death and context dependent activation of the EGF pathway regulate gliogenesis in the Drosophila olfactory system Mechanisms of Development 2004 121 65 78 14706701 10.1016/j.mod.2003.10.002
Lee T Luo L Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis Neuron 1999 22 451 461 10197526 10.1016/S0896-6273(00)80701-1
Ang LH Kim J Stepensky V Hing H Dock and Pak regulate olfactory axon pathfinding in Drosophila Development 2003 130 1307 1316 12588847 10.1242/dev.00356
Jefferies GSXE Vyas R Berdnik D Ramaekers A Stocker RF Tanaka NK Luo L Developmental origin of wiring specificity in the olfactory system of Drosophila Development 2003 131 117 130 14645123 10.1242/dev.00896
Aigouy B Van de Bor V Boeglin M Giangrande A Time-lapse and cell ablation reveal the role of cell interactions in fly glia migration and proliferation Development 2004 131 5127 38 15459105 10.1242/dev.01398
Shilo BZ Signaling by the Drosophila epidermal growth factor receptor during development Experimental cell Research 2003 284 140 149 12648473 10.1016/S0014-4827(02)00094-0
Bergmann A Tugentmann M Shilo BZ Steller H Regulation of cell number by MAPK-dependent control of apoptosis: a mechanism for trophic survival signaling Dev Cell 2002 2 159 170 11832242 10.1016/S1534-5807(02)00116-8
Hidalgo A Interactive nervous system development: control of cell survival in Drosophila Biochem Soc Trans 2002 31 50 55 12546652
Sepp KJ Auld VJ RhoA and Rac1 GTPases mediate the dynamic rearrangement of actin in peripheral glia Development 2003 130 1825 35 12642488 10.1242/dev.00413
Vosshall LB Wong AM Axel R An olfactory sensory map in the fly brain Cell 2000 102 147 159 10943836 10.1016/S0092-8674(00)00021-0
Bhalerao S Sen A Stocker R Rodrigues V Olfactory neurons expressing identified receptor genes project to subsets of glomeruli within the antennal lobe of Drosophila melanogaster J Neurobiol 2003 54 577 592 12555270 10.1002/neu.10175
Higgins MR Gibson NJ Eckholdt PA Nighorn A Copenhaver PF Nardi J Tolbert LP Different isoforms of fasciclin II are expressed by a subset of developing olfactory receptor neurons and by olfactory-nerve glial cells during formation of glomeruli in the moth Manduca sexta Dev Biol 2002 244 134 154 11900464 10.1006/dbio.2002.0583
Rangarajan R Gong Q Gaul U Migration and function of glia in the developing Drosophila eye Development 1999 126 3285 3292 10393108
Condron B Spatially distinct FGF-mediated signaling directs glial morphogenesis Development 1999 126 4635 4641 10498697
Perrimon N Perkins LA There must be 50 ways to rule the signal: the case of the Drosophila EGF receptor Cell 1997 89 13 16 9094709 10.1016/S0092-8674(00)80177-4
|
16281986
|
PMC1310525
|
CC BY
|
2021-01-04 16:30:32
|
no
|
BMC Dev Biol. 2005 Nov 11; 5:25
|
utf-8
|
BMC Dev Biol
| 2,005 |
10.1186/1471-213X-5-25
|
oa_comm
|
==== Front
BMC Cell BiolBMC Cell Biology1471-2121BioMed Central London 1471-2121-6-391628750310.1186/1471-2121-6-39Research ArticleGranulocyte heterochromatin: defining the epigenome Olins Donald E [email protected] Ada L [email protected] Department of Biology, Bowdoin College, Brunswick, ME 04011, USA2005 15 11 2005 6 39 39 15 8 2005 15 11 2005 Copyright © 2005 Olins and Olins; licensee BioMed Central Ltd.2005Olins and Olins; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Mammalian blood neutrophilic granulocytes are terminally differentiated cells, possessing extensive heterochromatin and lobulated (or ring-shaped) nuclei. Despite the extensive amount of heterochromatin, neutrophils are capable of increased gene expression, when activated by bacterial infection. Understanding the mechanisms of transcriptional repression and activation in neutrophils requires detailing the chromatin epigenetic markers, which are virtually undescribed in this cell type. Much is known about the heterochromatin epigenetic markers in other cell-types, permitting a basis for comparison with those of mature normal neutrophilic granulocytes.
Results
Immunostaining and immunoblotting procedures were employed to study the presence of repressive histone modifications and HP1 proteins in normal human and mouse blood neutrophils, and in vitro differentiated granulocytes of the mouse promyelocytic (MPRO) system. A variety of repressive histone methylation markers were detectable in these granulocytes (di- and trimethylated H3K9; mono-, di- and trimethyl H3K27; di- and trimethyl H4K20). However, a paucity of HP1 proteins was noted. These granulocytes revealed negligible amounts of HP1 α and β, but exhibited detectable levels of HP1 γ. Of particular interest, mouse blood and MPRO undifferentiated cells and granulocytes revealed clear co-localization of trimethylated H3K9, trimethylated H4K20 and HP1 γ with pericentric heterochromatin.
Conclusion
Mature blood neutrophils possess some epigenetic heterochromatin features that resemble those of well-studied cells, such as lymphocytes. However, the apparent paucity of HP1 proteins in neutrophils suggests that heterochromatin organization and binding to the nuclear envelope may differ in this cell-type. Future investigations should follow changes in epigenetic markers and levels of HP1 proteins during granulopoiesis and bacterial activation of neutrophils.
==== Body
Background
The epigenome of a specific tissue constitutes the total set of chromatin modifications existing above the level of DNA base sequence and mitotically inherited, conveying stability to the differentiated state. In mammalian cells, these epigenetic modifications consist primarily of DNA methylation, histone post-translational modifications and variants, and nucleosome remodeling mechanisms [1-5]. With the increased availability of reagents and techniques for defining epigenetic modifications, numerous studies have been published describing the epigenomes of various cell types. Examples of mammalian cells that have been studied include mouse resting B lymphocytes [6], and embryonic erythrocytes and fibroblasts [7].
Granulopoiesis, the terminal differentiation of blood granulocytes (primarily neutrophils or "polymorphs") occurs within the bone marrow and is well-described [8]. In humans the process takes about two weeks, starting from the myeloblast stage (ovoid nuclei with minimal heterochromatin), exhibiting one week of differentiation and mitosis, followed by one week of post-mitotic nuclear and cytoplasmic differentiation [9]. During the post-mitotic phase the non-dividing nucleus displays progressive chromatin condensation and nuclear shape changes. The normal human neutrophil nucleus has 3–4 lobes [8]; mouse neutrophils frequently possess ring-shaped nuclei [10,11]. These modulations of neutrophil nuclear shape and the considerable amount of heterochromatin located adjacent to the nuclear envelope (NE) depend upon normal amounts of the integral NE protein lamin B receptor (LBR; for a recent review on the structure of the NE, see [12]). Without sufficient levels of LBR, the neutrophil nucleus does not exhibit the normal lobulation or ring-shape and the heterochromatin undergoes clumping removed from the NE [13,14]. Other factors involved in the differentiation of neutrophil nuclear shape include NE lamin composition and microtubule integrity (for a description of our current hypothesis, see [15]).
A recent study comparing the nuclear composition of human neutrophils with a variety of myeloid leukemias [16] concluded that normal mature neutrophils exhibit a deficiency of mono-, di- and trimethylated histone H3 lysine 9 (H3K9), combined with an absence of heterochromatin protein 1 (HP1) α, β and γ; whereas myeloid leukemias possessed all of these markers. This observation is somewhat puzzling, since LBR has been shown to interact with HP1 proteins [17,18], and HP1 has been suggested to mediate the association between heterochromatin and LBR at the NE [19,20]. Furthermore, since methylated H3K9 is a well-studied repressive epigenetic modification [21] and trimethylated H3K9 is concentrated at pericentric and centric constitutive heterochromatin [22,23], the apparent absence of methylated H3K9 and HP1 would imply a unique combination of factors in the epigenome of granulocytes.
In the present investigation, we demonstrate that human and mouse granulocytes do possess methylated H3K9 (and other methylated histones), as well as low amounts of HP1 γ. The same spectrum of epigenetic markers (including detectable levels of HP1 α, β and γ) was observed in undifferentiated and granulocytic forms of retinoic acid (RA) treated mouse promyelocytic cells (MPRO) [24], which undergo complete and normal differentiation in vitro [25]. Therefore, excepting the reduced amounts of HP1 proteins, the global epigenome of granulocytes appears to be more consistent with other mammalian cell-types and various myeloid leukemias, than has been previously suggested [16]. In the present study, mouse blood and MPRO granulocytes revealed clear co-localization of trimethylated H3K9, trimethylated H4K20 and HP1 γ with pericentric heterochromatin. This observation agrees with earlier observations of co-localizated trimethylated H3K9 and trimethylated H4K20 in pericentric heterochromatin of various cultured mouse cells [23,26].
Results
Peripheral blood granulocytes possess various types of histone methylation
Wright-Giemsa stained smears of human or mouse peripheral blood reveal the well-described nuclear morphologies of neutrophilic granulocytes (Figure 1). Human neutrophils normally exhibit 3–4 lobes; mouse neutrophils frequently show ring-shaped nuclei with numerous NE nodules. In situ hybridization studies have clearly demonstrated that nodules in female human granulocytes (called "drumsticks") contain the inactive X chromosome [27]. However, the situation appears to be somewhat different with mouse granulocytes, since nodules are more frequent than the number of X chromosomes (unpublished observations).
Figure 1 Wright-Giemsa stained smears of human and mouse blood. Selected neutrophils are shown (top row, human; bottom row, mouse). Note that human neutrophils exhibit nuclear lobes frequently connected by thin strands. Mouse neutrophil nuclei are usually twisted ring-shapes, frequently exhibiting several small NE nodules. Scale bars: 10 μm.
Slides were prepared from purified human granulocytes for immunostaining with a variety of anti-methylated histone antibodies. Some of the results from conventional epifluorescent imaging are presented in Figure 2. The purified granulocytes were centrifuged onto polylysine-coated slides, fixed with PFA, permeabilized and immunostained. All of the anti-methylated histone antibodies gave positive staining; but with varying intensities and varying patterns of localization. Anti-di- and trimethyl H3K9 and anti-mono-, di- and trimethyl H3K27 appeared to stain granulocyte nuclei. The pattern of staining by anti-trimethyl H3K9 was especially intriguing, yielding several intensely stained spots within each granulocyte nucleus. Anti-di- and trimethyl H4K20 revealed a mixture of weak nuclear and cytoplasmic staining.
Figure 2 Immunostaining of human blood neutrophils with various methylated histone antibodies. Columns: DAPI (DNA stain); Cy3, antibody staining viewed by conventional epifluorescent microscopy. Abbreviations: monoMe, diMe and triMe, monomethylated, dimethylated and trimethylated lysine. Scale bar: 10 μm.
Immunoblots were performed on acid extracts of purified human blood granulocytes derived from one normal male and one normal female. A gallery of representative blots, including acid extracted HeLa cells, is presented in Figure 3. Similar granulocyte protein loads were employed in all blots; Hela protein loads were also unchanged. All of the anti-methylated histone antibodies gave positive staining of varying strengths and with varying intensities relative to the HeLa extract. An indication of reaction strength was obtained by noting the exposure times of the films. Consideration of both immunostaining and immunoblotting data yielded a qualitative assessment of the antibody reactions, and is summarized in Table 1. With respect to anti-trimethyl H3K9, it is conceivable that the contrasting results, comparing immunostaining with immunoblotting, reflects high local concentrations within the pericentric heterochromatin, but low total amounts of the protein modification in the nucleus.
Figure 3 Immunoblotting of human granulocyte extracts with various methylated histone antibodies. Samples: F, normal female; M, normal male; H, HeLa. Panels: a, dimethyl H3K9; b, trimethyl H3K9; c, monomethyl H3K27; d, dimethyl H3K27; e, trimethyl H3K27; f, dimethyl H4K20; g, trimethyl H4K20.
Table 1 Summmary Of Granulocyte Immunochemistry*.
ANTIBODY HUMAN BLOOD GRANULOCYTES MOUSE BLOOD GRANULOCYTES MOUSE MPRO CELLS**
dimethyl H3K9 + N + N + N
trimethyl H3K9 + N g + N p + N p
monomethyl H3K27 + N + N + N
dimethyl H3K27 + N + N + N
trimethyl H3K27 + N + N + N
dimethyl H4K20 +/- N, C +/- N +/- N
trimethyl H4K20 + N, C + N p + N p
HP1 α 0‡ 0 +/- N, C
HP1 β 0‡ 0 +/- N, C
HP1 γ +/- N, C + N p + N p
*The results listed represent a qualitative assessment, combining data from immunostaining and immunoblotting. In the case of normal mouse blood, due to inadequate enrichment of granulocytes, no immunoblotting was performed.
** Identical immunostaining and immunoblotting results were obtained with undifferentiated and RA differentiated MPRO cells.
Abbreviations: N, nuclear staining; N g, nuclear coarse granules; N p, nuclear pericentric heterochromatin; N, C, nuclear and cytoplasmic staining.
Reaction intensity: +, moderate to strong; +/-, weak; 0, negligible.
‡ Moderate to strong staining in the cytoplasm only.
Heparinized whole mouse blood, diluted with DMEM medium was centrifuged onto polylysine-coated slides, fixed with PFA, permeabilized and immunostained. Granulocytes were stained with a variety of anti-methylated histone antibodies and imaged by conventional epifluorescent microscopy (Figure 4). Varying levels of reactivity with all the antibodies was noted (summarized on Table 1). Of particular interest was the clear co-localization of anti-trimethylated H3K9 and trimethylated H4K20 with DAPI-bright pericentric heterochromatin, consistent with earlier observations on mouse cells [22,23,26,28]. Confocal images of mouse blood granulocytes doubly stained for lamin B and anti-trimethylated H3K9 or trimethylated H4K20 (Figure 5) demonstrate that NE nodules appear to contain these pericentric heterochromatin markers. No immunoblots were performed on mouse granulocytes, since we were unable to obtain purified fractions suitable for extraction.
Figure 4 Immunostaining of mouse blood neutrophils with methylated histone antibodies. Columns: DAPI (DNA stain); Cy3, antibody staining viewed by conventional epifluorescent microscopy. Abbreviations: monoMe, diMe and triMe, monomethylated, dimethylated and trimethylated lysine. Scale bar: 10 μm
Figure 5 Immunostaining of mouse blood neutrophils with methylated histone antibodies: confocal sections. Top row: anti-trimethyl H3K9 (left); anti-lamin B (middle); merged images (right). Bottom two rows: anti-trimethyl H4K20 (left); anti-lamin B (middle); merged images (right). Slides were fixed in methanol. Scale bar: 10 μm
MPRO granulocytes possess various types of histone methylation
The same panel of methylation-specific anti-histone antibodies was tested on undifferentiated and RA-differentiated MPRO cells by immunofluorescent and immunoblotting techniques. Figure 6 presents the results of immunostaining, with comparative staining by anti-lamin B and DAPI. As with mouse blood granulocytes (Figures 4 and 5), MPRO granulocytes (or undifferentiated cells) were strongly stained by anti-trimethyl H3K9 and anti-trimethyl H4K20 in bright spots that co-localized to DAPI-bright pericentric regions. More generalized nuclear staining was obtained with anti-dimethyl H3K9, anti-mono-, di- and trimethyl H3K27 and, less intensely, with anti-dimethyl H4K20.
Figure 6 Immunostaining of undifferentiated and granulocytic MPRO cells with various histone methylation antibodies. Panels: A, undifferentiated cells; B, granulocyte cells (4 days of RA). Columns: DAPI (DNA stain); Cy3, anti-methylated histone staining; Lamin B, FITC staining. Rows: monoMe, diMe and triMe, monomethylated, dimethylated and trimethylated lysines. Scale bar: 10 μm.
Immunoblotting experiments on acid extracts of MPRO cells (undifferentiated "O", and granulocytes "RA") are presented in Figure 7. As with the previous immunoblotting data on extracts of human granulocytes (Figure 3), the same protein loads were used in all gels and the reaction intensities shown are from different exposure times. The qualitative assessment of antibody reactivity, combining immunostaining and immunoblotting data is presented in Table 1. No systematic changes in reaction intensities were observed comparing undifferentiated with RA-treated granulocytic MPRO cell extracts. Also, as described for human granulocyte extracts, anti-trimethyl H3K9 displayed very weak immunoblotting reactivity. Again we suggest that the low cellular level of trimethyl H3K9 may yield strong immunostaining because of a high local concentration in the pericentric heterochromatin.
Figure 7 Immunoblotting of MPRO granulocyte extracts with various histone methylation antibodies. Samples: O, undifferentiated; RA differentiated granulocytes; H, HeLa. Panels: a, dimethyl H3K9; b, trimethyl H3K9; c, monomethyl H3K27; d, dimethyl H3K27; e, trimethyl H3K27; f, dimethyl H4K20; g, trimethyl H4K20.
Peripheral blood granulocytes are deficient in HP1 proteins
Because of difficulties that we have experienced employing commercial anti-HP1 antibodies to immunostain blood granulocytes, we decided to test the antibodies on mouse NIH 3T3 cells, where numerous authors have demonstrated localization of anti-HP1 on DAPI-bright interphase pericentric heterochromatin [7,28-30]. Results with our best antibodies for immunostaining (Chemicon anti-HP1 α and γ) of NIH 3T3 cells are presented in Figure 8. Our images closely resemble previously published data [7,30], indicating a co-localization of HP1 α and γ on pericentric heterochromatin.
Figure 8 Immunostaining of mouse NIH 3T3 cells with HP1 antibodies. Columns: DAPI (DNA staining); Cy3, antibody staining of HP1 viewed by conventional epifluorescent microscopy; Lamin B, FITC staining. Note the clear co-localization of HP1 staining with the DAPI-bright pericentric heterochromatin. Scale bar: 10 μm
Examples of HP1 α, β and γ immunostained human "buffy coat" preparations are presented in Figure 9. (Euromedex anti-HP1 antibodies were employed in this experiment.) Granulocytes of the buffy coat showed no nuclear staining by anti-HP1 α and β and only weak staining with anti-HP1 γ. Anti-HP1 α and β appeared to yield a weak reaction with granulocyte cytoplasmic granules, which was also seen using anti-HP1 γ. These granular structures do not derive from the secondary antibody (data not shown). Mononuclear cells (lymphocytes and monocytes) of the buffy coat provided an internal positive control for nuclear staining by anti-HP1 antibodies.
Figure 9 Immunostaining of human blood buffy coat with various HP1 antibodies. Columns: Cy3, antibody staining viewed by conventional epifluorescent microscopy; DAPI (DNA staining). Note that the monocytic nuclei exhibit staining with all monoclonal anti-HP1 antibodies (Chemicon); whereas, the granulocytic nuclei only reveal staining with anti-HP1 γ. Anti-HP1 α, β and γ appear to stain cytoplasmic granules. Scale bar: 10 μm.
Immunoblotting experiments with whole cell extracts of human granulocytes (Figure 10) generally agreed with the immunostaining experiments; i.e., no clear evidence of HP1 α or β, an indication of trace amounts of HP1 γ. Different sources of anti-HP1 β indicated the possibility of a lower molecular weight cross-reacting antigen (the cytoplasmic granules?), which was not studied further.
Figure 10 Immunoblotting of human granulocyte extracts with various HP1 antibodies. Samples: F, normal female; M, normal male; H, HeLa. Panels: a, monoclonal anti-HP1 α (Upstate); b, monoclonal anti-HP1 β (Chemicon); c, rabbit anti-HP1 β (Upstate); d, monoclonal anti-HP1 γ (Chemicon). Note the trace reaction of anti-HP1 γ with the granulocyte extract (panel d).
Immunostaining experiments with anti-HP1 antibodies were performed on buffy coat preparations from whole mouse blood (Figure 11). Results clearly indicated negligible staining by anti-HP1 α or β of mouse granulocytes, with clear nuclear staining of blood monocytic nuclei. However, anti-HP1 γ yielded granulocyte nuclear staining, somewhat brighter in pericentric heterochromatin; monocytic nuclei were also stained.
Figure 11 Immunostaining of mouse blood buffy coat with various HP1 antibodies. Columns: DAPI (DNA staining); Cy3, antibody staining viewed by conventional epifluorescent microscopy. Note that the monocytic nuclei exhibit staining with all monoclonal anti-HP1 antibodies (Chemicon). Granulocytic nuclei only reveal staining with anti-HP1 γ, which appears more concentrated at pericentric heterochromatin. Scale bar: 10 μm.
In summary (see Table 1), we conclude that peripheral blood granulocytes possess negligible amounts of HP1 α or β, and low levels of HP1 γ. On the other hand, blood mononuclear cells do clearly stain for all HP1 isoforms.
MPRO granulocytes possess HP1 proteins
Immunostaining experiments were performed on undifferentiated and granulocytic MPRO cells (Figure 12). The most convincing nuclear staining was with anti-HP1 γ, which yielded a combination of diffuse nuclear staining and focal staining, which co-localizes with DAPI-bright pericentric heterochromatin (panels A and B) and anti-trimethyl H4K20 (panel C). Both anti-HP1 α and β yielded weak staining in the nucleus and the cytoplasm. Immunoblotting experiments were conducted upon extracts of undifferentiated and RA-treated (4 days) granulocytic MPRO cells (Figure 13). An extract of Hela cells was included as a control. Convincing reactions with MPRO extracts were observed with anti-HP1 β and γ. [Anti-HP1 α (Upstate) yielded no convincing reactivity with MPRO or HeLa extracts.] MPRO HP1 β appears to possess a larger molecular weight than measured in HeLa extracts (MPRO, ~33 kD; Hela, ~28 kD) and does not appear to decrease during the differentiation process. The apparent molecular weight of HeLa and MPRO HP1 γ, as determined on 12% acrylamide SDS gels, was ~23–24 kD. Since the protein loads for the MPRO lanes was approximately the same (Figure 13, panel a), the declining reactivity comparing O and RA, suggests that the cellular level of HP1γ decreases during the granulocytic differentiation process. In summary, immunostaining and immunoblotting studies support the presence of HP1 γ in MPRO granulocyte nuclei, with weaker evidence pertaining to HP1 α and β (Table 1).
Figure 12 Immunostaining of MPRO cells with various HP1 antibodies. Panels: A, undifferentiated; B, granulocyte forms of MPRO. Columns: FITC, antibody staining viewed by conventional epifluorescent microscopy; DAPI (DNA staining). Cells were stained with anti-HP1 α, β and γ. Note the co-localization of anti-HP1 γ with DAPI-bright pericentric heterochromatin. Panel C: confocal images demonstrating co-localization of anti-trimethylated H4K20 with anti-HP1 γ stained granulocyte pericentric heterochromatin. Scale bars: 10 μm.
Figure 13 Immunoblotting of MPRO extracts with various HP1 antibodies. Samples: H, HeLa; O, MPRO undifferentiated cells; RA, MPRO granulocytes (differentiated 4 days). Panels: a, Coomassie blue stained membrane; b, monoclonal anti-HP1 β (Chemicon); c, monoclonal anti-HP1 γ (Chemicon). Note that MPRO HP1 β appears to be of higher mol wt, than that of HeLa cells. Also note the apparent reduction of the amount of MPRO HP1 γ during RA-induced differentiation.
Discussion
Blood neutrophilic granulocytes are regarded as terminally differentiated cells which are destined to undergo apoptosis, but can be activated into a short-lived migratory phagocytotic form by the presence of bacterial or yeast infections [8]. Mature neutrophils possess considerable quantities of heterochromatin [9], suggesting transcriptional repression. However, recent studies have clearly demonstrated that neutrophil activation by bacteria or bacterial lipopolysaccharide results in rapid and extensive changes in gene expression [31-33]. Our laboratory has focused upon the cellular mechanisms of nuclear shape change and heterochromatin distribution occurring during granulopoiesis [11,13-15,34,35].
The cellular mechanisms producing progressive chromatin condensation during granulopoiesis are not understood. It seems reasonable to assume that described features of constitutive and facultative heterochromatin [21,23,26,36-41] apply to the situation of condensing granulocyte chromatin. Some of these features include: 1) hypoacetylation of histone lysines; 2) methylation of specific histone lysine residues (e.g., H3K9, H3K27 and H4K20); 3) DNA methylation; 4) chromatin binding by repressive proteins, such as HP1 α, β and γ. Our present analysis of granulocyte nuclei only focuses upon two global epigenetic features, repressive histone methylation and HP1.
Repressive epigenetic features: histone methylation and HP1
Only one published study [16] has examined global epigenetic features of granulocyte heterochromatin. This study concluded that normal human neutrophils are deficient in methylated H3K9 and HP1 α, β and γ. Based upon our present study, we agree that there is a paucity of HP1 α and β in normal human and mouse neutrophils, but we can detect low amounts of nuclear HP1 γ. Futhermore, undifferentiated and granulocytic forms of MPRO (mouse promyelocytic) cells reveal detectable levels of HP1 α, β and γ. It might be argued that MPRO granulocytes are not normal differentiated cells, but resemble leukemic cells, which do contain all isoforms of HP1 [16]. However, MPRO cells are not derived from leukemic mice and the in vitro granulocytic differentiation induced by RA is believed to yield mature granulocytes [25], with a gene expression pattern closely mimicking normal granulopoiesis [42,43].
A possible explanation for the apparent absence of HP1 α and β in normal human neutrophils is degradation during granulocyte preparation and extraction for immunoblotting. Proteolysis is a definite problem when working with granulocyte preparations, even in the presence of high quantities of protease inhibitors (unpublished observations). Whatever the reasons, our evidence indicated that low levels of HP1 γ are detectable in normal granulocytes. The evident paucity of HP1 in the nuclei of mature neutrophils suggests that models postulating an essential role for HP1 in the establishment of heterochromatin and its adherence to the NE [20] may not be applicable to these cell-types. On this issue, we are in complete agreement with the earlier published discussion [16]. We cannot rule out the possibility that HP1 proteins present early during granulopoiesis fulfil a role in the establishment of heterochromatin, which subsequently might be maintained by low amounts of HP1 γ in the mature granulocytes.
The major disagreement between our study and the previous publication [16] concerns the existence of various isoforms of methylated H3K9. We describe immunofluorescent and immunoblotting data supporting the presence of repressive methylated H3K9, H3K27 and H4K20 in normal human and mouse granulocytes and MPRO undifferentiated and granulocytic forms. The most dramatic images were obtained with trimethyl H3K9 and trimethyl H4K20 immunostained mouse blood and MPRO granulocytes. In these situations, clear co-localization with DAPI-bright pericentric heterochromatin was demonstrated. Anti-trimethyl H3K9 yielded similar intensely stained spots within human granulocyte nuclei. But the pattern of DAPI staining in human granulocytes is quite different than observed in mouse granulocyte nuclei (compare Figures 2 and 4). In human granulocyte nuclei, the DAPI-stained heterochromatin forms a thick layer under the nuclear envelope and central masses within the nuclear lobes. Several of the methylated histone antibodies (anti-dimethyl H3K9 and anti-mono-, di- and trimethyl H3K27) appeared to generally stain human granulocyte heterochromatin.
Formation of heterochromatin in the apparent absence of HP1 has been reported previously [7]. Nucleated chicken erythrocytes possess negligible amounts of HP1 α, β and γ, reduced trimethyl H3K9 and very low amounts of trimethyl H3K27. However, reasonable levels of dimethyl H3K9 were observed. In the same study, the authors demonstrated significant levels of HP1 α, β and γ and trimethyl H3K9, but reduced trimethyl H3K27 in mouse embryonic erythrocyte nuclei. More recently, in a study of resting mouse lymphocytes [6], the authors describe low levels of HP1 β, mono-, di- and trimethylated H3K9, trimethylated H3K27 and di- and trimethylated H4K20. Activation of the lymphocytes resulted in increases in the levels of all these epigenetic marks. In a variety of mouse tissue culture cells, co-localization of trimethyl H3K9 and trimethyl H4K20 on pericentric heterochromatin has been clearly demonstrated [23,26]. A model for sequential trimethylation of H3K9 and H4K20 on pericentric heterochromatin was presented [26], involving a stabilizing role for HP1 α and β binding to trimethyl H3K9 and recruiting Suv4-20h to trimethylate H4K20. It remains to be demonstrated whether HP1 γ could also serve as the postulated intermediary, in the absence of HP1 α and β.
Histone variants and DNA methylation
An additional epigenetic feature of granulocyte nuclei, described by us in an earlier publication [35], pertains to histone H1 subtypes and phosphorylation. In a comparison of human neutrophils to leukemic HL-60 cells, we concluded that granulocytic differentiation results in dephosphorylation of subtypes H1.4, H1.5 and H1.2. Furthermore, subtype H1.3 was observed in acid extracts of normal human granulocytes, but not in those of undifferentiated or RA differentiated HL-60 cells.
There is virtually no published information on the possible role of global DNA methylation in the establishment of granulocyte heterochromatin. Given what is known from other cell types concerning CpG methylation of pericentric heterochromatin [44,45], it seems reasonable to assume a similar role for DNA methylation in granulocytic cells. However, studies on the patterns and inhibition of DNA methylation during granulocytic differentiation of HL-60 cells suggest that methylation is not essential [46,47].
Neutrophil nuclear deformability
The present experiments, combined with earlier observations, provide a basis for understanding the mechanism of granulocyte nuclear drumstick or nodule formation observed in human or mouse blood smears. It appears that nodules often contain pericentric heterochromatin (mouse) or the inactive X chromosome (human [27]). We have previously suggested that the NE of granulocytes is highly deformable due to low levels of lamins A and B1 [11,15]. This view is based upon observations on HL-60 cells [15,34,48] and earlier studies of mouse granulocytes [49,50]. Absence of lamin A in the NE has been shown to increase NE deformability in mechanical strain tests [51], consistent with conclusions from laminopathies [12]. There is now clear and convincing evidence that the composition and integrity of the lamin polymer strongly influences the NE strength [52]. Our view is that drumsticks and nodules are artifacts of the flattening of granulocyte nuclei, reflecting the more rigid heterochromatic foci pushing against a deformable NE.
It seems reasonable to speculate that the altered composition of the granulocyte NE (i.e., paucity of lamins A and B1, and elevation of LBR [35]) is an essential part of the mechanism of nuclear lobulation [15] or formation of ring-shaped nuclei [10,11]. Both the increased deformability of the NE and nuclear lobulation (or ring-shape) may facilitate migration of activated neutrophils through blood vessel endothelia and tight tissue spaces towards a site of infection. The importance of granulocyte nuclear lobulation for chemotactic migration through narrow pores and tight spaces was demonstrated in a study of individuals with Pelger Huet anomaly [53]. We have shown NE deficiency of LBR in human Pelger Huet anomaly [13] and mouse ichthyosis [14], resulting in granulocyte nuclear hypolobulation. One additional functional consequence of increased granulocyte NE deformability may be to facilitate the formation of neutrophil extracellular traps (NETs) [54]. These structures appear to be an extruded complex of chromatin and elastase that traps and degrades invading bacteria.
Conclusion
Mature blood neutrophils possess considerable heterochromatin containing a variety of repressive histone methylation markers: di- and trimethylated H3K9; mono-, di- and trimethyl H3K27; di- and trimethyl H4K20). In addition, normal neutrophils exhibit negligible amounts of HP1 α and β, but reveal detectable levels of HP1 γ. Clear co-localization of trimethylated H3K9, trimethylated H4K20 and HP1 γ on pericentric heterochromatin was demonstrated in normal mouse blood neutrophils and granulocytic forms of mouse MPRO cells.
Methods
Cells enrichment and cultivation
Human peripheral blood preparations (from one normal male and one normal female) were enriched for granulocytes and monocytes by density gradient centrifugation with HISTOPAQUE 1119 and 1077 (Sigma-Aldrich, Inc.), following procedures described by the supplier. Granulocytic and monocytic fractions were counted and examined for purity using Wright-Giemsa stain (Sigma-Aldrich, Inc., St. Louis MO). For acid extraction of histones (see below), the purified granulocytes were washed in RPMI medium and used directly. For immunostaining (see below), the granulocyte and monocyte fractions were pooled yielding a "buffy coat", which was centrifuged onto polylysine-coated slides.
Mouse blood was obtained in several ways. Separate blood pools of adult NMRI male or female mice, collected in EDTA-coated syringes, were centrifuged in HISTOPAQUE gradients. However, we routinely observed that the granulocyte and monocyte bands were cross-contaminated. Therefore, the two fractions were combined to yield a "buffy coat", which was suitable for immunostaining, but not for acid extraction of histones. In other immunostaining experiments with adult male CD1 mice, small quantities (~80–100 μl) of blood were collected in heparinized micro-hematocrit capillary tubes (Fisher Scientific, Pittsburgh PA). The contents were expelled into 1.0 ml of DMEM and centrifuged onto polylysine-coated slides.
MPRO (mouse promyelocytic) cells are a male cell line derived from normal bone marrow and grown in medium containing GM-CSF [24,25]. Following the addition of 10 μM RA, MPRO cells differentiate during four days into apparently normal granulocytes with a mixture of ring-shaped and lobulated nuclear forms [11]. Acid and SDS extracts were prepared for immunoblots during the course of RA-induced differentiation; immunostaining was performed on cells centrifuged onto polylysine-coated slides.
Antibodies and immunostaining
Antibodies were obtained from a variety of private and commercial sources. Our experience, especially with commercial antibodies, is that they frequently do not react as described and some exhibit different reactivities when the same antibody is obtained in different lots. The most reliable anti-methylated histone antibodies were generously provided by Thomas Jenuwein (Research Institute of Molecular Pathology, University of Vienna, Vienna, Austria) and by David Allis (The Rockefeller University, New York City), and are currently available from Upstate (Charlottesville, VA). These are all rabbit antibodies, with the following names and Upstate catalogue numbers: anti-dimethyl H3K9, #07-441; anti-trimethyl H3K9, #07-442; anti-monomethyl H3K27, #07-448; anti-dimethyl H3K27, #07-452; anti-trimethyl H3K27, #07-449; anti-dimethyl H4K20, #07-031; anti-trimethyl H4K20, #07-463. Antibodies against the HP1 isoforms were purchased from both Upstate and Chemicon International (Temecula, CA) at various times and with varying success. We never obtained acceptable immunostaining with Upstate anti-HP1 α (#05-689) and (#07-346) or anti-HP1 β (#07-333), although the anti-HP1 β was useful for immunoblots. Only the Chemicon anti-HP1 antibodies provided useful immunostaining. Mouse anti-HP1 α (#MAB3584) and anti-HP1 γ (#MAB3450) yielded clear immunostaining of NIH 3T3 cells (Figure 8), but only the anti-HP1 γ was suitable for immunoblotting. Chemicon anti-HP1 β (MAB 3448, lot # 24120319) did not work for immunofluorescence or immunoblotting, although the same antibody obtained earlier (Euromedex, Mundolsheim, France; catalogue # 1MOD-1A9-AS), while visiting the German Cancer Research Center, gave clean immunofluorescent staining on NIH 3T3 cells. Goat anti-lamin B was obtained from Santa Cruz Biotechnology (Santa Cruz, CA; catalogue # sc-6216). We have purchased this antibody several times since 1997. Recently this antibody is described as anti-lamin B1. Early lots of this antibody give rim staining of human granulocyte NE; the more recent lots have not given rim staining. We suspect that the earlier lots reacted with both lamins B1 and B2, whereas the more recent lots are more specific for lamin B1. The results presented in this paper (Figures 5, 6 and 8) utilize the earlier lots.
Immunostaining procedures have been described previously [11,15,35]. In brief, cells were centrifuged onto fresh polylysine-coated slides and the cells fixed in 4% formaldehyde (PFA) in PBS for 15 min, or anhydrous methanol (-20°, 10 min). PFA-fixed cells (most experiments) were permeabilized with 0.1% Triton X-100, washed in PBS and blocked with 5% normal donkey serum in PBS. Methanol-fixed cells (Figure 5) went directly into PBS, followed by blocking. Primary antibody dilutions followed suggestions by the supplier. Secondary antibodies were obtained from Jackson ImmunoResearch Laboratory (West Grove PA).
Cell extracts and immunoblotting
The immunoblotting procedure has been described earlier [35]. Most of the SDS-PAGE experiments were performed with 15 or 12% precast BioRad gels. Secondary HRP-conjugated antibodies were obtained from Jackson ImmunoResearch Laboratory. Several different ECL exposures were collected on X-ray film and subsequently scanned with a BioRad Chemi Doc.
Abbreviations
RA, retinoic acid; NE, nuclear envelope; MPRO, mouse promyelocytic cells; LBR, lamin B receptor
Authors' contributions
ALO performed the microscopy and prepared the figures. DEO performed the tissue culture, immunostaining and immunoblotting. Both authors were involved in the conception of the study and have read and approved the final manuscript.
Acknowledgements
This work was supported by an R15 grant from NIHHLBI and support from the Department of Biology, Bowdoin College. Some of these studies were initiated while ALO and DEO were visitors at the German Cancer Research Center (Heidelberg, Germany), hosted by P. Lichter and H. Herrmann. We express our gratitude to them. These studies would not have been possible without the generosity of T. Jenuwein (Research Institute of Molecular Pathology, University of Vienna, Vienna, Austria) and D. Allis (The Rockefeller University, New York City), and members of their laboratories with whom we communicated about our technical successes and failures. To all of them we are most grateful. We thank P. Newburger (University of Massachusetts Medical Center at Worchester) for the gift of an acid extract of human granulocytes. We also thank M. Nessling (German Cancer Research Center) for providing us with aliquots of Euromedex anti-HP1 α, β and γ.
==== Refs
Cho KS Elizondo LI Boerkoel CF Advances in chromatin remodeling and human disease Curr Opin Genet Dev 2004 14 308 315 15172675 10.1016/j.gde.2004.04.015
Egger G Liang G Aparicio A Jones PA Epigenetics in human disease and prospects for epigenetic therapy Nature 2004 429 457 463 15164071 10.1038/nature02625
Jiang YH Bressler J Beaudet AL Epigenetics and human disease Annu Rev Genomics Hum Genet 2004 5 479 510 15485357 10.1146/annurev.genom.5.061903.180014
Lund AH van Lohuizen M Epigenetics and cancer Genes Dev 2004 18 2315 2335 15466484 10.1101/gad.1232504
Cheung P Lau P Epigenetic regulation by histone methylation and histone variants Mol Endocrinol 2005 19 563 573 15677708 10.1210/me.2004-0496
Baxter J Sauer S Peters A John R Williams R Caparros ML Arney K Otte A Jenuwein T Merkenschlager M Fisher AG Histone hypomethylation is an indicator of epigenetic plasticity in quiescent lymphocytes Embo J 2004 23 4462 4472 15510223 10.1038/sj.emboj.7600414
Gilbert N Boyle S Sutherland H de Las Heras J Allan J Jenuwein T Bickmore WA Formation of facultative heterochromatin in the absence of HP1 Embo J 2003 22 5540 5550 14532126 10.1093/emboj/cdg520
Lee GR Foerster J Lukens J Paraskevas F Greer JP Rogers GM Wintrobe's Clinical Hematology 1999 10th Baltimore, Williams & Wilkins
Bainton DF Ullyot JL Farquhar MG The development of neutrophilic polymorphonuclear leukocytes in human bone marrow J Exp Med 1971 134 907 934 4106490 10.1084/jem.134.4.907
Biermann H Pietz B Dreier R Schmid KW Sorg C Sunderkotter C Murine leukocytes with ring-shaped nuclei include granulocytes, monocytes, and their precursors J Leukoc Biol 1999 65 217 231 10088605
Olins AL Olins DE The mechanism of granulocyte nuclear shape determination: possible involvement of the centrosome Eur J Cell Biol 2005 84 181 188 15819399 10.1016/j.ejcb.2004.12.021
Gruenbaum Y Margalit A Goldman RD Shumaker DK Wilson KL The nuclear lamina comes of age Nat Rev Mol Cell Biol 2005 6 21 31 15688064 10.1038/nrm1550
Hoffmann K Dreger CK Olins AL Olins DE Shultz LD Lucke B Karl H Kaps R Muller D Vaya A Aznar J Ware RE Sotelo Cruz N Lindner TH Herrmann H Reis A Sperling K Mutations in the gene encoding the lamin B receptor produce an altered nuclear morphology in granulocytes (Pelger-Huet anomaly) Nat Genet 2002 31 410 414 12118250
Shultz LD Lyons BL Burzenski LM Gott B Samuels R Schweitzer PA Dreger C Herrmann H Kalscheuer V Olins AL Olins DE Sperling K Hoffmann K Mutations at the mouse ichthyosis locus are within the lamin B receptor gene: a single gene model for human Pelger-Huet anomaly Hum Mol Genet 2003 12 61 69 12490533 10.1093/hmg/ddg003
Olins AL Olins DE Cytoskeletal influences on nuclear shape in granulocytic HL-60 cells BMC Cell Biol 2004 5 30 15317658 10.1186/1471-2121-5-30
Lukasova E Koristek Z Falk M Kozubek S Grigoryev S Kozubek M Ondrej V Kroupova I Methylation of histones in myeloid leukemias as a potential marker of granulocyte abnormalities J Leukoc Biol 2005 77 100 111 15507473
Ye Q Worman HJ Interaction between an integral protein of the nuclear envelope inner membrane and human chromodomain proteins homologous to Drosophila HP1 J Biol Chem 1996 271 14653 14656 8663349 10.1074/jbc.271.25.14653
Ye Q Callebaut I Pezhman A Courvalin JC Worman HJ Domain-specific interactions of human HP1-type chromodomain proteins and inner nuclear membrane protein LBR J Biol Chem 1997 272 14983 14989 9169472 10.1074/jbc.272.23.14983
Ye Q Barton RM Worman HJ Nuclear lamin-binding proteins Subcell Biochem 1998 31 587 610 9932507
Singh PB Georgatos SD HP1: facts, open questions, and speculation J Struct Biol 2002 140 10 16 12490149 10.1016/S1047-8477(02)00536-1
Peters AH Kubicek S Mechtler K O'Sullivan RJ Derijck AA Perez-Burgos L Kohlmaier A Opravil S Tachibana M Shinkai Y Martens JH Jenuwein T Partitioning and plasticity of repressive histone methylation states in mammalian chromatin Mol Cell 2003 12 1577 1589 14690609 10.1016/S1097-2765(03)00477-5
Lehnertz B Ueda Y Derijck AA Braunschweig U Perez-Burgos L Kubicek S Chen T Li E Jenuwein T Peters AH Suv39h-mediated histone H3 lysine 9 methylation directs DNA methylation to major satellite repeats at pericentric heterochromatin Curr Biol 2003 13 1192 1200 12867029 10.1016/S0960-9822(03)00432-9
Martens JH O'Sullivan RJ Braunschweig U Opravil S Radolf M Steinlein P Jenuwein T The profile of repeat-associated histone lysine methylation states in the mouse epigenome Embo J 2005 24 800 812 15678104 10.1038/sj.emboj.7600545
Tsai S Collins SJ A dominant negative retinoic acid receptor blocks neutrophil differentiation at the promyelocyte stage Proc Natl Acad Sci U S A 1993 90 7153 7157 8394011
Lawson ND Krause DS Berliner N Normal neutrophil differentiation and secondary granule gene expression in the EML and MPRO cell lines Exp Hematol 1998 26 1178 1185 9808058
Schotta G Lachner M Sarma K Ebert A Sengupta R Reuter G Reinberg D Jenuwein T A silencing pathway to induce H3-K9 and H4-K20 trimethylation at constitutive heterochromatin Genes Dev 2004 18 1251 1262 15145825 10.1101/gad.300704
Karni RJ Wangh LJ Sanchez JA Nonrandom location and orientation of the inactive X chromosome in human neutrophil nuclei Chromosoma 2001 110 267 274 11534818
Guenatri M Bailly D Maison C Almouzni G Mouse centric and pericentric satellite repeats form distinct functional heterochromatin J Cell Biol 2004 166 493 505 15302854 10.1083/jcb.200403109
Minc E Allory Y Worman HJ Courvalin JC Buendia B Localization and phosphorylation of HP1 proteins during the cell cycle in mammalian cells Chromosoma 1999 108 220 234 10460410 10.1007/s004120050372
Muchardt C Guilleme M Seeler JS Trouche D Dejean A Yaniv M Coordinated methyl and RNA binding is required for heterochromatin localization of mammalian HP1alpha EMBO Rep 2002 3 975 981 12231507 10.1093/embo-reports/kvf194
Newburger PE Subrahmanyam YV Weissman SM Global analysis of neutrophil gene expression Curr Opin Hematol 2000 7 16 20 10608499 10.1097/00062752-200001000-00004
Subrahmanyam YV Yamaga S Prashar Y Lee HH Hoe NP Kluger Y Gerstein M Goguen JD Newburger PE Weissman SM RNA expression patterns change dramatically in human neutrophils exposed to bacteria Blood 2001 97 2457 2468 11290611 10.1182/blood.V97.8.2457
Tsukahara Y Lian Z Zhang X Whitney C Kluger Y Tuck D Yamaga S Nakayama Y Weissman SM Newburger PE Gene expression in human neutrophils during activation and priming by bacterial lipopolysaccharide J Cell Biochem 2003 89 848 861 12858349 10.1002/jcb.10526
Olins AL Buendia B Herrmann H Lichter P Olins DE Retinoic acid induction of nuclear envelope-limited chromatin sheets in HL-60 Exp Cell Res 1998 245 91 104 9828104 10.1006/excr.1998.4210
Olins AL Herrmann H Lichter P Kratzmeier M Doenecke D Olins DE Nuclear envelope and chromatin compositional differences comparing undifferentiated and retinoic acid- and phorbol ester-treated HL-60 cells Exp Cell Res 2001 268 115 127 11478838 10.1006/excr.2001.5269
Craig JM Heterochromatin--many flavours, common themes Bioessays 2005 27 17 28 15612037 10.1002/bies.20145
Grewal SI Elgin SC Heterochromatin: new possibilities for the inheritance of structure Curr Opin Genet Dev 2002 12 178 187 11893491 10.1016/S0959-437X(02)00284-8
Sims RJ Nishioka K Reinberg D Histone lysine methylation: a signature for chromatin function Trends Genet 2003 19 629 639 14585615 10.1016/j.tig.2003.09.007
Lachner M O'Sullivan RJ Jenuwein T An epigenetic road map for histone lysine methylation J Cell Sci 2003 116 2117 2124 12730288 10.1242/jcs.00493
Maison C Almouzni G HP1 and the dynamics of heterochromatin maintenance Nat Rev Mol Cell Biol 2004 5 296 304 15071554 10.1038/nrm1355
Bird A DNA methylation patterns and epigenetic memory Genes Dev 2002 16 6 21 11782440 10.1101/gad.947102
Lian Z Wang L Yamaga S Bonds W Beazer-Barclay Y Kluger Y Gerstein M Newburger PE Berliner N Weissman SM Genomic and proteomic analysis of the myeloid differentiation program Blood 2001 98 513 524 11468144 10.1182/blood.V98.3.513
Lian Z Kluger Y Greenbaum DS Tuck D Gerstein M Berliner N Weissman SM Newburger PE Genomic and proteomic analysis of the myeloid differentiation program: global analysis of gene expression during induced differentiation in the MPRO cell line Blood 2002 100 3209 3220 12384419 10.1182/blood-2002-03-0850
Yan Q Cho E Lockett S Muegge K Association of Lsh, a regulator of DNA methylation, with pericentromeric heterochromatin is dependent on intact heterochromatin Mol Cell Biol 2003 23 8416 8428 14612388 10.1128/MCB.23.23.8416-8428.2003
Su RC Brown KE Saaber S Fisher AG Merkenschlager M Smale ST Dynamic assembly of silent chromatin during thymocyte maturation Nat Genet 2004 36 502 506 15098035 10.1038/ng1351
Momparler RL Bouchard J Samson J Induction of differentiation and inhibition of DNA methylation in HL-60 myeloid leukemic cells by 5-AZA-2'-deoxycytidine Leuk Res 1985 9 1361 1366 2417065 10.1016/0145-2126(85)90123-7
Lubbert M Salser W Prokocimer M Miller CW Thomason A Koeffler HP Stable methylation patterns of MYC and other genes regulated during terminal myeloid differentiation Leukemia 1991 5 533 539 1649355
Yabuki M Miyake T Doi Y Fujiwara T Hamazaki K Yoshioka T Horton AA Utsumi K Role of nuclear lamins in nuclear segmentation of human neutrophils Physiol Chem Phys Med NMR 1999 31 77 84 10816760
Rober RA Gieseler RK Peters JH Weber K Osborn M Sauter H Induction of nuclear lamins A/C in macrophages in in vitro cultures of rat bone marrow precursor cells and human blood monocytes, and in macrophages elicited in vivo by thioglycollate stimulation
Cells of the cellular immune and hemopoietic system of the mouse lack lamins A/C: distinction versus other somatic cells Exp Cell Res 1990 190 185 194 2209722 10.1016/0014-4827(90)90184-C
Rober RA Sauter H Weber K Osborn M Cells of the cellular immune and hemopoietic system of the mouse lack lamins A/C: distinction versus other somatic cells J Cell Sci 1990 95 ( Pt 4) 587 598 2200797
Lammerding J Schulze PC Takahashi T Kozlov S Sullivan T Kamm RD Stewart CL Lee RT Lamin A/C deficiency causes defective nuclear mechanics and mechanotransduction J Clin Invest 2004 113 370 378 14755334 10.1172/JCI200419670
Schirmer EC Gerace L The stability of the nuclear lamina polymer changes with the composition of lamin subtypes according to their individual binding strengths J Biol Chem 2004 279 42811 42817 15284226 10.1074/jbc.M407705200
Park BH Dolen J Snyder B Defective chemotactic migration of polymorphonuclear leukocytes in Pelger-Huet anomaly Proc Soc Exp Biol Med 1977 155 51 54 323871
Brinkmann V Reichard U Goosmann C Fauler B Uhlemann Y Weiss DS Weinrauch Y Zychlinsky A Neutrophil extracellular traps kill bacteria Science 2004 303 1532 1535 15001782 10.1126/science.1092385
|
16287503
|
PMC1310526
|
CC BY
|
2021-01-04 16:31:30
|
no
|
BMC Cell Biol. 2005 Nov 15; 6:39
|
utf-8
|
BMC Cell Biol
| 2,005 |
10.1186/1471-2121-6-39
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-661630574410.1186/1471-2202-6-66Research ArticleSubstance P selectively decreases paired pulse depression in the rat hippocampal slice Wease Kerrie N [email protected] Stephen N [email protected] Department of Biomedical Sciences, Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD, Scotland, UK2005 23 11 2005 6 66 66 27 6 2005 23 11 2005 Copyright © 2005 Wease and Davies; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Although being widespread in the hippocampus, the role tachykinins play in synaptic transmission is unclear. The effect of substance P on field potentials evoked by stimulation of the Schaffer collateral-commissural fibres and recorded from the CA1 region of the rat hippocampal slice were studied.
Results
Perfusion of substance P (8 μM) had no effect on the fEPSP or population spike. Substance P did however cause a selective reduction in the paired pulse depression of population spikes evoked by paired stimulation at interpulse intervals of 20–80 msec. A comparison of the actions of other tachykinin receptor agonists gave an order of potency of substance P > [β-Ala8]-neurokinin A (4–10) > senktide. The effect of substance P was reduced by the neurokinin-1 receptor antagonist SR140333, but not by the neurokinin-2 or neurokinin-3 receptor antagonists, MDL 29,913 or [Trp7, β-Ala8]-neurokinin A (4–10).
Conclusion
The order of potency of the agonists, and the effects of the antagonists, both indicate that the effect of substance P on paired pulse depression is mediated by neurokinin-1 receptors.
==== Body
Background
The mammalian tachykinins are a group of peptides sharing the common C-terminal sequence Phe-X-Gly-Leu-Met-NH2. The three principal tachykinins are substance P, neurokinin A and neurokinin B, and although these are preferred agonists for the neurokinin-1, neurokinin-2 and neurokinin-3 receptors respectively, they are not completely selective for any one receptor subtype [1,2]. Tachykinin receptors are distributed throughout the CNS, with all three receptor subtypes being expressed in the adult rat hippocampus [3-6]. A dense network of fibres containing substance P innervates the stratum oriens, stratum radiatum and alveus of the rat hippocampus. These may arise from both extrinsic sources such as the septum and hypothalamus, and from intrinsic GABA-containing interneurones [7,8].
Although being widespread in the hippocampus, the role tachykinins play in normal synaptic transmission is unclear. Using extracellular recordings from the mouse hippocampal slice, substance P and its analogue substance P methyl ester have been reported to cause a decrease in the amplitude and slope of the field excitatory postsynaptic potential (fEPSP) recorded from the CA1 stratum pyramidale [9]. The effect was blocked by the selective neurokinin-1 receptor antagonist SR140333, suggesting the action was NK-1 receptor mediated. The effect of substance P methyl ester was blocked by bicuculline, an antagonist for GABAA receptors, and not by glutamate receptor antagonists. The authors concluded the depressant effect of substance P and substance P methyl ester required an intact GABAergic system, with substance P causing facilitation of GABAergic neurotransmission, thereby increasing inhibitory synaptic transmission [9].
The aim of this present study was to use extracellular field recordings to a) identify the effect of substance P on synaptic transmission in the CA1 region of the rat hippocampus, and b) to use selective pharmacological agonists and antagonists to determine which tachykinin receptors were involved.
Results
Substance P had no effect on fEPSP's
fEPSPs were recorded from the CA1 region of the rat hippocampus using single pulse stimulation of the Schaffer collateral commissural fibres at 30 s intervals. Perfusion of 15 μM substance P for 10 min had no significant effect on the amplitude of the fEPSP (106 ± 5% of control at the end of drug perfusion, figure 1(a) and 1(b), not significant) or the slope of the fEPSP (113 ± 2% of control at the end of drug perfusion, figure 1(a) and 1(c), not significant).
Contrary to previous experiments performed in the mouse hippocampus [9]), we therefore found no effect of substance P on fEPSPs recorded in the rat hippocampus. Existing immunohistochemical and electrophysiological data point to the fact that substance P receptors are found solely on inhibitory interneurones in the hippocampus [8,10]. In our recording conditions, GABAergic transmission plays a minimal role in determining the slope or amplitude of the fEPSP. We therefore turned to recording synaptic responses in which GABAergic transmission clearly has an effect. Synaptic stimulation of CA1 pyramidal neurones evokes a powerful feedback inhibition, which is mediated by GABAA receptors [11]. Paired pulse stimulation can be used to evoke a second response during this phase inhibition and the extent of paired pulse depression can be used as an index of the strength of GABAergic transmission [12]. We therefore investigated next the action of substance P on paired pulse depression of population spikes.
Substance P decreased paired pulse depression
There is some variability in the extent of paired pulse depression observed between individual slices. In general, those slices deemed healthy by the criterion of not showing a secondary population spike to a single stimulus, also show good paired pulse depression. Overall, ie prior to any selection of slices, when pulses were delivered 20 ms apart, the mean amplitude of the second population spike (PS2) was 55 ± 11% of the amplitude of the first (PS1) (n = 10). For the following experiments however, unless otherwise stated, only those slices in which PS2 was 70% or less of PS1 when using a 20 ms interpulse interval were used. Slices were stimulated every 30 s with paired pulses delivered 20 ms apart. A control period of at least 15 min was established, in which both PS1 amplitude and extent of paired pulse depression were stable, and then substance P was applied. Perfusion of 8 μM substance P for 10 minutes was found to cause an increase in the amplitude of PS2 while having no effect on PS1 (figure 2(b) and 2(c), n = 7). The mean amplitude of PS2 during the control period was 37 ± 9% of PS1, and this increased to 88 ± 8% of PS1 (p > 0.05) at the end of drug perfusion. This increase in PS2 was not accompanied by any significant effect on the amplitude of PS1, which was 114 ± 8% of control at the end of drug perfusion (p = 0.1 not significant). The peak effect of substance P on paired pulse depression was observed within 10 min of perfusion, and the effect was reversible towards control levels over 20–30 min.
In a different set of experiments, we next examined whether the effect of substance P was limited to those interpulse intervals that correspond with the time course of GABAergic feedback inhibition. Using paired pulse stimulation at interpulse intervals of 20, 40, 80 and 150 ms, the effect of 8 μM substance P perfused for 10 min was studied. The effect of substance P was evident only at the shorter interpulse intervals of 20–80 ms (figure 2(d), n = 9). The greatest increase in amplitude of PS2 was observed at the 40 ms interpulse interval where the control amplitude of PS2 was 43 ± 12% of PS1, but increased to 78 ± 12% of PS1 at the end of drug perfusion (p = 0.005). Substance P also caused a smaller but still significant increase in the amplitude of PS2 at the 20 ms interpulse interval (PS2 amplitude increased from 23 ± 7 to 45 ± 13% of PS1, p = 0.03) and the 80 ms interpulse interval (PS2 amplitude increased from 92 ± 14 to 124 ± 4% of PS1, p = 0.03). In contrast, substance P had no significant effect at the 150 ms interpulse interval (p = 0.2). Further to this, in the minority of slices which, even at short interpulse intervals showed no paired pulse depression, perfusion of 8 μM substance P had no effect on the amplitude of the first or second population spike (n = 3, data not shown).
Effect of other tachykinin agonists on paired pulse depression
Substance P is most potent at the neurokinin-1 receptor, however, it can also activate the other tachykinin receptors (neurokinin-2 and neurokinin-3). Establishing an order of potency of selective agonists in mimicking the effect of substance P is therefore a useful indicator of which receptor mediates the effect.
Substance P methyl ester has been reported to be a potent neurokinin-1 receptor agonist [13]. Using paired pulse stimulation with an interpulse interval of 20 ms, perfusion of 0.5 μM substance P methyl ester for 10 min had no effect on PS1, but significantly increased the amplitude of PS2, though the effect appeared smaller than that produced by 8 μM substance P. Substance P methyl ester produced an increase in PS2 amplitude from 21 ± 7% to 39 ± 10% of PS1 at the end of drug perfusion (p < 0.05, figure 3, n = 8), while having no effect on PS1 (PS1 amplitude was 114 ± 9% of control at the end of drug perfusion).
[β8Ala]-neurokinin A (4–10) is an neurokinin-2 receptor preferring agonist [14]. Using paired pulse stimulation with an interpulse interval of 20 ms, perfusion of 10 μM [β8Ala]-neurokinin A (4–10) for 10 min had a small and statistically insignificant effect on PS2. PS2 amplitude increased from 31 ± 13% to 51 ± 13% of PS1 at the end of drug perfusion (not significant, figure 3, n = 4). It too had no effect on PS1 (amplitude of PS1 was 99 ± 9% of control at the end of drug perfusion). In order to directly compare the effect of substance P in the same slices, after a 30-min washout, 8 μM substance P was perfused for 10 min. This increased the amplitude of PS2 from 34 ± 4% to 67 ± 30% of PS1 (data not shown).
Senktide is a neurokinin-3 receptor preferring agonist [15] and was found to have no effect on the amplitude of PS1 or PS2. Using paired pulse stimulation with an interpulse interval of 20 ms, 10 μM senktide perfused for 10 min had no significant effect on the amplitude of PS2 (control amplitude of PS2 was 33 ± 8% of PS1 compared to 43 ± 5% of PS1 at the end of drug perfusion, not significant, n = 4, figure 3). In the same slices, after a 30-min washout, 8 μM substance P was perfused for 10 min and it again increased PS2 amplitude from 33 ± 7% of PS1 to 60 ± 5% of PS1 at the end of drug perfusion (p < 0.05, data not shown).
Effect of neurokinin-1, neurokinin-2 and neurokinin-3 receptor antagonists on the action of substance P
To further characterise the receptor via which substance P decreased paired pulse depression, we next used three selective antagonists.
SR140333, a non-peptide neurokinin-1 preferring antagonist, was obtained from Sanofi research [16]. To establish a within-slice control, 8 μM substance P was first perfused for 10 min and, using an interpulse interval of 20 ms, this again caused an increase in the amplitude of PS2, as previously seen. After a 30-min washout period, during which PPD returned to control levels, 12 μM SR140333 was perfused for 20 min prior to, and during, a second perfusion of substance P. SR140333 had no effect on either PS1 or PS2 when perfused alone. However, when 8 μM substance P was perfused in the presence of SR140333, the effect of substance P on PS2 was reduced (figure 4). Substance P increased the amplitude of PS2 from 21 ± 8% of PS1 to 75 ± 19% of PS1, and when applied in the presence of SR140333 it increased PS2 from 19 ± 8% of PS1 to 54 ± 17% of PS1 (n = 6, p > 0.05).
MDL 29,913 is a neurokinin-2 receptor antagonist [17]. Using the same protocol as described above, the effect of 8 μM substance P was established and allowed to recover, before 5 μM MDL29,913 was applied for 20 min prior to, and during, a second perfusion of substance P. MDL 29,913 was found to have no effect itself on PS1 or PS2, or to block the effect of substance P. Substance P perfused for 10 min caused an increase in PS2 from a control amplitude of 13 ± 8 of PS1 to 73 ± 26 of PS1 at the end of drug perfusion. When applied in the presence of MDL 29,913, substance P still increased the amplitude of PS2 to 76 ± 13 of PS1 (n = 4, figure 4).
[Trp7, β-Ala8]-neurokinin A (4–10) is a neurokinin-3 receptor-preferring antagonist [18]. Using the same protocol as described above, 5 μM [Trp7, β-Ala8]-neurokinin A (4–10) was applied for 20 min prior to, and for 10 min during, application of substance P. [Trp7, β-Ala8]-neurokinin A (4–10) was found to have no effect itself on PS1 or PS2, or to block the effect of substance P. Within slice controls showed that 8 μM substance P perfused for 10 min caused an increase in PS2 from a control value of 26 ± 13% of PS1 to 86 ± 3% of PS1 at the end of drug perfusion (n = 3, figure 4). When 8 μM substance P was perfused in the presence of 5 μM [Trp7, β-Ala8]-neurokinin A (4–10), substance P increased the amplitude of PS2 to 76 ± 2% of PS1.
Discussion
Substance P selectively decreases paired pulse depression
The most striking feature of our results is the selective effect of substance P on PS2, but not PS1. Tachykinin receptors are located on inhibitory interneurones and not pyramidal cells [8] and neurokinin-1 receptors in particular are located on the cell body and dendrites of GABA immunopositive interneurones [3]. The location of the neurokinin-1 receptors would suggest an involvement of substance P in the control of inhibition of pyramidal neurones, and maybe of other interneurones, but not in directly modulating excitatory transmission. This is consistent with the fact that substance P had no effect on the recorded fEPSP or on PS1, which are primarily mediated by AMPA receptors. This result is, however, in disagreement with previous work. Kouznetsova and Nistri [9] found that perfusion of substance P (2–4 μM) and its synthetic analogue, substance P methyl-ester, significantly depressed field potentials recorded from the CA1 region of the mouse hippocampus. The reason for this discrepancy may be due to species difference (mouse vs. rat), or alternatively to the baseline recording conditions, and specifically the level of GABAergic inhibition. Kouznetsova and Nistri [9] hypothesised that substance P exerted its depressant action via GABA interneurones and not directly on the pyramidal cells recorded from. In our experiments, we deliberately selected hippocampal slices that exhibited good paired pulse depression, and therefore robust GABAergic inhibition. If this inhibition was effectively maximal, then substance P may be unable to further enhance it. It is therefore significant that we have previously noted paired pulse depression (and therefore presumably GABAergic transmission) is much weaker in slices maintained in a submersion chamber of the type used by Kouznetsova and Nistri, than an interface chamber as used in our experiments.
Since we could not demonstrate any effect of substance P on synaptic responses to single pulse stimulation, we turned to the phenomenon of paired pulse depression. Using an interstimulus interval of 20 ms, substance P (8 μM) perfused onto slices that displayed paired pulse depression, increased the amplitude of PS2 with no effect on PS1. Paired pulse depression of population spikes is thought to be predominately caused by feedback inhibition and can be used as an index of the strength of GABAergic neurotransmission within the hippocampus [12]. As previously noted, neurokinin-1 receptors are located on interneurones of the hippocampus, and substance P acting at these receptors could regulate the release of GABA. A decrease in GABA release would decrease the amplitude of the GABAergic IPSP evoked in the pyramidal neurone, increasing the probability that a second stimulus would fire an action potential, thereby increasing PS2 and inhibiting paired pulse depression. This effect was found to occur only at shorter interstimlus intervals of below 80 ms, which corresponds with the time course of the intracellularly recorded GABAA receptor mediated IPSP evoked in the CA1 pyramidal neurones [19]. An effect on inhibitory synaptic transmission is supported by the anatomical localisation of substance P receptors to GABA-containing interneurones and not to glutamate containing principal (pyramidal) cells [7]. Furthermore, electrophysiological recordings show that neurokinin-1 receptor agonists depolarise interneurones and increase the frequency of spontaneous (action potential dependent) inhibitory post synaptic currents (IPSCs) recorded from pyramidal cells in the CA1 region of the hippocampus [10]. Whilst these experiments showed an increase, rather than a decrease in the frequency of spontaneous IPSCs, they did not investigate the effect of substance P on evoked IPSCs.
Effects of selective tachykinin receptor agonists and antagonists
A range of selective tachykinin receptor agonists was investigated. The neurokinin-1 receptor agonist substance P methyl ester was selected because it had been previously found to effectively mimic the effects of substance P in the mouse hippocampus, where it was effective in a concentration range of 10 nM–5 μM, with the maximum depressant action on field potentials observed using 0.1 μM [9]. In our experiments, perfusion of substance P methyl ester (0.5 μM), mimicked the effect of substance P and caused a significant increase in the amplitude of PS2. Due to the different concentrations used (8 μM and 0.5 μM), it is not possible to comment on the relative potencies of substance P and substance P methyl ester in our experiments. [β-Ala8]-neurokinin A (4–10) has been found to be a highly selective neurokinin-2 receptor agonist which has a 100-fold higher potency for neurokinin-2 receptors than for neurokinin-1 receptors [14]. 10 μM [β-Ala8]-neurokinin A (4–10) had a small effect on the amplitude of PS2 although this was not statistically significant. Any effect of [β-Ala8]-neurokinin A (4–10) on PS2 may be mediated via neurokinin-2 receptors, or more plausibly, [β-Ala8]-neurokinin A (4–10) may have some effect on neurokinin-1 receptors [20,21]. Such a weak interaction of neurokinin-2 receptor agonists with neurokinin-1 receptors has been reported in the entorhinal cortex, where [β-Ala8]-neurokinin A (4–10) mimicked the action of substance P in increasing GABA release from interneurones; an effect which was blocked by a neurokinin-1 receptor antagonist [10]. The neurokinin-3 receptor agonist senktide is one of the most potent of the neurokinin-3 receptor agonists [22]. Perfusion of 10 μM senktide had no effect on the amplitude of PS2, suggesting that the neurokinin-3 receptor is not involved in the decrease observed in paired pulse depression and has no effect on synaptic transmission measured here. A comparison of the effects of the agonists used gives an order of potency of substance P > [β-Ala8]-neurokinin A (4–10) > senktide. This is consistent with the effect of substance P on paired pulse depression being mediated by the neurokinin-1 receptor.
To further characterise the receptor involved, three tachykinin antagonist were used in an attempt to block the action of substance P. The selective neurokinin-1 receptor antagonist SR140333 (12 μM) significantly blocked the effect of substance P, although not completely. In the guinea pig ileum it was found that for SR140333 to have its full activity, a long contact time with the tissue was required [16] and this has been suggested to be longer than 120 min [23]. Contact time in the experiments performed here was a total of 30 min so an even better block may have been achieved with a longer perfusion time. Nevertheless, the ability of SR140333 to reduce the effect of substance P on paired pulse depression is consistent with the effect being neurokinin-1 receptor mediated. This is further supported with complete lack of effect of the neurokinin-2 receptor antagonist MDL29,913, and the neurokinin-3 receptor antagonist [Trp7, β-Ala8]-neurokinin A (4–10). An internal control was used in these experiments, which involved substance being applied twice to the same slice, firstly in the absence, and then in the presence of the antagonist. Under these conditions, desensitization of receptors might be expected to result in a smaller second response, independent of any antagonist effects. However, the fact that repeated perfusion of substance in the presence of the NK2 and NK3 antagonists produced substantially the same effect suggest that this is not the case, and that the reduced effect of substance P in the presence of SR140333 is due to antagonism of neurokinin-1 receptors.
The resulting order of potency of the agonists, and the effectiveness of the antagonists, therefore both suggest that the effect of substance P on paired pulse depression is mediated by neurokinin-1 receptors.
Future work
Whilst the results of our experiments are consistent with the action of substance P being mediated by neurokinin-1 receptors, a more definitive proof of this would be to perform similar experiments in neurokinin-1 receptor knockout mice to establish whether substance P still inhibits paired pulse depression in these animals. There is some evidence that central tachykinin receptors may have different properties to the better characterised peripheral tachykinin receptors, and the possibility remains that the central effects of substance P are mediated by a distinct gene product, albeit with similar properties.
The effect of substance P in selectively decreasing paired pulse depression is consistent with a decrease in GABAergic inhibitory feedback inhibition of CA1 pyramidal cells, and such a mechanism is supported by the anatomical localisation of neurokinin-1 receptors in the hippocampus. However, other factors also contribute to paired pulse depression [24] and therefore, to investigate this hypothesis further, future experiments should use intracellular or whole-cell patch recordings from both pyramidal cells, and from interneurones in the CA1 region. Whilst there is convincing evidence that substance P inhibits spontaneous GABA release from interneurones [10], the effect of substance P on evoked IPSPs has not been determined. Such experiments will give further insights into the role of substance P in the central nervous system.
Conclusion
The results show that perfusion substance P causes a selective reduction in paired pulse inhibition of population spikes evoked in the CA1 region of the rat hippocampal slice, and that this effect is mediated by NK1 receptors. This is consistent with the notion that NK1 receptors are present on the terminals of inhibitory interneurones and act to regulate GABA release.
Methods
Slice preparation
Young adult female Sprague Dawley rats (aged 4–6 weeks) were deeply anaesthetised using halothane and the brain removed and placed in chilled (4–5°C) oxygenated artificial cerebrospinal fluid (aCSF). The aCSF contained (in mM): NaCl 124, KCl 3, NaHCO3 26, NaH2PO4 1.25, D-glucose 10, MgSO4 1 and CaCl2 2 and was continuously bubbled with 95%O2/5% CO2. After dissecting free the hippocampus, 400 μm transverse slices were cut using a McIlwain tissue chopper. Slices were stored in a holding chamber at room temperature before being transferred to an interface type-recording chamber. Within the interface chamber, aCSF was continually perfused below the slice at a rate of 1–2 ml/min and at a constant temperature of 27–29°C.
Extracellular recordings
Extracellular field recordings were obtained from the CA1 region using a glass recording electrode filled with 3 M NaCl. The recording electrode was placed in stratum pyramidale for recording population spikes, or in stratum radiatum for fEPSP measurements. Population spikes and fEPSPs were evoked by a bipolar silver stimulating electrode placed in stratum radiatum towards the CA3 end of the CA1 region to stimulate the Schaffer collateral commissural fibres. Constant current stimulus pulses of 0.02 msec width and 2–11 V were set to elicit a response of approximately half-maximal amplitude. Paired pulse stimulation at interpulse intervals between 20 and 150 ms were used in order to identify effects on both the first population spike (PS1), and the extent of paired pulse depression of the second population spike (PS2). Synaptic responses were evoked every 30 s and collected and analysed using the LTP program [25].
Drugs and data analysis
All drugs were first dissolved in either water or dimethylsulphoxide (DSMO) according to the suppliers instructions (see below) to make a stock solution of at least 1000 times the final concentration. All stock solutions were kept frozen until needed. Application of drugs was achieved by dilution of stock solution into the aCSF, which was perfused onto the slice for the required time. Substance P and MDL 29,913 were obtained from Tocris (UK); substance P methyl-ester, [β-Ala8]-neurokinin A (4–10), senktide, spantide II and [Trp7, β-Ala8]-neurokinin A (4–10) were obtained from Bachem (UK); WIN 51708 was obtained from RBI Sigma (UK). SR140333 was a gift from Dr X. Emonds-Alt (Sanofi Research, Montpelier, France). Stock solutions of substance P, substance P methyl-ester, senktide and MDL 29,913 were all made up in water, whereas spantide II, [β-Ala8]-neurokinin A (4–10), [Trp7, β-Ala8]-neurokinin A (4–10), SR140333 and WIN 51708 were made up in DMSO. To facilitate pooling of data, fEPSP slopes or population spike amplitudes were normalised and expressed as a percentage of the mean slope or amplitude recorded during the entire 15 min control period before addition of drugs. In experiments using paired pulse stimulation, the extent of paired pulse depression was determined by expressing the amplitude of PS1 as a percentage of the amplitude of PS2. All graphs represent pooled data from 3–9 slices prepared from different animals and plot the mean ± standard error of the mean (s.e.m.). Statistical analysis of the raw (ie not normalised) data involved the use of a Students paired t-test to compare control fEPSP slope, population spike amplitude, or extent of paired pulse depression with that recorded at the end of drug perfusion. The control response was measured from the average of the last 5 consecutive responses before drug perfusion, and the drug response was measured from the average of the last 5 consecutive responses during drug perfusion. p < 0.05 was taken as significant.
Authors' contributions
KNW carried out all the experiments and performed the analysis of the data. SND conceived of the study and participated in its design and coordination. KNW and SND drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This work was performed under a grant from the PPP Healthcare Trust.
Figures and Tables
Figure 1 Perfusion of substance P (15 μM) had no effect on the amplitude or slope of the fEPSP recorded from the CA1 region of the rat hippocampal slice. (a) Example synaptic response recorded from a single slice using a single stimulus of the Schaffer collateral-commissural fibres every 30 s. The response on the left was recorded under control conditions whereas the trace on the right was recorded in the presence of substance P (15 μM). (b) and (c) Pooled time course data showing the lack of effect of substance P on the slope (b) and amplitude (c) of the fEPSP. Points represent mean ± s.e.m., n = 4. Scale bar represents 0.5 mV and 10 ms.
Figure 2 Perfusion of substance P (8 μM) had no effect on the amplitude of PS1 but reduced paired pulse depression. (a) Example synaptic response recorded from a single slice using paired pulse stimulation at an interstimulus interval of 20 ms. The response on the left was recorded under control conditions whereas the trace on the right is that recorded in the presence of substance P (8 μM). (b) Pooled time course data showing that perfusion of substance P for 10 min (black bar) had no effect on the amplitude of PS1. (c) Pooled time course data from the same set of experiments showing PS2 expressed as a % of PS1. Substance P (8 μM) perfused for 10 min (black bar) significantly increased PS2 amplitude and therefore decreased paired pulse depression. Points represent mean ± s.e.m., n = 7. (d) The effect of substance P on the full range of interstimulus intervals recorded during control conditions (open bars) and during drug perfusion (filled bars). * denotes p < 0.05, ** denotes p > 0.01, n = 9. Scale bar represents 1 mV and 10 ms.
Figure 3 The effect of tachykinin receptor agonists on paired pulse depression. All agonists were perfused for 10 min. (a) Bars represents the control paired pulse depression using an interstimulus interval of 20 ms (PS2 amplitude expressed as a % of PS1, open bars), compared to that recorded in the presence of each agonist (filled bars). Substance P methyl ester significantly increased the amplitude of PS2, p > 0.05, n = 9. (β8Ala) neurokinin A (4–10) and senktide failed to have any significant effect on the amplitude of PS2, n = 4 for both. Substance P significantly increased the amplitude of PS2, p > 0.05, n = 7. (b) Example synaptic response recorded from 3 different representative slices. Top row of traces are controls, bottom traces are those recorded during perfusion of each tachykinin agonist. Scale bar represents 1 mV and 10 ms.
Figure 4 The effect of tachykinin receptor antagonists, when perfused with substance P. (a) Pooled data showing for each antagonist; control paired pulse depression using an interstimulus interval of 20 ms (PS2 amplitude expressed as a % of PS1, open bars), the effect of 8 μM substance P (filled bars), the effect of antagonist alone (diagonally hatched bars) and the effect of substance P in the presence of SR140333, MDL29,913 or [Trp7, β-Ala8]-neurokinin A (4–10) (horizontally hatched bars). The neurokinin-1 receptor antagonist SR140333 (12 μM) significantly reduced the effect of substance P. * denotes P > 0.05, n = 6. The neurokinin-2 receptor antagonist MDl29,913 (12 μM, n = 4) and the NK-3 antagonist [Trp7, β-Ala8]-neurokinin A (4–10) (5 μM, n = 3), had no significant effecton the action of substance P (b) Example synaptic response recorded from 3 representative slices showing control responses (top row), responses recorded in the presence of substance P (middle row), and responses recorded in the presence of substance P and the antagonist (bottom row). Scale bar represents 1 mV and 10 ms.
==== Refs
Erpamer V The tachykinin peptide family Trends Neurosci 1981 4 297 269 10.1016/0166-2236(81)90093-X
Ingi T Kitajima Y Minamitake Y Nakanishi S Characterization of ligand-binding properties and selectivities of three rat tachykinin receptors by transfection and functional expression of their cloned cDNAs in mammalian-cells J Pharmacol Exp Ther 1991 259 968 975 1662278
Sloviter RS Ali-Akbarian L Horvath KD Menkens KA Substance P receptor expression by inhibitory interneurons of the rat hippocampus: enhanced detection using improved immunocytochemical methods for the preservation and colocalization of GABA and other neuronal markers J Comp Neurol 2001 430 283 305 11169468 10.1002/1096-9861(20010212)430:3<283::AID-CNE1031>3.0.CO;2-V
Hagan RM Beresford IJM Stables J Dupere J Stubbs CM Elliot PJ Sheldrick RLG Chollet A Kawashima E Mcelroy AB Ward P Characterization, CNS distribution and function of NK2 receptors studied using potent NK2 receptor antagonists Regul Pept 1993 46 9 19 8210508 10.1016/0167-0115(93)90005-S
Mileusnic D Magnuson DJ Hejna MJ Lorens JB Lorens SA Lee JM Age and species-dependent differences in the neurokinin B system in rat and human brains Neurobiol Aging 1999 20 19 35 10466890 10.1016/S0197-4580(99)00019-6
Rigby M O'Donnell R Rupniak NMJ Species differences in tachykinin receptor distribution: Further evidence that substance P (NK1) receptor predominates in human brain J Comp Neurol 2005 490 335 353 16127708 10.1002/cne.20664
Nakaya Y Kaneko T Shigemoto R Nakanishi S Mizuno N Immunohistochemical localisation of substance P receptor in the central nervous system of the adult rat J Comp Neurol 1994 347 249 274 7814667 10.1002/cne.903470208
Acsady L Katona I Gulyas AI Shigemoto R Freund TF Immunostaining for substance P receptor labels GABAergic cells with distinct termination patterns in the hippocampus J Comp Neurol 1997 378 320 336 9034894 10.1002/(SICI)1096-9861(19970217)378:3<320::AID-CNE2>3.0.CO;2-5
Kouznetsova M Nistri A Modulation by substance P of synaptic transmission in the mouse hippocampal slice Eur J Neurosci 1998 10 3076 3084 9786202 10.1046/j.1460-9568.1998.00318.x
Stacey AE Woodhall GL Jones RS Neurokinin-receptor-mediated depolarization of cortical neurons elicits an increase in glutamate release at excitatory synapses Eur J Neurosci 2002 16 1896 906 12453053 10.1046/j.1460-9568.2002.02266.x
Schwartzkroin PA Knowles WD Local interactions in the hippocampus Trends Neurosci 1983 6 88 92 10.1016/0166-2236(83)90045-0
Al-Hayani A Davies S Effect of cannabinoids on synaptic transmission in the rat hippocampal slice is temperature-dependent Eur J Pharmacol 2002 442 47 54 12020681 10.1016/S0014-2999(02)01493-0
Watson SP Sandberg BEB Hanley MR Iversen LI Tissue selectivity of substance P alkyl esters: suggesting multiple receptors Eur J Pharmacol 1983 87 77 84 6188620 10.1016/0014-2999(83)90052-3
Rovero P Pestellini V Patacchini R Giuliani S Santicioli P Maggi CA Meli A Giachetti A A potent and selective agonist for NK-2 tachykinin receptor Peptides 1989 10 593 595 2550911 10.1016/0196-9781(89)90148-4
Wörmser U Laufer R Hart Y Chorev M Gilon C Selinger Z Highly selective agonists for substance P receptor subtypes EMBO J 1986 5 2805 8 2431898
Emond-Alt X Doutremepuich J-D Heaulme M Neliat G Santucci V Steinberg R Vilain P Bichon D Ducoux J-P Proietto V Van Broeck D Soubrie P Le Fur G Breliere J-C In vitro and in vivo biological activities of SR140333, a novel potent non-peptide tachykinin NK1 receptor antagonist Eur J Pharmacol 1993 250 403 413 7509286 10.1016/0014-2999(93)90027-F
Van Giersbergen PL Shatzer SA Harbeson SL Rouissi N Nantel F Buck SH Multiple NK2 receptor subtypes are suggested by physiological and biochemical studies with neurokinin A (NKA) analogues and antagonists Ann NY Acad Sci 1991 632 483 484 1659300
Drapeau G Roussi N Nantel F Rhaleb NE Tousignant C Regoli D Antagonists for the neurokinin NK-3 receptor evaluated in selective receptor systems Reg Pept 1990 31 125 135 10.1016/0167-0115(90)90115-D
Davies CH Davies SN Collingridge GL Paired-pulse depression of monosynaptic GABA-mediated inhibitory postsynaptic responses in rat hippocampus J Physiol 1990 424 513 531 2167975
Hastrup H Schwartz TW Septide and neurokinin A are high-affinity ligands on the NK-1 receptor; evidence from homologous versus heterologous binding analysis FEBS Lett 1996 399 264 266 8985159 10.1016/S0014-5793(96)01337-3
Bremer AA Leeman SE Boyd ND The common C-terminal of substance P and neurokinin A contact the same region of the NK-1 receptor FEBS Lett 2000 486 43 48 11108840 10.1016/S0014-5793(00)02228-6
Petitet F Saffroy M Torrens Y Glowinski J Beaujouan JC A new selective bioassay for tachykinin NK3 receptors based on inositol monophosphate accumulation in the guinea pig ileum Eur J Pharmacol 1993 247 185 191 7506659 10.1016/0922-4106(93)90076-L
Martini-Luccarini F Reynaud JC Puizillout JJ Effects of tachykinins on identified dorsal vagal neurons: an electrophysiological study in vitro Neurosci 1996 71 119 31 10.1016/0306-4522(95)00418-1
Higgins MJ Stone TW The contribution of adenosine to paired-pulse inhibition in the normal and disinhibited hippocampal slice Eur J Pharmacol 1996 317 215 223 8997603 10.1016/S0014-2999(96)00731-5
Anderson WW Collingridge GL The LTP Program: A data acquisition program for on-line analysis of long-term potentiation and other synaptic events J Neurosci Meth 2001 108 71 83 10.1016/S0165-0270(01)00374-0
|
16305744
|
PMC1310527
|
CC BY
|
2021-01-04 16:03:49
|
no
|
BMC Neurosci. 2005 Nov 23; 6:66
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-66
|
oa_comm
|
==== Front
Respir ResRespiratory Research1465-99211465-993XBioMed Central London 1465-9921-6-1381630305310.1186/1465-9921-6-138ResearchTherapeutic lung lavages in children and adults Paschen Christian [email protected] Karl [email protected] Franz [email protected] Helmut [email protected] Matthias [email protected] Dr. von Haunersches Kinderspital, University of Munich, Lindwurmstr. 4a, D-80337 Munich, Germany2 ASKLEPIOS Fachkliniken, Zentrum für Pneumologie und Thoraxchirurgie, Robert-Koch-Allee 2, D-82131 München-Gauting, Germany3 Ruhrlandklinik, Department Respiratory and Sleep Medicine, University of Essen, Tüschener Weg 40, Germany2005 22 11 2005 6 1 138 138 27 8 2005 22 11 2005 Copyright © 2005 Paschen et al; licensee BioMed Central Ltd.2005Paschen et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Pulmonary alveolar proteinosis (PAP) is a rare disease, characterized by excessive intra-alveolar accumulation of surfactant lipids and proteins. Therapeutic whole lung lavages are currently the principle therapeutic option in adults. Not much is known on the kinetics of the wash out process, especially in children.
Methods
In 4 pediatric and 6 adult PAP patients 45 therapeutic half lung lavages were investigated retrospectively. Total protein, protein concentration and, in one child with a surfactant protein C mutation, aberrant pro-SP-C protein, were determined during wash out.
Results
The removal of protein from the lungs followed an exponential decline and averaged for adult patients 2 – 20 g and <0.5 to 6 g for pediatric patients. The average protein concentration of consecutive portions was the same in all patient groups, however was elevated in pediatric patients when expressed per body weight. The amount of an aberrant pro-SP-C protein, which was present in one patient with a SP-C mutation, constantly decreased with ongoing lavage. Measuring the optical density of the lavage fluid obtained allowed to monitor the wash out process during the lavages at the bedside and to determine the termination of the lavage procedure at normal protein concentration.
Conclusion
Following therapeutic half lung lavages by biochemical variables may help to estimate the degree of alveolar filling with proteinaceous material and to improve the efficiency of the wash out, especially in children.
BALPAPProtein
==== Body
Introduction
Pulmonary alveolar proteinosis (PAP) is a rare respiratory disease characterized by the accumulation of surfactant-derived material in the lung of patients [1]. Currently PAP is categorized into acquired, congenital, and secondary PAP [2,3].
The acquired form of PAP is clinically characterized by cough, dyspnea and progression to respiratory failure. The presence of anti-GM-CSF auto-antibodies in serum and bronchoalveolar lavage (BAL) is of diagnostic value for this entity[3,4]. The congenital form of PAP is characterized by an acute onset immediately after birth with respiratory distress and rapid progression[5]. Pathogenetically mutations of the genes encoding surfactant protein B [6,7] and C[8,9], the GM-CSF receptor β subunit[10], or ABC-A3 [11] may lead to the accumulation of proteinaceous alveolar material. Secondary PAP is uncommon and includes cases with lysinuric protein intolerance, acute silicosis and other inhalational syndromes, immunodeficiency disorders, and malignancies and hematopoietic disorders[3].
Therapeutic bronchoalveolar lavages (BAL) are the principle option to reduce the abnormal accumulation of PAS positive proteinaceous material that fills the alveolar space of patients with pulmonary alveolar proteinosis (PAP)[3]. Little is known about the kinetics of the protein wash out during therapeutic whole lung lavages. The lavages of one adult patient were investigated by Onodera et al. and showed a rapid declining curve of protein and phospholipid in the successive lavage fractions[12]. Doyle et al. also showed the decrease of cholesterol, surfactant protein A (SP-A), surfactant protein B (SP-B) and phospholipids in aliquots[13]. Recently Perez and Rogers reported in adult patients that chest percussion therapy and positional changes during whole lung lavage enhanced alveolar clearance[14]. In children almost no data are available on the wash out kinetics.
The aim of the present study was to investigate the volume, the total amount, concentration and pattern of protein washed out of the lungs during such procedures in pediatric and adult patients with PAP and in one patient with cholesterol pneumonitis. The value of simple determination of the optical density (OD) to monitor the progress of the lavage procedure and to help determine when to stop the procedure was evaluated. We found an exponential wash out of protein from the lungs and suggest to lavage until the effluent has an OD at 405 nm of 0.04 or less, as this ensures that protein concentrations present in the normal lung are achieved.
Patients and methods
A total of 45 lavages from patients with alveolar proteinosis were prospectively collected to study the wash out of surfactant material from the lungs during therapeutic lung lavage.
Pediatric patients with pulmonary alveolar proteinosis
PAP was diagnosed by the characteristic histologic pattern of alveolar filling with periodic acidic Schiff positive material in open lung biopsy in all children (patients J01, J02, J03, J04). In addition the effluent from the lavages was milky and showed the characteristic cytological pattern. Patient J01 was described previously to have a heterozygous SP-C mutation[9,15]. In the other children no SP-B or SP-C mutations were detected. GMCSF autoantibodies were negative in all these patients in serum and lavage. Further clinical details of the subjects are given in Tab. 1.
Table 1 Patient characteristics and overview on lavages performed
Patients Sex Body weight age at diagnosis age at follow up number of lavage sessions number of 500 ml portions per lavage total lavage volume recovered per lung volume/b.w
(kg) (y) (y) left right left lung right lung (ml) (ml/kg)
PAP ped. (J01) m 8.5 1.75 5.8 9 11 9 (6.5/10) 6 (4/7) 3258 (2780/4080) 383.3
PAP ped. (J02) f 14.5 1.75 7.8 6 6 7.5 (5.5/10) 7 (5.5/9.5) 3353 (2862/3913) 231.2
PAP ped. (J03) f 4.5 0.33 died at age 1/3 1 1 2 8 270.3 60.1
PAP ped. (J04) m 4.3 0.08 died at age 1/4 1 1 2 4 1018.3 236.8
median (25/75 percentile) 6.5 (4.4/11.5)++ 1.0 (0.2/2.3)++ 2007 (644.3/3199) 234 (145.7/310)
PAP adult (A01) f 70.0 39.5 alive 1 1 28 29 14102 201.5
PAP adult (A02) f 69.6 39 alive 1 0 13 0 6608 94.4
PAP adult (A03) f 69.7 49 alive 0 1 0 35 17659 252.3
PAP adult (A04) f 70.2 37 alive 0 1 0 27 27000 385.7
PAP adult (A05)# m 80.5 51.5 alive 1 1 20 24 22000 275
PAP adult (A06)# m 80.1 43.5 alive 1 1 31 39 35000 437.5
median (25/75 percentile) 70.1 (69.7/80.3) 42 (38/50) 19830 (10355/31000)$$ 263.7 (248/411.6)$
Cholesterol – Pneumonitis (L01) m 13 6.5 13.5. LTX 1 1 n.a. 2 577.5 44.4
control (C01) f 13 1.5 n.a. 0 1 n.a. n.a. 12 0.9
control (C02) m 20 5 n.a. 0 1 n.a. n.a. 48 2.4
control (C03) m 16 4 n.a. 0 1 n.a. n.a. 23 1.4
control (C04) m 11 3 n.a. 0 1 n.a. n.a. 17 1.6
control (C05) f 11 2.5 n.a. 0 1 n.a. n.a. 21 1.9
control (C06) m 8.5 1.5 n.a. 1 0 n.a. n.a. 21 2.5
control (C07) f 7 0.5 n.a. 1 0 n.a. n.a. 12.5 1.8
control (C08) m 10 1.5 n.a. 1 0 n.a. n.a. 14 1.4
control (C09) m 11 2 n.a. 1 0 n.a. n.a. 17 1.6
control (C10) m 36 10.5 n.a. 1 0 n.a. n.a. 65 1.8
median (25/75 percentile) 11 (9.3/18) 2.3 (1.5/4.5) 19 (13.3/35) 1.7 (1.4/2.2)
n.a. = not applicable. b.w. = body weight. Only diagnostic bronchoalveolar lavage. i.e. 4 ml/kg in 4 fractions; LTX = lung transplantation; data are presented as median (25/75 percentile); y = years; # = every portion consists of 1000 ml, total protein of 500 ml portions calculated; BAL = bronchoalveolar lavage; f = female; m = male; All three groups (PAP ped, PAP adult and controls) were compared by Friedmann (ANOVA), followed by Dunn's post-hoc-test: +: p < 0.05, ++: p < 0.01, +++: p < 0.001 indicate differences between pediatric PAP and adult PAP. §: p < 0.05, §§: p < 0.01, §§§: p < 0.001 indicate differences between pediatric PAP and controls. $:p < 0.05, $$: p < 0.01, $$$: p < 0.001 indicate differences between adult PAP and controls.
Adult patients with pulmonary alveolar proteinosis
PAP was diagnosed by open lung biopsy (patients A01, A02, A03) or by a combination of typical clinical and radiological findings on HRCT and a diagnostic BAL showing milky fluid and abundant extracellular periodic acidic Schiff positive material on cytopreps (Patients A04, A05, A06) [16]. Clinical details of the patients are given in Tab. 1. All 6 adults patients had idiopathic PAP with high titres of GMCSF autoantibodies.
A child with cholesterol pneumonitis and suspected alveolar proteinosis (labeled as CHOL)
The diagnosis of idiopathic cholesterol pneumonitis, associated with pulmonary alveolar proteinosis was made by open lung biopsy and the child was referred to our centre for therapeutic lavage. He had progressive respiratory distress and was oxygen dependent at that time. Two therapeutic lavages were done, one on each side. However the material obtained was not milky and thus the lavage procedure was terminated early, when almost clear fluid was recovered.
Control children
Lavages from ten healthy children who participated in a study on the biophysical activity of surfactant [17] were used in this study as a comparison group. The children had no history of chronic respiratory symptoms or recent upper or lower respiratory tract infection. Their clinical details are given in table 1. All children were undergoing elective surgery for non pulmonary illnesses. Bronchoalveolar lavages (BAL) were performed during general anaesthesia and tracheal intubation with an endhole catheter wedged in the right lower lobe and the lavage was performed as described below. The original study of these children by BAL had been approved by the ethics committee (Nr. 97079) and written informed consent was given [17]. For the present study those lavages were used to determine the protein levels. The analysis of the therapeutic lavages was done retrospectively on samples stored after informed consent. The ethics committee had approved the anonymous usage of these samples for further variables of the surfactant system.
Bronchoalveolar lavages and processing of the lavage fluid
Initially, in each patient, a diagnostic bronchoalveolar lavage was done. This was done either through the endhole catheter in the control children, through a bronchoscope wedged in the adult PAP patients or in the pediatric patients through a pulmonary artery catheter (Balloon Wedge Pressure Catheter, 60 cm, inner diameter 6 French = 2 mm, Arrow Inc., Reading, USA) in wedge-position on the right or left side. Normal saline (0.9% NaCl) warmed to body temperature (4 × 1 ml/kg body weight) was instilled in aliquots of 1 ml/kg bw, in adults 160 ml (8 times 20 ml) were instilled and recovered with a 20 ml syringe under manual control. The first aliquot of recovered fluid was treated separately and 2–4 ml was used for microbiological investigations. All consecutive aliquots were pooled and labeled "BAL" throughout this paper.
The therapeutic lavages in the children were done with up to 20 ml/kg b.w. aliquots of normal saline. In the small infants where it was not possible to position a double lumen endotracheal tube, a pulmonary artery catheter was introduced through an endotracheal tube and wedged in the main stem bronchus. The tightness of the fit of the balloon was continuously monitored throughout the procedure via a 1.8 or 2.3 mm flexible endoscope advanced outside the tube and positioned proximal to the balloon of the catheter. The fluid recovered was collected in consecutive 500 ml portions. In the adults, the therapeutic lavages were done similarly through one port of a double lumen endotracheal tube with 500–1000 ml aliquots of normal saline, whereas the other port was used to ventilate the contra lateral lung. The returned fluid was collected in consecutive 1000 ml aliquots.
Analysis of proteins
Total protein concentration was measured by the method of Bradford [18]. The abundance of an abberant proform of SP-C, present in the lavages of subject J01 was determined by one dimensional SDS polyacrylamide gel electrophoresis and western blotting[9,19].
For a rapid semi-quantitative assessment of the lavage protein content, absorption measurements were performed on the native lavage samples at a wave length of 405 nm. Spectra were obtained in a spectrophotometer for wavelengths from 200 nm to 800 nm (Ultrospec 1000, Amersham Pharmacia Biotech, Uppsala, Schweden).
Statistical analysis
Individual data points and where appropriate medians with interquartile range and range are given. Two groups were compared by Mann-Whitney test and several groups by Kruskal Wallis Anova followed by Dunn post hoc test for non-parametric variables. A p < 0.05 was considered significant. Statistical analysis was performed with Prism 4.0 (Graph Pad Software, San Diego, USA).
Results
Therapeutic lavages were done in 4 children with median age of 1 year at diagnosis of PAP, in 6 adults (median age 42 years) and in a 6.5 year old child with cholesterol pneumonitis.
The recovered half lung lavage volume in adults was on average about 20 l per lung and in infants 2 l per lung. However, corrected for body weight, the same volume of about 250 ml/kg b.w. was used for both groups (Tab. 1). Recovery of instilled fluid in all therapeutic lavage procedures was 100 ± 10 %.
The amount of protein removed from the lungs by the therapeutic lavages varied substantially between subjects, but not so much within a certain subject (J01 and J02 in Fig. 1 and Tab. 2). For adult patients the removed amount of protein varied between 2 – 20 g, while the removed amount for pediatric patients was between < 0.5 to 6 g. There were no significant differences between the right and left lung (Fig. 1).
Figure 1 Amount of protein removed from the lungs of patients with pulmonary alveolar proteinosis of the adult (6 subjects, PAPadult), 4 children with PAP (J01 to J04), a child with idiopathic cholesterol pneumonitis, associated histologically with PAP (CHOL) and in 10 control children (CON). In the latter only regular diagnostic bronchoalveolar lavages were done. Each symbol represents the amount of protein recovered from a single lung lavage. L and R donates left and right sides. PAPadult represents total of 9 half lung lavages from patients A01 to A06. Horizontal bars indicate medians.
Table 2 Protein recovered from the lungs
Patients concentration of protein (μg/ml) concentration of protein (μg/ml and kg body weight) amount of protein (mg) amount of protein (mg/kg body weight)
BAL half lung lavages BAL half lung lavages BAL half lung lavages BAL half lung lavages
PAP ped. (J01) 233 (195/285.1) 131 (107.5/163) 27.4 15.4 (12.6/19.2) 7.7 (7/10) 370 (287/405) 0.9 44 (34/48)
PAP ped. (J02) 306 (206.5/1149) 1010 (664/1199) 21.1 69.7 (45.8/82.7) 57 (39/134) 3214 (2262/4826) 3.9 222 (156/333)
PAP ped. (J03) 1034 2975 229.8 661 20.2 922 4.4 205
PAP ped. (J04) 352 269 81.9 62.6 5.1 223 1.2 52
median (25/75 percentile) 307 (207/1149) 236 (130/1010) 220 (147/252)§ 29.7 (16.4/127) 11.7 (8/42) 545.8 (347/2618) 3 (1/4) 68.7 (43.6/329.3)
PAP adult (A01) 1490 924 21.3 13.2 226.7 12828 3.24 186
PAP adult (A02) 322 274 4.60 3.9 54.1 1766 0.77 26
PAP adult (A03) 135 199 1.92 2.8 21.4 3488 0.31 50
PAP adult (A04) no data 569 no data 8.1 no data 15374 no data 220
PAP adult (A05)# no data 161.5 no data 2 no data 3563 no data 45
PAP adult (A06)# no data 1090 no data 13.6 no data 18820 no data 235
median (25/75 percentile) 846 (228/1491) 422 (180/1007) 4.6 5.8 (2.5/13.7)* 130 (38/227)$ 5650 (1034/16850)ns 0.8 77 (14.1/229.8)ns
Cholesterol-Pneumonitis (L01) 136.2 157 10.5 16.8 1.7 124 0.13 9.5
control (C01) 47 n.a. 3.58 n.a. 0.6 n.a. 0.05 n.a.
control (C02) 58 n.a. 2.88 n.a. 2.7 n.a. 0.14 n.a.
control (C03) 77 n.a. 4.82 n.a. 1.7 n.a. 0.11 n.a.
control (C04) 82 n.a. 7.41 n.a. 1.4 n.a. 0.13 n.a.
control (C05) 47 n.a. 4.30 n.a. 1.0 n.a. 0.09 n.a.
control (C06) 85 n.a. 9.96 n.a. 1.8 n.a. 0.21 n.a.
control (C07) 97 n.a. 13.90 n.a. 1.2 n.a. 0.17 n.a.
control (C08) 77 n.a. 7.71 n.a. 1.1 n.a. 0.11 n.a.
control (C09) 65 n.a. 5.92 n.a. 1.1 n.a. 0.10 n.a.
control (C10) 49 n.a. 1.36 n.a. 3.2 n.a. 0.09 n.a.
median (25/75 percentile) 71 (48.2/83) 5.4 (3.2/8.8) 1.3 (1/2.3)§§§§ 0.09
n.a. = not applicable. only diagnostic bronchoalveolar lavage. i.e. 4 ml/kg in 4 fractions; LTX = lung transplantation; data are presented as median (25/75 percentile); y = years; # = only every second portion available. total protein calculated; BAL = bronchoalveolar lavage; f = female; m = male. Two groups were compared by Mann-Whitney-test: ns: not significant,*: p < 0.05, **: p < 0.01, ***: p < 0.001 indicate differences between pediatric and adult PAP All three groups (PAP ped, PAP adult and controls) were compared by Friedmann (ANOVA), followed by Dunn's post-hoc-test: +: p < 0.05, ++: p < 0.01, +++: p < 0.001 indicate differences between pediatric PAP and adult PAP. §: p < 0.05, §§:p < 0.01, §§§: p < 0.001 indicate differences between pediatric PAP and controls. $: p < 0.05, $$: p < 0.01, $$$: p < 0.001 indicate differences between adult PAP and control
The average concentration of protein in the consecutive portions of the half lung lavages was the same in adult, pediatric patients and the patient with cholesterol pneumonitis. When expressed per kg – body weight, pediatric patients had elevated concentrations (Tab. 2).
In the BAL, i.e. the diagnostic lavage, as defined in Methods, the concentrations of protein in adult and pediatric patients were clearly elevated, compared to normal children (Tab. 2). Corrected for kg – body weight, only the pediatric patients had higher levels than the controls. This difference was only about 3 – fold, too small to be reliable for diagnostic purposes.
The kinetics of the wash out followed an exponential decay function for all adult patients and for J01, J02, and J04 (Fig. 2). In patient J03, due to an insufficient procedure, because of instability of the patient, there was no real wash out function visible. This patient had in addition a severe pulmonary infection, that led together with the PAP to respiratory insufficiency and death within 8 weeks. The lavage in the child with the cholesterol pneumonitits was stopped at 1 liter due to very poor recovery of proteinous material (Fig. 2), i.e. an almost clear effluent, suggesting that the histologically suggested alveolar proteinosis was not of significant extent.
Figure 2 Protein concentrations in the diagnostic BAL and the consecutive 500 ml portions of lung lavages from patients with juvenile PAP (J01 – J10), a patient with cholesterol pneumonitis and PAP (CHOL) and 3 adult patients with idiopathic PAP (A04 – A06). Each symbol represents the protein concentration of one 500 ml portion lavage fluid recovered from one side. The numbers of BAL done on each side are indicated in Table 1. Horizontal bars indicate medians. Note the different scales of the protein axis.
Using Western blot, clearly a wash out of an aberrant protein, i.e. pro SP-C, present in a child with PAP and SP-C mutation[15], was demonstrated. As a constant amount of protein was added to the gel, a continuous decrease of this aberrant protein, with ongoing washout, which affected all 3 aberrant pro SP-C bands equally, was observed (Fig. 3).
Figure 3 Western blot of 1 dimensional gel electrophoresis of BAL and 6 subsequent 500 ml portions of one lavage in patient J01. This patient was known to express an abberant pro-SP-C peptide in his lung. The blot was incubated with NPROSP-C Met10-Gln23 as first and goat anti rabbit as secondary antibody to show 3 specific pro SP-C bands.
An immediate estimate of the overall protein concentration would be very helpful for bed side monitoring of the lavage procedure. There was a reasonable correlation between direct OD readings, used to estimate the protein concentration from a previously made calibration curve and the precise protein concentration, as assessed by a colorimetric protein assay (Fig. 4a,b). There was consistent agreement within thumb nail error (± 100 %) (Fig. 4c). Receiver operator curves calculated for different cut-offs to stop the lavage procedure, showed a 100% specificity (i.e. the fraction correctly defined as negative) with a sensitivity (i.e. the fraction correctly defined as positive) of at least 60% at the protein concentration found in healthy subjects, i.e. 100 μg/ml or equivalent to an OD of 0.038 or less (Fig. 4b and 4d ).
Figure 4 Semiquantitative estimation of the lavage protein content by measurement of its absorption at 405 nm for each 500 ml or 1000 ml portion lavage fluid fro all PAP patients (J01 – J04, A01 – A06) and the patient with cholesterol pneumonitis. a: Relationship of protein concentration and absorption at 405 nm of lung lavage fluid. There was a significant correlation between protein concentration assessed in the lavage with the Bradford assay and the absorption directly measured in the photometer. b: Zoom in on the relationship of protein concentration and absorption at 405 nm of lung lavage fluid. The maximum value of protein concentrations observed in the healthy comparison children is indicated by a dotted horizontal line. With an OD value of less than 0.038, more than 90% of the subjects with PAP had a protein concentration in their lung effluent, that was below of the healthy subjects, i.e. a protein concentration of 100 μg/ml or less. c: Bland-Altman Plot for comparison of the two methods, i.e. direct measurement of the OD of the lavage aliquot and the corresponding protein concentration, assessed by the protein assay. d: Receiver Operator Curve analysis of a cut off of of a protein concentration of 100 μg/ml. The area under the curve quantifies the overall ability of the test to discriminate those individuals with the disease, i.e. increased lavage protein concentration, and those without the disease. An area of larger than 90% (here 91%) indicates an accurate test.
Discussion
In this study we provide detailed data on the concentrations, amounts and the wash out kinetics of proteins during therapeutic half lung lavages in infants and adults with PAP. A method was presented to easily monitor the wash out process during lavages and to determine when a physiological protein concentration is reached and a therapeutic lavage procedure may be stopped.
Since their introduction by Ramirez[20], Wasserman [21] and Seard [22], therapeutic lung lavages are the treatment of choice in patients with PAP[3,14,23,24]. While this procedures is well established and relatively easily performed in adults, therapeutic lavages in children are technically much more challenging. There are 4 reports in children[10,25-27], 7 in infants [5,28-33] and some in neonates[26,30,34]. Therefore it is not yet clear if therapeutic lung lavages are effective in treating infants with PAP. In addition there are almost no data on the protein washed out in children. Here we present the first data on such kinetics and on the amount of protein removed by whole lung lavage in small children.
In adult patients about 80 – fold higher amounts of total protein were recovered in comparison to normal whole lung lavage values which were estimated by calculation from rat lung washings[12]. Between 4 and 27.7 g were obtained, values that were similar to the 1.8 to 22 g, we found in this study. The control subjects in those studies, i.e. patients with interstitial pneumonia or alveolar cell carcinoma, had 2.8–3.4 g of protein recovered, which was about 10-fold elevated compared to rat lung washings[12]. The amount of protein removed from children with PAP was in the order of 0.4 g to 2.6 g (range 0.16 g to 5.5 g). However, when expressed per kg body weight, the same amount of protein was removed from the lungs of children and adults.
A central problem in all studies on whole lung lavages is the comparison group, as it is not appropriate to lavage normal subjects or other patients without therapeutic need. To circumvent this problem and to still be able to compare controls and PAP lavages directly, we used the regular diagnostic bronchoalveolar lavage (BAL) which was done in all subjects, before the therapeutic lavages were started, for comparison.
In our study we found protein concentrations in the diagnostic BAL that were increased 3 – fold in pediatric PAP patients and 10 – fold in adult PAP patients in comparison to controls. Despite the significant difference to control values, the result is of limited use for diagnosing PAP. There is substantial overlap with other lung diseases, like pulmonary fibrosis [35], pneumonitis[36] and bronchial asthma[37], where total protein may be elevated 2 – 5 fold, thus not allowing a clear diagnostic estimation. The protein concentrations of therapeutic lavages performed by others were 17 – 100 fold increased compared to patients with chronic bronchitis, asthma and a patient with interstitial fibrosis[38].
Until now, the protein wash out characteristic of the wash out process of a therapeutic lavage has been reported for only one patient[12]. For this reason the kinetics of the wash out is of interest. Here, for adult subjects, we show an exponential decay of protein during the procedures. For children comparable results were obtained, however at different levels of protein concentration (compare fig. 2, J01 and J02). The volumes used are about 1/10 of the ones used in adults, but when corrected for kg body weight, they were the same. A reasonable correlation between the protein concentration determined by the Bradford assay and the optical density of the lavage fluid was demonstrated. Thus, the method to monitor the estimated protein concentration in BAL fluids during lavage was evaluated further. When an OD of 0.04 or less was used as the cut off to stop the lavage procedure, the protein level was very likely to be less or in the range of the maximum protein concentration observed in healthy subjects.
Information on the progress of the wash out process from simple online and bedside monitoring may be very helpful, as can also be demonstrated in the patient with cholesterol pneumonitis. This subject had evidence from histological pattern for both cholesterol pneumonitis and PAP. The therapeutic lavage was stopped rather soon, as the effluent appeared relatively clear by visual inspection. However this may have been too early, because the protein concentration of the lavages determined after the procedure was finished, were always above 100 μg/ml. It has previously been reported in an adult patient with endogenous lipoid pneumonia due to Niemann-Pick Type B, that whole lung lavage may be successful with other diagnoses than PAP [39].
Of interest was that aberrant pro SP-C protein not normally present in lavage and found in one patient with a SP-C mutation [9], steadily decreased during the ongoing lavage, suggesting that this particular protein had accumulated over time and was efficiently removed from the alveolar space without significant replacement during the wash out.
In summary, there are considerable differences in the amount of protein washed out by whole lung lavages in children and adults with various forms of PAP. The progress of therapeutic lavage procedures and the kinetics of protein removed from the lungs during the lavage process may be continuously estimated by simple OD measurement of the effluent. This may help to make the lavage procedure more efficient, especially in young children and thus help to further optimize the technique in an age group where the procedure is technically very demanding.
==== Refs
Shah PL Hansell D Lawson PR Reid KBM Morgan C Pulmonary alveolar proteinosis: clinical aspects and current concepts on pathogenesis Thorax 2000 55 67 77 10607805 10.1136/thorax.55.1.67
Trapnell BC Whitsett JA Nakata K Pulmonary alveolar proteinosis N Engl J Med 2003 349 2527 2539 14695413 10.1056/NEJMra023226
Seymour JF Presneill JJ Pulmonary alveolar proteinosis: progress in the first 44 years Am J Resp Crit Care Med 2002 166 215 235 12119235 10.1164/rccm.2109105
Latzin P Tredano M Wüst Y de Blic J Nicolai T Bewig B al Anti-GM-CSF antibodies in pediatric pulmonary alveolar proteinosis Thorax 2004 60 39 44 15618581 10.1136/thx.2004.021329
Nogee LM de Mello DE Dehner LP Colten HR Brief-report: deficiency of pulmonary surfactant protein B in congenital alveolar proteinosis N Engl J Med 1993 328 406 410 8421459 10.1056/NEJM199302113280606
Tredano M van Elburg RM Kaspers AG Zimmermann LJ Houdayer C Aymard P Hull WM Whitsett JA Elion J Griese M Bahuau M Compound SFTPB 1549C-GAA (121ins2) and 457delC heterozygosity in severe congenital lung disease and surfactant protein B (SP-B) deficiency Hum Mutat 1999 14 502 509 10571948 10.1002/(SICI)1098-1004(199912)14:6<502::AID-HUMU9>3.0.CO;2-C
Nogee LM Wert SE Proffit SA Hull WM Whitsett JA Allelic heterogeneity in hereditary surfactant protein B (SP-B) deficiency Am J Respir Crit Care Med 2000 161 973 981 10712351
Nogee L Dunbar III AE Wert S Askin F Hamvas A J.A. W Mutations in the surfactant protein C gene associated with interstitial lung disease Chest 2002 121 20S 21S 11893657 10.1378/chest.121.3_suppl.20S
Brasch F Griese M Tredano M Johnen G Ochs M Rieger C al Interstitial lung disease in a baby with a de novo mutation in the SFTPC gene Eur Respir J 2004 24 30 39 15293602 10.1183/09031936.04.00000104
Dirksen U Nishinakamura R Groneck P Hattenhorst U Nogee L Murray R Burdach S Human Pulmonary Alveolar Proteinosis Associated with a Defect in GM-GSF/IL-3/IL-5 Receptor Common ß Chain Expression J Clin Invest 1997 100 2211 2217 9410898
Shulenin S Nogee LM Annilo T Wert SE Whitsett JA Dean M ABCA3 Gene Mutations in Newborns with Fatal Surfactant Deficiency N Engl J Med 2004 350 1296 1303 15044640 10.1056/NEJMoa032178
Onodera T Nakamura M Sato T Akino T Biochemical characterization of pulmonary washings of patients with alveolar proteinosis interstitial pneumonitis and alveolar cell carcinoma. Tohoku J Exp Med 1983 139 245 263 6304935
Doyle IR Davidson KG Barr HA Nicholas TE Quantity and structure of surfactant proteins vary among patients with alveolar proteinosis. Am J Respir Crit Care Med 1998 157 658 664 9476887
Perez A Rogers RM Enhanced alveolar clearance with chest percussion therapy and positional changes during whole-lung lavage for alveolar proteinosis Chest 2004 125 2351 2356 15189962 10.1378/chest.125.6.2351
Tredano M Griese M Brasch F Schumacher S de Blic J Marque S Houdayer C Elion J Couderc R Bahuau M Mutation of SFTPC in infantile pulmonary alveolar proteinosis with or without fibrosing lung disease Am J Med Genet 2004 126A 18 26 10.1002/ajmg.a.20670
Wang BM Stern EJ Schmidt RA Pierson DJ Diagnosing pulmonary alveolar proteinosis. A review and an update. Chest 1997 111 460 466 9041997
Griese M Felber J Reiter K Strong P Reid K Belohradsky BH Jäger G Nicolai T Airway inflammation in children with tracheostomy Pediatr Pulmonol 2004 37 356 361 15022133 10.1002/ppul.10432
Bradford MM A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 1976 72 248 254 942051 10.1016/0003-2697(76)90527-3
Beers MF Wali A Eckenhoff MF Feinstein SI Fisher JH Fisher AB An antibody with specificity for surfactant protein C precursors: identification of pro-SP-C in rat lung Am J Respir Cell Mol Biol 1992 7 368 378 1389209
Ramirez J Schultz RM Dutton MA Pulmonary alveolar proteinosis: a new technique and rationale for treatment Arch Intern Med 1963 112 419 431 14045290
Wasserman K Blank N Fletcher G Lung lavage (alveolar washing) in alveolar proteinosis Am J Med 1968 44 611 617 5642718 10.1016/0002-9343(68)90062-4
Seard C Wasserman K Benfield JR Cleveland RJ Costley DO Heimlich EM Simultaneous bilateral lung lavage (alveolar washing) using partial cardiopulmonary bypass Am Rev Respir Dis 1970 101 877 884 5419972
Kariman K Kylstra JA Spock A Pulmonary alveolar proteinosis: prospective clinical experience in 23 patients for 15 years Lung 1984 162 223 231 6492867
Selecky PA Wasserman K Benfield JR Lippmann M The clinical and physiological effect of whole-lung lavage in pulmonary alveolar proteinosis: a ten-year experience Ann Thorac Surg 1977 24 451 461 921378
Gray ES Letter: Autoimmunity in childhood pulmonary alveolar proteinosis Br Med J 1973 4 296 297 4753251
Mahut B Delcourt C Scheinmann P de Blic J Mani TM Fournet JC Bellon G Pulmonary alveolar proteinosis: Experience with eight pediatric cases and a review Pediatrics 1996 97 117 122 8545205
Danigelis JA Markarian B Pulmonary alveolar proteinosis including pulmonary electron microscopy Am J Dis Child 1969 118 871 875 5353015
Wilkinson RH Blanc WA Hagstrom JW Pulmonary alveolar proteinosis in three infants Pediactrics 1968 41 510 515
Mazyck EM Bonner JT Herd HM Symbas PN Pulmonary lavage for childhood pulmonary alveolar proteinosis J Pediatr 1972 80 839 842 5018397
Ito T Sato M Okubo T Ono I Akanabe J Infantile pulmonary alveolar proteinosis with interstitial pneumonia: bilateral simultaneous lung lavage utilizing extracorporeal membrane oxygenation and steroid therapy Tohoku J Exp Med 1999 187 279 283 10458485 10.1620/tjem.187.279
Sakai Y Abo W Yoshimura H Sano H Kuroki Y Satoh M Kaimori M Pulmonary alveolar proteinosis in infants Eur J Pediatr 1999 158 424 426 10333130 10.1007/s004310051107
Moulton SL Krous HF Merritt TA Odell RM Gangitano E Cornish JD Congenital pulmonary alveolar proteinosis: Failure of treatment with extracorporeal life support J Pediatr 1992 120 297 302 1341413
Hamvas A Nogee LM Mallory GB Spray TL Lung transplantation for treatment of infants with surfactant protein B deficiency J Pediatr 1997 130 231 239 9042125 10.1016/S0022-3476(97)70348-2
de la Fuente AA Voorhout WF deMello DE Congenital alveolar proteinosis in the Netherlands: a report of five cases with immunohistochemical and genetic studies on surfactant apoproteins Pediatr Pathol Lab Med 1997 17 221 231 9086529 10.1080/107710497174877
McCormack FX King TE Voelker DR Robinson PC Mason RJ Idiopathic pulmonary fibrosis Am Rev Resp Dis 1991 144 160 166 2064123
Günther A Siebert C Schmidt R Ziegler S Grimminger F Yabut M Temmesfeld B Walmrath D Morr H Seeger W Surfactant alterations in severe pneumonia, acute respiratory distress syndrome, and cardiogenic lung edema Am J Respir Crit Care Med 1996 153 176 184 8542113
Van de Graaf EA Jansen HM Lutter R Alberts C Kobesen J DeVries IJ Out TA Surfactant protein A in bronchoalveolar lavage fluid J Lab Clin Med 1992 120 252 263 1500824
Ramirez R Harlan WR Pulmonary alveolar proteinosis. Nature and origin of alveolar lipid Am J Med 1968 45 502 512 5678094 10.1016/0002-9343(68)90166-6
Nicholson AG Wells AU Hooper J Hansell DM Kelleher A Morgan C Successful treatment of endogenous lipoid pneumonia due to Niemann-Pick Type B disease with whole-lung lavage Am J Respir Crit Care Med 2002 165 128 131 11779742
|
16303053
|
PMC1310528
|
CC BY
|
2021-01-04 16:36:25
|
no
|
Respir Res. 2005 Nov 22; 6(1):138
|
utf-8
|
Respir Res
| 2,005 |
10.1186/1465-9921-6-138
|
oa_comm
|
==== Front
BMC Musculoskelet DisordBMC Musculoskeletal Disorders1471-2474BioMed Central London 1471-2474-6-561627448710.1186/1471-2474-6-56Research ArticleLumbar segmental instability: a criterion-related validity study of manual therapy assessment Abbott J Haxby [email protected] Brendan [email protected] Peter [email protected] Graeme [email protected] Cathy [email protected] Tracy [email protected] Clarity Clinical Research Consultants, New Zealand2 Computer Science Department, University of Otago, PO Box 56, Dunedin, New Zealand3 Department of Preventive and Social Medicine, University of Otago, PO Box 913, Dunedin, New Zealand4 Back In Motion Physiotherapy, Dunedin, New Zealand5 Physiotherapy Department, Dunedin Hospital, Otago District Health Board, Dunedin, New Zealand2005 7 11 2005 6 56 56 7 6 2005 7 11 2005 Copyright © 2005 Abbott et al; licensee BioMed Central Ltd.2005Abbott et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Musculoskeletal physiotherapists routinely assess lumbar segmental motion during the clinical examination of a patient with low back pain. The validity of manual assessment of segmental motion has not, however, been adequately investigated.
Methods
In this prospective, multi-centre, pragmatic, diagnostic validity study, 138 consecutive patients with recurrent or chronic low back pain (R/CLBP) were recruited. Physiotherapists with post-graduate training in manual therapy performed passive accessory intervertebral motion tests (PAIVMs) and passive physiological intervertebral motion tests (PPIVMs). Consenting patients were referred for flexion-extension radiographs. Sagittal angular rotation and sagittal translation of each lumbar spinal motion segment was measured from these radiographs, and compared to a reference range derived from a study of 30 asymptomatic volunteers. Motion beyond two standard deviations from the reference mean was considered diagnostic of rotational lumbar segmental instability (LSI) and translational LSI. Accuracy and validity of the clinical assessments were expressed using sensitivity, specificity, and likelihood ratio statistics with 95% confidence intervals (CI).
Results
Only translation LSI was found to be significantly associated with R/CLBP (p < 0.05). PAIVMs were specific for the diagnosis of translation LSI (specificity 89%, CI 83–93%), but showed poor sensitivity (29%, CI 14–50%). A positive test results in a likelihood ratio (LR+) of 2.52 (95% CI 1.15–5.53). Flexion PPIVMs were highly specific for the diagnosis of translation LSI (specificity 99.5%; CI 97–100%), but showed very poor sensitivity (5%; CI 1–22%). Likelihood ratio statistics for flexion PPIVMs were not statistically significant. Extension PPIVMs performed better than flexion PPIVMs, with slightly higher sensitivity (16%; CI 6–38%) resulting in a likelihood ratio for a positive test of 7.1 (95% CI 1.7 to 29.2) for translation LSI.
Conclusion
This study provides the first evidence reporting the concurrent validity of manual tests for the detection of abnormal sagittal planar motion. PAIVMs and PPIVMs are highly specific, but not sensitive, for the detection of translation LSI. Likelihood ratios resulting from positive test results were only moderate. This research indicates that manual clinical examination procedures have moderate validity for detecting segmental motion abnormality.
==== Body
Background
Musculoskeletal physiotherapists routinely assess lumbar spinal segmental motion and choose interventions on the basis of the findings of those assessments. However, the validity of clinical tests used to assess segmental motion has not been established. When physiotherapists examine the lumbar spine, common assessments include passive accessory intervertebral motion tests (PAIVMs) and passive physiological intervertebral motion tests (PPIVMs) [1,2]. Movement abnormalities, such as hypermobility, are believed to be detected by these assessments [1].
To date, the only evidence for the concurrent validity of manual testing for the presence of lumbar segmental instability (LSI) comes from two studies in which the presence of spondylolysis was considered a proxy for the presence of segmental hypermobility. The first comprised of a very small subgroup analysis (6 patients) of patients with spondylolysis, within a sample of 62 patients with non-specific LBP [3]. The results of that investigation indicated that PAIVMs and PPIVMs could identify the symptomatic level with 83% sensitivity and 98% specificity [3]. In the second study, manual assessment (combined information from both PPIVMs and PAIVMs) was 69% sensitive and 96% specific for detection of the lytic segment [4]. When analysis was restricted to subjects who reported visual analogue pain scores of greater than 4/10, sensitivity and specificity rose to 100% [4]. In addition, some preliminary evidence indicates that PAIVM testing may have predictive validity for the purpose of classifying patients in a 'stabilisation' category, who respond better to an exercise intervention intended to increase lumbar segmental stability [5].
As there is currently no evidence in the literature to establish the concurrent validity of manual therapy tests for the detection of excessive sagittal planar motion of the lumbar spine, the aims of this study were to estimate the accuracy of three common clinical assessment items for the detection of lumbar segmental hypermobility (PAIVMs, flexion PPIVMs, and extension PPIVMs), compared to a criterion standard of radiographic measurement of sagittal segmental rotation and translation.
Methods
Design
Physiotherapists with post-graduate training in musculoskeletal manual therapy recruited consecutive eligible patients presenting with a new episode of recurrent or chronic low back pain (R/CLBP). Recruiting took place in the physiotherapists' own clinics, between October 2001 and August, 2003. Patients were included if i) they presented with a new episode of low back pain and, ii) they had experienced similar low back pain before, the first episode of which was at least three months prior to the date of recruitment, or iii) they were experiencing persistent low back pain of at least three months duration. Patients were excluded if they i) had spinal surgery within the previous six months, or ii) had a history of traumatic fracture of the spine which resulted in permanent neurological deficit, iii) had a history of serious neurological or psychiatric disease, iv) were under 20 years of age, or v) were pregnant. This research was approved by the Otago and Canterbury Regional Ethics Committees (reference # 01/05/030 & 01/10/095) of the New Zealand Ministry of Health.
The physiotherapists assessed PAIVMs and PPIVMs, at each lumbar segment, nested within a comprehensive physical examination. PAIVMs consisted of postero-anterior central pressure applied to the spinous processes, with the patient lying prone [1,2] (figure 1). PPIVMs were assessed with the patient side-lying, and consisted of moving the patients' spine through sagittal forward-bending (flexion) and backward-bending (extension), while palpating between the spinous process of adjacent vertebrae to assess the motion taking place at each motion segment [1,2] (figures 2 &3). PAIVM ratings were assessed on a 3 point ordinal scale, with 0 indicating hypomobility, 1 indicating normal motion, and 2 indicating hypermobility. PPIVMs were rated on a 5 point ordinal scale, with 0 & 1 indicating hypomobility, normal anchored at 2, and 3 & 4 indicating hypermobility. While pain responses were assessed, they were recorded separately from the assessment of motion, and were not included in the analysis for this study, which was concerned only with the assessment of spinal motion. Consenting patients were referred to radiology for flexion-extension lateral radiographs.
Figure 1 The central posteroanterior passive accessory intervertebral motion (PAIVM) test. The patient lies prone. The clinician contacts the spinous process of the target vertebra with the hypothenar eminence, and delivers a gradual posteroanteriorly directed force.
Figure 2 The passive physiological intervertebral motion (PPIVM) test in flexion. The patient is positioned side-lying. The clinician palpates the interspace between the adjacent spinous processes of the target motion segment with one finger, while moving the lumbar spine from neutral into flexion via the patient's uppermost limb.
Figure 3 The passive physiological intervertebral motion (PPIVM) test in extension. The patient is positioned side-lying. The clinician palpates the interspace between the adjacent spinous processes of the target motion segment with one finger, while moving the lumbar spine from neutral to extension via the patient's uppermost limb.
The reference standard for normal and abnormal spinal mobility measures was defined using the kinematic data from a sample of asymptomatic volunteers with no significant history of LBP, and no LBP within the prior three years. A sample of 30 asymptomatic adults was recruited and radiographed using the same protocol as the patient cohort. This project was approved by the University of Otago Human Ethics Committee.
For both cohorts, the sagittal rotation and translation motion of segments L2-3, L3-4, L4-5, and L5-S1 was measured using the method of Bodguk & Schneider [6-8], by researchers blinded to the clinical examination findings and radiologists' reports. Radiographs of insufficient quality to allow the analysis of two or more segments were excluded.
Measurement procedures
Calculation of rotation and translation motion was performed using the ClaritySMART version 1.2 computer program [9]. Concurrent validity of rotation measurement by ClaritySMART v1.2 was tested against a reference standard (measurement using NIH Image [10]), and assessed using the intraclass correlation coefficient (ICC). Rotation measurement was tested against manual constructions (0.3 mm pencil on tracing paper; measurements using a 0.5 mm graduated ruler). These trials demonstrated near perfect concurrence for both rotation (ICC(3,4) of 0.98, 95% CI 0.92, 0.99), and translation (ICC(3,1) of 0.98, 95% CI 0.94, 0.99). Inter-rater reliability was excellent for both rotation (ICC(3,1) 0.96, 95% CI 0.87, 0.99) and translation (ICC(3,1) 0.83, 95% CI 0.46, 0.95).
Data analysis
The reference standard for presence of LSI in the C/RLBP cohort was abnormal segmental hypermobility in excess of 2 standard deviations (sd) beyond the mean of a sample of 30 pain-free individuals. Prevalence of LSI findings in the C/RLBP cohort (i.e. the number of segments that fall beyond the 2sd cut-point derived from the kinematic data of the asymptomatic sample) were calculated. The chi squared (χ2) goodness of fit test was used to test the hypothesis that abnormal segmental hypermobility (i.e. LSI) is found in a higher proportion of patients with R/CLBP than would be expected in an asymptomatic sample. Significance was set at p < 0.05.
In concordance with the reference standard, only clinical PAIVM ratings of grade 2 and PPIVM ratings of grade 4 were considered positive for LSI. LSI was considered absent for all other data. For analysis of clinical examination data, both clinical and radiographic data were then collapsed into two regions, corresponding to upper lumbar and lower lumbar. This was decided a priori, and considered necessary because there is considerable evidence that therapists are not sufficiently accurate in identifying specific segmental levels by palpation, although they are usually within one level (up or down) and are generally reliable at locating again a segment they had previously located [11-13]. This inaccuracy presented an unacceptable risk of misclassification, that collapsing into regions would attenuate. Furthermore, it is also clear that some physical assessment procedures affect mobility at multiple segments [14] and that segmental specificity does not appear to be important with regard to application of physical therapies for LSI, including manual therapy [5,15-22] (although one study has found otherwise [23]). Data were thus collapsed into the 2 × 2 tables. By-segment results are, however, provided [see Additional file 1] for readers to compare.
Missing data resulted in list-wise deletion of the clinical and radiographic data, on a per-lumbar region, per-analysis basis. The accuracy of the clinical examination items was tested by calculating sensitivity and specificity from 2 × 2 contingency tables. Likelihood ratios were then calculated from these data. These statistics were calculated in Microsoft Excel, using a program written by the primary investigator (JHA). The program calculated 95% confidence intervals (CI) using Wilson's method for sensitivity & specificity, and the score method for likelihood ratios [24]. Methods and results were reported according to the STARD guideline checklist [25].
Results
One hundred and thirty eight (138) consenting patients were recruited for clinical examination. One hundred and eight (108) were recruited in primary care; the remaining 30 presented to a hospital outpatient physiotherapy department. Ten patients failed to present to radiology for flexion-extension radiographs. Five sets of radiographs were of insufficient quality for analysis. Of the 123 included participants, 68 (55%) were males and 55 (45%) females. Further characteristics are described in Table 1. A STARD flow chart is provided in figure 4. No adverse events were reported.
Table 1 Description of the R/CLBP cohort
Mean sd Range N
Age 40.0 11.2 20–75 106
Body mass index 26.7 4.75 19.8–43.0 85
Years since first LBP episode 8.3 8.0 <1–33 104
Disability score (out of 18) 7.13 4.543 0–17 119
Pain level (out of 100) 42.7 25.7 0–100 117
Proportion with constant LBP .23 .420 - 106
Proportion not working due to LBP .12 .331 - 105
Delay between clinical examination and radiography (days) 5 5 -1 – 22 128
Notes: R/CLBP = recurrent or chronic low back pain. sd = Standard deviation; N = number with complete data. Disability score was assessed on the modified Roland-Morris RM18 [57]; Pain level was self-rated on a horizontal 10 cm visual analog scale.
Figure 4 STARD flow diagram.
Nine males and 24 females were available for recruitment into the asymptomatic sample. Three participants violated the exclusion criteria with regard to low back pain history, and were therefore ineligible. The asymptomatic sample therefore comprised of 9 males and 21 females, aged 23 to 60 years (mean 41.3, sd 12.8).
The 27 clinicians who collaborated on this study graduated with their first professional physiotherapy qualification between 1974 and 1996 (mean years since graduation 17, range 6 to 29). All had gained at least one post-graduate qualification in musculoskeletal physiotherapy which included training in manual therapy procedures for the spine, between 1983 to 2000 (mean years since graduation 8.7, range 2 to 19). They spent an average of 31 hours (interquartile range 21 to 40) per week treating patients, with LBP patients comprising, on average, 30% of their patient load (interquartile range 20 to 40).
Prevalence of lumbar segmental instability
Sagittal rotation LSI was not found in statistically significant numbers (6 of 468 segments, or 1.3%), which is smaller than the number that would be expected by chance alone in a normally distributed sample of this size. Sagittal translation LSI was found at a prevalence of 3.6% (17 of 468 segments) (χ2 p < 0.05). In this cohort, 5.6% of individuals had rotation LSI at least one segment, and 12.0% had translation LSI at least one segment.
Accuracy of manual therapy assessment
PAIVMs and PPIVMs were specific for the diagnosis of both rotation LSI and translation LSI, but showed poor sensitivity. The accuracy statistics for PAIVM and PPIVM tests appear in Tables 2 &3. Full 2 × 2 contingency tables are also provided [see Additional file 1]. A positive PAIVM test (grade 2 on a scale from 0 to 2) results in likelihood ratios (LR+) of 2.74 and 2.52 for rotation LSI and translation LSI respectively. Extension PPIVMs performed better than flexion PPIVMs due to their slightly higher sensitivity. A positive extension PPIVM test (grade 4 on a scale from 0 to 4) results in LR+ of 8.4 and 7.1 for rotation LSI and translation LSI, respectively. Likelihood ratios for flexion PPIVMs were not statistically significant.
Table 2 Accuracy of PAIVMs for detecting lumbar segmental instability
LSI Sensitivity (CI) Specificity (CI) LR+ (CI) LR- (CI)
Rotation LSI .33 (.12, .65) .88 (.83, .92) 2.74 (1.01, 7.42) .76 (.48, 1.21)
Translation LSI .29 (.14, .50) .89 (.83, .93) 2.52 (1.15, 5.53) .81 (.61, 1.06)
Notes: PAIVMs = central posteroanterior passive accessory intervertebral motion tests; LSI = lumbar segmental instability; CI = 95% confidence interval; LR+ = likelihood ratio for a positive test; LR- = likelihood ratio for a negative test; Accuracy was assessed by lumbar region (upper lumbar and lower lumbar), overall results are presented; Items in bold type are statistically significant at p < 0.05.
Table 3 Accuracy of PPIVMs for detecting lumbar segmental instability
Flexion PPIVMs
LSI Sensitivity (CI) Specificity (CI) LR+ (CI) LR- (CI)
Rotation LSI .05 (.01, .36) .99 (.96, .1.00) 4.12 (.21, 80.3) .96 (.83, 1.11)
Translation LSI .05 (.01, .22) .995 (.97, 1.00) 8.73 (.57, 134.7) .96 (.88, 1.05)
Extension PPIVMs
LSI Sensitivity (CI) Specificity (CI) LR+ (CI) LR- (CI)
Rotation LSI .22 (.06, .55) .97 (.94, .99) 8.40 (1.88, 37.55) .80 (.56, 1.13)
Translation LSI .16 (.06, .38) .98 (.94, .99) 7.07 (1.71, 29.2) .86 (.71, 1.05)
Notes: PPIVMs = passive physiological intervertebral motion tests; LSI = lumbar segmental instability; CI = 95% confidence interval; LR+ = likelihood ratio for a positive test; LR- = likelihood ratio for a negative test; Accuracy was assessed by lumbar region (upper lumbar and lower lumbar), overall results are presented; Items in bold type are statistically significant at p < 0.05.
Discussion
Despite their widespread use, the validity of PAIVMs and PPIVMs for assessing abnormal sagittal planar motion has not been previously established. We have found PAIVMs and PPIVMs to have high specificity, but poor sensitivity, for the diagnosis of both rotation LSI and translation LSI.
Like sensitivity and specificity, the likelihood ratio for a positive test (LR+) is more powerful when its value is high. Because of the many factors which must be taken into account when applying a diagnostic test to an individual patient (such as the setting the test is used in, purpose of applying the test, prevalence of the disorder, consequences of missing a diagnosis, and risk of harm from the indicated therapy), there are no set cut-off values for sensitivity, specificity, or likelihood ratios, however some authors provide general guidelines [26]. Tests returning LR+ values of 2 to 5 produce small but often useful changes in probability [26], while LR+ values of 5 to 10 (and greater) are more powerful. A test with a likelihood ratio of one is of no clinical utility. The results of this study indicate that a segment testing positive with a PAIVM test is approximately two-and-a-half times more likely to be hypermobile than not [27]. The results for PPIVMs were higher, indicating that a segment testing positive with an extension PPIVM test is approximately seven times more likely to be hypermobile than it is to be normal or hypomobile.
Likelihood ratios for negative tests from this research were less impressive than were the LR+ values, with values between 0.76 and 0.96. None were statistically significant. A LR- closer to zero is more powerful, whereas a LR- of one has no discriminative power. Tests returning LR- values of 0.2 to 0.5 produce small but useful changes in probability, while those with values less than 0.2 are more powerful [26]. This research indicates that a negative result for hypermobility with PAIVM or PPIVM tests is clinically uninformative.
The low prevalence of rotation LSI in this non-surgical, mostly primary care cohort indicate that sagittal rotation hypermobility does not appear to be associated with R/CLBP, as the number of segments hypermobile in rotation is less than the number that would be expected in a sample from a normally distributed asymptomatic population. Sagittal translation hypermobility was found in a significantly higher than expected proportion of patients with R/CLBP (12.0%), and therefore using a Gaussian definition of abnormality (i.e. beyond 2sd from a reference mean) [28] can be considered a valid clinical disorder. Only a small proportion of segments (3.6%) satisfied this Gaussian definition for sagittal translational LSI, however, indicating that it is neither common in this population nor strongly associated with C/RLBP. This may be considered surprising in the light of the emphasis on sagittal translation in the LSI literature [29,30]. This proportion does, however, compare well with clinicians' judgement using PAIVM tests. In the present study, therapists considered 5% of lumbar segments to have manual tests findings positive for LSI. This figure compares well to the 12% of patients with LBP reported to be hypermobile by therapists using PAIVM testing in other research [5]. With regard to the physical examination, though, it is also recognised that assessment of displacement kinematics alone may not be a sufficient basis for the diagnosis of LSI [31,32].
This study has a number of limitations which limit the interpretation of these results. Firstly, while the assessments were nested within a comprehensive clinical examination, and performed in the physiotherapists' own clinical setting, only these three physical assessments were studied in isolation. No attempt was made to identify clusters of assessments that may multiplicatively improve diagnostic accuracy. It is likely that these assessments would have much greater clinical utility within a cluster of other valid signs, symptoms, and history items [16,19]. Furthermore, it may be necessary to adjust the likelihood ratios of these and other tests researched in the future, to remove the influence of conditional dependence, using statistical methods such as logistic regression [33]. Secondly, the prevalence of LSI (using a Gaussian definition of abnormal motion) in this population is low. Defining LSI using a statistical model other than the Gaussian definition used here may result in different prevalence rates. We derived our cut-point for the definition of LSI from the results of our asymptomatic sample; validating the cut-points in another, independent sample would make these results more robust. Sensitivity and specificity, and hence likelihood ratios, may differ in a population with different prevalence rates, such as gymnasts or other athletes, patients with spondylolysis, or surgical candidates [34]. It is also well known that diagnostic tests achieve higher values in the secondary and tertiary care populations, where severity of disease is generally higher [34]. For this reason, too, values may differ in a population with a different spectrum of the target disorder(s), such as patients with spondylolisthesis or higher pain or disability scores. In the primary care low back pain population, the severity of low back conditions is generally low, making differential diagnosis more difficult. In the context of the present population, however, because mechanical low back pain is not life-threatening and the risks of physiotherapeutic interventions are very low [35], moderate index values are acceptable and may still be useful in the diagnosis of low back pain subgroups. Thirdly, analysis of segmental motion from flexion-extension radiographs was limited to sagittal segmental planar rotation and translation. These are properties of displacement kinematics, and as such identify only abnormalities in the quantity of motion. Other parameters of displacement kinematics, such as ratio of translation to rotation [36], instantaneous axis of rotation, and centre of reaction [6] may better characterise abnormalities of movement quality, rather than quantity. Motion abnormalities may also occur in the mid-range of movement and thus cannot be captured on flexion-extension radiographs, but may be detectable by videofluoroscopy. Furthermore, displacement kinematics are only one aspect of segmental motion (and may not be the most important aspect). The physical examination procedures employed by physiotherapists may assess important parameters other than displacement kinematics [32]. This study has not attempted to examine physical assessment of spinal motion velocity, acceleration, or temporal patterns of displacement, nor has it examined physical assessment of kinetics relevant to spinal segmental motion, such as stiffness, viscoelasticity, or force-displacement characteristics. Further research is warranted to fill in the gaps in the literature addressing these limitations.
This research has focussed on the diagnostic accuracy of PAIVMs and PPIVMs, and the multi-centre, pragmatic design of the study precluded assessment of their reliability. The reliability of these clinical assessments has been debated in the literature for many years [37,38]. While many studies have found reliability to be poor [39,40], others have reported considerably better reliability [41,42]. Contrary to popularly held opinion [43,44], it is not easy to conduct a valid and rigorous reliability study. The biostatistical literature points out quite clearly that there numerous difficulties and pitfalls to the study of reliability [45-52] which may threaten the validity of research results. Common methodological problems include violation of the assumptions necessary for the statistical tests used, selection of an inappropriate sample of subjects, lack of true variance in the levels or categories within the sample tested, low prevalence of results across the full spectrum of test scores, and skewed or assymetrical distribution of data. These factors all have a very large impact on the validity and interpretation of much of the literature available on the reliability of these physical examination items: much of the published research regarding reliability may be biased toward the null. It has been argued that tests can be useful for clinical decision-making, in spite of ostensibly low reliability [53], and that it is more important to establish validity of a test or measure [46]. For these reasons, it can be argued that reliability should only be studied in the context of validity [53]. Further research is warranted into these issues.
The first research published in the peer-reviewed literature to test the concurrent validity of these manual assessments for the detection of abnormal segmental rotation appeared in the literature only recently [54], and addressed lumbar segmental hypomobility. The findings of that research indicated that PAIVMs were moderately sensitive (75%) but not specific (35%) for the detection of hypomobility, while flexion PPIVMs were found to be specific (89%) but not sensitive (42%), with a LR+ of 3.9 [54]. Those findings, and others from the literature on predictive validity of hypomobility [16,55,56], are generally consistent with the present results, and represent a gathering body of evidence supporting the validity and clinical utility of these manual clinical assessments.
While the LR+ values reported in the present research are only of moderate strength, they may have some clinical utility. If a patient returns a positive test using the extension PPIVM, this would increase the probability that the lumbar segment being tested has translation LSI from 3.6% (the proportion of lumbar segments found to have LSI in this study) to 20.9%. Even assuming conditional independence of the tests, if the patient then returns a positive test using the central P-A PAIVM, post-test probability that the segment is hypermobile would rise to only 40%. This is, however, still too low for clinical or research usefulness, without further improvement in diagnostic certainty being available from other components of the clinical examination (such as the patients history and interview findings, other patient-derived information, and other physical signs). Research investigating the predictive validity of clinical examination findings has found that manual assessments of a similar nature to be a significantly useful addition to a clinical prediction rule, when combined in a test item cluster with other findings [16,55,56]. These factors mean that the LR+ values found in this study may be of a magnitude sufficient to be useful in clinical practice when combined with other information from the clinical examination.
Conclusion
This study provides the first evidence reporting the concurrent validity of manual assessments for detecting the excessive sagittal planar motion associated with LSI in vivo. PAIVMs and PPIVMs were specific, but not sensitive, for the detection of rotation LSI and translation LSI. Positive PAIVM and extension PPIVM tests had statistically significant likelihood ratios for identifying translational LSI. The validity of the manual therapists' assessments of excessive sagittal planar motion was only moderate, but as these results do not take into account other important parameters of segmental mobility, such as stiffness or viscoelasticity, this level of validity is still encouraging. Further investigation into the validity of the clinical examination for the detection of lumbar segmental motion disorders is warranted, such as whether greater accuracy may be achieved from clinical examination when manual assessments are combined with other information from the patients' history and physical examination.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JHA conceived, designed and coordinated the study, recruited the clinicians, recruited and examined some of the patients, carried out data analysis and prepared the manuscript. JHA retains copyright on all contents. BMcC assisted with measurement technology & data analysis, and manuscript preparation. PH provided statistical support. GM, CC, and TH assisted in clinician recruitment, patient recruitment and examination, data collection, and provided facilities. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Supplementary Material
Additional file 1
Click here for file
Acknowledgements
This project was supported in part by grants from the Department of Anatomy & Structural Biology, the Otago School of Medical Sciences, the University of Otago Research Fund, and the New Zealand Society of Physiotherapists Scholarship Trust Fund. JHA was supported in part by a University of Otago PhD Scholarship. Many thanks to Dr Susan Mercer for assistance with project design and coordination. Thanks to Barry Donaldson, Deidre Johnson, Carole Stevens, Sally Lovell-Smith, Geoff Anderson, Jane Ashby, Mary Connors, Karen Elliot, Rachael Hopkins, Richard Hopkins, Lindsay Jago, Karl Koch, Karl McDonald, Nicola Newlands, Robyn Owen, Michelle Sintmaartensdyk, Mike Stewart, and Sean Wilson, who all recruited & examined two or more patients. Thanks to Drs Georgia Stefanko & Richard Walsh for contributing to radiograph analysis. Thanks also to Marion de Lambert, Rachael Walker, Maggie James, and Karen Wilson for radiography, Sue Wallace, Pat Robertson, and Lesley Dixon for pregnancy screening, as well as consultant radiologists Drs Brett Lyons Andrew Slaven, and Neil Morrison for their willing collaboration.
==== Refs
Maitland GD Edwards BC Vertebral manipulation 1986 5 London; Boston: Butterworths
Magee DJ Orthopedic physical assessment 1997 3 Philadelphia: W.B. Saunders
Phillips DR Twomey LT comparison of manual diagnosis with a diagnosis established by a uni-level lumbar spinal block procedure Manual Therapy 1996 1 82 87 11386842 10.1054/math.1996.0254
Avery AF The reliability of manual physiotherapy palpation techniques in the diagnosis of bilateral pars defects in subjects with chronic low back pain Master of Applied Science thesis 1996 Perth, Western Australia: Curtin University of Technology
Fritz JM Childs JD Flynn TW Whitman JM Wainner RS Segmental mobility testing in the classification of low back pain J Orthop Sports Phys Ther 2004 34 A7 15493525
Bogduk N Amevo B Pearcy M biological basis for instantaneous centres of rotation of the vertebral column Proceedings of the Institute of Mechanical Engineers Part H: Journal of Engineering in Medicine 1995 209 177 183 8519407
Schneider G Instantaneous centres of rotation, centres of reaction, and true translation of lumbar segments: Normative data and reliability Master of Medical Science thesis 1999 Newcastle, NSW, Australia: University of Newcastle
Schneider G Bogduk N Evaluation of new method for determining translation of lumbar spinal segments Proceedings of the Spine Society of Australia Conference: Adelaide 2000
ClaritySMART Spinal Motion Analysis Research Technology
NIH Image
McKenzie AM Taylor NF Can physiotherapists locate lumbar spinal levels by palpation? Physiotherapy 1997 83 235 239 10.1016/S0031-9406(05)66213-X
Downey B Taylor N Niere K Can manipulative physiotherapists agree on which lumbar level to treat based on palpation? Physiotherapy 2003 89 74 81 10.1016/S0031-9406(05)60578-0
Harlick JC Is spinal lumbar palpation a valid clinical measure? New Zealand Journal of Physiotherapy 2001 29 33 34
Kulig K Landel R Powers CM Assessment of lumbar spine kinematics using dynamic MRI: a proposed mechanism of sagittal plane motion induced by manual posterior-to-anterior mobilization J Orthop Sports Phys Ther 2004 34 57 64 15029938
Chiradejnant A Maher CG Latimer J Stepkovitch N Efficacy of "therapist-selected" versus "randomly selected" mobilisation techniques for the treatment of low back pain: a randomised controlled trial Australian Journal of Physiotherapy 2003 49 233 241 14632622
Flynn T Fritz J Whitman J Wainner R Magel J Rendeiro D Butler B Garber M Allison S A clinical prediction rule for classifying patients with low back pain who demonstrate short-term improvement with spinal manipulation Spine 2002 27 2835 2843 12486357 10.1097/00007632-200212150-00021
Flynn TW Fritz JM Wainner RS Whitman JM The audible pop is not necessary for successful spinal high-velocity thrust manipulation in individuals with low back pain Archives of Physical Medicine & Rehabilitation 2003 84 1057 1060 12881834 10.1016/S0003-9993(03)00048-0
Fritz JM Whitman JM Flynn TW Wainner RS Childs JD Clinical factors related to the failure of individuals with low back pain to improve with spinal manipulation J Orthop Sports Phys Ther 2003 33 A4 A5
Childs JD Fritz JM Flynn TW Irrgang JJ Delitto A Johnson KK Majkowski GR Validation of a clinical prediction rule to identify patients likely to benefit from spinal manipulation J Orthop Sports Phys Ther 2004 34 A9 A10
Richardson CA Jull GA Hides JA Hodges PW Therapeutic exercise for spinal segmental stabilization in low back pain: scientific basis and clinical approach 1999 Edinburgh; Sydney: Churchill Livingstone
O'Sullivan PB Lumbar segmental 'instability': clinical presentation and specific stabilizing exercise management Manual Therapy 2000 5 2 12 10688954 10.1054/math.1999.0213
Mayer TG Robinson R Pegues P Kohles S Gatchel RJ Lumbar segmental rigidity: can its identification with facet injections and stretching exercises be useful? Archives of Physical Medicine & Rehabilitation 2000 81 1143 1150 10987151 10.1053/apmr.2000.9170
Chiradejnant A Latimer J Maher CG Stepkovitch N Does the choice of spinal level treated during posteroanterior (PA) mobilisation affect treatment outcome? Physiotherapy Theory and Practice 2002 18 165 174 10.1080/09593980290058544
Altman DG Machin D Bryant TN Gardner MJ eds Statistics with Confidence: Confidence Intervals and Statistical Guidelines 2000 2 BMJ Books
Bossuyt PM Reitsma JB Bruns DE Gatsonis CA Glasziou PP Irwig LM Lijmer JG Moher D Rennie D de Vet HC Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative BMJ 2003 326 41 44 12511463 10.1136/bmj.326.7379.41
Jaeschke R Guyatt GH Sackett DL Users' guides to the medical literature. III. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? The Evidence-Based Medicine Working Group JAMA 1994 271 703 707 8309035 10.1001/jama.271.9.703
Sackett DL Richardson WS Rosenberg W Haynes RB Evidence-Based Medicine: How to practice and teach EBM 1997 Edinburgh: Churchill Livingstone
Sackett DL Straus SE Richardson WS Rosenberg W Haynes RB Evidence-Based Medicine: How to practice and teach EBM 2000 2 Edinburgh: Churchill Livingstone
Nachemson A The role of spinal fusion: Question 8: How do you define instability? How is it diagnosed, and what surgical treatment policy do you follow? Spine 1981 6 306 307
White A Panjabi M Clinical Biomechanics of the Spine 1990 2 Philadelphia: JB Lippincott
Landel RF Kulig K Powers CM Accuracy of manual spinal segmental motion testing as determined by dynamic MRI J Orthop Sports Phys Ther 2003 33 A3 A4
Maher CG Simmonds M Adams R Therapists' conceptualization and characterization of the clinical concept of spinal stiffness Physical Therapy 1998 78 289 300 9520974
Holleman DR JrSimel DL Quantitative assessments from the clinical examination. How should clinicians integrate the numerous results? Journal of General Internal Medicine 1997 12 165 171 9100141 10.1046/j.1525-1497.1997.012003165.x
Jaeschke R Guyatt G Sackett DL Users' guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? Evidence-Based Medicine Working Group JAMA 1994 271 389 391 8283589 10.1001/jama.271.5.389
Flynn TW Move it and move on J Orthop Sports Phys Ther 2002 32 192 193
Frobin W Brinckmann P Leivseth G Biggemann M Reikeras O Precision measurement of segmental motion from flexion-extension radiographs of the lumbar spine Clinical Biomechanics 1996 11 457 465 11415660 10.1016/S0268-0033(96)00039-3
Matyas T Bach T The reliability of selected techniques in clinical arthrometrics Australian Journal of Physiotherapy 1985 31 175 195
Riddle DL Measurement of accessory motion: critical issues and related concepts Physical Therapy 1992 72 865 874 1454862
Maher CG Latimer J Adams R An investigation of the reliability and validity of posteroanterior spinal stiffness judgments made using a reference-based protocol Physical Therapy 1998 78 829 837 9711208
Inscoe EL Witt PL Gross MT Mitchell RU Reliability in evaluating passive intervertebral motion of the lumbar spine Journal of Manual & Manipulative Therapy 1995 3 135 143
Strender LE Sjoblom A Sundell K Ludwig R Taube A Interexaminer reliability in physical examination of patients with low back pain Spine 1997 22 814 820 9106324 10.1097/00007632-199704010-00021
Lundberg G Gerdle B The relationships between spinal sagittal configuration, joint mobility, general low back mobility and segmental mobility in female homecare personnel Scandinavian Journal of Rehabilitation Medicine 1999 31 197 206 10599896 10.1080/003655099444362
Crosbie J Physiotherapy research: A retrospective look at the future Australian Journal of Physiotherapy 2000 46 159 164 11676800
Refshauge K Reflections on the direction of research and PRI Physiotherapy Research International 2002 7 iii v 12109239
Streiner DL Learning how to differ: agreement and reliability statistics in psychiatry Can J Psychiatry 1995 40 60 66 7788619
Portney LG Watkins MP Foundations of clinical research: Applications to practice 1993 East Norwark, Connecticut: Appleton & Lange
Bartko JJ Measurement and reliability: statistical thinking considerations Schizophr Bull 1991 17 483 489 1947873
Brennan P Silman A Statistical methods for assessing observer variability in clinical measures BMJ 1992 304 1491 1494 1611375
Feinstein AR Cicchetti DV High agreement but low kappa: I. The problems of two paradoxes J Clin Epidemiol 1990 43 543 549 2348207 10.1016/0895-4356(90)90158-L
Fleiss JL The design and analysis of clinical experiments 1986 New York: Wiley
Lantz CA Application and evaluation of the kappa statistic in the design and interpretation of chiropractic clinical research J Manipulative Physiol Ther 1997 20 521 528 9345681
Cicchetti DV Feinstein AR High agreement but low kappa: II. Resolving the paradoxes J Clin Epidemiol 1990 43 551 558 2189948 10.1016/0895-4356(90)90159-M
Wainner RS Reliability of the clinical examination: How close is close enough? J Orthop Sports Phys Ther 2003 33 488 491 14524507
Abbott JH Mercer SR Lumbar segmental hypomobility: Criterion-related validity of clinical examination items (a pilot study) New Zealand Journal of Physiotherapy 2003 31 3 9
Fritz JM Whitman JM Flynn TW Wainner RS Childs JD Factors related to the inability of individuals with low back pain to improve with a spinal manipulation Phys Ther 2004 84 173 190 14744207
Childs JD Fritz JM Flynn TW Irrgang JJ Johnson KK Majkowski GR Delitto A A clinical prediction rule to identify patients with low back pain most likely to benefit from spinal manipulation: a validation study Ann Intern Med 2004 141 920 928 15611489
Stratford PW Binkley JM Measurement properties of the RM-18. A modified version of the Roland-Morris Disability Scale Spine 1997 22 2416 2421 9355224 10.1097/00007632-199710150-00018
|
16274487
|
PMC1310529
|
CC BY
|
2021-01-04 16:32:03
|
no
|
BMC Musculoskelet Disord. 2005 Nov 7; 6:56
|
utf-8
|
BMC Musculoskelet Disord
| 2,005 |
10.1186/1471-2474-6-56
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1561627766410.1186/1471-2164-6-156Research ArticleThe zebrafish progranulin gene family and antisense transcripts Cadieux Benoît [email protected] Babykumari P [email protected] David [email protected] Hugh PJ [email protected] Endocrine Laboratory, Royal Victoria Hospital, McGill University Health Centre, Montreal, Quebec, Canada2 Cancer Research Institute, UCSF, 2340 Sutter Street, N-231 San Francisco, CA 94143, USA3 Room L2.05, Royal Victoria Hospital, 687 Pine Avenue West, Montreal, Quebec, H3A 1A1, Canada2005 8 11 2005 6 156 156 29 8 2005 8 11 2005 Copyright © 2005 Cadieux et al; licensee BioMed Central Ltd.2005Cadieux et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Progranulin is an epithelial tissue growth factor (also known as proepithelin, acrogranin and PC-cell-derived growth factor) that has been implicated in development, wound healing and in the progression of many cancers. The single mammalian progranulin gene encodes a glycoprotein precursor consisting of seven and one half tandemly repeated non-identical copies of the cystine-rich granulin motif. A genome-wide duplication event hypothesized to have occurred at the base of the teleost radiation predicts that mammalian progranulin may be represented by two co-orthologues in zebrafish.
Results
The cDNAs encoding two zebrafish granulin precursors, progranulins-A and -B, were characterized and found to contain 10 and 9 copies of the granulin motif respectively. The cDNAs and genes encoding the two forms of granulin, progranulins-1 and -2, were also cloned and sequenced. Both latter peptides were found to be encoded by precursors with a simplified architecture consisting of one and one half copies of the granulin motif. A cDNA encoding a chimeric progranulin which likely arises through the mechanism of trans-splicing between grn1 and grn2 was also characterized. A non-coding RNA gene with antisense complementarity to both grn1 and grn2 was identified which may have functional implications with respect to gene dosage, as well as in restricting the formation of the chimeric form of progranulin. Chromosomal localization of the four progranulin (grn) genes reveals syntenic conservation for grna only, suggesting that it is the true orthologue of mammalian grn. RT-PCR and whole-mount in situ hybridization analysis of zebrafish grns during development reveals that combined expression of grna and grnb, but not grn1 and grn2, recapitulate many of the expression patterns observed for the murine counterpart. This includes maternal deposition, widespread central nervous system distribution and specific localization within the epithelial compartments of various organs.
Conclusion
In support of the duplication-degeneration-complementation model of duplicate gene retention, partitioning of expression between grna and grnb was observed in the intermediate cell mass and yolk syncytial layer, respectively. Taken together these expression patterns suggest that the function of an ancestral grn gene has been devolved upon four paralogues in zebrafish.
==== Body
Background
In mammals, a single progranulin gene, also known as proepithelin, acrogranin and PC-cell-derived growth factor (PCDGF), encodes a glycoprotein precursor exhibiting pleiotropic tissue growth factor activity (reviewed in [1-4]). Progranulin is secreted in an intact form [5-8], or undergoes proteolysis leading to the release of its constituent peptides, the granulins [9-11]. Individual granulins have an approximate molecular weight of 6 kDa, and are structurally defined by the presence of 12 cysteines arranged in a characteristic motif: X2–3CX5–6CX5CCX8CCX6CCX5CCX4CX5–6CX2 [12]. Comparison of the biosynthetic origin of granulin peptides in various mammals reveals that all are commonly derived from a precursor composed of one amino-terminal half followed by seven non-identical copies of the granulin motif.
A role of progranulin in mammalian embryogenesis has been suggested. The exogenous addition of recombinant progranulin to eight-cell stage mouse embryos grown in culture ex vivo accelerates the onset of cavitation, stimulates the rate of blastocoel expansion, and leads to an increase in the number of trophectoderm cells compared to controls [13]. Conversely, the use of a progranulin function-blocking antibody arrests growth and prohibits embryo implantation [13,14]. These results are consistent with the growth-promoting activity of progranulin upon epithelial cells in vitro.
Despite these advances, the evolutionary history and phylogenetic distribution of the progranulin gene outside the mammalian radiation remain largely unexplored. In order to shed light on this issue, and to establish a model for studying the functional contribution of progranulin to vertebrate development, we undertook the characterization of the biosynthetic origins of progranulins in the zebrafish. The widely documented evidence in favor of a pan-genomic duplication event at the base of the teleost radiation [15], commonly referred to as 3R, predicts that the single mammalian progranulin gene will likely be represented by two zebrafish co-orthologues [16,17]. Results reported here demonstrate that, in zebrafish, progranulins arise as members of an extended gene family represented by two distinct architectures, in excess of that predicted by 3R.
Comparative chromosomal mapping of the various progranulin genes (grns) was performed to assist in the establishment of an orthologous relationship to their mammalian counterpart and to provide a point of reference to discuss the evolutionary origins of the distinct progranulin architectures. In support of the duplication-degeneration-complementation (DDC) model [18] gene expression analysis of the zebrafish progranulins reveals spatio-temporal divergence among the different family members, possibly reflecting extensive functional devolution of an ancestral form. Also, the occurrence of natural antisense transcription to some grns suggests that gene dosage may have influenced the retention of extra grn paralogues in zebrafish.
Results
Evidence for a progranulin multigene family in teleosts
A previous study indicated that major forms of granulin peptides found in hematopoietic organs of carp, Cyprinus carpio, differ in their relative abundance [19]. Specifically, extracts of carp spleen contained granulin-1 only, while the head kidney contained granulins-1, -2, and -3, arguing that some granulins found in teleost fish, unlike those found in mammals, have different biosynthetic origins [19]. We revisited the issue of the sole occurrence of granulin-1 in carp spleen and confirmed the earlier report. However, granulin-1 was found to co-purify with another member of the granulin family (Figure 1, Panel A). Sequencing of this peptide suggested that it is a close homologue of mammalian granulin-A, sharing 58% identity with the human peptide (Figure 1, Panel B). The isolation of a carp granulin peptide homologous to mammalian granulin-A suggested that teleosts synthesize a protein equivalent to mammalian progranulin. Carp is a tetraploid species known to express multiple copies of closely related genes that can complicate the study of the origins of multigene families [20]. Therefore, we chose to study the structure and expression of granulin genes in the zebrafish, a closely-related diploid teleost of the cyprinoforme order [21].
Figure 1 Reversed-Phase HPLC purification of granulin-A and granulin-1 from extracts of two carp spleens. Panel A: HPLC fractions derived from an extract of two carp spleens and enriched in granulin-1 immunoreactivity were further purified by HPLC using 0.13% heptafluorobutyric acid as counterion. Two major components were identified, a peptide sharing 58% sequence conservation with human granulin-A (peak A) and granulin-1 (peak 1), each indicated by arrows. Panel B: Sequence comparison of carp granulin peptides with their respective candidate orthologues deduced from cloned zebrafish cDNA sequences (this study), and human granulin-A. Numbers correspond to amino acid position. Characteristic cysteines are shown in bold.
Zebrafish progranulins are represented by two distinct architectures
Progranulin-1 and progranulin-2
A PCR-based strategy using degenerate primers based on the amino acid sequence of carp granulin-1 led to the cloning of two zebrafish cDNAs sharing 92.3% identity and encoding structures homologous to carp granulin-1 and granulin-2, respectively (see Materials and Methods, Figure 2 and Additional File 1). In contrast with mammalian progranulins, the deduced architecture of these zebrafish precursors, named progranulin-1 (grn1) and progranulin-2 (grn2), consist of one full and one amino-terminal half granulin-like repeats only (Figure 2). Unlike progranulins A and B (see below), no potential N-glycosylation sites are found within the predicted progranulin-1 or progranulin-2 polypeptide structures. Subsequent cloning of the respective genes for grn1 and grn2 revealed that each is encoded on five exons (as shown for progranulin-1 in Additional Files 2 and 3), confirming that they are derived from distinct transcriptional units. The genomic sequences encoding grn1 and grn2 are interrupted by short introns in the exact equivalent positions to that observed within the mammalian grn gene [22,23]. Sequences at the splice junctions for the grn1, grn2 and ASgrn1-2 genes are depicted in Additional File 4.
Figure 2 Comparison of the deduced translated sequences for zebrafish progranulin-1 and progranulin-2. Two zebrafish cDNAs sharing 92.3% identity (grn1 and grn2), each possessing a 441 nucleotide-long open reading frame (ORF), encode deduced precursors consisting of one full and one amino-terminal half granulin peptide, with a calculated mass of 14.5 kDa and 14.8 kDa, respectively. Both carry an identical signal peptide (italics), whose predicted cleavage site is indicated by an arrow. The predicted sequences for zebrafish granulins 1 and 2 are highlighted. Numbers correspond to amino acid position.
Hybrid grn RNA
A cDNA encoding a hybrid progranulin identical in size to, and sharing 95.6% nucleotide sequence identity with both grn1 and grn2, was uncovered through our cloning strategy (see Materials and Methods). The nucleotide substitutions for hybrid progranulin are non-randomly distributed among the exons for grn1 and grn2, indicating that it is likely derived from the joining of the first two exons of grn1 with exons 3, 4 and 5 of grn2 (Figure 3 and Additional File 5). A hybrid of the opposite character was not detected (i.e. exons 1 and 2 of grn2 fused to exons 3, 4 and 5 of grn1). No evidence was found to suggest the existence of additional genomic sequences corresponding to grn1 or grn2, raising the possibility of a post-transcriptional mechanism underlying the origin of this chimeric structure.
Figure 3 Genomic organization of zebrafish grn1, grn2 and their complementary antisense gene. Zebrafish grn1 and grn2 genes are found in tandem in a head-to-tail orientation and share an identical exonic organization (exons 1–5, orange and blue boxes, respectively), but differ in their respective intron lengths (A, C, D). The spliced and polyadenylated non-protein coding ASgrn1-2 is encoded on four exons (pink boxes) and shares exon/intron complementarity to both grn1 and grn2 (see text for details). Shown on top is a schematic representation of the chimeric progranulin transcript, suggesting trans-splicing as a mechanism for its generation. Bars (-) above or below exons indicate the relative position of primer pairs (grn1+2 forward and reverse; ASgrn1-2 forward and reverse) used for discriminating between ASgrn1-2 and combined grn1/grn2 expression using RT-PCR.
Antisense progranulin1-2 gene
During the cloning of grn1, genomic sequences were used to perform BLAST searches for sequences deposited at NCBI [24]. An EST was detected (GenBank accession AW777232) whose sequence was an exact match to a portion of the grn1 gene, but in the reverse complement orientation. The full characterization of this EST, designated ASgrn1-2, revealed that it is spliced and shares exonic complementarity to exons 2 and 3, in addition to flanking intronic sequences, of both grn1 and grn2 genes (Figure 3 and Additional File 6). This observation suggested that grn2 is located upstream of grn1 in a head-to-tail organization, thus providing strong support for trans-splicing between the grn1 and grn2 primary transcripts as a mechanism for the genesis of the hybrid progranulin RNA (Figure 3).
In addition to its partial complementarity to grn1 and grn2, the last exon of the ASgrn1-2 gene (corresponding to nucleotides 1015 to 1989 of the cDNA) shares a high degree of sequence conservation with the tzf transposon [25], a subclass belonging to the Tc1/mariner superfamily of class II DNA mobile elements [25]. However, this mobile element is in the reverse complement orientation within the antisense transcript, and has undergone extensive mutations resulting from nucleotide insertions, deletions and point mutations (Additional File 7). Thus, ASgrn1-2 never possessed the ability to encode a translatable transposase protein, nor does it have a clearly predictable ORF in view of the presence of several termination codons in all three possible reading frames. For these reasons, this naturally occurring antisense transcript is considered to belong to the category of non-coding RNA genes.
The zebrafish co-orthologues of mammalian progranulin
Based on the finding of a carp granulin peptide whose sequence was homologous to mammalian forms of granulin (Figure 1, panel B), it was reasoned that these observations could be extended to the zebrafish. Using a strategy identical to that used for cloning a cDNA encoding granulin-1, we isolated a partial cDNA encoding a deduced zebrafish granulin-A peptide (data not shown). BLAST searches using this sequence retrieved two distinct ESTs from GenBank databases (accession AW174591 and AW184435), the former being identical to our cloned sequence. Sequencing of these cDNAs, referred to as progranulin-a (grna) and progranulin-b (grnb) respectively, revealed that each encodes a deduced granulins precursor bearing 10 and 9 tandemly repeated and non-identical granulin domains (Figure 4), demonstrating that the zebrafish progranulin repertoire is not limited to the simplified grn gene structures.
Figure 4 Sequence comparison of zebrafish progranulin-A and -B with human progranulin. Amino acid sequences were deduced from the cloned grna and grnb zebrafish cDNAs. Unlike human progranulin, which carries one-half and seven granulin peptide motifs, zebrafish progranulin-a and progranulin-b harbour 10 and 9 full copies of the granulin motif (boxed), and possess distinct putative signal peptides (italics). Sequences were aligned using the ClustalW method, and gaps were introduced as dashed lines for optimal alignment. Identical residues are indicated by an asterisk. Numbers on the right represent amino acid position. Human granulin motif nomenclature is listed (right).
Zebrafish progranulin-A and progranulin-B are 48.6% identical over their aligned sequences, and each is similarly related to human progranulin with 44.8% and 42.9% identity, respectively. As expected, sequence conservation is seen primarily within the aligned granulin domains of the zebrafish precursors (Figure 4). However, the deduced granulins within the zebrafish grnA and grnB precursors cannot be aligned strictly on the basis of the mammalian nomenclature (i.e. in the order granulin-G, -F, -B, -A, -C, -D, -E) [5,26-28]. Indeed, progranulin-A contains five granulin peptides (domains 4, 6, 7, 8, 9) whose sequences bear close sequence similarity with one-another and with that of human granulin-A. In addition, zebrafish progranulin-A encodes a deduced granulin structure (domain 2) displaying the modified cysteine motif found in human granulin G, which is characterized by the absence of cysteine residues 4 and 7 (Figure 4).
Northern blot analysis
The size of the cloned sense and antisense zebrafish progranulin transcripts was assessed by northern blot analysis (Additional File 8). With the exception of the putative ASgrna (see below), the observed transcript sizes were in agreement with those predicted from cloned sequences. However, many pre-mRNA transcripts of higher than expected molecular weight were identified for several family members. Only grnb demonstrated the presence of splice variants of unknown composition.
Chromosomal mapping of zebrafish grns
The chromosomal localization of each grn gene was determined using the LN54 radiation-reduced mapping panel [29]. Primers are listed in Additional File 9. grn1 and grn2 are closely linked on linkage group (LG) 19, 5.98 centiRays (cR) from EST clone fb47h01 in the vicinity of the simple sequence length polymorphism (SSLP) marker z6661 consistent with their physical proximity deduced from the cloning of a chimeric transcript (Figure 5). Grna is located on LG3, 9.92 cR from SSLP marker z22516, in a region showing a clear syntenic correspondence with the chromosomal position of murine and human grn (Figure 5). In contrast, grnb maps next to SSLP marker z9325, and is 9.76 cR from EST fc18g06, located on LG24 (data not shown), and is not part of a known block of conserved synteny with zebrafish LG3 or human chromosome 17.
Figure 5 Chromosomal assignment of zebrafish grn genes. Zebrafish grna is located close to genes (HoxB cluster, dlx8, pyy) that form an extensive bloc of conserved synteny with human chromosome 17 (Hsa17), indicating an orthologous relationship to human progranulin (grn). Zebrafish grnb, in contrast, maps to LG24 in a region devoid of syntenic correspondence to zebrafish LG3 or Hsa17 (data not shown). Grn1 and grn2 map to LG19, in a region that finds scattered synteny to two human chromosomes (Hsa 6 and Hsa 7). The presence of grn1 and grn2 on a zebrafish chromosome bearing the HoxA cluster, npy and dlx6 genes (i.e. paralogues of genes linked to zebrafish grna and human grn), suggests that grn1 or grn2 may have originated in concert with the mechanism leading to emergence of duplicated Hox clusters at the base of the vertebrate radiation. Map position on zebrafish chromosomes (LG) is presented in centiRays where 1 centiRay = 148 kilobases, the estimated average breakpoint frequency for the LN54 RH panel.
Assessment of zebrafish grn gene expression by RT-PCR
Zebrafish grn gene expression in adult tissues
Semi-quantitative RT-PCR analyses were performed to examine the relative expression of the individual members of the zebrafish grn gene family in adult tissues. A list of primers used and size of the respective amplicons is listed in Additional File 10. Both grna and grnb are expressed in all adult tissues examined, including the gills, heart, multiple visceral organs, and at modest levels in the brain (Figure 6, panel A). Comparison of the expression of the smaller zebrafish paralogues (grn1 and grn2) relative to their mutual antisense gene in selected adult zebrafish organs (Figure 6, panel B), showed that the combined expression of grn1 and grn2 was qualitatively similar to that of grna and grnb (Figure 6, panels B and C). In contrast to the widespread combined expression of grn1 and grn2, low levels of ASgrn1-2 transcripts are detected in the blood and intestine (Figure 6, panel B). In agreement with the observed carp peptide expression profiles [19], the zebrafish spleen expresses grn1 only (Figure 6, panel C). In tissues that display overlapping expression, grn1 is the predominant form in the heart and intestine, while the eyes express higher levels of grn2 (Figure 6, panel C). Surprisingly, low abundance of the hybrid grn transcript, the authenticity of which was confirmed by sequencing of the amplicon, is detected exclusively in the intestine (Figure 6, panel C).
Figure 6 RT-PCR Analysis of zebrafish grns in adult tissues. Panel A: Zebrafish grna and grnb are ubiquitously expressed in various organs. Panel B: A comparison of the combined expression of grn1 and grn2 (grn1+2) relative to their antisense transcript. An increased number of cycles was used in the PCR to allow for the detection of ASgrn1-2 transcripts. Panel C: grn1, grn2 and hybrid grn are differentially regulated with the latter expressed only in the intestine. Results shown were generated using the forward and reverse1 primer pair (Additional File 10). Identical results were obtained using the forward and reverse2 primer pair (not shown). Hybrid grn was amplified using a grn1 forward and grn2 reverse primer combination. No product was obtained using a grn2 forward and grn1 reverse primer pair (not shown). Number of cycles for each reaction is indicated. Amplified PCR products were analyzed by electrophoresis next to a 100-bp DNA ladder. No template and amplification of actin mRNA were used as negative and positive controls, respectively. Similar results were obtained with two other experiments.
Developmental expression of Zebrafish grn genes
Since divergence in expression patterns between duplicated genes is suspected to promote their retention [18], the temporal regulation of grn gene expression between the two grn gene classes during development was investigated by RT-PCR. Transcripts for both grna and grnb are maternally provided although grnb is more abundant (Figure 7, panel A). This trend continues following commencement of zygotic transcription until mid-epiboly (shield stage) where both transcripts are present at similar levels (Figure 7, panel A). In contrast, combined grn1 and grn2 expression is first noticeable during the late pharyngula period by 48 hours post-fertilization (hpf) (Figure 7, panel A). When using an increased number of cycling, the combined expression of grn1 and grn2 is detected as early as 30 hpf, while antisense transcript levels remain too low for detection using conventional ethidium bromide staining (Figure 7, panel B). Southern blot analysis demonstrates that antisense transcription occurs, albeit weakly, at 72 hpf and becomes more evident by 120 hpf (Figure 7, panel B).
Figure 7 RT-PCR analysis of zebrafish grn expression during development. Panel A: grna and grnb transcripts are detected throughout all stages of development, whereas grn1 and grn2 expression is first detected by 48 hours post-fertilization. Maternal expression of grnb is more abundant than grna. Panel B: Combined expression of grn1 and grn2 relative to their antisense transcript. Ethidium bromide stain reveals the presence of sense transcription only (top). Detection of the antisense transcript is revealed by using a 32P-labelled oligo as probe that recognizes both sense and antisense amplicons after Southern transfer. Note weak expression of grn1 and/or grn2 at earlier stages of development. Numbers of cycles used for the PCR are indicated. No template and actin were used as negative and positive controls, respectively. Developmental stages are as follows according to Kimmel et al. 1995: cleavage (16-cell); high (mid-blastula, 3 hpf); sphere-dome (late blastula, 4–4.3 hpf); shield (50% epiboly, 6 hpf); tailbud (10 hpf); 3–7 somites (11–12 hpf); 12–14 somites (14–16 hpf); early (24 hpf) and late (48 hpf) pharyngula; hatching (72 hpf); ealy larval period (96 and 120 hpf). Gene specific primers and amplicon sizes are listed in Additional File 10.
Assessment of zebrafish grn gene expression by whole mount in situ hybridization
In order to evaluate the relationship between sense and antisense transcription, and to gain insights into the potential contributions of the four grn paralogues to development, the spatio-temporal expression of zebrafish grns was monitored by whole-mount in situ hybridization. For all stages examined, sense and antisense riboprobes for sonic hedgehog were used as experimental controls since the tissue expression of this gene is discrete and well documented [30]. No non-specific hybridization signal for sonic hedgehog was detected (data not shown). Unless stated otherwise, the respective sense riboprobes to grna and grnb did not give rise to detectable signals.
Grna and grnb
The low abundance of transcripts for grna relative to grnb revealed by RT-PCR during early embryogenesis (Figure 7, panel A) is reflected in the relative signal intensity for these transcripts by whole-mount in situ hybridization from the 4-cell to late segmentation stage (18–20 hpf) (Figure 8, panel A and B, a–f). Weak, ubiquitous grna expression can first be detected by 12 hpf (6 somite stage), with slightly stronger signal in the hypoblast (Figure 8, panel A, c, d). Grnb expression remains similarly ubiquitous, however defined regions of heightened expression are evident including in the hypoblast, central nervous system (CNS), optic epithelial layers, ear primordium, lateral plate mesoderm (LPM) and tailbud (Figure 8, panel B, c, d and data not shown). As the embryo matures to late segmentation stages (18–20 hpf), low levels of grna expression become confined to the eyes, tectum and tailbud (Figure 8, panel A, e, f), whereas grnb expression undergoes further regionalization within the eyes and CNS, the caudal region of the notochord and surrounding adaxial cells, and in the yolk syncytial layer (YSL) (Figure 8, panel B, e, f). During the pharyngula period (24–48 hpf) and hatching stage (72 hpf), overlapping expression patterns between grna and grnb include the pharyngeal and anterior visceral endoderm, the skin epidermis (Figure 8, panels A and B, g–i), and the pronephric tubules, albeit at modest levels (data not shown). In addition, both genes are expressed temporally within the apical ectodermal ridge (AER) between 36 and 72 hpf (Figure 8, panels A and B, h, i and data not shown). Divergent expression of the zebrafish co-orthologues of mammalian progranulin is also notable during this period (24–72 hpf). For instance, only grna is strongly expressed in the intermediate cell mass (ICM) caudal region, the lens and retina as well as the tectum of the 24 hpf embryo (Figure 8, panel A, g). Grna is also uniquely expressed at low levels within the forming head vasculature and aorta at 48 hpf (Figure 8, panel A, h–i and data not shown), and exhibits relatively increased leukocytic expression from 48 to 72 hpf (Figure 8, panel A, h–k). In contrast to that observed for grna, grnb is expressed within the YSL and often found concentrated at the end of the yolk extension. Sustained high levels of grnb were observed within the brain at all stages examined (Figure 8, panel B, g–k) and in the swim bladder by 72 hpf (Figure 8, panel B, k). Patterns observed during the hatching period generally persist and are accentuated in the 5 day-old larva (Figure 9, panels A and B). In addition, grna can be detected in the epithelial lining of various visceral organs, in particular the pharynx, intestine, swim bladder and pronephric ducts (Figure 9, panel A, a). Weak staining in the dorsal aorta but not the posterior cardinal vein can sometimes be noticed in whole-mount and through sectioning of the animal (Figure 9, panel A, a). Likewise, grnb is widely expressed in the visceral region of the larval stage animal, in particular the intestine, pancreas and YSL (Figure 8 and 9). Weak expression is also detected for both genes in the olfactory epithelium and in the presumptive thymus as bilateral patches located caudal to the eyes (data not shown).
Figure 8 Developmental expression analysis of zebrafish grna and grnb mRNAs by whole mount in situ hybridization. An ontogeny of expression conducted for grna (A) and grnb (B) revealed similar expression patterns for these genes from fertilization to late segmentation stage (a–f), with grna being weaker than grnb. At the 4-cell stage (a) and 50% epiboly (b) ubiquitous expression is observed for grnb only. A lateral view of the 6-somite stage embryo (c) reveals discernable ubiquitous expression above background levels for grna, and increased grnb expression in the epithelium of the eye primordium and CNS, as well as the caudal region. In a dorsal view of the same animals (d), caudal expression in the axial mesoderm (arrow) is observed for both genes, whereas only grnb is detected in the paraxial mesoderm (arrowheads). Lateral (e) and frontal (f) views of the late somitogenesis stage embryo (18–20 hpf) show continued expression in the eye primordium, CNS and tailbud for both genes. In addition, grnb can be detected in the YSL (arrow) (e) and the adaxial cells (arrowheads) (f) flanking the axial mesoderm. At 24 hpf (g), grna expression is found in the tectum and eye retina, in a diffuse pattern in the anterior endoderm (arrowhead) and in a punctuate pattern within the ventral tail region of the ICM (arrow), whereas elevated expression persists for grnb in the forebrain, midbrain and ventral hindbrain region, the eyes, as well as in the YSL, concentrated at the tip of the yolk extension (arrow). In a lateral view at 48 hpf (h), grna expression in the ICM extends rostrally, is detected in the head vasculature, and is now apparent in the skin epithelium, whereas in a lateral view (i) both grna and grnb are transiently expressed in the AER of the pectoral fin buds (arrows). At 72 hpf (j), grna, but not grnb, is expressed in presumed dispersed leukocytes, while in a dorsolateral view (k), grnb can be detected in the swim bladder (arrow). AE, anterior endoderm; AER, apical ectodermal ridge of the pectoral fin buds; ICM, intermediate cell mass; LPM. lateral plate mesoderm; YSL, yolk syncytial layer.
Figure 9 Expression analysis of zebrafish grna and grnb mRNAs by whole mount in situ hybridisation at 5 dpf. At 120 hpf grna (A) exhibits widespread expression in the visceral region, including the pronephric kidneys (arrowheads in sections 1 and 3) and intestine (arrow in section 2), while grnb (B) remains expressed in the YSL and pancreas (arrows in section 1) and is strong in the proctodeum region of the intestine (arrow in section 2). At this stage, a hybridization signal for the sense riboprobe to grna, but not grnb, is detected in the brain, intestine and pronephric ducts (b). Numbered arrows denote the position of corresponding sections shown below (magnified 10×). PD, pronephric ducts; YSL, yolk syncytial layer.
The sense riboprobe corresponding to grna but not grnb detects staining in the brain, intestine and pronephros at this stage (Figure 9, panel A and B, b), suggesting either non-specific hybridization or the presence of an antisense transcript. This observation prompted a search for sequences deposited at GenBank corresponding to parts of the grna cDNA sequence, but in the reverse complement orientation. Three unidirectionally cloned ESTs (accessions CD585878, CD585963 and CD596001) were all revealed upon sequencing to correspond to a 914 nucleotide long cDNA sharing perfect complementarity to nucleotides 2701–3614 of the grna cDNA (Additional File 11, panel A). This sequence corresponds to the 3'UTR region of the grna gene and is not bisected by an intron at the genomic level. This precludes the conclusion that this candidate antisense transcript is not an artifact of cloning. To confirm its directionality, cDNA was synthesized for subsequent PCR amplification using a primer located within the 3'UTR exon of grna that shared complementarity to ASgrna (sense relative to grna) or using another primer that was located downstream of the cloned ASgrna sequence and within a known intron for grna (Additional File 11, panel A). This RT-PCR strategy suggests that the ASgrna deduced from cloned EST sequences may represent a splice variant, and that antisense transcription extends further in the 3'direction (Additional File 11, panel B).
Grn1 and grn2
As expected from the RT-PCR data (Figure 7, panel A), expression for grn1 or grn2 is not detected by in situ hybridization in early development (data not shown). Expression of grn1 is detected in the intestine and the pharyngeal region of the 3-day old animal (Figure 10, panel A) and at very low levels in the pronephros (data not shown). Grn2 does not share this expression pattern and is detected at low levels in the proctodeum (Figure 10, panel B) and is often detected in a few sporadic peripheral leukocytes (data not shown). In contrast hybrid grn is detected at relatively high levels in the proctodeum. (Figure 10, panel C). This specific expression pattern is confirmed by examining the signal obtained using sense riboprobes corresponding to grn1 and grn2 that, after prolonged exposure, detect weak signals from their complementary transcript (ASgrn1-2) in the pharyngeal region of these animals (Figure 10, panels D and E). Notably, the use of a sense riboprobe corresponding to hybrid grn (Figure 10, panel F) does not replicate the expression pattern observed for ASgrn1-2 (Figure 10, panel D and E).
Figure 10 Expression analysis of grn1, grn2, hybrid grn, and ASgrn1-2 in the hatching stage zebrafish embryo by whole mount in situ hybridization. Panel A: grn1 is expressed in the intestine and pharyngeal region (arrow), and at low levels in the pronephric ducts. Panel B: In contrast, grn2 mRNA is only weakly detected in the pharyngeal region (arrow) and the proctodeum (arrowhead), and is occasionally found in dispersed leukocytes (not shown). Panel C: The abundance of the trans-spliced product (hybrid grn) is stronger than grn2 in the proctodeum (arrowhead), but absent in the pharyngeal region. Panels D–F: The corresponding sense riboprobes to grn1 and grn2, but not to hybrid grn, detect ASgrn1-2 expression in the pharyngeal region (arrows). These expression patterns were reproduced in at least three independent experiments.
In the 5 day-old larva, specific grn1 expression is observed in the intestine and the swim bladder, whereas low levels can be seen in the pharynges (Figure 11A–C). Elevated levels of expression for grn2 in the brain, the pharyngeal jaw region, and in presumed peripheral leukocytes (Figure 11D–F), contrast with the pattern observed for grn1. However, both genes share higher levels of expression in the anterior or head kidneys and pronephric ducts (Figure 11B and 11E). In situ hybridization of cross sections of the animals indicates that grn1 and grn2 expression, like that of grna and grnb, occurs in the epithelial lining of the visceral organs (data not shown). The chimeric transcript remained restricted to the proctodeum where it is expressed more abundantly than grn2, and was not detected in leukocytes at this stage (Figure 11G). In turn, the antisense gene to grn1 and grn2 is expressed in the brain, swim bladder, and the middle segment of the intestine (Figure 11H).
Figure 11 Expression analysis of grn1, grn2, hybrid grn, and ASgrn1-2 in the 5 day-old zebrafish larva by mRNA in situ hybridization. Panels A–C: grn1 is expressed in the intestine (arrow), swim bladder, and more abundantly in the head kidneys (arrowheads) than in the pronephric ducts. Panels D–F: grn2 is expressed similarly to grn1 in the head kidneys (arrowheads) and pronephric ducts, but is undetected in the intestine. In contrast, grn2 is strongly expressed in the brain and the branchial jaw region (compare A with D, and C with F), is distributed in a punctuate pattern along the ventral region of the animal in presumed myeloid progenitors (asterisks in D), and often found in randomly dispersed leukocytes (large cells in F). Panel G: Hybrid grn is found exclusively in the proctodeum. Panel H: The sense riboprobe to ASgrn1-2 (devoid of the tzf sequence) recapitulates the combined expression patterns for grn1 and grn2. Panels (I–N): Sense riboprobes to grn1 (I,J) or grn2 (K,L), but not hybrid grn (M), show that antisense transcription occurs in the jaw region (arrows in I and K), the swim bladder, and in the mid-region of the intestine, in a pattern identical to that observed for the antisense probe corresponding to ASgrn1-2 (N). B, E: dorsal views; C, F, I, K: ventral views. For each target mRNA, the use of corresponding sense and antisense (AS) riboprobes is indicated. OC, presumed ossification center; HK, head kidney; PD, pronephric duct; SB, swim bladder.
Discussion
Zebrafish progranulins: simplified molecular forms and orthologues of the mammalian gene
The granulin peptide family was originally discovered as a component of the granule fraction of mammalian phagocytic leukocytes. A series of related cysteine-rich peptides (designated granulins A, B, C and D) were purified from extracts of human neutrophils [9]. The definition of the structure of human progranulin as a glycoprotein bearing multiple copies of the granulin motif made it apparent that granulin peptides are generated through proteolytic cleavage of this precursor within the phagolysosomal compartment of the neutrophil. This explains the roughly equimolar ratios observed for members of the granulin family that are co-packaged within this subcellular compartment. The hematopoietic tissues (spleen and head kidney) of the carp (Cyprinus carpio) were also shown to be abundant sources of three granulin-like peptides (granulins-1, -2 and -3) [19]. However, the non-stoichiometric ratios observed for carp granulins-1,-2 and -3 suggested that the granulin gene family in this teleost species may expand beyond the prototypic single grn gene found in mammals. Specifically, carp spleen contains mainly granulin-1, whereas granulins-1, -2 and -3 were found in extracts of the head kidney (Belcourt et al. 1993). To simplify the identification of teleost granulin gene family members, the zebrafish (Danio rerio) was chosen based on its usefulness as a model of vertebrate development and disease.
The presence of extra gene paralogues in teleost fish has been extensively documented, often [15,31,32] but not invariably supporting [33,34] the hypothesis that the actinopterygian (ray-finned) lineage underwent an additional round of genome duplication (3R) subsequent to diverging from the sarcopterygian (lobe-finned) lineage approximately 450 mya. Specific examples include the Hox clusters [16,35], the annexins [36], the claudins [37] and the Nodal-related genes squint and cyclops [38]. Further, comparative chromosomal mapping studies have shown that duplicated zebrafish genes often reside on distinct chromosomes that exhibit extensive blocs of conserved synteny with their mammalian counterpart [39,40]. Using this approach, it has been estimated that approximately 20% of human genes may be represented by two co-orthologues in the genome of zebrafish [41].
As predicted by 3R, we demonstrate that grns are members of an extended gene family in zebrafish. We have identified two deduced precursors, progranulins A and B, harbouring 10 and 9 granulin peptide repeats respectively that bear close structural and sequence relationship to human progranulin (Figure 4). Grna was localized to a region of LG3 known to show syntenic correspondence to where human grn is found on human chromosome 17 (Figure 5). Grnb was localized to LG 24 rather than being positioned on LG12, predicted to bear synteny with LG3 [40]. Despite this apparent discrepancy, a co-orthologous relationship between grna and grnb relative to mammalian grn is supported by their sequence conservation as well as their extensive overlapping expression patterns observed during development.
Two smaller grn genes, grn1 and grn2, each encoding one full and one amino-terminal half copies only of the granulin motif, were also characterized (Figure 2). Interestingly, a common ancestry of the smaller zebrafish grn genes with grna and grnb, and thus mammalian grn, is implied by conservation of the strict exonic organization that these genes display. However, it is unlikely that grn1 and grn2 arose from the postulated whole genome duplication event corresponding to 3R since they were found to be localized in tandem on LG19 (Figure 5). It is notable that both grn1 and grn2 are linked to a Hox cluster and dlx gene paralogues, similar to that observed for grna. This suggests that a smaller grn may have originated coincidentally with the duplication of a Hox-bearing chromosome in a primordial species. Evidence suggests that this putative structure would then have been retained within the teleosts but lost within the sarcopterygian line of evolution leading to mammals.
Analysis of progranulin expression by RT-PCR
The DDC model predicts that an important driving force behind the retention of duplicated genes is the devolution of an ancestral function onto the resultant pair through quantitative and qualitative changes in gene expression which, when combined, may reflect the sum of the ancestral expression pattern [18]. Thus, it was of interest to determine the extent of expression partitioning and overlap between the two grn gene classes during development by RT-PCR and to compare these patterns to those known for the mammalian counterpart.
As an initial survey of the differential expression patterns of the zebrafish grns, we conducted semi-quantitative RT-PCR analyses using adult tissues and staged embryos. Similar to the well documented widespread expression pattern of human [26,27,42], rat [28], mouse and guinea pig [5]grns in several tissues and cell lines of epithelial, mesenchymal, and hematopoietic origin, both grna and grnb were observed to be expressed ubiquitously in several adult zebrafish organs. In contrast, grn1 and grn2 exhibit a more restricted pattern of expression. It was previously shown that carp granulin-1 and granulin-2 peptides are differently distributed in the spleen and head kidneys of the carp [19]. Interestingly, RT-PCR experiments demonstrate that the expression of the homologous structures in zebrafish were similarly uncoupled at the level of mRNA, (Figure 6, panel C), reflected in a lack of detectable zebrafish grn2 expression in adult spleen. Also, grn1 appears to be the predominant form in the heart, while the eyes express higher levels of grn2. Whether the widespread pattern of expression for zebrafish grns is due in part to leukocyte entrapment in some organs, cells known to express granulins in carp [43] and goldfish [44] cannot be determined using this experimental approach.
Other notable differences in the expression of the paralogous and orthologous pairs of genes were observed. First, maternal transcripts for grna and grnb are readily detectable, with grnb showing higher abundance than grna until early somitogenesis when the two genes become expressed at similar levels (Figure 7, panel A). In contrast, although very low levels of grn1 mRNA are also detected in the newly-fertilized egg, the combined expression of grn1 and grn2 only becomes detectable by 30 hpf, at a stage when most organogenesis is well advanced (Figure 7, panel A). Thus, the absence of grn1 and grn2 expression prior to the onset of zygotic expression argues that these genes are functionally dispensable in early embryogenesis.
Analysis of progranulin expression by whole mount in situ hybridization
Early expression
In order to determine possible roles for the grns during development, their spatio-temporal distribution was examined by whole mount in situ hybridization. Overall, grna and grnb expression patterns share conserved features with their murine orthologue in the early embryo. Transcripts for both zebrafish co-orthologues are maternally deposited and remain ubiquitously expressed subsequent to the onset of zygotic transcription (mid-blastula transition – 3 hpf). Similarly, in a pattern that reflects the replacement of maternal mRNAs with zygotically expressed transcripts [45], murine grn mRNA levels fall rapidly after egg fertilization, reaching negligible levels as early as the 2-cell stage, but rise again to detectable levels by the eight-cell stage [13] Notably, this precedes the morula stage and subsequent blastocyst stage when the epithelium is first formed. It is interesting to note that during zebrafish epiboly, which comprises the morphogenetic movements of the blastoderm towards the vegetal pole (dome to tailbud stage – 4.5 hpf to 10 hpf), grna and grnb are still ubiquitous but more intense in the outer enveloping monolayer of cells (EVL), which ultimately will give rise to an epithelium covering the blastoderm. Expression of these grns in the EVL and later in the skin ectoderm is reminiscent of the more elevated levels of expression for grn in the apical surface of the mouse blastocyst epithelium, the trophectoderm, relative to the inner cell mass population [13].
CNS
Although several regions display increased expression of grna and grnb during brain segmentation (Figure 8, panel A and B, c–f), regional specificity is more apparent at 24 hpf. Distinct or non-overlapping patterns observed include grna expression within the tectum and a more expansive grnb expression pattern encompassing the midbrain-hindbrain boundary, tegmentum and telencephalon (Figure 8, panel B, g). Despite the expression of both grna and grnb in the epithelial lining of the eyes and lens, transient expression within the retina and tectum is noticeable for grna only, suggesting that this orthologue may affect the development of the retinotectal projections. Similar functions may also be implied for mammalian grn, given its expression within the retinal glia during murine development [46]. Interestingly, murine grn expression is abundant throughout the central and peripheral nervous system, similar to that observed for grnb at later developmental time points within the zebrafish CNS (Figure 8 and 9). The generally unrestricted gene expression pattern suggests a role for progranulin in cell survival or proliferation or as a competence factor. Indeed, mammalian grn has been demonstrated to be a potent glial cell mitogen in vitro [47] and is consistently up-regulated in malignant human gliomas [47,48]. Grn2 and ASgrn1-2 but not grn1 demonstrate similar unrestricted expression within the zebrafish brain (Figure 11d,h). Taken together these expression patterns suggest that CNS development in the zebrafish may involve a functional interplay between the various molecular forms of granulin.
The only known functional and physiological contribution that grn gene expression is known to make during neural maturation is its involvement in the sexual differentiation of the male rat brain. It has been shown that male sexual behaviour is associated with steroid-dependent grn expression in the male neonatal hypothalamus [49,50]. Similar distinctions in grn gene expression were not determined in the present study given that sexual differentiation of zebrafish gonads occurs much later and spans roughly 21–28 dpf [51].
Endoderm
The expression of the zebrafish grn gene family members also displays overlapping and distinctive patterns with respect to the endoderm and tissues derived from this germ layer. At 24 hpf both grna and grnb can be found within the pharyngeal and foregut endoderm, whereas only the latter is located within the YSL (Figure 8, panels A and B, g). Both transcripts maintain a degree of dispersed endodermal expression until 120 hpf where grna is located within the epithelial lining of the stomach and anterior intestine while grnb can be found within the YSL, pancreas and proctodeum (Figure 8 and 9).
Unlike the expression of grna and grnb, the paralogues grn1, grn2 and hybrid granulin are highly abundant and for the most part restricted to pharyngeal and visceral endodermal derivatives from 72 hpf onward. Although grn1 and grn2 demonstrate some endodermal tissue-specific expression, these transcripts often co-localize (Figure 11a,b,d,e). The restricted expression of hybrid granulin within the proctodeum is particularly striking (Figure 11g). Consistent with the manner in which the hybrid transcript is formed, both grn1 and grn2 are likewise expressed in the proctodeum. However, expression of these two transcripts overlaps within a large portion of the intestine and stomach where no hybrid grn transcript is found. This suggests that wherever the hybrid grn is observed, the generation of this chimeric peptide must be a highly regulated process and that a specific function is implied.
In contrast to the observed grn distribution in zebrafish, murine grn is not detected within developing endodermal derivatives, with the exception of adult deep crypt enterocytes [42,46]. This suggests that mammalian dependence on grn expression may be developmentally restricted to endodermal-epithelial transitions in the gut or subsequent maintenance of this organ. Alternatively, it is possible that endodermal expression of zebrafish grns reflects a species-specific requirement.
Hematopoietic tissue
Of particular interest in regards to functional equivalence between species and functional separation between duplicated co-orthologues, is the almost complete partitioning of grn hematopoietic expression onto grna and to a much lesser extent grn2. In zebrafish, primitive hematopoiesis occurs within the anterior ICM from where nucleated erythroblasts originate and myeloid cells that can be seen circulating at 24 hpf [52]. Hematopoietic stem cells (HSCs) are then believed to populate the dorsal aorta and yolk sac which represent the zebrafish equivalent of the mammalian aorta-gonad-mesonephros, the tissue presumed to be responsible for later definitive erythropoiesis [53,54]. The posterior ICM, located within the ventral tail and positive for molecular markers of all three hematopoietic lineages, may represent a secondary zebrafish HSC population or region required for HSC maturation [55]. Unlike the murine model, zebrafish definitive hematopoiesis undergoes a migratory transition from the dorsal aorta/ventral tail to the kidney (roughly 96 hpf), without the involvement of liver or bone marrow.
In accordance with expression patterns mentioned previously, grna only acquires distinctive tissue-specificity at 24 hpf, as is the case for its prevalence within the caudal ICM (Figure 8, panel A, g–j), restricting its involvement to definitive hematopoietic waves. Grna can be found at low levels within the dorsal aorta at 48 hpf (data not shown) and is highly expressed within the caudal-ventral tail region throughout all stages post-24 hpf (Figure 8 and 9), suggesting its involvement in multiple hematopoietic lineages. Significantly, sustained grna expression in this hematopoietic organ is coupled with the appearance of grna-expressing leukocytes dispersed throughout the animal with levels that peak at approximately 72 hpf (Figure 8, panel A, j–k). Thus grna expression in presumed granulocytes all over the body of the animal may suggest its involvement in the innate immune response of the host. Grn2 expression is also found within peripheral leukocytes, but in a sporadic pattern that is distinct from that observed for grna. Whether these differences reflect leukocyte sub-populations or activation states for these cells has not been addressed. Commensurate with the transition of ICM to kidney as the major site of hematopoiesis, grna can be found within the pronephric ducts (Figure 9) along with grn1 and grn2, which are also present in the head kidney (Figure 11, panel B and E).
Mammalian grn exhibits a similar expression pattern, particularly in neutrophils. Furthermore, murine grn can modulate the inflammatory response during wound healing, acting as both a chemokinetic factor and inhibitor of neutrophil degranulation and respiratory burst [56,57]. It remains to be determined whether grna supports a similar role in zebrafish.
Chimeric transcription
During the initial degenerate primer amplification of cDNAs encoding grn1 and 2, a third cDNA was cloned and identified as sharing strict identity with portions of both grn1 and 2. Interestingly, this granulin-hybrid showed 100% identity with exons 1 and 2 of grn-1, and with exons 3, 4 and 5 of grn-2, suggesting that this hybrid cDNA may represent a splicing of granulin-1 and 2 primary transcripts (Figure 3). Chimeric transcripts usually result from one of the following mechanisms: chromosomal translocations, transcription of neighboring genes as a single transcription unit or alternative splicing in trans. In all cases, joining of exons is predicted to occur through the recognition of canonical splice acceptor and donor sites.
A hybrid granulin structure has been previously reported through cloning of cDNA sequences in the rat [27]. Specifically, a structural splice variant of progranulin cDNA was retrieved and predicted to encode a granulin domain consisting of the amino-terminal domain of granulin-C fused to the carboxyl-terminal domain of granulin-D, consistent with the removal of an exon from the larger primary transcript [22,28]. The zebrafish hybrid grn described here likely originates through a mechanism other than alternative splicing from a larger primary transcript since the grn2 gene is located 5' to the grn1 gene (Figure 3). This topology was confirmed through the cloning and structural analysis of the partially complementary ASgrn1-2 gene. Also, we found no evidence for the presence of additional grn1-like genomic sequences located upstream of the grn2 gene, or elsewhere in the genome by Southern analysis (data not shown). In particular, no equivalent of carp grn3 was found in zebrafish. These observations suggest that the presence of hybrid grn in zebrafish likely occurs through a splicing reaction in trans between grn1 and grn2 pre-mRNAs, similar to the mechanism originally documented in trypanosomatids [58]. Although rare, scrambled or intergenic RNA molecules consisting of exons originating from distinct genes through a trans-splicing reaction have also been documented in vertebrates. For instance, acyl-CoA:cholesterol acyltransferase-1 (ACAT-1) and the CYP3A family of P450 cytochrome genes produce hybrid mRNA variants in humans [59,60]. Trans-splicing of the voltage-gated sodium channel in response to nerve growth factor stimulation, further suggests that this alternative mode of splicing can be a regulated process [61]. Indeed, there is evidence suggesting that splicing in trans may be facilitated through the recognition of regulatory elements within transcript sequences called splicing enhancers that require binding of SR proteins for activity [62]. Whether grn1 or grn2 harbour a necessary enhancer sequence that could explain the directionality of hybrid grn is currently not known. We believe the hybrid granulin represents the first example of trans-splicing with regards to the modification of a growth factor gene product.
Antisense transcription
Although the majority of deposited zebrafish EST library sequencing confirmed the existence of grn1 and grn2, one particular sequence (AW777232) corresponded to the exact reverse complement to both grn1 and grn2 within the same exon/intron spanning region (exons 2–3 and intervening intron), and was named ASgrn1-2 accordingly. In addition, ASgrn1-2 harbours sequences for an extensively mutated tzf transposon (Tc1/mariner superfamily) in its last exon, but in the reverse complement orientation (Figure 3 and Additional File 7). Despite its polyadenylation, the lack of an ORF classifies ASgrn1-2 as a non-coding RNA. To our knowledge, ASgrn1-2 represents the first example of a single spliced transcript with antisense complementarity to two tandemly organized paralogous protein-coding genes.
The existence of ASgrn1-2 has potential implications in aspects of grn1 and grn2 function. Classically, antisense transcripts often function as inhibitors of the expression of their associated gene, through repression of transcription or promotion of mRNA degradation. There is the possibility that ASgrn1-2 is involved in the formation of hybrid grn wherein this complementary transcript may provide a scaffold to sequester grn1 and grn2 mRNA transcripts within the same intracellular locale, preventing or alternatively facilitating the trans-splicing reaction.
Regarding the RT-PCR data, despite evidence of clear tissue-specific and temporal ASgrn1-2 expression in adult tissues and various developmental stages respectively, this transcript shows no clear reciprocal relationship to grn1 or grn2 (Figures 6 and 7). These expression patterns were confirmed at all stages examined using sense riboprobes for grn1, grn2, and verified for specificity by using the corresponding sense riboprobe to hybrid grn as negative control (Figures 10 and 11, and data not shown). At 5 dpf, ASgrn1-2 is expressed in the presumed hyoid ossification center, stomach and rostral intestine, as well as swim bladder (Figure 11i–n). Although grn1 and grn2 are found within most of these tissues (particularly grn1), specific expression of ASgrn1-2 in the hyoid region undergoing osteogenesis may suggest the required down-regulation of the expression of its counterpart genes during bone development. Therefore a reciprocal relationship may exist between these genes under strict spatio-temporal regulation. Although ASgrn1-2 may mediate a degradation independent mode of gene regulation, such as alternative splicing, it clearly does not perform as a universal negative regulator of grn1 and grn2 expression.
At least one type of transcription modulation may be associated with ASgrn1-2 based on whole mount in situ hybridization analysis. There is a clear reciprocal relationship between hybrid grn expression and absence of ASgrn1-2 transcription. At 120 hpf, hybrid grn is restricted to the distal intestine and despite the expression of its substrate transcripts within several other locales, no other grn1/grn2 rich setting is devoid of ASgrn1-2 (Figure 11). This pattern further suggests that ASgrn1-2 expression prevents formation of the hybrid grn RNA.
The existence of ASgrn1-2 suggested that an equivalent entity may exist for one or both co-orthologues, grna and grnb. Indeed, the sense probe for grna in whole mount in situ hybridization analysis produced a consistent and reproducible signal within the intestine and pronephric ducts at 5 dpf (Figure 9, panel A, b). Several unidirectionally cloned cDNAs (accession numbers (CD585878, CD585963, and CD596001) were found to correspond to the reverse complement sequence of grna within the 3'UTR. Northern blot analysis using sense grna demonstrated a putative 4 kb transcript and directional cDNA synthesis followed by RT-PCR indicated this putative antisense transcript extended within the known grna intronic sequence. However, repeated attempts to clone the full-length transcript by RACE were unsuccessful. Nevertheless, these observations provide evidence for the existence of a second naturally occurring grn complementary transcript, namely ASgrna.
The granulin motif – phylogenetic and functional implications
The granulin motif is so distinctive that it is a relatively simple matter to search the protein and nucleotide databases to obtain an impression of its phylogenetic origins. Figure 12 shows a diagrammatic representation of the nature of proteins bearing the granulin motif that can be confidently predicted through searching the cDNA, EST and/or genome sequence databases using BLAST [24]. The numbers of full granulin modules present in each putative protein is shown in cartoon form together with an impression of the phylogenetic origins of the species involved. The source of the sequence data used to predict these structures is listed as accession numbers in the legend to Figure 12. The distribution of some of the proteins bearing the granulin motif within assorted species has been partially annotated and can be viewed at various web sites [63,64]. With some interesting exceptions (e.g. Drosophila melanogaster), the granulin motif can be found throughout eukaryotes, both in plants and multicellular animals. In contrast there is no representation within fungi or unicellular organisms. The presence of a protein bearing the granulin motif within the slime mold (Dictyostelium discoideum) is particularly revealing since this organism is thought to be a modern representative of an amoeboid organism that was a transition species between unicellular and multicellular eukaryotes [65]. The appearance of this organism is thought to predate the divergence of plants and animals about 1.5 billion years ago. This suggests that a primordial gene bearing a single granulin motif evolved once during this transition period. The plant motif is found in just one context as a carboxyl-terminal domain of a cysteine protease that is found in many members of the viridiplantae including Arabidopsis thaliana [66] (Figure 12). Together this suggests that the founding granulin gene was represented by a single copy and composed of a single granulin motif.
Figure 12 Diagrammatic representation of the structures and evolutionary origins of the granulin in multicelluar organisms. Evolutionary distances in millions of years derived from Hedges 2002 [77]. The estimated rounds of vertebrate genome duplication events are indicated (1R, 2R, 3R). The various progranulin structures were derived from various databases as outlined below. Land plant – (Arabidopsis thaliana) – Papain-like thiol protease bearing a carboxyl-terminal granulin domain (AAK71314); Slime mold – (Dictyostelium discoideum) – 1 copy progranulin from a single EST (AU267401) Trematode worm (Schistosoma japonicum) – 9 copy progranulin built from a combination of four ESTs (AY810079, BU790215, BU799560, BU771494); Nematode worm – (Caenorhabditis elegans) – 3 copy progranulin from single EST (NM_060580) – overall architecture confirmed by genome sequence (Z81595); Annelid worm (earthworm – Lumbricus rubellus) – 1 copy progranulin from a single EST (CO046860); Primitive chordate – (Sea squirt, Ciona intestinalis) – 7 copy progranulin predicted from draft genomic sequence (AABS01000126) and overall architecture confirmed by ESTs. Domains 2 to 7 nearly identical (BW368775, BW311239). Amphibian – (Frog, Xenopus laevis) co-orthologues progranulins A and B consisting of one-half domain followed by 12 full domains and one-half domain followed by nine full domains respectively [78]. These are structurally closely related and are a result of a recent tetraploidization event 30 mya [79]. Avian (chicken – Gallus gallus) – 4 copy progranulin built from three ESTs (BM440305, BU297352 and BX265765) – overall architecture confirmed by genome sequence (LOC426606); Human (Homo sapiens) – Progranulin composed of one half domain followed by 7 domains (UniProt entry: P28799); Teleost – Takifugu rubripes – 2, 3 and 11 copy progranulins predicted from draft genome sequences (M000077, S002118, S0001020) with overall structures confirmed by ESTs (CA846088, CA332411, AL842916, CA588603). Teleost – Danio rerio: from this study: co-orthologue progranulins A and B composed of 10 (NM_001001949) and 9 (AY289606) respectively; two smaller progranulins consisting of 1 and one half granulin repeats (AF273479, AF273480).
In contrast to the conserved protease/granulin gene architecture found in plants, members of the animal kingdom have expanded their granulin repertoire, not via genomic or segmental duplication, but rather through tandem multiplication of the granulin ORF. For instance the Xenopus leavis and the early chordate Ciona intestinalis granulin genes encode of five and six tandemly repeated near identical granulin motifs, respectively. The variance in the number of granulin intragenic regions demonstrates a degree of plasticity during grn gene expansion. Intragenic multiplication is unlikely to have occurred as a single ancestral event; rather grn gene expansion has taken place independently in various species to varying extents. This mechanism of conserved domain repetition is not unique to grn genes. Indeed, the same type of genetic expansion is likely responsible for the repetition of immunoglobulin, EGF and lectin domains in numerous proteins.
The presence of single grn genes in protostomes is not surprising since the origins of these species predate the estimated genomic duplications within the vertebrate radiation (1R and 2R, Figure 12); these large-scale events appear not to have given rise to grn gene expansion. Specifically, all members of the sarcopterygian lineage (derived from lobe-finned fish) harbour a single granulin gene of varying motif number, indicating that intragenic multiplication remains the preferred and tolerated means of granulin expansion for most vertebrates, including mammals. The same intolerance to gene duplication has not encompassed the actinopterygians (ray-finned fish), including Danio rerio and Takifugu rubripes. These species have undoubtedly expanded their granulin gene repertoire through tandem intragenic expansion as well as genome duplication (3R, Figure 12), to yield the co-orthologues grna and grnb. Interestingly, the existence of grn1 and grn2 does not necessarily conform to these methods of gene expansion, suggesting that a specific and unidentified means of gene expansion, possibly involving ASgrn1-2, may exist.
Conclusion
Although the existence of two co-orthologues of mammalian progranulin in zebrafish is likely a result of genome-wide duplication, similar genetic events have occurred within chordates prior to divergence of the ray-finned (teleost) and lobe-finned (mammalia) radiations. Indeed two rounds of genome-wide duplication are believed to have occurred [67]. More precisely, the first genome duplication probably occurred in a common ancestor of all agnathans and gnathostomes after its divergence from cephalochordates, ~594 mya (million years ago). The second round is presumed to have occurred ~488 mya, within the lineage leading to jawed vertebrates after the jawless line diverged, presumably before the split between cartilaginous and bony fish. Despite this, all mammals studied thus far have retained only a single copy of the progranulin gene, whereas two rounds of genome duplication would theoretically create four progranulin genes. It is therefore interesting to consider the biological rationale behind retention of grna and grnb following the teleost genome duplication, an event not permitted within other vertebrates, in conjunction with the appearance of two extra paralogues, grn1 and grn2. Regulation by gene dosage through complementary transcription may have allowed for the retention of the smaller paralogues, while putative antisense transcription to grna may be necessary for precisely regulating the spatio-temporal activity of this growth factor.
Overall, the expression patterns of zebrafish progranulins faithfully replicate those observed for the mouse counterpart in a similar context [42,46]. Importantly, this indicates that the use of zebrafish will enable modeling of the contributions of progranulin activity to vertebrate development through investigating both grna and grnb. These studies will be uncomplicated by the presence of grn1 and grn2, whose expression patterns largely do not overlap with the co-orthologues. Overall, the expression patterns for the grns indicate that these growth factors may subserve multiple functions in vivo that are consistent with the known role of their mammalian counterpart in cell growth, motility and survival.
Materials and methods
Tissue extraction and granulin peptide purification
Carp (Cyprinus carpio) were purchased live at a local fish market (Waldman Plus, Montreal, QUE). Peptides from fish spleens were extracted using C18 Sep-Pak cartridges (Waters Canada Ltd. Mississauga, ONT) and separated using reversed-phase high-performance liquid chromatography (RP-HPLC) on a C18 Bondapak column (Waters) as previously described [19,68]. Column fractions were screened for cysteine content by amino acid analysis and granulin-1/2 immunoreactivity by radioimmunoassay [19]. Fractions positive for both criteria were further purified by RP-HPLC using solvents containing 0.13% (v/v) heptafluorobutyric acid as counter-ion, and subsequently purified to homogeneity using the original solvent system containing 0.1% (v/v) trifluoroacetic acid. Molecular weight of purified peptides was determined using a Voyager matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometer (Perceptive Biosystems, Framingham, MA) located at the Sheldon Biotechnology Centre of McGill University.
Microsequencing of carp granulin-A
The putative carp granulin-A was alkylated according to a previously published protocol [19]. Approximately 20 μg of the alkylated carp granulin-A was digested with sequencing grade chymotrypsin (Roche Diagnostics Canada, Laval, QUE) according to the manufacturer's instructions, and the resulting fragments separated by RP-HPLC on a C18 Bondapak column. Amino-terminal sequence analysis of carp granulin-A and its chymotryptic fragments was undertaken using a Procise sequencer (Applied Biosystems, Foster City, CA) located at the Sheldon Biotechnology Centre of McGill University.
Fish husbandry
Wild type zebrafish were purchased from Scientific Hatcheries (Huntington Beach, CA) and maintained on a 14 h/10 h light/dark cycle at 28.5°C in a laboratory aquarium (Allantown Aquaneering, Allantown, NJ). Fish were fed twice daily, and bred as described elsewhere [69]. Embryos for developmental studies were collected from tanks and staged according to conventional criteria [70] and by hours post-fertilization (hpf).
Library screening and cloning of zebrafish progranulins
The zebrafish grn1 cDNA was cloned using a PCR strategy (Additional File 1). The carp granulin-1 amino acid sequence was used to design degenerate forward DF1 (5'-GTI ATY CAY TGY GAY GC-3') and reverse DR1 (5'-CAR CAR TGR ATI CCR TC-3') and DR2 (5'-TCR CAR TGR TAI CCR TG-3') primers for use in the polymerase chain reaction (IUPAC codes are used to refer to the bases in primer sequences). The template for the initial amplification reaction was a 5'-STRETCH plus cDNA library cloned in lambda gt10 vector (Clontech BD Biosciences, Mississauga, ONT). cDNA for this library was prepared from 1-month-old zebrafish using a combination of oligo-dT and random priming. 0.25 μl of library (approximately 108 pfu/ml) was used in a final reaction volume of 100 μl for each new amplification attempt. The annealing temperature was determined empirically in order to maximize yield of product. PCR amplifications were performed with Taq DNA polymerase, unless specified otherwise, using a Hybaid thermal cycler from Bio/Can Scientific Inc. (Etobicoke, ONT). Amplified products were isolated by agarose gel electrophoresis, purified with the QIAquick Gel Extraction Kit (Qiagen Inc. Mississauga, ONT) and sequenced after cloning into TOPO pCR2.1 (Invitrogen, Carlsbad, CA). An initial reaction using the DF1 and DR1 primer pair yielded several products. 5 μl of this reaction was subjected to re-amplification using DF1 primer in combination with the nested (anchored) DR2 primer, which revealed a product of 126-bp encoding a partial sequence for granulin-1 (Additional File 1, step 1). New grn1 primers F126 (5'-ACTGTGTGTCCAGACGG-3') and R215 (5'-CCATCCCTGCAACACTG-3') were then designed based on this sequence and were used, respectively, in combination with flanking gt10 primers in order to obtain the 5'- and 3'-untranslated region (UTR) cDNA sequences (Additional File 1, steps 2 and 3). Finally, the entire ORF was amplified with Pwo DNA polymerase (Roche Diagnostics), using forward F1 (5'-ATGTTCCCAGTGTTGATG-3') and reverse R (STOP) (5'-GCTTACAACTCCAACCCG-3') primers (Additional File 1, Step 4). This PCR was performed in a final volume of 100 μl, containing 0.25 μl of library, 50 mM KCl, 10 mM Tris-HCl, (pH 8.8), 1.5 mM MgCl2, 0.1% Triton X-100, 0.2 mM concentration of each dNTP, 0.5 unit of Pwo DNA polymerase, and 100 pmol of each primer. An initial denaturation step was carried out at 94°C for 3 min. Annealing temperatures of 54°C, 56°C and 58°C were used sequentially for 10 cycles each. Typical denaturation, annealing, and amplification reactions were carried out at 94°C for 30 sec, 54°C for 1 min, and 68°C for 1 min, respectively. A final extension step of 10 min at 72°C was carried out after adding 0.25 unit Taq DNA polymerase. An amplification product specific for grn1 was sequenced on both strands. The 5'-UTR, and a portion of the 3'-UTR for grn1, were amplified using grn1-specific primers in conjunction with a lambda gt10 primer. Distinct cDNAs encoding progranulin-2 and a chimeric progranulin were uncovered through this approach. Each transcript was confirmed through sequencing of independent amplification reactions using template cDNA derived from either adult organs or embryos of mixed stages. Following a strategy similar to that used for the isolation of zebrafish grn1 cDNA, primers based on the purified carp granulin-A peptide sequence were designed to clone partial cDNAs for zebrafish grna and grnb, respectively (not shown). BLAST searches using the cloned sequences retrieved two unique ESTs at NCBI sharing an exact match with grna and grnb, respectively (accession numbers AW174591 and AW184435). These respective ESTs were purchased from RZPD GmbH (Heidelberg, Germany; clone ID: UCDMp574E2318Q2 and UCDMp574I0223Q2) and sequenced on both strands to create a final assembly of the full-length cDNAs encoding zebrafish progranulin-A and progranulin-B. In addition to our cloning strategy, the rapid amplification of cDNA ends (RACE) was performed with the GeneRacer kit (Invitrogen, Burlington, ONT) using total RNA isolated from adult zebrafish intestine. For grn1 and grn2 transcripts, a reverse primer that corresponded to nucleotides 195–215 based on the cloned ORF of both transcripts (5'-CCATCCCTGCAACACTGACCC-3'), was used to perform the 5' RACE, while a forward primer corresponding to nucleotides 1–22 of each transcript (5'-ATGTTCCCAGTGTTGATGTTAC-3') was used to perform the RACE in the 3' direction. Similarly a 5' UTR sequence for grna was obtained using the map reverse primer (see below). Repeated RACE attempts in both directions for ASgrna were unsuccessful.
Cloning of the zebrafish grn1 gene
A zebrafish genomic library constructed in P1 artificial chromosome (PAC) [71] and represented on filters at high-density (RZPD GmbH) was screened for the presence of the grn1 gene using standard procedures. The cDNA bearing the grn1 ORF was labeled with [α-32P] dCTP by random priming using the Oligolabeling kit (Amersham Biosciences, Baie d'Urfe, QUE) for use as probe, and purified using a Sephadex G-50 column (Amersham Biosciences). Kodak X-OMAT AR film was used for autoradiography (Fisher Scientific Ltd, Whitby, ONT). Three positive clones (706K2254Q, BUSMP706K14116Q2, 706F20133Q2) were detected by autoradiography, and the first two were confirmed to carry at least part of the grn1 gene by PCR, using the F1 (5'-ATGTTCCCAGTGTTGATG-3') and R215 (5'-CCATCCCTGCAACACTG-3') primer pair which does not discriminate between grn1 and grn2, and sequencing. DNA from a positive clone (706K2254Q) was purified with the Plasmid Midi Kit (Qiagen). 1.5 μg of this DNA was subjected to restriction digest with EcoRI to generate fragments suitable for cloning into pBluescript II KS (Stratagene, La Jolla, CA), and was followed by transformation in TOP 10F' electrocompetent cells (Invitrogen). Screening of colonies transferred onto nitrocellulose membranes (Xymotech, Montreal, QUE), employing the same probe used for the original library screening, was performed in the following prehybridization and hybridization conditions: 2 × SSC, 0.5% SDS, 0.05% Na Pyrophosphate at 65°C. Membranes were washed twice in 1 × SSC, 0.1% SDS, 0.05% Na Pyrophosphate at 60°C for 15 min, followed by two washes in 0.1% SSC, 0.1% SDS, 0.05% Na Pyrophosphate at 60°C for 10 min. Plasmid DNA from a positive clone was purified using the high pure plasmid isolation kit (Roche Diagnostics, Laval, QUE). An insert of ~9-kb was fully sequenced and revealed the presence of the promoter region and approximately half of the grn1 gene. The remaining gene sequence was found in a ~6-kb insert clone isolated by re-screening the colony lifts with 32P-labeled reverse R(STOP) oligonucleotide (5'-GCTTACAACTCCAACCCG-3') as probe. A PCR was performed using primers flanking this EcoRI site, and sequenced to confirm that the isolated 9 kb and 6 kb clones represent continuous sequences within the original PAC clone.
Retrieval of antisense transcripts from NCBI
While in the process of analyzing cloned grn1 genomic sequences (data not shown) through BLAST searches for corresponding sequences at GenBank, an EST harbouring sequences corresponding to unspliced grn1, but in the reverse complement orientation, was noticed. This clone (accession number AW777232) was purchased (RZPD, clone ID: DKFZp717B091Q2) and further analyzed through sequencing. A putative transcript exhibiting perfect complementarity to the 3'UTR region of zebrafish grna (ASgrna) was deduced from sequencing four unidirectionally cloned ESTs deposited at GenBank (CD585878, CD585963, CD596001 and CD588938) that originated from an oligo-dT-primed cDNA synthesis from adult kidney marrow RNA (Song et al. 2004; kindly provided by Dr. Chen, Shanghai Institute of Biological Science). Unidirectional cDNA synthesis using total RNA derived 5 day-old larvae was synthesized using a sense primer relative to the 3'UTR exon (ASgrna 2) or to a known intron (ASgrna 3) of grna (Additional File 10 and 11). These primers were then used in conjunction with the following primer (ASgrna 1) in subsequent RT-PCR to confirm antisense transcription to grna.
Chromosomal assignment and syntenic analysis
Zebrafish grns were mapped using the LN54 Radiation Hybrid Panel as previously described [29]. Primers for each gene are noted in Additional File 9. Each PCR reaction was carried in a final volume of 20 μl containing 100 ng "hybrid DNA", 500 mM KCl, 100 mM Tris-HCl (pH 8.3), 15 mM MgCl2, 0.2 mM each dNTP, 1 unit Taq DNA polymerase, and 5 pmol of each oligo. Denaturation, annealing and amplification were performed at 94°C for 30 sec, 55°C (grn1 and grn2) or 60°C (grna and grnb) for 30 sec, and 72°C for 30 sec, respectively, followed by an extension step of 7 min at 72°C. To determine syntenic relationships between zebrafish and human genomes, mapped zebrafish genes flanking a given zebrafish grn gene were identified using the consolidated zebrafish maps available from ZFIN [72] and data from LocusLink [73].
Gene expression profiling by RT-PCR
Total RNA from various adult tissues and developmental stages was isolated using Trizol LS reagent (Gibco BRL, Burlington, ONT), treated with DNaseI and used in first strand synthesis using the Revert Aid H-synthesis kit (MBI Fermentas Inc. Burlington, ONT). PCR conditions used for each family member consisted of an initial denaturation at 94°C for 2 min, followed by 30–40 cycles at 94°C for 45 sec, gene-specific annealing temperature (Additional File 10) for 1 min and extension at 72°C for 1 min, with a final single cycle extension at 72°C for 7 min. To discriminate between grn1, grn2 and hybrid grn, an (NH4)2SO4 buffer with 1 mM MgCl2 (MBI Fermentas) was used for RT-PCR in conjunction with cloned template controls (pBluescript, Clonetech) for all three forms using the same reaction parameters. Two independent reverse primers for grn1 and grn2 yielded products of expected size and hybrid grn was produced using grn1 forward with either of the grn2 reverse primers. All PCR products were resolved on 2% agarose gels, ethidium bromide stained and visualized on Polaroid 667 Film. The authenticity of all PCR products was confirmed by sequencing after cloning into TOPO/pCR2.1.
Northern blot analysis
Full-length mRNA transcript size was assessed for each progranulin family member by Northern analysis of poly-A enriched mRNA (Micro Poly (A) Purist Small Scale Purification Kit; Ambion) derived from whole adult or 5 dpf animals [74]. Hybridization (Ultrahyb; Ambion, Austin, TX) was carried out using non-radioactive biotin-labeled cRNA probes (Psoralen-Biotin Non-iosotopic labeling Kit; Ambion) and detected with Brightstar BioDetect Nonisotopic Detection Kit (Ambion) according to manufacturers instructions. Band size was determined using pre-labeled biotin markers (Ambion).
Whole-mount mRNA in situ hybridization
In situ hybridization for progranulin family gene expression was carried out essentially as previously described [75]. Briefly, digoxigenin-labeled RNA probes for each full-length cDNA, with the exception of ASgrn-1/2 which corresponded to exons 1–3 only, were hybridized at 70°C using various developmental stages from cleavage to larval. In some cases, polyvinyl alcohol was added to the staining solution in order to minimize the occurrence of background, especially when the reaction was required to proceed for several days [76]. Stained whole-mount and sectioned embryos were mounted in glycerol and visualized under a Leica MZFLIII stereomicroscope (Richmond Hill, ONT). Pictures were taken with a Leica DC350F camera and processed with Adobe Photoshop 7.0 software.
Sequence Accession Numbers
GenBank accession numbers of all zebrafish proranulin genes and antisense transcripts described in this paper are as fellows: grn1, AF273479; grn2, AF273480; hybrid grn, AF273481; ASgrn1-2, AY289607; grna, AF375477, ASgrna, AY826190; grnb, AY289606 .
List of abbreviations
AER, apical ectodermal ridge; ASgrn, antisense granulin; CNS, central nervous system; DDC, duplication-degeneration-complementation; EVL, enveloping monolayer of cells; grn, granulin; hpf, hours post-fertilization; HSC, hematopoietic stem cell; ICM, intermediate cell mass; LG, linkage group; LPM, lateral plate mesoderm; mya, million years ago; SSLP, single sequence length polymorphism; YSL, yolk syncitial layer.
Authors' contributions
Benoît Cadieux established and justified the zebrafish animal model, carried out all the molecular genetic studies and prepared the initial draft of the manuscript. Babykumari P. Chitramuthu carried out the whole-mount in situ hybridization studies and Northern blot analysis, participated in the design of the study and helped draft the manuscript, David Baranowski participated in the design of the study and helped draft the manuscript. Hugh P. J. Bennett conceived of the study, co-ordinated its design and helped draft and finalize the manuscript. All co-authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Cloning strategy for the cDNA encoding the precursor for zebrafish granulin-1. Panel A: The full-length cDNA for progranulin-1 is represented at the top of the diagram. Black rectangles represent the ORF, and blank rectangles represent the respective 5' and 3' untranslated regions. The dashed lines represent lambda phage (vector) sequences. Numbers on the left represent the sequential order of PCRs undertaken (see Materials and Methods section). Panel B: Deduced amino acid sequence for the precursor encoding granulin-1, consisting of one and one-half repeats of the granulin consensus motif. Characteristic cysteines are underlined and in bold. A predicted leader sequence is shown in italics. The full granulin-1 peptide sequence (35–91) is separated from the amino-terminal half peptide (116–147) by an intervening sequence. Stop codon is represented by *. Numbers represent amino acid position.
Click here for file
Additional File 2
Genomic architecture of the zebrafish grn1 gene. Exons (1–5) and deduced amino acid sequences are shown in uppercase letters and are boxed, while introns (A–D) and flanking sequences are shown in lowercase. 5' extension of exon-1 is based on a deposited EST sequence (accession number BG884011) and 5' RACE. A potential TATA box in the promoter region is in bold and italics. The translation initiation codon (ATG) and polyadenylation signal sequence (AATAAA) are in italics. The EcoRI restriction site (gaattc) used for cloning the genomic fragments for this gene is located in intron C, and is underlined and bold. An identical exonic architecture was found for grn2 (data not shown).
Click here for file
Additional File 3
Conserved exonic organization of zebrafish grn1 and grn2 genes relative to mammalian progranulin. Like granulin repeat units found in mammalian progranulin the nucleotide sequences encoding zebrafish granulin-1 (and granulin-2, not shown here) is derived from the joining of two spliced exons with phase 0 boundaries. This characteristic splicing occurs at nucleotide positions corresponding to four amino acids after cysteine 6 within the amino-terminal region (exon 2), and two amino acids before cysteine 7 in the carboxyl-terminal end (exon 3). The relative sizes of exons (1–5) and introns (A–D) is indicated.
Click here for file
Additional File 4
Splice junctions of the zebrafish grn1, grn2 and the non-protein coding ASgrn1-2 genes. The consensus sequence for splice donor and acceptor sites is shown on the top line (Breathnach and Chambon, 1981). The nucleotide sequences surrounding the sites for introns A–D for the respective grn1 and grn2 genes, as well as for introns A–C of the ASgrn1-2 gene, are shown. Exons are in uppercase, introns in lowercase. Phase of introns interrupting open reading frames are indicated.
Click here for file
Additional File 5
Cloning of a chimeric transcript that encodes a hybrid progranulin. Panel A: Nucleotide sequence alignment of the ORFs for progranulin-1, chimeric progranulin, and progranulin-2. RACE confirmed that the cloned cDNAs possess identical 5' and 3' untranslated regions (not shown). After verifying the cDNA nucleotide sequences for grn1 and grn2 with corresponding exonic segments (boxed), it was found that for the chimeric (hybrid grn) transcript, all except one nucleotide substitution (T instead of C, exon 4), are conserved and non-randomly distributed among corresponding exonic sequences of either the grn1 (highlighted in blue, exon 2) or grn2 (highlighted in orange, exons 3–5) genes. The translation initiation (ATG) and termination (TAA) codons are in bold. Arrows indicate location of primers used for RT-PCR analyses (bold; see Additional File 10). Panel B: Sequence alignment of the deduced translated sequences for progranulin-1, progranulin-2, and hybrid progranulin. The candidate chimeric transcript consists of the amino-terminal portion of grn1 (exons 1 and 2) and of the carboxyl-terminal portion of progranulin-2 (exons 3 to 5). The position of introns (A–D) located in the respective grn1 and grn2 genes are indicated by arrowheads. The granulin-1 peptide and amino-terminal half-domain are underlined.
Click here for file
Additional File 6
Nucleotide sequence of a transcript antisense to zebrafish grn1 and grn2 genes. The 1989 nucleotide ASgrn1-2 cDNA is encoded on four exons (boxed) and bears a polyadenylation signal (AATAAA) (bold and italics). Sequences corresponding to exons 2 and 3 of ASgrn1-2 are complementary to regions encompassing the second and third exons (uppercase and bold) and intronic sequences (lowercase) of the grn1 and grn2 genes, respectively. The nucleotide sequence of exon 4 corresponds to a mutated transposase gene of the tzf transposon sub-class of the Tc1/mariner superfamily of mobile elements, but in the reverse complement orientation. ASgrn1-2 is a non-protein coding RNA based on the absence of a predictable open reading frame.
Click here for file
Additional File 7
The genomic region encompassing exon 4 of the ASgrn1-2 gene corresponds to a defective transposon of the tzf family, in the reverse complement orientation. A consensus 1621 nucleotide sequence for the tzf transposon, deduced by the majority rule from aligned sequences obtained from GenBank EST entries (U51226-U51230) and published data (Lam et al. 1996), is shown on top in uppercase, with the characteristic 200 nucleotides inverted repeats (underlined) and terminal TA dinucleotides (bold). The translation initiation (ATG) and termination (TGA) codons of the transposase gene are in bold and boxed. The reverse complement sequence of the region encompassing exon 4 of the ASgrn1-2 gene (boxed) is aligned underneath the transposon sequence, with mismatches represented in lowercase. Deletions (-) and insertions (+) relative to the intact transposon sequence that have occurred within exon 4 and flanking intronic sequences of the ASgrn1-2 gene, rendering this mobile element inactive, are indicated.
Click here for file
Additional File 8
Northern analysis of zebrafish grn sense and antisense transcripts. Northern blot analyses using strand-specific cRNA probes were performed to confirm the size of cloned cDNA sequences and resolve the issue of antisense transcription to grna. Panel A: grna is expressed as a predominant transcript of 3.7 kb in size, consistent with cloned sequences (3649 bp), and can also be found as a larger transcript exceeding 6 kb. Panel B: Four grnb transcripts of approximately 1.5 kb, 1.8 kb, 2.9 kb and 5.2 kb in size, respectively, are detected. Note that the 2.9 kb band, showing strongest intensity, is in agreement with cloned sequences for this mRNA (2820 bp). Panel C: A riboprobe targeting ASgrna detects a faint transcript of approximately 4 kb in size, suggesting that the cloned sequences for this antisense transcript (914 bp) may represent a splice variant. Panel D: The grn1 transcript is approximately 0.8 kb in size, as expected from cloned sequences (see Additional Files 2 and 3), but also expressed at lower levels as a larger transcript of approximately 1.7 kb. Panel E: Unlike grn1, grn2 is not expressed as a transcript other than 0.8 kb in size. Panel F: A band of low intensity can be detected at 0.8 kb for the hybrid grn RNA, suggesting that no cross-hybridization occurs with grn1 and grn2 mRNAs using this riboprobe. The hybrid grn riboprobe also detects a band of approximately 1.7 kb similar in intensity to that seen for grn1. Panel G: The overall abundance of the ASgrn1-2 transcript is too low for detection. A–C: 15 μg poly-A enriched RNA; D–G: 15 μg total RNA, each derived from a whole adult.
Click here for file
Additional File 9
Primers used in the linkage group assignment of the zebrafish progranulin genes. Before applying to the LN 54 mapping panel, conditions for each primer combination were optimized by PCR with the use of zebrafish genomic DNA derived from the AB wild type strain. The authenticity and specificity of each amplicon was verified by sequencing after cloning into the pCRII plasmid. For the assignment to zebrafish linkage groups, each PCR amplification experiment was performed at least twice.
Click here for file
Additional File 10
Sequences of primers and predicted sizes of PCR amplicons. All primer pairs are located on consecutive exons of their corresponding gene.
Click here for file
Additional File 11
Partial characterization of a putative antisense transcript to zebrafish grna. Panel A: Schematic representation of the location of ASgrna, a unidirectionally cloned transcript sharing complementarity with part of the last exon of the zebrafish grna gene. The relative position of primers used for cDNA synthesis (primers 2 and 3, respectively) and subsequent RT-PCR are indicated (see Materials and Methods). Panel B: RT-PCR analysis of cDNA synthesized using either primer 2 or primer 3, respectively, with (+) or without (-) reverse-transcriptase. Primers depicted in A are used in the following combination for the RT-PCR: pair 1/2 (342 bp product) and pair 1/3 (478 bp product). The presence of an amplicon for both primer pairs using each +RT cDNA template, but not using template derived from -RT reactions, indicates that ASgrna extends in the 3'end direction and overlaps with intronic sequences of grna. Positive controls for PCR conditions using genomic DNA as template, as well as a no template control are indicated, respectively. The authenticity of the amplicons was determined through sequencing. Each step of the experimental procedure was performed twice.
Click here for file
Acknowledgements
This work was supported by an operating grant (MOP-53105) from the Canadian Institute of Health Research. We are indebted to Drs. Marc Ekker and Marie-Andrée Akimenko (University of Ottawa) for their hospitality and for sharing their technical expertise with the whole mount in situ hybridisation technique during the initial phases of this project, as well as for performing the linkage analyses on the LN54 mapping panel. We also thank Ms. Jo-Ann Bader (Molecular Oncology Group, Royal Victoria Hospital, Montreal) for providing assistance with sectioning and staining tissue sections.
==== Refs
Bateman A Bennett HP Granulins: the structure and function of an emerging family of growth factors J Endocrinol 1998 158 145 151 9771457 10.1677/joe.0.1580145
He Z Bateman A Progranulin (granulin-epithelin precursor, PC-cell-derived growth factor, acrogranin) mediates tissue repair and tumorigenesis J Mol Med 2003 81 600 612 12928786 10.1007/s00109-003-0474-3
Ong CH Bateman A Progranulin (granulin-epithelin precursor, PC-cell derived growth factor, acrogranin) in proliferation and tumorigenesis Histol Histopathol 2003 18 1275 1288 12973694
Serrero G Autocrine growth factor revisited: PC-cell-derived growth factor (progranulin), a critical player in breast cancer tumorigenesis Biochem Biophys Res Commun 2003 308 409 413 12914763 10.1016/S0006-291X(03)01452-9
Baba T Hoff HB Nemoto H Lee H Orth J Arai Y Gerton GL Acrogranin, an acrosomal cysteine-rich glycoprotein, is the precursor of the growth-modulating peptides, granulins, and epithelins, and is expressed in somatic as well as male germ cells Mol Reprod Dev 1993 34 233 243 8471244 10.1002/mrd.1080340302
Zhou J Gao G Crabb JW Serrero G Purification of an autocrine growth factor homologous with mouse epithelin precursor from a highly tumorigenic cell line J Biol Chem 1993 268 10863 10869 8496151
He Z Bateman A Progranulin gene expression regulates epithelial cell growth and promotes tumor growth in vivo Cancer Res 1999 59 3222 3229 10397269
Xu SQ Tang D Chamberlain S Pronk G Masiarz FR Kaur S Prisco M Zanocco-Marani T Baserga R The granulin/epithelin precursor abrogates the requirement for the insulin-like growth factor 1 receptor for growth in vitro J Biol Chem 1998 273 20078 20083 9685348 10.1074/jbc.273.32.20078
Bateman A Belcourt D Bennett H Lazure C Solomon S Granulins, a novel class of peptide from leukocytes Biochem Biophys Res Commun 1990 173 1161 1168 2268320 10.1016/S0006-291X(05)80908-8
Shoyab M McDonald VL Byles C Todaro GJ Plowman GD Epithelins 1 and 2: isolation and characterization of two cysteine-rich growth-modulating proteins Proc Natl Acad Sci U S A 1990 87 7912 7916 2236009
Sparro G Galdenzi G Eleuteri AM Angeletti M Schroeder W Fioretti E Isolation and N-terminal sequence of multiple forms of granulins in human urine Protein Expr Purif 1997 10 169 174 9226711 10.1006/prep.1997.0726
Hrabal R Chen Z James S Bennett HP Ni F The hairpin stack fold, a novel protein architecture for a new family of protein growth factors Nat Struct Biol 1996 3 747 752 8784346 10.1038/nsb0996-747
Diaz-Cueto L Stein P Jacobs A Schultz RM Gerton GL Modulation of mouse preimplantation embryo development by acrogranin (epithelin/granulin precursor) Dev Biol 2000 217 406 418 10625564 10.1006/dbio.1999.9564
Qin J Diaz-Cueto L Schwarze JE Takahashi Y Imai M Isuzugawa K Yamamoto S Chang KT Gerton GL Imakawa K Effects of Progranulin on Blastocyst Hatching and Subsequent Adhesion and Outgrowth in the Mouse Biol Reprod 2005 15901638
Taylor JS Braasch I Frickey T Meyer A Van de Peer Y Genome duplication, a trait shared by 22000 species of ray-finned fish Genome Res 2003 13 382 390 12618368 10.1101/gr.640303
Amores A Force A Yan YL Joly L Amemiya C Fritz A Ho RK Langeland J Prince V Wang YL Westerfield M Ekker M Postlethwait JH Zebrafish hox clusters and vertebrate genome evolution Science 1998 282 1711 1714 9831563 10.1126/science.282.5394.1711
Postlethwait JH Yan YL Gates MA Horne S Amores A Brownlie A Donovan A Egan ES Force A Gong Z Goutel C Fritz A Kelsh R Knapik E Liao E Paw B Ransom D Singer A Thomson M Abduljabbar TS Yelick P Beier D Joly JS Larhammar D Rosa F Westerfield M Zon LI Johnson SL Talbot WS Vertebrate genome evolution and the zebrafish gene map Nat Genet 1998 18 345 349 9537416 10.1038/ng0498-345
Force A Lynch M Pickett FB Amores A Yan YL Postlethwait J Preservation of duplicate genes by complementary, degenerative mutations Genetics 1999 151 1531 1545 10101175
Belcourt DR Lazure C Bennett HP Isolation and primary structure of the three major forms of granulin-like peptides from hematopoietic tissues of a teleost fish (Cyprinus carpio) J Biol Chem 1993 268 9230 9237 8486624
David L Blum S Feldman MW Lavi U Hillel J Recent duplication of the common carp (Cyprinus carpio L.) genome as revealed by analyses of microsatellite loci Mol Biol Evol 2003 20 1425 1434 12832638 10.1093/molbev/msg173
Schreeb KH Groth G Sachsse W Freundt KJ The karyotype of the zebrafish (Brachydanio rerio) J Exp Anim Sci 1993 36 27 31 8312385
Bhandari V Bateman A Structure and chromosomal location of the human granulin gene Biochem Biophys Res Commun 1992 188 57 63 1417868 10.1016/0006-291X(92)92349-3
Baba T Nemoto H Watanabe K Arai Y Gerton GL Exon/intron organization of the gene encoding the mouse epithelin/granulin precursor (acrogranin) FEBS Lett 1993 322 89 94 8482392 10.1016/0014-5793(93)81544-A
Altschul SF Gish W Miller W Myers EW Lipman Basic local alignment search tool. J Mol Biol 1990 215 403 410 2231712 10.1006/jmbi.1990.9999
Lam WL Lee TS Gilbert W Active transposition in zebrafish Proc Natl Acad Sci USA 1996 93 10870 10875 8855274 10.1073/pnas.93.20.10870
Bhandari V Palfree RG Bateman A Isolation and sequence of the granulin precursor cDNA from human bone marrow reveals tandem cysteine-rich granulin domains Proc Natl Acad Sci U S A 1992 89 1715 1719 1542665
Plowman GD Green JM Neubauer MG Buckley SD McDonald VL Todaro GJ Shoyab M The epithelin precursor encodes two proteins with opposing activities on epithelial cell growth J Biol Chem 1992 267 13073 13078 1618805
Bhandari V Giaid A Bateman A The complementary deoxyribonucleic acid sequence, tissue distribution, and cellular localization of the rat granulin precursor Endocrinology 1993 133 2682 2689 8243292 10.1210/en.133.6.2682
Hukriede NA Joly L Tsang M Miles J Tellis P Epstein JA Barbazuk WB Li FN Paw B Postlethwait JH Hudson TJ Zon LI McPherson JD Chevrette M Dawid IB Johnson SL Ekker M Radiation hybrid mapping of the zebrafish genome Proc Natl Acad Sci U S A 1999 96 9745 9750 10449765 10.1073/pnas.96.17.9745
Concordet JP Lewis KE Moore JW Goodrich LV Johnson RL Scott MP Ingham PW Spatial regulation of a zebrafish patched homologue reflects the roles of sonic hedgehog and protein kinase A in neural tube and somite patterning Development 1996 122 2835 2846 8787757
Jaillon O Aury JM Brunet F Petit JL Stange-Thomann N Mauceli E Bouneau L Fischer C Ozouf-Costaz C Bernot A Nicaud S Jaffe D Fisher S Lutfalla G Dossat C Segurens B Dasilva C Salanoubat M Levy M Boudet N Castellano S Anthouard V Jubin C Castelli V Katinka M Vacherie B Biemont C Skalli Z Cattolico L Poulain J De Berardinis V Cruaud C Duprat S Brottier P Coutanceau JP Gouzy J Parra G Lardier G Chapple C McKernan KJ McEwan P Bosak S Kellis M Volff JN Guigo R Zody MC Mesirov J Lindblad-Toh K Birren B Nusbaum C Kahn D Robinson-Rechavi M Laudet V Schachter V Quetier F Saurin W Scarpelli C Wincker P Lander ES Weissenbach J Roest Crollius H Genome duplication in the teleost fish Tetraodon nigroviridis reveals the early vertebrate proto-karyotype Nature 2004 431 946 957 15496914 10.1038/nature03025
Naruse K Tanaka M Mita K Shima A Postlethwait J Mitani H A medaka gene map: the trace of ancestral vertebrate proto-chromosomes revealed by comparative gene mapping Genome Res 2004 14 820 828 15078856 10.1101/gr.2004004
Robinson-Rechavi M Marchand O Escriva H Bardet PL Zelus D Hughes S Laudet V Euteleost fish genomes are characterized by expansion of gene families Genome Res 2001 11 781 788 11337474 10.1101/gr.165601
Niimura Y Nei M Comparative evolutionary analysis of olfactory receptor gene clusters between humans and mice Gene 2005 346 13 21 15716120 10.1016/j.gene.2004.09.025
Amores A Suzuki T Yan YL Pomeroy J Singer A Amemiya C Postlethwait JH Developmental roles of pufferfish Hox clusters and genome evolution in ray-fin fish Genome Res 2004 14 1 10 14707165 10.1101/gr.1717804
Farber SA De Rose RA Olson ES Halpern ME The zebrafish annexin gene family Genome Res 2003 13 1082 1096 12799347 10.1101/gr.479603
Loh YH Christoffels A Brenner S Hunziker W Venkatesh B Extensive expansion of the claudin gene family in the teleost fish, Fugu rubripes Genome Res 2004 14 1248 1257 15197168 10.1101/gr.2400004
Feldman B Gates MA Egan ES Dougan ST Rennebeck G Sirotkin HI Schier AF Talbot WS Zebrafish organizer development and germ-layer formation require nodal-related signals Nature 1998 395 181 185 9744277 10.1038/26013
Barbazuk WB Korf I Kadavi C Heyen J Tate S Wun E Bedell JA McPherson JD Johnson SL The syntenic relationship of the zebrafish and human genomes Genome Res 2000 10 1351 1358 10984453 10.1101/gr.144700
Woods IG Kelly PD Chu F Ngo-Hazelett P Yan YL Huang H Postlethwait JH Talbot WS A comparative map of the zebrafish genome Genome Res 2000 10 1903 1914 11116086 10.1101/gr.10.12.1903
Postlethwait JH Woods IG Ngo-Hazelett P Yan YL Kelly PD Chu F Huang H Hill-Force A Talbot WS Zebrafish comparative genomics and the origins of vertebrate chromosomes Genome Res 2000 10 1890 1902 11116085 10.1101/gr.164800
Daniel R He Z Carmichael KP Halper J Bateman A Cellular localization of gene expression for progranulin J Histochem Cytochem 2000 48 999 1009 10858277
Belcourt DR Okawara Y Fryer JN Bennett HP Immunocytochemical localization of granulin-1 to mononuclear phagocytic cells of the teleost fish Cyprinus carpio and Carassius auratus J Leukoc Biol 1995 57 94 100 7829976
Barreda DR Hanington PC Walsh CK Wong P Belosevic M Differentially expressed genes that encode potential markers of goldfish macrophage development in vitro Dev Comp Immunol 2004 28 727 746 15043942 10.1016/j.dci.2003.11.005
Schultz RM Regulation of zygotic gene activation in the mouse. BioEssays 1993 15 531 538 8135766 10.1002/bies.950150806
Daniel R Daniels E He Z Bateman A Progranulin (acrogranin/PC cell-derived growth factor/granulin-epithelin precursor) is expressed in the placenta, epidermis, microvasculature, and brain during murine development Dev Dyn 2003 227 593 599 12889069 10.1002/dvdy.10341
Liau LM Lallone RL Seitz RS Buznikov A Gregg JP Kornblum HI Nelson SF Bronstein JM Identification of a human glioma-associated growth factor gene, granulin, using differential immuno-absorption Cancer Res 2000 60 1353 1360 10728698
Markert JM Fuller CM Gillespie GY Bubien JK McLean LA Hong RL Lee K Gullans SR Mapstone TB Benos DJ Differential gene expression profiling in human brain tumors Physiol Genomics 2001 5 21 33 11161003
Suzuki M Yoshida S Nishihara M Takahashi M Identification of a sex steroid-inducible gene in the neonatal rat hypothalamus Neurosci Lett 1998 242 127 130 9530922 10.1016/S0304-3940(98)00008-1
Suzuki M Nishiahara M Granulin precursor gene: a sex steroid-inducible gene involved in sexual differentiation of the rat brain Mol Genet Metab 2002 75 31 37 11825061 10.1006/mgme.2001.3274
Maack G Segner H Life-stage-dependent sensitivity of zebrafish (Danio rerio) to estrogen exposure Comp Biochem Physiol C Toxicol Pharmacol 2004 139 47 55 15556065 10.1016/j.cca.2004.09.004
Orkin SH Zon LI Genetics of erythropoiesis: induced mutations in mice and zebrafish Annu Rev Genet 1997 31 33 60 9442889 10.1146/annurev.genet.31.1.33
Galloway JL Zon LI Ontogeny of hematopoiesis: examining the emergence of hematopoietic cells in the vertebrate embryo Curr Top Dev Biol 2003 53 139 158 12510667
Medvinsky A Dzierzak E Definitive hematopoiesis is autonomously initiated by the AGM region Cell 1996 86 897 906 8808625 10.1016/S0092-8674(00)80165-8
Bennett CM Kanki JP Rhodes J Liu TX Paw BH Kieran MW Langenau DM Delahaye-Brown A Zon LI Fleming MD Myelopoiesis in the zebrafish, Danio rerio Blood 2001 98 643 651 11468162 10.1182/blood.V98.3.643
He Z Ong CH Halper J Bateman A Progranulin is a mediator of the wound response Nat Med 2003 9 225 229 12524533 10.1038/nm816
Zhu J Nathan C Jin W Sim D Ashcroft GS Wahl SM Lacomis L Erdjument-Bromage H Tempst P Wright CD Ding A Conversion of proepithelin to epithelins: roles of SLPI and elastase in host defense and wound repair Cell 2002 111 867 878 12526812 10.1016/S0092-8674(02)01141-8
Agabian N Trans splicing of nuclear pre-mRNAs Cell 1990 61 1157 1160 2142018 10.1016/0092-8674(90)90674-4
Li BL Li XL Duan ZJ Lee O Lin S Ma ZM Chang CC Yang XY Park JP Mohandas TK Human acyl-CoA:cholesterol acyltransferase-1 (ACAT-1) gene organization and evidence that the 4.3-kilobase ACAT-1 mRNA is produced from two different chromosomes. J Biol Chem 1999 274 11060 11071 10196189 10.1074/jbc.274.16.11060
Finta C Zaphiropoulos PG The human cytochrome P450 3A locus. Gene evolution by capture of downstream exons Gene 2000 260 13 23 11137287 10.1016/S0378-1119(00)00470-4
Akopian AN Okuse K Souslova V England S Ogata N Wood JN Trans-splicing of a voltage-gated sodium channel is regulated by nerve growth factor FEBS Lett 1999 445 177 182 10069396 10.1016/S0014-5793(99)00126-X
Bruzik JP Maniatis T Enhancer-dependent interaction between 5' and 3' splice sites in trans Proc Natl Acad Sci U S A 1995 92 7056 7059 7624368
The Pfam database of protein families and HMMs
European Bioinformatics Institute -Interpro
Weijer CJ Dictyostelium morphogenesis Curr Opin Genet Dev 2004 14 392 398 15261655 10.1016/j.gde.2004.06.006
Funk V Kositsup B Zhao C Beers EP The Arabidopsis xylem peptidase XCP1 is a tracheary element vacuolar protein that may be a papain ortholog Plant Physiol 2002 128 84 94 11788755 10.1104/pp.128.1.84
Wang Y Gu X Evolutionary patterns of gene families generated in the early stage of vertebrates J Mol Evol 2000 51 88 96 10903375
Bennett HPJ Browne CA Solomon S Purification of the two major forms of rat pituitary corticotropin using only reversed-phase liquid chromatography Biochemistry 1981 20 4530 4538 6271167 10.1021/bi00519a004
Mullins MC Hammerschmidt M Hafter P Nusslein-Volhard C Large-scale mutagenesis in the zebrafish: in search of genes controlling development in a vertebrate Curr Biol 1994 4 189 202 7922324 10.1016/S0960-9822(00)00048-8
Kimmel CB Ballard WW Kimmel SR Ullmann B Schilling TF Stages of embryonic development of the zebrafish. Dev Dyn 1995 203 253 310 8589427
Amemiya CT Zon LI Generation of a zebrafish P1 artificial chromosome library Genomics 1999 58 211 213 10366454 10.1006/geno.1999.5827
ZFIN
LocusLink
Sambrook J Fritsch EF Maniatis T Molecular cloning - A laboratory manual. Second edition 1989 Cold Spring Harbor Laboratory Press
Thisse C Thisse B Schilling TF Postlethwait JH Structure of the zebrafish snail1 gene and its expression in wild-type, spadetail and no tail mutant embryos Development 1993 119 1203 1215 8306883
DeBlock M Debrouwer D RNA-RNA in situ hybridisation using DIG-labeled probes: the effect of high molecular weight polyvinyl alcohol on the alkaline phosphatase indoxyl-nitoblue tetrazolium reaction.
Anal Biochem 1993 215 86 89 7507651 10.1006/abio.1993.1558
Hedges SB The origin and evolution of model organisms Nat Rev Genet 2002 3 838 849 12415314 10.1038/nrg929
Pera EM Hou S Strate I Wessely O De Robertis EM Exploration of the extracellular space by a large-scale secretion screen in the early Xenopus embryo Int J Dev Biol 2005 49 781 796 16172975 10.1387/ijdb.052003ep
Bisbee CA Baker MA Wilson AC Haji-Azimi I Fischberg M Albumin phylogeny for clawed frogs (Xenopus) Science 1977 195 785 787 65013
|
16277664
|
PMC1310530
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Nov 8; 6:156
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-156
|
oa_comm
|
==== Front
BMC BiochemBMC Biochemistry1471-2091BioMed Central London 1471-2091-6-241628588110.1186/1471-2091-6-24Research ArticleStudies of the intermediary metabolism in cultured cells of the insect Spodoptera frugiperda using 13C- or 15N-labelled tracers Adam Petra [email protected]ütlich Markus [email protected] Hartmut [email protected] Adelbert [email protected] Wolfgang [email protected] Lehrstuhl für Organische Chemie und Biochemie, Technische Universität München, Lichtenbergstr. 4, D-85747 Garching, Germany2 Forschungsinstitut für molekulare Pharmakologie, Robert-Rössle-Str. 10, D-13125 Berlin, Germany2005 14 11 2005 6 24 24 25 7 2005 14 11 2005 Copyright © 2005 Adam et al; licensee BioMed Central Ltd.2005Adam et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Insect cells can serve as host systems for the recombinant expression of eukaryotic proteins. Using this platform, the controlled expression of 15N/13C labelled proteins requires the analysis of incorporation paths and rates of isotope-labelled precursors present in the medium into amino acids. For this purpose, Spodoptera frugiperda cells were grown in a complex medium containing [U-13C6]glucose. In a second experiment, cultures of S. frugiperda were grown in the presence of 15N-phenylalanine.
Results
Quantitative NMR analysis showed incorporation of the proffered [U-13C6]glucose into the ribose moiety of ribonucleosides (40 – 45%) and into the amino acids, alanine (41%), glutamic acid/glutamine (C-4 and C-5, 30%) and aspartate/asparagine (15%). Other amino acids and the purine ring of nucleosides were not formed from exogenous glucose in significant amounts (> 5%). Prior to the incorporation into protein the proffered 15N-phenylalanine lost about 70% of its label by transamination and the labelled compound was not converted into tyrosine to a significant extent.
Conclusion
Growth of S. frugiperda cells in the presence of [U-13C6]glucose is conducive to the fractional labelling of ribonucleosides, alanine, glutamic acid/glutamine and aspartic acid/asparagine. The isotopolog compositions of the ribonucleosides and of alanine indicate considerable recycling of carbohydrate intermediates in the reductive branch of the pentose phosphate pathway. The incorporation of 15N-labelled amino acids may be hampered by loss of the 15N-label by transamination.
==== Body
Background
Cultured insect cells can serve as efficient host systems for the recombinant expression of certain eukaryotic proteins that fail to be expressed in bacterial host systems. Advantageously, insect cells provide eukaryotic type chaperones, permit the expression of long open reading frames, and are not subject to the limitations caused by bacterial codon preferences [1]. Stable isotope labelling is crucial for NMR structure analysis of proteins using multinuclear spectroscopic techniques. Moreover, differential isotope labelling can significantly enhance the scope of various other spectroscopic techniques such as EPR, infrared and Raman spectroscopy for the analysis of proteins. Differential 15N labelling of proteins in Baculovirus-infected insect cells has been achieved via 15N-labelled amino acids present in the growth medium [2,3].
This study was designed to explore in some detail the metabolic network of Spodoptera frugiperda cells under the specific growth conditions for 15N/13C labelling. Our data demonstrate that cultured S. frugiperda cells grown on commercial culture media rely on exogenous supply rather than on intracellular synthesis of most proteinogenic amino acids, although central metabolic intermediates (i.e., pentose phosphate, oxaloacetate, pyruvate and acetyl-CoA) acquire label from the proffered [U-13C6]glucose and transamination activity was observed in the experiment with exogenous 15N-phenylalanine.
Results
In order to assess the utilization of glucose for the biosynthesis of amino acids and ribonucleotides, we cultured S. frugiperda cells in a complex culture medium supplemented with a mixture of [U-13C6]glucose and unlabelled glucose at a ratio of 1:99. The cells were harvested, and lipids were removed by solvent extraction. Cellular RNA was hydrolysed by alkali treatment, and the resulting ribonucleotides were isolated and dephosphorylated. The resulting nucleosides were further purified by HPLC. The residue obtained after alkali treatment consisted predominantly of protein and was hydrolyzed by hydrochloric acid. Amino acids were isolated chromatographically from the hydrolysate.
1H and 13C NMR spectra were determined for all isolated metabolites. 13C Abundances were determined by quantitative NMR spectroscopy (see Methods). 13C Signals of isotopologs carrying two or more adjacent 13C atoms were characterized by satellite signals arising by 13C13C couplings. As an example, the 13C signals of the ribose carbon atoms of cytidine showed satellites indicating coupling to one or two adjacent 13C atoms (Fig. 1). More specifically, the signals of carbon atoms 1', 2', 3' and 5' appeared as pseudotriplets where the central lines represent molecules with a single 13C atom, whereas the satellites indicate molecules with blocks of two 13C atoms. The signature of the ribose carbon atom 4' displayed a central line, a doublet and a double-doublet indicating the presence of three different 13C-labelled isotopologs, namely [4'-13C1]-, [4', 5'-13C2]-, and [3', 4', 5'-13C3]cytidine, respectively.
Figure 1 13C NMR signals of cytidine from S. frugiperda cells grown with a 1:99 mixture of [U-13C6]glucose and unlabelled glucose added to SF-900 II medium. 13C-Coupling patterns are indicated.
From the relative fractions of each respective satellite pair in the total 13C NMR signals (% 13C13C in Table 1) the abundance (in mol%) of the different isotopologs in the ribose moiety of cytidine (1) was calculated (Fig. 2). Notably, the abundance of multiply 13C-substituted molecules can be determined independently from the signal pattern of two or more different carbon atoms. The accuracy of the measurements can therefore be assessed statistically. The standard deviations document the accuracy of the method (Fig. 2).
Table 1 Isotopolog compositions of nucleosides from S. frugiperda cells grown with a mixture of [U-13C6]glucose and natural abundance glucose.
Compound Position 13C NMR Chemical Shift (δ, ppm) Coupling Constants JCC (Hz) %13Ca %13C13Cb
Cytidine (1) 2 159.7 1.1
4 168.4 55.5 (5) 1.3 7.3 (5)
5 98.9 55.3 (4) 1.2 8.5 (4)
6 144.4 1.2
1' 92.2 42.7 (2') 1.6 22.1 (2')
2' 76.6 42.5 (1') 1.6 22.8 (1')
3' 71.9 38.7 (4') 1.5 25.3 (4')
4' 86.4 38.7 (3'), 42.0 (5') 1.6 23.7 (3', 5'), 2.8 (5')
5' 63.5 42.0 (4') 1.5 27.4 (4')
Adenosine (2) 2 154.7 1.1
4 150.6 n.d.
5 121.3 n.d.
6 157.8 2.7
8 142.7 1.1
1' 90.4 42.7 (2') 1.7 23.5 (2')
2' 75.8 42.5 (1') 1.8 22.8 (1')
3' 72.7 38.5 (4') 1.7 26.0 (4')
4' 87.9 41.8 (5'), 38.3 (3') 1.6 24.6 (3', 5'), 2.9 (5')
5' 63.6 41.3 (4') 1.7 25.9 (4')
Guanosine (3) 6 156.4 1.3
2 153.8 1.1
4 150.1 1.1
5 115.7 1.0
8 135.6 1.2
1' 86.5 42.9 (2') 1.7 22.8 (2')
2' 73.7 42.9 (1') 1.7 22.3 (1')
3' 70.2 38.4 (4') 1.7 24.1 (4')
4' 85.2 41.8 (5'), 38.0 (3') 1.6 26.0 (3', 5'), 2.8 (5')
5' 61.3 42.0 (4') 1.7 26.5 (4')
aabsolute 13C abundance.
brelative fraction of satellite pairs in the global 13C NMR intensity of the index carbon atom. The coupling partners of the index carbon atom are indicated in parentheses.
Figure 2 Isotopolog composition of 13C-labelled nucleosides from S. frugiperda cells grown with a 1:99 mixture of [U-13C6]glucose and unlabelled glucose added to SF-900 II medium. The filled dot represents a 13C1 isotopolog detected at an abundance well above the natural abundance contribution. Bold lines indicate isotopologs with adjacent 13C-labelled carbon atoms that were transferred from the same molecule of [U-13C6]glucose. The numbers give the molar enrichments of the indicated isotopologs.
The isotopolog compositions of the ribose side chains of adenosine (2) and guanosine (3) were closely similar to that of cytidine (1) (cf. Table 1 and Fig. 2). The sugar units of these nucleosides are all derived from the cellular pentose phosphate pool and the close similarity of their labelling patterns as shown by the low standard deviations (Fig. 3) again document the precision of the experimental data. The average labelling pattern revealed the presence of [1,2-13C2]- and [3,4,5-13C3]ribose moieties at relatively high abundances of 0.41 and 0.42 mol%, respectively. [4,5-13C2]Ribose moieties were present at substantially lower abundance (0.06 mol%). [U-13C5]Ribose moieties and other multiply 13C-labelled ribose isotopologs were virtually absent (less than 0.05 mol%). As described in more detail below, it follows that the ribose pool was primarily alimented by the reductive branch of the pentose phosphate pool, whereas the direct pathway via oxidative decarboxylation of the applied [U-13C6]glucose played an almost negligible role. From the molar fraction of 13C-labelled glucose in the medium (approximately 1%) and the molar fractions of multiply 13C-labelled isotopologs (see above), a specific incorporation of the proffered [U-13C6]glucose into the ribosyl moiety of nucleosides can be estimated as 40 – 45% (Table 2).
Figure 3 Average isotopolog composition of 13C-labelled ribose units in ribonucleosides from S. frugiperda cells grown with a 1:99 mixture of [U-13C6]glucose and unlabelled glucose added to SF-900 II medium. Bold lines indicate isotopologs with adjacent 13C-labelled carbon atoms that were transferred from the same molecule of [U-13C6]glucose. The numbers give the molar contributions of the indicated isotopologs.
Table 2 Specific incorporation of proffered glucose into metabolites of S. frugiperda cells grown in SF-900 II medium.
Compound Specific incorporation in%
Nucleosides
Ribose moiety 40–45
Pyrimidine base 10–15
Purine base < 5
Amino acids
Alanine 45–50
Glutamate/Glutamine (C-4/C-5) 30
Aspartate/Asparagine 10–15
Others < 5
The 13C NMR signals of C-4 and C-5 in the pyrimidine ring of cytidine (1) also showed 13C13C satellites. The quantitative analysis revealed the presence of a [4,5-13C2]-isotopolog at a molar abundance of 0.1 mol% (Fig. 2) which was significantly above the natural abundance level (0.01 mol%). Moreover, the 1H NMR signal of H-6 (Fig. 4) showed 13C coupling satellites indicating a 13C abundance of 1.2 mol% for the [6-13C1]-isotopolog, i.e. 0.1 mol% excess over the natural 13C abundance of 1.1 mol% (see filled circle in Fig. 2). The presence of these isotopologs demonstrates that a fraction of the pyrimidine was biosynthesised de novo from the supplemented 13C-labelled glucose via aspartate (see below). The specific incorporation of the proffered 13C-glucose into the pyrimidine moiety can be estimated as 10–15% (Table 2).
Figure 4 1H NMR signal of H-6 of cytidine from S. frugiperda cells grown with a 1:99 mixture of [U-13C6]glucose and unlabelled glucose added to SF-900 II medium. The amplitude of the satellites is 50-fold enlarged by comparison with the central signal. The coupling pattern is indicated.
The purine ring systems of adenosine and guanosine showed no significantly increased levels of molecular species carrying two or more 13C atoms (Table 1). 1H NMR analysis showed no increased 13C abundance for the position 8 methine groups of both nucleosides. Consequently, the data demonstrate that purines were obtained from the culture medium and were not biosynthesised to a significant extent (> 5%) under the culture conditions (Table 2).
Most amino acids obtained from the protein hydrolysate (i.e. leucine, phenylalanine, tyrosine, lysine, histidine, arginine, serine, threonine, valine, proline, methionine and isoleucine) showed only the low intensity 13C13C coupling satellites typical for natural abundance compounds. In contrast to the spectra of the amino acids mentioned above, the 13C NMR signals of alanine (4) were characterized by intense satellite signals due to couplings between adjacent 13C atoms. More specifically, the signal for C-2 showed a doublet indicating coupling to 13C-3, as well as a double-doublet indicating simultaneous coupling to 13C-3 and 13C-1. The relative fractions of these satellite pairs in the overall signal intensity of the C-2 signal accounted for 3.8 and 22.2%, respectively, which correspond to an abundance of 0.07 and 0.41 mol% for the [2,3-13C2]- and [U-13C3]-isotopolog (see also Table 3 and Fig. 5). On this basis, the specific incorporation of the 13C-glucose into alanine can be estimated as 45 – 50% (Table 2).
Table 3 Isotopolog compositions of amino acids from S. frugiperda cells grown in a mixture of [U-13C6]glucose and natural abundance glucose.
Compound Position 13 Chemical Shift (δ, ppm) Coupling Constants JCC (Hz) %13Ca 13C13Cb
Alanine (4) 1 175.7 n.d. n.d. n.d.
2 51.4 59.6 (1), 34.9 (3) 1.8 3.8 (3), 22.2 (1, 3)
3 18.4 35.1 (2) 1.7 26.5 (2)
Aspartate (5) 1 182.9 n.d. n.d. n.d.
2 61.3 60.4 (1) 1.8 7.4 (1)
3 45.9 55.3 (4) 1.9 7.2 (4)
4 185.1 55.7 (3) 1.8 7.6 (3)
Glutamate (6) 1 176.1 1.3
2 56.2 1.0
3 27.9 1.1
4 32.5 55.1 (5) 1.8 17.1 (5)
5 179.6 55.1 (4) 1.7 17.2 (4)
aabsolute 13C abundance.
brelative fraction of satellite pairs in the global 13C NMR intensity of the index carbon atom. The coupling partners of the index carbon atom are indicated in parentheses.
n.d., not determined due to poor signal quality.
Figure 5 Isotopolog composition of 13C-labelled amino acids from S. frugiperda cells grown with a 1:99 mixture of [U-13C6]glucose and unlabelled glucose added to SF-900 II medium. Bold lines indicate isotopologs with adjacent 13C-labelled carbon atoms that were transferred from the same molecule of [U-13C6]glucose. The numbers give the molar contributions of the indicated isotopologs.
Significantly enhanced levels of doubly 13C-labelled isotopologs were also found in glutamate (5) and aspartate (6) (Fig. 5). [4,5-13C2]Glutamate had a concentration of 0.30 mol% indicating a specific incorporation of approximately 30%. [1,2-13C2]- and [3,4-13C2]aspartate were found at abundances of 0.13 and 0.14 mol%, respectively. The abundances of these double-labelled species were closely similar to the abundance of [4,5-13C2]cytidine (see above).
A second experiment was designed to investigate the incorporation of a 15N-labelled amino acid from the culture medium by S. frugiperda cells. The standard culture medium for SF-9 cells contains numerous amino acids in relatively large amounts [4]. Moreover, the yeast extract added to the medium contains peptides which may become available to the growing cells after proteolysis. The following experiment was therefore performed with a customized amino acid and sugar free medium which was supplemented with amino acids as indicated in Experimental Procedures. In a preliminary study, we monitored the proliferation rate of cells growing with media containing different amounts of phenylalanine; the addition of 100 mg per litre was found to be sufficient for maximum growth activity.
An incorporation experiment was then performed with medium containing 100 mg of [15N]phenylalanine per litre. All other medium components were present in standard amounts (cf. Experimental Procedures). After two passages, the cells were harvested and subjected to acid hydrolysis. Phenylalanine and tyrosine were isolated chromatographically and were analyzed by 15N and 13C NMR spectrometry, as well as by mass spectrometry.
The 15N NMR spectrum showed a signal at 35.8 ppm which was assigned to [15N]phenylalanine by internal standardization. The 15N abundance was determined by 13C NMR spectroscopy. The 13C NMR signals for C-2 and C-3 were accompanied by up-field shifted satellite signals due to 15N isotope shifts. The sizes of the isotope shifts for C-2 and C-3 (50.9 ppb and 37.6 ppb, respectively) as well as the 13C15N coupling constant of 3.7 Hz (1JCN) were in accordance with published values [5,6]. The relative fractions of the signal intensities of the up-field shifted satellites in the overall 13C NMR signal intensities of C-2 and C-3 accounted for 25 ± 2%. This value was confirmed by GC/MS of the N-trifluoroacetyl-n-butylester of phenylalanine [7]. A relative abundance of 30 ± 8% was determined for the 15N-labelled fragment (m/z = 216). Tyrosine isolated from cell protein was not 15N-labelled.
Discussion
Intermediary metabolism constitutes a complex network with hundreds to thousands of nodes depending on the genetic complexity of the experimental system. The nodes in the central area of the intermediary metabolic network involving the reciprocal transformation of simple carbohydrates, carboxylic and dicarboxylic acids are typically connected by short links. Notably, the sizes and complexity of networks in different organisms can now be compared in some detail on the basis of whole genome sequence data of more than 100 species. The comprehensive quantitative description of such a network requires, in principle, the quantitative assessment of forward and backward reactions mediated by hundreds to thousands of enzymes under intracellular conditions; studies in cell extracts after disruption of cellular integrity can hardly serve as substitute. However, an in vivo approach can be performed with cells or organisms on the basis of perturbation/relaxation analysis of the isotopolog distribution in target metabolites [8-16].
This approach can be briefly summarized as follows. Naturally occurring organic matter is a highly complex mixture of isotopologs comprising all stable isotopes of hydrogen, carbon and nitrogen. The natural abundance of 13C is about 1.1%. In the quasi-equilibrium mixture of natural organic material of low molecular weight, the isotopologs carrying more than one 13C atom are of low abundance; as an example, the approximate natural abundances of some 13C-isotopologs of ribose are shown in Table 4.
Table 4 Approximate natural abundance of selected ribose isotopologs carrying 13C.
Isotopolog Abundance [mol%]
[1-13C1] 1.1
[2-13C1] 1.1
[3-13C1] 1.1
[1,2-13C2] 0.012
[4,5-13C2] 0.012
[1,2,3-13C3] 0.00013
[3,4,5-13C3] 0.00013
[U-13C5] 0.000000018
The introduction of a 13C-labelled compound into any biological system constitutes a local perturbation of the quasi-equilibrium state of isotopolog distribution in the respective metabolic network. Such a perturbation can spread in the network by way of enzymatic interconversion. The comprehensive analysis of the network-wide metabolic relaxation processes resulting from such an initial perturbation requires, in principle, the analysis of isotopolog compositions at virtually every node of the network at various times after the onset of the perturbation. In metabolic networks of typical prokaryotes and eukaryotes comprising hundreds to thousands of nodes, this is a relatively tall order. However, the task can be simplified quite considerably in light of two considerations. (i) Since metabolic networks are highly crosslinked via enzyme-catalyzed reactions, it is sufficient to measure the isotopolog distribution at a relatively small number of nodes. The isotopolog composition at numerous adjacent nodes can then be estimated on the basis of certain well-known enzyme catalyzed reactions. (ii) Following a perturbation of the isotopolog equilibrium, the relaxation process is typically not conducive to a true equilibrium state, since anabolic processes lead to a quasisteady state; only catabolic processes are conducive to actual reequilibration.
In practical terms, primary metabolites such as amino acids and nucleic acid constituents are biosynthetically derived from pools of small molecules by anabolic reactions. These primary metabolites are assembled from simple carbohydrates, carboxylic and dicarboxylic acids which are present in low amounts and are at the same time subject to rapid turnover. On the other hand, the products of primary metabolism (amino acids, nucleic acid components) become embedded into polymers (proteins, nucleic acids) where their turnover rate is relatively low. Thus, they reflect the isotopolog distribution of the central intermediary pools (carbohydrates, carboxylic acids) at the time of their formation. Based on these considerations, the transient label distribution in the central metabolite pools can be easily gleaned from analysis of the monomeric building blocks of proteins and nucleic acids (for review, see [13]).
The formation of pentose derviatives from [U-13C6]glucose by oxidative decarboxylation via the oxidative branch of the pentose phosphate pathway, if active to a significant extent, should have afforded [U-13C5]ribose phosphate. The virtual absence of that species in the ribonucleosides analyzed indicates that other metabolic processes (i.e. glycolytic cycling prior to any oxidative decarboxylation of glucose) must have played a crucial role. The observed isotopolog pattern can be best explained by cooperation of the reductive branch of the pentose phosphate cycle with glycolysis and regeneration of glucose from triose pool intermediates. Passage through the reductive branch of the pentose phosphate pathway results in breaking of the bond between C-3 and C-4 of glucose. The same is true for glycolysis. Both pathways afford [U-13C3]glyceraldehyde 3-phosphate from carbon atoms 4 through 6 of the proffered [U-13C6]glucose. The utilization of that intermediate by transketolase affords a [3,4,5-13C3]pentulose. On the other hand, grafting of a [13C2]-fragment derived from [U-13C6]glucose to an unlabelled glyceraldehyde phosphate moiety affords [1,2-13C2]pentulose phosphate isotopologs (species 6 in Fig. 6A).
Figure 6 Metabolic processes involved in pentose/pentulose formation from [U-13C6]glucose in a large excess of unlabelled glucose (A), via the pentose phosphate cycle and (B), via glycolytic cycling followed by the pentose phosphate cycle. (C), Metabolic processes involved in the formation of the detected aspartate isotopologs. Bold lines connect 13C-atoms in a given molecular species.
Glucogenesis using [13C3]triose phosphates and unlabelled triose phosphates in excess (from the unlabelled glucose fraction) is conducive to the formation of [1,2,3-13C3]- and [4,5,6-13C3]hexose phosphates. Oxidative decarboxylation of these species also affords [1,2-13C2]- and [3,4,5-13C3]pentose phosphates, respectively (Fig. 6B).
The formation of minor amounts of [4,5-13C2]pentoses can be explained by the transfer of a [13C2]fragment from a [1,2-13C2]pentulose phosphate (species 6) to unlabelled erythrose phosphate by the action of transketolase. The resulting [1,2-13C2]fructose phosphate (species 7) can then be converted to [2,3-13C2]triose phosphate which affords [4,5-13C2]pentose phosphate by recycling in the pentose phosphate cycle (Fig. 6A). The labelling patterns of the carbohydrate moieties indicate that 13C-label is transferred efficiently to the triose phosphate pool by the joint action of glycolysis and the pentose phosphate cycle.
On the basis of the pyruvate/alanine transamination reaction, the labelling pattern of alanine can be taken as a reference for the labelling pattern of pyruvate. The isotopolog composition of alanine (4) was in perfect agreement with the labelling pattern of the three carbon moiety comprising C-3, C-4 and C-5 in the ribose unit of ribonucleosides (cf. Figs. 3 and 5). As discussed above, the latter moiety is biosynthetically equivalent to the triose phosphate pool and, on this basis, the isotopolog compositions of triose phosphate and pyruvate are apparently identical. It can be concluded that pyruvate is predominantly or exclusively biosynthesized from the triose phosphate pool via phosphoenolpyruvate and that alternative routes leading to pyruvate (e.g., via decarboxylation of oxaloacetate) do not contribute significantly to the de novo pyruvate synthesis under the experimental settings.
The presence of [4,5-13C2]glutamate indicates its formation from [4,5-13C2]2-ketoglutarate (species 8) which is assembled from oxaloacetate and [1,2-13C2]acetyl-CoA via the citrate cycle (Fig. 6C). Following the reactions of the citrate cycle, [4,5-13C2]2-ketoglutarate is converted into a mixture of [1,2-13C2]- and [3,4-13C2]oxaloacetate (species 9) via the symmetric intermediates succinate and fumarate. Transamination of these oxaloacetate species yields the detected aspartate isotopologs. Reaction of [3,4-13C2]oxaloacetate with acetyl-CoA should have afforded [1,2-13C2]2-ketoglutarate via the citrate cycle. Notably, this species is not reflected in the labelling pattern of analyzed glutamate. One possible explanation for this finding is the existence of two compartmented oxaloacetate and/or 2-ketoglutarate pools (i) providing substrates for the citrate cycle and (ii) providing substrates for transamination. Further experiments are necessary to clarify this hypothesis.
By comparison with alanine, the label concentration in the acidic amino acids derived from citrate cycle intermediates is significantly lower. This could be due to the incorporation of exogenously supplied glutamate and aspartate into protein. This is not surprising since the culture medium used contained relatively large amounts of asparagine, aspartate, glutamate and glutamine.
The virtual absence of label in many amino acids, with the exception of alanine, glutamate and aspartate, indicates that the de novo synthesis under the experimental conditions was limited to those amino acids which can be obtained by single step transamination reactions from keto acids which are abundantly present in intermediary metabolism (pyruvate, oxaloacetate and ketoglutarate) (Fig. 7). Under the experimental conditions, remodelling of carbon skeletons for the purpose of amino biosynthesis proceeded at a very low level, if at all.
Figure 7 Metabolic network of S. frugiperda cells grown in SF-900 II medium.
Recently, the incorporation of 13C-glucose into recombinant Abl kinase expressed in Baculovirus-infected insect cells was studied by Strauss et al. [17]. The incorporation rates of glucose into amino acids (i.e., only into alanine, glutamate/glutamine and aspartate/asparagine) were similar to the data presented in our study.
In line with the finding that transamination is the only significant activity with regard to the biosynthesis of aliphatic amino acids, the experiment with [15N1]phenylalanine shows that transamination occurs on a relatively high level. Although the proffered [15N1]isotopolog was apparently the only source of phenylalanine, the sample isolated from cell protein retained only about 30% of 15N label. This is in notable contrast to an earlier study, where incorporation rates of > 90% have been reported for exogenous 15N-phenylalanine into human Abl kinase expressed in Baculovirus-infected insect cells [3,17]. The reasons for this apparent discrapency might be (i) the difference of the protein/protein fraction analyzed by NMR spectroscopy in the two studies, (ii) different concentrations of the supplied 15N-phenylalanine, and/or (iii) the difference of the experimental setup with the culture systems.
More specifically, growing cells of the insect were studied in our study. All cellular proteins were then hydrolyzed and amino acids were subjected to NMR analysis. In the studies of Strauss et al. [3,17] insect cells were transfected with Baculovirus and recombinant protein (i.e., c-Abl kinase) was analyzed. The concentrations of exogenous 15N-phenylalanine were 10-fold different in the two studies and it cannot be excluded that minor amounts of unlabelled phenylalanine were still present in the culture medium (i.e., "SF-900 II medium without amino acids and without sugar", Gibco) used in our study.
Isolated tyrosine did not contain detectable amounts of 15N. This suggests that hydroxylation of phenylalanine does not take place to a significant extent, although phenylalanine hydroxylase is known to be present in insects [18-20].
Conclusion
Our findings are relevant for the design of protein labelling with recombinant S. frugiperda cells. The following aspects need to be considered.
(i) The standardized media contain amino acids in relatively large excess over the real metabolic needs of the cells. Obviously, the concentration of isotope labelled amino acids in the medium should be optimized (i.e., reduced to an appropriate level).
(ii) Differential 13C/15N labelling of most amino acids (i.e. leucine, phenylalanine, tyrosine, lysine, histidine, arginine, serine, threonine, valine, proline, methionine and isoleucine) can be achieved via the respective labelled amino acids present in the growth medium (see also, [2,3,17]). However, labelling with 15N could be hampered by the relatively high transamination activity in the cells (cf. also [21]). Hence, the carbon skeleton of specific [13C,15N]-labelled amino acids would be incorporated intact, but the nitrogen label would be subject to extensive dilution by nitrogen from other, non-labelled amino acids. It may be possible to counteract this isotope loss by the addition of a mixture of [15N]-amino acids (e.g., glutamine and asparagine).
(iii) Fractional 13C labelled proteins comprising [U-13C3]alanine, [4,5-13C2]glutamate, [1,2-13C2]- and [3,4-13C2]aspartate can be obtained by addition of [U-13C6]glucose to the growth medium.
Methods
Materials
[U-13C6]Glucose and [15N]phenylalanine were obtained from Isotec, Miamisburg, Ohio, USA. SF-900 II medium was obtained from Gibco, Karlsruhe, Germany.
Cell culture
SF-9 cells were grown in SF-900 II medium [4]. The cells were kept at a density of 2 × 105 to 2 × 106 cells/ml at 28°C in spinner flasks. Expansion cultures were performed in spinner flasks at 28°C and 80 rpm or in a shaking flask at 28°C and 60 rpm. In order to study the metabolic utilization of glucose, the cells were grown in SF-900 II medium supplemented with [U-13C6]glucose and unlabelled glucose at a ratio of 1:99 (w/w) for two passages.
In order to study the metabolic utilization of exogenous phenylalanine, SF-9 cells were grown in "SF-900 II medium without amino acids and without sugar" (Gibco) with the following supplements: D-glucose (10 g/l), D-maltose (1 g/l), D-sucrose (1.65 g/l), β-alanine (300 mg/l), L-arginine (800 mg/l), L-asparagine (1.3 g/l), L-aspartate (1.3 g/l), L-cystin, sodium salt (150 mg/l), L-glutamate (1.5 g/l), L-glutamine (2 g/l), glycine (200 mg/l), L-histidine (200 mg/l), 2-hydroxyproline (700 mg/l), L-isoleucine (750 mg/l), L-leucine (250 mg/l), L-lysine hydrochloride (700 mg/l), L-methionine (1 g/l), 15N-L-phenylalanine (100 mg/l), L-proline (500 mg/l), DL-serine (400 mg/l), L-threonine (200 mg/l), L-tryptophan (100 mg/l), L-tyrosine (disodium salt dihydrate) (360 mg/l), and L-valine (500 mg/l).
Isolation of amino acids and ribonucleosides
S. frugiperda cells were harvested, washed with saline, and lyophilized. The dry powder (3.4 g) was extracted with 42 ml of a methylene chloride/methanol/water mixture (1.1:1:1.1, v/v), as described [22]. The residue was suspended in 14 ml of 1 M NaOH. The mixture was kept at room temperature overnight and was than centrifuged. Nucleotides were isolated from the supernatant as described earlier [23]. Treatment with alkaline phosphatase afforded ribonucleosides which were purified as described earlier [23]. The pellet was hydrolysed by treatment with azeotropic hydrochloric acid for 24 h, and amino acids were isolated chromatographically as described [23]. It should be noted that glutamine and asparagine are converted into glutamate and aspartate, respectively, under the acidic conditions of the hydrolysis. Therefore, analysed glutamic acid and aspartic acid also reflect the contributions of glutamine and asparagine, respectively.
Derivatization of amino acids
An aliquot of the hydrolysate was lyophilized and dissolved in 200 μl of HCl in n-butanol. The solution was incubated at 100°C for 1 h. Butanol was removed under a stream of nitrogen. The residue was incubated with 50 μl of trifluoroacetic acid anhydride at room temperature. The mixture was evaporated to dryness at 80°C. The residue was dissolved in 500 μl of ethyl acetate. An aliquot was analyzed by gas chromatography/mass spectrometry (column: DB5-MS, J+W scientific, CA, USA; start temperature, 90°C for 3 min; heating rate, 10°C min-1 up to 280°C; 8 min at 280°C).
NMR spectroscopy
Guanosine was dissolved in 0.5 ml of DMSO-D6, adenosine and cytidine were dissolved in 0.5 ml of D2O, respectively, and amino acids were dissolved in 0.5 ml of D2O/DCl. 1H, 13C, and 15N NMR spectra were acquired with a DRX 500 spectrometer from Bruker Instruments, Karlsruhe, at transmitter frequencies of 500.13, 125.76, and 50.7 MHz, respectively. The data were processed with standard Bruker software (XWINNMR 3.0). Composite pulse decoupling was used for 13C NMR measurements. No 1H decoupling was applied for 15N NMR measurements. All spectra were recorded using a flip angle of 30° and a relaxation delay of 2.0 s. Quadrature detection and quadrature phase cycling were applied in all NMR measurements. The resulting free induction decays were processed by zero filling and multiplication with a Gaussian window function (lb, -1; gb, 0.1). 3-(Trimethylsilyl)-1-propanesulfonate served as an external standard for 1H and 13C NMR measurements. 15N Chemical shifts are reported relative to the 15N NMR signal of N-5 of [U-15N4]6,7-dimethyl-8-ribityllumazine at 327.0 ppm. 1H and 13C NMR signal assignments were taken from [23].
Assessment of isotopolog composition
The isotopolog compositions of nucleosides and amino acids were determined by quantitative NMR spectroscopy (Tables 1 and 3) [24]. Relative 13C abundances were obtained from signal intensities in one-dimensional 13C NMR spectra. Specifically, the signal integrals (I*) were determined for each 13C NMR signal of the metabolite under study. Using the same experimental settings, the signal integrals (Iref) of the same compound at natural 13C abundance were determined. The ratios of the signal integrals of each respective carbon atom (I*/Iref) were then calculated. Absolute 13C abundances were determined for certain carbon positions from the 13C coupling satellites in the 1H NMR spectra (for an example, see Fig. 4). Relative 13C abundances (I*/Iref) were then referenced to the absolute enrichments.
In NMR spectra of multiply labelled samples displaying 13C13C coupling (for an example, cf. Fig. 1), each satellite signal in the 13C NMR spectra was integrated separately. The relative fraction of each respective satellite pair in the total 13C NMR signal integral of a given carbon atom was calculated (% 13C13C in Tables 1 and 3). The relative fractions of satellite pairs were then referenced to the overall 13C abundance of a respective carbon atom which was determined as described above. This resulted in molar contributions (mol%) for each respective isotopolog.
Authors' contributions
PA isolated the metabolites, MG grew the cells, HO and WE were responsible for the NMR analysis, WE and AB drafted the manuscript.
Acknowledgements
This work was supported by grants from the Deutsche Forschungsgemeinschaft, the Fonds der Chemischen Industrie and the Hans Fischer Gesellschaft. We thank Fritz Wendling, Angela Grygier and Angelika Werner for expert help with the preparation of the manuscript.
==== Refs
Luckow VA Summers MD High level expression of nonfused foreign genes with Autographa californica nuclear polyhedrosis virus expression vectors Virology 1989 170 31 39 2497580 10.1016/0042-6822(89)90348-6
Brüggert M Rehm T Shanker S Georgescu J Holak TA A novel medium for expression of proteins selectively labeled with 15N-amino acids in Spodoptera frugiperda (Sf9) insect cells J Biomol NMR 2003 25 335 348 12766395 10.1023/A:1023062906448
Strauss A Bitsch F Cutting B Fendrich G Graff P Liebetanz J Zurinin M Jahnke W Amino-acid-type selective isotope labeling of proteins expressed in Baculovirus-infected insect cells useful for NMR studies J Biomol NMR 2003 26 367 372 12815263 10.1023/A:1024013111478
Weiss SA Godwin GP Gorfien SF Whitford WG Serum-free media Bioprocess Technology 1993 17 179 194 7763506
Blomberg F Maurer W Rüterjans H Nitrogen-15 nuclear magnetic resonance investigations on amino acids Proc Natl Acad Sci USA 1976 73 1409 1413 5722
Rabenstein DL Mariappan SVS Determination of 15N isotope effects on the acid-base equilibria of amino groups in amino acids by carbon-13 NMR J Org Chem 1993 58 4487 4489 10.1021/jo00068a057
Leimer KR Rice RH Gehrke CW Complete mass spectra of N-trifluoroacetyl-n-butyl esters of amino acids J Chromatogr 1977 141 121 144 330555 10.1016/S0021-9673(00)99131-3
Eisenreich W Strauß G Werz U Bacher A Fuchs G Retrobiosynthetic analysis of carbon fixation in the phototrophic eubacterium Chloroflexus aurantiacus Eur J Biochem 1993 215 619 632 8354268 10.1111/j.1432-1033.1993.tb18073.x
Szyperski T Biosynthetically directed fractional 13C-labeling of proteinogenic amino acids. An efficient analytical tool to investigate intermediary metabolism Eur J Biochem 1995 232 433 438 7556192 10.1111/j.1432-1033.1995.tb20829.x
Schmidt K Marx A deGraaf AA Wiechert W Sahm H Nielsen J Villadsen J 13C tracer experiments and metabolite balancing for metabolic flux analysis: comparing two approaches Biotechnol Bioeng 1998 58 254 257 10191397 10.1002/(SICI)1097-0290(19980420)58:2/3<254::AID-BIT19>3.0.CO;2-C
Fiaux J Andersson CJY Holmberg N Bülow L Kallio PT Szyperski T Bailey JE Wüthrich K 13C NMR flux ratio analysis of Escherichia coli central carbon metabolism in microaerobic bioprocesses J Am Chem Soc 1999 121 1407 1408 10.1021/ja983786y
Park SM Klapa MI Sinskey AJ Stephanopoulos G Metabolite and isotopomer balancing in the analysis of metabolic cycles: II. Applications Biotechnol Bioeng 1999 62 392 401 9921151 10.1002/(SICI)1097-0290(19990220)62:4<392::AID-BIT2>3.0.CO;2-S
Bacher A Rieder C Eichinger D Arigoni D Fuchs G Eisenreich W Elucidation of biosynthetic pathways and metabolic flux patterns via retrobiosynthetic NMR analysis FEMS Microbiol Rev 1999 22 567 598 10.1016/S0168-6445(98)00034-5
Kruger NJ Ratcliffe RG Roscher A Quantitative approaches for analysing fluxes through plant metabolic networks using NMR and stable isotope labelling Phytochem Rev 2003 2 17 30 10.1023/B:PHYT.0000004255.59062.88
Glawischnig E Gierl A Tomas A Bacher A Eisenreich W Starch biosynthesis and intermediary metabolism in maize kernels. Quantitative analysis of metabolite flux by NMR Plant Physiol 2003 130 1717 1727 12481054 10.1104/pp.006726
Eisenreich W Ettenhuber C Laupitz R Theus C Bacher A Isotopolog perturbation techniques for metabolic networks. Metabolic recycling of nutritional glucose in Drosophila melanogaster Proc Natl Acad Sci USA 2004 101 6764 6769 15096588 10.1073/pnas.0400916101
Strauss A Bitsch F Fendrich G Graff P Knecht R Meyhack B Jahnke W Efficient uniform isotope labeling of Abl kinase expressed in Baculovirus-infected insect cells J Biomol NMR 2005 31 343 349 15929001 10.1007/s10858-005-2451-3
Silva FJ Bel Y Botella LM Cotton RG Ferre J Immunological detection of phenylalanine hydroxylase protein in Drosophila melanogaster Biochem J 1992 287 85 89 1417795
Bel Y Wiesner P Kayser H Candidate target mechanisms of the growth inhibitor cyromazine: studies of phenylalanine hydroxylase, puparial amino acids, and dihydrofolate reductase in dipteran insects Arch Insect Biochem Physiol 2000 45 69 78 11093244 10.1002/1520-6327(200010)45:2<69::AID-ARCH3>3.0.CO;2-3
Ruiz-Vázques P Silva FJ Aberrant splicing of the Drosophila melanogaster phenylalanine hydroxylase pre-mRNA caused by the insertion of a B104/roo transposable element in the Henna locus Insect Biochem Mol Biol 1999 29 311 318 10333570 10.1016/S0965-1748(99)00002-8
Drews M Doverskog M Ohman L Chapman BE Jacobsson U Kuchel PW Haggstrom L Pathways of glutamine metabolism in Spodoptera frugiperda (Sf9) insect cells: evidence for the presence of the nitrogen assimilation system, and a metabolic switch by 1H/15N NMR J Biotechnol 2000 78 23 27 10702908 10.1016/S0168-1656(99)00231-X
Bligh EG Dyer WJ A rapid method of total lipid extraction and purification Can J Biochem Physiol 1959 37 911 917 13671378
Eisenreich W Schwarzkopf B Bacher A Biosynthesis of nucleotids, flavins, and deazaflavins in Methanobacterium thermoautotrophicum J Biol Chem 1991 226 9622 9631 2033055
Eisenreich W Bacher A Setlow JK Elucidation of biosynthetic pathways by retrodictive/predictive comparison of isotopomer patterns determined by NMR spectroscopy 2000 22 Genetic Engineering, Principles and Methods. Kluwer Academic/Plenum Publisher, New York 121 153 11501374
|
16285881
|
PMC1310531
|
CC BY
|
2021-01-04 16:26:24
|
no
|
BMC Biochem. 2005 Nov 14; 6:24
|
utf-8
|
BMC Biochem
| 2,005 |
10.1186/1471-2091-6-24
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1131625962410.1186/1471-2458-5-113Study ProtocolProtocol for: Sheffield Obesity Trial (SHOT): A randomised controlled trial of exercise therapy and mental health outcomes in obese adolescents [ISRCNT83888112] Daley Amanda J [email protected] Robert J [email protected] Neil P [email protected] Jerry KH [email protected] The Department of General Practice and Primary Care, The Medical School, Clinical Sciences Building, University of Birmingham, UK, B15 2TT2 The Centre for Sport and Exercise Science, Sheffield Hallam University, Sheffield, S10 2BP, UK3 Sheffield Children's NHS Trust, Sheffield, S10 2TH, UK4 Academic Unit of Child Health, The Children's Hospital, Sheffield, S10 2TH, UK2005 31 10 2005 5 113 113 4 10 2005 31 10 2005 Copyright © 2005 Daley et al; licensee BioMed Central Ltd.2005Daley et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
While obesity is known to have many physiological consequences, the psychopathology of this condition has not featured prominently in the literature. Cross-sectional studies have indicated that obese children have increased odds of experiencing poor quality of life and mental health. However, very limited trial evidence has examined the efficacy of exercise therapy for enhancing mental health outcomes in obese children, and the Sheffield Obesity Trial (SHOT) will provide evidence of the efficacy of supervised exercise therapy in obese young people aged 11–16 years versus usual care and an attention-control intervention.
Method/design
SHOT is a randomised controlled trial where obese young people are randomised to receive; (1) exercise therapy, (2) attention-control intervention (involving body-conditioning exercises and games that do not involve aerobic activity), or (3) usual care. The exercise therapy and attention-control sessions will take place three times per week for eight weeks and a six-week home programme will follow this. Ninety adolescents aged between 11–16 years referred from a children's hospital for evaluation of obesity or via community advertisements will need to complete the study. Participants will be recruited according to the following criteria: (1) clinically obese and aged 11–16 years (Body Mass Index Centile > 98th UK standard) (2) no medical condition that would restrict ability to be active three times per week for eight weeks and (3) not diagnosed with insulin dependent diabetes or receiving oral steroids. Assessments of outcomes will take place at baseline, as well as four (intervention midpoint) and eight weeks (end of intervention) from baseline. Participants will be reassessed on outcome measures five and seven months from baseline. The primary endpoint is physical self-perceptions. Secondary outcomes include physical activity, self-perceptions, depression, affect, aerobic fitness and BMI.
==== Body
Background
The prevalence of obesity has reached alarming levels in Britain with several studies [1] reporting that the number of young people who are overweight and obese has increased notably over the past decade. This dramatic increase in overweight has not been confined to British children and adolescents; pediatric overweight is also increasing in other western countries [2,3]. While obesity is known to have many physiological consequences, the psychopathology of this condition has not featured prominently in the literature. Overweight children have increased odds of experiencing poor health related quality of life, particularly in the domains of psychosocial health, self-esteem and physical functioning [4]. Severely obese children and adolescents have been reported to have lower health related quality of life than children and adolescents who are healthy and to experience a similar quality of life as those diagnosed with cancer [5]. Cross sectional data has also demonstrated a relationship between depressive symptoms and body mass index (BMI) scores in pre-adolescent girls [6]. Overweight adolescents are more likely to be socially isolated than their normal weight counterparts [7]. Of more concern however is the evidence [8] that teasing about weight body has been consistently associated with high depressive symptoms and thoughts about attempting suicide in school children. Collectively, these findings have created a need for clinicians and researchers to address issues that are related to the long-term well-being of clinically obese young people. Without intervention, many of these negative feelings can persist into adulthood and further diminish quality of life and psychosocial health. The psychological needs of obese adolescents are unlikely to be met fully by clinicians or health professionals and exercise is not typically part of most rehabilitation programmes for obese young people. Evidently, it is important that interventions be offered to obese adolescents as part of their rehabilitation process so that they are able to participate in society to the same extent as their non-obese peers.
The health benefits of a physically active lifestyle are well documented [9] and body weight has been found to be associated with increased risk of hyperlipidemia, hypertension, insulin resistance and diabetes in later life [10]. Furthermore, physical activity for young people can contribute to the enhancement of psychological and social well-being [11]. Indeed, the value of such activities should not be underestimated given that research has consistently demonstrated that involvement in physical activity and exercise can positively improve the mental and social health of young people [12-16]. While the use of exercise as an intervention to promote psychological health outcomes has not been extensively investigated using RCT methodologies with obese children, some preliminary evidence [17,18] has indicated that participation in weight loss camps that involve physical activity can influence these outcomes. Such studies are few in number however, and typically have been poorly controlled. Where studies [19] have included psychological variables, these have not been designated as primary outcomes measures. A significant number of studies [19-23] have been based around weight loss programmes, such as restricting calorific intake and low fat diets, which, in themselves, might be a sources of distress in young people who are obese. Furthermore, although few studies have included waiting-list control groups [22], to our knowledge no published randomised controlled trial (RCT) has included an equal contact attention-control in an attempt to account for any attention effects that might be associated with different types of lifestyle or behavioural change interventions in this population. Research is also lacking on ways to tailor interventions to the needs and interests of clinically obese young people, which is perhaps partly attributable to the lack of detailed information provided by previous authors concerning their intervention approaches.
Interventions that address both the physical and psychological concerns associated with obesity are warranted. Using a randomised attention controlled methodology the Sheffield Obesity Trial (SHOT) was designed to evaluate the efficacy of exercise therapy as an intervention for improving both mental and physical health outcomes in obese young people. The primary trial hypothesis was that the exercise therapy intervention would lead to changes in participants' mental health and physical activity behaviour. By implication, these changes in physical activity behaviour might also translate into reductions in participants' BMI scores at follow-up.
Methods/design
Study aims
The primary aim of SHOT is to examine the effects of a supervised exercise therapy intervention in young people who are obese.
Study design and setting
SHOT is a pragmatic randomised controlled trial where obese young people aged 11–16 years are randomised to receive exercise therapy, usual care or an attention-control intervention. The study sample will consist of adolescents who have been referred to a children's hospital in the United Kingdom (UK) for evaluation of obesity or via community and media advertisements publicising the study. Participants recruited by community adverts will have their medical eligibility to enter the trial confirmed by one of the study paediatricians (third and fourth authors). In order to facilitate recruitment and retention a £25 sport store voucher will be given to participants at the end of the intervention phase and a contribution of £2.50 towards travel expenses will be made per visit.
Ethical considerations
Full ethical approval for this study has been obtained from the South Sheffield Local Research Ethics Committee. Written informed consent from all participants and their parents will be sought prior to their enrolment into the study. Participants will be asked to attend the dedicated project exercise facility for a familiarisation session with their parents before entering the trial
Study interventions
The exercise therapy sessions will take place in a dedicated project exercise therapy room housed at an English University. All exercise therapy sessions will take place one-to-one with the second author and last approximately 1 hr. Participants will be offered a range of aerobic exercise modalities, such as stepping, cycling, seated rowing, the dance mat and walking, and asked to exercise intermittently for 30 minutes, three times per week for eight weeks. The intermittent exercise will consist of a 4 minute warm up followed by four 4 minute bouts of moderate intensity exercise at 40–59% of heart rate reserve (%HRR) with 2 minute rests between each bout and 4 minute warm down. Mini games will also be included, primarily designed with fun in mind; they will also provide participants with the opportunity to experience personal development throughout the programme and introduce a small self-referenced competitive element into the sessions. Heart rate will be measured during the last minute of each 4 minute bout of exercise. Once participants have completed the eight week exercise intervention they will be given an individualised (moderate intensity) home exercise programme to follow for a further six weeks. It is hoped that the follow-up phase will help participants to move towards becoming autonomous exercisers and empower them to continue to commit to a lifestyle that involves regular aerobic exercise. During the exercise therapy sessions participants' rating of perceived exertion will be measured using the Pictorial Children's Effort Rating Table (PCERT) [24,25]. This instrument uses pictures as well as descriptive language, and has been found to reflect the changing physiological demands of given exercise tasks; higher ratings as measured by the PCERT corresponded with increases in exercise intensity [25]. Participants will be asked to estimate the exertion they feel on a 10-point scale as illustrated in Figure 1.
Figure 1 The Pictorial Children's Effort Rating Table (PCERT).
Motivating obese children to exercise cannot be achieved in the same way as for children of normal weight [26]. Not only are obese children physiologically different from children of normal weight, but they also have demonstrated significant emotional differences [27]. Additionally, as obese individuals tend both to be sedentary and to have had poor experiences with exercise [28], short bouts of intermittent exercise are considered most appropriate for this population. One of the common barriers to achieving long-term exercise habits in obese young people is the duration of physical activity that is often expected of them. Moreover, this can be a daunting task for these children and such high exercise demands are unlikely to be enjoyable or sustainable; this is particularly likely to be the case when obese individuals are in the early phases of adopting a physically active lifestyle. As the primary goal of this study was mental health outcomes, and not necessarily weight loss per se, intermittent exercise was considered most likely to provide opportunities for participants to experience a sense of accomplishment. The Department of Health [29] in England has recently advocated the use of short bouts of exercise accumulated throughout the day to gain health benefits.
Exercise counselling will be an integral part of the exercise sessions. It is hoped that exercise counselling will provide participants assigned to the exercise therapy group with the necessary knowledge and the psychological skills and tools to sustain changes in their exercise behaviour. This trial will use the Transtheoretical Model (TTM) [30] as the guiding framework for the exercise counselling to promote positive exercise attitudes and experiences. In line with TTM, weeks 1–4 will focus on cognitively based intervention strategies such as cognitive reappraisal and consciousness raising. During weeks 5–8, more behaviourally based interventions will be introduced, for example, goal setting, self-monitoring and finding social support. Participants will follow a broad structured curriculum of topics over the course of the intervention. Detailed descriptions of the strategies and techniques to be used during exercise counselling are outlined in Table 1. Weight loss per se will not be explicitly discussed with participants and no weight loss targets will be set, although sensible eating habits will be discussed and encouraged as part of the exercise therapy intervention.
Table 1 Exercise Counselling Protocol
Weeks Process of change Exercise Counselling framework: Examples of skills and techniques used
1–2 Cognitive Consciousness Raising a. Review first session:
• How did it feel? Was it difficult/easy?
• Did you enjoy it?
Dramatic Relief • Importance of exercise, why do we need to warm up & cool down
Decisional Balance • Heart rate monitoring, what to wear, what & when to drink
• What to expect in the coming weeks.
• Any questions
b. Healthy eating
• What is it?
• When should I eat?
• What type of foods are good/not so good?
• Hand out standard dietary information sheet
c. Benefits of exercise
• How often?
• How hard?
• Where and when?
3–4 Cognitive Self Re-evaluation d. Which physical exercises do I prefer?
Decisional Balance • Previous exercise experiences, why this worked / failed.
• What other exercises might you like to try?
Consciousness Raising e. Do you know?
• Benefits of exercise
• Importance of healthy eating
f. Are you enjoying the sessions?
• What do you like?
• What do you dislike?
• What would you change?
• Is it what you had expected?
g. Active and healthy living
• Food groups, choices, portion sizes.
• The breakfast challenge
• Review healthy eating card.
5–6 Cognitive and Behavioural Self Re-evaluation h. Evaluate sessions so far
• How do you feel about exercise now?
Goal setting/Self-regulation • How comfortable do you feel exercising?
• Which exercises do you enjoy the most?
Social Support i. Introduce goal setting
• What is it?
• How might it help?
• Set one exercise goal and one healthy eating goal for the week
j. Findings support for exercise
• Thinking of others who might encourage participation in exercise
• Finding someone to talk to when exercising is difficult
• Consider ways in which to exercise with other people
7–8 Behavioural Goal setting/Self-regulation k. Review goals
• Did you achieve them?
• If yes then well done!
Stimulus Control • If not then why not? What can we do to help change this?
Reinforcement Management and Self-Liberation l. Cues for action
• Thinking of tasks that might prompt participation in exercise
m. Thinking about moving on from the programme
• Home programme phase
• Future exercise options
n. Looking and planning ahead. SWOT analysis
• What will help me to exercise in the future?
• What will stop me?
o. What have I achieved so far
• Review exercise
• Review healthy eating
• What do I want to achieve from here?
• Thinking positively and taking positive action
• What has been learned
An attention-control intervention (body-conditioning) has been included in this trial in an attempt to control for any attention effects that might occur in participants assigned to the exercise group. This is particularly important in the current study because the exercise therapy group will receive one-to-one sessions with the researcher. Any attention-control condition must be relevant and meaningful, particularly when used with young people. We have tried to achieve this by presenting an alternative 'exercise' group that does not involve aerobic exercise but an apparently different type of exercise in the form of body conditioning activities. Thus, like the exercise therapy group, participants assigned to the attention-control group will attend the project exercise facility for 1 hr three times per week for eight weeks. HR will be maintained below 40% HRR. Attention-control sessions will include activities such as stretching, posture, twister, as well as static juggling and catching tasks. The format of the attention-control sessions will be similar to the exercise therapy sessions; involving four 4 minute body-conditioning activities with 2 minute rest between activities. During the remainder of the session, other sedentary activities and games such as pool, darts and table football will be included to help facilitate adherence to the intervention and to make the sessions interesting and engaging. The attention group will be given a home body-conditioning programme to follow for six weeks. The attention-control group will be asked to otherwise continue with their lifestyle as normal throughout the study and will not receive exercise counselling.
The usual care comparison group will be asked to continue with their lives as usual; they will be given the opportunity to complete exercise sessions at the centre once they had completed the study.
Determining eligibility for the study
Participants will be recruited according to the following criteria: (1) clinically obese and aged 11–16 years (Body Mass Index Centile > 98th UK standard) [31]; (2) no medical condition that would restrict ability to be active three times per week for eight weeks; and (3) not diagnosed with insulin dependent diabetes or receiving oral steroids.
Randomisation
A researcher from an independent University will perform the randomisation procedures by allocating participants to groups according to a computer generated random list.
Outcome measures
Physical self-perceptions
Physical self-perceptions served as the primary outcome measure. The Physical Self-perception Profile (CY-PSPP) was originally developed by Fox and Corbin [32] and later adapted for use with children by Whitehead [33]. The inventory contains six 6-item subscales; Sport/Athletic Competence, Attractive Body Adequacy, Condition, Strength, Physical Self-worth. The Children and Youth Physical Self-perception Profile (CY-PSPP) assesses the extent to which young people view themselves as competent in variety of physical domains. Each is devised in a structured alternative format on a scale between 1 (low score) and 4 (high score).
Self perceptions
Items measuring social acceptance, scholastic competence and global self-worth are to taken from Harter's Self Perception Profile for Adolescents [34]. The social acceptance subscale assessed the degree to which the adolescent feels accepted by their peers, feels popular, has lots of friends, and feels that he/she is easy to like. The scholastic competence items tap participants' perception of their competence or ability within the school context. The global self worth subscale assesses the extent to which participants like themselves as a person and the way they are living their lives. Each subscale contains six items devised in a structured alternative format on a scale between 1 (low competence) and 4 (high competence).
Depression
Depression will be assessed using the Children's Depression Inventory (CDI) [35]. The CDI is a 27-item self-rated symptom-orientated scale suitable for school-aged youngsters and adolescents. For each item, the child is asked to endorse one of three statements that best describe how he or she has typically felt over the past two weeks. Each response is scored as 0 (asymptomatic), 1 (somewhat symptomatic), or 2 (clinically symptomatic), contributing to a total CDI score that can range from 0–54.
Affective responses
In the absence of any exercise specific measures of affect for use with clinical child populations, items used by Ebbeck and Weiss [36] in sports settings will be included in this study. Participants responded to two subscales that assessed positive and negative affective responses over the previous week. On a scale between 1 (not at all or very slightly) to 5 (extremely) participants are asked to indicate the degree to which a series of positive and negative adjectives described how they have felt over the previous week.
Physical activity
The Physical Activity Questionnaire for Adolescents [37] will be used to collect detailed information about participants' involvement in different physical activities. Specifically, participants are asked about their involvement in (1) various physical activities in their spare time, (2) physical education, (3) lunchtime physical activities, (4) extra-curricular physical activities, (5) evening physical activities and (6) weekend activities. Each physical activity component is scored on a scale between 1 (not involved) to 5 (involved 5–7 times per week.
Anthropometrics, flexibility and aerobic fitness
As treadmill protocols engage a larger muscle mass than cycling and peak VO2 scores are more likely to be limited by central rather than peripheral factors [38], the poorly fit category of the modified Balke protocol [39], will be used to assess aerobic fitness. Distance walked in miles was recorded. Height was measured to the nearest completed 0.1 cm using a wall-mounted stadiometer. Weight was measured to the nearest 0.1 kg using a balance scale. From these values BMI (weight(kg)/(height(m)2) will be calculated and the standard deviation Score (SDS – or Z score) derived from the UK 1990 Data [40]; a child whose BMI exceeds the 98th percentile for age and sex according to UK reference data in 1990 will defined as obese for the purposes of this study. Severe obesity will be defined as a BMI SDS of >+3.5 (or an adult-equivalent BMI of >40). Participants' trunk and hamstring flexibility will be assessed using the modified Acuflex I flexibility test, which allows for variations in participants' arms and legs.
Assessment of outcomes
The main study outcomes will be assessed at baseline, as well as four (intervention midpoint) and eight weeks (end of intervention) from baseline. Participants will be reassessed on outcome measures at the end of the home intervention phase (approximately five months from baseline). To provide evidence on the possible longer-term effects of exercise therapy, a final assessment will be completed approximately seven months from baseline (see Figure 2). The second author will perform all assessments and deliver the exercise therapy and body conditioning sessions. Demographic data, including age and ethnicity and current physical activity participation will be collected at baseline. Adherence to the exercise therapy and attention-control interventions will also be monitored.
Figure 2 Intervention flow chart.
Sample size considerations
As there has been a lack of published studies in this field power calculations have been based upon related review studies in the field of exercise and mental health. Power calculations are based upon physical self-perceptions as the primary outcome measure; (predicted effect size = 0.6, providing 80% power, p < 0.05) indicating 30 participants would be needed for each group (n = 90). A 25% dropout rate has been assumed at eight weeks from baseline indicating that 122 participants might need to be recruited.
Statistical analysis
Differences in primary and secondary outcomes between control and intervention groups will be compared using intention to treat analysis. Imputation methods will be used to assess data losses through level drop out and loss to follow up. All results will be reported as means and 95% confidence intervals.
Time plan for the study
Participant recruitment began in April 2002 and by January 2006 all participants will have completed the trial and follow-up assessments of outcomes.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
Amanda Daley and Jerry Wales were responsible for identifying the research question and contributing to drafting the research protocol. Robert Copeland has contributed to the development of the protocol as member of the research team. All authors were responsible for the drafting of this paper and have read and approved the final version.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was supported by a research award to the first and fourth authors from The Health Foundation: grant number 2402/957.
==== Refs
Chinn S Rona R Prevalence and trends in overweight and obesity in three cross sectional studies of British children, 1974–1994 BMJ 2001 322 24 26 11141148 10.1136/bmj.322.7277.24
Hedley AA Ogden CL Johnson CL Carrol MD Curtin LR Flegal KM Prevalence of overweight and obesity among US children, adolescent and adults, 1999–2002 JAMA 2004 291 2847 2850 15199035 10.1001/jama.291.23.2847
World Health Organisation Obesity: preventing and managing the global epidemic 1987 Geneva. WHO [WHO Technical Report Series: No 894]
Friedlander SL Larkin EK Rosen C Palermo T Redline S Decreased quality of life associated with obesity in school-aged children Arch Ped Adolesc Med 2003 157 1206 1211 10.1001/archpedi.157.12.1206
Schwimmer J Burwinkle T Varni J Health-related quality of life of severely obese children and adolescents JAMA 2003 289 1813 1819 12684360 10.1001/jama.289.14.1813
Erikson S Robinson T Haydel F Killen J Are overweight children unhappy? Arch Ped Adolesc Med 2003 157 746 752 10.1001/archpedi.157.8.746
Strauss R Pollack HA Social marginalization of overweight children Arch Ped Adolesc Med 2000 154 931 935
Eisenberg M Neumark-Sztainer D Story M Associations of weight-based teasing and emotional well-being among adolescents Arch Ped Adolesc Med 2003 157 733 738 10.1001/archpedi.157.8.733
Chakravarthy MV Joyner MJ Booth FW An obligation for primary care physicians to prescribe physical activity to sedentary patients to reduce the risk of chronic health conditions Mayo Clin Proc 2002 77 165 173 11838650
Berenson GS Srinivasan SR Bao W Newman WP Tracy RE Wattigney WA Association between multiple cardiovascular risk factors and atherosclerosis in children and young adults: The Bogalusa Heart Study N Engl J Med 1998 338 1650 1656 9614255 10.1056/NEJM199806043382302
Biddle S Cavill N Sallis J Biddle S, Sallis J, Cavill N Policy framework for young people and health-enhancing physical activity Young and active? Young people and health-enhancing physical activity: Evidence and implications 1998 London: Health Education Authority 3 16
Mutrie N Parfitt G Biddle SJH, Cavill N, Sallis J physical activity and its link with mental, social and moral health in young people In Young and Active? Young People and Heath-enhancing physical Activity: Evidence and Implications London. Health Education Authority 49 68
Brown JD Seigel JM Exercise as a buffer of life stress: A prospective study of adolescent health in the pediatric population Clinics in Sports Med 1998 1 515 527
Calfas K Taylor WC Effects of physical activity on psychological variables in adolescents Ped Exer Sci 1994 6 406 423
Caruso C Gill DL Strengthening physical self-perceptions through exercise The J of Sports Med Physical Fitness 1992 32 416 427
Steptoe A Butler N Sports participation and emotional well-being in adolescents Lancet 1996 347 1789 1792 8667922 10.1016/S0140-6736(96)91616-5
Gately P Cooke CR Butterfly R Knight C Carroll S The acute effects of an 8-week diet, exercise and educational camp program Ped Ex Sci 2000 12 413 423
Walker LLM Gately PJ Bewick BM Hill AJ Children's weight-loss camps: psychological benefit or jeopardy I J of Obesity 2003 27 748 754 10.1038/sj.ijo.0802290
Mellin LM Slinkard LA Irwin CE Adolescent obesity intervention: validation of the SHAPEDOWN program J Am Diet Assoc 1987 87 333 338 3819254
Nova A Russo A Sala E Long-term management of obesity in paediatric office practice: experimental evaluation of two different types of intervention Ambul Child Health 2001 7 239 47 10.1046/j.1467-0658.2001.00135.x
Stolley MR Fitzgibbon ML Effects of an obesity prevention program on the eating behaviour of African American mothers and daughters Health Educ Behav 1997 24 152 64 9079575
Epstein LH Wing RR Koeske R Effects of diet plus exercise on weight change in parents and children J Consult Clin Psychol 1984 52 429 437 6747061 10.1037//0022-006X.52.3.429
Saelens B Sallis J Wilfley D Patrick K Cella J Buchta R Behavioral Weight Control for Overweight Adolescents Initiated in Primary Care Obesity Research 2002 10 22 32 11786598
Lamb KL Exercise regulation during cycle ergometry using the CERT and RPE scales Ped Exercise Sci 1996 8 337 350
Yelling M Lamb K Swaine I Validity of a pictorial perceived exertion scale for effort estimation and effort production during stepping exercise in adolescent children European Physical Education Review 2002 8 157 175
McWhorter JW Wallmann HW Alpert PT The obese child: motivation as a tool for exercise J Pediatr Health Care 2003 17 11 7 12533726 10.1067/mph.2003.25
Sothern M Hunter S Suskind R Brown R Udall J Blecker U Motivating the obese child to move: the role of structured exercise in pediatric weight management Southern Medical Journal 1999 92 577 584 10372851
American College of Sports Medicine (ACSM) Guidelines for exercise testing and prescription 2000 6 Philadelphia: Lea & Febiger 214 216
Department of Health At least five a week: Evidence on the impact of physical activity and its relationship to health 2004
Prochaska JO DiClemente CC Stages and processes of self-change of smoking: Towards an integrative model of change J of Consul and Clin Psychol 1983 51 390 395 10.1037//0022-006X.51.3.390
Cole TJ Freeman JV Preece MA Body mass index reference curves for the UK, 1990 Arch Dis Child 1995 73 25 9 7639544
Fox KR Corbin CB The physical self-perception profile: Development and preliminary validation J of Sport Ex Psychol 1989 11 408 430
Whitehead JR A study of children's physical self-perceptions using an adapted physical self-perception profile questionnaire Ped Ex Sc 1995 7 132 151
Harter S Manual for the Self-perception Profile for Adolescents 1995 Denver, Colo: University of Denver
Kovas M Children's Depression Inventory 1992 Multi-Health Systems Inc, USA 1 101
Ebbeck V Weiss M Determinants of children's self-esteem: An examination of perceived competence and affect in sport Ped Ex Sci 1998 10 285 298
Kowalski K Crocker PE Kowalski NP Convergent validity of the physical activity questionnaire for adolescents Ped Ex Sci 1997 9 342 352
Shephard RJ Tests of maximum oxygen intake. A critical review Sports Med 1984 1 99 124 6385195
Rowland TW Rowland TW Aerobic exercise testing protocols Pediatric laboratory exercise testing 1993 Champaign, IL: Human Kinetics 19 41
Freeman JV Cole TJ Chinn S Jones PRM White EM Preece MA Cross sectional stature and weight reference curves for the UK, 1990 Arch Dis Child 1995 73 17 24 7639543
|
16259624
|
PMC1310532
|
CC BY
|
2021-01-04 16:28:58
|
no
|
BMC Public Health. 2005 Oct 31; 5:113
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-113
|
oa_comm
|
==== Front
BMC Fam PractBMC Family Practice1471-2296BioMed Central London 1471-2296-6-461627114610.1186/1471-2296-6-46Research ArticleEffects of screening and brief intervention training on resident and faculty alcohol intervention behaviours: a pre- post-intervention assessment Seale J Paul [email protected] Sylvia [email protected] John M [email protected] IS [email protected] Barbara [email protected] Department of Family Medicine, Mercer University School of Medicine and Medical Center of Central Georgia, 3780 Eisenhower Parkway, Macon GA 31210, USA2 Institute of Public Health, Georgia State University, One Park Place South, Sixth Floor, Suite 660, Atlanta, GA 30302, USA2005 4 11 2005 6 46 46 10 6 2005 4 11 2005 Copyright © 2005 Seale et al; licensee BioMed Central Ltd.2005Seale et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Many hazardous and harmful drinkers do not receive clinician advice to reduce their drinking. Previous studies suggest under-detection and clinician reluctance to intervene despite awareness of problem drinking (PD). The Healthy Habits Project previously reported chart review data documenting increased screening and intervention with hazardous and harmful drinkers after training clinicians and implementing routine screening. This report describes the impact of the Healthy Habits training program on clinicians' rates of identification of PD, level of certainty in identifying PD and the proportion of patients given advice to reduce alcohol use, based on self-report data using clinician exit questionnaires.
Methods
28 residents and 10 faculty in a family medicine residency clinic completed four cycles of clinician exit interview questionnaires before and after screening and intervention training. Rates of identifying PD, level of diagnostic certainty, and frequency of advice to reduce drinking were compared across intervention status (pre vs. post). Findings were compared with rates of PD and advice to reduce drinking documented on chart review.
Results
1,052 clinician exit questionnaires were collected. There were no significant differences in rates of PD identified before and after intervention (9.8% vs. 7.4%, p = .308). Faculty demonstrated greater certainty in PD diagnoses than residents (p = .028) and gave more advice to reduce drinking (p = .042) throughout the program. Faculty and residents reported higher levels of diagnostic certainty after training (p = .039 and .030, respectively). After training, residents showed greater increases than faculty in the percentage of patients given advice to reduce drinking (p = .038), and patients felt to be problem drinkers were significantly more likely to receive advice to reduce drinking by all clinicians (50% vs. 75%, p = .047). The number of patients receiving advice to reduce drinking after program implementation exceeded the number of patients felt to be problem drinkers. Recognition rates of PD were four to eight times higher than rates documented on chart review (p = .028).
Conclusion
This program resulted in greater clinician certainty in diagnosing PD and increases in the number of patients with PD who received advice to reduce drinking. Future programs should include booster training sessions and emphasize documentation of PD and brief intervention.
==== Body
Background
Patients with hazardous and harmful drinking patterns are commonly encountered by primary care clinicians worldwide. Studies from the U.S., Europe and Australia indicate that 10–40% of patients seen in primary care settings engage in hazardous or harmful drinking [1-5]. This group of patients, sometimes referred to as "risky drinkers," includes patients who meet diagnostic criteria for alcohol abuse and alcohol dependence, as well as patients who exceed recommended "safe drinking guidelines" of the National Institute for Alcohol Abuse and Alcoholism and are at increased risk for alcohol-related problems [2]. Numerous randomized controlled trials have demonstrated that screening and brief intervention (SBI) are effective in reducing alcohol consumption among such drinkers [6-8], yet SBI still remains underutilized in primary care practices. Several older studies, many of them based on chart review data, suggested that problem drinking (PD) was largely undetected and untreated in primary care [9-12]. Two recent studies using other measurement techniques (clinician exit questionnaires and direct observation) have indicated that discussions of alcohol use by clinicians and patients in primary care are more frequent than previously thought, occurring in 9–10% of primary care encounters [13,14]. Nonetheless, studies conducted in the U.S., Australia, the United Kingdom and Finland indicate that clinicians frequently fail to screen for PD, and fail to address PD in at least one-third to one-half of cases, even when the diagnosis is known [13,15-20]. Indeed, 72% of U.S. primary physicians surveyed in 1999 reported that they preferred not to counsel early problem drinkers themselves but rather to refer them to a nurse trained in behavioural interventions [21]. Residents are less likely to perform brief interventions than faculty physicians [22], and only 13–20% of problem drinkers report receiving advice to reduce drinking, a key element of most effective SBI programs [17,22,23]. While studies have demonstrated that providing experiential training can increase primary care clinicians' rates of providing brief advice to problem drinkers [24-27], studies of the effect of resident training have yielded mixed results [28,29]. We previously reported initial findings from the Healthy Habits Project, a training program designed to increase SBI rates in a family medicine residency program using a combination of clinician training and clinic systems intervention. Based primarily on chart review, findings indicated that following program implementation, alcohol interventions increased from 12.5% to 47.7% of patients who screened positive for hazardous and harmful drinking, and that clinicians who were prompted with positive screening results gave advice to reduce drinking to 72% of patients [30]. This report provides further information, obtained using clinician exit questionnaires, regarding the impact of this SBI training program on resident and faculty physician alcohol intervention attitudes and behaviours. We hypothesized that the SBI training program would result in the following changes for both resident and faculty clinicians: (1) greater recognition of PD, (2) increased certainty in identifying PD, and (3) increased advice to reduce drinking.
Methods
The Healthy Habits Project utilized a combination of clinician training and a clinic-wide systems intervention program to increase alcohol screening and brief intervention in a family practice residency clinic in the southeastern U.S. Details of the program's systems interventions and training procedures have been previously described [30]. Briefly, the clinic is staffed by residents (28 physicians completing three years of post medical school training) and by faculty (8 family physicians and 2 physician assistants). All of the clinic's clinicians (residents and faculty) participated and gave written consent, and efforts were made to screen all adult patients during a 12-month period. The study was approved by the Institutional Review Board of the Medical Center of Central Georgia. Alcohol screening and intervention procedures were modeled after the University of Connecticut's Cutting Back Screening and Brief Intervention Program [26,31]. A clinic team implemented alcohol screening using the three-question AUDIT-C, a validated screening instrument for hazardous and harmful drinking [32-34], embedded in a health questionnaire distributed by registration clerks. After scoring the AUDIT-C, nurses asked screen-positive patients to complete the ten-question Alcohol Use Disorders Identification Test (AUDIT) [35]. All clinicians underwent 3 hours of training in which they were instructed to score the 10-question AUDIT and conduct brochure-guided brief interventions with all screen-positive patients. A key component of the intervention was negotiating a clinician-patient contract to reduce alcohol consumption during the ensuing 30 days. Training included a lecture, demonstration interviews, and role-playing exercises. Clinicians were asked to reschedule a follow-up visit within 30 days. Regular feedback sessions with clinic staff, nurses, and clinicians were implemented to encourage compliance with protocols. Periodic program evaluation by the project's implementation committee resulted in minor modifications of protocols which created three separate implementation phases: A (Months 1–3), B (Months 4–6), and C (Months 7–12). During Phase B, in an attempt to boost overall patient screening rates, monthly feedback sessions were scheduled with clinic staff and nurses. During Phase C, as a further attempt to increase screening rates, alcohol screening questions were integrated into the clinic's mandatory annual clinical information update.
Assessments
Clinician exit questionnaires (CEQ's)
A pre-post research design was used to measure the project's impact on clinicians' level of certainty in identifying problem drinking and their self-report of advice given to patients to reduce their drinking. Four sets of CEQ's were collected (longitudinally during the baseline month and Implementation Phase A, then for 1–2 weeks at the end of Phases B and C). Clinicians were asked to complete a three-question instrument (see Appendix 1) in which they reported whether they thought the patient they had just seen had a drinking problem, their degree of certainty of this diagnosis on a five-point Likert-type scale, and whether they had suggested that the patient stop drinking or cut back. All CEQ's were completed at the conclusion of individual patient encounters. The first, second, and fourth collections were performed by a research assistant, who approached individual clinicians in the hall immediately following two patient visits during each half-day clinic session. During the third collection period (Phase B), exit questionnaires were attached to each patient's routing form, and clinicians were requested to complete them on each patient seen.
Chart reviews
During two one-month periods (at baseline and during the project's final month), chart reviews of every fourth adult patient seen were conducted by one of two investigators (JPS, BB), who reviewed clinic notes from the patient's database, problem list, index visit, and all office visits during the previous year. Patients were considered "diagnosed" if alcohol abuse, alcohol dependence, heavy drinking, or similar terms were listed in the assessment area of a clinic note or the patient's problem list. Documentation of advice to quit drinking, cut back, or attend formal alcohol treatment was considered "intervention."
Statistical analysis
Statistical programs available in SPSS for Windows were used for analysis [36]. Pre- and post-assessments of the clinician's diagnostic impression, degree of certainty, and advice to decrease drinking were compared. Data were analyzed separately for faculty and residents, and then for all clinicians. Data from the three implementation phases of the study were analyzed separately and as aggregate data. Additional analyses were performed after removing CEQ's completed by two of the study's co-investigators (JPS, JB) to look for possible bias. All data were analyzed using Pearson's Chi-squared test. For the tables with expected cell frequencies less than five, Fisher's exact test was employed. Chi-square test for trend was used to determine diagnostic changes over time for faculty and residents. Analysis of variance (ANOVA) was used to compare differences between faculty and residents across baseline and study phases for the study's three primary outcome measures – mean number of subjects recognized with problem drinking, level of certainty in recognition, and number of patients given advice to reduce drinking. ANOVA was also used to assess possible impact of clinician gender and age on these three outcome measures. Test of linear trends was used to assess whether clinical diagnoses of problem drinking increased or decreased over time for patients who were evaluated by faculty, residents and all clinicians. The customary p-values of < 0.05 were used to indicate statistical significance.
Results
Thirty-eight clinicians completed exit questionnaires from a total of 1,052 patient encounters (164 at baseline, 888 during the implementation period). Residents were significantly younger than faculty clinicians (mean age 34 vs. 44 years, p < .0001; see Table 1). Although a higher percentage of residents were female, when compared to faculty, (64% vs. 30%), differences were not statistically significant (p = .078). During the third data collection period, in which exit questionnaires were self-administered, questionnaires were completed on 44% of adult patients seen during a one-week period. Chart reviews were conducted on 178 charts from the one-month baseline period and 200 charts from Month 12 of implementation.
Table 1 Baseline comparability: demographics of residents and faculty
Variable Faculty (n = 10) Residents (n = 28) Total Participants (n = 38)
Age, years mean (SD) 44.2 (SD 6.4) 34.0 (SD 6.6) 36.7 (SD 7.9)
% Female 30.0 64.3 55.3
Recognition of problem drinking
Overall, there was no statistically significant difference in mean number of subjects recognized as problem drinkers by faculty and residents across baseline and study phases. Clinicians reported problem drinking in 9.8% (16/164) of patients at baseline and 7.4% (66/888) of patients during the project's three implementation phases (p = .308). These rates are similar to the rates of risky drinking obtained by questionnaire screening using the AUDIT-C (8.6% during the baseline period and 8.0% during the implementation period [30]). Faculty impressions of problem drinking remained relatively stable throughout the project, while residents' diagnostic impressions showed a trend toward decline over time (see Table 2; p = .052). Despite equivalent percentages of patients who screened positive for hazardous drinking using the AUDIT-C at baseline (8.6%) and during the final phase of the study (8.8%), recognition rates for PD measured by CEQ were significantly higher than rates documented in patient charts during both periods (baseline: 9.8% vs. 1.2%, p = .035, and final study phase: 6.1% vs. 1.5%, p = .008).
Table 2 Changes in numbers and percent of patients diagnosed as problem drinkers across study phases
Diagnoses by faculty
N (%) Diagnoses by residents
N (%) Diagnoses by all clinicians
N (%)
Study Phase
Baseline 77 (9.1) 86 (10.5) 164 (9.8)
Phase A 222 (9.5) 279 (7.5) 501 (8.4)
Phase B 141 (5.7) 90 (6.7) 256 (6.3)
Phase C 78 (10.2) 43 (0) 131 (6.1)
p-value .782 .052 .148
p-value is for linear trends
Certainty in identifying problem drinking
Clinicians' mean level of certainty regarding the presence or absence of hazardous or harmful drinking among all patients was high both before (4.23, +/- 0.98) and after SBI implementation (4.21, +/- 0.97). ANOVA showed no differences between certainty levels for faculty and residents across all patient encounters (p = .446). Differences were observed, however, when analyses were limited to patients felt to have a drinking problem. Faculty level of certainty in patients with PD was greater than residents' level of certainty both before SBI program implementation (4.14 vs. 3.56) and after implementation (4.38 vs. 3.96), p = .028. After program implementation, levels of certainty for patients with PD increased for both faculty (4.38 vs. 4.14, p = .039) and residents (3.96 vs. 3.56, p = .030).
Advice to reduce drinking
Overall, comparative analysis of mean number of patients given advice to reduce drinking by faculty and residents across baseline and study phases did not demonstrate statistically significant changes. CEQ responses indicated that clinicians gave brief advice to 6.1% (10/164) of all patients seen at baseline and 8.6% (75/874) of patients seen during the project's implementation phase (p = .287); see Figure 1. Analyses revealed no impact of gender or clinician age on clinician advice rates. Intervention rates were highest during Phase A, the first three months following training. Rates decreased modestly for all providers during Phase B, then in Phase C showed declines for residents and increases for faculty. Comparisons of the mean number of all subjects receiving advice to reduce drinking by faculty vs. residents across study phases found that faculty were significantly more likely to advise patients to reduce drinking (p = .042). When baseline brief advice rates are compared with those during the combined intervention periods, residents showed greater increases in brief advice rates (from 4.7 % to 7.8%) than faculty (from 7.8 % to 9.3%); p = .041. Intervention rates increased among faculty who were co-investigators in the study (11.5% vs. 16.7%, p = .049) but showed no significant change among faculty who were not (9.8% vs. 7.0%, p = .421). When the analysis was limited to patients felt by clinicians to be problem drinkers, advice to reduce drinking increased from 50% (8/16) of problem drinkers during the baseline phase to 75% (49/65) during the three implementation phases (p = .038). While the overall number of patients receiving advice to reduce drinking was less than the number of patients thought to be problem drinkers during the baseline period (10/16, or 62%), the number of patients receiving advice to reduce drinking during the implementation phase actually exceeded the number of patients thought to be problem drinkers (75/66, or 114%).
Figure 1 Changes in clinician advice to reduce drinking (all patients).
Effect of clinician age and gender
Finally, ANOVA showed no statistically significant effect of age or gender of the faculty and residents on the study's three primary outcome measures – mean number of subjects recognized with problem drinking, level of certainty in recognition, and number of patients given advice to reduce drinking.
Discussion
Clinician attitudes and behaviours related to brief intervention
To our knowledge, this study is the third residency-based SBI training program to demonstrate positive changes in clinician attitudes and behaviours related to alcohol intervention. Researchers at the University of Massachusetts previously reported increases in readiness to intervene and in actual performance of brief interventions performed by residents and faculty physicians in a similar program which also provided clinician training, routine screening, and clinician prompting with screen-positive patients [24,37]. Wilk and Jensen [28] reported increases in brief interventions by residents in interviews with unannounced standardized patients following brief intervention training. In our study, both faculty and residents showed significant increases in their certainty in diagnosing PD after training. Residents showed greater increases in diagnostic certainty and intervention rates than faculty, an important finding in light of evidence from our study and others [22] that residents are less likely than faculty to perform brief interventions. Because previous studies have demonstrated that not all alcohol-related discussions include advice to reduce drinking, a key element of SBI cited by the US Preventive Services Task Force [38], our evaluation was focused on the percentage of patients who actually received advice to reduce drinking. During the project's intervention phase, clinicians reported a modest increase in providing advice to reduce drinking (from 6.8% to 8.6%). While this increase did not reach statistical significance (p = .287), a significant increase was seen in the percent of perceived problem drinkers receiving such advice (50% to 75%, p = .047). These findings, although based on small numbers of encounters with problem drinkers, are consistent with previous studies indicating that clinicians who have received SBI training are more confident in their ability to conduct brief interventions and more likely to intervene with problem drinkers [26,39,40]. In contrast to some earlier studies which found younger clinicians to be more willing to intervene than older clinicians [39,41,42], this study found no impact of age or gender on confidence in diagnoses of problem drinking or advice given to reduce drinking. Reasons for this finding are unclear, but could be related to the relatively young age of the overall group (mean age 37, with only two clinicians over age 50) or to the fact that clinicians of all ages received intensive training, which has been shown to correlate with greater clinician confidence and performance levels [40,42,43]. Further research is needed to determine which of the training program's multiple components – experiential training, implementation of routine alcohol screening performed by nurses, prompting clinicians with positive screening results and assessment data, or compliance feedback regarding intervention rates – were most critical in achieving increased intervention rates.
Overall, our study indicated that faculty prescribed reduction in drinking to more patients than residents did, but that resident intervention rates showed greater increases after training than faculty rates. Resident interventions did, however, show a non-significant trend toward decline toward the end of the one-year study. While this could represent a loss of training effect over time, evaluation results also may have been confounded by conducting the final exit interview evaluation in July, when skilled third-year residents had just graduated and newly-promoted residents were struggling to manage increased patient volumes. Interestingly, intervention prompt forms reviewed for the previous report on this project [30] indicate that there was no decline in resident interventions when prompted with positive screening results during this period: brief interventions were performed in 75% (6/8) of cases. Regardless of the reason for the decline, findings suggest that reinforcement methods such as booster sessions are needed to maintain behaviours taught in the initial training sessions.
One of this study's most encouraging findings is the fact that the number of patients receiving advice to reduce drinking after SBI training actually exceeded the number of patients felt to be problem drinkers. This finding suggests that the SBI program was successful in legitimizing and normalizing conversations about alcohol, such that clinicians felt more at ease in addressing alcohol use in a variety of clinical scenarios, and indicates that at-risk drinkers as well as problem drinkers received brief advice to reduce their drinking.
Recognition of problem drinking
In contrast to our expectation, clinician exit questionnaires did not reflect increased recognition of problem drinking after program implementation. While this could be due to the relatively high levels of clinician recognition at baseline, it could also reflect decreased clinician vigilance once routine screening protocols were in place. This finding, if confirmed in other studies, has implications for future SBI training programs, especially in light of the fact that most "routine" screening systems are not effective in detecting all risky drinkers. Clinician training should include reminders that significant numbers of risky drinkers may remain unscreened or escape detection via questionnaire screening, whose sensitivity rarely exceeds 85%, and clinicians must remain alert to clinical clues related to hazardous or problem drinking.
Documentation of problem drinking
One striking finding of this study is the marked difference between the perceived level of problem drinking (7.4–9.8% of patients) and the documentation of such diagnoses in the medical record (1.2–1.5%), suggesting that clinicians document perceived problem drinking in less than 25% of cases. Significant underdocumentation of alcohol disorders has been noted in several other previous studies [15,44,45], pointing out an important methodological flaw in many previous studies which have presumed that problem drinkers are "undiagnosed" based on chart review data [9-12]. This finding suggests that future studies assessing clinician recognition and intervention rates should utilize more sensitive measures such as clinician or patient exit interviews or direct observation, and that future training efforts should address documentation of alcohol disorders in the medical record.
Limitations of this pilot study
It is important to take into account possible methodological limitations of this pilot study. First, the sample size was not large, particularly during the baseline assessment period. However, the numbers in each group were sufficient to detect between-group differences of at least 10% or greater in recognition of problem drinking and at least 10% or greater in advice to reduce drinking. Secondly, a change in methodology during the study's second implementation phase could have confounded study results. During this period, exit questionnaires were attached to each patient's routing form, and clinicians were requested to complete them on each patient seen. The return rate was low (44%), and selection bias may have occurred, perhaps in favour of patients whom clinicians identified as problem drinkers or in favour of patients who received an intervention. Nonetheless, both clinicians' recognition (6.7% of patients) and intervention rates (8.0%) during this phase are within the range of the other phases of the study and do not suggest selection bias. Thirdly, this study lacked a criterion diagnosis for confirming problem drinking in patients considered by clinicians to be problem drinkers. While some patients may have been incorrectly diagnosed, rates of problem drinking recognition by clinicians throughout this study are similar to this study's previously-published problem drinking estimates obtained by AUDIT-C questionnaire screening, and only slightly lower than the estimated U.S. problem drinking prevalence of 11% in primary care [1]. Future studies comparing clinician's impressions with the results of standardized diagnostic interviews for problem drinking could help to clarify this issue.
Conclusion
This SBI training program resulted in greater clinician certainty in diagnosing PD and modest but significant increases in the number of patients with PD who received advice to reduce their drinking. The program shows promise for helping translate SBI findings into residency and clinical practice. Trends toward lower rates of identification of PD and intervention by residents during the program's later phases suggest a need for booster training sessions. Increased emphasis on documentation of problem drinking and brief intervention is also needed.
Appendix 1: exit questionnaire for clinicians
1). Do you think this patient has a drinking problem? Yes__ No
2). What is your degree of certainty?
Uncertain 1 2 3 4 5 Certain
3. Did you talk with this patient today about cutting back or quitting?
No___ Yes____
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JPS conceived of and designed the study, supervised acquisition of the data, participated in interpretation of data and drafted the manuscript. SS assisted in designing and implementing the study, interpreting the data, and editing the manuscript. JMB assisted in designing and implementing the study, interpreting the data and editing of the manuscript. ISO performed the statistical analysis and participated in drafting and editing the manuscript. BB participated in data collection, study design and implementation, and data analysis.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors would like to thank Dr. Dan Vinson for his help in study design and the directors of the Cutting Back program for sharing their technical expertise and the materials used in their program. We also thank Dr. Piyush Patel for his assistance in data collection. Financial support for this project was provided by the Clinical Research Center of the Medical Center of Central Georgia and by Department of Health and Human Services Grant #1 D12 HP 00159.
==== Refs
Fleming MF Graham AW Screening and brief interventions for alcohol use disorders in managed care settings Recent Dev Alcohol 2001 15 393 416 11449755
National Institute on Alcohol Abuse and Alcoholism Helping Patients with Alcohol Problems: A Health Practitioner's Guide 2003 National Institutes of Health, Public Health Service, U. S. Department of Health and Human Services NIH Publication No. 03-03769
Saitz R Clinical practice. Unhealthy alcohol use N Engl J Med 2005 352 596 607 15703424 10.1056/NEJMcp042262
World Health Organization Alcohol consumption in the European Region – consumption, harm and policies 2001 Copenhagen, World Health Organization Regional Office for Europe
Richmond RL G-Novak K Kehoe L Calfas G Mendelsohn COP Wodak A Effect of training on general practitioners' use of a brief intervention for excessive drinkers Aust N Z J Public Health 1998 22 206 9 9744177
Moyer A Finney JW Swearingen CE Vergun P Brief interventions for alcohol problems: A meta-analytic review of controlled investigations in treatment-seeking and non-treatment-seeking populations Addiction 2002 97 279 292 11964101 10.1046/j.1360-0443.2002.00018.x
Fleming M Manwell LB Brief intervention in primary care settings. A primary treatment method for at-risk, problem, and dependent drinkers Alcohol Res Health 1999 23 128 137 10890807
Bertholet N Daeppen JB Wietlisbach V Fleming M Burnand B Reduction of alcohol consumption by brief alcohol intervention in primary care: systematic review and meta-analysis Arch Intern Med 2005 165 986 95 15883236 10.1001/archinte.165.9.986
Moore RD Malitz FE Underdiagnosis of alcoholism by residents in an ambulatory medical practice J Med Educ 1986 61 46 52 3941421
Woodall HE Alcoholics remaining anonymous: resident diagnosis of alcoholism in a family practice center J Fam Pract 1988 26 293 6 3346632
Curtis JR Geller G Stokes EJ Levine DM Moore RD Characteristics, diagnosis, and treatment of alcoholism in elderly patients J Am Geriatr Soc 1989 37 310 6 2921452
Cerise FP Scarinci IC Thibodaux R Cannatella M Stark S Brantley PJ Alcoholism among indigent inpatients: identification and intervention by internal medicine residents South Med J 1998 91 27 32 9438398
Vinson DC Elder N Werner JJ Vorel LA Nutting PA Alcohol-related discussions in primary care: A report from ASPN J Fam Pract 2000 49 28 33 10678337
Stange KC Zyzanski SJ Jaen CR Callahan EJ Kelly RB Gillanders WR Shank JC Chao J Medalie JH Miller WL Crabtree BF Flocke SA Gilchrist VJ Langa DM Goodwin MA Illuminating the 'black box'. A description of 4454 patient visits to 138 family physicians J Fam Pract 1998 46 377 89 9597995
Conigliaro J Lofgren RP Hanusa BH Screening for problem drinking: impact on physician behavior and patient drinking habits J Gen Intern Med 1998 13 251 6 9565388 10.1046/j.1525-1497.1998.00075.x
Kaner EF Heather N Brodie J Lock CA McAvoy BR Patient and practitioner characteristics predict brief alcohol intervention in primary care Br J Gen Pract 2001 51 822 827 11677706
Aalto M Pekuri P Seppa K Primary health care professionals' activity in intervening in patients' alcohol drinking during a 3-year brief intervention implementation project Drug Alcohol Depend 2003 69 9 14 12536061 10.1016/S0376-8716(02)00228-4
Rydon P Redman S Sanson-Fisher RW Reid AL Detection of alcohol-related problems in general practice J Stud Alcohol 1992 53 197 202 1583898
Richmond RL Anderson P Research in general practice for smokers and excessive drinkers in Australia and the UK. III. Dissemination of interventions Addiction 1994 89 49 62 8148744
Kaner EF Lock CA McAvoy BR Heather N Gilvarry E A RCT of three training and support strategies to encourage implementation of screening and brief alcohol intervention by general practitioners Br J Gen Pract 1999 49 699 703 10756610
Spandorfer JM Israel Y Turner BJ Primary care physicians' views on screening and management of alcohol abuse: inconsistencies with national guidelines J Fam Pract 1999 48 899 902 10907628
Bradley KA Epler AJ Bush KR Sporleder JL Dunn CW Cochran NE Braddock CH 3rdMcDonell MB Fihn SD Alcohol-related discussions during general medicine appointments of male VA patients who screen positive for at-risk drinking J Gen Intern Med 2002 17 315 26 12047727 10.1046/j.1525-1497.2002.10618.x
Aalto M Pekuri P Seppa K Primary health care professionals' activity in intervening in patients' alcohol drinking: a patient perspective Drug Alcohol Depend 2002 66 39 43 11850134 10.1016/S0376-8716(01)00179-X
Adams A Ockene J Wheeler E Hurley T Alcohol counseling: Physicians will do it J Gen Intern Med 1998 13 692 698 9798817 10.1046/j.1525-1497.1998.00206.x
Richmond RL G-Novak K Kehoe L Calfas G Mendelsohn CP Wodak A Effect of training on general practitioners' use of a brief intervention for excessive drinkers Aust N Z J Public Health 1998 22 206 9 9744177
Babor TF Higgins-Biddle JC Higgins PS Gassman RA Gould BE Training medical providers to conduct alcohol screening and brief interventions Subst Abus 2004 25 17 26 15201108 10.1300/J465v25n01_04
Roche AM Stubbs JM Sanson-Risher RW Saunders J A controlled trial of educational strategies to teach medical students brief intervention skills for alcohol problems Prev Med 1997 26 78 85 9010901 10.1006/pmed.1996.9990
Wilk AI Jensen NM Investigation of a brief teaching encounter using standardized patients: teaching residents alcohol screening and intervention J Gen Intern Med 2002 17 356 60 12047732 10.1046/j.1525-1497.2002.10629.x
Saitz R Horton NJ Sullivan LM Moskowitz MA Samet JH Addressing alcohol problems in primary care: a cluster randomized, controlled trial of a systems intervention. The screening and intervention in primary care (SIP) study Ann Intern Med 2003 138 372 82 12614089
Seale JP Shellenberger S Tillery WL Boltri J Vogel R Barton B McCauley M Implementing Alcohol Screening & Intervention in a Family Medicine Residency Clinic Substance Abuse 2005
Babor TF Higgins-Biddle J Dauser D Higgins P Burleson J Alcohol screening and brief intervention in primary care settings: implementation models and predictors J Stud Alcohol 2005 66 361 8 16047525
Bush K Kivlahan DR McDonnell MB Fihn SD Bradley KA The AUDIT alcohol consumption questions (AUDIT-C): an effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test Arch Intern Med 1998 158 1789 1795 9738608 10.1001/archinte.158.16.1789
Gordon AJ Maisto SA McNeil M Kraemer KL Conigliaro R Kelly ME Conigliaro J Three questions can detect hazardous drinkers J Fam Pract 2001 50 313 320 11300982
Bradley KA Bush KR Epler AJ Dobie DJ Davis TM Sporleder JL Maynard C Burman ML Kivlahan DR Two brief alcohol-screening tests from the Alcohol Use Disorders Identification Test (AUDIT): Validation in a female Veterans Affairs patient population Arch Intern Med 2003 163 821 829 12695273 10.1001/archinte.163.7.821
Saunders JB Aasland OG Babor TF de la Fuente JR Grant M Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption – II Addiction 1993 88 791 804 8329970
SPSS for Windows Release 11.5.0 [computer program] 2002 Chicago: SPSS
Ockene JK Wheeler EV Adams A Hurley TG Hebert J Provider training for patient-centered alcohol counseling in a primary care setting Arch Intern Med 1997 157 2334 41 9361574 10.1001/archinte.157.20.2334
Whitlock EP Polen MR Green CA Orleans T Klein J U.S. Preventive Services Task Force Behavioral counseling interventions in primary care to reduce risky/harmful alcohol use by adults: a summary of the evidence for the U.S. Preventive Services Task Force Ann Intern Med 2004 140 557 68 15068985
Clement S The identification of alcohol-related problems by general practitioners Br J Addict 1986 81 257 64 3458494
Kaner EF Wutzke S Saunders JB Powell A Morawski J Bouix JC WHO Brief Intervention Study Group Impact of alcohol education and training on general practitioners' diagnostic and management skills: findings from a World Health Organization collaborative study J Stud Alcohol 2001 62 621 7 11702801
Roche AM Richard GP Doctors' willingness to intervene in patients' drug and alcohol problems Soc Sci Med 1991 33 1053 61 1771432 10.1016/0277-9536(91)90010-A
Roche AM Increasing primary care providers' willingness to intervene in alcohol- and drug-related problems: A review Substance Abuse 1996 17 201 217
Fucito L Gomes B Murnion B Haber P General practitioners' diagnostic skills and referral practices in managing patients with drug and alcohol-related health problems: implications for medical training and education programmes Drug Alcohol Rev 2003 22 417 24 14660131 10.1080/09595230310001613930
McDonald AJ 3rdWang N Camargo CA Jr US emergency department visits for alcohol-related diseases and injuries between 1992 and 2000 Arch Intern Med 2004 164 531 7 15006830 10.1001/archinte.164.5.531
Seppa K Lahtinen T Antila S Aalto M Alcohol drinking among emergency patients – alcometer use and documentation Alcohol Alcohol 2004 39 262 5 15082466
|
16271146
|
PMC1310533
|
CC BY
|
2021-01-04 16:29:14
|
no
|
BMC Fam Pract. 2005 Nov 4; 6:46
|
utf-8
|
BMC Fam Pract
| 2,005 |
10.1186/1471-2296-6-46
|
oa_comm
|
==== Front
BMC MedBMC Medicine1741-7015BioMed Central London 1741-7015-3-171631647210.1186/1741-7015-3-17Research ArticleLong-term use of non-steroidal anti-inflammatory drugs and the risk of myocardial infarction in the general population García Rodríguez Luis A [email protected]ález-Pérez Antonio [email protected] Centro Español de Investigación Farmacoepidemiológica (CEIFE), Madrid, Spain2005 29 11 2005 3 17 17 23 6 2005 29 11 2005 Copyright © 2005 García Rodríguez and González-Pérez; licensee BioMed Central Ltd.2005García Rodríguez and González-Pérez; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent data indicate that chronic use of coxibs leads to an increased occurrence of thrombotic cardiovascular events. This raises the question as to whether traditional non-steroidal anti-inflammatory drugs (tNSAIDs) might also produce similar hazards. Our aim has been to evaluate the association between the chronic use of tNSAIDs and the risk of myocardial infarction (MI) in patients.
Methods
We performed a nested case-control analysis with 4,975 cases of acute MI and 20,000 controls, frequency matched to cases by age, sex, and calendar year.
Results
Overall, current use of tNSAID was not associated with an increased risk of MI (RR:1.07;95%CI: 0.95–1.21). However, we found that the relative risk (RR) of MI for durations of tNSAID treatment of >1 year was 1.21 (95% CI, 1.00–1.48). The corresponding RR was 1.34 (95% CI, 1.06–1.70) for non-fatal MI. The effect was independent from dose. The small risk associated with long-term use of tNSAIDs was observed among patients not taking low-dose aspirin (RR: 1.29; 95% CI, 1.01–1.65). The effect of long-term use for individual tNSAIDs ranged from a RR of 0.87 (95% CI, 0.47–1.62) with naproxen to 1.38 (95% CI, 1.00–1.90) with diclofenac.
Conclusion
This study adds support to the hypothesis that chronic treatment with some tNSAIDs is associated with a small increased risk of non-fatal MI. Our data are consistent with a substantial variability in cardiovascular risks between individual tNSAIDs.
==== Body
Background
The recent withdrawal of rofecoxib together with new data showing that other cyclo-oxygenase-2 (COX-2) inhibitors might also be associated with an increased risk of developing adverse cardiovascular events, mainly acute myocardial infarction (MI), has created a climate of uncertainty surrounding the safety of these drugs and traditional non-steroidal anti-inflammatory drugs (tNSAIDs) [1,2]. In particular, this latter concern has been fostered by the preliminary communication of a press release suggesting that naproxen may be associated with an increased risk of cardiovascular and cerebrovascular events [3]. Two recently published studies found a similar result [4,5]. A feature in these newly released reports was that an increased risk only became apparent after prolonged administration of the suspected agent (in particular, coxibs), and notably in patients at a relatively low initial risk of cardiovascular disease.
Recently, we published a large epidemiologic study that evaluated the association between tNSAIDs as a class, as well as individual tNSAIDs, and the risk of MI [6]. Our overall estimate for tNSAIDs was compatible with either no association or a small increased risk, while the corresponding estimate for naproxen was compatible with either no association or a small reduced risk. We reported only briefly on the effect of duration of therapy in that paper, but now present more detailed information on the impact of duration of treatment on the relationship between tNSAIDs and MI.
Methods
We have used the same data set and similar analytical models to those previously reported [6]. In brief, the design was of a prospective cohort study with nested case-control analysis. Overall 4,975 cases of acute MI and death from coronary heart disease (CHD) aged 50 to 84 years were identified between 1997 and 2000, using the UK General Practice Research Database. A total of 20,000 controls were randomly sampled and frequency matched to cases by age, sex, and calendar year. Both cases and controls were required to be enrolled with their general practitioner for at least two years before entering the study. Using this data set, we estimated the effect of duration of tNSAIDS as a class and 3 individual tNSAIDs (diclofenac, ibuprofen and naproxen) on the risk of MI. We computed estimates of odds ratios using unconditional logistic regression to estimate the relative risks (RR)[7], and 95% confidence interval (CI) of MI associated with current use of tNSAIDs compared to non-use. Estimates were adjusted for sex, age, calendar year, anemia, smoking status, alcohol use, diabetes, hypertension, hyperlipidemia, body mass index, rheumathoid arthritis, osteoarthritis, prior cardiovascular disease, use of steroids, anticoagulants, aspirin, and paracetamol. We performed several sensitivity analyses to see whether the observed effect was independent of variations in the operational definition of exposure as well as duration. In the main analysis, we identified NSAID prescriptions before the index date for cases and controls, and categorized exposure to NSAIDs as in the original publication: "current," when the supply of the most recent prescription lasted until index date or ended in the 30 days before the index date; "recent," when it ended between 31 and 180 days before the index date; "past," when it ended between 6 months and 2 years before the index date; and "non-use," when there was no recorded use in the 2 years before the index date. We repeated the same regression models using a more restrictive time-window for current use of 7 days prior to the index date. We studied the effect of duration among current users, and evaluated duration of use adding the periods of "consecutive" prescriptions, defined as an interval of <1 month (main analysis) between the end of supply of one prescription (assuming adherence) and the date of prescription of the subsequent one. In secondary analyses of duration, we varied the interval to be either <1 week or <2 months. Finally, we performed 2 additional analyses: first one on "new current users" (those current users who did not receive a prescription for an NSAID in the 6 months prior to starting on NSAIDs); and the second being past users of NSAIDs (patients with NSAID use that ended between 31 days and 2 years before the index date) as the reference group.
Results
Overall, current intake of tNSAIDs was associated with a RR of 1.07 (95% CI, 0.95–1.21; Figure 1). The corresponding estimate when defining current exposure as use within a week prior to index date was 1.02 (95% CI, 0.90–1.17). The risk for treatment duration <1 year was no different from non-use, and for treatment duration >1 year, the RR was 1.21 (95% CI, 1.00–1.48): test for duration response trend, p .04. This estimate was slightly increased when using past users of NSAID as the reference group (RR:1.34;95%CI: 1.10–1.64). The median duration of use among long-term users was 2.8 years among cases and 2.6 among controls. A similar pattern of small increased risk of MI with longer duration was present with diclofenac, while this trend was not apparent for ibuprofen. On the other hand, the estimate of current use of naproxen was compatible with either no association or a small reduced risk of MI. The effect was present after a treatment of 1 month and persisted over longer durations of use (RR: 0.86; 95% CI, 0.58–1.27). Estimates of duration effects for tNSAIDs overall and naproxen, in particular, with 3 different assumptions about intervals between tNSAID use applied in the definition of the duration variables (Table 1). The increased risk with long-term treatment of tNSAIDs was most apparent when gaps between consecutive prescriptions were not permitted to exceed 7 days (RR: 1.31; 95% CI, 0.94–1.81). The varying intervals had no major impact on the estimates of duration-response associated with naproxen. We performed a similar sensitivity analysis for exposure to inhaled steroids (a priori not associated with MI). The estimates were similar in the 3 different scenarios (in order of increasing laxitude in the gap definition: 0.97 (0.62–1.49);0.95 (0.72–1.26);1.12 (0.91–1.37)). The duration effect of NSAIDs according to the concomitant use of low-dose aspirin and daily dose of tNSAIDs is shown in Table 2. The excess risk associated with long term use of tNSAIDs was observed among patients not taking concomitantly low-dose aspirin (RR: 1.29; 95% CI, 1.01–1.65), while long term use of tNSAIDs did not appear to increase the risk of MI among patients taking cardioprotective aspirin (RR: 0.90; 95% CI, 0.61–1.32). Data were too scarce to evaluate the interaction between aspirin use and chronic treatment with individual NSAIDs. No major variation in the risk of MI associated with chronic use of tNSAIDs was observed between users of low-medium dose versus high dose (Table 2). The corresponding estimates of RR associated with NSAID duration of >1 year for fatal and non-fatal MI were 1.02 (95% CI, 0.76–1.37) and 1.34 (95% CI, 1.06–1.70), respectively. The corresponding estimates of non-fatal MI for diclofenac, ibuprofen, and naproxen were 1.82 (1.27–2.62), 1.19 (0.70–2.03), and 1.00 (0.48–2.10), respectively. "New current users" with duration of use longer than 1 year had a RR of 1.31 (95%CI: 0.95–1.80).
Figure 1 Duration of use of NSAIDs and individual NSAIDs among current users (use within a month) and risk of MI. Overall, current use of tNSAIDs was associated with a RR of 1.07 (95% CI, 0.95–1.21). The corresponding estimates for diclofenac, ibuprofen and naproxen were 1.17 (0.98–1.40), 1.05 (0.86–1.28) and 0.89 (0.64–1.25), respectively. Duration of use was computed adding the periods of "consecutive" prescriptions, defined as an interval of less than one month between the end of supply of one prescription and the date of prescription of the subsequent one. Estimates are adjusted for sex, age, calendar year, anemia, smoking status, alcohol use, diabetes, hypertension, hyperlipidemia, BMI, RA, OA, prior cardiovascular disease, use of steroids, anticoagulants, aspirin, paracetamol, and NSAIDs. The duration response trends for NSAIDs, diclofenac, ibuprofen and naproxen were P = .04, P = .02, P = .47, P = .54 respectively.
Table 1 Duration of use of NSAIDs and naproxen among current users (use within a month) and risk of MI according to different definitions of the interval between consecutive prescriptions*
Interval = 7 days Interval = 30 days Interval = 60 days
OR (95%CI) † OR (95%CI) † OR (95%CI) †
NSAIDs
Dura 0–30 days 0.97 (0.83–1.14) 1.00 (0.82–1.21) 1.13 (0.92–1.40)
Dura 31–365 days 1.14 (0.97–1.34) 1.04 (0.89–1.23) 1.00 (0.83–1.19)
Dura >365 days 1.31 (0.94–1.81) 1.21 (1.00–1.48) 1.10 (0.93–1.30)
Naproxen‡
Dura 0–30 days 0.81 (0.49–1.35) 0.95 (0.52–1.75) 0.91 (0.46–1.79)
Dura > 30 days 0.94 (0.61–1.45) 0.86 (0.58–1.27) 0.88 (0.60–1.28)
* Duration of use was computed adding the periods of "consecutive" prescriptions, defined as varying intervals between the end of supply of one prescription and the date of prescription of the subsequent one.
† Estimates are adjusted for sex, age, calendar year, anemia, smoking status, alcohol use, diabetes, hypertension, hyperlipidemia, BMI, RA, OA, prior cardiovascular disease, use of steroids, anticoagulants, aspirin, paracetamol, and NSAIDs.
‡ Due to the limited number of observations in the duration strata of 31–365 days and the lack of heterogeneity in estimates of risk between this strata and the duration strata of greater than one year, we collapsed the two of them into one single duration strata.
Table 2 Duration of NSAID use among current users (use within a month) and risk of MI stratified by aspirin use and NSAID daily dose*
Cases (%) Controls (%) OR (95%CI)
Aspirin non users†
NSAID duration
Dura 0–30 days 100 (28.2) 481 (31.2) 1.03 (0.81–1.30)
Dura 31–365 days 149 (42.0) 680 (44.1) 1.04 (0.85–1.27)
Dura >365 days 106 (29.8) 381 (24.7) 1.29 (1.01–1.65)
Aspirin current users†
NSAID duration
Dura 0–30 days 46 (26.4) 91 (28.8) 0.94 (0.64–1.40)
Dura 31–365 days 77 (44.3) 126 (39.9) 1.08 (0.78–1.50)
Dura >365 days 51 (29.3) 99 (31.3) 0.90 (0.61–1.32)
NSAID low-medium dose‡
NSAID duration
Dura 0–30 days 92 (28.5) 350 (31.3) 1.04 (0.81–1.34)
Dura 31–365 days 136 (42.1) 484 (43.4) 1.05 (0.84–1.32)
Dura >365 days 95 (29.4) 282 (25.3) 1.30 (1.00–1.71)
NSAID high dose‡
NSAID duration
Dura 0–30 days 70 (27.2) 269 (30.8) 0.98 (0.73–1.31)
Dura 31–365 days 110 (42.8) 382 (43.8) 1.10 (0.86–1.41)
Dura >365 days 77 (30.0) 222 (25.4) 1.21 (0.90–1.64)
* Duration of use was computed adding the periods of "consecutive" prescriptions, defined as an interval of less than one month between the end of supply of one prescription and the date of prescription of the subsequent one.
† Estimates are adjusted for sex, age, calendar year, anemia, smoking status, alcohol use, diabetes, hypertension, hyperlipidemia, BMI, RA, OA, prior cardiovascular disease, use of steroids, anticoagulants and paracetamol.
‡ Estimates are adjusted for sex, age, calendar year, anemia, smoking status, alcohol use, diabetes, hypertension, hyperlipidemia, BMI, RA, OA, prior cardiovascular disease, use of steroids, anticoagulants, aspirin and paracetamol. Cut-off values for dose in mg were: aceclofenac 100, acemetacin 120, apazone 600, diclofenac 75, etodolac 400, fenbufen 900, fenoprofen 1200, flurbiprofen 150, ibuprofen 1200, indomethacin 75, ketoprofen 100, mefenamic 1000, meloxicam 7.5, nabumetone 1000, naproxen 500, piroxicam 10, sulindac 200, tenoxicam 10, and tiaprofenic acid 450.
Discussion
This additional analysis shows that use of tNSAIDs as a class for duration of continuous use of <1 year in general practice appears not to have a clinically relevant influence on the occurrence of MI. Our data are also compatible with the possibility that long-term therapy with tNSAIDs may confer a small excess risk of developing MI of ~20%, in particular of non-fatal MI. Our risk estimate is similar to the one provided by the authors of the first published epidemiologic study that analyzed the association between tNSAIDs and MI, who reported an estimate of relative risk for tNSAID duration longer than one year of 1.25; (95% CI, 0.90–1.72)[8]. The excess risk of MI among chronic users appears to be independent of NSAID daily dose and concentrated among patients not taking low-dose cardioprotective aspirin (eg. patients on average with a low cardiovascular profile). Our sensitivity analysis on the impact of varying operational definitions of duration for tNSAIDs as a class yielded small differences in the corresponding estimates associated with long-term treatment. The most stringent criteria of requiring a gap or interval of no greater than 1 week were the circumstances under which the increased risk with long-term tNSAID therapy was most noticeable, whereas the most liberal gap (up to 2 months) revealed a relative risk closer to no increased risk. This last result could be due in part to misclassification of the "true" continuous duration sequence in real life.
Even though our study was not designed to make direct comparissons between NSAIDs, we observed that at least three individual NSAIDs (the 3 most widely used) appeared to have different risk profile within a seemingly heterogeneous group of tNSAIDS. A clear duration-response was found for diclofenac, with an emerging risk associated with chronic use extending over 1 year. Secondly, we found no clear trend with ibuprofen duration – albeit that could, in part, be explained by some misclassification of exposure due to over the counter use of ibuprofen. Finally, there was no evidence of an increased risk associated with naproxen use and our results are more compatible with a minor reduced risk of MI. Misclassification due to over the counter use could also be an issue for naproxen. The risk with naproxen was relatively insensitive to varying the interval used to calculate the length of treatment duration. A potential small protective effect associated with naproxen has been reported in most epidemiological studies [6], but not all [9]. These results are at odds with the unpublished suggestion of a cardiovascular hazard from naproxen in a clinical trial stopped prematurely last year and under controversial circumstances [3]. The apparent level of protection afforded by regular intake of naproxen in our study may be due to suppression of platelet thromboxane production, when administered twice a day [11]. This is the mechanism by which aspirin affords cardioprotection. However, unlike aspirin, some, but not all patients, treated chronically with naproxen maintain suppression of thromboxane throughout the whole dosing interval at a level compatible with effective platelet inhibition [10,11]. Furthermore, many individuals classified as "chronic users" of naproxen in this general population may not have been compliant with the required dosing regimen (twice a day) all throughout the period of prescribed drug intake further diluting the benefit of an "aspirin effect". Thus, one would not expect to observe a benefit similar to the one of low-dose aspirin in the range of a 20–25% reduction in the incidence of events. No other widely used tNSAID shares the same pharmacodynamic/pharmacokinetic properties of naproxen on platelets and consequently other tNSAIDs are not expected to confer any degree of cardioprotection. Due to scarce numbers of chronic users for other individual NSAIDs, we could not estimate their respective duration-response relationship. Yet, one can not assume a class effect based on the heterogeneity observed between the 3 individual NSAIDs. In particular, these observations question the emerging strategy of comparing naproxen with "non-naproxen tNSAIDs". Direct comparison of our findings with those of recent coxib/tNSAID randomized clinical trials (RCTs) are hampered by the use in RCTs of significantly higher average daily doses of tNSAIDs together with more strict compliance than the one occuring in "real life" situations [1-3].
We analyzed the risk of inhaled steroids (a drug class considered a priori not to be associated one way or the other with the risk of MI) to confirm the internal validity of our sensitivity analysis of the operational definition of duration. The estimates of RR for long-term duration of inhaled steroids changed minimally using the 3 different intervals, ranging between 1.0 and 1.1. A limitation in the interpretation of our results is the imprecision of most estimates of risk, given that the magnitude of the "true" association we have purported to evaluate is most likely weak, either of increased risk with long-term duration of NSAIDs or reduced risk with naproxen. In this scenario, the common limitations in all observational studies (potential residual or unmeasured confounding) could be affecting our results. Over-the-counter (OTC) medications are not recorded in our source of information. Yet, it should be noted that OTC long-term use of tNSAIDs or cardioprotective aspirin is uncommon in the UK and could have slightly underestimated our measures of association, in particular ibuprofen.
Conclusion
The overall experience with tNSAIDs in our study is generally congruent with a neutral effect on cardiovascular disease although there is a suggestion of a small excess risk with chronic exposure, especially for non-fatal MI. The summary estimate for the tNSAID group is composed of substantial variation in risk between the three studied individual tNSAIDs suggesting a biologically plausible heterogenity in cardiovascular risk [1,12]. Our study suggests either no effect or a small reduction of cardiovascular risk during sustained treatment with naproxen, a small increased risk with diclofenac, and an undetectable risk with ibuprofen. Larger observational studies and, if at all possible, randomised clinical trials will be necessary to address this hypothesis of mechanistic heterogenity amongst tNSAIDs with respect to cardiovascular risk.
Abbreviations
cyclo-oxygenase-2 :COX-2
myocardial infarction: MI
traditional non-steroidal anti-inflammatory drugs : tNSAIDs
coronary heart disease: CHD
relative risk: RR
randomized clinical trials: RCTs
over-the-counter: OTC
Competing interests
We used data from a previous study that was supported in part by a research grant from Pharmacia. Two employees of Pharmacia (now Pfizer) coauthored that paper. In the present study the funding body had no role.
Authors' contributions
Both LAGR and AGP participated in the conception, design, analysis, interpretation of data, and drafting of the manuscript. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank the general practitioners for their excellent collaboration and the Boston Collaborative Drug Surveillance Program for providing access to the General Practice Research Database. We also thank Drs Garret A. FitzGerald, Steven Lanes, Cristina Varas and Susana Pérez-Gutthann for their comments on earlier versions of the manuscript.
==== Refs
FitzGerald GA Coxibs and cardiovascular disease N Engl J Med 2004 351 1709 1711 15470192 10.1056/NEJMp048288
National Institutes of Health, U.S. Department of Health and Human Services. NIH News: NIH Halts Use of COX-2 Inhibitor in Large Cancer Prevention Trial. On NIH website [updated 2004 Dec 17]
Solomon SD McMurray JJ Pfeffer MA Pfeffer MA Wittes J Fowler R Finn P Anderson WF Zauber A Hawk E Bertagnolli M Adenoma Prevention with Celecoxib (APC) Study Investigators: Cardiovascular risk associated with celecoxib in a clinical trial for colorectal adenoma prevention New Engl J Med 2005 352 1071 1080 15713944 10.1056/NEJMoa050405
Johnsen SP Larsson H Tarone RE McLaughlin JK Norgard B Friis S Sorensen HT Risk of hospitalization for myocardial infarction among users of Rofecoxib, Celecoxib, and Other NSAIDs Arch Int Med 2005 165 978 984 15883235 10.1001/archinte.165.9.978
Hippisley-Cox J Coupland C Risk of myocardial infarction in patients taking cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs: population based nested case-control analysis BMJ 2005 330 1366 1372 15947398 10.1136/bmj.330.7504.1366
García Rodríguez LA Varas-Lorenzo C Maguire A González-Pérez A Nonsteroidal Antiinflammatory Drugs and the Risk of Myocardial Infarction in the General Population Circulation 2004 109 3000 3006 15197149 10.1161/01.CIR.0000132491.96623.04
Walker AM Observation and inference An introduction to the methods of epidemiology 1991 Newton Lower Falls: Epidemiology Resources Inc
García Rodríguez LA Varas C Patrono C Differential effects of aspirin and non-aspirin nonsteroidal antiinflammatory drugs in the primary prevention of myocardial infarction in postmenopausal women Epidemiology 2000 11 382 387 10874543 10.1097/00001648-200007000-00004
Graham DJ Campen D Hui R Spence M Cheetham C Levy G Shoor S Ray WA Risk of acute myocardial infarction and sudden cardiac death in patients treated with cyclo-oxygenase 2 selective and non-selective non-steroidal anti-inflammatory drugs: nested case-control study Lancet 2005 365 475 481 15705456
Capone ML Tacconelli S Sciulli MG Grana M Ricciotti E Minuz P Di Gregorio P Merciaro G Patrono C Patrignani P Clinical pharmacology of platelet, monocyte, and vascular cyclooxygenase inhibition by naproxen and low-dose aspirin in healthy subjects Circulation 2004 109 1468 1471 15037526 10.1161/01.CIR.0000124715.27937.78
Reilly IA FitzGerald GA Inhibition of thromboxane formation in vivo and ex vivo: implications for therapy with platelet inhibitory drugs Blood 1987 69 180 186 3790723
Catella-Lawson F Reilly MP Kapoor SC Cucchiara AJ DeMarco S Tournier B Vyas SN FitzGerald GA Cyclooxygenase inhibitors and the antiplatelet effects of aspirin N Engl J Med 2001 345 1809 1817 11752357 10.1056/NEJMoa003199
|
16316472
|
PMC1310534
|
CC BY
|
2021-01-04 16:27:56
|
no
|
BMC Med. 2005 Nov 29; 3:17
|
utf-8
|
BMC Med
| 2,005 |
10.1186/1741-7015-3-17
|
oa_comm
|
==== Front
BMC UrolBMC Urology1471-2490BioMed Central London 1471-2490-5-131628197010.1186/1471-2490-5-13Research ArticleImmediate endoscopic management of complete iatrogenic anterior urethral injuries: A case series with long-term results Maheshwari Pankaj N [email protected] Hemendra N [email protected] Department of Urology, R. G. Stone Urological Research Institute, Mumbai, India2 Presently consultant urologist, Bombay Hospital, Indore, India2005 9 11 2005 5 13 13 19 5 2005 9 11 2005 Copyright © 2005 Maheshwari and Shah; licensee BioMed Central Ltd.2005Maheshwari and Shah; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Urethral injury produces partial or complete disruption of the urethral integrity. Advances in endourology have made endoscopic management of most of these injuries feasible without greatly compromising the final result. We report our institutional experience of immediate endoscopic realignment of complete iatrogenic anterior urethral injury.
Methods
From May 1997 to May 2003, seven patients with complete anterior urethral disruption were managed by immediate endoscopy guided splinting of urethra. Retrograde urethroscopy, combined with fluoroscopic guidance and in some cases antegrade cystoscopy through a suprapubic stab cystostomy was performed. A guide wire was negotiated across the disruption. Later, a 16 F Foley catheter was placed for 1–3 weeks. Patients were followed up at 1, 3, 6 and 12 months and then yearly to assess the long-term outcome of endoscopic management.
Results
Immediate endoscopic realignment was achieved in all patients. Three patients developed recurrence at six months; that was treated by optical urethrotomy. Only one patient developed multiple recurrences over an average follow-up of 49.2 months (range 7 to 74 months). He was offered open end-to-end urethroplasty at twenty months after third recurrence. Thus immediate endoscopic realignment avoided any further intervention in four patients (57.14%); while after an additional optical urethrotomy, urethroplasty could be avoided in six patients (87.2%).
Conclusion
Immediate endoscopic realignment of traumatic urethral disruption is a feasible, safe and effective treatment modality for management of patients with iatrogenic complete anterior urethral injuries.
==== Body
Background
Anterior urethral rupture is rare and is usually iatrogenic, during catheterisation or urethral dilatation. Trauma resulting from inflating a Foley balloon in the urethra occurs frequently. It may also result from blunt trauma e.g. saddle injury.
The definitive management of anterior urethral injury remains controversial [1]. Initial supra pubic cystostomy and delayed urethral re-construction had been considered as a reference standard for managing posterior urethral injuries [2]. This approach has problems like need of a supra pubic drainage for prolonged period (6 weeks to 3 months) as well as formation of an inevitable urethral stricture requiring reconstructive urethroplasty.
Recent advances in endourological techniques have made primary realignment feasible to perform with minimal manipulation. This technique realigns the urethra over a urethral catheter that splints the urethra till the healing occurs [3]. We herewith retrospectively analyse our experience with immediate endoscopic realignment of iatrogenic complete anterior urethral injuries.
Methods
Patients
From May 1997 to May 2003, seven men were diagnosed at our institute with complete iatrogenic anterior urethral injury. Patients had acute retention of urine and/or bleeding per urethra and were treated on emergency basis. A detailed clinical evaluation was performed and previous treatment records were reviewed to know the exact pathology in the urethra. Patients were subjected to primary endoscopic realignment of their urethral injury. An informed consent was taken for supra pubic cystostomy in all cases.
Operative technique
Under spinal or epidural anaesthesia patients were placed in lithotomy position with the legs on adjustable leg supports. A dynamic ascending urethrogram was performed under fluoroscopic guidance to confirm extravasation of contrast. Complete rupture was suspected when there was failure to delinate proximal urethra during urethrography.
Initial urethroscopy was done with 17 F cystoscopic sheath and a 0° telescope. Normal saline irrigation was kept to as minimum as possible. Clots in the urethra were gently evacuated to reach the site of urethral disruption. Once complete lack of urethral mucosal integrity was confirmed endoscopically following steps were used to delineate proximal urethra:
Step 1- Retrograde endoscopic delineation of proximal urethra
Bladder was punctured suprapubically with 16 F initial puncture needle and methylene blue was injected in bladder. A suprapubic pressure was applied with the simultaneous urethroscopy to identify possible efflux of methylene blue at the site of trauma. An attempt was made to gently negotiate 0.035"hydrophilic glide wire through the area effluxing methylene blue. If attempt succeeded, position of glide wire in bladder was confirmed fluoroscopically. A stiff zebra wire replaced the glide wire. This served as a guide for subsequent urethroscopy.
Step 2- Antegrade fluoroscopic guided delineation of proximal urethra
If the glide wire could not be negotiated retrogradely; an angled tip glide wire was passed from initial puncture needle into the bladder. Under fluoroscopic guidance multiple attempts were made to negotiate glide wire into posterior urethra through bladder neck. Once glide wire reached the posterior urethra a 6 F open-ended ureteric catheter was passed over it into posterior urethra & a glide wire was exchanged for a guide wire.
Step 3- Antegrade endoscopic delineation of proximal urethra
If fluoroscopically guided attempts to negotiate guide wire in to the posterior urethra failed; it was supplemented with endoscopic guidance. The supra pubic tract was dilated up to 24 F using Alken dilators and an Amplatz sheath was than used as a conduit to pass a rigid cystoscope into bladder. Bladder neck was than explored with antegrade cystoscopy and an open ended ureteric catheter was passed into the posterior urethra.
Simultaneous retrograde urethroscopy was used to explore the site of urethral trauma. The glide wire passed through the suprapubically placed open-ended ureteric catheter was pushed down to the site of urethral trauma. This glide wire was easily identified during the retrograde urethroscopy, grasped with cystoscopic grasping forceps and pulled out through the external meatus. The 6 F open-ended ureteric catheter was introduced percutaneously over the glide wire till it appeared at the meatus. Later the glide wire was exchanged with super stiff zebra wire. This zebra wire then acted as a guide for 16 F Foley catheter, which was placed perurethrally in the bladder. If the supra pubic tract was dilated, an 18 F Foley catheter was placed suprapubically as a safety measure.
Postoperative course and follow-up
The suprapubic catheter, if placed was removed on first postoperative day before the patient was discharged from hospital. The urethral catheter was left in place for 1 to 3 weeks, depending on severity of injury. A dynamic urethrogram was performed by the side of catheter to confirm absence of extravasation of contrast before catheter removal. Patients were followed up at 1, 3, 6 and 12-month interval and than yearly. They were monitored with history, physical examination, uroflow rate measurement and ultrasonography guided determination of post void residual urine volume. Development of obstructive urinary symptoms, decrease in flow rate and/or increase in post-void residual urine volume was evaluated by retrograde urethography or Cystoscopy.
Results
From May 1997 to May 2003, seven men with age ranging from 36 years to 68 years (mean age 52 years) were diagnosed at our institute with complete iatrogenic anterior urethral injury. Three of these patients had the balloon of a Foley catheter inflated in the bulbar urethra and were referred with catheter in situ draining blood. Three patients were known case of urethral stricture and underwent urethral dilatation followed by bleeding per urethra. One patient had undergone optical urethrotomy for a 1 cm bulbar urethral stricture. However during surgery, glide wire accidentally came out after partial incision of the stricture and later on proximal urethra could not be identified. The patient was later referred to other institute for further management. The detailed demographic data, cause of urethral trauma, primary urethral pathology, duration between injury and endoscopic realignment, operation time, operative procedure required, duration of hospital stay, duration of perurethral catheter & followup course of each patient is mentioned in table 1.
Table 1 Patients demographic, surgical & follow-up data.
Sr. No. Age Cause of urethral Trauma Past urethral Pathology Duration between trauma and surgery (Hrs.) OR time (min) Procedure needed for endoscopic alignment SPC needed Hospital Stay (hours) Duration of perurethral catheterization (weeks) Duration of followup (months) Recurrence
1 57 Visual internal urethrotomy Bulbar urethral stricture 8 45 Suprapubic Amplatz Yes 36 3 74 At 6 month
2 48 Urethral dilatation Proximal penile stricture 4 15 Retrograde glide No 14 2 66 Nil
3 68 Catheterization Acute retention after piles surgery 16 40 Suprapubic Amplatz Yes 28 1 58 Nil
4 53 Catheterization Acute urinary retention 4 30 Suprapubic Amplaz Yes 28 1 56 Nil
5 42 Urethral dilatation Stricture Urethra 2 35 Suprapubic Amplatz Yes 36 2 47 At 4 months
6 36 Urethral Dilatation Stricture Urethra 18 30 Suprapubic needle access No 18 3 36 At 6, 11 & 20 months followup. Open end to end Urethroplasty done
7 60 Catheterization Acute retention after bilateral hernia surgery 6 35 Suprapubic Amplatz Yes 24 1 7 Nil
All patients had complete loss of urethral continuity. Bulbar urethra was affected in four patients and the remaining three patients had proximal penile urethral rupture. Glide wire was negotiated retrogradely only in one patients & the remaining six patients needed antegrade (suprapubic) access. Rigid antegrade cystoscopy was needed through suprapubically placed Amplatz sheath in five patients.
Procedure was successful in all the seven patients. Suprapubic catheter kept in five patients was removed on first post-operative day. Per urethral catheter was kept for one week in three patients, two weeks in two patients and for three weeks in remaining two patients. No patient had extravasation of contrast during dynamic retrograde urethrogram performed before catheter removal. All patients were asymptomatic and had insignificant post-void residue on sonography during 1-month follow-up. The peak uroflow rate was 19.7 ml/sec. (range 15–24 ml/sec) & the post-void residual urine was 38 ml (range 10 to 55 ml) on ultrasonography measurement. By six months of follow-up three patients developed stricture at the site of trauma. Of these two patients had injury during dilatation for the stricture urethra. All these three patients were treated by optical urethrotomy and were taught self-calibration of urethra for an average period of 6 months. With the average follow-up of 49.2 months (range 7–74 months) all, except one patient are asymptomatic with no further recurrence of stricture urethra. Only one patient developed multiple recurrences & was offered open end-to-end urethroplasty at twenty months after third recurrence. Hence four of our patients (57.14%) were disease free during their follow-up after their immediate endoscopic realignment for urethral trauma. After an additional optical urethrotomy, urethroplasty could be avoided in six patients (87.2%).
Discussion
Controversy continues regarding proper management of traumatic urethral disruption. The suggested surgical treatment modalities include a) immediate primary simple realignment over a stenting catheter b) immediate primary suture repair c) immediate suprapubic cystostomy alone, with delayed elective urethroplasty for the resulting stricture. Many urologists believe that delayed urethral reconstruction is the safest method [2]. However, placement of suprapubic catheter significantly impairs the patient's quality of life.
Advances in endoscopic instrumentation and techniques have expanded our armamentarium for safe and effective treatment of urethral strictures. In a recent review on management of urethral trauma, it is stated that a stenting urethral catheter may be useful in management of anterior urethral injury in some cases [4]. However the exact role of primary endoscopic realignment in management of complete iatrogenic urethral injury was not specified.
Towler JM et al initially described realignment of traumatic urethra over a splinting catheter by instrumentation and endoscopy via both suprapubic and urethral routes in an effort to decrease patient morbidity [5]. Moudouni SM et al described early endoscopic realignment of posterior traumatic urethral disruption [6]. On follow up of 68 months they concluded that the urethral continuity could be established without any increase in the incidence of impotence, stricture formation or incontinence. In case of failure, endoscopic realignment doesn't compromise the results of secondary urethroplasty. In another study Jepson BR et al employed a variety of endourological techniques to achieve urethral continuity in 8 patients [3]. The average time to realignment was 9.5 days. With a mean follow up of 50.4 months four patients required subsequent internal urethrotomies with eventual voiding stabilisation over the course of 11 months. Olapede J et al used either both antegrade & retrograde cystoscopy (with or without fluoroscopy), or flexible retrograde urethroscopy alone for primary endoscopic realignment of posterior urethral trauma [7]. Similar good results were reported in various studies from different institution [8-11]. Thus the technique of immediate endoscopic realignment of urethral injuries is widely employed for management of posterior urethral injuries.
Since the anterior urethral injuries are rare, accepted treatment methods are commonly based on relatively few patients. Benchekroun A et al on reviewing their 23 patients with anterior urethral trauma recommended delayed end to end anastomosis for total rupture, reserving immediate repair in patients with associated penile fracture [12]. Dobrowlski ZF et al in their series of 255 patients with anterior urethral injury found iatrogenic trauma during catheterisation or cystoscopy to be commonest (206 patients) [13]. The treatment was conservative in 193 patients and surgical in 62. Partial rupture of anterior urethra will heal after urinary diversion alone with excellent results [14]. However the most common complication of complete urethral rupture is development of urethral stricture [15]. The goal of immediate primary realignment lies in the ability to perform the procedure in as timely a fashion as possible with minimal disruption to the already traumatized tissue. Realignment of the urethra will produce more anatomically aligned stricture and obviate the need for long-term suprapubic tube drainage [15]. Londergan TA et al employed early fluoroscopic realignment for traumatic urethral injury [9]. Of 2 patients with anterior urethral trauma, the procedure was successful in one patient. The patient later required visual internal urethrotomy at 35-month follow-up. Keeping the results of primary endoscopic realignment of posterior urethral injuries in mind, Ying-Hao S et al were first to employ technique of urethroscopic realignment for treating 16 patients with ruptured bulbar urethra [1]. They felt that since the gap in the anterior urethra after disruption is relatively short and the proximal end of disrupted anterior urethra is more immobile, a combined suprapubic intervention was not required to re-establish anterior urethral continuity. However only 4 patients in their series had complete injury. They found procedure feasible in all these patients with only 1 patient requiring intermittent self-dilatation once weekly. They treated all patients on outpatient basis under local anaesthesia. Based on their results, they recommended primary endoscopic urethral reunion as an ideal, cost effective treatment in treating male patients with bulbar urethral disruption. Nakajima et al employed a thin trocar needle for endoscopic realignment of complete disruption of bulbar urethra resulting from a straddle injury [16]. They confirmed exact suprapubic trocar position with flexible antegrade cystoscopy and placed a guide wire in proximal urethra and found the procedure safe, simple and reproducible. Ku JH et al supplemented above results by comparing immediate versus delayed treatment [15]. Of 95 patients studied, 46 had complete disruption. Primary endoscopic realignment was successful in 65 patients. The remaining patients underwent delayed repair. In their experience, the incidence of urethral stricture after complete disruption was statistically low in patients undergoing immediate realignment as compared with those patients treated with delayed repair (31.3% vs. 68.8%). They suggested that better outcomes could be obtained when immediate urethral realignment is successful in patients with bulbar urethral disruption.
In view of encouraging results of primary endoscopic realignment of posterior urethral disruption, we employed the similar technique for treating our patients with iatrogenic anterior urethral injuries from May 1997 onwards. Our results are encouraging and except for three patients who developed stricture at 6 months follow-up, rest all patients had a patent urethra without any urinary symptoms. We believe that the development of stricture in these three patients was attributable to presence of urethral stricture before urethral trauma rather than the long-term result of immediate endoscopic realignment that they underwent for urethral trauma. Two out of these three patients had documented stricture for which they had undergone dilatation in past. During dilatation, trauma resulted causing urethral disruption. Patients who developed stricture during follow-up were easily managed by optical urethrotomy, avoiding the need for open urethroplasty &/or long-term supra-pubic urinary diversion. Based on our experience we feel that immediate endoscopic realignment of iatrogenic anterior urethral disruption is feasible, safe and effective. The major drawback of our study is small number of patients and lack of comparison with patients who were managed by delayed method. A prospective randomised trial with larger number of patient is necessary to evaluate definitive role of immediate endoscopic realignment in management of complete anterior urethral injuries.
Conclusion
Immediate endoscopic realignment of complete iatrogenic urethral disruption is a feasible, safe and effective treatment modality for management of these patients.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
HS participated in the design of the study, performed the review of records, data acquisition, statistical analysis and helped to draft the manuscript. PM revised the manuscript critically for important intellectual content and finally approved the manuscript. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
none
==== Refs
Ying-Hao S Chuan-Liang X Xu G Guo-Qiang L Jian-Guo H Urethroscopic realignment of ruptured bulbar urethra J Urol 2000 164 1543 5 11025700 10.1097/00005392-200011000-00019
Tune MH Tefekli AH Kaplancan T Esen T Delayed repair of posttraumatic posterior urethral distraction injuries: long term results Urology 2000 55 837 41 10840087 10.1016/S0090-4295(00)00464-7
Jepson BR Boullier JA Moore RG Parra RO Traumatic posterior urethral injury and early primary endoscopic realignment: evaluation of long-term follow-up Urology 2000 53 1205 1210 10367853 10.1016/S0090-4295(99)00003-5
Chapple C Barbagli G Jordan G Mundy AR Netto NR Pansadoros V McAninch JW Consensus statement on urethral trauma BJU Int 2004 93 1195 1202 15180604 10.1111/j.1464-410x.2004.04805.x
Towler JM Eisen SM A new technique for the management of urethral injuries Br J Urol 1987 60 162 6 3664205
Moudouni SM Patard JJ Manunta A Guiraud P Lobel B Guillé F Early endoscopic realignment of post-traumatic posterior urethral disruption Urology 2001 57 628 632 11306365 10.1016/S0090-4295(00)01068-2
Olapade-Olaopa EO Adebayo SA Atalabi OM Popoola AA Ogunmodede IA Enabulele UF Rigid retrograde endoscopy under regional aneasthesia: a novel technique for the early realignment of traumatic posterior urethral disruption Afr J Med Med Sci 2002 31 277 80 12751573
Guille F Cipolla B Leveque JM Guirassy S Olivo JF Lobel B Early endoscopic realignment of complete traumatic rupture of the posterior urethra Br J Urol 1991 68 178 80 1884146
Londergan TA Gundersen LH van Every MJ Early fluoroscopic realignment for traumatic urethral injuries Urology 1997 49 101 3 9000194 10.1016/S0090-4295(96)00429-3
Kielb SJ Voeltz ZL Wolf JS Evaluation and management of traumatic posterior urethral disruption with flexible cystourethroscopy J Trauma 2001 50 36 40 11253761
Gheiler EL Frontera RJ Immediate primary realignment of prostato-membranous urethral disruptions using endourologic techniques Urology 1997 49 596 99 9111631 10.1016/S0090-4295(97)80002-7
Benchekroun A Alami M Ghadouane M Zanoud M Nouini Y Benslimane L Belahnech Z Faik M Anterior urethral injury Report of 23 cases Annals d'Urologie 2000 36 150 53
Dobrowolski ZF Weglarz W Jakubik P Lipczynski W Dobrowolska B Treatment of posterior and anterior urethral trauma BJU Int 2002 89 752 4 11966638 10.1046/j.1464-410X.2002.02719.x
Pontes JE Pierce JM Anterior urethral injuries: four years of experience at the Detroit General Hospital J Urol 1978 120 563 4 568672
Ku JH Kim ME Jeon YS Lee NK Park YH Management of bulbous urethral disruption by blunt external trauma: the sooner, the better? Urology 2002 60 579 83 12385910 10.1016/S0090-4295(02)01834-4
Nakajima K Dejuchi M Ishii N Kawakami T Nozawa E Hara H Miura K Ishii N Endoscopic management of a traumatic disruption of the bulbous urethra using a thin trocar Int J Urol 2001 8 202 4 11260357 10.1046/j.1442-2042.2001.00284.x
|
16281970
|
PMC1310535
|
CC BY
|
2021-01-04 16:30:02
|
no
|
BMC Urol. 2005 Nov 9; 5:13
|
utf-8
|
BMC Urol
| 2,005 |
10.1186/1471-2490-5-13
|
oa_comm
|
==== Front
BMC Med Res MethodolBMC Medical Research Methodology1471-2288BioMed Central London 1471-2288-5-351626908110.1186/1471-2288-5-35Research ArticleParametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data Vickers Andrew J [email protected] Integrative Medicine Service, Biostatistics Service, Memorial Sloan Kettering Cancer Center, Howard 1312a, 1275 York Avenue, NY, NY 10021, USA2005 3 11 2005 5 35 35 22 3 2005 3 11 2005 Copyright © 2005 Vickers; licensee BioMed Central Ltd.2005Vickers; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
It has generally been argued that parametric statistics should not be applied to data with non-normal distributions. Empirical research has demonstrated that Mann-Whitney generally has greater power than the t-test unless data are sampled from the normal. In the case of randomized trials, we are typically interested in how an endpoint, such as blood pressure or pain, changes following treatment. Such trials should be analyzed using ANCOVA, rather than t-test. The objectives of this study were: a) to compare the relative power of Mann-Whitney and ANCOVA; b) to determine whether ANCOVA provides an unbiased estimate for the difference between groups; c) to investigate the distribution of change scores between repeat assessments of a non-normally distributed variable.
Methods
Polynomials were developed to simulate five archetypal non-normal distributions for baseline and post-treatment scores in a randomized trial. Simulation studies compared the power of Mann-Whitney and ANCOVA for analyzing each distribution, varying sample size, correlation and type of treatment effect (ratio or shift).
Results
Change between skewed baseline and post-treatment data tended towards a normal distribution. ANCOVA was generally superior to Mann-Whitney in most situations, especially where log-transformed data were entered into the model. The estimate of the treatment effect from ANCOVA was not importantly biased.
Conclusion
ANCOVA is the preferred method of analyzing randomized trials with baseline and post-treatment measures. In certain extreme cases, ANCOVA is less powerful than Mann-Whitney. Notably, in these cases, the estimate of treatment effect provided by ANCOVA is of questionable interpretability.
==== Body
Background
Introductory statistics textbooks typically advise against the use of parametric methods, such as the t-test, for the analysis of randomized trials unless data approximate to a normal distribution. Altman, for example, states that "parametric methods require the observations within each group to have an approximately Normal distribution ... if the raw data do not satisfy these conditions ... a non-parametric method should be used" [1]. In some cases, central limit theorem is invoked such that parametric methods are said to be applicable if sample size is suitably large: "for reasonably large samples (say, 30 or more observations in each sample) ... the t-test may be computed on almost any set of continuous data" [2].
The rationale for recommending non-parametric over parametric methods, unless certain conditions are met, is rarely made explicit. But techniques for statistical inference from randomized trials can only fail in one of two ways: they can inappropriately reject the null hypothesis of no difference between groups (false positive or Type I error) or inappropriately fail to reject the null (false negative or Type II error). Hence any recommendation to favor one technique over another must be based on the relative rates of these two errors.
Empirical statistical research has clearly demonstrated that the t-test does not inflate Type I (false positive) error. In a typical study, Heeren et al examined the properties of the t-test to analyze small two-group trials where data are ordinal, such as from a five point scale, and thus non-normal [3]. They found that where there was truly no difference between groups, the t-test would reject the null hypothesis close to 5% of the time.
Thus concern over the relative advantages of parametric and non-parametric methods has focussed on Type II error [4]. Typically, researchers have created a large number of data sets, in which observations were created from a distribution incorporating a difference between groups. Each data set is then analyzed by both parametric and non-parametric methods in order to calculate the proportion of times the null hypothesis is rejected (that is, the power) [5-7].
The results have been fairly consistent. Where data are sampled from a normal distribution, the t-test has very slightly higher power than Mann-Whitney, the non-parametric alternative. However, when data are sampled from any one of a variety of non-normal distributions, Mann-Whitney is superior, often by a large amount. Bridge and Sawilowsky, for example, concluded that" "the t-test was more powerful only under a distribution that was relatively symmetric, although the magnitude of the differences was trivial. In contrast, the [Mann-Whitney] held huge power advantages for data sets which presented skewness" [7]. Many workers have linked results showing the superiority of non-parametric methods for non-normal distributions to claims that data rarely follow a normal distribution (as Micceri puts it: "The unicorn, the normal curve and other improbable creatures" [8]). This has led to implicit recommendations that non-parametric techniques should be considered the method of choice [7].
It is arguable, however, that these prior investigations are flawed. The t-test and Mann-Whitney are used for continuous variables such as blood pressure, depression, weight or pain. Most commonly, we are interested in seeing how these variables change following an intervention. This reflects clinical practice where the patient presents with a problem and asks the doctor to help improve it. In a typical study, a patient with hypertension, obesity or chronic headache is randomized to drug or placebo to see whether the drug is effective for reducing blood pressure, weight or pain. The researchers might report that, say, blood pressure fell by 5 mm in the placebo group but by 14 mm in the drug group. Indeed, trials in which we are interested only in post-treatment scores, and where change is not of interest, are rather rare, being primarily confined to iatrogenic symptoms such as post-operative pain or chemotherapy vomiting.
There are two implications for methodologic research on the relative value of parametric and non-parametric techniques. First, we should worry about the distribution of change scores. It seems likely that change from baseline would approximate more closely to a normal distribution than the post-treatment score. This is because change scores are a linear combination and the Central Limit Theorem therefore applies. As a simple example, imagine that baseline and post-treatment score were represented by a single throw of a die. The post-treatment score has a flat (uniform) distribution, with each possible value having an equal probability (figure 1a). The change score has a more normal distribution: there is a peak in the middle at zero – the chance of a zero change score is the same as the chance of throwing the same number twice, that is 1 in 6 – with more rare events at the extremes – there is only a 1 in 18 chance of increasing or decreasing score by 5 (Figure 1b).
Figure 1 Distribution of scores for a single die roll and the difference between two die rolls. The change score tends towards a more normal distribution.
Moreover, where an endpoint is measured at baseline and again at follow-up, the t-test is not the recommended parametric method. Analysis of covariance (ANCOVA), where baseline score is added as a covariate in a linear regression, has been shown to be more powerful than the t-test [9-11]. It has several additional advantages: it adjusts for any chance baseline imbalances; it can be extended to incorporate randomization strata as co-variates, which has been shown to increase power [12]; it can also be extended to incorporate time effects where measures are repeated.
In this paper, I report results from a study making the more rational comparison between parametric and non-parametric methods: ANCOVA and Mann-Whitney. Such a comparison does not appear to have been reported previously. I aimed to compare relative power of the two methods under a variety of distributions. As a secondary objective, I aimed to determine whether ANCOVA provided an unbiased estimate for the difference between groups where data did not follow a normal distribution. A third, overarching aim was to investigate the distribution of change scores between repeat assessments of a non-normally distributed variable.
Methods
The starting point for this study was to obtain archetypal data sets for analysis. I will follow Bridge [7] in choosing empirical rather than theoretical distributions. I examined the distribution of a large number of empirical data sets and cross-referenced these with those described by Micceri, who systematically obtained 440 data sets from the psychological and educational domains [8]. The most common distribution appeared one with moderate positive skew. As an exemplar, I used a headache severity index from a large (n = 401) randomized trial of headache prophylaxis [13] (Figure 2). This distribution was also used with scores reversed, to create a distribution with moderate negative skew. A second pain data set, this time from a trial on athletes with shoulder pain [14], provides an example of a more uniform distribution (Figure 3). Data on Ki67, an antigen that is a marker for cell proliferation, were obtained from a randomized comparison of two hormonal treatments for breast cancer [15]. The distribution for Ki67 is comparable to Micceri's "extreme asymmetry distribution" (Figure 4). For extreme negative skew, I used data from the physical functioning scale of the SF36 (Figure 5), again taken from the headache trial. As a comparison group, data were also drawn from a normal distribution with a mean of 5 and a standard deviation of 1.
Figure 2 Distribution of post-treatment and change scores from original and simulated data for headache severity ("moderate positive skew" distribution).
Figure 3 Distribution of post-treatment and change scores from original and simulated data for shoulder pain ("uniform" distribution).
Figure 4 Distribution of post-treatment and change scores from original and simulated data for Ki67, a biomarker of cell proliferation ("extreme asymmetry" distribution).
Figure 5 Distribution of post-treatment and change scores from original and simulated data for physical functioning scale of the SF36 ("extreme negative skew" distribution).
For each of the distributions, I created a polynomial that converted normal data to a distribution with an approximately similar shape. For example, the distribution with moderate positive skew in Figure 2 was simulated by sampling x from the normal and creating a new variable equal to 14.8+16.5x+7.5x2-1.15x3, rounded, like the original scale, to the nearest 0.25. The simulation distributions were compared to the empirical distributions by visual inspection and comparison of the standard deviation, skewness and kurtosis.
To run the simulations, a bivariate normal (mean 0, standard deviation 1) with a specified correlation was created for a trial of a given sample size equally divided in two groups. The polynomial was applied and a treatment effect introduced. The treatment effect was one of two forms: a shift, for example, scores in the treatment group were reduced by two points; and a ratio, for example, treatment group scores were reduced by 20%. Results were then analyzed by Mann-Whitney and ANCOVA, with p-values obtained by asymptotic approximation for the Mann-Whitney test. In some simulations, t-tests and ANCOVA of log-transformed data were applied. The t-test and Mann-Whitney used the follow-up score if correlation was less than 0.5 and the change score otherwise. This maximizes the power of these tests [11] and might be seen as favoring unadjusted tests on the grounds that the correlation between baseline and follow-up scores is not known when the protocol for statistical analysis is written. Note that the correlation cited in the results is the correlation between baseline and follow-up in the control group. Some previous workers have used the overall correlation using both groups when investigating the properties of ANCOVA [11]. The difference between these two values was small in the context of our simulations, for example, a correlation of 0.5 in the control group was equivalent to a correlation of 0.476 for both groups combined.
Simulations were repeated 1000 times for each combination of sample size (10, 20, 30, 40, 60, 100, 200, 400, 800) and correlation (0.1, 0.2, 0.3 ... 0.9) using Stata 8.2 (Stata Corp., College Station, Texas). The exception was extreme asymmetry data for the Ki67 biomarker. The baseline and post-treatment distributions had quite different shapes and different polynomials were used to model each. This constrained the range of possible correlations, hence only the empirical correlation observed in the original study was used, 0.4, with 5000 iterations.
Results were compared between different methods using the "relative efficiency" (RE) measure. This gives the relative number of patients required for a study analyzed using parametric methods so that power was equivalent to the non-parametric alternative. Hence an RE of 1.25 indicates that a particular trial analyzed by parametric statistics would have to accrue 25% more patients than if it were to be analyzed non-parametrically; an AE of 0.80 would indicate that the parametric method was superior by an equivalent amount. The RE is calculated from observed power of the tests, that is, the proportion of simulations in which the p-value was less than the α of 5%. Where (1-βnp) and (1-βp) are the observed powers from the simulations for the non-parametric and parametric test respectively, RE is given by the formula:
Note that, although it is arguable that the null hypotheses for different tests, say the t-test and Mann-Whitney, are technically different, the conclusions drawn by investigators of a randomized trial given a particular p-value will be the same, regardless of the analytic method used. Hence direct comparison of the power of different tests is justified in this setting.
Results
The figures show the distributions of post-treatment and change scores from the original data and associated simulations. Visual comparison of subfigures (a) with (b), and (c) with (d), suggests that the polynomials used for the simulations produce distributions that are reasonably similar to the related empirical distribution. Comparing subfigures (a) to (c), and (b) with (d), it is apparent that, as hypothesized, the change between baseline and follow-up scores tends towards the normal distribution. These visual impressions are confirmed in Table 1, which shows estimates of the shape parameters for the distributions. The shape parameters for the empirical and simulated data are similar, and skewness is much closer to zero for the change score compared to the follow-up score.
Table 1 Shape parameters for the distributions produced by the simulations compared to those from the original empirical data. Parameters for the moderate negative skew are as for the moderate positive skew, except that the sign for skew is reversed.
Distribution Post-treatment scores
Standard deviation Skewness Kurtosis
Empirical Simulation Empirical Simulation Empirical Simulation
Moderate positive skew 17.01 17.02 1.62 1.63 5.71 6.00
Uniform 13.49 13.49 -0.11 -0.11 2.08 2.01
Extreme asymmetry 8.96 9.78 3.03 2.90 13.77 13.97
Extreme negative skew 22.17 21.88 -1.74 -1.79 5.88 5.85
Change scores
Standard deviation Skewness Kurtosis
Empirical Simulation Empirical Simulation Empirical Simulation
Moderate positive skew 10.40 18.80 0.35 0.01 4.49 5.15
Uniform 12.31 14.56 0.00 0.43 3.05 2.98
Extreme asymmetry 10.62 10.97 0.88 1.08 5.38 6.37
Extreme negative skew 15.51 15.02 0.00 0.75 6.31 8.52
As a second check on the simulations, Table 2 compares the power of t-test and Mann-Whitney. The data for post-treatment scores were obtained by combining all data from simulations where correlation was less than 0.5; the change scores were from data where correlation was 0.5 or more. These results broadly replicate those of previous workers and therefore provide support for the methods of the current study. In particular, the increase in relative efficiency of the t-test under normality (or uniform) is trivial compared to its loss in relative power under asymmetry. Two aspects of Table 2 have not been reported previously. First, RE can vary depending on whether the treatment effect is a shift or a ratio change. Second, the power of Mann-Whitney and t-test are more similar (RE closer to 1) for change scores, presumably because change scores are more normally distributed. An exception is for extreme asymmetry, where Mann-Whitney has extremely poor power for change scores.
Table 2 Relative power of t-test and Mann-Whitney given as relative efficiency. Values less than 1 indicate greater power of t-test; greater than 1 indicates superiority of Mann-Whitney. Results are combined across sample sizes and correlations.
Distribution Post-treatment scores Change scores
Moderate positive skew: shift 0.9348 0.9835
Moderate positive skew: ratio 1.1382 1.0436
Moderate negative skew: shift 1.1833 1.0187
Moderate negative skew: ratio 0.9301 0.9825
Uniform: shift 0.9339 0.9846
Uniform: ratio 0.9488 0.9929
Extreme negative skew: shift 1.3769 1.1140
Extreme negative skew: ratio 1.6675 1.2046
Extreme asymmetry: shift 7.1461 0.5370
Extreme asymmetry: ratio 9.0091 0.6432
Normal: shift 0.9660 0.9726
Normal: ratio 0.9740 0.9760
Table 3 gives RE for each combination of sample size and correlation for the moderate positive skew data, where the treatment effect was a shift. ANCOVA is generally superior to Mann-Whitney. Smaller sample sizes and correlations near the extremes reduce the advantage of ANCOVA. Table 4 shows the RE for each of the different distributions combining data for correlations between 0.4 and 0.7, which constitutes a typical range for correlations described in the literature [16]. Mann-Whitney is superior for some very small sample sizes, but RE is non-trivially larger than 1 across sample sizes only for the extreme negative skew distribution with a ratio treatment effect. In table 5, data are given by correlation, combining sample sizes. The table has one particularly notable feature: for some distributions, RE's drop dramatically between correlation of 0.4 and 0.5. This is apparently because the endpoint analyzed changed from the post-treatment score to the change score at correlations of 0.5 and above. This was to maximize power following previous work on the power of unadjusted tests based on the normal [9,11]. As it seems possible that the relative power of analyzing change and post-treatment scores may differ between the normal and asymmetric case, the data were reanalyzed using post-treatment scores only (see Table 6). In the case of extreme negative skew, the simulation was repeated with ANCOVA on log-transformed data. Cleary, analyzing only post-treatment score, irrespective of correlation, improves the efficiency of Mann-Whitney considerably, but it is still inefficient compared to log-transformed ANCOVA. That said, log-transformed ANCOVA is slightly anti-conservative: when the simulation was repeated with no treatment effect, the null hypothesis was rejected for 5.23% (rather than the nominal 5%) of trials.
Table 3 Relative efficiency of ANCOVA and Mann-Whitney for the moderate positive skew data. Values less than 1 indicate greater power of ANCOVA; greater than 1 indicates superiority of Mann-Whitney. In blank cells, the power of one or both tests was 100%.
Sample size Correlation between baseline and post-treatment score
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
10 0.8125 0.9773 0.6017 0.9403 0.9722 0.8834 1.2129 1.0851 1.1320
20 1.0160 1.1093 0.8172 0.9348 0.7742 0.9905 0.8124 1.0058 1.0231
30 1.0000 0.9167 0.7785 0.8617 0.8027 0.7756 0.8689 0.8760 1.0096
40 0.8866 0.8596 0.8120 0.7332 0.7365 0.7556 0.9172 0.9067 0.929
60 0.8925 0.8996 0.7752 0.7632 0.7418 0.8277 0.8728 0.8841 0.9892
100 0.8822 0.8594 0.7816 0.7259 0.7071 0.8277 0.8639 0.8702 0.9259
200 0.8484 0.8030 0.7611 0.6920 0.6979 0.7591 0.8793 0.8888 -
400 0.8512 0.8292 0.7392 0.7113 0.6707 0.8029 0.8336 - -
800 0.8781 0.9087 - - - - - - -
Table 4 Relative efficiency of ANCOVA and Mann-Whitney combining correlations 0.4 – 0.7. Values less than 1 indicate greater power of ANCOVA; greater than 1 indicates superiority of Mann-Whitney. In blank cells, the power of one or both tests was 100%.
Distribution Sample size
10 20 30 40 60 100 200 400 800
Moderate positive skew: shift 1.0221 0.8751 0.8292 0.8004 0.8085 0.7887 0.7549 0.7497 -
Moderate positive skew: ratio 1.5001 0.9832 1.0161 0.8441 0.7973 0.8079 0.7755 0.7689 0.8389
Moderate negative skew: shift 1.0045 0.9793 0.8080 0.8300 0.8088 0.7810 0.7772 0.7494 0.7404
Moderate negative skew: ratio 1.7878 1.3025 1.1354 1.0737 0.9367 0.8763 0.8949 0.8766 0.8612
Uniform: shift 0.8611 0.8162 0.8360 0.7854 0.7787 0.7938 0.7560 0.7404 0.8137
Uniform: ratio 0.8285 0.8462 0.7789 0.7685 0.7759 0.7799 0.7401 0.7747 -
Extreme negative skew: shift 1.2952 1.0213 0.9250 0.9802 1.0610 1.0431 1.0477 1.0479 -
Extreme negative skew: ratio 1.5027 1.1288 1.1332 1.2639 1.3322 1.3808 1.3442 1.2769 -
Normal: shift 0.9601 1.0049 0.8356 0.7336 0.7850 0.7865 0.7797 0.7560 0.7516
Normal: ratio 0.8230 0.9140 0.8959 0.8114 0.7702 0.8166 0.7855 0.7961 0.7781
Table 5 Relative efficiency of ANCOVA and Mann-Whitney combining all sample sizes. Values less than 1 indicate greater power of ANCOVA; greater than 1 indicates superiority of Mann-Whitney.
Distribution Correlation between baseline and post-treatment score
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Moderate positive skew: shift 0.9342 0.9257 0.8707 0.8532 0.8569 0.9030 0.9447 0.9538 0.9963
Moderate positive skew: ratio 1.1560 1.1300 1.1074 1.0800 0.8170 0.8656 0.9065 0.9231 0.9720
Moderate negative skew: shift 0.9392 0.9058 0.8837 0.8444 0.8799 0.9114 0.9421 0.9593 0.9885
Moderate negative skew: ratio 1.1766 1.1946 1.1467 1.0980 0.8851 0.9319 0.9671 1.0241 1.0390
Uniform: shift 0.9492 0.9170 0.8869 0.8480 0.8586 0.8977 0.9281 0.9723 1.0021
Uniform: ratio 0.9575 0.9384 0.8862 0.8499 0.8604 0.8849 0.9213 0.9467 0.9895
Extreme negative skew: shift 1.4019 1.3718 1.3339 1.3045 0.8987 0.9425 0.9475 0.9405 0.9345
Extreme negative skew: ratio 1.6914 1.6185 1.6101 1.5835 1.0153 1.0367 1.0247 1.0344 1.0062
Normal: shift 0.9700 0.9393 0.9142 0.8445 0.8372 0.8840 0.9065 0.9485 0.9734
Normal: ratio 0.9551 0.9506 0.9126 0.8791 0.8536 0.8859 0.9072 0.9293 0.9654
Table 7 compares the power of Mann-Whitney to ANCOVA on raw and log-transformed data for the distribution with extreme asymmetry. For this distribution, the non-parametric test is generally superior, though there is no simple relationship to sample size. Again, non-parametric analysis of change scores is dramatically less efficient that use of post-treatment scores. To check these data, the methods were used on the original data (n = 185). The p-values for Mann-Whitney on post-treatment scores, Mann-Whitney on change scores, ANCOVA on raw scores and ANCOVA on log-transformed scores were, respectively: 0.0001, 0.672, 0.216 and 0.0003.
Table 7 Relative efficiency of ANCOVA and Mann-Whitney for the extreme asymmetry distribution. Values less than 1 indicate greater power of ANCOVA; greater than 1 indicates superiority of Mann-Whitney. In blank cells, the power of one or both tests was 100%.
Sample size Post-treatment score Change score
ANCOVA v. Mann-Whitney log ANCOVA v. Mann-Whitney ANCOVA v. Mann-Whitney log ANCOVA v. Mann-Whitney
Shift Ratio Shift Ratio Shift Ratio Shift Ratio
10 3.0404 4.1864 0.8862 1.1179 1.2586 1.9736 0.5567 0.539
20 1.2037 2.6045 0.8073 1.2589 0.5480 0.7372 0.3473 0.2983
30 1.1503 1.7717 0.9409 1.2707 0.3084 0.3860 0.2472 0.2701
40 1.1730 1.4786 1.0233 1.4105 0.2643 0.2772 0.2446 0.2421
60 1.1853 1.2015 1.1118 1.4062 0.2115 0.2121 0.1898 0.2586
100 1.2682 1.0648 1.1842 1.4789 0.2224 0.1753 0.2065 0.2545
200 1.2880 0.9257 1.2570 1.5437 0.2078 0.1496 0.1985 0.2544
400 1.3576 0.9089 1.3112 1.5308 0.1961 0.1358 0.1816 0.2418
800 1.4222 - 1.4116 - 0.2038 - 0.1783 0.2444
Table 8 compares the estimates of treatment effects from ANCOVA with the parameter used to specify the treatment effect. For the distributions with extreme skew, the simulations were repeated without truncation, that is, ignoring maximum and minimum scores. ANCOVA appears to be unbiased where the treatment effect is a shift. Where the treatment effect is a ratio, the estimate given by ANCOVA is effectively the shift expected by a patient with the mean baseline score. The size of the bias under ratio change does not seem to be large and could be adjusted for by incorporating a term for baseline score by treatment interaction.
Table 8 Ratio of ANCOVA estimate of treatment effect to true treatment effect.
Moderate positive skew: shift 0.9955
Moderate positive skew: ratio 0.9890
Moderate negative skew: shift 1.0003
Moderate negative skew: ratio 0.9823
Uniform: shift 1.0028
Uniform: ratio 1.0005
Extreme asymmetry: shift 1.0067
Extreme asymmetry: ratio 0.9122
Extreme negative skew: shift 0.9973
Extreme negative skew: ratio 1.0055
Discussion
This study complements previous work on the relative power of parametric and non-parametric statistics by examining the common situation where an outcome is measured before and after a randomly assigned treatment. The study also appears to be novel in its incorporation of different types of treatment effect: shift and ratio.
The immediate conclusions challenge the conventional wisdom of the textbooks. There is no simple and obvious manner in which non-parametric methods becomes superior once the distribution of data shifts away from normal. It is true that under normality parametric methods are trivially more efficient. But for non-normal data, the relative power of parametric and non-parametric statistics varies from distribution to distribution and depends on whether the size of the treatment effect depends on baseline score (i.e. a ratio effect). Moreover, there is no simple relationship between relative power and sample size and no clear rationale for the frequently cited threshold of 30 – 50 patients per group indicating acceptability of parametric statistics.
In general, ANCOVA outperformed Mann-Whitney for most distributions under most circumstances. This is heartening because ANCOVA has a major advantage over any non-parametric method: it provides an estimate for the size of the difference between group, that is, an effect size. Clinicians and patients generally want to know not just whether a treatment helps, but how much it helps, so they can determine whether it is worth the time, effort, risks and expense. The CONSORT group, which issues recommendations on the reporting of randomized trials, has stated that the results of a trial should stated as "a summary of results for each group, and the estimated effect size and its precision (e.g., a 95% confidence interval)". They go on to state that "although p-values may be provided ... results should not be re ported solely as p-values" [17]. ANCOVA directly provides the effect size, which appears to be unbiased; Mann-Whitney only the p-value. It is true that an estimate, such as a difference between medians with associated confidence interval, can be calculated separately from the Mann-Whitney and reported alongside the p-value. Nonetheless, the need to use separate techniques for estimation and inference must be seen as a disadvantage. Moreover, the parametric methods are also often to be preferred because estimates using medians may have little relevance for decision making. A good example comes from health economics [18]: we want to know the difference between the mean costs of two treatments because multiplying this difference by the number of patients we expect to treat gives us the expected financial impact of choosing one treatment over the other; the difference in median costs has no practical application.
Accordingly, in apparent distinction to much of the prior methodologic literature, ANCOVA should be the method of choice for analyzing randomized trials with baseline measures. Not only does it do something essential, provide an estimate, that Mann-Whitney cannot, but it appears more powerful in most circumstances. The exception is instructive: Mann-Whitney consistently outperformed ANCOVA only for a data set with extreme skew obtained from a biomarker study. Yet with such extreme skew, the estimate provided by ANCOVA – the average reduction in the biomarker – is of questionable interpretability. Rather than conclude that treatment lead to a 1.5 point drop in Ki67, it seems more appropriate to say that 32% of patients in the treatment group had zero Ki67 at follow-up compared to 14% of controls. In other words, there appears to be a link between the power of ANCOVA and the usefulness of the estimate it provides.
It should be remembered that the relative advantage of ANCOVA is primarily restricted to analysis of randomized trials. It has been argued [19] that ANCOVA with baseline scores should not be used for non-randomized trials on the grounds where baseline scores are not expected to be equivalent. For example, in measuring how anxiety of adolescent boys and girls changes after a stimulus, use of ANCOVA would address the question: "What would be the difference in changes between boys and girls given an equivalent baseline score?". Yet we would not anticipate that baseline anxiety levels of boys and girls would be the same.
This paper has not examined lumpy or multimodal distributions [8]. Yet given that the relative power of parametric methods seems primarily affected by asymmetry – compare the normal and uniform with the skewed distributions – the results cited here should apply to such distributions. This paper also did not examine semi-parametric methods, such as ANCOVA on ranks. There is some evidence that these methods are preferable to fully parametric alternatives for skewed distributions [20] and there remains the possibility of using standard ANCOVA for obtaining estimates of treatment effects and the semi-parametric test for inference.
Table 6 Relative efficiency of ANCOVA and Mann-Whitney combining all sample sizes. Mann-Whitney is always analyzed using the post-treatment score. Values less than 1 indicate greater power of ANCOVA; greater than 1 indicates superiority of Mann-Whitney.
Distribution Correlation between baseline and post-treatment score
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Moderate postive skew: shift 0.9397 0.9184 0.8831 0.8433 0.8012 0.7384 0.6701 0.5627 0.4214
Moderate postive skew: ratio 1.1430 1.1415 1.1036 1.0849 1.0541 0.9759 0.9090 0.7662 0.5947
Moderate negative skew: shift 0.9293 0.9281 0.8844 0.8326 0.7973 0.7256 0.6631 0.5556 0.4077
Moderate negative skew: ratio 1.1724 1.1871 1.1453 1.1103 1.0568 0.9867 0.9004 0.7665 0.5783
Uniform: shift 0.9324 0.9136 0.8926 0.8609 0.7741 0.7341 0.6483 0.5632 0.4147
Uniform: ratio 0.9475 0.9385 0.8957 0.8497 0.8191 0.7507 0.6659 0.5723 0.4297
Extreme negative skew: shift 1.4043 1.3724 1.3504 1.2859 1.2534 1.1947 1.0992 0.9472 0.7523
Extreme negative skew: ratio 1.6803 1.6120 1.6265 1.5843 1.4941 1.4206 1.2697 1.0932 0.858
Extreme negative skew: shift. ANCOVA log transformed 0.9709 0.9600 0.9638 0.9043 0.8940 0.8633 0.8109 0.7443 0.6680
Extreme negative skew: ratio. ANCOVA log transformed 0.9502 0.9298 0.9317 0.9077 0.8662 0.8282 0.7834 0.7161 0.6408
Normal: shift 0.9712 0.9258 0.9081 0.8618 0.7841 0.7272 0.6423 0.5349 0.3896
Normal: ratio 0.9550 0.9557 0.9183 0.8692 0.8139 0.7652 0.6527 0.5427 0.4131
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
No outside funding was obtained for this study. Original data for the Ki67 study was kindly provided by Dr Matthew Ellis; data for the shoulder pain study was provided by Dr Konrad Streitberger.
==== Refs
Altman DG Practical Statistics for Medical Research London: Chapman and Hall (monograph) 1991.
Jekel JF Katz DL Elmore JG Epidemiology, Biostatistics and Preventive Medicine 2001 Philadelphia, W.B. Saunders Company
Heeren T D'Agostino R Robustness of the two independent samples t-test when applied to ordinal scaled data Stat Med 1987 6 79 90 3576020
Sawilowsky SS Comments on using alternative to normal theory statistics in social and behavioural science. Canadian Psychology 1993 34 432 439
Zimmerman DW Zumbo BD The effect of outliers on the relative power of parametric and nonparametric statistical tests. Perceptual and Motor Skills 1990 71 339 349
Sawilowsky SS Clifford-Blair R A more realistic look at the robustness and Type II error properties of the t test to departures from population normality. Psychological Bulletin 1992 111 352 360 10.1037/0033-2909.111.2.352
Bridge PD Sawilowsky SS Increasing physicians' awareness of the impact of statistics on research outcomes: comparative power of the t-test and and Wilcoxon Rank-Sum test in small samples applied research J Clin Epidemiol 1999 52 229 235 10210240 10.1016/S0895-4356(98)00168-1
Micceri T The unicorn, the normal curve, and other improbable creatures. Psychological Bulletin 1989 105 156 166 10.1037/0033-2909.105.1.156
Senn S Statistical Issues in Drug Development 1997 Chichester, John Wiley
Vickers AJ The use of percentage change from baseline as an outcome in a controlled trial is statistically inefficient: a simulation study BMC Med Res Methodol 2001 1 6 11459516 10.1186/1471-2288-1-6
Frison L Pocock SJ Repeated measures in clinical trials: analysis using mean summary statistics and its implications for design Stat Med 1992 11 1685 1704 1485053
Kalish LA Begg CB Treatment allocation methods in clinical trials: a review Stat Med 1985 4 129 144 3895341
Vickers AJ Rees RW Zollman CE McCarney R Smith CM Ellis N Fisher P Haselen RV Acupuncture for chronic headache in primary care: large, pragmatic, randomised trial BMJ 2004 328 744 15023828 10.1136/bmj.38029.421863.EB
Kleinhenz J Streitberger K Windeler J Gussbacher A Mavridis G Martin E Randomised clinical trial comparing the effects of acupuncture and a newly designed placebo needle in rotator cuff tendinitis Pain 1999 83 235 241 10534595 10.1016/S0304-3959(99)00107-4
Ellis MJ Neoadjuvant endocrine therapy as a drug development strategy Clin Cancer Res 2004 10 391S 395S 14734497 10.1158/1078-0432.CCR-031202
Vickers AJ How many repeated measures in repeated measures designs? Statistical issues for comparative trials BMC Med Res Methodol 2003 3 22 14580266 10.1186/1471-2288-3-22
Moher D Schulz KF Altman DG The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials Ann Intern Med 2001 134 657 662 11304106
Thompson SG Barber JA How should cost data in pragmatic randomised trials be analysed? BMJ 2000 320 1197 1200 10784550 10.1136/bmj.320.7243.1197
Cribbie RA Jamieson J Structural equation models and the regression bias for measuring correlates of change. Educational and Psychological Measurement 2000 60 893 907 10.1177/00131640021970970
Conover WJ Iman RL Analysis of covariance using the rank transformation Biometrics 1982 38 715 724 7171697
|
16269081
|
PMC1310536
|
CC BY
|
2021-01-04 16:32:51
|
no
|
BMC Med Res Methodol. 2005 Nov 3; 5:35
|
utf-8
|
BMC Med Res Methodol
| 2,005 |
10.1186/1471-2288-5-35
|
oa_comm
|
==== Front
BMC GastroenterolBMC Gastroenterology1471-230XBioMed Central London 1471-230X-5-341628197510.1186/1471-230X-5-34Research ArticleDecreased expression of cytochrome P450 protein in non-malignant colonic tissue of patients with colonic adenoma Bergheim Ina [email protected] Christiane [email protected] Alexandr [email protected] Hohenheim University (140), Department Physiology of Nutrition, Stuttgart, Germany2 Department of Pharmacology and Toxicology, University of Louisville Health Sciences Center, Louisville, KY, USA2005 10 11 2005 5 34 34 21 4 2005 10 11 2005 Copyright © 2005 Bergheim et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Cytochrome P450 (CYP) enzymes in epithelial cells lining the alimentary tract play an important role in both the elimination and activation of (pro-)carcinogens. To estimate the role of cytochrome P450 in carcinogenesis of the colon, expression patterns and protein levels of four representative CYPs (CYP2C, CYP2E1, CYP3A4 and CYP3A5) were determined in colon mucosa of normal and adenomatous colonic tissue of patients with adenomas and disease-free controls.
Methods
Expression of CYP2C, CYP2E1, CYP3A4, and CYP3A5 in colon mucosa of normal and adenomatous colonic tissue of patients with adenoma and disease-free controls was determined by RT-PCR. Protein concentration of CYPs was determined using Western blot.
Results
With the exception of CYP3A5, expression of CYP mRNA was similar among groups and tissues (e.g. normal colon mucosa and adenoma). CYP3A5 mRNA expression was significantly higher in adenoma in comparison to normal tissue of patients with adenoma (~48%). When comparing protein concentrations of CYPs measured in adenomas with neighboring normal colonic mucosa no differences were found. However, in normal tissue of patients with adenomas, protein levels of CYP2C8, CYP3A4 and CYP3A5, but not that of CYP2E1, were significantly lower than in biopsies obtained from disease-free controls. Specifically, in normal colonic mucosa of patients protein concentrations of CYP2C8, CYP3A4, and CYP3A5 were ~86%, ~69%, and ~54%, respectively, lower than in disease-free controls.
Conclusion
In conclusion, among other factors, the altered protein levels of certain CYPs (e.g. CYP2C8, CYP3A4 and CYP3A5) in colon mucosa might contribute to the development of neoplasia in the colon.
==== Body
Background
Cytochrome P450 enzymes (CYPs) play a key role in the oxidative, peroxidative and reductive metabolism of endogenous and exogenous compounds such as drugs, dietary components and alcohols (for review see [1].). However, CYPs not only function in the detoxification of xenobiotics but may also be involved in the activation of potential (pro-) carcinogens. Furthermore, it has been suggested that the local expression of CYPs in tumors is very important for the management of cancer since CYPs expressed in tumors may be involved in activation and/or inactivation of chemotherapeutic drugs (for review see [2].). The alimentary tract is exposed to a large variety of xenobiotics, including potential (pro-) carcinogens. Indeed, it has been proposed that extrahepatic tissue might play an important role in the CYP-mediated metabolism of xenobiotic compounds and therefore eventually the susceptibility of certain organs to the development of neoplasia (e.g. colon); however, knowledge on regulation and localization of CYP-mediated xenobiotic metabolism outside the liver is limited.
CYP is most abundant in liver; however, the main CYP families participating in the metabolism of xenobiotics (e.g. CYP1, 2, 3) are also expressed in extrahepatic tissues (for review see [3].). Kaminsky and Fasco [4]. have proposed that members of cytochrome P450 families 3 (e.g. CYP3A) and 2 (e.g. CYP2C), which are present in the small intestinal epithelium at high concentrations [5], protect the small intestine from carcinogenesis. Furthermore, it has been suggested that the absence of some of these microsomal enzymes in the colon may be involved in the comparably high incidence of carcinoma in this organ [6]. However, the availability of data on human colonic cytochrome P450 protein expression in unaffected as well as in adenomatous tissue of the colon is limited. Furthermore, some of the published findings are contradictory. For example, expression of mRNA for CYP3A4 and CYP3A5 has been found in human colorectal epithelium [7]. and in intestinal cell lines [8,9]. CYP2E1 protein has been detected in colon carcinoma cell lines and in the colon [10,11]. but not in human peritumoral or tumoral tissue [12]. Expression of CYP2C7, corresponding to human CYP2C8, and CYP2C (including CYP2C8-CYP2C19) in human and rat colon tissues has been reported [11,13,14].
CYPs might play a critical role not only in the development but also in treatment of colonic neoplasia. However, knowledge on CYPs in colon is limited. Therefore, the aim of the present study was to determine expression and protein concentrations of four representative CYPs (e.g. CYP2C, CYP2E1, CYP3A4, and CYP3A5) in macroscopically normal colonic tissue of subjects with and without colonic adenomas. As the expression of different xenobiotic-metabolizing enzymes may be altered throughout the development of neoplasia, which is generally accepted to follow the adenoma-carcinoma sequence, expression of these cytochrome P450 genes was also measured in colonic adenoma and compared to that of unaffected neighboring colonic tissue of the same subjects.
Methods
Subjects
The study protocol was approved by the Ethics Committee of the medical association Stuttgart, Germany, and followed the ethical guidelines of the 1975 Declaration of Helsinki. Informed consent was obtained from all subjects. Colonic adenoma and surrounding normal colon mucosa biopsies (each 5–10 mg) were obtained during routinely performed coloscopies from 25 patients, aged 50 – 70 years (Table 1). In addition, biopsy specimens of normal colon mucosa were obtained from 27 disease-free controls during routine endoscopy (age 50 – 70 years) displaying no signs of colonic neoplasia or other diseases of the colon. Immediately after removal of biopsies, colon mucosa specimens were frozen in liquid nitrogen and stored at -80°C. All patients completed a questionnaire concerning factors that may influence the expression of cytochrome P450 such as medication, smoking, and alcohol consumption (Table 2). Neither anthropometrics nor medication differed significantly between the patient group and that of disease-free controls. Patients and disease-free controls did not consume drugs known to interfere with/ or to induce cytochrome P450 enzymes investigated in this study.
Tissues and isolation of total RNA and protein
All colon mucosa specimens were immediately frozen in liquid nitrogen after excision and stored at -80°C. Both total RNA and protein were isolated using Trizol reagent (Invitrogene, Gaithersburg, MD, USA).
Immunoblot analysis
Antibodies used for the detection of CYP2E1 and CYP3A4 were a generous gift of Dr. M. Ingelman-Sundberg, Karolinska Institute, Stockholm, Sweden. Primary antibodies for the measurements of CYP2C8 and CYP3A5 were purchased from Chemicon, Inc. (Frankfurt, Germany). Microsomes and Supersomes were from cell lines/clones over-expressing human CYP2C8, CYP2E1, CYP3A4, and CYP3A5, respectively. They were used as internal standards for quantification and to test for cross reactivity of antibodies (Gentest Corporation, Woburn, MA, USA). Protein concentration was determined by the method of Bradford [15], using a commercial preparation (BioRad, Munich, Germany).
Twenty to 30 μg of total protein were separated by SDS-polyacrylamid gel electrophoresis and transferred to nitrocellulose membranes. To enable quantification of CYP protein, a serial dilution of the appropriate standard was proceeded identically. To ensure equal loading of samples, membranes were pre-stained with Ponceau red before blocking and incubation with antibodies. Membranes were blocked in 5% non-fat milk in Tris-buffered saline-Tween 20 (TBST, 0.01% v/v Tween 20) and probed with dilutions of primary antibodies in TBS, followed by an incubation with the secondary antibody. The protein/antibody complex was visualized by enhanced chemiluminescence (SuperSignal® West Dura, Pierce, KTM, Bad Godesberg, Germany). Blots were evaluated (Camera LAS 1000, Fuji, USA) and densitometric analysis was performed using the software AIDA (Raytest, Isotopenmessgeraete, Straubenhardt, Germany). Signal intensities of the samples were adjusted to the intensities of the serially diluted standards. To ensure equal loading some blots were probed for β-actin (Sigma, St Louis, USA).
Reverse transcription and PCR
The integrity and concentration of RNA was analyzed in a 1.2% agarose gel. First-strand complementary DNA was synthesized from 200 ng of total RNA using a First-Strand cDNA Synthesis Kit (Invitrogen, Gaithersburg, MD, USA). Sequences of primers are summarized Table 3. The PCR reaction consisted of 0.6 μl of cDNA, 10 x PCR buffer, 200 μM dNTPs (Boehringer, Mannheim, Germany), BSA (0.25 mg/ml), DMSO (2% v/v), 0.5 μM of specific primer and 0.5 U Taq-polymerase (Promega, Madison, WI, USA), and water to a final volume of 10 μl. For amplifications of the four cytochrome P450 cDNAs, PCR-conditions were as follows: 3 s at 94°C, 3 s at 45°C, 30 s at 72°C, for 32 cycles. Amplification of histone 3.3 was performed applying the following conditions: 3 s at 94°C, 3 s at 45°C, and 30 s at 72°C for 30 cycles. All PCR amplifications were carried out in triplicate in a Rapid Cycler (Idaho Tec., USA) within the linear range of the reaction. PCR products were separated in a 1.5% agarose gel, stained with ethidium bromide and photographed using a digital camera from Biometra (Goettingen, Germany). To ensure the success of PCR, human liver cDNA was used as a positive control.
Statistical analysis
Results are presented as means ± standard error of the mean (SEM) unless otherwise indicated. Fisher's exact test was used to compare lifestyle data. The Mann-Whitney U-test was used for the comparison of relative mRNA concentration and protein levels measured in normal colon mucosa obtained from patients with adenoma and disease-free controls. Wilcoxon's t-test was used for the comparison of relative mRNA expression and protein concentration measured in normal colon mucosa and colonic adenoma of the same subjects. Differences accepted as significant had a significance level (P) of less than 0.05.
Results
Expression of CYP2C, CYP2E1, CYP3A4, and CYP3A5 mRNA in colon of patients with colonic adenoma and disease-free controls
High quality, non-degraded mRNA was obtained from normal tissue of 20 disease-free controls and unaffected tissue as well as adenomas of 18 patients with colonic adenoma. Expression of histone 3.3, which was used as housekeeping gene, was detected in all samples. Figure 1 depicts representative agarose gels of RNA integrity and results of RT-PCR measurements. Results of CYP2C, CYP2E1, CYP3A4, and CYP3A5 mRNA expression are summarized in Figures 1C and 1D. When comparing the relative mRNA concentration of CYP2C, CP2E1, CYP3A4, and CYP3A5 between adenomatous tissue and normal tissue of patients with adenoma, significant differences were only found for the expression of CYP3A5. Specifically, CYP3A5 mRNA expression was significantly higher by ~48% in adenomatous tissue in comparison to normal colonic tissue. No differences were found when comparing CYP expression in normal tissue of patients with adenoma and disease-free controls.
Protein levels of CYP2C8, CYP2E1, CYP3A4, and CP3A5 in colon of patients with colonic adenoma and disease-free controls
Western blot analyses of CYP2E1, CYP3A4, and CYP3A5 were performed with specimens obtained from 25 patients with colonic adenoma and 27 disease-free controls. Since higher protein concentrations were needed for the detection of CYP2C8, protein concentration of CYP2C8 was only determined in biopsies obtained form 19 cases and 12 disease-free controls. No differences were found when comparing mean protein levels of CYP2C8, CYP2E1, CYP3A4, and CYP3A5 measured in normal and neoplastic tissue of patients with adenoma. Representative Western blots and quantitative analysis of blots are depicted in Figure 2.
In addition, protein levels of CYPs were also determined in colon mucosa biopsies of disease-free controls and compared with those determined in macroscopically normal tissue of patients with colonic adenoma. Figure 3 depicts representative Western blots and quantitative analysis of protein.
Protein levels of CYP2E1 in mucosal biopsies obtained from normal colon mucosa of patients with adenoma did not differ significantly from those measured in disease-free controls. In contrast, mean protein level of CYP2C8 was significantly lower in unaffected, macroscopically normal colon mucosa obtained from patients with adenoma than in samples obtained from disease-free controls. Specifically, protein concentration of CYP2C8 in colon mucosa obtained from patients with adenoma was ~85% lower than in tissue obtained from disease-free controls. Similar results were found when comparing protein levels of CYP3A4 and CYP3A5 between patients with adenoma and disease-free controls. CYP3A4 protein concentration was ~69% lower in colon mucosa of patients with adenoma in comparison to disease-free controls. Protein level of CYP3A5 was ~54% lower in biopsies obtained from patients with adenoma when compared to disease-free controls.
Relation of protein levels and mRNA expression pattern of CYPs in normal colonic mucosa of patients with adenoma and disease-free controls
To further investigate whether differences found in protein levels of CYPs between normal tissue of patients with adenoma and disease-free controls are related to the mRNA expression of CYPs, protein levels of CYP2C8, CYP3A4, and CYP3A5 of subjects with detectable mRNA expression were compared to those with undetectable mRNA expression of the respective CYP. CYP2E1 was excluded from this comparison, since CYP2E1 expression was only detected in three disease-free controls and two patients with adenoma. No comparison was performed with adenomatous tissue since the sample sizes were too small (n < 5) for most of the CYPs investigated. Results are summarized in Figure 4.
Despite the fact that expression of CYP2C was detected in ~50 % of patients with adenoma and disease-free controls, no differences were found when comparing protein levels of patients with adenoma and disease-free controls with detectable and undetectable mRNA expression. Similar, protein levels of CYP3A4 in disease-free controls with undetectable mRNA expression did not differ from those with detectable mRNA expression of CYP3A4. In contrast, CYP3A4 protein levels of patients with adenoma with undetectable mRNA expression of CYP3A4 were significantly higher when compared to those with detectable mRNA expression. Specifically, protein levels were ~68 % higher in patients with adenoma with undetectable mRNA expression. In disease-free controls, protein concentration of CYP3A5 did not differ between disease-free controls with detectable und undetectable CP3A5 mRNA expression. Similar to CYP3A4, protein levels of CYP3A5 were ~46 % higher in patients with undetectable mRNA expression in comparison to those with detectable mRNA expression of CP3A5. However, due to a large inter-individual variability differences were not statistically significant (p = 0.098).
Discussion
Protein levels of CYPs are decreased in normal tissue of patients with colonic adenoma
Carcinogens and pro-carcinogens present in the diet are critical environmental factors influencing the development of carcinoma in the large intestine. Cytochrome P450-mediated metabolism of (pro-)carcinogens has been shown – depending upon the compound – to either result in detoxification or toxification (for review see [1]) and might therefore have potential impact on the development of neoplasia. Furthermore, most human CYPs have been found to be genetically polymorphic and these polymorphisms may affect the enzyme expression and activity subsequently leading to an increased risk to develop several forms of cancer but also effecting treatment (for review see [16]). Only a few extensive studies on the mRNA and protein expression of cytochrome P450 in colonic tissue and colonic adenoma have been performed thus far. The presence of CYP2C-, CYP2E- and CYP3A- in normal and in neoplastic colonic mucosa along with substantial inter-individual variability has been reported by others before [12,14,17,18]. However, some of the available data are contradictory and most studies either determined mRNA expression or protein levels [14,18]. For example, Yokose et al. [14] showed the presence of CYP2C (CYP2C8-CYP2C19) protein in both unaffected and neoplastic human colon mucosa using immunohistochemical methods. In contrast, Western blot analyses of de Waziers et al. [18] and Massaad et al. [12], who both used conventional immunoperoxidase staining procedures, showed neither cytochrome P450 2C8-10 nor 2E1 to be present in normal colon mucosa and in colon carcinoma. McKay et al. [17], who used immunohistochemical methods, detected CYP3A protein more frequently in neoplastic tissue in colon mucosa specimens of patients with neoplasia than in the morphologically normal mucosa of these patients. Expression of CYP3A4 and CYP3A5 has been reported for human colon mucosa [19,20]. In the present study, the expression and protein levels of four representative CYPs were determined in macroscopically normal colon mucosa and adenoma of patients with adenoma and disease-free controls. At the level of mRNA expression, only the expression of CYP3A5 was found to be significantly altered between neoplastic tissue and normal, unaffected mucosa of patients with adenoma. However, as reported by others before, mRNA expression of all four CYPs varied extensively between individuals even though the expression of the housekeeping gene histone 3.3 was detected in all samples and was used for normalization. At the level of protein, CYP2C8, CYP3A4, and CYP3A5 protein concentration was found to be significantly lower in normal tissue of patients with adenoma than in colon mucosa of disease-free controls. Even though previous studies have indicated that the abundance of CYP protein is altered between neoplasia and normal tissue [17], in the present study protein levels of CYPs were found to be similar in neoplasia and normal tissue of patients with adenoma. Taken together, these data suggest that CYP expression not only varies extensively between individuals but that protein levels of some CYPs (e.g. CP2C8, CYP3A4, and CYP3A5) are considerably lower in normal tissue of patients with adenoma in comparison to those of disease-free controls.
CYP protein levels and mRNA expression are not related in normal colonic tissue
It has been suggested that expression of CYPs is not solely regulated at the level of gene transcription [11]. This is supported by the results of animals studies, reporting a dissociation of mRNA expression and protein levels of CYP2C7 (corresponding to human CYP2C8) and CYP2E1 in rat colon mucosa as well as CYP3A4 and CYP3A5 in duodenum and kidney. Furthermore, in vitro studies performed in rat hepatocytes indicate that CYP2E1 is regulated by posttranscriptional ligand-dependent stabilization of the enzyme [21]. Similar mechanisms have been described for CYP3A in rats and humans [22,23]. Using cultured hepatocytes it also has been showed that only ~60–70% of mRNA encoding for CYP2E1 is translated [24]. Indeed, in the present study, no relation with respect to protein levels of subjects with detectable and undetectable mRNA expression of the CYPs was found. Instead, protein levels of CYP2C8, CYP3A4, and CYP3A5 did not differ in disease-free controls. In contrast, in normal tissue of patients with adenomas protein levels of CYP3A4 and CYP3A5 were contrary to mRNA expression pattern. Taken together, these data suggest that in colon mucosa the expression of CYP2C8, CYP3A4, and CYP3A5 is not solely regulated at the level of transcription and that the mechanism of regulation, at least for some CYPs (e.g. CYP3A4 and CYP3A5), might differ between patients with colonic adenoma and healthy subjects.
Conclusion
In summary, inter-individual variability along with a substantial dissociation of mRNA expression pattern and protein levels seems to be a characteristic of CYP expression in the colon as has been reported in part by others before [12,14,17,18]. However, in the present study protein levels of CYP2C8, CYP3A4, and CYP3A5 were found to be significantly lower in normal unaffected colonic mucosa of patients with colonic adenoma in comparison with disease-free controls. Although the metabolic implications of these differences remain to be determined, reduced levels of some CYPs might result in an altered metabolism of xenobiotics and therefore contribute to the development of neoplasia in the large intestine. Future studies will address this possibility.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
IB has made substantial contributions to acquisition of data, the biochemical analysis, and the drafting of article. CB has made substantial contribution to conception and design as well as the interpretation of data. AP has been involved in the design, the drafting of the article, and revised it critically for intellectual content. All authors have given final approval of the version to be published.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The antibodies against human CYP2E1 and CYP3A4 were kindly provided by Dr. M. Ingelman-Sundberg. This work was supported by a grant from the Deutsche Krebshilfe (70-1881-B01) to AP and CB. The authors thank Gavin E. Arteel for his valuable comments.
Figures and Tables
Figure 1 CYP3A5, CYP2E1, CYP3A4, and CYP2C mRNA expression in colon mucosa of patients with adenoma and disease-free controls. (A) Representative agarose gel of mRNA integrity. Lane 1 and 12 = positive control, lane 2–11 = RNA extracted from colonic mucosa biopsies. (B) Representative photomicrograph of RT-PCR products of two subjects. All measurements were carried out in triplicate. Lanes 1, 6, 11: Histone 3.3; Lanes 2, 7, 12: CYP3A5; Lanes 3, 8, 13: CYP2E1; Lanes 4, 9, 14: CYP3A4; Lanes 5, 10, 15 = CYP2C, bp = base pairs. (C) Quantitative analysis of CYP2C, CYP2E1, CYP3A4, and CYP3A5 mRNA expression in normal colon mucosa and in colonic adenoma of patients. Results are normalized to histone 3.3 expression. Data are means ± SEM, ap < 0.05 compared to normal tissue. (D) Quantitative analysis of CYP2C, CYP2E1, CYP3A4, and CYP3A5 mRNA expression in normal colon mucosa of patients with colonic adenoma and disease-free controls. Results are normalized to histone 3.3 expression. Data are means ± SEM, ap < 0.05 compared to disease-free controls.
Figure 2 Protein levels of CYP2C8, CYP2E1, CYP3A4, and CYP3A5 in macroscopically normal tissue and in adenoma of patients with colonic adenoma. (A) Representative Western blots of CYP2C8, CYP2E1, CYP3A4, and CYP3A5 in macroscopically normal colon mucosa (= N) and adenoma (= A) of patients with colonic adenomas and (B) quantitative analysis of blots are shown. CYP2E1, CYP3A4, and CYP3A5 protein levels were determined in 25 patients with adenoma. CYP2C8 protein levels were determined in 19 cases. Data are means ± SEM
Figure 3 Diminished protein levels of CYP2C8, CYP3A4, and CYP3A5 in normal colon mucosa of patients with colonic adenoma. (A) Representative Western blots of CYP2C8, CYP2E1, CYP3A4, and CYP3A5 in normal colon mucosa from patients with colonic adenomas and disease-free controls. (B) Representative Western blot of β-actin protein in normal mucosa of patients with colonic adenoma and disease-free controls. (C) Quantitative analysis of blots. CYP2E1, CYP3A4, and CYP3A5 protein levels were determined in 25 patients with adenoma und 27 disease-free controls. CYP2C8 protein levels were determined in 19 cases and 12 disease-free controls. Data are means ± SEM, ap < 0.05 compared to disease-free controls.
Figure 4 Relation of CYP2C8, CYP3A4, CYP3A5 protein levels and mRNA expression pattern in normal colon mucosa of cases and disease-free controls. Comparison of protein levels of CYP2C8, CYP3A4, and CYP3A5 of patients and disease-free controls, respectively, with detectable mRNA expression and those with no detectable mRNA expression of the respective CYP. Data are mean ± SEM. ap < 0.05 compared to disease-free controls.
Table 1 Origin of tissue specimen and histology of colonic adenomatous tissue samples.
Patients with adenoma Disease-free controls
Normal tissue Adenoma Normal tissue
Ascending colon 8 7 8
Descending colon 8 12 5
Sigmoid colon and rectum 9 6 14
Tubular adenoma - 12 -
Tubular villious adenoma - 10 -
Villious adenoma - 3 -
Table 2 Clinical data of patients with adenoma and disease-free controls
Patients with adenoma Disease-free controls
Number (female/male) 25 (9/16) 27 (12/15)
Age 61 ± 5 59 ± 6
Body-mass index (BMI) 26.9 ± 4.8 26.4 ± 4.4
Cigarette usage: yes/no (number/d) 8/17 (2.9 ± 6.9) 5/22 (2.1 ± 6.5)
Alcohol consumers (more than 10 g/d): yes/no 8/17 4/23
Average alcohol intake (g/d) of consumers 15.1 ± 16.5 10.5 ± 13.5
Medication (yes/none) 15/10 17/10
All data are expressed as means ± SD.
Table 3 Primers used for RT-PCR analysis
Sense primer Antisense primer PCR product (bp) Reference
Histone 3.3 GCGTGCTAGCTGGATGTCTT CCACTGAACTTCTGATTCGC 150 [25].
CYP2C8-19 GCTAAAGTCCAGGAAGAGATTGA TCCTGCTGAGAAAGGCATGAAGT 332 [26].
CYP2E1 AGCACAACTCTGAGATATGG ATAGTCACTGTACTTGAACT 365 [26].
CYP3A4 CCAAGCTATGCTCTTCACCG TCAGGCTCCACTTACGGTGC 324 [27].
CYP3A5 TGTCCAGCAGAAACTGCAAA TTGAAGAAGTCCTTGCGTGTC 472 [27].
==== Refs
Sheweita SA Drug-metabolizing enzymes: mechanisms and functions Curr Drug Metab 2000 1 107 132 11465078 10.2174/1389200003339117
Oyama T Kagawa N Kunugita N Kitagawa K Ogawa M Yamaguchi T Suzuki R Kinaga T Yashima Y Ozaki S Isse T Kim YD Kim H Kawamoto T Expression of cytochrome P450 in tumor tissues and its association with cancer development Front Biosci 2004 9 1967 1976 14977602
Ding X Kaminsky LS Human extrahepatic cytochromes P450: function in xenobiotic metabolism and tissue-selective chemical toxicity in the respiratory and gastrointestinal tracts Annu Rev Pharmacol Toxicol 2003 43 149 173 12171978 10.1146/annurev.pharmtox.43.100901.140251
Kaminsky LS Fasco MJ Small intestinal cytochromes P450 Crit Rev Toxicol 1991 21 407 422 1801846
Zhang QY Dunbar D Ostrowska A Zeisloft S Yang J Kaminsky LS Characterization of human small intestinal cytochromes P-450 Drug Metab Dispos 1999 27 804 809 10383924
Peters WH Kock L Nagengast FM Kremers PG Biotransformation enzymes in human intestine: critical low levels in the colon? Gut 1991 32 408 412 1902809
Kolars JC Lown KS Schmiedlin-Ren P Ghosh M Fang C Wrighton SA Merion RM Watkins PB CYP3A gene expression in human gut epithelium Pharmacogenetics 1994 4 247 259 7894497
Chun YJ Lee S Yang SA Park S Kim MY Modulation of CYP3A4 expression by ceramide in human colon carcinoma HT-29 cells Biochem Biophys Res Commun 2002 298 687 692 12419308 10.1016/S0006-291X(02)02541-X
Pfrunder A Gutmann H Beglinger C Drewe J Gene expression of CYP3A4, ABC-transporters (MDR1 and MRP1-MRP5) and hPXR in three different human colon carcinoma cell lines J Pharm Pharmacol 2003 55 59 66 12625868
White TB Hammond DK Vásquez H Strobel HW Expression of two cytochromes P450 involved in carcinogen activation in a human cell line Mol Cell Biochem 1991 102 61 69 2052000 10.1007/BF00232158
Hakkak R Korourian S Ronis MJ Ingelman-Sundberg M Badger TM Effects of diet and ethanol on the expression and localization of cytochromes P450 2E1 and P450 2C7 in the colon of male rats Biochem Pharmacol 1996 51 61 69 8534269 10.1016/0006-2952(95)02154-X
Massaad L de Waziers I Ridbrag V Janot F Beaune PH Morizet J Gouyette A Chabot GG Comparison of Mouse and Human Colon Tumors with Regard to Phase I and Phase II Drug-metabolizing Enzyme Systems Cancer Research 1992 52 6567 6575 1423302
Massaad L Chabot GG Toussaint C Koscielny S Morizet J Bissery MC Gouyette A Influence of tumor size on the main drug-metabolizing enzyme systems in mouse colon adenocarcinoma Co38 Cancer Chemother Pharmacol 1994 34 497 502 7923560
Yokose T Doy M Taniguchi T Shimada T Kakiki M Horie T Matsuzaki Y Mukai K Immunohistochemical study of cytochrome P450 2C and 3A in human non-neoplastic and neoplastic tissues Virchows Arch 1999 434 401 411 10389623 10.1007/s004280050359
Bradford MM A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding Anal Biochem 1976 72 248 254 942051 10.1016/0003-2697(76)90527-3
Agundez JA Cytochrome P450 gene polymorphism and cancer Curr Drug Metab 2004 5 211 224 15180491 10.2174/1389200043335621
McKay JA Murray GI Weaver RJ Ewen SW Melvin WT Burke MD Xenobiotic metabolising enzyme expression in colonic neoplasia Gut 1993 34 1234 1239 8406161
De Waziers I Cugnenc PH Berger A Leroux JP Beaune PH Drug-metabolizing enzyme expression in human normal, peritumoral and tumoral colorectal tissue samples Carcinogenesis 1991 12 905 909 2029756
McKinnon RA Burgess WM Gonzalez FJ McManus ME Metabolic differences in colon mucosal cells Mutat Res 1993 290 27 33 7694095
Mercurio MG Shiff SJ Galbraith RA Sassa S Expression of cytochrome P450 mRNAs in the colon and the rectum in normal human subjects Biochem Biophys Res Commun 1995 210 350 355 7755610 10.1006/bbrc.1995.1668
Eliasson E Mkrtchian S Ingelman-Sundberg M Hormone- and substrate-regulated intracellular degradation of cytochrome P450 (2E1) involving MgATP-activated rapid proteolysis in the endoplasmic reticulum membranes J Biol Chem 1992 267 15765 15769 1639811
Feierman DE Melnikov Z Zhang J The paradoxical effect of acetaminophen on CYP3A4 activity and content in transfected HepG2 cells Arch Biochem Biophys 2002 398 109 117 11811955 10.1006/abbi.2001.2677
Zangar RC Hernandez M Novak RF Posttranscriptional elevation of cytochrome P450 3A expression Biochem Biophys Res Commun 1997 231 203 205 9070249 10.1006/bbrc.1997.6054
Kocarek TA Zangar RC Novak RF Post-transcriptional regulation of rat CYP2E1 expression: role of CYP2E1 mRNA untranslated regions in control of translational efficiency and message stability Arch Biochem Biophys 2000 376 180 190 10729204 10.1006/abbi.2000.1704
Futscher BW Blake LL Gerlach JH Grogan TM Dalton WS Quantitative polymerase chain reaction analysis of mdr1 mRNA in multiple myeloma cell lines and clinical specimens Anal Biochem 1993 213 414 421 8238918 10.1006/abio.1993.1440
Hakkola J Pasanen M Purkunen R Saarikoski S Pelkonen O Maenpaa J Rane A Raunio H Expression of xenobiotic-metabolizing cytochrome P450 forms in human adult and fetal liver Biochem Pharmacol 1994 48 59 64 8043031 10.1016/0006-2952(94)90223-2
Kivisto KT Griese EU Fritz P Linder A Hakkola J Raunio H Beaune P Kroemer HK Expression of cytochrome P 450 3A enzymes in human lung: a combined RT- PCR and immunohistochemical analysis of normal tissue and lung tumours Naunyn Schmiedebergs Arch Pharmacol 1996 353 207 212 8717162
|
16281975
|
PMC1310537
|
CC BY
|
2021-01-04 16:03:28
|
no
|
BMC Gastroenterol. 2005 Nov 10; 5:34
|
utf-8
|
BMC Gastroenterol
| 2,005 |
10.1186/1471-230X-5-34
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2711628392410.1186/1471-2105-6-271Research ArticleThe yeast kinome displays scale free topology with functional hub clusters Lee Robin EC [email protected] Lynn A [email protected] Ottawa Health Research Institute, Molecular Medicine Program, Ottawa, Canada2 University of Ottawa, Department of Cellular and Molecular Medicine, Ottawa, Canada2005 9 11 2005 6 271 271 12 5 2005 9 11 2005 Copyright © 2005 Lee and Megeney; licensee BioMed Central Ltd.2005Lee and Megeney; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The availability of interaction databases provides an opportunity for researchers to utilize immense amounts of data exclusively in silico. Recently there has been an emphasis on studying the global properties of biological interactions using network analysis. While this type of analysis offers a wide variety of global insights it has surprisingly not been used to examine more localized interactions based on mechanism. In as such we have particular interest in the role of key topological components in signal transduction cascades as they are vital regulators of healthy and diseased cell states.
Results
We have used publicly available databases and a novel software tool termed Hubview to model the interactions of a subset of the yeast interactome, specifically protein kinases and their interaction partners. Analysis of the connectivity distribution has inferred a fat-tailed degree distribution with parameters consistent with those found in other biological networks. In addition, Hubview identified a functional clustering of a large group of kinases, distributed between three separate groupings. The complexity and average degree for each of these clusters is indicative of a specialized function (cell cycle propagation, DNA repair and pheromone response) and relative age for each cluster.
Conclusion
Using connectivity analysis on a functional subset of proteins we have evidence that reinforces the scale free topology as a model for protein network evolution. We have identified the hub components of the kinase network and observed a tendency for these kinases to cluster together on a functional basis. As such, these results suggest an inherent trend to preserve scale free characteristics at a domain based modular level within large evolvable networks.
==== Body
Background
The Barabási and Albert scale free network model is a mathematical precept that describes the innate connectivity and distribution within complex networks. These scale free networks defy the traditional random graph model of Erdös and Renyi and display a connectivity distribution where the occurrence of highly interacting components of the network, defined as nodes decay as a power law, P(k) ~ k-γ [1-3]. In turn, growth of a scale free network is characterized by a preferential attachment scheme in which new nodes attach to older more connected nodes with a higher probability [2,4,5]. This model facilitates a rich-get-richer schema and allows for the existence of a very important class of highly connected hubs [1,6]. These hubs are largely responsible for the non-Gaussian connectivity distribution of scale free networks and are commonly orders of magnitude more connected than the average node. The existence of the hubs also provide a robust environment that is tolerant of random attack and failure but is very sensitive to hub perturbation [3,7-10].
This scale free topology has been demonstrated in a variety of man-made networks such as the World Wide Web and the actor collaboration network [1,2]. Scale free principles have also been noted in biologic systems such as the yeast protein-protein interaction dataset and the metabolic protein network [3,6]. Nevertheless, the suitability of the static scale free construct across diverse biologic systems has been challenged as a universal principle. For example, the yeast protein interaction network has been described as a date and party hub scale free network, in which these hubs are defined by variable or consistent interactions, respectively [10]. More critically, mathematical models of network growth have shown that preferential attachment may follow a random geometric topology rather than a scale free distribution [11]. Another study uses a learning algorithm to infer duplication-mutation-complementation as the central topology mechanism in the Drosophila melanogaster protein interaction network [12]. Indeed, it has been reported that the essential proteins within the larger yeast protein interaction network form an exponential connectivity distribution rather than a scale free distribution [13]. These observations raise intriguing possibilities, one of which suggests that broader scale free systems may evolve from a compilation of sub networks of different topology. Alternatively, this non-scale free structure may be an anomaly that originates from examining essential hubs versus non-essential hubs in the framework of an already established network.
Within this context, phosphorylation dependent signal transduction pathways provide an interesting venue to examine network behavior. In eukaryotic organisms, kinase directed phosphorylation is one of the most common forms of post-translational modification and as such this protein class is noted as a vital regulator of cellular function [14-16]. In addition, kinase families are well conserved across diverse phyla, suggesting that network organization may be similarly conserved. However, phosphorylation pathways are commonly studied as linear events connecting stimulus to response through a simple ladder of molecular interactions, a concept that is based largely on experimental perturbation and observation of directly connected proteins.
As such, identification of the select kinase hubs and interaction profiling should offer an insight into the functional complexities of cellular signaling in yeast and higher eukaryotes. Here, we examined the subset of the S. cerevisiae interaction data, which include protein kinases and their direct protein interactions. In all cases, analysis was performed on filtered datasets available in public databases to identify likely hub kinases and their interactivity. We confirmed scale free behaviour of this dataset using connectivity analysis and observed parameters as applied to a novel computer program/visualization tool we termed Hubview [17]. Interactions between the 19 most connected kinases, which we identified as super-hubs, were mapped along with less connected hub kinases. From this map we were able to discern three distinct clusters of kinase proteins, with each cluster retaining a common biologic function, i.e. cell cycle control, DNA repair/recombination and the pheromone/mating response. Together these observations suggest that scale free topology of the yeast kinome co-evolved with the emergence of distinct biologic domains.
Results and discussion
To study the topological properties of kinase mediated phosphorylation it was necessary to isolate the signaling component of the S. Cerevisiae proteome which we refer to as the Kinase-Partner Interaction set (KPI). The KPI node set was assembled from the concatenation of kinases from the database of interacting proteins (DIP) kinase search, the yeast kinases identified by Hunter and Plowman [18] and their non-kinase interaction partners. The interactions of the KPI nodes were considered bidirectional, as no directionality can be consistently inferred in most experimental conditions, and consisted of kinase-kinase and kinase-non-kinase interactions only (i.e. any potential interactions between non-kinases have been filtered). The core and complete KPI consisted of 607 nodes with 834 interactions and 1085 nodes and 1481 interactions respectively. Analysis using maximum likelihood estimation (MLE) of the degree distributions (Figure 1) resulted in derived γ values of γ = 2.32 in the core KPI and γ = 2.38 in the complete KPI which is in the biologically robust range of 2 < γ < 3 [19]. The Kolmogorov-Smirnov test for Power Law Distribution [20] of both MLEs (Core: N = 500, K = 0.021; Complete: N = 1000, K = 0.015) support the hypothesis that the KPI networks are indeed both power-law distributions and hence scale free in topology. The γ-values found for the KPIs are very consistent with reports from complete protein interaction data analysis and the deterministic scale free model [21,22] which confirms that the selection criteria for the KPI is not biased to any connectivity class. A study of metabolic networks has shown that the largest most connected part of a network (in the case of metabolic networks the largest component is less than 33% the size of the full network) tends to dominate the parameters found through topological analysis [23]. Here the degree distribution is challenged as a global property and treated as a local property of the network. It is worth noting that while the KPI does contain a limited number of segregated modules, the size of the largest component accounts for roughly 95% of the network and the degree distribution does represent a global property of the network.
Figure 1 Degree distributions giving the probability that a given protein will interact with exactly k other proteins for: a. the core KPI with γcore = 2.32 b. the complete KPI with γcomplete= 2.38. In both cases γ determined by maximum likeliness estimation (MLE) and goodness of fit determined by Kolomogorov-Smirnov (KS) test. The self organization of scale free topology is normally associated with much larger datasets yet we still find the scale free characteristics.
The KPI interaction data was analyzed by our visualization tool, Hubview. The hub-star-satellite view separates nodes with degrees higher than a user defined cut-off and their substrates of unary degree, groups the rest of the nodes within a sphere, and places the hub-stars around the sphere as satellites. The core and complete KPI were viewed with a cut-off of 10 and 15 respectively (Figure 2) and in both cases resulted in 28 satellites responsible for about 69% and 71% of the interactions respectively. The average node degree for both the core and complete KPI was found to be <k> ≅ 1.3.
Figure 2 a. Hubview Fruchterman-Rheingold visualization of Core KPI (607 nodes, 834 interactions) b. Hub-Star-Satellite output of Hubview of complete KPI (1085 nodes with 1481 interactions) with hub degree cut-off of 15 yields 28 hubs.
The putative hubs identified by Hubview were compiled into a list of 33 distinct nodes and ranked by average degree where the degree found in the core KPI was given twice the weight (Table 1). Defining the actual cut-off degree for a hub is a subjective task, here we defined 13 (10*<k>) as the cut-off for high confidence in super-hub status. This cut-off retains 19 proteins as high confidence hubs that still maintain ~64% of KPI interactions which suggests that less understood signaling systems in higher eukaryotes may be studied with higher efficiency by identifying likely hub kinases (using expression and activity profiling) and mapping the complete set of their immediate interaction partners.
Table 1 Summary of hub kinases as identified by Hubview: Weighted mean calculated by giving double weight to degrees listed in the core KPI dataset. Hubs with knockout lethal phenotype listed as identified by Giaver et al [35].
Name DIP Node Number Knockout Lethal Weighted Degree Confidence as a super-hub
CDC28 DIP:1039N yes 150 ± 40 High
CKA1 DIP:48N no 50 ± 10 High
HRR25 DIP:157N yes 40 ± 10 High
SLT2 DIP:1448N no 37 ± 5 High
YCK1 DIP:719N no 34 ± 7 High
KSS1 DIP:60N no 33 ± 5 High
SNF1 DIP:18N no 24 ± 1 High
PHO85 DIP:1493N no 22 ± 2 High
RAD53 DIP:2322N yes 22 ± 2 High
CKB2 DIP:262N no 19 ± 5 High
CDC7 DIP:1235N yes 19 ± 6 High
RIM11 DIP:1566N no 19 ± 3 High
SSN3 DIP:2574N no 17 High
DBF2 DIP:2319N no 17 ± 2 High
DUN1 DIP:1772N no 16 ± 6 Uncertain
MKK2 DIP:1447N no 16 ± 2 High
CDC5 DIP:2321N yes 14 ± 1 High
STE11 DIP:861N no 14 ± 1 High
PKC1 DIP:1516N yes 14 ± 1 High
TPK3 DIP:550N no 14 ± 1 High
CKA2 DIP:1038N no 13 ± 5 Uncertain
SPS1 DIP:6598N no 13 ± 2 Uncertain
CLA4 DIP:2276N no 13 Uncertain
YAK1 DIP:1374N no 12 ± 4 Uncertain
STE20 DIP:712N no 12 ± 1 Uncertain
FUS3 DIP:714N no 12 ± 1 Uncertain
CHK1 DIP:1253N no 12 ± 2 Uncertain
KIN2 DIP:6276N no 12 ± 2 Uncertain
BUD32 DIP:5008N no 11 ± 9 Uncertain
SWE1 DIP:2410N no 11 Low
CKB1 DIP:282N no 11 ± 7 Uncertain
GIN4 DIP:2260N no 11 ± 1 Low
BCY1 DIP:551N no 10 ± 3 Low
The 124 members of the protein kinase superfamily list [18] were cross referenced with the list of essential yeast proteins [24] to identify the yeast kinases with known knockout lethal phenotypes. Of the 124 kinases only 16 were deemed to be lethal deleterious mutants yielding a 13% chance of lethality in an instance of random single kinase deletion. In contrast, 6 of the 19 hubs named as high confidence in table 1 are listed as essential resulting in a 32% chance of lethality attributed to random deletion of one of the 19 high confidence hubs. This marked increase in lethality associated with directed hub attack is consistent with existing studies of scale free networks [3] and indicates a likely tendency for hub kinases to be preserved in an evolutionary perspective.
24 of the 33 hubs listed in table 1 were found to interact with one another. The interplay between these 24 connected hubs forms a kinase signaling backbone (figure 3a) through which 3 distinct groups of interacting hubs (forthwith these interacting hubs are referred to as hub clusters) can be identified. Presumably, the hub clusters would provide vital functions as whole as in most cases the constituent hubs are not directly essential themselves. The structure of this backbone may offer some insight in identifying synthetic lethality strategies, i.e. CKA1 and CKA2 knockouts are both viable but double deletion mutant has a lethal phenotype [25]. Backbone hubs have been ordered by degree to illustrate a possible correlation between degree and phylogenetic age where, by direct consequence of the growth and preferential attachment conditions in scale free systems, more connected hubs are likely to be older than less connected hubs [26]. A cross genome study of four organisms from different regions of the phylogenetic tree has been used to identify connectivity and emergence time of yeast proteins [5]. The results of this study support the preferential attachment and growth criteria as outlined by the scale free theory. Older proteins appear to be more connected than younger proteins. Another explanation of the degree arrangement is that the average size or degree of a cluster is associated with the evolutionary age of the clusters functional class [27]. This perspective is based on a similar study using a more rigorous phylogenetic profiling technique. The results suggest a modified form of scale free preferential attachment whereby proteins bind preferentially within their own functional class and not globally or promiscuously. By this model a younger protein may be more connected than an older one simply because it is part of an older and more connected functional grouping which emerged during an earlier phylogenetic period. Here the average connectivity of the functional group is proportional to the age of that group i.e. older eukaryotic proteins are shown to be more connected than yeast specific proteins. This perspective is very plausible as it suggests that proteins of similar function will interact within the same pool.
Figure 3 a. Interplay between 24 of the hubs identified by HubView constituting the signaling backbone. Red hubs are deemed as lethal knockout phenotypes as described in the systematic deletion project [35]. The nodes are separated by degree along the ordinate to illustrate a possible relationship between the degree of a hub and its phylogenetic age while the arrangement along the abscissa is purely aesthetic. The proposed function of the 3 clusters from left to right are: DNA damage/repair, cell cycle propagation and pheromone response. b. The dendrogram to the right confirms functional interactions for a number of the backbone hubs using redundancy clustering of the entire KPI as described by Samanta & Liang [28]. Only clusters containing a hub kinase are depicted in the dendrogram.
In response to the latter interpretation we examined the basic purpose of the individual hubs and observed a common functional theme concomitant with each cluster. The largest cluster, containing cdc28, is functionally associated with cell cycle propagation through the various phases. The second cluster, with CKA1 as a peak, is generally associated with kinase proteins that manipulate response to DNA damage and the final KSS/MAPK cluster is involved with the regulation of the pheromone response. These results seem to offer a reasonable order to the emergence of specialized functions central to all eukaryotes i.e., the cell division cycle predates the DNA verification mechanisms, which in turn predates the youngest reproductive module, the mating response.
The entire core and complete kinomes were clustered using the probabilistic method described by Samanta and Liang [28]. This method identifies functional relationships between proteins through redundancy of interaction partners. A number of the associations in the backbone clusters were confirmed using this algorithm (figure 3b). Interestingly the proteins in the cell cycle propagation cluster did not appear as functionally redundant in the clustering. Presumably the three clusters converge downstream to some extent but at the hub level this indicates that these components offer highly specialized non redundant services to the cell cycle cluster likely due to the ancient nature of their function. This method can also be used to identify likely synthetic lethality as many viable knockouts are rescued through redundant interactions. The full results of the clustering is available as supporting information (see additional file 1) or can be generated using Hubview.
In addition to the scale free topology, modeling of the yeast kinome using the Hubview cascade crawler function revealed other notable characteristics. Specifically, individual clusters containing hub kinases also include kinases that interact both inside and outside the scope of the immediate functional cluster. This characteristic was generally not observed with non-hub clusters. For example, the cluster of kinases involved in the MAPK cascade (a functional cluster with hub kinases) retain interactions with a number of non-MAPK kinases i.e. single points that interact both within and outside of the MAPK class. This is a feature we refer to as an open loop signal (Figure 4). Identifying open points within a cluster provides the user with probable targets for regulation of that functional cluster or even likely paths for signaling crosstalk. Open loop kinase cascades appear to reflect robust cellular responses that require multiple alterations and as such would require direct communication and signal propagation between numerous key regulatory factors/kinases themselves. However, non hub kinase clusters such as the TOR kinase cascade do not retain direct interactions between unclustered kinases and as such conform to a closed loop structure (Figure 4). Closed loop structures are likely to be kinase directed cascades that perform a very discrete cellular function in response to a limited or very specific initiating event.
Figure 4 Cascade crawler output of Hubview. Gray spheres represent non-kinases while all other colored spheres represent kinases of varying degree. a) Depiction of closed loop TOR signaling: neither TOR protein is directly connected to another kinase indicating highly specified reaction. b) Depiction of open loop MAPK signaling: white spheres denote non-MAPK kinases that interact within and outside of the MAPK clusters representing possible regulatory and cross-talk channels associated with more complicated cellular behaviour.
The network of essential yeast proteins has been compiled and identified as an exponential distribution [13]. This distribution is normally associated with more stochastic evolutionary mechanisms. It has been argued that this network may represent an ancestral core about which the rest of the yeast interactome has formed [13]. The existence of an exponential core does not directly contradict the scale free topology observed in the protein interaction network but may simply exist as a scaffold for scale free mechanisms to adhere to. This possibility is interesting as it may also suggest that different parts of the interactome may have evolved by different evolutionary pressures causing unequal distribution of topological properties within the same interactome.
A recent investigation concerning the effects of sampling on topology adds a small shard of doubt to studies of protein network topology. In this study the effects of various large scale experiments were simulated by first generating different networks of known topology and then sampling interactions in a scale mimicking yeast two hybrid and co-affinity purification [29]. They found that under some conditions that non-scale free topologies (i.e. Erdös and Renyi network with <K> = 10), when sampled, can generate sub-networks with scale free properties. Here the kinome benefits from the fact that it is a widely studied mechanistic class and many of the interactions, especially in the core kinome, have been identified in smaller scale experiments and not exclusively large scale experiments. This suggests that the much smaller kinome network may not suffer as much as networks derived solely from large scale experiments. The results of this study certainly insist on the caveat that the results of our KPI network cannot be extrapolated to the complete yeast protein interaction network with any amount of confidence.
Conclusion
Our analysis suggests that the yeast kinome is an evolved scale free system. Moreover, these observations suggest the intriguing possibility that the scale free topology of the global protein-protein interaction network or any larger biologic network may be the composite of smaller evolving topologies (such as the kinome), all of which are subject to their own selective pressures.
Methods
Interaction database
Both the core and complete yeast interaction data of the manually curated DIP [30] were used as interaction data sets. The complete dataset consists largely of high throughput interaction data [19,31-33]. The core DIP dataset consists of interactions found in small scale experiments, two or more independent larger scale experiments and, when paralogous interaction data exist, the Paralogous Verification Method PVM [31,32]. The core dataset is believed to correctly identify the core of interacting proteins in yeast and provides a minimal interaction view of the yeast interactome. For our purposes the complete yeast interaction set is viewed as a hypothetical maximal interaction set. The many false positives, negatives and unlikely biologic interactions [19] available in the complete dataset are still valuable as they may be representative of interactions in a diseased state based on possible spatial and temporal protein delocalization. The DIP is available online at .
Interaction filter
Both datasets were filtered to include only kinases and direct interaction partners with kinases as found in the DIP node search in conjunction with kinases listed in the protein kinase superfamily found by Hunter and Plowman [18]. The resulting Kinase-Partner Interaction dataset (KPI) consisted of 607 nodes with 834 interactions in the case of the core dataset and 1085 nodes with 1481 interactions in the case of the complete dataset.
Hubview description
We developed a program called Hubview to help us analyze the KPI network and visualize the hubs and hub interactions found in the datasets. The degree distribution of the loaded network can be obtained by pressing the probability distribution button. The main program and OpenGL network interface utilize an undirected binary adjacency matrix which is then interpreted in real-time 3D. Yeast specific information such as the naming convention (DIP number, ORF and common name) and protein type (kinase or non-kinase) is hard coded into Hubview minimizing the amount of data required to generate an interaction network. The 3D representation is geared towards identifying nodes with degrees higher than a user-defined cut-off and displaying them in either a hub-star-satellite view whereby hub degree and inter-hub interactions are plainly visible or a Fruchterman-Rheingold (FR) force-directed placement arrangement [34] which offers a less tangled, more visually appealing interpretation.
Briefly, the FR algorithm causes the system to untangle itself through iterative simulation of mechanical and electrostatic forces. A connection between a pair of nodes is treated as though a spring were connecting those nodes creating attractive forces between all connected pairs. Repulsive electrostatic forces are also generated by considering each node as a negative point charge. The nodes in the analogous system move in 3 dimensional space according to the attractive and repulsive forces. The final arrangement is displayed once the system has evolved through a set number of iterations resulting in an intelligible and appealing graph.
The hub-star-satellite view is generated by placing all nodes randomly within a sphere of radius ri. All nodes with connectivity higher than the user defined cutoff are identified as hubs and projected outside of the sphere to a radial position, rf, outside the confines of the initial sphere (rf > ri). Any substrates of the new hub with unary degree are also moved to positions spherically centered near the newly placed hub generating a hub-star-satellite. The algorithm ends once all hubs are processed similarly. The advantage of this view type is that it allows interactions between hubs to be quickly and easily identified as all visually interfering substrates remain pooled within the initial sphere.
Another useful visualization method included in Hubview is the cascade crawler function. This view type is geared towards depiction of smaller cascades (the immediate and remote neighbors of a chosen protein) within the complete network. The cascade crawler function is controlled by a point and click interface whereby the user can define a specific protein(s) as a starting point and display all of its substrates by clicking on it. Clicking subsequent nodes will display their interaction partners in turn. Using this function along with the FR algorithm one can develop appealing visual interpretations of specific cascades and interactions (figure 4).
Hubview also utilizes the clustering method proposed by Samanta & Liang [28]. The main suggestion of this algorithm is that if two proteins in a network share a significantly larger number of common interaction partners than what is expected from a similar random network then the pair of proteins likely share a close functional relationship. This process assigns a P value between every pair of proteins in the network representing the probability that an association between proteins is random i.e. a higher P score means that the pair is not functionally associated. The algorithm then merges the pair sharing the lowest P value into a cluster and recalculates P values for all possible pairs again treating the newly formed cluster as though it were a single protein. This process repeats until all P values are higher than a user defined cutoff. Once a network is loaded one can access this method by clicking the cluster button. Here a cutoff value can be defined which represents the probability that a particular association is random and a dendrogram is produced (which can be saved as a .BMP file), Samanta & Liang reported successful clustering of a large portion of the yeast interactome (N = 4692) using a cutoff value of up to 2 × 10-4 [28] indicating that this cutoff can be considered sharp and biologically relevant in our much smaller KPI networks (Ncore = 607 and Ncomplete = 1085).
Topology analysis
To counter the distortion associated with log-log data transformation the γ-value associated with the degree distribution of the KPI was analyzed using maximum likelihood estimation of the zeta function (MLE) and goodness of fit confirmed by the Kolmogorov-Smirnov test for power law distributions [20]. Briefly, the γ parameter associated with the pure power law,
P(k)=k−γζ(γ) [1]
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGqbaucqGGOaakcqWGRbWAcqGGPaqkcqGH9aqpdaWcaaqaaiabdUgaRnaaCaaaleqabaGaeyOeI0cccaGae83SdCgaaaGcbaGae8NTdONaeiikaGIae83SdCMaeiykaKcaaiaaxMaacaWLjaGaei4waSLaeGymaeJaeiyxa0faaa@3FED@
is best approximated by the solution of:
∂ζ(γ)∂γζ(γ)=−∑i=1NLog(ki)N [2]
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaadaWcaaqaamaaliaabaGaeyOaIylccaGae8NTdONaeiikaGIae83SdCMaeiykaKcabaGaeyOaIyRae83SdCgaaaqaaiab=z7a6jabcIcaOiab=n7aNjabcMcaPaaacqGH9aqpcqGHsisldaWcaaqaamaaqahabaGaemitaWKaem4Ba8Maem4zaCMaeiikaGIaem4AaS2aaSbaaSqaaiabdMgaPbqabaaabaGaemyAaKMaeyypa0JaeGymaedabaGaemOta4eaniabggHiLdGccqGGPaqkaeaacqWGobGtaaGaaCzcaiaaxMaacqGGBbWwcqaIYaGmcqGGDbqxaaa@526C@
Where:
- ζ(γ) is the Riemann Zeta function
∑k=1∞k(−γ) [3]
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaadaaeWbqaaiabdUgaRnaaCaaaleqabaGaeiikaGIaeyOeI0cccaGae83SdCMaeiykaKcaaaqaaiabdUgaRjabg2da9iabigdaXaqaaiabg6HiLcqdcqGHris5aOGaaCzcaiaaxMaacqGGBbWwcqaIZaWmcqGGDbqxaaa@3E44@
- ki is the ith non-zero observed degree of the P(k) vs. k distribution.
- γ is the power law exponent [20]
Protein essentiality
Phenotypic profiles of gene-deletion mutants (nearly 96% of known ORFs) have been systematically constructed and analyzed by a PCR-based gene deletion strategy [35]. A list of essential ORFs has been generated [24] and can be used to predict a lethal protein knockout or disruption phenotype.
Authors' contributions
RECL participated in the projects design and coordination, carried out programming, informatics, and drafted the manuscript. LAM conceived the study, participated in its design and coordination and drafted the manuscript.
Supplementary Material
Additional File 1
The software Hubview has been successfully tested and used on a number of recent generation PCs with the Windows XP operating system. Suggested systems should have more than 256 Megs of ram and an OpenGL compliant video card with onboard ram. The software requires the windows operating system. Installation: Simply unzip all files into the same folder. Supplementary Material.doc: Microsoft Word Document, Complete results of redundancy clustering. Hubview.zip: Winzip archive, Reviewer copy of Hubview program used to develop data for this manuscript.
Click here for file
Acknowledgements
The authors would like to thank Drs. Pasan Fernando, Lawrence Puente and Lukasz Salwinski for helpful discussions. This work was supported by grants to L.A.M. from the Canadian Institutes of Health Research (CIHR) and the Heart and Stroke Foundation of Canada. L.A.M. is the Mach-Gaennelsen chair in cardiac research at the Ottawa Health Research Institute.
==== Refs
Albert R Jeong H Barabási AL Internet: Diameter of the World-Wide Web Nature 1999 401 130 131 10.1038/43601
Barabasi AL Albert R Emergence of scaling in random networks Science 1999 286 509 512 10521342 10.1126/science.286.5439.509
Jeong H Mason SP Barabasi AL Oltvai ZN Lethality and centrality in protein networks Nature 2001 411 41 42 11333967 10.1038/35075138
Barabasi AL Albert R Jeong H Mean-Field theory for scale-free random networks Physica A 1999 272 173 187 10.1016/S0378-4371(99)00291-5
Eisenberg E Levanon EY Preferential attachment in the protein network evolution Phys Rev Lett 2003 91 138701 14525344 10.1103/PhysRevLett.91.138701
Jeong H Tombor B Albert R Oltvai ZN Barabasi AL The large-scale organization of metabolic networks Nature 2000 407 651 654 11034217 10.1038/35036627
Callaway DS Newman ME Strogatz SH Watts DJ Network robustness and fragility: percolation on random graphs Phys Rev Lett 2000 85 5468 5471 11136023 10.1103/PhysRevLett.85.5468
Sole RV Montoya JM Complexity and fragility in ecological networks Proc R Soc Lond B Biol Sci 2001 268 2039 2045 10.1098/rspb.2001.1767
Shargel B Sayama H Epstein IR Bar-Yam Y Optimization of robustness and connectivity in complex networks Phys Rev Lett 2003 90 68701 10.1103/PhysRevLett.90.068701
Han JD Bertin N Hao T Goldberg DS Berriz GF Zhang LV Dupuy D Walhout AJ Cusick ME Roth FP Vidal M Evidence for dynamically organized modularity in the yeast protein-protein interaction network Nature 2004 430 88 93 15190252 10.1038/nature02555
Przulj N Corneil DG Jurisica I Modeling interactome: scale-free or geometric? Bioinformatics 2004 20 3508 3515 15284103
Middendorf M Ziv E Wiggins CH Inferring network mechanisms: The Drosophila melanogaster protein interaction network Proc Natl Acad Sci U S A 2005 15728374
Pereira-Leal JB Audit B Peregrin-Alvarez JM Ouzounis CA An Exponential Core in the Heart of the Yeast Protein Interaction Network Mol Biol Evol 2004
Manning G Whyte DB Martinez R Hunter T Sudarsanam S The protein kinase complement of the human genome Science 2002 298 1912 1934 12471243 10.1126/science.1075762
Hunter T Signaling--2000 and beyond Cell 2000 100 113 127 10647936 10.1016/S0092-8674(00)81688-8
Holmberg CI Tran SE Eriksson JE Sistonen L Multisite phosphorylation provides sophisticated regulation of transcription factors Trends Biochem Sci 2002 27 619 627 12468231 10.1016/S0968-0004(02)02207-7
Hubview
Hunter T Plowman GD The protein kinases of budding yeast: six score and more Trends Biochem Sci 1997 22 18 22 9020587 10.1016/S0968-0004(96)10068-2
von Mering C Krause R Snel B Cornell M Oliver SG Fields S Bork P Comparative assessment of large-scale data sets of protein-protein interactions Nature 2002 417 399 403 12000970 10.1038/nature750
Goldstein ML Morris SA G.G. Y Problems with fitting to the power-law distribution Eur Phys J B 2004 41 255 258 10.1140/epjb/e2004-00316-5
Farkas I Derenyi I Jeong H Neda Z Oltvai ZN Ravasz E Schubert A Barabasi AL Vicsek T Networks in life: Scaling properties and eigenvalue spectra Physica A 2002 314 25 34 10.1016/S0378-4371(02)01181-0
Yook SH Oltvai ZN Barabasi AL Functional and topological characterization of protein interaction networks Proteomics 2004 4 928 942 15048975 10.1002/pmic.200300636
Ma HW Zeng AP The connectivity structure, giant strong component and centrality of metabolic networks Bioinformatics 2003 19 1423 1430 12874056 10.1093/bioinformatics/btg177
Yeast Deletion Project
Barz T Ackermann K Pyerin W Perturbation of protein kinase CK2 uncouples executive part of phosphate maintenance pathway from cyclin-CDK control FEBS Lett 2003 537 210 214 12606059 10.1016/S0014-5793(03)00112-1
Wuchty S Evolution and topology in the yeast protein interaction network Genome Res 2004 14 1310 1314 15231746 10.1101/gr.2300204
Kunin V Pereira-Leal JB Ouzounis CA Functional evolution of the yeast protein interaction network Mol Biol Evol 2004 21 1171 1176 15071090 10.1093/molbev/msh085
Samanta MP Liang S Predicting protein functions from redundancies in large-scale protein interaction networks Proc Natl Acad Sci U S A 2003 100 12579 12583 14566057 10.1073/pnas.2132527100
Han JD Dupuy D Bertin N Cusick ME Vidal M Effect of sampling on topology predictions of protein-protein interaction networks Nat Biotechnol 2005 23 839 844 16003372 10.1038/nbt1116
Database of Interacting Proteins
Deane CM Salwinski L Xenarios I Eisenberg D Protein interactions: two methods for assessment of the reliability of high throughput observations Mol Cell Proteomics 2002 1 349 356 12118076 10.1074/mcp.M100037-MCP200
Salwinski L Miller CS Smith AJ Pettit FK Bowie JU Eisenberg D The Database of Interacting Proteins: 2004 update Nucleic Acids Res 2004 32 Database issue D449 51 14681454 10.1093/nar/gkh086
Xenarios I Rice DW Salwinski L Baron MK Marcotte EM Eisenberg D DIP: the database of interacting proteins Nucleic Acids Res 2000 28 289 291 10592249 10.1093/nar/28.1.289
Fruchterman TMJ Reingold EM Graph Drawing by Force-directed Placement Softw Exp Pract 1991 21 1129 1164
Giaever G Chu AM Ni L Connelly C Riles L Veronneau S Dow S Lucau-Danila A Anderson K Andre B Arkin AP Astromoff A El-Bakkoury M Bangham R Benito R Brachat S Campanaro S Curtiss M Davis K Deutschbauer A Entian KD Flaherty P Foury F Garfinkel DJ Gerstein M Gotte D Guldener U Hegemann JH Hempel S Herman Z Jaramillo DF Kelly DE Kelly SL Kotter P LaBonte D Lamb DC Lan N Liang H Liao H Liu L Luo C Lussier M Mao R Menard P Ooi SL Revuelta JL Roberts CJ Rose M Ross-Macdonald P Scherens B Schimmack G Shafer B Shoemaker DD Sookhai-Mahadeo S Storms RK Strathern JN Valle G Voet M Volckaert G Wang CY Ward TR Wilhelmy J Winzeler EA Yang Y Yen G Youngman E Yu K Bussey H Boeke JD Snyder M Philippsen P Davis RW Johnston M Functional profiling of the Saccharomyces cerevisiae genome Nature 2002 418 387 391 12140549 10.1038/nature00935
|
16283924
|
PMC1310538
|
CC BY
|
2021-01-04 16:27:47
|
no
|
BMC Bioinformatics. 2005 Nov 9; 6:271
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-271
|
oa_comm
|
==== Front
BMC GeriatrBMC Geriatrics1471-2318BioMed Central London 1471-2318-5-131628009110.1186/1471-2318-5-13Research ArticlePrevalence of physical and verbal aggressive behaviours and associated factors among older adults in long-term care facilities Voyer Philippe [email protected] René [email protected] Ginette M [email protected] Johanne [email protected] Nathalie [email protected]édard Annick [email protected] Faculty of nursing, Laval University, Quebec City, Canada2 Faculty of Medicine, Laval University, Quebec City, Canada3 Faculty of Medicine, University of Sherbrooke, Sherbrooke, Canada4 Faculty of Medicine, University of Montréal, Montreal, Canada5 School of Psychology, Laval University, Quebec City, Canada2005 10 11 2005 5 13 13 25 3 2005 10 11 2005 Copyright © 2005 Voyer et al; licensee BioMed Central Ltd.2005Voyer et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Verbal and physical aggressive behaviours are among the most disturbing and distressing behaviours displayed by older patients in long-term care facilities. Aggressive behaviour (AB) is often the reason for using physical or chemical restraints with nursing home residents and is a major concern for caregivers. AB is associated with increased health care costs due to staff turnover and absenteeism.
Methods
The goals of this secondary analysis of a cross-sectional study are to determine the prevalence of verbal and physical aggressive behaviours and to identify associated factors among older adults in long-term care facilities in the Quebec City area (n = 2 332).
Results
The same percentage of older adults displayed physical aggressive behaviour (21.2%) or verbal aggressive behaviour (21.5%), whereas 11.2% displayed both types of aggressive behaviour. Factors associated with aggressive behaviour (both verbal and physical) were male gender, neuroleptic drug use, mild and severe cognitive impairment, insomnia, psychological distress, and physical restraints. Factors associated with physical aggressive behaviour were older age, male gender, neuroleptic drug use, mild or severe cognitive impairment, insomnia and psychological distress. Finally, factors associated with verbal aggressive behaviour were benzodiazepine and neuroleptic drug use, functional dependency, mild or severe cognitive impairment and insomnia.
Conclusion
Cognitive impairment severity is the most significant predisposing factor for aggressive behaviour among older adults in long-term care facilities in the Quebec City area. Physical and chemical restraints were also significantly associated with AB. Based on these results, we suggest that caregivers should provide care to older adults with AB using approaches such as the progressively lowered stress threshold model and reactance theory which stress the importance of paying attention to the severity of cognitive impairment and avoiding the use of chemical or physical restraints.
==== Body
Background
Among the entire spectrum of behavioural and psychological symptoms of dementia, aggressive behaviour (AB) is the most disturbing and distressing behaviour displayed by older patients in long-term care facilities. According to Patel and Hope [1], AB refers to an overt act, which is not accidental, involving the delivery of noxious stimuli to (but not necessarily aimed at) an object or towards the self or others. Choux and colleagues [2] further specified that the AB may be verbal or physical behaviour that harms or threatens another person. Physical aggression includes hitting, kicking, scratching, pushing, biting, punching, grabbing, throwing objects, pinching, cutting, and stabbing. Verbal aggression is typically considered as insulting, obscene or profane language or sexual advances.
The purpose of this study was to describe the phenomenon of AB among older patients in long-term care facilities.
Prevalence of AB among older patients in long-term care facilities
Prevalence of aggressive behaviour among residents in long-term care facilities varies widely from 7% to 91% [3-7]. According to a recent literature review [8], it was estimated that on average 24% of cognitively impaired residents are agitated or aggressive. Based on our own review of studies published between 1999 and 2001, 24 to 95% of long term care residents display AB, 10 to 95% exhibit physical aggressive behaviour and 10 to 91%, verbal aggressive behaviour [9,10]. In brief, prevalence rates of AB, although varying widely, may be very high.
Consequences of aggressive behaviour
Aggressive behaviour affects mainly older patients themselves and their informal and formal caregivers. It can also lead to increased health care costs. Older patients displaying ABs and other disruptive behaviours are more likely to receive psychotropic drugs and to be physically restrained [7]. The negative consequences of both interventions are well known and include worsening of cognitive impairment and reduced physical strength, endurance and flexibility [11]. AB is also associated with depression and loss of functional independence [12].
Family members and friends are affected by ABs in long-term care facilities [13,14]. They can be embarrassed by these behaviours and eventually reduce the frequency of their visits.
Health care providers' stress is augmented by AB displayed by long-term care residents [7,15]. ABs are more likely to occur during the activities of daily living (ADL) [16], more specifically when the older patients are mobilized, transferred, dressed, fed, bathed and groomed. AB compromises the delivery of care and can lead to psychological distress, emotional exhaustion, depression and occupational injuries among nursing staff [2,17,18]. Finally, increased health care costs arising from staff turnover and absenteeism are also associated with the prevalence of AB among older patients in long term care facilities [16].
Factors associated with aggressive behaviour
Individual and environmental factors have been associated with AB among older adults. Among the individual factors, being of male gender and among the younger segment of the older adult population (65 to 70 years old) are the most consistent demographic characteristics associated with AB [19,20]. Other individual factors include dementia (cortical and subcortical) and delirium, especially its hyperactive form [4,5,9,21]. Among older patients with dementia, severity of cognitive impairment has been linked to AB. Older patients with severe cognitive impairment are associated with more frequent AB episodes [19]. Psychiatric diseases (depression, mania, schizophrenia, anxiety and post traumatic stress disorders) and some specific symptoms (delusions, illusions, hallucinations) have also been linked to AB [5,19]. Personality traits, especially premorbid aggressive personality traits and aggression, are also risk factors for assaultive behaviour [2,6]. Researchers underline the potential role of pain, discomfort, sensory deficiencies (vision, hearing) and unmet basic needs (nutrition, hydration, exercise, sleep, etc) that could also lead to AB [19,21,22].
Both social and physical environmental factors, such as social visitation and physical restraint, may trigger an outburst of AB and, as has previously been mentioned, ABs occur more frequently during ADL provided by nursing staff. Researchers have hypothesized that older patients react in this way because their personal space or privacy is perceived as violated and they feel threatened [2,3,23]. Thus, AB would appear to be a defensive reaction. This stresses the importance of both the personal communication skills and knowledge and understanding of nursing staff providing care to older patients in institutions.
Environmental factors may also lead to ABs when they cause pain (e.g. drawing blood samples), discomfort or frustration (e.g. locked doors) or when they are interpreted as threatening. Similarly associated factors include noise, uncomfortable temperature, inadequate lighting and moving older patients to unfamiliar places [5,19].
To summarize, numerous factors have been associated with AB and since no one study has addressed all of them, it is difficult to determine those that are more important than others. This lack of knowledge concerning AB in long-term care facilities raises four important issues. First, studies on AB in long-term care facilities have generally been conducted on a small sized sample, some were case studies and few details are usually provided about methods for selecting participants [2,10,16,20,24,25], i.e. many studies appear to be vulnerable to selection bias. Second, many have examined less than four of the risk factors for AB [10,23-27], although there are obviously some exceptions to this observation [6,20]. Third, among those studies reporting prevalence rates for AB, few have described in detail the frequencies of specific ABs displayed in long-term care facilities [6,28]. Fourth, results of recent studies that have separated AB into verbal and physical behaviours suggest that factors associated with these two behaviour types may be different [25,27,28]. Based on their results, we are of the opinion that there is a sound basis for dividing AB into two different categories, namely verbal and physical AB.
Objectives of the study
The main objective of this study was to describe the phenomenon of AB among older patients in long-term care facilities. Specific objectives were: 1 – to determine the prevalence of verbal and physical AB and the frequencies for each behaviour, and 2 – to identify the factors associated with a) both verbal and physical aggressive behaviours (BAB), b) physical aggressive behaviour (PAB) and, c) verbal aggressive behaviour (VAB).
Methods
This is a secondary analysis of a cross-sectional study involving 2 633 older adults that was originally carried out to identify factors associated with the use of physical restraints in all 28 long-term care facilities in the Quebec City area.
In the province of Quebec, public long-term care facilities generally house older people with significant physical and mental problems [29,30]. The admission process to these facilities is managed by a single central committee, which evaluates admission requests with respect to medical diagnoses, severity of the loss of autonomy, and extent of the health care needs [31].
Sample
In the original study, individuals were eligible if they were aged 65 or over and living in long-term care facilities, notably nursing homes, in the Quebec City area. Residents in six long-term care units within a large-scale psychiatric institution (n = 301) were excluded from this analysis (n = 2 332), as most of these residents had a life-long history of a psychiatric illness, usually used more psychotropic drugs and displayed AB in a context quite different from that of the average older patient in a long-term care facility.
Data Collection
Original data were collected from two sources as follows: first, structured simultaneous interviews with two nurses who were familiar with the residents in their health-care unit, and second, a systematic review of the medical files by research assistants. This strategy aimed at minimizing the risk that research assistants would influence the course of the interviews with nurses (information bias). Individual variables included in this study are: age, sex, length of stay, functional autonomy, cognitive status, psychological distress, isolation and withdrawal behaviour, sleep disturbance. Environmental variables are: social visitation, physical restraint and neuroleptic and benzodiazepine drugs.
Variables measured during the interviews with nurses
Dependant variable: Aggressive behaviour
Types of AB were evaluated using the validated French version of the Cohen Mansfield Agitation Inventory (CMAI) [32,33], an instrument widely used in the field of psychogeriatrics. This instrument measures 29 disruptive behaviours classified into four groups: 1) aggressive physical behaviour, 2) non-aggressive physical behaviour, 3) aggressive verbal behaviour, and 4) non-aggressive verbal behaviour. These behaviour types are rated on the basis of observations gathered during the two weeks preceding data collection. In this analysis, we focused solely on two dimensions of the scale, namely, aggressive physical behaviour and aggressive verbal behaviour. If participants showed at least one instance of aggressive physical behaviour, they were classified as displaying said behaviour. Likewise, the older person who evinced aggressive verbal behaviour was classified as displaying that behaviour. In both cases, if the older persons showed no aggressive behaviour during the 2 weeks preceding the interview, they were classified as not displaying AB. Psychometric properties of the French adaptation of the CMAI have been previously studied and qualified as good: interrater reliability (r = 0.72; p < 0.05), temporal stability (r = 0.72; p < 0.05), internal consistency (Cronbach's alpha varying between 0.75 and 0.77; p < 0.05), concomitant reliability (r = 0.74; p < 0.05) and construct validity [34].
Independent variables
The MOSES (Multidimensional Observation Scale for Elderly Subjects) is an instrument that provides an overall evaluation of older adults. It consists of 40 closed questions regarding quantified observations carried out in the week preceding the data collection in five domains: (1) functional autonomy; (2) cognitive status; (3) psychological distress; (4) disruptive behaviour; (5) isolation and withdrawal behaviour. In this study, four domains (1, 2, 3 and 5) were used. We did not use the disruptive behaviour subscale because we preferred to use a better known scale for this variable (the CMAI, see above), since its extensive use offers excellent comparability. The MOSES was developed and validated with a sample of 2,391 patients in long-term care, aged 65 and over [35]. The MOSES scale has an internal consistency of 0.80 (p < 0.05). The internal consistencies of each domain of the scale are also satisfactory: functional autonomy 0,81, cognitive status 0,86, psychological distress 0,79, disruptive behaviours 0,78, isolation and withdrawal behaviour 0,77. Correlations with the Zung Depression, Robertson Short Mental Status, Kingston Dementia and the Physical and Mental Impairment-of-function Evaluation scales have confirmed the validity of the MOSES scale [35].
All of the MOSES variables were categorized by two clinician-researchers (PV, RV) according to two criteria. First, we looked at their clinical relevance and meaning. We wanted to ensure that any clinician could easily distinguish the categories created. Second, we looked at the mean, median and variance of data for each variable to be categorized. We wanted to ensure that the created categories would include a similar number of subjects or a sufficient number of subjects in each category. For instance, autonomous or semi-autonomous residents have to be easily distinguished from totally dependent residents by any health care provider.
(1) Functional autonomy is a composite variable integrating six items from the MOSES scale: dressing, bathing, aesthetic care, use of toilets, physical mobility, and getting in and out of bed. Each item was assigned one of four ratings, from 1 (entirely independent) to 4 (entirely dependent). For each resident, the ratings for the six items were added up to constitute a functional autonomy score varying between 6 and 24. For the study analyses, the total score was dichotomised in autonomous/semi-autonomous (rating of 6 to 20), and dependent (rating of 21 to 24).
(2) Cognitive status is a composite variable based on seven items from the MOSES scale: understanding communication, talking, recognizing personnel, perception of place, perception of time, memory of recent events, and memory of important past events. Each item was assigned one of four ratings, from 1 (no impairment) to 4 (severe impairment), and summed together for a total score varying from 7 to 28. For analysis, the score was divided into three categories: a) no cognitive impairment (rating of 7), b) mild-moderate cognitive impairment (rating between 8 and 21), c) severe cognitive impairment (rating between 22 and 28).
(3) Psychological distress (i.e. the combination of symptoms of depression and anxiety) was assessed by the following seven MOSES items: looking sad and depressed, reporting sadness and depression, sounding sad and depressed, looking worried and anxious, reporting worry and anxiety, crying, and pessimism about the future. Each item was rated (1 = never; 2 = sometimes; 3 = often; 4 = always). Participants were identified as psychologically distressed if they either often or always displayed at least one symptom of psychological distress during the previous week.
(5) Isolation and withdrawal behaviour were assessed usingeight items from the MOSES scale: prefers solitude, initiates social contacts, responds to social contacts, maintains friendships with other residents, shows interest in daily and external events, keeps busy, and helps other residents. Each of these items was rated from 1 (socially active) to 4 (socially isolated) for a total score varying from 8 to 32 for each participant. This variable was dichotomised: No, if the older person was socially active or slightly active (rating from 8 to 23) and Yes, if the older person was socially isolated or slightly isolated (rating from 24 to 32). Following variables were not measured by the MOSES.
Social visitation
The number of hours of visits by family and friends received by older adults on a monthly basis was calculated by averaging the number of hours of visits over the previous 12 months. This scale has been developed by our team.
Use of physical restraints during the 24 hours preceding the interview
Physical restraints include ties, straps or belts (which can be tied to the legs, ankles, arms or waist), jackets, gloves, geriatric chairs equipped with security tables, or other devices designed to limit the mobility of the older person and over which he or she has no control. Bedrails, half doors and locked doors forming a barrier or obstacle to keep the older person in a given area, were not considered as physical restraints. This scale has very good validity and reliability [36].
Sleep disturbance was determined by the presence of four symptoms of sleeping problems during the previous week. Two nurses were asked to use a Likert-type scale (1 = never to 4 = always) to rate whether the subject a) had trouble falling asleep, b) woke up and had trouble falling back to sleep during the night, c) woke up too early in the morning, and d) did not appear rested in the morning. According to DSM-IV-R [37], a subject is diagnosed with sleep disturbance if he or she displays difficulty initiating or maintaining sleep, or displays non-restorative sleep and it causes significant distress or impairment in daytime functioning. For the study, we classified participants as having a sleep disturbance if there was evidence of symptom (d) (either often or always) and one of the remaining symptoms (a, b, c). In other words, subjects were considered to have sleep disturbance if they had one of these combinations: (a + d), or (b + d), or (c + d). This scale was developed by our team for this study.
Variables measured by review of medical files
A structured questionnaire was used to collect information on: (1) socio-demographic characteristics of participants: age, gender, and length of stay in the care unit, (2) use of benzodiazepines and neuroleptics. To be considered consumers of benzodiazepine, participants had to have a regular prescription for or have consumed an as-needed (PRN) dose of this drug during the previous week. The identical procedure was followed for neuroleptics. The medication was coded according to the Anatomic, Therapeutic, Chemical (ATC) classification system [38]. Benzodiazepines which includes the short-acting and long-acting forms, are coded NO5BA in this system. Neuroleptics, both conventional and atypical, are coded NO5A.
The protocol of this study was approved by the Laval University Research Ethics Board.
Statistical analysis
Characteristics of the participants and ABs (verbal, physical and both) were described by frequencies and percentages. Bivariate analyses to determine crude odds ratios were used to evaluate the degree of association between independent variables and ABs (verbal, physical and both). Statistically significant variables in bivariate analyses were then assessed for multicolinearity according to the method outlined by Besley and colleagues [39]. The isolating and withdrawal behaviours and the functional autonomy variables correlated strongly with cognitive impairment and were discarded in the final model. Logistic regression analysis was used to examine the contribution of each independent variable to ABs (verbal, physical and both). All analyses were carried out using the Statistical Analysis System (SAS) software, version 8.0.
Results
Among the 2 332 residents who were included in this analysis, 494 (21.2%) displayed physical aggressive behaviour (PAB), 497 (21.5%) verbal aggressive behaviour (VAB) and 258 patients (11.2%) displayed both behaviours (BAB) (see Tables 1, 2, 3). Among all PAB measured in this study, hitting, pushing and kicking are the most common (Table 2). Among VAB, verbal aggression or insults were the most frequent (Table 3).
Table 1 Factors associated with both forms of aggressive behaviour (verbal and physical).
Characteristic Total Both forms of Aggressive behaviour Bivariate analyses Regression analyses
N: 2309 (23 miss.) Yes n:258 (11.2%) No n: 2051 (88.8%) Crude odds ratio (95% confidence interval) Adjusted odds ratio (95% confidence interval)
Individual factors
Age (years)
≥ 85 ‡ 1063 (45.6%) 110 (42.6%) 946 (46.1%) 1.00 1.00
75 to 84 893 (38.3%) 118 (45.7%) 764 (37.3%) 1.33* (1.01–1.75) 1.22 (0.91–1.63)
65 to 74 376 (16.1%) 30 (11.6%) 341 (16.6%) 0.76 (0.50–1.15) 0.61 (0.39–0.96)
Gender
Female ‡ 1759 (75.4%) 165 (63.9%) 1579 (77.0%) 1.00 1.00
Male 573 (24.6%) 93 (36.1%) 472 (23.0%) 1.89* (1.43–2.48) 2.13* (1.58–2.86)
Length of stay in LTCC (years)
0 to 2 ‡ 1093 (46.9%) 124 (48.1%) 951 (46.34%) 1.00
3 to 4 432 (18.5%) 53 (20.5%) 375 (18.3%) 1.08 (0.77–1.53)
≥ 5 807 (34.6%) 81 (31.4%) 725 (35.4%) 0.86 (0.64–1.15)
Functional autonomy
Autonomous / semi-autonomous‡ 1207 (51.9%) 122 (47.3%) 1076 (52.5%) 1.00
Dependent 1117 (48.1%) 136 (52.7%) 974 (47.5%) 1.23 (0.95–1.60)
Cognitive status
No impairment ‡ 471 (20.3%) 15 (5.8%) 454 (22.2%) 1.00 1.00
Mild-moderate impairment 847 (36.5%) 91 (35.3%) 749 (36.5%) 3.69* (2.11–6.44) 2.87* (1.62–5.09)
Severe impairment 1006 (43.3%) 152 (58.9%) 847 (41.3%) 5.44* (3.16–9.37) 3.77* (2.12–6.72)
Withdrawal behaviour
No ‡ 1219 (52.5%) 98 (37.9%) 1111 (54.2%) 1.00
Yes 1105 (47.6%) 160 (62.0%) 939 (45.8%) 1.93* (1.48–2.52)
Sleep disturbance
No ‡ 2180 (93.8%) 230 (89.1%) 1936 (94.4%) 1.00 1.00
Yes 144 (6.2%) 28 (10.8%) 114 (5.6%) 2.07* (1.34–3.20) 1.95* (1.23–3.08)
Psychological distress
No ‡ 1838 (78.8%) 187 (72.5%) 1631 (79.5%) 1.00 1.00
Yes 494 (21.2%) 71 (27.5%) 420 (20.5%) 1.47* (1.10–1.98) 1.36* (1.00–1.85)
Environmental factors
Use of benzodiazepine drugs
No ‡ 1344 (57.6%) 136 (52.7%) 1195 (58.3%) 1.00
Yes 988 (42.4%) 122 (47.3%) 856 (41.7%) 1.25 (0.97–1.62)
Use of neuroleptic drugs
No ‡ 1683 (72.2%) 138 (53.5%) 1525(74.35%) 1.00 1.00
Yes 649 (27.8%) 120 (46.5%) 526 (25.6%) 2.52* (1.94–3.28) 2.12* (1.61–2.81)
Social visitation.
0 to 3 hours ‡ 777 (33.6%) 99 (38.4%) 671 (32.9%) 1.00 1.00
4 to 15 hours 736 (31.8%) 84 (32.6%) 649 (31.8%) 0.89 (0.65–1.21) 0.94 (0.68–1.29)
≥ 16 hours 803 (34.7%) 75 (29.1%) 722 (35.4%) 0.71* (0.52–0.98) 0.90 (0.64–1.26)
Use of physical restraints
No ‡ 1576 (67.6%) 137 (53.1%) 1421 (69.3%) 1.00 1.00
Yes 756 (32.4%) 121 (46.9%) 630 (30.7%) 1.99* (1.53–2.59) 1.43* (1.07–1.92)
* Statistically significant p < 0.05
p‡ Reference Category
miss. = missing
Table 2 Types and frequencies of physical aggressive behaviours.
Physical aggressive behaviours Frequency (percent %) n = 2332 (22 miss.)
Hitting
Never 2022 (87.5%)
Rarely 105 (4.5%)
Sometimes 79 (3.4%)
Often 58 (2.5%)
Always 46 (2%)
Pushing
Never 2120 (91.8%)
Rarely 53 (2.3%)
Sometimes 55 (2.4%)
Often 36 (1.5%)
Always 46 (2%)
Kicking
Never 2157 (93.4%)
Rarely 52 (2.2%)
Sometimes 44 (1.9%)
Often 28 (1.2%)
Always 29 (1.3%)
Scratching
Never 2216 (95.9%)
Rarely 32 (1.4%)
Sometimes 27 (1.2%)
Often 15 (0.6%)
Always 20 (0.9%)
Spitting
Never 2235 (96.8%)
Rarely 23 (1%)
Sometimes 15 (0.6%)
Often 7 (0.3%)
Always 30 (1.3%)
Throwing things
Never 2236 (96.8%)
Rarely 36 (1.5%)
Sometimes 20 (0.9%)
Often 8 (0.3%)
Always 10 (0.4%)
Biting
Never 2270 (98.3%)
Rarely 20 (0.9%)
Sometimes 7 (0.3%)
Often 5 (0.2%)
Always 8 (0.3%)
Hurting self
Never 2292 (99.2%)
Rarely 8 (0.3%)
Sometimes 3 (0.1%)
Often 3 (0.1%)
Always 4 (0.2%)
Intentional falling
Never 2295 (99.3%)
Rarely 9 (0.4%)
Sometimes 3 (0.1%)
Often 1 (0.04%)
Always 2 (0.1%)
Table 3 Types and frequencies of verbal aggressive behaviours
Verbal aggressive behaviours Frequency (%) n = 2332 (12 miss.)
Verbal aggression or insult
Never 1902 (81.9%)
Rarely 141 (6.1%)
Sometimes 109 (4.7%)
Often 86 (3.7%)
Always 82 (3.5%)
Verbal threat
Never 2072 (89.3%)
Rarely 76 (3.3%)
Sometimes 76 (3.3%)
Often 48 (2.1%)
Always 48 (2.1%)
Verbal sexual advances
Never 2303 (99.3%)
Rarely 8 (0.3%)
Sometimes 5 (0.2%)
Often 4 (0.2%)
Always 0 (0%)
miss. = missing
In multivariate analyses, the individual factors significantly associated with BAB are: male gender (OR = 2.13), mild-moderate and severe cognitive impairment (respectively OR = 2.87 and 3.77), sleep disturbance (OR = 1.95), and psychological distress (OR = 1.36). Environmental factors associated with BAB are: neuroleptic drug use (OR = 2.12) and physical restraints (OR = 1.43) (Table 1).
About one fifth (21.2%) of older adults in long-term care displayed PAB. In the multivariate analysis (Table 4), individual factors associated with PAB are as follows: aged over 74 years (OR = 0.69), male gender (OR = 2.04), mild-moderate or severe cognitive impairment (respectively OR = 2.59 and 8.26), sleep disturbance (OR = 2.03) and psychological distress (OR = 1.31). Neuroleptic drug use (OR = 1.74) is the only environmental factor associated with PAB.
Table 4 Factors associated with physical aggressive behaviours.
Characteristic Aggressive physical behaviours Bivariate analyses Regression analyses
Yes n: 494 (21.2%) No n: 1816 (77.8%) Crude odds ratio (95% confidence interval) Adjusted odds ratio (95% confidence interval)
Individual factors
Age (years)
≥ 85 ‡ 228 (46.2%) 829 (45.6%) 1.00 1.00
75 to 84 203 (41.1%) 679 (37.4%) 1.09 (0.88–1.35) 1.06 (0.83–1.34)
65 to 74 63 (12.8%) 308 (16.9%) 0.74 (0.55–1.01) 0.69* (0.49–0.98)
Gender
Female ‡ 339 (68.6%) 1406 (77.4%) 1.00 1.00
Male 155 (31.4%) 410 (22.6%) 1.57* (1.26–1.95) 2.04* (1.59–2.62)
Length of stay in LTCC (years)
0 to 2 ‡ 225 (45.6%) 851 (46.9%) 1.00
3 to 4 101 (20.5%) 327 (18.0%) 1.17 (0.89–1.53)
≥ 5 168 (34.0%) 638 (35.1%) 0.99 (0.80–1.25)
Functional autonomy
Autonomous / semi-autonomous‡ 187 (37.8%) 1012 (55.7%) 1.00 1.00
Dependent 307 (62.12%) 803 (44.3%) 2.07* (1.69–2.54) 0.79 (0.60–1.05)
Cognitive status
No impairment ‡ 23 (4.7%) 446 (24.6%) 1.00 1.00
Mild-moderate impairment 119 (24.1%) 722 (39.9%) 3.20* (2.02–5.08) 2.59* (1.61–4.17)
Severe impairment 352 (71.3%) 647 (35.7%) 10.57* (6.82–16.39) 8.26* (5.13–13.30)
Withdrawal behaviour
No ‡ 150 (30.4%) 1060 (58.4%) 1.00
Yes 344 (69.9%) 755 (41.6%) 3.22* (2.60–3.99)
Sleep disturbance
No ‡ 446 (90.3%) 1721 (94.8%) 1.00 1.00
Yes 48 (9.7%) 94 (5.2%) 1.97* (1.37–2.83) 2.03* (1.35–3.04)
Psychological distress
No ‡ 367 (74.3%) 1452 (79.9%) 1.00 1.00
Yes 127 (25.7%) 364 (20.0%) 1.38* (1.09–1.74) 1.31* (1.02–1.69)
Environmental factors
Use of benzodiazepine drugs
No ‡ 286 (57.9%) 1046 (57.6%) 1.00
Yes 208 (42.1%) 770 (42.4%) 0.99 (0.81–1.21)
Use of neuroleptic drugs
No ‡ 285 (57.7%) 1379 (75.9%) 1.00 1.00
Yes 209 (42.3%) 437 (24.1%) 2.32* (1.88–2.85) 1.74* (1.38–2.19)
Social visitation.
0 to 3 hours ‡ 198 (40.2%) 572 (31.6%) 1.00 1.00
4 to 15 hours 159 (32.3%) 574 (31.7%) 0.81 (0.64–1.03) 0.94 (0.68–1.29)
≥ 16 hours 136 (27.6%) 662 (36.6%) 0.60* (0.47–0.77) 0.90 (0.64–1.26)
Use of physical restraints
No ‡ 241 (48.8%) 1318 (72.6%) 1.00 1.00
Yes 253 (51.2%) 498 (27.4%) 2.78* (2.27–3.41) 1.79 (1.37–2.33)
* Statistically significant p < 0.05
‡ Reference Category
miss. = missing
One fifth of the subjects under study (21.5%) displayed VAB. Table 5 shows the individual factors linked to VAB to be: male gender (OR = 1.64), functional dependency (OR = 0.78), mild-moderate or severe cognitive impairment (respectively OR = 1.85 and 1.48), and sleep disturbance (OR = 1.76). Environmental factors for VAB are benzodiazepine and neuroleptic drug use (respectively OR = 1.28 and 1.62). A summary of the factors associated with BAB, PAB, and VAB in the study is provided in Table 6.
Table 5 Factors associated with verbal aggressive behaviours.
Characteristic Aggressive verbal behaviours Bivariate analyses Regression analyses
Yes n: 497 (21.5%) No n: 1823 (78.9%) Crude odds ratio (95% confidence interval) Adjusted odds ratio (95% confidence interval)
Individual factors
Age (years)
≥ 85 ‡ 187 (37.6%) 870 (47.7%) 1.00 1.00
75 to 84 220 (44.3%) 668 (36.6%) 1.53* (1.23–1.91) 1.40 (1.12–1.76)
65 to 74 90 (18.1%) 285 (15.6%) 1.47* (1.11–1.95) 1.18 (0.87–1.60)
Gender
Female ‡ 329 (66.2%) 1422 (78.0%) 1.00 1.00
Male 168 (33.8%) 401 (22.0%) 1.81* (1.46–2.25) 1.64* (1.30–2.06)
Length of stay in LTCC (years)
0 to 2 ‡ 237 (47.7%) 845 (46.4%) 1.00
3 to 4 96 (19.3%) 336 (18.4%) 1.02 (0.78–1.33)
≥ 5 164 (33.0%) 642 (35.2%) 0.91 (0.73–1.14)
Functional autonomy
Autonomous / semi-autonomous‡ 282 (56.8%) 923 (50.7%) 1.00 1.00
Dependent 214 (43.1%) 899 (49.3%) 0.78* (0.64–0.95) 0.78* (0.62–0.99)
Cognitive status
No impairment ‡ 76 (15.3%) 395 (21.7%) 1.00 1.00
Mild-moderate impairment 221 (44.5%) 624 (34.3%) 1.82* (1.37–2.43) 1.85* (1.37–2.50)
Severe impairment 199 (40.1%) 803 (44.1%) 1.27 (0.95–1.70) 1.48* (1.06–2.07)
Withdrawal behaviour
No ‡ 265 (53.4%) 951 (52.2%) 1.00
Yes 231 (46.6%) 871 (47.8%) 0.95 (0.78–1.16)
Sleep disturbance
No ‡ 449 (90.5%) 1725 (94.7%) 1.00 1.00
Yes 47 (9.5%) 97 (5.3%) 1.86* (1.29–2.68) 1.76* (1.21–2.56)
Psychological distress
No ‡ 373 (75.1%) 1454 (79.7%) 1.00 1.00
Yes 124 (24.9%) 369 (20.2%) 1.31* (1.04–1.65) 1.19 (0.93–1.51)
Environmental factors
Use of benzodiazepine drugs
No ‡ 261 (52.5%) 1075 (58.9%)) 1.00 1.00
Yes 236 (47.5%) 748 (41.0%) 1.30* (1.07–1.59) 1.28* (1.05–1.58)
Use of neuroleptic drugs
No ‡ 313 (62.9%) 1359 (74.5%) 1.00 1.00
Yes 184 (37.0%) 464 (25.4%) 1.72* (1.40–2.13) 1.62* (1.30–2.02)
Social visitation.
0 to 3 hours ‡ 167 (33.7%) 606 (33.4%) 1.00
4 to 15 hours 176 (35.5%) 560 (30.9%) 1.15 (0.91–1.46)
≥ 16 hours 153 (30.8%) 648 (35.7%) 0.86 (0.68–1.11)
Use of physical restraints
No ‡ 325 (65.4%) 1242 (68.1%) 1.00
Yes 172 (34.6%) 581 (31.9%) 0.13 (0.92–1.40)
* Statistically significant p < 0.05
‡ Reference Category
miss. = missing
Table 6 Summary of factors associated with aggressive behaviours according to multivariate analysis.
Associated risk factors BAB PAB VAB
Individual factors
Age (years) 85 and over +
Male gender + + +
Functionally independent +
Mild-moderate or severe cognitive impairment + + +
Sleep disturbance + + +
Psychologically distressed + +
Environmental factors
Use of benzodiazepine drugs +
Use of neuroleptic drugs + + +
Use of physical restraints +
+ = statistically significant
Discussion
Prevalence of AB
The first goal of this study was to determine the prevalence of verbal and physical aggressive behaviour among older residents in long-term care settings in the Quebec City area. In this study, aggressive behaviour (either physical or verbal) was displayed by 21% of the older residents and 11.2% of them exhibited both forms. In other studies, Cohen-Mansfield et al. [3] reported a prevalence rate for AB of 8 to 91% in institutional settings, Lyketsos and colleagues [40] found a prevalence rate of 23.7% for aggressive/agitated behaviour (Neuropsychiatric Inventory) among community-dwelling seniors and long term care residents, and Schreiner [23] reported a prevalence rate of 45.4% of physical or verbal aggressive behaviour (CMAI; last two weeks) among 391 cognitively impaired long term care residents. Marx et al. [6] reported a lower rate of 32% for aggressive behaviour (physical and verbal) in long-term care facilities (CMAI; last two weeks) and lastly, Giancola et al. [4] reported prevalence rates of 14 to 21% for physical aggressive behaviour and 10–14% for verbal aggressive behaviour in two nursing homes. While our results are in general agreement with these reported prevalence rates there is, nevertheless, substantial discrepancy between prevalence rates among studies. This is probably due in part to their use of different measures of AB [23,40]. However we also cannot ignore the fact that some differences may be due to the varying quality of care provided in those long-term care facilities or to the inclusion criteria applied in the studies.
In our study, we found that 21% of older residents in long-term care settings displayed physical or verbal aggressive behaviour, confirming the significance of the phenomenon and the need to address it. Based on both their high frequency and their potentially distressing effect on both the resident and the caregivers, it appears that some specific behaviours, such as hitting and insults, deserve more attention from researchers. Such behaviour types have also been highlighted by researchers in previous studies [10,23].
Although there is a relatively high prevalence of ABs (21%), the majority are not displayed often. Indeed, less than 3% are displayed "often" or "always". We understand from these low frequency levels that, in general, older residents do not favour one behaviour over another when exhibiting AB. There is also the possibility that specific contexts of care giving are more conducive to aggressive behaviour. However we were not able to take this into account in our study. For instance, AB during bathing has previously been observed in half of older patients with dementia [41], an observation that has led researchers to target specific care-giving contexts to tackle the problem among older patients with dementia. Sloane et al. [42] tested two experimental bathing interventions compared to a control group leading to impressive results. When compared to the control group, aggressive behaviour in the person-centred shower group and in the towel-bath group declined significantly (53% and 60% respectively). This would appear a likely avenue for future research, to target specific contexts where AB is more likely to occur in order to develop intervention suited to those contexts. Another study [43] was successful in decreasing agitation among older residents in institutions by targeting many factors and contexts (see Clinical implications). These results suggest that a broad approach to nursing care is also of value in the prevention of AB in long-term care facilities and that specific interventions should be developed targeting those care activities at increased risk for AB.
Our second study goal was to identify the factors associated with BAB, PAB and VAB. Overall, they were quite similar. This is somewhat in line with the results of previous research studies, although they yielded opposite results [27,28]. Ryden et al. [27] found that among 116 residents, the use of psychotropic drugs, physical restraint and living on secured units were differently associated with VAB and PAB. In our study, use of benzodiazepine drugs was associated with VAB only and physical restraint only with BAB. It should be noted that Ryden's study focused exclusively on residents with frequent aggressive behaviours (9–10 aggressive behaviours a day, using the Ryden Aggression Scale-2), which is higher than the AB frequency in our study population. Using the CMAI, Schreiner [23] reported a higher prevalence of PAB among men than women (p = 0.05; n = 391), but did not observe such a difference for VAB, whereas we found men exhibited more PAB and VAB than did women. By examining patterns of co-occurrence of both ABs among residents (n = 240; 98% men; 32% had a psychiatric diagnosis), Souder et al [28] found PABs more likely to occur with non-aggressive physical behaviour, and unlikely to occur with verbally disruptive behaviours. These researchers did not look at other risk factors for AB. In our study, 11% of participants displayed BAB.
As can be concluded from these studies, much more research is needed in the field to determine whether the factors associated with BAB, PAB and VAB differ. Improved study comparability in the future requires research studies of comparable design and instruments. At present, differences in conclusions among studies could result from methodological differences, such as the instruments used, the Ryden Aggression Scale-2 [27], the CMAI [23], or the disruptive behaviour scale [28]. Nonetheless, certain results from our findings do deserve further attention. They are discussed below.
AB and cognitive impairment
As shown in Tables 1, 4 and 5, AB was more likely to occur among older participants with mild-moderate or severe cognitive impairment than among those with no cognitive impairment. These findings are in agreement with those of Ryden et al. [27] and Marx et al. [6]. Menon et al. [26] also indicated that physical and verbal aggression increases with the severity of the cognitive impairment. According to Hall and O'Connor's literature review [19], the association between severity of cognitive impairment and aggressive behaviour has received strong support and may be caused by the communication deficiencies accompanying severe cognitive impairment.
This association between AB and cognitive impairment also provides support for the Progressively Lowered Stress Threshold (PLST) model [44,45], according to which, a person with dementia has a declining ability to adjust to environmental demands as the cognitive losses progress. Demands not adapted to the resident's stress threshold become stressors directed at the resident, leading to the development of anxiety symptoms. If the caregiver does nothing to reduce these stressors, the resident will then display agitation such as verbal or physical aggressive behaviours. This model is useful for explaining the association between severe cognitive impairment and aggression and as such, it warrants further attention from researchers and clinicians since it brings insight to our understanding of AB [45]. Said model has also been found to improve the quality of care provided by caregivers in the community [46,47] and therefore it would be important to also test its usefulness among older residents with severe cognitive impairment in long-term care facilities.
An interesting outcome of our study not in accordance with the PLST model is the non-linear association between cognitive impairment and VAB. According to the PLST, participants with severe cognitive impairment would have displayed more VAB than those with mild-moderate cognitive impairment. Our results in fact, showed that older residents with mild-moderate cognitive impairment were more likely to display VAB than those with severe cognitive impairment. Matteau et al. [25] explored the relationship between language deterioration and disruptive vocalization in demented residents (Alzheimer's, vascular or mixed type) living in nursing homes. They showed that those with language deficiencies were more likely to display frequent verbal behaviour in a large variety of distinct forms. Thus, disruptive vocalization could be a consequence of communicative difficulties [14,24,48]. Therefore, it is possible that among residents with severe cognitive impairment, their language limitations are so severe that they can no longer express their concerns through verbal agitation, including VAB. On the other hand, those residents with mild-moderate cognitive impairment, who possess residual verbal capacity, would be more prone to display their concerns via VAB, as was suggested by our results. As put forward by Hall and O'Connor [19], the association between communication impairment and VAB has implications for communication skill training for nursing staff.
AB and psychotropic drug use
The use of neuroleptics was significantly associated with PAB, and both benzodiazepine and neuroleptics were linked to VAB. Other studies [7,10,14,20,40] have found AB and psychotropic drugs such as neuroleptics and benzodiazepines to be associated. Conventional (e.g. haloperidol, thioridazine, chlorpromazine) and atypical (e.g. risperidone, olanzapine, quetiapine) neuroleptics are frequently used in the treatment of agitation and AB among older residents. Three meta-analyses concluded that, despite their wide use, neuroleptics might reduce the frequency of disruptive behaviour by only 18% [49] to 26% in older patients with dementia [50]. Lonergan and colleagues [51] report that haloperidol is not more effective than a placebo in controlling agitation among the elderly suffering from dementia and was slightly more effective than a placebo for AB. In addition, these drugs are associated with frequent adverse effects such as extra-pyramidal symptoms, drowsiness and anticholinergic manifestations [52,53]. According to a one-year longitudinal study [54], researchers reported that change in disruptive behaviour occurs among nursing home residents regardless of the use of neuroleptic drugs, but that it occurs more frequently among those receiving neuroleptic medication. In fact, users of neuroleptics showed greater changes in both developing and resolving disruptive behaviour during the year than those not receiving the drugs. In short, given their limited effectiveness and the high risk of side effects, including permanent consequences such as tardive dyskinesia, the use of neuroleptic drugs should be a last resort
Benzodiazepines should also be used with caution among older residents. Liebson [9] reported that these drugs exacerbate cognitive deficits in cognitively impaired residents. Finally, we are not aware of any controlled clinical trial on the effectiveness of benzodiazepine for treating aggressive behaviour among older residents with dementia. It is worth mentioning that since sleep disturbance and psychological distress (both associated with AB) are indications for the use of benzodiazepine drugs, there are, undoubtedly, a certain percentage of participants who were taking these drugs to treat said symptoms and not AB. Nonetheless, future studies should be directed toward the development of new drugs and alternative treatments such as the towel bath [42] or staff training programs [55].
AB and physical restraints
Physical restraints are sometimes used in an attempt to control aggressive or other risky behaviours. Their use results in loss of personal autonomy and self-esteem, and may lead to AB among residents. As found in a previous study [27], physical restraint (an environmental factor) was associated with AB among older residents. Tinetti et al. [56] reported that disruptive behaviour (which includes AB) was the reason most often cited for nursing staff's resorting to a physical restraint. However, physical restraint is not a solution for such behaviour [11]. Based on the reactance theory, we even suggest that physical restraint increases AB. Reactance theory, as proposed by Brehm, suggests that individuals pursue freedom and want control over their lives [57]. Any attempt to remove this sense of control or freedom from an individual will result in defensive behaviour. The removal of fundamental rights for a long period can lead to aggressive behaviour [58]. Therefore, one of the first interventions to be applied in the context of AB would be to reduce the use of physical restraints in long-term care settings.
Clinical implications
Based on our results, we would like to suggest a preventive intervention program for AB inspired by the work of Inouye in the field of delirium [59]. Inouye et al. have been able to reduce the prevalence of delirium by implementing preventive interventions for every factor associated with delirium (e.g.: dehydration, malnutrition, sleep disturbance, hearing impairment, physical restraint, etc.). In the same way, it might be useful to adopt this approach for AB, since it too is associated with several factors. However a future study would need to test any such program to determine its relevance and effectiveness. According to our results, a multi-component intervention program would target the following five factors in order to reduce AB among older residents in long-term care facilities: cognitive impairment, sleep disturbance, psychological distress, benzodiazepine and neuroleptic drugs and physical restraints. Cognitive impairment, although it cannot be cured, can be alleviated somewhat through the use of an appropriate communication method. When the approach selected is suitable (validation therapy, reality orientation, reminiscence, etc.), it may reduce the occurrence of aggressive behaviours.
These multiple targets might discourage even the most well-intentioned caregivers and while it may appear that our recommendations are heavy artillery for treating AB with little real clinical application, a recent intervention study on agitation in long-term care [43] has been successful in targeting multiple risk factors for agitation, while remaining realistic in terms of clinical practice. We are of the opinion that this intervention, entitled BACE (Balancing Arousal Controls Excesses), appears promising, at least with regard to its approach to behavioural problems.
Limits and contribution of the study
This study has some limitations. First, since it was cross-sectional in nature, the findings cannot be regarded as providing a cause-effect relationship. Second, aggressive behaviour, like all human behaviour, is influenced by individual (e.g. Parkinson's Disease, stroke) and environmental factors (e.g. staff's approach, environmental changes, quality of medical and nursing care), all of which could not be included in this study. Third, this study did not collect data on pain, a factor related to AB among long-term care residents. Lastly, this research is based on staff-reported data. The fact that the staff members in the institutions were very busy and had limited time available for the study may have affected the quality of the data collected. However, to reduce the impact of any potential bias, we interviewed two nurses simultaneously. Nevertheless, this work has two important strengths related to comprehensiveness: It should be noted in particular that our study population comprised all the residents in all the long-term care facilities in the Quebec City area (except for those in specialized psychiatric settings), and this resulted in a large sized sample. Finally, this study has made an important contribution by differentiating VAB and PAB when conducting analyses.
Conclusion
Findings of the study suggest that overall, AB is associated with many individual factors (younger age, male gender, functional dependency, cognitive impairment, sleep disturbance, psychological distress) and environmental factors (benzodiazepine and narcoleptic drug use, physical restraints). Future prevention and treatment studies on AB are encouraged to pay attention to these factors and to be multi-dimensional in nature so as to better reflect our understanding of their association with AB. Reactance theory and the Progressively Lowered Stress Threshold model appear to us interesting frameworks that can improve nursing care in long-term care and reduce the prevalence and the burden of AB.
List of Abbreviations Used
MOSES: Multidimensional Observation Scale for Elderly Subjects
PLST: Progressively Lowered Stress Threshold model
AB: aggressive behaviour
ABs: the different manifestations of aggressive behaviour
PAB: physical aggressive behaviour
VAB: verbal aggressive behaviour
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
PV designed the secondary analysis study, supervised statistical analysis, outlined the first draft and edited subsequent drafts. RV participated in the design of the original study and edited the final draft. GA revised the literature review, wrote the first draft and edited subsequent drafts. JD, NC, AB edited first and subsequent drafts. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors thank Zhora Benoussissa for her statistical assistance and the Evaluation of long-term care interventions division of the Aging Research Network of the Quebec Health Research Fund for its financial support.
==== Refs
Patel V Hope RA A rating scale for aggressive behaviour in the elderly-the RAGE Psychol Med 1992 22 211 221 1574558
Chou KR Kaas MJ Richie MF Assaultive behavior in geriatric patients J Gerontol Nurs 1996 22 30 38 8954390
Cohen-Mansfield J Werner P Culpepper WJIII Wolfson MA Wandering and Aggression The Practical Handbook Of Clinical Gerontology 1996 California: Sage Publications 374 397
Giancola P Zeichner A Aggressive Behavior in Elderly: A Critical Review Clin Geronto 1993 13 3 21 10.1300/J018v13n02_02
Patel V Aggressive Behavior in elderly People with Dementia: A Review Int J Geriatr Psychiatry 1993 8 457 472 10.1002/gps.930080603
Marx M Cohen-Mansfield J Werner P A Profile Of The Aggressive Nursing Home Resident Behavior, Health, and Aging 1990 1 65 73
Talerico KA Evans LK Strumpf NE Mental health correlates of aggression in nursing home residents with dementia Gerontologist 2002 42 169 177 11914460
Brodaty H Low LF Aggression in the elderly J Clin Psychiatry 2003 64 36 43 12672263
Liebson E Aggression in Dementia Home Health Care Consultant 2001 8 26 31
Sourial R McCusker J Cole M Abrahamowicz M Agitation in demented patients in an acute care hospital: prevalence, disruptiveness, and staff burden Int Psychogeriatr 2001 13 183 197 11495393 10.1017/S1041610201007578
Voyer P Martin LS Improving geriatric mental health nursing care: making a case for going beyond psychotropic medications Int J Ment Health Nurs 2003 12 11 21 14685955 10.1046/j.1440-0979.2003.00265.x
Wancata J Benda N Meise U Windhaber J Non-cognitive symptoms of dementia in nursing homes: frequency, course and consequences Soc Psychiatry Psychiatr Epidemiol 2003 38 637 643 14614552 10.1007/s00127-003-0683-7
Camberg L Woods P Ooi WL Hurley A Volicer L Ashley J Odenheimer G McIntyre K Evaluation of Simulated Presence: a personalized approach to enhance well-being in persons with Alzheimer's disease J Am Geriatr Soc 1999 47 446 452 10203120
Draper B Snowdon J Meares S Turner J Gonski P McMinn B McIntosh H Latham L Draper D Luscombe G Case-controlled study of nursing home residents referred for treatment of vocally disruptive behavior Int Psychogeriatr 2000 12 333 344 11081953 10.1017/S1041610200006438
Ballard C Gray A Ayre G Psychotic symptoms, aggression and restlessness in dementia Rev Neurol (Paris) 1999 155 44 52
Rossby L Beck C Heacock P Disruptive behaviors of a cognitively impaired nursing home resident Arch Psychiatr Nurs 1992 6 98 107 1596118 10.1016/0883-9417(92)90005-4
Evers W Tomic W Brouwers A Effects of aggressive behavior and perceived self-efficacy on burnout among staff of homes for the elderly Issues Ment Health Nurs 2001 22 439 454 11885158 10.1080/01612840151136975
Chambers N We have to put up with it-don't we? The experience of being the registered nurse on duty, managing a violent incident involving an elderly patient: a phenomenological study J Adv Nurs 1998 27 429 436 9515657 10.1046/j.1365-2648.1998.00542.x
Hall KA O'Connor DW Correlates of aggressive behavior in dementia Int Psychogeriatr 2004 16 141 158 15318761 10.1017/S1041610204000286
Shah A Chiu E Ames D Harrigan S McKenzie D Aggression. Characteristics of aggressive subjects in Australian (Melbourne) nursing homes Int Psychogeriatr 2000 12 145 161 10937536 10.1017/S1041610200006281
Cox C Dealing with the aggressive nursing home resident J Gerontol Soc Work 1993 19 179 193
Algase D Beck C Kolanowski A Whall A Berent S Richards K Beattie E Need-driven dementia-compromised behavior: An alternative view of disruptive behavior Am J Alzheimers Dis 1996 11 10 19
Schreiner AS Aggressive behaviors among demented nursing home residents in Japan Int J Geriatr Psychiatry 2001 16 209 215 11241727 10.1002/1099-1166(200102)16:2<209::AID-GPS314>3.0.CO;2-9
Cohen-Mansfield J Werner P Marx MS The social environment of the agitated nursing home resident Int Journal Geriatr Psychiatry 1992 7 789 798 10.1002/gps.930071104
Matteau E Landreville P Laplante L Laplante C Disruptive vocalizations: a means to communicate in dementia? Am J Alzheimers Dis Other Demen 2003 18 147 153 12811989
Menon AS Gruber-Baldini AL Hebel R Kaup B Loreck D Zimmerman SI Burtoon L German P Magaziner J Relationship between aggressive behaviors and depression among nursing home residents with dementia Int J Geriatr Psychiatry 2001 16 139 146 11241718 10.1002/1099-1166(200102)16:2<139::AID-GPS284>3.0.CO;2-5
Ryden MB Feldt KS Oh HL Brand K Warne M Weber E Nelson J Gross C Relationships between aggressive behavior in cognitively impaired nursing home residents and use of restraints, psychoactive drugs, and secured units Arch Psychiatr Nurs 1999 13 170 178 10478495 10.1016/S0883-9417(99)80003-X
Souder E Heithoff K O'Sullivan PS Lancaster AE Beck C Identifying patterns of disruptive behavior in long-term care residents J Am Geriatr Soc 1999 47 830 836 10404927
Auger L Durand PJ Mercier P Paré L Étude des caractéristiques de la clientèle en liste d'attente pour le réseau institutionnel Conseil régional de la santé et des services sociaux Région 03 1989 Québec: Ministère de la santé et des services sociaux
Ministère de la santé et des services sociaux du Québec Le système de santé et des services sociaux au Québec, 2001 Québec: Gouvernement du Québec
Bernard S Étude descriptive de l'utilisation des contentions physiques auprès des personnes âgées hébergées en centres d'hébergements et de soins de longue durée 1999 Québec: Département de médecine sociale et préventive. Université Laval
Belanger M Agitation et atteinte cognitive: Étude transversale en établissement d'hébergement 1993 Université Laval, Département de Médecine sociale et préventive. Québec
Cohen-Mansfield J Marx MS Rosenthal AS A description of agitation in a nursing home J Gerontol 1989 44 77 84
Deslauriers S Landreville P Dicaire L Verreault R Validité et fidélité de l'inventaire d'agitation de Cohen-Mansfield Can J Aging 2001 20 373 384
Helmes E Csapo KG Short JA Standardization and validation of the Multidimensional Observation Scale for Elderly Subjects (MOSES) J Gerontol 1987 42 395 405 3598087
Laurin D Voyer P Verreault R Durand PJ Physical restraint use among nursing home residents: A comparison of two data collection methods BMC Nurs 2004 3 5 15488144 10.1186/1472-6955-3-5
American Psychiatric Association Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-Revised) Washington 2000
WHO Collaborating Center for Drug Statistics Methodology Anatomical Therapeutic Chemical (ATC) Classification Index Geneva 2000
Belsley DA Kuh E Welsch RE Regression diagnostics: identifying influential data and sources of collinearity New York 1980
Lyketsos CG Steinberg M Tschanz JT Norton MC Steffens DC Breitner JCS Mental and behavioral disturbances in dementia: Findings from the Cache County Study on Memory in Aging Am J Psychiatry 2000 157 708 714 10784462 10.1176/appi.ajp.157.5.708
Béland C Comportements aggressifs physiques associés à la démence: interactions entre les résidents et les soignants lors des soins d'hygiène Mémoire de maîtrise 2004 Université Laval, Faculté des études supérieures
Sloane PD Hoeffer B Mitchell CM McKenzie DA Barrick AL Rader J Stewart BJ Talerico KA Rasin JH Zink RC Koch GG Effect of person-centered showering and the towel bath on bathing-associated aggression, agitation, and discomfort in nursing home residents with dementia: a randomized, controlled trial J Am Geriatr Soc 2004 52 1795 1804 15507054 10.1111/j.1532-5415.2004.52501.x
Kovach CR Dohearty P Schlidt AM Cashin S Silva-Smith AL Effect of the BACE intervention on agitation of people with dementia Gerontologist 2004 44 797 806 15611216
Hall GR Buckwalter KC Progressively lowered stress threshold: a conceptual model for care of adults with Alzheimer's disease Arch Psychiatr Nurs 1987 1 399 406 3426250
Smith M Gerdner LA Hall GR Buckwalter KC History, development, and future of the progressively lowered stress threshold: a conceptual model for dementia care J Am Geriatr Soc 2004 52 1755 1760 15450057 10.1111/j.1532-5415.2004.52473.x
Gerdner LA Hall GR Buckwalter KC Caregiver training for people with Alzheimer's based on a stress threshold model Image J Nurs Sch 1996 28 241 246 8854546
Stolley JM Reed D Buckwalter KC Caregiving appraisal and interventions based on the progressively lowered stress threshold model Am J Alzheimers Dis Other Demen 2002 17 110 120 11954669
Ragneskog H Gerdner LA Josefsson K Kihlgren M Probable reasons for expressed agitation in persons with dementia Clin Nurs Res 1998 7 189 206 9633339 10.1111/j.1365-2702.1998.00147.x
Schneider LS Pollock VE Lyness SA A metaanalysis of controlled trials of neuroleptic treatment in dementia J Am Geriatr Soc 1990 38 553 563 1970586
Lanctot KL Best TS Mittman N Liu BA Oh PI Einarson TR Naranjo CA Efficacy and safety of neuroleptics in behavioral disorders associated with dementia J Clin Psychiatry 1998 59 550 561 quiz 562–563. 9818639
Lonergan E Luxenberg J Colford J Haloperidol for agitation in dementia Cochrane Database Syst Rev 2001 CD002852 11687166
Devanand DP Marder K Michaels KS Sackeim HA Bell K Sullivan MA Cooper TB Pelton GH Mayeux R A randomized, placebo-controlled dose-comparison trial of haloperidol for psychosis and disruptive behaviors in Alzheimer's disease Am J Psychiatry 1998 155 1512 1520 9812111
Brodaty H Ames D Snowdon J Woodward M Kirwan J Clarnette R Lee E Lyons B Grossman F A randomized placebo-controlled trial of risperidone for the treatment of aggression, agitation, and psychosis of dementia J Clin Psychiatry 2003 64 134 143 12633121
Burton LC Rovner BW German PS Brant LJ Clark RD Neuroleptic use and behavioral disturbance in nursing homes: a 1-year study Int Psychogeriatr 1995 7 535 545 8833277 10.1017/S1041610295002274
Ingersoll DB Schroepfer T Pryce J The effectiveness of a solution-focused approach for problem behaviors among nursing home residents J Gerontol Soc Work 1999 32 49 64 10.1300/J083v32n03_04
Tinetti ME Liu WL Marottoli RA Ginter SF Mechanical restraint use among residents of skilled nursing facilities. Prevalence, patterns, and predictors JAMA 1991 265 468 471 1985233 10.1001/jama.265.4.468
Fogarty JS Reactance theory and patient noncompliance Soc Sci Med 1997 45 1277 1288 9381240 10.1016/S0277-9536(97)00055-5
Meddaugh DI Reactance. Understanding aggressive behavior in long-term care J Psychosoc Nurs Ment Health Serv 1990 28 28 33 2110250
Inouye SK Bogardus ST Charpentier PA Leo-Summers L Acampora D Holford T Cooney LM A multicomponent intervention to prevent delirium in hospitalized older patients The New Engl J of Med 1999 340 669 76 10053175 10.1056/NEJM199903043400901
|
16280091
|
PMC1310539
|
CC BY
|
2021-01-04 16:30:32
|
no
|
BMC Geriatr. 2005 Nov 10; 5:13
|
utf-8
|
BMC Geriatr
| 2,005 |
10.1186/1471-2318-5-13
|
oa_comm
|
==== Front
PLoS GenetPLoS GenetpgenplgeplosgenPLoS Genetics1553-73901553-7404Public Library of Science San Francisco, USA 1635525210.1371/journal.pgen.001007005-PLGE-RA-0242R2plge-01-06-01Research ArticleBioinformatics - Computational BiologyEvolutionStatisticsGenetics/Population GeneticsHomo (Human)Clines, Clusters, and the Effect of Study Design on the Inference of Human Population Structure Clines, Clusters, and Human Population StructureRosenberg Noah A 1*Mahajan Saurabh 2Ramachandran Sohini 3Zhao Chengfeng 4Pritchard Jonathan K 5Feldman Marcus W 31 Department of Human Genetics, Bioinformatics Program, and the Life Sciences Institute, University of Michigan, Ann Arbor, Michigan, United States of America
2 Department of Computer Science, University of Southern California, Los Angeles, California, United States of America
3 Department of Biological Sciences, Stanford University, Stanford, California, United States of America
4 Mammalian Genotyping Service, Marshfield Clinic Research Foundation, Marshfield, Wisconsin, United States of America
5 Department of Human Genetics, University of Chicago, Chicago, Illinois, United States of America
Allison David EditorUniversity of Alabama at Birmingham, United States of America* To whom correspondence should be addressed. E-mail: [email protected] 2005 9 12 2005 1 6 e7018 8 2005 24 10 2005 Copyright: © 2005 Rosenberg et al.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.Previously, we observed that without using prior information about individual sampling locations, a clustering algorithm applied to multilocus genotypes from worldwide human populations produced genetic clusters largely coincident with major geographic regions. It has been argued, however, that the degree of clustering is diminished by use of samples with greater uniformity in geographic distribution, and that the clusters we identified were a consequence of uneven sampling along genetic clines. Expanding our earlier dataset from 377 to 993 markers, we systematically examine the influence of several study design variables—sample size, number of loci, number of clusters, assumptions about correlations in allele frequencies across populations, and the geographic dispersion of the sample—on the “clusteredness” of individuals. With all other variables held constant, geographic dispersion is seen to have comparatively little effect on the degree of clustering. Examination of the relationship between genetic and geographic distance supports a view in which the clusters arise not as an artifact of the sampling scheme, but from small discontinuous jumps in genetic distance for most population pairs on opposite sides of geographic barriers, in comparison with genetic distance for pairs on the same side. Thus, analysis of the 993-locus dataset corroborates our earlier results: if enough markers are used with a sufficiently large worldwide sample, individuals can be partitioned into genetic clusters that match major geographic subdivisions of the globe, with some individuals from intermediate geographic locations having mixed membership in the clusters that correspond to neighboring regions.
Synopsis
By helping to frame the ways in which human genetic variation is conceptualized, an understanding of the genetic structure of human populations can assist in inferring human evolutionary history, as well as in designing studies that search for disease-susceptibility loci. Previously, it has been observed that when individual genomes are clustered solely by genetic similarity, individuals sort into broad clusters that correspond to large geographic regions. It has also been seen that allele frequencies tend to vary continuously across geographic space. These two perspectives seem to be contradictory, but in this article the authors show that they are indeed compatible.
First the authors demonstrate that the clusters are robust, in that if sufficient data are used, the geographic distribution of the sampled individuals has little effect on the analysis. They then show that allele frequency differences generally increase gradually with geographic distance. However, small discontinuities occur as geographic barriers are crossed, allowing clusters to be produced. These results provide a greater understanding of the factors that generate the clusters, verifying that they arise from genuine features of the underlying pattern of human genetic variation, rather than as artifacts of uneven sampling along continuous gradients of allele frequencies.
Citation:Rosenberg NA, Mahajan S, Ramachandran S, Zhao C, Pritchard JK, et al. (2005) Clines, clusters, and the effect of study design on the inference of human population structure. PLoS Genet 1(6): e70.
==== Body
Introduction
It has recently been demonstrated in several studies that to a large extent, without prior knowledge of individual origins, the geographic ancestries of individuals can be inferred from genetic markers [1–5]. In one of the most extensive of these studies to date, considering 1,056 individuals from 52 human populations, with each individual genotyped for 377 autosomal microsatellite markers, we found that individuals could be partitioned into six main genetic clusters, five of which corresponded to Africa, Europe and the part of Asia south and west of the Himalayas, East Asia, Oceania, and the Americas [3]. Some individuals from boundary locations between these regions were inferred to have partial ancestry in the clusters that corresponded to both sides of the boundary. In many cases, subclusters that corresponded to individual populations or to subsets of populations were also identified.
To further ascertain the degree of difficulty in obtaining the genetic clusters, several articles have considered the influence of properties of the study design on the extent of clustering [3,4,6–10]. These studies have shown that the clustering patterns are robust, provided that at least about 60–150 markers are used [3,4,7,9], or about 40 or fewer if markers are preselected to have a high information content about ancestry [6]. They have also observed that although clustering patterns are influenced by sample size for small samples, the cluster membership estimates obtained for individuals in analysis of subsamples of larger datasets are close to those seen in analysis of the full data [9]. Additionally, they have found clustering results obtained with different statistical techniques to be quite similar [7,8].
Other factors besides sample size and number of markers, however, may influence clustering patterns. Serre and Pääbo [10] argued that the geographic dispersion of the sample and the assumption made about whether or not allele frequencies are correlated across populations had substantial influences on genetic clustering. They suggested that individuals are less strongly placed into clusters when the sample is more geographically uniform, and when allele frequencies are assumed to be uncorrelated. Consequently, they claimed that the geographic clusters obtained by Rosenberg et al. [3] were artifacts of the sampling design and of the use of a model of correlation among allele frequencies across populations. However, much of the geographic dispersion analysis of [10] was based on two datasets with 89 and 90 individuals and 20 loci, in general too little data for clustering to be apparent [3,4,9]. The remainder of their geographic analysis, as well as the source of their comments about uncorrelated frequencies, was a comparison to the Rosenberg et al. [3] results of several analyses of 261 individuals chosen to be equally distributed across the 52 populations studied. Serre and Pääbo's analyses assumed allele frequencies to be uncorrelated across populations, whereas Rosenberg et al. had assumed that they were correlated. Thus, although a difference in results was seen between the analyses in [10] and those in [3], the attribution of this difference specifically to a difference in geographic dispersion or to a difference in assumptions about allele frequency correlations is problematic, because both of these variables differed between studies, as did the number of individuals.
Figure 1 Distribution of the Geographic Dispersion Statistic (An) for Sets of 100 Points Randomly Sampled from a Sphere, Randomly Sampled from the Land Area of the Earth (from among the Points Plotted in Figure 5 of [11]), and Randomly Sampled from the Reported Locations of Individuals in the Dataset
Each distribution is obtained by binning the values of An for 100,000 sets of points.
In this article, we perform an extensive evaluation of the role of study design on genetic clustering, considering both geographic dispersion and allele frequency correlation, as well as sample size, number of loci, and number of clusters. The dataset employed is an expansion of our original data [3] to 993 markers, including 783 microsatellites [11] and 210 insertion/deletion polymorphisms. Analysis of multilocus genotypes in the larger dataset reveals essentially the same set of clusters as was produced with the original 377 markers. The number of loci, sample size, and number of clusters are observed to have considerable influence on clustering. In agreement with the suggestion of [10], the assumption made about allele frequency correlations is also seen to have a substantial impact. Because large allele frequency correlations exist across populations, however, the basis for the supposition by [10] that allele frequencies are uncorrelated is questionable. Finally, the level of geographic dispersion of the sample is seen to have only a relatively small effect on the clustering results, and this variable is not consistent in the direction in which it influences the level of clustering. Therefore, we find no reason to interpret our inferred clusters as artifacts of the sampling design in our original study, and we conclude with an illustration of how the clusters can have arisen from small discontinuities in genetic distance across geographic barriers.
Results
We utilized the unsupervised clustering algorithm implemented in STRUCTURE [12,13] to group individuals into genetic clusters in such a way that each individual is given an estimated membership coefficient for each cluster, corresponding to the fraction of his or her genome inferred to have ancestry in the cluster. This method requires that the number of clusters be prespecified, and assumes either a particular model of allele frequency correlations across clusters [12,13] or that allele frequencies are uncorrelated. The correlated frequencies model—the F model in [13]—supposes that the various clusters represent populations that have descended with genetic drift from a common ancestral population, so that alleles in different clusters have correlated frequencies due to shared ancestry. The uncorrelated frequencies model, on the other hand, is based on an assumption that allele frequencies are not expected to be similar across populations, and does not hypothesize an ancestral relationship among the clusters [12]. The reasoning underlying the correlated frequencies model is that for closely related populations, as measured by statistics such as Fst, allele frequencies tend to be correlated. Including correlation in the population structure model typically gives STRUCTURE greater power to detect similar but distinct populations (Figure 2 of [13]).
Figure 2 Inferred Population Structure Based on 1,048 Individuals and 993 Markers, Assuming Correlations among Allele Frequencies across Clusters
Each individual is represented by a thin line partitioned into K colored segments that represent the individual's estimated membership fractions in K clusters. Each plot, produced with DISTRUCT [23], is based on the highest-likelihood run of ten runs: the two runs that were used in further analysis, and the eight runs described under “Cluster Analysis using STRUCTURE.” As in [3], four of ten runs with K = 3 separated a cluster corresponding to East Asia instead of one corresponding to Europe, the Middle East, and Central/South Asia. Two of ten runs with K = 5 separated Surui instead of Oceania. The highest-likelihood run of the ten runs with K = 6, shown in the figure, had a different pattern from the other nine runs (not shown). These other runs, instead of subdividing native Americans into two clusters, subdivided a cluster roughly similar to the Kalash cluster seen in [3], except with a less pronounced separation of the Kalash population. The clusteredness scores for the plots shown with K = 2, 3, 4, 5, and 6 are 0.50, 0.76, 0.84, 0.86, and 0.87, respectively.
A total of 367,220 runs of STRUCTURE were performed on subsets of a dataset consisting of 1,048 individuals from the Human Genome Diversity Project–Centre d'Etude du Polymorphisme Humain (HGDP-CEPH) Human Genome Diversity Panel [14] and 993 microsatellite and insertion/deletion polymorphisms. These runs utilized five choices for the number of clusters (two, three, four, five, and six), seven choices for the number of loci (ten, 20, 50, 100, 250, 500, and 993), four choices for the sample size (100, 250, 500, and 1,048), and two choices for the allele frequency correlation model (correlated and uncorrelated, as described by [12,13]). For each choice of the number of loci other than 993, runs were performed with each of ten prespecified sets of loci randomly selected from among the full set of markers, and for each choice of the sample size other than 1,048, runs were performed with each of 100 prespecified sets of individuals.
Figure 3 Mean Clusteredness versus Number of Loci
Each point shows the mean clusteredness of 2,000 runs with the specified sample size and allele frequency correlation model: two replicates for each of ten sets of loci for each of 100 sets of individuals (for 1,048 individuals, it is the mean of 20 runs, as only one set of individuals was used; for 1,048 individuals and 993 loci, it is the mean of two runs, as only one set of loci was used). Error bars denote standard deviations. The x-axis is plotted on a logarithmic scale.
The 100 sets of individuals used were selected to have a wide range of levels of geographic dispersion (Figure 1), as measured by the dispersion statistic An (see Materials and Methods). Because the sets all utilized the sampling locations of the diversity panel, their An values were bounded by the minimal and maximal levels of dispersion possible in this sample. However, with a sample size of 100, the sets that had the lowest values of An—and were therefore most uniformly distributed geographically—had comparable An values to some sets of 100 points randomly chosen from the land area of the earth. For each collection of settings—the lists of individuals and loci, and the choices for the number of clusters and the allele frequency correlation model—two replicate STRUCTURE runs were performed. The “clusteredness” (see Materials and Methods) of the collection of estimated membership coefficients was then calculated for each of the 367,220 runs. This statistic measures the extent to which a randomly chosen individual is inferred to have ancestry in only one cluster (clusteredness = 1), with the other extreme being equal membership in all clusters (clusteredness = 0). Use of this statistic relies on the observation that when populations are unstructured or when insufficient data are used, STRUCTURE typically distributes the membership coefficients of all individuals evenly across clusters rather than assigning each individual a membership coefficient of one for one cluster (the same cluster for all individuals) and zero for all other clusters (see the top right plot in Figure 4 of [6] and the top left plot in Figure 6 of [9]).
Figure 4 Mean Clusteredness versus Geographic Dispersion as Measured by An
Each point shows the mean clusteredness of 20 runs with the specified number of loci and allele frequency correlation model: two replicates for each of ten sets of loci (for 993 loci, it is the mean of two runs, as only one set of loci was used). From left to right, the three groups of points in each plot respectively represent sets of 100, 250, and 500 individuals.
Figure 5 Inferred Population Structure Based on Two Different Sets of 100 Individuals, Using 993 Markers and the Correlated Allele Frequencies Model
The two sets of 100 individuals represent extremes of the distribution of An: the plots on the left are based on a more geographically random sample, and those on the right are based on a less random sample. Each plot is based on the higher-likelihood run among the two runs performed with the given combination of loci and individuals. In all plots, individuals and populations are in the same order as in Figure 2. Black vertical lines at the bottom of the figure separate populations from the different geographic regions described in [3], with the asterisk representing Oceania.
Figure 6 Genetic and Geographic Distance for Pairs of Populations
Red circles indicate comparisons between pairs of populations with majority representation in the same cluster in the K = 5 plot of Figure 2; blue triangles indicate pairs with one population from Eurasia and one from East Asia; brown squares indicate pairs with one population from Africa and the other from Eurasia; and green diamonds indicate pairs with one population from East Asia and the other from either Oceania or America. Comparisons involving one of Hazara, Kalash, and Uygur and other populations from Eurasia or East Asia are marked 1, 2, and 3, respectively. No comparisons are shown between any of these three groups and any African population.
Representative estimates of the population structure based on the full dataset are shown in Figure 2. These estimates are quite similar to what was previously obtained using 377 loci [3], with the main difference being that the sixth cluster sometimes corresponds to a subdivision of native Americans into more northerly and more southerly populations rather than to a separation of the isolated Kalash population of Pakistan.
To examine the influence of the study design parameters on clusteredness, we separately considered each variable, holding the others constant. This analysis included linear regressions of clusteredness on each variable for each possible combination of values of the other variables. We also analyzed the full collection of runs to determine the relative contributions of the quantities considered to variability in clusteredness.
Number of Loci
Holding the number of clusters, sample size, and allele frequency correlation model fixed, the general trend was that clusteredness was noticeably smaller for ten and 20 loci, and was larger for 50 or more loci (Figure 3). This was usually true regardless of the choice of the number of clusters, sample size, or correlation model. For 39 of 40 combinations of these three variables, the regression coefficient of the logarithm of the number of loci was significantly different from zero at the p < 0.001 level, indicating a noticeable effect of the number of loci on clusteredness (the 40th combination had p = 0.002). For all 40 combinations, the regression coefficient was positive, indicating an increase in clusteredness with increasing number of loci, and the mean coefficient of determination (R
2) across the 40 regressions equaled 0.454.
Number of Clusters
When the number of loci, sample size, and correlation model were held constant, K = 2 (that is, two clusters) generally produced smaller clusteredness than did the larger values of K (Figures 3 and 4; Table 1). For the correlated allele frequencies model, K = 5 and K = 6 tended to have higher clusteredness than did K = 3 and K = 4, whereas the reverse was true for the uncorrelated model (Figure 4). This trend was reflected in the regression coefficients for K: with the correlated model, for 27 of 28 combinations of the number of loci and the sample size, the regression coefficient was positive, whereas it was positive for only 11 of 28 combinations with the uncorrelated model (Table 2). In 51 of 56 combinations, the regression coefficient was significantly different from zero at p < 0.001; 34 of these involved positive and 17 involved negative regression coefficients. Reflecting the general monotonic trend in clusteredness with K in the correlated model but not in the uncorrelated model, the average R
2 was larger across the 28 combinations with the correlated model (0.382) than it was for the 28 combinations with the uncorrelated model (0.147).
Table 1 Clusteredness Mean and Standard Deviation for the Correlated and Uncorrelated Allele Frequency Models
Table 2 Influence of the Number of Clusters K on Clusteredness
Sample Size
Holding the number of loci, number of clusters, and correlation model fixed, clusteredness was generally higher for the samples of size 250 and 500 than it was for the samples of size 100 (Figures 3 and 4; Table 1). For 65 of 70 combinations of the number of loci, the number of clusters, and the correlation model, the regression coefficient for sample size was both significantly different from zero at p < 0.001 and positive (Table 3). The five cases for which the regression coefficient was negative, not significantly different from zero at p < 0.001, or both all involved K = 2. The average R
2 across the 70 combinations equaled 0.511.
Table 3 Influence of the Sample Size on Clusteredness
Geographic Dispersion of Individuals
With the correlation model and the numbers of loci, clusters, and individuals held constant, the inferred population structure was generally similar for different values of An (Figure 5, for example). Population structure estimates differed substantially for different values of An mainly in situations where one but not the other dataset had a very small sample from one of the main clusters in the full dataset. For example, Oceania is well-represented and corresponds to a cluster for the more geographically random dataset in Figure 5 (left side), but is not well-represented and does not correspond to a cluster for the less random dataset (Figure 5, right side).
Often, geographic dispersion had a negative rather than a positive influence on clusteredness (see Figure 4), so that less uniformly distributed samples produced lower clusteredness. This effect was reflected in the regression coefficient for An, which was negative for 174 of 210 combinations of the number of loci, sample size, number of clusters, and correlation model (Table 4). Of the 36 combinations with positive regression coefficients, 12 had regression coefficients that were significantly different from zero at p < 0.001. However, the decrease of clusteredness with increasing An in the remaining 174 cases was often quite small; in 46 of these 174 cases, the regression coefficient was not significantly different from zero at p < 0.001, and the average R
2 across the 210 regressions was only 0.045.
Table 4 Influence of the Geographic Dispersion An on Clusteredness
Allele Frequency Correlation Model
With the numbers of loci, clusters, and individuals held constant, the correlation model had a noticeable influence on clusteredness, with the correlated model usually producing higher clusteredness than the uncorrelated model (see Figures 3 and 4; Table 1). This effect was generally seen regardless of the number of loci (Table 1). In 101 of 105 combinations in which the sample size was 100, 250, or 500, the Wilcoxon test for a difference in clusteredness under the correlated versus under the uncorrelated model was significant at p < 0.001. In 97 of these 101 combinations, the correlated model had higher mean clusteredness across runs than did the uncorrelated model. For 1,048 individuals, fewer runs were performed, and p < 0.001 for only 14 of 35 combinations; as with smaller sample sizes, however, in 32 of the 35 combinations with 1,048 individuals, clusteredness was greater for the correlated model. Considering all sample sizes, all nine cases in which clusteredness was smaller for the correlated model involved K = 2.
Analysis of Variance of Clusteredness
With each sample size, considering all 122,000 STRUCTURE runs with the given sample size, the R
2 values for regressions of clusteredness on individual variables were greatest for the number of loci and the allele frequency correlation model, and smallest for the number of clusters and the geographic dispersion (Table 5). Combining all 367,220 runs, the sample size also produced an effect comparable to that seen for the number of loci and the correlation model, while the contributions of the number of clusters and the geographic dispersion remained smaller.
Table 5 Values of R
2 for Regressions of Clusteredness on Study Design Variables
Discussion
In this article, we have systematically analyzed the influence of five variables on the genetic clustering of individuals from genome-wide markers: number of loci, sample size, number of clusters, geographic dispersion of the sample, and assumptions about allele frequency correlation. Each of these variables was found to have an effect on clustering. Holding all other variables constant, geographic dispersion had a relatively modest effect on clusteredness, with a considerably smaller R
2 than number of loci, sample size, or number of clusters. Additionally, geographic dispersion was generally less consistent in the direction in which it affected clusteredness, although in contrast to what was expected based on the results of [10], samples with higher An (that is, samples that were less geographically random) produced lower clusteredness more often than they produced higher clusteredness.
Unlike geographic dispersion, the number of loci and sample size both had strong direct relationships with clusteredness for nearly all combinations of the other variables. Excluding a few scenarios that utilized two clusters, the correlation model produced significantly greater clusteredness for nearly all combinations of the other variables, when a large number of STRUCTURE runs were performed. The number of clusters influences the way in which individual membership coefficients are distributed, but its effect on the clusteredness statistic was found to be smaller than that of the number of loci or the sample size. The effect of the number of clusters depended on the choice of correlation model: in the correlated model, clusteredness generally increased with K, whereas in the uncorrelated model, clusteredness was not monotonic in K.
Two main claims of Serre and Pääbo [10] merit direct comparison with our results. First, on the basis of STRUCTURE runs of two samples with 89 and 90 individuals, 20 loci, and the uncorrelated allele frequencies model, Serre and Pääbo argued that use of a sample with a more random geographic distribution led to reduced clusteredness. Although we were expecting to corroborate this observation, which was not based on the HGDP-CEPH Human Genome Diversity Panel sample studied here, our analysis under similar conditions did not support it. Moving across the range of An for the 100 samples of size 100, when 20 loci were used with the uncorrelated model (Figure 4), there was a trend opposite to that expected, in that clusteredness decreased with increasing geographic nonrandomness: for a sample size of 100 and 20 loci, regression coefficients for An were negative with p < 0.001 for each value of K and both correlation models (see Table 4). In other words, in a test similar to that performed by [10], the effect of reduced clusteredness with increasing geographic nonrandomness was not seen when 100 samples were studied, rather than two samples, as in [10].
Second, in three analyses with the uncorrelated allele frequencies model, each of which used 261 individuals, Serre and Pääbo observed a reduction in clusteredness compared with analyses using 261 individuals and the correlated model, and compared with analyses based on 1,066 individuals and either model. They attributed the different results in these scenarios to the use of the uncorrelated frequencies model. We found, however, that with either the correlated or the uncorrelated allele frequencies model, holding all other variables constant, when 100 samples of size 250 were considered, clusteredness differed for the samples of size of 250 compared with those of size 1,048 (Figure 3). Therefore, the difference in results obtained by [10] is likely to derive from a combination of both the difference in models and the difference in sample size.
Even if the frequency correlation model actually provided the sole explanation for the weaker clustering in their analysis, we question the basis for assuming that allele frequencies are uncorrelated across populations. Allele frequencies should be expected to be correlated, on the basis of the shared descent of all human populations from the same set of ancestral groups. Clearly, as has been shown in simulations [13], the choice of correlation model has a substantial influence on clustering results (Figures 3 and 4; Table 1). However, as the correlated and uncorrelated models should only be expected to produce different results if data contain a high level of correlation—which is taken into account by the correlated model but not by the uncorrelated model—it is precisely when allele frequencies have strong correlations across populations that the two models will produce different results. Thus, the high correlation coefficients we have estimated for allele frequencies ([9]; Table 6) both explain the difference in results between the correlated and uncorrelated models, and suggest that the correlated model, which we used in [3] and in Figure 2, provides a more appropriate model for human genetic variation.
Table 6 Correlation Coefficients of Allele Frequencies
In summary, the observation of [10] of stronger clustering with increased geographic nonrandomness was not seen in our analysis of a larger number of samples. Additionally, geographic dispersion was seen to be the least influential of the five study design variables that we considered. By using fewer loci and individuals in their various tests, and by assuming an uncorrelated allele frequencies model, Serre and Pääbo chose study design parameters in such a way that clustering was less pronounced than had been previously observed. In no way does this alter the fact that when a sufficiently large sample and number of loci are used, together with the more appropriate correlated allele frequencies model, individuals do cluster into populations that correspond largely to geographic regions. Indeed, the observation of essentially the same clusters with a larger dataset further supports the robustness of our original analysis.
Clines or Clusters?
Serre and Pääbo [10] argue that human genetic diversity consists of clines of variation in allele frequencies. We agree and had commented on this issue in our original paper ([3], p. 2382): “In several populations, individuals had partial membership in multiple clusters, with similar membership coefficients for most individuals. These populations might reflect continuous gradations across regions or admixture of neighboring groups.” At the same time, we find that human genetic diversity consists not only of clines, but also of clusters, which STRUCTURE observes to be repeatable and robust.
How can these seemingly discordant perspectives on human genetic diversity be reconciled? Figure 6 shows a plot of genetic distance and geographic distance for pairs of populations. To illustrate the effects of moving continuously across geographical space, only pairs from within clusters or from geographically adjacent clusters are shown. That is, for the five clusters with K = 5 in Figure 2 of the present study and in Figure 1 of [3]—corresponding to Africa, Eurasia (Europe, Middle East, and Central/South Asia), East Asia, Oceania, and the Americas—an intercluster population pair is plotted only if it includes one population from Africa and one from Eurasia, one from Eurasia and one from East Asia, or one from East Asia and one from Oceania or the Americas.
For population pairs from the same cluster, as geographic distance increases, genetic distance increases in a linear manner, consistent with a clinal population structure. However, for pairs from different clusters, genetic distance is generally larger than that between intracluster pairs that have the same geographic distance. For example, genetic distances for population pairs with one population in Eurasia and the other in East Asia are greater than those for pairs at equivalent geographic distance within Eurasia or within East Asia. Loosely speaking, it is these small discontinuous jumps in genetic distance—across oceans, the Himalayas, and the Sahara—that provide the basis for the ability of STRUCTURE to identify clusters that correspond to geographic regions.
Two exceptions to the pattern include the Hazara and Uygur populations, from Pakistan and western China, respectively, whose genetic distances scale continuously with geographic distance both for populations in Eurasia and for those in East Asia. These populations were evenly split across the clusters corresponding to Eurasia and East Asia, and thus, unlike most other populations, they do not reflect a discontinuous jump in genetic distance with geographic distance. Finally, a third population of interest in the plot is the Kalash population (of Pakistan), whose genetic distances to other populations are large at all geographic distances, illustrating the distinctiveness of the group as the only member of its own genetic cluster in some STRUCTURE analyses with K = 6 [3].
Excluding points that involve Hazara, Kalash, or Uygur, a linear regression on geographic distance for the points in Figure 6 has R
2 = 0.690. When an additional binary variable B is added—equaling one if an ocean, the Himalayas, or the Sahara must be crossed to travel between two populations, and zero otherwise—R
2 increases to 0.729. The regression equation is Fst = 0.0032 + 0.0049D + 0.0153B, where D is distance in thousands of kilometers. By dividing the regression coefficients for B and D, it can be observed that crossing one of the barriers adds an equivalent amount of genetic distance as traveling approximately 3,100 km on the same side of the barrier. The effect of a barrier is to add 0.0153 to Fst beyond the value predicted by geographic distance alone. As 0.0153 is not a large value of genetic distance, and because the addition of the B term produces only a modest increase in R
2, the discontinuities that give rise to genetic clusters—as we have stated previously [3]—constitute a relatively small fraction of human genetic variation.
Our evidence for clustering should not be taken as evidence of our support of any particular concept of “biological race.” In general, representations of human genetic diversity are evaluated based on their ability to facilitate further research into such topics as human evolutionary history and the identification of medically important genotypes that vary in frequency across populations. Both clines and clusters are among the constructs that meet this standard of usefulness: for example, clines of allele frequency variation have proven important for inference about the genetic history of Europe [15], and clusters have been shown to be valuable for avoidance of the false positive associations that result from population structure in genetic association studies [16]. The arguments about the existence or nonexistence of “biological races” in the absence of a specific context are largely orthogonal to the question of scientific utility, and they should not obscure the fact that, ultimately, the primary goals for studies of genetic variation in humans are to make inferences about human evolutionary history, human biology, and the genetic causes of disease.
Materials and Methods
Data.
The dataset analyzed here consists of 1,048 individuals from the HGDP-CEPH Human Genome Diversity Panel [14]. Each individual was genotyped by the Mammalian Genotyping Service for 993 polymorphisms spread across all 22 autosomes: 783 microsatellites (with 3.7% missing data) and 210 insertion/deletion markers (with 7.7% missing data). Of these loci, 377 of the microsatellites were previously studied by [3] in most of the individuals analyzed here. The remaining microsatellites were drawn from Marshfield Screening Sets #13 and #52 [17], and the insertion/deletion markers were drawn from those studied by [18]. All 783 of the microsatellites were previously studied by [11].
The set of individuals used here differs slightly from that studied by [3]. It corresponds exactly to the set in [11], with two alterations. First, 21 Surui individuals excluded by [11] are included here, and second, eight individuals grouped into the southwestern Bantu and southeastern Bantu populations in [11] are grouped here as a single population labeled Bantu (southern Africa). Thus, we analyzed 53 populations.
Geographic dispersion.
The geographic dispersion of a set of n points on a sphere can be measured by the statistic
where ψij is the angle between the ith and jth points measured at the center of the sphere. The quantity An is a test statistic for the null hypothesis that the n points are uniformly distributed on the sphere (p. 149 of [19]). Larger values of An indicate sets of points that are less uniformly distributed. To evaluate ψij for a pair of points i and j, rectangular coordinates (x, y, z) are obtained from (latitude, longitude) coordinates (a, b) using (xi, yi, zi) = (cos(ai) cos(bi), cos(ai) sin(bi), sin(ai)) and (xj, yj, zj) = (cos(aj) cos(bj), cos(aj) sin(bj), sin(aj)). By the law of cosines,
Method 1 of [20] was used to generate the rectangular coordinates for random points uniformly distributed on the sphere. For each sample size (n = 100, 250, or 500), 100,000 sets of points were considered in obtaining the distribution of An.
To determine the distribution of An for sets of points uniformly distributed on the land area of the earth, 4,210 lattice points on land were identified for a lattice of 200 longitudes and 79 latitudes on the earth's surface (Figure 5 of [11]). From these points, for each sample size (n = 100, 250, or 500), 100,000 sets of points were drawn (with replacement), and An was calculated for each set.
To obtain the distribution of An for sets of points randomly chosen from the dataset, for each sample size (n = 100, 250, or 500), 100,000 random subsets of the 1,048 individuals were selected (without replacement), and An was computed for each subset. Latitude and longitude coordinates were taken from Supplementary Table 1 of [14]. In cases where latitudes and longitudes were given as ranges, the centroid of the specified region was calculated, with the longitude being the average of the endpoints of the range and the latitude being the inverse sine of the average of the sines of the endpoints of the range. Of the 100,000 random subsets of individuals, the 100 sets located at quantiles c + 1/2 with respect to the distribution of An were utilized in further analyses, where c ranged over integers from zero to 99.
Clusteredness.
To measure the average “clusteredness” of individuals, or the extent to which individuals were estimated to belong to a single cluster rather than to a combination of clusters, we computed for each STRUCTURE run the quantity
where qik denotes the estimated membership coefficient for the ith individual in the kth cluster, I denotes the total number of individuals, and K denotes the total number of clusters. The factor K/(K − 1) was included so that a change in K would not produce a systematic change in clusteredness.
Cluster analysis using STRUCTURE.
All runs of the STRUCTURE program [12] employed for analyzing the study design variables utilized 1,000 iterations after a burn-in period of 5,000 iterations. To evaluate whether this length was sufficient for convergence, we performed longer runs, all with a burn-in period of 5,000, and we compared results based on later iterations with those of the first 1,000 iterations after the burn-in. For each of K = 2, 3, 4, 5, and 6, eight runs were performed using the full dataset and the correlated allele frequencies model. Estimates of membership coefficients were separately obtained using the first 1,000 iterations after completion of the burn-in, iterations 15,001–20,000 after the burn-in, and iterations 45,001–50,000. Using a symmetric similarity coefficient [21], each of these three stages in each run was compared to each stage in the other seven runs with the same value of K, as well as to the other two stages from the same run. In all cases except for one of the runs with K = 6, similarity scores were 0.96 or greater, indicating that membership coefficient estimates were nearly identical both for different runs with the same K as well as for the three stages of the same run. Thus, it was determined that estimates would not be substantially different if runs longer than 1,000 iterations after a burn-in period of 5,000 were used. For each K, the results obtained from the eight runs at 1,000 iterations after completion of the burn-in were among the ten runs considered in choosing the highest-likelihood runs to display in Figure 2.
Statistical tests.
Linear regression was used to test the influence of study design variables on clusteredness. To control for the effects of the other variables, each regression utilized only STRUCTURE runs in which variables other than the one being tested were held constant. For example, to examine the influence of the number of clusters on clusteredness, 56 separate regressions were performed, one for each combination of the number of loci (seven possibilities), the sample size (four possibilities), and the allele frequency correlation model (two possibilities). Similarly, 40 regressions of clusteredness on the base-10 logarithm of the number of loci were performed, as were 70 regressions of clusteredness on sample size and 210 regressions of clusteredness on An. Note that in the case of An, since there was no variability in An across different runs with the full 1,048 individuals, the number of regressions reflects seven choices for the number of loci, five choices for the number of clusters, two choices for the allele frequency correlation model, and only three choices for the sample size. For each regression, the F-test was used to test the null hypothesis that the regression coefficient for the dependent variable equaled zero.
In the case of the allele frequency correlation model, the runs with the correlated and uncorrelated models were compared using the Wilcoxon two-sample test instead of with linear regression. Because there were seven numbers of loci, five numbers of clusters, and four numbers of individuals, 140 separate tests were performed.
For each sample size, regressions of clusteredness on individual variables were also performed using all 122,000 runs with the given sample size. Additional regressions were also performed using all 367,220 runs. These regressions used the base-10 logarithm of the number of loci.
Genetic and geographic distance.
For the comparison of genetic and geographic distance, calculations were performed as in [11], using Fst for genetic distance—computed as Fst = −ln(1 − θ), with the estimate of θ taken from equation 5.12 of [22]—and waypoint routes avoiding large bodies of water for geographic distance. A slight difference from the analysis in [11] was that the great circle distance for a pair of points i and j was computed using rψij where r is the radius of the earth (6,371 km) and ψij is measured in radians, rather than with equation 1 of [11]. Only the microsatellite data were used for this analysis, and the Karitiana, Maya, and Surui were omitted from the comparisons: Maya due to likely admixture [3], and Karitiana and Surui to keep the ranges of the axes in the plot small enough for the patterns of interest to be visible. See [11] for additional related plots.
We thank J. Long, J. Molitor, C. Roseman, H. Tang, E. Ziv, and an anonymous reviewer for suggestions that have greatly improved the manuscript. This work was supported by National Institutes of Health GM28016 to MWF, by the Stanford Genome Training Program (T32 HG00044 from the National Human Genome Research Institute), by a Burroughs Wellcome Fund Career Award in the Biomedical Sciences to NAR, and by a grant from the University of Southern California. The Mammalian Genotyping Service is supported by the National Heart, Lung, and Blood Institute (HV48141). The data used in this study are a subset of the genotypes available at http://research.marshfieldclinic.org/genetics, and the exact data employed in our analysis are available at http://rosenberglab.bioinformatics.med.umich.edu.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. The initial approach was conceived by NAR, with help from JKP and MWF. The construction of subsamples with different levels of geographic dispersion was performed by SM and NAR. CZ contributed to the design and construction of the marker panels and to initial analysis with STRUCTURE; the full STRUCTURE analysis was designed by NAR and SM, with help from JKP, and was performed by SM with help from NAR. The regression analyses were designed by NAR with help from MWF, and were performed by NAR. The genetic/geographic distance analysis was designed by SR and NAR and was performed by SR. NAR wrote the paper with help from SR, JKP, and MWF.
Abbreviations
HGDP-CEPHHuman Genome Diversity Project–Centre d'Etude du Polymorphisme Humain
==== Refs
References
Bowcock AM Ruiz-Linares A Tomfohrde J Minch E Kidd JR 1994 High resolution of human evolutionary trees with polymorphic microsatellites Nature 368 455 457 7510853
Mountain JL Cavalli-Sforza LL 1997 Multilocus genotypes, a tree of individuals, and human evolutionary history Am J Hum Genet 61 705 718 9326336
Rosenberg NA Pritchard JK Weber JL Cann HM Kidd KK 2002 Genetic structure of human populations Science 298 2381 2385 12493913
Bamshad MJ Wooding S Watkins WS Ostler CT Batzer MA 2003 Human population genetic structure and inference of group membership Am J Hum Genet 72 578 589 12557124
Tang H Quertermous T Rodriguez B Kardia SLR Zhu XF 2005 Genetic structure, self-identified race/ethnicity, and confounding in case-control association studies Am J Hum Genet 76 268 275 15625622
Rosenberg NA Li LM Ward R Pritchard JK 2003 Informativeness of genetic markers for inference of ancestry Am J Hum Genet 73 1402 1422 14631557
Turakulov R Easteal S 2003 Number of SNPS loci needed to detect population structure Hum Hered 55 37 45 12890924
Corander J Waldmann P Marttinen P Sillanpää MJ 2004 BAPS 2: Enhanced possibilities for the analysis of genetic population structure Bioinformatics 20 2363 2369 15073024
Ramachandran S Rosenberg NA Zhivotovsky LA Feldman MW 2004 Robustness of the inference of human population structure: A comparison of X-chromosomal and autosomal microsatellites Hum Genomics 1 87 97 15601537
Serre D Pääbo S 2004 Evidence for gradients of human genetic diversity within and among continents Genome Res 14 1679 1685 15342553
Ramachandran S Deshpande O Roseman CC Rosenberg NA Feldman MW 2005 Support from the relationship of genetic and geographic distance in human populations for a serial founder effect originating in Africa Proc Natl Acad Sci U S A 102 15942 15947 16243969
Pritchard JK Stephens M Donnelly P 2000 Inference of population structure using multilocus genotype data Genetics 155 945 959 10835412
Falush D Stephens M Pritchard JK 2003 Inference of population structure using multilocus genotype data: Linked loci and correlated allele frequencies Genetics 164 1567 1587 12930761
Cann HM de Toma C Cazes L Legrand MF Morel V 2002 A human genome diversity cell line panel Science 296 261 262 11954565
Cavalli-Sforza LL Piazza A Menozzi P 1994 The history and geography of human genes Princeton (New Jersey) Princeton University Press 1,088 p.
Pritchard JK Donnelly P 2001 Case-control studies of association in structured or admixed populations Theor Pop Biol 60 227 237 11855957
Ghebranious N Vaske D Yu A Zhao C Marth G 2003 STRP screening sets for the human genome at 5 cM density BMC Genomics 4 6 12600278
Weber JL David D Heil J Fan Y Zhao C 2002 Human diallelic insertion/deletion polymorphisms Am J Hum Genet 71 854 862 12205564
Fisher NI Lewis T Embleton BJJ 1987 Statistical analysis of spherical data Cambridge Cambridge University Press 352 p.
Marsaglia G 1972 Choosing a point from the surface of a sphere Ann Math Stat 43 645 646
Nordborg M Hu TT Ishino Y Jhaveri J Toomajian C 2005 The pattern of polymorphism in Arabidopsis thaliana
PLoS Biol 3 1289 1299 DOI: 10.1371/journal.pbio.0030196
Weir BS 1996 Genetic data analysis II: Methods for discrete population genetic data, 2nd ed Sunderland (Massachusetts) Sinauer Associates 445 p.
Rosenberg NA 2004 Distruct: A program for the graphical display of population structure Mol Ecol Notes 4 137 138
|
16355252
|
PMC1310579
|
CC BY
|
2021-01-05 07:59:38
|
no
|
PLoS Genet. 2005 Dec 9; 1(6):e70
|
utf-8
|
PLoS Genet
| 2,005 |
10.1371/journal.pgen.0010070
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-641628198410.1186/1471-2148-5-64Research ArticleEvolution of virulence: coinfection and propagule production in spore-producing parasites Lively Curtis M [email protected] Department of Biology, Indiana University, Bloomington, IN 47405, USA2005 10 11 2005 5 64 64 19 5 2005 10 11 2005 Copyright © 2005 Lively; licensee BioMed Central Ltd.2005Lively; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The evolution of within-host growth rates by parasites is expected to depend on a trade-off between propagule production and virulence. The presence of coinfections, however, is thought to alter this trade-off, and hence alter the evolutionarily stable strategy (ESS) for the parasite. Here I consider a model wherein the number of coinfections that are identical by descent can depend on the parasite's reproductive strategy. Transmission success was treated as being either a negative-linear or a negative-exponential function of the total number of propagules produced by all coinfections.
Results
Increasing the number of unrelated coinfections either selected for a decrease in reproductive output by the parasite (linear case), or had no effect on the ESS (exponential case). Nonetheless, the total number of propagules produced within each host increased in both cases. Increasing the relatedness among coinfections, however, selected for reductions in parasite reproduction in both cases.
Conclusion
Unrelated coinfection may increase overall parasite virulence, but the result stems from adding more infections rather than to more aggressive growth by the individual infections. However, all else being equal, if the coinfections are more related than expected by chance alone, then the total reproductive output by all coinfections would be expected to be reduced, resulting in reduced virulence.
==== Body
Background
Present theory for the evolution of parasite virulence is built upon the idea that there is a trade-off between the advantage of within-host replication and the disadvantage that such replication has on host survivorship [1-4]. Several factors have been shown to affect this trade-off, and thus change the attractor of local evolutionary dynamics [review in [5]]. For example, the generation of new strains during the course of infection by mutation, and/or the direct addition of new coexisting strains (coinfections), can select for increased rates of parasite replication by decreasing relatedness among strains and increasing the among-strain, within-host competition [4-11]. However, in contrast to this widely accepted view that coinfection selects for increase parasite virulence, three more recent models have shown that adding coinfections could instead select for reduced rates of replication by parasites [12-14]. Using kin selection models, Chao et al. [13] found that soft selection could lead to the evolution of reduced virulence in coinfections, and Brown et al. [12] found that a "collective action" by coinfecting parasites could lead to the evolution of reduced virulence. Using computer simulations of an epidemiological model, Schjørring and Koella [14] found that sub-lethal effects of parasites could also lead to reduced virulence. As such, it would seem that the details are important in determining whether or not coinfection increases virulence.
I constructed a kin-selection model to examine the effects of coinfection and relatedness on propagule production in spore-producing parasites. The model is different from previous kin-selection models in that it examines the effect of competition between the transmission stages produced by all infected individuals in a large population of susceptible hosts. The model assumes that an annual host comes into contact with parasite spores as a juvenile. If multiple spores infect the same host at about the same time, they produce coexisting infections within the host. (The number of coinfections is assumed here to be property of susceptible hosts, rather that a function of parasite reproductive rates.) During the within-host growth phase, the parasites replicate at a rate such that N propagules are produced by the end. The propagules do not directly interfere with each other, thus two infections could potentially produce twice as many propagules during the within-host growth period as one infection. Following within-host replication, the propagules metamorphose into spores that become competent for release into the external environment; the release occurs when the annual host dies, rather than is an steady stream as they are produced [following [15]]. After their release, the spores are "free-living" until the following year's cohort of juvenile hosts emerges. I assume that the probability that a propagule becomes a free-living spore depends on the total number of propagules produced during the within-host growth period by all coinfections within the host. This latter assumption is similar to that made by both Chao et al. [13] and Brown et al. [12]. Propagule production reduces the probability of spore formation and/or dissemination; but the effect of infection on the host varies by system and depends on the relationship (generally negative) between total propagule number and host reproductive output. Nonetheless, I assume that the host is not killed during the within-host growth period prior to spore formation. Finally, I assume that the spores do not survive more than one year in the environment. The effect of carryover among years has been treated elsewhere [10,16,17].
Results
Consider an asexual population of haploid parasites [see [18]]. Most of the host population is infected with parasites having the wild type, a, allele; and each of these parasites produce Na propagules during their within-host growth phase. One host, however, is infected with one or more parasites having a mutant allele, A, which leads to the production of NA propagules during the within-host growth phase. The expected fitness of this rare strain of parasite, WA, is the number of propagules that escape the host and become free-living spores, SA, times the product of the number of hosts, H, and the number of infections per host, K, divided by the total number of spores in the population, Stot:
Here SA is equal to the number of propagules produced, NA, times the probability that each propagule produced is successfully released as a spore, T; hence SA = NAT. The variable T is negatively related to the total number of propagules, Ntot, produced by all infections within a host, K. Below I consider two cases for the relationship between transmission, T, and total propagule number, Ntot: exponential and linear.
Exponential case
I first consider the case where each propagule has the same proportional negative effect on T as all other propagules. Thus
T = exp{-α[NA + (K - 1)(NA(f + (1 - f)p) + Na(1 - f)(1 - p))]}, (2)
where α gives the effect of each propagule on the expected probability of spore formation and transmission to the environment; f gives the probability that a coinfection is identical by descent [following [19]], and p gives the frequency of the A allele in the population. The expression in square brackets is the total number of propagules produced, Ntot, within the focal host containing the mutant parasite strain. Finally, Stotis the total number of spores produced in the focal host, Sfocal, plus the number produced in all other hosts, Sother(Stot = Sfocal + Sother), where
Sfocal = TNtot, (3)
Taking the limit as host population size, H, goes to infinity and p goes to 0 [following [20]], the expected fitness of the mutant bearing the A allele converges on
where is the average number of propagules produced in the focal host.
The proportion of infections in the focal host that are identical by descent and express the A allele could be affected by a change in the reproductive rate of the parasite. Thus f might be a function of NA. All other variables (i.e., Na and K) were treated as constants. By the chain rule, the change in fitness given a slight heritable change in NA is equal to
where βf,N gives the regression of f on propagule number, N. This method is based on the method of Taylor and Frank [11], but it considers how changes in the probability of identity by descent changes with the phenotype of the focal individual, rather that how the group mean phenotype changes with the phenotype of the focal individual. Nonetheless, as the group mean in the present model only changes with changes in f and NA, the approaches are similar and they yield identical results.
Using standard methods [21], the equilibrium value for the number of propagules produced (N*) is found by solving for
The equilibrium is also evolutionarily and convergence stable [22-26], respectively, if:
Solving equation (7), the equilibrium number of propagules produced during the period of within-host reproduction is:
which is both evolutionarily and convergence stable.
When there is only a single infection within each host (thus K = 1), the equilibrium value, N*, reduces to α-1, which is the value that maximizes R0[27]. I refer to α-1 as the baseline value. Given the result in equation (9), parasites will be selected to produce fewer propagules than the baseline value when f(K-1) > 0, which is when the number of coinfections that are identical by descent is greater than zero. Conversely, parasites will be selected to produce more propagules than the baseline value when f(K-1) < 0, which is not biologically possible. Hence in this model, coinfection does not lead to selection to increase in the within-host growth rate. However, coinfection could lead to a reduction in the within-host growth rate if multiple individuals share the same mutation.
At equilibrium, the total number of propagules produced within a host, Ntot is simply the number of coinfections, K, times N*. Hence
In the present model, relatedness, R, is equal to the frequency of infections within the focal host that share the mutation. Thus
As such, Ntot reduces to 1/(Rα). Similarly, N* reduces to 1/(RKα). In these terms, coinfection will result in a decrease in propagule production when N* is less than the baseline value, which is when R>1/K. Because 1/K gives the relatedness expected by chance in a well-mixed population of spores, selection is expected to favor a reduction in the rate of propagule production when relatedness is greater than that expected by chance alone. If instead, relatedness is equal to that expected by chance (R = 1/K), adding coinfections should have no effect on the parasite's ESS. Only if relatedness is less than that expected by chance (i.e., R<1/K) would the parasite be selected to increase its within-host growth rate to be greater than the baseline value of α-1.
What about transmission? How does the sum of propagule production by multiple coinfections affect the probability of successful spore formation and transmission from the infected host into the environment? The transmission probability at equilibrium is
Note that for K = 1 (and therefore R = 1), the result reduces to e-1, which converges on a previous result assuming a single infection per host [27]. For K>1, T* will be e-1 as long as R = 1. If on the other hand, for R<1, the transmission probability at equilibrium is less than e-1. For example, in a well mixed population of spores, R = 1/K, and T* = e-K. Thus, in general, unrelated coinfections reduce the overall transmission probability (and may increase virulence), but the result stems from a greater number of infections per host, not from an increase in the within-host growth rate of the parasite.
The total number of spores that emerge from an infected host at the parasite's equilibrium is simply E* = KN*T*, which simplifies to
Greater total spore production per host is thus expected at equilibrium as relatedness increases.
Linear case
The results above apply to the situation where the addition of each propagule has the same proportional effect as any previously added propagule. For comparison, it is useful to consider the situation where each propagule has the same absolute effect, giving a linear reduction in transmission probability with total propagule number. (This also the assumption in previous kin-selection models [5,13,14]). Consider for example the case where
T = 1 - α{NA + (K - 1)[NA(f + (1 - f)p) + Na(1 - f)(1 - p)]}. (14)
The expression in parentheses is the total number of propagules produced, Ntot, within the focal host containing the mutant parasite strain. Finally, Stot is the total number of spores produced in the focal host, Sfocal, plus the number produced in all other hosts, Sother, where
Sfocal = TNtot, (15)
and Sother = (H - 1)KNa(1 - αNaK). (16)
As host population size, H, goes to infinity and p goes to 0, the expected fitness of the mutant bearing the A allele converges on
Working as above, the equilibrium value is
which is both evolutionarily and convergence stable. The result shown in equation (18) is the same as that derived by Chao et al. [see eq. [5] in ref. [13]]; the result is also conceptually similar to the result derived by Brown et al. [12].
As previously, the benchmark value, R0, is found by setting K = 1, which gives N* = 1/(2α); this result converges to that first shown by Frank [5]. Coinfection results in increasing the rate of within-host reproduction when the right hand side of equation (18) is greater than the benchmark value for singleton infections, which is when 2>(K+1+f(K-1)), which is not biologically possible given that the minimum value for K is 1. Conversely, coinfection results in selection to reduce the rate of propagule production when 2<(K+1+f(K-1)), which is whenever there are coinfections (i.e., K>1). Nonetheless, holding the total number of coinfections constant, there will be selection to increase propagule production as f decreases; but it will always be less than the value that maximizes R0.
The total number of propagules produced within a host at equilibrium, Ntot is the number of coinfections, K, times N*. Hence
The transmission probability at equilibrium is T* = 1 - αNtot, which simplifies to
Thus, as for the exponential case, increasing relatedness among coinfections increases transmission at equilibrium, and reduces virulence. Finally, the total number of spores that emerge from an infected host at the parasite's equilibrium is E* = KN*T*, which simplifies to
Increasing relatedness among coinfections therefore results in greater total spore production at equilibrium.
Discussion
The results are consistent with the recent studies [12-14] suggesting that coinfection in spore-producing parasites would not necessarily result in selection for increased rates of within-host replication (Fig. 1). For the exponential case, increasing the number of coinfections selects for a decrease in propagule production whenever relatedness is greater than expected by chance alone (R>1/K). Otherwise (i.e., R = 1/K), the stable growth rate for each infection is unaffected by increasing the number of coinfections. For the linear case, increasing the number of coinfections selects for a decrease in the within-host growth rate by each infection, even when the probability that a coinfection is identical by descent is equal to zero (f = 0) and relatedness is equal to that expected in a well-mixed population of spores (R = 1/K). Nonetheless, the total number of propagules produced by all the coexisting coinfections does increase with the number of coinfections, unless relatedness (R) is equal to one. Thus overall virulence may increase with increasing numbers of coinfections, but this is not due to more aggressive growth by each of the individual infections.
Figure 1 Graphical results for the exponential and linear cases. Circles: R = 1/K. Squares: R = 1. The top row gives the evolutionarily stable number of spores produced during the within-host growth phase, N*. The middle row gives the total number of spores produced by all coinfections in a host at the parasite's ESS, which is equal to the produce of the number of coinfections, K, and the equilibrium number of spores produced by each infection, N*. The bottom row gives the per propagule probability of successful spore formation and release from the host, T*.
For example, suppose that there is only one infection. For the linear case, the propagule production for that infection would be equal to 1/(2α), and the transmission probability per propagule would be equal to one half. Now suppose there are two coexisting unrelated infections that each make at 1/(2α) propagules. In this case the total number of propagules produced, Ntot, would be 1/α, and the transmission probability (T = 1-αNtot) would be zero. Hence there would be strong between-host selection to reduce the number of propagules produced by each infection. The results above suggest that the continuously stable strategy for this example would be 1/(3α) for each infection, which yields a transmission probability of 0.333. This value is clearly less than that observed for a single infection, so the total impact of adding a coinfection is negative; but the reduction comes from adding the coinfection, not from more rapid reproduction by each coinfection.
These results are in contrast to previous models, which have shown that adding a coinfection selects for an increase in the parasite's growth rate [5-7]. The reason for the difference in results is not transparent, but may be due to the different assumptions. For example, the Nowak and May model [7] assumed that virulence is determined by the most aggressively growing strain, while the models above assume that the probability of spore formation and release (which could be correlated with virulence) is determined by the total number of propagules produced by all coinfections [see also [13,14]]. On the other hand, the results may stem from my assumption that the infection does not kill the host prior to spore formation. Schjorring and Koella [14] showed that coinfection in lethal parasites selected for greater parasite growth rates, but that coinfection by parasites with sub-lethal effects resulted in selection for reduced rates of parasite growth. Finally, the difference might stem from my simplifying assumption [following [5,14]] that the number of coinfections at equilibrium is a property of the host's biology, and not determined by the within-host growth rates by parasites.
My feeling, however, is that the different result stems (at least in part) from the fact that the present models include competition between all the spores produced in a large population of infected hosts. Thus the importance of between-host competition may outweigh the importance of within-host competition, and thus select for a reduced rate of reproduction. Consider, for example, the difference in assumptions between Frank's model [5] and the models presented above. Frank (page 71) considers parasites that are horizontally transmitted by a vector. The vector ingests a fixed volume of blood from an infected host, and transmits the parasite's transmission stages to an uninfected host. The relative fitness of a coinfecting parasite thus depends on the proportion of its transmission stages that occupy the blood, and hence there is selection to increase its rate of reproduction. In contrast, I assumed that the transmission stages (spores) become mixed together following their release from the hosts, and that parasite fitness is determined by how many spores are shed by the focal infection relative to the number of spores shed by all the individual infections in the parasite population. As such, the competition is shifted from being very local (within a single host) to more global (among all hosts), and selection is shifted from favoring a more aggressive reproductive strategy to favoring a more cooperative strategy.
In any case, the results of the present study are consistent with previous models showing that relatedness among coinfections would lead to selection to reduce the rate of within-host replication [5,9-13]. If all the coinfections are identical by descent, then each infection would be expected to produce an average of one Kth of the propagules expected in populations where only singleton infections are possible. The total number of propagules produced would then be expected to be equal to the number expected for a single infection per host.
The model was formulated here by examining the effect of total propagule production on the expected probability of spore formation and transmission into the environment. It assumes that each additional propagule produced by all coexisting infections reduces this probability; but the reduction may or may not be completely mediated through the effects that the propagules have on host survivorship following spore formation. The propagules may interfere with each other's success through ways other than reducing host survival. The actual effect of the infection on host fitness (virulence) is therefore not necessarily described by the same function that relates total propagule production to spore transmission; but virulence is nonetheless expected to be negatively correlated with the total number of propagules produced. As such, the results suggest that coinfection should lead to an increase in virulence, unless all the coinfections are identical by descent; but the increase is not due to more aggressive growth by each infection relative to that expected for solo infections.
Conclusion
The addition of unrelated coinfections may increase overall virulence; but the result stems from adding coinfections, rather than to more aggressive growth by the individual infections. However, holding coinfection number constant, increased relatedness among coinfections selects for less aggressive parasite growth, potentially resulting in a reduced impact for the overall infection.
Authors' contributions
CL constructed the model and wrote the paper.
Acknowledgements
I thank Aneil Agrawal, Farrah Bashey-Visser, Britt Koskella, Troy Day, Michelle Tseng, Jen Cianciolo, Sylvain Gandon, Stu West, Sofia Adolfsson and three anonymous reviewers for helpful comments on the manuscript. This work was supported by the US National Science Foundation (Population Biology Program: DEB-9904840; and Ecology of Infectious Diseases Program: DEB-03268742).
==== Refs
Levin BR Pimentel D Selection of intermediate rates of increase in parasite-host systems American Naturalist 1981 117 308 315 10.1086/283708
Anderson RM May RM Coevolution of hosts and parasites Parasitology 1982 85 411 426 6755367
May RM Anderson RM Epidemiology and genetics in the coevolution of parasites and hosts Proceedings of the Royal Society of London B, Biological Sciences 1983 219 281 313
Bremermann HJ Pickering J A game-theoretical model of parasite virulence Journal of Theoretical Biology 1983 100 411 426 6834864 10.1016/0022-5193(83)90438-1
Frank SA Models of parasite virulence Quarterly Review of Biology 1996 71 37 78 8919665 10.1086/419267
van Baalen M Sabelis MW The dynamics of multiple infection and the evolution of virulence American Naturalist 1995 146 881 910 10.1086/285830
May RM Nowak MA Coinfection and the evolution of virulence Proceedings of the Royal Society of London B, Biological Sciences 1995 261 209 215
Bonhoeffer S Nowak MA Mutation and the evolution of virulence Proceedings of the Royal Society of London B, Biological Sciences 1994 258 133 140
Frank SA A kin selection model for the evolution of virulence Proceedings of the Royal Society of London B, Biological Sciences 1992 250 195 197
Gandon S The curse of the pharaoh hypothesis Proceedings of the Royal Society of London B, Biological Sciences 1998 265 1545 1552 10.1098/rspb.1998.0470
Taylor PD Frank SA How to make a kin selection model Journal of Theoretical Biology 1996 180 27 37 8763356 10.1006/jtbi.1996.0075
Brown SP Hochberg ME Grenfell BT Does multiple infection select for raised virulence? Trends in Microbiology 2002 10 401 405 12217504 10.1016/S0966-842X(02)02413-7
Chao L Hanley KA Burch CL Dahlberg C Turner PE Kin selection and parasite evolution: Higher and lower virulence with hard and soft selection Quarterly Review of Biology 2000 75 261 275 11008699 10.1086/393499
Schjorring S Koella JC Sub-lethal effects of pathogens can lead to the evolution of lower virulence in multiple infections Proceedings of the Royal Society of London Series B-Biological Sciences 2003 270 189 193 10.1098/rspb.2002.2233
Ebert D Weisser WW Optimal killing for obligate killers: the evolution of life histories and virulence of semelparous parasites Proceedings of the Royal Society of London B, Biological Sciences 1997 264 985 991 10.1098/rspb.1997.0136
Day T Virulence evolution via host exploitation and toxin production in spore-producing pathogens Ecology Letters 2002 5 471 476 10.1046/j.1461-0248.2002.00342.x
Bonhoeffer S Lenski RE Ebert D The curse of the pharaoh: the evolution of virulence in pathogens with long living propagules Proceedings of the Royal Society of London B, Biological Sciences 1996 263 715 721
Nuismer SL Otto SP Host-parasite interactions and the evolution of ploidy Proceedings of the National Academy of Sciences, USA 2004 101 11036 11039 10.1073/pnas.0403151101
Templeton AR Milkman R Adaptation and the integration of evolutionary forces Perspectives on Evolution 1982 Sunderland: Sinauer 15 31
Day T Taylor PD Unifying genetic and game theoretic models of kin selection for continuous traits Journal of Theoretical Biology 1998 194 391 407 9778445 10.1006/jtbi.1998.0762
Maynard Smith J Evolution and theory of games 1982 Cambridge: Cambridge University Press
Eshel I Evolutionary and continuous stability Journal of Theoretical Biology 1983 103 99 111 10.1016/0022-5193(83)90201-1
Taylor PD Evolutionary stability in one-parameter models under weak selection Theoretical Population Biology 1989 36 125 143 10.1016/0040-5809(89)90025-7
Rousset F Genetic structure and selection in subdivided populations 2004 Princeton: Princeton University Press
Christiansen FB On the conditions for evolutionary stability for a continuously varying character American Naturalist 1991 138 37 50 10.1086/285203
Day T Taylor PD Evolutionary dynamics and stability in discrete and continuous games Evolutionary Ecology Research 2003 5 1 9
Lively CM Propagule interactions and the evolution of virulence Journal of Evolutionary Biology 2001 14 317 324 10.1046/j.1420-9101.2001.00268.x
|
16281984
|
PMC1310603
|
CC BY
|
2021-01-04 16:29:17
|
no
|
BMC Evol Biol. 2005 Nov 10; 5:64
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-64
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-591627113810.1186/1471-2148-5-59Research ArticleWhy do snails have hairs? A Bayesian inference of character evolution Pfenninger Markus [email protected]áková Magda [email protected] Dirk [email protected]èpraz Aline [email protected] Abteilung Ökologie & Evolution, J.W. Goethe-Universität, BioCampus Siesmayerstraße, 60054 Frankfurt/Main, Germany2 Deparment of Zoology, Charles University, Viniènà 7, 128 44 Praha 2, Czech Republic3 Department of Biology, University of Konstanz, Postbox 5560 M618, 78457 Konstanz, Germany4 Département d'Ecologie et Evolution, Université de Lausanne, Bâtiment de Biologie, Dorigny, 1015 Lausanne, Switzerland2005 4 11 2005 5 59 59 14 7 2005 4 11 2005 Copyright © 2005 Pfenninger et al; licensee BioMed Central Ltd.2005Pfenninger et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Costly structures need to represent an adaptive advantage in order to be maintained over evolutionary times. Contrary to many other conspicuous shell ornamentations of gastropods, the haired shells of several Stylommatophoran land snails still lack a convincing adaptive explanation. In the present study, we analysed the correlation between the presence/absence of hairs and habitat conditions in the genus Trochulus in a Bayesian framework of character evolution.
Results
Haired shells appeared to be the ancestral character state, a feature most probably lost three times independently. These losses were correlated with a shift from humid to dry habitats, indicating an adaptive function of hairs in moist environments. It had been previously hypothesised that these costly protein structures of the outer shell layer facilitate the locomotion in moist habitats. Our experiments, on the contrary, showed an increased adherence of haired shells to wet surfaces.
Conclusion
We propose the hypothesis that the possession of hairs facilitates the adherence of the snails to their herbaceous food plants during foraging when humidity levels are high. The absence of hairs in some Trochulus species could thus be explained as a loss of the potential adaptive function linked to habitat shifts.
==== Body
Background
Evolutionary theory predicts that costly structures must convey a fitness advantage to their bearers in order to be maintained over evolutionary time [1]. Flightlessness in birds and insects, limblessness in lizards and sightlessness in cave-dwelling organisms are some prominent examples of phenotypic regression due to the loss of adaptive function (reviewed in [2]). Molluscs in general and gastropods in particular display a fascinating diversity of elaborate shell structures [3,4] and have attracted considerable research efforts to explain them in adaptive terms [5-7]. The proposed roles invoked mechanical stability [8], defence against predators [9], sexual selection [10] and climatic selection [11]. However, the potential selective advantage of hair-like shell ornamentation of certain land snail species remains unknown.
These hairs can reach varying densities (up to 20 per squaremilimetre) and lengths (up to three millimetres). In some cases hardly visible, they confer an almost furry impression to the shell in others. These semi-rigid structures are part of the periostracum, a thin protein layer (conchiolin) secreted by the snail to cover the calcareous shell [12]. Building hairs requires the snail to have specialised glandular tissue and complex strategies to form them. Consequently, this trait can be assumed to be costly and should thus present a selective advantage to its bearers in order to be conserved.
Haired shells occur in several species of the Stylommatophoran families Polygyridae, Helicidae and Hygromiidae. These families are only distantly related [13], suggesting that this features has evolved several times independently. Haired shells are almost exclusively observed in species living in moist microhabitats, like layers of fallen leaves, broad-leaved vegetation, damp meadows or wet scree [14]. Such a correlation suggests an adaptive significance of the trait in such a habitat [1]; it was thus speculated that the hygrophobic hairs facilitate the movement in wet environments by relieving surface tension [14,15]. A correlation between haired shells and humid habitats is thus expected. In order to test this, we employed the recent Bayesian extensions of the comparative method, allowing to take mapping and phylogenetic uncertainty simultaneously into account [16]. With a diversity hotspot in South Germany, Eastern France and Switzerland, the land snail genus Trochulus s. str. (common name: Hairy snails) is particularly suited to address our question: its species exhibit variability in both hairiness and ecology. This study present the first comprehensive molecular phylogeny for the genus Trochulus Chemnitz, 1786 (until recently Trichia, Hartmann 1840) based on mitochondrial and nuclear loci. Finally, we tested experimentally whether the possession of haired shells indeed facilitates locomotion.
Results
Lineage identification and phylogenetic relations
The initial phylogenetic analysis on a COI data set of the presumed Trochulus species resolved 18 terminal clades, each with 0.99 posterior probabilities or higher (Figure 1). The uncorrected sequence divergence among those clades ranged from 0.029 to 0.173 (Table 2). Out of these lineages, nine could be assigned to existing taxa, because the species were sampled from the type locality and/or were morphologically unmistakable. The nine remaining clades, however, could not be unequivocally attributed to a taxonomic name. All eighteen identified lineages were used as molecularly defined operational taxonomic units in the subsequent analyses [17].
Figure 1 Unrooted consensus tree of 90,000 trees sampled by the Markov-chain in Bayesian analysis for the COI-fragment. Numbers on nodes indicate the Bayesian posterior probability.
Table 2 Pairwise uncorrected COI sequence divergence among lineages and species (mean ± s.d.).
A B C villosulus montanus clandest. caelatus D E F G H I biconicus nov. spec. villosus
B 0.111 ± 0.015 striolatus/plebeius
C 0.118 ± 0.015 0.091 ± 0.013 clade
villosulus 0.104 ± 0.014 0.073 ± 0.013 0.091 ±
0.014
montanus 0.143 ± 0.017 0.144 ± 0.017 0.138 ±
0.017 0.117 ± 0.016
clandestinus 0.135 ± 0.017 0.129 ± 0.016 0.140 ±
0.017 0.125 ± 0.016 0.083 ± 0.014 Jura clade
caelatus 0.144 ± 0.016 0.152 ± 0.018 0.141 ±
0.017 0.130 ± 0.016 0.092 ± 0.014 0.093 ± 0.013
D 0.146 ± 0.017 0.105 ± 0.016 0.125 ±
0.016 0.127 ± 0.017 0.142 ± 0.017 0.126 ± 0.016 0.138 ± 0.017
E 0.148 ± 0.017 0.109 ± 0.016 0.135 ±
0.016 0.117 ± 0.016 0.134 ± 0.016 0.104 ± 0.014 0.107 ± 0.015 0.079 ± 0.013 sericeus/hispidus
F 0.152 ± 0.017 0.112 ± 0.015 0.126 ±
0.015 0.119 ± 0.016 0.146 ± 0.016 0.113 ± 0.015 0.114 ± 0.015 0.072 ± 0.013 0.029 ± 0.008 clade
G 0.142 ± 0.017 0.139 ± 0.017 0.136 ±
0.018 0.109 ± 0.016 0.132 ± 0.017 0.117 ± 0.016 0.147 ± 0.016 0.120 ± 0.017 0.111 ± 0.016 0.105 ± 0.015
H 0.164 ± 0.017 0.153 ± 0.018 0.170 ±
0.017 0.169 ± 0.018 0.162 ± 0.017 0.126 ± 0.015 0.147 ± 0.017 0.116 ± 0.015 0.100 ± 0.014 0.101 ± 0.014 0.128 ± 0.016
I 0.150 ± 0.017 0.151 ± 0.017 0.155 ±
0.017 0.144 ± 0.017 0.148 ± 0.016 0.124 ± 0.015 0.133 ± 0.016 0.106 ± 0.014 0.102 ± 0.014 0.099 ± 0.013 0.106 ± 0.014 0.061 ± 0.010
biconicus 0.158 ± 0.018 0.136 ± 0.017 0.141 ±
0.018 0.128 ± 0.018 0.164 ± 0.020 0.154 ± 0.018 0.167 ± 0.019 0.145 ± 0.017 0.145 ± 0.018 0.145 ± 0.018 0.143 ± 0.018 0.179 ± 0.018 0.167 ± 0.017
nov. spec. 0.153 ± 0.017 0.137 ± 0.017 0.151 ±
0.017 0.142 ± 0.017 0.151 ± 0.016 0.128 ± 0.015 0.137 ± 0.016 0.100 ± 0.015 0.082 ± 0.014 0.081 ± 0.013 0.106 ± 0.015 0.080 ± 0.013 0.057 ± 0.011 0.161 ±
0.018
villosus 0.144 ± 0.018 0.139 ± 0.018 0.154 ±
0.018 0.152 ± 0.018 0.160 ± 0.018 0.153 ± 0.018 0.160 ± 0.018 0.168 ± 0.019 0.149 ± 0.018 0.151 ± 0.018 0.173 ± 0.019 0.171 ± 0.019 0.165 ± 0.018 0.163 ±
0.019 0.168 ± 0.019 villosa/alpicola
alpicolus 0.142 ± 0.016 0.141 ± 0.015 0.151 ±
0.018 0.153 ± 0.017 0.159 ± 0.014 0.151 ± 0.019 0.162 ± 0.019 0.167 ± 0.018 0.147 ± 0.016 0.150 ± 0.019 0.173 ± 0.019 0.173 ± 0.020 0.162 ± 0.016 0.161 ±
0.020 0.170 ± 0.020 0.006 ±
0.004
The Bayesian phylogenetic analysis of the entire data set (COI, 16S and ITS-1) showed the monophyly of the genus Trochulus within the Hygromiinae with high posterior probability, except for T. lubomirskii, which seems to be only distantly related to this genus (Figure 2). In addition to the early branching T. villosus/alpicolus clade, the genus is composed of three well supported subclades: first, a clade containing the T. striolatus/plebeius-like lineages together with T. villosulus, a second clade with ecologically divers species confined to the Jura mountains and the neighbouring Mittelland-plain and finally, a T. hispidus/sericeus-like clade, containing also T. biconicus and a new species.
Figure 2 Consensus tree of 90,000 trees sampled by the Markov-chain in Bayesian analysis for the total data set (1383 bp of COI, 16S and ITS1). Numbers on nodes indicate the Bayesian posterior probabilities.
Correlation of shell hairiness with habitat
The PCA on habitat humidity describing variables resulted in two meaningful axes, representing 79.7% and 13.4% of the total variation. The first component opposed sampling sites in shady woods and sites in sun exposed, open areas. This axis can therefore be interpreted as an evaporation gradient. The second axis is a gradient of the summer precipitation on one hand and the humidity demand of the vegetation on the other (Figure 3). It can thus be considered as a humidity gradient. The sampling sites appear as two distinct clusters that could be classified as either moist or dry (Table 1). The outlier (TA) was also considered to be humid, according to its high humidity levels. For each population, at least ten adult individuals were scored for the presence or absence of hairs (mixed populations were not found). Non-haired populations exclusively corresponded to species described in the literature as having smooth shells (Table 1). When plotting the hairiness of each population on the PCA, a complete congruence between humidity and hairiness became apparent: haired shells tended to occur at sites with low evaporation and/or high precipitation while smooth shells were found at places with high evaporation and/or lower precipitation (Figure 3).
Figure 3 Two first components of the PCA of the sampled localities over 9 environmental variables. Black dots: populations with haired individuals. Open squares: populations with hair-less individuals. Sampling sites above the dotted line are considered moist whereas those under it are dry.
Table 1 Table of sampling sites, presumed taxon, habitat characterisation and presence or absence of hairs.
Sampling site Abbreviation Geographical position Presumed taxon Habitat Humidity Hairs
Burgsinn, Bayern, Germany D3 50°09' 31"N 09°40'34"E T. striolatus/plebeius wood moist yes
Habichtstal, Bayern, Germany D4 50°02'54"N 09°25'41"E T. striolatus/plebeius wood moist yes
Dommershausen, Rheinland-Pfalz, Germany D5 50°07'45"N 07°23'47"E T. striolatus/plebeius wood moist yes
Bingen, Rheinland-Pfalz, Germany D6 49°55'56"N 07°58'57"E T. striolatus/plebeius wood moist yes
Eltville, Hessen, Germany D7 50°0059"N 08°04'28"E T. striolatus/plebeius T. sericeus/hispidus wood moist yes
Büchsenberg, Baden-Württemberg, Germany D8 48°04'55"N 07°37'23"E T. striolatus/plebeius wood moist yes
St. Seine l'Abbaye, Côte d'Or, France MdO 47°26'04"N 04°46'55"E T. sericeus/hispidus wood - yes
La Neirigue, Fribourg, Switzerland CH3 46°42'16"N 06°55'12"E T. clandestinus riverbank vegetation moist no
Barrage des Rossens, Fribourg, Switzerland CH6 46°43'33"N 07°06'55"E T. sericeus/hispidus wood moist yes
Gorges de la Jogne, Fribourg, Switzerland CH8 46°36'45"N 07°07'10"E T. sericeus/hispidus gorge moist yes
Ste Croix, Vaud, Switzerland CH10 46°50'44"N 06°32'02"E T. montanus open wood dry no
La Côte aux Fées, Neuchâtel, Switzerland CH11+12 46°50'74"N 06°32'42"E T. montanus grassland dry no
Col des Mosses, Vaud, Switzerland CH13 46°25'49"N 07°08'22"E T. villosus wood moist yes
Vallée du Rhône, Vaud, Switzerland CH15 46°19'43"N 06°13'39"E T. sericeus/hispidus wood moist yes
Sensetal, Bern, Switzerland CH18 46°49'46"N 07°19'19"E T. clandestinus riverbank vegetation dry no
Birseschlucht, Bern, Switzerland CH21 47°17'54"N 07°23'00"E T. caelatus cliff dry no
Birseschlucht, Bern, Switzerland CH24 47°16'56"N 07°23'13"E T. sericeus/hispidus wood moist yes
Château d'Oex, Vaud, Switzerland CHAT 46°16'23"N 07°21'42"E T. nov. spec. alpine meadow dry no
Bannalppass, Nidwalden, Switzerland TA 46°53'40"N 08°27'15"E T. alpicolus alpine meadow moist yes
Bannalppass, Nidwalden, Switzerland TB 46°53'43"N 08°27'21"E T. biconicus alpine meadow dry no
Velká Javořina, Velká nad Veličkou, Czech Republic TVJ 48°51'26"N 17°39'11"E T. villosulus wood moist yes
Bohuslavice u Zlína, Czech Republic PLB 49°09'19"N 17°37'29"E T. lubomirskii meadow - yes
Character state evolution
As the occurrence in moist habitats was systematically linked to the presence of hairs in Trochulus s.str., only a single analysis was necessary for both characters. The Bayesian analysis of character evolution suggested with high posterior probability that the most recent common ancestor of the genus Trochulus most likely possessed hairs and lived in a moist habitat (Figure 4). The analysis also revealed considerable mapping- and/or phylogenetic uncertainty in the reconstruction of crucial ancestral nodes (nodes 1–3 in Figure 4). The average Bayesian parameter estimate for the character change ratio was 2.50 ± 0.11 (mean ± s.d.), indicating that a loss of hairs associated with a transition from wet to dry habitats occurred more frequently than vice versa. This was in concordance with the parsimony reconstruction of character state changes on all different topologies of the 99% credibility set of trees. A minimum number of three independent losses of hairs / habitat transitions had a higher probability (0.59) than the only other observed pattern of two losses/one gain or three losses/no gain (0.41).
Figure 4 Bayesian reconstruction of ancestral states (hairs/no hairs, moist/dry habitat, respectively) on the topology of the Bayesian consensus tree (restricted to the Trochulus-clade).
Functional analysis
The analysis of variance showed that on a water-covered leaf surface, hairy shells required a significantly higher minimum force to overcome the adhesion (F = 720, d.f. = 2, p < 0.00001). There was no difference on a dry surface (F = 0.47, d.f. = 2, p = 0.37; Figure 5).
Figure 5 Mean (+/- s.d.) minimum necessary force to move haired and smooth shells on wet and dry leaf surfaces.
Discussion
Considering the limited number of sites sampled, we found a relatively large number of lineages, most of which could not be attributed to described species. This suggests that many other more or less morphologically similar entities may exist throughout the range of the genus. The existence of cryptic lineages could explain at least in part the current taxonomic uncertainty in Trochulus [18-20]. For example, several subspecies have been described for T. striolatus [21], which may well represent distinct evolutionary lineages such as described here. Given that the sequence divergence among the nine unidentified lineages is of the same magnitude as among described, morphologically and ecologically distinct species (Figure 1, Table 2), it can be reasonably assumed that the cryptic lineages within the striolatus/plebeius and hispidus/sericeus clades correspond to good species. Even under the assumption of an exceptionally fast molecular clock in land snails of up to 5% sequence divergence per one million years [22], the lineages in the striolatus/plebeius clade, for example, persisted for at least two million years as independent evolutionary entities. The existence of more or less cryptic lineages or species is not an unusual finding in land snails [23-25]. In contrast to the high divergence of the unidentified lineages, the comparatively small genetic distance between T. villosus and T. alpicolus indicated a questionable specific distinction between these two taxa. Detailed phylogeographic analyses in addition to morphometric and ecological studies will be necessary to disentangle the species limits of these cryptic Trochulus complexes, clarify the taxonomy and reveal their evolutionary history. In addition, the species T. lubomirskii, which was placed by Schileyko [26] into the subgenus Plicuteria, may not belong to the genus Trochulus at all.
A haired shell appears as the ancestral state in the genus Trochulus. This inference is strengthened by the observation that some of the hair-less species do possess some as juveniles. During the evolutionary history of the genus Trochulus, hairs appear to have been lost several times independently (Fig. 3, Table 1) and this was always correlated with a shift in habitat (i.e. hairs are only present in moist habitats, mostly woodlands). This suggests that hairs potentially have an adaptive function in humid habitats and once the presumed selective pressure for the maintenance of these costly protein structures is relieved, they are lost. Such a correlation makes certain potential adaptive explanations for hairiness unlikely: defence against predators or mechanical stability have no obvious reasons to co-vary with the humidity characteristics of a habitat.
The facilitation of locomotion by decreasing the adhesion to water films in humid environments had been previously hypothesised to be the selective advantage of a haired shell [14,15]. However, the results of our experiments have shown that the opposite is true. The presence of hairs significantly increased the minimum force necessary to move shells over wet surfaces. Having thus shown that the initial hypothesis [14] is at least in this case not applicable, we propose an alternative: haired shells may confer an selective advantage by increasing the adhesion to the water film on the unstable, moving leaves of their feeding plants during foraging (Figure 5). Indeed, snails are mostly active during phases of high ambient humidity [27] when leaves are covered with a water film due to rain, fog or dew. This water film is usually in contact with the shell during locomotion (Figure 6). Observation shows that Trochulus species in moist habitats preferentially forage on large-leaved herbaceous plants like Adenostyles, Urtica, Homogyne or Tussilago [28]. Hence, falling off the leaf and needing to crawl up again to this feeding site (that can be one meter above ground) represents a considerable effort given the exceedingly costly and ineffective locomotion of land snails [29]. In dry habitats on the contrary, snail species avoid the hard plant matter typical for this habitat and preferentially feed on dead material lying on the ground [28,30], where a mechanism increasing shell adhesion offers no obvious advantage to its bearer. This interpretation is supported by the fact that phylogenetically distantly related haired species, such as Helicodonta obvoluta and Isognomostoma isognomostoma, are found in the same habitats and have in general similar life-styles [31]. However, as long as the positive effect of increased adherence to food plants on the individual fitness is not proven, this remains a hypothesis and does not preclude additional or even other adaptive functions of haired shells.
Figure 6 Active T. villosus foraging on a leaf. Note that the water-film on the leaf is adhering to the shell.
Conclusion
The present comparative analysis suggested that hairs on the shell confer a selective advantage in humid habitats only and are thus lost in drier habitats. In other words, the variability of hairiness within the genus Trochulus could be explained in terms the loss of its adaptive function in a selectively different environment.
Methods
Taxon sampling
Analyses were undertaken on twelve of the about 15 currently recognised species presumed to belong to the genus Trochulus s. str. Chemnitz, 1786 (Hygromiidae, Stylommatophora). However, the exact number of existing species is not known, because the species limits of the widely distributed T. hispidus and T. sericeus on the one hand and T. plebeius and T. striolatus on the other are equivocal [19,20], the validity of several described taxa is disputed [18,32] and newly discovered species are not yet formally described (Pfenninger, unpublished data). Since initial analyses showed the existence of cryptic lineages, several populations for each of the putative species were sampled (Table 1). Four species from other genera of the subfamily Hygromiinae and two species of the family Helicidae were used as potential outgroups [33] (GenBank accession numbers AY546263, AY546343, AY546303, AY546284, AY546364, AY546324, AY546283, AY546363, AY546323, AY546291, AY546371, AY546331).
DNA sequencing, lineage identification and phylogenetic analysis
Entire snails were crushed and vortexed in 10% w/v laundry detergent solution for storage at room temperature and tissue digestion [34]. For 78 individuals, a 512 bp segment of the cytochrome oxidase subunit I gene (COI) was amplified with PCR and sequenced. For selected individuals representing the major evolutionary lineages inferred in the previous analysis, a 362 bp fragment of the large subunit mitochondrial ribosomal gene (16S) and 509 bp of the internal transcribed spacer 1 (ITS-1) from the nuclear ribosomal cluster were additionally amplified and sequenced. An amount of 0.2 to 1 ng total DNA (quantified on a 1% agarose gel using a λ Hind III marker) were used as template in polymerase chain reaction (PCR). Specific PCRs were performed with the primers, amplification conditions and temperature profiles shown in Table 2. Primers were used for both specific PCR and subsequential automated direct sequencing. PCR products were purified using E.N.Z.A. Cycle Pure Kit (peqlab, Erlangen, Germany). Ten ng per sample were subjected to cycle sequencing using the ABI Prism Big Dye terminator kit (Perkin-Elmer, Norwalk, CT, USA). Sequencing reactions were electrophoresed on an ABI 377 automated DNA sequencer. In order to verify the results, gene products were sequenced in both directions and the two strands were aligned with SEQUENCE NAVIGATOR 1.0.1 (Perkin-Elmer, Norwalk, CT, USA). Sequences were deposited in GenBank under accession numbers DQ217794-DQ217831. The orthologous DNA sequences were initially aligned using the default settings of CLUSTAL X [35] and optimised by eye. The most likely models of sequence evolution and their parameters according to the Akaike information criterion were inferred for each DNA data partition using MODELTEST v. 3.4 [36]. In an initial analysis, we used the COI data set to identify evolutionary lineages. A 99.9% credible set of phylogenetic trees was estimated with the program MRBAYES [37] by sampling the tree space using a Metropolis coupled Monte Carlo Markov chain, implementing a TN+I+Γ model of COI sequence evolution (where TN denotes Tamura-Nei, Γ is the shape parameter of the gamma distribution and I the proportion of invariant sites). Initial runs as well as a posterior inspection of the likelihoods in the final run showed that a burn-in phase of 10,000 generations was largely sufficient for both analyses to allow the likelihood values to reach convergence. The chain was run for 10,000,000 generations and sampled every 100th generation. An unrooted majority consensus tree was computed from the sampled trees, excluding the trees sampled in the burn-in phase. The procedure was repeated for the phylogenetic data set where the Markov chain was run with separate models of sequence evolution for each data partition (GTR (general time reversible)+I+G for 16S and TVM (transversional model)+ Γ for ITS-1). Outgroup status was assigned to Helixaspersa [33].
Correlation of habitat humidity with shell hairiness
The direct estimation of humidity levels for sampling sites is difficult without long-term observation. However, the precipitation regime, habitat structure and vegetation at a sampling site can give clues on the degree of humidity experienced by the snails. For this behalf, five variables were recorded for all but one population belonging to Trochulus s.str. species. To characterise the microhabitat conditions, the mean light- and humidity indicator values [38] of the three most abundant herbaceous plant species at each sampling site were recorded (variables LIGHTIND and HUMIND). The evaporation regime is strongly influenced locally by the exposure to sun and wind, which was accounted for by characterising each sampling site as either i) entirely shadowed (2), partially or sometimes shadowed (1) and never shadowed (0) (variable SHADOW) and either ii) situated in a closed wood (2), open wood or forest edge (1) or not in a wood (0) (variable WOOD). Ultimately, the humidity conditions of a site depend on the precipitation in the area. As Trochulus species are active mainly during summer, we have recorded the average long-term precipitation from April to September (variable SUMMERPREC). This information was extracted from the climate layers with a spatial resolution of 0.5 min implemented in the computer program DIVA-GIS version 4.2 for the spatial analysis of biodiversity [39]. The variables were summarised in a principal component analysis (PCA).
For all Trochulus s.str. populations investigated, the presence or absence of hairs on the shell of at least 10 adult individuals was recorded. As the hairs may wear off during adulthood (although rarely completely), the lack of the typical hair pits in the fine sculpture of the shell was taken as evidence for their principal absence. The presence or absence of hairs of the respective populations was then plotted on the PCA ordination.
Bayesian estimation of ancestral character states
In a first approach, we derived the posterior probability distribution of ancestral character states and their rate of change from 3000 trees sampled at random from the 99.9% credibility set of phylogenetic trees, using the Bayesian approach as implemented in the program MULTISTATEBAYES [40]. Applying an uninformative (uniform) prior on the rate parameter distribution, a Markov chain was run for 1,000,000 generations after it reached convergence. The estimated rate parameter ratio for both directions of character change as well as the reconstructed ancestral states for each internal node of the tree investigated was sampled every 200th generation. This procedure estimates i) the probability that the ancestral node existed in the first place and ii) the probabilities of both character states at the respective node. These three probabilities sum up to 1, thus simultaneously taking phylogenetic and character mapping uncertainty into account. In a second approach, the most parsimonious number of character state changes was reconstructed for each of the 99.9% credibility set of phylogenetic trees using the ANCESTRAL STATE RECONSTRUCTION module in MESQUITE [41]. The different reconstructions were then weighted according to the posterior probability of the corresponding tree [42].
Adhesion experiments
The minimum force necessary to move Trochulus shells (upwards oriented apex) with or without hairs over dry and wet, horizontal leaf surfaces was measured. For this behalf, we have chosen the largest species, T. villosus. It would have been desirable to use shells of other lineages as well, however, it was not possible to measure the force necessary to move smaller shells with the necessary accuracy. Twelve T. villosus shells were glued to thin nylon strings. The strings were led over a roll with a small aluminium basket fastened on the other end. Small weights were incrementally added to the basket until the shell began to slide. This was replicated five times for each shell on both water film covered and on dry surfaces. Then, the hairs were mechanically removed to obtain smooth shells and the procedure was repeated. For each condition, differences in minimum force needed to move the shells with or without hairs were tested for significance with an ANOVA design.
Authors' contributions
MP designed the study, collected parts of the material, performed the analyses and drafted the manuscript. MH contributed to the samples and carried out part of the molecular work. DS also contributed to the samples and participated in the statistical analyses. AD contributed samples, participated in the design of the study and helped to draft the manuscript. All authors were involved in preparation of the manuscript and approved the final version.
Table 3 Primers used for specific PCR and direct sequencing, amplification conditions and temperature profiles.
Primer Sequence amplification conditions temperature profile
COI universal [43] 5'-GGTCAACAATCATAAAGATATTGG-3' 5'-TAAACTTCAGGGTGACCAAAAAATCA-3' total volume 25 μl with: 0.17 mM dNTPs 3 mM MgCl2 in 1 × PCR buffer 0.13 μM of each primer 1 unit Taq polymerase (Invitrogen) 1 cycle of 2.5 min at 94°C
40 cycle 30s at 90°C
1 min at 48°C
1 min at 72°C
1 cycle of 10 min at 72°C
16S universal [44] 5'-CGGCCGCCTGTTT ATCAAAAACAT-3' 5'-GGAGCTCCGGTTTGAACTCAGATC-3' total volume 15 μl with: 0.1 mM dNTPs 2.5 mM MgCl2 in 1 × PCR buffer 0.2 μM of each primer 0.5 unit Taq polymerase (Invitrogen) 1 cycle of 2.5 min at 90°C
10 cycles of 50s at 92°C
30s at 44°C
40s at 72°C
36 cycles of 30s at 92°C
40s at 48°C
40s at 72°C
1 cycle of 3 min at 72°C
ITS-1 mollusc specific [45] 5'-TAACAAGGTTTCCGTAGGTGAA-3' 5'GCTGCGTTCTTCATCGATGC-3' total volume 15 μl with: 0.3 mM dNTPs 2.5 mM MgCl2 in 1 × PCR buffer 0.18 μM of each primer 0.5 unit Taq polymerase (Invitrogen) 1 cycle of 3 min at 94°C
40 cycles of 30s at 92°C
30s at 52°C
1 min at 72°C
1 cycle of 5 min at 72°C
Acknowledgements
We thank Holger Geupel for technical assistance. MP acknowledges the financial support of the A. Messer Stiftung. Jacques Hausser, Aris Parmakelis, Christian Albrecht and an anonymous referee gave valuable comments on the manuscript. We thank Margret Gosteli and the Museum of Natural History in Bern for samples and additional information.
==== Refs
Harvey P Pagel MD The comparative method in evolutionary biology 1991 Oxford: Oxford University Press
Porter ML Crandall KA Lost along the way: the significance of evolution in reverse Trends Ecol Evol 2003 18 541 547 10.1016/S0169-5347(03)00244-1
Vermeij GJ A natural history of shells 1995 Princeton: Princeton University Press
Godan D Molluscs: Their significance for science, medicine, commerce and culture 1999 Berlin: Parey
Palmer AR Quantum changes in gastropod shell morphology need not reflect speciation Evolution 1985 39 699 705
Hellberg ME Balch DP Roy K Climate-driven range expansion and morphological evolution in a marine gastropod Science 2001 292 1707 1710 11387473 10.1126/science.1060102
West KCA Morphology and behavior of crabs and gastropods from Lake Tanganyika, Africa: Implications for lacustrine predator-prey coevolution Evolution 1991 45 589 607
Britton JC The relationship between position on shore and shell ornamentation in 2 size-dependent morphotypes of Littorina striata, with an estimate of evaporative water-loss in these morphotypes and in Melarhaphe neritoides Hydrobiologia 1995 309 129 142
Wilson AB Glaubrecht M Meyer A Ancient lakes as evolutionary reservoirs: evidence from the thalassoid gastropods of Lake Tanganyika Proc R Soc Lond Ser B-Biol Sci 2004 271 529 536 10.1098/rspb.2003.2624
Schilthuizen M Sexual selection on land snail shell ornamentation: a hypothesis that may explain shell diversity BioMedCentral Evolutionary Biology 2003 3 13 10.1186/1471-2148-3-13
Goodfriend GA Variation in land-snail shell form and size and its causes – a Review Systematic Zoology 1986 35 204 223
Saleuddin ASM Ultrastructural studies on formation of periostracum in Helix aspersa (Mollusca) Calcified Tissue Research 1976 22 49 65 1000343 10.1007/BF02010346
Wade CM Mordan PB Clarke B A phylogeny of the land snails (Gastropoda: Pulmonata) Proc R Soc Lond B 2000 268 413 422
Suvorov AN Some mechanisms of adaptation to the wet microhabitats in higher geophila (Mollusca, Pulmonata) Zhurnal Obshchei Biologii 1999 60 177 188
Solem A The shell makers 1974 New York: John Wiley & Sons
Ronquist F Bayesian inference of character evolution Trends Ecol Evol 2004 19 475 481 16701310 10.1016/j.tree.2004.07.002
Baxter ML The promise of a DNA taxonomy Philosophical Transactions of the Royal Society of London – Series B: Biological Sciences 2004 359 669 679 10.1098/rstb.2003.1447
Falkner G Trichia (Trichia) graminicola n.sp. aus Südbaden (Gastropoda: Helicidae) Archiv für Molluskenkunde 1973 103 209 227
Forcart L New researches on Trichia hispida (Linnaeus) and related forms Proceedings of the first European Malacological Congress 1962 79 93
Perrin N Honsberger P Pontet A Approche biochimique et morphologique des espèces du genre Trichia (Helicidae, Gastropoda) de Suisse occidentale Revue Suisse de Zoologie 1984 92 483 495
Falkner G Beiträge zur Nomenklatur der europäischen Binnenmollusken, VIII: Nomenklaturnotizen zu europäischen Hygromiidae (Gastropoda: Stylommatophora) Heldia 1995 2 97 107
Thomaz D Guiller A Clarke B Extreme divergence of mitochondrial DNA within species of pulmonate land snails Proc R Soc Lond Ser B-Biol Sci 1996 263 363 368
Pinceel J Jordaens K Pfenninger M Backeljau T Rangewide phylogeography of a terrestrial slug in Europe: evidence for Alpine refugia and rapid colonization after the Pleistocene glaciations Mol Ecol 2005 14 1133 1150 15773941 10.1111/j.1365-294X.2005.02479.x
Pfenninger M Magnin F Phenotypic evolution and hidden speciation in Candidula unifasciata ssp (Helicellinae, Gastropoda) inferred by 16S variation and quantitative shell traits Mol Ecol 2001 10 2541 2554 11742553 10.1046/j.0962-1083.2001.01389.x
Wilke T Duncan N Phylogeographical patterns in the American Pacific Northwest: lessons from the arionid slug Prophysaon coeruleum Mol Ecol 2004 13 2303 2315 15245403 10.1111/j.1365-294X.2004.02234.x
Schileyko AA On the systematics of Trichia s. lat. (Pulmonata: Helicoidea: Hygromiidae) Malacologia 1978 17 1 56
Cook A Barker GM Behavioural Ecology: doing the right thing, in the right place at the right time The Biology of Terrestrial Molluscs 2001 Oxon: CABI Publishers 447 488
Frömming E Biologie der Mitteleuropäischen Landgastropoden 1954 Berlin: Dunker & Humbold;
Denny M Locomotion: the cost of gastropod crawling Science 1980 208 1288 1290
Kerney MP Cameron RAD Jungbluth JH Die Landschnecken Nord- und Mitteleuropas 1983 Hamburg; Berlin: Paul Parey
Speiser B Barker GM Food and feeding behaviour The Biology of Terrestrial Molluscs 2001 Oxon: CABI Publishers 259 288
Gittenberger E Neuteboom WH On Trichia alpicola (Eder, 1921) from Switzerland (Mollusca: Gastropoda Pulmonata:Hygromiidae) and the spiral sculpture on its shell Zoologische Mededelingen 1991 65 247 250
Steinke D Albrecht C Pfenninger M Molecular phylogeny and character evolution in the Western Palaearctic Helicidea s.l. (Gastropoda, Stylommatophora) Molecular Phylogenetics & Evolution 2004 33 724 734 15288050 10.1016/j.ympev.2004.03.004
Bahl A Pfenninger M A rapid method of DNA isolation using laundry detergent Nucleic Acids Res 1996 24 1587 1588 8628700 10.1093/nar/24.8.1587
Thompson JD Gibson TJ Plewniak F Jeanmougin F Higgins DG The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools Nucleic Acids Res 1997 25 4876 4882 9396791 10.1093/nar/25.24.4876
Posada D Crandall KA MODELTEST: testing the model of DNA substitution Bioinformatics 1998 14 817 818 9918953 10.1093/bioinformatics/14.9.817
Huelsenbeck JP Ronquist F Bayesian inference of phylogeny Bioinformatics 2001 17 754 755 11524383 10.1093/bioinformatics/17.8.754
Ellenberg H Weber HE Düll R Wirth V Werner W Paulißen D Zeigerwerte von Pflanzen in Mitteleuropa 1992 2 Göttingen: Goltze KG
Hijmans RJ Guarino L Cruz M Rojas E Computer tools for spatial analysis of plant genetic resource data: 1. DIVA-GIS Plant Genetic Resources Newsletter 2001 127 15 19
Pagel M Meade A Barker DM Bayesian estimation of ancestral character states on phylogenies Syst Biol 2004 53 673 684 15545248 10.1080/10635150490522232
Maddison WP Maddison DR Mesquite: a modular system for evolutionary analysis. Version 1.05 2004
Huelsenbeck JP Rannala B Larget B A Bayesian framework for the analysis of cospeciation Evolution 2000 54 352 364 10937213
Folmer O Black M Hoeh W Lutz R Vrijenhoek R DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates Mol Mar Biol Biotech 1994 3 294 299
Palumbi S Martin A Romano S McMillan W Stine O Grabowski G The simple fool's guide to PCR version 20 1991 Honolulu: University of Hawaii
Armbruster GFJ van Moorsel CHM Gittenberger E Conserved sequence patterns in the non-coding ribosomal ITS-1 of distantly related snail taxa J Molluscan Stud 2000 66 570 573 10.1093/mollus/66.4.570
|
16271138
|
PMC1310604
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Nov 4; 5:59
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-59
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2771630574510.1186/1471-2105-6-277Research ArticleDiscover protein sequence signatures from protein-protein interaction data Fang Jianwen [email protected] Ryan J [email protected] Yinghua [email protected] Gerald H [email protected] Bioinformatics Core Facility, University of Kansas, Lawrence, KS 66045, USA2 Information and Telecommunication Technology Center, University of Kansas, Lawrence, KS 66045, USA3 Molecular Graphics and Modeling Laboratory, University of Kansas, Lawrence, KS 66045, USA2005 23 11 2005 6 277 277 18 7 2005 23 11 2005 Copyright © 2005 Fang et al; licensee BioMed Central Ltd.2005Fang et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The development of high-throughput technologies such as yeast two-hybrid systems and mass spectrometry technologies has made it possible to generate large protein-protein interaction (PPI) datasets. Mining these datasets for underlying biological knowledge has, however, remained a challenge.
Results
A total of 3108 sequence signatures were found, each of which was shared by a set of guest proteins interacting with one of 944 host proteins in Saccharomyces cerevisiae genome. Approximately 94% of these sequence signatures matched entries in InterPro member databases. We identified 84 distinct sequence signatures from the remaining 172 unknown signatures. The signature sharing information was then applied in predicting sub-cellular localization of yeast proteins and the novel signatures were used in identifying possible interacting sites.
Conclusion
We reported a method of PPI data mining that facilitated the discovery of novel sequence signatures using a large PPI dataset from S. cerevisiae genome as input. The fact that 94% of discovered signatures were known validated the ability of the approach to identify large numbers of signatures from PPI data. The significance of these discovered signatures was demonstrated by their application in predicting sub-cellular localizations and identifying potential interaction binding sites of yeast proteins.
==== Body
Background
The development of high-throughput technologies for discovering interactions between proteins has made it possible to screen entire proteomes and produce large protein-protein interaction (PPI) datasets. Different methods of PPI detection, including yeast two-hybrid assays [1-3], mass spectrometry of coimmunoprecipitated protein complexes [4,5], and correlated messenger RNA profiles [6,7], discover PPIs of variable reliability and the majority of putative PPIs are of low confidence. Despite the presence of false positives, the wealth of PPI data generated over the past several years is the source of many publicly available databases, such as the Database of Interacting Proteins (DIP [8]) and the MIPS mammalian protein-protein interaction [9]. The availability of these large datasets is now enabling researchers to predict undiscovered PPIs and hypothesize the function and sub-cellular localization of proteins.
PPI data has been used to analyse domain-domain interactions (DDIs), based upon the widely accepted hypothesis that proteins interact with one another via conserved domains (Figure 1). Large-scale PPI databases are used to identify correlated domains that are implicated in the binding of protein partners. When one of these sequence signatures is observed in a newly discovered protein, it is possible to predict its interactions with other proteins based on the knowledge base of correlated domains. DDIs were thus used to predict the function and PPIs of newly discovered proteins [10]. Deng et al. [11] used maximum likelihood estimation to discover DDIs, which were then used to predict the likelihood of interaction for any protein pair. Other recent forms of DDI analysis include the use of interacting domain profile pairs [12], and a domain combination based probabilistic framework [13].
Figure 1 A scheme illustrates the procedure of inferring DDIs from PPIs. Colored shapes represent sequence signatures. Suppose protein H (the host) interacts with four guest proteins (G1, G2, G3, G4) and all signatures in the schema are known with the exception of the one represented by purple hexagon. In this case only interactions with G1 and G2 are useful in inferring DDIs. In this study we used MEME program to identify all signatures shared by guests.
Very recently, PPI data, in conjunction with structural information, were used to produce a set of putative binding motif pairs [14]. The significance of motif discovery stems from the idea that the actual binding sites most directly responsible for the binding of proteins are probably smaller than whole domains. Thus, the discovery of these smaller sequence signatures allows researchers to structurally characterize PPIs with more precision.
This study was also based on the assumption that PPIs result from the interactions of conserved sequence signatures. Unlike Li and Li's work [14], our method of PPI data mining did not use structural data, which are well known to be biased towards small, globular proteins. In this paper, a set of guest proteins represents those proteins known from PPI database to share a common interacting partner, i.e. a host protein. If a protein interacts with itself, it is a host as well as a guest. Signatures shared by sets of guest proteins were initially discovered using the program MEME [15] on a large PPI dataset. Searches of sequence signature databases for the identified motifs revealed that 84 distinct motifs had not been characterized previously. The significance of these newly discovered signatures was then demonstrated by their application in predicting the sub-cellular localization of yeast proteins and identifying potential interacting sites.
Results
A sequence signature is defined as a "highly conserved region", a sequence pattern that is found repeatedly in a group of related protein sequences [15]. By this definition, a sequence signature could be a protein family, functional domain, functional site, or any conserved region of unknown function, and thus the actual physical manifestation of a signature can vary greatly in size. In our study, sequence signatures were derived from MEME motifs. We wrote numerous Perl scripts and used a MySQL relational database to facilitate the processes of data collection, program execution, and data analysis.
Discovery of sequence signatures
The 1923 batch executions of MEME yielded 3108 sequence signature models shared by the 1555 distinct guest proteins of 944 host proteins from Saccharomyces Cerevisiae (baker's yeast) (see details in Methods). Of the 6770 distinct PPIs actually involved in building these signature models, 1509 (22.3%) were identified as high confidence interactions in the PPI dataset. When compared to the percentage of high confidence PPIs in the input files (20.7%), the percentage of high confidence PPIs used to construct motif models represents a statistically significant difference (p-value = 0.0013, two-tailed t-test).
Signature model length varied from 10 to 300 residues: the minimum and maximum lengths specified for each MEME execution. Only 25 models (<1%) were as long as 300 residues, which indicated that the maximum length used in this study was appropriate. MEME splits one sequence signature in two if its length is greater than the specified maximum. Thus, less than 2% of the 3108 models were the result of splitting sequence signatures. The average model length was 33.6 residues, with a standard deviation of 40.3. It should be pointed out that there was redundancy among these signatures because different host proteins may interact with similar sets of guest proteins. We did not attempt to identify distinct signatures because that was not the main goal of the present study. Instead, we identified distinct novel sequence signatures (84 distinct signatures out of 172 initial results, see below for details). Thus, we estimated that overall about half of these signatures were distinct.
Occurrence of discovered sequence signatures in the yeast genome
Using the 3108 signature models discovered in MEME as input, MAST [16] was used to scan the entire genome of S. cerevisiae for occurrences of these sequence signatures in proteins that were not used to build the motif models. 1,993 protein sequences contained one or more of the sequence signatures, a 28% increase over the 1,555 proteins used to construct the signatures. Although this increase indicates that the newly discovered sequence signatures have some potential predictive value, any predictions based on these sequence signatures would be limited to approximately one-third of the S. cerevisiae genome. A broader application will be feasible only when more reliable PPI data are available.
Novelty of discovered sequence signatures
Using the standalone version of InterProScan, the consensus sequences of 2337 of the discovered motif models were found to match signatures listed in one of the InterPro member databases. When the online version of InterProScan was used, an additional 599 sequence signatures were matched to un-integrated entries of the InterPro member databases. 172 novel sequence signatures remained. FASTA searches, which were the basis for the grouping of similar/identical sequence signatures, resulted in the creation of 84 distinct, novel sequence signatures. The length of these novel sequence signatures ranged from 10 to 36 residues. Table 1 provides a list of several of these novel signatures. A complete list can be found on the supplementary website . Interestingly, when InterProScan was used to match consensus sequence signatures to the Pfam database alone, only 545 (~18%) of the signatures were matched to known signatures.
Table 1 Novel sequence signature examples.
Signature id Host Consensus sequence Length
YDL166C_1 YDL166C EVLCCQLPKWCGFFQM 16
YML094_4 YML094 QRQGKLEVPGYVDIVKTSSGNEMPPQ 26
YOL094C_3 YOL094C LWVEKYRPKNLDEVCGN 17
YGL063W_2 YGL063W VKAVEGRKKGKEGKASQLVDLKFALAEDKV 30
YOR335C_5 YOR335C AQSVGCRVDFKNPHDIIEGINAGEIE 26
Localization prediction
Using signature sharing information, the sub-cellular localizations of 108 proteins were predicted based on the known locations of 5416 budding yeast proteins (see details in Methods). 52 predictions agreed with the ontology annotations of the SGD and 24 disagreed (~68% accuracy). The accuracy of the remaining 32 (Table 2) predictions could not be assessed, as the locations of these proteins have yet to be determined empirically. It is reasonable to believe these predictions would have similar prediction accuracy.
Table 2 Predicted localizations without known annotations from SGD. Evidence notation: 1: the ORF is a host, all or most guests are in the same location. 2: a guest, its host and all or most siblings are in the same location; 3: also a guest, but the location of host is unknown, all or most siblings are in the location. If there are multiple predictions for one ORF, the evidence and/or host names are concatenated in the corresponding columns.
ID ORF Predicted_location Evidence(s) Host name(s)
1 Q0105 cytoplasm 1
2 YAL046C cytoplasm, nucleus 1
3 YAR073W cytoplasm 2 YMR217W
4 YBL041W cytoplasm, nucleus 1,2 YJL001W, YPR103W, YGR135W, YML092C, YGR253C, YER094C, YGL011C
5 YBL092W cytoplasm, nucleus 2 YGR034W, YDL191W
6 YBR257W cytoplasm, nucleus 2 YHR203C, YJR014W, YJR145C
7 YCR031C cytoplasm, nucleus 2 YGR034W
8 YCR072C cytoplasm, nucleus 1
9 YDL075W cytoplasm, nucleus 3 YDR292C
10 YDR064W cytoplasm, nucleus 1,2 YGR262C, YAL035W
11 YDR109C cytoplasm, nucleus 2 YJR024C
12 YDR287W cytoplasm, nucleus 2 YEL041W
13 YEL041W cytoplasm, nucleus 1,2 YDL236W, YHL046C
14 YER094C cytoplasm, nucleus 1,2,3 YFR050C, YGL011C, YPR103W, YBL041W, YJL001W, YML092C, YGR253C, YGR135W
15 YGL063W cytoplasm, nucleus 1,2 YDR158W, YDR007W
16 YGL224C cytoplasm, nucleus 2 YMR009W, YDL219W, YJR024C
17 YHR016C cytoplasm, actin 2 YBL007C
18 YHR044C cytoplasm, nucleus 2 YDR074W
19 YJL213W cytoplasm 2 YGR094W
20 YKL104C cytoplasm, nucleus 2 YDR127W, YPL160W, YDR211W, YDR394W, YER110C
21 YLR209C nucleolus, nucleus 1
22 YLR359W cytoplasm 2 YGL234W
23 YMR084W cytoplasm, nucleus 1,2 YDR211W
24 YMR130W cytoplasm, nucleus 2 YJR024C
25 YMR217W cytoplasm 1
26 YOL114C cytoplasm, nucleus 2 YPL160W
27 YOR054C cytoplasm, nucleus 2 YDR454C, YBR252W
28 YOR093C cytoplasm 2 YBR208C
29 YOR111W actin 2 YDL161W
30 YPL003W cytoplasm, nucleus 2 YDR054C
31 YPL171C cytoplasm, nucleus 2 YKR031C
32 YPL217C nucleolus,nucleus 2 YLR197W, YHR052W, YDR449C
Homology modeling and detection of putative interacting sites
The exact biological meanings of these novel sequence signatures can only be determined by web-lab experiments. One possible role of these signatures is to serve as the binding sites for protein-protein interactions. A binding site should have significant exposure to solvent. In order to assess this possibility, we built homology models for those yeast proteins containing novel signatures and having good model templates [see Additional file 1]. Using DSSP software program [17], we calculated the proposition of residues of signatures appearing on the surface (residues with solvent exposed surface ≥ 25 Å2). Statistical analysis (two-sided Fisher's exact test) confirmed that residues of signatures occurred on the surface more frequently than would be expected by chance (P < 0.04, Fisher's exact test). Thus we hypothesized these signatures are potential binding sites and plan to use site-directed mutagenesis and NMR spectrometry to verify the bioinformatics results.
Discussion
Although independent, the PPI data mining method presented here is similar to that proposed by Li and Li [14]. Their research focused on motif pairs located on protein surfaces, and motif discovery was, in part, based on three-dimensional structures of proteins. Our method did not rely on PDB structural information, which is known to be biased towards small, globular proteins. Even without the additional structural information, many of the novel sequence signatures discovered in this study appear in the surfaces of proteins. Thus they are likely interacting sites.
Approximately 94% of the sequence signatures discovered in this study matched known sequence patterns, confirming the ability of this method to discover sequence signatures involved in various biological functions. It is our contention that the 84 novel sequence signatures reported in this study likely play biological roles such as interacting sites, and we are planning wet-lab experiments to investigate their functions.
The lengths of the novel sequence signatures are quite short, ranging from 10 to 36 residues. This is not surprising, as the yeast genome has been the subject of a remarkable number of studies and the majority of long sequence signatures are likely already known. Additionally, longer sequence signatures tend to contain gaps, and will thus be interpreted as multiple shorter signatures by MEME. Nevertheless, the discovery of short, novel sequence signatures, based on medium- and high-confidence PPIs, suggests that short sequence signatures do play biologically significant roles.
Only 545 (~18%) of the discovered sequence signatures matched known signatures in Pfam: a significantly smaller number than the 2936 signatures matched to one or more InterPro member databases. This result highlights a potential shortcoming of PPI predictions based on the analysis of DDIs inferred from Pfam data alone (e.g. ref [11]). The use of a single domain databases, such as pfam database with the average length of 145 amino acids [18] might cause a researcher to miss many important short sequence signatures, thereby decreasing prediction accuracy.
The use of PPI data to predict the sub-cellular localization of proteins is based on an intuitively simple idea: proteins that are found in the same location within a cell are more likely to interact with one another than proteins that are not. Ten subcellular compartments were actually used in our study. The resulting accuracy of PPI-based prediction of sub-cellular localization is reasonably good in this study and, at ~68%, represents a substantial increase in accuracy relative to what would be achieved (37%) if cytoplasm, the most populated compartment, was predicted for all systems. Our accuracy is comparable to that achieved in other recent studies. For example, using a hybrid system of gene ontology, functional domain and pseudo amino acid composition approaches, Chou and Cai obtained 70% of overall success identification rate [19,20]. Our accuracy rate was inferior to others that used fewer localization categories (for example, 88% accuracy rate based on cross validation was achieved when only four localization categories were used in ref [21]), but it is perfectly natural that a more ambitious categorization scheme such as ours should have a greater margin of error. Also we should emphasize that our approach represents a very intuitive and simple scheme based on PPI induced sequence signatures alone, in contrast to complicated hybrid systems employed in previous studies. Admittedly, our approach can only be used in predicting the localization of proteins involving in currently known PPIs, thus a broader application will be feasible only when more PPI data are available.
One of the major challenges to mining PPI data is the presence of numerous false positives, resulting from the deficiencies of current high-throughput screening techniques. The PPI data produced by some screening techniques such as yeast two-hybrid systems has been estimated to contain as much as 91% false interactions [22]. The 11,161 PPIs used as input to MEME were identified as medium or high confidence interactions, of which 20.7 % were high confidence. Of the PPIs actually used to build sequence signatures, 22.3% were high confidence interactions, a statistically significant increase of 7.7% over the original dataset. The disproportionate use of high-confidence PPIs to build sequence signatures supports the validity of the original reliability assignments, and suggests a method by which one may increase confidence in putative PPIs. Nevertheless, the quality of the results generated by all forms of PPI data mining remains constrained by the quantity and quality of the PPI data currently available. Consequently, the reliability of predictions based on PPI data is expected to increase as PPI databases increase in accuracy, size and taxonomic range.
Conclusion
In conclusion, we have reported a novel procedure by which sequence signatures were discovered based on a large PPI dataset from Saccharomyces cerevisiae. The majority of these sequence signatures were matched with known sequence signatures present in the InterPro member databases. Nevertheless, 84 distinct sequence signatures were novel, and may be involved in the interactions of the proteins containing them. The sub-cellular localizations of 108 proteins of the yeast genome were predicted, based on the known locations of other proteins and PPI dataset. Of the 108 localization predictions, 52 agreed with SGD annotations, and 24 disagreed. The localization of remaining 32 proteins was experimentally unknown. However, it is reasonable to believe these predictions would have similar prediction accuracy.
Wet-lab experiments to determine the biological function of the discovered novel sequence signatures are being planned. We are also in the process of developing an algorithm that will enable the discovery of gap-containing sequence signatures based on PPI data. The PPI data mining method presented here is imminently applicable to other genomes associated with large PPI datasets. For example, we conducted similar study on the E. Coli genome and were able to identify 22 novel signatures (the results of which can be found in the complementary website).
Methods
Dataset
PPI data specific to the genome of Saccharomyces Cerevisiae (baker's yeast) were used because the quantity of PPI data available for yeast exceeds that of any other model organism. The ~6000 proteins of the yeast proteome could potentially produce more than 18 million distinct, guest-host interactions, though the actual number of PPIs is certainly much smaller, probably less than 100,000 [23,24]. However, PPIs are dynamic, and the empirical discovery of these interactions is time and location dependent. The current list of putative PPIs between proteins of the yeast proteome, therefore, does not represent all PPIs that occur in the cells of yeast.
The dataset used here was reported by von Mering et al. [23]. It contained 78380 non-redundant PPIs from yeast, which were assigned to three categories of reliability: 2455 high confidence, 9400 medium confidence, and 66535 low confidence. PPIs of this dataset were discovered by various experimental and computational methods including yeast two-hybrid systems, mass spectrometry technologies.
In an attempt to minimize the occurrence of false positives, only those PPIs assigned a reliability of high or medium confidence were used (2617 host proteins involved in 11855 interactions). Because MEME requires input in the form of set of two or more related proteins, 694 host proteins that interacted with only one protein were also excluded. Of the remaining 1923 host proteins, only 25 were involved in more than 100 distinct PPIs, including the most interactive protein, YPR110C, which was involved in 118 putative PPIs.
MEME and MAST
MEME (v.3.0.10) was used to search for signatures shared by each group of guest proteins. MEME implements an unsupervised learning algorithm and ultimately produces one or more probabilistic signature models based on this input. The statistical significance of each signature model is quantified as an expectation value (E-value), which is an estimate of the number of signatures that would possess a higher log-likelihood ratio given randomly-generated training sequences. All signatures discovered by MEME are gapless, and the best width, number of occurrences, and description of each motif are based on statistical models.
For each of the 1923 host proteins associated with two or more guest proteins, a multiple sequence FASTA file was created from the amino acid sequences of its guest proteins. In every instance, MEME was executed with the following options: a minimum motif width of 10, maximum motif width of 300, maximum E-value of 0.1, and 5 as the maximum number of motifs.
MEME output files were then used as input for MAST (v3.0). MAST was used to search the entire yeast proteome for the sequence signatures described in the MEME output files. MAST output consists of the sequence name of each high-scoring match as well as the E-value of each match. For all MAST executions, the maximum E-value was set to 0.1. The results of MAST searches were used to assess the sequence coverage of sequence signatures identified by MEME and the usefulness of MEME output to PPI prediction.
Signature model comparison
InterPro [25] is an integrated collection of the most commonly used databases of protein families, domains, and functional sites. The program InterProScan allows a user to search for sequence signatures in any number of these databases simultaneously [26]. Only LAMA can be used to compare MEME results to the BLOCKS database [27], but no tools currently exist for comparison to other sequence signature databases. Therefore, the consensus sequence of each motif model identified by MEME was searched for in all InterPro member databases, using the standalone version of InterProScan (release 4.0) and a local copy of InterPro (release 8.1). Signatures that were unsuccessfully matched with any entries in the local InterPro database were input to the online version of InterProScan to identify matches to known signatures that were not integrated into the InterPro database (i.e., thus unavailable in the local database). Those signatures that remained unmatched were considered novel. Because different host proteins may share the same set of guest proteins, some of these novel signatures were identical or similar. Thus, FASTA [28] searches were performed, using each potentially novel signature as a query sequence, and the set of all potentially novel signatures as a local database. We tested several E-values (0.1, 0.5, 1) and found that 0.5 was the best for distinguishing signatures. Higher threshold E-values led to the identification of signature pairs as similar when only one or two contiguous residues were identical, while lower values excluded the detection of signatures that were clearly similar. To compare the coverage of the individual InterPro member databases, each consensus sequence signature was also assessed using the Pfam database only.
Querying sequence signature databases with the consensus sequence of a MEME model, rather than the model itself, is similar to the approach proposed by Kahsay et al. [29], which facilitated the comparison of two Hidden Markov Models. To verify the appropriateness of using consensus sequences in lieu of the actual models, we queried the consensus sequences of several signature models along with each of their component sequences against InterPro databases. We found the hits of the consensus sequences were consistent to those of their component sequences. For example, the consensus sequence of the signature YPL049_1 matched to all significant signatures that two component sequences had. The only difference was that two residues of the consensus sequence additionally matched to an un-integrated signature. This match was insignificant considering that the length of the signature was 65 residues.
Prediction of protein subcellular localization
Two proteins that interact with one another are likely found in the same subcellular location [23]. Thus PPI data can be used to predict the subcellular localizations of proteins. However, PPI data alone are currently not sufficient to predict subcellular localization due to the generally low reliability of current PPI data. In this study, we added an additional layer of confidence to predictions of subcellular localization by including our knowledge of sequence signatures shared by the guests of a host protein. For a guest protein with unknown localization, if its host protein and at least half of its fellow guest proteins shared a subcellular location, that guest protein was predicted to share this location as well. Similarly, if the localization of a host protein was unknown, and more than half of its guest proteins shared a common subcellular localization and one or more sequence signatures, the host protein was predicted to exist in the localization (Table 3).
Table 3 An example of protein location prediction. The host YGL115W has four guest proteins that share four statistically significant signatures. The host and all its guests with known location were found in cytoplasm. Thus the location of YGL208W was predicted as cytoplasm. The prediction was then confirmed with the ontology annotation in SGD database. The p-value of the occurrence is the probability that a single random subsequence of the length of the motif matches the motif.
Guest Motif ID P-value Guest location
YER027C YGL115W_1 3.17E-76 cytoplasm
YGL208W YGL115W_1 7.48E-75
YDR422C YGL115W_1 4.78E-48 cytoplasm
YER027C YGL115W_2 3.87E-56 cytoplasm
YGL208W YGL115W_2 8.48E-57
YDR422C YGL115W_2 3.64E-37 cytoplasm
YER027C YGL115W_3 6.83E-77 cytoplasm
YGL208W YGL115W_3 6.37E-71
YDR028C YGL115W_3 9.81E-38 cytoplasm
YER027C YGL115W_4 5.62E-22 cytoplasm
YGL208W YGL115W_4 7.23E-24
YDR477W YGL115W_4 1.89E-14 cytoplasm
Predictions of subcellular localization were based on the known localizations of 4156 budding yeast proteins [30], where there are 22 categories of subcellular location. Predictive accuracy was evaluated by comparing predicted locations to the known locations of these proteins as reported in the ontology annotation of the Saccharomyces Genome Database (SGD, ).
Homology modeling
NCBI's online BLAST engine was used to search PDB database for protein sequences similar to the selected yeast protein sequences. The best match was selected as a template structure and its PDB file was downloaded from the PDB database. All homology modeling was carried out with MOE (Molecular Operating Environment 2004.03, The Chemical Computing Group Inc., 2004). The query sequences and their templates were first aligned in MOE. Ten intermediate models were then created, each was finely energy-minimized for steric interactions using the AMBER-94 forcefield with the solvation option turned on. The best structure prediction was then selected according to energy ranking.
Authors' contributions
JWF designed the project. JWF and RJH carried out the study and drafted the manuscript. YHD and GHL participated in the study and manuscript preparation.
Supplementary Material
Additional File 1
Homology models of five yeast proteins. The following files are available in the complementary website : MEME output files for all novel signatures, PDB files of five homology models, a complete list of identified novel signatures and a list of these signatures grouped by similarity, a complete list of protein location prediction, and the distribution of the number of interaction partners.
Click here for file
Acknowledgements
This work was Supported by the K-INBRE Bioinformatics Core, NIH grant number P20 RR016475 and NIH Grant RR-P20 RR17708. We thank Dr. Xue-wen Chen from KU ITTC for useful discussions. We also thank the editor and two anonymous reviewers for their thoughtful comments and suggestions.
==== Refs
Uetz P Giot L Cagney G Mansfield TA Judson RS Knight JR Lockshon D Narayan V Srinivasan M Pochart P Qureshi-Emili A Li Y Godwin B Conover D Kalbfleisch T Vijayadamodar G Yang MJ Johnston M Fields S Rothberg JM A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae Nature 2000 403 623 627 10688190 10.1038/35001009
Ito T Chiba T Ozawa R Yoshida M Hattori M Sakaki Y A comprehensive two-hybrid analysis to explore the yeast protein interactome Proceedings of the National Academy of Sciences of the United States of America 2001 98 4569 4574 11283351 10.1073/pnas.061034498
Tong AHY Drees B Nardelli G Bader GD Brannetti B Castagnoli L Evangelista M Ferracuti S Nelson B Paoluzi S Quondam M Zucconi A Hogue CWV Fields S Boone C Cesareni G A combined experimental and computational strategy to define protein interaction networks for peptide recognition modules Science 2002 295 321 324 11743162 10.1126/science.1064987
Gavin AC Bosche M Krause R Grandi P Marzioch M Bauer A Schultz J Rick JM Michon AM Cruciat CM Remor M Hofert C Schelder M Brajenovic M Ruffner H Merino A Klein K Hudak M Dickson D Rudi T Gnau V Bauch A Bastuck S Huhse B Leutwein C Heurtier MA Copley RR Edelmann A Querfurth E Rybin V Drewes G Raida M Bouwmeester T Bork P Seraphin B Kuster B Neubauer G Superti-Furga G Functional organization of the yeast proteome by systematic analysis of protein complexes Nature 2002 415 141 147 11805826 10.1038/415141a
Ho Y Gruhler A Heilbut A Bader GD Moore L Adams SL Millar A Taylor P Bennett K Boutilier K Yang LY Wolting C Donaldson I Schandorff S Shewnarane J Vo M Taggart J Goudreault M Muskat B Alfarano C Dewar D Lin Z Michalickova K Willems AR Sassi H Nielsen PA Rasmussen KJ Andersen JR Johansen LE Hansen LH Jespersen H Podtelejnikov A Nielsen E Crawford J Poulsen V Sorensen BD Matthiesen J Hendrickson RC Gleeson F Pawson T Moran MF Durocher D Mann M Hogue CWV Figeys D Tyers M Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry Nature 2002 415 180 183 11805837 10.1038/415180a
Cho RJ Campbell MJ Winzeler EA Steinmetz L Conway A Wodicka L Wolfsberg TG Gabrielian AE Landsman D Lockhart DJ Davis RW A genome-wide transcriptional analysis of the mitotic cell cycle Molecular Cell 1998 2 65 73 9702192 10.1016/S1097-2765(00)80114-8
Hughes TR Marton MJ Jones AR Roberts CJ Stoughton R Armour CD Bennett HA Coffey E Dai HY He YDD Kidd MJ King AM Meyer MR Slade D Lum PY Stepaniants SB Shoemaker DD Gachotte D Chakraburtty K Simon J Bard M Friend SH Functional discovery via a compendium of expression profiles Cell 2000 102 109 126 10929718 10.1016/S0092-8674(00)00015-5
Salwinski L Miller CS Smith AJ Pettit FK Bowie JU Eisenberg D The Database of Interacting Proteins: 2004 update Nucleic Acids Research 2004 32 D449 D451 14681454 10.1093/nar/gkh086
Pagel P Kovac S Oesterheld M Brauner B Dunger-Kaltenbach I Frishman G Montrone C Mark P Stumpflen V Mewes HW Ruepp A Frishman D The MIPS mammalian protein-protein interaction database Bioinformatics 2005 21 832 834 15531608 10.1093/bioinformatics/bti115
Sprinzak E Margalit H Correlated sequence-signatures as markers of protein-protein interaction Journal of Molecular Biology 2001 311 681 692 11518523 10.1006/jmbi.2001.4920
Deng MH Mehta S Sun FZ Chen T Inferring domain-domain interactions from protein-protein interactions Genome Research 2002 12 1540 1548 12368246 10.1101/gr.153002
Wojcik J Schachter V Protein-Protein interaction map inference using interacting domain profile pairs Bioinformatics 2001 17 S296 S305 11473021
Han DS Kim HS Jang WH Lee SD Suh JK PreSPI: a domain combination based prediction system for protein-protein interaction Nucleic Acids Research 2004 32 6312 6320 15576357 10.1093/nar/gkh972
Li HQ Li JY Discovery of stable and significant binding motif pairs from PDB complexes and protein interaction datasets Bioinformatics 2005 21 314 324 15374856 10.1093/bioinformatics/bti019
Bailey TL Elkan C Unsupervised Learning of Multiple Motifs in Biopolymers Using Expectation Maximization Machine Learning 1995 21 51 80 10.1007/BF00993379
Bailey TL Gribskov M Methods and statistics for combining motif match scores Journal of Computational Biology 1998 5 211 221 9672829
Kabsch W Sander C Dictionary of Protein Secondary Structure - Pattern-Recognition of Hydrogen-Bonded and Geometrical Features Biopolymers 1983 22 2577 2637 6667333 10.1002/bip.360221211
Bornberg-Bauer E Beaussart F Kummerfeld S Teichmann S Weiner J The evolution of domain arrangements in proteins and interaction networks Cellular and Molecular Life Sciences 2005 62 435 445 15719170 10.1007/s00018-004-4416-1
Cai YD Chou KC Predicting 22 protein localizations in budding yeast Biochemical and Biophysical Research Communications 2004 323 425 428 15369769 10.1016/j.bbrc.2004.08.113
Chou KC Cai YD Predicting protein localization in budding yeast Bioinformatics 2005 21 944 950 15513989 10.1093/bioinformatics/bti104
Bhasin M Raghava GPS ESLpred: SVM-based method for subcellular localization of eukaryotic proteins using dipeptide composition and PSI-BLAST Nucleic Acids Research 2004 32 W414 W419 15215421
Mrowka R Patzak A Herzel H Is there a bias in proteome research? Genome Research 2001 11 1971 1973 11731485 10.1101/gr.206701
von Mering C Krause R Snel B Cornell M Oliver SG Fields S Bork P Comparative assessment of large-scale data sets of protein-protein interactions Nature 2002 417 399 403 12000970 10.1038/nature750
Bader GD Hogue CWV Analyzing yeast protein-protein interaction data obtained from different sources Nature Biotechnology 2002 20 991 997 12355115 10.1038/nbt1002-991
Mulder NJ Apweiler R Attwood TK Bairoch A Bateman A Binns D Bradley P Bork P Bucher P Cerutti L Copley R Courcelle E Das U Durbin R Fleischmann W Gough J Haft D Harte N Hulo N Kahn D Kanapin A Krestyaninova M Lonsdale D Lopez R Letunic I Madera M Maslen J McDowall J Mitchell A Nikolskaya AN Orchard S Pagni M Pointing CP Quevillon E Selengut J Sigrist CJA Silventoinen V Studholme DJ Vaughan R Wu CH InterPro, progress and status in 2005 Nucleic Acids Research 2005 33 D201 D205 15608177 10.1093/nar/gki106
Zdobnov EM Apweiler R InterProScan - an integration platform for the signature-recognition methods in InterPro Bioinformatics 2001 17 847 848 11590104 10.1093/bioinformatics/17.9.847
Frenkel-Morgenstern M Voet H Pietrokovski S Enhanced statistics for local alignment of multiple alignments improves prediction of protein function and structure Bioinformatics 2005 21 2950 2956 15870168 10.1093/bioinformatics/bti462
Pearson WR Flexible sequence similarity searching with the FASTA3 program package Methods in Molecular Biology 2000 132 185 219 10547837
Kahsay RY Wang GL Gao G Liao L Dunbrack R Quasi-consensus-based comparison of profile hidden Markov models for protein sequences Bioinformatics 2005 21 2287 2293 15797916 10.1093/bioinformatics/bti374
Huh WK Falvo JV Gerke LC Carroll AS Howson RW Weissman JS O'Shea EK Global analysis of protein localization in budding yeast Nature 2003 425 686 691 14562095 10.1038/nature02026
|
16305745
|
PMC1310605
|
CC BY
|
2021-01-04 16:27:48
|
no
|
BMC Bioinformatics. 2005 Nov 23; 6:277
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-277
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2691628008410.1186/1471-2105-6-269SoftwareErmineJ: Tool for functional analysis of gene expression data sets Lee Homin K [email protected] William [email protected] Kiran [email protected] Paul [email protected] Columbia Genome Center, Columbia University, New York NY 10032, USA2 Department of Computer Science, Columbia University, New York NY 10025, USA3 Department of Philosophy, University of Arizona, Tucson, AZ 85721, USA2005 9 11 2005 6 269 269 18 8 2005 9 11 2005 Copyright © 2005 Lee et al; licensee BioMed Central Ltd.2005Lee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
It is common for the results of a microarray study to be analyzed in the context of biologically-motivated groups of genes such as pathways or Gene Ontology categories. The most common method for such analysis uses the hypergeometric distribution (or a related technique) to look for "over-representation" of groups among genes selected as being differentially expressed or otherwise of interest based on a gene-by-gene analysis. However, this method suffers from some limitations, and biologist-friendly tools that implement alternatives have not been reported.
Results
We introduce ErmineJ, a multiplatform user-friendly stand-alone software tool for the analysis of functionally-relevant sets of genes in the context of microarray gene expression data. ErmineJ implements multiple algorithms for gene set analysis, including over-representation and resampling-based methods that focus on gene scores or correlation of gene expression profiles. In addition to a graphical user interface, ErmineJ has a command line interface and an application programming interface that can be used to automate analyses. The graphical user interface includes tools for creating and modifying gene sets, visualizing the Gene Ontology as a table or tree, and visualizing gene expression data. ErmineJ comes with a complete user manual, and is open-source software licensed under the Gnu Public License.
Conclusion
The availability of multiple analysis algorithms, together with a rich feature set and simple graphical interface, should make ErmineJ a useful addition to the biologist's informatics toolbox. ErmineJ is available from .
==== Body
Background
A difficulty experienced by many (if not all) users of gene expression microarrays is making sense of the complex results. After analyzing each gene in a data set, an experimenter is often left to the task of summarizing the results with little assistance. It is common for experimenters to ask questions at the level of molecular pathways or other functionally relevant groupings of genes. While "ad hoc" manual annotation of data sets is a common approach, there are numerous advantages to using a computational and statistical approach to analyze groups of genes.
The most common means of performing this analysis is to ask whether certain Gene Ontology (GO) [1] terms are "over-represented" in a set of genes selected by fold-change or statistically-motivated approaches such as a t-test. This is easily implemented by using the properties of the hypergeometric distribution (often referred to as Fisher's exact test for two categories) or its binomial approximation. In our work, these methods are more generically referred to as "over-representation analysis" or ORA. In addition, as the GO is just one way of organizing genes, we refer to the general goal of these methods as "gene set analysis", where a gene set is any grouping of genes not derived from the data itself, typically based on biologically-motivated criteria.
The need to perform ORA has led to the emergence of a variety of tools. A list of many such tools is available from the Gene Ontology Consortium [2], and a large number of them were recently reviewed [3]. However, to our knowledge these tools all implement ORA methods; other methods or algorithms are not available, with the exception of the Perl script Catmap [4]. Thus these tools primarily differentiate themselves through user interface features, ease of use, supported data types, and speed [3]. Most tools surveyed by [3] were reported to have one or more significant limitations, including slow performance, an inability to analyze gene annotations other than those directly annotated (that is, other levels of the GO hierarchy are not considered), requiring web access to use, are difficult to install (limiting their usefulness to biologists), or lack the ability to visualize the GO hierarchy [3].
In this paper we describe ermineJ, a stand-alone tool that implements methods described by [5] and [4] in addition to ORA, has a rich feature set, and does not have the limitations cited above. One of the offered analysis methods in particular is complementary to ORA analysis, which we now call Gene Set Resampling or GSR (the "experiment" score in Pavlidis et al. (2002)). In GSR, the gene-by-gene scores (e.g., t-test p-values) are not thresholded. Instead, for each gene set an aggregate score is computed, such as the geometric mean of the p-values for genes in the category, and the significance of that score determined by random sampling of the data. We have recently presented some evidence that GSR can provide better results than ORA in some situations [6].
ErmineJ also has methods for analysis of genes based on rankings (the receiver operator characteristic, or ROC) [4]. ROC can be thought of as a version of ORA where all possible thresholds are considered simultaneously. Like GSR, the ROC method utilizes non-thresholded gene scores, but considers only their ranking, which might be considered more robust than using the raw gene scores. Finally, ErmineJ offers an analysis based on the correlation of gene expression profiles, gene group correlation analysis (GCA) [5]. GCA can be used as an alternative to the use of ORA for the determination of whether genes in particular functional categories are "clustering together".
The first version of ermineJ was made available in 2003. Recently we have completely revamped the user interface and updated the feature set, releasing ermineJ 2.0 in October 2004 and 2.1 in June 2005.
Implementation
ErmineJ is implemented entirely in the Java programming language [7]. It uses the Java Swing libraries to create a graphical user interface that can run on many different platforms. Architecturally, an effort has been made to separate analytical and algorithmic concerns from user presentation concerns. Besides being a design best practice, the architecture was also driven by the need to support command-line interfaces as well as application programming interfaces to the methods. The structure of ermineJ also lends itself to fairly easy extensibility, so new algorithms can be added to the software as requirements change. The analysis algorithms in ermineJ were previously described [4,5].
In addition to using the Java SDK, ermineJ depends on a number of free third-party libraries, most notably the Colt library [8]. Colt is a high-performance numerical computing library that includes implementations of many linear algebra and statistical methods, as well as useful data structures which we rely on heavily in our software. Other libraries ermineJ uses include various Jakarta Commons libraries [9], and the Xerces XML parsing engine [10], which we use to parse the Gene Ontology XML description. Many of the low-level numerical and utility routines (e.g., for file parsing and string manipulation) are tested in an extensive unit test suite.
Results and discussion
Inputs
All interfaces to ermineJ use the same basic inputs. The first is a description of the Gene Ontology in XML format, obtained from the GO consortium web site [11]. The second is a description of the microarray platform (the "array annotation file", which contains tab-delimited text), which associates probe identifiers with Gene Ontology terms and additionally associates each probe with a gene (used in the statistical analysis to account for repeated genes, as described below) and descriptions that are useful for viewing in the context of the results. The third required input is the user's own data. For ORA, GSR and ROC applications, this takes the form of a list of gene scores, one for every probe set on the array design. Alternatively (for expression profile correlation analysis), the input can be the expression profile matrix, as might be used as an input to a clustering tool. The gene scores can be p-values or another score such as fold-change. ErmineJ is purposefully largely agnostic about the meaning of the gene scores, and focused on the distributional properties of the scores.
We maintain on the order of 30 different mouse, human and rat array annotation files for different platforms, as well as generic files for RefSeq [12] genes that can be used to construct annotation files for other platforms (available from our web site [13]). The native annotation file format is very simple and new files can easily be constructed with a modicum of bioinformatics skill. ErmineJ can also read Affymetrix "CSV" (comma-separated-value) annotation files available from the manufacturer's web site. We gladly entertain requests to add support for other arrays. When an annotation file is read in, the software automatically associates each probe with all parent terms of each directly annotated terms. For example, all genes annotated with the term "regulation of cell size" are also associated with the higher-level terms "cellular morphogenesis" and "morphogenesis". This feature is only supported by some of the tools reviewed by [3].
There are a number of parameters to set and decisions the user must make in order to run the software. The choice of analysis method is the most obvious, and each method has a few other settings that the user can choose to change. For example, for ORA analysis a threshold score must be defined. This is in contrast to most ORA software packages which take as input a list of "genes of interest"; instead, ermineJ takes as input all the gene scores for the experiment. This lets ermineJ avoid the problem of selecting the correct "null" gene set [3]: it is defined strictly by the genes analyzed in the experiment but not meeting the user-defined score threshold.
For GSR, the method used to compute the score for a gene set is a key parameter. The two options currently supported are the mean and the median. During the analysis, GSR uses the selected method to compute a summary of the gene scores for each resampled or real gene set, and this aggregate score is used to represent the gene set. Choosing the median will tend to yield slightly more conservative results, as individual genes with very high scores are not given as much weight as in the mean computation.
Some settings are used for multiple methods. For example, when a gene is represented more than once in the data set, a decision has to be made as to how to treat these "replicates" (which might not be replicates per se but represent different transcripts). The options supported are to use the "best" score among the replicates to represent them as a group; to use the mean; or to treat them as separate entities. Use of the "best" option is somewhat anti-conservative, but is reasonable when most "replicates" are in fact assaying different biological entities. In contrast, treating replicates completely separately is not generally advised as it can lead to spurious positive findings in cases of true replicates, as the gene set gets "adulterated" with multiple copies of the same high-scoring gene. For this reason the last option is not available from the GUI, though it can be accessed from the other interfaces. Another important setting is the range of gene set sizes to analyze. Gene sets that are very small are unlikely to be very informative, because the goal of the analysis is to study genes in groups, while large gene sets may be too non-specific to provide useful information. In addition, analyzing too many gene sets reduces the power of the analysis due to multiple testing costs. In practice we often use a range of 5–100 or 5–200.
In addition to the pre-defined gene sets as defined by the Gene Ontology, users are free to input their own gene sets. These are defined in simple text files that are placed in a directory that ermineJ checks at startup. These text files can be created "off-line" or within the ermineJ GUI. In addition, users can modify gene sets from within ermineJ. This functionality can be used to correct errors or omissions in the Gene Ontology annotations, though care must be exercised to avoid introducing biases into the results.
Types of analysis
Gene-score based methods
The ORA, GSR and ROC methods are closely related in that they are based on the gene-by-gene scores, with the goal of finding gene sets that are some sense "enriched" in high-scoring genes (which typically might be "differentially expressed genes"). ORA is sometimes used to analyze genes which are selected by clustering, rather than a continuous score. In this situation, GSR and ROC are not appropriate. However, the correlation method is specifically designed to address this situation. GSR and ROC have the benefit of not requiring a threshold to divide genes into "selected" and "non-selected" genes. The choice of the threshold for ORA can have a substantial effect on the results obtained, because the "selected genes" change [4].
Correlation analysis
Gene group correlation analysis (GCA) is based on the similarity of the expression profiles of genes in a gene set: loosely speaking, how well they "cluster together". Thus we propose that GCA can be used as an alternative to using ORA to analyze clusters. There are some differences to be noted between the typical application of ORA to clusters and the ermineJ correlation analysis. GCA is group-centric, not cluster-centric. Thus we ask whether the correlation among the members is higher than expected by chance, not whether a given set of correlated genes is enriched for the genes in the group; GCA does not involve clustering. This is not a trivial distinction, because while the highest scores will be obtained for gene groups that have uniform and high correlations among all the members, groups that have two or more "sub-clusters" can also obtain high scores. In the current implementation of GCA, the absolute value of the correlation is always used, which allows. In future versions we may expose this as a user-settable option, as well as implementing other possible correlation metrics other than the current Pearson correlation.
In all methods, for each gene set analyzed, ermineJ computes a score and, based on that score and the gene sets size, a p-value representing the "significance" of that gene set with respect to the null hypothesis. The definition of the raw score and the null hypothesis depends on the method being used. Note that the raw scores are of limited use because it cannot be evaluated in the absence of information about the gene set size. However, they can provide the user with a helpful indication the strength of the result, not just its statistical significance.
For ORA, the null hypothesis is that the genes in the gene set are distributed randomly between the "selected" genes and the "non-selected" genes. The raw score reported by ErmineJ is the number of genes in the set which pass the threshold for gene selection. For GSR, the null hypothesis is that the mean (or median) gene score (which forms the gene set score; for p-values negative-log-transformed values are used) is drawn from the global (data-wide) distribution of possible gene set mean (or median) gene scores, as determined by resampling [5]. For ROC analysis, the null hypothesis is that the genes in the gene set are distributed randomly in the ranking; p-values are computed using the fact that the ROC is equivalent to the Wilcoxon rank-sum test [4]. The raw gene set score is simply the area under the receiver operator characteristic curve [14], which ranges from values of 0.5 (random ranking) to 1.0 (all genes in the gene set at the top of the ranking). Finally, for correlation analysis, the null hypothesis is that the mean pairwise correlation of profiles in the gene set is drawn from the global distribution of gene set correlation scores, as determined by resampling [5]. The raw score is the mean absolute value of the pair-wise correlation of the genes in the set (comparisons of a probe to itself, or to other probes for the same gene, are always ignored).
ErmineJ includes implementations of three multiple test correction methods (though currently only one of these, Benjamini-Hochberg false discovery rate (FDR) [15], is made available through the GUI). The additional options, available from the command line, are Bonferroni correction and a resampling-based family-wise error rate correction [16]. The FDR is used in the GUI as a rapid and reasonable guide to which gene sets are likely to be of highest interest.
The ermineJ GUI
Most users of ermineJ will access it through its graphical interface. The GUI of ermineJ was designed to be simple to use and provides "wizards" to guide users through common tasks such as running an analysis. Many settings made by the user during operation of the software are remembered between sessions, facilitating repeated analysis of the same data files and maintaining the user's preferred window sizes, for example. A complete manual is provided and is accessible via an on-line help function, as web pages on our web site, or in portable document format (PDF).
Some aspects of the ermineJ graphical user interface is illustrated in Figures 1, 2, 3. The main panel of the software can be viewed either as a table of gene sets (Figure 1A) or in a hierarchical (tree) view (Figure 1B). These views are linked so changes in one are reflected in the other. To facilitate navigation of these displayed, gene sets can be searched by the name of the gene set or by the names of genes they contain. User-defined gene sets are displayed in contrasting colors. Not shown in the figures is the initial startup screen in which the user chooses the gene annotation file to use for the session.
Figure 1 A: The main panel of ErmineJ after several analyses have been performed. Gene sets selected at low FDR levels are indicated in color. B: The tree-view panel of ErmineJ, illustrating the ability to browse gene sets in the GO hierarchy. The icons at each node have specific meanings. For example, the yellow "bull's-eye" icon indicates a gene sets selected at an FDR of 0.05 or less. Purple diamonds indicate nodes that have "significant" sub-nodes.
Figure 2 A gene set details view. The controls at the top allow adjustment of the size and contrast of the heat map. The gene scores (in this case p-values) are shown in the second text column. The grey and blue graph, shown only for experiments using p-values, shows the expected (grey) and actual (blue) distribution of p-values in the gene set. This display is provided as an additional aid to evaluation of the results. The last two columns provide information about each gene. The targets of the hyperlinks are configurable by the user.
Figure 3 Examples of screens from ErmineJ Wizards. A: Analysis wizard. This illustrates options to set the range of gene set sizes to analyze, and the method of treating "replicates" of genes. See text for details of the latter. B: Gene set modification wizard. In this screen the user is selecting genes to delete from a gene set. The list of all probe available on the platform is available in the left panel. A "find" function simplifies the location of genes and probes.
Double-clicking on a gene set in the main panel opens a new window that displays the genes in the gene set, along with the expression profiles in a "heat-map" view (if the user has provided the profile data; Figure 2). The appearance of the heat map is configurable through menus and toolbar controls. The data displayed in the table, as well as the image of the matrix, can be saved to disk using additional menu options. The hyperlinks to external web sites can be configured by the user to point to a web site of their choosing, again through a menu option. All of these capabilities are available even if the user has not performed any analysis, so ErmineJ can be used as a "gene set browser" as well as for analysis.
An important feature of the GUI is the capability to rapidly define and edit gene sets, which is accomplished in a "wizard" that takes the user through the process set-by-step. Alternatively, the user can simply populate the gene set directory with files they have obtained from other sources, for example created in bulk with a Python script or obtained from another user. As far as we know, no tool surveyed by [3] affords the user the ability to define or modify the categories. ErmineJ also allows the user to choose which of the GO aspects (Biological Process, etc.) to use in the analysis.
The GUI version of ermineJ can be installed on the user's computer or run via Java WebStart. The latter option simply involves clicking on a link in the user's web browser, and ensures that the users have the most up-to-date version of the software. The drawback of using WebStart is that the user must be connected to the internet to use the software. With a local installation, no internet connection is needed.
Running an analysis
Running an analysis using the ErmineJ GUI involves using a "wizard" to set the parameters (Figure 3). The user is asked to choose an analysis method, select the data file to analyze, choose any user-defined gene sets to include in the analysis, and set the various parameters required for the particular analysis. All settings are documented via "tool tips" and in the manual.
Once an analysis is initiated, the user is informed of its progress via a status bar. An analysis can be cancelled any time. On completion, the results are added to the tabular and tree views (Figure 1). Multiple results can be displayed simultaneously in the tabular view, allowing easy comparison of different runs. The tree view can display only a single analysis result set at a time, but offers a pull-down menu to selected among the results sets to display. In the tree and tabular views, high-scoring (i.e., significant) gene sets are highlighted in color. The tree view uses a simple system of icons for each node to indicate whether a significant node is contained within a given higher level node. Finally, the results of an analysis can be saved to a tab-delimited file for use in other software or to be reloaded by ermineJ at a later time.
Other interfaces
In addition to the GUI, ermineJ offers a command line interface (CLI) and a simple application programming interface (API). The CLI exposes some features of ermineJ that are not available in the GUI, such as different methods for multiple test correction. The CLI is suitable for scripting runs of ermineJ. For example, a simple Perl script can be used to automate runs of ermineJ with different settings or on different data sets. In contrast, the API was introduced to allow programmers to include the analyses available in ermineJ in their own software. The API currently provides more limited access to the functionality of the software than the command line version, but will be expanded in future versions.
Performance
We tested the performance of ermineJ using the HG-U133_Plus_2 Affymetrix array design. This is a particularly large array design with over 54,000 probe sets, and represents a something of a worst-case scenario with respect to performance. With our current annotation set, 4844 different GO categories (gene sets) are available for analysis in this array design. We limited our analysis to gene sets with between 5 and 100 genes, leaving about 2700 gene sets. The times reported below are for analyzing the complete set of over 54,000 probe sets with respect to these 2700 gene sets on a on a 1.7 GHz Pentium laptop.
With this array, ermineJ has an initial startup phase that lasts 15–20 seconds, most of which is consumed by time it takes for the gene annotation file to be read in and processed for analysis. The time for analysis once startup is completed depends on the method used. For ORA, a complete analysis is completed in 8 seconds (average of 3 runs; times are wall clock seconds timed from within the software). While it is difficult to directly compare our benchmarks with previously published benchmarks because the number of gene sets analyzed and the size of the "null" gene set was not reported, and the times reported might in some cases include initial startup times [3], the fastest reported methods on the largest data sets tested completed ORA analyses in under 10 seconds. This indicates that ErmineJ is at least competitive with and possibly faster than the fastest previously reported tools.
GSR analysis took about 370 seconds if a full resampling is performed (100,000 resampling trials per gene set size in our tests). However, ermineJ implements an approximation, where limited resampling is used to estimate the parameters of a normal distribution. This normal is used to compute the p-values for each gene set. It also takes advantage that, especially for larger class sizes, the shape of the resampled distribution is very similar for similar class sizes, so not all of them need to be computed. In this mode the analysis takes approximately 80 seconds. ROC analysis, which does not involve resampling, took about 100 seconds. Correlation analysis is the most computationally intensive resampling method; even with the approximations enabled it currently takes about 400 seconds to run on the test data set (which contained 12 microarrays). This is because computing correlations is computationally intensive, compared to the methods which use pre-computed gene scores such as p-values.
ErmineJ is fairly memory-intensive, because it holds in memory a complex data structure describing the annotations, as well as the microarray data and information about the results for thousands of gene sets and tens of thousands of genes. For the large HG-U133_Plus_2 design, after startup ermineJ occupies approximately 85 Mb of RAM (determined using a Java heap profiler under Windows). After running the correlation analysis, this grew to 105 Mb, reflecting the loading of the complete expression profile set and the results. Therefore we recommend running ermineJ on machines that have at least 256 Mb of RAM.
Future plans
At this writing, the current version of ermineJ is 2.1.6. New features planned for the software include expanding the API and allowing more flexible creation of user-defined gene sets, including allowing support of alternative nomenclatures such as the Plant Ontology [17]. We also plan to provide annotation files for more platforms and organisms.
We have been interested in the possibility of including other resampling-based methods such as GSEA [18] or the similar resampling method implemented in Catmap [4] in ermineJ. The primary reason to consider these methods is that they examine the distribution of gene scores by resampling over the samples, which is more correct than merely resampling over the genes. This is because the null hypotheses in the gene score analysis are some variation on a random distribution of genes within the ranking of genes. This assumption can be badly violated for gene sets containing highly correlated genes (such as the ribosomal protein genes); such genes will tend to have correlated rankings, and in some situations (particularly when the gene p-value distribution is close to uniform), spurious false positives can occur [4]. The ORA, GSR and ROC methods are all susceptible to this problem, though we stress that this is only an serious issue for gene sets that show high correlations not related to the experimental design.
It would be challenging to provide a general-purpose implementation of GSEA or Catmap that is easily accessible to biologists with limited computational skills. These methods require either that users can provide the gene scores for hundreds (if not thousands) of resampled data sets [4], a task that is difficult to accomplish for the targeted user base of ermineJ, or computation of gene scores by the software. Because each experimental design might have a different mechanism for computing gene scores (fold-change, t-test, ANVOA, Cox regression, etc), it would be difficult to provide a fully flexible tool without including a full-fledged statistical analysis package as well. A feasible solution we are considering is to cover the most frequently-encountered situations (e.g., t-test and one-way ANOVA).
Conclusion
ErmineJ is a fast, full-featured, user-friendly, multi-platform open source application for analysis of gene sets. It implements multiple algorithms for performing the analysis, and permits easy modification and creation of new gene sets. These features afford users considerable flexibility in testing different methods and parameters. Perhaps the greatest current limitation to its usability at this date is the availability of gene annotation files for non-Affymetrix array designs we have not encountered frequently. Users who wish to develop annotation files for their platform should contact us for assistance.
Availability and requirements
• Project name: ErmineJ
• Project home page:
• Operating system(s): Platform independent
• Programming language: Java
• Other requirements: Java 1.4 or higher; 256 Mb RAM recommended.
• License: GNU GPL and LPGL for helper library.
• Any restrictions to use by non-academics: None
List of Abbreviations
ORA: Over-representation analysis
GSR: Gene score resampling
ROC: Receiver operator characteristic
GCA: Gene group correlation analysis
GSEA: Gene Set Enrichment Analysis
FDR: False discovery rate
GO: Gene Ontology
GUI: Graphical User Interface
API: Application Programming Interface
CLI: Command Line Interface
Authors' contributions
PP was the project lead and chief architect of the software, and contributed to the source code. HKL, WB and KK all contributed to the source code.
Acknowledgements
We thank Shahmil Merchant and Edward Chen for contributions to an early version of ErmineJ, and William Noble for supporting the development of the methods, and Neil Segal for providing the microarray data used in the screen shots. We also thank testers and users who provided bug reports and suggestions for improvements.
==== Refs
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Gene Ontology Tools
Khatri P Draghici S Ontological analysis of gene expression data: current tools, limitations, and open problems Bioinformatics 2005 21 3587 3595 15994189 10.1093/bioinformatics/bti565
Breslin T Eden P Krogh M Comparing functional annotation analyses with Catmap BMC Bioinformatics 2004 5 193 15588298 10.1186/1471-2105-5-193
Pavlidis P Lewis DP Noble WS Exploring gene expression data with class scores Pac Symp Biocomput 2002 474 485 11928500
Pavlidis P Qin J Arango V Mann JJ Sibille E Using the gene ontology for microarray data mining: a comparison of methods and application to age effects in human prefrontal cortex Neurochem Res 2004 29 1213 1222 15176478 10.1023/B:NERE.0000023608.29741.45
Java
Colt
Jakarta Commons
Xerces
Gene Ontology Consoritum
Maglott D Ostell J Pruitt KD Tatusova T Entrez Gene: gene-centered information at NCBI Nucleic Acids Res 2005 33 D54 8 15608257 10.1093/nar/gki031
Microarray annotation files
Duda RO Hart PE Stork DG Pattern classification 2001 2nd New York , Wiley xx, 654
Benjamini Y Hochberg Y Controlling the False Discovery Rate: a Practical and Powerful Approach to Multiple Testing Journal of the Royal Statistical Society B 1995 57 289 300
Westfall PH Young SS Resampling-based multiple testing 1993 New York , John Wiley & Sons, Inc. 340
The Plant Ontology
Subramanian A Tamayo P Mootha VK Mukherjee S Ebert BL Gillette MA Paulovich A Pomeroy SL Golub TR Lander ES Mesirov JP Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles Proc Natl Acad Sci U S A 2005 102 15545 15550 16199517 10.1073/pnas.0506580102
|
16280084
|
PMC1310606
|
CC BY
|
2021-01-04 16:27:48
|
no
|
BMC Bioinformatics. 2005 Nov 9; 6:269
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-269
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-631628008110.1186/1471-2148-5-63Methodology ArticleExploration of phylogenetic data using a global sequence analysis method Chapus Charles [email protected] Christine [email protected] Scott [email protected] Alain [email protected] Bernard [email protected] Patrick [email protected] Equipe de Bioinformatique Génomique et Moléculaire, INSERM U 726, Case 7113, Tour 53-54, 2 place Jussieu, 75005 Paris, France2 Inserm U494, 91 bd de l'Hopital 75634 Paris CEDEX 13, France3 Dept. of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138 USA4 Current address: Dept. of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138 USA2005 9 11 2005 5 63 63 15 4 2005 9 11 2005 Copyright © 2005 Chapus et al; licensee BioMed Central Ltd.2005Chapus et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Molecular phylogenetic methods are based on alignments of nucleic or peptidic sequences. The tremendous increase in molecular data permits phylogenetic analyses of very long sequences and of many species, but also requires methods to help manage large datasets.
Results
Here we explore the phylogenetic signal present in molecular data by genomic signatures, defined as the set of frequencies of short oligonucleotides present in DNA sequences. Although violating many of the standard assumptions of traditional phylogenetic analyses – in particular explicit statements of homology inherent in character matrices – the use of the signature does permit the analysis of very long sequences, even those that are unalignable, and is therefore most useful in cases where alignment is questionable. We compare the results obtained by traditional phylogenetic methods to those inferred by the signature method for two genes: RAG1, which is easily alignable, and 18S RNA, where alignments are often ambiguous for some regions. We also apply this method to a multigene data set of 33 genes for 9 bacteria and one archea species as well as to the whole genome of a set of 16 γ-proteobacteria. In addition to delivering phylogenetic results comparable to traditional methods, the comparison of signatures for the sequences involved in the bacterial example identified putative candidates for horizontal gene transfers.
Conclusion
The signature method is therefore a fast tool for exploring phylogenetic data, providing not only a pretreatment for discovering new sequence relationships, but also for identifying cases of sequence evolution that could confound traditional phylogenetic analysis.
==== Body
Background
Phylogenetic classifications traditionally rely on phenotypic traits and the paleontological record [1]. As a result of the large amount of DNA sequences now available in the databases, molecular phylogeny has become an essential companion in studying evolutionary relationships among species [2]. As usually practiced, it allows constructing phylogenetic trees based on differences between homologous sequences or genes [3]. A basic and indispensable step in phylogenetic study is alignment of the set of homologous sequences [4]. However, distantly related sequences can be difficult to align and under these conditions, different algorithms often lead to different phylogenetic results [5,6]. There are other problems linked to the use of biological sequences in phylogenetic analysis, including sampling of representative sequences, biological processes such as lateral gene transfer, fusion events and recombination (see Brocchieri et al [5] for a review).
New approaches of molecular phylogeny, taking into account new characteristics of sequences, have been recently developed. Such methods include using other aspects of molecular data such as structural properties of proteins [7], the presence and organization of genes along genomes [8-11], occurrence of characteristic patterns [12,13] and the frequencies of short nucleotide or peptide relative abundance [14-18]. These methods contribute to the understanding of species evolution from different points of view, particularly in terms of our understanding of genome evolution. What is intriguing about these methods is that they often yield phylogenetic results comparable to those of traditional methods, frequently employing data sets much larger than traditional phylogenetic analyses. As such, they deserve the attention of those wishing to extract maximal information from comparative genomic data sets.
We expand on a method to characterize DNA sequences: the sequence signature. Sequence signature is defined as the whole set of frequencies of short oligonucleotides (words, until ten nucleotides long currently) of a sequence [19]. The principal characteristics of sequence signature used for phylogenetic studies are species-specificity of sequence signature and conservation of signature in any part of the genome [20] allowing researchers to compare sequences from diverse regions of the genome. It has already been established that distances between species signatures of the same taxonomic group are smaller than between signatures of species belonging to different groups [19,21]. A difference of signatures between two sequences could arise from shifts in the pattern of point substitution, but could also involve interactions among adjacent nucleotides, natural selection, DNA repair processes and conformational constraints (super coiling, nucleosome formation, bend DNA) [22]. A phylogenetic analysis of signatures could therefore reflect underlying genomic changes that shift motif frequencies, thereby yielding higher-order homologies available for phylogenetic analysis. The method has already been used for taxonomic classification of some species groups [23-25]. One advantage of such a method consists mainly in avoiding the alignment step, and can be used on numerous sequences of varying size. In addition, distance matrices, such as those applicable to genomic signatures, generally permit fast building of trees. Perhaps most importantly, genomic signatures provide a means of comparing large-scale patterns in genomes and can help evaluate trends in genome evolution across a phenetic tree. However, no systematic analysis of the reliability of the signature approach has been performed on homologous sequences. It has been demonstrated that long word frequencies describes DNA sequence information more accurately [19,25], but with their much larger number, long words are difficult to apply to short sequences because word frequencies are poorly estimated. Wang et al. [25] have also qualitatively analyzed the impact of the choice of the divergence metrics on phylogenetic results. However, no quantitative analyses or simulations have been presented yet on this subject.
In this paper, statistical studies of the ability of a signature approach for reconstructing phylogenies are investigated, specifically in order to determine the optimum word length and the influence of the divergence metric on the results. One of the tests we employ allows us to determine whether the signature distance can be considered tree-like, possessing hierarchical information [26]. Working with homologous, fully alignable sequences, we tested the method on simulated sequences whose true topologies are known and also analyzed two published examples of DNA sequences that propose novel interspecific relationships. Overall we find that there is a strong correspondence between signature trees and those generated by conventional means. As a means of improving large multi-gene studies [27,28], we also propose the use of signatures for rapid, large-scale sequence analysis specifically to detect subsets of genes supporting similar species phylogenies and to identify cases of horizontal transfer. In an analysis of 16 complete γ-proteobacteria genomes, we also illustrate how the signature method can also be used on data sets in which some of gene sequences are missing.
Results and discussion
Word length and metrics
In order to determine if the distance between signatures can be relevant in phylogenetic analysis, the signature distances between 2 sequences were plotted as function of their observed sequence identity (Fig 1). We simulated a large set of sequences (100 sequences per point) derived from a reference sequence (random mutations with no homoplasy). The signature of the different sequences – the reference sequence and the whole set of modified sequences – were calculated and compared by Euclidian metric in order to obtain distances to the reference. The same plot was obtained with the χ2 metric. These two metrics lead to quite similar results. The χ2 distance exhibited somewhat more information (steeper slope, better dynamics of the plot) than the Euclidean distance and was consequently used. As shown in figure 1, there is a monotonous increase in distance as the observed sequence identity between sequences increased, suggesting the metrics used to compare signatures may be a valid approach to evaluate differences between sequences.
Figure 1 Signature distance as a function of sequence identity. Distances obtained from 5 kb sequences. (6 letter-words, Euclidian metric). Each point represents the mean of 100 sequence comparisons. The standard deviation of each point is shown.
We then tried to determine how tree-like were the trees inferred by the signature method, and if the distances in our signature matrices reflected tree distances. To do that, we used the distance matrices and the trees of the RAG1 study (see below for a discussion of these results). Various criteria for evaluating treeness, such as arboricity and stress, have been used as proposed by Guénoche and Garetta [26] to answer this question. Considering the three sums involved in the four point condition in quadruples [29], arboricity measures the percentage of quadruples for which the middle sum is closer to the largest one than to the smallest one. Stress corresponds to the square root of the quadratic difference between tree and matrice distances divided by the average distance value. These criteria are numerical and topological. All the criteria have been calculated on the signature-based distance matrices. These distance matrices are obtained using different word lengths (between 1 and 10), because we do not have an a priori knowledge of the optimum length.
We found that when word length increases, the arboricity index increases, indicating that the distance improves as a phylogenetic measure (Fig 2). This improvement is clear between 2- and 5-letter words and remains stable for increasing word length. This is in agreement with previous results showing that long words provide better specificity and thus a better taxonomic classification [21]. However, the use of signatures requires that each word occurs frequently enough to provide a good statistical estimation of the true word frequency difference between signatures. The values of the criteria have been also computed for distance matrices of the conventional distance method (Fig 2). From 5-letter words and longer, the criteria from the signature-based distance are better than those of the conventional distance method, especially for the stress criterion. It appears that the different criteria (metric and topological) reached stability and quality for word length around 6-letters. This value of 6 for the word length seems a good trade-off between sequence size and word length and was consequently chosen for additional analyses in this study.
Figure 2 Dynamics of signature distance matrices. Distance matrices were obtained from the RAG1 vertebrate study (see below). There are two types of criteria: metric (for example Vaf, stress) and topological (Arboricity, rate of well designed quadruples, rate of elementary quadruples). Vaf (variance accounted for): quadratic difference divided by the variance of distance. Rate of well designed quadruples: quadruples having the same topology according the two distance matrices; Rate of elementary quadruples, Arboricity; see [26]. On the y-axis, the criteria values obtained from the method of distance are plotted. For the stress, this value is indicated also by a dot line.
Are trees for different word lengths converging on a stable tree or is the tree based on each n-letter word different? To compare trees, the tree dissimilarity criterion (dT) of Robinson-Foulds [30-32], a widely used tree comparison metric, was computed for trees based on n- and the (n+1)-letter word for n = 1 to 9. The dissimilarity distance has also been calculated between n-letter word signature trees and trees obtained by ML and distance methods from conventional aligned sequences (Fig. 3).
Figure 3 Robinson-Foulds distance analysis of trees. The distances were computed from trees of the RAG1 study (see below). For each world length between 1 and 10, a signature tree was computed and compared to the NJ, ML and random trees. For comparison of random trees and signature trees, 100 random trees were built. In this latter case, the dT is approximately 86 (the maximum value possible with this number of species). As a reference, dT between the NJ and ML trees is plotted as dashed line. The dT of the n-/(n+1)-letter word trees was computed for the Euclidean and χ2 metrics.
dT decreases when word length increases (Fig 3), indicating a convergence of the trees towards a stable topology that is reached for 6-letter word whatever the metric used, then for longer word a plateau is observed. The 5- or 6-letter word signature trees are comparable to those obtained by NJ or ML. The dT observed between the signature/NJ or ML trees and those between conventional NJ/ML trees are similar for 5-letter word and higher confirming our choice in 6-letter word for the study.
Simulation of sequences
We decided to compare signatures trees to known trees using simulated sequences from a known phylogeny. Our simulation tests used a protocol similar to the work of Kumar [33] and Gascuel [34]). 100 phylogenetic trees were chosen randomly among a dataset of the 2000 random trees, proposed by Gascuel to test phylogeny methods [35]. These simulation sets are composed of 24-taxon or 96-taxon trees. For each tree T, we used SEQGEN [36] to generate 10 data files with sequences of length 1 kb, 3 kb and 5 kb. These sequences were obtained by simulating the evolution of nucleotides along T according to the Kimura two-parameter model with a transition/transversion rate of 2 and a model of site-specific rate heterogeneity following a gamma distribution (with parameter α = 0.75). We obtained for each length of sequence and each number of taxons 1000 data files.
Two reconstruction methods were applied to the simulated sequences: the signature method, using 4, 5 and 6-letter words and the Euclidian and the χ2 metrics, and the distance method using conventional alignments. We used three different evolutionary model: Kimura two-parameter model (same model than the one used to generated the sequences), a simpler model Jukes-Cantor and a more complex HKY85. All the models have been used with a rate of heterogeneity parameter α equal to 0.75. The results are shown in Table 1.
Table 1 Simulation results with 1000 trees. The values correspond to the proportion of wrong branches in the inferred trees. Two distance metrics (χ2 and Euclidean) were used with three word lengths. For the distance method, three different evolutionary model have been used : JC, K2P et HKY85.
24 taxa 96 taxa
sequence length 1 kb 3 kb 5 kb 3 kb
eucl – 4-letter word 17.8 16.3 16.4 20.5
eucl – 5-letter word 13.8 12.0 11.9 16.0
eucl – 6-letter word 12.9 10.7 10.6 14.9
χ2 – 4-letter word 17.6 16.4 16.4
χ2 – 5-letter word 14.3 12.1 12.0
χ2 – 6-letter word 14.4 11.4 10.9
Jukes-Cantor 11.1 6.3 5.2 9.3
Kimura 2-parameter 10.5 6.1 5.0 9.2
HKY 85 10.5 6.1 5.0 9.2
The methods are compared by their ability to infer the "true" tree, i.e. the topology of the tree that has been used to generate the sequences. We used the topological distance dT of Robinson-Foulds between the inferred tree and the true one. The bipartition distance of Robinson-Foulds [30] is equal to the number of bipartition present in one of the two trees and not in the other. The results are presented in term of percentage of misinferred branches. This percentage is equal to the topological distance divided by the maximum number of different bipartition between two trees: 2N-6 where N is the number of taxa.
In both methods, the Neighbor-Joining reconstruction algorithm was used. The differences in the results come principally from the choice of the distance. The Kimura two-parameter can be designed as the "true" distance, because the parameter of the distance are exactly the same as those chosen to generate the sequences. So normally the Kimura distance must be the branch length of the original trees. The fact that the results obtained by the distance method are not perfect can be attributed to the reconstruction algorithm Neighbor Joining (see Gascuel [37]). HKY85 is a model that includes the Kimura 2-parameter (K2P) model, so the result should be the same.
The proportion of wrong branches decreases in the signature method when word length increases (Table 1). At the same time, the longer the sequences, the better the results with the signature method. However, the proportion of correct branches obtained from the signature is not as high as for the distance method. As expected, the results of HKY85 are the same than those of Kimura 2-parameter. The results of the Jukes-Cantor model are similar to those of the signature for 1 k sequences. But for longer sequences, the signature method is less effective than the JC method. The result of K2P can be explained by the fact that the distance method uses exactly the model used to generate the data. This fact also explains why the results of the signature method improve less with the increase of the sequence length than those of the distance method. The fact that, for the moment, no evolutionary model can be design to the signatures limits the estimation of distances between the signatures. An improvement will be to find how the signature evolves with time as function of nucleotide substitution models. Increases in sequence length facilitate estimation of distance by conventional methods, because the substitution model is known. With the signature, 3 kb sequences are sufficient to obtain a representative signature of the species using 6 letter words. As a result, the increase in accuracy between 3 kb and 5 kb is not significant.
Despite the fact that no evolutionary model has been used with the signature, the results obtained from the signature method are reasonable. With 6-letter words, only 10 % of the internal branches are incorrect. It can be compared to the results presented by Gascuel [37]. The results of the signature method are not as good as the distance method, but they are nevertheless rather accurate. In general, the median size of genes is around 1 k. If we use longer sequences, it will be in the case of non-homologous sequences. For long sequences, no conventional method can be applied.
Vertebrate phylogeny
We used RAG1, a highly conserved gene that produces small distances between sequences to infer the vertebrate phenetic tree [38]. The analysis of the 46 sequences in the dataset had shown that four sequences were complete and the other contained only the conserved core, with length ranging from 1 kb for core sequences to 3 kb for complete ones. This large difference in length induced a bias in the signatures of the four complete sequences, and so in the obtained trees. For comparison with published works [38], we only used the conserved core of RAG1 gene.
A phylogenetic tree was inferred for 46 vertebrate sequences by maximum parsimony, distance (nucleic and protein sequences) and the signature method (Fig 4). Trees produced by classical and signature methods show that position of various vertebrate clades (birds, sharks, mammals, fishes, batrachians) is in agreement with paleontological data. The distance tree obtained using protein sequences exhibited some obvious errors: birds presented a stable group but were placed within mammals (data not shown). Moreover, the relationships between species within each taxonomic group are frequently incongruent with other data. The MP method leads to several most parsimonious trees that are summed up into a consensus tree. On the one hand, the major taxonomic groups can be recovered and are placed correctly; on the other hand, positions of species inside these groups are often poorly inferred (for instance, the relationships between mammals are unresolved).
Figure 4 Phylogeny of vertebrate species. Three methods were applied to the RAG1 gene from 46 species. Distance method: alignment with ClustalW, (Kimura 2-parameter distance), reconstruction by NJ algorithm. MP: use of same alignment. PAUP* has been used with default parameters. Signature method: 6-letter words – χ2 metric. The tree is inferred by NJ method. The bootstrap coefficients for distance and signature method are indicated.
In the signature tree, species are placed within classes in agreement with taxonomy. For example, in the signature analysis, the relationships within birds are congruent with conventional analysis [39]. With regard to mammals, the signature method is the only method that correctly recovers bats as a monophyletic group, with the exception of Felis catus. But the cat, Felis catus, is misplaced by every method, and so its incorrect placement cannot be attributed to a specific phylogenetic method. Mammal relationships appear much more problematic when analyzed by conventional phylogenetic methods than with the signature method. The polyphyly of tetrapods may be explained by the paucity of batrachian sequences, which could lead to an unreliable position for this clade. The monophyly of taxonomic classes, as well as relationships within each class appear quite robust as measured by bootstrap values.
To determine how strong the phylogenetic signal is present in the signature topology, a congruence analysis of phylogenetic trees [40] can be performed. The topologies obtained by ML, MP (the two best trees), NJ and signature (4- to 6-letter word for the Euclidean and the χ2 metrics) methods, are compared by determining the likelihood of each topology. We establish that the signature trees have a phylogenetic signal similar to the alignment-based ones. The signature trees with long words are more congruent than those using small words. The 6-letter word χ2 signature-tree is congruent with the ML tree and the congruence signature/ML is the same than the congruence NJ/ML (Table 2).
Table 2 Difference in log Likelihood. The differences are computed between the ML tree and the other trees.
Tree Δ-ln L
Maximum Likelihood best
Parsimony 9.38
Distance method 58.95
signature
χ2 – 4-letter 445.87
χ2 – 5-letter 297.8
χ2 – 6-letter 65.67
Mean random trees 9132.77
Plant phylogeny
This study, based on an article of Soltis et al. [41], used 18S rRNA for 93 plant species whose sequences are available from the "Green Plant Phylogeny Research Coordination Group" . The species can be grouped into nine main clades (Angiosperms (flowering plants), Conifers, Gnetales, Cycads (palm trees), Hornworts, Liverworts, Ferns, Mosses, Lycophytes), with some additional isolated species and an outgroup.
The signature tree presents significant similarities with the published tree [41]. The angiosperms, conifers, gnetales, cycads and ferns form stable monophyletic groups (high bootstrap coefficients (Fig 5)). The principal result of the article – that the angiosperms are at the root of conifers, gnetales, palm trees and ginkgo (Angiosperm + ((Cycad + Ginkgo) + (Conifern + Gnetale))) – are confirmed by our study and another molecular study [42]. This phylogenetic organization is original as Gnetales are more often linked to Angiosperms by morphological data [43-47] (see Doyle [48] for review).
Figure 5 Phylogenetic tree of plants obtained by comparison of 18S rRNA signatures. (6-letter words – χ2 metric). The bootstrap coefficients (500 sets) of principal groups are indicated. The species class names are indexed by a code: A – Angiosperm, C – Conifer, G – Gnetale, Cyca – Cycad, F – Fern, M – Moss, L – Lycophyte, Lw – Liverwort, Hw – Hornwort. (see annex for the correspondence code/species).
Recent analyses based on molecular data [49] confirms this result (Soltis [41] and Källersjö [49]). In addition, Equisetum and Psilotaceae are placed with the Ferns. This grouping is found in other studies [50,51] and these species are presented as sister group of Ferns. The sister group relationship of Psilotaceae and Ophioglossaceae is also found [52]. Contrary to the results obtained by Soltis, [41] the ferns are polyphyletic in the signature tree.
The outgroup separates the plants into two groups: the seed plants and the other land plants. To confirm the position of this outgroup, 18S rRNA sequences of Homo sapiens, Saccharomyces cerevisiae and Schizosaccharomyces pombe have been added (Data not shown). The outgroup is still confirmed as well as the tree split. This separation of land/flowering plants, the separation of the Lycophytes and the fact that the moss and liverwort do not form a monophyletic clade have been found also by Soltis when a NJ analysis was performed [41]. Thus, the signature method leads to a similar topology as the NJ method with alignment.
Multigene trees
Phylogenetic trees carry two types of signal: species evolution and gene evolution. For a variety of reasons, gene trees can be different from the tree of species from which they are sampled [53]. In addition, signals coming from different genes could lead to different inferred phylogenetic relationships between species [54].
In order to deal with this problem, several genes can be used to build a multigene tree [27,28]. The addition of signals coming from various genes can under some conditions reinforce the information on species evolution. In general, the alignment of each gene can be determined, and alignments concatenated prior to tree building. The signature has many properties that facilitate the calculation of multigene tree.
Another problem deals with the selection of genes participating into the multigene tree. In general, several steps of selection occur to eliminate horizontal transferred genes, duplications or those leading to aberrant phylogeny (see [27,28] for an example of these steps). Signatures are an ideal pretreatment tool for identifying horizontally transferred genes [55], and selecting those genes that conform to evolutionary relationships of the species under consideration. Moreover, due to the rapidity of the treatment with the signature, a very large number of genes can be tested at once.
We propose applying the signature method to infer a consensus tree of multiple genes. Two methods are possible. First, assuming that each gene brings the same quantity of information to the phylogeny for each species, an average signature is computed from several genes. The set of average signatures is then analyzed by the signature method. Another approach is to assume that each gene brings a quantity of phylogenetic information that is correlated with its length. In this approach, the sequences are concatenated and signatures are computed on the set of concatenated sequences.
To carry out this study, we used 33 genes originating from ten species (nine Bacteria: Bacillus subtilis, Clostridium perfringens, Escherichia coli, Lactococcus lactis, Neisseria meningitidis, Salmonella typhimurium, Staphylococcus aureus, Vibrio cholerae, Xanthomonas axonopodis and one Archaebacteria:Archaeoglobus fulgidus – see Material & Methods).
Because the signature does not rely on statements of homology at the level of individual nucleotides, it is possible to compare signatures from different genes in order to quantify statistical patterns and information content among genes. To determine the relative influence of gene evolution versus species evolution in shaping phylogenetic patterns, all the sequences involved in this study (393 sequences) were compared together by means of a hierarchical classification (Fig 6). The hierarchical classification is an unsupervised method allowing the detection of proximities between complex objects. The main result here is the grouping of gene signatures by species (Fig 6), and the species relationships present some differences with the consensus tree. These relationships are more in agreement with the known topology. V. cholerae, E. coli and S. typhimurium form a stable group, but inside this group, the signatures are grouped by genes (Fig 7). The signature of V. cholerae is very close to those of E. coli/S. typhimurium, as well as in the consensus distance matrix. We clearly face a problem of reconstruction of the Neighbor-Joining algorithm. For E. coli and S. typhimurium, the differentiation between these two species is quite recent and the homologous genes are very conserved. This leads to an alternate clustering of genes. In the Gram+, the C. perfringens signatures are very different to the other and place at the root of the Gram+. This confirms the species specificity of the signature, which was known to be present even in short DNA fragments [20]. The signatures of single genes conserve the characteristics of the species from which they are sampled.
Figure 6 Hierarchical classification of 393 6-letter word signatures. The signatures of a given species have the same color code. For each species group, the name of the species is indicated at left. The EF-Tu gene that also forms a stable group is also highlighted. Finally, arrows point out the horizontal transfer (HT) candidates that are discussed in this article.
Figure 7 Detailed view of the hierarchical classification of 393 6-letter word signatures. A detail focusing on the group with E. coli, S. Typhimurium and V. cholerae is shown. The symbols on the left of the names indicate the genes analyzed.
By contrast, an example where gene conservation is very strong is for EF-Tu gene; the signatures of nearly all the species are grouped together at the root of the V. cholerae/E. coli/S. typhimurium group. As it can be observed in the phylogenetic trees (signature and method of distance, results not shown), the A. fulgidus and C. perfringens copies of the gene are quite different, enough to their species signal to be stronger than the EF-Tu signal.
Some gene signatures cluster with species other than their own in the hierarchical tree. This could result from horizontal gene transfer. For instance, the phosphomannomutase gene of S. typhimurium is placed at the root of the S. aureus group. In the phosphomannomutase NJ tree and the signature tree, the relationships between the Gram- and the Gram+ bacteria are incongruent with other data and presumably wrong. Despite that, the other phosphomannomutase signatures are correctly assigned to their host species. The misplacement of this gene may indicate a horizontal transfer in S. typhimurium from an unknown donor. Two other potential horizontal transfers can be found deep inside species group: the elongation factor 2 signature of N. meningitidis and the ornithine carbamoyltransferase signature of S. aureus respectively inside the V. cholerae group and inside the C. perfringens group. In each case the signature is near the signature of the homologous gene of that species. So the gene signal is strong enough to displace the signature inside a different species group. To see if the original sequences are horizontal transfers, we examined two horizontal transfer databases: HGT-DB [56] and HGT Analysis Database [57]. In HGT-DB, the phosphomannomutase sequence of S. Typhimurium is tagged as horizontal transfer [56], but not the other two original sequences detected by the hierarchical classification. Thus our novel result suggests original sequences that need to be studied more precisely before being incorporated into a multigene study.
In all the methods, after removal of dubious genes the consensus tree separates the bacteria into the Gram+ and Gram- groups (Fig 8). But for individual genes this topology is seldom obtained. For Gram+ bacteria, the MP and signature methods lead to a (B. subtilis + (L lactis + (S. aureus + C perfringens))) grouping, but ML and distance methods place B. subtilis deep inside the Gram+ group. For Gram- bacteria, E. coli and S. typhimurium are always grouped and the majority of the methods (exception maximum of parsimony) place N meningitidis and X axonopodis together. The principal difference is the place of V. cholerae within the Gram-. The ML and MP trees place V. cholerae at the root of E. coli and S. typhimurium. The signature method places V. cholerae at the root of Gram- Bacteria.
Figure 8 Consensus trees for ten species. The four methods shown are the signature (6-letter words – χ2 metric) method, distance method, MP and ML. For each method except ML, the bootstrap coefficients (100 sets) are indicated.
To compare the result of the different studies and to determine the dispersion of the phylogenetic trees, we used the dissimilarity distance between the consensus tree and the whole set of gene trees for distance, MP, ML and the signature method (Fig 9) [32]. The distribution of dissimilarity distances indicates that the signature result is independent of the chosen gene and that each individual gene tree is similar to the consensus tree. In this latter case, the variations mainly arise from the placement of V. cholerae, either at the root of Gram- or E. coli/S. typhimurium clades By contrast, the distance method leads to variable results: no distance tree has a dT lower than 6 when compared to the consensus tree. To a lesser degree, the MP and ML trees exhibit a large dispersion (Table 3). Thus a single gene signature tree is less dissimilar from the consensus tree than a conventional one.
Figure 9 Dissimilarity distances between the consensus tree and the sets of genes retained. The dT distances have been computed for the method of distance, ML, MP and signature methods (6-letter word and χ2 metric).
Table 3 Statistical analysis of the distribution of dissimilarity distances as a function of method used.
Method Mean dT Standard deviation
distance 8.47 2.15
parsimony 5.37 2.98
maximum likelihood 5.65 3.28
signature 3 2.3
The consensus signature method allows us to analyze genes present only for some species. We added 9 genes to the data set (see Materials & Methods), thereby increasing the amount of sequence used to 65 kb per species. The signatures of these genes are amalgamated into the species signatures. The tree obtained (data not shown) exhibits the same topology as the consensus tree obtained with the whole set of genes per species computed previously.
The robustness of the consensus tree topology was assessed by computing 100 bootstrap trees. The bootstrap coefficient was 100% for all branches (Fig 8). Another way to test the robustness of the multigene tree is to vary the number of genes per species included in it, as in a jackknife procedure [58]. In this case, 30, 50, 75 and 90 % of the genes available per species are randomly selected. From the selected genes, an average signature is computed for each species. Distances between these average signatures are used to obtain a signature tree. This procedure is performed 100 times per percentage to yield a bootstrap tree. Results show that the topology of the consensus tree is always the same. However, in some cases the bootstrap coefficients are not maximal (table 4).
Table 4 Bootstrap values as function as the number of genes analyzed in the multigene study.
Percentage of used genes 30% 50% 75% 100%
Bootstrap coefficient 100 % except for two clades:
– (E. coli + S. typhimurium) = 91 %
– (N. meningitides + X. axonopodis) = 96 % 100% for all branches 100% for all branches 100% for all branches
In the individual phylogenetic trees, the variations in topologies are so important whatever the method used (except signature) (Fig 9, Table 3), that they do not allow us to confirm whether these sequences have in fact undergone a horizontal transfer.
Phylogeny of γ-proteobacteria
We have shown that using signatures and comparing non-homologous sequences such as are found in complete genomes made it possible to determine the relationship between species. To extend the results obtained with 10 prokaryotes genomes, we explore phylogenetic relationship of a well-studied taxonomic group: the γ-proteobacteria [28]. We selected 16 species whose complete genomes are available. These species can be classified in 6 taxonomic groups (Table 5). Pride et al. [15] used corrected signatures to infer phylogenetic trees. The signatures were corrected by zero order Markov model to normalize the base composition of the different species. Pride et al. [15] determined that this correction permits to obtain a signature tree the most congruent with the 16S rRNA tree. In order to compare the results to a reference, the 16S rRNA sequences have been used to infer a tree by the ML method (Fig 10A). A comparison of trees using signatures corrected and not corrected for base compositional biases is shown in Figures 10B and 10C.
Table 5 Species names and taxonomic groups of γ-proteobacteria.
Species name Taxonomic group
Shewanella oneidensis Alteromonadale
Buchnera aphidicola Enterobacteriale
Escherichia coli Enterobacteriale
Salmonella typhi Enterobacteriale
Salmonella typhimurium Enterobacteriale
Shigella flexneri Enterobacteriale
Yersinia pestis Enterobacteriale
Haemophilus influenzae Pasteurellale
Pasteurella multocida Pasteurellale
Pseudomonas aeruginosa Pseudomonaceae
Pseudomonas putida Pseudomonaceae
Vibrio cholerae Vibrionale
Vibrio vulnificus Vibrionale
Xanthomonas axonopodis Xanthomonadale
Xanthomonas campestris Xanthomonadale
Xylella fastidiosa Xanthomonadale
Figure 10 A- Tree of γ-proteobacteria obtained from the MP method for the 16S rRNA sequences. Each color corresponds to a taxonomic group. B- Tree of γ-proteobacteria obtained from non-corrected signatures (6-letter word signatures and City Block metric). Each color corresponds to a taxonomic group. C- Tree of γ-proteobacteria obtained from the signatures corrected by a zero order Markov model signatures (6-letter word signatures and City Block metric). Each color corresponds to a taxonomic group.
The 16S rRNA tree permits the establishment of reference relationships between the γ-proteobacteria. Some taxonomic groups are recovered: Xanthomonadales, Pseudonomaces as well as Pasteurelles. The tree groups Xanthomonadales and Pseudodomaces, and places B aphidicola close to Pasteurellale but with a long branch. This long branch can explain the incongruent placement of Pasteurelles in Enterobacteria for the ML tree (the phenomenon of long branch attraction [59]).
The tree calculated using the base compositionally-corrected signatures of complete genomes is more in agreement with the 16S rRNA tree. A group of Enterobacteria similar to that found in the 16S rRNA signature tree was obtained. However the monophyly of Xanthomonadales is not recovered in any of the trees obtained from signatures of complete genomes. X fastidiosa is placed at the root of the group (Xanthomonadale + Pseudomaceae). Another difference between trees for complete genomes and those of the 16S RNA is a grouping of Pasteurellales, Vibrionales and S. oneidensis found in the signature tree.
In the complete genome trees, B aphidicola appears misplaced. It is always positioned apart from the Enterobacterial clade, despite its belonging to this group taxonomically. An analysis of genome signatures of B aphidicola revealed that this species exhibits a very different signature from those of the other γ-proteobacteria (result not shown). This result is not due to a bias in signature method arising from the size of B aphidicola genome, because a tree obtained by randomly selecting the same sequence length in the 15 other genomes (650 kb) leads to the same topology (result not shown). We suggest that the source of this anomoly is that B aphidicola is a symbiotic bacteria, andhas a very small genome (650 KB) compared with those of the other γ-proteobacteria (4 to 5 Mb). This genome reduction arises from its parasitic lifestyle and is the result of many independent losses of genes and genomic segments. B aphidicola experienced very strong evolutionary pressures that led to a profound shift in its signature, and also transferred numerous genes to its host [60]. Such symbiotic species are known to be difficult to place phylogenetically [61]. B. aphidicola also has a strong compositional bias (the genome of is nearly 75% AT rich). The other γ-proteobacteria are more GC rich. These problems appear when using whole genomes to infer a tree and are bypassed when using conserved genes or a selection of genes sharing a common history [28].
We used the method of Dufraigne et al. [55] to detect in the B aphidicola genome sequences that may have arisen by horizontal transfer. We divided the entire genome into 5 kb sequence windows. For each window, a 4-letter word signature was computed. The method developed by Dufraigne et al. allows us to detect which sequences have original signatures such as would be found in cases of horizontal tranfer. We removed this original sequences from the genome and a new 6-letter word signature was computed. The tree obtained is exactly the same as the base compositionally-corrected tree (Fig 10C).
Conclusion
In this paper we have illustrated the exploration of phylogenetic data with a global sequence analysis method, the signature method. Using a variety of genes, this method yields tree topologies similar to those obtained using traditional phylogenetic approaches. The results presented here suggest that trees obtained by this method could be used as an exploratory step in phylogenetic studies. The signature method can deliver a quick overview of phylogenetic relationships between species in data sets that can be challenging or time consuming for traditional alignment and phylogenetic analysis. As our simulations showed, the signature method sometimes yields phylogenies that are less accurate than those produced by conventional analyses, but this arises mainly from the fact that no evolutionary model is known for word frequencies comprising genomic signatures. The signature tree can be used as fast pretreatment in conjunction with classical methods such as ML. We also demonstrated that the signature distances are tree-like, reflect tree distances and that in the case of short sequences such as frequently assembled in studies of homologous sequences, the optimal word length seems to be 6. This length represents a trade-off between long words that represent more accurately the DNA sequences [21,25] and the size of the sequences.
The signature method is particularly useful as a first step in data exploration. The speed of the analysis permits detection of either misplacement of particular species, in some cases due to local composition fluctuations (horizontal transfer), or unexpected groupings of species that can be scrutinized further by biological means or conventional phylogenetic study. Thus, the signature method easily permits the researcher to use long and/or numerous genes in a study. When using numerous species, their phylogenetic proximities can be analyzed using their signatures by conventional statistical methods and the set of species split into subgroups. This method is also useful in combining information from different genes. The signature method permits the averaging of a great number of genes of any length to obtain a consensus and a unique signature per species and thereby take into account a great number of evolutionary events. The signature method does not rely on homology of DNA sites to compare sequences and it is possible to compare non-homologous sequences to infer a phylogenetic tree. Thus, many genes not present in every species can be added to this tree, giving more confidence in the species tree. This approach was already applied to birds [24], bacterial [15-18,21,62] or mitochondrial [25] phylogenetic studies. In contrast to conventional methods, the signature method utilizes information present in the sequences that may not be analyzable with conventional alignments, such as additional sequences at the beginning or the end of alignments.
For studies of complete genomes, detection of horizontal transfer using signatures, such as proposed by Dufraigne et al. [55], permits removal of sequences that will compromise phylogenetic analysis. Finally, signatures allow the rapid detection of horizontally transferred genes or simply misplaced genes that require additional attention via hierarchical clustering or other statistical classification methods.
Methods
Sequence signature
Sequence signature can be computed easily and very quickly thanks to an algorithm -the "Chaos game representation" (CGR)-, (about 1 Mb per second on a laptop computer) [63]. The signature can be displayed as an image, where each pixel represents a word and the darkness of the pixel increases with the frequency of the word in the sequence.
DNA sequences
We selected two genes to compare signature analysis of two different clades with results from the literature. These genes are long enough to get a significant signature and address the phylogeny of vertebrates and plants including a large number of species. The recombination activation gene RAG1 is used for inferring the phenetic tree of 46 species of vertebrates. Ribosomal RNA sequence analysis is the de facto standard for phylogenetic reconstruction. Here we use ribosomal 18S RNA to analyze 93 plant species. Finally, 42 genes, accounting for more than 50 kb of sequence, are used for a multigene study (see Annex), including nine Bacteria and one Archaea. To select the 42 genes, we utilized the SYSTERS database [64]. For all the selected species, the database returned 119 orthologous protein families shared by the whole set of species. These families were filtered by size of the corresponding DNA sequences (retained families contain sequences with mean lengths > 1 kb). From these, 33 complete sets and 9 partial sets of genes were obtained. The selected genes belong mainly to amino acid, nucleotide and protein synthesis and DNA metabolism families. All the sequences were extracted from GenBank or Genome Information Broker [65]. The complete genomes of 16 γ-proteobacteria were gathered from GenBank (see appendix). Simulated sequences from a known phylogeny were found on Gascuel's website [35].
Phylogenetic analysis and signature method
Two distance metrics (Euclidean and ?2) were used to quantify the differences between signatures. Other metrics (Manhattan, Mahalanobis, Correlation and Cosine) were investigated as well; these methods rarely performed better than our two focal methods, and often performed worse, so we do not consider them further. Distance matrices were obtained via the Euclidean and ?2 metrics. We used these matrices to infer trees with the Neighbor-Joining (NJ) reconstruction algorithm implemented in the PHYLIP package [66]. In order to estimate the robustness of the tree topology, we simulated by bootstrap [67] a whole new set of signatures from the initial set of motif frequencies, sampling with replacement (in general, 100 bootstrap trees were computed). Each dataset contains the same individuals from the initial data and N new variables (words) randomly drawn in order to replace the N variables from the initial set [67]. For each set of sequences, the phylogenetic tree was inferred and a consensus tree was calculated from each bootstrap replicate. Besides the signature method, three commonly used methods [3] were used to analyze aligned sequences from the same data sets: Neighbor-Joining (NJ) [68], maximum parsimony (MP) [69] and maximum of likelihood (ML) [2]. All three methods were implemented using the PAUP* [70] and PHYLIP packages. Alignments were obtained with ClustalW (default parameters)[4] and were similar to those used in their respective sources. For the different conventional methods, we have used the HKY85 model of sequence evolution, and gaps were treated as missing data in the MP analysis. For ML analyses, a gamma distribution of rate heterogeneity with simultaneous parameter estimation was used.
Appendix
Species annotation for the 18S rRNA sequences of plants
A1: Asarum canadense; A2: Sparganium eurycarpum; A3: Tetracentron sinense; A4: Trochodendron aralioides; A5: Austrobaileya scandens; A6: Sassafras albidum; A7: Akebia quinata; A8: Amborella trichopoda; A9: Camptotheca acuminata; A10: Gossypium hirsutum; A11: Celtis yunnanensis; A12: Canna coccinea; A13: Ceratophyllum demersum; A14: Dipsacus sp; A15: Liquidambar styraciflua; A16: Zea mays; A17: Nymphaea tuberosa; A18: Oncidium excavatum; A19: Phytolacca americana; A20: Pisum sativum; A21: Symphoricarpos albus; A22: Saururus cernuus; A23: Saxifraga integrifolia; A24: Saruma henryi; C1: Araucaria excelsa; C2: Cephalotaxus wilsoniana; C3: Juniperus chinensis; C4: Phyllocladus trichomonoides; C5: Pinus elliottii; C6: Pinus luchuensis; C7: Dacrycarpus imbricatus; C8: Amentotaxus formosana; C9: Torreya nucifera; C10: Taiwania cryptomerioides; C11: Podocarpus costalis; C12: Nageia nagi; C13: Taxus chinensis var. mairei; C14: Abies lasiocarpa; Cyca1: Cycas taitungensis; Cyca2: Zamia pumila; Equisetum: Equisetum hyemale; F1: Adiantum raddianum; F2: Blechnum occidentale; F3: Dicksonia antarctica; F4: Dicranopteris linearis; F5: Hypolepis muelleri; F6: Lonchitis hirsuta; F7: Osmunda cinnamomea; F8: Odontosoria chinensis; F9: Ophioglossum petiolatum; F10: Pteridium aquilinum; F11: Salvinia natans; F12: Vandenboschia davallioides; G1: Welwitschia mirabilis; G2: Ephedra sinica; G3: Ephedra torreyana; G4: Gnetum nodiflorum; G5: Gnetum urens; G6: Gnetum gnemon; Ginkgo: Ginkgo biloba; Hw1: Anthoceros agrestis; Hw2: Notothylas breutelii; Hw3: Phaeoceros laevis; L1: Huperzia lucidula; L2: Isoetes durieui; L3: Isoetes engelmannii; L4: Lycopodiella inundata; L5: Huperzia phlegmaria; L6: Huperzia taxifolia; L7: Lycopodium tristachyum; L8: Selaginella umbrosa; L9: Selaginella vogelii; Lw1: Marchantia polymorpha; Lw2: Fossombronia pusilla; Lw3: Pellia epiphylla; Lw4: Reboulia hemisphaerica; Lw5: Sphaerocarpos donnelli; Lw6: Scapania nemorea; Lw7: Riccardia pinguis; M1: Physcomitrella patens; M2: Atrichum undulatum; M3: Eurhynchium hians; M4: Funaria hygrometrica; M5: Leptobryum pyriforme; M6: Polytrichum formosum; M7: Physcomitrium pyriforme; M8: Sphagnum cuspidatum; O1a: Chara australis; O1b: Chara connivens; O1c: Chara foetida; O2a: Nitella flexilis; O2b: Nitella sp; Psilo1: Psilotum nudum; Psilo2: Tmesipteris tannensi.
Table 6 Genes used in multigene study:
1/ whole set of species:
Adenylosuccinate lyase 1.3 kb
Adenylosuccinate synthetase 1.3 kb
Alanyl-tRNA synthetase 2.6 kb
Argininosuccinate synthase 1.3 kb
Argininosuccinate lyase 1.4 kb
Arginyl-tRNA synthetase 1.7 kb
Aspartate aminotransferase 1.2 kb
Aspartyl-tRNA synthetase 1.8 kb
Carbamyl-phosphate synthase 3.2 kb
Cell division protein ftsZ 1.2 kb
Chorismate synthase 1.1 kb
CTP synthase 1.6 kb
DNA-directed RNA polymerase 2.6 kb
DNA topoisomerase I 2.0 kb
Elongation factor 2 2.1 kb
Enolase 1.3 kb
5-enolpyruvylshikimate-3-phosphate synthetase 1.3 kb
Glutamine synthetase 1.5 kb
Leucyl-tRNA synthetase 2.8 kb
Methionyl-tRNA synthetase 2.1 kb
Ornithine carbamoyltransferase 1.0 kb
Pantothenate metabolism flavoprotein 1.2 kb
D-3-phosphoglycerate dehydrogenase 1.2 kb
Phosphoglycerate kinase 1.2 kb
Phosphomannomutase 1.3 kb
Phosphoribosylformylglycinamidine synthase II 3.8 kb
Queuine tRNA-ribosyltransferase 1.1 kb
Ribonucleotide reductase 2.3 kb
Serine hydroxymethyltransferase 1.2 kb
Thermosome alpha subunit 1.6 kb
Threonyl-tRNA synthetase 2.0 kb
Translation elongation factor EF-Tu 1.3 kb
Valyl-tRNA synthetase 2.6 kb
Total length = 57.2 kb
2/ partial set of species:
Acetolactate synthase large subunit 1.7 kb
Cysteinyl-tRNA synthetase 1.4 kb
Galactosyltransferase 1.1 kb
GTP cyclohydrolase II 1.1 kb
Histidine kinase 2.0 kb
Phosphoenolpyruvate synthase 2.4 kb
dTDP-glucose 4,6-dehydratase 1.1 kb
Tryptophan synthase subunit beta 1.2 kb
X-pro aminopeptidase 1.3 kb
Authors' contributions
CC and DP conceived the study, drew the figures and wrote the first draft of the manuscript. CC was the main contributor of the bioinformatic analysis. CD participated in the bioinformatic study. SE participated in the method design and drafted the manuscript. AG and BF drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This research was supported by contract 120910 from the "Action inter EPST Bio-informatique 2001" of the French Research Ministry and contract A02114DS from the "Action inter EPST Bio-informatique 2003" of the French Research Ministry.
==== Refs
Lecointre G Le Guyader H Classification phylogénétique du vivant 2001 Paris, Belin 544
Felsenstein J Evolutionary trees from DNA sequences: a maximum likelihood approach J Mol Evol 1981 17 368 376 7288891
Li WH Sinauer A Molecular Evolution 1997 Sinauer , Sinauer 487
Higgins DG Thompson JD Gibson TJ Using CLUSTAL for multiple sequence alignments Methods Enzymol 1996 266 383 402 8743695
Brocchieri L Phylogenetic inferences from molecular sequences: review and critique Theor Popul Biol 2001 59 27 40 11243926
Russo CAM Takezaki N Nei M Efficiencies of different genes and different tree-building methods in recovering a known vertebrate phylogeny Mol Biol Evol 1996 13 525 536 8742641
Billoud B Guerrucci MA Masselot M Deutsch JS Cirripede phylogeny using a novel approach: molecular morphometrics Mol Biol Evol 2000 17 1435 1445 11018151
Wolf YI Rogozin IB Grishin NV Tatusov RL Koonin EV Genome trees constructed using five different approaches suggest new major bacterial clades BMC Evol Biol 2001 1 8 11734060
Korbel JO Snel B Huynen MA Bork P SHOT: a web server for the construction of genome phylogenies Trends Genet 2002 18 158 162 11858840
Tekaia F Lazcano A Dujon B The genomic tree as revealed from whole proteome comparisons Genome Res 1999 9 550 557 10400922
Daubin V Gouy M Perrière G A phylogenetic approach to bacterial phylogeny: evidence of a core of genes sharing a common history. Genome Research 2002 12 1080 1090 12097345
Gupta RS Protein phylogenies and signature sequences: a reappraisal of evolutionary relationships among Archaebacteria, Eubacteria and Eukaryotes. Microbiol Mol Biol Rev 1998 62 1435 1491 9841678
Rivera MC Lake JA Evidence that eukaryotes and eocyte prokaryotes are immediate relatives Science 1992 257 74 76 1621096
Karlin S Burge C Dinucleotide relative abundance extremes: a genomic signature Trends In Genetics 1995 11 283 290 7482779
Pride DT Meinersmann RJ Wassenaar TM Blaser MJ Evolutionary implications of microbial genome tetranucleotide frequency biases Genome Res 2003 13 145 158 12566393
Yu ZG Jiang P Distance, correlation and mutual information among portraits of organisms based on complete genomes Phys Lett A 2001 286 34 46
Qi J Wang B Hao BI Whole proteome prokaryote phylogeny without sequence alignment: a K-string composition approach J Mol Evol 2004 58 1 11 14743310
Stuart GW Moffett K Leader JJ A comprehensive vertebrate phylogeny using vector representations of protein sequences from whole genomes Mol Biol Evol 2002 19 554 562 11919297
Deschavanne PJ Giron A Vilain J Fagot G Fertil B Genomic signature: characterization and classification of species assessed by Chaos Game Representation of sequences Molecular Biology and Evolution 1999 16 1391–1399 10563018
Deschavanne P Giron A Vilain J Dufraigne C Fertil B Genomic signature is preserved in short DNA fragments BIBE2000 IEEE international Symposium on bio-informatics & biomedical engineering, Washington USA, 8-10 november 2000 2000 161 167
Chapus C Fertil B Edwars S Giron A Deschavanne P Spang R, Béziat P and Vingron M Classification of species based on DNA style: ; Berlin, Germany. 2003 147 148
Karlin S Bacterial DNA strand compositional asymmetry Trends Microbiol 1999 7 305 308 10431198
Karlin S Mràzek J Campbell AM Compositional biases of bacterial genomes and evolutionary implications J Bact 1997 179 3899 3913 9190805
Edwards SV Fertil B Giron A Deschavanne PJ A genomic schism in birds revealed by phylogenetic analysis of DNA strings Syst Biol 2002 51 599 613 12228002
Wang Y Hill K Singh S Kari L The spectrum of genomic signatures: from dinucleotides to chaos game representation Gene 2005 346 173 185 15716010
Guénoche A Garreta H Gascuel O and Sagot MF Can We Have Confidence in a Tree Representation?: 2001; Montpelier, France. Lecture Notes in Computer Science Goos G, Hartmanis J, van Leeuwen J 2000 2066 Springer 45 56
Daubin V Gouy M Perriere G A phylogenomic approach to bacterial phylogeny: evidence of a core of genes sharing a common history Genome Res 2002 12 1080 1090 12097345
Lerat E Daubin V Moran NA From gene trees to organismal phylogeny in prokaryotes: the case of the g-proteobacteria PLOS Biology 2003 1 101 109
Buneman P Hobson FR, Kendall DG and Tautu P Mathematics in the Archeological and Historial Sciences 1971 Edinburgh, University Press 387
Robinson DF Foulds LR Comparaison of phylogenetic trees Math Biosci 1981 53 131 147
Kuhner MK Felsenstein J A simulation comparison of phylogeny algorithms under equal and unequal evolutionary rates Mol Biol Evol 1994 11 459 468 8015439
Penny D Hendy MD The use of tree comparaison metrics Systematic Zoology 1985 34 75 82
Kumar S A stepwise algorithm for finding minimum evolution trees Mol Biol Evol 1996 13 584 593 8882501
Ranwez V Gascuel O Quartet-based phylogenetic inference: improvements and limits Mol Biol Evol 2001 18 1103 1116 11371598
Gascuel O http://www.lirmm.fr/~w3ifa/MAAS/US-MAAS.html
Rambaut A Grassly NC Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees Comput Appl Biosci 1997 13 235 238 9183526
Gascuel O Baxevanis A, Davison D, Page R, Stein L, Stormo G and Petsko G Getting a Tree Fast: Neighbor Joining and Distance Based Methods Current Protocols in Bioinformatics Current Protocols & Sons Wiley 2004 , Wiley & Sons 6.3.1 6.3.18
Greenhalgh P Steiner LA Recombination activating gene 1 (Rag1) in zebrafish and shark Immunogenetics 1995 41 54 55 7806278
Sibley CG Ahlquist JA Phylogeny and classification of birds: a study in molecular evolution 1990 New Haven, Yale University Press
Feil EJ Holmes EC Bessen DE Chan MS Day NP Enright MC Goldstein R Hood DW Kalia A Moore CE Zhou J Spratt BG Recombination within natural populations of pathogenic bacteria: short-term empirical estimates and long-term phylogenetic consequences Proc Natl Acad Sci U S A 2001 98 182 187 11136255
Soltis PS Soltis DE Wolf PG Nickrent DL Chaw SM Chapman RL The phylogeny of land plants inferred from 18S rDNA sequences: pushing the limits of rDNA signal? Mol Biol Evol 1999 16 1774 1784 10605118
Chaw SM Parkinson CL Cheng Y Vincent TM Palmer JD Seed plant phylogeny inferred from all three plant genomes: monophyly of extant gymnosperms and origin of Gnetales from conifers Proc Natl Acad Sci U S A 2000 97 4086 4091 10760277
Crane PR Phylogenetic analysis of seed plants and the origin of the angiosperms Ann Mo Bot Gard 1985 72 716 793
Doyle JA Seed plant phylogeny and the relationshipsof Gnetales Int J Plant Sci 1996 157 S3 S39
Doyle JA Molecules, morphology, fossils, and the relationship of angiosperms and Gnetales Mol Phylogenet Evol 1998 9 448 462 9667993
Doyle JA Donoghue MJ Seed plant phylogeny and the origin of the angiosperms: an experimental cladistic approach Bot Rev 1986 52 321 431
Doyle JA Donoghue MJ Zimmer EA Integration of morphological and ribosomal RNA data on the origin of the angiosperms Ann Mo Bot Gard 1994 81 419 450
Doyle JA Phylogeny of vascular plants Annu Rev Ecol Syst Biol 1998 29 567–569
Källersjö M Farris JS Chase W Bremer B Fay MF Humpries CJ Petersen G Seberg O Bremer K Simultaneous parsimony jackknife analysis of 2538 rbcl DNA sequences reveals upport for major clades of green plants, land plants, seed plants and flowering plants. Plant Syst Evol 1998 213 2599 2287
Kenrick P Crane PR The origin and early evolution of plants on land. nature 1997 389 33 39
Manhart JR Phylogenetic analysis of green plant rbcL sequences Mol Phylogenet Evol 1994 3 114 127 8075831
Manhart JR Chloroplast 16S rDNA sequences and phylogeneic relationships of ferns and fern allies Am Fern J 1995 85 182 192
Rokas A Williams BL King N Carroll SB Genome-scale approaches to resolving incongruence in molecular phylogenies Nature 2003 425 798 804 14574403
Gray MW Burger G Lang BF Mitochondrial evolution Science 1999 283 1476 1481 10066161
Dufraigne C Fertil B Lespinats S Giron A Deschavanne P Detection and characterization of horizontal transfers in prokaryotes using genomic signature Nucleic Acids Res 2005 33 e6 15653627
Garcia-Vallve S Guzman E Montero MA Romeu A HGT-DB: a database of putative horizontally transferred genes in prokaryotic complete genomes Nucleic Acids Research 2003 31 187 189 12519978
Nakamura Y Itoh T Matsuda H Gojobori T Biased biological functions of horizontally transferred genes in prokaryotic genomes Nature Genetics 2004 36 760 766 15208628
Lanyon SM Detecting internal inconsistencies in distance data Systematic Zoology 1985 34 397 403
Felsenstein J Cases in wich parsimony or compability methods will be positively misleading. Systematic Zoology 1978 27 401 410
Moran NA Mira A The process of genome shrinkage in the obligate symbiont Buchnera aphidicola Genome Biol 2001 2 RESEARCH0054 11790257
Itoh T Martin W Nei M Acceleration of genomic evolution caused by enhanced mutation rate in endocellular symbionts Proc Natl Acad Sci U S A 2002 99 12944 12948 12235368
Yu ZG Anh V Lau KS Measure representation and multifractal analysis of complete genomes Phys Rev E 2001 64 1 9
Jeffrey HJ Chaos game representation of gene structure Nucleic Acids Research 1990 18 2163 2170 2336393
Krause A Stoye J Vingron M The SYSTERS protein sequence cluster set Nucleic Acids Res 2000 28 270 272 10592244
GIB http://gib.genes.nig.ac.jp/
Felsenstein J PHYLIP (Phylogeny Inference Package) 3.6 2004
Felsenstein J Confidence limits on phylogenies : an approach using the bootstrap Evolution 1985 4 783 791
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Fitch WM On the problem of discovery the most parsimonious tree Am Nat 1977 111 223 257
Swofford DL PAUP*, Phylogenetic Analysis Using Parsimony (*and Other Methods) 2003 4 , Sinauer Associates, Sunderland, Massachusetts
|
16280081
|
PMC1310607
|
CC BY
|
2021-01-04 16:29:17
|
no
|
BMC Evol Biol. 2005 Nov 9; 5:63
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-63
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-631628008110.1186/1471-2148-5-63Methodology ArticleExploration of phylogenetic data using a global sequence analysis method Chapus Charles [email protected] Christine [email protected] Scott [email protected] Alain [email protected] Bernard [email protected] Patrick [email protected] Equipe de Bioinformatique Génomique et Moléculaire, INSERM U 726, Case 7113, Tour 53-54, 2 place Jussieu, 75005 Paris, France2 Inserm U494, 91 bd de l'Hopital 75634 Paris CEDEX 13, France3 Dept. of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138 USA4 Current address: Dept. of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138 USA2005 9 11 2005 5 63 63 15 4 2005 9 11 2005 Copyright © 2005 Chapus et al; licensee BioMed Central Ltd.2005Chapus et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Molecular phylogenetic methods are based on alignments of nucleic or peptidic sequences. The tremendous increase in molecular data permits phylogenetic analyses of very long sequences and of many species, but also requires methods to help manage large datasets.
Results
Here we explore the phylogenetic signal present in molecular data by genomic signatures, defined as the set of frequencies of short oligonucleotides present in DNA sequences. Although violating many of the standard assumptions of traditional phylogenetic analyses – in particular explicit statements of homology inherent in character matrices – the use of the signature does permit the analysis of very long sequences, even those that are unalignable, and is therefore most useful in cases where alignment is questionable. We compare the results obtained by traditional phylogenetic methods to those inferred by the signature method for two genes: RAG1, which is easily alignable, and 18S RNA, where alignments are often ambiguous for some regions. We also apply this method to a multigene data set of 33 genes for 9 bacteria and one archea species as well as to the whole genome of a set of 16 γ-proteobacteria. In addition to delivering phylogenetic results comparable to traditional methods, the comparison of signatures for the sequences involved in the bacterial example identified putative candidates for horizontal gene transfers.
Conclusion
The signature method is therefore a fast tool for exploring phylogenetic data, providing not only a pretreatment for discovering new sequence relationships, but also for identifying cases of sequence evolution that could confound traditional phylogenetic analysis.
==== Body
Background
Phylogenetic classifications traditionally rely on phenotypic traits and the paleontological record [1]. As a result of the large amount of DNA sequences now available in the databases, molecular phylogeny has become an essential companion in studying evolutionary relationships among species [2]. As usually practiced, it allows constructing phylogenetic trees based on differences between homologous sequences or genes [3]. A basic and indispensable step in phylogenetic study is alignment of the set of homologous sequences [4]. However, distantly related sequences can be difficult to align and under these conditions, different algorithms often lead to different phylogenetic results [5,6]. There are other problems linked to the use of biological sequences in phylogenetic analysis, including sampling of representative sequences, biological processes such as lateral gene transfer, fusion events and recombination (see Brocchieri et al [5] for a review).
New approaches of molecular phylogeny, taking into account new characteristics of sequences, have been recently developed. Such methods include using other aspects of molecular data such as structural properties of proteins [7], the presence and organization of genes along genomes [8-11], occurrence of characteristic patterns [12,13] and the frequencies of short nucleotide or peptide relative abundance [14-18]. These methods contribute to the understanding of species evolution from different points of view, particularly in terms of our understanding of genome evolution. What is intriguing about these methods is that they often yield phylogenetic results comparable to those of traditional methods, frequently employing data sets much larger than traditional phylogenetic analyses. As such, they deserve the attention of those wishing to extract maximal information from comparative genomic data sets.
We expand on a method to characterize DNA sequences: the sequence signature. Sequence signature is defined as the whole set of frequencies of short oligonucleotides (words, until ten nucleotides long currently) of a sequence [19]. The principal characteristics of sequence signature used for phylogenetic studies are species-specificity of sequence signature and conservation of signature in any part of the genome [20] allowing researchers to compare sequences from diverse regions of the genome. It has already been established that distances between species signatures of the same taxonomic group are smaller than between signatures of species belonging to different groups [19,21]. A difference of signatures between two sequences could arise from shifts in the pattern of point substitution, but could also involve interactions among adjacent nucleotides, natural selection, DNA repair processes and conformational constraints (super coiling, nucleosome formation, bend DNA) [22]. A phylogenetic analysis of signatures could therefore reflect underlying genomic changes that shift motif frequencies, thereby yielding higher-order homologies available for phylogenetic analysis. The method has already been used for taxonomic classification of some species groups [23-25]. One advantage of such a method consists mainly in avoiding the alignment step, and can be used on numerous sequences of varying size. In addition, distance matrices, such as those applicable to genomic signatures, generally permit fast building of trees. Perhaps most importantly, genomic signatures provide a means of comparing large-scale patterns in genomes and can help evaluate trends in genome evolution across a phenetic tree. However, no systematic analysis of the reliability of the signature approach has been performed on homologous sequences. It has been demonstrated that long word frequencies describes DNA sequence information more accurately [19,25], but with their much larger number, long words are difficult to apply to short sequences because word frequencies are poorly estimated. Wang et al. [25] have also qualitatively analyzed the impact of the choice of the divergence metrics on phylogenetic results. However, no quantitative analyses or simulations have been presented yet on this subject.
In this paper, statistical studies of the ability of a signature approach for reconstructing phylogenies are investigated, specifically in order to determine the optimum word length and the influence of the divergence metric on the results. One of the tests we employ allows us to determine whether the signature distance can be considered tree-like, possessing hierarchical information [26]. Working with homologous, fully alignable sequences, we tested the method on simulated sequences whose true topologies are known and also analyzed two published examples of DNA sequences that propose novel interspecific relationships. Overall we find that there is a strong correspondence between signature trees and those generated by conventional means. As a means of improving large multi-gene studies [27,28], we also propose the use of signatures for rapid, large-scale sequence analysis specifically to detect subsets of genes supporting similar species phylogenies and to identify cases of horizontal transfer. In an analysis of 16 complete γ-proteobacteria genomes, we also illustrate how the signature method can also be used on data sets in which some of gene sequences are missing.
Results and discussion
Word length and metrics
In order to determine if the distance between signatures can be relevant in phylogenetic analysis, the signature distances between 2 sequences were plotted as function of their observed sequence identity (Fig 1). We simulated a large set of sequences (100 sequences per point) derived from a reference sequence (random mutations with no homoplasy). The signature of the different sequences – the reference sequence and the whole set of modified sequences – were calculated and compared by Euclidian metric in order to obtain distances to the reference. The same plot was obtained with the χ2 metric. These two metrics lead to quite similar results. The χ2 distance exhibited somewhat more information (steeper slope, better dynamics of the plot) than the Euclidean distance and was consequently used. As shown in figure 1, there is a monotonous increase in distance as the observed sequence identity between sequences increased, suggesting the metrics used to compare signatures may be a valid approach to evaluate differences between sequences.
Figure 1 Signature distance as a function of sequence identity. Distances obtained from 5 kb sequences. (6 letter-words, Euclidian metric). Each point represents the mean of 100 sequence comparisons. The standard deviation of each point is shown.
We then tried to determine how tree-like were the trees inferred by the signature method, and if the distances in our signature matrices reflected tree distances. To do that, we used the distance matrices and the trees of the RAG1 study (see below for a discussion of these results). Various criteria for evaluating treeness, such as arboricity and stress, have been used as proposed by Guénoche and Garetta [26] to answer this question. Considering the three sums involved in the four point condition in quadruples [29], arboricity measures the percentage of quadruples for which the middle sum is closer to the largest one than to the smallest one. Stress corresponds to the square root of the quadratic difference between tree and matrice distances divided by the average distance value. These criteria are numerical and topological. All the criteria have been calculated on the signature-based distance matrices. These distance matrices are obtained using different word lengths (between 1 and 10), because we do not have an a priori knowledge of the optimum length.
We found that when word length increases, the arboricity index increases, indicating that the distance improves as a phylogenetic measure (Fig 2). This improvement is clear between 2- and 5-letter words and remains stable for increasing word length. This is in agreement with previous results showing that long words provide better specificity and thus a better taxonomic classification [21]. However, the use of signatures requires that each word occurs frequently enough to provide a good statistical estimation of the true word frequency difference between signatures. The values of the criteria have been also computed for distance matrices of the conventional distance method (Fig 2). From 5-letter words and longer, the criteria from the signature-based distance are better than those of the conventional distance method, especially for the stress criterion. It appears that the different criteria (metric and topological) reached stability and quality for word length around 6-letters. This value of 6 for the word length seems a good trade-off between sequence size and word length and was consequently chosen for additional analyses in this study.
Figure 2 Dynamics of signature distance matrices. Distance matrices were obtained from the RAG1 vertebrate study (see below). There are two types of criteria: metric (for example Vaf, stress) and topological (Arboricity, rate of well designed quadruples, rate of elementary quadruples). Vaf (variance accounted for): quadratic difference divided by the variance of distance. Rate of well designed quadruples: quadruples having the same topology according the two distance matrices; Rate of elementary quadruples, Arboricity; see [26]. On the y-axis, the criteria values obtained from the method of distance are plotted. For the stress, this value is indicated also by a dot line.
Are trees for different word lengths converging on a stable tree or is the tree based on each n-letter word different? To compare trees, the tree dissimilarity criterion (dT) of Robinson-Foulds [30-32], a widely used tree comparison metric, was computed for trees based on n- and the (n+1)-letter word for n = 1 to 9. The dissimilarity distance has also been calculated between n-letter word signature trees and trees obtained by ML and distance methods from conventional aligned sequences (Fig. 3).
Figure 3 Robinson-Foulds distance analysis of trees. The distances were computed from trees of the RAG1 study (see below). For each world length between 1 and 10, a signature tree was computed and compared to the NJ, ML and random trees. For comparison of random trees and signature trees, 100 random trees were built. In this latter case, the dT is approximately 86 (the maximum value possible with this number of species). As a reference, dT between the NJ and ML trees is plotted as dashed line. The dT of the n-/(n+1)-letter word trees was computed for the Euclidean and χ2 metrics.
dT decreases when word length increases (Fig 3), indicating a convergence of the trees towards a stable topology that is reached for 6-letter word whatever the metric used, then for longer word a plateau is observed. The 5- or 6-letter word signature trees are comparable to those obtained by NJ or ML. The dT observed between the signature/NJ or ML trees and those between conventional NJ/ML trees are similar for 5-letter word and higher confirming our choice in 6-letter word for the study.
Simulation of sequences
We decided to compare signatures trees to known trees using simulated sequences from a known phylogeny. Our simulation tests used a protocol similar to the work of Kumar [33] and Gascuel [34]). 100 phylogenetic trees were chosen randomly among a dataset of the 2000 random trees, proposed by Gascuel to test phylogeny methods [35]. These simulation sets are composed of 24-taxon or 96-taxon trees. For each tree T, we used SEQGEN [36] to generate 10 data files with sequences of length 1 kb, 3 kb and 5 kb. These sequences were obtained by simulating the evolution of nucleotides along T according to the Kimura two-parameter model with a transition/transversion rate of 2 and a model of site-specific rate heterogeneity following a gamma distribution (with parameter α = 0.75). We obtained for each length of sequence and each number of taxons 1000 data files.
Two reconstruction methods were applied to the simulated sequences: the signature method, using 4, 5 and 6-letter words and the Euclidian and the χ2 metrics, and the distance method using conventional alignments. We used three different evolutionary model: Kimura two-parameter model (same model than the one used to generated the sequences), a simpler model Jukes-Cantor and a more complex HKY85. All the models have been used with a rate of heterogeneity parameter α equal to 0.75. The results are shown in Table 1.
Table 1 Simulation results with 1000 trees. The values correspond to the proportion of wrong branches in the inferred trees. Two distance metrics (χ2 and Euclidean) were used with three word lengths. For the distance method, three different evolutionary model have been used : JC, K2P et HKY85.
24 taxa 96 taxa
sequence length 1 kb 3 kb 5 kb 3 kb
eucl – 4-letter word 17.8 16.3 16.4 20.5
eucl – 5-letter word 13.8 12.0 11.9 16.0
eucl – 6-letter word 12.9 10.7 10.6 14.9
χ2 – 4-letter word 17.6 16.4 16.4
χ2 – 5-letter word 14.3 12.1 12.0
χ2 – 6-letter word 14.4 11.4 10.9
Jukes-Cantor 11.1 6.3 5.2 9.3
Kimura 2-parameter 10.5 6.1 5.0 9.2
HKY 85 10.5 6.1 5.0 9.2
The methods are compared by their ability to infer the "true" tree, i.e. the topology of the tree that has been used to generate the sequences. We used the topological distance dT of Robinson-Foulds between the inferred tree and the true one. The bipartition distance of Robinson-Foulds [30] is equal to the number of bipartition present in one of the two trees and not in the other. The results are presented in term of percentage of misinferred branches. This percentage is equal to the topological distance divided by the maximum number of different bipartition between two trees: 2N-6 where N is the number of taxa.
In both methods, the Neighbor-Joining reconstruction algorithm was used. The differences in the results come principally from the choice of the distance. The Kimura two-parameter can be designed as the "true" distance, because the parameter of the distance are exactly the same as those chosen to generate the sequences. So normally the Kimura distance must be the branch length of the original trees. The fact that the results obtained by the distance method are not perfect can be attributed to the reconstruction algorithm Neighbor Joining (see Gascuel [37]). HKY85 is a model that includes the Kimura 2-parameter (K2P) model, so the result should be the same.
The proportion of wrong branches decreases in the signature method when word length increases (Table 1). At the same time, the longer the sequences, the better the results with the signature method. However, the proportion of correct branches obtained from the signature is not as high as for the distance method. As expected, the results of HKY85 are the same than those of Kimura 2-parameter. The results of the Jukes-Cantor model are similar to those of the signature for 1 k sequences. But for longer sequences, the signature method is less effective than the JC method. The result of K2P can be explained by the fact that the distance method uses exactly the model used to generate the data. This fact also explains why the results of the signature method improve less with the increase of the sequence length than those of the distance method. The fact that, for the moment, no evolutionary model can be design to the signatures limits the estimation of distances between the signatures. An improvement will be to find how the signature evolves with time as function of nucleotide substitution models. Increases in sequence length facilitate estimation of distance by conventional methods, because the substitution model is known. With the signature, 3 kb sequences are sufficient to obtain a representative signature of the species using 6 letter words. As a result, the increase in accuracy between 3 kb and 5 kb is not significant.
Despite the fact that no evolutionary model has been used with the signature, the results obtained from the signature method are reasonable. With 6-letter words, only 10 % of the internal branches are incorrect. It can be compared to the results presented by Gascuel [37]. The results of the signature method are not as good as the distance method, but they are nevertheless rather accurate. In general, the median size of genes is around 1 k. If we use longer sequences, it will be in the case of non-homologous sequences. For long sequences, no conventional method can be applied.
Vertebrate phylogeny
We used RAG1, a highly conserved gene that produces small distances between sequences to infer the vertebrate phenetic tree [38]. The analysis of the 46 sequences in the dataset had shown that four sequences were complete and the other contained only the conserved core, with length ranging from 1 kb for core sequences to 3 kb for complete ones. This large difference in length induced a bias in the signatures of the four complete sequences, and so in the obtained trees. For comparison with published works [38], we only used the conserved core of RAG1 gene.
A phylogenetic tree was inferred for 46 vertebrate sequences by maximum parsimony, distance (nucleic and protein sequences) and the signature method (Fig 4). Trees produced by classical and signature methods show that position of various vertebrate clades (birds, sharks, mammals, fishes, batrachians) is in agreement with paleontological data. The distance tree obtained using protein sequences exhibited some obvious errors: birds presented a stable group but were placed within mammals (data not shown). Moreover, the relationships between species within each taxonomic group are frequently incongruent with other data. The MP method leads to several most parsimonious trees that are summed up into a consensus tree. On the one hand, the major taxonomic groups can be recovered and are placed correctly; on the other hand, positions of species inside these groups are often poorly inferred (for instance, the relationships between mammals are unresolved).
Figure 4 Phylogeny of vertebrate species. Three methods were applied to the RAG1 gene from 46 species. Distance method: alignment with ClustalW, (Kimura 2-parameter distance), reconstruction by NJ algorithm. MP: use of same alignment. PAUP* has been used with default parameters. Signature method: 6-letter words – χ2 metric. The tree is inferred by NJ method. The bootstrap coefficients for distance and signature method are indicated.
In the signature tree, species are placed within classes in agreement with taxonomy. For example, in the signature analysis, the relationships within birds are congruent with conventional analysis [39]. With regard to mammals, the signature method is the only method that correctly recovers bats as a monophyletic group, with the exception of Felis catus. But the cat, Felis catus, is misplaced by every method, and so its incorrect placement cannot be attributed to a specific phylogenetic method. Mammal relationships appear much more problematic when analyzed by conventional phylogenetic methods than with the signature method. The polyphyly of tetrapods may be explained by the paucity of batrachian sequences, which could lead to an unreliable position for this clade. The monophyly of taxonomic classes, as well as relationships within each class appear quite robust as measured by bootstrap values.
To determine how strong the phylogenetic signal is present in the signature topology, a congruence analysis of phylogenetic trees [40] can be performed. The topologies obtained by ML, MP (the two best trees), NJ and signature (4- to 6-letter word for the Euclidean and the χ2 metrics) methods, are compared by determining the likelihood of each topology. We establish that the signature trees have a phylogenetic signal similar to the alignment-based ones. The signature trees with long words are more congruent than those using small words. The 6-letter word χ2 signature-tree is congruent with the ML tree and the congruence signature/ML is the same than the congruence NJ/ML (Table 2).
Table 2 Difference in log Likelihood. The differences are computed between the ML tree and the other trees.
Tree Δ-ln L
Maximum Likelihood best
Parsimony 9.38
Distance method 58.95
signature
χ2 – 4-letter 445.87
χ2 – 5-letter 297.8
χ2 – 6-letter 65.67
Mean random trees 9132.77
Plant phylogeny
This study, based on an article of Soltis et al. [41], used 18S rRNA for 93 plant species whose sequences are available from the "Green Plant Phylogeny Research Coordination Group" . The species can be grouped into nine main clades (Angiosperms (flowering plants), Conifers, Gnetales, Cycads (palm trees), Hornworts, Liverworts, Ferns, Mosses, Lycophytes), with some additional isolated species and an outgroup.
The signature tree presents significant similarities with the published tree [41]. The angiosperms, conifers, gnetales, cycads and ferns form stable monophyletic groups (high bootstrap coefficients (Fig 5)). The principal result of the article – that the angiosperms are at the root of conifers, gnetales, palm trees and ginkgo (Angiosperm + ((Cycad + Ginkgo) + (Conifern + Gnetale))) – are confirmed by our study and another molecular study [42]. This phylogenetic organization is original as Gnetales are more often linked to Angiosperms by morphological data [43-47] (see Doyle [48] for review).
Figure 5 Phylogenetic tree of plants obtained by comparison of 18S rRNA signatures. (6-letter words – χ2 metric). The bootstrap coefficients (500 sets) of principal groups are indicated. The species class names are indexed by a code: A – Angiosperm, C – Conifer, G – Gnetale, Cyca – Cycad, F – Fern, M – Moss, L – Lycophyte, Lw – Liverwort, Hw – Hornwort. (see annex for the correspondence code/species).
Recent analyses based on molecular data [49] confirms this result (Soltis [41] and Källersjö [49]). In addition, Equisetum and Psilotaceae are placed with the Ferns. This grouping is found in other studies [50,51] and these species are presented as sister group of Ferns. The sister group relationship of Psilotaceae and Ophioglossaceae is also found [52]. Contrary to the results obtained by Soltis, [41] the ferns are polyphyletic in the signature tree.
The outgroup separates the plants into two groups: the seed plants and the other land plants. To confirm the position of this outgroup, 18S rRNA sequences of Homo sapiens, Saccharomyces cerevisiae and Schizosaccharomyces pombe have been added (Data not shown). The outgroup is still confirmed as well as the tree split. This separation of land/flowering plants, the separation of the Lycophytes and the fact that the moss and liverwort do not form a monophyletic clade have been found also by Soltis when a NJ analysis was performed [41]. Thus, the signature method leads to a similar topology as the NJ method with alignment.
Multigene trees
Phylogenetic trees carry two types of signal: species evolution and gene evolution. For a variety of reasons, gene trees can be different from the tree of species from which they are sampled [53]. In addition, signals coming from different genes could lead to different inferred phylogenetic relationships between species [54].
In order to deal with this problem, several genes can be used to build a multigene tree [27,28]. The addition of signals coming from various genes can under some conditions reinforce the information on species evolution. In general, the alignment of each gene can be determined, and alignments concatenated prior to tree building. The signature has many properties that facilitate the calculation of multigene tree.
Another problem deals with the selection of genes participating into the multigene tree. In general, several steps of selection occur to eliminate horizontal transferred genes, duplications or those leading to aberrant phylogeny (see [27,28] for an example of these steps). Signatures are an ideal pretreatment tool for identifying horizontally transferred genes [55], and selecting those genes that conform to evolutionary relationships of the species under consideration. Moreover, due to the rapidity of the treatment with the signature, a very large number of genes can be tested at once.
We propose applying the signature method to infer a consensus tree of multiple genes. Two methods are possible. First, assuming that each gene brings the same quantity of information to the phylogeny for each species, an average signature is computed from several genes. The set of average signatures is then analyzed by the signature method. Another approach is to assume that each gene brings a quantity of phylogenetic information that is correlated with its length. In this approach, the sequences are concatenated and signatures are computed on the set of concatenated sequences.
To carry out this study, we used 33 genes originating from ten species (nine Bacteria: Bacillus subtilis, Clostridium perfringens, Escherichia coli, Lactococcus lactis, Neisseria meningitidis, Salmonella typhimurium, Staphylococcus aureus, Vibrio cholerae, Xanthomonas axonopodis and one Archaebacteria:Archaeoglobus fulgidus – see Material & Methods).
Because the signature does not rely on statements of homology at the level of individual nucleotides, it is possible to compare signatures from different genes in order to quantify statistical patterns and information content among genes. To determine the relative influence of gene evolution versus species evolution in shaping phylogenetic patterns, all the sequences involved in this study (393 sequences) were compared together by means of a hierarchical classification (Fig 6). The hierarchical classification is an unsupervised method allowing the detection of proximities between complex objects. The main result here is the grouping of gene signatures by species (Fig 6), and the species relationships present some differences with the consensus tree. These relationships are more in agreement with the known topology. V. cholerae, E. coli and S. typhimurium form a stable group, but inside this group, the signatures are grouped by genes (Fig 7). The signature of V. cholerae is very close to those of E. coli/S. typhimurium, as well as in the consensus distance matrix. We clearly face a problem of reconstruction of the Neighbor-Joining algorithm. For E. coli and S. typhimurium, the differentiation between these two species is quite recent and the homologous genes are very conserved. This leads to an alternate clustering of genes. In the Gram+, the C. perfringens signatures are very different to the other and place at the root of the Gram+. This confirms the species specificity of the signature, which was known to be present even in short DNA fragments [20]. The signatures of single genes conserve the characteristics of the species from which they are sampled.
Figure 6 Hierarchical classification of 393 6-letter word signatures. The signatures of a given species have the same color code. For each species group, the name of the species is indicated at left. The EF-Tu gene that also forms a stable group is also highlighted. Finally, arrows point out the horizontal transfer (HT) candidates that are discussed in this article.
Figure 7 Detailed view of the hierarchical classification of 393 6-letter word signatures. A detail focusing on the group with E. coli, S. Typhimurium and V. cholerae is shown. The symbols on the left of the names indicate the genes analyzed.
By contrast, an example where gene conservation is very strong is for EF-Tu gene; the signatures of nearly all the species are grouped together at the root of the V. cholerae/E. coli/S. typhimurium group. As it can be observed in the phylogenetic trees (signature and method of distance, results not shown), the A. fulgidus and C. perfringens copies of the gene are quite different, enough to their species signal to be stronger than the EF-Tu signal.
Some gene signatures cluster with species other than their own in the hierarchical tree. This could result from horizontal gene transfer. For instance, the phosphomannomutase gene of S. typhimurium is placed at the root of the S. aureus group. In the phosphomannomutase NJ tree and the signature tree, the relationships between the Gram- and the Gram+ bacteria are incongruent with other data and presumably wrong. Despite that, the other phosphomannomutase signatures are correctly assigned to their host species. The misplacement of this gene may indicate a horizontal transfer in S. typhimurium from an unknown donor. Two other potential horizontal transfers can be found deep inside species group: the elongation factor 2 signature of N. meningitidis and the ornithine carbamoyltransferase signature of S. aureus respectively inside the V. cholerae group and inside the C. perfringens group. In each case the signature is near the signature of the homologous gene of that species. So the gene signal is strong enough to displace the signature inside a different species group. To see if the original sequences are horizontal transfers, we examined two horizontal transfer databases: HGT-DB [56] and HGT Analysis Database [57]. In HGT-DB, the phosphomannomutase sequence of S. Typhimurium is tagged as horizontal transfer [56], but not the other two original sequences detected by the hierarchical classification. Thus our novel result suggests original sequences that need to be studied more precisely before being incorporated into a multigene study.
In all the methods, after removal of dubious genes the consensus tree separates the bacteria into the Gram+ and Gram- groups (Fig 8). But for individual genes this topology is seldom obtained. For Gram+ bacteria, the MP and signature methods lead to a (B. subtilis + (L lactis + (S. aureus + C perfringens))) grouping, but ML and distance methods place B. subtilis deep inside the Gram+ group. For Gram- bacteria, E. coli and S. typhimurium are always grouped and the majority of the methods (exception maximum of parsimony) place N meningitidis and X axonopodis together. The principal difference is the place of V. cholerae within the Gram-. The ML and MP trees place V. cholerae at the root of E. coli and S. typhimurium. The signature method places V. cholerae at the root of Gram- Bacteria.
Figure 8 Consensus trees for ten species. The four methods shown are the signature (6-letter words – χ2 metric) method, distance method, MP and ML. For each method except ML, the bootstrap coefficients (100 sets) are indicated.
To compare the result of the different studies and to determine the dispersion of the phylogenetic trees, we used the dissimilarity distance between the consensus tree and the whole set of gene trees for distance, MP, ML and the signature method (Fig 9) [32]. The distribution of dissimilarity distances indicates that the signature result is independent of the chosen gene and that each individual gene tree is similar to the consensus tree. In this latter case, the variations mainly arise from the placement of V. cholerae, either at the root of Gram- or E. coli/S. typhimurium clades By contrast, the distance method leads to variable results: no distance tree has a dT lower than 6 when compared to the consensus tree. To a lesser degree, the MP and ML trees exhibit a large dispersion (Table 3). Thus a single gene signature tree is less dissimilar from the consensus tree than a conventional one.
Figure 9 Dissimilarity distances between the consensus tree and the sets of genes retained. The dT distances have been computed for the method of distance, ML, MP and signature methods (6-letter word and χ2 metric).
Table 3 Statistical analysis of the distribution of dissimilarity distances as a function of method used.
Method Mean dT Standard deviation
distance 8.47 2.15
parsimony 5.37 2.98
maximum likelihood 5.65 3.28
signature 3 2.3
The consensus signature method allows us to analyze genes present only for some species. We added 9 genes to the data set (see Materials & Methods), thereby increasing the amount of sequence used to 65 kb per species. The signatures of these genes are amalgamated into the species signatures. The tree obtained (data not shown) exhibits the same topology as the consensus tree obtained with the whole set of genes per species computed previously.
The robustness of the consensus tree topology was assessed by computing 100 bootstrap trees. The bootstrap coefficient was 100% for all branches (Fig 8). Another way to test the robustness of the multigene tree is to vary the number of genes per species included in it, as in a jackknife procedure [58]. In this case, 30, 50, 75 and 90 % of the genes available per species are randomly selected. From the selected genes, an average signature is computed for each species. Distances between these average signatures are used to obtain a signature tree. This procedure is performed 100 times per percentage to yield a bootstrap tree. Results show that the topology of the consensus tree is always the same. However, in some cases the bootstrap coefficients are not maximal (table 4).
Table 4 Bootstrap values as function as the number of genes analyzed in the multigene study.
Percentage of used genes 30% 50% 75% 100%
Bootstrap coefficient 100 % except for two clades:
– (E. coli + S. typhimurium) = 91 %
– (N. meningitides + X. axonopodis) = 96 % 100% for all branches 100% for all branches 100% for all branches
In the individual phylogenetic trees, the variations in topologies are so important whatever the method used (except signature) (Fig 9, Table 3), that they do not allow us to confirm whether these sequences have in fact undergone a horizontal transfer.
Phylogeny of γ-proteobacteria
We have shown that using signatures and comparing non-homologous sequences such as are found in complete genomes made it possible to determine the relationship between species. To extend the results obtained with 10 prokaryotes genomes, we explore phylogenetic relationship of a well-studied taxonomic group: the γ-proteobacteria [28]. We selected 16 species whose complete genomes are available. These species can be classified in 6 taxonomic groups (Table 5). Pride et al. [15] used corrected signatures to infer phylogenetic trees. The signatures were corrected by zero order Markov model to normalize the base composition of the different species. Pride et al. [15] determined that this correction permits to obtain a signature tree the most congruent with the 16S rRNA tree. In order to compare the results to a reference, the 16S rRNA sequences have been used to infer a tree by the ML method (Fig 10A). A comparison of trees using signatures corrected and not corrected for base compositional biases is shown in Figures 10B and 10C.
Table 5 Species names and taxonomic groups of γ-proteobacteria.
Species name Taxonomic group
Shewanella oneidensis Alteromonadale
Buchnera aphidicola Enterobacteriale
Escherichia coli Enterobacteriale
Salmonella typhi Enterobacteriale
Salmonella typhimurium Enterobacteriale
Shigella flexneri Enterobacteriale
Yersinia pestis Enterobacteriale
Haemophilus influenzae Pasteurellale
Pasteurella multocida Pasteurellale
Pseudomonas aeruginosa Pseudomonaceae
Pseudomonas putida Pseudomonaceae
Vibrio cholerae Vibrionale
Vibrio vulnificus Vibrionale
Xanthomonas axonopodis Xanthomonadale
Xanthomonas campestris Xanthomonadale
Xylella fastidiosa Xanthomonadale
Figure 10 A- Tree of γ-proteobacteria obtained from the MP method for the 16S rRNA sequences. Each color corresponds to a taxonomic group. B- Tree of γ-proteobacteria obtained from non-corrected signatures (6-letter word signatures and City Block metric). Each color corresponds to a taxonomic group. C- Tree of γ-proteobacteria obtained from the signatures corrected by a zero order Markov model signatures (6-letter word signatures and City Block metric). Each color corresponds to a taxonomic group.
The 16S rRNA tree permits the establishment of reference relationships between the γ-proteobacteria. Some taxonomic groups are recovered: Xanthomonadales, Pseudonomaces as well as Pasteurelles. The tree groups Xanthomonadales and Pseudodomaces, and places B aphidicola close to Pasteurellale but with a long branch. This long branch can explain the incongruent placement of Pasteurelles in Enterobacteria for the ML tree (the phenomenon of long branch attraction [59]).
The tree calculated using the base compositionally-corrected signatures of complete genomes is more in agreement with the 16S rRNA tree. A group of Enterobacteria similar to that found in the 16S rRNA signature tree was obtained. However the monophyly of Xanthomonadales is not recovered in any of the trees obtained from signatures of complete genomes. X fastidiosa is placed at the root of the group (Xanthomonadale + Pseudomaceae). Another difference between trees for complete genomes and those of the 16S RNA is a grouping of Pasteurellales, Vibrionales and S. oneidensis found in the signature tree.
In the complete genome trees, B aphidicola appears misplaced. It is always positioned apart from the Enterobacterial clade, despite its belonging to this group taxonomically. An analysis of genome signatures of B aphidicola revealed that this species exhibits a very different signature from those of the other γ-proteobacteria (result not shown). This result is not due to a bias in signature method arising from the size of B aphidicola genome, because a tree obtained by randomly selecting the same sequence length in the 15 other genomes (650 kb) leads to the same topology (result not shown). We suggest that the source of this anomoly is that B aphidicola is a symbiotic bacteria, andhas a very small genome (650 KB) compared with those of the other γ-proteobacteria (4 to 5 Mb). This genome reduction arises from its parasitic lifestyle and is the result of many independent losses of genes and genomic segments. B aphidicola experienced very strong evolutionary pressures that led to a profound shift in its signature, and also transferred numerous genes to its host [60]. Such symbiotic species are known to be difficult to place phylogenetically [61]. B. aphidicola also has a strong compositional bias (the genome of is nearly 75% AT rich). The other γ-proteobacteria are more GC rich. These problems appear when using whole genomes to infer a tree and are bypassed when using conserved genes or a selection of genes sharing a common history [28].
We used the method of Dufraigne et al. [55] to detect in the B aphidicola genome sequences that may have arisen by horizontal transfer. We divided the entire genome into 5 kb sequence windows. For each window, a 4-letter word signature was computed. The method developed by Dufraigne et al. allows us to detect which sequences have original signatures such as would be found in cases of horizontal tranfer. We removed this original sequences from the genome and a new 6-letter word signature was computed. The tree obtained is exactly the same as the base compositionally-corrected tree (Fig 10C).
Conclusion
In this paper we have illustrated the exploration of phylogenetic data with a global sequence analysis method, the signature method. Using a variety of genes, this method yields tree topologies similar to those obtained using traditional phylogenetic approaches. The results presented here suggest that trees obtained by this method could be used as an exploratory step in phylogenetic studies. The signature method can deliver a quick overview of phylogenetic relationships between species in data sets that can be challenging or time consuming for traditional alignment and phylogenetic analysis. As our simulations showed, the signature method sometimes yields phylogenies that are less accurate than those produced by conventional analyses, but this arises mainly from the fact that no evolutionary model is known for word frequencies comprising genomic signatures. The signature tree can be used as fast pretreatment in conjunction with classical methods such as ML. We also demonstrated that the signature distances are tree-like, reflect tree distances and that in the case of short sequences such as frequently assembled in studies of homologous sequences, the optimal word length seems to be 6. This length represents a trade-off between long words that represent more accurately the DNA sequences [21,25] and the size of the sequences.
The signature method is particularly useful as a first step in data exploration. The speed of the analysis permits detection of either misplacement of particular species, in some cases due to local composition fluctuations (horizontal transfer), or unexpected groupings of species that can be scrutinized further by biological means or conventional phylogenetic study. Thus, the signature method easily permits the researcher to use long and/or numerous genes in a study. When using numerous species, their phylogenetic proximities can be analyzed using their signatures by conventional statistical methods and the set of species split into subgroups. This method is also useful in combining information from different genes. The signature method permits the averaging of a great number of genes of any length to obtain a consensus and a unique signature per species and thereby take into account a great number of evolutionary events. The signature method does not rely on homology of DNA sites to compare sequences and it is possible to compare non-homologous sequences to infer a phylogenetic tree. Thus, many genes not present in every species can be added to this tree, giving more confidence in the species tree. This approach was already applied to birds [24], bacterial [15-18,21,62] or mitochondrial [25] phylogenetic studies. In contrast to conventional methods, the signature method utilizes information present in the sequences that may not be analyzable with conventional alignments, such as additional sequences at the beginning or the end of alignments.
For studies of complete genomes, detection of horizontal transfer using signatures, such as proposed by Dufraigne et al. [55], permits removal of sequences that will compromise phylogenetic analysis. Finally, signatures allow the rapid detection of horizontally transferred genes or simply misplaced genes that require additional attention via hierarchical clustering or other statistical classification methods.
Methods
Sequence signature
Sequence signature can be computed easily and very quickly thanks to an algorithm -the "Chaos game representation" (CGR)-, (about 1 Mb per second on a laptop computer) [63]. The signature can be displayed as an image, where each pixel represents a word and the darkness of the pixel increases with the frequency of the word in the sequence.
DNA sequences
We selected two genes to compare signature analysis of two different clades with results from the literature. These genes are long enough to get a significant signature and address the phylogeny of vertebrates and plants including a large number of species. The recombination activation gene RAG1 is used for inferring the phenetic tree of 46 species of vertebrates. Ribosomal RNA sequence analysis is the de facto standard for phylogenetic reconstruction. Here we use ribosomal 18S RNA to analyze 93 plant species. Finally, 42 genes, accounting for more than 50 kb of sequence, are used for a multigene study (see Annex), including nine Bacteria and one Archaea. To select the 42 genes, we utilized the SYSTERS database [64]. For all the selected species, the database returned 119 orthologous protein families shared by the whole set of species. These families were filtered by size of the corresponding DNA sequences (retained families contain sequences with mean lengths > 1 kb). From these, 33 complete sets and 9 partial sets of genes were obtained. The selected genes belong mainly to amino acid, nucleotide and protein synthesis and DNA metabolism families. All the sequences were extracted from GenBank or Genome Information Broker [65]. The complete genomes of 16 γ-proteobacteria were gathered from GenBank (see appendix). Simulated sequences from a known phylogeny were found on Gascuel's website [35].
Phylogenetic analysis and signature method
Two distance metrics (Euclidean and ?2) were used to quantify the differences between signatures. Other metrics (Manhattan, Mahalanobis, Correlation and Cosine) were investigated as well; these methods rarely performed better than our two focal methods, and often performed worse, so we do not consider them further. Distance matrices were obtained via the Euclidean and ?2 metrics. We used these matrices to infer trees with the Neighbor-Joining (NJ) reconstruction algorithm implemented in the PHYLIP package [66]. In order to estimate the robustness of the tree topology, we simulated by bootstrap [67] a whole new set of signatures from the initial set of motif frequencies, sampling with replacement (in general, 100 bootstrap trees were computed). Each dataset contains the same individuals from the initial data and N new variables (words) randomly drawn in order to replace the N variables from the initial set [67]. For each set of sequences, the phylogenetic tree was inferred and a consensus tree was calculated from each bootstrap replicate. Besides the signature method, three commonly used methods [3] were used to analyze aligned sequences from the same data sets: Neighbor-Joining (NJ) [68], maximum parsimony (MP) [69] and maximum of likelihood (ML) [2]. All three methods were implemented using the PAUP* [70] and PHYLIP packages. Alignments were obtained with ClustalW (default parameters)[4] and were similar to those used in their respective sources. For the different conventional methods, we have used the HKY85 model of sequence evolution, and gaps were treated as missing data in the MP analysis. For ML analyses, a gamma distribution of rate heterogeneity with simultaneous parameter estimation was used.
Appendix
Species annotation for the 18S rRNA sequences of plants
A1: Asarum canadense; A2: Sparganium eurycarpum; A3: Tetracentron sinense; A4: Trochodendron aralioides; A5: Austrobaileya scandens; A6: Sassafras albidum; A7: Akebia quinata; A8: Amborella trichopoda; A9: Camptotheca acuminata; A10: Gossypium hirsutum; A11: Celtis yunnanensis; A12: Canna coccinea; A13: Ceratophyllum demersum; A14: Dipsacus sp; A15: Liquidambar styraciflua; A16: Zea mays; A17: Nymphaea tuberosa; A18: Oncidium excavatum; A19: Phytolacca americana; A20: Pisum sativum; A21: Symphoricarpos albus; A22: Saururus cernuus; A23: Saxifraga integrifolia; A24: Saruma henryi; C1: Araucaria excelsa; C2: Cephalotaxus wilsoniana; C3: Juniperus chinensis; C4: Phyllocladus trichomonoides; C5: Pinus elliottii; C6: Pinus luchuensis; C7: Dacrycarpus imbricatus; C8: Amentotaxus formosana; C9: Torreya nucifera; C10: Taiwania cryptomerioides; C11: Podocarpus costalis; C12: Nageia nagi; C13: Taxus chinensis var. mairei; C14: Abies lasiocarpa; Cyca1: Cycas taitungensis; Cyca2: Zamia pumila; Equisetum: Equisetum hyemale; F1: Adiantum raddianum; F2: Blechnum occidentale; F3: Dicksonia antarctica; F4: Dicranopteris linearis; F5: Hypolepis muelleri; F6: Lonchitis hirsuta; F7: Osmunda cinnamomea; F8: Odontosoria chinensis; F9: Ophioglossum petiolatum; F10: Pteridium aquilinum; F11: Salvinia natans; F12: Vandenboschia davallioides; G1: Welwitschia mirabilis; G2: Ephedra sinica; G3: Ephedra torreyana; G4: Gnetum nodiflorum; G5: Gnetum urens; G6: Gnetum gnemon; Ginkgo: Ginkgo biloba; Hw1: Anthoceros agrestis; Hw2: Notothylas breutelii; Hw3: Phaeoceros laevis; L1: Huperzia lucidula; L2: Isoetes durieui; L3: Isoetes engelmannii; L4: Lycopodiella inundata; L5: Huperzia phlegmaria; L6: Huperzia taxifolia; L7: Lycopodium tristachyum; L8: Selaginella umbrosa; L9: Selaginella vogelii; Lw1: Marchantia polymorpha; Lw2: Fossombronia pusilla; Lw3: Pellia epiphylla; Lw4: Reboulia hemisphaerica; Lw5: Sphaerocarpos donnelli; Lw6: Scapania nemorea; Lw7: Riccardia pinguis; M1: Physcomitrella patens; M2: Atrichum undulatum; M3: Eurhynchium hians; M4: Funaria hygrometrica; M5: Leptobryum pyriforme; M6: Polytrichum formosum; M7: Physcomitrium pyriforme; M8: Sphagnum cuspidatum; O1a: Chara australis; O1b: Chara connivens; O1c: Chara foetida; O2a: Nitella flexilis; O2b: Nitella sp; Psilo1: Psilotum nudum; Psilo2: Tmesipteris tannensi.
Table 6 Genes used in multigene study:
1/ whole set of species:
Adenylosuccinate lyase 1.3 kb
Adenylosuccinate synthetase 1.3 kb
Alanyl-tRNA synthetase 2.6 kb
Argininosuccinate synthase 1.3 kb
Argininosuccinate lyase 1.4 kb
Arginyl-tRNA synthetase 1.7 kb
Aspartate aminotransferase 1.2 kb
Aspartyl-tRNA synthetase 1.8 kb
Carbamyl-phosphate synthase 3.2 kb
Cell division protein ftsZ 1.2 kb
Chorismate synthase 1.1 kb
CTP synthase 1.6 kb
DNA-directed RNA polymerase 2.6 kb
DNA topoisomerase I 2.0 kb
Elongation factor 2 2.1 kb
Enolase 1.3 kb
5-enolpyruvylshikimate-3-phosphate synthetase 1.3 kb
Glutamine synthetase 1.5 kb
Leucyl-tRNA synthetase 2.8 kb
Methionyl-tRNA synthetase 2.1 kb
Ornithine carbamoyltransferase 1.0 kb
Pantothenate metabolism flavoprotein 1.2 kb
D-3-phosphoglycerate dehydrogenase 1.2 kb
Phosphoglycerate kinase 1.2 kb
Phosphomannomutase 1.3 kb
Phosphoribosylformylglycinamidine synthase II 3.8 kb
Queuine tRNA-ribosyltransferase 1.1 kb
Ribonucleotide reductase 2.3 kb
Serine hydroxymethyltransferase 1.2 kb
Thermosome alpha subunit 1.6 kb
Threonyl-tRNA synthetase 2.0 kb
Translation elongation factor EF-Tu 1.3 kb
Valyl-tRNA synthetase 2.6 kb
Total length = 57.2 kb
2/ partial set of species:
Acetolactate synthase large subunit 1.7 kb
Cysteinyl-tRNA synthetase 1.4 kb
Galactosyltransferase 1.1 kb
GTP cyclohydrolase II 1.1 kb
Histidine kinase 2.0 kb
Phosphoenolpyruvate synthase 2.4 kb
dTDP-glucose 4,6-dehydratase 1.1 kb
Tryptophan synthase subunit beta 1.2 kb
X-pro aminopeptidase 1.3 kb
Authors' contributions
CC and DP conceived the study, drew the figures and wrote the first draft of the manuscript. CC was the main contributor of the bioinformatic analysis. CD participated in the bioinformatic study. SE participated in the method design and drafted the manuscript. AG and BF drafted the manuscript. All authors read and approved the final manuscript.
Acknowledgements
This research was supported by contract 120910 from the "Action inter EPST Bio-informatique 2001" of the French Research Ministry and contract A02114DS from the "Action inter EPST Bio-informatique 2003" of the French Research Ministry.
==== Refs
Lecointre G Le Guyader H Classification phylogénétique du vivant 2001 Paris, Belin 544
Felsenstein J Evolutionary trees from DNA sequences: a maximum likelihood approach J Mol Evol 1981 17 368 376 7288891
Li WH Sinauer A Molecular Evolution 1997 Sinauer , Sinauer 487
Higgins DG Thompson JD Gibson TJ Using CLUSTAL for multiple sequence alignments Methods Enzymol 1996 266 383 402 8743695
Brocchieri L Phylogenetic inferences from molecular sequences: review and critique Theor Popul Biol 2001 59 27 40 11243926
Russo CAM Takezaki N Nei M Efficiencies of different genes and different tree-building methods in recovering a known vertebrate phylogeny Mol Biol Evol 1996 13 525 536 8742641
Billoud B Guerrucci MA Masselot M Deutsch JS Cirripede phylogeny using a novel approach: molecular morphometrics Mol Biol Evol 2000 17 1435 1445 11018151
Wolf YI Rogozin IB Grishin NV Tatusov RL Koonin EV Genome trees constructed using five different approaches suggest new major bacterial clades BMC Evol Biol 2001 1 8 11734060
Korbel JO Snel B Huynen MA Bork P SHOT: a web server for the construction of genome phylogenies Trends Genet 2002 18 158 162 11858840
Tekaia F Lazcano A Dujon B The genomic tree as revealed from whole proteome comparisons Genome Res 1999 9 550 557 10400922
Daubin V Gouy M Perrière G A phylogenetic approach to bacterial phylogeny: evidence of a core of genes sharing a common history. Genome Research 2002 12 1080 1090 12097345
Gupta RS Protein phylogenies and signature sequences: a reappraisal of evolutionary relationships among Archaebacteria, Eubacteria and Eukaryotes. Microbiol Mol Biol Rev 1998 62 1435 1491 9841678
Rivera MC Lake JA Evidence that eukaryotes and eocyte prokaryotes are immediate relatives Science 1992 257 74 76 1621096
Karlin S Burge C Dinucleotide relative abundance extremes: a genomic signature Trends In Genetics 1995 11 283 290 7482779
Pride DT Meinersmann RJ Wassenaar TM Blaser MJ Evolutionary implications of microbial genome tetranucleotide frequency biases Genome Res 2003 13 145 158 12566393
Yu ZG Jiang P Distance, correlation and mutual information among portraits of organisms based on complete genomes Phys Lett A 2001 286 34 46
Qi J Wang B Hao BI Whole proteome prokaryote phylogeny without sequence alignment: a K-string composition approach J Mol Evol 2004 58 1 11 14743310
Stuart GW Moffett K Leader JJ A comprehensive vertebrate phylogeny using vector representations of protein sequences from whole genomes Mol Biol Evol 2002 19 554 562 11919297
Deschavanne PJ Giron A Vilain J Fagot G Fertil B Genomic signature: characterization and classification of species assessed by Chaos Game Representation of sequences Molecular Biology and Evolution 1999 16 1391–1399 10563018
Deschavanne P Giron A Vilain J Dufraigne C Fertil B Genomic signature is preserved in short DNA fragments BIBE2000 IEEE international Symposium on bio-informatics & biomedical engineering, Washington USA, 8-10 november 2000 2000 161 167
Chapus C Fertil B Edwars S Giron A Deschavanne P Spang R, Béziat P and Vingron M Classification of species based on DNA style: ; Berlin, Germany. 2003 147 148
Karlin S Bacterial DNA strand compositional asymmetry Trends Microbiol 1999 7 305 308 10431198
Karlin S Mràzek J Campbell AM Compositional biases of bacterial genomes and evolutionary implications J Bact 1997 179 3899 3913 9190805
Edwards SV Fertil B Giron A Deschavanne PJ A genomic schism in birds revealed by phylogenetic analysis of DNA strings Syst Biol 2002 51 599 613 12228002
Wang Y Hill K Singh S Kari L The spectrum of genomic signatures: from dinucleotides to chaos game representation Gene 2005 346 173 185 15716010
Guénoche A Garreta H Gascuel O and Sagot MF Can We Have Confidence in a Tree Representation?: 2001; Montpelier, France. Lecture Notes in Computer Science Goos G, Hartmanis J, van Leeuwen J 2000 2066 Springer 45 56
Daubin V Gouy M Perriere G A phylogenomic approach to bacterial phylogeny: evidence of a core of genes sharing a common history Genome Res 2002 12 1080 1090 12097345
Lerat E Daubin V Moran NA From gene trees to organismal phylogeny in prokaryotes: the case of the g-proteobacteria PLOS Biology 2003 1 101 109
Buneman P Hobson FR, Kendall DG and Tautu P Mathematics in the Archeological and Historial Sciences 1971 Edinburgh, University Press 387
Robinson DF Foulds LR Comparaison of phylogenetic trees Math Biosci 1981 53 131 147
Kuhner MK Felsenstein J A simulation comparison of phylogeny algorithms under equal and unequal evolutionary rates Mol Biol Evol 1994 11 459 468 8015439
Penny D Hendy MD The use of tree comparaison metrics Systematic Zoology 1985 34 75 82
Kumar S A stepwise algorithm for finding minimum evolution trees Mol Biol Evol 1996 13 584 593 8882501
Ranwez V Gascuel O Quartet-based phylogenetic inference: improvements and limits Mol Biol Evol 2001 18 1103 1116 11371598
Gascuel O http://www.lirmm.fr/~w3ifa/MAAS/US-MAAS.html
Rambaut A Grassly NC Seq-Gen: an application for the Monte Carlo simulation of DNA sequence evolution along phylogenetic trees Comput Appl Biosci 1997 13 235 238 9183526
Gascuel O Baxevanis A, Davison D, Page R, Stein L, Stormo G and Petsko G Getting a Tree Fast: Neighbor Joining and Distance Based Methods Current Protocols in Bioinformatics Current Protocols & Sons Wiley 2004 , Wiley & Sons 6.3.1 6.3.18
Greenhalgh P Steiner LA Recombination activating gene 1 (Rag1) in zebrafish and shark Immunogenetics 1995 41 54 55 7806278
Sibley CG Ahlquist JA Phylogeny and classification of birds: a study in molecular evolution 1990 New Haven, Yale University Press
Feil EJ Holmes EC Bessen DE Chan MS Day NP Enright MC Goldstein R Hood DW Kalia A Moore CE Zhou J Spratt BG Recombination within natural populations of pathogenic bacteria: short-term empirical estimates and long-term phylogenetic consequences Proc Natl Acad Sci U S A 2001 98 182 187 11136255
Soltis PS Soltis DE Wolf PG Nickrent DL Chaw SM Chapman RL The phylogeny of land plants inferred from 18S rDNA sequences: pushing the limits of rDNA signal? Mol Biol Evol 1999 16 1774 1784 10605118
Chaw SM Parkinson CL Cheng Y Vincent TM Palmer JD Seed plant phylogeny inferred from all three plant genomes: monophyly of extant gymnosperms and origin of Gnetales from conifers Proc Natl Acad Sci U S A 2000 97 4086 4091 10760277
Crane PR Phylogenetic analysis of seed plants and the origin of the angiosperms Ann Mo Bot Gard 1985 72 716 793
Doyle JA Seed plant phylogeny and the relationshipsof Gnetales Int J Plant Sci 1996 157 S3 S39
Doyle JA Molecules, morphology, fossils, and the relationship of angiosperms and Gnetales Mol Phylogenet Evol 1998 9 448 462 9667993
Doyle JA Donoghue MJ Seed plant phylogeny and the origin of the angiosperms: an experimental cladistic approach Bot Rev 1986 52 321 431
Doyle JA Donoghue MJ Zimmer EA Integration of morphological and ribosomal RNA data on the origin of the angiosperms Ann Mo Bot Gard 1994 81 419 450
Doyle JA Phylogeny of vascular plants Annu Rev Ecol Syst Biol 1998 29 567–569
Källersjö M Farris JS Chase W Bremer B Fay MF Humpries CJ Petersen G Seberg O Bremer K Simultaneous parsimony jackknife analysis of 2538 rbcl DNA sequences reveals upport for major clades of green plants, land plants, seed plants and flowering plants. Plant Syst Evol 1998 213 2599 2287
Kenrick P Crane PR The origin and early evolution of plants on land. nature 1997 389 33 39
Manhart JR Phylogenetic analysis of green plant rbcL sequences Mol Phylogenet Evol 1994 3 114 127 8075831
Manhart JR Chloroplast 16S rDNA sequences and phylogeneic relationships of ferns and fern allies Am Fern J 1995 85 182 192
Rokas A Williams BL King N Carroll SB Genome-scale approaches to resolving incongruence in molecular phylogenies Nature 2003 425 798 804 14574403
Gray MW Burger G Lang BF Mitochondrial evolution Science 1999 283 1476 1481 10066161
Dufraigne C Fertil B Lespinats S Giron A Deschavanne P Detection and characterization of horizontal transfers in prokaryotes using genomic signature Nucleic Acids Res 2005 33 e6 15653627
Garcia-Vallve S Guzman E Montero MA Romeu A HGT-DB: a database of putative horizontally transferred genes in prokaryotic complete genomes Nucleic Acids Research 2003 31 187 189 12519978
Nakamura Y Itoh T Matsuda H Gojobori T Biased biological functions of horizontally transferred genes in prokaryotic genomes Nature Genetics 2004 36 760 766 15208628
Lanyon SM Detecting internal inconsistencies in distance data Systematic Zoology 1985 34 397 403
Felsenstein J Cases in wich parsimony or compability methods will be positively misleading. Systematic Zoology 1978 27 401 410
Moran NA Mira A The process of genome shrinkage in the obligate symbiont Buchnera aphidicola Genome Biol 2001 2 RESEARCH0054 11790257
Itoh T Martin W Nei M Acceleration of genomic evolution caused by enhanced mutation rate in endocellular symbionts Proc Natl Acad Sci U S A 2002 99 12944 12948 12235368
Yu ZG Anh V Lau KS Measure representation and multifractal analysis of complete genomes Phys Rev E 2001 64 1 9
Jeffrey HJ Chaos game representation of gene structure Nucleic Acids Research 1990 18 2163 2170 2336393
Krause A Stoye J Vingron M The SYSTERS protein sequence cluster set Nucleic Acids Res 2000 28 270 272 10592244
GIB http://gib.genes.nig.ac.jp/
Felsenstein J PHYLIP (Phylogeny Inference Package) 3.6 2004
Felsenstein J Confidence limits on phylogenies : an approach using the bootstrap Evolution 1985 4 783 791
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Fitch WM On the problem of discovery the most parsimonious tree Am Nat 1977 111 223 257
Swofford DL PAUP*, Phylogenetic Analysis Using Parsimony (*and Other Methods) 2003 4 , Sinauer Associates, Sunderland, Massachusetts
|
0
|
PMC1310608
|
CC BY
|
2021-01-04 16:33:02
|
no
|
BMC Psychiatry. 2005 Nov 30; 5:44
|
latin-1
|
BMC Psychiatry
| 2,005 |
10.1186/1471-244X-5-44
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1221630768510.1186/1471-2458-5-122Research ArticlePatient and health service delay in pulmonary tuberculosis patients attending a referral hospital: a cross-sectional study Kiwuwa Mpungu S [email protected] Karamagi [email protected] Mayanja Kizza [email protected] Clinical Epidemiology Unit, Makerere University, Faculty of Medicine, P.O Box 7072, Kampala, Uganda2 Department of Medicine, Makerere University, Faculty of Medicine, P.O Box 7072, Kampala, Uganda2005 24 11 2005 5 122 122 10 2 2005 24 11 2005 Copyright © 2005 Kiwuwa et al; licensee BioMed Central Ltd.2005Kiwuwa et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Delays in diagnosis and initiation of effective treatment increase morbidity and mortality from tuberculosis as well as the risk of transmission in the community. The aim of this study was to determine the time taken for patients later confirmed as having TB to present with symptoms to the first health provider (patient delay) and the time taken between the first health care visit and initiation of tuberculosis treatment (health service delay). Factors relating to these 'delays' were analyzed.
Methods
A cross-sectional survey, of 231 newly diagnosed smear-positive tuberculosis patients was conducted in Mulago National referral Hospital Kampala, from January to May 2002. Socio-demographic, lifestyle and health seeking factors were evaluated for their association with patient delay (>2 weeks) and health service delay (>4 weeks), using odds ratios with 95% confidence intervals (CI) including multivariate logistic regression.
Results
The median total delay to treatment initiation was 12 weeks. Patients often presented to drug shops or pharmacies (39.4%) and private clinics (36.8%) more commonly than government health units (14%) as initial contacts. Several independent predictors of 'patient delay' were identified: being hospitalized (odds ratio [0R] = 0.32; 95% CI: 0.12–0.80), daily alcohol consumption (OR = 3.7; CI: 1.57–9.76), subsistence farming (OR = 4.70; CI: 1.67–13.22), and perception of smoking as a cause of TB (OR = 5.54; CI: 2.26–13.58). Independent predictors of 'health service delay' were: >2 health seeking encounters per month (OR = 2.74; CI: 1.10–6.83), and medical expenditure on TB related symptoms >29 US dollars (OR = 3.88; CI: 1.19–12.62). Perceived TB stigma and education status was not associated with either form of delay.
Conclusion
Delay in diagnosis of TB is prolonged at the referral centre with a significant proportion of Health service delay. More specific and effective health education of the general public on tuberculosis and seeking of appropriate medical consultation is likely to improve case detection. Certain specific groups require further attention. Alcoholics and subsistence farmers should be targeted to improve accessibility to TB treatment. Continuing medical education about TB management procedures for health providers and improvement in the capacity of TB control services should be undertaken.
pulmonary tuberculosishealth-seeking behaviortreatment delay
==== Body
Background
Uganda is one of the global TB high burden countries ranked 20th in the world with an estimated incidence rate of tuberculosis in 2001 of 324 cases per 100,000 and a notification rate of 171 per 100,000 of the population. The estimated proportion of tuberculosis cases that were HIV positive was 35% in 2001. The number of reported TB cases is increasing with 19,016 cases registered in 1991 and 41,831 in 2001[1]. This increase is possibly due to population growth, improvement in case finding or a true increase, as sentinel sero-prevalence surveys in Ugandan hospitals had shown that 45% to 65% of newly diagnosed TB patients were HIV positive[2]. The detection of new sputum smear positive cases was 52%, which is far below the (WHO) case detection target of 70%[3].
In Mulago National Referral TB treatment center located in Kampala, the capital city of Uganda, clinic records show that the majority of tuberculosis patients have had symptoms for two or more months before seeking TB treatment. Delays in diagnosis and initiation of effective treatment increase morbidity and mortality from tuberculosis as well as the risk of transmission in the community. The stigma attached to TB disease may hinder access to timely health care[4].
In patients with TB, the low detection and cure rates may be attributed to limited information about factors contributing to delayed diagnosis in Uganda. Recent work in Mukono [5], a mainly rural district revealed that there is considerable delay in diagnosis of TB. However factors contributing to delay in diagnosis and treatment are likely to vary depending on the populations in their local settings. The aim of this study was to determine the time interval between symptom onset and initiation of TB treatment[6,7] and analyze the determinants of this interval. From this study, it was determined which factors serve as barriers or facilitators for seeking treatment early, how these factors are interrelated and which ones are liable to behavioral change or social encouragement by health education and health promotion.
Methods
Ethical considerations
Institutional consent was obtained from the Faculty Research and Ethics Committee of the Medical School (institutional ethical board for the medical school).
Study setting
The study was carried out at the National Tuberculosis and Leprosy control program (NTLP) center of Mulago hospital situated in the city centre. This treatment center is the principal facility providing in-patient and outpatient TB care in Kampala district. The TB center which is one of Mulago hospital's various departments has an inpatient bed capacity of about 100 beds. The hospital provides services to Kampala's population of about 1,208,544 (Uganda population census 2001) and surrounding districts plus upcountry referrals. TB patients are usually referred to the center from other hospitals, medical centers and private practitioners in and outside Kampala. Most health care units are strategically located within the city, and the transport system has good coverage. The TB control program initiated decentralization of TB management services to the district and sub-district levels, in 1997. The total number of new TB patients including children registered at the center in 2001 was about 4000.
Study participant selection
We performed a cross-sectional study during a four-month period from January to May 2002 during which a total of 1,382 TB patients were registered at the treatment center. Eligible participants were patients presenting to the center aged 18 years or older, with newly diagnosed sputum smear positive pulmonary tuberculosis. Consenting eligible patients were consecutively enrolled and interviewed by one of three trained research assistants including the principal author, within two days of the pulmonary tuberculosis diagnosis, using a semi-structured questionnaire.
For hospitalized patients, the TB register at the treatment centre was used to identify eligible patients. Such patients were then located on the ward for consent to participate in the study. Most interviews were conducted in the main local language of the area (Luganda), or in English. Re-treatment patients and those with sputum negative disease were excluded from the study. The questionnaire was pre-tested among TB patients prior to the start of study, with modifications incorporated in the final version. The questionnaire collected information on the following attributes including: (a) Demographic and socio-economic variables: age, sex, occupation, education level, and family size; (b) Physical factors: symptoms and their duration, sputum results and chest radiographic findings); (c) Psychological factors: patient beliefs, perceptions, attitudes and knowledge about tuberculosis; (d) Institutional or health service factors: distance to health facilities, costs of travel and medical expenditure on treatment of TB related symptoms. Each interview schedule lasted approximately 40 minutes per participant and allowed for careful probing of responses to minimize recall bias. Other data sources were case notes and referral letters. A chest physician interpreted chest radiographic findings and regular supervision of the interviewers was conducted throughout the study period.
Participants were asked to estimate the time in months or weeks they had been experiencing the major presenting symptoms – evening fevers, night sweats, chest pain, difficulty in breathing, weight loss, loss of appetite or generalized weakness along with the cough episode leading to initiation of tuberculosis treatment (total delay). Patient delay was the estimated time interval between symptom onset and the first medical consultation and the time taken from then until the diagnosis was confirmed and treatment started was considered as health service delay[6,7]. In our study a health provider was defined as any person consulted by the patient about his/her sickness who prescribed any form of medication. These included dispensers, pharmacists, medical staff and herbalists or traditional healers. The estimated sample size of 231 patients was obtained assuming, from prior local studies that 80% would have a total delay of more than 4 weeks [8]. A total delay of more than four weeks was considered, as national guidelines suggest that patients coughing for more than 3 weeks should be investigated further. Limitations of our study include imprecise estimates of delay due to recall bias or in some cases, lack of case notes and referral letters to verify the accuracy of information given, and the single center studied.
Data management and analysis
In order to ensure data quality, the data was double entered into a computer, and the two copies of the data verified. Delay in weeks was presented as medians and proportions. Patient delay (classified as presentation after 2 weeks of the onset of symptoms) and health service delay (when patient was delayed >4 weeks between initial contact with health provider and treatment onset) were compared for different sub-groups using odds ratios and 95% CI. To identify factors independently associated with patient delay and health service delay, a multivariate logistic regression analysis with delay time dichotomized as described above was performed. When appropriate, non-parametric tests (Chi-square, Mann-Whitney and Kruskal-Wallis tests) were used where assumptions of normality were unmet. Statistical significance was taken as P < 0.05.
Results
Socio-demographic characteristics of the participants
Two hundred thirty one eligible adult newly diagnosed smear positive TB patients were recruited into the study. The mean age was 30.7{Standard deviation (SD) = 9.4}, with a median age of 30 years. The mean age of the males 32.2 (SD = 9.2) years and females 28.7 (SD = 9.2) years were significantly different (p < 0.05). Seventy-two (25.5%), were hospitalized at the time of diagnosis. The mean family size was 4(SD = 2.4) with a range of (1–11). One hundred and thirty-nine had at most post-primary education regardless of gender. Occupation wise 28 (12.1%) were subsistence farmers, 56 (24.2%) skilled workers and 115 (49.8%) unskilled workers including traders and housewives. Of those who did not earn any income, 15 (11.4%) were more likely to be male compared to 36 (36.4%) of the females (P < 0.001). The average medical expenditure on treatment of TB related symptoms was 30 US dollars range (0–412) (Table 1).
Table 1 Socio-demographic characteristics and distribution of the various time delays among the participants*
Characteristic Patients, No. (%)
Age, mean (SD), y 30.7 (9.4)
Sex
Male 132 (57.1)
Female 99 (42.9)
Hospital admission status
Hospitalised 72 (25.5)
Outpatients 159 (74.5)
Highest education level attained
None 21 (9.1)
Primary school 118 (51.1)
Secondary 58 (25.1)
Post-secondary 13 (5.6)
Tertiary 21 (9.1)
No. of persons in Household
(1) 24 (10.4)
(>1) 207 (89.6)
Weekly Alcohol consumption
Daily 35 (15.2)
Sometimes 47 (20.3)
Rarely 26 (11.3)
Never 123 (53.2)
Main Occupation
Subsistence farming 28 (12.1)
Unskilled workers 115 (49.8)
Skilled workers 56 (24.2)
Students 20 (8.7)
Unemployed 12 (5.2)
Medical expenditure on TB related symptoms, US $
0–29 123 (53.2)
>29 108 (46.8)
Distance from home to treatment centre, km
<25 197 (85.3)
>25 34 (14.7)
Distribution of the time delays
Patient delay Health service delay Total delay
Median, wk 1.0 9.0 12.0
Interquartile Range, wk 0.9–4.0 5.3–15.3 8.0–20.0
Delay >2 wk 70 (30.3) 207 (89.6) 231 (100)
Delay >4 wk 40 (14.7) 188 (81.4) 210 (90.9)
*Values are number % unless otherwise indicated
Clinical characteristics of the participants at presentation
The mean performance status (Karnofsky score) which is a measure of the degree of debility, of the respondents was 80 (SD = 10). Of our participants, symptoms experienced included; hemoptysis 63 (27.3%), fever 78.4%, chest pain 68.4%, anorexia, 57.6%, exertional dyspnoea 74%, fatigue 77.9%, and cervical lymphadenopathy 6.5%.
Fourteen (25%) of the hospitalised patients had ≤ 3 affected lung zones compared to 22 (13%) of the out-patients (OR = 2.20; CI: 1.05–4.72). Participants who ever had hemoptysis before diagnosis had a longer delay to initiation of TB treatment (Mann-Whitney test P-value = 0.019). Disease severity as measured by AFB (Alcohol Acid Fast Bacilli) sputum smear grades, was also associated with the duration of delay to treatment. (Kruskall-Walis test P < 0.001). It was also observed that presence of cavitary disease was associated with prolonged delay to initiation of treatment (Man-Whitney test P < 0.001).
HIV sero-status information assessed by self-report was only available in 56 (24%) of the respondents, in which an association with delay could not be established. However self reported HIV positive sero-status results were more common among hospitalised participants 18 (90%) compared to 10 (27.8%) in the out-patients (OR = 23; CI: 4.57–119.77).
Self reported HIV positive patients were more likely to be female 20 (69.0%) vs. male 8 (29.6%), (OR = 5.30; CI 1.69–16.40); and were un-likely to have cavitary disease 24 (85.7 %) vs. 13 (46.4%) of the self reported HIV sero-negatives. (OR = 6.92; CI: 1.90–25.23).
Distribution of the various time delays
The time delays experienced by the patients are described in Table 1 as well. Health service delay was the most frequent type of delay observed and the greatest contributor to total delay. The total delay to treatment was more than one month in 210 (90.9%) and exceeded 2 months in 65.8% of the cases.
Place of first presentation
Table 2 shows the first health facility visited by the participants for their illness and the corresponding time delays. Seventy six percent of the patients had first presented to a private clinic or pharmacy/drug shop. Health service delay was longest (13.7 weeks) for those who first approached government health centres for their sickness. Of those who initially presented to government hospitals for treatment, 46.2% were investigated for TB using sputum microscopy; and 69.2% of them, were unlikely to change to another health facility. The prevalence of self-referral to the NTLP centre was 33% and family members were seen to be an important source of influence in seeking treatment in 53% of the cases.
Table 2 First health facility visited by the participants for their illness and corresponding median time delays in weeks
First health facility visited Number % Patient delay Health service delay Total delay
Pharmacy/Drug shop 91 39.4 1.0 10.6 12.0
Private clinic 85 36.8 1.0 8.4 12.0
Traditional healer 21 9.1 3.0 9.9 12.0
Government hospital 13 5.6 4.0 7.1 12.0
Health center 10 4.3 3.0 13.7 22.0
NTLP center 9 3.9 10.0 0.0 10.0
Non Governmental hospital* 2 0.9
*Delay times not stated because cases were very few.
Respondents' awareness of TB symptoms and perceived causes
The most frequently mentioned TB symptom was persistent cough (40.7%) and wasting (25.1%) and fever (9.5%). One hundred and fifteen (50%) patients were not aware of any possible causes of TB, and only 7.4% mentioned that germs cause TB. As noted, there was a low level of awareness of TB symptoms and it's causes in this study population. About 56% of the respondents acknowledged that TB patients are stigmatised in their communities. For instance, 40% of participants would not want to be near a TB case.
Factors associated with patient delay
Table 3 presents un-adjusted and adjusted odds ratios for patient delay (>2 weeks) for a number of socio-demographic, lifestyle and health seeking factors. Of all the factors investigated, we found that hospitalized patients had a shorter delay to seeking treatment compared to out-patients (OR = 0.38, 95% CI, 0.18–0.81). Notably subsistence farmers delayed longer seeking help for treatment compared to the rest of the respondents (OR = 4.70; CI: 1.67–13.22); and were more likely not to have had post-primary education 27 (96.4%), compared to the other participants 112 (55.2%), (OR = 21.94; CI: 2.93–164.55). Other independent predictors of patient delay included: daily alcohol consumption (0R = 3.70; CI: 1.57–9.67) and misperceptions that TB is caused by smoking (OR = 5.54; CI: 2.26–13.58). Compared with males, females had a shorter delay seeking help for symptoms, although this finding did not reach statistical significance. Severity of initial symptoms and age had no effect on patient delay.
Table 3 Relationship between socio-demographic, lifestyle and health seeking factors to Patient delay. Both unadjusted and adjusted odds ratios are shown (n = 231)
Variable n % Patient delay >2 weeks Unadjusted odds ratio (95% CI) n* Adjusted odds ratio† (95% CI)
Age 18–40, y 205 30.2 0.98 (0.40–2.40) NA
Male 132 35.6 1.83 (1.02–3.29) NA
Hospitalised 59 16.9 0.38 (0.18–0.81) 56 0.32 (0.12–0.80)‡
Marital status separated/single 125 30.4 1.01 (0.58–1.77) NA
Post primary education Level 92 25.0 0.65 (0.36–1.18) NA
>1 Household persons 207 28.5 0.47 (0.20–1.11) NA
Daily Alcohol consumption 38 51.5 2.93 (1.41–6.11) 34 3.70 (1.57–9.76)‡
Subsistence farming 28 46.4 2.22 (1.02–5.00) 26 4.70 (1.67–13.22)‡
Perceived smoking as cause of TB 34 55.9 3.63 (1.72–7.66) 34 5.54 (2.26–13.58)‡
>2 Health seeking encounters per month 76 13.2 0.24 (0.12–0.50) 72 0.15 (0.06–0.36)‡
*Number included in forward stepwise logistic regression method.
†Odds ratios for the variables appearing at the final step of forward stepwise selection.
‡P < 0.05; CI indicates confidence interval; NA not applicable.
Factors associated with health Service Delay
The median health service delay was 9 weeks range (0.1–135.6). Only fifty-eight (36%) of the patients were diagnosed and initiated on treatment within one month following initial contact with a health provider. Of the factors investigated, these were the three independent predictors of health service delay >4 weeks, after adjusting for other variables (Table 4) namely: medical expenditure on treatment of TB related symptoms > 29 US dollars, (OR = 3.88), >2 multiple health seeking encounters per month (OR = 2.74) and residing alone (OR = 0.31). Presence of hemoptysis at onset was associated with health service delay although this aspect did not reach statistical significance.
Table 4 Relationship between socio-demographic, lifestyle and health seeking factors to Health service delay. Both unadjusted and adjusted odds ratios are shown (n = 231)
Variable n % Health service delay >4 weeks Unadjusted odds ratio (95% CI) n* Adjusted odds ratio† (95% CI)
Age 18–40, y 205 81.0 0.77 (0.25–2.37) NA
Male 132 81.1 0.95 (0.49–1.86) NA
Hospitalised 59 86.4 1.63 (0.71–3.75) NA
Marital status separated/single 125 82.1 1.09 (0.56–2.12) NA
Post primary education level 92 83.5 1.40 (0.72–2.73) NA
Single Household person 24 66.7 0.41 (0.16–1.03) 23 0.31 (0.11–0.84)‡
Daily Alcohol consumption 38 85.7 1.44 (0.53–3.97) NA
Subsistence farming 28 89.3 2.05 (0.59–7.11) NA
Perceived smoking as cause of TB 34 87.5 1.62 (0.20–13.56) NA
>2 Health seeking encounters per month 76 89.5 2.48 (1.09–5.65) 72 2.74 (1.10–6.83)‡
Medical expenditure on TB related symptoms (>29 US $) 60 88.3 2.02 (0.85–4.82) 60 3.88 (1.19–12.62) ‡
Hemoptysis at onset 33 93.9 4.05 (1.02–15.90) NA
*Number included in forward stepwise logistic regression method.
†Odds ratios for the variables appearing at the final step of forward stepwise selection.
‡P < 0.05; CI indicates confidence interval; NA not applicable.
Discussion
Delay in diagnosis and initiation of TB treatment
This study from Kampala Uganda highlights the prolonged delay from onset of patients' symptoms until initiation of TB treatment at Mulago hospital. Health service delay (74%) represented the main component of the overall total delay, reflecting an inadequacy of the clinical services to diagnose TB among symptomatic individuals. The median total delay to treatment of 12 weeks observed in this study is similar to that seen in other studies[6,9,10]. In Uganda [5], Oola (2001) found a median total delay of 17 weeks among both smear positive and smear negative TB patients.
Seventy percent of the respondents reported visiting a health provider within 2 weeks of onset of signs and symptoms. This short patient delay could reflect good awareness of health problems and easy access to health care in the study population. However 21 (9%) of patients sought the initial treatment 2 months after symptom onset. Often misinterpretation of initial symptoms results in actions (patients' reliance on self medication and pharmacy or drug shop attendance) that may delay seeking of appropriate medical care. The greater proportion of health service delay found in our study is similar to prior findings from studies in Ghana, Botswana and Gambia[7,9,10], but contrary to findings from South-Africa and Tanzania[11,12]. This long health service delay conceivably reflects insufficient knowledge of the signs and symptoms of TB among different types of health providers and the general population. The shorter patient delay found in our study, contrasts with findings of Oola (2001), in which patient delay was prolonged[5]. Most of the patients (61.5%) in Oola's study were rural based, compared to the more urban setting of the present study. Rural residence has been found to be a risk factor for late diagnosis because of poorer access to health care and differences in education levels between rural and urban areas[10,11].
Factors associated with patient delay
In this present study, perception of smoking as a cause of TB was associated with prolonged patient delay. This may be because cough appearing in smokers is often attributed to smoking, resulting in delays seeking help for TB related symptoms. By contrast, in Philippines[13], there was no relationship between perceived causes of TB and health-seeking due to potential recall and selection biases. Education status had no effect on delay to treatment initiation as established elsewhere[10,11]. However subsistence farmers were found to have long patient delays probably due to lack of education and poverty. Similarly, alcoholics had prolonged patient delays. Gender had little impact on delay[14], though other studies have confirmed prolonged delays to initiation of treatment among females[5,7,11]. However larger studies should be undertaken to explore the effect of gender on access to health care. Severity of initial symptoms had no effect on delay possibly due to recall bias and individual variations in perception of disease.
Factors associated with health service delay
Our study revealed that the private health system is the common first choice of care for TB patients in Kampala district, with approximately 75% of individuals presenting to a private clinic, pharmacy or drug shop initially. In contrast, 75% of the patients in South Africa used the public health system as the initial contact[11]. This may be partly related to the health-care service quality. The public health sector in Uganda offers a free health service in agreement with the government health policy, though existing under-funding of this sector leads to insufficient utilities and consequently discourages it's optimal use. Longer delays to treatment were experienced among patients originating from peri-urban localities in which limited access to primary health care services significantly contributes to late diagnosis[7,10].
Multiple health seeking encounters contributed to the prolonged duration of health service delay along with the associated medical costs. A low level of clinical suspicion of TB by health providers and failure to order proper investigations or refer patients to 'higher level' health units contributes in a major way to health service delay. Patients often need to return to the same providers of healthcare or seek advice from others, because of persistent symptoms[6,15].
Clinical features associated with delay to treatment
Our study found that patients who required hospitalisation prior to TB diagnosis had shorter delays to treatment and less advanced disease, compared to out-patients. This shorter delay to treatment is partly explained by the finding that hospitalisation was HIV associated. In Ghana [7] hospitalised patients had more advanced disease and prolonged delays to treatment, though HIV status was not documented. A disparity in the HIV epidemic transitions in both countries is a possible explanation for this difference. The presence of limited cavitary disease among known HIV positive patients is known to be related to diminished immune function. Illness severity as measured by chest radiography and sputum smear grades was found to be associated with the duration of delay to treatment. Advanced disease has been found to correlate with mortality and chronic morbidity [7].
Conclusion
Patient and health service delays contribute significantly to delays in patients accessing treatment. Substantial reduction in case detection delays may be achieved through; 1) More specific and effective health education of the general public on tuberculosis and seeking of appropriate medical consultation; 2) Targeting specific groups of alcoholics and subsistence farmers to improve accessibility to TB treatment 3) Continuing medical education about TB management procedures for health providers; 4) Improvement in the capacity of TB control services. Similar research in the future could serve as a basis for monitoring improvement in the quality of the tuberculosis control program in Uganda.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
MSK conceived of the study, wrote the research proposal, conducted the research, did data entry, performed statistical analysis and wrote the manuscript. KC was involved in the write up of the proposal, data analysis and draft of the manuscript. MKH was involved in the write up of the proposal, data analysis and draft of the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors wish to thank all the staff of the National Referral Tuberculosis treatment centre who participated in the study particularly Mr. Augustine Banyanga, Mr Okot Gabriel and the research assistants Dr. Henry Wamala and Dr. Helen Byakwaga. This study was supported by a WHO-TDR Grant.
==== Refs
Uganda MoH National Tuberculosis and leprosy control Program Annual report 2002
Ministry of Health Uganda: Guidelines for National demonstration and Training Districts implementing DOTS, Ministry of Health NTLP Working draft Kampala 1997
World Health Organisation. Global tuberculosis control: WHO Report 2003 WHO/CDS/TB/2003316 2003 Geneva, WHOM
Ngamvithayapong J Winkvist A Diwan V High AIDS awareness may cause tuberculosis patient delay: results from an HIV epidemic area, Thailand AIDS 2000 14 1413 1419 10930157 10.1097/00002030-200007070-00015
Oola J Factors influencing Delayed diagnosis of tuberculosis in Mukono District Uganda 2001 Institute of Public Health
Liam CK Tang BG Delay in the diagnosis and treatment of pulmonary tuberculosis in patients attending a university teaching hospital Int J Tuberc Lung Dis 1997 1 326 332 9432388
Lawn SD Afful B Acheampong JW Pulmonary tuberculosis: diagnostic delay in Ghanaian adults Int J Tuberc Lung Dis 1998 2 635 640 9712277
Guwatudde D Nakakeeto M Jones-Lopez EC Maganda A Chiunda A Mugerwa RD Ellner JJ Bukenya G Whalen CC Tuberculosis in household contacts of infectious cases in Kampala, Uganda Am J Epidemiol 2003 158 887 898 14585767 10.1093/aje/kwg227
Steen TW Mazonde GN Pulmonary tuberculosis in Kweneng District, Botswana: delays in diagnosis in 212 smear-positive patients Int J Tuberc Lung Dis 1998 2 627 634 9712276
Lienhardt C Rowley J Manneh K Lahai G Needham D Milligan P McAdam KP Factors affecting time delay to treatment in a tuberculosis control programme in a sub-Saharan African country: the experience of The Gambia Int J Tuberc Lung Dis 2001 5 233 239 11326822
Pronyk RM Makhubele MB Hargreaves JR Tollman SM Hausler HP Assessing health seeking behaviour among tuberculosis patients in rural South Africa Int J Tuberc Lung Dis 2001 5 619 627 11467368
Wandwalo ER Morkve O Delay in tuberculosis case-finding and treatment in Mwanza, Tanzania Int J Tuberc Lung Dis 2000 4 133 138 10694091
Auer C Sarol J JrTanner M Weiss M Health seeking and perceived causes of tuberculosis among patients in Manila, Philippines Trop Med Int Health 2000 5 648 656 11044280 10.1046/j.1365-3156.2000.00615.x
Godfrey-Faussett P Kaunda H Kamanga J van Beers S van Cleeff M Kumwenda-Phiri R Tihont V Why do patients with a cough delay seeking care at Lusaka urban health centres? A health systems research approach Int J Tuberc Lung Dis 2002 6 796 805 12234135
Needham DM Foster SD Tomlinson G Godfrey-Faussett P Socio-economic, gender and health services factors affecting diagnostic delay for tuberculosis patients in urban Zambia Trop Med Int Health 2001 6 256 259 11348515 10.1046/j.1365-3156.2001.00709.x
|
16307685
|
PMC1310609
|
CC BY
|
2021-01-04 16:28:59
|
no
|
BMC Public Health. 2005 Nov 24; 5:122
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-122
|
oa_comm
|
==== Front
BMC Complement Altern MedBMC Complementary and Alternative Medicine1472-6882BioMed Central London 1472-6882-5-201629723610.1186/1472-6882-5-20Research ArticleA gap between acceptance and knowledge of herbal remedies by physicians: The need for educational intervention Clement Yuri N [email protected] Arlene F [email protected] Kristi [email protected] Tricia [email protected] Savrina [email protected]é Maurice [email protected] Oneil [email protected] Kerry [email protected] Compton E [email protected] Faculty of Medical Sciences, The University of the West Indies, Trinidad and Tobago2005 18 11 2005 5 20 20 29 7 2005 18 11 2005 Copyright © 2005 Clement et al; licensee BioMed Central Ltd.2005Clement et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The unprecedented global increase in the use of herbal remedies is set to continue apace well into the foreseeable future. This raises important public health concerns, especially as it relates to safety issues including adverse effects and herb-drug interactions. Most Western-trained physicians are ignorant of the risks and benefits of this healthcare modality and assessment of acceptance and knowledge would identify appropriate intervention strategies to improve physician-patient communication in this area.
Methods
A cross-sectional survey was done using an interviewer-administered pilot tested de novo questionnaire at six public hospitals in Trinidad between May–July 2004. The questionnaire utilized weighed questions to quantify acceptance (maximum score = 14 points) and knowledge (maximum score = 52 points). Acceptance and knowledge scores were analyzed using the ANOVA and Tukey's tests.
Results
Of 192 physicians interviewed, most (60.4%) believed that herbal remedies were beneficial to health. Respondents had relatively high acceptance levels (mean = 5.69 ± 0.29 points or 40% of total possible score) and poor knowledge (mean = 7.77 ± 0.56 points or 15% of total possible score). Seventy-eight physicians (40.6%) admitted having used herbs in the past, and 60 of these (76.9%) were satisfied with the outcome. Although 52 physicians (27.1%) recommended the use of herbs to their patients only 29 (15.1%) were able to identify at least one known herb-drug interaction.
Conclusion
The use of herbal remedies is relatively high in Trinidad, as throughout the world, and most patients self-medicate with or without the knowledge of their attending physician. Surprisingly, we demonstrated relatively high acceptance levels and use of herbs among physicians in Trinidad. This interesting scenario of high acceptance levels and poor knowledge creates a situation that demands urgent intervention. We recommend educational intervention to narrow the gap between acceptance and knowledge so that physicians would be adequately equipped to communicate with their patients on this modality. The integration of herbal medicine into the curriculum of medical schools, continuing education programs and the availability of reputable pharmacopoeias for referencing at public health institutions are useful instruments that can be used to close this gap and promote improved physician-patient communication.
==== Body
Background
More than 80% of the population in the developing world use traditional medicine, which includes herbal remedies, for the management of health [1]. In the developed world the dramatic revolution in healthcare was facilitated by the discovery of pharmacologically active chemical entities (supported by evidence-based safety and efficacy testing) and this has shifted the emphasis away from traditional herbal medicine. Consequently, many Western-trained physicians have little formal training and lack knowledge on the benefits and risks of herbal remedies [2,3].
There has been an unprecedented explosion in the popularity of herbal preparations during the last few decades, especially in developed countries [4]. This phenomenon has stimulated considerable public health concern among physicians who are sometimes uncertain about the safety of herbs, especially when used concomitantly with allopathic drugs [5,6]. Despite these concerns, the global prevalence of use of medicinal herbs continues to rise as patients self-medicate with or without informing their physicians [1]. In this setting, the attitudes and knowledge of physicians would impact on the doctor-patient relationship and affect the overall quality of healthcare delivery, particularly with respect to issues such as possible adverse herb effects and herb-drug interactions [7-12].
There is significant use of herbal remedies in the Caribbean [13,14] and recent studies in Trinidad show relatively high prevalence of use for symptomatic relief in asthma [15] and therapeutic management in diabetes mellitus [16]. Merritt-Charles et al [17] reported an 86% lifetime prevalence of use among outpatients at a surgical facility on the island. This high prevalence of use dictates that the most appropriate intervention strategies be implemented to facilitate improved healthcare delivery, especially as it relates to physicians' knowledge of herbal medicines.
Few studies report on the knowledge, attitudes and practices of physicians regarding complementary and alternative medicine, with little emphasis on herbal medicine [18,19]. This study was undertaken primarily to determine the level of acceptance and knowledge regarding herbal medicine by physicians at public hospitals in Trinidad.
Methods
This descriptive study was cross-sectional in design and used a de novo pilot-tested questionnaire during the period May to July 2004. Weighted questions were used to quantitatively assess physicians' acceptance and knowledge of herbal remedies. A sample of public health sector physicians was recruited from the six public hospitals in Trinidad. The nature and purpose of the study were explained on an individual basis, and following the physicians' willingness to participate they signed their informed consent. Respondents were permitted to withdraw from the study at anytime after the interview had commenced. The study was approved by the Ethics Committee, Faculty of Medical Sciences at the University of the West Indies, Trinidad.
Setting and sample
The sample of physicians was recruited from the six public hospitals, namely the Port-of-Spain General Hospital (POSGH), the San Fernando General Hospital (SFGH) and the Eric Williams Medical Sciences Complex (EWMSC). These are the three major hospitals and employ most physicians in the public healthcare sector on the island. The other hospitals were the specialist obstetrics and gynecology Mount Hope Women's Hospital (MHWH), the St. Ann's Psychiatric Hospital (SAPH) and the minor Sangre Grande Regional Hospital (SGRH).
Quota sampling was used to obtain the sample size; the total number of physicians required was proportionally distributed among the hospitals based on the population of physicians at each hospital; i.e. the larger hospitals represented a larger percentage of the sample size. Furthermore, at these larger general hospitals, a representative proportion of physicians from all specialized department was obtained, and this was based on the total number of physicians working at each department. At these departments physicians were interviewed by convenient sampling until the quota was achieved. At the smaller specialist hospitals convenient sampling was used, without stratification, until the quota was achieved. With this method of interviewee selection we expected a 100% response rate. For inclusion in the study physicians must have been employed at the hospital for at least three months.
Interview instrument
The pilot-tested de novo questionnaire was interviewer-administered and determined demographic details including gender, nationality, country of study, duration of employment at the hospital, position at the hospital, years of practicing medicine and level of qualification.
Acceptance (or positive attitude) was assessed using questions that evaluated beliefs, feelings and actions regarding herbal medicines (Appendix 1). Questions were developed de novo to assess these cognitive, evaluative and behavioural aspects of acceptance with a maximum possible score of 14 points. We rationalized that the behavioural component would be the best indicator of acceptance, as it is sometimes possible for beliefs and feelings to be discordant with actions. We estimated that the cognitive and evaluative aspects contributed less weight to acceptance and these items were assigned fewer points (4 out of 14 points). Items on behaviour included personal use, past recommendation and prescription of medicinal herbs to patients, and these were allocated a larger proportion of the overall points (10 out of 14 points).
Knowledge was evaluated using open-ended weighed questions to identify five (5) Caribbean and five (5) non-Caribbean medicinal herbs, their uses, contraindications and important herb-drug interactions, together with other questions determined knowledge of herbal pharmacopoeias and clinical studies on herbal medicines with a maximum score of 52 (Appendix 2). Established pharmacopoeias [20-22] and reputable websites [23,24] were used to determine correct responses related to medicinal herbs (Caribbean and non-Caribbean), their indications and contraindications. A database for important herb-drug interactions was compiled from published reviews [8,11,12].
Statistical analysis
We postulated that most physicians in Trinidad would reject the use of herbal medicines, as did 82% of oncologists who rejected complementary and alternative medicines in the study by Hyodo et al [6]. We used this published prevalence rate to calculate a sample size of 192 physicians, with a confidence level of 95% [25]. This represented about 20% of physicians in the public health sector in Trinidad. The scores for acceptance and knowledge of herbal remedies were expressed as mean ± standard error of mean. ANOVA test was used to determine statistical significance between acceptance scores and knowledge scores and gender, nationality, country of study, years practicing medicine, hospital and department at hospital. Results were considered statistically significant when p < 0.05. In instances where p < 0.05, Tukey's test was used to determine statistical differences within and between groups. The data was analyzed using the Statistical Program for Social Sciences (SPSS) for Windows computer program (Version 11.0, Chicago, IL).
Results
Demography
One hundred and ninety two (192) survey questionnaires were completed (100% response rate) and the demographic details of the sample are given in Table 1. Most respondents were from the major hospitals (82.3%) and were male (74.0%). Most physicians were either native to the Caribbean or Latin America (62.5%) or trained in the region (55.2%). About one-third of the physicians interviewed were either nationals of, or trained in India/Asia and Nigeria. Most physicians (75.0%) had less than ten years medical practice experience. Most respondents (70.2%) were employed in the pediatrics, surgery, general medicine, obstetrics and gynecology and psychiatry departments at the hospitals.
Table 1 Demography and factors affecting knowledge and acceptance of herbal medicines of public health sector physicians in Trinidad
Demographic factor N (%) Mean knowledge scores (maximum = 52) Mean acceptance scores (maximum = 14)
Gender
Male 142 (74) 7.55 ± 0.65 5.60 ± 0.34
Female 50 (26) 8.40 ± 1.14 4.83 ± 0.57
Years practicing medicine
Less than 5 95 (49.5) 7.68 ± 0.82 6.08 ± 0.40
6 to 10 49 (25.5) 6.84 ± 1.03 4.68 ± 0.56
More than 10 48 (25.0) 8.91 ± 1.20 5.95 ± 0.63
Nationality
Caribbean/Latin American 120 (62.5) 9.48 ± 0.78 *1 5.49 ± 0.36
Indian/Asian 47 (24.5) 4.42 ± 0.71 6.36 ± 0.68
Nigerian 21 (10.9) 4.76 ± 1.13 5.14 ± 0.68
European/North American 4 (2.1) 11.88 ± 4.72 6.75 ± 3.03
Country of study
Caribbean/Latin America 106 (55.2) 9.24 ± 0.82 *2 5.56 ± 0.39
India/Asia 48 (25.0) 4.69 ± 0.77 6.27 ± 0.65
Nigeria 22 (11.5) 4.77 ± 1.08 4.98 ± 0.67
Europe/North America 6 (3.1) 11.25 ± 3.91 8.58 ± 1.73
More than 1 country 10 (5.2) 11.50 ± 2.84 4.20 ± 1.19
Hospital
SFGH 56 (29.2) 6.52 ± 0.91 4.67 ± 0.52
POSGH 58 (30.2) 7.85 ± 1.11 5.41 ± 0.49
EWMSC 44 (22.9) 7.39 ± 1.12 6.72 ± 0.65
SAPH 18 (9.4) 10.00 ± 1.95 5.64 ± 1.09
SGRH 9 (4.7) 8.61 ± 2.86 8.00 ± 1.45
MHWH 7 (3.7) 12.86 ± 3.38 6.93 ± 1.15
Department at hospital
Paediatrics 30 (15.6) 10.00 ± 1.53 6.30 ± 0.81
Surgery 29 (15.1) 5.17 ± 0.98 5.21 ± 0.75
General Medicine 30 (15.6) 7.17 ± 1.63 5.28 ± 0.68
Obstetrics & Gynaecology 21 (10.9) 9.64 ± 1.82 5.62 ± 0.81
Psychiatry 19 (9.9) 10.53 ± 1.92 5.42 ± 1.05
Orthopaedics 10 (5.2) 6.25 ± 2.31 6.35 ± 1.23
Radiology 9 (4.7) 6.94 ± 2.49 6.78 ± 1.75
Others 24 (12.5) 5.94 ± 1.33 5.33 ± 0.75
Overall scores 192 7.77 ± 0.56 5.69 ± 0.29
Asterisks (*) show statistical significance using ANOVA. *1 p = 0.023; *2 p = 0.007.
Assessment of acceptance
The mean acceptance score was 5.69 ± 0.29 (maximum score = 14), Table 1. Gender, nationality, country of study, hospital site, specialty and experience did not influence acceptance scores. Most physicians (60.4%) believed that herbal medicines were beneficial to health. Seventy-eight physicians (40.6%) reported that they had used this healthcare modality in the past, and 60 of these 78 respondents (or 76.9%) indicated that they were satisfied with the outcome.
Fifty-two respondents (27.1%) had previously recommended the use of medicinal herbs to their patients, and were also able to identify the herbs and herbal products in the treatment and management of diseases such as peptic ulcers, prostate enlargement and hepatitis, which supported their acceptance of this modality. However, only 14 physicians (7.3%) had ever advised their patients to consult an herbalist. Forty-three or 22.4% of the sample indicated that they would recommend medicinal herbs, if the option were available, to patients who were refractory to treatment with conventional allopathic medicines. Thirty physicians (15.6%) indicated that they accepted herbal medicines as a viable healthcare option, as they were aware of traditional medicinal practices, such as Ayuvedic medicine, and clinical trials that supported safety and efficacy of herbal remedies. Most physicians (58.3%) also expressed their willingness to allow their patients to participate in randomized controlled clinical trials to validate the safety and efficacy of medicinal herbs.
For 73 physicians (39.1%) the primary reason for rejecting herbal medicines was due to the sparse scientific information available from clinical trials to support the safety and efficacy of medicinal herbs in healthcare management. To a much lesser extent other reasons for rejecting herbal medicines included an absence of impartation of such knowledge during their medical training, non-relevance to specific specialties and the medico-legal issues of prescribing this modality in the public healthcare delivery system in Trinidad. Other physicians indicated that they personally did not believe that herbs were either safe or beneficial and one respondent commented "... and some are not scientifically proven to work and can give patients a false hope." However, most respondents (81.3%) believed that continuing education in herbal medicine was important to facilitate greater doctor-patient interaction in this mushrooming area of healthcare management.
Assessment of knowledge
Mean knowledge score was 7.77 ± 0.56 (maximum score = 52), Table 1. Gender, hospital site or specialty did not influence knowledge; however, nationality and country of study significantly affected knowledge on medicinal herbs, p < 0.05. Tukey's test showed that physicians native to and trained in the Caribbean and Latin America had significantly higher knowledge scores than their Indian/Asian and Nigerian counterparts. A trend of moderate increase in knowledge with years of medical experience was observed, but this was not statistically significant.
Ninety-six respondents (50%) could identify at least two (2) Caribbean medicinal herbs and their traditional uses, whereas only 54 (28.1%) could identify at least two (2) non-Caribbean medicinal herbs and their uses. Most respondents were unable to identify at least one contraindication for either Caribbean (82.3%) or non-Caribbean herbs (87.5%) and only 29 physicians (15.1%) were able to correctly identify any known herb-drug interaction. Most physicians (55.7%) reported asking patients about their medicinal herb use in the history taking of drug use.
The most popular herbs identified are listed in Table 2, and include several medicinal plants of Caribbean and non-Caribbean origins such as lemongrass (Cymbopogon citratus DC. Stapf), noni (Morinda citrifolia Linn.) and saffron (Curcuma longa L.). Interestingly, the medicinal use of marijuana (Cannabis sativa) was recognized in third place with 26 physicians (13.5%) recognizing its usefulness.
Table 2 Most common medicinal herbs and their reputed uses cited by public health sector physicians in Trinidad
Common name Botanical name Traditional medicinal uses identified No. (%) of physicians citing
Lemon grass Cymbopogon citratus (DC.) Stapf. Fever, common cold, blood purifier, headache 34 (17.7)
Gingko Gingko biloba L. Enhance memory, vertigo, dementia, impotence 29 (15.1)
Marijuana Cannabis sativa L. Asthma, glaucoma, anorexia, multiple sclerosis, epilepsy, nausea/vomiting, analgesic antidepressant 26 (13.5)
Garlic Allium sativum L. Hypertension, fever, joint pain, cough, analgesic, anti-parasitic, kidney problems 24 (12.5)
Aloes Aloe vera (Aloe barbadensis Miller) Superficial wounds/cuts, dysmenorrhea, acid peptic disease, acne, skin purifier, peptic ulcer, constipation 23 (12.0)
Senna Cassia senna L. Laxative, antihelmentic, 'colon cleanser', 'blood purifier' 19 (9.9)
Noni Morinda citrifolia Linn. Hypertension, acne, diabetes, pain, immune system, prostate 18 (9.4)
Ginseng Panax ginseng C.A. Mayer Enhance stamina and memory, aphrodisiac 17 (8.9)
Echinacea Echinacea purpurea L. Moench Boost immune system, upper respiratory tract infections, common cold, acne 15 (7.8)
St. John's wort Hypercium perforatum L. Depression 15 (7.8)
Evening primrose oil Oenothera biennis Nutt. Hormonal imbalance in menopause, breast pathology and mastalgia 11 (5.7)
Saw palmetto Serenoa repens (Bartram) Small Benign prostatic hypertrophy 11 (5.7)
Wonder-of-the-world Kalanchoe pinnata (Lam.) Pers. l Swelling, earache, rash, mumps, cuts 9 (4.7)
Saffron Curcuma longa L.l Purgative, antiseptic, acne common cold, fever, cough antihistamine, menstrual problems, osteoarthritis 9 (4.7)
Only 18 physicians (9.4%) had access to information on herbal medicine at their place of work, and 30 (15.6%) had ever attended conferences or workshops where a paper on herbal medicine was presented or discussed. Although 105 physicians (54.7%) were aware of herbal pharmacopoeias, only 13 of these 105 (or 12.4%) could name at least one.
Discussion
This study showed for the first time the marked disparity between acceptance and knowledge of herbal medicines by public health sector doctors in Trinidad. The mean acceptance score was about 40% of the total possible score, whereas the mean knowledge score was about 15% of the total possible score. Generally, these healthcare providers had relatively high acceptance levels with poor knowledge. The gap between acceptance and knowledge of herbal remedies by physicians may indicate the differential between traditional/cultural beliefs and the lack of access to information.
We propose that this disparity could be partly explained by the origin and composition of Trinidad's 1.3 million inhabitants [26]. More than 80% of the population is composed of descendants of enslaved Africans and indentured Asian Indian labourers who arrived in the Caribbean during the last few centuries [27]. These two major ethnic groups now coexist in almost equal proportions on the island. There is also a minority presence of peoples of Middle Eastern, European and Chinese origin. A significant proportion of the population is comprised of individuals of 'mixed' heritage arising out of the intermarriage among all ethnic groups. These recently transplanted peoples have attempted to maintain their traditional medicinal practices, and in some cases medicinal plants were introduced into the flora of the island [28-31]. Although most Trinidadians today are aware of the benefits of some traditionally and culturally used medicinal herbs, there is generally a lack of transmission of substantial traditional knowledge from generation to generation. This progressive loss of traditional knowledge is further exacerbated by the displacement of traditional medicinal practices by Western medicine in the modern Trinidadian society. Nonetheless, most Trinidadians accept medicinal herbs as a viable option in healthcare management [15-17].
Unlike earlier studies that used closed questions, where true-or-false responses were required for a limited number of popular medicinal herbs, most of the knowledge questions in our survey instrument were open-ended. We postulated that this approach would provide a wider breadth to assess 'true' knowledge without limiting responses and also eliminate the likelihood of respondents guessing the correct answers. Besides, our finding of poor knowledge among physicians in Trinidad was consistent with other studies assessing medical practitioners [18,19] and other healthcare professionals such as nurses [32] and pharmacists [33].
Over 40% of physicians interviewed reported using medicinal herbs in the past, with more than three-fourths of these being satisfied with the outcome. Our results were markedly higher than in the Norwegian study where only 12% of physicians reported the use of alternative medicines, which included herbal medicines [5]. Our results were similar to a US study where 66% of pediatricians supported the view that complementary and alternative medicines could ameliorate symptoms or hasten recovery [34].
About one-fifth of our sample population suggested that this therapeutic option should be explored when conventional allopathic medicines fail, and this finding was similar to that reported in a recent UK study [35] where physicians' personal attributes and training influenced the likelihood of recommending herbal medicines. Hyodo et al [6] reported that 13% of oncologists noted CAM-associated improvement in their patients, and 9.9% indicated that there was sufficient evidence to support use of this modality. Most medical practitioners in Trinidad were willing to go a step further by agreeing to allow their patients to be recruited for randomized controlled clinical trials that would validate (or otherwise) the safety and efficacy of Caribbean "bush teas".
Most respondents, particularly native West Indians, were able to identify the traditional medicinal uses of Caribbean herbs; and fewer were able to identify non-Caribbean herbs and their indications. In this study, knowledge of contraindications and herb-drug interactions was very poor and was similar to a US study which also demonstrated a knowledge deficit as it related to adverse effects of herbs [36]. We noted that physicians of Indian/Asian and Nigerian origins were not familiar with medicinal herbs in the Caribbean and this adversely affected their knowledge scores. Despite this shortcoming, most physicians of Indian origin were aware of Ayuvedic and other traditional medicine practices, and of clinical trials with herbal remedies which supported their acceptance of this healthcare modality. Interestingly, many medicinal herbs presently growing in the Caribbean were transplanted by the Asian and African diaspora during the last few centuries.
More than half of the physicians interviewed reported having asked their patients specifically about herbal medicine use when taking a drug history, and this was similar to an Israeli study where 58% of physicians always or frequently asked their patients about their use of complementary medicine [37]. Our results were significantly higher than the 20% of pediatricians in a US study who queried use in their patients [34]. We did not determine whether this information was recorded in the patients' charts or whether physicians attempted to advise or dissuade patients from using this modality. Cohen et al [38] showed that although physicians may have asked their patients about herb use, only about one-third of these doctors actually documented this information in patients' charts. The subsequent impact of this query on the quality of healthcare is unknown.
As the use of medicinal herbs continues to increase worldwide, there has been a parallel surge in research to isolate pharmacologically active pure compounds from medicinal plants, and clinical investigations are being done to establish the safety profile and efficacy of some of the more popular traditionally used herbal remedies [39-42]. The reports of these investigations are appearing in several international herbal pharmacopoeias including the European Scientific Cooperative On Phytotherapy (ESCOP) [20], the German Commission E [21] and TRAMIL [22] giving such information as botanical names, common names, traditional uses and therapeutic indications, chemical constituents, contra-indications and pharmacological properties of selected herbs. Evidence-based information is available and could be compiled and structured in a format to impart workable knowledge to medical students and practicing physicians. The availability of these herbal medicine resources at the worksite, as reference material, is crucial for creating an environment conducive to more efficient physician-patient interaction in the area.
At present, formal training in herbal medicine at the regional university (The University of the West Indies) where most of our physicians are trained, does not form part of the curriculum and subsequently graduates possess no formal knowledge in the area. Our study showed that most respondents agreed that continuing education in herbal medicines was necessary for effective patient consultations. Recent studies have shown that herbal medicine educational interventions taught as structured programs via different media significantly improved physicians' knowledge, confidence and their interactions with patients [18,43]. Frenkel et al [44] showed that 72% of family practice physicians who participated in a structured patient-centred educational program on CAMs reported significant positive attitudinal changes.
Our survey was conducted at public hospitals throughout Trinidad, and obviously excluded physicians who worked exclusively at private institutions and probably different responses would have been elicited from these physicians. Although the use of quota sampling with stratification was advantageous for our research purposes, interviewer bias may have been introduced by the non-random convenient selection of respondents, resulting in a sample that may not have been truly representative. Although open-ended questions were especially useful for gathering responses on knowledge, where replies were too numerous to code (for example, more than 45 medicinal plants and several traditional indications were identified in this study), there were some limitations. Open-ended questions require more thought and are taxing on the respondent and this may have affected the quality of responses. Also, responses to these open-ended items may have been summarized and the true meaning distorted by the interviewer or by the coding process used for data entry.
Notwithstanding these limitations, the burden of our results suggests that there is an urgent need for educational intervention with regard to herbal medicine in the training of our physicians. We propose that there be an integration of herbal medicine into the current medical curriculum so that future physicians would be better able to communicate with their patients on this healthcare modality. Continuing education programs are also recommended so that practicing physicians would have the opportunity to upgrade their knowledge in this rapidly expanding area of significant public health concern. In the interim, public health institutions should be equipped with reputable herbal pharmacopoeias and electronic databases to answer questions that would arise during the course of clinical practice.
Conclusion
Our findings showed that medical practitioners in the public healthcare sector in Trinidad generally accepted herbal remedies as a viable option although they lack sufficient knowledge on the uses and potential risks associated with this modality. This result directly contradicted our initial hypothesis that herbal remedies would be rejected (or poor acceptance) and that this would correlate with poor knowledge. This creates an interesting scenario where the gap between acceptance and knowledge provides an ideal opportunity to facilitate the introduction of educational programs and policies that would increase the knowledgebase of these healthcare professionals. Well-informed physicians would be more confident in their interactions with patients and this would improve the quality of healthcare delivery, as more meaningful communication on important issues such as adverse effects and herb-drug interactions would be facilitated. The increasing trend in the use of herbs is set to continue well into the foreseeable future and the enhanced knowledgebase of physicians would redound to the benefit of patients who would appreciate a non-judgmental environment when discussing healthcare needs.
Appendix 1. The eight weighted acceptance questions (Total score = 14 points)
1. Do you believe that herbal medicines are beneficial in healthcare management? (2 points for affirmative answer)
Yes □
No □
2. Have you ever recommended the use of herbal medicines? (2 points for affirmative answer)
Yes □
No □
3. Have you ever prescribed herbs to patients for medicinal purposes? (2 points for affirmative answer)
Yes □
No □
4. Have you ever recommended patients to an herbalist? (2 points for affirmative answer)
Yes □
No □
5. Have you ever personally used herbs? (2 points for affirmative answer)
Yes □
No □
6. Do you think that the use of herbal medicines should be limited only to patients who
have failed conventional therapy? (1 point for affirmative answer)
Yes □
No □
7. Do you think that continuing education in herbal medicines is important? (1 point for affirmative answer)
Yes □
No □
8. Would you be willing to allow your patients to participate in clinical trials to evaluate
the efficacy of Caribbean 'bush' teas? (The normal trial protocol has been followed) (2 points for affirmative answer)
Yes □
No □
Appendix 2: The ten weighted knowledge questions (Total score = 52 points)
1. Can you identify any five (5) Caribbean herbs and their common usages? (2 points for each herb and its correct use; 10 points total)
2. Can you list five (5) contraindications of named Caribbean herbs? (2 points for each herb and its correct contraindication(s); 10 points total)
3. Can you identify any five (5) imported herbs and their common usages? 2 points for each herb and its correct use; 10 points total)
4. Can you list five (5) contraindications of named imported herbs? (2 points for each herb and its correct contraindication(s); 10 points total)
5. Can you list any two (2) herb-drug interactions? (2 points for each correct answer; 4 points total)
6. Do you specifically ask your patients about their use of herbal medicine when taking a drug history? 1 point for affirmative answer)
Yes □
No □
7. Are you aware that several international herbal pharmacopoeias exist? (1 point for affirmative answer)
Yes □
No □
8. If yes, could you give at least one (1) example of a herbal pharmacopoeia? (3 points for correct pharmacopoeia)
9. Are you aware that there are several completed and ongoing international clinical trials on the efficacy and safety of herbal medicines? (1 point for affirmative answer)
Yes □
No □
10. If yes, could you identify one example? (2 points for correct example)
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
YNC was the P.I. and was responsible for the study concept, development of methodology, coordinating of the research activities, analyzing the data, and writing the manuscript. KK, TB, SB, MF and KN were involved in methodological development, data collection, data input and analysis. OM was involved in methodological development. AFW was involved in data input and statistical analysis. CES was involved in an advisory capacity in the study concept and development of methodology phases of the research. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We wish to thank Ms. Asha Dookie for her initial assistance in drafting the protocol and questionnaire design. Dr. Celia Poon King and Mr. Rudy Singh are acknowledged for their assistance in epidemiology and statistical assistance respectively. The Chief of Medical Staff of the participating hospitals, Drs. R.P. Singh, Dale Hassranah, Esau Joseph, Rowland Moze, Suresh Pooran and Ian Hypolite, are especially thanked for their expeditious responses to our requests for approval to conduct the study. We thank all the physicians who willingly participated in the study and without whose input this report would not have been possible. Mrs. Patricia Clement and Dr. Lazara Montané Jaime are acknowledged for their valuable and insightful review of the manuscript.
==== Refs
World Health Organization WHO traditional medicine strategy 2002–2005 2002 WHO, Geneva
Pirotta MV Cohen MM Kotsirilos V Farish SJ Complementary therapies: have they become accepted in general practice? Med J Aust 2000 172 105 109 10735019
Suchard JR Suchard MA Steinfeldt JL Physician knowledge of herbal toxicities and adverse herb-drug interactions Eur J Emerg Med 2004 11 193 197 15249804 10.1097/01.mej.0000134721.72356.f7
Tindle HA Davis RB Phillips RS Eisenberg DM Trends in use of complementary and alternative medicine by US adults: 1997–2002 Altern Ther Health Med 2005 11 42 49 15712765
Risberg T Kolstad A Johansen A Vingerhagen K Opinions on and use of alternative medicine among physicians, nurses and clerks in northern Norway In Vivo 1999 13 493 498 10757043
Hyodo I Eguchi K Nishina T Endo H Tanimizu M Mikami I Takashima S Imanishi J Perceptions and attitudes of clinical oncologists on complementary and alternative medicine, a nationwide survey in Japan Cancer 2003 97 2861 2868 12767101 10.1002/cncr.11402
Sardesai VM Herbal medicines: Poisons or potions? J Lab Clin Med 2002 139 343 348 12066132 10.1067/mlc.2002.124582
Fugh-Berman A Herb-drug interactions Lancet 2000 355 134 138 10675182 10.1016/S0140-6736(99)06457-0
Klepser TB Klepser ME Unsafe and potentially safe herbal therapies Am J Health Syst Pharm 1999 56 125 138 10030529
Brazier NC Levine MA Drug-herb interaction among commonly used conventional medicines: a compendium for health care professionals Am J Ther 2003 10 163 169 12756423 10.1097/00045391-200305000-00003
Williamson EM Drug interactions between herbal and prescription medicines Drug Saf 2003 26 1075 1092 14640772
Izzo AA Ernst E Interactions between herbal medicines and prescribed drugs: a systematic review Drugs 2001 61 2163 2175 11772128
Gardner JM Grant D Hutchinson S Wilks R The use of herbal teas and remedies in Jamaica West Indian Med J 2000 49 331 335 11211545
Michie CA The use of herbal remedies in Jamaica Ann Trop Paediatr 1992 12 31 36 1376584
Clement YN Williams AF Aranda D Chase R Watson N Mohammed R Stubbs O Williamson D Medicinal herb use among asthmatic patients attending a specialty care facility in Trinidad BMC Complementary and Alternative Medicine 2005 5 3 15713232 10.1186/1472-6882-5-3
Mahabir D Gilliford MC Use of medicinal plants for diabetes in Trinidad and Tobago Rev Panam Salud Publica 1997 1 174 179 9128111
Merritt-Charles L Chen D Perera G Koppada A Hunte S Use of herbal medicines by surgical patients at the Eric Williams Medical Sciences Complex West Indian Med J 2003 52 23A 12806750
Mikail CN Hearney E Nemesure B Increasing physician awareness of the common uses and contraindications of herbal medicines: utility of a case-based tutorial for residents J Altern Complement Med 2003 9 571 576 14499033 10.1089/107555303322284866
Silverstein DD Spiegel AD Are physicians aware of the risks of alternative medicines? J Community Health 2001 26 159 174 11478563 10.1023/A:1010303528081
E/S/C/O/P Monographs: The scientific Foundation for Herbal Medicinal Products 2003 2 New York: Thieme
Blumenthal M ed The complete German Commission E monographs: Therapeutic guide to herbal medicine 1998 Austin, Texas: American Botanical Council
Robineau L ed Towards a Caribbean Pharmacopoeia TRAMIL 4 Workshop: Scientific Research and Popular Use of Medicinal Plants in the Caribbean 1989 Santo Domingo, Dominican Republic: Enda-caribe and Universidad Nacional Autónoma de Honduras
Raintree nutrition
Medicinal Plant database
Bland M An introduction to statistics 1995 Oxford: Oxford University Press
Central Statistical Office Population and Housing Census 1993 11 Office of the Prime Minister, Port of Spain, Trinidad
Williams EU History of the people of Trinidad and Tobago 1962 London: Andre Deutsh Ltd
Mischel F Faith healing and medical practice in the southern Caribbean Southwestern Journal of Anthropology 1959 15 407 417
Mahabir K Medicinal and Edible Plants Used by East Indians in Trinidad and Tobago 1991 El Dorado, Trinidad: Chanka Publishing House
St John L Remedies and Recipes of my ancestry 1997 Princes Town, Trinidad: Lystra Elder-St. John
Seaforth CE Adams CD Sylvester Y A guide to medicinal plants of Trinidad and Tobago 1983 London: Commonwealth Secretariat
Sand-Jecklin K Badzek L Nurses and nutraceuticals: Knowledge and use Journal of Holistic Nursing 2003 21 383 397 14692071 10.1177/0898010103258603
Chang ZG Kennedy DT Holdford DA Small RE Pharmacists' Knowledge and Attitudes Toward Herbal Medicine The Annals of Pharmacotherapy 2000 34 710 715 10860130 10.1345/aph.19263
Kemper KJ O'Connor KG Pediatricians' recommendation for complementary and alternative medical (CAM) therapies Ambul Pediatr 2004 4 482 487 15548098 10.1367/A04-050R.1
Aziz Z Herbal medicines: predictors of recommendation by physicians J Clin Pharm Ther 2004 29 241 246 15153085 10.1111/j.1365-2710.2004.00562.x
Kemper KJ Amata-Kynvi A Dvorkin L Whelan JS Woolf A Samuels RC Hibberd P Herbs and other dietary supplements: healthcare professionals' knowledge, attitudes and practices Altern Ther Health Med 2003 9 42 49 12776474
Giveon SM Liberman N Klang S Kahan E A survey of primary care physicians' perception of their patients' use of complementary medicine Complement Ther Med 2003 11 254 260 15022662 10.1016/S0965-2299(03)00114-6
Cohen RJ Ek K Pan CX Complementary and alternative medicine (CAM) use by older adults: a comparison of self-report and physician chart documentation J Gerontol A Biol Sci Med Sci 2002 57 M223 237 11909887
Barnes J Abbot NC Harkness EF Ernst E Articles on complementary medicine in the mainstream medical literature: an investigation of MEDLINE, 1966 through 1996 Arch Intern Med 1999 159 1721 1725 10448774 10.1001/archinte.159.15.1721
Phillipson JD 50 years of medicinal plant research – every progress in methodology is a progress in science Planta Med 2003 69 491 495 12865964 10.1055/s-2003-40656
Hall M Nazir N Quantification of the scientific research in the United States about popular herbal remedies referenced in PubMed Altern Ther Health Med 2005 11 34 37 15943130
Lee KH Current developments in the discovery and design of new drug candidates from plant natural product leads J Nat Prod 2004 67 273 283 14987069 10.1021/np030373o
Kemper KJ Amata-Kynvi A Sanghavi D Whelan JS Dvorkin L Woolf A Samuels RC Hibberd P Randomized trial of an internet curriculum on herbs and other dietary supplements for healthcare professionals Acad Med 2002 77 882 889 12228083
Frenkel M Ben-Arye E Hermoni D An approach to educating family practice residents and family physicians about complementary and alternative medicine Complement Ther Med 2004 12 118 125 15561521 10.1016/j.ctim.2004.07.044
|
16297236
|
PMC1310610
|
CC BY
|
2021-01-04 16:31:46
|
no
|
BMC Complement Altern Med. 2005 Nov 18; 5:20
|
utf-8
|
BMC Complement Altern Med
| 2,005 |
10.1186/1472-6882-5-20
|
oa_comm
|
==== Front
BMC Pregnancy ChildbirthBMC Pregnancy and Childbirth1471-2393BioMed Central London 1471-2393-5-151627766510.1186/1471-2393-5-15Research ArticlePrevalence of pre- and postpartum depression in Jamaican women Wissart Janice [email protected] Omkar [email protected] Santosh [email protected] Department of Basic Medical Sciences, Faculty of Medical Sciences, The University of the West Indies, Mona, Kingston 7, Jamaica (WI)2 Department of Obstetrics, Gynaecology and Child Health, Faculty of Medical Sciences, The University of the West Indies, Mona, Kingston 7, Jamaica (WI)2005 8 11 2005 5 15 15 13 7 2005 8 11 2005 Copyright © 2005 Wissart et al; licensee BioMed Central Ltd.2005Wissart et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Maternal depression during pregnancy has been studied less than depression in postpartum period. The aims of this study were to find out the prevalence of prepartum and postpartum depression and the risk factors associated in a cohort of Afro-Jamaican pregnant women in Jamaica.
Methods
The Zung self-rating depression scale instrument was administered to 73 healthy pregnant women at 28 weeks gestation and at 6 weeks postpartum for quantitative measurement of depression. Blood samples were collected at 8, 28, 35 weeks gestation and at day 1 and 6 weeks postpartum to study the thyroid status.
Results
Study demonstrated depression prevalence rates of 56% and 34% during prepartum and postpartum period, respectively. 94% women suffering depression in both periods were single. There were significant variations in both FT3 and TT4 concentrations which increased from week 8 to week 28 prepartum (p < 0.05) and then declined at the 35th week (p < 0.05 compared with week 28) and 1 day post delivery study (p < 0.05 compared with week 35). The mean values for TSH increased significantly from week 8 through week 35. The mean values at 1 day postpartum and 6 week postpartum were not significantly different from the 35 week values. For FT3, TT4 and TSH there were no significant between group differences in concentrations. The major determinants of postpartum depression were moderate and severe prepartum depression and change in TT4 hormone concentrations.
Conclusion
High prevalence of depression was found during pre- and postpartum periods. Single mothers, prepartum depression and changes in TT4 were factors found to be significantly associated with postpartum depression.
==== Body
Background
Maternal depression during pregnancy has been studied less than depression in postpartum period. A recent study, however, shows that depression seems to be as common during pregnancy as after delivery [1]. Depending upon various scales used to measure depression, the prevalence rate has been found to be between 5% and 26% for antenatal depression [2-6] and between 7% and 30% for postpartum depression [7-10]. The etiology of both pre- and postpartum depression still remains illusive. Various psycho-social and endocrine factors have been connected to both pre- and postpartum depression [2,3,11-14]. In addition, postpartum depression has been linked to a variety of endocrine root causes- especially postpartum thyroid dysfunction [15-17]. Antenatal depression has also been considered to increase the risk for postnatal depression [11,3]. This paper presents our findings on the prevalence of pre- and postpartum depression and risk factors responsible in a cohort of pregnant Afro-Jamaican women.
Methods
One hundred and forty (140) healthy Afro-Jamaican pregnant women attending antenatal clinic in the Department of Obstetrics, Gynecology and Child Health, University Hospital of the West Indies, Jamaica (WI) during the period May 2000 to February 2001 consented to participate in this study. Women with a history of thyroid disease, any medical illness, depression or substance abuse were excluded. The mean age was 27 years, parity varied from primigravidae to 4 live births and the mean length of gestation was 8 weeks at booking. A relatively large number of these mothers (67) had to be excluded (45 failed to keep their antenatal appointments, 5 had miscarriages, 4 had initial thyroid dysfunction, 7 had premature deliveries, 1 developed severe hypertension during pregnancy and 5 were delivered by cesarean section). Thus a total of 73 pregnant women completed this study.
This study was duly approved by the UWI/UHWI Ethics Committee.
Depression assessment
Each mother was given a questionnaire at booking to obtain a clinical profile and relevant demographic data. The Zung self-rating depression scale (SDS) questionnaire was administered at 28 weeks gestation and at 6 weeks postpartum. A participant was considered to have no psychopathy if Zung was <50, minimal to mild depression if Zung score was 50–59, moderate to marked depression if Zung score was 60–69, and severe to extreme depression if Zung score was 70 and over. Depending on the depression status, the cohort was subdivided into four subgroups (i) women depressed during prepartum only (ii) women depressed prepartum and who continued to be depressed postpartum (iii) women depressed postpartum only and (iv) women showing no signs of depression in either period.
Hormonal profile
Blood samples were collected at 8, 28, 35 weeks of gestation and at 1 day and six weeks after childbirth. Serum total thyroxine (TT4), free triiodothyronine (FT3), thyrotropin (TSH) were determined using standard radioimmunoassay kits (Diagnostic Products Corporation, Los Angeles, California, USA). The sensitivity value for each assay was 0.25 pg/dl; 0.2 pg/ml; and 0.03 :IU/ml, respectively. All samples for each test were assayed in the same batch.
Statistical analysis
Data are expressed as means ± SE or counts as appropriate. The data were analyzed by repeated measures analysis of variance (RMANOVA) with the between group factor being the four subgroups (i) women depressed during prepartum only (ii) women depressed prepartum and who continued to be depressed postpartum (iii) women depressed postpartum only and (iv) women showing no signs of depression in either period and the measurements done over time (8 wks 28 wks 35 wks 1 day postpartum and 6 wks postpartum) as the repeated measures factor. In analyses where there were significant interactions between the group factor and the repeated measures factor, we compared differences between the depression categories at each experiment and differences between the means of measured variables within each clinical category at each experiment, by Tukey method. In these comparisons the error term was the root mean square error from the RMANOVA analysis with its associated degrees of freedom.
The Zung self-rating depression scale (SDS) was used to classify participants into depression categories. For this analysis these categories were treated as an ordinal scale. To assess effects of changing thyroid hormones on the odds of changing depression categories postpartum we used only the measurements done at 28 wk prepartum and 6 weeks postpartum. We assessed the odds of being in a particular depressed category in the postpartum period for participants using an ordinal logistic model adjusting for prepartum depression category and differences in thyroid hormones level (6 wk postpartum – wk 28 prepartum values) as covariate. Inferential tests were considered statistically significant if p < 0.05 (two tail).
Data analysis was performed using Stata version8 for Windows (Statacorp, College station, TX).
Results
Pre- and postpartum depression
Self-rating depression scale administered to 73 women indicated that 41 women (56.16%) were having depression at 28 weeks prepartum. Out of these 41, 23 (31.5%) had mild, 13 (17.8%) had moderate and 5 (6.9%) had severe depression. Out of these 41, only 18 (24.66%) women were found to be depressed at 6 weeks postpartum. Remaining 23 (31.5%) women recovered anywhere between 28 weeks prepartum and 6 weeks postpartum period. The data further indicated that 4 out of the 5 women who were severely depressed at 28 weeks prepartum were only moderately depressed at 6 weeks postpartum; and the remaining had no depression. In addition, 7 (9.59%) different women were found to be depressed only at 6 weeks postpartum. Therefore, a total of 25 (34.25%) women suffered mild to marked depression during the postpartum period (Table 1).
Table 1 Results of Zung self-rating depression scale (SDS) questionnaire administered to 73 women at 28 weeks of gestation and at 6 weeks postpartum.
SDS Index Equivalent Clinical Global Impression 28 weeks Gestation Number of Cases 6 weeks Postpartum Number of cases
Below 50 Normal range, no psychopathology 32 (43.8%) 48 (65.75%)
50–59 Presence of minimal to mild depression 23 (31.5%) 18 (24.66%)
60–69 Presence of moderate to marked depression 13 (17.8%) 7 (9.59%)
70 and over Presence of severe to extreme depression 5 (6.9%) None
Demography and depression
Analysis of the data on marital status demonstrated that 31 (75.6%) out of 41 women who were depressed at 28 weeks prepartum were single or legally not married. 17 (94.4%) out of 18 women who continued to have depression at 6 weeks postpartum were also single. Similarly, 5 (71.4%) out of 7 women who suffered depression in postpartum period only, were single. Age, parity, miscarriages and employment status was not associated with depression (Table 2).
Table 2 Demographic characteristics by depression groups
Variables Never Depressed
N = 25 Prepartum Depression only
N = 23 Prepartum & Postpartum Depression
N = 18 Postpartum Depression only
N = 7
Mean Age (Yrs) 27 27 25.9 28.4
Single 14(56%) 14(61%) 17(94%) 5 (71%)
Primigravidae 13 (52%) 11 (48%) 9 (50%) 3 (43%)
Multiparous 12 (48%) 12 (52%) 9 (50%) 4 (57%)
Employed 15 (60%) 15 (65%) 8 (44%) 6 (86%)
Unemployed 10 (40%) 8 (35%) 10 (56%) 1 (14%)
Thyroid status and depression
There were significant changes in the mean thyroid hormones concentrations over time (Fig 1). For both FT3 and TT4, mean values increased from week 8 to week 28 prepartum (p < 0.05) and then declined at the 35th week (p < 0.05 compared with week 28) and 1 day post delivery study (p < 0.05 compared with week 35). There was a small increase in the mean FT3 values at 6 weeks postpartum compared to the 1 day postpartum values but this was not statistically significant. In contrast, the mean values for TT4 decreased further compared to the 1 day postpartum values (p < 0.05). The mean values for TSH increased significantly from week 8 through week 35. The mean values at 1 day postpartum and 6 week postpartum were not significantly different from the 35 week values. For FT3, TT4 and TSH there were no significant between group differences in concentrations. However there was a significant group by study interaction for FT3. At weeks 8 and 28 prepartum, the women who were never depressed tended to have higher mean FT3 values compared to women who were depressed. At week 35, the mean values for FT3 of women who were never depressed were significantly lower (p < 0.05) than mean FT3 values for women who were depressed prepartum.
Figure 1 Thyroid hormones by depression groups. * Significantly different from previous values p < 0.05.
There were no significant relationships between thyroid hormone concentrations and Zung scores at 28 weeks or at 6 week post partum. The major determinants of postpartum depression were moderate and severe prepartum depression and change in TT4 hormone concentrations (Table 3). Thus the odds of being depressed postpartum increased by factors of 4.8 in the moderately depressed category and 221 in the severely depressed category relative to not being depressed respectively adjusting for changes in TT4. For every unit increase in the magnitude of the difference in mean TT4 values between week 28 prepartum and 6 week postpartum the odds of being more depressed increases by a factor of 0.75 (p < 0.003).
Table 3 Prepartum determinants of a more severe depression post partum category relative to less severe depression postpartum category
Prepartum Depression Category Factor change in odds* p value
Mild 1.7 ns
Moderate 4.8 <0.002
Severe 221 <0.001
Difference in T4 values 0.75 <0.003
* Relative to not depressed category
Discussion
The study demonstrated a higher prevalence rate of 56% maternal depression at 28 weeks prepartum as compared to 34 % at 6 weeks postpartum confirming that depression during pregnancy is as common as after delivery [1]. Our prevalence rates for both pre- and postpartum depression are higher than the rates reported earlier [1-3,7,8,10]. This may be attributed to the fact that greater percentages of women during both periods suffered from minimal to mild depression only, a score that might have been ignored by previous authors, who may have regarded it as clinically non-significant depression. Further analysis of the data on postpartum depression showed that prepartum depression was the major risk factor for postpartum depression [3,18].
Though the postnatal depression was observed only at 6 weeks postpartum, it is possible that the onset could have been within the first week after the delivery as reported earlier [9].
This study revealed that single mothers are more prone to depression both during pre- and postpartum periods. This can be interpreted in two ways (i) either that being single implies lack of a stable supporting union which influences depression and/or (ii) there is a predominance of single mothers in Jamaica (50 of the 73 mothers in this study were single). In both circumstances, whatever the cause of depression, the majority of subjects would be single. Employment, parity and the previous miscarriages did not seem to have any effect in producing depression.
In the sub-group of women representing 9.6% of the cohort, who developed depression in the postpartum period, relative hypothyroidism was observed during the late gestation and early postpartum periods. In addition, changes in mean TT4 levels were significantly related to postpartum depression. This finding is supported by previous studies [15-17] that suggest that postpartum thyroid dysfunction may be responsible for postpartum depression. However, this is in contrast to the study by Lucas et al [19] that has reported no link between postpartum thyroid dysfunction and postpartum depression.
Conclusion
The results of our study limited only to corporate area, suggested almost equally high incidence of depression in single Jamaican mothers both during pre- and postpartum periods. In addition, relative hypothyroidism developed between late gestation and postpartum period could have been responsible for postpartum depression in a sub-group of mothers. In the light of the results, it is suggested that women who develop depression during pregnancy should be monitored for thyroid functions and social support be provided to single mothers to avoid the risk of postpartum depression. A wider cross- sectional study in Jamaica is further needed to confirm these results.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JW M.Phil. Student who carried out this work.
OP chief supervisor, conceived and contributed to study design, reviewed the statistical analysis, interpretation of results and preparation of manuscript.
SK provided access to the subjects in the hospital, supervisory committee member, provided input on pregnancy matters.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
We thank Dr. Marvin Reid, Director, Sickle Cell Unit, TMRI, UWI, Mona, Kingston 7, Jamaica for his assistance in data analysis.
==== Refs
Evans J Heron J Francomb H Oke S Golding J Cohort study of depressed mood during pregnancy and after child birth BMJ 2001 323 257 260 11485953 10.1136/bmj.323.7307.257
Kitamura T Sugawara M Sugawara K Toda MA Shima S Psychosocial study of depression in early pregnancy Br J Psychiatry 1996 168 732 738 8773816
Johanson R Chapman G Murray D Johnson I Cox J The North Staffordshire Maternity Hospital prospective study of pregnancy associated depression J Psychosom Obstet Gynaecol 2000 21 93 97 10994181
Pajulo M Savonlahti E Sourander A Helenius H Piha J Antenatal depression, substance dependency and social support J Affect Disord 2001 65 9 17 11426516 10.1016/S0165-0327(00)00265-2
Luoma I Tamminen T Kaukonen P Laippala P Purra K Salmelin R Longitudinal study of maternal depressive symptoms and child well-being J Am Acad Child Adolesc Psychiatry 2001 40 1367 1374 11765281 10.1097/00004583-200112000-00006
Moses-Kolko EL Roth EK Antepartum and postpartum depression: healthy mom, healthy baby J Am Med Womens Assoc 2004 59 181 191 15354371
Da Costa D Larouche J Drista M Brender W Psychosocial correlates of prepartum and postpartum depressed mood J Affect Disord 2000 59 31 40 10814768 10.1016/S0165-0327(99)00128-7
Morris-Rush JK Freda MC Bernstein PS Screening for postpartum depression in an inner-city population Am J Obstet Gynecol 2003 188 1217 1219 12748483 10.1067/mob.2003.279
Yamashita H Yoshida K Screening and intervention for depressive mothers of new-born infants Seishin Shinkeigaku Zasshi 2003 105 1129 1135 14639935
Bloch M Daly RC Rubinow DR Endocrine factors in the etiology of postpartum depression Compr Psychiatry 2003 44 234 246 12764712 10.1016/S0010-440X(03)00034-8
Areias ME Kumar R Barros H Figueiredo E Correlates of postnatal depression in mothers and fathers Br J Psychiatry 1996 69 36 41 8818366
Glasser S Barell V Boyko V Ziv A Lusky A Shoham A Postpartum depression in an Israeli cohort; demographic, psychosocial and medical risk factors J Psychosom Obstet Gynaecol 2000 21 99 108 10994182
McCoy SJ Beal JM Watson GH Endocrine and postpartum depression. A selected review J Reprod Med 2003 48 402 408 12856509
Hendrick V Altshuler LL Suri R Hormonal changes in the postpartum and implications for postpartum depression Psychosomatics 1998 39 93 101 9584534
Amino N Tada h Hidaka Y Screening for postpartum thyroid dysfunction in the general population is beneficial J clin Endocrinol Metab 1999 84 1813 1816 10372667 10.1210/jc.84.6.1813
Kent GN Stuckey BGA Allen JR Lambert T Gee V Postpartum thyroid dysfunction: clinical assessment and relationship to psychiatric morbidity Clin Endocrinol 1999 51 429 438 10.1046/j.1365-2265.1999.00807.x
Barca MF Knobel M Tomimiri E Cardia MS Medeiros-Neto G Prevalence and characteristics of postpartum thyroid dysfunction in Sao Paulo Brazil Clin Endocrinol 2000 53 21 31 10.1046/j.1365-2265.2000.01034.x
Heron J O'Connor TG Evans J Golding J Glover V The ALSPAC Study Team The course of anxiety and depression through pregnancy and the postpartum in a community sample J Affect Disord 2004 80 65 73 15094259 10.1016/j.jad.2003.08.004
Lucas A Pizarro E Granada ML Salinas I Sanmarti A Postpartum thyroid dysfunction and postpartum depression: are they two linked disorders Clin Endocrinol 2001 55 809 814 10.1046/j.1365-2265.2001.01421.x
|
16277665
|
PMC1310611
|
CC BY
|
2021-01-04 16:32:05
|
no
|
BMC Pregnancy Childbirth. 2005 Nov 8; 5:15
|
utf-8
|
BMC Pregnancy Childbirth
| 2,005 |
10.1186/1471-2393-5-15
|
oa_comm
|
==== Front
BMC Health Serv ResBMC Health Services Research1472-6963BioMed Central London 1472-6963-5-741630767910.1186/1472-6963-5-74Research ArticleWhat factors explain the number of physical therapy treatment sessions in patients referred with low back pain; a multilevel analysis Swinkels Ilse CS [email protected] Raymond H [email protected] Peter P [email protected] den Bosch Wil JH [email protected] Joost [email protected] den Ende Cornelia HM [email protected] NIVEL Netherlands Institute for Health Services Research; PO Box 1568, 3500 BN Utrecht, the Netherlands2 Department of General Practice, Radboud University Medical Centre Nijmegen; Nijmegen, the Netherlands3 Department of Rehabilitation Medicine, Institute for Research in Extramural Medicine, VU University Medical Centre Amsterdam, Amsterdam, the Netherlands2005 24 11 2005 5 74 74 10 5 2005 24 11 2005 Copyright © 2005 Swinkels et al; licensee BioMed Central Ltd.2005Swinkels et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
It is well-known that the number of physical therapy treatment sessions varies over treatment episodes. Information is lacking, however, on the source and explanation of the variation. The purposes of the current study are: 1) to determine how the variance in the number of physical therapy treatment sessions in patients with non-specific low back pain (LBP) in the Netherlands is distributed over patient level, therapist level and practice level; and 2) to determine the factors that explain the variance.
Methods
Data were used from a national registration network on physical therapy. Our database contained information on 1,733 patients referred with LBP, treated by 97 therapists working in 41 practices. The variation in the number of treatment sessions was investigated by means of multilevel regression analyses.
Results
Eighty-eight per cent of the variation in the number of treatment sessions for patients with LBP is located at patient level and seven per cent is located at practice level. It was possible to explain thirteen per cent of all variance. The duration of the complaint, prior therapy, and the patients' age and gender in particular are related to the number of physical therapy treatment sessions.
Conclusion
Our results suggest that the number of physical therapy treatment sessions in patients with LBP mainly depends on patient characteristics. More variation needs to be explained, however, to improve the transparency of care. Future research should examine the contribution of psychosocial factors, baseline disability, and the ability to learn motor behavior as possible factors in the variation in treatment sessions.
==== Body
Background
It is well-known that the number of physical therapy treatment sessions varies over treatment episodes [1-10] and it is important for health care policy makers, physical therapists and patients to gain greater understanding of the sources of this variation. Greater understanding will increase the transparency of care and can provide novel insights into the quality of care. On grounds of equity, an 'ideal' situation is one where health status is the main determinant of treatment choice and hence of variation. As a consequence, the variation is appropriate when it occurs due to 'need' factors like the patient's clinical health status [11], but it is questionable whether the variation is appropriate when it occurs due to factors like social structure, health beliefs, or enabling resources (such as accessibility). Elimination of inappropriate variation is necessary for quality improvement in physical therapy practice [12] and it is important to know exactly where variance is located if proper quality measures are to be implemented. The variance may be on different levels, including patient level, therapist level and practice level.
Few investigations have been made as yet into the reasons for the variation in the number of treatment sessions [5,13,14] and none of these distinguished between variation at patient level, variation at therapist level and variation at practice level. Hendriks et al. (2000) showed that the therapist's age, a specialization in manual therapy, and practice size were associated with fewer treatment sessions [5], but it remained unclear to what extent the amount of variation was explained and how it was distributed over the different levels. Other studies showed that the patient's age [5,14], the duration of the complaints [14], the therapist's diagnostic findings, the medical diagnosis [14], and additional claims for other health care services [13] were positively related to the number of treatment sessions. Information on the amount of variation at different levels is lacking in the above-mentioned studies and much of the variation remains unexplained, which means that these studies do not fulfill the need for clarification of variation in utilization of physical therapy. To the knowledge of the authors, the current study is the first one in which different levels are taken into account to estimate not only the variation, but also its location.
The aims of the current study are as follows: 1) to determine how the variance in the number of physical therapy treatment sessions in patients with non-specific low back pain (LBP) is distributed over patient level, therapist level and practice level; and 2) to determine the factors that explain the variation, with factors relating to all three levels being taken into account. We addressed our research questions to patients with LBP, since they form the largest population in physical therapy practices.
Methods
Registration network: sampling
Data from the National Information Service for Allied Health Care (called LiPZ in Dutch) are used for the current study. The National Information Service for Allied Health Care is a registration network of Dutch physical therapists working in private practices all over the country, and this network has been collecting health care-related data on a continuous basis since 2001. Physical therapists were invited to participate in the registration network in early 2001, the selection of therapists being based on practice-size and region. The therapists invited to take part were a sample of all private physical therapy practices as listed in a national database [15]. Our objective was a registration network of 40 practices and therapists could only participate if one of two specific software programs was used in their practice. Physical therapists with a homogeneous patient population (> 50% of the treatment episodes consisting of one patient category, such as children) were excluded from the network. Twenty per cent of the therapists invited to participate were eligible to take part. Frequently mentioned reasons for not participating were 'not enough time' and 'personal reasons'. When dropouts occurred, an a-selective procedure was used to invite new physical therapists to participate. A response rate of 20% is acceptable, considering the kind of research, since a long-term commitment and a computerized practice are factors that lower the response rate. Despite the relatively low response rate, comparisons with other available data show that the participating practices, therapists, and data collected appear to be representative for the Netherlands [16-18]. Over 140 physical therapists working in more than 60 practices have participated since 2001. Participants are offered some financial remuneration and they also receive benchmark data on an annual basis. Relevant information on the Dutch health care system is provided in Table 1.
Table 1 The organization of physical therapy in The Netherlands
In the Dutch health care system, physical therapists are only accessible after referral by a physician and over 90% of patients attending a physical therapist have been directly referred by their GP. The remaining 10% are referred by a medical specialist. People in the Netherlands have either public or private health insurance, depending on their level of income. Public insurance cover for physical therapy is nationally regulated and in 2002 and 2003 this meant that people with public insurance (66% of the population) and low back pain were covered for 9 treatment sessions per episode per year. People with public insurance were able to obtain additional private insurance that covered them for more than 9 treatment sessions. Private insurance cover (the other third of the population) for physical therapy was not regulated at national level. Every physical therapy session lasts about 25 minutes and physical therapists are paid per session, irrespective of the type of diagnosis and intervention. In the Netherlands, nearly all therapists working in primary care are organised in private practices.
The Dutch situation will change in 2006; the differentiation between public and private health care insurances will disappear and physical therapy will be accessible without a referral.
Registration network: methods
Dutch therapists in private practice generally use a software program to record their patients and treatments, and for reimbursement. In addition to the information regularly recorded, therapists participating in the network use special software to record supplementary information on all their patients. The selection of the data was based on the Dutch physical therapy guideline for clinical reporting, a guideline that specifies the data that are relevant for physical therapy practices. Participants submit their data on a monthly basis and the data are entered in the database after standardized quality control has been performed to check for missing or inconsistent data. The practice receives feedback in the case of missing or inconsistent data, and corrected data are entered in the database in the next month.
A written questionnaire, completed annually by all participants, provides information on characteristics of the participating therapists and practices and also includes questions about the attitude towards quality improvement. The feasibility of the Dutch LBP guideline [19] is specifically addressed. The question "Could you please rate your opinion of the feasibility of the Low Back Pain guideline? The rating can range from 0 (very bad) to 10 (excellent)" was used as indicator of the attitude of physical therapists towards the physical therapy guideline for Low Back Pain. All relevant variables collected are listed in Table 2.
Table 2 Overview of data collection
Variables Measurement Used in analyses as
Demographic
Gender Male; Female Categorical
Age Date of birth Continuous: years old at start treatment episode
Health insurance Public health insurance (Puhi), private health insurance (Prhi) Categorical: Puhi; Prhi; Unknown
Education Highest level of education: Primary school, secondary education, higher education, university Categorical: Low (primary); Middle (secondary, higher); High (university); Unknown
Urbanization rate 1 very high, 2 high, 3 moderate, 4 low, 5 very low Categorical: High (3+2+1); Low (5+4); Missing values (1.3%) recoded as high urbanization rate
Complaints
Specialization of referring physician GP or Medical specialist Categorical: GP; Medical specialist
Reason for referral As given by letter by the referrer; coded with ICPC (26) by researchers Selection of patients with ICPC-code L03.00
Duration of complaint at start episode < 2 days; 2–7 days; 1 week – 1 month; 1–3 months; 3–6 months; 6 months – 1 year; 1–2 years; > 2 years; unknown Categorical: < 1 month; > 1 month; Missing values (1.0%) recoded as < 1 month
Recurrent complaint (appearing after a complaint-free episode of at least four weeks and at most two years) Yes; No; Unknown Categorical: Yes; No; Missing values (3.7%) recoded as no
Previous physical therapy for the same or other complaints in the last two years Yes; No; Unknown Categorical: Yes; No; Missing values (6.1%) recoded as no
Treatment
Treatment goals Based on the International Classification of Functioning, Disability and Health (27); One main goal (out of 24) at the level of physical functions; One main goal at the level of activities (out of 11) 5 dichotomous variables; Missing values (0.6%) recoded as changing body position
Interventions Based on Dutch classification; three interventions at most (out of 25) applied in at least 50% of the sessions 5 dichotomous variables; Missing values (16.1%) integrated in reference category
Therapists
Gender Male; Female Categorical: Male; Female; Missing values (2.1%) recoded as male
Age Date of birth Categorical: <45 years at January 1st, 2003; > 45 years at January 1st, 2003; Missing values (5.1%) recodes as > 45 years
Hours working per week Patient-related number of hours Categorical: <20 hours; 20–40 hours; >40 hours; Missing values (4.1%) recoded as 20–40 hours
Registration in quality register for manual therapy Yes; No Categorical: Yes; No
Additional training in LBP Yes; No Categorical: Yes; No; Missing values (4.1%) recoded as yes
Additional training in LBP guideline Yes; No Categorical: Yes; No; Missing values (4.1%) recoded as no
Feasibility of LBP guideline 1 item on questionnaire 10-point scale (1 = very bad; 10 = excellent; 7 = satisfactory) Categorical: <7 points; >7 points; Unknown
Time since graduation Date of graduation Categorical: < 20 years since graduation; > 20 years since graduation; Missing values (6.2% recodes as > 20 years.
Practices
Group size Number of therapists Categorical: Single handed; Group practice
Study sample
Data from therapists who treated patients referred with non-specific LBP during the period July 2002-September 2003 were selected from the database for the current study; these data were supplied by 97 therapists in 41 practices. The therapists treated an average of 1.6 new patients with LBP per month (average in a 30-hour week). Twenty-four per cent of the 41 participating practices were solo practices (Table 3). The majority of the physical therapists were male; the mean age of the therapists was 43.5 years (sd 9.3). The therapists selected did not differ significantly from all Dutch physical therapists.
Table 3 Characteristics of therapists (n = 97) and practices (n = 41) in the sample and in the Netherlands (12,695 therapists)
Sample Dutch population of physical therapists [15]
% (N) Mean (SD) % (N) P
Physical therapist Male 57.7 (56) 50.1 (6,359) 0.14
< 45 years 50.5 (49) 56.4 (7,049) 0.32
Registration quality register – manual therapy 12.4 (12)
Number of patient-related hours per week
< 20 hours 23.7 (23)
20–40 hours 61.9 (60)
> 40 hours 14.4 (14)
Additional training LPB 58.8 (57)
Additional training LBP guideline 39.2 (38)
Feasibility of the LBP guideline
< 7 points 39.2 (38)
> 7 points 43.3 (42)
Unknown 17.5 (17)
Number of new patients with LBP per therapist per month 1.6 (1.1)
Practice Single-handed 24.4 (10)
Where the patient population is concerned, all patients aged 18 years or older who had been referred with LBP without radiation (ICPC-code L03.00) between July 2002 and September 2003 were selected from the database (n = 1,760). Patient data were collected until April 2004, at which time 1,733 of these 1,760 patients had a completed treatment episode (98.5%). Data relating to these 1,733 patients were used in the current study.
Ethical approval was not required, since patients only received the customary care and there were no experimental interventions for the purposes of the present study. Patients were nevertheless informed about the research project by posters and leaflets in the waiting rooms in the practices and patients could refuse to participate.
Outcome variable and predictor variables
The outcome variable was the total number of treatment sessions per treatment episode. This variable was used as a continuous variable.
The predictor variables are listed in Tables 3 and 4. Age, gender, education level, health care insurance and urbanization rate are included as demographic variables. Variables relating to the complaints are also included, viz. the duration of the complaints, recurrent complaints, prior physical therapy or exercise therapy, and specialization of the referring physician. An interaction term consisting of gender and duration of the complaints was also added, since the gender distribution in patients with acute complaints was not equal to that in patients with chronic complaints. Treatment variables included variables on the treatment goals and the interventions. At the start of a treatment episode, therapists indicated one main treatment goal from a list of 11 predefined goals at activities level and/or one main treatment goal from a list of 24 predefined goals at physical functions level. Five treatment goals that were indicated in more than ten percent of the patients are included in the analyses as dichotomous variables. At the end of the treatment episode, physical therapists recorded a maximum of three interventions (from a list of 25 predefined interventions) that were applied in at least 50% of the treatment sessions. Interventions recorded in more than ten percent of the patients are included in the analysis as dichotomous variables (n = 5). Variables relating to gender, age, working hours per week, additional training in LBP and additional training in guideline-use for patients with LBP, the feasibility of the guideline LBP, registration in the quality register for manual therapy and group size were included at therapist and practice levels. Table 4 provides an overview of the characteristics of the variables at patient level.
Table 4 Characteristics of patients (n = 1,733)
% (No) Mean (SD)
Demographic
Age in years 48.7 (16.3)
Male 45.2 (783)
Education Low 30.8 (534)
Middle 26.3 (456)
High 13.5 (234)
Unknown 29.4 (509)
Health insurance Public 56.7 (983)
Private 24.2 (420)
Unknown 19.0 (330)
High urbanization rate (> 1,000 addresses per km2)1 58.9 (1,021)
Complaints
Duration complaint < 1 month (acute) 48.2 (835)
> 1 month (chronic) 51.8 (898)
Recurrent complaint2 47.0 (815)
Previous physical therapy3 47.1 (817)
Referred by general practitioner 95.2 (1,650)
Treatment
Treatment goal Maintaining body position (yes) 18.1 (313)
Changing body position (yes) 19.0 (329)
Functions of mobility (yes) 39.6 (687)
Functions of muscles (yes) 14.3 (247)
Pain (yes) 11.4 (197)
Interventions Massage (yes) 34.4 (596)
Manual manipulation (yes) 37.9 (657)
Physical modalities (yes) 12.0 (208)
Exercise therapy (yes) 65.8 (1,141)
Information/advice (yes) 27.1 (469)
Number of treatment sessions 9.9 (6.6)
1 [27]
2 recurrent complaint is defined as a complaint appearing after a complaint-free episode lasting at least four weeks and at most two years
3 for the same or other complaint
Data analysis
Descriptive statistics were calculated for the characteristics of the patients, the therapists, and the practices, and for the number of treatment sessions per treatment episode. Data were aggregated at the level of treatment episodes. Software-program SPSS 11.5 was used for the descriptive analysis. Missing value analyses showed four categorical variables with over 10% missing cases and a category designated as "unknown" was added for those variables. In the case of the other variables, the missing values were recoded to the mean (continuous variables) or most frequent value (categorical variables).
Data were analyzed by means of multilevel regression analysis to determine which variables were associated with the number of treatment sessions per treatment episode. Multilevel analysis was used because the data had an intrinsically hierarchical nature; the patients (level 1) are nested in the sample of physical therapists (level 2), who are nested in physical therapy practices (level 3). The data were not based on independent observations, therefore, which violates a major assumption of traditional regression analysis. Multiple levels are taken into account in multilevel analysis and variation can be split between levels.
Bivariate correlations between all predictor variables were examined to check for high correlations before starting the multilevel analysis. The therapists' age and the time since their graduation showed a correlation of 0.80 and so only the therapists' age was included in the analysis.
The multilevel analysis was carried out using MLwiN 1.1 software. The order of adding predictor variables to the model was determined by their level, as described above.
The analysis was carried out in 2 steps. An "intercept-only model" was made first. This is a model without any predictor variables, which establishes the contribution of each level to the variation in the number of treatment sessions. In the next step, all predictor variables were added. The multilevel analysis was done with three dependent variables: viz. the raw number of sessions, a log-linear transformation and a dataset in which the extreme values had been left out.
Indicator coding was used for categorical predictor variables, with the first category in each group treated as the reference group. The continuous predictor variables "patient's age" and "number of patients per therapist per month" were centered around their mean. The contribution of each predictor variable was expressed in a regression coefficient (B) and a standard error (SE). If their quotient is greater than 1.96 or smaller than -1.96, the coefficient is statistically significant (level of significance is 0.05) [20].
Results
The three different analyses yielded similar results. Since analyses containing log-transformation will be difficult for the reader to interpret, only the results on the raw number of sessions will be shown.
Number of treatment sessions per treatment episode
The mean number of physical therapy treatment sessions in patients referred with non-specific LBP was 9.9 (SD 6.6; median 9.0; minimum 1; maximum 67).
Variance components in intercept-only model
As shown in Table 5, most of the variance in treatment sessions was located among patients (88.4%); 4.4% of the total variance was located among therapists and 7.2% was located among practices. The mean number of treatment sessions, adjusted for therapists and practices, was 10.0. Using the intercept and the variance component at practice level, the mean number of treatment sessions in 95% of the practices was calculated to be between 6.6 and 13.4.
Table 5 Distribution of variation in the number of physical therapy treatment sessions in patients with non-specific LBP among different levels (practices, therapists, and patients). Results of the intercept-only model (n = 1,733)
(SE) % P
Intercept 10.03 (0.37)
Deviance 11,299.52
Variance
Practice level 3.06 (1.28) 7.2% 0.016
Therapist level 1.88 (0.91) 4.4% 0.038
Patient level 37.75 (1.314) 88.4% <0.001
Total 42.70 100.0%
Contribution of predictor variables in the final model
The contribution of the various predictor variables in the last step of the analyses is expressed in regression coefficients and standard errors in Table 6.
Table 6 Hierarchical regression analysis of predictors of the number of physical therapy treatment sessions in patients with non-specific LBP (n = 1,733)
B (SE)
Intercept 9.300 (1.28)
Patient level
Age (years) ***0.04 (0.01)
Female (ref. Male) ***1.90 (0.42)
Education level: Middle (ref. low) 0.62 (0.40)
High (ref. low) -0.76 (0.52)
Unknown (ref. low) -0.28 (0.42)
Health insurance: Private (ref. public) *-0.84 (0.38)
Unknown (ref. public) 0.34 (0.42)
High urbanization rate (ref. low) 0.10 (0.53)
Complaint level
Chronic complaints (ref. acute) ***2.27 (0.44)
Female*chronic complaints **-1.79 (0.59)
Recurrent complaint (ref. no) -0.34 (0.34)
Previous therapy (ref. no) ***1.17 (0.35)
Referral by medical specialist (ref. GP) ***4.18 (0.77)
Treatment level
Treatment goal Maintaining body position 0.09 (0.49)
Changing body position 0.31 (0.46)
Functions of mobility 0.08 (0.43)
Functions of muscles 0.32 (0.51)
Pain 1.23 (0.64)
Interventions Massage 0.71 (0.37)
Manual manipulation -0.38 (0.36)
Physical modalities *1.13 (0.48)
Exercise therapy **1.03 (0.35)
Information/advice -0.33 (0.38)
Therapist and practice level
Female (ref. male) *-1.23 (0.57)
Aged > 45 years (ref. < 45 years) ***-2.01 (0.51)
Manual therapist (ref. no) *-1.44 (0.60)
Patient-related working hours per week 20–40 (ref. < 20) -0.48 (0.61)
> 40 (ref. < 20) *1.80 (0.87)
Additional training in LBP (ref. no) **-1.47 (0.51)
Additional training in LBP guideline (ref. no) 0.39 (0.47)
Feasibility LBP guideline > 7 (ref. < 7) 0.09 (0.48)
Unknown (ref. < 7) -0.31 (0.65)
Number of LBP patients per therapist per month -0.14 (0.18)
Group practice (ref. single-handed) 0.14 (0.80)
Deviance 11,106.46
* = P < 0.05; ** = P < 0.01; *** = P < 0.001
The influence of the characteristics with regard to the complaints appeared to be most powerful when all predictor variables were included for hierarchical linear regression analysis. Three out of four variables were related to the number of treatment sessions. Patients with sub-acute or chronic complaints received 2.3 sessions more compared to patients with acute complaints when all other variables were held constant; patients who were referred by a medical specialist received 4.2 sessions more compared to patients referred by a general practitioner; and patients who had prior therapy for the same or other complaints received 1.2 sessions more compared to patients who did not have prior therapy.
Demographic variables also had a statistically significant relationship to the number of treatment sessions. Older patients, female patients, and patients with public health insurance were treated more often than other patients. The level of education did not have a statistically significant relationship to the number of treatment sessions when all other predictor variables were controlled.
Treatment goals did not have a statistically significant relationship to the number of physical therapy treatment sessions. Two out of five interventions did show an association with the number of treatment sessions; patients in whom exercise therapy or physical modalities are part of the treatment are treated in one session more than other patients.
Although most of the variance was located among patients, characteristics of the therapists were also shown to be related to the number of treatment sessions. Patients treated by a manual therapist received 1.4 sessions fewer than patients treated by other physical therapists. Therapists with additional training in LBP treated their patients in 1.5 sessions less than therapists without additional training. Female therapists and older therapists treated their patients in fewer sessions than younger and male therapists. Finally, therapists working more than 40 hours a week treated their patients in more sessions than therapists working less than 20 hours a week.
Explained variance in final model
Compared to the intercept-only model, the final model explained 13.4% of the variance (Table 7); 8.7% was explained at patient level, where most of the variance was located. The variance at practice level decreased by 22.2%, while the variance at therapist level disappeared almost entirely (decrease 93.2%). In the final model, in which all predictor variables were added to the model, 93% of the variance was located among patients (not in table); the remaining variance was mainly located among practices.
Table 7 Distribution of variation in the number of physical therapy treatment sessions in patients with non-specific LBP among different levels (practices, therapists, and patients). Results of step two in the analyses (n = 1,733)
Variance (SE) % of explained variance in relation to the intercept-only model
Practice level 2.38 (0.83) 22.2
Therapist level 0.13 (0.43) 93.2
Patient level 34.48 (1.20) 8.7
Total 36.98 13.4
Discussion
This study confirms that there is substantial variation in the number of physical therapy treatment sessions for patients with LBP and most of this variance is located among patients. A combination of various factors explains 13.4% of the variance in the number of physical therapy treatment sessions.
To our knowledge, this is the first study in which the variation in the number of physical therapy treatment sessions for LBP among patients, therapists and practices has been estimated simultaneously. The findings have major implications for the quality of care agenda in physical therapy.
Most of the variance by far is located at patient level. Demographic factors and factors relating to the complaints explained the major part of the variance, compared with factors relating to the treatment and the therapists. The positive association between the patient's age and the number of treatment sessions is in accordance with the literature [5,14], as is the effect of the patient's gender [14]. The duration of the complaint, prior therapy, and the specialization of the referrer are also related to the number of treatment sessions. Although there might be other explanations as well, this finding is in agreement with the assumption that these factors are related to the severity of the complaint. On grounds of equity, it is appropriate that the severity of the complaint is related to the number of physical therapy treatment sessions. The same is true of the relationship between the interventions and the number of treatment sessions, since it has been suggested that the contents of care are related to the severity of the complaints [2,7]. Jette et al. (1996) were able to show that outcomes were associated with the use of some types of physical therapy treatment in patients with spinal impairments [8]. As the outcome of care was not investigated in the current study, it might be interesting to carry out further investigation into the relationship between the content of the treatment, the number of treatment sessions and the outcome.
Factors at therapist level, such as their age, gender and specialty, were also associated with the number of treatment sessions, as were demographic factors and factors relating to the treatment and the complaints. It is questionable whether associations with factors at therapist level are desirable. It is suggested in the literature that practice style differences flourish in an environment of professional uncertainty [21,22]. The Dutch physical therapy guideline for LBP was published in 2001 to reduce professional uncertainty [19]. The effects of this publication on physical therapy practice might not be completely visible, since our results are based on data from patients treated between 2002 and 2004. The corresponding variable, however, does not show a relationship to the number of treatment sessions. Furthermore, the variation located at practice level might indicate a (conscious or unconscious) practice policy regarding the number of treatment sessions. This is in accordance with the assumption that individual practitioners are embedded within medical groups and that shared circumstances channel the behavior of the group members, as stated by Westert et al. (1999) [22].
In the current study, it proved possible to explain 13% of the variance. Although this percentage seems rather low, it is consistent with other studies carried out in health care professions [14,23,24]. Dunlop et al. (2000) studied the role of socio-economic status in the differential use of physician services and were able to explain between 9% and 20% of the variance in the various analyses [23]. Kersnik et al. (2001) investigated predictors of frequent attendance in general practice and explained 20% of the variance [24]. Finally, Zuijderduin et al. (1995) studied factors related to the number of treatment sessions and were able to explain 16% of the variance in the number of treatment sessions [14].
It is necessary to gain more insight into the variation in the number of treatment sessions in order to increase the transparency of physical therapy care and to increase its quality. What we particularly need to know is whether the unexplained variation is appropriate or not, as quality policy should be aimed at decreasing variance caused by inappropriate factors. Unexplained variation could consist of appropriate factors, such as psychosocial characteristics. Coping style, for example, is predictive of the ability to control or adjust pain [25] and a higher ability to control pain might result in a lower number of physical therapy treatment sessions. Furthermore, some LBP patients have high levels of fear avoidance beliefs, which result in avoidance behavior. Avoidance behavior is perceived to be a maladaptive response, as it is associated with negative psychological consequences (e.g. exaggerated pain perception) and negative physiological consequences (e.g. decreased range of spine motion) [26]. This reaction is likely to be associated with a higher number of treatment sessions. The extent to which these factors are indeed related to the number of physical therapy treatment sessions is unclear as yet, however. In addition to psychosocial factors, the ability to learn motor behavior might also influence the number of physical therapy treatment sessions. Patients with a low ability to learn motor behavior will need more treatment sessions than patients with a high ability to learn motor behavior. Furthermore, a patient with a high baseline disability will need more treatment sessions than a patient with a low baseline disability. On the other hand, inappropriate factors, such as demands made by a patient that have no clinical relevance, might also be part of the unexplained variation. It will be a challenge for future investigations to study the effects of the above-mentioned characteristics as well.
The mean number of treatment sessions is ten in the current study, but comparisons with international literature suggest that the mean number of treatment sessions varies. One study in Northern Ireland showed a median number of five treatment sessions for patients with LBP [4], while a study in the United States of America showed a mean number of eleven treatment sessions [6]. In the Dutch situation, the mean number of treatment sessions is located around the number that is eligible for reimbursement by public health insurance funds.
The limitations of the current study include its reliance on therapists to accurately record relevant data, but we expect only minimal inaccuracies in the data for two reasons. Firstly, the participating therapists charge the health care funds electronically for the treatment sessions provided. In the current study, a quantity of the data collected has been filtered out of this reimbursement data. Secondly, missing data or wrong data are corrected by means of standardized quality control. Another limitation of the study is the possibility that the participating therapists are a subgroup of Dutch therapists, i.e. therapists working in computerized practices and therapists that were willing to participate. Basic characteristics of the participants, however, like gender, age, and years since graduation, are comparable to all Dutch therapists.
Conclusion
In summary, our results suggest that the number of physical therapy treatment sessions in patients referred with non-specific LBP mainly depends on characteristics at patient level. The greater part of the clinical variation was not explained, however, which means that additional research focusing on psychosocial factors is necessary for a progressive increase in the transparency of care.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
IS participated in the design of the study, performed the statistical analyses and drafted the manuscript. RW conceived of the study, participated in the design and helped to draft the manuscript. PG participated in the design of the study. WB, JD and EE participated in the design of the study and helped to draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was funded by the Netherlands Health Care Insurance Board (CvZ).
==== Refs
Armstrong MP McDonough SM Baxter GD Clinical guidelines versus clinical practice in the management of low back pain Int J Clin Pract 2003 57 9 13 12587934
Dekker J van Baar ME Curfs EC Kerssens JJ Diagnosis and treatment in physical therapy: an investigation of their relationship Phys Ther 1993 73 568 579 8356106
Foster NF Thompson KA Baxter GD Allen JM Management of Nonspecific Low Back Pain by Physiotherapists in Britain and Ireland. A descriptive Questionnaire of Current Clinical Practice Spine 1999 24 1332 1342 10404576 10.1097/00007632-199907010-00011
Gracey JH McDonough SM Baxter GD Physiotherapy Management of Low Back Pain. A Survey of Current Practice in Northern Ireland Spine 2002 27 406 411 11840108 10.1097/00007632-200202150-00017
Hendriks HJM Physiotherapist's Consultation in General Practice PhD thesis 2000 Maastricht University
Jette AM Smith K Haley MS Davis KD Physical Therapy Episodes of Care for Patients with Low Back Pain Phys Ther 1994 74 101 110 8290616
Jette AM Delitto A Physical therapy treatment choices for musculoskeletal impairments Phys Ther 1997 77 145 154 9037215
Jette DU Jette AM Physical therapy and health outcomes in patients with spinal impairments Phys Ther 1996 76 930 941 8790272
Mielenz TJ Carey TS Dyrek DA Harris BA Garrett JM Darter JD Physical therapy utilization by patients with acute low back pain Phys Ther 1997 77 1040 1051 9327819
van der Valk R Dekker J van Baar ME Physical Therapy for Patients with Back Pain Physiotherapy 1995 81 345 351 10.1016/S0031-9406(05)66795-8
Andersen RM Revisiting the Behavioral Model and access to medical care: does it matter? J Health Soc Behav 1995 36 1 10 7738325
Hannan TJ Variation in health care – the roles of the electronic medical record Int J med Inform 1999 54 127 136 10219952 10.1016/S1386-5056(98)00175-0
Mitchell JM de Lissovoy G A comparison of resource use and cost in direct access versus physician referral episodes of physical therapy Phys Ther 1997 77 10 18 8996459
Zuijderduin W Dekker J Abrahamse H Determinanten van de omvang van de behandeling in de extramurale fysiotherapie. [Determinants of the number of sessions in a physical therapeutic treatment] Tijdschr Soc Gezondheidsz 1995 73 274 281
Hingstman L Kenens R Cijfers over fysiotherapeuten in de eerste lijn 2001 [Data on physical therapists in primary care 2001] Utrecht: NIVEL 2002
Dekker J van Ravensberg D van den Ende E Oostendorp R De beperkende maatregel fysiotherapie, oefentherapie Cesar en oefentherapie-Mensendieck en het Amsterdams Dienstenmodel: samenvatting van het evaluatie-onderzoek Deelrapport 4 [The restrictive measure for physical therapy, remedial therapy (Cesar and Mensendieck) and the Amsterdam 'services model': summary of the evaluation research Part 4] Utrecht: Nivel/Npi 1998
Hingstman L Kenens R van der Windt W Talma HF Meihuizen HE de Voogd-Hamelink AM Rapportage arbeidsmarkt zorg en welzijn 2001 Hoofdrapport [Report on the labour market for health care and welfare Main report] Tilburg: OSA publicatie ZW 21 2001
Verheij R Jabaaij L de Bakker D Abrahamse H van den Hoogen H Braspenning J van Althuis T Rutten R LINH jaarrapport 2001 cijfers uit het Landelijk Informatie Netwerk Huisartsenzorg: contacten, verwijzingen en voorschrijven in de huisartsenpraktijk [LINH 2001 annual report: encounters, referrals and prescribing in General Practice] Utrecht: Nivel 2002
Bekkering GE Hendriks HJM Koes BW Oostendorp RAB Ostelo RWJG Thomassen JMC van Tulder MW Dutch Physiotherapy Guidelines for Low Back Pain Physiotherapy 2003 89 82 96 10.1016/S0031-9406(05)60579-2
Hox JJ Applied multilevel analysis 1995 Amsterdam: TT-Publikaties
Long MJ An explanatory model of medical practice variation: a physician resource demand perspective J Eval Clin Pract 2002 8 167 174 12060414 10.1046/j.1365-2753.2002.00343.x
Westert GP Groenewegen PP Medical Practice Variations: changing the theoretical approach Scand J Public Health 1999 27 173 180 10482075 10.1080/140349499445112
Dunlop S Coyte PC McIsaac W Socio-economic status and the utilization of physicians' services: results from the Canadian National Population Health Survey Soc Sci Med 2000 51 123 133 10817475 10.1016/S0277-9536(99)00424-4
Kersnik J Švab I Vegnuti M Frequent attenders in general practice: quality of life, patient satisfaction, use of medical services and GP characteristics Scand J Prim Health Care 2001 19 174 177 11697559 10.1080/028134301316982405
Rosenstiel AK Keefe FJ The use of coping strategies in chronic low back pain patients' relationship to patient characteristics and current adjustment Pain 1983 17 33 44 6226916 10.1016/0304-3959(83)90125-2
George SZ Bialosky JE Fritz JM Physical therapist management of a patient with acute low back pain and elevated fear-avoidance beliefs Phys Ther 2004 84 538 549 15161419
Statistics Netherlands
|
16307679
|
PMC1310612
|
CC BY
|
2021-01-04 16:31:49
|
no
|
BMC Health Serv Res. 2005 Nov 24; 5:74
|
utf-8
|
BMC Health Serv Res
| 2,005 |
10.1186/1472-6963-5-74
|
oa_comm
|
==== Front
BMC GenetBMC Genetics1471-2156BioMed Central London 1471-2156-6-521626643410.1186/1471-2156-6-52Research ArticleAnalysis of four DLX homeobox genes in autistic probands Hamilton Steven P [email protected] Jonathan M [email protected] Elaine J [email protected] Nöel [email protected] Marc [email protected] John LR [email protected] Department of Psychiatry, University of California, San Francisco, CA, USA2 Center for Human Genetics, University of California, San Francisco, CA, USA3 Genomics Core Facility, University of California, San Francisco, CA, USA4 Nina Ireland Laboratory, University of California, San Francisco, CA, USA5 Department of Biology, University of Ottawa, Ontario, Canada2005 2 11 2005 6 52 52 21 4 2005 2 11 2005 Copyright © 2005 Hamilton et al; licensee BioMed Central Ltd.2005Hamilton et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Linkage studies in autism have identified susceptibility loci on chromosomes 2q and 7q, regions containing the DLX1/2 and DLX5/6 bigene clusters. The DLX genes encode homeodomain transcription factors that control craniofacial patterning and differentiation and survival of forebrain inhibitory neurons. We investigated the role that sequence variants in DLX genes play in autism by in-depth resequencing of these genes in 161 autism probands from the AGRE collection.
Results
Sequencing of exons, exon/intron boundaries and known enhancers of DLX1, 2, 5 and 6 identified several nonsynonymous variants in DLX2 and DLX5 and a variant in a DLX5/6intragenic enhancer. The nonsynonymous variants were detected in 4 of 95 families from which samples were sequenced. Two of these four SNPs were not observed in 378 undiagnosed samples from North American populations, while the remaining 2 were seen in one sample each.
Conclusion
Segregation of these variants in pedigrees did not generally support a contribution to autism susceptibility by these genes, although functional analyses may provide insight into the biological understanding of these important proteins.
==== Body
Background
Autism is a severe heterogeneous neurobehavioral syndrome that becomes apparent in the first years of life [1-3] Autism is often viewed as a type of mental retardation, as most autistics have IQs lower than 70. However, autism is distinguished from other mental retardation syndromes by disproportionately severe deficits in language and social skills. Persons with some autistic features but with preserved language are often referred to as having Asperger's Syndrome [4].
There has been much interest and work investigating the genetic basis of autism [5]. Twin studies have shown that autism is a strongly inherited disorder [2,6,7], as monozygotic twins are concordant for this syndrome substantially more frequently than are dizygotic twins. For reasons that are not yet understood, autism affects boys about four times more often than girls. Non-genetic etiological factors are under careful consideration [8], given the controversy over changing estimates of the incidence of autism over the past decades [9].
Recently, we hypothesized that some forms of autism may be due to a disproportionate high level of excitation (or disproportionately weak inhibition) in neural circuits that mediate language and social behaviors [10]; related models have also been postulated [11,12]. A "noisy" (hyperexcitable, poorly functionally differentiated) cortex is inherently unstable, and susceptible to epilepsy, a common malady of autistic people [13]. Consistent with this model, a subset of patients with a mutation of the X-linked homeobox gene ARX have autistic features [14,15]. Analysis of the expression and function of murine Arx has revealed that this transcription factor controls proliferation of cortical progenitor cells and it regulates the development of the basal ganglia and the tangentially-migrating GABAergic neurons, derived from this region, that become cortical local circuit neurons [16]. Furthermore, Arx is expressed in mature cortical local circuit neurons [16,17]. Arx expression in GABAergic neurons in the basal ganglia and cortex could underlie the epilepsy and movement disorders in humans bearing Arx mutations.
The highly conserved Dlx1, 2, 5 & 6 homeobox transcription factors also control the development of the basal ganglia and cortical local circuit neurons [18-23]. This gene family regulates the expression of glutamic acid decarboxylase [24,25] (and Cobos and Rubenstein, unpublished). Furthermore, these genes control the expression of Arx in cells derived from basal ganglia progenitor domains [17]. Finally, mice lacking Dlx1 have defects in subsets of cortical local circuit neurons that lead to their apoptosis and subsequent onset of epilepsy [26]. Therefore, murine Dlx genes have a central role in controlling the development and function of forebrain GABAergic (inhibitory) neurons. The human DLX genes are organized as bigene clusters on chromosomes 2q31.1 (DLX1/2) and 7q21.3 (DLX5/6) and their expression is controlled by intra- and extragenic enhancers [27-30] These genes are close to loci that have been linked with autism (2q32, and 7q22-31) [31-34].
In a genome scan using 152 affected sibling pairs, a multipoint maximum lod score (MLS) of 3.74 was reported near D2S2188 [31]. This region on chromosome 2 has also been reported to be linked to autism in one study of 51 multiplex families [35], while other studies show linkage slightly more telomeric to this region [36,37]. Chromosome 7q, also an important region in linkage studies of autism [38], contains DLX5 and DLX6. The International Molecular Genetic Study of Autism Consortium (IMGSAC) reported a single locus MLS of 2.93 in 83 sibling pairs at D7S477, some 4.1 Mb telomeric to the DLX5/6 cluster on 7q22 [31]. When 152 sibling pairs were analyzed, a multipoint MLS of 3.20 was calculated. An earlier IMGSAC study reported linkage ~35 Mb telomeric to the DLX5/6 region [39].
Converging biological and genetic evidence thus suggest a role for DLX proteins in autism. While previous studies have failed to identify non-synonymous mutations in DLX1 and DLX2 [32], or close linkage of a polymorphism in DLX6 in autistic patients [40], we sequenced the exons, exon/intron boundaries, and known regulatory elements of the DLX1/2 and DLX5/6 genes in 161 autism probands and 58 non-autistic siblings collected as part of the Autism Genetic Resource Exchange (AGRE) [41], an initiative coordinated by Cure Autism Now (CAN). We identified three non-synonymous variants in DLX2 and two non-synonymous variants in DLX5 that were either very rare or not present in other populations. A number of other SNPs were identified in all four genes.
Results
We resequenced the coding regions (total 3,153 bp, mean 263 bp/coding region, range 82–400 bp) and flanking non-coding regions (total 3831 bp, mean 319 bp/non-coding region, range 176–755 bp) of human DLX1, DLX2, DLX5, and DLX6. Additionally, we resequenced 2,679 bp of the regions between each bigene cluster known to contain highly conserved tissue specific enhancers [28]. Finally, we sequenced 1,372 bp of an upstream regulatory element (URE2) located ~12 kb upstream of DLX1 (Ghanem and Ekker, unpublished observations).
The sample consisted of 161 autism probands and 58 non-autistic siblings obtained from AGRE, for a total of 2.08 Mb of sequence. The DNA variants discovered are summarized in Table 1. The relative location of each variant within the DLX1/2 and DLX5/6 clusters is depicted in Figure 1. Variants are referred to by our simple name (e.g., DLX2 SNP-1) for the purpose of clarity within this manuscript, while technically accurate nomenclature [42,43] or dbSNP rs#, if available, are also referenced in Table 1. Allele frequencies for these variants are detailed in Additional File 2. Thirty-three variants were observed, consisting of 31 single nucleotide polymorphisms (SNPs) and two insertion/deletion polymorphisms. Coding region variants consisted of three synonymous SNPs (one in DLX2, two in DLX5) and four non-synonymous SNPs (two in DLX2, two in DLX5). Additionally, an in-frame insertion of AGC in exon 1 of DLX2 leads to the insertion of an additional serine after serine 46, lengthening a poly-serine stretch from six to seven residues. DLX1 and DLX6 did not have any coding region variants in this study. One SNP each in DLX1 and DLX2, as well as two SNPs in DLX6, have been previously deposited in dbSNP. Interestingly, DNA variants were ~2.3 fold less common in the four DLX1/2 and DLX5/6 intergenic enhancers and DLX1/2 URE2. For 33 of 34 variants, for which corresponding chimpanzee genomic sequence was available, the major human allele matched the chimpanzee allele. For the single exception, in the brain specific DLX1/2 intergenic enhancer, the minor allele seen in our sample matches the allele seen in chimp, mouse, rat, and chicken. We observed none of three SNPs in DLX1 and all three SNPs in DLX2 observed in a recent resequencing effort in 48 autism probands [32].
Table 1 DNA variants identified in DLX genes
Gene Location SNP # Name dbSNP Contig Build34 Change AA Pos AA Change
NT_005403.14
DLX1-2
URE2 - 1 NT_005403.14:g.23146815 23146815 173139942 C -> T - -
- 2 NT_005403.14:g.23147084 23147084 173140211 A -> G - -
- 3 NT_005403.14:g.23147129 23147129 173140256 G -> A - -
- 4 NT_005403.14:g.23147332 23147332 173140459 G -> A - -
DLX1 5' 1 -111G>T 23159712 173152839 G -> T - -
(AY257976) 3' 2 *170T>C 23162572 173155699 T -> C - -
3' 3 *182C>G rs3821186 23162584 173155711 C -> G - -
DLX1-2 BR - 1 NT_005403.14:g.23165866 23165866 173158993 C -> A - -
DLX1-2 AR - InDel-1 NT_005403.14:g.23168271 23168271 173161398 G -> - - -
DLX2 5' 1 -36G>A rs743605 23176719 173169846 G -> A - -
(NM_004405) Exon 1 InDel-1 138_139insAGC 23176546 173169673 AGC 46_47 Ser46_Leu47insSer
Exon 1 2 394G>A 23176290 173169417 G -> A 132 Glu -> Lys
Intron 1 3 401-156C>T 23175947 173169074 C -> T - -
Exon 2 4 525A>G rs2228184 23175667 173168794 A -> G 175 Gln -> Gln
Intron 2 5 585+100C>G 23175507 173168634 C -> G - -
Exon 3 6 670G>A 23175005 173168132 G -> A 224 Ala -> Thr
3' 7 987*1C>T 23174687 173167814 C -> T - -
3 8 987*65C>T 23174623 173167750 C -> T - -
3' 9 987*154C>T 23174534 173167661 C -> T - -
NT_007933.13
DLX5 Exon 1 1 252C>G 21887602 96265415 C -> G 84 Ala -> Ala
(NM_005221) Exon 1 2 306C>T 21887548 96265361 C -> T 102 His -> His
Intron 2 3 541-44C>G 21884339 96262152 C -> G - -
Intron 2 4 541-31C>T 21884326 96262139 C -> T - -
Intron 2 5 541-10C>T 21884305 96262118 C -> T - -
Exon 3 6 685T>C 21884151 96261964 T -> C 229 Ser -> Pro
Exon 3 7 702C>A 21884134 96261947 C -> A 234 Ser -> Arg
DLX5-6 I1 - 1 NT_007933.13:g.21875347 21875347 96253160 A -> G - -
DLX6 Intron 1 1 82+57C>T 21869700 96247513 C -> T - -
(NM_005222) Intron 1 2 82+58C>T 21869701 96247514 C -> T - -
Intron 1 3 82+59C>T 21869702 96247515 C -> T - -
Intron 1 4 82+103C>T 21869746 96247559 C -> T - -
Intron 1 5 82+124C>T 21869767 96247580 C -> T - -
Intron 1 6 82+160C>T 21869803 96247616 C -> T - -
Intron 1 7 83-85A>C rs1207727 21870783 96248596 A -> C - -
3' 8 *9A>G rs3213654 21873286 96251099 A -> G - -
aVariant naming by accepted mutation nomenclature convention (references 53 and 54).
bContig accessions: NT_005403.14 for DLX1/2 region; NT_007933.13 for DLX5/6 region.
cNucleotide changes are listed major allele to minor allele, with chimp allele bolded.
dChimp sequence not available for DLX2 InDel-1.
ers2228184 is also known as rs12619503.
Figure 1 Schematic of the genomic region for four DLX genes. The regions depicted are a) chromosome 2q31.1 (33.0 kb) for DLX1 and DLX2; b) chromosome 7q21.3 (20.1 kb) for DLX5 and DLX6. For both regions, the orientation has the telomere to the right of the figure. The top part of each figure shows the position and relative size of the amplicons sequenced, with the resulting SNPs (S) or insertion/deletion (ID) depicted with a thin vertical line. Below this is a schematic of the transcript in it's genomic context, with coding regions in gray and non-coding regions cross-hatched. Finally, a VISTA alignment of the genomic regions is depicted, comparing the human July 2003 build with the mouse February 2003 build, with regions with >75% sequence identity shaded. Exons for DLX6 do not show up as blue, as it was not annotated on the UCSC RefSeq track.
We focused on the non-synonymous SNPs and the DLX2 serine insertion for further analysis given the plausibility of the functional significance of these amino acid changes. The DLX2 InDel-1 was observed on 3 autism chromosomes. DLX2 SNPs-2 and 6 and DLX5 SNP-6 were each seen on individual chromosomes, while DLX5 SNP-7 was observed on 4 chromosomes. These observations result in allele frequencies of 0.009, 0.003, 0.003, 0.003, and 0.012, respectively, in our autism sample. We compared the allele frequencies for each of these five observed DLX2 and DLX5 variants in the AGRE sample with the frequency of these DLX variants in a different population. For this purpose we sequenced 188 DNAs from the Coriell DNA Polymorphism Discovery Resource (PDR). This population is not phenotypically defined, but does represent a diverse array of populations for the detection of uncommon DNA polymorphisms. An additional two SNPs (SNPs 3 and 4) in DLX5 were observed in the PDR sample as part of this effort, but not in the autistic sample or the non-autistic siblings (Additional File 2). Additionally, we genotyped the four non-synonymous SNPs in 95 Caucasian and 95 African-American samples, also obtained from the Coriell Institute. No heterozygous genotypes were seen for SNPs 2 and 6 of DLX2. For SNPs 6 and 7 of DLX5, single heterozygous samples were seen, with both occurring in the Caucasian sample. Inspection of this data indicates that allele frequencies for the 4 protein-changing variants did not differ between autistic probands and their siblings, nor between autistic probands and the PDR and human variation panels.
In general, the variants were uncommon. Fifteen were observed only once in the autism probands. Ten were seen in 2–4 autistic individuals. Four of the variants were more common in the autism probands, with minor allele frequencies ranging between 3.4% and 46.5%.
Additional File 3 depicts the genotypes of non-synonymous DLX2 and DLX5 variants and the DLX5/6 intergenic enhancer variant in AGRE pedigrees in which they were detected. The segregation patterns for these variants do not clearly support the hypothesis that they are autism susceptibility variants. DLX2 InDel-1 and DLX2 SNPs-2 and DLX5 SNP-6 are found in affected and unaffected siblings within the same pedigrees. The minor variant for DLX2 SNP-6, which was also seen in the Caucasian panel, is introduced to one of two affected cousins by a father who is not related by blood to the second cousin. SNP-7 in DLX5, which predicts a substitution of Arginine for Serine at position 234, was seen in 3 pedigrees (Additional File 3, figure 1e), but also in the PDR and Caucasian panels. Four autism probands and one sibling diagnosed as broad-spectrum autism were heterozygous for this SNP. Two normal siblings and one autistic proband were homozygous for the wild type allele. It is transmitted to both affected siblings in two families, and to an affected sibling and unaffected sibling in a third. The rich phenotypic data collected by AGRE provided an opportunity to assess any relationship between phenotype and the variants described above. The average non-verbal IQ for all affected persons in the pedigrees in which the four non-synonymous SNPs occurred was 96.5. The two lowest scores, 68 and 85, occurred in two sibs (one with autism, the other with "not quite autism") who both were heterozygous for the DLX2 Serine insertion/deletion and both had early trigonocephaly which normalized. The most notable observation is with family AU0028, in which a heterozygous mother transmits the variant DLX5 SNP-7 to one of her two affected children. Both children have medical histories positive for generalized tonic-clonic seizures, although only the heterozygous child, required medication treatment. Both children also exhibited obsessive-compulsive disroder and attention deficity hyperactivity disorder (ADHD) symptoms, although only the heterozygous child received an ADHD diagnosis. The child heterozygous for the variant also exhibited failure to thrive and an early gait abnormality, which resolved. Given the discordance between genotypes of the affected siblings, it is hard to argue for the role of the DLX5 variant in these specific phenotypes.
We investigated the evolutionary conservation of this particular amino acid change. As shown in Additional File 4, the Serine at position 234 in humans is conserved in two other mammals and one amphibian, and is substituted with Glycine in one bird species. Phylogenetic analysis shows close DLX5 homology among six vertebrate species, particularly among mammals. Similarly, the Serine at position 229 [which is substituted with Proline (SNP-6)] is conserved among all species examined, except for zebrafish, where the residue is Proline.
Discussion
In the present report, we focused on the DLX bigene clusters given their importance in forebrain development [18,25] and potential neurophysiological processes underlying autism [10]. We sequenced the coding regions and flanking non-coding regions for DLX1, DLX2, DLX5, and DLX6 in 161 autistic probands and 58 non-autistic siblings. We also sequenced four intergenic enhancers (two each between each cluster [28,30]) and an enhancer sequence ~13 kb upstream of DLX1 (Ghanem and Ekker, unpublished observations). In the gene regions, we identified 28 variants, four of which were previously deposited in public SNP databases. We found five variants that are predicted to change or insert an amino acid in the protein in DLX2 and DLX5. Three synonymous SNPs were also found in these genes. Interestingly, no coding sequence variants, synonymous or non-synonymous, were seen in DLX1 or DLX6. The low frequency of the non-synonymous variants preclude meaningful assessment of correlation between variant and disease in the families analyzed. For example, SNP-2 in DLX2 (Glu→Lys), SNP-6 in DLX2 (Ala→Thr), and SNP-6 in DLX5 (Ser→Pro) were each seen in single pedigrees (Additional File 1). In all three cases, one affected offspring in the pedigree has the variant, while either a second affected member does not, or a non-affected sibling also has the variant. The insertion of Serine residue between Serine 36 and Leucine 47 in DLX2 (InDel-1) occurred in three families, but occurred equally as frequently in autistic individuals as in non-autistic siblings, and was seen in 12 of 376 chromosomes in the polymorphism discovery sample. Interestingly, mouse and rat show tri-peptide insertions (Asn-Ser-Ser and Asn-Ser-Asn, respectively) at this site when compared to human and chimp sequence. Finally, a variant predicting a serine to arginine change in DLX5 (SNP-7), was seen in three pedigrees. In two pedigrees, each containing three offspring, two affected individuals were heterozygous for the variant, while a non-autistic sibling was homozygous for the wild type allele. In the third pedigree containing an affected sibling pair, one individual was heterozygous and the other was homozygous for the wild type allele. This variant was seen in one of 374 chromosomes in the polymorphism discovery sample, as well as in one of 380 chromosomes from the human variation panel samples. The non-synonymous changes from SNPs-2 and 6 in DLX2 and SNP-6 in DLX5 were not seen in 376 polymorphism discovery chromosomes or in the human variation panels. Our study design, involving resequencing of DLX family genes in autistic probands, has identified potentially interesting variants within these genes, but cannot provide statistically meaningful inferences about the effect on populations given our ascertainment scheme using non-independent probands from multiplex families and a rather small number of non-affected siblings.
Given the location of DLX1/2 and Dlx5/6 in relation to autism linkage intervals, other groups have examined whether DNA variants in these genes are significantly associated with autism. The IMGSAC consortium conducted a sequence survey of DLX1 and DLX2 in 48 autistic probands [32], which identified three variants in DLX1 that were not detected in our much larger sample and three of the DLX2 variants seen in our sample. One explanation for the non-overlap in findings is the low allele frequency of most variants found in both studies. For example, our SNP-1 in DLX1 was in a region also sequenced by the IMGSAC group, but our allele frequency was 0.2%, making it highly unlikely that it would be detected by assaying 96 chromosomes [32]. The high frequency SNPs-1 and 4 occurred in both samples. Another reason may be the origin of samples. The AGRE samples are predominantly of U.S. origin [41], while the IMGSAC samples are from a variety of geographically diverse countries [39], raising the possibility of population specific variants. This is particularly the case with rare variants, which are less likely to be shared across populations. Finally, differences in coverage of the gene (coding and non-coding regions), depth of coverage (i.e., sample size), or variant detection technology (e.g., direct sequencing in this report, versus variant detection with denaturing high performance liquid chromatography followed by direct sequencing by Bacchelli et al.) may explain the discrepancy in variants described. For example, our DLX1 SNPs-2 and 3 lie outside of the region assayed by IMGSAC.
In another recent study in 99 AGRE pedigrees and 308 other pedigrees, two SNPs in DLX2 were investigated [44]. One of these, rs2228184, corresponding to our DLX2 SNP-4, a synonymous coding sequence variant, showed marginal association to autism. This common variant was equally common in autistic probands and their unaffected siblings in our study (Additional File 2). The second SNP in the study by Rabionet et al., which was not associated with autism, occurs outside of the region we sequenced. These published data and our own do not provide support for the possibility that common variation in the DLX loci is associated with autism.
Another study focusing on DLX6 reported the existence of a CAG repeat in exon 1 of DLX6 after assaying 90 Caucasian samples [45]. Although we sequenced the same region, we did not detect this variation. In any case, the uncommon nature of the DLX variants reported here, even in aggregate, are unlikely to provide the basis for any linkage signal in the DLX gene clusters on chromosomes 2 and 7.
A striking observation in our data was the prominent lack of sequence diversity in the five non-coding regions we investigated. In the 4,000 bp encompassing the four intergenic enhancers and DLX1/2 upstream regulatory element, we found only seven variants. However, given the sequence conservation of these functional elements [28], this result is not surprising. Indeed, three of the four intergenic elements are included in 481 genomic segments greater than 200 bp with 100% conservation of human sequence with mouse and rat [46]. In other words, 0.6% of known ultraconserved sequences can be found in the 0.001% of the genome representing the DLX1/2 and DLX5/6 clusters. These deeply conserved sequences that were not exonic were significantly enriched near genes involved with transcriptional regulation, and in particular, those with Homeobox domains (p < 10-14) [46].
Of the variants identified, eight occurred in the coding regions of DLX2 and DLX5. The four variants that change the identity of an amino acid are non-conservative modifications: DLX2 SNP-2 (Glutamic acid to Lysine), DLX2 SNP-6 (Alanine to Threonine), DLX5 SNP-6 (Serine to Proline), and DLX5 SNP-7 (Serine to Arginine). In both DLX2 and DLX5, amino acid substitutions lie in conserved regions of the proteins. DLX2 SNP-2 is just N-terminal of the homeodomain in a region conserved among the human DLX2,3,5 subgroup. This DLX2 residue is conserved between human and chimp, dog, rat, and mouse, and fugu, while chicken, African clawed frog, and zebrafish contain the conservative Aspartic acid at the same position. The amino acids changed by DLX2 SNP-6, DLX5 SNP-6 and DLX5 SNP-7 are adjacent to a Proline-rich domain C-terminal to the homeodomain. The amino acids substituted by DLX5 SNP-6 and SNP-7 are invariant in 5 mammal species, chicken (except for the Serine changed to Arginine by SNP-7), and frog. DLX2 InDel-1 leads to the insertion of seventh Serine residue into a six residue polyserine tract within the conserved DLX2,3,5 DllA domain [21,47]. The functional significance of such a change is unknown.
The functional significance of the three synonymous SNPs (DLX2 SNP-4, DLX5 SNP-1 and DLX5 SNP-2) is uncertain, but cannot be summarily dismissed. For example, such "silent" variants can alter binding sites (exonic splice enhancers, ESE) for proteins involved in RNA splicing [48]. Using ESEfinder, a web-based application designed to analyze exonic sequences to identify potential ESEs responsive to the human SR proteins [49], each of these three variants alter the predicted strength or presence of recognition sites for one or more of several highly conserved and structurally related splicing factors termed Serine/Arginine-rich (SR) proteins (data not shown). For instance, the C to T substitution for DLX5 SNP-2 synonymous change abolishes binding sites for two of three SR proteins located in the region surrounding the SNP. The functional significance of this in silico observation is unknown, but highlights the potential importance of DNA variation that does not necessarily alter the primary structure or proteins. As is always the case with the analysis of rare variants, until the functionality of these variants is demonstrated, either through statistical differences in allele frequencies at the population level or through direct functional studies, these variants should not be considered disease mutations.
While the identification of variants that generate non-conserved amino acid changes in DLX2 and DLX5 in autistic people suggests that the DLX genes could contribute to autism susceptibility, there are limitations to our study. First the identified DLX variants identified here are rare variants that could be expected to naturally occur across the human population. While this is a possibility given the low likelihood that any random gene is an autism susceptibility locus [50], we believe that biological and genetic linkage data elevate the a priori probability that the DLX genes analyzed here may be autism genes. Furthermore, the nature of the amino acid changes suggests that they could alter the function of the DLX proteins, although direct demonstration of this is currently lacking. A second limitation is that the rarity of the DLX variants precludes the possibility that they account for a significant portion of the genetic susceptibility to autism. A further weakness is that our study did not allow the large-scale population-based case control design that would allow a better estimation of the probability that these variants contribute to causing autism. A third limitation involves the use of unaffected siblings in our mutation screen. These siblings were rigorously phenotyped, but not having clinical autism does not rule out the possibility that they have milder traits representing aspects of the autism phenotype. Thus, it is possible that variants shared by affected and "unaffected" siblings may be functionally significant. In terms of co-occurring medical conditions, we found no evidence for such in five families segregating non-synonymous amino acid variants. A fourth limitation is that the population sample we used to compare the allele frequencies of the autism variants with a "normal" population was not optimal. We used the DNA Polymorphism Discovery Resource sample, which is designed to mirror the sequence diversity of the human population. While this sample allowed us to examine a cross section of global genetic diversity, the use of this sample introduces two major limitations for the interpretation of our data. One is the lack of knowledge of phenotypes in the PDR and human variation samples raises the possibility that we may falsely conclude that a variant seen both in autistic probands and the Coriell samples is not involved with autism when in fact the Coriell samples with the variant unbeknownst to us may have autism or a related phenotype. The second involves the relative enrichment of the PDR sample for non-Caucasian samples when compared to our autism sample, which is predominantly Caucasian. This under-sampling of Caucasians (and consequent lack of power to detect rare variants) in the PDR sample may cause us to falsely attribute a rare variant as autism-related, when it may in reality be merely Caucasian-specific. This may indeed be the case for the two DLX5 non-synonymous variants, which were not seen in the PDR sample, but were seen in the Caucasian panel. Thus by including the Caucasian and African-American human variation panels, we have observed that the variants may be Caucasian-specific, albeit at low allele frequencies.
Despite the caveats described above, we suggest that the non-synonymous DLX SNPs could contribute to autism susceptibility for several reasons. Mice lacking DLX1 have epilepsy [26], a common feature in autistic patients. Heterozygosity of transcription factor mutations is well-known to cause human disease [51]. In mice the dosage of the DLX genes is known to be important in controlling the differentiation of forebrain GABAergic neurons and morphogenesis of craniofacial structures, including the middle and inner ear. Indeed, heterozygosity of DLX2 alters morphogenesis of the skull (Depew and Rubenstein, unpublished), although it is uncertain whether heterozygosity of a DLX gene alters brain function. Given recent evidence that the DLX5 locus is partially imprinted in humans [52], heterozygosity for DLX5 alleles could have profound ramifications. In our own pedigrees, we note that two of three pedigrees segregating the DLX5 non-synonymous SNP-7 show maternal transmission, while the single DLX5 SNP-6 pedigree showed paternal transmission (Additional File 3). Increases in Dlx5 expression have been found in mice lacking MECP2 (the Rett Syndrome gene), which are associated with alterations in long-range chromatin organization [53]. Therefore, several recent findings are increasing the likelihood that changes in DLX function/expression are involved in neuropsychiatric disorders.
The fact that 4.4% of autistic probands had non-synonymous DLX2 and DLX5 variants (5% when including the DLX5/6 intergenic enhancer variant) could reflect the multifactorial etiology of autism. Finally, perhaps the variants in DLX2, DLX5 and ARX [16], all of which alter the development of forebrain GABAergic neurons [18-25], are providing a clue that an increase in the ratio of excitation/inhibition underlies some forms of autism [10]. Furthermore, it suggests that one should study other genes within genetic pathways that control the ratio of excitation/inhibition in neural circuits that regulate cognition, memory and emotion [10].
Conclusion
We carried out in depth resequencing of the exons, exon/intron boundaries and known enhancers of the human homeobox genes DLX1, 2, 5 and 6 and identified four nonsynonymous variants in DLX2 and DLX5 in 4% of families tested. We also observed a variant of unknown significance in the highly conserved DLX5/6intragenic enhancer. Without a larger population controlled for phenotype, we cannot assert that these variants are more common in autism. While it is possible that these potentially functional rare variants may alter DLX gene function, our observations do not support a significant contribution to autism susceptibility. More detailed functional analysis or population analysis (e.g., more comprehensive SNP genotyping performed on a massively large trio sample with statistical evidence for genetic association for some of the identified variants) is needed before these variants in the DLX genes can be connected to autism.
Methods
Description of sample
Autism sample details: We used 161 autism probands and 58 non-autistic siblings from 95 families in the Autism Genetic Resource Exchange (AGRE) collection [41]. The phenotypic characterization is comprised of the Autism Diagnostic Interview-Revised (ADI-R), a semi-structured clinical instrument for assessing autism [54] based on DSM-IV criteria. In addition, AGRE assigns the phenotypes of Not Quite Autism (NQA) and Broad Spectrum. The former identifies individuals who are no more ≤1 point from meeting autism criteria on any or all of the 3 content domains (i.e., social, communication, and/or behavior); or, individuals who meet criteria on all 3 content domains, but do not meet criteria on the age of onset domain. Broad Spectrum characterizes persons who show patterns of impairment along the spectrum of pervasive developmental disorders, comprising individuals with mild to severe impairment. Details regarding the interview process and consent are available from the AGRE web site . Permission to access the AGRE sample and phenotypic data was obtained. The self-reported ethnicities of the pedigrees are: 68 Caucasian, 16 unknown, 7 mixed parentage, 2 African-American, and 2 Asian-American. In general, samples were chosen in which there were >1 affected person per pedigree. Subjects were chosen blinded to clinical data, including molecular data, beyond the primary phenotype. At the cost of not detecting an unknown number of additional variants, we chose to sequence more than one autism proband from many of the families as a way of more rapidly identifying variants that are segregating with disease in the families. This does not allow direct comparison with the undiagnosed populations described in the following section as one would see in a case-control association study, which was not the design for this study. The correlated genotypes within families would inflate estimates of the allele frequencies of discovered variants if compared to independent controls. Although we do not formally test for differences in allele frequencies, this could be done by randomly choosing a proband from families with more than one proband for any statistical test.
Undiagnosed population sample details: The prevalence of DNA variations detected in the autism sample was assayed in several populations chosen for representation of the major populations of humans. The DNA Polymorphism Discovery Resource (PDR) sample set, obtained from the Coriell Institute, contains samples from United States residents who have ancestors from the major geographic regions of the world: Europe, Africa, the Americas, and Asia. The European-American group includes non-Hispanic whites; the African-American group includes non-Hispanic blacks; the Americas group includes Mexican-Americans and Native Americans; and the Asian-American group includes individuals whose ancestors came from several countries in East and South Asia. We used 188 of the 450 available samples. Additionally, we used 95 Caucasian and 95 African-American samples from the Coriell Human Variation Panels, which do not overlap with the PDR collection.
Sequence analysis
Sequencing of DLX genes was based on gene structure information provided by GenBank accessions AY257976 (DLX1), NM_004405 (DLX2), NM_005221 (DLX5), and NM_005222 (DLX6). Target sequence for the DLX1/2 and DLX5/6 intergenic enhancers was identified from consensus sequence data [28,30]. Genomic context and intron/exon boundaries were provided by the UCSC (, July 2003, hg16, NCBI Build 34) and ENSEMBL (, February 2004, v19.34b.2) genome browsers.
Sequences were uploaded into VectorNTI 8 (InforMax, Frederick, Maryland) and PCR primers were designed using Primer3 with the human repeat mispriming library [55]. Primers, primer concentrations, and PCR conditions are listed in Additional File 1. All liquid handling was carried out on a TECAN Genesis robot (TECAN-US, Research Triangle Park, NC). PCR reaction volumes were 10 μl, using one of two PCR reagents. The first was Platinum Taq polymerase (Invitrogen, Carlsbad, CA), containing 10 ng DNA template, 50 mM KCl, 20 mM Tris-Hcl (pH 8.4), 200 uM dNTPs, 1.5 mM MgCl2, 1.25 mM Betaine, and 0.25 units Platinum Taq DNA polymerase. The second was with AmpliTaq Gold, containing 10 ng DNA template, and 5 μl of the 2× AmpliTaq Gold Master Mix (Applied Biosystems, Foster City, CA). Primers were added at the concentrations listed in Additional File 1. Reactions were cycled in 96 well GeneAmp 9700 PCR machines (Applied Biosystems). All amplicons could be amplified with one of two protocols. A "short touchdown" protocol involved 5 minutes at 95°C, followed by 10 cycles of 94°C (0:20), 61°C (0:20, decreasing 0.5°C every cycle), and 72°C (0:45), then followed by 35 cycles of 94°C (0:20), 56°C (0:20), and 72°C (0:45), followed by 10 minutes at 72°C. A "long touchdown" procedure involved 5 minutes at 95°C, followed by 14 cycles of 94°C (0:20), 63°C (0:20, decreasing 0.5°C every cycle), and 72°C (0:45), then followed by 35 cycles of 94°C (0:20), 56°C (0:20), and 72°C (0:45), followed by 10 minutes at 72°C. Excess PCR primers and nucleotides were removed by adding a 2 μl solution containing 1 unit each of shrimp alkaline phosphatase and exonuclease I and incubating at 37°C for 1 hour, followed by denaturation of the enzymes by incubation 90°C for 15 minutes. Sequencing was carried out in 5 μl reactions in a 384 well GeneAmp 9700 PCR machine using the BigDye Terminators v3.1 Cycle Sequencing Kit (Applied Biosystems) containing 400 nM sequencing primer, 1 μl PCR template, 0.25 ul dye terminators, and 0.875 μl 5× sequencing buffer. Reactions were cycled by running at 96°C for one minute, followed by 25 cycles of 96°C (0:10), 50°C (0:05), and 60°C (4:00). Unincorporated dye terminators and residual salts were removed by use of the Montáge SEQ96 Sequencing Reaction Cleanup Kit (Millipore, Billerica, MA). Samples were electrophoresed on 3700 or 3730xl DNA Analyzer capillary electrophoresis platforms (Applied Biosystems). Bases were called using the ABI KB Basecaller. Sequencher (Gene Codes, Ann Arbor, MI) was used to edit the called bases. All variants will be submitted to dbSNP at the National Center for Biotechnology Information and the AGRE web site upon acceptance of this manuscript for publication. Predicted protein alignments were carried out in the AlignX utility in VectorNTI, with construction of phylogenetic trees using the neighbor joining method [56].
SNP genotyping
For SNPs 2 and 6 in DLX2 and SNPs 6 and 7 of DLX5, we designed genotyping assays to determine the presence of these SNPs in the Caucasian and African-American Human Variation Panels. For SNPs 2 and 6 in DLX2 and SNP 7 in DLX5, custom 5'-nuclease (TaqMan) assays were designed by and purchased from Applied Biosystems (primers are listed in Additional File 1). Genotyping was performed by cycling a 40-fold dilution of the primer/probe mix with 2× Universal MasterMix and 20 ng of genomic template in a final reaction volume of 5 μl. Samples were incubated at 95°C for 10 minutes, and then were cycled 40 times at 92°C for 15 seconds followed by 60°C for 1 minute. Samples were then brought to room temperature before fluorescence was read on an Applied Biosystems 7900 Sequence Detection System. DLX5 SNP-6 was genotyped using fluorescence polarization detection of template-directed dye terminator incorporation (FP-TDI), an assay based on single base extension [57]. Briefly, the first step involves polymerase chain reactions (PCR) of 5 microliters (μl) containing 500 nM of the forward and reverse primers (Additional File 1), 20 ng genomic DNA template, 200 μM dNTPs (Roche, Indianapolis, IN), 1 M anhydrous betaine (Acros Organics, Geel, Belgium), 50 mM KCl, 20 mM Tris-HCl (pH 8.4), 1.5 mM MgCl2, and 0.25 units Platinum Taq DNA polymerase (Invitrogen, Carlsbad, CA). All primers and TDI probes were designed using Primer3 software and were manufactured by Invitrogen (Carlsbad, CA) [55]. Cycling conditions were: 95°C for 5 minutes, followed by a touchdown protocol of 10 cycles 94°C for 20 seconds, then 61°C for 20 seconds (with -0.5°C increments) and 72°C for 45 seconds, followed by 35 cycles of 94°C for 20 seconds, 56°C for 20 seconds, 72°C for 45 seconds, with a subsequent 10 minute incubation at 72°C. Primers and dNTPs were degraded by addition of a 2 μl solution of E. coli exonuclease I and shrimp alkaline phosphatase, with cycling of 37°C for 90 minutes with inactivation at 95°C for 15 minutes. The final step was the addition of a 13 μl solution containing a final concentration of 0.38 μM TDI probe, 2 μl of 10× TDI Reaction Buffer, 0.5 μl of AcycloTerminator Mix (containing R110 and TAMRA labeled AcycloTerminators, corresponding to the polymorphic base), and 0.025 μl of AcycloPol DNA polymerase (PerkinElmer). This mixture was cycled at 95°C for 2 min, followed by 20 cycles of 94°C for 15 sec and 55°C for 30 sec. Genotypes were read on a Victor 2 plate reader (PerkinElmer). Positive controls for both TaqMan and FP-TDI consisted of sequence-verified samples from the initial sequencing survey.
Authors' contributions
SPH participated in the design of the experiments, carried out the SNP genotyping, data analyses, multi-species sequence alignment, and co-wrote the manuscript. JMW and EJC designed and carried out the DNA sequencing and sequence analysis. NG and ME guided the analysis on the DLX regulatory elements and provided unpublished information their sequence and function. JLRR conceived of the study, participated in its design and coordination and co-wrote the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional file 2
Allele frequencies of variants detected by resequencing in four DLX genes and DLX1/2 intergenic enhancers. The allele frequency of the minor allele is presented for 161 autism probands (autism sample), 58 normal siblings of 58 autism probands (non-autism sample), and 188 samples from the NIGMS/NHGRI Polymorphism Discovery Resource (PDR sample). Selected variants were assayed in the PD sample, as described in the text. Variants from other autism studies are designated in the Variant column, with numbers of samples tested shown in the Autism Sample column. *, Bacchelli et al., 2003. [reference [32] **, Nabi et al., 2003 [reference [40]. The "intron 1 C/T" SNP currently maps to intron 2, and is also known as rs3801290, and was assayed in 221 affected sibling pairs and 210 discordant sibling pairs in 196 AGRE families. †, a variant reported at dbSNP within the region we sequenced, but that was not seen in our dataset. Of note, DLX2 SNP-1, a common SNP (rs743605) was found to differ significantly in allele frequency between the autistic probands and the polymorphism discovery sample (χ2 = 4.43, df = 1, p = 0.04). However, this SNP has also been genotyped as part of the HapMap project in a sample of 30 CEPH trios, and the minor allele frequency was found to be 0.47, essentially identical to our autism sample. This discrepancy is likely due to the heterogeneous population composition of the polymorphism discovery sample, which is approximately 26% each of European-American, African-American, and Asian-American populations, with smaller contributions of Mexican-American and Native-American samples (Collins, et al., 1998, Genome Res. 8, 1229–1231).
Click here for file
Additional file 3
1a AGRE pedigrees segregating InDel-1 DLX2. -, no insertion. AGC, insertion of AGC. ■, autism. ◒, NQA (not quite autism). 1b AGRE pedigrees segregating SNP-2 DLX2. ■, autism. 1c AGRE pedigrees segregating SNP-6 DLX2. ■, autism. 1d AGRE pedigrees segregating SNP-6, DLX5. ■, autism. 1e AGRE pedigrees segregating SNP-7 in the third exon of DLX5. ■, autism; ◓, broad spectrum autism. 1f AGRE pedigrees segregating SNP-1 from the DLX5/6 intergenic enhancer. ■, autism.
Click here for file
Additional file 4
Alignment of DLX5 protein sequences in six vertebrates. a) The region corresponding to amino acid residues 219–244 in the human sequence is depicted. Sequences are listed by GenBank accession and species (HS, Homo sapiens; MM, Mus musculus; RN, Rattus norvegicus; GG, Gallus gallus; XL, Xenopus laevis; and DR, Danio rerio). Type color depicts alignment status (red on yellow, completely conserved; blue on cyan, consensus derived from block of similar residues; green, residue weakly similar to consensus residue; black, non-similar to consensus residue; black on green, consensus derived from majority residue). The location of the residue affected by DLX5 SNPs is shown by the arrows (SNP-6, Ser/Pro; SNP-7, Ser/Arg). b) Phlyogenetic tree for the entire DLX5 protein sequence of six vertebrates calculated using neighbor joining method (reference 51, Saitou and Nei, 1987). The distances in parentheses represent the degree of divergence between sequences.
Click here for file
Additional file 1
Sequencing and genotyping conditions. PCR primers and conditions for generating templates for DNA sequencing for four DLX genes, two intergenic DLX1/2 enhancers (DLX1-2 BR and DLX1-2 AR), two DLX5/6 intergenic enhancers (DLX5/6 I-1 and I-2), and DLX1/2 upstream regulatory element (URE2). PCR protocols are short touchdown (ST) or long touchdown (LT), as described in Methods. The DNA polymerases (Pol) were Platinum Taq (P) or AmpliTaq Gold (AT). For SNP genotyping, primers for TaqMan (TM) or FP-TDI (FP) are shown. PCR primers and hybridization probes are shown for Taqman, with fluors (VIC or FAM) indicated along with bolded polymorphic base. Note inverted nature of DLX5 SNP-6. For DLX5 SNP-6, FP-TDI single base extension probe is shown. Note amplicon for this SNP is same as sequencing amplicon for DLX5 exon 3. Conditions for SNP genotyping are detailed in Methods.
Click here for file
Acknowledgements
The authors would like to acknowledge AGRE for access to their samples, Vlad Kustanovich at AGRE for assistance with phenotypic information, as well as the financial support of Alexsis de Raadt-St. James, Mark Hoffman, the Althea Foundation, Nina Ireland, NIMH RO1 MH49428 (JLRR), NIMH K05 MH65670 (JLRR), and CIHR MOP-14460 (ME).
==== Refs
Rapin I Katzman R Neurobiology of autism Ann Neurol 1998 43 7 14 9450763 10.1002/ana.410430106
Rutter M Genetic studies of autism: from the 1970s into the millennium J Abnorm Child Psychol 2000 28 3 14 10772346 10.1023/A:1005113900068
Tanguay PE Pervasive developmental disorders: a 10-year review J Am Acad Child Adolesc Psychiatry 2000 39 1079 1095 10986804 10.1097/00004583-200009000-00007
Gillberg C Billstedt E Autism and Asperger syndrome: coexistence with other clinical disorders Acta Psychiatr Scand 2000 102 321 330 11098802 10.1034/j.1600-0447.2000.102005321.x
Wassink TH Brzustowicz LM Bartlett CW Szatmari P The search for autism disease genes Ment Retard Dev Disabil Res Rev 2005 10 272 283 15666342 10.1002/mrdd.20041
Folstein SE Rosen-Sheidley B Genetics of autism: complex aetiology for a heterogeneous disorder Nat Rev Genet 2001 2 943 955 11733747 10.1038/35103559
Veenstra-VanderWeele J Cook E JrLombroso PJ Genetics of childhood disorders: XLVI. Autism, part 5: genetics of autism J Am Acad Child Adolesc Psychiatry 2003 42 116 118 12500084 10.1097/00004583-200301000-00018
Lawler CP Croen LA Grether JK Van de WJ Identifying environmental contributions to autism: provocative clues and false leads Ment Retard Dev Disabil Res Rev 2004 10 292 302 15666339 10.1002/mrdd.20043
Yeargin-Allsopp M Rice C Karapurkar T Doernberg N Boyle C Murphy C Prevalence of autism in a US metropolitan area JAMA 2003 289 49 55 12503976 10.1001/jama.289.1.49
Rubenstein JLR Merzenich MM Model of autism: increased ratio of excitation/inhibition in key neural systems Genes, Brain and Behavior 2003 2 255 267 10.1034/j.1601-183X.2003.00037.x
Hussman JP Suppressed GABAergic inhibition as a common factor in suspected etiologies of autism J Autism Dev Disord 2001 31 247 248 11450824 10.1023/A:1010715619091
Levitt P Eagleson KL Powell EM Regulation of neocortical interneuron development and the implications for neurodevelopmental disorders Trends in Neurosciences 2004 27 400 406 15219739 10.1016/j.tins.2004.05.008
Tuchman R Rapin I Epilepsy in autism Lancet Neurol 2002 1 352 358 12849396 10.1016/S1474-4422(02)00160-6
Turner G Partington M Kerr B Mangelsdorf M Gecz J Variable expression of mental retardation, autism, seizures, and dystonic hand movements in two families with an identical ARX gene mutation Am J Med Genet 2002 112 405 411 12376946 10.1002/ajmg.10714
Strømme P Mangelsdorf ME Scheffer IE Gecz J Infantile spasms, dystonia, and other X-linked phenotypes caused by mutations in Aristaless related homeobox gene, ARX Brain and Development 2002 24 266 268 12142061 10.1016/S0387-7604(02)00079-7
Kitamura K Yanazawa M Sugiyama N Miura H Iizuka-Kogo A Kusaka M Omichi K Suzuki R Kato-Fukui Y Kamiirisa K Matsuo M Kamijo S Kasahara M Yoshioka H Ogata T Fukuda T Kondo I Kato M Dobyns WB Yokoyama M Morohashi K Mutation of ARX causes abnormal development of forebrain and testes in mice and X-linked lissencephaly with abnormal genitalia in humans Nat Genet 2002 32 359 369 12379852 10.1038/ng1009
Cobos I Broccoli V Rubenstein JLR The vertebrate ortholog of Aristaless is regulated by Dlx genes in the developing forebrain 2004
Anderson SA Eisenstat DD Shi L Rubenstein JL Interneuron migration from basal forebrain to neocortex: dependence on Dlx genes Science 1997 278 474 476 9334308 10.1126/science.278.5337.474
Anderson SA Qiu M Bulfone A Eisenstat DD Meneses J Pedersen R Rubenstein JL Mutations of the homeobox genes Dlx-1 and Dlx-2 disrupt the striatal subventricular zone and differentiation of late born striatal neurons Neuron 1997 19 27 37 9247261 10.1016/S0896-6273(00)80345-1
Marin O Anderson SA Rubenstein JL Origin and molecular specification of striatal interneurons J Neurosci 2000 20 6063 6076 10934256
Panganiban G Rubenstein JLR Developmental functions of the Distal-less/Dlx homeobox genes Development 2002 129 4371 4386 12223397
Stuhmer T Puelles L Ekker M Rubenstein JL Expression from a Dlx gene enhancer marks adult mouse cortical GABAergic neurons Cereb Cortex 2002 12 75 85 11734534 10.1093/cercor/12.1.75
Yun K Fischman S Johnson J Hrabe dA Weinmaster G Rubenstein JL Modulation of the notch signaling by Mash1 and Dlx1/2 regulates sequential specification and differentiation of progenitor cell types in the subcortical telencephalon Development 2002 129 5029 5040 12397111
Anderson S Mione M Yun K Rubenstein JL Differential origins of neocortical projection and local circuit neurons: role of Dlx genes in neocortical interneuronogenesis Cereb Cortex 1999 9 646 654 10498283 10.1093/cercor/9.6.646
Stuhmer T Anderson SA Ekker M Rubenstein JL Ectopic expression of the Dlx genes induces glutamic acid decarboxylase and Dlx expression Development 2002 129 245 252 11782417
Cobos I Calcagnotto ME Vilaythong AJ Noebels J Baraban SC Rubenstein JLR Mice Lacking the Dlx1 Transcription Factor Exhibit Subtype-Specific Loss of Interneurons, Reduced Synaptic Inhibition and Epilepsy 2004
Stock DW Ellies DL Zhao Z Ekker M Ruddle FH Weiss KM The evolution of the vertebrate Dlx gene family Proc Natl Acad Sci U S A 1996 93 10858 10863 8855272 10.1073/pnas.93.20.10858
Ghanem N Jarinova O Amores A Long Q Hatch G Park BK Rubenstein JLR Ekker M Regulatory Roles of Conserved Intergenic Domains in Vertebrate Dlx Bigene Clusters Genome Res 2003 13 533 543 12670995 10.1101/gr.716103
McGuinness T Porteus MH Smiga S Bulfone A Kingsley C Qiu M Liu JK Long JE Xu D Rubenstein JL Sequence, organization, and transcription of the Dlx-1 and Dlx-2 locus Genomics 1996 35 473 485 8812481 10.1006/geno.1996.0387
Zerucha T Stuhmer T Hatch G Park BK Long Q Yu G Gambarotta A Schultz JR Rubenstein JL Ekker M A highly conserved enhancer in the Dlx5/Dlx6 intergenic region is the site of cross-regulatory interactions between Dlx genes in the embryonic forebrain J Neurosci 2000 20 709 721 10632600
International Molecular Genetic Study of Autism Consortium A genomewide screen for autism: strong evidence for linkage to chromosomes 2q, 7q, and 16p Am J Hum Genet 2001 69 570 581 11481586 10.1086/323264
Bacchelli E Blasi F Biondolillo M Lamb JA Bonora E Barnby G Parr J Beyer KS Klauck SM Poustka A Bailey AJ Monaco AP Maestrini E Screening of nine candidate genes for autism on chromosome 2q reveals rare nonsynonymous variants in the cAMP-GEFII gene Mol Psychiatry 2003 8 916 924 14593429 10.1038/sj.mp.4001340
Barnby G Monaco AP Strategies for autism candidate gene analysis Novartis Found Symp 2003 251 48 63 14521187
Hutcheson HB Bradford Y Folstein SE Gardiner MB Santangelo SL Sutcliffe JS Haines JL Defining the autism minimum candidate gene region on chromosome 7 Am J Med Genet 2003 117B 90 96 10.1002/ajmg.b.10033
Philippe A Martinez M Guilloud-Bataille M Gillberg C Rastam M Sponheim E Coleman M Zappella M Aschauer H Van Maldergem L Penet C Feingold J Brice A Leboyer M van Malldergerme L Genome-wide scan for autism susceptibility genes. Paris Autism Research International Sibpair Study [published erratum appears in Hum Mol Genet 1999 Jul;8(7):1353] Hum Mol Genet 1999 8 805 812 10196369 10.1093/hmg/8.5.805
Buxbaum JD Silverman JM Smith CJ Kilifarski M Reichert J Hollander E Lawlor BA Fitzgerald M Greenberg DA Davis KL Evidence for a susceptibility gene for autism on chromosome 2 and for genetic heterogeneity Am J Hum Genet 2001 68 1514 1520 11353400 10.1086/320588
Shao Y Raiford KL Wolpert CM Cope HA Ravan SA Ashley-Koch AA Abramson RK Wright HH DeLong RG Gilbert JR Cuccaro ML Pericak-Vance MA Phenotypic homogeneity provides increased support for linkage on chromosome 2 in autistic disorder Am J Hum Genet 2002 70 1058 1061 11875756 10.1086/339765
Badner JA Gershon ES Regional meta-analysis of published data supports linkage of autism with markers on chromosome 7 Mol Psychiatry 2002 7 56 66 11803446 10.1038/sj.mp.4001922
International Molecular Genetic Study of Autism Consortium A full genome screen for autism with evidence for linkage to a region on chromosome 7q. International Molecular Genetic Study of Autism Consortium Hum Mol Genet 1998 7 571 578 9546821 10.1093/hmg/7.3.571
Nabi R Zhong H Serajee FJ Huq AH No association between single nucleotide polymorphisms in DLX6 and Piccolo genes at 7q21-q22 and autism Am J Med Genet 2003 119B 98 101 10.1002/ajmg.b.10012
Geschwind DH Sowinski J Lord C Iversen P Shestack J Jones P Ducat L Spence SJ The autism genetic resource exchange: a resource for the study of autism and related neuropsychiatric conditions Am J Hum Genet 2001 69 463 466 11452364 10.1086/321292
den Dunnen JT Antonarakis SE Mutation nomenclature extensions and suggestions to describe complex mutations: a discussion Hum Mutat 2000 15 7 12 10612815 10.1002/(SICI)1098-1004(200001)15:1<7::AID-HUMU4>3.0.CO;2-N
den Dunnen JT Antonarakis SE Nomenclature for the description of human sequence variations Hum Genet 2001 109 121 124 11479744 10.1007/s004390100505
Rabionet R Jaworski JM shley-Koch AE Martin ER Sutcliffe JS Haines JL DeLong GR Abramson RK Wright HH Cuccaro ML Analysis of the autism chromosome 2 linkage region: GAD1 and other candidate genes Neuroscience Letters 2004 372 209 214 15542242 10.1016/j.neulet.2004.09.037
Pfeffer U Ferro P Pavia V Trombino S dell'Eva R Merlo G Levi G The coding region of the human DLX6 gene contains a polymorphic CAG/CCG repeat Int J Oncol 2001 18 1293 1297 11351265
Bejerano G Pheasant M Makunin I Stephen S Kent WJ Mattick JS Haussler D Ultraconserved Elements in the Human Genome Science 2004 1098119
Akimenko MA Ekker M Wegner J Lin W Westerfield M Combinatorial expression of three zebrafish genes related to distal-less: part of a homeobox gene code for the head Journal of Neuroscience 1994 14 3475 3486 7911517
Cartegni L Chew SL Krainer AR Listening to silence and understanding nonsense: exonic mutations that affect splicing Nat Rev Genet 2002 3 285 298 11967553 10.1038/nrg775
Cartegni L Wang J Zhu Z Zhang MQ Krainer AR ESEfinder: a web resource to identify exonic splicing enhancers Nucl Acids Res 2003 31 3568 3571 12824367 10.1093/nar/gkg616
Sullivan PF Eaves LJ Kendler KS Neale MC Genetic case-control association studies in neuropsychiatry Arch Gen Psychiatry 2001 58 1015 1024 11695947 10.1001/archpsyc.58.11.1015
Jimenez-Sanchez G Childs B Valle D Human disease genes Nature 2001 409 853 855 11237009 10.1038/35057050
Kimura MI Kazuki Y Kashiwagi A Kai Y Abe S Barbieri O Levi G Oshimura M Dlx5, the mouse homologue of the human-imprinted DLX5 gene, is biallelically expressed in the mouse brain J Hum Genet 2004 49 273 277 15362572 10.1007/s10038-004-0139-2
Horike Si Cai S Miyano M Cheng JF Kohwi-Shigematsu T Loss of silent-chromatin looping and impaired imprinting of DLX5 in Rett syndrome Nat Genet 2005 37 31 40 15608638 10.1038/ng1570
Lord C Rutter M Le CA Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders J Autism Dev Disord 1994 24 659 685 7814313 10.1007/BF02172145
Rozen S Skaletsky HJ Krawetz S, Misener S Primer3 on the WWW for general users and for biologist programmers Bioinformatics Methods and Protocols 2000 Totowa, NJ: Human Press 365 386
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Chen X Levine L Kwok PY Fluorescence polarization in homogeneous nucleic acid analysis Genome Res 1999 9 492 498 10330129
|
16266434
|
PMC1310613
|
CC BY
|
2021-01-04 16:30:39
|
no
|
BMC Genet. 2005 Nov 2; 6:52
|
utf-8
|
BMC Genet
| 2,005 |
10.1186/1471-2156-6-52
|
oa_comm
|
==== Front
BMC NeurosciBMC Neuroscience1471-2202BioMed Central London 1471-2202-6-691631863010.1186/1471-2202-6-69Research ArticleGenomic responses in rat cerebral cortex after traumatic brain injury von Gertten Christina [email protected] Amilcar Flores [email protected] Staffan [email protected] Tiit [email protected] Ann-Christin Sandberg [email protected] Department of Clinical Neuroscience, Karolinska Institutet, Section of Clinical CNS research, Karolinska University Hospital, SE-171 76 Stockholm, Sweden2 Department of Molecular Medicine, Karolinska Institutet, Stockholm, Sweden2005 30 11 2005 6 69 69 26 7 2005 30 11 2005 Copyright © 2005 von Gertten et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Traumatic brain injury (TBI) initiates a complex sequence of destructive and neuroprotective cellular responses. The initial mechanical injury is followed by an extended time period of secondary brain damage. Due to the complicated pathological picture a better understanding of the molecular events occurring during this secondary phase of injury is needed. This study was aimed at analysing gene expression patterns following cerebral cortical contusion in rat using high throughput microarray technology with the goal of identifying genes involved in an early and in a more delayed phase of trauma, as genomic responses behind secondary mechanisms likely are time-dependent.
Results
Among the upregulated genes 1 day post injury, were transcription factors and genes involved in metabolism, e.g. STAT-3, C/EBP-δ and cytochrome p450. At 4 days post injury we observed increased gene expression of inflammatory factors, proteases and their inhibitors, like cathepsins, α-2-macroglobulin and C1q. Notably, genes with biological function clustered to immune response were significantly upregulated 4 days after injury, which was not found following 1 day. Osteopontin and one of its receptors, CD-44, were both upregulated showing a local mRNA- and immunoreactivity pattern in and around the injury site. Fewer genes had decreased expression both 1 and 4 days post injury and included genes implicated in transport, metabolism, signalling, and extra cellular matrix formation, e.g. vitronectin, neuroserpin and angiotensinogen.
Conclusion
The different patterns of gene expression, with little overlap in genes, 1 and 4 days post injury showed time dependence in genomic responses to trauma. An early induction of factors involved in transcription could lead to the later inflammatory response with strongly upregulated CD-44 and osteopontin expression. An increased knowledge of genes regulating the pathological mechanisms in trauma will help to find future treatment targets. Since trauma is a risk factor for development of neurodegenerative disease, this knowledge may also reduce late negative effects.
==== Body
Background
Traumatic brain injury (TBI) is a complex disorder, often with a severe or deadly outcome. The primary phase with tissue disruption initiates secondary injury processes in a delayed phase causing pathophysiological changes in the brain. As a consequence of the initial mechanical impact to the head, cerebral metabolism, blood flow and ion homeostasis are altered for a period of hours to days and even months [1]. During the secondary injury high levels of glutamate, Ca2+ and lactate are released, and cytokines are produced, leading to an inflammatory response, which all contribute to further tissue damage [2]. Beside the destructive processes, endogenous neuroprotective events in repair and regeneration also occur [3]. However, harmful processes dominate and eventually lead to tissue loss due to cell death. A challenge lies in understanding the molecular mechanisms behind the pathological processes, and the complex interplay between the different genes and their pathways operating after TBI. Despite a growing literature and extensive research on TBI, current clinical treatments are insufficient to support the repair processes and obstruct secondary injuries why preventive measures might be the most efficient way to improve outcomes.
The disrupted homeostasis in energy and oxygen supply leads to activation of several systems. Transcription, growth, plasticity, differentiation, signalling, inflammation, and cell death genes are affected in different trauma models. The injury alters mRNA and/ or protein levels for e.g. growth factors like NGF, BDNF, and the IGF system [4,5], apoptotic and anti-apoptotic factors like caspases, bax and bcl-2 [6,7], immediate early genes such as c-fos, c-jun and JunB [8], inflammatory markers like interleukines [9] and heat shock proteins [10].
The microarray technique gives an opportunity to simultaneously look at gene expression changes in a large number of genes. It is thus a means to cope with multiple processes, which may well be a prerequisite to handle the complexity of TBI. Trauma reports in mice and rat have mainly focused on early time points around 2 – 72 hours after trauma [11-17]. Later time-points after TBI have not been studied in the rat, although late changes in mice show similarities to those induced by ischaemia in rats [18]. Genes operating at later time-points may well be linked to the continuous brain tissue damage and secondary injuries, which develops after impact. As molecular responses to trauma are time-dependent, we have compared RNA expression after TBI in individual rats 1 and 4 days post injury (dpi). We used the weight-drop technique to produce a cerebral cortical contusion (CCC) [19], which mimics the clinical situation of focal contusion in patients. Alterations in RNA levels in the injured cortex were compared to the uninjured cortex and analysed with a cDNA microarray containing 6200 probes. We found more affected genes 1 dpi than 4 dpi with little overlap existing between the two time points. Significant differences between 1 and 4 dpi were seen in groups of upregulated genes involved in e.g. transcription, metabolism and cell proliferation. Genes involved in proteolysis and immune responses were significantly overrepresented in the delayed phase, which probably are important in secondary injuries.
Results
cDNA microarray
A cDNA microarray containing 6200 gene probes was used to study changes in gene expression induced by CCC in the ipsilateral cortex (including site of injury) in comparison to the contralateral uninjured cortex. At 1 dpi, the expression of 150 genes was significantly increased [see Additional file 1] while 61 genes were downregulated [see Additional file 2]. In contrast, 4 dpi displayed 56 upregulated genes [see Additional file 1], while 7 were downregulated [see Additional file 2]. Only 20 genes were upregulated both 1 and 4 dpi, while none of the downregulated genes were in common for the two time points (Fig. 1).
The genes were grouped in functional categories using the functional classification proposed by the Gene Ontology Consortia, to allow for analysis of the response 1 and 4 dpi. A comparison of frequency distribution between gene ontology (GO) categories related to biological function, between the whole set of expressed genes and upregulated genes (Fig. 2), showed that both 1 and 4 dpi displayed a significant overrepresentation (p < 0.05) of upregulated genes in the categories 'cell differentiation' (GO:0030154), 'cellular defence response' (GO:0006968) and 'response to stimuli' (GO:0050896). Additionally at 1 dpi, 'cell growth' (GO:0016049), 'transport' (GO:0006810), 'development' (GO:0007275), 'cell death' (GO:0008219) and 'regulation of cell cycle' (GO:0000074) showed significant upregulation. Four days post injury, significant upregulation of 'immune response' (GO:0006955) and 'proteolysis and peptidolysis' (GO:0006508) were observed. Comparing the groups of downregulated genes to the whole set of expressed genes resulted in no significant category with the frequency distribution test, neither for 1 dpi nor 4 dpi. Other regulated genes were found in most functional categories indicating that response to injury is complex and affects many different biological processes.
Next, we analysed if there were statistically significant differences between the distributions of genes in functional categories when comparing 1 and 4 dpi (Fig. 2). Significantly (p < 0.05) different distribution of upregulated genes among GO categories were detected for 'cell communication' (GO:0009987), 'cell death' (GO:0008219), 'cell proliferation' (GO:0008283), 'metabolism' (GO:0008152) and 'transcription' (GO:0006350) clearly indicating the qualitative differences in response to trauma at the two different time points. There were no statistical differences in the GO categories between downregulated genes comparing 1 and 4 dpi. However, only seven genes were found to be downregulated at 4 dpi.
Regulated genes were also grouped manually, without taking multiple functions for a gene in consideration. 'Transcription and translation' was a large group at 1 dpi, whereas at 4 dpi genes in the groups 'proteases and their inhibitors' and 'ECM (extra cellular matrix) and cytoskeleton' were well represented [see Additional file 1]. Among regulated genes with decreased expression, 'transporters, channels and binding proteins' and 'metabolism' were large groups at day 1 [see Additional file 2].
PCR, in situ hybridisation and immunohistochemistry
CD-44, osteopontin (Opn), TIMP-1 and -2 (tissue inhibitors of matrix metalloproteinases), S-100 (injury marker), angiotensinogen (precursor for angiotensin), insulin-like growth factor (IGF)-II, and vitronectin were further analysed by PCR confirming genes as differentially expressed, in line with the microarray results (Fig. 3). Among the set of regulated genes, Opn and CD-44 have potentially important regulatory roles in brain injury. CD-44 is a receptor for Opn and interestingly, CD-44-/- mice are protected from ischaemia induced injury [20]. The fact that both ligand and receptor are upregulated suggests a possible auto-/paracrine regulating loop at the site of the injury involving these genes. To further explore these findings, the expression of Opn and CD-44 were studied by in situ hybridisation and immunohistochemistry. Opn mRNA expression was confined to the injury site (Fig. 4) at both 1 and 4 dpi consistent with the microarray result. Its immunoreactivity was co-localised to ED-1, a marker for macrophages and activated microglia (Fig. 5). CD-44 showed a similar local mRNA expression pattern in the damaged region (Fig. 4) with a matching immunoreactivity (Fig. 5) 4 dpi. Nuclear staining with Hoescht showed cytoplasmatic CD-44 expression. Neither in situ hybridisation nor immunohistochemistry revealed contralateral expression for either CD-44 or Opn.
Discussion
In this study we have identified 211 genes that were regulated at 1 dpi and 63 at 4 dpi following a mild focal experimental contusion. The changes at 1 dpi were either initiated by traumatic events that affected gene activation directly, such as depolarisation and increase of intracellular calcium, or as a reaction to traumatically disrupted membranes, ischaemia and increased metabolic demands. At the early time-point, many genes involved in metabolism were affected, such as lactate dehydrogenase and PDH phosphate, which would increase the rate of pyruvate utilisation. Likewise, genes involved in cell growth, cellular morphogenesis, development, transport, cell differentiation, cellular defence response and regulation of cell cycle were significantly upregulated. This was also the case for 'response to stimulus' genes, confirming TBI as a strong inducer of genomic responses. Genes in this group included early mediators such as hsp, Nfkb1, p53; many of which have been identified upregulated previously in experimental trauma [21]. Activation of these genes agrees well with the prevailing theories of response to trauma: ictus leads to activation of reactions that require energy while mitochondrial dysfunction may limit aerobic metabolism.
It is maybe not surprising that more genes were regulated at day 1 than day 4 after injury. Several of the early responding genes were transcription factors with potential to initiate further gene regulation. It is probable that early responses were, to some extent, chaotic while 4 dpi findings represented a situation with less ongoing reparative and/or destructive cellular processes. It may be futile to search for specific destructive pathways with a hope to inhibit these in the early stages: broad gene activation may be non-specific and reflect that the trauma-energy has overcome the activation energy requirements for many different reactions. Our experimental injury model, CCC, creates a focal contusion lesion but does not cause major neurological deficits, prolonged unconsciousness or death, and is therefore considered a mild injury. The experimental injury resembles clinical contusions in patients who have a favourable condition early after injury, but are at risk of deterioration from progression of the focal contusions. This is a typical cause of patients who "talk and die" [22], where energy transfer at the impact has not been sufficient to destroy main vital functions of the brain, but has initiated reactions that destroy neural function in a delayed manner. This is in line with the observation that genes with a role in cell death were markedly overrepresented at 1 dpi and differed significantly from 4 dpi, which could initiate neuronal apoptosis that occurs for a prolonged time after injury in this model [7].
Our study showed significant differences between 1 and 4 dpi for genes with increased expression involved in cell communication, cell proliferation, metabolism and transcription. However, similarly to 1 dpi, 4 dpi also displayed significant upregulation of genes involved in cellular differentiation and response to stimuli. An analysis of the stimulus response 4 dpi showed significance in immune response, which was not present at 1 dpi. The view of the brain as immunologically privileged has been re-evaluated during the last years. Neuroinflammation, either acute in trauma or chronic in neurodegenerative diseases, has been revealed as a pathological mechanism [23]. The defence response in TBI could generate chronic harmful stimuli resulting in neurodegeneration, while it also paves the way for repair. The immune response is likely to be involved in the secondary injury mechanisms working at the later time point.
Opn, strongly upregulated both 1 and 4 dpi, is involved in inflammation, formation of the ECM and cell-matrix interactions. Opn modulates processes like mitochondrial respiration [24] and nitric oxide inhibition [25], affected in TBI. The diverse roles of Opn depend on receptor interaction and phosphorylation state [26]. Opn actions are mediated by integrins and CD-44, a cell surface glycoprotein. Interestingly, CD-44 also showed increased expression levels at 4 dpi. As both Opn and CD-44 are involved in inflammation and neurodegeneration [27] we chose to study these molecules in more detail. Opn mRNA expression and immunoreactivity were confined to the injury site, which co-localised with the macrophage and activated microglia marker ED-1. Opn can induce migration of astrocytes in vitro, and has therefore been proposed as an "astrokine" [28]. In our model this could mean that macrophages attract astrocytes to the damaged area, which would be in line with the above proposition. Furthermore, Opn may contribute to the nonpermissive adult CNS milieu, in the glial scar, and inhibiting axon outgrowth [29]. CD-44 mRNA was found locally around and in the lesion area 1 and 4 dpi. This has not been reported previously in this experimental contusion model, but agrees with CD-44 expression after cortical incision in mice [30]. We saw a corresponding immunoreactivity, resembling Opn immunoreactivity, in the penumbra and injured area, indicating interactions between Opn and CD-44. Our findings corroborate studies of upregulated Opn and CD-44 expression in other brain injury models [12,31]. Opn administration was recently reported to be neuroprotective in stroke [32] whereas increased CD-44 mRNA seems to be harmful in ischaemia [20]. It is possible that Opn interacting with CD-44 could activate a pathway leading to further injury, while Opn-integrin interactions would confer neuroprotection. However, other genes affected following injury may also influence the result of Opn expression.
Wound healing in the brain as in other injured tissues requires synthesis of new ECM molecules. The upregulation of fibronectin and collagen expression at 4 dpi could reflect a temporary state with a new matrix, attempting to create an environment more suitable for cell migration. However, we also observed downregulation of vitronectin expression. Vitronectin is an extracellular glycoprotein [33], which could impair cell migration and attachment, and may be necessary for a favourable cell environment. ECM has more functions than mere structural support. For many cell types, ECM acts as a survival factor as cells deprived of matrix attachment are prone to die by apoptosis, and a distorted matrix network also may hinder actions of trophic factors due to defective presentation. Cell attachment generates intracellular signals ending in regulation of cytoskeletal organisation and gene expression [34]. The bridging molecules ezrin and moesin, which were upregulated 1 dpi, and 1 and 4 dpi respectively, probably influence extracellular signals to the cytoskeleton. A way to maintain homeostasis by stabilising the cytoskeleton, might be upregulation of the intermediate filament vimentin (here increased 1 and 4 dpi).
Proteolytic enzymes cause structural ECM changes. Here, CCC altered the expression of many genes for matrix- and matrix modelling molecules reflected in the prominent category 'proteolysis and peptidolysis' specifically at 4 dpi. In this category we found cathepsins but also proteases such as myelencephalon specific protease, which, to our knowledge, is described regulated in this experimental model for the first time. Though proteases are required for re-organisation and plasticity in order to break down obstructing debris, there might be an imbalance in expression between them and their inhibitors, like upregulation of TIMP-1 and TIMP-2 mRNA, or downregulation of neuroserpin. Proteases challenge blood brain barrier (BBB) function. The upregulated MMP-9 (matrix metalloproteinase), involved in BBB breakdown [35], is one marker of secondary injury processes. Further on, the precursor of angiotensin, angiotensinogen, which is involved in BBB reconstitution, was here downregulated 1 dpi [36].
Our experimental contusion resulted in altered mRNA levels for genes in the category 'transcription response' with significant differences between 1 and 4 dpi. Many genes, among them ATF (activating transcription factor)-4, rad (ras associated with diabetes) and retinoic acid receptor, were upregulated at the early time-point, while only a few genes remained altered at the later phase. A novel finding here was the increased expression at 1 dpi of CCAAT/enhancer binding protein (C/EBP)-δ belonging to C/EBP (bZIP) family, which couples extracellular signal transduction pathways to numerous cellular processes and is a potential tumour suppressor gene [37]. Activated C/EBP-δ regulates the neuroprotective IGF-1 [38] and neurotoxic iNOS expression [39], both affected by TBI. Interestingly, another family member, C/EBP-β, is involved in brain injury [40] and neuronal survival [41]. Moreover, C/EBP-β and -δ expression are upregulated by inflammatory stimuli. It is therefore possible that C/EBP-δ also would influence neuronal survival. Another regulated transcription factor belonging to the immediate early gene family was v-myc (homologue to c-myc). Myc is a key molecular integrator of cell cycle machinery and metabolism [42], and many of its target genes were seen upregulated after CCC in this study. Prothymosin-α (PT-α) is a c-myc target [43], associated with cell growth, transcription [44] and recently also suggested to be involved in apoptosis [45]. To our knowledge, we here show for the first time that it's expression level is upregulated following trauma. This knowledge combined with the fact that TBI is a risk factor for neurodegernative disorders with excessive cell death suggests that PT-α may well be a potential candidate for further studies with regard to TBI. Recently, PT-α was shown to associate with STAT-3, an acute phase response factor, resulting in nuclear translocation [46]. In this study increased STAT-3 mRNA levels were noted 1 dpi, corroborating other studies [15,29].
Microarray analyses create large amounts of data and would ideally allow understanding of common regulatory pathways. Database developments with possibilities to better compare similarities and differences between various experimental models will be advantageous when forming new hypotheses. Whether the similarities or differences between affected genes turn out to be the most important factors is empirical: trauma needs to be studied from all its angles.
Conclusion
This study of genomic responses to trauma comparing 1 dpi to 4 dpi showed significant time-dependent differences, such as the early response in the category 'transcription response' and a more delayed response in the category 'immune defence'. The upregulation of CD-44 and Opn, localised to injury site, probably have important roles in the inflammatory process following TBI. Outcome of their interaction needs to be further studied in order to understand beneficial or detrimental roles. Understanding the pathological mechanisms behind secondary insults and the interplay of operating genes, may help to find treatment targets for brain injury and reduce delayed negative effects, such as neurodegenerative diseases.
Methods
Experimental injury model
All studies were conducted in accordance with the guidelines of the regional ethics committee for animal research at the Karolinska University Hospital, Stockholm, Sweden. Totally, 15 male Sprague Dawley rats (BW 250 g; B&K Universal AB, Stockholm, Sweden) were included in the study. Before surgery all rats were anaesthetised with an intramuscular injection of 0.2 ml Hypnorm™ -Dormikum (1:1:2; Hypnorm™ (fentanyl citrate 0.315 mg/ml, fluanisone 10 mg/ml, Janssen Pharmaceutica, Beerse, Belgium): Dormikum (1 mg/ml, Roche AB, Stockholm, Sweden): dH2O). Prior to the skin incision, 0.1 ml Xylocaine® (5 mg/ml, Astra, Södertälje, Sweden) was injected subcutaneously in the sagittal midline of the skull and the rats were placed in a stereotactic frame.
Cerebral cortical contusion
Contusions were performed on 13 rats using the weight-drop model described by Feeney et al. [19]. In brief, a craniotomy was made 2.5 mm posterior and 2.5 mm lateral to bregma, and a footplate was placed so that it rested upon the surface of the dura. A stainless tube guided a 24 g weight, which was dropped from a height of 9.3 cm, compressing the tissue at a maximum of 3 mm. After impact, the scalp was sutured and the animals were allowed to recover. Sham operation, craniotomy without contusion, was performed on two animals. The rats were sacrificed at one day (n = 6) and four days (n = 9) by decapitation in Hypnorm™ anaesthesia. Brains (1 dpi n = 3; 4 dpi n = 4) were removed, and impact area with surrounding cortex, ipsilateral and contralateral was dissected out, before quickly frozen in isopentan containing dry ice, prior to RNA isolation. The remaining brains (1dpi n = 3, 4 dpi n= 3, sham n = 2) were removed for sectioning and directly frozen in isopentan – dry ice. Coronal 14 μm cryosections were cut through the center of the impact using a Leica cryostat (CM 3000, Leica Instruments GmbH, Nussloch, Germany). The sections were thaw-mounted onto Super Frost/Plus™ object glasses (Menzel-Gläser, Braunschweig, Germany) and stored at -20°C prior to use.
RNA extraction
Total RNA was isolated using RNeasy Qiagen kit (VWR International AB, Stockholm, Sweden) according to manufacturer's protocol. Brain tissue was homogenised using a polytron. The RNA was dissolved in diethyl pyrocarbonate (DEPC) treated dH2O and quantified by spectrophotometry at A260 and A280. Quality was verified on a 3-(N-morpholino)-propanesulfonic acid (MOPS)-formaldehyde-agarose gel.
cDNA microarrays
The cDNA microarrays were described earlier [47] but now extended to comprise about 6200 clones. Clones were selected from the TIGR Rat gene Index [48], Research Genetics [49] and from the lab obtained during differential cloning experiments. Arrays were pre-hybridised in 1% bovine serum albumin (BSA), 5 × SSC (1 × SSC: 0.15 M NaCl, 0.015 M sodium citrate) and 0.1% sodium dodecyl sulfate (SDS), at 42°C for 1–2 hours, washed in milli-Q H2O and dried immediately before the probe was applied.
cDNA labelling, purification, and hybridisation
Total RNA (30 μg) was used from individual animals in each hybridisation. Labelled cDNA was produced using an oligo-dT primer and Cy3-/ Cy5-uridine 5'-triphosphate labelled nucleotides (PerkinElmer, MA) in reverse transcription using Superscript II (Life Technologies Inc., NY). Cy3- and Cy5-labelled cDNAs were pooled and purified using Microcon 30 columns (Millipore, MA) and then adjusted to a final volume of 25 μl with hybridisation buffer [3.4 × SSC, 0.3% SDS, 20 μg mouse Cot-1 DNA (Invitrogen, CA), 20 μg poly A RNA, 20 μg yeast tRNA]. After heating at 98°C for 2 min and cooling to room temperature, the probe was added to the array and covered by a plastic cover-slip, put in a sealed hybridisation chamber (Corning Inc., NY), and hybridised at 65°C for 15–18 hours. Then the array was washed, dried and immediately scanned with a GMS 418 scanner (Affymetrix, CA). Analyses were made on individual animals and with dye-swap to account for dye-biased effects; 1 dpi n= 3 (two animals with ipsilateral total RNA labelled with Cy5 and one animal with ipsilateral total RNA labelled with Cy3) and 4 dpi n= 4 (two animals with ipsilateral total RNA labelled with Cy5 and two with ipsilateral total RNA labelled with Cy3).
Data processing and analysis
Image analysis was performed with GenePix Pro software (Axon instruments, CA). Automatic and manual flagging were used to localise absent or very weak spots (< 2 times above background), which were excluded from analysis. The signal from each spot was calculated as the average intensity minus the average local background. Expression ratios of Cy5/Cy3 (or Cy3/Cy5 in case of dye-swap) were normalised using a method that takes into account and corrects for intensity-dependent artefacts in the measurements; the locally weighted linear regression (Lowess) method in the SMA package (Statistics for Microarray Analysis)[50,51]. SMA is an add-on library written in the public domain statistical language R.
The significance of expression ratios was statistically evaluated using the SAM (Significance Analysis of Microarrays) technique [52]. Similarly to the familiar p-value, a q-value was assigned each of the detectable genes in the array. The q-value measures the lowest false discovery rate (FDR) at which the gene is called significant. A 2% FDR was used to identify regulated genes. On top of the SAM criteria, a mean ratio cut off (log2 ratio injured/uninjured > 0.7 corresponding to > 1.6-fold regulation) was applied to describe ratios as up-/down-regulated and was applied to each hybridised microarray. Using these settings one could expect three genes falsely identified as regulated among the 150 upregulated 1 dpi and one gene among those identified 4 dpi. Data has been deposited in Gene Expression Omnibus [53].
We used the web based tool eGOn v 1.0 (explore Gene Ontology, developed at the Norwegian University of Science and Technology) [54] to functionally classify the transcripts. A total of 3414 genes 1 dpi and 2242 genes 4 dpi were deposited in eGOn of which 1272 genes for 1 dpi and 904 genes for 4 dpi were annotated by eGOn and categorised into gene ontology (GO) categories related to biological function. Each list of differentially expressed genes were compared to all genes expressed, for 1 respectively 4 dpi with the two sided one-sample binomial test implemented in eGOn. Comparison was made to test if the proportion of genes upregulated at 1 dpi was different from the proportion of genes upregulated 4 dpi, in eGOn, which use McNemars test based on an implementation using the binomial distribution. Additionally, also using eGOn, upregulated genes were compared to downregulated genes for 1 and 4 dpi separately with Fishers exact test.
The regulated genes presented in tables 1 and 2 were grouped based on information from Gene ontology, DAVID (Database for Annotation and Visualisation and Integrated Discovery) [55], PubMed and other existing array reports. A gene can be annotated for several functions, which is not displayed in the tables where a gene is assigned to only one group. However, the statistical testing in eGOn has taken multiple functions for genes in consideration, why one and the same gene can be found in several of the categories made by eGOn. Therefore the groups in the tables do not correspond to the groups of biological function made by eGOn.
RT-PCR
cDNA templates for RT-PCR were generated from five μg of total RNA treated with DNase I (Roche Diagnostics Scandinavia AB, Bromma, Sweden) in a reaction with 1 U DNase I/μg RNA, 25 mM Tris (pH 8.0), 25 mM NaCl, 5 mM MgCl2 and 0.15 U rRNasin RNA inhibitor (Promega, Madison, WI) for 25 min at 37°C. Single stranded cDNA synthesis was made using PowerScript Reverse Transcriptase (Clontech, CA) according to manufacturer's protocol. RT-PCR was performed using Taq Dynazyme (Finnzymes, Espoo, Finland) under standard conditions (1 × Dynazyme buffer, 0.2 mM dNTPs (Life Technologies), 0.5 U Taq Dynazyme, 1 μM of each specific primer and 2 μl of cDNA) using a 4-min hot start at 94°C followed by 30 cycles of 94°C for 45 sec, 59°C for 45 sec, 72°C for 1 min, followed by 10 min final extension at 72°C. Glucose-6-phosphate dehydrogenase (G6PD) was co-amplified as an internal control in each reaction. PCR products were analysed by 1.5% agarose gel electrophoresis (Sigma, St Louis, MO), and visualised using ethidium bromide fluorescence. All primer pairs (Table 1) were obtained from MedProbe (MedProbe AS, Oslo, Norway).
In situ hybridisation
Synthetic oligonucleotide probes (Table 1) were synthesised and purified by reverse phase chromatography by Medprobe. The oligonucleotides were labelled at the 3' end with α-35S-dATP ((NEG034H) du Medical NEN, Bruxelles, Belgium) using terminal-deoxynucleotidyl-transferase TdT (Takara, Amersham Pharmacia Biotech, Uppsala, Sweden) at 37°C for 1 hour and purified using mini Quick Spin Oligo Columns (Roche Diagnostics Scandinavia AB). The specific activities obtained ranged from 1–4 × 109 cpm/μg oligonucleotide. 14 μm sections were air-dried for one hour and covered with a hybridisation buffer containing 50% formamide, 4 × SSC, 1 × Denhardt's solution (0.02% polyvinyl-pyrrolidone, 0.02% bovine serum albumin, and 0.02% Ficoll), 1% sarcosyl, 0.02 M phosphate buffer, 10% dextran sulphate (Amersham Pharmacia Biotech), 500 μg/ml heat-denatured salmon sperm DNA (Sigma) and 200 mM DTT (Amersham Pharmacia Biotech), and 1 × 107 cpm/ml labelled probe. The slides were incubated in a chamber humidified with 4 × SSC and 50% formamide for 16–20 hours at 42°C. After hybridisation, the sections were rinsed in 1 × SSC at 40°C, 4 × 15 min in 1 × SSC at 55°C, 1 × SSC and dH2O, 1 min each at room temperature followed by dehydration with 60% and 95% ethanol. Sections were air-dried and exposed on BioMax MR X-ray film (Eastman Kodak, Rochester, NY) at room temperature for 6 days.
Immunohistochemistry
Frozen sections were air-dried, rehydrated in 1 × phosphate-buffered saline (PBS) and fixed in 4% buffered paraformaldehyde for 7 min at room temperature, rinsed in 1 × PBS, and blocked with normal goat serum (1:150) in 1% BSA for 30 min at room temperature. Labelling was made overnight with a monoclonal mouse anti rat CD-44 antibody 1:100 (Serotec, Oxford, UK), a mouse monoclonal anti rat MPIIIB101 (osteopontin) antibody 1:50 (Developmental Studies Hybridoma Bank, IA), or mouse anti rat ED-1 antibody 1:1000 (Serotec) at 4°C. Sections were rinsed in 1 × PBS and incubated with fluorescent or indocarbocyanine (Cy3)-conjugated goat anti-mouse 1:1000 (Jackson Immunoresearch Lab. Inc, PA) or fluorescein isothiocyante (FITC)-conjugated goat anti mouse 1:150 (Jackson Immunoresearch Lab.) for 1 hour at room temperature. Nuclear staining with Hoescht was performed on slides with CD-44 immunolabelling. After washing, the slides were mounted with glycerol:PBS. Evaluation of staining was performed by fluorescence microscopy with Leica filter cube N2.1 (excitation filter: 515–560 nm, suppression filter edge wavelength: 590 nm) for detection of Cy3-labelling and Leica filter cube L4 (excitation filter: 450-490 nm, suppression filter edge wavelength: 515–560 nm) for FITC labelling. Photomicrographs for double labelling illustrations were obtained by changing filter cube without altering section position or focus.
Authors' contributions
SH performed the experimental neurosurgery with assistance of CvG. ACSN and CvG sampled and sectioned the tissue, and performed in situ hybridisation. CvG prepared the material and performed the cDNA microarray experiments, RT-PCR, and immunohistochemical stainings. AFM and CvG analysed the microarray data. CvG, AFM, TM, ACSN designed the study and drafted the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Upregulated genes 1 and 4 dpi following cerebral cortical contusion. The table shows clone identity (Clone ID), accession number (Acc no), gene name, and fold change 1 and 4 days post injury (dpi) after a cerebral cortical contusion (CCC). Bold numbers are regulated genes with fold change > 1.6 at a false discovery rate ≤ 2%. * = similar to, EST = expressed sequence tag. Negative signs denote downregulated fold changes.
Click here for file
Additional File 2
Downregulated genes 1 and 4 dpi following cerebral cortical contusion. The table shows clone identity (Clone ID), accession number (Acc no), gene name, and fold change 1 and 4 days post injury (dpi) after a cerebral cortical contusion (CCC). Bold numbers are regulated genes with fold change > 1.6 at a false discovery rate ≤ 2%. * = similar to, EST = expressed sequence tag. Negative signs denote downregulated fold changes.
Click here for file
Acknowledgements
This research is supported by Swedish Medical Research Council 13485 and 14824 and Karolinska Institutet. The MPIIIB101 (osteopontin) developed by M. Solursh and A. Franzen was obtained from the Developmental Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242.
Figures and Tables
Figure 1 Venn diagram illustrating regulated genes after cerebral cortical contusion 1 and 4 dpi. Overlapping area shows genes for which expression was altered at both time points. Number of regulated genes is noted in parenthesis.
Figure 2 Functional classification of regulated genes 1 and 4 days after cerebral cortical contusion. Up- and downregulated genes with fold changes >1.6 at a 2% false discovery rate, annotated and categorised by the gene ontology database eGOn. Note a significant (p < 0.05) overrepresentation of upregulated genes in relation to total number of expressed genes (*) in the functional groups of cell differentiation, defence response and response to stimulus, for both 1 and 4 days post injury (dpi), while 1 dpi also showed significance (p < 0.05) for genes involved in cell death, cell growth, cellular morphogenesis, development, regulation of cell cycle and transport. Additionally, genes with increased expression showed significant difference (p < 0.05) between 1 and 4 dpi in the groups of cell communication, cell death, cell proliferation, metabolism and transcription (¤) analysed by eGOn.
Figure 3 Confirmation of regulated genes after microarray analysis. Genes with altered expression analysed by reverse transcriptase-polymerase chain reaction (RT-PCR), comparing ipsilateral (i) to contralateral (c) hemisphere. OPN – osteopontin, TIMP-1/-2 – tissue inhibitor of matrix metalloproteinase, PAT – angiotensinogen, IGF-II – insulin like growth factor – II, and VTN – vitronectin.
Figure 4 Localisation of CD-44 and osteopontin mRNA expression. Autoradiographs following in situ hybridisation of rat brain sections after contusion showing CD-44 and OPN – osteopontin mRNA expression. Note intense mRNA signal in contusion area with no visible signal in contralateral hemisphere.
Figure 5 Fluorescence photomicrographs showing CD-44 and osteopontin immunoreactivity, 4 dpi. High-power photomicrographs of injured rat brains close to impact show immunofluorescence staining for CD-44 and osteopontin (OPN). Nuclear staining with Hoescht illustrates, in the merged picture, (CD-44/Hoescht) the cytoplasmatic staining for CD-44. Immunofluorescence staining for ED-1, a macrophage marker, demonstrates, in the merged picture, (OPN/ED-1) co-localisation with osteopontin. Scale bar = 50 μm.
Table 1 Primers and oligonucleotide probes used for confirmation of results.
Gene Primer/oligonucleotide probe sequence (nts) Product (bp) Acc. no
PAT F: 5'-GAAGCTAGAGGCTGAGGATC- 3'
R: 5'-GTGCAGTCTCCCTCCTTCAC- 3' 233 L00091
S-100 F: 5'-GGACCTGAGAGTGCTCATGG- 3'
R: 5'-GCATGCAATGATGAGCCCCG-3' 222 J03627.1
IGF-II F: 5'-GACTGAGTTGGGGCAAATAC- 3'
R: 5'-CAGGTGTTAGGAAGGTGCTC- 3' 211 AA899788
CD-44 F: 5'-CCGACCTTCCCACTTCACAG- 3'
R: 5'-TCTCCTCGCAGGACCAGAAG- 3'
5'- ATCACTGGTGGCCAGGGTGCTCCCA ATAAAGAAGGCGTCATCCC-3' (44-mer) 200 AA817820
OPN F: 5'-CTGCCAGCACACAAGCAGAC- 3'
R: 5'-ACTCCTTGGACTGCTCCAGG- 3'
5'-TCCTGATCAGAGGGCACGCTCAGAC GCTGGGCAACTGGGATGACC-3' (45-mer) 323 M14656
TIMP-1 F: 5'-CGAGACCACCTTATACCAGCG-3'
R: 5'-CAGGAAGCTGCAGGCAGTGAT-3' 216 L31883
TIMP-2 F: 5'-TGCACCCGCAACAGGCGTTTT-3'
R: 5'--TTCCTCCAACGTCCAGCGAGA-3' 224 L31884
VTN F: 5'-ATCGACGCTGCCTTCACTCG-3'
R: 5'-TGGCGCCATCAGAGGATCTG-3' 373 U44845
Nts = nucleotides, bp = base pairs, acc. no = accession number
==== Refs
Ray SK Dixon CE Banik NL Molecular mechanisms in the pathogenesis of traumatic brain injury Histol Histopathol 2002 17 1137 1152 12371142
Lenzlinger PM Morganti-Kossmann MC Laurer HL McIntosh TK The duality of the inflammatory response to traumatic brain injury Mol Neurobiol 2001 24 169 181 11831551 10.1385/MN:24:1-3:169
Rice AC Khaldi A Harvey HB Salman NJ White F Fillmore H Bullock MR Proliferation and neuronal differentiation of mitotically active cells following traumatic brain injury Exp Neurol 2003 183 406 417 14552881 10.1016/S0014-4886(03)00241-3
Oyesiku NM Evans CO Houston S Darrell RS Smith JS Fulop ZL Dixon CE Stein DG Regional changes in the expression of neurotrophic factors and their receptors following acute traumatic brain injury in the adult rat brain Brain Res 1999 833 161 172 10375691 10.1016/S0006-8993(99)01501-2
Sandberg Nordqvist AC von Holst H Holmin S Sara VR Bellander BM Schalling M Increase of insulin-like growth factor (IGF)-1, IGF binding protein-2 and -4 mRNAs following cerebral contusion Brain Res Mol Brain Res 1996 38 285 293 8793117 10.1016/0169-328X(95)00346-T
Morrison B Eberwine JH Meaney DF McIntosh TK Traumatic injury induces differential expression of cell death genes in organotypic brain slice cultures determined by complementary DNA array hybridization Neuroscience 2000 96 131 139 10683418 10.1016/S0306-4522(99)00537-0
Wennersten A Holmin S Mathiesen T Characterization of Bax and Bcl-2 in apoptosis after experimental traumatic brain injury in the rat Acta Neuropathol (Berl) 2003 105 281 288 12557016
Raghupathi R McIntosh TK Regionally and temporally distinct patterns of induction of c-fos, c- jun and junB mRNAs following experimental brain injury in the rat Brain Res Mol Brain Res 1996 37 134 144 8738144 10.1016/0169-328X(95)00289-5
Winter CD Pringle AK Clough GF Church MK Raised parenchymal interleukin-6 levels correlate with improved outcome after traumatic brain injury Brain 2004 127 315 20. Epub 2003 Nov 25. 14645145 10.1093/brain/awh039
Truettner J Schmidt-Kastner R Busto R Alonso OF Loor JY Dietrich WD Ginsberg MD Expression of brain-derived neurotrophic factor, nerve growth factor, and heat shock protein HSP70 following fluid percussion brain injury in rats J Neurotrauma 1999 16 471 486 10391364
Matzilevich DA Rall JM Moore AN Grill RJ Dash PK High-density microarray analysis of hippocampal gene expression following experimental brain injury J Neurosci Res 2002 67 646 663 11891777 10.1002/jnr.10157
Tang Y Lu A Aronow BJ Wagner KR Sharp FR Genomic responses of the brain to ischemic stroke, intracerebral haemorrhage, kainate seizures, hypoglycemia, and hypoxia Eur J Neurosci 2002 15 1937 1952 12099900 10.1046/j.1460-9568.2002.02030.x
Long Y Zou L Liu H Lu H Yuan X Robertson CS Yang K Altered expression of randomly selected genes in mouse hippocampus after traumatic brain injury J Neurosci Res 2003 71 710 720 12584729 10.1002/jnr.10524
Natale JE Ahmed F Cernak I Stoica B Faden AI Gene expression profile changes are commonly modulated across models and species after traumatic brain injury J Neurotrauma 2003 20 907 927 14588109 10.1089/089771503770195777
Raghavendra Rao VL Dhodda VK Song G Bowen KK Dempsey RJ Traumatic brain injury-induced acute gene expression changes in rat cerebral cortex identified by GeneChip analysis J Neurosci Res 2003 71 208 219 12503083 10.1002/jnr.10486
Rall JM Matzilevich DA Dash PK Comparative analysis of mRNA levels in the frontal cortex and the hippocampus in the basal state and in response to experimental brain injury Neuropathol Appl Neurobiol 2003 29 118 131 12662320 10.1046/j.1365-2990.2003.00439.x
Li HH Lee SM Cai Y Sutton RL Hovda DA Differential gene expression in hippocampus following experimental brain trauma reveals distinct features of moderate and severe injuries J Neurotrauma 2004 21 1141 1153 15453985 10.1089/0897715041953777
Kobori N Clifton GL Dash P Altered expression of novel genes in the cerebral cortex following experimental brain injury Brain Res Mol Brain Res 2002 104 148 158 12225869 10.1016/S0169-328X(02)00331-5
Feeney DM Boyeson MG Linn RT Murray HM Dail WG Responses to cortical injury: I. Methodology and local effects of contusions in the rat Brain Res 1981 211 67 77 7225844 10.1016/0006-8993(81)90067-6
Wang X Xu L Wang H Zhan Y Pure E Feuerstein GZ CD44 deficiency in mice protects brain from cerebral ischemia injury J Neurochem 2002 83 1172 1179 12437588 10.1046/j.1471-4159.2002.01225.x
Leker RR Shohami E Cerebral ischemia and trauma-different etiologies yet similar mechanisms: neuroprotective opportunities Brain Res Brain Res Rev 2002 39 55 73 12086708 10.1016/S0165-0173(02)00157-1
Mathiesen T Kakarieka A Edner G Traumatic intracerebral lesions without extracerebral haematoma in 218 patients Acta Neurochir (Wien) 1995 137 155 63, discussion 163. 8789656 10.1007/BF02187188
Streit WJ Mrak RE Griffin WS Microglia and neuroinflammation: a pathological perspective J Neuroinflammation 2004 1 14 15285801 10.1186/1742-2094-1-14
Gao C Guo H Wei J Kuo PC Osteopontin inhibits expression of cytochrome c oxidase in RAW 264.7 murine macrophages Biochem Biophys Res Commun 2003 309 120 125 12943672 10.1016/S0006-291X(03)01534-1
Rollo EE Laskin DL Denhardt DT Osteopontin inhibits nitric oxide production and cytotoxicity by activated RAW264.7 macrophages J Leukoc Biol 1996 60 397 404 8830797
Weber GF Zawaideh S Hikita S Kumar VA Cantor H Ashkar S Phosphorylation-dependent interaction of osteopontin with its receptors regulates macrophage migration and activation J Leukoc Biol 2002 72 752 761 12377945
Kim MD Cho HJ Shin T Expression of osteopontin and its ligand, CD44, in the spinal cords of Lewis rats with experimental autoimmune encephalomyelitis J Neuroimmunol 2004 151 78 84 15145606 10.1016/j.jneuroim.2004.02.014
Ellison JA Barone FC Feuerstein GZ Matrix remodeling after stroke. De novo expression of matrix proteins and integrin receptors Ann N Y Acad Sci 1999 890 204 222 10668427
Kury P Zickler P Stoll G Hartung HP Jander S Osteopontin, a macrophage-derived matricellular glycoprotein, inhibits axon outgrowth Faseb J 2004 29 29
Jones LL Liu Z Shen J Werner A Kreutzberg GW Raivich G Regulation of the cell adhesion molecule CD44 after nerve transection and direct trauma to the mouse brain J Comp Neurol 2000 426 468 492 10992250 10.1002/1096-9861(20001023)426:3<468::AID-CNE9>3.0.CO;2-I
Shin T Ahn M Kim H Moon C Kang TY Lee JM Sim KB Hyun JW Temporal expression of osteopontin and CD44 in rat brains with experimental cryolesions Brain Res 2005 1041 95 101 15804504 10.1016/j.brainres.2005.02.019
Meller R Stevens SL Minami M Cameron JA King S Rosenzweig H Doyle K Lessov NS Simon RP Stenzel-Poore MP Neuroprotection by osteopontin in stroke J Cereb Blood Flow Metab 2005 25 217 225 15678124 10.1038/sj.jcbfm.9600022
Niquet J Gillian A Ben-Ari Y Represa A Reactive glial cells express a vitronectin-like protein in the hippocampus of epileptic rats Glia 1996 16 359 367 8721676 10.1002/(SICI)1098-1136(199604)16:4<359::AID-GLIA8>3.0.CO;2-V
Zalewska T Ziemka-Nalecz M Sarnowska A Domanska-Janik K Transient forebrain ischemia modulates signal transduction from extracellular matrix in gerbil hippocampus Brain Res 2003 977 62 69 12788514 10.1016/S0006-8993(03)02742-2
Rosenberg GA Estrada EY Dencoff JE Matrix metalloproteinases and TIMPs are associated with blood-brain barrier opening after reperfusion in rat brain Stroke 1998 29 2189 2195 9756602
Kakinuma Y Hama H Sugiyama F Yagami K Goto K Murakami K Fukamizu A Impaired blood-brain barrier function in angiotensinogen-deficient mice Nat Med 1998 4 1078 1080 9734405 10.1038/2070
Huang AM Montagna C Sharan S Ni Y Ried T Sterneck E Loss of CCAAT/enhancer binding protein delta promotes chromosomal instability Oncogene 2004 23 1549 1557 14716301 10.1038/sj.onc.1207285
Ji C Chang W Centrella M McCarthy TL Activation domains of CCAAT enhancer binding protein delta: regions required for native activity and prostaglandin E2-dependent transactivation of insulin-like growth factor I gene expression in rat osteoblasts Mol Endocrinol 2003 17 1834 43. Epub 2003 Jun 5. 12791769 10.1210/me.2002-0235
Won JS Im YB Key L Singh I Singh AK The involvement of glucose metabolism in the regulation of inducible nitric oxide synthase gene expression in glial cells: possible role of glucose-6-phosphate dehydrogenase and CCAAT/enhancing binding protein J Neurosci 2003 23 7470 7478 12930785
Cortes-Canteli M Wagner M Ansorge W Perez-Castillo A Microarray analysis supports a role for ccaat/enhancer-binding protein-beta in brain injury J Biol Chem 2004 279 14409 17. Epub 2004 Jan 21. 14736879 10.1074/jbc.M313253200
Marshall J Dolan BM Garcia EP Sathe S Tang X Mao Z Blair LA Calcium channel and NMDA receptor activities differentially regulate nuclear C/EBPbeta levels to control neuronal survival Neuron 2003 39 625 639 12925277 10.1016/S0896-6273(03)00496-3
Dang CV c-Myc target genes involved in cell growth, apoptosis, and metabolism Mol Cell Biol 1999 19 1 11 9858526
Boan F Vinas A Buceta M Dominguez F Sanchez L Gomez-Marquez J Prothymosin alpha, a mammalian c-myc-regulated acidic nuclear protein, provokes the decondensation of human chromosomes in vitro Cytogenet Cell Genet 2001 93 171 174 11528108 10.1159/000056980
Karetsou Z Kretsovali A Murphy C Tsolas O Papamarcaki T Prothymosin alpha interacts with the CREB-binding protein and potentiates transcription EMBO Rep 2002 3 361 6. Epub 2002 Mar 15. 11897665 10.1093/embo-reports/kvf071
Jiang X Kim HE Shu H Zhao Y Zhang H Kofron J Donnelly J Burns D Ng SC Rosenberg S Wang X Distinctive roles of PHAP proteins and prothymosin-alpha in a death regulatory pathway Science 2003 299 223 226 12522243 10.1126/science.1076807
Yang CH Murti A Baker SJ Frangou-Lazaridis M Vartapetian AB Murti KG Pfeffer LM Interferon induces the interaction of prothymosin-alpha with STAT3 and results in the nuclear translocation of the complex Exp Cell Res 2004 298 197 206 15242774 10.1016/j.yexcr.2004.04.008
Flores-Morales A Stahlberg N Tollet-Egnell P Lundeberg J Malek RL Quackenbush J Lee NH Norstedt G Microarray analysis of the in vivo effects of hypophysectomy and growth hormone treatment on gene expression in the rat Endocrinology 2001 142 3163 3176 11416039 10.1210/en.142.7.3163
TIGR Rat gene Index
Research Genetics
SMA package (Statistics for Microarray Analysis)
Quackenbush J Microarray data normalization and transformation Nat Genet 2002 32 496 501 12454644 10.1038/ng1032
Tusher VG Tibshirani R Chu G Significance analysis of microarrays applied to the ionizing radiation response Proc Natl Acad Sci U S A 2001 98 5116 21. Epub 2001 Apr 17. 11309499 10.1073/pnas.091062498
Gene Expression Omnibus [GEO: GSE3396]
eGOn v 1.0 (explore Gene Ontology)
DAVID (Database for Annotation and Visualisation and Integrated Discovery)
|
16318630
|
PMC1310614
|
CC BY
|
2021-01-04 16:03:49
|
no
|
BMC Neurosci. 2005 Nov 30; 6:69
|
utf-8
|
BMC Neurosci
| 2,005 |
10.1186/1471-2202-6-69
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-1041628865310.1186/1471-2334-5-104Research ArticleMechanisms of escape phenomenon of spinal cord and brainstem in human rabies Juntrakul Sasiwimol [email protected] Preecha [email protected] Shanop [email protected] Supaporn [email protected] Thiravat [email protected] Molecular Biology Laboratory for Neurological Diseases, Department of Medicine, Chulalongkorn University Hospital, Rama 4 Road, Bangkok, Thailand2 Department of Pathology, Chulalongkorn University Hosital, Rama 4 Road, Bangkok, Thailand2005 16 11 2005 5 104 104 10 8 2005 16 11 2005 Copyright © 2005 Juntrakul et al; licensee BioMed Central Ltd.2005Juntrakul et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Rabies virus preferentially involves brainstem, thalamus and spinal cord in human furious and paralytic rabies beginning in the early stage of illness. Nevertheless, rabies patient remains alert until the pre-terminal phase. Weakness of extremities develops only when furious rabies patient becomes comatose; whereas peripheral nerve dysfunction is responsible for weakness in paralytic rabies.
Methods
Evidence of apoptosis and mitochondrial outer membrane permeabilization in brain and spinal cord of 10 rabies patients was examined and these findings were correlated with the presence of rabies virus antigen.
Results
Although apoptosis was evident in most of the regions, cytochrome c leakage was relatively absent in spinal cord of nearly all patients despite the abundant presence of rabies virus antigen. Such finding was also noted in brainstem of 5 patients.
Conclusion
Cell death in human rabies may be delayed in spinal cord and the reticular activating system, such as brainstem, thus explaining absence of weakness due to spinal cord dysfunction and preservation of consciousness.
==== Body
Background
Clinical presentations of rabies in humans can be categorized as classic (furious and paralytic) and non-classic rabies [1,2]. The latter is almost always associated with bat and some dog variants whereas the classic forms are associated with dog variants. There is no specific genetic pattern of rabies virus associated with either furious or paralytic forms based on an analysis of glyco-, phospho-, nucleoprotein genes [3]. Analysis of regional distribution of rabies virus antigen in the central nervous system (CNS) revealed similar pattern [4]. Rabies virus antigen preferentially localizes in the spinal cord and brainstem and thalamus, basal ganglia if the survival period is 7 days or less regardless of the clinical forms [4]. Such brainstem and thalamus predilection is also evident in animals [5].
Serial electrophysiologic studies of peripheral nerve in furious rabies patients revealed a sub-clinical evidence of anterior horn cell dysfunction in the spinal cord [6,7]. These patients did not exhibit any demonstrable weakness of the arms and legs. It is only at the time when furious rabies patients become comatose that weakness of all limb musculatures can be demonstrated. In paralytic rabies patients, limb weakness is explained by peripheral nerve and not the anterior horn cell dysfunction [6,7]. This is also confirmed by prominent inflammation and demyelination in the peripheral nerve of these paralytic rabies patients [6-10].
These findings raised important questions why clinical weakness due to spinal cord dysfunction does not develop in rabies patients. Along with this "escape phenomenon" of rabies virus infected spinal cord, it is also intrigued that rabies patients do not have depressed consciousness during the most entire clinical course despite an enormous amount of rabies virus since the early stage in the brainstem and thalamus, structures which are crucial in maintaining alertness and form an integral part of reticular activating system.
Spinal cord motoneuron resists to cytolysis and apoptosis in spinal cord and anterior horn cell culture system with rabies virus infection [11]. In vivo, despite the massive infection of the spinal cord in infected rat neonates, only a few motoneurons were apoptotic. Moreover, axon of rabies infected motoneuron was able to elongate at a comparable rate in virus-infected and noninfected cultures indicating that metabolic activity was maintained in these infected cells. In contrast, a large proportion of hippocampus neurons were apoptotic shortly after infection. These suggest that spinal cord motoneurons survive rabies virus infection because the viral induction of apoptosis is delayed in these neurons.
Apoptosis or programmed cell death can be induced by multiple insults which proceeds through the mitochondrial pathway – mitochondrial outer membrane permeabilization (MOMP) in which cytochrome c appears to be a major inducer of the entire cascade though the activation of caspase-9 and -3 [12].
The objective of our study was to determine whether brainstem and spinal cord in rabies patients are lacking of either apoptosis or mitochondrial outer membrane permeabilization (MOMP) or both which, in turn, may explain this escape phenomenon. We also compared the degree of rabies virus infection and apoptosis and MOMP at various CNS regions of furious and paralytic rabies patients.
Materials and methods
Materials
Formalin fixed paraffin embedded CNS tissue of 10 rabies patients processed between 1987 and 2005 were included in this study. Five rabies patients presented as furious and the remaining had paralysis. Characteristics of these patients were summarized in Table 1. All patients did not receive any intensive care support and had evidence of hypoxia and cardiovascular collapse during the last 24–36 hours before death. Post-mortem examinations were performed within 24 h of death. Brain and spinal cord were fixed in formalin for 7 days. Sections of frontal, temporal, hippocampus, parietal, occipital, thalamus, basal ganglia, cerebellum, midbrain, pons, medulla, cervical, thoracic, lumbosacral enlargement were subsequently embedded in paraffin, sectioned and examined for the presence of rabies virus antigen and apoptosis and MOMP.
Table 1 Characteristics of patients with rabies.
Patient No. Age (yrs), Sex Incubation time and site of bite Survival period (days)* History of Immunization
Furious
H1 9, male 1 month; dog bite on right buttock 5 -
H3 22, female 3 months; dog bite on right ankle 8 -
H5 55, female 2 months; dog bite on right hand, left leg, left breast 4 -
H6 14, male 1 month; dog bite on buttock 5 -
H8 31, male 3 months before; dog bite on right foot 5 -
Paralysis
H2 15, female 3 months; dog bite on finger 16 -
H4 81, male 2 months; dog bite on left calf 7 +
H7 61, male 4 months; dog bite on right leg 13 -
H9 43, female 3 months before; dog bite on left hand 9 -
H10 18, male uncertain; dog bite on right leg 13 -
* = interval between onset and death
Methods
1. Slide preparation
Three μm-thick paraffin sections of formalin-fixed tissue were mounted on silane-coated slides [2% 3-aminopropyltriethoxysilane (Sigma, USA)].
2. Immunoperoxidase staining for rabies antigen and cytochrome c
Sections were stained by DAKO EnVision™ System kit, HRP (DAKO Corporation, CA, USA) after deparaffinized by xylene and ethanol and then antigen was retrieved by pressure cooker with citrate buffer for 1 min. Briefly, sections were incubated for 10 min with 3%H2O2 to eliminate endogenous peroxidase, washed in phosphate-buffered saline (PBS), incubated for 20 min with 3% horse serum to block nonspecific staining, and incubated for 60 min with anti-rabies nucleocapsid polyclonal antibody (Bio-rad, France) at a dilution of 1:80 or anti-cytochrome c monoclonal antibody (Santa Cruz Biotechnology, USA) at a dilution of 1: 4000. After two 3-min rinses in PBS, sections were incubated in DAKO EnVision™ System kit, HRP reagent (DAKO Corporation, CA, USA) as secondary antibody for 30 min. Slides were washed with PBS again and incubated for 10 min with peroxidase substrate [diaminobenzidine (DAB; Sigma, USA) 0.5 mg/ml and 30%H2O2 in Tris-HCl buffer with 1 M Imidazole]. After rinsed by tap water, the stain was couterstained with hematoxylin.
3. Detection of apoptosis
To evaluate whether cell death was due to apoptosis, we used the ApopTag® Plus Peroxidase In Situ Apoptosis Kit (Intergen Company, USA) as well as TUNEL assay (Terminal deoxynucleotidyl transferase (TdT)-mediated dUTP-digoxigenin nick end labelling assay) for detection. The kit detects apoptotic cells by peroxidase staining detection of the digoxigenin-labeled 3'-OH DNA ends generated by DNA fragmentation, and typically localized in morphologically identifiable nuclei and apoptotic bodies.
Sections were stained by ApopTag® Peroxidase kit after deparaffinized by xylene and ethanol and then permeabilized cell by proteinase K. Briefly, sections were incubated for 5 min with 3%H2O2 to quench endogenous peroxidase and apply equilibration buffer. Sections were then incubated in a humidified chamber at 37°C for 1 hour with working strength TdT enzyme (TdT enzyme mediated digoxigenin-dUTP), kept in a coupling jar containing working strength stop/wash buffer, and incubated for 10 min at room temperature. Slides were washed with PBS, incubated with anti-digoxigenin peroxidase conjugate in humidified chamber for 30 min at room temp and washed again. Developing color was done with peroxidase substrate as previously described. Finally, slides were washed in water and couterstained with hematoxylin.
Quantitation and controls
In order to clarify how cells were shown to be neuronal and not another cell types. The specimens were serially sectioned and stained as follows: hematoxylin and eosin (H&E), rabies antigen, cytochrome c, TUNEL and neurofilament protein (as a neuronal marker) (DAKO Corporation, CA, USA). Number or density of neurons in each slide was examined by H&E and was found to be correlated with neurofilament immunostaining technique.
The number of rabies antigen-positive cells and cells with MOMP and TUNEL positive cells in various areas was graded on a 0 – 4 scale from none to most abundant [4] by 3 readers (SJ, SS and TH) independently and where disagreement occurred the respective cases were re-examined and a consensus reached. Scale measurements of 0 – 4 were based on the followings: (0)-no antigen positive neuron in all fields; (1) 1 – 25 % antigen positive neuron(s) in the whole section; (2) 26 – 50% antigen positive neurons in the whole section; (3) 51 – 75% antigen positive neurons in the whole section; (4) 76 – 100 % antigen positive neurons in the whole section.
Brain sections from a patient with lung cancer with no CNS complications served as negative controls.
Results
Regional CNS distribution of rabies virus antigen and apoptosis
Rabies virus antigen
The overall regional distribution of rabies viral antigen was roughly similar in terms of number and location to that previous report [4]. Rabies antigen-containing neurons were found predominantly in the brain stem and spinal cord, dorsal and ventral horn neurons, thalamus and basal ganglia particularly in patients who had survival periods of 7 days or less regardless of clinical forms (Patient nos. 1, 5, 6 and 8 in furious and 4 in paralytic group) (Table 2) (Figure 1). Those who died later than 7 days (Patient nos.3 in furious and 2, 7, 9, and 10 in paralytic group) had rabies viral antigen disseminated throughout the whole neuraxis (Table 3) (Figure 2).
Figure 1 Rabies antigen positive neurons in brainstem at midbrain region of a paralytic rabies patient (no. H4 – see text and Table 1).
Figure 2 Rabies antigen positive neurons in spinal cord at thoracic region of a furious rabies patient (no. H3 – see text and Table 1).
Table 2 Distribution of rabies virus and cytochrome c in CNS of human rabies patients who survived 7 days or less. (Numbers in bold and italic designated discrepancy between rabies antigen positive- and cytochrome c positive neurons in particular region).
Patient No. Survival period Antigen Frontal Temporal Hippocampus Parietal Occipital Thalamus Basal-ganglia Cerebellum Midbrain Pons Medulla Cervical Thoracic Lumbar Sacrum
Furious
H1 5 Rabies 0 0 0 0 0 3 2 1 3 3 3 3 4 4 nd**
Cyto C* 0 0 1 0 0 3 1 4 0 0 0 0 1 0 nd
H5 4 Rabies 2 2 2 3 2 3 3 3 4 4 4 3 3 4 4
Cyto C 2 1 4 3 4 4 4 2 4 4 4 0 1 2 2
H6 5 Rabies 2 0 1 3 0 2 3 2 3 3 3 3 3 3 3
Cyto C 1 3 2 4 4 4 4 3 4 4 4 2 0 0 0
H8 5 Rabies 1 3 4 3 2 4 4 2 3 4 4 nd nd nd nd
Cyto C 1 4 4 4 4 4 3 2 3 4 4 nd nd nd nd
Paralysis
H4 7 Rabies 2 1 2 2 2 3 2 2 4 4 3 3 3 3 nd
Cyto C 2 0 2 2 3 0 3 2 4 4 3 1 1 nd nd
* = cytochrome c ** = not done (sample not available)
Table 3 Distribution of rabies virus and cytochrome c in CNS of human rabies patients who survived longer than 7 days (Numbers in bold and italic designated discrepancy between rabies antigen positive- and cytochrome c positive neurons in particular region).
Patient No. Survival period Antigen Frontal Temporal Hippocampus Parietal Occipital Thalamus Basal-ganglia Cerebellum Midbrain Pons Medulla Cervical Thoracic Lumbar Sacrum
Furious
H3 8 Rabies 2 3 3 4 3 3 4 2 4 4 4 4 4 4 4
Cyto C* 0 3 4 0 2 4 4 1 4 4 4 2 1 2 3
Paralysis
H2 16 Rabies 4 4 4 4 4 4 4 4 4 4 4 4 4 4 nd**
Cyto C 0 0 0 0 1 3 3 1 1 1 0 0 0 0 nd
H7 13 Rabies 4 3 2 3 3 4 nd 3 4 4 4 4 3 4 4
Cyto C 2 3 nd 4 3 3 nd 1 4 2 4 0 0 nd 1
H9 9 Rabies 4 2 nd 4 4 nd 4 4 4 nd 3 nd nd nd nd
Cyto C 1 0 nd 2 0 nd 0 3 1 nd 1 nd nd nd nd
H10 13 Rabies 4 4 4 3 3 4 4 4 4 4 4 4 4 4 4
Cyto C 3 3 1 2 2 4 4 4 1 4 4 4 3 1 0
*= cytochrome c ** = not done (sample not available)
MOMP (cytochrome c assay)
Evidence of MOMP was detected by demonstration of cytochrome c antigen in cytoplasm (Figure 3). In furious group, there was a discrepant result between the degree of rabies positive and cytochrome c positive neurons in spinal cord of patients with a survival period of 7 days or less (Patients no. 1, 5, 6) (Table 2). This was also noted in paralytic patient no. 4 who survived 7 days. Furious patient no. 8 did not have spinal cord specimens available. Of these 5, one of them (Patient no. 1) had brainstem (midbrain, pons, and medulla) negative for MOMP.
Figure 3 Cytochrome c positive neurons in brainstem at midbrain region of a paralytic rabies patient (no. H4 – see text and Table 1).
Among 4 rabies patients who survived longer than 7 days with spinal cord specimens available, paralytic patient nos. 2 and 7 had relative absence of cytochrome c positive neurons as compared to rabies in all corresponding spinal cord regions (Table 3). Furious and paralytic patient nos. 3 and 10 had less degree of cytochrome c positive neurons in 3 out of 4 regions of spinal cord (Figure 4). Such discrepancy was noted in 2 or 3 regions of brainstem in 2 of 5 patients (paralytic patient nos. 9 and 2 respectively). Less degree of cytochrome c positive neurons was also evident in 1 of 3 regions of brainstem in paralytic patient nos. 7 and 10 (pons in one and midbrain in another).
Figure 4 Cytochrome c positive neurons in spinal cord at thoracic region of a furious rabies patient (no. H3 – see text and Table 1).
TUNEL assay
Apoptotic cells as demonstrated by TUNEL assay were found throughout the whole neuraxis in all patients (Figure 5). There was no significant correlation between short or long survival period, amount of rabies antigen positive neurons and degree of apoptosis in various CNS regions (data not shown).
Figure 5 TUNEL staining. Apoptotic cells as demonstrated by TUNEL assay were found throughout the whole neuraxis in all patients.
Discussion
Our study showed that that some of neuronal cells especially in spinal cord and brain stem regions had a delay in apoptotic process especially that mediated via cytochrome c of the mitochondrial pathway. Although TUNEL assays did not reveal any differences among neurons at various regions, this was not surprising since varieties of unavoidable factors, such as hypoxia and ischemia, also contributed to apoptosis.
A relative absence of MOMP was found in spinal cord of rabies patients (8 of 8) regardless of clinical forms and survival period despite the presence of abundant amount of rabies virus antigen. Such phenomenon was also evident in one or more regions of brainstem in 5 of 10 patients (patient nos. 1, 2, 7, 9 and 10). Biting site did not correlate with either clinical forms or the abundance or absence of MOMP and rabies virus antigen.
Both in vitro and animal model observations in rabies agree to a similar conclusion that necrotic process is usually lacking with apoptosis becomes dominant findings [13,14]. The degree of apoptosis correlated with amount of expression of rabies G protein in infected neurons [15-17]. Nonfatal or abortive infection and the process of viral clearance is mediated by local recruitment of T cells, as well as the development of apoptosis of infecting neurons and surrounding cells [17]. Downregulation of G protein expression in neuronal cells contributes to pathogenesis by preventing apoptosis [18]. Although apoptosis may be a protective rather than a pathogenetic mechanism because less pathogenic viruses induced more apoptosis than more pathogenic viruses [16,19,20], all of our patients were bitten by street rabies virus variant transmitted by rabid dogs. Our previous study did not reveal any evidence of specific variants in association with the development of furious or paralytic presentation [3].
It remains intriguing why neurons, particularly those in spinal cord and brainstem are resistant to the effect of rabies infection. Previous in vitro model suggest this refers to an inherent property of spinal motoneurons themselves [21]. This may also be true in brainstem neurons. Furthermore, we found that amount of rabies virus in the brain should not be the sole contributing factor in determining the functional degree of brain functional alterations. Biopsy specimen of temporal lobe from a paralytic rabies patient who remained alert and rational showed large amount of rabies virus antigen on direct fluorescent test [22]. Magnetic resonance imaging showed abnormalities in the brain of a furious rabies patient who at that time did not exhibit any brain symptoms and signs [8]. He only had a local neuropathic pain at bitten left arm. Numerous studies point to the alterations at the levels of neurotransmitters, cytokines, ion channels, cellular RNA and protein synthesis and brain electroencephalographic patterns as well as role of neurotoxicity [7,22-38].
Conclusion
In rabies virus infection, mechanisms involved in cell death or survival of neurons are complex [13,39,40]. Preservation of the neuronal network by inhibition of apoptosis and limitation of the inflammation and the destruction of T cells that invade the CNS is crucial for neuroinvasion [41]. This in addition to uncharacterized properties of certain neuronal cells may explain why spinal cord and brainstem where rabies virus was found heavily and early in the disease course, yet still retain their functions. We hope that by knowing what are unique among these types of neurons in term of response to rabies virus infection may give us ideas how to preserve neuronal functions and postpone death until native immunity (or any novel therapeutics) may arise for rabies virus clearance [38].
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
SJ carried out laboratory work and histopathological examination, participated in data analysis and involved in drafting the manuscript. PR developed and optimized laboratory protocol and condition and involved in drafting the manuscript. SS participated in histopathological examination and data analysis and interpretation and involved in drafting the manuscript. SW participated in data analysis and interpretation and in drafting the manuscript. TH designed the study and coordination and involved in histopathological examination and data analysis and interpretation and writing the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported in part by grant from National Science and Technology Development Agency
==== Refs
Hemachudha T Laothamatas J Rupprecht CE Human rabies: a disease of complex neuropathogenetic mechanisms and diagnostic challenges Lancet Neurol 2002 1 101 109 12849514 10.1016/S1474-4422(02)00041-8
Hemachudha T Phuapradit P Rabies Curr Opin Neurol 1997 10 260 267 9229136
Hemachudha T Wacharapluesadee S Lumlertdaecha B Orciari LA Rupprecht CE La-Ongpant M Juntrakul S Denduangboripant J Sequence analysis of rabies virus in humans exhibiting encephalitic or paralytic rabies J Infect Dis 2003 188 960 966 14513414 10.1086/378415
Tirawatnpong S Hemachudha T Manutsathit S Shuangshoti S Phanthumchinda K Phanuphak P Regional distribution of rabies viral antigen in central nervous system of human encephalitic and paralytic rabies J Neurol Sci 1989 92 91 99 2769305 10.1016/0022-510X(89)90178-0
Bingham J van der Merwe M Distribution of rabies antigen in infected brain material: determining the reliability of different regions of the brain for the rabies fluorescent antibody test J Virol Methods 2002 101 85 94 11849687 10.1016/S0166-0934(01)00423-2
Mitrabhakdi E Shuangshoti S Wannakrairot P Lewis RA Susuki K Laothamatas J Hemachudha T Difference in Neuropathogenetic Mechanisms in Human Furious and Paralytic Rabies (in press). J Neurol Sci 2005 16226769
Hemachudha T Wacharapluesadee S Mitrabhakdi E Wilde H Morimoto K Lewis RA Pathophysiology of human paralytic rabies J Neurovirol 2005 11 93 100 15804967 10.1080/13550280590900409
Laothamatas J Hemachudha T Mitrabhakdi E Wannakrairot P Tulayadaechanont S MR imaging in human rabies AJNR Am J Neuroradiol 2003 24 1102 1109 12812933
Sheikh KA Ramos-Alvarez M Jackson AC Li CY Asbury AK Griffin JW Overlap of pathology in paralytic rabies and axonal Guillain-Barre syndrome Ann Neurol 2005 57 768 772 15852372 10.1002/ana.20482
Chopra JSBAKMJMKPSR Paralytic rabies: A clinicopathological study Brain 1980 103 789 802 7437890
Guigoni C Coulon P Rabies virus is not cytolytic for rat spinal motoneurons in vitro J Neurovirol 2002 8 306 317 12161815 10.1080/13550280290100761
Green DR Kroemer G The pathophysiology of mitochondrial cell death Science 2004 305 626 629 15286356 10.1126/science.1099320
Fu ZF Jackson AC Neuronal dysfunction and death in rabies virus infection J Neurovirol 2005 11 101 106 15804968 10.1080/13550280590900445
Ubol S Kasisith J Pitidhammabhorn D Tepsumethanol V Screening of pro-apoptotic genes upregulated in an experimental street rabies virus-infected neonatal mouse brain Microbiol Immunol 2005 49 423 431 15905604
Faber M Pulmanausahakul R Hodawadekar SS Spitsin S McGettigan JP Schnell MJ Dietzschold B Overexpression of the rabies virus glycoprotein results in enhancement of apoptosis and antiviral immune response J Virol 2002 76 3374 3381 11884563 10.1128/JVI.76.7.3374-3381.2002
Yan X Prosniak M Curtis MT Weiss ML Faber M Dietzschold B Fu ZF Silver-haired bat rabies virus variant does not induce apoptosis in the brain of experimentally infected mice J Neurovirol 2001 7 518 527 11704884 10.1080/135502801753248105
Galelli A Baloul L Lafon M Abortive rabies virus central nervous infection is controlled by T lymphocyte local recruitment and induction of apoptosis J Neurovirol 2000 6 359 372 11031689
Dietzschold B Morimoto K Hooper DC Mechanisms of virus-induced neuronal damage and the clearance of viruses from the CNS Curr Top Microbiol Immunol 2001 253 145 155 11417132
Morimoto K Hooper DC Spitsin S Koprowski H Dietzschold B Pathogenicity of different rabies virus variants inversely correlates with apoptosis and rabies virus glycoprotein expression in infected primary neuron cultures J Virol 1999 73 510 518 9847357
Prehaud C Lay S Dietzschold B Lafon M Glycoprotein of nonpathogenic rabies viruses is a key determinant of human cell apoptosis J Virol 2003 77 10537 10547 12970438 10.1128/JVI.77.19.10537-10547.2003
Hemachudha T Phanuphak P Sriwanthana B Manutsathit S Phanthumchinda K Siriprasomsup W Ukachoke C Rasameechan S Kaoroptham S Immunologic study of human encephalitic and paralytic rabies. Preliminary report of 16 patients Am J Med 1988 84 673 677 2456691 10.1016/0002-9343(88)90103-9
Koprowski H Zheng YM Heber-Katz E Fraser N Rorke L Fu ZF Hanlon C Dietzschold B In vivo expression of inducible nitric oxide synthase in experimentally induced neurologic diseases Proc Natl Acad Sci U S A 1993 90 3024 3027 7681993
Dumrongphol H Srikiatkhachorn A Hemachudha T Kotchabhakdi N Govitrapong P Alteration of muscarinic acetylcholine receptors in rabies viral- infected dog brains J Neurol Sci 1996 137 1 6 9120481 10.1016/0022-510X(95)00275-7
Gourmelon P Briet D Court L Tsiang H Electrophysiological and sleep alterations in experimental mouse rabies Brain Res 1986 398 128 140 3801886 10.1016/0006-8993(86)91258-8
Gourmelon P Briet D Clarencon D Court L Tsiang H Sleep alterations in experimental street rabies virus infection occur in the absence of major EEG abnormalities Brain Res 1991 554 159 165 1933298 10.1016/0006-8993(91)90184-W
Iwata M Unno T Minamoto N Ohashi H Komori S Rabies virus infection prevents the modulation by alpha(2)- adrenoceptors, but not muscarinic receptors, of Ca(2+) channels in NG108-15 cells Eur J Pharmacol 2000 404 79 88 10980265 10.1016/S0014-2999(00)00621-X
Iwata M Komori S Unno T Minamoto N Ohashi H Modification of membrane currents in mouse neuroblastoma cells following infection with rabies virus Br J Pharmacol 1999 126 1691 1698 10372810 10.1038/sj.bjp.0702473
Fu ZF Weihe E Zheng YM Schafer MK Sheng H Corisdeo S Rauscher FJ Koprowski H Dietzschold B Differential effects of rabies and borna disease viruses on immediate- early- and late-response gene expression in brain tissues J Virol 1993 67 6674 6681 8411369
Ceccaldi PEFMPEATHFG Rabies virus selectively alters 5-HT1 receptor subtypes in rat brain Eur J Pharmacol 1993 245 129 138 8491253 10.1016/0922-4106(93)90120-X
Akaike T Weihe E Schaefer M Fu ZF Zheng YM Vogel W Schmidt H Koprowski H Dietzschold B Effect of neurotropic virus infection on neuronal and inducible nitric oxide synthase activity in rat brain J Neurovirol 1995 1 118 125 9222348
Prosniak M Zborek A Scott GS Roy A Phares TW Koprowski H Hooper DC Differential expression of growth factors at the cellular level in virus-infected brain Proc Natl Acad Sci U S A 2003 100 6765 6770 12736376 10.1073/pnas.0430999100
Prosniak M Hooper DC Dietzschold B Koprowski H Effect of rabies virus infection on gene expression in mouse brain Proc Natl Acad Sci U S A 2001 98 2758 2763 11226313 10.1073/pnas.051630298
Marquette C Van Dam AM Ceccaldi PE Weber P Haour F Tsiang H Induction of immunoreactive interleukin-1 beta and tumor necrosis factor-alpha in the brains of rabies virus infected rats J Neuroimmunol 1996 68 45 51 8784259 10.1016/0165-5728(96)00056-2
Marquette C Ceccaldi PE Ban E Weber P Tsiang H Haour F Alteration of interleukin-1 alpha production and interleukin-1 alpha binding sites in mouse brain during rabies infection Arch Virol 1996 141 573 585 8645096 10.1007/BF01718318
Hemachudha T Panpanich T Phanuphak P Manatsathit S Wilde H Immune activation in human rabies Trans R Soc Trop Med Hyg 1993 87 106 108 8465378 10.1016/0035-9203(93)90446-W
Van Dam AM Bauer J Man AHWK Marquette C Tilders FJ Berkenbosch F Appearance of inducible nitric oxide synthase in the rat central nervous system after rabies virus infection and during experimental allergic encephalomyelitis but not after peripheral administration of endotoxin J Neurosci Res 1995 40 251 260 7745618 10.1002/jnr.490400214
Willoughby REJ Tieves KS Hoffman GM Ghanayem NS Amlie-Lefond CM Schwabe MJ Chusid MJ Rupprecht CE Survival after treatment of rabies with induction of coma N Engl J Med 2005 352 2508 2514 15958806 10.1056/NEJMoa050382
Hemachudha T Human rabies: clinical aspects, pathogenesis, and potential therapy Curr Top Microbiol Immunol 1994 187 121 143 7859488
Dietzschold B Schnell M Koprowski H Pathogenesis of rabies Curr Top Microbiol Immunol 2005 292 45 56 15981467
Baloul L Lafon M Apoptosis and rabies virus neuroinvasion Biochimie 2003 85 777 788 14585545 10.1016/S0300-9084(03)00137-8
|
16288653
|
PMC1310615
|
CC BY
|
2021-01-04 16:28:17
|
no
|
BMC Infect Dis. 2005 Nov 16; 5:104
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-104
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1491629318910.1186/1471-2407-5-149Research ArticleAbnormal expression and processing of the proprotein convertases PC1 and PC2 in human colorectal liver metastases Tzimas George N [email protected] Eric [email protected] Sarah [email protected]ên Duc Thang [email protected] Abdel M [email protected] Victoria [email protected] Yi [email protected]étien Michel [email protected] Nabil [email protected] Peter [email protected] Transplant and Hepato-pancreatobiliary Research Group, McGill University Health Center, McGill University, 687 Pine Avenue West, S.10.26, Montreal, Canada H3A1A12 Department of Surgical Services, Kypselis General Hospital, 24 Drossopoulou Street, Athens 11257, Greece3 Organelle Signaling Laboratory, Department of Surgery, McGill University Health Center, McGill University, 687 Pine Avenue West, S.10.26, Montreal, Canada H3A1A14 Ottawa Health Research Institute, University of Ottawa, Ottawa, Canada Y1K 4K95 Department of Pathology, McGill University Health Center, McGill University, 1650 Cedar Avenue, Montreal, Canada H3G 1A46 Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, University of Montreal, Montreal, Canada H2W 1R7, QC2005 17 11 2005 5 149 149 11 3 2005 17 11 2005 Copyright © 2005 Tzimas et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The family of proprotein convertases has been recently implicated in tumorigenesis and metastasis in animal models. However, these studies have not yet been completely corroborated in human tumors.
Methods
Using RT PCR, immunoblot and immunohistochemistry we assessed the presence and the processing patterns of the convertases PC1 and PC2 as well as the PC2 specific chaperone 7B2 in human liver metastases originating from colorectal cancer and compared them to unaffected and normal liver. Furthermore, we assessed the presence and processing profiles of PC1, PC2 and 7B2 in primary colon cancers.
Results
mRNA, protein expression, and protein cleavage profiles of proprotein convertases 1 and 2 are altered in liver colorectal metastasis, compared to unaffected and normal liver. Active PC1 protein is overexpressed in tumor, correlating with its mRNA profile. Moreover, the enhanced PC2 processing pattern in tumor correlates with the overexpression of its specific binding protein 7B2. These results were corroborated by immunohistochemistry. The specific and uniform convertase pattern observed in the metastases was present only in a fraction of primary colon cancers.
Conclusion
The uniformly altered proprotein convertase profile in liver metastases is observed only in a fraction of primary colon cancers, suggesting possible selection processes involving PCs during metastasis as well as an active role of PCs in liver metastasis. In addition, the exclusive presence of 7B2 in metastatic tumors may represent a new target for early diagnosis, prognosis and/or treatment.
==== Body
Background
Liver metastasis remains the major cause of therapeutic failure in colorectal carcinoma. Amongst the 300,000 people that are given the diagnosis of colorectal cancer (CRC) every year in Europe and North America, 60% will eventually develop liver metastases, and several of them will succumb from their disease. Metastasis microenvironment conditioned by the tumor itself, may represent an alternative to the "seed-soil" theory. Indeed recent reports have demonstrated the pivotal role of the proprotein convertases (PCs) (1) in tumor growth and metastasis of HT-29 human colon carcinoma cells (2), lung (3) or breast cancer cells (4). To date eight members of this family have been identified, namely furin, PC1/PC3, PC2, PC4, PACE4, PC5/PC6, PC7/LPC/PC8 and SKI-1/S1P [1-7]. Amongst the PCs, PC1 and PC2 are unique due to their processing activity specific to the regulated secretory pathway, and their localization in dense core secretory granules. Both proteins are synthesized in the Endoplasmic Reticulum (ER) as zymogens (primary cleavage of the zymogen of proPC1 occurs in the ER while that of proPC2 occurs in the granules), transported through the Golgi apparatus, and the Trans Golgi Network (TGN) into mature secretory granules, and both are proteolytically activated along the way. The proper trafficking and processing of proPC2 into its active form is dependent on 7B2, a chaperone-like protein [8]. This bifunctional polypeptide has been shown to promote proPC2 cleavage to the active PC2 form with its N-terminal domain, but also to inhibit further PC2 processing with its C-terminal domain. The exact biological significance of this protein has not been elucidated yet, although its deletion in mice leads to a Cushing's-like syndrome [9]. Finally, recently a specific PC1 inhibitor, ProSAAS has been identified [10], but its physiological function remains unclear.
The purpose of this study is to examine the expression and processing patterns of PC1 and PC2 in human CRC liver metastases and to compare these profiles with those observed in unaffected and normal liver. Furthermore, we assessed the processing patterns of these convertases in primary colon cancers. Finally, we attempt to shed light into the PC1 and PC2 processing mechanisms and their potential impact on CRC liver metastasis.
Methods
Human liver specimens
Tumor specimens and their unaffected counterparts (Fig. 7D) were collected from 14 patients (proper informed consent obtained) that underwent liver resection for CRC metastases. Normal liver specimens were obtained from organ donors. All the specimens were harvested immediately after resection and were snap frozen in liquid nitrogen.
Human colon cancer specimens
In a similar fashion, colon cancer specimens with unaffected colon were collected from 5 patients (proper informed consent obtained) that underwent colon resection for primary colon cancer (pathology proven adenocarcinoma). The specimens were harvested immediately after resection and were snap frozen in liquid nitrogen.
PCR primers
PC1 primers were designed as follows f5'-TGGCTTGCTAAATGCCAAAGCTC-3'/r5'-ATCCACCATCTTCTCCACCCC-3' (annealing temperature 60°C), PC2 primers were f5'-GTCCTTGATGCAGGTGCCATC-3'/r5'-ACTCCTTCAGCACCCCCTTC-3' (annealing temperature 62°C) and 7B2 primers were f5'-CAC CAG GCC ATG AAT CTT-3'/r5'-CTG GAT CCT TAT CCT CAT CTG-3' (annealing temperature 61°C. GAPDH was used as standard and for normalization (f5'-CCC TTC ATT GAC CTC AAC TAC ATG GT-3'/r5'-GAG GGG CCA TCC ACA GTC TTC TG-3') (annealing temperature 58°C).
Antibodies
Rabbit antisera raised either against PC1 (amino acids 621–726, Figure 1B), PC2 (amino acids 529–637, Figure 2B) or 7B2 (amino acids 32–39, Figure 3B) were used.
RNAs and proteins extraction
RNA was extracted using Trizol (Invitrogen) according to the manufacturer's instructions. Semi-quantitative RT-PCR analysis was performed as previously described [11]. For protein extraction, approximately 0.5 mg of sample tissues were homogenized as previously described [12]. TX-100 was added to a 1% final concentration and tissue extract solubilized for 30' at 4°C. Lysates were then centrifuged at 10000 g for 2 × 30' at 4°C. Protein concentrations in the supernatants were determined using the Bradford method. For every immunoblot analysis 100 μg of protein were used.
Immunohistochemistry
4 μm-thick, formalin-fixed, paraffin-embedded tissue sections, were treated as previously described [13]. The slides were incubated with anti-PC1, PC2 and 7B2 antibodies at a 1:2000 dilution, revealed with biotinylated goat-anti-rabbit IgG-B (1:200), combined with Histocain-plus kit, followed by 3-amino-9-ethylcarbazole (AEC) chromogen and counterstaining with hematoxylin.
Statistical analysis
Data are expressed as means ± SD. Comparisons between the specimens were performed using student's t-test. A value of p < 0.005 was considered statistically significant.
Results
PC1 and PC2 mRNA are expressed in both normal and unaffected liver but their expression is differentially regulated in tumor
We initially assessed the presence of the mRNA encoding PC1 and PC2 in liver CRC metastatic tumors versus unaffected and normal livers. PC1 and PC2 RT-PCR amplification products were of the expected size, 553 bp and 422 bp respectively. PC1 mRNA expression was two-fold higher in areas of tumor compared to areas of unaffected and normal liver (P < 0.04, Figure 1A). In contrast, PC2 mRNA was overexpressed two-fold in areas of unaffected and normal liver compared to areas of metastasis (P < 0.003, Figure 2A).
PC1 and PC2 protein expression and maturation are altered in tumor
To correlate mRNA and protein expression, we examined total sample lysates by immunoblotting. Relative amounts of protein were quantified by scanning densitometry of the immunoblots. Interestingly, the anti-PC1 antibody (Fig. 1B) revealed two immunoreactive species of 84 and 66 kDa, respectively (Fig. 1C), in tumor (T), unaffected (U), and normal (N) liver. The higher molecular mass isoform (84 kDa) corresponded to the full-length active PC1 whereas the C-terminally immunoreactive 66 kDa isoform was likely to be an N-terminally truncated, inactive form of PC1 (Fig. 1B) (14,15). In tumors the total amount of PC1 (p84+p66) was ~2.5-fold more elevated than in unaffected and normal samples (Fig. 1D, left panel). Moreover the ratio of the 84 kDa active form over the 66 kDa form was also ~2.5 times higher than in both unaffected and normal liver (Fig. 1D, right panel). Immunoblot analysis with anti-PC2 antibody (Fig. 2B) revealed two species of 75 and 66 kDa respectively, representing the inactive proPC2 isoform and the cleaved, active PC2 (Fig. 2C) [14]. In contrast to PC1, the amount of total PC2 (p75+p66) was higher in unaffected and normal liver than in metastasis (Fig. 2D, left panel). Nevertheless a similar pattern of processing was observed, i.e. in metastasis the fully active 66 kDa form predominates over the inactive 75 kDa pro-form, while in unaffected and normal liver the proPC2 isoform was more abundant. Again the PC2 processing ratio of active PC2 (p66) over the inactive proPC2 (p75) in tumor versus unaffected and normal liver indicated a ~10-fold increase in PC2 maturation in liver metastasis (Fig. 2D, right panel). In conclusion, alteration of PC1 and PC2 processing, lead to increased accumulation of active isoforms of both convertases in liver metastasis.
The PC2 chaperone, 7B2, is expressed only in liver CRC metastases
Since PC2 processing is controlled by its specific binding protein 7B2, we quantified 7B2 expression profiles in normal and unaffected livers compared to liver metastasis. RT-PCR amplification of 7B2 led to an expected 454 bp product and showed a ~3-fold (P < 0.003) increase of 7B2 mRNA in metastasis compared to unaffected or normal liver (Fig. 3A). By immunoblotting, we were not able to detect any 7B2 in both unaffected and normal liver, while an intense immunoreactive band was observed at the expected 7B2 molecular mass of 21 kDa [8]) in metastasis samples (Fig. 3C).
Immunohistochemical validation
Using the antibodies described above, we characterized PC1, PC2 and 7B2 localization by immunohistochemistry. PC1 staining was cytoplasmic and more abundant in areas of tumor than in adjacent unaffected liver, where only a few cells were positively stained (Fig. 4B, 4C), thus correlating with the mRNA and immunoblot data. PC2 was present in areas of tumor, and adjacent unaffected parenchyma (Fig. 5B, 5C). In addition, 7B2 was detected by very strong cytoplasmic staining in areas of liver metastasis while adjacent unaffected liver did not stain (Fig. 6B, 6C). Finally, although normal liver stained for PC1 (Fig. 4A) and PC2 (Fig. 5A), 7B2 remained undetectable (Fig. 6A).
The expression and processing pattern of PC1 and PC2 observed in metastases is only found in a subset of primary colon cancers
Using the antibodies described above, we assessed the presence as well as the processing profile of PC1 and PC2 in primary colon cancers using immunoblot. PC1 was detected in various amounts in primary colon cancers but was undetectable in their unaffected counterparts (Fig. 7A). Immunoblot analysis revealed two immunoreactive species of 84 and 66 kDa, respectively. The higher molecular mass isoform (84 kDa) corresponded to the full-length active PC1 whereas the C-terminally immunoreactive 66 kDa isoform was likely to be an N-terminally truncated, inactive form of PC1. The ratio of the two isoforms varied among specimens with 3 out of 5 specimens expressing more abundant the 84 kDa isoform. PC2 was also detected in various amounts in primary colon cancers and was undetectable in their unaffected counterparts (Fig. 7B). Furthermore, the active 66 kD isoform was more abundant in 2 out of 5 specimens. In these specific specimens (2 out of five), immunoblot analysis revealed an intense immunoreactive band at the expected 7B2 molecular mass of 21 kDa (Fig. 7C). These results demonstrate a strong heterogeneity in the expression/processing of PC1, PC2 and 7B2 in primary colon cancers compared to liver metastases, As a consequence they suggest that the expression/processing pattern of these convertases observed in liver metastases may represent an important feature of these cancers.
Discussion
In this study, we assessed the presence and the cleavage/processing patterns of the two major convertases of the regulated secretory pathway, PC1 and PC2, as well as the PC2 chaperone 7B2. This study was performed in human liver metastases specimens from colorectal primaries and in unaffected liver samples from the same patients that underwent liver resection, as well as in normal livers. We also assessed these convertase profiles in primary colon cancers.
To our knowledge this is the first report describing the expression of PC1, PC2 and 7B2 in human liver tissues, and in human colon cancers. More importantly, we noted that (i) at the mRNA level, PC1 is overexpressed in metastatic tumor versus unaffected and normal liver, while PC2 expression is downregulated (Fig. 1A and 2A respectively); (ii) consistently, at the protein level, PC1 (p84+p66) is overexpressed (Fig. 1D), while PC2 (p75+p66) is downregulated in metastatic tumor (Fig. 2D); (iii) both active PC1 and PC2 are predominant in metastatic tumor (Fig. 1D and 2D respectively); (iv) consistently with the enhanced PC2 zymogen processing pattern, in tumor, 7B2 is overexpressed (Fig. 3A and 3C); (v) the above results are corroborated by immunohistochemistry (Fig. 4, 5, 6).
We also found that the specific PC2 and 7B2 profile observed in metastatic cancers was observed only in a fraction of primary colon cancers. These data support a specific negative feedback mechanism regulating PC2 mRNA expression in liver metastases, when PC2 proteolytic activity is overwhelming. Therefore in metastatic tumors, abundant active PC2 may lead to a downregulation of its mRNA, and vice versa in unaffected and normal liver. Furthermore, the homogeneous liver metastasis PC2 protein profile supports the hypothesis of a tightly regulated active PC2 production in the primary as well as in metastatic tumors, with 7B2 playing a leading role in this mechanism. Interestingly, PC1, PC2 and 7B2 are considered as markers of endocrine and neuroendocrine phenotypes [5,8,12]. The fact that they are also detected in human anal canal [13] from which also cancers and eventually metastases can develop suggests that a neuroendocrine differentiation program could take place during colon carcinogenesis and liver metastasis. Recently, a specific RE1-lk silencer element in the promoter of PC2 was identified [17] and binding of transcription-silencing factors to this element may contribute to repression of the PC2 gene in non-neuroendocrine cells.
Our results are in agreement with recent experimental animal evidence highlighting the potential implication of convertases in tumorigenesis and metastasis. Indeed, Khatib et al. have shown that convertase inhibition in the HT-29 colon cancer cell line is followed by decreased invasiveness and tumorigenicity [14]. In addition it has been shown that furin, another member of the convertase family, is implicated in tumor progression in human head and neck malignancies [15]. Convertase overexpression can also alter the growth behavior and the drug responsiveness in a human breast cancer cell line model [16]. In the past, 7B2 has been implicated with several types of neuroendocrine tumors, such as neuroendocrine bronchial tumors, nonfunctioning pancreatic islet tumors and ACTH-secreting pituitary tumors [17,18], mainly participating at the processing of tumor-secreted active peptides. However its role in tumorigenesis and metastasis, especially in colorectal cancer, remains largely unknown.
Other groups have also recently showed the implication of several members of the convertase family in the carcinogenesis process. Specifically Siegfried et al have clearly shown that members of the convertase family process VEGF-C, a known tumorigenic growth factor, in animal models [19]. The same group has demonstrated that convertases are involved in the processing of pro-platelet derived growth factor A, a hallmark of carcinogenesis [20], thus concluding that convertase inhibitors might be used eventually in the treatment of neoplasia.
Our data support the accumulation of active PC1 and PC2 in metastasis, and are in agreement with previous reports [3,16]. The main question although remains whether alterations of PC1, PC2 and 7B2 expression profiles are the cause or consequence of the metastatic phenotype. PC1 and PC2 process pro-neurotensin to its active form, which has been involved in colonic tumorigenesis [21,22], or pro-pancreatic peptide, proGHRH, proglucagon, prosomatostatin, and pro-insulin to insulin [23,24], which are known trophic factors for the gut and possibly involved in tumorigenesis as well. Indeed, recently it was shown that PC1 null mice are dwarfed, thus implicating PC1 in the processing of GHRH and subsequent growth [23]. Therefore change in PC1 and PC2 expression and activation may alter the profiles of secretory proteins, which in turn could increase cell growth potential. These changes could lead to selection of primary colon cancer cells destined to metastasize to the liver or even render the rest of the liver susceptible to future/further metastasis. Indeed, the fact that the PC2 profile observed in unaffected liver is partially different from the one observed in normal liver, supports the hypothesis that the tumor may induce modifications in the rest of the liver. Whether PC1, PC2 and 7B2 are directly implicated in such a model is under investigation, and it would be interesting also to evaluate the profile of these convertases in different also metastatic sites such as in the lung or in the brain.
In conclusion, the present study shows that PC1 and PC2 convertase expression and cleavage are altered in CRC liver metastases. 7B2, whose overexpression in tumor is thought to play a key role in the above processes, could represent a potential diagnostic, prognostic or even therapeutic target.
Furthermore, the metastasis/tumor associated convertase profile is observed only in a fraction of primary colon cancers thus suggesting a potential selection process for tumors that eventually will develop metastasis and may be associated with worse clinical outcome. These observations require prospective validation that is underway.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
George N Tzimas: study conception, performed the majority of PCR and immunoblot, wrote the manuscript
Eric Chevet: performed immunoblots and critically revised the manuscript
Sarah Jenna: performed immunoblots and PCR, revised the manuscript
Duc Thang Nguyên: performed part of PCRs
Abdel Majid Khatib: performed immunoblots
Victoria Marcus: evaluated immunohistochemistry
Yi Zhang: performed immunohistochemistry
Michel Chrétien: critically revised the manuscript
Nabil Seidah: provided antibodies and primers, revised the manuscript
Peter. Metrakos: provided specimens and reviewed the manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This study was performed under the guidelines of McGill University Health Center. The technical support by Caroline Rochon and Shuquing Liu was very much appreciated. The secretarial support of Mrs. Monaghan cannot be overemphasized. George N. Tzimas is the recipient of the Lois and Byron Dolgin Liver Transplant Fellowship, and the Michael Cohen Liver Transplant Fellowship. Eric Chevet is a Junior scholar from the Fonds de la Recherche en Santé du Québec.
Figures and Tables
Figure 1 Expression of PC1 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Quantification of PC1 mRNA expression respectively, in 3 normal (N), 14 tumor (T) and 14 unaffected (U) liver samples. Quantification was obtained by the ratio of the optical density of PC1 PCR amplification products over GAPDH. Standard error of mean (SEM) is indicated. Asterisk indicates a statistically significant difference (P < 0.04). B. Schematic representation of PC1 structure. The antigenic region against which the antibody was raised is mentioned (YAb). Cleavage sites are indicated (arrows). C. PC1 cleavage profile in normal (N1-N3, top panel), unaffected (U1-U2) and tumor (T1-T2, bottom panel) samples. Normal liver samples (N1-N3) derived from three different organ donors. T1/U1 and T2/U2 came from two independent patients. PC1 immunoblotting was done on a total of 3 normal, 14 metastasis and 14 unaffected liver samples, and was repeated three times. D. Relative amounts of total PC1 (C, left panel) protein, in tumor (T), unaffected (U) and normal liver (N), expressed as a ratio over normal liver samples. Ratios of p84/p66 PC1 isoforms (C, right panel) in the same samples are calculated. SEM is indicated. Simple and double asterisks indicate statistically significant differences (P < 0.05).
Figure 2 Expression of PC2 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Quantification of PC2 mRNA expression respectively, in 3 normal (N), 14 tumor (T) and 14 unaffected (U) liver samples. Quantification was obtained by the ratio of the optical density of PC2 PCR amplification products over GAPDH. Standard error of mean (SEM) is indicated. Asterisk indicates a statistically significant difference (P < 0.003). B. Schematic representation of PC2 structure. The antigenic region against which the antibody was raised is mentioned (YAb). Cleavage sites are indicated (arrows). C. PC2 cleavage profile in normal (N1-N3, top panel), unaffected (U1-U2) and tumor (T1-T2, bottom panel) samples. Normal liver samples (N1-N3) derived from three different organ donors. T1/U1 and T2/U2 came from two independent patients. PC1 immunoblotting was done on a total of 3 normal, 14 metastasis and 14 unaffected liver samples, and was repeated three times. D. Relative amounts of total PC2 (C, left panel) protein, in tumor (T), unaffected (U) and normal liver (N), expressed as a ratio over normal liver samples. Ratios of p75/p66 PC2 isoforms (C, right panel) in the same samples are calculated. SEM is indicated. Simple and double asterisks indicate statistically significant differences (P < 0.05).
Figure 3 Expression of 7B2 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Quantification of 7B2 mRNA expression respectively, in 3 normal (N), 14 tumor (T) and 14 unaffected (U) liver samples. Quantification was obtained by the ratio of the optical density of 7B2 PCR amplification products over GAPDH. Standard error of mean (SEM) is indicated. Asterisk indicates a statistically significant difference (P < 0.003). B. Schematic representation of 7B2 structure. The furin cleavage site is shown (arrow). The antigenic region against which the antibody was raised is indicated (YAb). C. 7B2 immunoblots in normal liver (N1-N3, top panel), in tumor (T1, T2, bottom panel) and unaffected (U1, U2, bottom panel) liver, indicating presence of 7B2 protein only in tumor. Corresponding gels stained with Coomassie blue G250 (CS) are shown. Experiments were repeated three times.
Figure 4 Immunohistochemistry for PC1 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Light microscopy immunohistochemistry of normal liver (N), using 100 × magnification. Arrowheads indicate the scarcely positively stained hepatic cells. B. Light microscopy PC1 immunohistochemistry of liver metastasis (T) and adjacent unaffected parenchyma (U), using 20 × magnification. Arrowheads indicate positively stained tumor cells. C. Light microscopy PC1 immunohistochemistry of liver metastasis (T) using 400 × magnification. Arrowheads indicate positively stained tumor cells.
Figure 5 Immunohistochemistry for PC2 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Light microscopy PC2 immunohistochemistry of normal liver (N), using 400 × magnification. Arrowheads indicate positively stained hepatic cells. B. Light microscopy PC2 immunohistochemistry of liver metastasis (T) and adjacent unaffected parenchyma (U), using 25 × magnification. Arrowheads indicate positively stained cells mainly in the unaffected liver. C. Light microscopy PC2 immunohistochemistry of liver metastasis (T) using 200 × magnification. Arrowheads indicate positively stained tumor cells.
Figure 6 Immunohistochemistry for 7B2 in normal and unaffected liver compared to colorectal (CRC) liver metastases. A. Light microscopy 7B2 immunohistochemistry of normal liver (N), using 400 × magnification. No positive cells were identified. B. Light microscopy 7B2 immunohistochemistry of liver metastasis (T) and adjacent unaffected parenchyma (U), using 100 × magnification. Arrowheads indicate positively stained cells uniquely in the metastasis. C. Light microscopy 7B2 immunohistochemistry of liver metastasis (T) using 400 × magnification. Arrowheads indicate dense positively stained tumor cells.
Figure 7 Expression of PC1, PC2 and 7B2 in primary colon cancer and unaffected colon. A. PC1 immunoblots in primary colon cancer in unaffected colon and in an additional primary colon cancer, note the different PC1 processing pattern between the two colon cancers. Experiments were repeated three times. B. Schematic representation of PC1 structure. C. PC2 immunoblots in primary colon cancer in unaffected colon and in an additional primary colon cancer. Note the absence of the 66 Kda band in the second primary colon cancer. Experiments were repeated three times. D. Schematic representation of PC2 structure. E. 7B2 immunoblots in primary colon cancer in unaffected colon and in an additional primary colon cancer, indicating presence of 7B2 protein only in one tumor. Experiments were repeated three times. Corresponding gels stained with Coomassie blue G250 (CS) are shown. F. Schematic representation of 7B2 structure.
==== Refs
Seidah NG Chretien M Proprotein and prohormone convertases: a family of subtilases generating diverse bioactive polypeptides Brain Res 1999 848 45 62 10701998 10.1016/S0006-8993(99)01909-5
Fahnestock M Zhu W Expression of human prohormone convertase PC2 in a baculovirus-insect cell system DNA Cell Biol 1999 18 409 417 10360841 10.1089/104454999315295
Miranda L Wolf J Pichuantes S Duke R Franzusoff A Isolation of the human PC6 gene encoding the putative host protease for HIV-1 gp160 processing in CD4+ T lymphocytes Proc Natl Acad Sci U S A 1996 93 7695 7700 8755538 10.1073/pnas.93.15.7695
Mbikay M Raffin-Sanson ML Tadros H Sirois F Seidah NG Chretien M Structure of the gene for the testis-specific proprotein convertase 4 and of its alternate messenger RNA isoforms Genomics 1994 20 231 237 8020970 10.1006/geno.1994.1158
Benjannet S Rondeau N Day R Chretien M Seidah NG PC1 and PC2 are proprotein convertases capable of cleaving proopiomelanocortin at distinct pairs of basic residues Proc Natl Acad Sci U S A 1991 88 3564 3568 2023902
Seidah NG Mowla SJ Hamelin J Mamarbachi AM Benjannet S Toure BB Basak A Munzer JS Marcinkiewicz J Zhong M Barale JC Lazure C Murphy RA Chretien M Marcinkiewicz M Mammalian subtilisin/kexin isozyme SKI-1: A widely expressed proprotein convertase with a unique cleavage specificity and cellular localization Proc Natl Acad Sci U S A 1999 96 1321 1326 9990022 10.1073/pnas.96.4.1321
Seidah NG Benjannet S Hamelin J Mamarbachi AM Basak A Marcinkiewicz J Mbikay M Chretien M Marcinkiewicz M The subtilisin/kexin family of precursor convertases. Emphasis on PC1, PC2/7B2, POMC and the novel enzyme SKI-1 Ann N Y Acad Sci 1999 885 57 74 10816641
Mbikay M Seidah NG Chretien M Neuroendocrine secretory protein 7B2: structure, expression and functions Biochem J 2001 357 329 342 11439082 10.1042/0264-6021:3570329
Westphal CH Muller L Zhou A Zhu X Bonner-Weir S Schambelan M Steiner DF Lindberg I Leder P The neuroendocrine protein 7B2 is required for peptide hormone processing in vivo and provides a novel mechanism for pituitary Cushing's disease Cell 1999 96 689 700 10089884 10.1016/S0092-8674(00)80579-6
Cameron A Fortenberry Y Lindberg I The SAAS granin exhibits structural and functional homology to 7B2 and contains a highly potent hexapeptide inhibitor of PC1 FEBS Lett 2000 473 135 138 10812060 10.1016/S0014-5793(00)01511-8
Schiller M Raghunath M Kubitscheck U Scholzen TE Fisbeck T Metze D Luger TA Bohm M Human dermal fibroblasts express prohormone convertases 1 and 2 and produce proopiomelanocortin-derived peptides J Invest Dermatol 2001 117 227 235 11511298 10.1046/j.0022-202x.2001.01412.x
Scopsi L Gullo M Rilke F Martin S Steiner DF Proprotein convertases (PC1/PC3 and PC2) in normal and neoplastic human tissues: their use as markers of neuroendocrine differentiation J Clin Endocrinol Metab 1995 80 294 301 7829629 10.1210/jc.80.1.294
Horsch D Day R Seidah NG Weihe E Schafer MK Immunohistochemical localization of the pro-peptide processing enzymes PC1/PC3 and PC2 in the human anal canal Peptides 1997 18 755 760 9213372 10.1016/S0196-9781(96)00029-0
Khatib AM Siegfried G Prat A Luis J Chretien M Metrakos P Seidah NG Inhibition of proprotein convertases is associated with loss of growth and tumorigenicity of HT-29 human colon carcinoma cells: importance of insulin-like growth factor-1 (IGF-1) receptor processing in IGF-1-mediated functions J Biol Chem 2001 276 30686 30693 11402025 10.1074/jbc.M101725200
Bassi DE Mahloogi H Al-Saleem L Lopez De Cicco R Ridge JA Klein-Szanto AJ Elevated furin expression in aggressive human head and neck tumors and tumor cell lines Mol Carcinog 2001 31 224 232 11536372 10.1002/mc.1057
Cheng M Xu N Iwasiow B Seidah N Chretien M Shiu RP Elevated expression of proprotein convertases alters breast cancer cell growth in response to estrogen and tamoxifen J Mol Endocrinol 2001 26 95 105 11241161 10.1677/jme.0.0260095
Iguchi H Demura R Yasuda D Wakasugi H Effect of LHRH on plasma 7B2 in patients with gonadotropin-producing pituitary adenomas Horm Metab Res 1992 24 31 33 1612556
Iguchi H Yasuda D Yamada Y Funakoshi A Wakasugi H Bloom SR Chretien M 7B2, a possible marker for nonfunctioning pancreatic islet cell tumor Horm Metab Res 1991 23 486 489 1662184
Siegfried G Basak A Cromlish JA Benjannet S Marcinkiewicz J Chretien M Seidah NG Khatib AM The secretory proprotein convertases furin, PC5, and PC7 activate VEGF-C to induce tumorigenesis J Clin Invest 2003 111 1723 1732 12782675 10.1172/JCI200317220
Siegfried G Khatib AM Benjannet S Chretien M Seidah NG The proteolytic processing of pro-platelet-derived growth factor-A at RRKR(86) by members of the proprotein convertase family is functionally correlated to platelet-derived growth factor-A-induced functions and tumorigenicity Cancer Res 2003 63 1458 1463 12670890
Rovere C Barbero P Kitabgi P Evidence that PC2 is the endogenous pro-neurotensin convertase in rMTC 6-23 cells and that PC1- and PC2-transfected PC12 cells differentially process pro-neurotensin J Biol Chem 1996 271 11368 11375 8626691 10.1074/jbc.271.19.11368
Maoret JJ Anini Y Rouyer-Fessard C Gully D Laburthe M Neurotensin and a non-peptide neurotensin receptor antagonist control human colon cancer cell growth in cell culture and in cells xenografted into nude mice Int J Cancer 1999 80 448 454 9935189 10.1002/(SICI)1097-0215(19990129)80:3<448::AID-IJC19>3.0.CO;2-N
Zhu X Zhou A Dey A Norrbom C Carroll R Zhang C Laurent V Lindberg I Ugleholdt R Holst JJ Steiner DF Disruption of PC1/3 expression in mice causes dwarfism and multiple neuroendocrine peptide processing defects Proc Natl Acad Sci U S A 2002 99 10293 10298 12145326 10.1073/pnas.162352599
Zhu X Orci L Carroll R Norrbom C Ravazzola M Steiner DF Severe block in processing of proinsulin to insulin accompanied by elevation of des-64,65 proinsulin intermediates in islets of mice lacking prohormone convertase 1/3 Proc Natl Acad Sci U S A 2002 99 10299 10304 12136131 10.1073/pnas.162352799
|
16293189
|
PMC1310616
|
CC BY
|
2021-01-04 16:03:08
|
no
|
BMC Cancer. 2005 Nov 17; 5:149
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-149
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1471625314410.1186/1471-2164-6-147Research ArticleLimitations of mRNA amplification from small-size cell samples Nygaard Vigdis [email protected] Marit [email protected]øland Anders [email protected] Mette [email protected] Ola [email protected] Eivind [email protected] Department of Tumor Biology, Institute for Cancer Research, The Norwegian Radium Hospital, Montebello, 0310 Oslo, Norway2 Norwegian Computing Center, P.O. Box 114 Blindern, 0314 Oslo, Norway3 Department of Mathematical Sciences, Norwegian University of Science and Technology, 7491 Trondheim, Norway2005 27 10 2005 6 147 147 3 6 2005 27 10 2005 Copyright © 2005 Nygaard et al; licensee BioMed Central Ltd.2005Nygaard et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Global mRNA amplification has become a widely used approach to obtain gene expression profiles from limited material. An important concern is the reliable reflection of the starting material in the results obtained. This is especially important with extremely low quantities of input RNA where stochastic effects due to template dilution may be present. This aspect remains under-documented in the literature, as quantitative measures of data reliability are most often lacking. To address this issue, we examined the sensitivity levels of each transcript in 3 different cell sample sizes. ANOVA analysis was used to estimate the overall effects of reduced input RNA in our experimental design. In order to estimate the validity of decreasing sample sizes, we examined the sensitivity levels of each transcript by applying a novel model-based method, TransCount.
Results
From expression data, TransCount provided estimates of absolute transcript concentrations in each examined sample. The results from TransCount were used to calculate the Pearson correlation coefficient between transcript concentrations for different sample sizes. The correlations were clearly transcript copy number dependent. A critical level was observed where stochastic fluctuations became significant. The analysis allowed us to pinpoint the gene specific number of transcript templates that defined the limit of reliability with respect to number of cells from that particular source. In the sample amplifying from 1000 cells, transcripts expressed with at least 121 transcripts/cell were statistically reliable and for 250 cells, the limit was 1806 transcripts/cell. Above these thresholds, correlation between our data sets was at acceptable values for reliable interpretation.
Conclusion
These results imply that the reliability of any amplification experiment must be validated empirically to justify that any gene exists in sufficient quantity in the input material. This finding has important implications for any experiment where only extremely small samples such as single cell analyses or laser captured microdissected cells are available.
==== Body
Background
Standard protocols for microarray analysis are generally based on samples with more than 1–5 μg of total RNA. However, there is an increasing interest in transcription profiling of small samples, as large amounts of material can be difficult, if not impossible, to obtain in both clinical and experimental settings. Fine needle aspirates (FNA) (~1–2 μg) and fine needle core biopsies (~2 μg of total RNA) offer feasible, atraumatic clinical sampling procedures of limited material. Advances in technology designed for selective collection of specialized cells such as laser capture microdissection (LCM), yields homogenous minute material for further analysis. Following standard protocols in a pilot study, Assersohn et al. [1] had to exclude 85% of the FNA of breast cancer samples from further microarray analysis due to insufficient material. The most common strategy to circumvent the large material requirement using a standard procedure has been to amplify the starting mRNA. The procedure for global mRNA amplification can be performed either linearly, using T7-based in vitro transcription [2-4] or exponentially, using a PCR-based amplification [5] or a combination of both [6]. These methods all include an initial reverse transcription. For the linear method, the yield is 103 to 106 fold amplification when using from one to three rounds of amplification, while an even higher fold up-scaling is possible using exponential amplification. Using either method one can technically amplify RNA from a single cell for gene expression analysis. The protocols for linear and exponential amplification have been subjected to optimizations [7-9], and a number of variations have been described, such as SMART-PCR [10] and Terminal continuation (TC) RNA amplification [11]. However, the most important issue for any amplification protocol to be used in combination with quantitative analysis of gene expression, is that the relative transcript abundance present in the initial mRNA sample is maintained throughout the procedure. In a previous study, we found that the gene expression ratios were not completely preserved between linearly amplified (from 200 ng total RNA) and non-amplified material [12]. Microarray experiments are subjected to variability from a range of sources, and under the experimental conditions applied, we showed that the impact of amplification was increased noise in the form of ratio distortions for some genes [12]. In this study we applied an ANOVA approach to examine the effects of variation as we reduced the amount of total RNA in the first round of aRNA production. We explored the lower range of RNA input that was technically feasible in our hands. A main concern is the lower sensitivity limits with respect to reliable data obtained from minute samples. This highly relevant issue remains under-documented in the literature. The presently published reports are focused on descriptive analysis of results from amplification from minute samples.
Furthermore, estimates of sensitivity limits have in these applications generally been defined as the minimum detectable abundance level that can be quantified in the respective protocols. Quantitative measures of reliability of the obtained data are most often lacking. In the two seminal papers first presenting detailed mRNA amplification protocols for use with microarray technology, the general consensus was a decrease in correlation coefficients and concordance when evaluating diminishing input RNA amounts [3,4]. Increased discrepancies between the data sets when comparing the input total RNA range of 200 ng, 10 ng and 2 ng against 10 ug input were observed [4]. Goley et al. [13] showed that serial dilution of RNA subjected to amplification resulted in increased dissimilarities when compared to unamplified RNA. Several studies have shown that starting with minute samples in the initial amplification round, the number of genes detected with sufficient signal for further analysis decreased [14-16]. Reducing the number of cells under investigation to approximately 1000 cells, Mohr et al. [16] observed that only 19% of the 9850 genes displayed a signal at least two times more than the background signal. A total of 79% of the genes detected had signals above background, indicating that the majority of genes were in the low to moderate expression range. For the examination of lower sensitivity levels, there are examples of studies that have applied spiked transcripts of known concentration [17,18]. Using a PCR based amplification strategy for single cell analysis, a limit of 80 copies per cell was reported to be the lower range for registration of two fold changes in input RNA [18].
For the purpose of applying mRNA amplification and small sized samples in future studies, we chose to define the lower sensitivity limit of an mRNA amplification protocol as the number of mRNA copies needed not only to be detectable but more importantly, produce reliable data. Unlike previous studies, our goal was to quantitatively analyze the reliability of expression data for each transcript present on our arrays as we reduced input RNA material and thus to deduce a transcript dependent cut-off limit with respect to reliability. We applied a novel model-based method, TransCount [19], that estimates absolute transcript concentrations from expression data for each examined sample. We observed the role of stochastic perturbations on the amplification of small number of reactant mRNA molecules. Stochastic noise dominating the signal in amplified mRNA analysis has not been acknowledged in previous reports. With the strategy of computing correlation coefficients between transcript concentrations for two different samples, the sensitivity limits of other amplification methods can also be validated. More importantly, the procedure provides a rational basis for the selection of genes that possess true predictive power.
Results
Descriptive statistics
Total RNA was isolated from three aliquots of 10 000 MT-1 cells and amplified as undiluted reference material. Cells were diluted to obtain aliquots of 1000 and 250 test cell samples. Total RNA isolated from parallel aliquots of 10 000 cells not used in this study, yielded an average of 115 ng when measured using a Nano Chip assay on the Agilent Bioanalyzer. On the same Nano Chip, total RNA from 1000 cells was not detectable. We therefore used 115 ng from 10 000 cells as a reference for relevant RNA calculations in this paper. It is important to emphasize that these downstream RNA calculations, such as amplification factors, only provide estimates and do not reflect sampling errors present during cell aliquot preparations that have undoubtedly influenced the actual yields measured. Hence, we infer that the total RNA yield from 1000 and 250 HeLa cells is approximately 11,5 ng and 2,88 ng respectively (Table 1). In addition to RNA from the cells, all samples were spiked with synthetic RNA transcripts obtained from the Lucidea Universal Scorecard kit to monitor the amplification procedure and used for inference purposes to bootstrap the transcript concentration calculations. These transcripts also served as carriers throughout the multistep amplification procedure. The fraction of mRNA in the total RNA is in the range of 1–3% and by applying the theoretical average of 2% mRNA we have summarized the estimated mRNA input values for each cell size sample (Table 1). We calculated that the average fold amplification factor after two rounds in these three cell size samples (10 000, 1000 and 250) was 1.0 × 104, 0.69 × 104 and 0.89 × 104, respectively (Table 1). The purity of the aRNA samples was measured by absorbance readings. With the exception of one 250 cell batch (ratio = 1.8), the ratios obtained by absorbance measurements were between 2.0–2.6 for all samples. The size distribution of the aRNA products were visualized using an Agilent Bioanalyzer assay. The distribution peaked at approximately 500 bp for all samples (data not shown). Products from 1000 and 250 cells were not larger than 1000 bp, while for the reference sample of 10 000 cells, they were detectable up to 2000 bp. Distinctive peaks representing the most concentrated Scorecard transcripts could be observed on the graphs (graphs not shown). Using a dye swap strategy, we analyzed data from six hybridizations with 10 000 cells versus 1000 cells and six hybridizations with 10 000 cells versus 250 cells, making a total of twelve arrays (Figure 1). On average, 25% of the spots (n = 11791) were filtered from arrays hybridized with material amplified from 1000 cells in the test channel, compared to 37% of the genes filtered when amplified RNA from 250 cells was in the test channel. The data for each filtering per array is presented in Table 2. Examining the filtering of weak spots for each set of cell size samples separately, we found that the average number of signals scored was 20% and 33% less for 1000 and 250 cell samples, respectively, when the reference sample (10 000 cells) was set to 100% (data not shown). These figures are partially confounded by reduced target material labelled in the test samples.
Table 1 Estimates of input RNA quantities and resulting yield for 10 000 cells (reference) and 1000 and 250 cells (test), respectively. Calculation of the average fold yield of aRNA after two rounds of amplification are based on the assumption that 2% of total RNA represents mRNA.
Cell sample size ~Total RNA ~mRNA (2% of total RNA) Synthetic mRNA Sum mRNA Average aRNA yield Average amplification factor
10 000 ~115 ng ~2.3 ng 3.0 ng 5.3 ng 54 μg 1.02 × 104
1000 ~11.5 ng ~0.23 ng 0.3 ng 0.53 ng 3.66 μg 0.69 × 104
250 ~2.88 ng ~0.058 ng 0.075 ng 0.133 ng 1.18 μg 0.89 × 104
Table 2 Filtering per array.
Arrays hybridized with sample size 1000 cells Arrays hybridized with sample size 250 cells
Array number 2 12 1 3 4 11 mean (SD) 6 9 7 5 8 10 mean (SD)
Genes flagged by Genepix 1189 1188 2344 1105 1272 456 1259 (556.2) 1731 1726 1155 1009 4466 3798 2314 (1327)
10.08% 10.08% 19.98% 9.37% 10.79% 3.87% 10.7% 14.68% 14.64% 9.80% 8.56% 37.88% 32.21% 19.6%
Genes flagged manually 10 17 42 27 21 51 28 (14.3) 21 11 104 359 29 69 98.8 (120.6)
0.08% 0.14% 0.36% 0.23% 0.18% 0.43% 0.2% 0.18% 0.09% 0.88% 3.04% 0.25% 0.59% 0.8%
Additionally filtered by spot-background>2× standard deviation of background 113 261 3459 2455 2229 1390 1651 (1198) 1602 562 2020 1292 3886 2184 1924 (1023)
0.01% 2.21% 29.34% 20.82% 18.90% 11.79% 13.9% 13.59% 4.77% 17.13% 10.96% 32.96% 18.52% 16.3%
Sum 1312 1466 5934 3587 3522 1897 2953 (1614) 3354 2299 3279 2660 8381 6051 4337 (2173)
10.17% 12.43% 49.68% 30.42% 29.87% 16.09% 24.8% 28.45% 19.50% 27.81% 22.56% 71.09% 51.32% 36.8%
Array quality index
The potential problem of using inadequate or poor quality input RNA in the test channel is relatively uniform signal intensities because of low fluorescence signals. In such cases the reference channel, which is normally designed to provide a large dynamic range of consistent signal intensities, will be the driving force of the ratio calculations and hence there will be a correlation between gene expression ratios and the signal intensities of the reference channel. To examine this potential problem we used two parameters of array quality index as described by Assersohn et al., [1]. We first examined the SD of the log 10-based signal intensities in the test channel (material from 1000 or 250 cell samples) as a measure of the dynamic range. The span was between 0.59 – 0.3 for all the arrays (Table 3). For 1000 and 250 cell samples the mean value was 0.48 and 0.38 respectively. A defined minimum threshold value for the SD of the signal intensities was not firmly established in the study by Asserhohn et al., [1]. However they applied 0.25 as a minimum requirement. In a preliminary test hybridization, we used 0.2 μg aRNA diluted from one of the reference samples as target material in one of the channels and for comparison, we calculated the SD of the signal intensities for this channel to be 0.5. In the twelve arrays analyzed in this study, the SD's of the signal intensities obtained from the test channel were comparable or higher (0.45 – 0.59) in 50% of the hybridizations. No array had a SD lower 0.3. Hence, we found these magnitudes to be sufficient for the arrays to be used in further analysis and verified by calculating the correlation between log10-ratios and the log10-intensities of signals in the reference channel to ensure that the gene expression ratios were not dominated by the reference target which was prepared from a larger amount of input RNA. The correlations calculated for the twelve arrays were in the order of 0.13 – 0.48, with a mean value of 0.24 and 0.33 for the 1000 and 250 cell samples respectively (Table 3). In comparison, the correlation obtained from two self -self experiments with either even and uneven target preparation in a pilot study was 0.01 and 0.117 respectively. Hence, due to lack of evident correlation between gene expression ratios and reference channel signal intensities, we can exclude the potential domination of the reference channel due to uneven targeting on the arrays used in this study.
Multiple hypothesis testing
From these self-self experiments, the expected log2-ratio values from the processed data were 0 for each gene in every microarray experiment, assuming no influence of material quantity in the initial steps of the amplification. However, the reduced number of genes detected above background levels indicated that low signals corresponding to moderate or low-copy transcripts were potentially more affected by background noise and thus less likely to correlate when performing expression level comparisons to a reference. In our previous report, we found that transcript dependent bias was not apparent in the optimal starting amount (200 ng total RNA) of the amplification protocol [12]. However, by drastically reducing input RNA in the amplification reaction, we observed in this study that the levels of detection were altered and variability introduced. To assess the observed variability in gene expression ratios, we used multiple hypothesis testing to identify genes with log2-ratios statistically significantly different from 0. The statistical analyses were based on filtered, transformed, normalized and dye-swap averaged data using a moderated t-test. P-values were calculated based on empirical Bayes t-test [20]. For the 10 000 versus 1000 cells experiments, we estimated the proportion of genes not differentially expressed to be 66.3%. Using a false discovery rate (FDR) cut-off of 0.05, we found 179 features (1.8%) that were significantly differentially expressed. For the 10 000 versus 250 cells experiments we estimated the proportion of genes not differentially expressed to be 57.1% and a FDR cut-off of 0.05 resulted in a list of 639 features (6.1%) that were statistically significant differential gene expression. Interestingly, between the two sets of significantly differentially expressed genes, there were 110 features in common. We found no connection between these 110 features with respect to signal intensity range or length of probe on array. We extracted the estimated transcript concentrations for these features and found the values for 82 features, representing 70 genes, as some were printed in duplicates on the array. As expected there was a large concentration span. For the 10 most concentrated genes, with the exception of one gene, the high transcript abundance was not maintained in the test samples compared to the reference samples as deduced from the respective raw data. For the remaining 60 genes, the general trend was equal or higher signal intensities in the test samples with respect to the reference channel. After normalization, these genes were overexpressed in the test samples, and hence, significantly differentially expressed.
ANOVA analysis
We assumed that the variance observed between test samples and the reference was due to limited material in the test samples. To investigate this assumption we identified all potential factors contributing to variation and constructed an ANOVA model to estimate their importance, and thus isolate the effect of reduced starting material. The estimated contributions from the sources of variation modelled are presented in Table 4. We noted that μ, C, D and A were small as expected since the ratios are normalized. The dye-gene (DG) interaction was of moderate size, possibly as a result of difference in amount labelled in the two channels. The cell sample size-gene (CG) interaction was small, indicating that the noise due to differences between sample size 1000 and 250 cells was quite small. In accordance with the results from multiple hypothesis testing, the gene effect (G) was relatively high. For the interaction between replicates and gene (BG, B1G, B2G), we obtained the gradient of noise level in the expected order, from the lowest in 10 000 cells to the highest in 250 cells. There was a relatively small difference between the 1000 and 250 replicates.
Table 4 Parameter estimates in the ANOVA model. This is a mixed-effects model, as the first three effects are fixed and the others are random. The noise in these experiments was largely due to gene and to the interaction between replicates and gene. Reduction in samples size yielded increased noise since
Fixed Effect Explanation Estimated value
μ Fixed overall level -0.13
C Sample size; 1000 or 250 cells 0.055
D Dye ratio; cy3/cy5 or cy5/cy3 0.10
Random effects E~N(0, σ2E) Explanation Estimated standard deviation ()
A Array; 1,...,12 0.14
G Gene; 1,...10643 0.47
CG Interaction: cell size sample and gene 0.044
BG Interaction: replicate (10 000 cells) and gene 0.17
B1G Interaction: replicate (1000 cells) and gene 0.41
B2G Interaction: replicate (250 cells) and gene 0.45
DG Interaction: dye ratio and gene 0.18
ε Model and measurement error 0.33
Absolute transcript concentration estimates and cut-off limits for unreliable data
Larger replicate-gene (B1G and B2G) interactions obtained from the ANOVA analysis imply a greater degree of unreliable data present in the data set obtained from reduced samples. To strengthen possible biological findings in small samples, it is of great value to remove uncertain measurements. Empirically, observations suggested that these measurements were characterized by low signals related to low transcript abundance. We applied the TransCount model as an approach to eliminate unreliable data based on quantitative measures of template input in the initial amplification and for the threshold determination. With TransCount we found estimates of transcript concentrations for each gene per sample. For the reference (10 000 cells), a conversion factor was calculated for obtaining absolute concentrations. The conversion factor was generated by first estimating the concentrations estimates of the synthetic Scorecard templates, and secondly, deducing the factor by using linear regression. The absolute concentrations of each of 8116 genes present in the reference sample (10 000 cells) were estimated. For values above 0 (n = 8085), the range was between 0.3 and 40 000 transcripts per gene in an MT-1 cell. These quantities represent the contents of one cell prior to amplification. Multiplying by 10 000 cells, the amplification factor and the fraction of aRNA used for labeling, the range was between 5.8 × 105 and 7.8 × 1010 transcripts per gene in the labelled cDNA pool that was applied to the array and represented the reference channel (Table 5). For this calculation, we assumed efficient conversion from aRNA to cDNA. Using TransCount, we obtained, in addition to the transcript concentrations for each gene and sample, the posterior joint probability distribution of all concentrations. This distribution could be used for computing various distributions and probabilities. We used it for calculating the distribution of the Pearson correlation coefficient between the reference sample concentrations and the 1000 cell sample concentrations for each gene (Fig. 2a). Similarly, the same procedure was performed for the reference and 250 cell samples (Fig. 2b). The graphs clearly showed that the correlation coefficients were copy number dependent. High copy numbers yielded high correlation coefficients. As expected, fewer transcripts per gene were necessary to obtain high correlations among genes amplified from 1000 cells compared to genes amplified from 250 cells. In Fig. 2a we observed a region where certain molecules were influenced by stochastic effects resulting in correlation coefficients alternating between poor and good. To define a cut-off value that ensured the exclusion of unreliable data in downstream analysis, we calculated the probability of positive correlation for each gene based on the distribution of the correlation coefficients. For the 1000 (250) cell sample, the probability of positive correlation was at least 0.99 when concentrations were 120.881 (1806.214) or more. Hence, in experiments using 1000 cells, the least number of templates required for further analysis was 121 copies per cell. For 250 cells the threshold was 1806 copies per cell. To convert the limit from minimum transcripts per gene in a cell to minimum number of transcripts per gene in the cDNA pool applied to the array, we multiplied by cell sample size, amplification factor and fraction of aRNA in the labeling reaction (Table 5). This calculated threshold was six times higher when initially starting from 250 cells compared to 1000 cells. With respect to the serially diluted synthetic templates, the two most diluted calibration spikes (cYIR09 and cYIR10) in the samples with 1000 cells were below the threshold, while for 250 cell samples, the five most diluted calibration spikes (cYIR06-cYIR10) were below threshold. The number of genes left for analysis when applying the cut-off values 121- and 1806 copies per cell were 3149 and 390, respectively. This represented an eight times difference of data loss when the sample was reduced from 1000 to 250 cells. Hence, for 250 cells as input value in the amplification procedure, only a relatively small fraction of the genes queried on the array were suitable for further functional analysis.
Table 5 Conversion of gene transcripts per cell to gene transcripts applied to the array. The number of molecules per gene hybridized to the array varied between 5.8 × 105 – 7.8 × 1010 with respect to the reference sample. The reliability threshold in terms of minimum number of molecules per gene applied to arrays for 250 cells, 6.8 × 108, was six times higher than the threshold for 1000 cells.
Cell sample size Description Copies per cell Amplification factor Fraction aRNA labeled Equivalent number of molecules applied to array
10 000 reference transcript concentration range 0.3 – 40 000 1.02 × 104 0.019 5.8 × 105 – 8.1 × 1010
1000 reliability threshold 121 0.69 × 104 0.137 1.1 × 108
250 reliability threshold 1806 0.89 × 104 0.169 6.8 × 108
Figure 2 a and b. Correlation of transcript concentration estimates. Using the TransCount method, we obtained for each gene the distribution of the correlation coefficients between the reference sample transcript concentrations and the 1000 (250) cell sample transcript concentrations. Summary values for a certain gene in the (mean) 1000 cell samples versus the (mean) reference samples, are plotted against the estimated concentration for the (mean) reference sample in Fig. 2a. The black solid line is the median correlation coefficient values. The blue and green dashed lines are the 2.5% and 97.5% quantiles, respectively. The vertical dashed black line is the reliability threshold 121, i.e. the value for which the probability of positive correlation is at least 0.99. Similarly, information about the distribution for the (mean) 250 cell sample and (mean) reference is summarized in Fig 2b. In this case, the reliability threshold is 1806. The number of genes per concentration is shown below each respective plot. For a certain concentration c, the number of genes is counted from the interval
We examined the signal intensity distribution of the genes that were filtered according to the cut-off value, but not by the weak spot filter criteria used when pre-processing the data. We chose to investigate the genes with at least two observations out of the three dye-swap duplicates. As expected, the majority were in the low signal range, with a slightly wider distribution range for data obtained from 250 cells (Fig. 3). A filtering criterion of <1500 in mean signal intensity would remove 95% of the genes categorized as unreliable by TransCount when amplifying from 1000 cells. This intensity criterion was chosen by visual inspection of Fig. 3. However, due to the greater spread of intensities in the 250 cell samples, a similar criterion was more difficult to determine. A removal of 95% of the genes required a signal >3270.
Figure 3 Signal intensity distribution of genes below the statistically defined threshold. Accumulation of genes in the low to moderate detection range when examining the mean signal intensity distribution of genes below the reliable threshold, but not filtered by the weak spot criteria. These genes were observed in at least two of the three dye-swap duplicates.
Discussion
To examine the conservation of ratio profiles between increasingly smaller cell size samples, we performed a series of analyses on microarray data generated from an experimental design using one source of material, but with varying amounts RNA into the amplification procedure prior to hybridization. Many studies have applied serially diluted RNA from the same total RNA sample to evaluate amplification procedures [3-5,13]. However, this does not approximate experimental conditions for separately processing minute samples from start to finish, as is the case in patient sample handling. Hence, the experimental procedure started with isolation of total RNA from 9 independent and cell counted samples drawn from the same cell culture. As the samples varied in size, two RNA isolation kits were applied that each were specially designed to cover a certain quantification range of cells. Both kits originated from the same manufacturer. They were based on the same chemical components and only differing in the RNA binding capacity of the silica fiber matrix and subsequent elution volume. Our aim was to maximize total RNA yield as suboptimal RNA isolation from the small test samples could result in biased data. In order to optimize the generation of amplified products and the degree of correlation between parallel samples, we thus chose to include the entire sample in the amplification procedure setup. We made conservative estimates of amplification factors, knowing that they are prone to sampling errors in the actual number of initial cells in each sample, in addition to variable efficiency in isolation of total RNA. For the lower limit of input material, the sample size was set to 250 cells, as this is, in our experience, the lower range of the amplification protocol with respect to generating sufficient aRNA for labeling probes in a dye-swap strategy. The aRNA yields from the small test samples were a determining factor for the amount used in the target labeling reactions. We applied a relatively equal percentage of the aRNA yield from the two test samples sizes for the respective labeling reactions. The labeling amount for the reference target was set to a constant value across all arrays and thus uneven target amounts were used being aware of the confounding contribution to the data filtering process. It is clear that in realistic clinical settings, patient sample size variability may be extensive both between patients and within the same patient and obtaining sufficient aRNA to perform microarray analysis from particular samples may be challenging. The essential requirement for inclusion of scarce aRNA material in data analysis is sufficient array quality. We used array quality index parameters to ensure adequate quality of the target used in the test sample channel and lack of reference channel domination of the gene expression ratios (Table 3). An alternative approach to further increase the yield of scarce material in an attempt to equate reference target material amount in the desired number of hybridizations, would be to introduce a third round of amplification. Scherer et al. [21] reported that an additional third round only had a modest effect on reproducibility. However, a third round would increase the work load, as it is optimal to compare aRNA based on equal number of amplification rounds, and also risk plateau effects with high copy gene saturation especially if using a reference sample in the experimental design. In fact, two rounds of amplification sufficiently upscaled many genes to saturation levels in the reference samples used in this study. Extending the in vitro transcription step is not a recommended alternative for increasing aRNA amounts due to increase in non-template based products. In vitro transcription past four hours has been shown to have negative effects in that the aRNA becomes degraded and thus reduces the quality of post-amplification procedures such as microarray analysis [22].
Table 3 Array quality index. The SD of the log10-intensities in channel 1 give an indication of the dynamic range obtained from the test samples (1000 or 250 cell samples). The correlation coefficients between gene expression log10-ratios of the experiment and log10-intensities of channel 2 (reference sample) were calculated to confirm that the gene expression ratios are not determined by the signal intensities of the reference channel.
Arrays hybridized with sample size 1000 cells Arrays hybridized with sample size 250 cells
Array number 2 12 1 3 4 11 mean 6 9 7 5 8 10 mean
Standard deviation of test channel signal intensity 0.6 0.49 0.3 0.39 0.52 0.59 0.48 0.45 0.53 0.3 0.34 0.3 0.38 0.38
Correlation ratio vs. reference channel signal intensities 0.13 0.23 0.35 0.29 0.26 0.2 0.24 0.27 0.28 0.26 0.25 0.48 0.43 0.33
The amplification protocol we used applies to cDNA arrays and modifications, as alternative approaches are required for oligo arrays due to the antisense orientation of the aRNA products.
Descriptive analyses of the twelve arrays showed a reduction of the number of genes with sufficient expression levels for further analysis in the test channel representing the aRNA generated from 1000 or 250 cells, compared to the reference channel. However, the average number of features scored on arrays with material from 250 cells (6300 genes) was comparable to the amount of data obtained using 30 μg of unamplified total RNA (6400 genes) from the same cell source on comparable array prints. Loss of data and lowered number of genes detected in earlier reports indicate that there is a bias for inaccurate representation of low copy number genes in the amplified pool of aRNA from minute samples. Another important issue is that low signals corresponding to moderate or low-copy transcripts may be more affected by background noise and thus less likely to correlate when performing comparisons. Although candidate genes may be extracted, the low signal-to-background ratio indicates that investigators need to be aware of the potential for inconsistent data interpretation.
To study the proportion of inconsistent data when using amplified material, correlation coefficient calculations have been the main choice of statistically based analyses. This is especially true for estimations of inconsistency in data generated from amplified material compared to non-amplified or from different amplification protocols [13,23-26], However, more sophisticated statistical analytical methods have also appeared [27-30]. To analyze the fidelity that the amplification procedure has on differential gene expression between two different samples, a comparison of t-scores or posterior distribution of fold change for individual genes have been applied [28-30]. The reported results from these analyses were restricted to a small subset of genes, more specifically the outliers, e.g. the top ranked 10 genes or genes with a fold change of >2. Without generating specific gene subsets, we used a t-test in our previous study to indicate the preservation of ratio when amplifying the input material and illustrated by plotting the ratio difference vs. the p-value for each gene. Likewise, a similar approach was used by Schlingemann et al., [31] to show how ratio preservation was maintained when serially diluting the RNA amount used in their sense-oriented amplification protocol. For the purpose of our investigation in this study, we globally estimated the portion of genes differentially expressed in the reference and test samples based on p-values calculated from a moderated t-test. Using the same sample source, the expected log2-ratio value was 0. We found that the number of differently expressed genes increased when comparing the smallest test sample size (250 cells) with the reference. The increase in differentially expressed genes indicated increased variability when reducing sample size.
In this study we applied an ANOVA model to estimate how much variability was introduced due to technical aspects when amplifying from very limited material. Ideally, variable sample size should not affect ratio measurement uncertainties. However, we showed there was a non-negligible increase in variation when sample size decreased. In this study, the variation was captured by the interaction between replicate sample sizes of 1000 cells (250 cells) and gene, BG. This estimate was of considerable proportion with respect to the various sources of noise in the ANOVA model. We found it necessary to address, characterize and filter data affected by this variation.
The main factors possibly influencing loss and variance of gene expression data when reducing input RNA are sequence-dependent bias and transcript abundance bias. In our previous work we found no statistical evidence to support sequence dependent bias [12]. In exponential amplification methods, the situation may be different as DNA polymerase is more prone to sequence bias than RNA polymerase. Transcript abundance bias was not evident in the optimal starting amount (200 ng total RNA) of the amplification protocol and we found the degree of transcript representation in the resulting aRNA pool with respect to the initial total RNA transcript population to be satisfactory for the use in gene expression analysis. We also observed that amplification prior to microarray hybridization increased the data output when comparing to standard total RNA labeling and hybridization. This was due to improved signal intensities of genes with low transcript numbers. Hence, we were able to extract information from a substantially larger number of genes using amplified material compared to non-amplified. However, by drastically reducing RNA in the reaction we are initiating amplification from samples with fewer copies of rare transcripts and this may increase the chance of data loss and variation due to stochastic distribution of low copy number templates. We may classify mRNA transcript abundance into 3 groups; high, moderate and low abundance. There are about 12 000 different transcripts per cell and over 90% of these are represented by the low-copy abundance class, having 1–15 copies of the transcript per cell [32]. According to SAGE data, more than 83% of transcript were present in one copy per cell [33]. In an approach to examine the effects of abundance in a comparison between amplified and non-amplified data and replicates within the two groups, Sheidl et al. [34] divided the genes into 10 groups based on their intensity and Pearson correlation coefficients were calculated for each group respectively. The results demonstrated a clear effect of abundance bias on the correlation. For low copy number transcripts, the coefficient approached 0.2 as intensity reached background levels. The high variability in the low abundance region of both amplified and non-amplified experiments show that this is in an unreliable region of transcript concentration irrespective of the inclusion of an amplification step [34]. A point of relevance for comparative microarray studies can be drawn from a previous assessment of stochastic effect in quantitative PCR [35,36]. When the queried template is present in low copy numbers in both the test and reference sample, the variation in ratio between the two templates is dependent on the stochastic distribution of both transcripts, resulting in a multiplication of stochastic effects. These observations can further be related to the term Monte Carlo effect coined by Karrer et al. [37]. The Monte Carlo effect is described by small, random and template concentration dependent differences in amplification efficiency. Each template has a certain probability of being amplified or lost, resulting in inconsistent detection. The lower the abundance of any template, the smaller the probability is that its true abundance will be maintained in the amplified product. It is therefore evident that the role of stochastic effects on the amplification of small number of mRNA molecules needs to be recognized. These effects have not previously been identified in relation to mRNA amplification procedures and subsequent analysis. Gene expression measurements affected by stochastic fluctuations may result in highly misleading quantitative information. It is of great advantage to the investigator to be able to define the limit of accuracy in terms of the number of template copies needed to avoid such obscured gene expression measurements. Data acquired above this limit should theoretically be sufficiently reproducible for further biological interpretation. In a study addressing microarray sensitivity using non-amplified material, the minimum transcript concentration that provided reliable measurements was defined to be the interception between the distribution of signal intensities, with the distribution of signal intensities from differentially expressed genes [38]. As mentioned above, to analyze fidelity of amplification on differential gene expression between two different samples, comparisons of t-scores or posterior distributions of fold change for individual genes have more recently been applied [28-30]. However, reports from the analyses were in these cases restricted to a small subset of genes. Maintenance of fidelity across levels of gene expression was not investigated in these studies. Until now, few studies have presented quantitative strategies to establish the estimated number of transcripts necessary prior to amplification to consistently maintain the relative abundance present in the un-amplified sample. The use of spiked control transcripts of known concentration is one method to approach this question. Using a PCR based amplification strategy for the analysis of single cells, a limit of 80 copies per cell was reported [18]. For two rounds of PCR-based amplification of mRNA and subsequent single cell transcript profiling, the analysis of spike controls in a single channel experiment led to a cut off limit corresponding to 1–2 template copies per cell [39]. Fifty cycles were used in the first PCR round followed by thirty in the second. However, the use of two rounds of PCR is to our knowledge limited. Another approach is to verify and compare relative expression levels obtained with microarray data with other quantitative technologies to possibly set a cut-off limit. Quantitative RT-PCR is a commonly applied technique. However, verification is performed on a gene-by-gene basis and is not suited for a high throughput verification of low expressing genes. Neither is fluorescent correlationspectrometry (FCS), a direct and sensitive technology for quantification of single molecule (gene) in solution [40]. A medium throughput technique is the standardization of competitiveRT-PCR (StaRT-PCR) [41]. This multiplex PCR technique allowed the simultaneous expression and quantification measurements of 25 genes. However, the method requires careful design of competitive templates for each gene. Other emerging PCR based solutions, such as SmartProbe [42], where pre-designed primers and probes may be used to provide quantitative information on gene expression in a more high-throughput format. To avoid the obstacles inherent in most of the technologies mentioned above, and given the limited sample material, we propose the use of high throughput absolute quantification of the respective genes and a subsequent comparison with correlation coefficients obtained in our experimental design for establishment of a limit of accuracy for the sample cell size in question. The first step requires acquisition of absolute abundance data for the genes to be queried in the material source. Other high throughput methods for quantification of gene expression levels other than cDNA and oligonucleotide arrays are represented by techniques such as serial expression of gene expression (SAGE) [43] and massive parallel signature sequencing (MPSS) [44]. The advantage of SAGE is that it provides absolute quantification of transcripts compared to relative quantification using microarrays. The disadvantage with SAGE is the large number of clones that must be sequenced and the method is not as widespread as microarray technology. MPSS also allows a direct quantification like SAGE. In the MPSS procedure, the sequencing step is modified, such that parallel sequencing is performed directly on the solid bead. However, the technology requires special equipment and is only available through Lynx Genetics. In this study, we have presented the successful use of TransCount as a high throughput alternative approach for absolute transcript concentration per gene queried on the array. TransCount can handle experiments based on amplified as well as non-amplified material. The results of this study are based on cDNA array data but the model is also applicable to data obtained with oligoarrays.
To characterize the genes affected by variation, we ranked the genes in the data set according to estimated concentration of the transcripts in the reference sample (10 000 cells) and plotted the correlation coefficient on the y-axis. The plots clearly showed that the degree of variability was dependent on the template concentrations. This is comparable with the intensity-based analyses by Sheidl et al. [34]. To reduce variability and to correctly assess low abundance expression ratios, they suggest replicates of the experiment. Replicated hybridizations are not always an option when scarce material is applied, and although some measurements are strengthened by more observations such as those in the vicinity of background levels, uncertainties will still revolve around others, especially those affected by stochastic fluctuations. We proposed a filtering strategy based on the results of TransCount. In a sample consisting of 1000 cells, we found that at least 121 copies of a gene had to be present per cell for reliable preservation of expression level and subsequent detection. Likewise, the threshold was 1806 per cell when only 250 cells were used. In total, these thresholds amount to 121 000 and 451 500 templates per sample size investigated. This is a difference of 3.7 fold, which is comparable to the dilution factor of 4. The majority of unreliable genes was rejected if a minimum signal intensity filter of 1500 was required in the hybridizations where 1000 cells were used (see Fig. 3). In other words, unreliable genes are present in the moderate signal intensity range that generally is not considered to be a high variance range, as these signals are well distinguished from background levels. A similar minimum signal requirement was not easily determined for 250 cells as unreliable genes were not only low/moderately expressed, but also extended towards high expression. The reliability limit set by TransCount for the respective sample sizes was therefore not potentially influenced by unbalanced, aRNA target preparation, as large portions of the unreliable genes were sufficiently above background levels indicating that the threshold is in fact determined during the initial phase of the amplification procedure when mRNA templates are copied into cDNA. This fact applies to all other amplification protocols, even those with higher amplification yields than the one applied in this study. Further, this emphasizes the need for an alternative approach to remove biologically irrelevant data such as the one presented in this article where stochastic effects were acknowledged and minimum limit for reliable detection was devoid of such effects. Identification of which and how many genes were available for analysis if the sample size was further reduced was difficult to estimate with certainty given the limited number of samples sizes in our experimental design. However, assuming a linear trend, we can predict that for a sample size of 100 cells, at least 1 210 000 templates per gene to be monitored have to be present in the pre-amplification mixture for reliable measurements. This figure translates to 12 100 copies per cell, and for our particular RNA source, we are left with 35 genes for analysis. In this range, we should consider what is the reasonable minimum number of cells from which we can obtain any informative qualitative results even if only studying high expression genes. If our genes of interest are below the threshold, then microarray analysis will in fact not bear relevant information. The cut-off value, the number and identification of genes left for downstream analyses will vary depending on RNA source. It is evident that single cell analysis is beyond the sensitivity of this assay.
Conclusion
For application of microarray technology combined with amplification of mRNA from nano- to picograms of mRNA, there is a pitfall of conferring biological relevance to unreliable data due to transcript abundant bias alteration of true gene expression. We demonstrated that with our strategy, and the use of TransCount, we can define limits with respect to number of transcripts necessary for meaningful interpretation of expression data from conditions using reduced input RNA, thus avoiding obscured expression variability. As the input cell number decreased, the necessary number of transcripts per cell increased. The potential impact of this should be reflected in the design of future studies. Assembling transcription concentration data from the source material of interest using TransCount should allow a prediction of which genes will provide meaningful results when minute samples of the same source are applied.
Using the framework presented in this study also allows evaluation of alternative amplification protocols, such as various PCR-based strategies
Furthermore, based on our findings regarding the sensitivity limit of our amplification protocol in use, only moderate to high expressing genes can be regarded in experiments with <1000 cells. All data from low expressing genes should be disregarded due to inaccuracy. For minute samples, any template diluted past a certain threshold copy number determined by assay sensitivity, will experience large variation in amplification whose origin is the stochastic nature of the biochemical reaction. In the event that the gene of interest is subjected to abundance dependent bias variation, only qualitative information can be obtained. In agreement with Stenman and Orpana [35], we conclude that quantitative expression profiling of single cells is only potentially possible for genes expressed at high levels that show reproducible expression in replicate experiments. Single cell analyses through mRNA amplification are burdened by stochasticity not only in the procedure itself, but also at the level of transcription as intrinsic noise [45], thereby severely limiting the precision of gene expression measurements. It is evident that sophisticated technologies for the selective collection of specialized cells are a step ahead of commonly available, high throughput quantitative expression profiling technologies. Caution is warranted when extrapolating biological relevance from the increasing number of expression profiling results published based on extremely low cell numbers.
Methods
cDNA arrays
The 13 k human cDNA arrays applied in this study were printed in house with cDNA clones from the Research Genetics 40 k cDNA clone library (Invitrogen, Carlsbad, CA.). The cDNAs were prepared from the clone library by PCR using M13 universal primers. The purified PCR products were resuspended in 3 × SSC and spotted on amino silane coated slides (CMT GAPS, Corning Life Sciences, Corning, NY) using a Micro Grid II robotic printer (Bio Robotics, Cambridge, UK). After printing the slides were cross-linked by UV to immobilize the double stranded probes. For details on the arrays, we refer to the website for the microarray core facility at The Norwegian Radium Hospital [46].
RNA purification
The human carcinoma cell line, HeLa, was used throughout the study. The cells were maintained in RPMI media (Bio Whittaker Europe) supplemented with 10% calf serum (PAA Laboratories, Linz, Austria). Phosphate-buffered saline (PBS) was used to dilute detached cells from one 25 cm2 culture flask and make aliquots of 10 000, 1000 or 250 cells. The cells were stored in lysis buffer (Stratagene, La Jolla, CA) at -70°C. Total RNA from 10 000 cells aliquots were isolated using Microprep kit (Stratagene), while the Nanoprep kit (Stratagene) was used for isolation of total RNA from 1000 and 250 cell samples. The eluted RNA was immediately applied in the subsequent amplification procedure.
RNA amplification
Two rounds of RNA amplification were carried out as described earlier [12]. For the 1000 and 250 cell samples, all volumes in the amplification procedure were reduced by a factor of 0.5 with the exception of in vitro transcription reactions. They were performed in the standard volume quantity suggested by the manufacturer. The reason for decreasing the reaction volumes was to reduce the difference in the primer to template ratio and thus minimize possible production of primer-dependent products. Briefly, the eluted total RNA was speed vacuumed with a primer annealing mix consisting of a dT/T7 primer, synthetic reference RNA (Lucidea Universal Scorecard reference RNA, Amersham Pharmacia Biotech AB, Uppsala, Sweden) in the case of 10 000 cell samples, and synthetic test RNA (Lucidea Universal Scorecard test RNA, Amersham Pharmacia Biotech AB) plus 40 ng of tRNA in the 1000 and 250 samples. Both reference and test Scorecard RNA contained one set of calibration control templates serially diluted (cYIR01-cYIR10), and one set of ratio control templates (rYIR1-rYIR8). We diluted 3 × 1 μl of scorecard RNA by the following scheme: 10 × dilution of Lucidea reference RNA was added to RNA from 10 000 cells, while 100 × and 400 × dilution of Lucidea test RNA was added to RNA from 1000 and 250 cells, respectively. Second strand synthesis was initiated by RNase digestion. The purified double stranded cDNA served as template for the first round of aRNA transcription. In the second round of amplification, the first strand cDNA was synthesized by priming with random hexameres. Second strand cDNA synthesis was initiated by annealing a dT/T7 primer to the aRNA/cDNA heteroduplex with partially digested aRNA. A second in vitro transcription reaction followed. The concentration of aRNA was determined by OD260 reading in 50 mM NaOH. An mRNA Nano Chip (Agilent Technologies, Palo Alto, CA) was used to examine the aRNA products on an Agilent 2100 Bioanalyzer.
RNA labeling
The amounts of aRNA used in labeling reactions were; 1 μg of aRNA from 10 000 cells, 0.5 μg aRNA from 1000 cells, and 0.2 μg aRNA from 250 cells. An indirect labeling procedure (FairPlay Microarray labeling kit, Stratagene) was performed according to the manufacturer's protocol. Random hexamers (8 μg) were used to prime the cDNA synthesis. Labeled Cy5- and Cy3-cDNA was eluted from their respective columns using 50 μl Tris-HCl (pH = 8.5), then mixed with 20 μg of human cot-DNA (Invitrogen, Carlsbad, CA) and 300 μl 0.5 × TE. The mixture was concentrated to ~10 μl using Microcon YM-30 columns (Amicon, Millipore Corporation, Bedford, MA) before the addition of SlideHyb#3 hybridization buffer (Ambion Inc, Austin, TX) to a total volume of 112 μl.
Hybridization and scanning
The hybridization mixture was heated to 100°C for 3 min and subsequently centrifuged at 13 K r.p.m. for 10 min before applying to the microarray slides fitted in hybridization chambers to a GeneTac automatic hybridization station (Genomic Solutions, Ann Arbor, MI) for overnight hybridization at 65°C. Prior to scanning, the slides were washed in the following solutions: 2 × SSC and 0.1% SDS, 1 × SSC and 0.1% SDS on the hybridization stations, while two washes in 0.05 × SSC were performed manually. The slides were dried by centrifugation. Scanning was performed with an Agilent DNA Microarray Scanner, model BA (Agilent Technologies, Palo Alto, CA) at 100% laser value and variable PMT values for optimal signal acquisition. Data from the images were acquired using GenePix Pro 4.0 software (Axon Instruments Inc., Union City CA).
Description of experiments
Three aliquots of 10 000 cells were amplified in parallel and used as reference against three amplified aliquots of 1000 and 250 cells, respectively. Hence, using a dye swap strategy, six arrays were hybridized using aRNA from replicates of 10 000 cells as a reference against aRNA from 1000 cell sample replicates. Likewise, six arrays were hybridized using 250 cell sample replicates versus reference. Thus, in total there were twelve arrays. The experimental design is presented in Fig. 1. The microarray slides were all from the same print batch.
Figure 1 Experimental design. Three replicates of each cell size sample were amplified. Each reference replicate B (10 000 cells) was hybridized to test samples B1 (1000 cells) and B2 (250 cells), respectively, in a dye-swap strategy. The arrows represent arrays, alternating in direction to indicate dye-swap. In total, six arrays were used for each test sample size versus reference.
Data preparation
From the data for each microarray we first manually removed technically flawed spots, and secondly, we removed spots automatically flagged by the GenePix software as not found. No additional data preparation was included prior to analysis based on TransCount because this method handles even very noisy data, and normalization is incorporated into the model. For the array quality index, multiple hypothesis testing and ANOVA analysis, also spots where the spot intensity (uncorrected foreground intensity) was lower than the background intensity plus two standard deviations of the background in any of the two channels were removed. An overview of the number of filtered genes is given in Table 2. In addition, systematic errors were corrected by normalizing the data using the locally weighted scatterplot smoother, lowess, as described in Yang et al. [47].
Array quality index
To ensure that the intensity in the reference channel (10 000 cell samples) did not dominate the log ratios due to differences in the amount labelled of the two comparative targets, we explored the following two parameters using filtered and normalized data. First, the SD of the log10-intensities of the test channel (target prepared from either 1000 cell or 250 cell samples) was calculated to give an indication of the dynamic range. Secondly, the correlation between log10-ratios versus the log10-intensities of the reference channel (10 000 cell sample) was calculated for the genes on each array to detect reference channel domination. The two parameters for each array are listed in Table 3. Scorecard genes were removed prior to this analysis.
Multiple hypothesis testing
To identify genes with log2-ratios significantly different from 0, p-values were first calculated for each gene using a moderated t-test [20]. The moderated t-test applied is based on empirical Bayes analysis and is equivalent to shrinkage (or expansion) of the estimated sample variances towards a pooled estimate, resulting in a more stable inference when the number of microarray experiments is small. Separate analyses were performed for the experiments involving 10 000 cells vs. 1000 cells and 10 000 vs. 250 cells. Based on these p-values, the proportion of genes not differentially expressed was estimated using the convex decreasing density estimator of Langaas et al. [48]. The method is built upon the assumption that the distributions of the p-values for the genes that are not differentially expressed follow a uniform distribution. Finally, adjusted p-values were calculated using the Benjamini-Hochberg step-up procedure [49], taking the estimated proportion of genes not differentially expressed into account. Using a cut-off of the adjusted p-values at 0.05 gives an approximate level of False Discovery Rate (FDR) at 0.05.
The moderated t-test and the convex decreasing density estimator were implemented in the Limma R package available as part of the Bioconductor project [50].
Analysis of variance modeling
To investigate the different sources of variability, we set up an ANOVA-based statistical model. Related models are found in Kerr et al. [51], Wolfinger et al. [52], Jin et al. [53] and Nygaard et al. [12]. Let log2 denote the log2-transform of the normalized measured ratio (signal from 1000 or 250 cells divided by signal from 10 000 cells) for gene g on array a, for cell sample size c, for replicates b, b1 or b2, having dye ratio case d. We explain the log2-transformed ratio by the following model:
with c = 0,1, d = 0,1, a = 1,...,12, g = 1,...,10643, b = 0,1,2, b1 = 0,1,2, b2 = 0,1,2, where μ is the overall mean, Aa is the overall array effect of array a, Cc is the overall cell sample size effect of cell sample size c, Dd is the overall dye effect of dye ratio case d, and Gg is the overall gene effect of gene g. Furthermore, CGcg is the interaction between cell sample size and gene, so if this effect is significant, the cell sample size has different effect for different genes. (-1)d DGg represents the gene-specific dye ratio effect. BGbg ( and similarly) models different effect of genes for different replicates i.e. BGbg is the effect of gene g and replicate b of the 10 000 cell sample size. μ, Cand D are fixed effects, the others are random effects. B, B1 and B2 could have been included as random effects in the model, but the number of repetitions was too small. They should therefore have been fixed effects instead. However, as they are confounded with each other and C and D, and as we expect them to be 0, they have not been included as single effects in the model. The parameters in the mixed-effects ANOVA model were estimated using Gibbs-sampling. We refer to Follestad et al. [54] for estimation details.
TransCount-based absolute transcript concentration estimation
The TransCount model is based on the idea of following the mRNA molecules through the microarray experiment, from cDNA synthesis to hybridization and subsequent washing, incorporating available information about the experiment. The process is modelled as a stepwise selection, where each molecule has a certain probability of being kept in the experiment. Using a Bayesian technique, the highly multivariate joint posterior distribution of all transcript concentrations is estimated. Details about the model and estimation method are found in the article by Frigessi et al. [19].
The model was first applied to spikes in the reference sample (10 000 cells). Expression data from synthetic Scorecard calibration control templates pre-diluted in a serial manner by the manufacturer was incorporated into the TransCount model to obtain the respective absolute concentration measurements. We chose three spikes based on signal intensities situated clearly within the boundaries of saturation and background levels to calculate a conversion factor. This factor was obtained using linear regression and information about the known spike concentrations per cell. Furthermore, the conversion factor was used to estimate absolute transcript concentrations for each gene in the reference cell sample. These estimates represented the number of transcripts per cell before amplification. The transcript concentrations for the test samples 1000 and 250 were also estimated using the TransCount method. The results from TransCount were used to calculate the distribution of the Pearson correlation coefficients between transcript concentrations for different sample sizes. The correlation for a certain concentration c was computed from genes with concentrations in a small interval around c.
Authors' contributions
VN performed all the laboratory work, assembled the expression data and drafted the manuscript. MH, AL and ML carried out the statistical analysis, TransCount estimations and wrote the statistical part of the manuscript. OM was in charge of the microarray production. EH conceived the study, guided the practical work, data analyses and editing of manuscript.
Acknowledgements
Vigdis Nygaard is a Research Fellow of the Norwegian Cancer Society. The Norwegian Microarray Consortium provided financial support for the collaboration with the Norwegian Computing Center.
==== Refs
Assersohn L Gangi L Zhao Y Dowsett M Simon R Powles TJ Liu ET The Feasibility of Using Fine Needle Aspiration from Primary Breast Cancers for cDNA Microarray Analyses Clin Cancer Res 2002 8 794 801 11895911
Van Gelder RN von Zastrow ME Yool A Dement WC Barchas JD Eberwine JH Amplified RNA synthesized from limited quantities of heterogeneous cDNA Proc Natl Acad Sci U S A 1990 87 1663 1667 1689846
Wang E Miller LD Ohnmacht GA Liu ET Marincola FM High-fidelity mRNA amplification for gene profiling Nat Biotechnol 2000 18 457 459 10748532 10.1038/74546
Baugh LR Hill AA Brown EL Hunter CP Quantitative analysis of mRNA amplification by in vitro transcription Nucleic Acids Res 2001 29 E29 11222780 10.1093/nar/29.5.e29
Iscove NN Barbara M Gu M Gibson M Modi C Winegarden N Representation is faithfully preserved in global cDNA amplified exponentially from sub-picogram quantities of mRNA Nat Biotechnol 2002 12172558
Puskas LG Zvara A Hackler LJ Van Hummelen P RNA amplification results in reproducible microarray data with slight ratio bias Biotechniques 2002 32 1330 4, 1336, 1338, 1340 12074164
Jeffrey SS Fero MJ Borresen-Dale AL Botstein D Expression array technology in the diagnosis and treatment of breast cancer Mol Interv 2002 2 101 109 14993355 10.1124/mi.2.2.101
Moll PR Duschl J Richter K Optimized RNA amplification using T7-RNA-polymerase based in vitro transcription Anal Biochem 2004 334 164 174 15464965 10.1016/j.ab.2004.07.013
Naderi A Ahmed AA Barbosa-Morais NL Aparicio S Brenton JD Caldas C Expression microarray reproducibility is improved by optimising purification steps in RNA amplification and labelling BMC Genomics 2004 5 9 15005798 10.1186/1471-2164-5-9
Endege WO Steinmann KE Boardman LA Thibodeau SN Schlegel R Representative cDNA libraries and their utility in gene expression profiling Biotechniques 1999 26 542 8, 550 10090998
Che S Ginsberg SD Amplification of RNA transcripts using terminal continuation Lab Invest 2004 84 131 137 14647400 10.1038/sj.labinvest.3700005
Nygaard V Loland A Holden M Langaas M Rue H Liu F Myklebost O Fodstad O Hovig E Smith-Sorensen B Effects of mRNA amplification on gene expression ratios in cDNA experiments estimated by analysis of variance BMC Genomics 2003 4 11 12659661 10.1186/1471-2164-4-11
Goley EM Anderson SJ Menard C Chuang E Lu X Tofilon PJ Camphausen K Microarray analysis in clinical oncology: pre-clinical optimization using needle core biopsies from xenograft tumors BMC Cancer 2004 4 20 15151703 10.1186/1471-2407-4-20
Ohyama H Zhang X Kohno Y Alevizos I Posner M Wong DT Todd R Laser capture microdissection-generated target sample for high-density oligonucleotide array hybridization Biotechniques 2000 29 530 536 10997267
Klur S Toy K Williams MP Certa U Evaluation of procedures for amplification of small-size samples for hybridization on microarrays Genomics 2004 83 508 517 14962677 10.1016/j.ygeno.2003.09.005
Mohr S Bottin MC Lannes B Neuville A Bellocq JP Keith G Rihn BH Microdissection, mRNA amplification and microarray: a study of pleural mesothelial and malignant mesothelioma cells Biochimie 2004 86 13 19 14987796 10.1016/j.biochi.2003.11.008
Luzzi V Holtschlag V Watson MA Expression profiling of ductal carcinoma in situ by laser capture microdissection and high-density oligonucleotide arrays Am J Pathol 2001 158 2005 2010 11395378
Theilgaard-Monch K Cowland J Borregaard N Profiling of gene expression in individual hematopoietic cells by global mRNA amplification and slot blot analysis J Immunol Methods 2001 252 175 189 11334978 10.1016/S0022-1759(01)00340-4
Frigessi A van de Wiel MA Holden M Svendsrud DH Glad IK Lyng H Genome-wide estimation of transcript concentrations from spotted cDNA microarray data. Nucleic Acids Res 2005 16204447
Smyth GK Linear models and emperical bayes methods for assessing differential expression in microarray experiments Statistical Applications in Genetics and Molecular Biology 2004 3 Article 3 16646809
Scherer A Krause A Walker JR Sutton SE Seron D Raulf F Cooke MP Optimized protocol for linear RNA amplification and application to gene expression profiling of human renal biopsies Biotechniques 2003 34 546 50, 552-4, 556 12661160
Spiess AN Mueller N Ivell R Amplified RNA degradation in T7-amplification methods results in biased microarray hybridizations BMC Genomics 2003 4 44 14606961 10.1186/1471-2164-4-44
Rudnicki M Eder S Schratzberger G Mayer B Meyer TW Tonko M Mayer G Reliability of t7-based mRNA linear amplification validated by gene expression analysis of human kidney cells using cDNA microarrays Nephron Exp Nephrol 2004 97 e86 95 15292679 10.1159/000078642
Kenzelmann M Klaren R Hergenhahn M Bonrouhi M Grone HJ Schmid W Schutz G High-accuracy amplification of nanogram total RNA amounts for gene profiling Genomics 2004 83 550 558 15028277 10.1016/j.ygeno.2003.09.026
Li Y Li T Liu S Qiu M Han Z Jiang Z Li R Ying K Xie Y Mao Y Systematic comparison of the fidelity of aRNA, mRNA and T-RNA on gene expression profiling using cDNA microarray J Biotechnol 2004 107 19 28 14687968 10.1016/j.jbiotec.2003.09.008
Marko NF Frank B Quackenbush J Lee NH A robust method for the amplification of RNA in the sense orientation BMC Genomics 2005 6 27 15740627 10.1186/1471-2164-6-27
Patel OV Suchyta SP Sipkovsky SS Yao J Ireland JJ Coussens PM Smith GW Validation and application of a high fidelity mRNA linear amplification procedure for profiling gene expression Vet Immunol Immunopathol 2005 105 331 342 15808310 10.1016/j.vetimm.2005.02.018
Gold D Coombes K Medhane D Ramaswamy A Ju Z Strong L Koo JS Kapoor M A comparative analysis of data generated using two different target preparation methods for hybridization to high-density oligonucleotide microarrays BMC Genomics 2004 5 2 14709180 10.1186/1471-2164-5-2
Hu L Wang J Baggerly K Wang H Fuller GN Hamilton SR Coombes KR Zhang W Obtaining reliable information from minute amounts of RNA using cDNA microarrays BMC Genomics 2002 3 16 12086591 10.1186/1471-2164-3-16
Li L Roden J Shapiro BE Wold BJ Bhatia S Forman SJ Bhatia R Reproducibility, fidelity, and discriminant validity of mRNA amplification for microarray analysis from primary hematopoietic cells J Mol Diagn 2005 7 48 56 15681474
Schlingemann J Thuerigen O Ittrich C Toedt G Kramer H Hahn M Lichter P Effective transcriptome amplification for expression profiling on sense-oriented oligonucleotide microarrays Nucleic Acids Res 2005 33 e29 15718295 10.1093/nar/gni029
Alberts B Bray D Lewis J Raff M Roberts K Watson JD Robertson M Molecular Biology of the Cell 1994 3rd New York, Garland Publishing, Inc.
Velculescu VE Madden SL Zhang L Lash AE Yu J Rago C Lal A Wang CJ Beaudry GA Ciriello KM Cook BP Dufault MR Ferguson AT Gao Y He TC Hermeking H Hiraldo SK Hwang PM Lopez MA Luderer HF Mathews B Petroziello JM Polyak K Zawel L Zhang W Zhang X Zhou W Haluska FG Jen J Sukumar S Landes GM Riggins GJ Vogelstein B Kinzler KW Analysis of human transcriptomes Nat Genet 1999 23 387 388 10581018 10.1038/70487
Scheidl SJ Nilsson S Kalen M Hellstrom M Takemoto M Hakansson J Lindahl P mRNA expression profiling of laser microbeam microdissected cells from slender embryonic structures Am J Pathol 2002 160 801 813 11891179
Stenman J Orpana A Accuracy in amplification Nat Biotechnol 2001 19 1011 1012 11689839 10.1038/nbt1101-1011b
Stenman J Lintula S Rissanen O Finne P Hedstrom J Palotie A Orpana A Quantitative detection of low-copy-number mRNAs differing at single nucleotide positions Biotechniques 2003 34 172 177 12545556
Karrer EE Lincoln JE Hogenhout S Bennett AB Bostock RM Martineau B Lucas WJ Gilchrist DG Alexander D In situ isolation of mRNA from individual plant cells: creation of cell-specific cDNA libraries Proc Natl Acad Sci U S A 1995 92 3814 3818 7731989
Reverter A McWilliam SM Barris W Dalrymple BP A rapid method for computationally inferring transcriptome coverage and microarray sensitivity Bioinformatics 2005 21 80 89 15308544 10.1093/bioinformatics/bth472
Tietjen I Rihel JM Cao Y Koentges G Zakhary L Dulac C Single-cell transcriptional analysis of neuronal progenitors Neuron 2003 38 161 175 12718852 10.1016/S0896-6273(03)00229-0
Korn K Gardellin P Liao B Amacker M Bergstrom A Bjorkman H Camacho A Dorhofer S Dorre K Enstrom J Ericson T Favez T Gosch M Honegger A Jaccoud S Lapczyna M Litborn E Thyberg P Winter H Rigler R Gene expression analysis using single molecule detection Nucleic Acids Res 2003 31 e89 12907741 10.1093/nar/gng089
Willey JC Crawford EL Jackson CM Weaver DA Hoban JC Khuder SA DeMuth JP Expression measurement of many genes simultaneously by quantitative RT-PCR using standardized mixtures of competitive templates Am J Respir Cell Mol Biol 1998 19 6 17 9651175
Mouritzen P Nielsen PS Jacobsen N Noerholm M Lomholt C Pfundheller HM Ramsig NB Kauppinen S Tolstrup N The ProbeLibraryTM - Expression profiling 99% of all human genes using only 90 dual-labeled real-time PCR probes Biotechniques 2004 37 492 495
Velculescu VE Zhang L Vogelstein B Kinzler KW Serial analysis of gene expression Science 1995 270 484 487 7570003
Brenner S Johnson M Bridgham J Golda G Lloyd DH Johnson D Luo S McCurdy S Foy M Ewan M Roth R George D Eletr S Albrecht G Vermaas E Williams SR Moon K Burcham T Pallas M DuBridge RB Kirchner J Fearon K Mao J Corcoran K Gene expression analysis by massively parallel signature sequencing (MPSS) on microbead arrays Nat Biotechnol 2000 18 630 634 10835600 10.1038/76469
Elowitz MB Levine AJ Siggia ED Swain PS Stochastic gene expression in a single cell Science 2002 297 1183 1186 12183631 10.1126/science.1070919
The Norwegian Radium Hospital Microarray Core Facility
Yang YH Dudoit S Luu P Lin DM Peng V Ngai J Speed TP Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation Nucleic Acids Res 2002 30 e15 11842121 10.1093/nar/30.4.e15
Langaas M Lindqvist BH Ferkingstad E Estimating the portion of true null hypotheses, with application to DNA microarray data J Royal Statistical Soc Series B 2005 67 555 572 10.1111/j.1467-9868.2005.00515.x
Benjamini Y Hochberg Y Controlling the false discovery rate: a practical and powerful approach to multiple testing. JRSS 1995 57 289 300
Gentleman RC Carey VJ Bates DM Bolstad B Dettling M Dudoit S Ellis B Gautier L Ge Y Gentry J Hornik K Hothorn T Huber W Iacus S Irizarry R Leisch F Li C Maechler M Rossini AJ Sawitzki G Smith C Smyth G Tierney L Yang JY Zhang J Bioconductor: open software development for computational biology and bioinformatics Genome Biol 2004 5 R80 15461798 10.1186/gb-2004-5-10-r80
Kerr MK Martin M Churchill GA Analysis of variance for gene expression microarray data J Comput Biol 2000 7 819 837 11382364 10.1089/10665270050514954
Wolfinger RD Gibson G Wolfinger ED Bennett L Hamadeh H Bushel P Afshari C Paules RS Assessing gene significance from cDNA microarray expression data via mixed models J Comput Biol 2001 8 625 637 11747616 10.1089/106652701753307520
Jin W Riley RM Wolfinger RD White KP Passador-Gurgel G Gibson G The contributions of sex, genotype and age to transcriptional variance in Drosophila melanogaster Nat Genet 2001 29 389 395 11726925 10.1038/ng766
Follestad T Langaas M Rue H Holden M Løland A glme: a C-program for parameter estimation using Gibbs-sampling in large linear mixd-effects models, with applications to DNA microarray data. 2004 Oslo, Norwegian Computing Center 1 20
|
16253144
|
PMC1310617
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Oct 27; 6:147
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-147
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1171628196910.1186/1471-2458-5-117DebateLithuanian health care in transitional state: ethical problems Jakušovaitė Irayda [email protected] Žilvinas [email protected]Žekas Romualdas [email protected] Department of Philosophy and Social Sciences, Kaunas University of Medicine, Mickeviciaus 9, 44307 Kaunas, LT, Lithuania2 National Board of Health of Lithuania, Gedimino 53, Vilnius, LT, Lithuania2005 9 11 2005 5 117 117 27 12 2004 9 11 2005 Copyright © 2005 Jakušovaitė et al; licensee BioMed Central Ltd.2005Jakušovaitė et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Throughout the economic and political reforms in post-communist countries, significant changes have also occurred in public morality. One of the tasks of the Lithuanian health policy is to create mechanisms for strengthening the significance of ethical considerations in the decision-making processes concerning health care of individuals and groups of individuals, as well as considering the positions of physicians and the health care system itself in a general way. Thus, health care ethics could be analyzed at two levels: the micro level (the ethics of doctor-patient relationships) and the macro level (the ethics of health policy-making, which can be realized by applying the principles of equal access, reasonable quality, affordable care and shared responsibilities). To date, the first level remains dominant, but the need arises for our attention to refocus now from the micro level to the patterns of managing and delivering care, managing the health care resources, and conducting business practices.
Discussion
In attempting to increase the efficiency of health services in Lithuania, a common strategy has been in place for the last fifteen years. Decentralization and privatization have been implemented as part of its policy to achieve greater efficiency. Although decentralization in theory is supposed to improve efficiency, in practice the reform of decentralization has still to be completely implemented in Lithuania. Debates on health policy in Lithuania also include the issue of private versus public health care. Although the approach of private health care is changing in a positive way, it is obvious that reduced access to health services is the most vulnerable aspect. In the Lithuanian Health Program adopted in July 1998, the target of equity was stressed, stating that by 2010, differences in health and health care between various socio-economic groups should be reduced by 25%.
Summary
The restructuring of health care system in Lithuania should be based on a balance between decentralization and centralization, and between public and private health care sectors. Successful transition requires a balanced role of the government. Today it is obvious in Lithuania that continuous encouragement to make sacrifices was not enough to induce the system to function well, and in an ethical manner.
==== Body
Background
In undertaking any health care reform, it is important to clarify the ethical problems they may engender [1-4]. Lithuania as well as Eastern and Central European countries in transition have postulated the importance of ethical considerations in changes in the health sector. In Lithuania, ethics has traditionally focused on the problems of the right and wrong conduct of individuals [5,6]. Transition to a more technological world confronts us with a new kind of human behavior, namely the management of collective systems. Today institutions, not just individuals, practice health care. Thus, we need to reorient health care ethics towards the issues of organizational activity and management instead of emphasizing the traditional discussions and moralizing. Behavioral changes must occur, but this can only be achieved within the framework of institutional and political support to ensure equal access, reasonable quality, and affordable care and shared responsibility in health care. It is a paradox that patients and physicians have to defend their rights against the advent of very complex health care systems, the aim of which should be to help them. Thus, we need to speak about the moral problems of health care system in the context of individual (micro) and social (macro) ethics [7]. Only synchronous functioning of these levels can ensure an ethical solution of the problems. When redistributing the gross public product, healthcare in Lithuania receives 3,6% of the gross national product. Despite this – and according to a summary of the achievements of the system – the effectiveness of the health care system of Lithuania occupies 52nd position in the global rating out of 191 countries. However, our country is only in 131st–133rd position with regard to the equity of financing of the system [8]. Such dissonance may not remain for long, since deficit in the financing sector will result in a crisis of human resources, lack of infrastructure and progress in the field of medicine, and will retard the reduction of health inequalities in Lithuania.
After the restoration of its independence in 1989, Lithuania inherited a centralized system that mainly delivered inefficient health care management and resource allocation. It opted for restructuring and decentralization as strategies that would to increase the efficiency of our health services. Decentralization of the health care system was achieved by segregating primary health care (family physicians), secondary health care (physicians – specialists), and tertiary health care levels (high specialization university clinics). The development and reformation of primary health care was seen as a key factor in the entire health care reform [9]. However, the experience of other countries shows that in practice, decentralization does not necessarily enhance efficiency. According to Reich [10], first "a local community still has a range of preferences within it, so that problems of allocative efficiency remain. Second, local government can still be secretive and unrepresentative, so that voter demands are not reflected in local bureaucratic decisions. Third, decentralized government can still be highly bureaucratic government, with high degrees of technical inefficiency, especially if local units have limited managerial capacity and expertise. Fourth, the small scale of decentralized units can produce a loss in economies of scale. In short, decentralization in practice creates multiple problems that may overwhelm any efficiency gains expected in theory". All these problems are obvious in Lithuania.
Many expectations are associated with the emergence of the private health care sector in Lithuania. The appearance of the private sector has raised many debates. Although the approach of private health care is changing in a positive way, it is obvious that in the private health care decreased access to health services is the most vulnerable, as the experiences of other countries have shown [11-14]. We agree with Segall's statement [15] that "privatization in health care does not lend itself to a quick technical fix. It is a complex process, essentially political and ethical in character, and involves the interplay of a number of considerations among which are not only those of equity and efficiency".
The aim of this article is to define the main social ethical problems in Lithuanian health care system in transition and to discuss the possible solutions to these problems in the context of the experiences of other countries. This has involved analysis of original research papers and official documents of Lithuanian health care system combined with the authors' personal opinion.
Discussion
All countries of Central and Eastern Europe have now expressed their wish for a total change in their health care systems. Changes in these countries include: the introduction of market economy mechanisms in health care, an increased focus on population needs in guiding health care systems, and the possibility of introducing a more general type of care at the primary level [16]. Ethical issues should be taken into account during the process of the implementation and evaluation of these changes in the healthcare system.
Balance between decentralized and centralized management in health care
An attempt to increase efficiency of our health services has been a common goal over the last fifteen years in Lithuania. Decentralization was one of the strategies to achieve it. The Lithuanian National Board of Health encourages the decentralization and privatization of the institutions of primary health care and the centralization of the institutions of secondary health care.
From 1994 to 1995, one of the political decisions was to devolve health care services by shifting administration from the Ministry of Health to the ten counties. At the same time, funding responsibilities were moved from the Lithuanian Ministry of Health to the State Sickness Fund. The counties are in charge of the enforcement of the state health programs in their respective regions. The municipalities are responsible for providing primary health care to their local population and are engaged in running small and medium size hospitals within their localities. Moreover, municipalities have a wide range of responsibilities in the implementation of local health programs and the improvement of public health activities. Decentralization of management transfers responsibility to where the work is actually done, allowing for the search for optimal solutions for the achievement of the results of health promotion in real conditions. In essence, this would mean the development of a new managerial model that would clearly define the rights, responsibilities and accountability of the participants in the health promotion process. Despite increasing cooperation between administrators, providers and consumers, the last remains the weakest in the bargaining process. Municipalities and county administrations do not have enough capacity to plan the services under their responsibility and appear to lack the authority to enforce their decisions. Municipalities responsible for maintaining and developing their respective health care infrastructure do not allocate sufficient resources for this purpose. A serious obstacle to health care reform in Lithuania is a continuing lack of managerial skills and low interest in the application of professional expertise in decision-making, [17]. In addition, municipalities that make decisions related to the development of primary healthcare services impede private capital investment. Municipal Councils that make such decisions demonstrate significant lobbying of state institutions. Lithuanian Law of Health Care Institutions [18] allows permitting or depriving the establishment of private health care institutions. Thus, there is an obvious gap between the aims of the healthcare reform policy and the expectations of the patients.
The situation of centralization in health care is much the same. Although the centralization of the institutions of secondary health care in Lithuania was successful in improving the quality of services, many issues remain unsolved. The annual report of the National Health Council states that the network of stationary healthcare institutions is excessive and irrational. Due to the lack of medical technologies and human resources, healthcare institutions of a lower level cannot ensure quality of healthcare. At present, certain hospitals perform one or two complex operations or procedures per year. In the absence of a sufficient number of operations or procedures, discussions about the quality of service become complex. The report indicates that Lithuania should adopt the model that proved to be optimal in the world and Europe, i.e. merging of hospitals rather than closing them. This would preserve human resources, modern technologies, and the experience of the merged hospitals. The main issue of the opponents of this model is with the assurance of the accessibility of healthcare services in a geographical/territorial sense [19].
It is very difficult to explain to people in the regions that although they have mandatory health insurance, they cannot sometimes get one or other of the services they had before. This situation raises social problems as well; not every patient can afford paying for the trip to the needed specialist who works in a higher-level institution in the district center or some big city. But the problem is more complex. There are over 29,990 hospital beds in Lithuania. Comparing hospital cases per 1000 inhabitants with the same in Sweden, it is becoming clear that in Sweden, 3600 hospital beds would be enough for the same treatment [20]. This situation exists because a lot of patients are treated in Lithuanian hospitals, while in Sweden and other countries many more would be treated in out patient departments (LOR, eye, nervous, many mental and other illnesses). In addition, our secondary health service is highly institutionalized in Lithuania; there is no out-patient care or nursing [21]. This model of treatment raises the threat of ageism (disturbance of human dignity, the "slippery slope" tendency).
From management to ethics
Improvement of management is one of the preconditions for tackling ethical problems in health care. The improvement of administration/management would ensure better satisfaction of the needs of patients for the same expenditure. However, the actual administration of health care institutions does not meet the patients' needs. According to the data of different surveys, both physicians and patients unanimously agree that the existing administration of health care institutions is inappropriate – >20% of respondents claimed they were unhappy with the quality of treatment, and ~30% of respondents were not satisfied with the attention paid and the quality of treatment in public health care institutions, while <5% of the respondents indicated difficulty in registration, bad quality of services and not insufficient attention paid in private health care. About 60% of patients were not satisfied with primary health care administration, mainly stressing the appointment system as not appropriate, rudeness of the employees at the reception desk, insufficient time for patient's examination, and a lack of effective distribution of functions. Physicians, on their part, admit that usually health care institutions have no "team" with clearly distributed responsibilities. Thus, a physician at the same time fills the papers and works as a psychologist, social worker, and service provider. The majority of physicians and patients have pointed out the shortage of social workers and their insufficient qualification [22]. The results of the qualitative study helped to identify the main reasons for patients' dissatisfaction, which were have divided into three levels:
1. the system level (dissatisfaction with the health care reform, bureaucracy, difficulty in getting the specialist, and high cost of services),
2. the institutional level (deficiencies in provision and quality of service, long queues, waiting, lack of medical equipment, and inadequate quality of the health care service),
3. the individual level (deficiencies in physicians' attitudes and skills and work, lack of attention, information, and responsibility, and negligence and rudeness).
These three groups indicate the level of responsibility for issues to be addressed. So, we need to speak not only about the problems of individual ethics, but also about moral problems of health care systems and organizational ethics [23]. Changes in the management of primary health care increase patient satisfaction with the doctor. The results of the population survey performed in Lithuania revealed that 69,9% of the respondents were satisfied with their primary care doctor [24]. A similar situation exists in Estonia, where 68% of patients were quite satisfied or very satisfied with their general practitioner or primary health care doctor [25]. Qualitative decision-making in primary health care should help to avoid many ethical conflicts.
Reimbursement of medicines – the right to quality medications
Another problem pointed out by both physicians and patients is the quota of reimbursed medicines and services. Until 1990, the entire pharmaceutical sector was state-owned. Medicines were subsidized by the state. In 1991, Lithuania decided to harmonize its standards with those of Western Europe, which favored the opening of the Lithuanian market to more expensive drugs produced in the European Union. At the same time, it has prohibited cheaper imports from the former USSR and other countries, as these did not meet European GMP standards. Because of the strong lobbyism of rich manufacturers of brand-name medicines, it is difficult for generic medicines to take a stronger position within the market.
According to the data of the State Sickness Fund, the expenditures of Lithuanian population for medicines in 2003 were ~65%, while the expenditure of the State Sickness Fund for reimbursed medicines was 35% [19]. Drugs are delivered free to the in-patient sector, but the reimbursement system for the drugs prescribed in the out-patient sector is complicated. Pharmaceutical companies, when operating through chief specialists in certain fields (e.g. chief pulmonologists, chief cardiologists, etc.), as well as through ordinary physicians, influence the patients' attitude, orientating towards treatment with the newest and the most expensive medicines. The state, on its part, having limited resources, attempts to regulate this process (through the procedure of reimbursement for the medicines, through the introduction of drug prescription quotas for physicians and healthcare institutions, etc.), so that the financial possibilities are not exceeded. Proposals were made for a more frequent use of generic medicines, which are cheaper. This brings about significant ethical conflicts between the Ministry of Health and the pharmaceutical companies, as well as between physicians and patients. Physicians are now in a paradoxical situation where having more patients and prescribing more subsidized medicines means the institution might sink into debt. Physicians working in primary health care institutions arrive at a paradoxical conclusion: the best physician is the one whose patients are healthy and rarely visit the institution; therefore they do not require subsidized medicines and expensive examinations. It is not in the interests of the health care institution to have the patients visit the institution, since the less work they do, the more money is allocated to the institution [22].
With its relatively small market of medicines and underdeveloped system of reimbursement, Lithuania can hardly expect the lowest prices of all reimbursed medicines compared with other countries of the EU. An increase in the population's extra pay for reimbursed medicines results in a decrease in the accessibility of new and effective medications [19]. This is an important problem of social ethics.
There has been a lack of surveys in Lithuania needed to estimate for common drugs usage. Every country has its own specificity. The comparative study performed among Estonian and Finnish general practitioners to evaluate the need for common drugs showed that different therapeutic traditions influence the list of essential medications [26].
Private versus public health care
The period of 1993–1994 was marked with public debate on the issues of private versus public health care institutions in Lithuania. The founders of the institutions that belong to the Lithuanian National Health System (LNHS) are the State, the Ministry of Health, counties, municipalities, and private persons (independent contractors). All these institutions receive financing from the Sickness Funds. There are healthcare institutions whose founders are private persons who have not signed contracts with the Sickness Funds, and thus the services of such institutions are fully paid by the patients. In Lithuania, there are few private healthcare institutions that do not belong to the LNHS. The number of rich people in Lithuania is also low, and people, having paid for obligatory health insurance are unwilling to pay for the services a second time. Private healthcare institutions that belong to the LNHS function much more effectively.
The approach to private health care is changing. According to a survey conducted in October 1995, <25% of health professionals and politicians stressed the private health provision and the market in general as being of future importance in planning and resource allocation in the Lithuanian health sector [19]. Since 1996, the possibility of choosing the general practitioner voluntarily for the society members was provided. When the law of health insurance was submitted, their conditions to get the financial support from the Sickness Founds and to establish the private institutions of family doctors have appeared, although the beginning was rather slow. The European Union PHARE project that took place in 1999–2000 has given a more prompt impulse to the reform. In 1999 and 2000, the National Health Board of Lithuania accepted the resolutions, which claimed the agreement to decentralize and privatize the institutions of primary health care [27]. Unfortunately, the rate of decentralization and the establishment of private general practitioners (GP) have decreased. The heads of public health care institutions had not felt the competition of private GPs at first. Only when the mostly active and perspective doctors have moved to the private institutions taking patients "with them" did the budgets of institutions decline significantly, and there were statements that the whole health care system was being damaged. The appearance of the elements of a market in health care delivery was considered to be chaotic. Municipalities have the right to grant or to refuse the role of the founder of service provider. All service providers are being distributed into "homeys" and "strangers". The boards of municipalities are tending to set the priorities for public institutions and will never let private institutions be established in the more settled areas, where the activity of the institution would be rather profitable (cost-effective). The main reason, they claim, is that there is a state-owned institution in this area already. In this way the municipality plays the main role of market regulation. This was found to have very negative consequences for the coming of private capital into the health sector. It also reduces private initiatives and enterprise, limits citizens' choices of health care providers, although this is based on – and ensured by – certain laws. According to representative survey data, the reason given is that 87,7% of Lithuanians use public health care services, only 12,3% used state health care services, and 12,3% visited private health care clinics or hospitals that had no agreements with the Sickness Funds, requiring full coverage for their services from the patients [24]. However, in some municipalities that have signed agreements with the Sickness Funds, 40% and even more of the population chose private providers of healthcare services. According to the data of the State Sickness Fund, the number of primary healthcare institutions increased from 306 in 2002 to 322 in 2003. The number of state-owned institutions remained unchanged, while the number of private institutions during this period increased from 108 to 124 [19]. Most people who choose the private health care sector have considerably greater incomes comparing to those who choose public health care sector. Indeed, there is nothing unethical about allowing or even encouraging people who can afford to buy their own health care, especially if they continue to pay into the public fund. It is unethical to restrict the person's free choice to private health care. As the experience of other countries shows, "health is in an area of ubiquitous market failure. In the rush to harness efficiency gains from the market forces, policy-makers in developing and transition economics should not blindly apply marketization to the health sector. Successful transition requires a balanced governmental role" [14].
The state strategy of Lithuania for reducing health care inequalities
In many European countries, the principle of equity has governed many health services and policy decisions. Equity is an ethical concept, which means that health care resources are allocated according to needs, health services are received according to the needs, and payment for health services is made according to the ability to pay. It implies access, quality, and acceptability in health services for all [28,29]. Equity in health care cannot mean total equality of health, but it can mean the reduction and, ultimately, elimination of avoidable inequalities in health. The experience of developed countries shows that the highest level of health is seen not in the richest countries (societies), but rather in those where the difference of family income between the rich and the poor is the smallest.
The first report on health care inequalities in Lithuania was published in 1998. Data from the mortality register indicates that education, socioeconomic group, and marital status were significant predicting factors in health inequality. Higher level of education, higher income, and urban places of residence gave strong positively correlation with self-reported health status and better health, especially with regard to smoking and alcohol consumption. Large inequalities in neonatal health according to the mother's level of education and marital status were discovered. Finally, socio-economic inequalities were found as having an influence on health care accessibility, with lower socioeconomic status predicating decreased access to the services. Large differences were observed in relation to socio-economic status in other Baltic states. In Lithuania, we face the same problem as in Estonia when the reform of the primary health care was recently evaluated in terms of efficiency, but not in terms of equity [30]. Thus, one of the objectives of the Lithuanian Health Program adopted in July 1998 was to ensure equity in health and health care. The Lithuanian Health Program contains the particular target of equity, which states that by the year 2010, the differences in health and health care between various socioeconomic population groups should be reduced by 25%. Parliament adopted a resolution stipulating that the actions should focus on ensuring equal rights of access to health for all by decreasing health inequalities among rural and urban populations, populations with different education and income levels, and between age groups by active cooperation of the State, local self-government institutions and non-governmental organizations. The following table presents the indicators of socioeconomic inequalities in health and health frequency of monitoring sources [31].
The problem of illegal (informal) payments
When talking about the topical problems of health care ethics and patients' rights, one cannot avoid the problem of illegal payments, or so-called "under-the-counter" payments. The concept of these payments becomes more problematic. The participants in discussions used such terms as "additional payments", "gratuities", and "gifts". According to the Lithuanian Civil Code, payment that does not exceed the minimum standard of living is not considered to be a bribe. It appears that people perceive a bribe to be payment to a physician in cash, while cakes, champagne, sweets or coffee (the most popular forms of gratuity) are not considered bribes. The non-civic practices thriving in health care ("knowing the right people" and bribery) should be replaced by behavior models based on normative acts.
According to the quantitative sociological research "Patients' rights in contemporary health care" performed in 2002 by the Bioethics Society and the Center of Civil Initiative, many physicians think that the reasons that make people give bribes to physicians are of a more psychological nature [21]. Since neither the physicians nor the patients indicated a case where the physician tells the patient the amount of money that has to be paid prior to surgery or some other service, or where patient does not receive the necessary assistance because he or she did not pay the physician personally, it is considered that people usually offer bribes because this makes them to feel more secure. However, the results of the survey performed in Lithuania in 2005 showed that informal payment can lead to better access to higher quality health care [32].
It is obvious that patients do not receive sufficient information about the changes in or the provision of the health care system. Patients do not know under what conditions health care services are free, and for which ones they have to pay. They fear that they will have to pay additionally for certain services, and are often in a quandary over whether they should offer a bribe. There is also a lack of transparency at the highest levels, where the money of mandatory health insurance is distributed. One can hardly find a person who would know how the money from the taxes (including health care fees) is distributed; in other words, people do not know what they can get for the money they have paid in taxes. Corruption manifests itself in the sphere of public purchase and privatization. Cases of gratuities to committee members who can influence contest results are also frequent. It is a public secret that sometimes pharmacy companies give bribes when attempting to have a specific medicine registered or included in the list of subsidized medicines, or when trying to attract the physicians to prescribe a specific medicine.
Is it possible to defeat corruption in the health care system? Corruption is a well-known by-product of governments in rich, poor and so-called developed countries [10]. In Lithuania, the State Sickness Fund follows the National anti-corruption rules and uses anti-corruption measures in health care. We believe that in order to defeat corruption, it is not enough to strengthen the control mechanisms. It is even more important to reduce the incentives to give or take the bribes. While it is impossible to eliminate corruption altogether, it is necessary to take all measures to limit its extent. The experience of other countries shows that formalization of some unofficial fees with careful monitoring of their impact may help to fight corruption. The main strategies within this area should be the full recognition (development of the system?) of patients' rights, simple procedures for complaints, transparent contracts of employment, and targeting those facilities in which collusion among professionals leads to "networked" health care fees [33].
It is important to emphasize that corruption is first of all an economic problem. As the countries of the former Soviet Union and Eastern Europe have made the transition from state control of all aspects of economic life, the evolution of health care markets has been at best erratic, and the need for a better managed transition has become apparent. The existence of informal payments is prima facie evidence that publicly set prices are insufficient to induce supply, and that threats of sanctions against providers who do not offer services at these prices are insufficient. The appropriate response for governments of course is to set producer prices, but in times of budgetary squeeze and excess capacity, this is infeasible. Without rationalization of the wholesale side of supply, informal payments will continue to play a major role in resource allocation, and negative effects on equity and access will continue [34]. In Lithuania, only the first steps are being made towards the management of economic measures.
Summary
Ethical considerations should be taken into account while tackling the problems of health care reform. Over the last 30 years, Lithuanian health care institutions have focused primarily on individuals, paying insufficient attention to the problems of social ethics. The latter refocuses our attention from the micro level to the patterns of management and provision of health care. Balance in the restructuring of the health care system between decentralization and centralization in Lithuania, between public and private sectors in health care, the improvement of management, and the evaluation of these changes help to solve many problems of social ethics not only in the aspect of efficiency, but in terms of equity.
Competing interests
The author(s) declare that they have no competing interests.
Table 1 Indicators of socioeconomic inequalities in health and health care, frequency of monitoring and data sources
Health and health care indicators Socioeconomic indicators Monitoring frequency Data sources
1. Mortality (by gender and age groups): all causes, cardiovascular diseases, malignant neoplasms, external causes, respiratory diseases, other causes Place of residence: urban – rural Annually Mortality register
Administrative regions Every 3 yrs
Education Every 10 yrs, based on census data
Marital status
2. Life expectancy Place of residence: urban – rural Annually
Administrative regions Every 3 yrs
3. Avoidable mortality Place of residence: urban – rural Every 5 yrs
Administrative regions
4. Low birth weight newborns and stillbirths Mothers' education Every 3 yrs Newborn register
Mothers marital status
5. Health behavior of adult population (by gender and age groups): Self-assessed health; outpatient visits per year; stress during the last month; depression; oral health; smoking; alcohol use; nutrition; physical activity; traffic safety Education Every 4 yrs Health behavior Monitoring among the adult Population
Place of residence: urban/rural Every 2 yrs
Income Every 2 yrs
6. Health behavior of schoolchildren (by gender): Self-assessed health; outpatient visits per year; stress during the last month; depression; oral health; smoking; alcohol use; drug use; nutrition; physical activity; traffic safety Family well-being index Every 2 – 4 yrs Health behavior monitoring in schoolchildren
Parents' profession
Place of residency
7. Health care accessibility Profession Every 3 yrs Surveys on representative samples of the population
Education
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Callahan D Ends and means: the goals of health care Ethical dimensions of health policy 2002 New York: Oxford University Press
Roberts MJ Reich MR Ethical analysis in public health Lancet 2002 23 1055 9 11937202 10.1016/S0140-6736(02)08097-2
Fung H Tse N Yeoh EK Health Care Reform and Societal Values J Med Philos 1999 6 638 652 10709774 10.1076/jmep.24.6.638.2550
Šimek J Špalek V Ethical, Sociological and Philosophical aspects of Health Care System Health Care Reforms in Central and Eastern Europe: Outcomes and Challenges 2000 7 13
Jakušovaitė I Darulis Ž Medicinos ir sveikatos priežiūros tikslai ir vertybės (Goals and values of medicine and health care) Medicina 2004 9 828 832
Jakušovaitė I Peičius E Vertybių kaita ir organizacinė etika Lietuvos sveikatos priežiūros sistemoje (Value changes and organizational ethics in Lithuanian health care system) Visuomenės sveikata 1998 4 18 22
Jakušovaitė I Visuomenės sveikata ir etika (Public health and ethics) Visuomenės sveikatos mokslas ir studijos: atsakas į laikmečio iššūkius (Public health and ethics Science and studies of Public Health: response to the challenges of the time) 2004 Kaunas: Kaunas Medical University Press 171 181
Health Systems: improving performance World Health Report Geneva 2003
Health Care Systems in transition: Lithuania European Observatory on Health Care Systems 2000 World Health Organization, Regional Office for Europe, Copenhagen
Reich MR Reshaping the state from above, from within, from below: implications for public health Soc Sci Med 2002 54 1669 1675 12113448 10.1016/S0277-9536(01)00334-3
Callahan D Krizova E, Simek J Medicine and the market Health care reforms in the Central and Eastern Europe: outcomes and challenges Prague 2000
Ovretveit J Beyond the private debate. The mixed economy of health Health Policy 1996 35 75 93 10157043 10.1016/0168-8510(95)00770-9
Jankauskienė D Žemgulienė J Sveikatos priežiūros paslaugų prieinamumas (Access to health care services) Medicina 1998 3 301 338
Liu Y Hsiao WC Eggleston K Equity in health and health care: the Chinese experience Soc Sci Med 1999 49 1349 1356 10509825 10.1016/S0277-9536(99)00207-5
Segall M District health systems in a neoliberal world: a review of five key policy areas Int J Health Plann Manage 2003 18 S5 S26 14661938 10.1002/hpm.719
Lember M A policy of introducing a new contract and funding system of general practice in Estonia Int J Health Plann Manage 2002 17 41 53 11963443 10.1002/hpm.651
Pirmasis reformų dešimtmetis: sveikatos priežiūros sektorius socialinių – ekonominių pokyčių kontekste (First decade of reforms: health care sector in the context of social and economical changes). Vilnius 2000
Lietuvos Respublikos Sveikatos priežiūros įstaigų įstatymo 17 straipsnio papildymo įstatymas, 2001 lapkričio mėn. 13 d., Nr. IX-593, Valstybės žinios, 2001, 97-3419. (Complement on the 17th paragraph of the Law on Health Care Institutions in the Republic of Lithuania, November 13, 2001, No. IX-593, State News, 2001, 97-3419)
2004 metų Nacionalinės sveikatos tarybos metinis pranešimas. Lietuvos sveikatos programos įgyvendinimas: pasiekimai ir problemos. Vilnius; 2005. (The Annual Report of the National Health Board of the year 2004. The Implementation of the Lithuanian Health Programme: achievements and problems). Vilnius; 2005
Hellers G Lietuvos sveikatos projekto konsultanto pranešimo stenograma. Nacionalinės sveikatos tarybos posėdis, (The stenograph of the report of the consultant of the Lithuanian health project. The session of the committee of the National Health Council). Vilnius; 2001-12-20
Hitaitė L Pagyvenusiu Kauno rajono pacientu slauga namuose. (Home nursing of the elderly of Kaunas District) Bachelor thesis 2004 Klaipėda University, Department of Health Sciences
Pacientų teisės Lietuvoje: situacijos analizė ir visuomenės aktyvumas (Patient's rights in Lithuania: analysis of situation and activity of the society). Vilnius 2002
Bankauskaitė V Saarelma O Why are people dissatisfied with medical care services in Lithuania? A qualitative study using responses to open-ended questions Int J Qual Hlth Care 2003 15 23 9 10.1093/intqhc/15.1.23
Peičius E Lithuanian citizens' participation in health care decision-making PhD thesis 2005 Kaunas University of Medicine, Department of Philosophy and Social Sciences
Polluste K Kalda R Lember M Primary health care system in transition: the patient's experience Int J Qual Hlth Care 2000 6 503 509 10.1093/intqhc/12.6.503
Lember M Pikk A Matilla K Virjo I Kermes R Isokoski M How Estonian and Finnish primary care doctors rate their need for common drugs Eur J Clin Pharmacol 1997 52 437 440 9342578 10.1007/s002280050316
Grabauskas V Lietuvos sveikatos politika XXI amžiuje III Nacionalinė sveikatos politikos konferencija (Lithuanian health policy in 21st century III Conference of a National Health Policy) Vilnius 2000 3
Braveman P Gruskin S Poverty, equity, human rights and health Bull World Health Organ 2003 81 539 45 12973647
Blas E Hearst N Health sector reform and equity – learning from evidence? Health Policy Plan 2002 17 1 4 12477735 10.1093/heapol/17.suppl_1.1
Habicht J Kunst AE Social inequalities in health care services utilisation after eight year of health care reforms: a cross-sectional study of Estonia, 1999 Social Science and Medicine 2005 60 777 787 15571895 10.1016/j.socscimed.2004.06.026
Socio-economic inequalities in health in Lithuania: indicators for monitoring and latest situation Report Vilnius 2002
Plieskis M The accessibility of the primary health care and the citizens' satisfaction with the services in the context of the development of family medicine PhD thesis 2005 Kaunas University of Medicine, Department of Social Medicine
Ensor T Informal payments for health care in transition economies Social Science and Medicine 2004 58 237 246 14604610 10.1016/S0277-9536(03)00007-8
Jack W Lewis M Falling short of expectations: public health interventions in developing and transition economies Social Science and Medicine 2004 58 223 225 14604608 10.1016/S0277-9536(03)00005-4
|
16281969
|
PMC1310618
|
CC BY
|
2021-01-04 16:28:59
|
no
|
BMC Public Health. 2005 Nov 9; 5:117
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-117
|
oa_comm
|
==== Front
BMC Infect DisBMC Infectious Diseases1471-2334BioMed Central London 1471-2334-5-1021627449010.1186/1471-2334-5-102Research ArticleCorrelation between human papillomavirus infection and bladder transitional cell carcinoma Barghi MR [email protected] A [email protected] SMM Hosseini [email protected] B [email protected] Assistant Professor Of Urology, Shohada Tajrish Hospital, Shahid Beheshti University Of Medical Sciences, Urology and Nephrology Research Center, Tehran, Iran2 Resident Of Urology, Shohada Tajrish Hospital, Shaeed Beheshti University Of Medical Sciences, Urology and Nephrology Research Center, Tehran, Iran3 Research Consultant, Assistant Professor Of Infectious Diseases And Tropical Medicine, Master of Public Health, Urology and Nephrology Research Center, Shaheed Beheshti University Of Medical Sciences, Tehran, Iran4 Cellular and Molecular Biology Research Center, Urology and Nephrology Research Center, Shaheed Beheshti University of Medical Sciences, Urology and Nephrology Research Center, Tehran, Iran5 Urology and Nephrology Research Center (UNRC), Shohada Tajrish Hospital, Shaheed Beheshti University of Medical Science, Tehran Iran2005 8 11 2005 5 102 102 23 4 2005 8 11 2005 Copyright © 2005 Barghi et al; licensee BioMed Central Ltd.2005Barghi et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
To determine the association of human papillomavirus infection (HPV) and transitional cell carcinoma (TCC).
Methods
Using polymerase chain reaction, fifty-nine bladder tissue specimens of patients with transitional cell carcinoma of bladder compared with 20 bladder samples of cases with non-neoplastic disorders.
Results
Male to female ratio was similar in the two groups (50/9 vs. 16/4, P = 0.62). Mean age was 67 ± 10.8 years and 52 ± 20.3 years in the case and control groups, respectively (P = 0.6). Of the 59 tissue specimens with diagnosis of transitional cell carcinoma, HPV DNA was detected in 21 (35.6%) samples, while it was present in only one sample (5%) in the control group (P = 0.008). HPV18 was the most common type of virus with the incidence rate of 17/21(81%).
Conclusion
HPV might play a causative role in transitional cell carcinoma of bladder in our geographic area.
==== Body
Background
The etiology of transitional cell carcinoma (TCC), which represents 90 percent of bladder malignancies, is not quite clear, while squamous cell carcinoma (5%) of the bladder is well associated with some factors like urinary stones and prolonged infections. Some genital malignancies like vulvar or cervical neoplasms are associated with previous infection with human papillomavirus (HPV). Recently, conflicting findings have been reported on association of HPV infection and TCC [1-5].
Using PCR, the overall prevalence of human papillomavirus (HPV) DNA in penile carcinoma is about 40–45%, which is similar to the detection rate of HPV-DNA in vulvar carcinoma [6]. Males may be carriers of oncogenic HPVs and male partners may markedly contribute to the risk of developing cervical cancer in their female partners [7]. HPV detection rate in cases of invasive cervical cancer can be as high as 90 % [8]. Because of vicinity of bladder to mucosal surface of urogenital tract, HPV may play a significant role in TCCs.
While Kamel et al found HPV DNA in 57% of cases of TCC [1], Tekin's study [9] and Mvula's report [10] did not support the etiologic role of HPV in bladder carcinogenesis. In the present study, we compared the rate of HPV positivity in formalin-fixed, paraffin-embedded tissues from the urinary bladder of patients with transitional cell carcinomas and non-neoplastic bladder tissue samples. To our knowledge, this is one of limited number of studies that used non-neoplastic bladder tissue specimens as controls.
Methods
The medical records and bladder tissue specimens of 59 consecutive patients with transitional cell carcinoma of bladder who underwent transurethral resection of bladder tissue from October1999 to May 2002 were reviewed and compared with those of 20 consecutive patients with non-neoplastic disorders who were served as controls. So, a total of 79 biopsy tissues of urinary bladder were re-evaluated and analyzed by polymerase chain reaction (PCR). The designing and conducting of this study was in accordance with the ethical standards of UNRC ethics committee. Table 1 demonstrates histopathological diagnosis of specimens in the control group in detail. Since there are infrequent indications for bladder tissue sampling in non-neoplastic conditions, comparing to malignant lesions, we found only 20 non-neoplastic bladder tissue specimens.
Table 1 histologic examination of tissue specimens from control group
Finding Frequency
Inflammatory tissue (non specific cystitis) 5
Granuloma formation 1
Eosinophilic cystitis 2
Squamous metaplasia 1
Fibrosis 3
Hemorrhagic cystitis 1
Severe chronic cystitis + Atypical epithelial hyperplasia 1
Transitional papilloma
Chronic cystitis 1
Glandular cystitis + squamous metaplasia 2
Fibromuscular tissue 1
Normal mucosa 1
1
Sample
Specimens were taken from paraffin-embeded bladder tissues stored in laboratory of our hospital- Tajrish Shohada medical center. All specimens were collected and transported to molecular biology research center of shaheed Beheshti University of Medical Sciences under a well-controlled condition. Biopsy specimens were digested in lyses buffer (0.33 M sucrose, 10 mM Tris bas, 5 mM MgCl2, 2% Titon X-100) at 37°C for 4 hours. Phenol-chloroform extraction and ethanol precipitation were performed. The precipitated DNA was suspended in distilled water and used for amplification.
Primers
Two primer pairs were designed of HPV L1 gene (encodes the major capsid protein): one primer pair amplify a 269-bp fragment of HPV 6, 11, 31 and 33, and another primer pair amplify a 564-bp fragment of HPV 16, 18 and 35; and HPV types were defined by PCR products restriction analysis. PCR: Amplifications were carried out in 50-μl volumes. The reaction mixture contained 1.5 mM MgCl2, 0.1 mM of each dNTP, 20 pico mol of each primer, 0.1 μg DNA, 1.25 unit of Taq DAN polymerase and 1× PCR buffer. PCR conditions are: 30 cycles of denaturing at 94°C for 30 sec., annealing temperature at 45°C for 30 sec. and extension at 72°C for 30 sec. PCR reaction was performed using automated thermal cycler machine (model personal, Eppendorf Co, Germany).
Electrophoresis
PCR products were electrophoresis on 2% agarose gel. Gels were stained by ethidium bromide and photographed by UV transilluminator at 254 nm.
Statistical analysis was performed using Mann-Whitney, chi-square and fisher's exact tests as appropriate with significance considered at P < 0.05.
Results
The study population comprised 59 cases and 20 controls whose demographic and clinical characteristics are given in table 2. There was no difference in male to female ratio in the two groups (50/9 vs. 16/4, P = 0.62). Mean age was 67 ± 10.8 (range 38 to 85) years and 52 ± 20.3 (range18 to 81) years in the case and control groups, respectively (P = 0.6).
Table 2 demographic and clinical characteristics
Case Control
Age 67 ± 10.8(38–85) 52 ± 20.3(18–81)
Sex
Male 50(84.7%) 16(80%)
Female 9(15.3%) 4(20%)
UII 3(5.1%) 0
Metastasis
Bone 2 -
Lung 2 -
Local lymph nodes 1 -
Both liver & kidney 2 -
Site of lesion in bladder
Dome 6 -
Left lateral 21 -
Right lateral 14 -
Bilateral 10 -
Trigone 1 -
Trigone & right lateral 2 -
Bladder neck 2 -
Diffuse 2 -
Ureteral orifice 1 -
Other simultaneous problems
Hypertension 7(11.8%) 2(10%)
IHD 5(8.5%) 0
CLL 0 1(5%)
CRF 0 1(5%)
DM 6(10.1%) 0
Asthma 1(1.7%) 0
Hepatitis 1(1.7%) 0
Renal Stone 3(5%) 0
Tuberculosis 1(1.7%) 0
Hyperthyroidism 2(3.4%) 1(5%)
Neurogenic tumor 0 1(5%)
Urethral stricture 0 1(5%)
Number of tumor lesions in cystoscopy
1 26(44.1%) -
2 9(15.3%) -
3 2(3.4%) -
4 3(5.1%) -
>4 19(32.2%) -
Of the 59 tissue specimens with diagnosis of transitional cell carcinoma, HPV DNA was detected in 21 (35.6%) samples. Human papillomavirus DNA was positive in only one sample (5%) in the control group (P = 0.008) that was histolologically diagnosed as severe chronic cystitis with some degrees of atypical hyperplasia, that might be considered as a pre-malignant condition.
Type specific primers also analyzed the positive samples. In the case group, sequences of HPV 6 genome were detected in 2/21 patients (9.5%), one male and 1 female; HPV 18 was found in 14/21 patients (66.7%),13 males and 1 female ; and HPV 33 sequences were detected in2/21(9.5%), 1 male and 1 female. In 3 HPV-positive male cases (14.3%) more than one type was found; 2 patients with HPV6/HPV18, and 1 witht HPV18/HPV33. Thus, HPV18 was the most common type of virus with the incidence rate of 17/21(81%). HPV18 was also found in one male patient in the control group. HPV types 11, 16, 31 were present in no tissue specimens. Table 3 shows HPV types in two groups, distinctly.
The histopathological stages and grades of tissue samples with diagnosis of transitional cell carcinoma considering infection with HPV are given in Table 4. No significant correlation existed between tumor stages and presence of HPV. Additionally, there was no significant difference between low-grade (grade 1 and 2) and high grade (grade 3) tumors with regard to HPV infection (36% Vs.33.3%, P = 0.59).
Table 3 positive samples analyzed by type specific-primers
HPV-type/Group HPV6 HPV18 HPV33 HPV6/HPV18 HPV18/HPV33
Case 2(9.5%) 14(66.7%) 2(9.5%) 2(9.5%) 1(4.8%)
Control - 1(100%) - - -
Table 4 distribution of tumor stage and grade among cases.
Stage or grade/HPV infection Stage Grade
Tis Ta T1 T2a T2b T3 T4 I II III
infected 1 (4.8%) 5 (23.8%) 12 (57.1%) 3 (14.3%) 0 0 0 8 (38.1%) 10 (47.6%) 3 (14.3%)
Non infected 0 20 (52.6%) 12 (31.6%) 5 (13.1%) 1 (2.6%) 0 0 13 (34.2%) 19 (50%) 6 (15.8%)
Smoking rate in the case group was significantly higher than control group (54.2% vs 10%, P = 0.00). In the case group, 12 out of 21 (57%) infected patients and 20 out of 38 (52.6%) non-infected were cigarette smokers, respectively (P = 0.7). Ten cases (16.9%) but no patient in the control group were addicted to opium (P = 0.05). On the other hand, four infected (18.2%) and 6 non-infected (10.5%) subjects were addicted to opium (P = 0.36).
Discussion
One of the most common malignancies, especially in developing countries, is transitional cell carcinoma of the bladder. Several chemical agents have been suspected to have a role in its development. HPV plays an etiological role for genital tumors, but the exact effect of this virus in transitional cell carcinoma of bladder is still vague. More than a decade ago, a report of HPV-16 positive bladder carcinoma in a patient with attenuated natural killer cell function and aplastic anemia was published [11]. In the same year, another paper briefed a transplant patient with bladder tumor in whom HPV 11 was detected [12]. Agliano and colleagues investigated the presence of HPV types 16 and 18 DNA in formalin-fixed, paraffin-embedded tissues from the urinary bladder (46 transitional cell carcinomas and 10 non-neoplastic normal urinary samples) of non-immune deficient cases. HPV16 and/or HPV18 genomes were detected in 23 of 46 (50%) bladder carcinomas and in none of 10 (0%) non-neoplastic urinary samples[13]. In 1995, Kamel et al. analyzed 47 bladder carcinomas for the presence of DNA-HPV subtypes 6, 11, 16, 18, 31 and 33 by nucleic acid in situ hybridization. HPV DNA was found in 27/47 (57%) bladder carcinomas, with multiple subtypes in 20 cases (76). To our knowledge, our study has one of the largest control groups of non-neoplastic bladder tissue specimens. The high prevalence of HPV infection in the present study demonstrates an association between HPV and TCC of the bladder.
Human papillomavirus seems to be related to the etiology of bladder tumor because of its high prevalence in samples obtained from bladder tumors in the present study and some previous researches. Our study demonstrated HPV-positivity rate in 35.6% of cases and 5% of controls. It seems that carcinoma development may be triggered by HPV infection. Inactivation of the tumor suppressor pRB by the human papillomavirus (HPV) oncoprotein E7 is a mechanism by which HPV promotes cell growth [14]. Human papillomavirus type 16 proteins, E6 and E7, have been shown to cause centrosome amplification and lagging chromosomes during mitosis, leading to chromosomal instability. Genomic instability is thought to be an essential part of the conversion of a normal cell to a cancer cell [15] Kamel and colleagues demonstrated concurrent HPV positivity and abnormal p53 protein accumulation in 18 out of 47 cases, 14 showing the presence of HPV subtypes 16 and/or 18 DNA[1]. In our study, HPV 18 was the most frequent type(81%) requiring more specific epidemiologic studies and experimental investigations.
HPV may have a great role in progression of TCCs toward higher stages and/or grades by inactivation of the tumor suppressors or other unknown mechanisms.
Larue et al. reported that sensitivity of detection of HPV is largely dependent on a series of technical factors such as tissue fixation, DNA preparation and amplification conditions. In their study, presence of HPV correlated with grade but not stage of the tumors. [16]. Our specimens were formalin-fixed, paraffin-embedded tissues from the urinary bladder. Using fresh or frozen tumor material will probably increase the sensitivity. The high incidence demonstrated in the present study might be higher if fresh specimens were available and used in order to overcome the probability of qualitative loss of DNA in the material.
High rate of HPV positivity in present study suggests that other sexually transmissible viruses may play some roles in development and progression of transitional cell carcinoma of the bladder. As Gazzaniga et al demonstrated, there may be a high synergism between Human Papillomaviruses type 16 and 18 (HPV 16, HPV 18), Epstein-Barr virus (EBV), cytomegalovirus (CMV) and herpes simplex virus type 2 (HSV-2) and bladder carcinogenesis[17]. In this regard, further investigation with a large number of patients sounds to be required.
On the basis of high rate of HPV positivity in TCC cases (35.6%), comparing with control group(5%), the present report supports an etiologic role of HPV in bladder carcinogenesis.
Conclusion
This quasi-experimental study is the first report from Iran on HPV infection and transitional cell carcinoma. Our findings, in spite of some previous reports, suggest that HPV may play a causative role in transitional cell carcinoma of bladder in our geographic area. Public education regarding HPV transmission and high-risk behavior may be helpful for decreasing the incidence of urogenital malignancies.
Abbreviations
HPV: Human Papilloma Virus
Competing interests
The author(s) declare that they have no competing interest.
Authors' contributions
M.R.B. and A.H. introduced the cases and performed surgical operations. S.M.M.H.M. designed the study and performed statistical analyses. B. K. performed laboratory examinations and PCR. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
This work was supported by a grant from Urology Nephrology Research Center (UNRC).
==== Refs
Kamel D Paakko P Pollanen R Vahakangas K Lehto VP Soini Y Human papillomavirus DNA and abnormal p53 expression in carcinoma of the urinary bladder APMIS 1995 103 331 338 7654357
Sano T Sakurai S Fukuda T Nakajima T Unsuccessful effort to detect human papillomavirus DNA in urinary bladder cancers by the polymerase chain reaction and in situ hybridization Pathol Int 1995 45 506 512 7551011
Aynaud O Tranbaloc P Orth G Lack of evidence for a role of human papillomaviruses in transitional cell carcinoma of the bladder J Urol 1998 159 86 89 9400443
Chang F Lipponen P Tervahauta A Syrjanen S Syrjanen K Transitional cell carcinoma of the bladder: failure to demonstrate human papillomavirus deoxyribonucleic acid by in situ hybridization and polymerase chain reaction J Urol 1994 152 1429 1433 7933176
Sinclair AL Nouri AM Oliver RT Sexton C Dalgleish AG Bladder and prostate cancer screening for human papillomavirus by polymerase chain reaction: conflicting results using different annealing temperatures Br J Biomed Sci 1993 50 350 354 8130696
Gross G Pfister H Role of human papillomavirus in penile cancer, penile intraepithelial squamous cell neoplasias and in genital warts Med Microbiol Immunol (Berl) 2003 193 35 44 12838415 10.1007/s00430-003-0181-2
Castellsague X Bosch FX Munoz N The male role in cervical cancer Salud Publica Mex 2003 45 S345 53 14746027
Park TC Kim CJ Koh YM Lee KH Yoon JH Kim JH etal Human papillomavirus genotyping by the DNA chip in the cervical neoplasia DNA Cell Biol 2004 23 119 125 15000752 10.1089/104454904322759939
Tekin MI Tuncer S Aki FT Bilen CY Aygun C Ozen H Human papillomavirus associated with bladder carcinoma? Analysis by polymerase chain reaction Int J Urol 1999 6 184 186 10226835 10.1046/j.1442-2042.1999.06435.x
Mvula M Iwasaka T Iguchi A Nakamura S Masaki Z Sugimori H Do human papillomaviruses have a role in the pathogenesis of bladder carcinoma? J Urol 1996 155 471 474 8558638 10.1097/00005392-199602000-00013
Kitamura T Yogo Y Ueki T Murakami S Aso Y Presence of human papillomavirus type 16 genome in bladder carcinoma in situ of a patient with mild immunodeficiency Cancer Res 1988 48 7207 2847865
Querci della Rovere G Oliver RT McCance DJ Castro JE Development of bladder tumour containing HPV type 11 DNA after renal transplantation Br J Urol 1988 62 36 38 2841992
Agliano AM Gradilone A Gazzaniga P Napolitano M Vercillo R Albonici L etal High frequency of human papillomavirus detection in urinary bladder cancer Urol Int 1994 53 125 129 7645137
Fan X Liu Y Chen JJ Activation of c-Myc contributes to bovine papillomavirus type 1 E7-induced cell proliferation J Biol Chem 2003 278 43163 65 12937171 10.1074/jbc.M306008200
Patel D Incassati A Wang N McCance DJ Human papillomavirus type 16 e6 and e7 cause polyploidy in human keratinocytes and up-regulation of g(2)-m-phase proteins Cancer Res 2004 64 1299 306 14973072 10.1158/0008-5472.CAN-03-2917
LaRue H Simoneau M Fradet Y Human papillomavirus in transitional cell carcinoma of the urinary bladder Clin Cancer Res 1995 1 435 40 9816001
Gazzaniga P Vercillo R Gradilone A Silvestri I Gandini O Napolitano M Prevalence of papillomavirus, Epstein-Barr virus, cytomegalovirus, and herpes simplex virus type 2 in urinary bladder cancer J Med Virol 1998 55 262 7 9661833 10.1002/(SICI)1096-9071(199808)55:4<262::AID-JMV2>3.0.CO;2-Z
|
16274490
|
PMC1310619
|
CC BY
|
2021-01-04 16:28:16
|
no
|
BMC Infect Dis. 2005 Nov 8; 5:102
|
utf-8
|
BMC Infect Dis
| 2,005 |
10.1186/1471-2334-5-102
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1411625578210.1186/1471-2407-5-141Research ArticleCEBPG transcription factor correlates with antioxidant and DNA repair genes in normal bronchial epithelial cells but not in individuals with bronchogenic carcinoma Mullins D'Anna N [email protected] Erin L [email protected] Sadik A [email protected] Dawn-Alita [email protected] Youngsook [email protected] James C [email protected] Department of Medicine, Medical University of Ohio, Room 0012 Ruppert Health Building, 3000 Arlington Avenue, Toledo, OH 43614, USA2005 29 10 2005 5 141 141 7 7 2005 29 10 2005 Copyright © 2005 Mullins et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Cigarette smoking is the primary cause of bronchogenic carcinoma (BC), yet only 10–15% of heavy smokers develop BC and it is likely that this variation in risk is, in part, genetically determined. We previously reported a set of antioxidant genes for which transcript abundance was lower in normal bronchial epithelial cells (NBEC) of BC individuals compared to non-BC individuals. In unpublished studies of the same NBEC samples, transcript abundance values for several DNA repair genes were correlated with these antioxidant genes. From these data, we hypothesized that antioxidant and DNA repair genes are co-regulated by one or more transcription factors and that inter-individual variation in expression and/or function of one or more of these transcription factors is responsible for inter-individual variation in risk for BC.
Methods
The putative transcription factor recognition sites common to six of the antioxidant genes were identified through in silico DNA sequence analysis. The transcript abundance values of these transcription factors (n = 6) and an expanded group of antioxidant and DNA repair genes (n = 16) were measured simultaneously by quantitative PCR in NBEC of 24 non-BC and 25 BC individuals.
Results
CEBPG transcription factor was significantly (p < 0.01) correlated with eight of the antioxidant or DNA repair genes in non-BC individuals but not in BC individuals. In BC individuals the correlation with CEBPG was significantly (p < 0.01) lower than that of non-BC individuals for four of the genes (XRCC1, ERCC5, GSTP1, and SOD1) and the difference was nearly significant for GPX1. The only other transcription factor correlated with any of these five target genes in non-BC individuals was E2F1. E2F1 was correlated with GSTP1 among non-BC individuals, but in contrast to CEBPG, there was no significant difference in this correlation in non-BC individuals compared to BC individuals.
Conclusion
We conclude that CEBPG is the transcription factor primarily responsible for regulating transcription of key antioxidant and DNA repair genes in non-BC individuals. Further, we conclude that the heavy smokers selected for development of BC are those who have sub-optimal regulation of antioxidant and DNA repair genes by CEBPG.
==== Body
Background
BC is currently the leading cause of cancer-related death in the United States, causing 28% of all cancer deaths [1]. Although cigarette smoking is the primary risk factor, only 10–15% of heavy smokers (greater than 20 pack years) develop BC [1-3]. Antioxidant and DNA repair enzymes that provide protection from the effects of cigarette smoke are expressed in the progenitor cells for BC, normal bronchial epithelial cells (NBEC) [1]. Inherited inter-individual variation in the function of these genes plays a role in determining risk for BC [4-6]. Antioxidant enzymes protect NBEC from reactive oxygen species produced by interaction with and metabolism of xenobiotics such as pollution and cigarette smoke [4-7] as well as those produced by normal cellular metabolism. Reactive oxygen species cause many damaging reactions including denaturation of proteins, cross-linking of lipids and proteins and modification of nucleic acid bases, which can lead to cancer [7]. DNA repair enzymes repair the frequent damage to DNA caused by oxidant stress as well as other stresses, including bulky adducts derived from carcinogens in cigarette smoke [8].
We previously reported that an interactive transcript abundance index comprising antioxidant genes was lower in NBEC of BC individuals compared to non-BC individuals, suggesting that BC individuals are selected on the basis of poor antioxidant protection [9]. In that study, there was a tendency towards correlation in transcript abundance between several pairs of antioxidant or DNA repair genes in non-BC individuals, but not in BC individuals. Gene pairs included in that observation were GSTP1/GPX1, CAT/GPX3, and GPX3/SOD1.
Correlation is one typical characteristic of co-regulated genes. Another is shared transcription factor recognition sites in the regulatory regions of those genes [10]. Based on the above findings, it was hypothesized first, that there is inter-individual variation in regulation of key antioxidant and DNA repair genes by one or more transcription factors and second, that individuals with sub-optimal regulation are selected for development of BC if they smoke cigarettes. To test these hypotheses, transcription factor recognition sites common to the regulatory regions of the above correlated gene pairs were identified through in silico DNA sequence analysis, and their transcript abundance measured simultaneously with an expanded group of ten antioxidant and six DNA repair genes.
Methods
NBEC sample procurement
Brush biopsy samples of normal bronchial epithelium were obtained for research studies at the time of diagnostic bronchoscopy according to previously described methods [9,11]. Normal bronchial epithelium in the lung not involved with cancer was brushed prior to biopsy of the suspected cancerous area. Samples were collected in a manner satisfying all requirements of the Institutional Review Board for the Medical University of Ohio. Each BC diagnosis and subtype identification was determined by histopathological examination in the Department of Pathology at the Medical University of Ohio. NBEC samples from a total of 49 individuals, including 24 non-BC individuals and 25 BC individuals, were evaluated in this study. The biographical characteristics of these individuals are presented in Table 1.
Transcript abundance measurement
Total RNA samples extracted from NBEC were reverse transcribed using M-MLV reverse transcriptase and oligo dT primers as previously described [9,11]. Standardized RT (StaRT)-PCR was used for transcript abundance measurement in these studies. With StaRT-PCR, an internal standard for each gene within a standardized mixture of internal standards (SMIS) is included in each PCR reaction. After amplification, products were electrophoresed on an Agilent 2100 Bioanalyzer using DNA Chips with DNA 1000 Kit reagents for visualization according to the manufacturer's protocol (Agilent Technologies Deutschland GmbH, Waldbronn, Germany).
The StaRT-PCR technology is licensed to Gene Express, Inc. (Toledo, OH). Many of the reagents are available commercially and were obtained through Gene Express, Inc. for this study. StaRT-PCR reagents for each of the measured genes that were not commercially available, including primers and SMIS, were prepared according to previously described methods [11,12]. Sequence information for the primers is provided in Table 2.
Including an internal standard within a SMIS in each measurement controls for all known sources of variation during PCR, including inhibitors in samples, and generates virtually-multiplexed transcript abundance data that are directly comparable across multiple experiments and institutions [13]. The performance characteristics of StaRT-PCR are superior to other forms of commercially available quantitative PCR technology in the areas critical to this study. With respect to these studies, the key property of a quantitative PCR method is not whether the PCR products are measured kinetically or at endpoint, but rather whether there are internal standards in each measurement or not. The overall performance characteristics of StaRT-PCR, including extensive validation of the method in independent laboratories have been presented in several recent articles and chapters [13-15]. With respect to the genes measured in this study, for each gene the StaRT-PCR reagents had lower detection threshold of less than 10 molecules, linear dynamic range of more than six orders of magnitude (less than 10 to over 107 molecules), and signal-to-analyte response of 100%. In addition, the presence of an internal standard controls for inter-sample variation in presence of PCR inhibitors (which often are gene-specific) and ensures no false negatives (if the PCR fails the internal standard PCR product is not observed and there are no data to report). False positives are eliminated through use of a control PCR reaction with no cDNA in it.
Statistical analysis
More than 6,000 transcript abundance measurements were conducted in multiple experiments over two years to assess the six transcription factors and sixteen antioxidant and DNA repair genes in NBEC samples from 49 individuals (24 non-BC individuals and 25 BC individuals).
Correlation of each of the six transcription factors with each of the antioxidant or DNA repair genes was determined by Pearson's correlation following logarithmic transformation. The transformation was necessary due to the wide biological variation in expression of each gene among the individuals. Significance level was defined as p < 0.01 following Bonferroni adjustment for multiple comparison, specifically comparison of each of six transcription factors to each of the antioxidant or DNA repair genes. Comparison for significant differences between pairs of correlation coefficients was done by Fisher's Z-transformation test [16].
Analysis of the relationship between virtually-multiplexed transcript abundance data for each gene with age was assessed by Pearson's correlation, with gender by t-test, and with smoking history by ANOVA followed by Duncan's test.
Transcription factor recognition site analysis
The El Dorado (Build 35) program from the Genomatix software package was used to locate the correlated genes within the genome and define 1101 base pairs of the promoter regions (1000 base pairs upstream of and 100 base pairs into the transcription start site) for each gene (Genomatix Software GmbH, Munich, Germany, [17]). The 1101 base pair sequences obtained from the El Dorado program then were used as the target sequences for putative transcription factor recognition site identification using the MatInspector Version 4.2 program, which yielded sites for 11 transcription factors (Genomatix Software GmbH, Munich, Germany, [17]). The parameters used were the standard (0.75) core similarity and the optimized matrix similarity [18]. StaRT-PCR reagents were optimized for ten of these transcription factors, including CEBPB, CEBPE, CEBPG, E2F1, E2F3, E2F4, E2F5, E2F6, EVI1, and PAX5. Four transcription factors were expressed at low and invariant levels among multiple NBEC samples and were therefore excluded from the study. The remaining six, CEBPB, CEBPG, E2F1, E2F3, E2F6, and EVI, were evaluated for correlation with an expanded group of ten antioxidant and six DNA repair genes.
Results
Virtually-multiplexed transcript abundance data were obtained for each gene in each of the 49 samples, except for E2F1 measurement in sample 147 (Table 3). A gene-specific inhibitor in sample 147 prevented amplification of E2F1. Neither the internal standard, nor the native cDNA PCR product was observed. The presence of gene-specific PCR inhibition was observable in some other samples as reduction in peak heights in internal standard PCR products relative to that expected for the number of internal standard molecules present at the beginning of the PCR reaction. However, in each such case, the PCR amplification was efficient enough to enable quantification.
Bivariate analysis
In non-BC individuals there was significant (p < 0.01) correlation between CEBPG and eight of the 16 antioxidant or DNA repair genes, specifically XRCC1, ERCC5, GSTP1, SOD1, GPX1, ERCC1, CAT and ERCC2 (Table 4). In contrast, in BC individuals samples CEBPG was not correlated with any of the antioxidant or DNA repair genes. These relationships were not observed with any of the other transcription factors studied.
For XRCC1, ERCC5, GSTP1, and SOD1 the correlation with CEBPG was significantly lower in BC individuals compared to non-BC individuals and the difference was nearly significant for GPX1 (Fig. 1b). Scatter plots of the relationship between CEBPG and XRCC1 in non-BC individuals or BC individuals (Fig. 2a,b) are representative of the other four genes. Neither CEBPG, nor XRCC1, ERCC5, GSTP1, SOD1 or GPX1 was significantly correlated with age, gender, or smoking history in non-BC individuals, BC individuals, or the combined group.
In non-BC individuals, based on the r2 values from Pearson's correlation analysis, CEBPG accounts for much of the variance in expression of XRCC1 (69%), ERCC5 (62%), GSTP1 (55%), SOD1 (44%), and GPX1 (52%). E2F1 accounts for some of the remaining variance. For example, when samples from all 49 non-BC individuals and BC individuals were assessed as a single group, E2F1 was significantly correlated with ERCC5, GSTP1 and SOD1 (Table 4). Further, in non-BC individuals, E2F1 was correlated with GSTP1 (Fig. 1c) and the correlation was lower in BC individuals. However, the difference in correlation between non-BC individuals and BC individuals was not significant. None of the other transcription factors were correlated with XRCC1, ERCC5, GSTP1, SOD1, or GPX1 (Fig. 1a,d,e,f).
Comparison of gene expression with demographic characteristics
E2F1 and GSTZ1 each were positively correlated with age. GSTM1-5 was the only gene with a difference in expression by gender. There was a difference in ERCC2 expression between former and never smokers.
Discussion
In this study, we tested two hypotheses. First, that there is inter-individual variation in regulation of key antioxidant and DNA repair genes by one or more transcription factors. Second, that individuals with sub-optimal regulation are selected for development of BC if they smoke cigarettes.
These hypotheses are supported by the findings that a) there was large inter-individual variation in transcript levels of CEBPG and each of the target genes and in non-BC individuals, b) CEBPG transcript abundance values were significantly correlated by bivariate analysis with the transcript abundance values of four key antioxidant and DNA repair genes in non-BC individuals, and c) that there was no correlation between CEBPG and these genes in BC individuals.
These results support the hypothesis that each of the antioxidant or DNA repair genes correlated with CEBPG in non-BC individuals is regulated by CEBPG. This is supported by the specificity of the CEBPG correlation. That is, there was lack of correlation between any of the other five transcription factors assessed and these target genes. Of particular note is the lack of correlation of the target genes with CEBPB, which binds to the same recognition site as CEBPG, and shares its recognition site within each of the antioxidant or DNA repair genes. However, there are alternative explanations for the observed correlation of CEBPG with antioxidant and DNA repair genes in non-BC individuals. One possibility is that CEBPG and each of the correlated antioxidant or DNA repair genes is regulated by a transcription factor that is as yet undiscovered, and/or has a recognition site that is not yet known and was not in the Genomatix software database.
There also is more than one possible explanation for the observed lack of correlation between CEBPG and antioxidant or DNA repair genes in BC individuals. For example, the non-BC individual and BC individual groups are not perfectly matched with respect to age, gender or smoking history (Table 1) and each of these factors could contribute to the observed difference in correlation between groups. However, the lack of association of transcript abundance level for CEBPG, XRCC1, ERCC5, GSTP1, SOD1, or GPX1 with age, gender or smoking history argues against such an explanation. One way to examine this possibility is through additional, larger, more closely matched studies. Another possible explanation is that any differences in NBEC from BC individuals compared to non-BC individuals resulted from development of BC, instead of being a hereditary cause of increased risk for cancer. The best way to determine this will be to conduct a prospective study. In such a study, individuals matched for smoking history will be monitored for development of BC over time. The correlation of transcript abundance values for CEBPG relative to transcript abundance values for each of the antioxidant or DNA repair genes will be assessed. It is expected that the greatest incidence of BC will be among the heaviest smokers. Among the matched heaviest smokers, it is expected that CEBPG will be significantly correlated with each of the antioxidant or DNA repair genes among the non-BC individuals but not correlated in BC individuals.
Thus, there are multiple possible explanations for the observed findings. However, based on the preponderance of data thus far available, we conclude that CEBPG is responsible for optimal transcriptional regulation of key antioxidant or DNA repair genes in NBEC and that there is inter-individual variation in the regulation of each of these genes by CEBPG. If this conclusion is correct, the individuals at greatest risk for BC will be those with the most extreme smoking history combined with sub-optimal regulation of the largest number of antioxidant and DNA repair genes. This, in turn, leads to increased representation among BC individuals of individuals with lack of correlation between CEBPG and each of the affected antioxidant and/or DNA repair genes.
CEBPG is a truncated CEBP transcription factor [19] and possesses the sequences necessary for DNA binding and heterodimer formation, but lacks the sequences necessary for transactivation [20]. CEBPG forms heterodimers with other CEBP family members and in other tissues this leads to increased [21] or decreased [20] transcription of the regulated gene. CEBPG is known to have stimulatory effect on the IL-6 and IL-8 promoters in B cell lines [21], and can also act as a dominant negative regulator of CEBPA and CEBPB in fibroblast and B cell lines [20].
The data from CEBPG knockout mice support a role for CEBPG in protecting lungs from oxidant damage. CEBPG-/- knockout mice are healthy at birth but begin to die within 24 hours, and histological examination reveals emphysematous lungs [22]. In humans, risk for emphysema is associated with antioxidant capacity [23], and there is a strong correlation between risk for emphysema and risk for BC.
However, it will be important to obtain direct experimental evidence in NBECs for the role of CEBPG in regulating the antioxidant and DNA repair genes included in this study. Correlation between CEBPG and target gene transcript levels may not be associated with correlation at the protein level.
In this study, E2F1 correlation with DNA repair and antioxidant genes was less than the correlation observed with CEBPG, and the E2F1 correlation was observed in both non-BC individuals as well as BC individuals. The maintained correlation of E2F1 with DNA repair and antioxidant genes in BC individuals suggests that this function is more tightly controlled in the population and does not play a role in determination of risk for BC. E2F1 has previously been reported to regulate transcription of DNA repair enzyme genes in other cell types, including primary human fibroblasts and mouse epidermal cells [24,25]. Clearly this would have survival value since DNA repair gene up-regulation in response to E2F1 provides additional DNA repair when the DNA is replicating and is particularly vulnerable to damage.
Epidemiologic assessment of the correlation between a particular variation in DNA sequence, or polymorphism, and risk for BC has been a dominant paradigm for many years. Thus far, these efforts have met with scant success [26]. A common limitation in design of such studies is that they involve assessment of a single polymorphism or occasionally, a few polymorphisms. Further, although the polymorphism assessed typically resides within a gene known to protect bronchial epithelium from carcinogens, oxidants, or DNA damage, the selection of the particular polymorphism for study is largely empiric, and not based on known functional properties. These are problems because multiple infrequent polymorphisms at different sites may all contribute to risk and unless the key polymorphisms can be identified through a functional test, a statistically valid assessment would require much larger study populations [27].
The findings of this study support a novel approach to identifying clinically useful biomarkers. According to the paradigm used in this study, a) a normal phenotype results from regulated transcription of a group of genes by one or more transcription factors, b) the corresponding risk-conferring or disease phenotype results from sub-optimal interaction among those same genes, and c) each phenotype is identifiable and distinguishable through virtually-multiplexed transcript abundance analysis. The data presented here support the utility of this paradigm in identifying genes associated with risk for BC.
The next step will be to identify polymorphisms that affect regulation of XRCC1, ERCC5, GSTP1, SOD1, and GPX1 by CEBPG. Such polymorphisms should yield biomarkers suitable for more readily accessible samples, such as peripheral blood or buccal smears. A biomarker combining polymorphisms that affect regulation with those that affect function of antioxidant and DNA repair genes is likely to be the most accurate for identifying individuals at risk for BC. Biomarkers that accurately identify individuals at risk for BC will improve efficacy of chemoprevention and early detection clinical trials.
The observed inter-sample variation in the presence of gene-specific inhibitors of PCR provides evidence supporting the need for inclusion of an internal standard in each quantitative PCR transcript abundance measurement. Including such internal standards in the form of standardized mixtures of internal standards improves the reproducibility of transcript abundance measurement and enables development of a standardized database comprising virtually-multiplexed transcript abundance data. Virtually-multiplexed transcript abundance data are highly suited to identification of genes that have correlated transcript abundance values. Correlation at the transcript abundance level is an important property of genes that are co-regulated at the transcription level.
Conclusion
We conclude that in non-BC individuals, CEBPG regulates transcription of key antioxidant or DNA repair genes in NBEC and that in smokers who develop BC, CEBPG regulation is sub-optimal for a sufficient number of antioxidant and/or DNA repair genes to cause increased risk.
Competing interests
JCW and ELC each have significant equity interest in Gene Express, Inc., which produces and markets StaRT-PCR™ reagents used in this study.
Authors' contributions
DNM participated in the design of the study, performed the TF identification, carried out TA measurements, coordinated and participated in the statistical analysis and drafted the manuscript. ELC contributed the preliminary data, participated in the design of the study and carried out TA measurements. SAK performed the statistical analysis for interpretation of data. DAH and YY consented patients and obtained the primary samples necessary for the study according to IRB regulations. JCW conceived of the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
These studies and manuscript preparation were supported by grants from the National Cancer Institute (NCI), including CA85147 and CA 95806, and the George Isaac Cancer Research Fund. StaRT-PCR reagents for measurement of some of the genes were provided at no cost by Gene Express, Inc. Neither the NCI nor Gene Express, Inc. had any role in study design, collection, analysis, or interpretation of data, writing of the manuscript, or in the decision to submit the manuscript for publication.
Figures and Tables
Figure 1 Correlation of each transcription factor with XRCC1, ERCC5, GSTP1, SOD1, or GPX1. (a-f) Each panel presents the correlation coefficients (r values) for one transcription factor in relation to each of the five genes: (a) CEBPB, (b) CEBPG, (c) E2F1, (d) E2F3, (e) E2F6, (f) EVI1. The p value for each significant correlation is provided above the bar. For CEBPG, presented in panel b, the difference in r value between non-BC individuals and BC individuals was significant or nearly significant for each correlated gene, and the p value for each comparison is provided below the corresponding pair of bars.
Figure 2 Scatter plot representation of bivariate correlation of CEBPG with XRCC1. (a, b) CEBPG/XRCC1 data from Figure 1b presented as scatter plots: (a) non-BC individuals, (b) BC individuals.
Table 1 Demographic data of patients from whom the NBEC samples were obtained.
Subject # Group Age Gender Histology Smoking Hx1 Ethnicity
63 NBCI2 77 M 75 W3
64 NBCI 47 F 45 W
136 NBCI 38 M 25 A.A.4
139 NBCI 44 F 17.5 W
150 NBCI 70 F 45 W
156 NBCI 46 F NS5 A.A.
157 NBCI 60 M >100 W
194 NBCI 57 M 3 W
210 NBCI 40 M 34 W
257 NBCI 69 F 20 W
261 NBCI 73 F NS W
282 NBCI 83 F 60 W
285 NBCI 69 F NS W
296 NBCI 43 F 20 W
305 NBCI 50 F 40 W
315 NBCI 64 M N/A6 W
330 NBCI 39 M NS W
331 NBCI N/A N/A N/A N/A
334 NBCI 51 M >50 W
336 NBCI 31 F NS W
337 NBCI 32 M 22 N/A
339 NBCI 59 M 50 W
361 NBCI 73 F NS H7
363 NBCI 50 M 20 A.A.
34 BCI8 80 M NSCLC9 40 W
71 BCI 63 M NSCLC 100 W
85 BCI 73 F SQ10 >100 W
88 BCI 85 M SQ 75 W
99 BCI 63 M NSCLC 45 W
118 BCI 72 M SQ 30 W
146 BCI 64 F SCLC11 45 W
147 BCI 76 M SCLC 75 W
158 BCI 88 M SCLC 115.5 W
167 BCI 60 F NSCLC 50 W
171 BCI 67 M SCLC 100 W
191 BCI 75 M SQ 54 W
211 BCI 71 M SQ 50 W
212 BCI 65 M SQ 67.5 W
247 BCI 75 F SQ 50 W
255 BCI 60 F NSCLC 30 W
259 BCI 68 M CS12 137.5 W
271 BCI 58 M AC13 94.5 W
287 BCI 65 F NSCLC 50 W
300 BCI 56 M SQ 34 W
306 BCI 46 M SQ 30 W
314 BCI 69 F BC14 NS W
329 BCI 76 F PD15 >37.5 W
335 BCI 75 M SCLC 58 A.A.
B3 BCI 63 M SQ 60 W
1Pack years; 2Non-bronchogenic carcinoma individual; 3White; 4African-American; 5Non-smoker; 6Not available; 7Hispanic; 8Bronchogenic carcinoma individual; 9Non-small cell lung cancer; 10Squamous carcinoma; 11Small cell lung cancer; 12Carcinoma-in-situ; 13Adenocarcinoma;14Bronchogenic Cancer, histology not specified; 15Poorly differentiated carcinoma.
Table 2 Sequence for each primer used for StaRT-PCR virtually-multiplexed transcript abundance measurement or for internal standard preparation (CT) [15].
Gene Accession # Primer Sequence Position Product
ACTB X00351 Forward 5' ATC CTC ACC CTG AAG TAC CC 3' 231
Reverse 5' CCA TCT CTT GCT CGA AGT CC 3' 704 493 bp
CT 5' CCA TCT CTT GCT CGA AGT CCG CCA GCC AGG TCC AGA CGC A 3' 568 377 bp
CAT X04076 Forward 5' CCA GAA GAA AGC GGT CAA GA 3' 1492
Reverse 5' AAC CTT CAT TTT CCC CTG GG 3' 1822 350 bp
CT 5' AAC CTT CAT TTT CCC CTG GGC CAG TGA TGA GCG GGT TAC A 3' 1699 247 bp
CEBPB NM_005194 Forward 5' TGT CCA AAC CAA CCG CAC AT 3' 1412
Reverse 5' AGC AAC AAG CCC GTA GGA AC 3' 1657 265 bp
CT 5' AGC AAC AAG CCC GTA GGA ACA CGC GTT CAG CCA TGT TTA A 3' 1571 199 bp
CEBPG U20240 Forward 5' CGG TTG AAA AGC AAG CAG AAA GCA 3' 488
Reverse 5' GAT CCC AGA AAA TAG CCT CCA ATG 3' 814 350 bp
CT 5' GAT CCC AGA AAA TAG CCT CCA ATG AAC ATT CAA GCC ACA AGC TC 3' 726 282 bp
E2F1 M96577 Forward 5' TGA TAC CCC AAC TCC CTC TA 3' 2076
Reverse 5' AAA GCA GGA GGG AAC AGA GC 3' 2452 396 bp
CT 5' AAA GCA GGA GGG AAC AGA GCA CTG CAG GGA CCA CAG G 3' 2363 327 bp
E2F3 Y10479 Forward 5' TGA AAG CCC CTC CAG AAA CAA G 3' 1019
Reverse 5' GCA GCA GGG GAG GCA GTA AGT T 3' 1336 339 bp
CT 5' GCA GCA GGG GAG GCA GTA AGT TGG GGA GGC CAG AGG AGA AAG GT 3' 1253 278 bp
E2F6 AF059292 Forward 5' GGG CCT GCT GCC ATC AAA AAT A 3' 99
Reverse 5' CCG CTT TCG GAC TCC CAG TTT 3' 283 205 bp
CT 5' CCG CTT TCG GAC TCC CAG TTA GCG ATA CAT CAA AAC GAG G 3' 184 125 bp
ERCC1 M13194 Forward 5' CTG GAG CCC CGA GGA AGC 3' 739
Reverse 5' CAC TGG GGG TTT CCT TTG 3' 1049 328 bp
CT 5' CAC TGG GGG TTT CCT TGG AAG GCC AGA TCT TCT CTT 3 928 240 bp
ERCC2 X52221 Forward 5' GGC CTT CTT CAC CAG CTA C 3' 1608
Reverse 5' GTA GTC CGT CTT GCC CCT G 3' 2004 415 bp
CT 5' GTA GTC CGT CTT GCC CCT GTG GAA CTG GTC CCG CAG GT 3' 2597 346 bp
ERCC4 U64315 Forward 5' AGT GCA TCT CCA TGT CCC GCT ACT A 3' 2213
Reverse 5' CGA TGT TCT TAA CGT GGT GCA TCA A 3' 2578 390 bp
CT 5' CGA TGT TCT TAA CGT GGT GCA TCA ACA GGC TGT GGC TTG CTT TGT 3' 2433 265 bp
ERCC5 D16305 Forward 5' AAG GAA AGA GAA AGA AGC AGC AGC CA 3' 3087
Reverse 5' CAA ACA CAG ATC TGG CGG TCA CGA GG 3' 3501 440 bp
CT 5' CAA ACA CAG ATC TGG CGG TCA CGA GGA GCT TCC TTC ACT GAG TTC TGC GAA T 3' 3401 366 bp
EVI1 NM_005241 Forward 5' CGC CGG ATA TCC ACG AAG A 3' 302
Reverse 5' ATG CTG AGA GCG AAT GTG C 3' 711 428 bp
CT 5' ATG CTG AGA GCG AAT GTG CTT AAA TGC CTT GGG ACA CT 3' 587 323 bp
GPX1 Y00433 Forward 5' CCT GGT GGT GCT CGG CTT CC 3' 522
Reverse 5' CAA TGG TCT GGA AGC GGC GG 3' 852 350 bp
CT 5' CAA TGG TCT GGA AGC GGC GGA CCG GAG ACC AGG TGA TGA G 3' 757 279 bp
GPX3 D16360 Forward 5' GCA GAG CCG GGG ACA AGA GAA 3' 113
Reverse 5' CTG CTC TTT CTC TCC ATT GAC 3' 471 379 bp
CT 5' CTG CTC TTT CTC TCC ATT GAC GCT CTT CCT GTA GTG CAT TCA 3' 298 227 bp
GSTM1,2,4,5 J03817 Forward 5' GGG ACG CTC CTG ATT ATG AC 3' 122
Reverse 5' GCA AAC CAT GGC CGC TTC CC 3' 442 340 bp
CT 5' GCA AAC CAT GGC CGC TTC CCT TCT CCA AAA TGT CCA CAC G 3' 301 219 bp
GSTM3 J05459 Forward 5' GTG CGA GTC GTC TAT GGT TC 3' 23
Reverse 5' AGT TGT GTG CGG AAA TCC AT 3' 342 339 bp
CT 5' AGT TGT GTG CGG AAA TCC ATT GCT CTG GGT GAT CTT GTT C 3' 230 247 bp
GSTP1 X08058 Forward 5' TCC GCT GCA AAT ACA TCT CC 3' 305
Reverse 5' TGT TTC CCG TTG CCA TTG AT 3' 616 331 bp
CT 5' TGT TTC CCG TTG CCA TTG ATT AGG ACC TCA TGG ATC AGC A 3' 485 220 bp
GSTT1 X79389 Forward 5' GCT CTA CCT GGA CCT GCT GT 3' 12
Reverse 5' GGA ACA CAG GGA ACA TCA CC 3' 351 359 bp
CT 5' GGA ACA CAG GGA ACA TCA CCT AGA GCA GGA TGG CCA CAC T 3' 199 227 bp
GSTZ1 U86529 Forward 5' TCA CCC CCT ACC CTA CCA TCA GC 3' 806
Reverse 5' ATT TCA GCG CGG GCA TTC TTT 3' 1267 482 bp
CT 5' ATT TCA GCG CGG GCA TTC TTT CCG CAT TCT CAT CTC AGC CTC AC 3' 1161 399 bp
mGST1 J03746 Forward 5' GTC GGA GCA CGG ATC TAC CAC A 3' 404
Reverse 5' TTC CTC TGC TCC CCT CCT ACC TA 3' 623 242 bp
CT 5' TTC CTC TGC TCC CCT CCT ACC TAT TTT CAG CAA CCT GTA AGC C 3' 505 144 bp
SOD1 X02317 Forward 5' TGA AGG TGT GGG GAA GCA TTA 3' 153
Reverse 5' TTA CAC CAC AAG CCA AAC GAC 3' 492 360 bp
CT 5' TTA CAC CAC AAG CCA AAC GAC TGA TGC AAT GGT CTC CTG AGA 3' 384 273 bp
XPA D14533 Forward 5' CTC GGC GAC GGC GGC TGC GGC TAC TGG AG 3' 178
Reverse 5' TGT CGG ACT TCC TTT GCT TCT TCT AAT GC 3' 629 480 bp
CT 5' TGT CGG ACT TCC TTT GCT TCT TCT AAT GCT CTT TTT TCT AAA TCA CAG TCT 3' 487 360 bp
XRCC1 M36089 Forward 5' CCC CTG AAG AGA CCA AAG CA 3' 1906
Reverse 5' CCA TTG AAG GCT GTG ACG TA 3' 2241 355 bp
CT 5' CCA TTG AAG GCT GTG ACG TAT CAG GGA CTG GCA GAT G 3' 2142 276 bp
Table 3 Virtually-multiplexed transcript abundance data.
SUBJECT # GROUP CEBPB CEBPG E2F1 E2F3 E2F6 EVI1 CAT ERCC1 ERCC2 ERCC4 ERCC5
63 NBCI 7.2E+03 6.4E+02 2.7E+02 1.9E+02 3.4E+02 6.1E+01 2.0E+04 1.1E+05 3.8E+03 1.1E+02 4.3E+04
64 NBCI 7.9E+03 1.7E+03 2.0E+03 2.0E+01 1.7E+02 2.5E+01 2.5E+04 3.0E+05 3.4E+03 8.1E+01 2.0E+05
136 NBCI 6.2E+03 2.1E+02 7.0E+02 1.0E-02 5.0E+01 6.0E+01 2.9E+03 7.4E+03 5.9E+02 5.6E+01 1.9E+04
139 NBCI 4.5E+03 3.4E+03 2.3E+04 5.6E+02 1.0E+03 1.0E+03 6.1E+05 1.2E+06 2.2E+04 3.9E+03 4.6E+05
150 NBCI 8.5E+03 7.4E+02 1.6E+02 1.1E+02 4.1E+01 3.2E+02 3.5E+04 1.7E+05 5.7E+03 6.9E+02 7.2E+04
156 NBCI 2.1E+04 1.2E+03 7.5E+02 ND1 ND 1.4E+02 1.5E+04 1.8E+05 2.0E+03 2.9E+02 3.0E+04
157 NBCI 2.3E+04 4.1E+03 3.1E+03 2.1E+02 6.1E+02 1.4E+02 3.5E+05 5.6E+05 8.4E+03 1.6E+03 2.1E+05
194 NBCI 6.5E+03 2.1E+03 2.9E+02 2.6E+02 8.5E+02 4.7E+02 4.5E+04 6.1E+05 4.7E+03 4.4E+02 8.9E+04
210 NBCI 1.0E+04 2.1E+03 7.6E+02 4.0E+02 6.1E+02 3.6E+02 7.6E+04 7.6E+04 3.4E+03 7.7E+02 7.2E+04
257 NBCI 1.1E+04 1.8E+03 2.7E+02 8.9E+02 1.7E+03 9.7E+02 1.0E+05 2.6E+05 1.6E+03 7.6E+02 7.4E+04
261 NBCI 7.6E+03 1.3E+03 2.5E+02 1.7E+02 1.6E+02 9.4E+01 4.4E+04 2.7E+05 5.1E+03 4.6E+02 1.3E+05
282 NBCI 6.4E+03 1.2E+03 5.4E+02 4.1E+01 2.9E+01 1.2E+02 4.1E+04 1.0E+05 6.0E+03 8.1E+02 4.0E+04
285 NBCI 2.6E+03 4.4E+02 2.5E+03 1.1E+03 ND 1.2E+02 3.7E+04 9.1E+04 1.6E+03 7.7E+03 1.6E+04
296 NBCI 1.9E+04 1.0E+03 1.0E+03 ND 2.8E+01 ND 6.9E+04 2.5E+05 3.6E+03 1.7E+02 7.1E+04
305 NBCI 1.8E+03 9.1E+01 6.1E+02 ND 8.7E+01 2.6E+02 2.0E+04 4.9E+04 4.3E+02 2.8E+02 1.3E+04
315 NBCI 3.5E+03 1.3E+03 1.2E+03 2.0E+02 7.5E+01 6.1E+02 4.7E+04 2.1E+05 2.7E+03 2.4E+03 6.2E+04
330 NBCI 2.7E+03 2.4E+02 4.0E+02 1.1E+02 3.5E+02 4.0E+02 3.6E+04 8.3E+04 1.5E+03 1.7E+03 2.8E+04
331 NBCI 7.3E+03 1.3E+03 8.0E+02 ND ND ND 8.6E+04 2.6E+05 6.5E+02 6.4E+02 6.2E+04
334 NBCI 3.7E+03 6.1E+02 1.1E+03 2.3E+01 1.9E+01 4.0E+01 7.8E+04 1.4E+05 2.5E+03 1.3E+03 5.1E+04
336 NBCI 3.6E+03 8.4E+02 2.8E+03 4.3E+02 1.2E+02 2.5E+02 8.9E+04 2.1E+05 7.9E+03 3.1E+03 9.5E+04
337 NBCI 5.0E+03 9.3E+02 1.1E+03 1.6E+02 2.5E+02 ND 6.5E+04 1.9E+05 3.3E+03 1.3E+03 4.2E+04
339 NBCI 2.5E+03 3.2E+02 5.2E+02 3.4E+01 4.3E+01 6.3E+01 5.0E+04 8.7E+04 2.3E+03 7.7E+02 3.2E+04
361 NBCI 7.9E+03 2.5E+03 6.7E+02 5.8E+02 1.4E+03 3.5E+02 6.7E+04 2.5E+05 1.5E+03 2.8E+03 4.7E+04
363 NBCI 6.2E+03 1.7E+03 1.4E+03 1.1E+02 1.9E+02 5.2E+01 7.6E+04 1.4E+05 3.7E+03 7.3E+02 7.2E+04
34 BCI 1.9E+03 1.7E+03 5.7E+02 2.4E+02 1.2E+02 3.7E+02 6.7E+04 5.6E+05 6.1E+03 1.5E+03 1.2E+05
71 BCI 7.6E+03 1.2E+03 1.3E+03 1.7E+01 1.8E+02 2.1E+01 9.5E+04 6.7E+05 1.8E+04 2.4E+02 2.2E+05
85 BCI 1.0E+04 9.7E+02 8.7E+02 6.3E+01 ND ND 2.6E+04 4.9E+04 1.1E+03 3.9E+02 2.3E+04
88 BCI 1.5E+03 4.0E+02 4.2E+02 1.1E+02 ND ND 3.1E+04 2.9E+04 2.7E+03 3.6E+02 3.9E+04
99 BCI 1.4E+04 3.3E+03 2.6E+03 2.2E+02 3.1E+03 1.4E+02 2.1E+05 2.0E+05 3.7E+03 9.3E+02 1.4E+05
118 BCI 1.5E+03 1.2E+03 2.8E+02 ND ND 1.4E+02 2.4E+04 3.5E+04 1.9E+03 1.2E+03 1.7E+04
146 BCI 1.9E+04 1.2E+03 2.3E+02 1.2E+02 1.6E+03 6.0E+01 3.1E+04 1.4E+05 4.9E+03 2.7E+02 6.2E+04
147 BCI 2.8E+03 1.3E+03 N/A2 3.6E+02 2.8E+02 3.8E+02 6.1E+04 3.0E+05 2.7E+03 1.1E+03 1.8E+04
158 BCI 4.9E+03 8.8E+02 1.9E+02 2.8E+01 4.0E+02 6.8E+01 2.2E+04 1.4E+05 1.7E+03 1.5E+02 1.3E+04
167 BCI 6.7E+03 9.2E+02 4.2E+02 1.8E+02 8.2E+02 3.8E+02 8.9E+04 1.1E+05 2.2E+03 2.8E+02 4.1E+04
171 BCI 1.0E+04 3.9E+03 3.2E+03 5.1E+02 8.9E+01 2.4E+02 5.3E+05 8.3E+05 1.3E+04 6.3E+02 1.6E+05
191 BCI 1.6E+04 2.4E+03 1.5E+02 1.5E+02 7.1E+01 1.2E+02 2.0E+04 9.6E+04 2.5E+03 3.4E+02 1.2E+04
211 BCI 3.8E+03 1.1E+03 6.5E+02 6.3E+02 3.2E+02 2.9E+02 8.3E+04 3.5E+05 2.8E+03 3.2E+02 9.1E+04
212 BCI 1.8E+04 2.8E+03 6.0E+02 3.4E+02 2.0E+02 2.2E+02 3.4E+04 2.1E+05 5.2E+03 3.0E+02 3.6E+04
247 BCI 6.5E+03 7.5E+02 6.4E+02 1.8E+02 ND 1.2E+02 7.8E+04 4.4E+04 5.5E+02 2.7E+03 1.3E+04
255 BCI 1.9E+04 1.1E+03 6.1E+02 1.8E+01 2.3E+02 7.6E+01 1.1E+05 6.5E+05 1.2E+04 1.2E+03 3.8E+05
259 BCI 8.4E+03 6.4E+02 5.1E+02 ND 6.7E+01 4.0E+01 4.7E+04 1.5E+05 2.6E+03 3.8E+02 7.9E+04
271 BCI 4.1E+03 6.6E+02 1.4E+03 6.5E+01 2.7E+02 1.9E+02 9.1E+04 1.5E+05 1.6E+03 ND 1.5E+05
287 BCI 8.7E+03 1.1E+03 9.5E+01 7.4E+01 4.0E+01 2.6E+02 4.7E+04 1.2E+05 6.5E+03 1.1E+02 4.1E+04
300 BCI 4.9E+03 4.4E+02 4.1E+02 4.4E+01 ND 5.1E+01 3.4E+04 6.9E+04 1.0E+03 1.0E+03 2.6E+04
306 BCI 6.5E+03 6.8E+02 4.9E+02 4.4E+01 ND 1.7E+01 3.7E+04 2.5E+05 2.8E+03 3.2E+02 4.1E+04
314 BCI 4.4E+03 9.3E+02 4.8E+02 3.1E+02 ND 1.9E+02 5.3E+04 6.2E+04 4.5E+03 5.2E+02 3.1E+04
329 BCI 1.4E+04 3.5E+02 2.4E+02 ND ND ND 7.9E+04 1.5E+05 1.4E+03 ND 1.5E+05
335 BCI 4.2E+03 3.7E+02 2.2E+03 3.3E+02 2.4E+02 2.4E+02 4.8E+04 1.9E+05 9.8E+03 3.1E+02 3.3E+04
B3 BCI 7.4E+03 9.3E+02 1.4E+02 2.4E+02 1.8E+02 3.8E+02 3.4E+04 1.6E+05 2.2E+03 2.8E+02 4.1E+04
SUBJECT # GROUP GPX1 GPX3 GSTM15 GSTM3 GSTP1 GSTT1 GSTZ1 MGST1 SOD1 XPA XRCC1
63 NBCI 8.4E+05 1.5E+03 7.3E+03 3.9E+03 1.9E+06 ND 3.1E+03 1.7E+05 1.6E+05 2.5E+03 2.1E+04
64 NBCI 4.4E+05 3.1E+02 1.5E+04 2.6E+03 3.5E+06 6.4E+03 2.0E+03 1.6E+05 6.5E+05 2.2E+03 4.2E+04
136 NBCI 2.0E+05 5.2E+02 6.3E+03 2.2E+03 5.3E+05 ND 8.9E+02 3.2E+04 5.3E+04 1.1E+03 4.4E+03
139 NBCI 1.8E+06 3.2E+03 5.2E+04 8.3E+03 3.2E+07 ND 2.8E+04 8.4E+05 2.2E+06 2.6E+04 1.6E+05
150 NBCI 2.1E+05 2.8E+03 1.1E+04 1.1E+03 1.5E+06 ND 3.1E+03 3.1E+04 2.0E+05 2.0E+03 3.1E+04
156 NBCI 4.8E+05 3.9E+03 1.9E+04 6.5E+03 2.9E+06 ND 4.9E+03 5.8E+04 1.3E+05 4.0E+03 2.5E+04
157 NBCI 2.4E+06 2.0E+03 2.6E+04 3.3E+03 1.8E+07 8.5E+03 3.6E+03 3.6E+05 1.8E+06 6.2E+03 1.7E+05
194 NBCI 5.4E+05 4.0E+03 1.2E+04 1.1E+03 7.8E+06 7.7E+03 3.2E+03 9.6E+04 5.8E+05 3.7E+03 2.4E+05
210 NBCI 3.2E+05 3.3E+03 4.1E+03 2.6E+03 3.9E+06 1.8E+03 3.7E+03 7.2E+04 3.3E+05 4.8E+03 2.9E+04
257 NBCI 6.3E+05 2.5E+03 1.1E+04 1.2E+02 2.8E+06 8.7E+03 2.9E+03 2.3E+04 1.2E+05 2.6E+03 4.9E+04
261 NBCI 7.6E+05 4.7E+03 1.6E+04 7.4E+03 3.1E+06 1.5E+03 1.8E+03 1.2E+05 4.5E+05 4.9E+03 7.2E+04
282 NBCI 4.6E+05 2.5E+03 1.3E+04 2.3E+03 1.9E+06 1.5E+04 1.3E+03 6.5E+04 4.5E+05 2.0E+03 3.1E+04
285 NBCI 4.3E+05 5.3E+03 8.4E+04 6.2E+02 3.5E+06 7.3E+03 9.2E+03 3.0E+04 2.3E+05 ND 2.0E+04
296 NBCI 8.1E+05 1.5E+03 7.4E+03 4.1E+03 3.6E+06 8.1E+03 4.5E+03 2.0E+05 5.4E+05 2.6E+03 5.2E+04
305 NBCI 1.4E+05 6.5E+02 4.0E+03 3.2E+03 1.6E+06 ND 6.9E+02 4.1E+04 1.8E+05 1.3E+03 5.7E+03
315 NBCI 4.2E+05 7.6E+03 5.1E+03 3.1E+03 7.9E+06 7.2E+03 4.3E+03 7.3E+04 4.5E+05 1.0E+04 3.8E+04
330 NBCI 1.2E+05 2.0E+03 5.8E+03 3.5E+03 1.1E+06 3.1E+01 1.3E+03 2.9E+04 2.3E+05 1.4E+03 1.7E+04
331 NBCI 6.1E+05 3.4E+03 1.3E+04 3.6E+03 7.3E+06 1.0E+04 1.8E+03 1.2E+05 1.3E+06 1.9E+03 5.7E+04
334 NBCI 6.5E+05 4.0E+03 2.7E+04 1.7E+04 3.7E+06 4.6E+03 3.0E+03 6.8E+04 5.9E+05 2.9E+03 2.6E+04
336 NBCI 4.7E+05 2.7E+03 4.4E+04 1.6E+03 3.4E+06 1.2E+04 8.3E+03 6.2E+04 5.4E+05 4.2E+03 1.2E+05
337 NBCI 2.8E+05 2.8E+03 3.8E+03 4.3E+03 1.5E+06 5.3E+03 5.6E+03 4.9E+04 3.3E+05 1.7E+03 4.4E+04
339 NBCI 3.1E+05 1.6E+04 3.0E+04 1.1E+03 3.5E+06 6.6E+03 2.0E+03 6.6E+04 2.4E+05 2.5E+03 2.2E+04
361 NBCI 3.7E+05 8.4E+02 1.5E+04 1.3E+03 7.8E+06 3.1E+03 2.7E+03 4.8E+04 6.4E+05 6.1E+03 7.7E+04
363 NBCI 6.1E+05 2.7E+03 2.2E+04 2.0E+03 8.1E+06 9.9E+03 2.4E+03 9.7E+04 7.0E+05 7.0E+03 3.6E+04
34 BCI 8.8E+05 3.1E+03 3.4E+03 5.1E+02 1.8E+06 ND 1.9E+03 6.1E+04 2.9E+05 3.1E+03 5.6E+04
71 BCI 8.8E+05 1.3E+04 3.2E+04 2.2E+03 6.7E+06 ND 2.5E+03 4.1E+05 7.5E+05 3.1E+03 5.6E+05
85 BCI 2.3E+05 1.1E+03 2.6E+04 4.1E+03 1.2E+06 1.0E+04 8.1E+02 4.8E+04 1.5E+05 3.5E+03 1.8E+04
88 BCI 1.3E+05 1.5E+03 3.5E+03 1.1E+03 6.9E+05 ND 5.6E+02 2.4E+04 1.9E+05 2.4E+03 1.6E+04
99 BCI 9.0E+05 2.1E+03 1.5E+04 5.5E+03 9.6E+06 6.7E+03 8.0E+03 1.2E+05 9.4E+05 8.5E+03 4.7E+04
118 BCI 8.0E+04 3.8E+03 1.1E+04 4.8E+03 1.7E+06 4.0E+03 4.5E+02 4.4E+04 1.7E+05 1.1E+03 6.1E+03
146 BCI 4.1E+05 2.9E+04 4.7E+04 4.7E+02 2.0E+06 7.1E+03 1.6E+04 6.6E+04 1.7E+05 3.8E+04 6.5E+04
147 BCI 4.1E+05 2.5E+03 8.2E+03 7.3E+02 8.8E+05 1.7E+03 1.3E+03 2.3E+04 6.5E+05 2.3E+02 7.4E+04
158 BCI 2.6E+04 2.6E+03 1.4E+04 4.7E+03 2.3E+06 2.9E+03 1.2E+03 2.3E+04 3.7E+04 1.9E+03 2.3E+04
167 BCI 1.9E+05 2.5E+03 5.3E+03 1.8E+03 2.4E+06 1.4E+04 2.3E+03 9.5E+04 4.0E+05 2.9E+03 2.8E+04
171 BCI 1.6E+06 1.7E+03 1.6E+04 4.0E+03 7.8E+06 6.0E+03 6.5E+03 2.5E+05 1.4E+06 1.4E+04 6.0E+04
191 BCI 4.5E+05 1.9E+03 3.5E+03 2.5E+02 1.7E+06 3.6E+03 5.6E+02 2.1E+04 1.6E+05 2.5E+03 2.1E+04
211 BCI 6.0E+05 1.7E+04 1.1E+04 1.3E+02 1.0E+07 5.5E+03 4.1E+03 2.4E+04 7.1E+05 4.3E+03 8.3E+04
212 BCI 6.5E+04 1.6E+03 9.6E+03 1.1E+03 4.0E+06 8.1E+03 9.5E+02 7.5E+04 9.4E+04 3.1E+03 3.6E+04
247 BCI 2.3E+05 2.0E+02 1.4E+04 2.1E+03 1.3E+06 7.1E+03 1.1E+03 2.4E+04 1.3E+05 1.3E+03 1.1E+04
255 BCI 1.3E+06 7.3E+03 1.5E+04 2.5E+03 2.8E+06 3.8E+04 8.1E+03 1.1E+05 2.8E+06 2.0E+03 1.0E+05
259 BCI 3.0E+06 5.9E+03 1.4E+04 7.0E+03 9.2E+06 5.1E+03 5.1E+03 1.1E+05 5.0E+05 4.8E+03 2.1E+04
271 BCI 7.4E+05 9.5E+02 3.8E+03 2.6E+03 2.4E+06 3.7E+03 3.1E+03 7.1E+04 6.3E+05 4.1E+03 3.0E+04
287 BCI 5.8E+05 8.7E+03 1.1E+04 2.3E+03 1.9E+06 1.5E+04 6.3E+02 8.8E+04 1.2E+05 4.7E+03 3.3E+04
300 BCI 2.7E+05 1.5E+03 7.7E+03 5.0E+03 3.7E+06 7.1E+03 1.9E+03 6.7E+04 3.2E+05 8.0E+02 1.3E+04
306 BCI 4.2E+05 2.0E+02 9.7E+03 2.3E+03 2.2E+06 ND 4.8E+03 5.6E+04 1.1E+06 4.5E+02 1.9E+04
314 BCI 1.6E+05 3.7E+03 1.7E+04 1.1E+03 1.2E+06 1.9E+02 1.5E+03 2.7E+04 2.7E+05 2.8E+03 3.4E+04
329 BCI 2.3E+05 8.9E+03 1.6E+04 5.3E+03 7.2E+06 7.8E+03 1.4E+03 5.7E+04 8.9E+05 1.9E+03 4.8E+04
335 BCI 4.1E+05 6.5E+03 2.1E+04 7.1E+03 4.3E+06 1.8E+02 7.0E+03 6.9E+04 5.2E+05 3.9E+03 4.4E+04
B3 BCI 4.0E+05 3.2E+03 6.0E+03 3.8E+02 1.6E+06 7.7E+03 1.8E+03 1.8E+04 2.6E+05 3.7E+03 1.7E+04
Virtually-multiplexed transcript abundance data for each gene (in the form of molecules/106 β-actin molecules) from all experiments were included in the same Standardized Expression Database (SED). These data are now directly comparable to previously published virtually-multiplexed transcript abundance data from this laboratory [15], or to Virtually-multiplexed transcript abundance data collected by others who use the NCI-funded (R24 CA 95806) Standardized Expression Measurement (SEM) Center. The data presented here represent more than 6,000 Virtually-multiplexed transcript abundance measurements conducted in multiple experiments. The sixteen antioxidant or DNA repair genes and each of the six transcription factors except for E2F1 were measured in each NBEC sample from 49 individuals (24 non-BC individuals and 25 BC individuals).
1ND, Not detectable (When 60 molecules of Competitive Template (CT) were added to the reaction, CT was detected but native template was not.)
2NA, Not available. E2F1 was measured in all of the samples except for one BC individual (24 non-BC individuals and 24 BC individuals).
Table 4 Bivariate analysis of virtually-multiplexed transcript abundance data values for each antioxidant or DNA repair gene versus each transcription factor.
Non-BC Individuals n = 24 BC Individuals n = 25 ALL n = 49
Antioxidant/DNA Repair Genes vs Transcription Factors r Value p Value r Value p Value r Value p Value
CAT vs CEBPB 0.13 1 0.18 1 0.15 1
CAT vs CEBPG 0.65 0.004 0.35 0.48 0.55 <0.0006
CAT vs E2F1* 0.54 0.04 0.68 0.002 0.56 <0.0006
CAT vs E2F3 0.48 0.12 0.18 1 0.37 0.06
CAT vs E2F6 0.26 1 0.3 0.84 0.25 0.48
CAT vs EVI1 -0.01 1 0.21 1 0.08 1
ERCC1 vs CEBPB 0.32 0.78 0.27 1 0.29 0.24
ERCC1 vs CEBPG 0.77 <0.0006 0.42 0.24 0.62 <0.0006
ERCC1 vs E2F1 0.35 0.54 0.39 0.36 0.37 0.06
ERCC1 vs E2F3 0.39 0.36 0.21 1 0.31 0.18
ERCC1 vs E2F6 0.17 1 0.63 0.005 0.42 0.02
ERCC1 vs EVI1 -0.02 1 0.38 0.36 0.17 1
ERCC2 vs CEBPB 0.25 1 0.19 1 0.22 0.84
ERCC2 vs CEBPG 0.63 0.006 0.39 0.3 0.53 <0.0006
ERCC2 vs E2F1 0.39 0.36 0.32 0.72 0.33 0.12
ERCC2 vs E2F3 0.58 0.02 0.22 1 0.42 0.02
ERCC2 vs E2F6 0.37 0.42 0.51 0.06 0.42 0.02
ERCC2 vs EVI1 0.19 1 0.29 0.96 0.23 0.66
ERCC4 vs CEBPB -0.35 0.6 -0.11 1 -0.16 1
ERCC4 vs CEBPG 0.24 1 0.37 0.42 0.25 0.48
ERCC4 vs E2F1 0.42 0.24 0.04 1 0.2 1
ERCC4 vs E2F3 0.6 0.01 0.33 0.6 0.33 0.12
ERCC4 vs E2F6 -0.04 1 0.04 1 0.07 1
ERCC4 vs EVI1 0.24 1 0.33 0.66 0.27 0.36
ERCC5 vs CEBPB 0.4 0.3 0.28 1 0.33 0.12
ERCC5 vs CEBPG 0.79 <0.0006 0.12 1 0.46 0.005
ERCC5 vs E2F1 0.44 0.18 0.45 0.18 0.44 0.01
ERCC5 vs E2F3 0.39 0.36 -0.11 1 0.13 1
ERCC5 vs E2F6 0.41 0.3 0.35 0.54 0.38 0.04
ERCC5 vs EVI1 0.07 1 -0.04 1 0.01 1
GPX1 vs CEBPB 0.49 0.06 0.24 1 0.32 0.12
GPX1 vs CEBPG 0.72 <0.0006 0.19 1 0.4 0.02
GPX1 vs E2F1 0.48 0.12 0.38 0.36 0.43 0.02
GPX1 vs E2F3 0.22 1 -0.004 1 0.08 1
GPX1 vs E2F6 0.06 1 0.36 0.48 0.28 0.3
GPX1 vs EVI1 -0.06 1 0.2 1 0.1 1
GPX3 vs CEBPB -0.18 1 0.14 1 0.02 1
GPX3 vs CEBPG 0.13 1 0.01 1 0.07 1
GPX3 vs E2F1 -0.03 1 -0.17 1 -0.12 1
GPX3 vs E2F3 0.32 0.78 -0.2 1 0.05 1
GPX3 vs E2F6 -0.26 1 0.44 0.18 0.19 1
GPX3 vs EVI1 0.06 1 0.08 1 0.06 1
GSTM1-5 vs CEBPB -0.08 1 0.43 0.18 0.17 1
GSTM1-5 vs CEBPG 0.25 1 0.02 1 0.16 1
GSTM1-5 vs E2F1 0.51 0.06 0.23 1 0.41 0.02
GSTM1-5 vs E2F3 0.29 0.96 -0.16 1 0.1 1
GSTM1-5 vs E2F6 -0.3 0.9 0.006 1 -0.1 1
GSTM1-5 vs EVI1 0.22 1 -0.12 1 0.07 1
GSTM3 vs CEBPB 0.01 1 0.06 1 0.04 1
GSTM3 vs CEBPG -0.007 1 -0.28 1 0.01 1
GSTM3 vs E2F1 0.29 1 0.35 0.54 0.34 0.12
GSTM3 vs E2F3 -0.31 0.84 -0.49 0.06 -0.4 0.03
GSTM3 vs E2F6 -0.11 1 -0.25 1 -0.16 1
GSTM3 vs EVI1 -0.27 1 -0.25 1 -0.25 0.54
GSTP1 vs CEBPB 0.19 1 0.38 0.36 0.28 0.36
GSTP1 vs CEBPG 0.74 <0.0006 0.18 1 0.51 0.001
GSTP1 vs E2F1 0.6 0.01 0.46 0.12 0.56 <0.0006
GSTP1 vs E2F3 0.32 0.78 -0.25 1 0.07 1
GSTP1 vs E2F6 0.1 1 0.35 0.48 0.26 0.42
GSTP1 vs EVI1 0.11 1 0.13 1 0.12 1
GSTT1 vs CEBPB 0.03 1 0.45 0.12 0.24 0.6
GSTT1 vs CEBPG 0.39 0.36 0.16 1 0.3 0.24
GSTT1 vs E2F1 0.07 1 -0.15 1 -0.05 1
GSTT1 vs E2F3 0.35 0.54 -0.1 1 0.17 1
GSTT1 vs E2F6 0.05 1 0.21 1 0.11 1
GSTT1 vs EVI1 -0.26 1 0.22 1 -0.04 1
GSTZ1 vs CEBPB 0.11 1 0.36 0.42 0.25 0.54
GSTZ1 vs CEBPG 0.51 0.06 0.08 1 0.28 0.3
GSTZ1 vs E2F1 0.64 0.004 0.5 0.06 0.58 <0.0006
GSTZ1 vs E2F3 0.42 0.24 0.14 1 0.25 0.54
GSTZ1 vs E2F6 -0.05 1 0.48 0.12 0.32 0.18
GSTZ1 vs EVI1 0.02 1 0.27 1 0.16 1
mGST vs CEBPB 0.31 0.78 0.35 0.48 0.32 0.12
mGST vs CEBPG 0.56 0.02 0.25 1 0.42 0.02
mGST vs E2F1 0.58 0.02 0.54 0.04 0.58 <0.0006
mGST vs E2F3 0.03 1 -0.15 1 -0.06 1
mGST vs E2F6 0.17 1 0.29 0.96 0.27 0.36
mGST vs EVI1 -0.16 1 0.07 1 -0.04 1
SOD1 vs CEBPB 0.13 1 0.15 1 0.14 1
SOD1 vs CEBPG 0.66 0.002 0.009 1 0.36 0.06
SOD1 vs E2F1 0.59 0.02 0.55 0.04 0.56 <0.0006
SOD1 vs E2F3 0.25 1 -0.07 1 0.09 1
SOD1 vs E2F6 0.12 1 0.14 1 0.14 1
SOD1 vs EVI1 -0.17 1 0.03 1 -0.06 1
XPA vs CEBPB 0.31 0.84 0.42 0.24 0.31 0.18
XPA vs CEBPG 0.36 0.54 0.33 0.66 0.34 0.12
XPA vs E2F1 -0.05 1 0.22 1 -0.02 1
XPA vs E2F3 -0.07 1 0.14 1 -0.01 1
XPA vs E2F6 0.55 0.04 0.46 0.12 0.4 0.02
XPA vs EVI1 0.04 1 0.07 1 0.04 1
XRCC1 vs CEBPB 0.36 0.48 0.28 1 0.32 0.18
XRCC1 vs CEBPG 0.83 <0.0006 0.27 1 0.591 <0.0006
XRCC1 vs E2F1 0.32 0.78 0.37 0.48 0.35 0.12
XRCC1 vs E2F3 0.47 0.12 0.22 1 0.35 0.06
XRCC1 vs E2F6 0.26 1 0.54 0.04 0.41 0.02
XRCC1 vs EVI1 -0.009 1 0.12 1 0.06 1
The correlation of logarithmically transformed virtually-multiplexed transcript abundance data (Table 3) from each of the six transcription factors with each of the sixteen antioxidant or DNA repair genes was determined. The transformation was necessary due to the wide range of expression of each gene among the samples. The correlation coefficient (r value) and level of significance (p value) for each correlation are presented. Significance (p < 0.01) was determined using a two-tailed test following Bonferroni adjustment for six multiple comparisons (comparison of each of six transcription factors to each of the antioxidant or DNA repair genes).
*values for E2F1 obtained in all but one BC individuals.
==== Refs
Weitberg AB Weitberg AB Cancer of the Lung: from molecular biology to treatment guidelines Current Clinical Oncology MarkmanM 2002 Totowa , Humana Press, Inc.
Baldwin PD Lung Cancer Clinical Journal of Oncology Nursing 2003 7 699 702 14705493
Page GP Green JL Lackland D Epidemiology of lung cancer with special reference to genetics, bioassays, women, and developing countries Semin Respir Crit Care Med 2000 21 365 374 16088748 10.1055/s-2000-9403
Anttila S Hirvonen A Vainio H Husgafvel-Pursiainen K Hayes JD Ketterer B Immunohistochemical localization of glutathione S-transferases in human lung Cancer Res 1993 53 5643 5648 8242618
Chu FF Esworthy RS Doroshow JH Doan K Liu XF Expression of plasma glutathione peroxidase in human liver in addition to kidney, heart, lung, and breast in humans and rodents Blood 1992 79 3233 3238 1339300
Ketterer B Harris JM Talaska G Meyer DJ Pemble SE Taylor JB Lang NP Kadlubar FF The human glutathione S-transferase supergene family, its polymorphism, and its effects on susceptibility to lung cancer Environ Health Perspect 1992 98 87 94 1486868
Freeman BA Crapo JD Biology of disease: free radicals and tissue injury Laboratory Investigation 1982 47 412 426 6290784
Seeberg EELBM The base excision repair pathway Trends in Biochemical Sciences 1995 20 391 397 8533150 10.1016/S0968-0004(00)89086-6
Crawford EL Khuder SA Durham SJ Frampton M Utell M Thilly WG Weaver DA Ferencak WJ Jennings CA Hammersley JR Olson DA Willey JC Normal bronchial epithelial cell expression of glutathione transferase P1, glutathione transferase M3, and glutathione peroxidase is low in subjects with bronchogenic carcinoma Cancer Res 2000 60 1609 1618 10749130
Zhong X Thornton K Reed E Computer based analyses of the 5'-flanking regions of selected genes involved in the nucleotide excision repair complex International Journal of Oncology 2000 17 375 380 10891549
Crawford EL Warner KA Khuder SA Zahorchak RJ Willey JC Multiplex standardized RT-PCR for expression analysis of many genes in small samples Biochemical and Biophysical Research Communications 2002 293 509 516 12054630 10.1016/S0006-291X(02)00243-7
Willey JC Coy E Brolly C Utell MJ Frampton MW Hammersley JR Thilly WG Olson DA Cairns K Xenobiotic metabolism enzyme gene expression in human bronchial epithelial and alveolar macrophage cells AmJRespCell MolBiol 1996 14 262 271
Willey JC Crawford EL Knight CR Warner KA Motten CA Herness EA Zahorchak RJ Graves TG Shimkets RA Gene expression profiling: methods and protocols-"Standardized RT-PCR and the Standardized Expression Measurement Center Methods in Molecular Biology 2004 258 Totowa , Humana Press, Inc. 14970455
Willey JC Arzimanoglou I, Balough M, Bonn G, Cohen D, Dishman R, Finkelstein S, Foster C, Logan GW, Porcella A, Regnier FE, Royyuru A, Sinskey AJ, Speicher DW Quality-controlled multi-gene expression measurement: an essential tool for the development of drugs and diagnostic tests PharmaGenomics 2004 2004 Advanstar Communications 20 33
Willey JC Crawford EL Knight CR Warner KA Motten CA Herness EA Zahorchak RJ Graves TG Shimkets RA Standardized RT-PCR and the Standardized Expression Measurement Center Methods in Molecular Biology 2004 258 Totowa , Humana Press, Inc 13 41 14970455
Workman J Mark H Comparison of Goodness of Fit Statistics for Linear Regression, Part I. Spectroscopy 2004 19 38 41
GmbH GS 2004
Quandt K Frech K Karas H Wingender E Werner T MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data Nucleic Acids Res 1995 23 4878 4884 8532532
Johnson PF Williams SC Yaniv M, Tronche F Liver Gene Expression 1994 Austin , R.G. Landes Company 231 258
Cooper C Henderson A Artandi S Avitahl N Calame K Ig/EBP (C/EBP gamma) is a transdominant negative inhibitor of C/EBP family transcriptional activators Nucleic Acids Res 1995 23 4371 4377 7501458
Gao H Parkin S Johnson PF Schwartz RC C/EBP gamma has a stimulatory role on the IL-6 and IL-8 promoters J Biol Chem 2002 277 38827 38837 12177065 10.1074/jbc.M206224200
Kaisho T Tsutsui H Tanaka T Tsujimura T Takeda K Kawai T Yoshida N Nakanishi K Akira S Impairment of natural killer cytotoxic activity and interferon gamma production in CCAAT/enhancer binding protein gamma-deficient mice J Exp Med 2005 190 1573 1582 10.1084/jem.190.11.1573
Repine JE Bast A Lankhorst I Oxidative stress in chronic obstructive pulmonary disease. Oxidative Stress Study Group AmJRespirCritCare Med 1997 156 341 357
Ren B Cam H Takahashi Y Volkert T Terragni J Young RA Dynlacht BD E2F integrates cell cycle progression with DNA repair, replication, and G(2)/M checkpoints Genes Dev 2002 16 245 256 11799067 10.1101/gad.949802
Berton TR Mitchell DL Guo R Johnson DG Regulation of epidermal apoptosis and DNA repair by E2F1 in response to ultraviolet B radiation Oncogene 2005 24 2449 2460 15735727 10.1038/sj.onc.1208462
Caporaso NE Why have we failed to find the low penetrance genetic constituents of common cancers? Cancer Epidemiol Biomarkers Prev 2002 11 1544 1549 12496041
Xi T Jones IM Mohrenweiser HW Many amino acid substitution variants identified in DNA repair genes during human population screenings are predicted to impact protein function Genomics 2004 83 970 979 15177551 10.1016/j.ygeno.2003.12.016
|
16255782
|
PMC1310620
|
CC BY
|
2021-01-04 16:03:08
|
no
|
BMC Cancer. 2005 Oct 29; 5:141
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-141
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1681630955910.1186/1471-2164-6-168Research ArticleGenes involved in complex adaptive processes tend to have highly conserved upstream regions in mammalian genomes Lee Soohyun [email protected] Isaac [email protected] Simon [email protected] Bioinformatics Program, Boston University, Boston, MA 02215, USA2 Department of Biomedical Engineering, Boston University, Boston, MA 02215, USA3 Center for Advanced Genomic Technology,. Boston University, Boston, MA 02215, USA4 Children's Hospital Informatics Program at Harvard-MIT Health Sciences and Technology, Boston, MA 02215, USA2005 27 11 2005 6 168 168 28 8 2005 27 11 2005 Copyright © 2005 Lee et al; licensee BioMed Central Ltd.2005Lee et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent advances in genome sequencing suggest a remarkable conservation in gene content of mammalian organisms. The similarity in gene repertoire present in different organisms has increased interest in studying regulatory mechanisms of gene expression aimed at elucidating the differences in phenotypes. In particular, a proximal promoter region contains a large number of regulatory elements that control the expression of its downstream gene. Although many studies have focused on identification of these elements, a broader picture on the complexity of transcriptional regulation of different biological processes has not been addressed in mammals. The regulatory complexity may strongly correlate with gene function, as different evolutionary forces must act on the regulatory systems under different biological conditions. We investigate this hypothesis by comparing the conservation of promoters upstream of genes classified in different functional categories.
Results
By conducting a rank correlation analysis between functional annotation and upstream sequence alignment scores obtained by human-mouse and human-dog comparison, we found a significantly greater conservation of the upstream sequence of genes involved in development, cell communication, neural functions and signaling processes than those involved in more basic processes shared with unicellular organisms such as metabolism and ribosomal function. This observation persists after controlling for G+C content. Considering conservation as a functional signature, we hypothesize a higher density of cis-regulatory elements upstream of genes participating in complex and adaptive processes.
Conclusion
We identified a class of functions that are associated with either high or low promoter conservation in mammals. We detected a significant tendency that points to complex and adaptive processes were associated with higher promoter conservation, despite the fact that they have emerged relatively recently during evolution. We described and contrasted several hypotheses that provide a deeper insight into how transcriptional complexity might have been emerged during evolution.
==== Body
Background
Transcription regulation is among the most sophisticated of regulatory processes, involving a complex combinatorial selection of cis- and trans-acting signals [1]. Proximal upstream regions of a gene in particular contain many cis-regulatory elements that regulate the expression of the gene by binding to various transcription factors. Many of the cis-regulatory motifs have been successfully identified by phylogenetic footprinting, which makes use of cross-species sequence conservation as a functional signature [2-6]. Based on this rationalization, we aimed to test if the complexity of transcriptional regulation depends on gene function, by looking at the sequence conservation at the proximal upstream region. This is the first whole genome study providing statistical evidence of significant conservation in upstream regions of human genes involved in certain biological processes and functions.
There have been studies on how gene function is related to degree of conservation and evolutionary rate in the protein-coding region. For example, Clark et al. [7] detected functional categories that showed accelerated evolution of the protein-coding region in human compared to chimp, based on site-specific dN/dS ratios. The functional categories include 'olfaction', 'signal transduction', 'cell adhesion', 'transport', 'developmental processes', 'ion channel' and 'extracellular matrix'. In [8], Dorus et al. reported that the protein coding regions of the genes linked to nervous system development show positive selection in the primate lineage compared to rodents, by using housekeeping genes as a control. Bustamante et al. [9] compared within-species polymorphism and between-species divergence to detect positively and negatively selected human genes. Functional categories with an excess of positively selected genes included 'sensory perception', 'defence/immunity protein' and in particular, 'transcription factor'. Some discrepancy exists among studies, since [9] detected 'ectoderm development', 'extracellular matrix' and 'voltage-gated potassium channel' as categories with an excess of negatively selected genes.
Little is known about how the proximal upstream regions evolve in those genes that are detected to be positively selected, even though it is an interesting question whether the noncoding regulatory regions will show a similar pattern as the coding regions. Interestingly, we detected a significantly higher upstream conservation in these adaptive genes, particularly those involved in development, cell communication, signal transduction, transcription factor and neurophysiological functions. One possibility is that there is opposite purifying selection on the upstream regulatory region and the protein-coding region. The other possibility is that a relatively high regulatory complexity exists in these genes and that their dense cis-regulatory elements provide higher promoter conservation. We speculate that the latter explanation is biologically more intuitive. According to [10], a complex network can rapidly achieve the ability to buffer mutations. One interesting connection we can make is that the genes in a complex network tend to evolve more rapidly in the sequence level because their mutations can be buffered more easily.
Thus, in this paper, we provide a possible insight on the combinatorial complexity of the transcriptional regulation and its evolutionary meanings of the most complex and adaptive processes such as development and cell communication.
Results
We identified functional categories that are enriched towards high proximal promoter conservation by computing a Spearman's rank correlation on upstream sequence alignment scores and Gene Ontology (GO) [11] terms.
The terms 'development', 'morphogenesis' and 'organogenesis' were found at the top of the list, followed by 'cell communication', 'signal transduction', 'transcription factor activity' and 'neurophysiological process'. Interestingly, more routine processes such as 'biosynthesis' and 'ribosome' turned out to be correlated with low upstream alignment scores (Table 1, 2, Additional file 1: Table 1). When we compared the alignment scores of negatively correlated terms vs. the most positively correlated terms, we found a 50% increase in the 1 kb, 2 kb and 5 kb alignment scores from former to the latter. The differences in alignment scores are clearly noticeable in the histograms of alignment scores of selected terms provided in Figure 1.
Table 1 Selected GO terms significantly enriched toward high 2 kb upstream alignment scores. P-values are Bonferroni-corrected. b,m,c represents GO hierarchy (b:biological process, m:molecular function, and c:cellular component). mean: mean alignment score. The mean alignment score for all the genes analyzed is 411.01.
GO accession GO term definition p-value mean # genes
GO:0007275 b development 2.69E-48 860.98 1432
GO:0009653 b morphogenesis 2.82E-47 898.06 949
GO:0009887 b organogenesis 1.50E-42 911.47 765
GO:0048513 b organ development 1.50E-42 911.47 765
GO:0007154 b cell communication 8.30E-38 799.48 2473
GO:0007165 b signal transduction 4.61E-24 791.28 1969
GO:0007399 b neurogenesis 7.04E-22 945.20 322
GO:0003700 m transcription factor activity 1.63E-20 892.63 673
GO:0050877 b neurophysiological process 4.29E-09 836.34 435
GO:0019226 b transmission of nerve impulse 7.07E-06 888.41 197
GO:0007268 b synaptic transmission 1.07E-05 889.12 192
GO:0016055 b Wnt receptor signaling pathway 4.19E-04 989.45 78
Table 2 GO terms significantly enriched toward low 2 kb upstream alignment scores. P-values are Bonferroni-corrected. b,m,c represents GO hierarchy (b:biological process, m:molecular function, and c:cellular component). mean: mean alignment score. The mean alignment score for all the genes analyzed is 411.01.
GO accession GO term description p-value mean # genes
GO:0005840 c ribosome 2.74E-05 501.18 154
GO:0030529 c ribonucleoprotein complex 2.69E-04 551.19 269
GO:0008270 m zinc ion binding 3.03E-04 639.02 1089
GO:0003723 m RNA binding 9.62E-04 579.31 410
GO:0003735 m structural constituent of ribosome 1.21E-03 531.85 174
GO:0046914 m transition metal ion binding 6.70E-03 647.13 1229
GO:0006952 b defense response 1.86E-02 612.35 694
GO:0044249 b cellular biosynthesis 2.00E-02 622.59 805
GO:0005739 c mitochondrion 2.71E-02 610.17 529
GO:0003824 m catalytic activity 3.43E-02 661.43 3900
GO:0009058 b biosynthesis 3.92E-02 625.43 840
Figure 1 Distribution of alignment scores of selected terms. Histograms of alignment scores of genes annotated with selected terms that were significantly correlated with high or low alignment scores. a: 1 kb upstream, red: neurogenesis (+), green: ribosome (-), b. 2 kb upstream, red: neurogenesis (+), green: ribonucleoprotein complex (-), c. 5 kb upstream, red: neurogenesis (+), green: ribonucleoprotein complex (-), where (+) and (-) represents positively and negatively correlated with high upstream alignment scores, respectively.
In order to confirm that the phenomenon we observed in human/mouse proximal promoter conservation is also observed with respect to other mammalian genomes and is not specific to human and mouse, we conducted a similar analysis based on human and dog genomes. The same key terms were found to be significantly correlated. (Additional file 2: Table 2). Also, we obtained very similar results by employing a different score function (SL) that penalizes sparsely distributed matches (details provided in the Method section). Thus, the significance is not an artifact of the global alignment scores that can be affected by random matches that are not functional motifs.
Among developmental genes, we noticed that 36 Hox or Hox homologue genes (listed in Additional file 3: Table 3) had very high alignment scores (mean = 699.28, 1359.2 and 3235.6, two-sample KS-test p-value (two-tailed) = 9.19 × 10-17, 1.68 × 10-17 and 2.65 × 10-17 for 1 kb, 2 kb and 5 kb upstream, respectively). Differential expression of Hox genes confers positional identities in developing cells by forming sharp boundaries along the antero-posterior axis, and the strong conservation of the promoter region is very well expected. Genes involved in embryonic patterning such as SHH and PTCH also showed high promoter conservation (2 kb scores 1262 and 1522, respectively). Figure 2 visualizes the upstream alignments of selected development genes with top alignment scores.
Figure 2 Visualization of human-mouse alignments of 2 kb upstream of selected development genes. The rightmost points are transcription start sites. a. HOXB8 (development, transcription factor), b. TBX2 (development, transcription factor), c. JARID2 (central nervous system development), d. NOG (neurogenesis, skeletal development). Blue: match, yellow: mismatch, white: gap.
Many transcription factors play important roles in development. There are a significantly large number of developmental genes among transcription factors, based on the GO annotation (215 out of 674, χ2 test p-value = 7.82 × 10-85), indicating a larger number of both trans-elements as well as cis-elements in the developmental gene network. Then, is it transcription factors that contribute most to the enrichment of development? In order to test this, we performed rank correlation tests using only developmental genes. From the analysis based on 1 kb scores, only the term 'transcription factor activity' was enriched at the top (mean = 597.37, p = 4.43 × 10-3). However, in similar analyses on 2 kb and 5 kb, no significant term was identified, indicating that genes other than transcription factors must also contribute to the significantly higher upstream conservation of development. Even for 1 kb upstream regions, the significant enrichment of 'development' is not exclusively due to transcription factor genes, because a rank correlation test without the 674 transcription factor genes also gave strong significance for development. (p = 5.24 × 10-28) (similar for 2 kb and 5 kb regions).
Cell communication is known to play an important role in development [12]. However, cell communication and signal transduction remained significant after rank tests without developmental genes, indicating the significant correlation was not because of the developmental genes that also participate in these processes.
It has been reported that ultraconserved non-coding regions (UCRs) are located around key developmental regulator genes [13,14], which may suggest a complex developmental regulatory network spanning large chromosomal regions. However, most of the UCRs are far from genes, and few proximal promoter regions overlap with UCRs, suggesting that the complexity of regulation suggested by proximal promoter conservation is independent of UCR-mediated regulation.
The alignment score can be affected by G+C content (G+C%). Mutational biases that depend on base composition may affect the baseline conservation and it is easier to get a match in an alignment with high or low G+C%. Indeed, G+C% was positively correlated with alignment scores and even with the terms 'development', 'cell communication' and 'regulation of transcription' (data not shown). It can also affect the relationship between function and promoter conservation, because promoter type depends on G+C%. Also, a high G+C% is indicative of a higher neutral substitution rate, due to a higher rate of mutation at methylated cytosines in CpG sites and some other factors [15,16]
We performed a partial rank correlation [17] between upstream alignment score and GO term label, controlling for G+C%, to eliminate the effect of G+C% in the correlation test. There was no significantly enriched term for 1 kb upstream after the partial correlation, but for 2 kb and 5 kb, the key terms detected remained, indicating G+C% generally had little effect on our results (Additional file 4: Table 4).
A high upstream G+C% may also reflect a potential CpG island, which plays a role in transcriptional repression in a variety of types of cells and tissues. To compare regulation by CpG island methylation and regulation via other cis-regulatory elements, we performed a rank correlation test between CpG dinucleotide frequency in 1 kb, 2 kb and 5 kb upstream regions of human genes and GO term label. The result was somewhat different from the alignment score/functional label analysis, in that terms related to metabolism, cell cycle and transcription were found to be positively correlated and terms related to response to external signal, immune response, membrane and extracellular matrix were negatively correlated with high CpG frequency (Additional file 5: Table 5).
Discussion
A commonly invoked heuristic for discovery of functional sites is locating regions of high similarity across multiple species. Thus, a relatively high proportion of such conserved regions may indicate an increased number of functional cis-elements. This in turn may suggest a more complex combinatorial circuitry in the transcriptional regulatory network, since higher density of functional cis-elements will allow more combinations of trans-acting signals as well.
In this context, our results indicate the existence of a relatively high complexity in the transcriptional regulation of development, cell communication, signal transduction, transcriptional regulation and neurophysiological function as compared to those of ribosomes and metabolism. This explanation is supported by prior gene-specific studies. For example, the sea urchin gene Endo16 has been found to have a dense distribution of cis-elements. Endo16 is expressed in the endoderm and may play a role in cell adhesion. Its upstream sequence has been well characterized to have complex cis-regulatory modules [18] and the proximal promoter region was shown to be highly conserved during evolution [19].
However, a caveat of the analysis is the interpretation of alignment data. A match in an alignment is not an accurate indicator of either purifying selection or functionality of that site, since non-functional sites under neutral evolution can remain unchanged by vertical inheritance or multiple/reverse mutation. However, the effect of neutral mutation can be ignored in this analysis. Neutral mutations that have occurred between human and mouse lineage can be considered to be saturated, because of the sufficiently high evolutionary distance between these species [20]. The G+C%-partialling analysis described in the Result section also suggests that the difference in neutral substitution rate has little effect on our results.
Stronger purifying selection on the promoter cis-elements of complex genes is another alternative explanation. It is not easy to tell if a conserved region is from a single highly important functional site or from overlaps of less important functional sites. An increased percent identity in a genomic region may indicate more important functional elements (stronger purifying selection) rather than a larger number of functional elements residing in that region.
In the classic Waddington's canalization theory, purifying (e.g. stabilizing) selection is a driving force for developmental genes to attain robustness against genetic and environmental changes [21]. In this context, developmental genes may undergo stronger purifying selection than others (Plotkin JB, personal communication). However, a recent view provided by [10] denies the necessity of stabilizing selective pressure to attain robustness for a system that can be represented as an interacting network. Developmental [10] and nervous system [22] can be represented as a network and thus the genes involved in these systems can intrinsically achieve the buffering ability. The finding that protein-coding regions of these genes undergo rapid evolution in humans [7-9] may indicate that the complexity of their network can buffer the mutations in the sequence level more easily. Our regulatory complexity hypothesis is consistent with the network implementation of canalization in developmental genes. Multiple regulatory sites interacting with increased number of trans-factors can achieve increased network connectivity. Although we find this explanation more biologically intuitive than the hypothesis that the higher purifying selection on the genes that has positive evolution in the protein-coding region, purifying selection and regulatory complexity might not be entirely disjoint and both factors can contribute to the conservation of promoter sequences as well as to their buffering capacity.
Another alternative hypothesis is that the transcriptional control of development genes is modulated by a family of transcription factors that require an increased specificity or a longer binding site. Indeed, many developmental regulators act as dimers that can take up larger areas on DNA. However, dimerization can also be considered a part of gene regulation (e.g. different heterodimeric combinations, dominant negative-type repression, etc.) and therefore would support the complex regulation hypothesis.
Recently, a similar study [23] was performed in yeast (that lacks developmental, neural or complex cell communication mechanisms) and it was reported that steroid, alcohol and carbohydrate metabolisms tend to be associated with higher promoter conservation in yeast. However, after a more careful examination of the entire GO term list provided in the Supplementary Material of [23], we found that the terms 'transcription factor activity', 'signal transduction', 'cell communication' and 'cellular morphogenesis' were associated with higher promoter conservation, whereas 'structural constituents of ribosome', 'DNA recombination' and 'RNA processing' had insignificant but negative enrichment, showing a consistent pattern with our results in mammals.
It has been suggested that a large fraction of genetic components in the evolution of development involves changes in transcriptional regulation [24]. One possible explanation of the link between complex regulation and adaptation is that adaptive changes in complex processes tend to occur in an incremental fashion, by slowly adding to its regulatory complexity, so that other important parts of the process remain functionally intact.
Conclusion
We identified a class of functions that are associated with either high or low promoter conservation in mammals. We detected a significant tendency that complex and adaptive processes were associated with higher promoter conservation, despite the fact that they have emerged relatively recently during evolution. We described and contrasted several hypotheses that provide a deeper insight into how transcriptional complexity might have been emerged during evolution.
Methods
Promoter alignment and GO term labelling
To measure proximal promoter conservation, upstream alignment scores of human genes were computed by counting the number of matches in the alignment between the sequence 1 kb, 2 kb and 5 kb upstream of the transcription start site and its syntenic mouse sequence. In the cases in which a human gene had no recorded mouse counterpart, the score was set to 0. The alignments were obtained directly from the UCSC genome browser [25] (version Jun-2003). Genes with multiple promoter assignments were not included in the analysis, to avoid bias due to alternative promoters or incorrect annotation of transcription start sites. Consequently, 14449, 14434, 14412 genes were assigned alignment scores for 1 kb, 2 kb and 5 kb upstream, respectively.
GO term labeling was done as follows: for each of the 17594 non-obsolete GO terms, each gene was labeled 1 if it is annotated with the term itself or any of its descendants and 0 otherwise. Thus, each term was represented by a binary vector whose size is the number of the genes. We used the latest version of GO annotation downloaded on March 2005. The alignment scores and GO annotations of all the genes analyzed can be found in Additional file 6: Table 6.
Rank correlation test
A Spearman rank correlation test [26] was performed on each GO term vector and the upstream alignment score vector, in order to test whether each GO term is associated with high or low proximal upstream conservation. Two-tailed p-values were calculated using Student's t-distribution. All the p-values provided in this paper and additional files are Bonferroni-corrected and the significance level used was α = 0.05.
Human vs. dog
We used multi-species (human, chimp, dog, mouse, rat, chicken, zebrafish and fugu) upstream alignments downloaded from the UCSC genome browser (version May 2004) and extracted the human and dog alignment. The scoring scheme and correlation method were the same as in the human-mouse analysis described above.
Modification of the alignment scores
Throughout the study described in the paper, we have used a simple global alignment score that might fail to capture a variety of features that represent patterns of conservation such as locally clustered functional regions. Thus, we tried a modified alignment score SL that penalizes sparsely distributed matches compared to locally enriched 'match blocks'.
SL = (# of matches) – (# of non-match blocks) + 1,
where a non-match block means consecutive runs of non-matches (mismatches or gaps) flanked by matches.
The last 1 is added to shift the score to be nonnegative. This score becomes 0 either when there is no match or all the matches are of length 1. The longer the blocks of matches are, the larger SL is.
Authors' contributions
SL carried out the analysis and drafted the initial manuscript. SL, IK and SK interpreted the data and wrote portions of the final manuscript. SK coordinated the research. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
Table 1 GO terms significantly correlated with upstream alignment scores. The table lists all the GO terms significantly correlated with 1 kb, 2 kb and 5 kb upstream alignment scores in the human-mouse analysis.
Click here for file
Additional File 2
Table 2 GO terms significantly correlated with upstream alignment scores in the analysis using dog. The table lists all the GO terms significantly correlated with 1 kb, 2 kb and 5 kb upstream alignment scores in the human-dog analysis.
Click here for file
Additional File 3
Table 3 List of Hox and Hox homolog genes. List of Hox and Hox homolog genes used in the analysis, with their 1 kb, 2 kb and 5 kb upstream alignment scores.
Click here for file
Additional File 4
Table 4 GO terms significantly correlated with upstream alignment scores after partialling out G+C%. The table lists all the GO terms significantly correlated with 1 kb, 2 kb and 5 kb upstream alignment scores in the human-mouse analysis after controlling G+C% using partial rank correlation.
Click here for file
Additional File 5
Table 5 GO terms significantly correlated with CpG dinucleotide frequency in the human upstream sequence. The table lists all the GO terms significantly correlated with CpG dinucleotide frequency in the 1 kb, 2 kb and 5 kb human upstream sequences. Genes are sorted by 1 kb upstream alignment score.
Click here for file
Additional File 6
Table 6 Human genes with their upstream alignment scores and Gene Ontology annotations. The 1 kb, 2 kb and 5 kb upstream alignment scores based on human-mouse comparison and Gene Ontology annotations of human genes.
Click here for file
Acknowledgements
We thank Dr. Stan Letovsky, Dr. Joshua Plotkin and Dr. Shamil Sunyaev at Harvard University for valuable advice and Michael Schaffer at Boston University for providing the visualization of upstream sequence alignments. We also thank the UCSC Genome Bioinformatics Group, for maintaining the UCSC genome browser. We thank the reviewers for providing thoughtful suggestions and Michael Schaffer, Sharona Thompson and John Rachlin at Boston University for revising the final manuscript. This work is supported in part by NSF grants DBI-0239435 and ITR-048715, NHGRI grant #1R33HG002850-01A1 and NIH grant U54 LM008748.
==== Refs
Pilpel Y Sudarsanam P Church GM Identifying regulatory networks by combinatorial analysis of promoter elements Nature Genet 2001 29 153 159 11547334 10.1038/ng724
Gumucio DL Heilstedt-Williamson H Gray TA Tarle SA Shelton DA Tagle DA Slightom JL Goodman M Collins FS Phylogenetic footprinting reveals a nuclear protein which binds to silencer sequences in the human gamma and epsilon globin genes Mol Cell Biol 1992 12 4919 4929 1406669
Lenhard B Sandelin A Mendoza L Engström P Jareborg N Wasserman WW Identification of conserved regulatory elements by comparative genome analysis J Biol 2003 2 13 12760745 10.1186/1475-4924-2-13
Liu Y Liu XS Wei L Altman RB Batzoglou S Eukaryotic regulatory element conservation analysis and identification using comparative genomics Genome Res 2004 14 451 458 14993210 10.1101/gr.1327604
Sandelin A Wasserman WW Lenhard B ConSite: web-based prediction of regulatory elements using cross-species comparison Nucleic Acids Res 2004 32 W249 W252 15215389 10.1093/nar/gkh012
Wasserman WW Palumbo M Thompson W Fickett JW Lawrence CE Human-mouse genome comparisons to locate regulatory sites Nature Genet 2000 26 225 228 11017083 10.1038/79965
Clark AG Glanowski S Nielsen R Thomas PD Kejariwal A Todd MA Tanenbaum DM Civello D Lu F Murphy B Ferriera S Wang G Zheng X White TJ Sninsky JJ Adams MD Cargill M Inferring nonneutral evolution from human-chimp-mouse orthologous gene trios Science 2003 302 1960 1963 14671302 10.1126/science.1088821
Dorus S Vallender EJ Evans PD Anderson JR Gilbert SL Mahowald M Wyckoff GJ Malcom CM Lahn BT Accelerated evolution of nervous system genes in the origin of Homo sapiens Cell 2004 119 1027 1040 15620360 10.1016/j.cell.2004.11.040
Bustamante CD Fledel-Alon A Williamson S Nielsen R Hubisz MT Glanowski S Tanenbaum DM White TJ Sninsky JJ Hernandez RD Civello D Adams MD Cargill M Clark AG Natural selection on protein-coding genes in the human genome Nature 2005 437 1153 1157 16237444 10.1038/nature04240
Siegal ML Bergman A Waddingtons' canalization revisited: developmental stability and evolution Proc Natl Acad Sci U S A 2002 99 10528 10532 12082173 10.1073/pnas.102303999
The Gene Ontology Consortium Gene Ontology: tool for the unification of biology Nature Genet 2000 25 25 29 10802651 10.1038/75556
Gilbert SF Developmental Biology 2000 6 Sinauer Associates 10993670
Sandelin A Bailey P Bruce S Engström PG Klos JM Wasserman WW Ericson J Lenhard B Arrays of ultraconserved non-coding regions span the loci of key developmental genes in vertebrate genomes BMC Genomics 2004 5 99 107 15613238 10.1186/1471-2164-5-99
Woolfe A Goodson M Goode DK Snell P McEwen GK Vavouri T Smith SF North P Callaway H Kelly K Walter K Abnizova I Gilks W Edwards YJK Cooke JE Elgar G Highly conserved non-coding sequences are associated with vertebrate development PLoS Biol 2005 3 e7 15630479 10.1371/journal.pbio.0030007
Hurst LE Williams EJB Covariation of GC content and the silent site substitution rate in rodents: implications for methodology and for the evolution of isochores Gene 2000 261 107 114 11164042 10.1016/S0378-1119(00)00489-3
Piganeau G Mouchiroud D Duret L Gautier C Expected relationship between the silent substitution rate and the GC content: implications for the evolution of isochores J Mol Evol 2002 54 129 133 11734906
Wherry RJ Contributions to correlational analysis 1984 Academic Press
Yuh C-H Bolouri H Davidson EH Genomic cis-regulatory logic: experimental and computational analysis of a sea urchin gene Science 1998 279 1896 1902 9506933 10.1126/science.279.5358.1896
Romano LA Wray GA Conservation of Endo16 expression in sea urchins despite evolutionary divergence in both cis and trans-acting components of transcriptional regulation Development 2003 130 4187 4199 12874137 10.1242/dev.00611
Smith NGC Webster MT Ellegren H Deterministic mutation rate variation in the human genome Genome Res 2002 12 1350 1356 12213772 10.1101/gr.220502
Waddington CH Canalization of development and the inheritance of acquired characters Nature 1942 150 563 565
Niven JE Channelling evolution: canalization and the nervous system PLoS Biol 2004 2 e19 14737195 10.1371/journal.pbio.0020019
Chin C Chuang JH Li H Genome-wide regulatory complexity in yeast promoters: Separation of functionally conserved and neutral sequence Genome Res 2005 15 205 213 15653830 10.1101/gr.3243305
Wary GA Transcriptional regulation and the evolution of development Int J Dev Biol 2003 47 7 8 10.1387/ijdb12
The UCSC Genome Browser
Kendall MG Rank correlation methods 1970 4 Griffin
|
16309559
|
PMC1310621
|
CC BY
|
2021-01-04 16:32:47
|
no
|
BMC Genomics. 2005 Nov 27; 6:168
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-168
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2731629319110.1186/1471-2105-6-273Research ArticleAncestral sequence alignment under optimal conditions Hudek Alexander K [email protected] Daniel G [email protected] School of Computer Science, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, N2L 3G1, Canada2005 17 11 2005 6 273 273 28 7 2005 17 11 2005 Copyright © 2005 Hudek and Brown; licensee BioMed Central Ltd.2005Hudek and Brown; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Multiple genome alignment is an important problem in bioinformatics. An important subproblem used by many multiple alignment approaches is that of aligning two multiple alignments. Many popular alignment algorithms for DNA use the sum-of-pairs heuristic, where the score of a multiple alignment is the sum of its induced pairwise alignment scores. However, the biological meaning of the sum-of-pairs of pairs heuristic is not obvious. Additionally, many algorithms based on the sum-of-pairs heuristic are complicated and slow, compared to pairwise alignment algorithms.
An alternative approach to aligning alignments is to first infer ancestral sequences for each alignment, and then align the two ancestral sequences. In addition to being fast, this method has a clear biological basis that takes into account the evolution implied by an underlying phylogenetic tree.
In this study we explore the accuracy of aligning alignments by ancestral sequence alignment. We examine the use of both maximum likelihood and parsimony to infer ancestral sequences. Additionally, we investigate the effect on accuracy of allowing ambiguity in our ancestral sequences.
Results
We use synthetic sequence data that we generate by simulating evolution on a phylogenetic tree. We use two different types of phylogenetic trees: trees with a period of rapid growth followed by a period of slow growth, and trees with a period of slow growth followed by a period of rapid growth.
We examine the alignment accuracy of four ancestral sequence reconstruction and alignment methods: parsimony, maximum likelihood, ambiguous parsimony, and ambiguous maximum likelihood. Additionally, we compare against the alignment accuracy of two sum-of-pairs algorithms: ClustalW and the heuristic of Ma, Zhang, and Wang.
Conclusion
We find that allowing ambiguity in ancestral sequences does not lead to better multiple alignments. Regardless of whether we use parsimony or maximum likelihood, the success of aligning ancestral sequences containing ambiguity is very sensitive to the choice of gap open cost. Surprisingly, we find that using maximum likelihood to infer ancestral sequences results in less accurate alignments than when using parsimony to infer ancestral sequences. Finally, we find that the sum-of-pairs methods produce better alignments than all of the ancestral alignment methods.
==== Body
Background
Multiple genome alignment is an important problem in bioinformatics. It is used in comparative studies to help find new genomic features such as genes and regulatory elements. Current multiple genome alignment programs [1,2] use progressive alignment [3], with a phylogenetic tree as a reference.
The primary operation of progressive alignment is the alignment of two multiple alignments. Most genome aligners have two main phases: anchoring and aligning between the anchors. Here, we focus on the algorithms used to align between anchors. Many popular alignment algorithms for DNA use the sum-of-pairs heuristic, where the score of a multiple alignment is the sum of the induced pairwise alignment scores. However, Just [4] has shown that finding the optimal alignment of two multiple alignments under the sum-of-pairs heuristic is NP-hard. Of course, since the problem is important, numerous heuristic algorithms [5-7] exist for alignment of alignment under sum of pairs.
The biological meaning of the sum-of-pairs of pairs heuristic is not obvious. Additionally, many heuristic algorithms are complicated and slow, compared to pairwise alignment algorithms [5-7]. An alternative approach to the strategy of aligning alignments under the sum-of-pairs heuristic is to first infer ancestral sequences for each alignment and then align the two ancestral sequences. In addition to being fast, this method has a clear biological basis that takes into account the evolution implied by the underlying tree.
Bray and Pachter use this approach to align alignments in MAVID [1]. MAVID uses maximum likelihood to infer ancestral sequences for alignments. In our previous work [8], we infer ancestral sequences for anchoring multiple alignments, but not for aligning alignments. Instead of maximum likelihood, we use parsimony to infer ancestral sequences, but we also allow these sequences to keep some ambiguity. The idea behind allowing ambiguity is to retain as much information about the underlying multiple alignment as possible. A natural question is whether aligning such ambiguous ancestral sequences leads to better alignments than aligning unambiguous ancestral sequences.
In this study, we explore this idea as well as other aspects of aligning alignments by ancestral sequence inference. We compare four ancestral alignment methods with two sum-of-pairs alignment algorithms. We infer ancestral sequences using parsimony and maximum likelihood, and study the effect of allowing ambiguity in these sequences. Since we are interested in the performance of these methods under optimal conditions, we use data generated by a very simple evolution simulation. For aligning full alignments with the sum-of-pairs heuristic, we use ClustalW [5] and a newer heuristic by Ma, Wang, and Zhang [6].
We find that alignment algorithms based on the sum-of-pairs heuristic are more accurate than all of our methods based on ancestral sequence alignment. However for alignment of inferred ancestral sequences, parsimony outperforms maximum likelihood in this application. Using maximum likelihood to infer ancestral sequences results in final alignment accuracies that are more unpredictable. Also, computing log-odds for ancestral sequences inferred with maximum likelihood is far more computationally intensive than computing log-odds scores for ancestral sequences inferred with parsimony. Finally, we find that allowing ancestral sequences to have ambiguity does not result in more accurate final alignments.
Results
To determine whether using ambiguous symbols in ancestral sequences inference improves multiple alignment, we have performed experiments on simulated sequences. We propose five hypotheses, explain our experimental method, and finally discuss results and give conclusions.
Our hypotheses
The first hypothesis is that by using ancestral sequences with ambiguity, we obtain more accurate multiple alignments. Ambiguous symbols may allow us to retain more information about the underlying multiple alignments, which may make it easier to identify matching positions. Combined with an appropriate log-odds scoring system, this extra information may allow for more accurate alignment of ancestral sequences, and by extension, for more accurate multiple alignments.
Our second hypothesis is that alignment of ancestral sequences is more sensitive to gap open costs than alignment of alignments using the sum-of-pairs heuristic. When aligning ancestral sequences, existing gaps in the underlying alignment are not considered when inserting a new gap, so the first position of a new gap always costs the gap open cost. Incorrect gap penalties may cause too many gaps to be inserted between ancestral sequences. During progressive alignment, errors at each step propagate leading to an incorrect final alignment. In contrast, when aligning alignments using the sum-of-pairs heuristic, the cost of adding a new gap depends on all underlying gaps as well as the gap open cost. An incorrect gap open cost affects the cost of gaps less and new gaps may still be correctly inserted based on the structure of the existing gaps.
Our third hypothesis is that the function used to estimate the gap open cost during progressive alignment is important to alignment accuracy when aligning ancestral sequences. When aligning ancestral sequences, the frequencies of gaps in the ancestral sequences depends on the amount of mutation between the sequences. Therefore, it is important to modify the gap open cost based on the distance between the ancestral sequences being aligned.
Our fourth hypothesis is that we expect that aligning alignments using the sum-of-pairs heuristics gives more accurate multiple alignments than aligning inferred ancestral sequences. There are two reasons for this. First, as stated in hypothesis two, using an incorrect gap open cost affects the sum-of-pairs heuristic less than it affects the alignment of ancestral sequences. Since choosing the correct gap open cost can be difficult in practice, we expect that aligning alignments using the sum-of-pairs heuristic results in a more accurate final alignment because it is less sensitive to this parameter. Additionally, the ancestral sequences we infer are not completely accurate, which compounds the errors made in the process of progressive alignment. Thus, the much slower run times of algorithms based on the sum-of-pairs heuristic are acceptable.
Finally, we expect that the maximum likelihood methods result in better multiple alignments than the parsimony methods. Unlike parsimony, maximum likelihood uses the edge distances on the phylogenetic tree. Thus we expect maximum likelihood to better infer ancestral sequences.
Experimental data
We use synthetic data in order to have correct alignments to test our methods against. Additionally, by generating our own data we ensure that the data is generated from the same model of evolution that is required for the alignment algorithm. Therefore, we consider the performance of the algorithms on this data to be the the best possible for algorithms of their type.
Phylogenetic trees
Despite our use of synthetic data, we want our data to mimic the basic properties of real biological sequence. Thus we generate random trees that resemble real trees, and assign mutation rates based on analysis of real sequences.
Specifically, we are interested in algorithm performance on two different types of random trees: trees with a period of heavy growth followed by a period of no growth, and trees with a period of light growth followed by a period of heavy growth. The first type of tree, which we refer to as early growth, is similar to the tree of placental mammals from Eizirik, Murphy, and O'Brien [9]. The second type of tree, which we refer to as late growth, is similar to the type of tree formed in a coalescent process of neutral mutation and speciation. We generate two sets of twenty phylogenetic trees, one set for each tree type. Each individual tree has eight taxa. We have chosen to limit our trees to trees with four taxa in each subtree of the root, both for ease of programming and to keep the time required to compute the log-odds scores for maximum likelihood reasonable.
We generate these in two steps. First, we generate one large tree of each type using a random birth-death process implemented in Phyl-O-Gen v1.2 [10]. To generate the early growth tree we start with a pure birth process with a birth rate of 0.35 events per million years. After we produce 100 lineages we switch to a birth-death process with both the birth and death rate set to 0.04 per million years. This episode lasts seven times as long as the first episode. We sample 66 lineages from the result, and this becomes our final tree, in Figure 3. We chose these values such that the tree resembles the tree of placental mammals both in topology and time scale. To generate the late growth tree we start with a pure birth process with a birth rate of 0.01 speciation events per million years. After eight speciation events, we switch to a birth rate of 0.20 speciation events per million years. We continue the process until we obtain 100 lineages. Again, we obtain a final tree by sampling 66 lineages from the result of the random process. This tree is shown in Figure 4. Here we choose values such that the tree has the same time scale and taxa as the early growth tree, but with a different topology.
Figure 1 Example random trees. Example of a random tree with early growth (A) and a random tree with late growth (B).
Figure 2 Plot of column scores against mean pairwise column scores. Plot of column accuracy versus mean pairwise column accuracy for alignments from all experiments. The column accuracy and the mean pairwise column accuracy have a roughly linear relationship.
Figure 3 Phylogenetic tree with early growth. Phylogenetic tree with a period of rapid growth followed by a period of slow growth. This tree resembles the tree of placental mammals and the distance from the root to taxa is approximately 100 million years.
Figure 4 Phylogenetic tree with late growth. Phylogenetic tree with a period of slow growth followed by a period of rapid growth. The distance from root to taxa is approximately 100 million years.
From each final tree, using the method of Kearney, Munro, and Phillips [11], we randomly sample subtrees of eight taxa subject to the constraint that the amount of simulated time from the root to each taxon is between 90 to 110 million years. We then filter the resulting trees and only keep trees where the the left and right children have four descendants each. Examples of two of these trees are in Figure 1.
Random sequences
For each of the eight taxon trees in our two data sets, we generate twenty random sequence sets by simulating evolution over the tree. We use a program written by us, but similar to ROSE [12], to generate our sequences. For a given input tree, our program starts with an initial random sequence and mutates that sequence into new sequences down the tree. The program simulates Jukes/Cantor mutation events [13] as well as geometrically distributed insertion and deletion events with a mean length of 5.6. At the end, we have a set of sequences suitable as input to a multiple alignment program, and we have the original multiple alignment of the sequences.
Since we wanted our mutations, insertions, and deletions to be as close to real sequences as possible, we calibrated our simulator with parameters estimated through analysis of homologous human and baboon sequences. This gives us two sets of 400 random input sequences, one for each set of trees.
We chose the CFTR region in human and baboon for this parameter estimation, so that we could ensure that our alignment is mostly correct. We obtained human and baboon sequences with repeats masked out from the NISC Comparative Vertebrate Sequencing project [14]. We aligned a 10 kB region from each sequence and trimmed the ends of the alignments to obtain a final good alignment. From this final alignment, we measure the number of mutations as well as the length and number of gaps. Assuming that humans and baboons diverged approximately 25 million years ago (MYA), [15], and using the equation
Pr[mutation] = 3/4 (1 - e-rαt), (1)
where t is the time in millions of years [13], we estimate the mutation rate α to be 7.1 × 10-4 mutations per site per million years. We observe a rate of 4.1 × 10-3 insertions or deletions per site. Assuming that insertions and deletions are equally likely and that we have only one possible insertion or deletion at a particular site, we find that Pr[insertion] = Pr[deletion] = 4.1 × 10-5 events per site per million years. Additionally, we create another two sets of random input sequences using the same sets of trees, but with double the mutation rate on each tree branch.
Experimental methods
We use the same insertion and deletion rates from our evolution simulator to compute the log-odds scores for the maximum likelihood methods. For a mean gap length of 5.6, we compute the optimal gap extension penalty to be 0.57 by standard methods [13]. When aligning alignments using the standard sum-of-pairs heuristic, any single gap cost is wrong for many pairs. Therefore, we should ideally use a different gap cost for each pair. However, as this increases the time complexity of Ma, Wang, and Zhang's algorithm, we instead use a single, unscaled gap cost for all pairwise alignments.
We test our first hypothesis by running all ancestral alignment methods and sum-of-pairs methods on all data sets using the optimal gap extension costs and a base gap open cost of 7. We use the Expected gap open cost estimation function since later tests show it to be better than the Max estimation function. To test our second hypothesis, we expand the previous test by exploring gap costs of 5 and 9. We test our third hypothesis by using two different scaling methods for each of the ancestral alignment methods. Our fourth and fifth hypothesis are also answered by the above three tests.
Measuring success
We take the fraction of correct columns in an alignment to be the measure the alignment's accuracy. A correct column is one which contains the exact same nucleotides, from the same positions in the sequences, as a column in the correct alignment; a column with the same bases as a correct column, but from different positions, is incorrect. For a given data set and algorithm, we use the mean alignment accuracy of all 400 sequence data sets as a measure of the algorithm's accuracy.
We also considered an accuracy measure based on the pairwise alignments induced by the multiple alignment. In this, we compute the alignment accuracy of all induced pairwise alignments as in the previous paragraph and take the mean of these.
Discussion
Before we discuss our results, we compare the two accuracy measures: column accuracy and mean pairwise column (MPC) accuracy. Tables 1 and 4 show that both measures have the same trends. Additionally, Figure 2 shows that the column accuracies and the MPC accuracies have a roughly linear relationship; the MPC accuracies are strictly higher than the column accuracies. Therefore, we do not give MPC accuracies for the experimental results in Tables 2 and 3. We now describe our experimental results with respect to our five hypotheses.
Table 1 Alignment accuracies using correct gap costs. Alignment accuracies using a gap open cost of 7 and the optimal gap extension cost of 0.57. P-values are computed using a paired Student's t-test.
Data Set Measure Parsimony Ambiguous Parsimony ML Ambiguous ML Ma et al. ClustalW
Early Growth Mean 88.43% 83.83% 87.86% 86.26% 91.74% 91.25%
Std. 2.89% 5.94% 2.79% 4.27% 1.69% 2.28%
P-value 1.6461 × 10-65 1.1992 × 10-17 N/A
Early Growth Double Length Mean 64.68% 57.24% 63.54% 59.89% 74.29% 72.35%
Std. 5.23% 8.25% 4.99% 7.59% 4.62% 4.30%
P-value 1.5922 × 10-103 1.9161 × 10-21 N/A
Late Growth Mean. 95.76% 96.01% 89.44% 87.52% 96.63% 96.79%
Std 1.29% 1.28% 5.44% 7.03% 0.99% 1.10%
P-value 7.7818 × 10-10 1.0404 × 10-12 N/A
Late Growth Double Length Mean 86.71% 88.12% 64.76% 55.67% 89.68% 89.54%
Std. 2.71% 2.85% 7.94% 12.62% 2.42% 2.11%
P-value 2.1699 × 10-43 2.5033 × 10-50 N/A
Table 2 Alignment accuracies for differing gap open costs. Alignment accuracies for various gap open costs using a gap extension cost of 0.57.
Data Set Gap Open Cost Parsimony Ambiguous Parsimony ML Ambiguous ML Ma et al. ClustalW
Early Growth 5 89.27% 88.41% 88.43% 88.09% 92.77% 90.48%
7 88.43% 83.83% 87.86% 86.26% 91.74% 91.25%
9 86.17% 76.04% 86.13% 82.58% 90.42% 91.48%
Change 3.10% 12.37% 2.3% 5.51% 2.35% 1.00%
Early Growth Double Length 5 63.43% 63.32% 63.01% 63.05% 78.19% 69.63%
7 64.68% 57.24% 63.54% 59.89% 74.29% 72.35%
9 61.40% 46.72% 60.02% 52.33% 68.44% 73.59%
Change 2.03% 16.6% 2.99% 10.72% 9.75% 3.96%
Late Growth 5 95.95% 96.35% 92.06% 91.70% 96.97% 96.50%
7 95.76% 96.01% 89.44% 87.52% 96.63% 96.79%
9 95.21% 95.39% 84.95% 80.71% 96.12% 96.91%
Change 0.74% 0.96% 7.11% 10.99% 0.85% 0.41%
Late Growth Double Length 5 86.21% 88.91% 73.21% 68.18% 91.43% 88.17%
7 86.71% 88.12% 64.76% 55.67% 89.68% 89.54%
9 85.58% 86.16% 54.36% 43.02% 87.08% 90.15%
Change 0.63% 2.75% 18.85% 25.16% 4.35% 1.98%
Table 3 Alignment accuracies using different gap open cost scaling functions. Alignment accuracies using two different gap cost scaling functions. The unsealed gap open cost is 7 and the unsealed gap extension cost is 1. The Max method scales gap open costs according to the maximum value in the scoring matrix. The Expected method scales gap open costs according to the expected score of a related symbol pair from the two ancestral sequences.
Data Set Gap Scaling Method Parsimony Ambiguous Parsimony ML Ambiguous ML
Early Growth Max 86.92% 85.07% 86.59% 86.13%
Expected 89.95% 89.79% 89.17% 89.31%
Early Growth Double Length Max 66.10% 64.51% 67.17% 66.10%
Expected 71.12% 73.48% 72.02% 73.81%
Late Growth Max 95.02% 95.21% 93.52% 92.31%
Expected 96.04% 96.35% 95.03% 94.68%
Late Growth Double Length Max 85.95% 86.39% 82.74% 75.98%
Expected 88.38% 89.91% 87.17% 85.35%
Table 4 Mean pairwise alignment accuracies using correct gap costs. Mean pairwise alignment accuracies using a gap open cost of 7 and the optimal gap extension cost of 0.57. P-values are computed using a paired Student's t-test.
Data Set Measure Parsimony Ambiguous Parsimony ML Ambiguous ML Ma et al. ClustalW
Early Growth Mean 95.64% 92.61% 96.19% 95.41% 97.61% 97.48%
Std. 1.43% 3.47% 1.08% 1.98% 0.48% 0.64%
P-value 1.3830 × 10-67 9.6503 × 10-17 N/A
Early Growth Double Length Mean 84.33% 78.29% 84.30% 82.19% 91.20% 90.58%
Std. 2.99% 5.50% 2.91% 4.95% 1.75% 1.51%
P-value 9.1586 × 10-114 3.5959 × 10-18 N/A
Late Growth Mean 98.13% 98.26% 94.73% 93.74% 98.65% 98.67%
Std. 0.62% 0.62% 3.14% 4.05% 0.39% 0.44%
P-value 5.9142 × 10-09 1.1169 × 10-10 N/A
Late Growth Double Length Mean 93.63% 94.45% 80.79% 75.51% 95.50% 95.44%
Std. 1.37% 1.47% 4.85% 7.91% 1.14% 0.95%
P-value 2.1676 × 10-44 6.5523 × 10-47 N/A
It is not clear that including ambiguity in ancestral sequences improves alignment. Looking at Tables 1 and 4, we see that on more than half the data sets, the ambiguous versions of parsimony and maximum likelihood have lower mean accuracies than their unambiguous counterparts. Additionally, ambiguous methods are less consistent in their scores, as evidenced by the larger standard deviations in the results for the ambiguous methods.
Table 2 shows mean alignment accuracies for the different data sets using a gap opening costs of 5, 7, and 9. For each data set and method, we measure the change in mean accuracy from a gap open cost of 5 to a gap open cost of 9. It is clear that the mean accuracy of the ambiguous methods changes far more than the mean accuracy of the unambiguous methods. Looking at the difference in mean accuracy of the ambiguous methods versus the unambiguous methods, we again see that the ambiguous methods are often lower than the unambiguous methods. Therefore, we conclude that the ambiguous methods are more sensitive to the gap open cost than the unambiguous methods.
Our experiment confirmed our hypothesis that the gap cost scaling function is very important to the resulting alignment accuracies. When changing from scaling based on the largest value in the ancestral sequence scoring matrix to the expected cost for related positions, we see a significant increase in alignment accuracy on all data sets and all methods. See Table 3 for results.
In Tables 1, 2, and 4, we see that both Ma, Wang, and Zhang's algorithm and ClustalW perform consistently better than the ancestral alignment methods. Also, Ma et al.'s algorithm performs better than ClustalW in most cases. However, the gap open cost affects Ma et al.'s algorithm and ClustalW differently. Depending on the choice of gap open cost, ClustalW may perform better than Ma et al.'s algorithm. Looking at all examined gap costs, Ma et al.'s algorithm achieves the highest alignment accuracy on each data set.
Surprisingly, the maximum likelihood methods performed worse than parsimony methods in the context of ancestral alignment. In Tables 1, 2, and 4, unambiguous maximum likelihood always performs worse than unambiguous parsimony. In some cases, such as on the Late Growth data set in Table 1, maximum likelihood performs drastically worse obtaining a mean alignment accuracy of 64.76% versus 86.71%. Looking at the mean alignment accuracy alone, it is not clear that ambiguous maximum likelihood is worse than ambiguous parsimony. However, on the late growth data set ambiguous parsimony does much worse than ambiguous parsimony. On other data sets, it performs similarly. Therefore, we conclude that ambiguous parsimony is more reliable than ambiguous maximum likelihood, if not more accurate.
Conclusion
We have tested four ancestral alignment methods as well as two sum-of-pairs alignment methods on simulated data. The data mimics evolution on two types of evolutionary trees: trees with a period of rapid growth followed by a period of slow growth, and trees with a period of slow growth followed by a period of rapid growth. The four ancestral alignment methods we have tested are unambiguous parsimony, ambiguous parsimony, unambiguous maximum likelihood, and ambiguous maximum likelihood. The sum-of-pairs alignment methods we have tested are the ClustalW [5] algorithm and the algorithm of Ma, Wang, and Zhang [6].
We have found that, contrary to our hypotheses, allowing ambiguity in ancestral sequences does not lead to better alignments. When we use ambiguous ancestral sequences, we find that the multiple alignment is more sensitive to our choice in gap costs than to the form of ancestral sequence chosen. Reinforcing this conclusion, we find that the gap open cost scaling function is also extremely important to obtaining good scores when aligning ancestral sequences. Finally, to our surprise, using maximum likelihood to infer ancestral sequences resulting in less accurate alignments than using parsimony. The reason for this is that the maximum likelihood method is far more sensitive to the underlying data and therefore resulting in alignments accuracies that have a large amount of variation. Also, on the data set generated from the tree that has a small amount of growth followed by a large amount of growth, the maximum likelihood based methods did particularly poorly compared to the parsimony based methods.
Finally, both the sum-of-pairs approaches did better than all the ancestral alignment methods, as expected. Additionally, we found that Ma, Wang, and Zhang's algorithm [6] outperformed ClustalW [5] by a small amount.
Methods
Our multiple alignment framework uses progressive alignment up a specified phylogenetic tree. At each internal node we perform an alignment of two multiple alignments. We test six different algorithms for aligning alignments: ClustalW, the recent algorithm of Ma, Wang, and Zhang, and four algorithms that align inferred ancestral sequences. Our four ancestral alignment algorithms explore the use of both parsimony and maximum likelihood to infer ancestral sequences, and also allow the use of both ambiguous and unambiguous ancestral sequences. We include ClustalW, as it is widely used in practice, and the algorithm of Ma, Wang, and Zhang, whose output more accurately approximates the optimal alignment under sum-of-pairs scoring.
In this section, we describe how we align two alignments using ancestral sequence inference, as well as our four ancestral sequence inference techniques and associated log-odds scoring frameworks.
Alignment by ancestral sequence inference
Given two multiple alignments, we first infer an ancestral sequence for each alignment. Then, we remove gaps in the inferred ancestral sequences and align the them with the classic Needleman-Wunsch dynamic programming algorithm [16,17], under an appropriate log-odds scoring framework. Finally, we map the alignment of the two ancestral sequences to an alignment between the two alignments by inserting a column of gaps into each alignment for each gap inserted into the respective ancestral sequence. Given a length n alignment of k sequences, we can infer an ancestral sequence in time Θ(kn). To align two length n alignments, one with k sequences and with ℓ sequences, requires Θ((k + ℓ)n + n2) time; this contrasts with the much larger run time of O (n2 (k + ℓ)) required by Ma, Wang and Zhang's algorithm [6].
We explore four different methods of inferring and aligning ancestral sequences. First, we infer sequences using parsimony, finding both ambiguous and unambiguous ancestral sequences. Second, we infer sequences using maximum likelihood, again finding both ambiguous and unambiguous ancestral sequences. We call these methods parsimony, ambiguous parsimony, maximum likelihood, and ambiguous maximum likelihood. For each method, we also compute a log-odds scoring framework, which we use when we align two ancestral sequences.
To infer an ancestral sequence for a multiple alignment, we assume we have a correct edge weighted phylogenetic tree relating the sequences in the alignment. We work with a Jukes/Cantor 1-parameter evolution model [13], but our approach easily extends to more realistic models of evolution. We represent ambiguity in ancestral sequences using the 15-letter IUPAC alphabet. For example, if a position in an ancestral sequence is either an A or a T, we encode it as the IUPAC symbol W.
Let Γ represent the IUPAC alphabet and let Σ = {A, C, G, T} be the DNA alphabet. We represent a gap with the '-' character and define a third alphabet ΣG = Σ∪{-}. We use the symbol b to represent alphabet symbols from Σ or ΣG and c to represent symbols from Γ.
Parsimony and ambiguous parsimony
The most parsimonious ancestral sequence is the sequence that requires the least number of mutations down the tree to obtain the sequences in the alignment. There can be multiple most parsimonious ancestral sequences for a given alignment. We take the union of these ancestral symbols at each position of the sequence, and thus obtain an ambiguous ancestral sequence. When we require an unambiguous ancestral sequence, we randomly choose a position at each point. We use an efficient algorithm by Fitch [18] to reconstruct both ambiguous and unambiguous ancestral sequences. We now briefly describe this algorithm.
Consider a single column of the alignment where the leaves of the tree are assigned the corresponding base, or gap from the column. Starting at the leaves of the tree we work upwards assigning an ambiguous symbol to each internal node. We do this with a consensus operation. If we consider each ambiguous symbol to be a set of possible bases, then the consensus of two symbols is the intersection of the sets, or the union if the intersection is empty. At a node z with children x and y, we assign z the consensus of the symbols at x and y. On completion, the symbol assigned to the root of the tree represents the set of most parsimonious ancestral bases for the column. Note that gaps are handled naturally: the consensus of a gap with any other symbol is the other symbol. Since each alignment column has at least one non-gap symbol, the reconstructed ancestral sequence contains no gaps.
To obtain an unambiguous ancestral base, we choose one of the bases represented by the root symbol uniformly at random. A consensus operation takes constant time, and so the above algorithm runs in O(n) time for a alignment column of n sequences.
Log-odds scoring framework
When aligning two ancestral sequences, we use the log-odds scoring framework from Brown and Hudek [8], explained in more detail in Hudek's thesis [19]. when aligning two ancestral sequences. We modify this framework slightly for unambiguous parsimony, and describe the framework and modification here.
Suppose we want to find an alignment at node z, with children x and y at which we have already computed multiple alignments. To do this, we align the inferred ancestral sequence at x to the inferred ancestral sequence at y. We score pairs of symbols from the two ancestral sequences using a log-odds framework based on the underlying phylogenetic tree.
We compute two probabilities: the probability Pr[d1,d2|related] of seeing the observed symbols d1 and d2 in ancestral sequence positions related by evolution according to the given tree, and Pr[d1, d2|unrelated], the probability of seeing observed symbols d1 and d2 in ancestral sequences positions unrelated by evolution. Here, d1 and d2 are either from the ambiguous or unambiguous alphabet depending on the type of ancestral sequence we use.
We start by computing Cz(c, b), the probability that we see an ambiguous symbol c ε Γ in the ancestral sequence for node z, given that the true value of the ancestral sequence is the DNA base b.
Theorem 1. For all nodes z and symbols c ε Γ and b ε Σ, the probability Cz(c, b) can be computed in O(m) time where m is the number of descendants of node z and we treat Γ and Σ as constant in size.
Proof. At the leaves, the ancestral symbol is the same as the associated alignment symbol. Therefore, we initially set Cu(c, b) = 1 for all c and b where c = b, and C(c, b) = 0 for all others.
Let Tv(b) be the event that the true value of the ancestral sequence at node v is the DNA base b. Let Cc be the set of pairs of symbols (cx, cy) from Γ, whose consensus symbol is c. Consider a node z with children x and y for which we have already computed Cv(c, b) for all values of c ε Γ and b ε Σ. We compute Cz(c, b) using the equation
Cz(c,b)=∑ bx,by∈∑(cx,cy)∈ccDz,b(x,bx,cx)Dz,b(b,y,by,cy), (2)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGdbWqdaWgaaWcbaGaemOEaOhabeaakiabcIcaOiabdogaJjabcYcaSiabdkgaIjabcMcaPiabg2da9maaqafabaGaemiraq0aaSbaaSqaaiabdQha6jabcYcaSiabdkgaIbqabaGccqGGOaakcqWG4baEcqGGSaalcqWGIbGydaWgaaWcbaGaemiEaGhabeaakiabcYcaSiabdogaJnaaBaaaleaacqWG4baEaeqaaOGaeiykaKIaemiraq0aaSbaaSqaaiabdQha6jabcYcaSiabdkgaIbqabaGccqGGOaakcqWGIbGycqGGSaalcqWG5bqEcqGGSaalcqWGIbGydaWgaaWcbaGaemyEaKhabeaakiabcYcaSiabdogaJnaaBaaaleaacqWG5bqEaeqaaOGaeiykaKIaeiilaWcalqaabeqaaiaaykW7caaMc8UaemOyai2aaSbaaWqaaiabdIha4bqabaWccqGGSaalcqWGIbGydaWgaaadbaGaemyEaKhabeaaliabgIGiolabggHiLdqaaiabcIcaOiabdogaJnaaBaaameaacqWG4baEaeqaaSGaeiilaWIaem4yam2aaSbaaWqaaiabdMha5bqabaWccqGGPaqkcqGHiiIZcqWGJbWydaWgaaadbaGaem4yamgabeaaaaWcbeqdcqGHris5aOGaaCzcaiaaxMaadaqadaqaaiabikdaYaGaayjkaiaawMcaaaaa@7860@
where
Du,b (v, bv, cv) = Pr[Tv(bv)|Tu(b)]Cv(cv, bv) (3)
is the the probability of seeing cv at node v given the true value is bv, times the probability that the true value at node v is bv given that the true value at node u is b. We compute Pr[Tv(bv)|Tu(b)] using the length of the edge (u, v) and the probabilistic mutation model. We compute Cv(c, b) for each node of the tree, and obtain Cz(c, b) in O(m) time, where m is the number of leaves. □
Ambiguous probabilities
When we use ambiguous ancestral sequences, the probability of seeing consensus letters c1 and c2 at a positions arising from a common ancestor is
Pr[c1,c2|related]=∑bx,by,bz∈∑Pr[Tx(bx),Ty(by)|Tz(bz)]Cx(c1,bx)Cy(c2,by). (4)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@8C17@
That is, over all choices of the true value at x, y, and z, we compute the conditional probabilities of seeing consensus pair (c1, c2) at positions related by ancestry.
At positions unrelated by ancestry, we assume that the true value for the ancestral sequence is equally likely to be any of the four DNA bases, though this model can be made more complex to model more realistic sequences. Therefore, the probability of seeing consensus letters c1 and c2 is
Pr[c1,c2|unrelated]=116∑bx∈ΣCx(c1,bx)∑by∈ΣCy(c2,by). (5)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7351@
Finally, the log-odds score for aligning consensus symbols c1 and c2 is
S(c1, c2) = log2 (Pr[c1, c2|related]/Pr[c1, c2|unrelated]), (6)
in bits.
Unambiguous probabilities
If we use unambiguous ancestral sequences, the situation is similar. Let V(b, c) be the probability that we randomly choose base b from the set of bases represented by consensus symbol c ∊ Γ. That is, if c is a symbol corresponding to a set of k symbols from Σ, and b is in this set, then V(b, c) = 1/k, otherwise it is zero. Then,
Pr[b1,b2|related]=∑bx,by,bz∈Σc1,c2∈ΓPr[Tx(bx),Ty(by)|Tz(bz)] Cx(c1,bx)V(b1,c1)Cy(c2,by)V(b2,c2) (7)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0dXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@A5AC@
is the probability that we see DNA base b1 at x and DNA base b2 at y given that we are looking at positions arising from a common ancestor.
At positions unrelated by ancestry, we assume that the true value for the ancestral sequence is equally likely to be any of the four DNA bases. Therefore, the probability of seeing DNA base b1 and b2 is
Pr[b1,b2|unrelated]=116∑c1∈Γbx∈ΣCx(c1,bx)V(b1,c1)∑c2∈Γby∈ΣCy(c2,by)V(b2,c2). (8)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@8F92@
The log-odds score for DNA symbols b1 and b2 is
S(b1, b2) = log2 (Pr[b1, b2|related]/Pr[b1, b2|unrelated]), (9)
in bits.
Maximum likelihood
For a given alignment column, we compute the most likely ancestral DNA base as in Felsenstein [20], but modified to consider gap characters in the column. We model the gap symbol as other symbols in the alphabet, but use special probabilities when considering changes to and from gap characters. Specifically, we use an insertion/deletion probability whenever we consider a mutation from a non-gap symbol to a gap symbol, or a gap symbol to a non-gap symbol. The probability of going from a gap symbol to a gap symbol is one minus the probability of an insertion.
Upon completion of the basic inference algorithm, we have a vector at the root that gives, for each position and each DNA base, the likelihood of that base. Assuming independence, the likelihood of a given ancestral sequence is the product of these. We obtain an unambiguous ancestral sequence from this by taking the base with maximum likelihood, randomly choosing between ties. To obtain an ambiguous maximum likelihood, which in effect is an approximation of the posterior Bayesian distribution of the ancestral symbol at that site, where we assume a uniform prior distribution over all alphabet symbols, we map the vector to an IUPAC symbol as follows. For each IUPAC symbol, we define a vector over the DNA alphabet where we have a one for each DNA symbol described by the IUPAC symbol, and a zero for all other DNA symbols. For example, the IUPAC symbol W represents an A or a T. We define associated vector (1,1,0,0), scaled to a probability distribution, where the numbers in the vector refer to A,T,C, and G, in that order. We then map the likelihood vector, also scaled to a probability vector (which, again, corresponds to computing the posterior probabilities, assuming a flat prior on all four DNA bases though this assumption can easily be removed), to an IUPAC symbol, choosing the IUPAC symbol with associated vector that has the closest euclidean distance to the likelihood vector scaled to a probability distribution.
Log-odds scoring framework
We desire a log-odds scoring framework similar to that developed for parsimonious ancestral sequences. While an appropriate scoring matrix can be obtained by sampling, we want to eliminate any sampling error from our study and so choose to compute the log-odds scoring framework directly.
Assume we are at node z with children x and y in a tree rooted at node r. First, consider the sub tree rooted at x. Let Lx,d
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBamXvP5wqSXMqHnxAJn0BKvguHDwzZbqegm0B1jxALjhiov2DaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciaacaGaaeqabaWaaeGaeaaakeaaimaacaWFmbWaaSbaaSqaaiabdIha4jabcYcaSiabdsgaKbqabaaaaa@3D3D@ be the set of all assignments of symbols from ΣG to the leaves of the sub tree rooted at x such that maximum likelihood infers symbol d at x.
Let Ox(a) be the probability that a particular assignment a of bases to the leaves of sub tree x occurs by evolution. We compute the probability that the true DNA base at node x is b by considering the probability all possible bases for the root of the tree, and simulating evolution down to b.
In the following, as we compute scores for unambiguous ancestral sequences, d1 and d2 refer to symbols from ΣG. To compute scores for ambiguous ancestral sequences, d1 and d2 refer to symbols from Γ. We compute Pr[d1,d2|related], the probability of seeing symbols d1 and d2 in positions related by a common ancestor, as
Pr[d1,d2|related]=∑ax∈Lx,d1ay∈Ly,d2 bz∈ΣGPr[Ox(ax)|Tz(bz)]Pr[Oy(ay)|Tz(bz)]Pr[Tz(bz)], (10)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBamXvP5wqSXMqHnxAJn0BKvguHDwzZbqegm0B1jxALjhiov2DaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@AFD9@
where we compute the probability that the true value at node z is b as
Pr[Tz(b)]=∑br∈ΣGPr[Tz(b)|Tr(br)] Pr[Tr(br)]. (11)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@6754@
That is, for each possible true value at root r, we compute the probability that the value mutates to b on the path from r to z.
We now compute Pr[d1,d2|unrelated], the probability that we see symbols d1 and d2 in positions unrelated by a common ancestor. First, let MLx(d) be the event that maximum likelihood infers symbol d at node x. We compute MLx(d) as
Pr[MLx(d)]=∑a∈Lx,d b∈ΣGPr[Ox(a)|Tx(b)]Pr[Tx(b)] (12)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBamXvP5wqSXMqHnxAJn0BKvguHDwzZbqegm0B1jxALjhiov2DaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@79ED@
where we compute Pr[Tx(b)] in the same way as Pr[Tz(b)] previously. We compute Pr[d1, d2|unrelated] as
Pr[d1, d2|unrelated] = Pr[MLx(d1)] Pr[MLy(d2)]. (13)
Finally, the log-odds score of symbols d1 and d1 is
S(d1, d2) = log2 (Pr[d1, d2|related]/Pr[d1, d2|unrelated]). (14)
To compute the above scores, we must examine every possible assignment of bases to leaves for both the left and right children of the root. For a particular tree, let k be the maximum of the number of taxa below the left child of the root and the right child of the root. We require time O(|ΣG|k) to compute the log-odds scores for this tree.
Gap open cost scaling
When aligning alignments with ancestral sequences, gap open costs play a major role. In a tree with differing edge lengths, the gap open cost should also be able to vary. Since existing gaps are not considered when inserting a new gap in ancestral alignments, the gap open cost has a large influence over the quality of the resulting alignment. It is therefore important that we estimate an appropriate gap open cost for each node of the tree.
Consider creating alignment for node z with children x and y. We scale the gap open cost according to the distance between the nodes x and y. We test two slightly different scaling functions. The first function multiplies the gap open cost by the largest score value in the log-odds scoring matrix for node z. We call this the Max method. The second function computes the expected score of two symbols from unrelated positions and uses this value to scale the gap open cost. We call this the Expected method.
Authors' contributions
Both AH and DB developed the fundamental ideas and hypotheses and the mathematical framework for deriving the alignment scores. AH developed and ran the experiments and implemented the alignment algorithms.
Acknowledgements
The baboon and human sequence data used for estimating the gap parameters were generated by the NIH Intramural Sequencing Center . Our work has been supported by the Natural Sciences and Engineering Research Council of Canada and by the Human Frontier Science Program.
==== Refs
Bray N Pachter L MAVID: Constrained Ancestral Alignment of Multiple Sequences Genome Research 2004 14 693 699 15060012
Brudno M Do C Cooper G Kim M LAGAN and Multi-LAGAN: Efficient Tools for Large-Scale Multiple Alignment of Genomic DNA Genome Research 2003 13 721 731 12654723
Feng D Doolittle R Progressive sequence alignment as a prerequisite to correct phylogenetic trees Journal of Molecular Evolution 1987 25 351 360 3118049
Just W Computational Complexity of Multiple Sequence Alignment with SP-Score Journal of Computational Biology 2001 8 615 623 11747615
Thompson J Higgins D Gibson T CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Research 1994 22 4673 4680 7984417
Ma B Wang Z Zhang K Alignment between Two Multiple Alignments Proceedings of CPM 2003 2003 Springer-Verlag 254 265
Kececioglu J Starrett D Aligning alignments exactly RECOMB '04: Proceedings of the eighth annual international conference on Computational molecular biology 2004 New York, NY, USA: ACM Press 85 96
Brown DG Hudek AK New Algorithms for Multiple DNA Sequence Alignment Proceedings of WABI 2003 2004 3240 314 325
Eizirik E Murphy W O'Brien S Molecular Dating and Biogeography of the Early Placental Mammal Radiation The Journal of Heredity 2001 92 212 219 11396581
Rambaut A Phyl-O-Gen v1.2 2003
Kearney P Munro I Phillips D Efficient Generation of Uniform Samples from Phylogenetic Trees Proceedings of WABI 2003 2003 Benson G, Page R 177 189
Stoye J Evers D Meyer F Rose: generating sequence families Bioinformatics 1998 14 157 163 9545448
Durbin R Eddy S Krogh A Mitchison G Biological sequence analysis 1998 Cambridge University Press
Thomas J Touchman J Blakesley R Bouffard G Beckstrom-Sternberg S Margulies E Blanchette M Siepel A Thomas P McDowell J Maskeri B Hansen N Schwartz M Weber R Kent W Karolchik D Bruen T Bevan R Cutler D Schwartz S Elnitski L Idol J Prasad A Lee-Lin S Maduro V Summers T Portnoy M Dietrich N Akhter N Ayele K Benjamin B Cariaga K Brinkley C Brooks S Granite S Guan X Gupta J Haghighi P Ho S Huang M Karlins E Laric P Legaspi R Lim M Maduro Q Masiello C Mastrian S McCloskey J Pearson R Stantripop S Tiongson E Tran J Tsurgeon C Vogt J Walker M Wetherby K Wiggins L Young A Zhang L Osoegawa K Zhu B Zhao B Shu C Jong PD Lawrence C Smit A Chakravarti A Haussler D Green P Miller W Green E Comparative analyses of multi-species sequences from targeted genomic regions Nature 2003 424 788 793 12917688
Goodman M The genomic record of Humankind's evolutionary roots American Journal of Human Genetics 1999 64 31 39 9915940
Needleman S Wunsch C A general method applicable to the search for similarities in the amino acid sequence of two proteins Journal of Molecular Biology 1970 48 443 453 5420325
Gotoh O An improved algorithm for matching biological sequences Journal of Molecular Biology 1982 162 705 708 7166760
Fitch W Toward defining the course of evolution: minimum change for a specified tree topology Systematic Zoology 1971 20 406 416
Hudek AK New Anchoring Techniques for Global Multiple Alignment of Genomic Sequences Master's thesis 2004 University of Waterloo
Felsenstein J Inferring Phylogenies 2003 Sinauer Associates
|
16293191
|
PMC1310622
|
CC BY
|
2021-01-04 16:27:48
|
no
|
BMC Bioinformatics. 2005 Nov 17; 6:273
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-273
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2741629724010.1186/1471-2105-6-274Methodology ArticleA stepwise framework for the normalization of array CGH data Khojasteh Mehrnoush [email protected] Wan L [email protected] Rabab K [email protected] Calum [email protected] British Columbia Cancer Research Centre, Vancouver, BC, Canada2 Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada2005 18 11 2005 6 274 274 20 6 2005 18 11 2005 Copyright © 2005 Khojasteh et al; licensee BioMed Central Ltd.2005Khojasteh et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In two-channel competitive genomic hybridization microarray experiments, the ratio of the two fluorescent signal intensities at each spot on the microarray is commonly used to infer the relative amounts of the test and reference sample DNA levels. This ratio may be influenced by systematic measurement effects from non-biological sources that can introduce biases in the estimated ratios. These biases should be removed before drawing conclusions about the relative levels of DNA. The performance of existing gene expression microarray normalization strategies has not been evaluated for removing systematic biases encountered in array-based comparative genomic hybridization (CGH), which aims to detect single copy gains and losses typically in samples with heterogeneous cell populations resulting in only slight shifts in signal ratios. The purpose of this work is to establish a framework for correcting the systematic sources of variation in high density CGH array images, while maintaining the true biological variations.
Results
After an investigation of the systematic variations in the data from two array CGH platforms, SMRT (Sub Mega base Resolution Tiling) BAC arrays and cDNA arrays of Pollack et al., we have developed a stepwise normalization framework integrating novel and existing normalization methods in order to reduce intensity, spatial, plate and background biases. We used stringent measures to quantify the performance of this stepwise normalization using data derived from 5 sets of experiments representing self-self hybridizations, replicated experiments, detection of single copy changes, array CGH experiments which mimic cell population heterogeneity, and array CGH experiments simulating different levels of gene amplifications and deletions. Our results demonstrate that the three-step normalization procedure provides significant improvement in the sensitivity of detection of single copy changes compared to conventional single step normalization approaches in both SMRT BAC array and cDNA array platforms.
Conclusion
The proposed stepwise normalization framework preserves the minute copy number changes while removing the observed systematic biases.
==== Body
Background
Microarray-based Comparative Genomic Hybridization (array CGH) is used to detect the aberrations in segmental copy numbers at chromosomal loci represented by DNA clones with known genomic locations [1]. CGH microarrays typically contain tens of thousands of spotted DNA sequences such as those derived from bacterial artificial chromosomes (BACs). Sample DNA from a test and a reference genome are labelled with different fluorescent dyes (usually Cyanine-3 and Cyanine-5 dyes) and then hybridized to the genomic microarray. The fluorescent signal intensity of each spot on the microarray serves as a relative measure of the amount of sample DNA bound to the DNA sequence of that spot. The ratio between the Cyanine-3 and the Cyanine-5 intensity of each spot reflects the relative quantities of the test and reference DNA samples.
The ratio of the two fluorescent signals at each spot is commonly used to detect copy number alteration. However, the ratios of the fluorescent signals are usually influenced by systematic effects from non-biological sources that can introduce biases in the estimates of these ratios. Such biases should be removed in order to draw conclusions on copy number status. The process of correcting for the systematic effects is often referred to as normalization.
Array CGH technology generally has more stringent performance requirements than gene expression microarray analysis. These requirements are to detect single DNA copy number changes in abnormal cells, typically within tumor samples. Detection sensitivity is complicated by the heterogeneous nature of tumor tissue with varying degrees of contamination from non-cancer cells. Due to the limitation in material availability performing replicate experiments is not always possible or desired.
In the context of developing a normalization protocol for array CGH, knowing the copy number status of DNA segments provides true values for calibration. The same copy number exists in different samples as the normal state for human cells is diploid. In contrast, gene expression level varies continuously for each gene and the expression level of the same gene is not expected to be identical in two different samples.
While a 2 fold change in signal may not represent a significant alteration in gene expression microarray analyses [3], for CGH arrays, a single copy gain compared to normal diploid DNA will result in a ratio of 3:2. A single copy loss would reduce the signal ratio to 1:2. Considering the contamination of tumor (abnormal) cells with non-cancer (normal) cells, the copy number ratio may be even smaller. So the challenge in normalization is to preserve the true copy number change signals while removing the systematic variations.
The purpose of this work is to correct for the systematic sources of variation while maintaining the true biological variations as small as a single copy number change in a sample of a heterogeneous cell population.
After an investigation of the systematic variations in the data from array CGH experiments, we tested existing normalization methods commonly used for gene expression data in order to deduce a stepwise normalization framework tailored to handling high density array CGH data. Here we demonstrate the efficacy of the stepwise normalization scheme through several quantitative characteristics of the data from several functional types of array CGH.
Results and discussions
Materials
Data from five sets of experiments were used in the development of the normalization strategy (Table 1). The first four datasets were generated from array CGH experiments performed using the SMRT (Sub Mega base Resolution Tiling) arrays. These arrays are tiling resolution BAC arrays with complete coverage of the human genome using 32,433 fingerprint-verified individually amplified BAC clones [4]. The experimental procedures for array CGH and generating spot images have been described previously [4]. The entire set of 32,433 solutions was spotted in triplicate onto two slides by a 4 × 12 pin arrayer. For the purpose of this study, only the data from the first array out of the two arrays were used.
Table 1 Data description. In this table, the array data of this study are summarized.
Array Reference DNA Sample DNA Data type for normalization performance evaluation Evaluation method
MM-1 to MM-4 Male genomic Male genomic Self-self hybridizations S.d. for each array
H526-1 to H526-8 H526 cell line Male genomic Replicate H526 cell line experiments 1. Correlation coefficient for each pair of arrays
2. ICC
3. S.d. for each spot
MF-1 and MF-2 Female genomic Male genomic Single copy change T-test
T1 to T5 Female genomic Male/Female mixture (see Additional file 1) Single copy loss with normal cell contamination T-test
T6 to T10 Male genomic Male/Female mixture (see Additional file 1) Single copy gain with normal cell contamination T-test
X1 to X5 Female DNA cell lines containing varying numbers of X chromosomes (see Additional file 2) varying levels of gene amplification and deletion for each of the X-chromosomal genes T-test
The fifth dataset is a public dataset downloaded from the Stanford Microarray Database . This datasets was generated from array CGH experiments performed using human cDNA microarrays, [12].
The first dataset (self-self hybridization data) was derived from hybridization of the same DNA sample, i.e., normal male genomic DNA was used for both test and reference materials but labelled with different dyes. The four microarrays used in this CGH experiment are referred to as MM-1 to MM-4 in the following text.
The second dataset (hybridization data from replicate experiments) was derived from comparison of a tumor cell DNA sample with well characterized chromosomal aberrations (lung cancer cell line H526) [4] against normal male DNA. The 8 arrays used in this experiment are denoted H526-1 through H526-8.
The third dataset (hybridization data from male and female DNA mimicking single copy deletion) was derived from comparison of normal male DNA versus normal female DNA, using arrays named MF-1 and MF-2.
The fourth dataset (hybridization data from samples mimicking heterogeneous cell populations) was derived from a series of array CGH experiments in which the samples to be compared were mixtures of male and female DNA affecting X chromosome dosage mimicking tumor samples with varying levels of normal cell contamination. Precise proportions of DNA were mixed to simulate increasing levels of heterogeneity as previously described [5]. Arrays T1 through T5 compared male DNA against female DNA generating a 1:2 ratio for X chromosome sites mimicking a single copy deletion. Contamination from normal cells was then simulated by spiking varying amounts of female DNA into the male DNA sample. Arrays T6 through T10 compared a 50/50 mixture of male and female DNA against a male DNA reference generating a 3:2 ratio for X chromosome sites mimicking single copy amplifications. Contamination from normal cells was simulated by spiking varying amounts of female DNA into the male/female DNA mixture.
The fifth dataset was derived from hybridization of genomic DNAs from cell lines containing varying numbers of X chromosomes to simulate varying levels of gene amplification and deletion for each of the X-chromosomal genes present in the cDNA array [12]. The five experiments comprising the fifth data set are denoted X1 through X5.
Systematic variations
After a thorough investigation of the systematic variations in the data from our array CGH experiments, four kinds of bias were identified. Below we explain each bias type.
Intensity bias
This bias is evident in the frequently used M-A plots which are plots of the log ratio M = log2(Ir/Ig) = log2(Ir) - log2(Ig) against the mean of the log intensities A = 1/2(log2(Ir) + log2(Ig)), where Ir and Ig are the intensities of the cyanine-5 and cyanine-3 channels respectively. In our data, this bias predominantly appears as curvature in the low intensity end of the M-A plot.
Spatial bias
The representation of log ratios based on the corresponding spot location on the microarray is another type of plot which can be used to reveal spatially variable bias. We refer to this plot as M-XY plot. The spatially smoothed M-XY plot reveals the general trend of log ratios against their locations on the array (Fig. 1). For randomly distributed genomic loci across an array this plot should be a flat plane.
Figure 1 A smoothed M-XY plot illustrating spatial bias. The plot displays representation of log2 ratios based on the corresponding spot location on the microarray, the plot is smoothed with a moving median filter.
Spatial heterogeneity was thought to be caused by the different print tips used in printing the targets on the arrays [6]. However, our data show that the spatial heterogeneity is not caused by print tips effects because the spatial patterns are not organized in a block wise fashion (as they would be due to bias introduced by specific print tips). In fact, the patterns appear as a continuous function across the entire array.
Plate bias
This is a spatial pattern that can be seen in the data after the spatial gradient has been removed by the spatial normalization step mentioned above. This pattern is repeated in all subgrids in the M-XY plot and corresponds to the plate groups (groups of spots on the microarray that are all printed from the same microplate).
Plate bias is evident when box-plots of log2 ratios from each plate group are compared. These box plots show a systematic difference among the log2 ratios of the different plate groups. The median log2 ratio of each plate group is expected to be near zero, i.e. positive and negative deviations should cancel out in each plate group, unless the copy numbers of the clones in a plate biologically differ between the test and the control samples. We do not believe this is the case in our experiments.
This bias is caused by the fact that different clones that are produced in different microplates may have experienced slightly different physical conditions during the polymerase chain reaction (PCR) or in subsequent purification steps [7]. This variation in the efficiency of spot solution synthesis appears to affect different plate groups resulting in a plate level bias.
Background bias
The measured intensity for each microarray spot contains a contribution from the background fluorescence within the spot. This introduces a bias in the ratios of the spots' intensities. In the M-A plot this bias appears as deviation from zero in the log2 ratios of the lower intensity spots.
Methods of bias removal
In order to remove these types of biases, we evaluated the following stepwise normalization procedure:
1. The spatial trend is estimated by computing, for each spot on the array, the median of log2 ratios for the spots within a spatial neighbourhood window of size 11 rows by 11 columns centred on that spot. The spatial bias that is estimated for each spot in this way is then subtracted from the log2 ratio of that spot. This step is referred to as "Spatial" normalization.
2. The plate bias is removed by calculating the median of the log2 ratios for all spots in the same plate group and subtracting it from the log2 ratios for all those spots. This step is referred to as "Plate" normalization.
3. The intensity bias is estimated using robust LOWESS curve fitting [8]. After this bias is estimated, assuming the bias is multiplicative; the bias is subtracted from the log ratios. This step is denoted as "Intensity LOWESS" normalization.
4. To remove the background bias, one of the following two different approaches is usually taken: either the estimate of the background intensity is subtracted from the estimated foreground intensity of each spot before taking the ratios, or it is not subtracted. In the latter case, the introduced bias is dealt with by treating it as intensity dependent bias. We evaluated both of these approaches in our experiments (see below).
Below we show that the above stepwise procedure is effective in removing the mentioned types of systematic variations. We demonstrate the efficacy of our procedure by comparing several quantitative characteristics of data normalized by our proposed strategy to those of non-normalized data and data normalized by other techniques listed in Table 2.
Table 2 Summary of normalization methods. Each of the normalization methods in this table will be denoted by its number through out the text. For full description of methods refer to "Methods of bias removal" section in Results and Discussion and the "Normalization methods" section in Methods.
Method no. Normalization method Description
Background subtracted
1 No normalization Raw ratios
Global method
2 Global median Ratio Ratios scaled by their median
Intensity dependant methods
3 Intensity LOWESS, 10% span Global Intensity LOWESS, span = 10%
4 Intensity LOWESS, 25% span Global Intensity LOWESS, span = 25%
5 Intensity LOWESS, 40% span Global Intensity LOWESS, span = 40%
Spatial methods
6 Print tip mean Ratio Ratios of each print-tip group scaled by the mean ratio of that group
7 Spatial median of log2 ratios for the spots within a spatial neighbourhood window of size 11 rows by 11 columns centred on that spot
8 Spatial + Median Plate Ratio Method 8 followed by plate normalization
Combined intensity dependent and spatial methods
9 Print Tip Intensity LOWESS, span = 40% LOWESS performed on the ratios from each print-tip group
10 Intensity LOWESS + Spatial Stepwise Method 4 and 8
Three step normalization
11 Intensity LOWESS + Spatial + Median Plate Ratio Stepwise Methods 4 and 9
12 Spatial + Median Plate Ratio + intensity LOWESS Stepwise Methods 9 and 4
Background not subtracted
13 No Normalization See Method 1, but without background subtraction
Global method
14 Global median Ratio See Method 2, but without background subtraction
Intensity dependant methods
15 Global Intensity LOWESS, span = 10% See Method 4, but without background subtraction
Spatial methods
16 Print tip Mean Ratio See Method 3, but without background subtraction
17 Spatial See Method 8, but without background subtraction
Combined intensity dependent and spatial methods
18 Intensity LOWESS + Spatial See Method 10, but without background subtraction
Three step normalization
19 Intensity LOWESS + Spatial + Median Plate Ratio See Method 11, but without background subtraction
Normalization of self-self array CGH data
The self-self experiments (arrays MM-1 through MM-4) were used to study the effect of normalization on removing the bias from the data and increasing the accuracy of the measurements. The 19 methods of normalization listed in Table 2 were evaluated on the data obtained from these arrays.
Since the same male genomic DNA serves as both sample and reference DNA, the copy numbers detected in both the Cyanine-3 and Cyanine-5 channels are expected to be the same at all loci, resulting in a zero theoretical value for the log2 ratio of intensities at all spots on the array. The effects of normalization on removing the bias were examined by calculating the standard deviation (s.d.) of the log2 ratios for each array in the experiment, evaluating each of the 19 methods listed in Table 2. Then all 19 standard deviations were scaled against the standard deviation of the raw ratios before normalization (i.e. against the s.d. value from the first method of Table 2). For each normalization method, the scaled s.d. values were then averaged across the four arrays. Figure 2 shows these average standard deviations.
Figure 2 Normalization of self-self hybridization data. Relative standard deviation (s.d.) of log2 ratios averaged across arrays MM-1 through MM-4 using all data points are shown in blue. The repeated analysis of relative s.d. after removal of the weakest 10% of spots is shown in red. The numbers on the horizontal axis refer to the methods used for normalization listed on Table 2.
The three different window sizes of 10%, 25% and 40% of the data points, used for LOWESS intensity normalization (methods 4-6 in Table 2) did not have a significant effect on the effectiveness of normalization.
Among 12 normalization methods that are performed on the ratios of background subtracted intensities, the stepwise strategy (method 12) results in the lowest s.d. for all four arrays. Also, among 7 normalization methods that are performed on the ratios of non-background subtracted intensities, the stepwise strategy (method 19) results in the smallest s.d.
When the three-step proposed normalization is performed on the ratios of non-background subtracted intensities, it yields better performance, in terms of reducing the s.d. of log2 ratios, than when it is applied to the ratios of background-subtracted intensities.
To further explore the effect of the background intensities, the standard deviations were recalculated for these four arrays with the lowest intensity spots removed from each data set. The difference between the s.d. of the ratios after normalization for the case of background subtracted and the case of non-background subtracted intensities became smaller on the reduced datasets. As an example, the new s.d. values when 10% of the lowest intensity spots are removed, are plotted in Fig. 2. This suggests that subtracting background increases the variability of ratios of lower intensity spots and the variability of higher intensity spots are not affected much by subtracting or not subtracting the background.
Normalization of hybridization data from replicate experiments
In order to see how normalization affects the consistency of the data from replicate experiments, 8 replicate experiments were performed. H526-1 through H526-8 represent independent array CGH experiments using the same source of sample DNA (isolated from the well studied lung cancer cell line H526).
The Standard deviations of the log2 ratios of the same spot across the 8 replicate arrays were calculated and averaged across all the spots for each normalization method. The results are shown in Fig. 3A. The standard deviation measure attains its smallest value after method 12 or 19 is performed on the data. When the three-step normalization is performed on the ratios of non-background subtracted data (method 19), its performance is slightly better than when it is performed on ratios of background subtracted intensities (method 12).
Figure 3 Normalization of hybridization data from replicate experiments. 8 replicate array CGH experiments were done comparing sample DNA from H526 cell line and the reference normal male genomic DNA. A. Graph shows the average of the standard deviations of log2 ratios for the same spot across 8 replicate arrays. B. shows the ICC and Average correlation coefficient of replicate arrays. Horizontal axis represents the method number listed in Table 2.
The Pearson's Correlation Coefficient [9] was calculated for the data from each pair of the replicate arrays, with 28 possible pairings. The average of the 28 correlation coefficients for each single method was then calculated (Fig. 3B).
The Intraclass Correlation Coefficient (ICC) [9] was calculated for the set of data obtained from the 8 replicate arrays normalized using each of the methods described above. The results are also summarized in Fig. 3B. The ICC and Pearson correlation coefficient show similar results across the methods. Both ICC and Correlation coefficient attain their highest values after the three-step normalization method. This applies to both the ratios of non-background subtracted intensities and ratios of background subtracted intensities. ICC and Correlation coefficient are slightly higher when background subtraction is not performed on spot intensities measures.
Normalization of hybridization data from male and female DNA
To evaluate the effect of normalization on improving detection of single copy loss, two array CGH experiments were conducted comparing male (XY) genomic DNA against female (XX) genomic DNA. The copy numbers of autosomal loci (clones on chromosome 1 through 22) are equal, while the X loci exhibit a 1:2 ratio, simulating a single copy loss.
The normalization methods described above were applied to the data obtained from these two experiments. To determine which method results in the best separation of clones with normal copy from those with a single copy loss, a two-sample two-tailed T-test was performed on each array data normalized by each method. The T-test evaluates the difference between the means of two groups of log ratios. The first group consists of log ratios for clones from chromosomes 1 through 22 and the second group consists of log ratios for clones from chromosome X. The value of the T statistic is shown in Fig. 4 for both arrays and for each normalization method. A larger value for the T-statistic indicates better separation between the means of the two samples.
Figure 4 Normalization of hybridization data from male and female DNA. For each of arrays MF-1 and MF-2, a T-test was performed on the two groups of log ratios, i.e. log ratios for the autosomal clones and those for the X chromosome clones. Values of T-statistic after each normalization method are shown. Horizontal axis represents the method number listed in Table 2.
For the data from array MF-1, the largest T-statistic was obtained after our three-step normalization procedure was performed on the ratios of background subtracted intensities. For this array, the normalization methods performed on the ratios of the non-background subtracted intensities were not as effective.
For the MF-2 array data, the normalization methods do not significantly change the value of the T-statistic. The three-step normalization performed on the ratio of non-background subtracted intensities slightly increases the T-statistic. In fact the correlation coefficient of the log ratios and the estimated intensity bias and the correlation coefficient of the log ratios and the estimated spatial bias were both quite low for this array compared to the other arrays (below 15%). Also the background intensities for this array were quite low compared to the other arrays. This suggests that the reason for the lack of significant change in the T-statistic values after normalization is that the data from this particular array did not have significant bias.
Normalization of hybridization data from samples mimicking heterogeneous cell populations and single copy alterations
Array CGH is often used to detect genetic alterations in tumor cells. However, tumours generally consist of heterogeneous cell populations including a variety of infiltrating non-cancerous cells. Contamination from normal cells may affect the ability to detect copy number aberrations. In the case of a single copy gain, contamination from diploid normal cells dampens the expected 3:2 signal ratio produced by the single copy gained sequences in the tumour cells due to the averaging effect in the mixed cell population. In the case of a single copy loss, normal cell contamination increases the average copy number, deviating from the expected 1:2 ratio. In a previous study, this effect on detection sensitivity was evaluated by mixing male (XY) and female (XX) DNA in precise proportions to mimic 0%, 15%, 30%, 50% and 75% normal cell contamination affecting the dosage of the X chromosome [5].
In this study, we wish to determine how our three-step normalization method affects the estimated log2 ratios for the clones with single copy number changes and increasing levels of heterogeneity. The stepwise normalization method was applied to the data from the titration series (arrays T1-T10) that simulated different contamination levels for both single copy gains and losses (Fig. 5).
Figure 5 Normalization of hybridization data from samples mimicking heterogeneous cell populations and single copy alterations. Array CGH data were generated for samples mimicking single copy loss (deletion) or single copy gain (amplification) with contamination of increasing proportion of reference DNA, indicated as percentage on the horizontal axis. The experimental procedure for the array CGH experiments was previously described [5]. Global median normalization (method 1), stepwise normalization (method 12), global median normalization with background subtraction (method 13), and 3 step normalization with background subtraction (method 19) were applied. T-statistic values computed before and after normalization for arrays T1-T10 are summarized.
We compared the data obtained after performing the three-step normalization procedure to data obtained after performing global median normalization on both the ratios of background subtracted intensities and the ratios of non-background subtracted intensities. For each array, a T-test was performed on the two groups of log ratios, i.e. log ratios for the autosomal clones and those for the X chromosome clones. T-values are shown in Fig. 5.
The T-statistic values are higher after normalization in all cases which assures us that the separation of the two groups is increased and the low-level copy number changes are preserved and even magnified. Comparing the T-statistic values for data with no normalization to the normalized results shows that normalization increases the sensitivity of detection of the single copy number changes up to 5 times. However, the T-statistic values are considerably lower for the ratios of non-background subtracted intensities as compared to the ratios of background subtracted intensities.
Functional normalization increases the separation between the distributions of the clones with normal and abnormal copy numbers and this facilitates the analysis of heterogeneous samples. For example, after normalization, the T-statistic for array T9 which simulates a single copy amplification with 50% contamination, becomes quite close to the T-statistic of array T6 which simulates a case with no contamination.
Normalization of hybridization data from cDNA arrays simulating varying levels of gene amplification and deletion for X-chromosomal genes on the array
To evaluate the performance of the stepwise normalization strategy on hybridization data from cDNA arrays, we used public data from hybridization of genomic DNAs from cell lines containing varying numbers of X chromosomes that simulate varying levels of gene amplification and deletion for each of the X-chromosomal genes present on the array (arrays X1 to X5).
We compared the data obtained after performing the three-step normalization procedure to data obtained after performing global median normalization on both the ratios of background subtracted intensities and the ratios of non-background subtracted intensities. For each array, a T-test was performed on the two groups of log ratios, i.e. log ratios for the autosomal clones and those for the X chromosome clones. The T-statistic values are shown in Fig. 6.
Figure 6 Normalization of hybridization data from cDNA arrays. Array CGH data were generated for samples simulating varying levels of gene amplification and deletion for X-chromosomal genes on the array. Global median normalization (method 1), stepwise normalization (method 12), global median normalization with background subtraction (method 13), and 3 step normalization with background subtraction (method 19) were applied. T-statistic values computed before and after normalization for arrays X1-X5 are summarized.
The T-statistic values are higher after normalization in all cases. The increase in the T-statistic values may be interpreted as the increase in the separation of the distributions of the log2 ratios from two groups of normal and altered genes.
Other considerations
Visual comparison of the genomic profiles
The use of the genomic location of the clones allows us to compare profiles before and after normalization and to use the visual correlation between observed and expected profiles as a measure of success. (This is not possible when analyzing gene expression array data.)
In Figures 6A and 6B, chromosome plots of the data from two of the replicate H526 arrays, generated by SeeGH software [10], are shown. Chromosome plots show the log2 of ratios for each of the target DNA clones, as a function of the location of the clone in the chromosome. Figure 7A shows the chromosome plots for chromosome 1 of arrays H526-1 and H526-5. Figure 7B shows the chromosome plots for chromosome 2 of arrays H526-1 and H526-5. For each array and each chromosome the log2 ratios are shown after global median normalization and after the three-step normalization. The variability of log2 ratios in array H526-5 is much higher than that of array H526-1. For the H526 genome, the regions of copy number changes are known [4]. As the figures show, for data from array H526-5 (low quality data), normalization reduced the unwanted variations. Consequently, after normalization the altered regions are clearer. An important point to note for data from array H526-1 (high quality data), where the variation of the log2 ratios is quite low even before normalization, is that normalization did not remove the true biological variation present in the sample.
Figure 7 Chromosome plots before and after normalization. Plot of log2 signal ratios for clones (from chromosome 1 in A and chromosome 2 in B) versus their location across the chromosome. The profiles from left to right are: H526-1 data with global median normalization (method 1), H526-1 data with stepwise normalization (method 12), H526-5 data with global median normalization (method 13), H526-5 data with stepwise normalization (method 19). Each dot on the SeeGH plot represents a BAC clone. A shift in signal ratio to the left of center line indicates a copy number reduction, while a shift to the right indicates a gain. Blue arrow points to a high level segmental amplification. The arrow in part B points to the micro-amplification.
Background subtraction
The issue of subtracting or not subtracting background intensities has been an open question in microarray data analysis. Some groups choose to use the raw intensities while others use the background subtracted intensities. Through our experiments we observed that not subtracting the background results in slightly less variability and more repeatability of the ratios. However, knowing the truth about the ratios of array CGH experiments enabled us to examine how subtracting and not subtracting the background intensities affect the ability to detect copy number changes. We observed that for the array CGH data from SMRT arrays the ability to detect the copy number changes when using the ratio of non-background subtracted intensities is degraded when compared to using the ratio of background subtracted intensities. However, for the array CGH data from cDNA arrays, the ability to detect the copy number changes when using the ratio of non-background subtracted intensities is increased. We believe that the fact that different methods of background estimation are used in these two cases and the differences in the average level of background intensities of the arrays have caused this inconsistency between the results. The data from SMRT arrays along with the image analysis methods used suggest that background subtraction improves normalization and should be performed for these data.
Conclusion
We evaluated the performance and effectiveness of an integration of novel and existing bias removal methods mainly used for gene expression arrays considering the stringent performance requirements of the array CGH experiments and using the characteristics of the array CGH data that provide the true biological values for calibration.
A normalization scheme is expected to remove the systematic variations in the data and leave the true biological variations unchanged. In evaluating the performance of the normalization methods, both these issues should be considered. Our method is shown to preserve even the low-level copy number changes while reducing the systematic biases. To the best of our knowledge this is the first study to examine the effectiveness of various normalization methods taking advantage of the knowledge of the underlying truth in known copy number status in genomic array CGH data – as opposed to using variable gene expression changes in normalizing expression microarray data. Our stepwise normalization framework estimates the intensity dependent, spatial and plate bias using regression-based techniques and removes the estimated biases from the raw log2 ratios. These biases were observed in two different array CGH platforms, the SMRT BAC arrays [4] and cDNA arrays [12]. Our results demonstrate that multi-step normalization outperforms conventional single step methods in reducing systematic biases in array CGH spot data from both BAC and cDNA platforms (such as those representing self-self hybridization, replicated experiments, single copy detection, and data mimicking tissue heterogeneity) and suggest that multiple systematic variations need to be addressed in the normalization of genomic array CGH data.
In this study we focused on within-array normalization and did not consider performing between-array normalization. This was based on the fact that because of tissue heterogeneity, there is usually some degree of contamination from normal cells into the tumor cells in array CGH experiment samples. As a result, it is not known that a single copy change results in how much change in the fluorescent ratios [5]. Because of this, it seems that the safest way to deal with the issue of unequal scales of data from different arrays would be to find the regions of gains or losses in DNA copy number according to data from one array CGH experiment and assign different levels of change to those different regions. These levels may then be compared across arrays.
Methods
Microarray image analysis
SMRT arrays
Hybridized arrays were imaged using a charge-coupled device based imaging system and analyzed using the SoftWorx Tracker spot analysis software (ArrayWorx eAuto, API, Issaquah, WA). The mean pixel intensity was used for the spot foreground intensities and the median pixel intensity was used for the spot background intensity. Background calculation was achieved using the "Cell method" in the SoftWorx Tracker program. In this method, a square of 125% size of spot spacing is drawn and centred on the centroid of the spot's contour. All pixels within the square which are not located within the two-pixel margin of the spot's contour are treated as background pixels for that spot.
cDNA arrays
The image analysis methods are described in [12]. For computing the fluorescence ratios, the mean pixel intensity was used for the spot foreground intensities and the median pixel intensity was used for the spot background intensity.
Normalization methods
The normalization methods that were used in this study are listed in Table 2. Among these methods, global intensity LOWESS, Median Plate Ratio and Spatial normalization have been described in the text above. For comparison purposes print tip LOWESS intensity normalization [6] is also implemented, performing LOWESS curve fitting on log ratios from each subgrid of the microarray.
LOWESS (Locally Weighted Scatter plot Smoothing) is a curve-fitting technique based on local regression [8]. Each smoothed value is determined by its neighbouring data points defined within the span. A regression weight function is defined for the data points contained within the span. In addition to the regression weight function, a robust weight function may be used, which makes the process resistant to outliers. In this study, we used a robust LOWESS with a first degree polynomial for regression.
Evaluation methods
Pearson's correlation coefficient
In the analysis of replicated experiments (H526-1 through H526-8), the Pearson correlation coefficient is calculated for all 28 possible pairings of the 8 replicate arrays. For each pair wise comparison, the log2 ratios for spots from one array form the first group, and the log2 ratios for the corresponding spots from the other array form the second group.
Intra-class correlation coefficient (ICC)
This is an ANOVA-based type of correlation. It measures the relative homogeneity within groups compared to their total variation. Suppose that we have k groups of measurements and each group consists of n replicate measurements. Xi,j, i = 1,..,k and j = 1,..,n represents the j-th measurement in the i-th group. If we define:
X¯=∑i=1k∑j=1nXi,jnk
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuWGybawgaqeaiabg2da9maalaaabaWaaabCaeaadaaeWbqaaiabdIfaynaaBaaaleaacqWGPbqAcqGGSaalcqWGQbGAaeqaaaqaaiabdQgaQjabg2da9iabigdaXaqaaiabd6gaUbqdcqGHris5aaWcbaGaemyAaKMaeyypa0JaeGymaedabaGaem4AaSganiabggHiLdaakeaacqWGUbGBcqWGRbWAaaaaaa@44C1@
X¯i=∑j=1nXi,jn
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacuWGybawgaqeamaaBaaaleaacqWGPbqAaeqaaOGaeyypa0ZaaSaaaeaadaaeWbqaaiabdIfaynaaBaaaleaacqWGPbqAcqGGSaalcqWGQbGAaeqaaaqaaiabdQgaQjabg2da9iabigdaXaqaaiabd6gaUbqdcqGHris5aaGcbaGaemOBa4gaaaaa@3E01@
RSS=∑i=1k(X¯i−X¯)2
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGsbGucqWGtbWucqWGtbWucqGH9aqpdaaeWbqaamaabmaabaGafmiwaGLbaebadaWgaaWcbaGaemyAaKgabeaakiabgkHiTiqbdIfayzaaraaacaGLOaGaayzkaaWaaWbaaSqabeaacqaIYaGmaaaabaGaemyAaKMaeyypa0JaeGymaedabaGaem4AaSganiabggHiLdaaaa@3FEE@
TSS=∑i=1k∑j=1n(Xi,j−X¯)2
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGubavcqWGtbWucqWGtbWucqGH9aqpdaaeWbqaamaaqahabaWaaeWaaeaacqWGybawdaWgaaWcbaGaemyAaKMaeiilaWIaemOAaOgabeaakiabgkHiTiqbdIfayzaaraaacaGLOaGaayzkaaWaaWbaaSqabeaacqaIYaGmaaaabaGaemOAaOMaeyypa0JaeGymaedabaGaemOBa4ganiabggHiLdaaleaacqWGPbqAcqGH9aqpcqaIXaqmaeaacqWGRbWAa0GaeyyeIuoaaaa@4911@
SSE = TSS - RSS
MSbetweengroups=RSSk
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGnbqtcqWGtbWudaWgaaWcbaGaemOyaiMaemyzauMaemiDaqNaem4DaCNaemyzauMaemyzauMaemOBa4wbaeqabeqaaaqaaiabdEgaNjabdkhaYjabd+gaVjabdwha1jabdchaWjabdohaZbaaaeqaaOGaeyypa0ZaaSaaaeaacqWGsbGucqWGtbWucqWGtbWuaeaacqWGRbWAaaaaaa@474A@
MSwithingroups=SSEnk−k
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWGnbqtcqWGtbWudaWgaaWcbaGaem4DaCNaemyAaKMaemiDaqNaemiAaGMaemyAaKMaemOBa4wbaeqabeqaaaqaaaaacqWGNbWzcqWGYbGCcqWGVbWBcqWG1bqDcqWGWbaCcqWGZbWCaeqaaOGaeyypa0ZaaSaaaeaacqWGtbWucqWGtbWucqWGfbqraeaacqWGUbGBcqWGRbWAcqGHsislcqWGRbWAaaaaaa@49AA@
then rICC is calculated from the following formula:
rICC=MSBetweengroups−MSWithingroupsMSBetweengroups+(n−1)∗MSWithingroups
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@8929@
The maximum positive value of the intra-class correlation coefficient is 1.0, but its maximum negative value is (-1/(n-1)). Intra-class correlation coefficient is large and positive when there is no variation within the groups, but the group means differ. Intra-class correlation coefficient is large and negative when the group means are the same but there is great variation within groups. A negative intra-class correlation occurs when between-group variation is less than within-group variation [9]. ICC was shown to be useful for the assessment of technical and biological variations in microarray experiments [11].
In evaluating the normalized data from the replicate arrays (H526-1 through H526-8), n is the total number of replicate arrays which is 8 and k is the total number of clones on each array, and Xi,j represents the estimated log2 ratio for the j-th clone on the i-th array.
T-test
A two-sample two-tailed T-test was used to determine whether two samples (with different numbers of observations) from a normal distribution (in x and y) could have the same mean when the standard deviations are unknown but assumed equal.
In the analysis performed on third, fourth, and fifth datasets (MF-1 to MF-2, T1 to T10, and X1 to X5), for each array dataset, log2 ratios for autosomal clones represent the first sample and log2 ratios for clones from chromosome X represent the second sample.
Authors' contributions
MK developed and implemented the methods, participated in the design of the study, and drafted the manuscript. WL provided expertise on the array CGH platform. RW participated in the coordination of the study. CM participated in the design, development and coordination of the study. All authors read and approved the manuscript.
Supplementary Material
Additional File 1
Supplemental table 1. A description of array CGH experiments that simulate varying degrees of normal diploid cells contamination in a population of cancer cells carrying a single copy alteration. A more detailed description can be found in [5].
Additional file 2 - Supplemental table 2
Click here for file
Additional File 2
Supplemental table 2. A Description of array CGH experiments involving hybridization of genomic DNAs from cell lines containing varying numbers of X chromosomes that simulate varying levels of gene amplification and deletion for each of the X-chromosomal genes present on the cDNA array. A more detailed description can be found in .
Additional file 2 - Supplemental table 2
Click here for file
Acknowledgements
The authors wish to thank Bradley Coe for his useful discussions, Spencer Watson for providing array CGH data and Ron J DeLeeuw for careful proofreading of the manuscript. This work was supported by funds from the Canadian Institute of Health Research, Genome British Columbia/Genome Canada and the National Institute of Dental and Craniofacial Research Grant R01 DE015965.
==== Refs
Snijers AM Nowak N Segraves R Blackwood S Brown N Conroy J Hamilton G Hindle AK Huey B Kimura K Law S Myambo K Palmer J Ylstra B Yue JP Gray JW Jain AN Pinkel D Albertson DG Assembly of microarrays for genome-wide measurement of DNA copy number Nat Genet 2001 29 263 4 11687795 10.1038/ng754
Schena M Shalon D Davis RW Brown PO Quantitative monitoring of gene expression patterns with a complementary DNA microarray Science 1995 270 467 470 7569999
Draghici S Statistical intelligence: effective analysis of high-density microarray data Drug Discov Today 2002 7 S55 63 12047881 10.1016/S1359-6446(02)02292-4
Ishkanian AS Malloff CA Watson SK DeLeeuw RJ Chi B Coe BP Snijders A Albertson DG Pinkel D Marra MA Ling V MacAulay C Lam WL A tiling resolution DNA microarray with complete coverage of the human genome Nat Genet 2004 36 299 303 14981516 10.1038/ng1307
Garnis C Coe BP Lam SL MacAulay C Lam WL High-resolution array CGH increases heterogeneity tolerance in the analysis of clinical samples Genomics 2005 85 790 3 15885505 10.1016/j.ygeno.2005.02.015
Smyth GK Speed T Normalization of cDNA microarray data Methods 2003 31 265 73 14597310 10.1016/S1046-2023(03)00155-5
Watson SK deLeeuw RJ Ishkanian AS Malloff CA Lam WL Methods for high throughput validation of amplified fragment pools of BAC DNA for constructing high resolution CGH arrays BMC Genomics 2004 5 6 14723794 10.1186/1471-2164-5-6
Cleveland WS Robust Locally Weighted Regression and Smoothing Scatter plots Journal of the American Statistical Association 1979 74 829 836
StatSoft, Inc Electronic Statistics Textbook 2004 Tulsa, OK: StatSoft
Chi B DeLeeuw RJ Coe BP MacAulay C Lam WL SeeGH – a software tool for visualization of whole genome array comparative genomic hybridization data BMC Bioinformatics 2004 5 13 15040819 10.1186/1471-2105-5-13
Pellis L Franssen-van Hal NL Burema J Keijer J The intraclass correlation coefficient applied for evaluation of data correction, labeling methods, and rectal biopsy sampling in DNA microarray experiments Physiol Genomics 2003 16 99 106 14570982 10.1152/physiolgenomics.00111.2003
Pollack JR Sorlie T Perou CM Rees CA Jeffrey SS Lonning PE Tibshirani R Botstein D Borresen-Dale AL Brown PO Microarray analysis reveals a major direct role of DNA copy number alteration in the transcriptional program of human breast tumors Proc Natl Acad Sci USA 2002 99 12963 8 12297621 10.1073/pnas.162471999
|
16297240
|
PMC1310623
|
CC BY
|
2021-01-04 16:27:47
|
no
|
BMC Bioinformatics. 2005 Nov 18; 6:274
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-274
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1531627114810.1186/1471-2164-6-153Methodology ArticleA rigorous method for multigenic families' functional annotation: the peptidyl arginine deiminase (PADs) proteins family example Balandraud N [email protected] P [email protected] EGJ [email protected] M [email protected] D [email protected] J [email protected] P [email protected] EA 3781, Evolution Biologique, Université de Provence, 3 pl. V. Hugo, 13331 Marseille Cedex 03, France2 INSERM UMR 639, Laboratoire d'Immunogénétique de la polyarthirte rhumatoïde, faculté de médecine la Timone, 13005 Marseille, France3 AFMB-UMR 6098-CNRS (U1 – U2) Glycogenomics and Biomedical Structural Biology, Case 932, 163 Avenue de Luminy, 13288 Marseille cedex 09, France2005 4 11 2005 6 153 153 2 8 2005 4 11 2005 Copyright © 2005 Balandraud et al; licensee BioMed Central Ltd.2005Balandraud et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
large scale and reliable proteins' functional annotation is a major challenge in modern biology. Phylogenetic analyses have been shown to be important for such tasks. However, up to now, phylogenetic annotation did not take into account expression data (i.e. ESTs, Microarrays, SAGE, ...). Therefore, integrating such data, like ESTs in phylogenetic annotation could be a major advance in post genomic analyses. We developed an approach enabling the combination of expression data and phylogenetic analysis. To illustrate our method, we used an example protein family, the peptidyl arginine deiminases (PADs), probably implied in Rheumatoid Arthritis.
Results
the analysis was performed as follows: we built a phylogeny of PAD proteins from the NCBI's NR protein database. We completed the phylogenetic reconstruction of PADs using an enlarged sequence database containing translations of ESTs contigs. We then extracted all corresponding expression data contained in EST database This analysis allowed us 1/To extend the spectrum of homologs-containing species and to improve the reconstruction of genes' evolutionary history. 2/To deduce an accurate gene expression pattern for each member of this protein family. 3/To show a correlation between paralogous sequences' evolution rate and pattern of tissular expression.
Conclusion
coupling phylogenetic reconstruction and expression data is a promising way of analysis that could be applied to all multigenic families to investigate the relationship between molecular and transcriptional evolution and to improve functional annotation.
==== Body
Background
The "in silico" functional annotation of proteins generated by large scale sequencing projects is an important challenge in biology. Here we propose a rigorous protocol for multigenic families' annotation.
1/Importance of phylogenetic reconstruction
Because important functions are conserved during evolution, the first step in analysis is to determine homologous sequences. More specifically, orthologs are more likely to share the same function while paralogs can undergo functional shifts. Developments and improvements in database similarity search programs such as BLAST [1] allowed the rapid identification of potential homologous sequences and therefore allowed functional prediction of several thousand of genes and proteins present in databases. However, the closest BLAST is often not the nearest neighbor [2]. Indeed, similarity-based approaches do not consider all the information from comparative and evolutionary biology. They do not differentiate between orthologs and paralogs among homologs. So, phylogenetic approaches, taking into account duplication and speciation events are necessary to robustly produce functional annotation of new, uncharacterized proteins [3-8].
2/Necessity to enlarge databases
For proteins' phylogenetic reconstruction, protein databases containing the proteomes of completely sequenced species along with individually submitted protein sequences are usually used (Ensembl protein db, NCBI Protein db etc...) [9,10]. Yet, the vast majority of species are not fully sequenced and most of their protein sequences are still unknown. However, a lot of transcriptional information is carried by growing gene expression databases, concerning normal or pathological tissues (Expressed Sequence Tags from NCBI, TIGR, GeneNote, Gepis etc...) [11-14]. These mRNAs could be used for (total or partial) reconstruction of unknown proteins in "not yet sequenced" species. In parallel, translation of EST contigs can be used to enlarge the spectrum of species containing homologs when one analyses a protein family.
3/Importance of expression patterns' determination for an accurate annotation
It should be noted that phylogenetic analysis can only give information at the biochemical function level. Furthermore, while orthologs can have very similar "molecular function", they can exhibit different "macroscopic functions", due to a transcriptional shift for example. To produce an accurate proteic functional annotation, one must have complete sequence information given by phylogenetic reconstruction, with expression patterns analysis. This is the second reason why using data from expression databases is interesting. Analysis of expression divergence between paralogs and orthologs have been recently published in Human and Mouse. It appears that gene expression profiles diverge between paralogs. Orthologs can diverge in their expression pattern too [15]. Moreover, orthologs that have undergone recent duplication have less strongly correlated expression profiles than those that have not [16]. Up to now, there is no study examining expression divergence between homologs that takes into account a broad spectrum of species and after a phylogenetic reconstruction.
4/Our approach
We present here a new way to functionally annotate proteins in silico, taking into account all these concepts. In a first step, we reconstructed the phylogeny of a protein family, using an enlarged database containing 1/full length proteins from NCBI NR protein database [10] and 2/translation of EST contigs from NCBI dbEST database [11]. We used a new software platform, FIGENIX [17], adapted to this kind of phylogenomic reconstruction. In a second step, we created an automated pipeline to couple these phylogenetic reconstructions with expression pattern data. We then compared the evolution rate of the different paralogous sequences with their patterns of expression.
We validated this approach with the peptidyl arginine deiminase multigenic family. These genes encode peptidyl arginine deiminase proteins (PADs) that are implied in Rheumatoid Arthritis (RA), a systemic autoimmune disease that primarily manifests as a chronic symmetric polyarthritis which gradually destroys articular cartilage and bone [18,19]. PADs transform Arginine residues into Citrulline in a calcium-dependant manner [20]. It was recently shown by a case-control linkage disequilibrium study that PAD type 4 is a susceptibility locus for rheumatoid arthritis in a Japanese population [21]. These data were not retrieved in a Caucasian population [22]. Up to now, no phylogenomic annotation coupled with expression data of this protein family has been proposed. We believe this analysis will shed new light on this incompletely characterized gene family.
Results
1/Phylogenetic reconstruction (See material and methods for process)
First, we built a phylogenetic tree with full length PAD proteins (from NCBI NR protein database) (Figure 1). We found five paralogs of PAD proteins (types 1, 2, 3, 4, 6) encoded by five different genes as previously described [23]. We found PAD members in Birds (Gallus gallus), Amphibians (Xenopus laevis) and Teleosteans (Danio rerio). This tree supported by high bootstrap values suggested the following steps for evolution of PAD genes in craniates: PADs' common ancestor was present as far back as in the last common ancestor of Teleostean and Mammals. PAD-2 first diverged by duplication from the common ancestor before the radiation of mammals.
Figure 1 Phylogenetic tree built with PAD and NR database, with Danio-rerio's PAD sequence in out-group. This tree is the fusion on the NJ topology, of three phylogenetic trees built based on Neighbour Joining (NJ) [36], Maximum Parsimony (MP) [39], and Maximum Likelihood (ML) [40] methods. The tree is labelled npl-A at the first root (for Nj, mP, mL, with All topology congruence tests passed). For each node, bootstrap values are reported for each npl method. A "*" means that the bootstrap value was under 50%. PAD protein from Danio-Rerio is co-orthologous to all others PADs. In mammals five groups of orthologs are found, named PAD-1, PAD-2, PAD-3, PAD-4 and PAD-6. PAD-2 paralogy group is the best conserved as it evolved slower than the others.
We next built a phylogeny with the same protein dataset completed by additional proteins built from translations of EST contigs. First, BLAST searches on EST databases allowed us to identify other species in which PAD proteins are likely present. We found PADs in additional Teleosteans (Fugu rubripes, Salmon, Ictularus punctatus, Onchorynchus-mykiss, Gasterosteus aculeatus) and in Cephalochordates (Branchiostoma Floridae). Because ESTs are not full length sequences, they can not be used in this raw state to reconstruct a phylogenetic tree. To correct for this, we built larger sequences by merging ESTs relative to the same gene into contigs. EST contigs were translated and included in a local database and used for phylogenetic reconstruction along with protein from NCBI NR database. Because ESTs contigs often do not correspond to full length transcripts, we had to build several trees with alignment only on parts of the protein (five artificial domains). A tree built for each domain is available in additional file online [see additional file]. Some ESTs could not be integrated into a contig (these ESTs are called singletons) and were used in this state as sequence queries to reconstruct a phylogeny. An example of a tree built with a singleton (an Amphioxus sequence) is shown in Figure 2.
Figure 2 Phylogenetic reconstruction with translated ESTs singleton. One of the phylogenetic trees built with PAD and enlarged database (NR database and local translated EST contigs database). The tree was built with translated EST from PAD Amphioxus as query sequence, which was not included in contigs and in the five arbitrary domains. Building method is the same as in Figure 1.
All the trees built with partial sequences were validated only if they were in agreement with the topology of the reference tree (Figure 1). This allowed us to classify all the contigs and singletons into paralogy or orthology groups. This analysis allowed us to propose a more detailed evolutionary story for the PAD genes: the common ancestor of the PAD gene was at least present in the common ancestor of the Euchordates (Cephalochordates and Craniates) (see tree of life in Additiona file online). It means that the PAD gene present in Amphioxus is co-orthologous to all PADs (as this is also the case for PAD genes in Teleostan). Analysis of trees' topologies allowed us to conclude that duplications of PADs certainly occurred after the speciation of craniata or more probably after the speciation of Sarcopterygian, but before the mammals' radiation. So, phylogenetic trees including translated EST contigs allowed us to enlarge the dataset of species and to classify these "new proteins" into a paralogy group.
When looking at the branches lengths, which are directly correlated with the sequence evolution rate, it appeared that some members of PAD genes evolved faster than the others. Branches lengths were longer within the PAD-6 paralogy group (Homo sapiens/Mus musculus/Rattus norvegicus PAD-6) whereas branches lengths were the shortest within the PAD-2 paralogy group (Homo sapiens/Mus musculus/Rattus norvegicus PAD-2). It simply means that the similarity between species is higher for PAD-2 than it is for PAD-6. This indicates that PAD-2s' group is the most highly conserved. Furthermore, the distance from the common ancestor is lower for PAD-2s' group than it is for the others, suggesting that PAD-2's group has been submitted to high negative selective pressure and has probably retained the ancestral biochemical function.
2/Coupling expression pattern with phylogenetic classification
The phylogenetic reconstruction allowed us to classify all the contigs or singletons in phylogenetic groups for the PADs phylogeny. We then developed software agents to extract transcriptional data (pattern of tissular expression) from the NCBI dbEST database [11]. We first normalized expression data (relative to the number of clones in each library), because expression data are not normalized in NCBI dbEST. We then built a table classifying ESTs according to their position in the tree (group of paralogs, orthologs or co-orthologs) and to the tissue in which they are expressed. Results are illustrated in a simplified Figure 7. Raw data are in Table 2 [see Additional file]. Given the current EST coverage, we focused on human and mouse. The bias could occur when, say, the coverage of the zebrafish is much lower than Human or Mouse. EST data are all represented in the table, but statistical analysis took into account only Human and Mouse data. For Homo sapiens and Mus musculus, data are congruent with expression patterns data available on UniGene [24].
Figure 7 PADs paralogs pattern of expression. Proteins are classified as belonging to one of the four paralogy groups (PAD 1, 2, 3, 4, 6) in columns. Lines correspond to tissue expression and cell categories. Numbers correspond to normalized number of hit (means of number of hit/number of clone × 1 000 000). Tissues are separated in adult tissues, cells and foetal tissues. In order to compare the expression levels from EST libraries between the different paralogous copies of PAD genes, we used the statistical test described by Audic and Claverie [25]. For each tissue type and cell category we compared the EST expression data (number of hits/number of clones in the considered library) for a given species between the different paralogous copies of PAD genes. NS: not significant. ** statistically significant difference.
Coupling phylogenetic classification with expression data permitted us to perform updated footprints of the transcriptional pattern for each paralogs group in this multigenic family. One group of paralogs, PAD-2, showed a broad tissular expression. As the expression pattern was different across paralogous PAD genes, we could only compare a few tissue types for differential levels of expression. The statistical test described by Audic and Claverie [25] was used. Significant differences (p > 95%) were shown in thymus, whole blood, lung, inner ear, mixed liver and spleen library and leucocytes (see Figure 7).
3/Coupling sequence evolution rate and expression pattern
When observing tissue distribution, we noticed that it was conversely correlated (Correlation coefficient R = 0.76) to sequence evolution rate (branches lengths) (see Figure 4). This was remarkable when we compared branches lengths and tissue distribution between PAD-2 (shortest branches, broadest tissue distribution) and PAD-6 (longest branches, more limited tissue distribution) (see Figure 3 and Figure 4).
Figure 3 Phylogenetic relationships and expression profiles in PADs family. This is a fusion tree of three methods (Neighbor joining, maximum parsimony and maximum likelihood). Bootstrap values are calculated for these three methods. Branches length is correlated to sequence evolution rate. Branches lengths between PAD-6 paralogy group members are longer than between PAD-2 paralogy group members. Numbers in small squares correspond to different adult tissues: 1/bone, 2/brain, 3/colon, 4/eye, 5/kidney, 6/liver, 7/lung, 8/mammary gland, 9/pancreas, 10/placenta, 11/skin, 12/thymus, 13/uterus 14/ovary, 15/inner ear, 16/olfactory tissue. Grey squares indicate the presence of expression in the corresponding tissue in updated analysis (January 2005). White squares indicate no expression has been found in this tissue. Red circles indicate duplication.
Figure 4 Correlation between branches length and PAD tissular expression: Branch lengths correlate with tissue expression as shown by the linear regression line. The coefficient of correlation R is 0.76.
Discussion
1/Using phylogenomics and data from EST databases
Using EST data to complete data from protein databases is commonly performed. Yet, if one uses only the best BLAST hits approach to classify ESTs data, there is a large risk of misclassification into paralogy or orthology groups. Thus, expression data for a given paralogy group could be undermined by the wrong assignment of EST data into a group. In our example, we found additional expressed sequences from Teleostean and Cephalochordates in EST databases. When performing analysis with reciprocal best BLAST hit on NR protein database; EST from Amphioxus (cephalochordates) and Teleostean is classified as an ortholog of PAD-2 only, but not as a co-ortholog of all PADs as is shown by phylogenetic analysis. This illustrates why phylogenetic reconstruction is essential for accurate EST analysis and able to better avoid misclassification. Our approach enlarges the dataset of homologs-containing species and robustly and accurately classifies multigenic family members into paralogy and orthology group. Comparison of our results to those of Chavanas et al. [23], showed that our method allowed a more complete reconstruction of these genes' evolutionary histories, including a broader spectrum of species (such as cephalochordates). Our evolutionary scenario of duplications which occurred after the radiation of mammals is consistent with the co-localization of PAD genes in the same genomic region (1p35 in Human, chromosome 4E1 in Mus musculus and chromosome 5q36 in Rattus norvegicus), which suggests cis-duplications.
Notice that one third of reliably-inferred alternative mRNA isoforms are candidates for nonsense-mediated mRNA dacay (NMD), an mRNA surveillance system [26]. These transcripts, rather than being translated into proteins, are expected to be degraded and may be subject to regulated unproductive splicing and translation. In our study, we did not take into account this NMD-type alternative splicing, and took all information (assembling all exons) to reconstruct the most parsimonious story of a gene family. however, when looking at the overall data, we could suppose that some EST found in some tissus are not translated into protein but are regulatory elements.
Phylogenetic trees show that some PAD family members evolve faster than others, with the longest branches lengths seen in the PAD-6 paralogy group and the shortest branches lengths in the PAD-2 paralogy group. These data suggested that PAD-2s have probably kept the ancestral function and that other PADs may have run a neofunctionalization, still unknown. For example, the biochemical condition (Calcium concentration or pH) for enzyme activation could be different according to each PADs. Such an hypothesis has just been supported by Yamamoto's group who showed small different enzymatic properties between PAD-2 and PAD-4 [27].
Any kind of expression data could be used to enhance the accuracy of the analysis. The method used by Xun Gu et al. and developed in Genetics [28] and PNAS [29] is particularly interesting because they used microarray data in yeast to develop a statistical framework for studying the evolution of genes after duplications. On the basis of a Bayesian-based method, they reconstructed the evolutional trace of expression diversity after gene duplication. Their conclusions are in agreement with ours, showing that the expression of duplicate genes tends to evolve asymmetrically, one copy maintaining the ancestral expression profile while the others evolve rapidly.
2/When analyzing transcriptional data (Figure 7), we confirmed that transcriptional shift occurred between paralogs as previously shown [15,30]. PAD-2 shows a widespread expression whereas PAD-6 is expressed essentially during embryonic development. PAD-3, -4 and -1 have a restricted expression pattern (respectively thymus, eye, skin, ear and uterus for PAD-3, thymus, bone marrow, skin, breast, lung spleen, aorta and vein for PAD-4 and eye and skin for PAD-1). For orthologous genes, it is very difficult to compare different tissues in different species because 1/analysis is never done under the same experimental conditions and 2/one can never be sure that tissues evolved under the same physiological conditions in the different species.
3/Our analysis allowed us to correlate (R = 0.76) sequence evolution rate and tissular expression distribution of paralogs (figure 3 and 4). It is interesting to notice that shorter branches lengths are correlated with ubiquitous tissular expression. One can hypothesize that one copy (PAD-2) has probably kept the ancestral function, with ubiquitous expression. Other copies may have slightly different biochemical function, and underwent shifts in their expression pattern which is different and more limited. These data are consistent with studies showing that housekeeping genes evolve more slowly than tissue specific genes, and are under stronger specific constraints [31,32]. Our data suggest that the same phenomena could exist in multigenic families but, it must be confirmed in other examples.
Conclusion
We believe our method of adding ESTs and expression data to phylogenetic analysis provides a new way to annotate multigenic families. More than classical phylogeny, it allows highlighting of transcriptional shifts between paralogs. It shows that functional shifts can occur in differential tissue expression rather than in biochemical function of the protein. It also shows a correlation between lower sequence evolution rate (branches length) and larger tissular expression distribution. For drug development, it points out the fact that when one analyses an orthologous protein in a species phylogenetically close to humans, one should keep in mind that tissular distribution of a protein can be different between species before extrapolation of the function to human.
This type of analysis is in its infancy and must be extended to other multigenic families and to all kinds of expression databases, including database where expression data are normalized such as in UniGene, SAGE and Micro arrays. In the future, the availability of more and more sequence information from different species will enable tracing a genome's evolutionary history, down to the expression data level, with more accuracy.
Methods (cf. Figure 5)
Figure 5 Methods. This picture illustrates the different steps of the software.
1-Phylogenetic reconstruction with full-length proteins, from NCBI NR database
PAD-2 was chosen as the query sequence for the first phylogenetic analysis. To build the phylogeny, the following steps, as described elsewhere [4] were used:
BLAST
Search BLAST hits on NR protein database [10] with e-value filter set to 1e-4.
Multiple alignment
Elimination of sequences disturbing alignments and doubles. Calculation of an alignment with CLUSTALW software [33]. Elimination of large gaps. Domains searching with HMMPFAM software [34]. Composition test for residues on Tree-Puzzle [35]. For each domain, elimination of non-monophyletic "repeats" in a tree built with NEIGHBOR JOINING algorithm on CLUSTALW software [36]. Elimination of sequences with divergent residue composition by using the amino-acid composition test from TREE-PUZZLE software [35].
Elimination of sites not under neutrality (fast evolving sites) by statistical methods for testing functional divergence after gene duplication [37].
Selection of congruent domains with HOMPART test from PAUP package [38], and building of a new alignment by merging preserved parts of domains' alignments.
Tree building
From this alignment, three phylogenetic trees were generated using the following methods: Neighbor Joining [36], Maximum Parsimony in Paup [39], Maximum Likelihood in Puzzle [40]. By comparing topologies of these trees with PSCORE command (Templeton winning sites test [41,42]) from PAUP package and Kishino-Hasegawa test [43] from TREE-PUZZLE package, production of a fusion of these trees in a unique consensus tree according to the results of these tests.
Comparison of this fusion tree with the NCBI tree of life, , allowed deducing paralogous and orthologous proteins, using the algorithm described by Zmasek et al. [44]. Trees were rooted with the midpoint method. A fusion of the congruent three trees is noted NPL-A at the first root (for Nj, mP, mL, with All topology congruence tests passed). Respective bootstrap values are noted in front of each node.
Analysis of the tree and rerooting
Our phylogenies are built with the midpoint method rooting. In order to avoid misclassification induced by fast evolving sites, sites which are not under neutrality are generally eliminated, using statistical methods for testing functional divergence after gene duplication [37]. In the phylogenetic "tree of reference", fast evolving sites could not be removed by Gu software because not enough sequences were present in the different paralogy groups. So, the PAD-6 paralogy group, which evolved faster than the others was automatically placed in out-group to equilibrate the tree. This was not necessarily reflective of the true evolutionary history of the PAD family. Indeed, when looking at the branches lengths between species belonging to the same paralogy group, particularly Mus musculus, Rattus norvegicus and Homo sapiens, we can deduce that duplication of the PAD ancestor occurred after the separation of Sarcopterygii and Actinopterygian. All of the species external to these groups in the tree of life were placed in out-group.
Indeed, if one placed PAD-6 in the out-group, the conclusion concerning PAD' evolutionary events would change. One would conclude that PADs duplicated before separation of Euchordates, but not after the emergence of Sarcopterigian. In this hypothesis, Cephalochordates and Teleosteans PADs reside in the PAD-2 paralogous group. This scheme is based on the hypothesis that PAD-6, -1, -3, -4 were all lost during evolution in Teleosteans and Cephalochordates. The tree built with PADI-6 in the out-group is farless parsimonious with regards to evolutionary events and is deducted from too many supposed gene losses. The same conclusions could be arrived at with other PAD groups placed in the out-group. So, a second tree, more parsimonious regarding evolutionary events was built with an out-group species rooting (Figure 1).
All this phylogenetic reconstructions were performed with the FIGENIX automated platform [17].
2-Building the tree including translations of ESTs contigs
In a second step, the tree was completed with sequences extracted from EST database.
BLAST
We first searched by BLAST request in NCBI dbEST database, sequences similar to PADs (result: 411 hits for e value 1e-4).
Creation of a local database
These short ESTs were put in a local database. This local database was then completed by mRNA (CDS) of completely sequenced proteins (NCBI protein db).
Contigs generation
Because ESTs are not full length mRNAs of known proteins, they cannot be included in a phylogenetic building in this raw state. We built a specific protein database using EST clustering. In order to reduce sequencing error, 10% of the EST sequence length was eliminated from each tip and only the 500 core central nucleotides were kept. Each species' EST group was analyzed with CAP3 (ESTs clustering software) [45] to produce RNA contigs (30 contigs constructed, 17 singletons). These RNA sequences were translated into proteins (6 open reading frames).
Phylogenetic reconstruction
Phylogenetic reconstruction was performed with the same method as described above. The phylogenetic analysis was at first built with the whole PAD-2 protein as the query sequence.
Some short sequences (not full length contigs) were eliminated during the phylogenetic process. For these sequences, phylogenetic trees were built using five arbitrary created domains of PAD-2 (three equal sequences and two overlapping sequences), used as the sequence query (Figure 6).
Figure 6 Five arbitrarily created domains of the protein PAD-2 (three equal sequences and two overlapping sequences). Each sequence was used as sequence query.
Using this method, most of the proteins translated from EST contigs could be analyzed and classified in a group of paralogs. However, 15 sequences were not overlapped by these arbitrary domains. These singletons were analyzed as sequence query in separated phylogenies. The trees were validated only if their topologies were similar to the topology of the reference tree (Figure 1). As an example, we provide a tree built with a protein from an Amphioxus PAD EST singleton (Figure 2). EST contigs incorporated in the trees are grouped with complete sequences (ex: contig: 14-1 in Mus musculus was close to complete PAD-3 protein of Mus musculus. It means that contig 14-1 contains ESTs from PAD-3 protein). Five trees built with this method are available on the supplementary material on line.
Transcriptional pattern analysis
Phylogenetic analysis allowed classification of each EST into a paralogy group. Information concerning tissue or organ provenance is given for each EST in the NCBI dbEST database [11]. We developed software "agents" in FIGENIX platform to automatically collect information corresponding to each EST. This expression information is: ESTs content of each contigs and tissue or organ expression corresponding to each EST. These data were pooled and classified in a summarizing table (Figure 7). For an accurate analysis of expression, we normalized expression data relative to the number of clones in each library (results are given in number of hit for 1 000 000 clones). Data were congruent with expression profile available for Homo sapiens and Mus musculus available in UniGene [24].
Differential expression significance and data consistency (See Figure 7 and Table 2 in supplementary material on line)
In order to compare the expression levels from EST libraries between the different paralogous copies of PAD genes, we used the statistical test described by Audic and Claverie [25]. For each tissue type and cell category, we compared the EST expression data (number of hits/number of clones in the considered library) for a given species between the different paralogous copies of PAD genes.
When a library was shared between two paralogous copies, we directly compared the two ratios. The threshold used for differential expression was set to p > 95% either for under or over expression.
When several different libraries were hit by a PAD gene for a given tissue or cell type, we first checked whether each library was consistent with all the other ones by applying the same statistical test [25]. When a given library was different from more than 50% of other libraries (using a threshold of p < 60% for over or under expression), we removed the significantly different library.
Only tissue and cells not associated with a pathological state were considered for comparison, as expression can be altered in cancers and other diseases. Similarly, genetically modified or selected mice variants were removed from the comparison as they may not reflect normal expression.
Authors' contributions
Nathalie Balandraud initiated the study, collected the data and performed most steps of bioinformatic and statistical analysis presented in the paper. She developed the initial framework for the analysis of the EST genomic sequences. She produced all the phylogenies, all the Figures in the paper and Table 1, and wrote the manuscript. Philippe Gouret designed and supervised all the bioinformatic study with the creation of «Figenix» software. Etienne G.J. Danchin participated in creation of «Figenix» software, in the annotation of the gene set, participated in result interpretation, designed the statistical tests applied in the study and in writing later versions of the manuscript. Mathieu Blanc built the software agent to retrieve EST tissue expression data. Daniel Zinn co-participated in the creation of the software agent for ESTs. Jean Roudier participated in the inititation of the study and in writing latest version of the manuscript. Pierre Pontarotti supervised the study and co-wrote the manuscript.
Supplementary Material
Additional File 1
Supplementary trees built with EST contigs. A simplified Tree of life. Table 2: normalized complete table
Click here for file
Acknowledgements
We thank the "Société Française de Rhumatologie" (SFR) that provided funds. We thank Dr. Nathalie Lambert and Dr. Daniel Gautheret for useful discussion and manuscript correction. We thank Nick Boespflug for manuscript correction.
==== Refs
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Koski LB Golding GB The closest BLAST hit is often not the nearest neighbor J Mol Evol 2001 52 540 542 11443357
Eisen JA Phylogenomics: improving functional predictions for uncharacterized genes by evolutionary analysis Genome Res 1998 8 163 167 9521918
Abi-Rached L Gilles A Shiina T Pontarotti P Inoko H Evidence of en bloc duplication in vertebrate genomes Nat Genet 2002 31 100 105 11967531 10.1038/ng855
Madsen O Scally M Douady CJ Kao DJ DeBry RW Adkins R Amrine HM Stanhope MJ de Jong WW Springer MS Parallel adaptive radiations in two major clades of placental mammals Nature 2001 409 610 614 11214318 10.1038/35054544
Murphy WJ Eizirik E Johnson WE Zhang YP Ryder OA O'Brien SJ Molecular phylogenetics and the origins of placental mammals Nature 2001 409 614 618 11214319 10.1038/35054550
Ohno S Gene duplication and the uniqueness of vertebrate genomes circa 1970-1999 Semin Cell Dev Biol 1999 10 517 522 10597635 10.1006/scdb.1999.0332
Sankoff D Gene and genome duplication Curr Opin Genet Dev 2001 11 681 684 11682313 10.1016/S0959-437X(00)00253-7
Ensembl
NCBI NR protein database
dbEST NCBI NCBI dbEST
TIGR
Shmueli O Horn-Saban S Chalifa-Caspi V Shmoish M Ophir R Benjamin-Rodrig H Safran M Domany E Lancet D GeneNote: whole genome expression profiles in normal human tissues C R Biol 2003 326 1067 1072 14744114
Zhang Y Eberhard DA Frantz GD Dowd P Wu TD Zhou Y Watanabe C Luoh SM Polakis P Hillan KJ Wood WI Zhang Z GEPIS--quantitative gene expression profiling in normal and cancer tissues Bioinformatics 2004
Huminiecki L Wolfe KH Divergence of spatial gene expression profiles following species-specific gene duplications in human and mouse Genome Res 2004 14 1870 1879 15466287 10.1101/gr.2705204
Yanai I Graur D Ophir R Incongruent expression profiles between human and mouse orthologous genes suggest widespread neutral evolution of transcription control Omics 2004 8 15 24 15107234 10.1089/153623104773547462
Gouret P Vitiello V Balandraud N Gilles A Pontarotti P Danchin EG FIGENIX: intelligent automation of genomic annotation: expertise integration in a new software platform BMC Bioinformatics 2005 6 198 16083500 10.1186/1471-2105-6-198
Alarcon GS Epidemiology of rheumatoid arthritis Rheum Dis Clin North Am 1995 21 589 604 8619090
Harris EDJ Rheumatoid arthritis. Pathophysiology and implications for therapy N Engl J Med 1990 322 1277 1289 2271017
Ishigami A Ohsawa T Asaga H Akiyama K Kuramoto M Maruyama N Human peptidylarginine deiminase type II: molecular cloning, gene organization, and expression in human skin Arch Biochem Biophys 2002 407 25 31 12392711 10.1016/S0003-9861(02)00516-7
Suzuki A Yamada R Chang X Tokuhiro S Sawada T Suzuki M Nagasaki M Nakayama-Hamada M Kawaida R Ono M Ohtsuki M Furukawa H Yoshino S Yukioka M Tohma S Matsubara T Wakitani S Teshima R Nishioka Y Sekine A Iida A Takahashi A Tsunoda T Nakamura Y Yamamoto K Functional haplotypes of PADI4, encoding citrullinating enzyme peptidylarginine deiminase 4, are associated with rheumatoid arthritis Nat Genet 2003 34 395 402 12833157 10.1038/ng1206
Barton A Bowes J Eyre S Spreckley K Hinks A John S Worthington J A functional haplotype of the PADI4 gene associated with rheumatoid arthritis in a Japanese population is not associated in a United Kingdom population Arthritis Rheum 2004 50 1117 1121 15077293 10.1002/art.20169
Chavanas S Mechin MC Takahara H Kawada A Nachat R Serre G Simon M Comparative analysis of the mouse and human peptidylarginine deiminase gene clusters reveals highly conserved non-coding segments and a new human gene, PADI6 Gene 2004 330 19 27 15087120 10.1016/j.gene.2003.12.038
UNIGENE
Audic S Claverie JM The significance of digital gene expression profiles Genome Res 1997 7 986 995 9331369
Lewis BP Green RE Brenner SE Evidence for the widespread coupling of alternative splicing and nonsense-mediated mRNA decay in humans Proc Natl Acad Sci U S A 2003 100 189 192 12502788 10.1073/pnas.0136770100
Nakayama-Hamada M Suzuki A Kubota K Takazawa T Ohsaka M Kawaida R Ono M Kasuya A Furukawa H Yamada R Yamamoto K Comparison of enzymatic properties between hPADI2 and hPADI4 Biochem Biophys Res Commun 2005 327 192 200 15629448 10.1016/j.bbrc.2004.11.152
Gu X Statistical framework for phylogenomic analysis of gene family expression profiles Genetics 2004 167 531 542 15166175 10.1534/genetics.167.1.531
Gu X Zhang Z Huang W Rapid evolution of expression and regulatory divergences after yeast gene duplication Proc Natl Acad Sci U S A 2005 102 707 712 15647348 10.1073/pnas.0409186102
Gu Z Rifkin SA White KP Li WH Duplicate genes increase gene expression diversity within and between species Nat Genet 2004 36 577 579 15122255 10.1038/ng1355
Duret L Mouchiroud D Determinants of substitution rates in mammalian genes: expression pattern affects selection intensity but not mutation rate Mol Biol Evol 2000 17 68 74 10666707
Zhang L Li WH Mammalian housekeeping genes evolve more slowly than tissue-specific genes Mol Biol Evol 2004 21 236 239 14595094 10.1093/molbev/msh010
Thompson JD Higgins DG Gibson TJ CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic Acids Research 1994 22 4673 4680 7984417
Eddy SR Profile hidden Markov models. Bioinformatics 1998 14 755 763 9918945 10.1093/bioinformatics/14.9.755
Schmidt HA Strimmer K Vingron M Von Haeseler A TREE-PUZZLE: maximum likelihood phylogenetic analysis using quartets and parallel computing. Bioinformatics 2002 18 502 504 11934758 10.1093/bioinformatics/18.3.502
Saitou N Nei M The neighbor-joining method: a new method for reconstructing phylogenetic trees Mol Biol Evol 1987 4 406 425 3447015
Gu X Statistical methods for testing functional divergence after gene duplication. Molecular Biology and Evolution 1999 1664 1674 10605109
Swofford D PAUP* Phylogenetic Analysis Using Parsimony (*and other methods) Version 4 Sinauer Associates. 2002 12504223
Felsenstein J Statistical properties of parsimony, in Inferring Phylogenies Sinaur Associates, Inc, Sunderland, Massachusetts 2004 97 122
Felsenstein J Evolutionary trees from DNA sequences: a maximum likelihood approach J Mol Evol 1981 17 368 376 7288891 10.1007/BF01734359
Templeton AR Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and apes. Evolution 1983 37 221 244
Templeton AR Convergent evolution and non-parametric inferences from restriction fragment and DNA sequence data. B. Statistical Analysis of DNA Sequence Data (Marcel Dekker: New York) 1983 151 179
Kishino HHM Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J Mol Evol 1989 29 179 179 10.1007/BF02100115
Zmasek CM Eddy SR A simple algorithm to infer gene duplication and speciation events on a gene tree Bioinformatics 2001 17 821 828 11590098 10.1093/bioinformatics/17.9.821
Huang X Madan A CAP3: A DNA sequence assembly program Genome Res 1999 9 868 877 10508846 10.1101/gr.9.9.868
|
16271148
|
PMC1310624
|
CC BY
|
2021-01-04 16:32:47
|
no
|
BMC Genomics. 2005 Nov 4; 6:153
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-153
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1471628750710.1186/1471-2407-5-147Case ReportLong-term disease-free survival in advanced melanomas treated with nitrosoureas: mechanisms and new perspectives Durando Xavier [email protected] Emilie [email protected]'Incan Michel [email protected] Anne [email protected] Jean-Claude [email protected] Philippe [email protected] Medical Oncology Unit, Centre Jean Perrin, Clermont-Ferrand, France2 Dermatology Department, Hôtel-Dieu, Clermont-Ferrand, France3 INSERM U484, Clermont-Ferrand, France2005 15 11 2005 5 147 147 17 12 2004 15 11 2005 Copyright © 2005 Durando et al; licensee BioMed Central Ltd.2005Durando et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Median survival of metastatic malignant melanoma is 6.0 to 7.5 months, with a 5-year survival of ~6.0%. Although long-term complete remissions are rare, few reports describe cases after chemotherapy. Fifty-three patients with metastatic melanoma were treated with Cystemustine, a chloroethyl nitrosourea (CENU) (60 or 90 mg/m2).
Case presentation
We describe 5 cases, presenting with complete response with long-term disease-free survival of long-term remission of 14, 12, 9, 7 and 6 years after Cystemustine therapy alone.
Conclusion
Long-term survival has already been described in literature, but in all cases they have been obtained after chemotherapy associated with or followed by surgery. But despite these noteworthy and encouraging but also rare results, it appears essential to increase cystemustine efficiency.
==== Body
Background
Melanoma has become an important public health issue because of its rising incidence in the Caucasian population. Even though, most cases are cured by surgery alone in the early stages of the disease, advanced melanoma has a poor prognosis. Therapeutic strategies of metastatic malignant melanoma are based on multiple treatment including chemotherapy (dacarbazine, platinum analogs, chloronitrosoureas (CENU), vinca alkaloïds or taxanes) and immunotherapy (interferon α, interleukine 2), both as single agents or in association.
Median survival of metastatic malignant melanoma is 6.0 to 7.5 months, with a 5-year survival of approximately 6% [1]. Although long-term complete remissions are rare, some authors have reported cases after chemotherapy treatment [2-6].
Cystemustine {N'-(2-chloroethyl)-N- [2-(methyl sulphonyl)ethyl]-N'-nitrosourea} is a CENU derived from 2-chloro-ethylnitroso-carbamoylcysteamine (CCNC); it is synthesised in INSERM U484, Clermont-Ferrand [7]. This new compound has demonstrated an equivalent and often better chemotherapeutic index for solid tumours than other chloroethylating agents currently in use [8].
We describe 5 cases treated with Cystemustine (60 or 90 mg/m2) for metastatic melanoma showing complete responses with long-term survival (survival time 24 months or longer).
Case Presentation
A series of 53 patients has been treated with Cystemustine administered as a 15 min iv infusion every 2 weeks. 30 patients were included in a phase II national and multicentric study, whereas other 23 patients received a Cystemustine compassionate treatment.
In the phase II study, patients provided written informed consent after they were informed about the objectives of the study. The protocol designs and relative modifications were fully approved by the ethic committee. The patients received Cystemustine 60 or 90 mg/m2. The median number of Cystemustine cycles was 3 (range, 2 to 7 cycles). 28 out of 30 patients were assessed for disease response (2 patients were included without measurable disease according to World Health Organisation (WHO) criteria. The overall response rate was 17.9%, with 3 complete responses and 2 partial responses. Two patients showing complete response had been treated with Cystemustine 90 mg/m2, the third patient received Cystemustine 60 mg/m2.
Patients given "compassionate treatment" received 60 mg/m2 of Cystemustine. The median number of Cystemustine cycles was 3 (range, 1 to 27 cycles). Among these 23 patients, 15 were assessed for disease response (i.e. patients with measurable disease according to WHO criteria). Four complete and two partial responses were observed. All patients were treated with Cystemustine 60 mg/m2.
Among these 53 patients, 5 complete responses with long-term survival occurred. In 4 cases, complete responses were obtained after Cystemustine treatment alone, whereas, for the fifth patient, one of the metastatic lesions was excised at the early beginning of the treatment. The characteristics of these 5 patients are summarised in the "patient characteristics" table 1.
Table 1 Patient characteristics.
Patient 1 2 3 4 5
Patient age/disease discovery (year) 55 33 63 44 43
Primary disease date April 1981 August 1985 January 1984 January 1997 August 1993
Primary disease location Right ankle Left calf Left ankle Left arm Left thorax
Clarck index III V IV III III
First recurrence date March 1988 June 1986 April 1987 January 1997 March 1998
Previous treatment for advanced
disease (before cystemustine) None Dacarbazine Vindesine (6 cycles)
Dacarbazine Vinblastine (6 cycles)
Dacarbazine Interferon α 2a (6 cycles) Interferon α 2a Radiotherapy None Dacarbazine (8 cycles)
Treatment start date March 1988 May 1990 August 1990 February 1997 Novembre 1998
Custemustine dose (mg/m2) 90 60 90 90 60
Metastatic sites Popliteal mass Lymph node Lung Bone Lymph node Lymph node Sub-cutaneous Lung
Treatment concomitant to Cystemustine Surgery (March 1988) None None None None
Recurrence of the disease None None None None None
Date of death November 2000 Alive March 2000 Alive Alive
Disease free survival (years) 12 14 10 7 6
Patient 1 was a 55 year woman presented with a level III Clark right ankle malignant melanoma excised on April 1981. In 1988, she displayed an inoperable popliteal mass associated with a large right inguinal adenopathy. Between March and August 1988, she received 13 cycles of Cystemustine 90 mg/m2. CR was documented by clinical and echographic exams in August 1988. The main toxicities of treatment were nausea, vomiting and thrombopenia. These last ones have led to dose reduction and treatment delay. This patient died of stroke on November 2000 without any recurrence of the disease.
Patient 2 was a 33 year woman presented with a level IV Clark left calf malignant melanoma excised on August 1985. Between 1986 and 1990, several relapse of melanoma were diagnosed. The patient has been successively treated with vindesine/dacarbazine, vindesine/dacarbazine, and dacarbazine/interferon α. Hormonotherapy with tamoxifene was then introduced with close clinical follow-up. In 1990, she presented with right tibiae metastasis and two pulmonary lesions. Treatment with Cystemustine 60 mg/m2 was introduced for 18 cycles. After 6 cycles, complete regression of pulmonary lesions was observed, but bone lesions were always present in spite of a significant improvement. Tumour assessment after 17 cycles revealed CR to treatment. The main toxicity was haematological, with neutropenia and thrombopenia, that necessitated course delay and dose reduction.
Patient 3 was a 63 year woman presenting with Clark level IV malignant melanoma on left ankle, excised in January 1984. Three years later, tumour recurrence was diagnosed on the left calf. An adjuvant treatment with interferon was introduced (but interrupted after the first infusion because of an allergic reaction), followed by 3 × 18 Gy cycles of radiotherapy in June 1987. In August 1988, a node was detected at the lower part of the left popliteal pit, and was excised in December 1988. In September 1990, supra-clavicular and lombo-aortic nodes were detected. The patient received 11 cycles of Cystemustine 90 mg/m2, and a CR was obtained after 8 cycles. Haematological toxicities occurred with severe thrombopenia, leading to course delays and platelet transfusion after the 8th infusion. Except for the haematological toxicity, treatment was well tolerated. However, a systematic exploration of respiratory function in January 2000 showed a severe hypoxia and confirmed the diminution of carbon monoxide capacity found 5 years earlier.
In July 1998, myelodisplasic syndrome was diagnosed. She died in March 2000 during severe aplasia, nevertheless her meanoma was still in complete remission for 9 years and a month.
Patient 4 was a 44 year male presenting Clark level III malignant melanoma located on left arm. Tumour excision was performed on January 1997 and completed with a node dissection. One node of 11 removed was metastatic. Imaging assessment found also a superficial 2 cm left axillary node as well as 2 – 3 infracentimetric metastatic subcutaneous nodes in the chest wall. The patient received 6 cycles of Cystemustine 90 mg/m2. Target size decreased after 2 cycles, and the patient was CR at the end of treatment. No relevant toxicity was reported, except an episode of febrile aplasia after the 6th cycle leading to red blood cells and platelets transfusion.
Patient 5 was a 43 year male presenting with a Clark level III cutaneous thoracic melanoma lesion excised on August 1993. Surgery was completed by a wide local excision margin of 3 cm of the scar. Disease assessment imaging revealed no metastasis. The patient was then followed every 3 months.
Several pulmonary node metastasis were found in March 1998. The patient received 8 cycles of dacarbazine, and because of new disease progression, he received six cycles of Cystemustine 60 mg/m2 every 2 weeks. The patient was considered in complete remission after 2 cycles of Cystemustine. No significant toxicity was reported during the treatment.
Among these five patients, three remain in long-term remission for 14, 7 and 6 years after Cystemustine monotherapy. Of the other 2 patients, one (patient 1) died from a stroke after a 12 years survival without disease recurrence. The second (patient 3) presented with a myelodysplastic syndrome treated with cytarabine. This patient died in March 2000 from severe aplasia, and at this time her melanoma was still in complete remission after 9 years.
Conclusion
Although some reports had already described long-term survival in patients with metastatic melanoma treated with chemotherapy, most of these results were observed for patients treated with chemotherapy in association, or with chemotherapy alone followed by surgery. In a phase III study, 580 patients were treated with dacarbazine (Deticene®) alone or in association with carmustine, lomustine, vincristine or an hydroxyurea. Hill et al. [2] reported 8 patients, with complete response who survived 6 years after the treatment. Ahmann et al. [9] gave results for 15 clinical studies. Among the 503 patients, 6 presented with complete response and long term survival of seven years or longer. Four patients have been treated with a semustine based regimen, one with a combination of vinblastine, bleomycin and cisplatin, and the last one with dianthydrogalacticol.
Petit et al. [5] described 5 out of 160 patients who went into long-term remission 7 years after fotemustine chemotherapy followed by surgery. In Samuel et al. [6], 40 patients with symptomatic metastatic melanoma were treated with procarbazine, vincristine and lomustine (POC). Among them, two remained in complete remission at 6 and 6.5 years. Berd et al. [3] treated 147 patients with a combination of carmustine, dacarbazine and cisplatin. Among the 17 complete response, 7 patients presented with long-term survival (19 to 82 months).
Coates et al. also described the case of 8 patients who remain in long-term remission 4 to 15 years after chemotherapy for visceral metastatic melanoma among some 1100 patients with visceral melanoma who have received chemotherapy, almost always with single agent dacarbazine or a nitrosourea treatments. For these patients, the mechanism remains uncertain, but the author suggested that chemotherapeutic agents can cause mutations that allow expression of antigenicity in tumour cells [4].
Despite these noteworthy and encouraging but also rare results, it appears essential to increase cystemustine efficiency. The main antitumor target of CENU is DNA with generation of O6-chloroethylguanine and cross-linking with cDNA strands [10]. The MGMT (O6-methylguanine-DNA methyltransferase) protein removes O6-alkylguanine by accepting the alkyl group on the cystein residue of its active site. MGMT level varies considerably in normal and tumor cells, and cells that exhibit a low MGMT level are more sensitive to CENU [11]. Methionine restriction is known to inhibit growth of human and animal tumor (in vitro, and in vivo) [12]. It decreases MGMT mRNA and activity in tumoral cells [13]. Furthermore, the potentiating effect of methionine depletion on Cystemustine treatment has been shown in B16 melanoma-bearing mice [14]. On the basis of these previous experimental results, we have initiated a phase-I clinical trial of dietary methionine restriction, in association with Cystemustine treatment for adults with recurrent metastatic melanomas.
In conclusion, we described case of five out 53 patients with metastatic melanoma, treated with Cystemustine, 60 or 90 mg/m2, and presented with complete response and long-term disease-free survival. Despite these encouraging results, it appears crucial to increase cystemustine efficiency.
Competing interests
The author(s) declare that they have no competing interests.
Author's contributions
XD was involved in patient treatment and participated in writing the manuscript.
ET was responsible for data management, and participated in writing the manuscript.
MD'I, AS and PC were involved in patient treatment and participated in drafting the manuscript.
JCM has provided Cystemustine and participated in drafting the manuscript.
All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Bajetta E Del Vecchio M Bernard-Marty C Vitali M Buzzoni R Rixe O Nova P Aglione S Taillibert S Khayat D Metastatic melanoma: chemotherapy Semin Oncol 2002 29 427 445 12407508 10.1053/sonc.2002.35238
Hill GJ 2ndKrementz ET Hill HZ Dimethyl triazeno imidazole carboxamide and combination therapy for melanoma. IV. Late results after complete response to chemotherapy (Central Oncology Group protocols 7131, and 7131A) Cancer 7130 53 1299 1305 6362841
Berd D Mastrangelo MJ Combination chemotherapy of metastatic melanoma J Clin Oncol 1995 13 796 797 7884439
Coates AS Segelov E Long term response to chemotherapy in patients with visceral metastatic melanoma Ann Oncol 1994 5 249 251 8186173
Petit T Borel C Rixe O Avril MF Monnier A Giroux B Weil M Khayat D Complete remission seven years after treatment for metastatic malignant melanoma Cancer 1996 77 900 902 8608481 10.1002/(SICI)1097-0142(19960301)77:5<900::AID-CNCR14>3.0.CO;2-7
Samuel LM Harvey VJ Mitchell PL Thompson PI Mak D Melville P Evans BD Phase II trial of procarbazine, vincristine and lomustine (POC) chemotherapy in metastatic cutaneous malignant melanoma Eur J Cancer 1994 30A 2054 2056 7857702 10.1016/0959-8049(94)00273-8
Godeneche D Madelmont JC Moreau MF Duprat J Plagne R Meyniel G Metabolism of 2-chloroethyl nitrosocarbamoylcystamine by rat liver subcellular fractions Drug Metab Dispos 1986 14 112 117 2868853
Filippeschi S Colombo T Bassani D De Francesco L Arioli P D'Incalci M Bartosek I Guaitani A Antitumor activity of the novel nitrosourea S10036 in rodent tumors Anticancer Res 1988 8 1351 1354 3064716
Ahmann DL Creagan ET Hahn RG Edmonson JH Bisel HF Schaid DJ Complete responses and long-term survivals after systemic chemotherapy for patients with advanced malignant melanoma Cancer 1989 63 224 227 2642731
Demidem A Morvan D Papon J De Latour M Madelmont JC Cystemustine induces redifferentiation of primary tumors and confers protection against secondary tumor growth in a melanoma murine model Cancer Res 2001 61 2294 2300 11280801
Paz MF Yaya-Tur R Rojas-Marcos I Reynes G Pollan M Aguirre-Cruz L Garcia-Lopez JL Piquer J Safont MJ Balana C Sanchez-Cespedes M Garcia-Villanueva M Arribas L Esteller M CpG island hypermethylation of the DNA repair enzyme methyltransferase predicts response to temozolomide in primary gliomas Clin Cancer Res 2004 10 4933 4938 15297393 10.1158/1078-0432.CCR-04-0392
Cellarier E Terret C Labarre P Ouabdesselam R Cure H Marchenay C Maurizis JC Madelmont JC Cholle P Armand JP Pharmacokinetic study of cystemustine, administered on a weekly schedule in cancer patients Ann Oncol 2002 13 760 769 12075746 10.1093/annonc/mdf098
Kokkinakis DM von Wronski MA Vuong TH Brent TP Schold SC Jr Regulation of O6-methylguanine-DNA methyltransferase by methionine in human tumour cells Br J Cancer 1997 75 779 788 9062396
Morvan D Papon J Madelmont JC Demidem A Methionine deprivation potentiates the effect of cystemustine treatment on B16 melanoma tumors in syngenic recipients AACR, San Francisco 2002 3822
|
16287507
|
PMC1310625
|
CC BY
|
2021-01-04 16:39:11
|
no
|
BMC Cancer. 2005 Nov 15; 5:147
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-147
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1691630956010.1186/1471-2164-6-169Research ArticleThe EH1 motif in metazoan transcription factors Copley Richard R [email protected] Wellcome Trust Centre for Human Genetics, Roosevelt Drive, Oxford OX3 7BN, UK2005 27 11 2005 6 169 169 26 8 2005 27 11 2005 Copyright © 2005 Copley; licensee BioMed Central Ltd.2005Copley; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The Engrailed Homology 1 (EH1) motif is a small region, believed to have evolved convergently in homeobox and forkhead containing proteins, that interacts with the Drosophila protein groucho (C. elegans unc-37, Human Transducin-like Enhancers of Split). The small size of the motif makes its reliable identification by computational means difficult. I have systematically searched the predicted proteomes of Drosophila, C. elegans and human for further instances of the motif.
Results
Using motif identification methods and database searching techniques, I delimit which homeobox and forkhead domain containing proteins also have likely EH1 motifs. I show that despite low database search scores, there is a significant association of the motif with transcription factor function. I further show that likely EH1 motifs are found in combination with T-Box, Zinc Finger and Doublesex domains as well as discussing other plausible candidate associations. I identify strong candidate EH1 motifs in basal metazoan phyla.
Conclusion
Candidate EH1 motifs exist in combination with a variety of transcription factor domains, suggesting that these proteins have repressor functions. The distribution of the EH1 motif is suggestive of convergent evolution, although in many cases, the motif has been conserved throughout bilaterian orthologs. Groucho mediated repression was established prior to the evolution of bilateria.
==== Body
Background
The Engrailed Homology 1 (EH1) motif is a short (<10 amino acids) region, initially found in engrailed (en) and other homeobox containing proteins, that mediates transcriptional repression via interaction with the WD40 repeat containing groucho (Gro) [1,2]. Shimeld [3] proposed that the EH1 motif of Smith and Jaynes was shared with various forkhead (FH/HNF-3) containing transcription factors. The short size of the motif, however, suggests that it may occur by chance in many different protein families. Shimeld did not demonstrate statistically significant sequence similarity between the motifs from the homeobox- and forkhead-containing families. However, the human orthologs of groucho (the transducin-like enhancer of split proteins) have been shown to interact with FOXA2 via a region of sequence containing an EH1 motif, clearly demonstrating the biological relevance of the sequence similarity [4].
In this article I search systematically for instances of the EH1 motif in homeobox and forkhead containing genes and go on to demonstrate that the EH1 motif is also found in proteins containing T-box, Doublesex Motif (DM) and Zn finger domains. I show that within metazoan genomes, the observed association of the motif with transcription factor function is statistically significant. The location of the motif in members of the same transcription factor family is often non-homologous, occurring both N- and C-terminal to the DNA binding domain, suggesting that the presence of the motif is, in part, due to convergent evolution, as proposed by Shimeld; the conservation within orthologs points to many of these convergences predating the last common ancestor of the bilateria.
Results and Discussion
Significant association of EH1 motif with transcription factor function
I searched for sequence motifs in homeobox containing transcription factors taken from the proteins of human, Drosophila melanogaster and Caenorhabditis elegans, by first masking known Pfam domains [5], and then using the expectation maximization algorithm implemented in the meme program [6]. The first non-subfamily specific motif identified corresponded to previously known examples and new instances of, the EH1 motif (see Figure 1a), in 100 sites, with an E-value of < 10-126. I then applied the same approach to Forkhead containing transcription factors, identifying 25 sites with a combined E-value of < 10-31 (Figure 2a). These motifs also appeared to conform to the consensus of the EH1 motif, as previously reported by Shimeld [3].
Figure 1 Alignments of putative EH1 motifs in a) Homeobox and b) Paired box domain containing proteins, subdivided by domain partners and orientation, with representative non-bilaterian sequences included. Alignments were derived from meme searches, as described in text. Conserved aromatic residues (FHYW) are coloured white on a red background ('a' in the consensus). Conserved aliphatic residues (ILV), black on a yellow background. ('I' in the consensus) Conserved big residues (EFHIKLMQRWY) blue on a light yellow background ('b' in the consensus). Conservation is calculated over the full alignment of sequences in figures 1 and 2. The figure was produced using the Chroma program [41]. Gene names are standard HUGO Gene Nommenclature Committee, flybase or wormbase symbols where available, otherwise accessions for their respective databases. When available Uniprot protein accessions are also given [42], along with the starting residue of the motif.
Figure 2 Alignments of putative EH1 motifs in a) Forkhead b) T-box c) ETS d) Doublesex and e) Zinc finger containing proteins. Alignment 'a' was derived from a meme search, as described in text. Sub-alignments b-e were derived from HMMER searches with the EH1hox HMM. Other details as for Figure 1.
To further investigate the significance of this similarity, I constructed hidden Markov models (HMM) of the motif (EH1hox & EH1fh) which I then searched against the complete set of predicted proteins from human, D. melanogaster &C. elegans. The highest scoring non homeobox containing domain match of EH1hox was a Forkhead protein (human FOXL1), and the second highest scoring non-Forkhead containing match of EH1fh was to a homeobox containing protein (D. melanogaster inv). In both cases, nearly all the high scoring hits were to proteins containing domains with transcription factor function (see Figure 3). Among the best scoring matches of the EH1hox searches were several T-box (TBOX), Doublesex Motif (DM), Zinc finger (ZnF_C2H2) and ETS containing proteins (domain names as per SMART, Figure 2b–e) [7,8]. Excluding hits to homeobox containing proteins, but otherwise including all scores, the overall significance of the association of transcription factor function with higher scores to the EH1hox HMM is P < 10-47, using a logistic regression model which tests association between score and transcription factor annotation (see methods and supplementary file 1 for raw data). The association remains significant when scores derived from Forkhead and PAX domain containing proteins are also excluded (P < 10-34). This indicates that, although the scores associated with any individual EH1-like motif may not be statistically significant, overall, we would not see so many EH1-like sequences co-occurring with DNA binding domains if their co-occurrence were governed simply by chance – there is, therefore, likely to be a functional reason for these partnerships. In the following sections, I review the higher scoring associations detected here in the light of known gene functions.
Figure 3 a) Distribution of HMMER bit scores for the database search of EH1hox HMM against the combined proteomes of human, D. melanogaster and C. elegans. Counts from scores from transcription factors (see methods) have been coloured red – i.e. the proportion of a bar coloured red is equal to the proportion of transcription factors. Scores from proteins containing a homeobox domain (interpro accession IPR001356), from which the EH1hox HMM was derived, have been excluded, b) as for 'a', but rescaled to show region of biological relevance. High scoring hits are greatly enriched in specific transcription factor families. For scores ≥ 5.0 bits, there are 68 transcription factors and 142 non-transcription factors; for scores <5.0 bits, 3075 transcription factors and 51513 non-transcription factors giving a chi-square test p-value statistic of P < 0.0001 – the statistical significance is discussed more fully in the text.
EH1 motifs in homeobox and forkhead containing proteins
The presence of EH1 motifs within various homeobox, and to a lesser extent, forkhead containing proteins has been widely reported, although not systematically studied [3]. I found EH1-like motifs co-occurring with 3 major groupings of homeobox sub-types: the extended-hox class, typified by Drosophila engrailed (en); the paired class, including Drosophila goosecoid (gsc), and the NK class, including Drosophila tinman (tin) [1,9,10] (see [11] for a description of these broad classes). Related to the paired class homeobox domains, a number of genes containing PAIRED domains only (i.e. the PAX domain of SMART [7]) were also found to contain EH1-like motifs (see Figure 1b). With only a few exceptions, outlined below, the EH1-like motif occurs N-terminal to the homeobox domain and C-terminal to the PAIRED domain when present. A number of these proteins have been shown to interact with groucho or its orthologs e.g. C. elegans cog-1 [12], vertebrate Nkx proteins [13], Drosophila engrailed (en) and goosecoid (gsc) [2,14], and in high throughput assays Drosophila invected (inv) and and ladybird late (Ibl) [15].
A handful of EH1-like motifs are found C-terminal to homeobox domains. Of these, the best characterized is C. elegans unc-4, which has been shown to interact with the groucho ortholog unc-37 [16]; the Drosophila ortholog unc-4 also interacts with groucho in high throughput experiments [15]. The C-terminal EH1-like motif is conserved in the closely related Drosophila paralog OdsH. The gene prediction for the human ortholog of unc-4 (ensembl gene identifier ENSG00000164853) appears to be artefactually truncated, but the mouse ortholog (Uncx4.1 ENSMUSG00000029546) and corrected human gene models, contain EH1-like motifs both N & C-terminal to the homeobox domain. Taken together with the fact that in the majority of related homeobox containing proteins the EH1-like motifs are N-terminal, this suggests that the N-terminal motif has been lost in Drosophila and C. elegans unc-4 orthologs.
EH1-like motifs also occur N- and C-terminal to Forkhead domains. The N-terminal class consists of the sloppy-paired genes (slp1 and slp2) of Drosophila and orthologous or closely related sequences: human FOXG1, and Drosophila CG9571; the C. elegans ortholog fkh-2 contains an EH1-like motif although a cysteine residue causes a low score. The C-terminal class consists of an apparent clade including the human FOXA, FOXB, FOXC and FOXD genes (Figure 2a), although if the EH1 motif was present in the common ancestor of this clade, multiple losses must have later occurred (see [17] for a Forkhead domain phylogeny). The situation is complicated somewhat by an EH1-like motif at the N-terminus of C. elegans unc-130 i.e. in the FOXD like family. The EH1 motif in slp1 has been shown to interact with groucho [18], and FOXA type genes have been shown to interact with human groucho orthologs [4].
EH1 motifs in novel domain contexts
Assuming a conservative per-domain cutoff score of 10.0 bits for true matches to the EH1hox model (see Figure 3), yields hits to proteins containing T-box domains (highest score 13.1 bits); Doublesex (DM) domains (highest score 11.6 bits) and C2H2 Zinc fingers (highest score 11.2 bits). Also of note was a further match at 9.4 bits, to an ETS domain containing protein. Prompted by these similarities I further investigated the presence of EH1-like motifs in these families, looking for high scoring matches to the EH1hox HMM that were conserved in closely related genes.
T-box containing proteins
I identified likely EH1 motifs co-occurring with T-Box domains in two distinct contexts (Figure 2b). The motif occurs C-terminal to the T-box in the Drosophila dorsocross proteins Doc1, Doc2 and Doc3. It is found N-terminal to the T-box in 11 proteins including mls-1 and mab-9 from C. elegans; H15, mid/nmr2 and bi/omd from Drosophila; in humans there are strong matches to TBX18, TBX20 and TBX22 and more marginal matches to TBX3 and TBX2. Although, to the best of my knowledge, none of these proteins has been shown to interact with groucho or its orthologs, several are known to act as transcriptional repressors: for instance, in murine heart development, Tbx20 represses Tbx2 which in turn represses Nmyc [19,20]; the Dorsocross genes from Drosophila repress wingless and ladybird [21], and Doc itself is repressed by mid/nmr2 [22]. The human proteins TBX1 and TBX10, and Drosophila org-1 which are closely related to those above, do not appear to contain EH1 motifs. The human T (brachyury) protein contains a motif broadly similar to the EH1 consensus: LQYRVDHLLSA in a comparable N-terminal location to those found in other T-box containing proteins. Although this motif scores poorly against EH1hox (-0.1 bits), the homologous regions from other T orthologs (for instance, the non-bilaterian sequences discussed below) provide a more persuasive case for the presence of a functioning EH1 motif in these proteins.
Zinc finger containing proteins
The highest scoring match of EH1hox to a C2H2 zinc finger containing protein, was ces-1 from C. elegans (bit score 11.2); this protein interacts with the groucho ortholog unc-37 [[23], #54] and can act as a repressor [24]. The putative EH1 motif is at the N-terminal end of ces-1. In contrast, the Drosophila proteins bowl and odd have EH1-like motifs at their C-terminal ends (with bit scores of 10.9 & 8.4 respectively). In neither case is there direct evidence from high throughput studies of an interaction with groucho, but both can function as repressors [25]. The human protein ZNF312 (bit score 8.6) is the ortholog of zebrafish Fezl, which contains an EH1 motif essential for repressor activity [26] – this motif is conserved in the human paralog ENSG00000128610 and likely Drosophila ortholog CG31670 (bit scores of 8.4 & 5.1) (Figure 2e).
Doublesex motif containing proteins
The Doublesex Motif (DM) was first found in proteins controlling sexual differentiation in Drosophila. Two DM containing proteins were confidently predicted to contain EH1-like motifs – human DMRT2 (bit score 11.6), and Drosophila dmrt11e (bit score 11.2) – these are likely orthologs; a C. elegans protein, C27C12.6 contained a weaker match (bit score 6.6) (Figure 2d). The molecular function of these proteins is unknown.
Other potential associations with transcription factor domains
Although scoring less highly than some non-transcription factor hits, another intriguing association is with the ETS domain. The three uncharatcerized C. elegans paralogs F19F10.5, F19F10.1 & C50A2.4 contain C-terminal matches to the EH1 motifs (bit scores 9.4, 2.3 & 7.4), and two other ETS proteins, C. elegans lin-1, and Drosophila Eip74EF, both have relatively high scoring matches (bit scores 6.5 & 6.6) (Figure 2c). A high scoring protein that is not annotated as a transcription factor (as it contains no interpro domains) is Drosophila Hairless (H) with a score of 8.3 bits. Experimental work has previously confirmed the presence of an EH1-like motif (SSYSIHSLLGG) within H that is responsible for its interaction with groucho [27]. The Drosophila protein Dorsal has been reported to interact with groucho via an EH1-like motif [28] – this region (NGPTLSNLLSF) is markedly different to those reported here, having a low score against EH1hox (-10.7 bits) and so may better be regarded as a, so far, unique type of groucho interaction motif.
Evolutionary considerations
Convergent evolution
The EH1 motif is found N- and C-terminal to homeobox, forkhead, T-box and Zn finger protein domains. Clearly, as the locations of the EH1 motif are non-homologous, the N- and C-terminal associations must have occurred independently. The short size of the motif makes it tempting to speculate that the motif itself may have arisen independently (i.e. in repeated cases it may have evolved within sequence that was already part of the gene, rather than via a recombination event). The strongest evidence for this is that, in general, the majority of domain combinations occur in a fixed N to C orientation, suggesting that recombination events combining domains are relatively rare [29,30]. The fact that we would here have many such events suggests that the alternative hypothesis of independent invention is more appropriate.
Pre-bilaterian origins of association with different transcription factors
Groucho is orthologous to the C. elegans unc-37 gene, and the four human paralogs TLE1-4 (Transducin Like Enhancer of split). An ortholog is also found in the cnidarian Hydra mangipapillata (e.g. the EST with gi 47137860, data not shown), and certain cnidarian homeobox containing genes also contain an EH1-like motif, suggesting groucho/EH1 mediated repression pre-dates the split between diplobasts and triplobasts; indeed, a sponge Bar/Bsh like homeobox containing protein (i.e. protein gi: 33641772) [31] also contains an EH1-like motif, as does paxb from the non-bilaterian placozoan Trichoplax adhaerens [32] and a Tlx-like protein from a ctenophore (gi: 38602653), suggesting the repression system was in place in the earliest animals (see [33] for a discussion of early metazoan evolution). I find high scoring EH1-like motifs in Forkhead domain containing proteins from sponges, cnidarians and ctenophores, in both the C-terminal (FOXA-D clade) (region II in [34]) and N-terminal (FOXG, sloppy paired clade) varieties (reported as 'HPFSI' in [35]). The presumed ortholog of 'T' from the Trichoplax adhaerens [36] contains an EH1-like motif (8.6 bits). These results suggest that groucho mediated repression using a variety of transcription factors was widespread in the last common ancestor of the metazoa.
Conclusion
Candidate EH1 motifs exist in combination with a variety of transcription factor domains, suggesting that these proteins have roles as repressors of transcriptional activity. The distribution of the EH1 motif is suggestive of a number of instances of convergent evolution, although in many cases the motif has been conserved throughout bilaterian orthologs. Together with the existence of a cnidarian Groucho ortholog, this leads to the conclusion that EH1/Groucho mediated repression was established prior to the evolution of bilateria.
Methods
Proteomes were derived from ensembl 32 (human NCBI 35, C. elegans wormbase 140, Drosophila BDGP 4) [37]. In cases of multiple splice variants, the one with the most exons was included (or the longest in the case of ties). Transcription factor activity was taken as the presence of the gene ontology accession GO:0003700 associated with an interpro domain predicted for the protein [38]. These data were also taken from ensembl. Although C2H2 subtype Zn fingers are not annotated by Interpro as transcription factors they are DNA binding and frequently have this role, so have been included in the transcription factor set. Bit scores reported in the text are for comparisons of the EH1hox HMM against the target sequence using the HMMER software package [39].
The association of transcription factor function (coded as a dichotomous variable, t, taking the values 1 [transcription factor] or 0 [non-transcription factor]) with the bit score, x, of the EH1hox HMM, was tested using a logistic regression model implemented in the glm() function of the R package [40]). I fitted the model
Prob(t = 1) = exp(a + bx)/(1 + exp(a + bx))
The coefficients a, b were estimated from the data by maximum-likelihood. The hypothesis of no association is equivalent to testing if b = 0.
Where inferences of orthology are made, they are based on clear-cut separation of BLAST scores or alignment-based phylogenies.
Authors' contributions
RRC performed the analysis and wrote the paper.
Supplementary Material
Additional File 1
Each non-homeobox containing protein in the database is classified as either being a transcription factor or not (see methods), and its score against the EH1hox HMM is given.
Click here for file
Acknowledgements
I am grateful to an anonymous referee for comments on TBX15 & brachyury. I thank the Wellcome Trust for financial support, Dr. Richard Mott for statistical advice, Drs. Martin Taylor and William Valdar for helpful suggestions.
==== Refs
Smith ST Jaynes JB A conserved region of engrailed, shared among all en-, gsc-, Nk1-, Nk2- and msh-class homeoproteins, mediates active transcriptional repression in vivo Development 1996 122 3141 3150 8898227
Tolkunova EN Fujioka M Kobayashi M Deka D Jaynes JB Two distinct types of repression domain in engrailed: one interacts with the groucho corepressor and is preferentially active on integrated target genes Mol Cell Biol 1998 18 2804 2814 9566899
Shimeld SM A transcriptional modification motif encoded by homeobox and fork head genes FEBS Lett 1997 410 124 125 9237613 10.1016/S0014-5793(97)00632-7
Wang JC Waltner-Law M Yamada K Osawa H Stifani S Granner DK Transducin-like enhancer of split proteins, the human homologs of Drosophila groucho, interact with hepatic nuclear factor 3beta J Biol Chem 2000 275 18418 18423 10748198 10.1074/jbc.M910211199
Bateman A Coin L Durbin R Finn RD Hollich V Griffiths-Jones S Khanna A Marshall M Moxon S Sonnhammer EL The Pfam protein families database Nucleic Acids Res 2004 D138 141 14681378 10.1093/nar/gkh121
Bailey TL Elkan C Fitting a mixture model by expectation maximization to discover motifs in biopolymers Proc Int Conf lntell Syst Mol Biol 1994 2 28 36
SMART – Simple Modular Architecture Research Tool
Letunic I Copley RR Schmidt S Ciccarelli FD Doerks T Schultz J Ponting CP Bork P SMART 4.0: towards genomic data integration Nucleic Acids Res 2004 D142 144 14681379 10.1093/nar/gkh088
Galliot B de Vargas C Miller D Evolution of homeobox genes: Q50 Paired-like genes founded the Paired class Dev Genes Evol 1999 209 186 197 10079362 10.1007/s004270050243
Jagla K Bellard M Frasch M A cluster of Drosophila homeobox genes involved in mesoderm differentiation programs Bioessays 2001 23 125 133 11169585 10.1002/1521-1878(200102)23:2<125::AID-BIES1019>3.0.CO;2-C
Banerjee-Basu S Baxevanis AD Molecular evolution of the homeodomain family of transcription factors Nucleic Acids Res 2001 29 3258 3269 11470884 10.1093/nar/29.15.3258
Chang S Johnston RJ JrHobert O A transcriptional regulatory cascade that controls left/right asymmetry in chemosensory neurons of C. elegans Genes Dev 2003 17 2123 2137 12952888 10.1101/gad.1117903
Muhr J Andersson E Persson M Jessell TM Ericson J Groucho-mediated transcriptional repression establishes progenitor cell pattern and neuronal fate in the ventral neural tube Cell 2001 104 861 873 11290324 10.1016/S0092-8674(01)00283-5
Jimenez G Verrijzer CP Ish-Horowicz D A conserved motif in goosecoid mediates groucho-dependent repression in Drosophila embryos Mol Cell Biol 1999 19 2080 2087 10022895
Giot L Bader JS Brouwer C Chaudhuri A Kuang B Li Y Hao YL Ooi CE Godwin B Vitols E A protein interaction map of Drosophila melanogaster Science 2003 302 1727 1736 14605208 10.1126/science.1090289
Winnier AR Meir JY Ross JM Tavernarakis N Driscoll M Ishihara T Katsura I Miller DM 3rd UNC-4/UNC-37-dependent repression of motor neuron-specific genes controls synaptic choice in Caenorhabditis elegans Genes Dev 1999 13 2774 2786 10557206 10.1101/gad.13.21.2774
Mazet F Yu JK Liberles DA Holland LZ Shimeld SM Phylogenetic relationships of the Fox (Forkhead) gene family in the Bilateria Gene 2003 316 79 89 14563554 10.1016/S0378-1119(03)00741-8
Andrioli LP Oberstein AL Corado MS Yu D Small S Groucho-dependent repression by sloppy-paired 1 differentially positions anterior pair-rule stripes in the Drosophila embryo Dev Biol 2004 276 541 551 15581884 10.1016/j.ydbio.2004.09.025
Stennard FA Costa MW Lai D Biben C Furtado MB Solloway MJ McCulley DJ Leimena C Preis JI Dunwoodie SL Murine T-box transcription factor Tbx20 acts as a repressor during heart development, and is essential for adult heart integrity, function and adaptation Development 2005 132 2451 2462 15843414 10.1242/dev.01799
Cai CL Zhou W Yang L Bu L Qyang Y Zhang X Li X Rosenfeld MG Chen J Evans S T-box genes coordinate regional rates of proliferation and regional specification during cardiogenesis Development 2005 132 2475 2487 15843407 10.1242/dev.01832
Reim I Lee HH Frasch M The T-box-encoding Dorsocross genes function in amnioserosa development and the patterning of the dorsolateral germ band downstream of Dpp Development 2003 130 3187 3204 12783790 10.1242/dev.00548
Reim I Mohler JP Frasch M Tbx20-related genes, mid and H15, are required for tinman expression, proper patterning, and normal differentiation of cardioblasts in Drosophila Mech Dev 2005 15922573
Li S Armstrong CM Bertin N Ge H Milstein S Boxem M Vidalain PO Han JD Chesneau A Hao T A map of the interactome network of the metazoan C. elegans Science 2004 303 540 543 14704431 10.1126/science.1091403
Thellmann M Hatzold J Conradt B The Snail-like CES-1 protein of C. elegans can block the expression of the BH3-only cell-death activator gene egl-1 by antagonizing the function of bHLH proteins Development 2003 130 4057 4071 12874127 10.1242/dev.00597
Campbell G Regulation of gene expression in the distal region of the Drosophila leg by the Hox11 homolog, C15 Dev Biol 2005 278 607 618 15680373 10.1016/j.ydbio.2004.12.009
Levkowitz G Zeller J Sirotkin HI French D Schilbach S Hashimoto H Hibi M Talbot WS Rosenthal A Zinc finger protein too few controls the development of monoaminergic neurons Nat Neurosci 2003 6 28 33 12469125 10.1038/nn979
Barolo S Stone T Bang AG Posakony JW Default repression and Notch signaling: Hairless acts as an adaptor to recruit the corepressors Groucho and dCtBP to Suppressor of Hairless Genes Dev 2002 16 1964 1976 12154126 10.1101/gad.987402
Flores-Saaib RD Jia S Courey AJ Activation and repression by the C-terminal domain of Dorsal Development 2001 128 1869 1879 11311166
Apic G Gough J Teichmann SA Domain combinations in archaeal, eubacterial and eukaryotic proteomes J Mol Biol 2001 310 311 325 11428892 10.1006/jmbi.2001.4776
Gough J Convergent evolution of domain architectures (is rare) Bioinformatics 2005 21 1464 1471 15585523 10.1093/bioinformatics/bti204
Hill A Tetrault J Hill M Isolation and expression analysis of a poriferan Antp-class Bar-/Bsh-like homeobox gene Dev Genes Evol 2004 214 515 523 15322878
Hadrys T Desalle R Sagasser S Fischer N Schierwater B The Trichoplax PaxB Gene: A Putative Proto-PaxA/B/C Gene Predating the Origin of Nerve and Sensory Cells Mol Biol Evol 2005 22 1569 1578 15858210 10.1093/molbev/msi150
Medina M Collins AG Silberman JD Sogin ML Evaluating hypotheses of basal animal phylogeny using complete sequences of large and small subunit rRNA Proc Natl Acad Sci USA 2001 98 9707 9712 11504944 10.1073/pnas.171316998
Adell T Muller WE Isolation and characterization of five Fox (Forkhead) genes from the sponge Suberites domuncula Gene 2004 334 35 46 15256253 10.1016/j.gene.2004.02.036
Yamada A Martindale MQ Expression of the ctenophore Brain Factor 1 forkhead gene ortholog (ctenoBF-1) mRNA is restricted to the presumptive mouth and feeding apparatus: implications for axial organization in the Metazoa Dev Genes Evol 2002 212 338 348 12185487 10.1007/s00427-002-0248-x
Martinelli C Spring J Distinct expression patterns of the two T-box homologues Brachyury and Tbx2/3 in the placozoan Trichoplax adhaerens Dev Genes Evol 2003 213 492 499 13680223 10.1007/s00427-003-0353-5
Hubbard T Andrews D Caccamo M Cameron G Chen Y Clamp M Clarke L Coates G Cox T Cunningham F Ensembl 2005 Nucleic Acids Res 2005 D447 453 15608235
Mulder NJ Apweiler R Attwood TK Bairoch A Bateman A Binns D Bradley P Bork P Bucher P Cerutti L InterPro, progress and status in 2005 Nucleic Acids Res 2005 D201 205 15608177
HMMER: sequence analysis using profile hidden Markov models
The R project for statistcal computing
Goodstadt L Ponting CP CHROMA: consensus-based colouring of multiple alignments for publication Bioinformatics 2001 17 845 846 11590103 10.1093/bioinformatics/17.9.845
Bairoch A Apweiler R Wu CH Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M The Universal Protein Resource (UniProt) Nucleic Acids Res 2005 D154 159 15608167
|
16309560
|
PMC1310626
|
CC BY
|
2021-01-04 16:32:46
|
no
|
BMC Genomics. 2005 Nov 27; 6:169
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-169
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2791630575110.1186/1471-2105-6-279DatabaseComputational identification of strain-, species- and genus-specific proteins Mazumder Raja [email protected] Darren A [email protected] Sudhir [email protected] Rathi [email protected] Cathy H [email protected] Department of Biochemistry and Molecular Biology, Georgetown University Medical Center, 3900 Reservoir Rd., NW, Washington, DC 20057-1414, USA2 DCWASA-DWT, 5000 Overlook Ave., SW, Washington, DC 20032, USA2005 23 11 2005 6 279 279 28 4 2005 23 11 2005 Copyright © 2005 Mazumder et al; licensee BioMed Central Ltd.2005Mazumder et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
The identification of unique proteins at different taxonomic levels has both scientific and practical value. Strain-, species- and genus-specific proteins can provide insight into the criteria that define an organism and its relationship with close relatives. Such proteins can also serve as taxon-specific diagnostic targets.
Description
A pipeline using a combination of computational and manual analyses of BLAST results was developed to identify strain-, species-, and genus-specific proteins and to catalog the closest sequenced relative for each protein in a proteome. Proteins encoded by a given strain are preliminarily considered to be unique if BLAST, using a comprehensive protein database, fails to retrieve (with an e-value better than 0.001) any protein not encoded by the query strain, species or genus (for strain-, species- and genus-specific proteins respectively), or if BLAST, using the best hit as the query (reverse BLAST), does not retrieve the initial query protein. Results are manually inspected for homology if the initial query is retrieved in the reverse BLAST but is not the best hit. Sequences unlikely to retrieve homologs using the default BLOSUM62 matrix (usually short sequences) are re-tested using the PAM30 matrix, thereby increasing the number of retrieved homologs and increasing the stringency of the search for unique proteins. The above protocol was used to examine several food- and water-borne pathogens. We find that the reverse BLAST step filters out about 22% of proteins with homologs that would otherwise be considered unique at the genus and species levels. Analysis of the annotations of unique proteins reveals that many are remnants of prophage proteins, or may be involved in virulence. The data generated from this study can be accessed and further evaluated from the CUPID (Core and Unique Protein Identification) system web site (updated semi-annually) at .
Conclusion
CUPID provides a set of proteins specific to a genus, species or a strain, and identifies the most closely related organism.
==== Body
Background
Over 200 pathogenic and non-pathogenic bacteria have been completely sequenced [1], including multiple strains from several species. The availability of sequence data from related genomes has facilitated comparative genomic analysis, which not only allows the study of major evolutionary processes, but also the determination of proteins conserved across — or unique to — different species [2-4]. Information on the presence or absence of genes is a powerful tool to gain knowledge about the metabolism, pathogenicity, physiology and behavior of different organisms [2-4]. It can also provide the basis for the detection of pathogens in a given sample and to distinguish between pathogenic and non-pathogenic relatives — critical to combating disease and to the emerging biodefense field.
Two recent studies (ORFanage [5] and Procom [6]) have focused on species- and clade-specific genes, both noting a substantial number of unique genes encoded in specific organisms. The ORFanage database includes about 32,000 unique ORFs (ORFans) from 84 fully sequenced microbial genomes. An ORFan is a protein that failed to hit any other protein encoded within those 84 genomes with an E-value better than 10e-3 (or 10e-5 for alignments of <80 residues) using BLAST [7]. The Procom database compared proteins from thirty completely sequenced eukaryotic genomes. The proteins were pair-wise compared using WU-BLASTP [8] with a threshold E-value of 1.
We developed a general protocol to identify proteins unique to different taxa, and applied it to a set of food- and water-borne pathogens. Specifically, we: a) identify proteins that are unique to a particular strain, species, or genus; b) extract the set of proteins common to two or more strains or species; and c) determine the organism most closely related to a particular genus, species or strain. The information generated from this study is available from the CUPID (Core and Unique Protein Identification) system via a flexible and easy-to-use web interface.
Construction and content
Definition of terms
Throughout this paper the following terms apply:
Organism
The leaf-most taxonomic node indicated for a given fully-sequenced genome. The node can be at the species, strain, or sub-strain taxonomic level.
Self
The source strain, species or genus of a BLAST query when considering strain-, species-, and genus-specific proteins, respectively. For example, a protein encoded by any Bacillus is considered self at the genus level when the query organism is Bacillus anthracis strain Ames.
The following terms refer to the protein sets generated by this study:
Core
Proteins that are present in all selected organisms. "Selected organisms" will include all the completely-sequenced genomes in the data set that are taxonomically identical to the query organism at the strain, species or genus level.
Unique
Proteins that have no related sequences in non-self organisms. Related sequences are mined from all sources (not just from completely-sequenced genomes). Related sequences from the query organism (paralogs) are ignored as self.
Core unique
Proteins that have related sequences in all selected organisms, but not in non-selected organisms.
The following terms refer to the evidence provided for the indicated uniqueness status on the results page. Unless otherwise indicated, the evidence is based on BLAST results:
Hits
The query retrieves a protein from a non-self organism with an e-value better than 0.001.
Identical protein
The query sequence is identical to a sequence from a non-self organism.
Reciprocal hit
The best non-self hit has the original query as its best non-self hit. Reciprocal best hits represent an approximation of orthology [3].
No external hits
The query fails to hit a protein from a non-self organism using 0.001 as a cutoff.
No reverse hit
The best hit does not retrieve the original query when testing for reciprocal hits in a reverse BLAST. Additional evidence is provided for such cases. Forwardgives the E-value for the hit between the query and subject. Maxrevgives the worst E-value reported when the subject is used as query.
Unclear
The best hit retrieves the original query in a reverse BLAST, but it is not the reciprocal best hit. The E-values for the forwardand reverseBLASTS are given.
Curator judgment
"Unclear" cases that were resolved after manual checking of the data.
PAM30
The results shown are for BLAST using the PAM30 matrix. The default matrix is BLOSUM62.
Data source and algorithm
Complete proteomes were retrieved from the NREF database at PIR [9], a comprehensive (~2.1 million sequences) non-redundant protein database. Identical sequences from a given species are merged in this database. Sequence searches were conducted using the BLAST program. An initial set of organism-specific proteins was generated computationally. Proteins that could not be confirmed as unique by computer analysis were manually checked by several methods [10], including multiple sequence alignment [11], hidden Markov models [12], and PSI-BLAST [7]. Proteins specific to a set of organisms were identified by determining which of the conserved core proteins [13,14] are unique (using the same procedure as used to determine organism-specific proteins). A flowchart of the process is shown in Figure 1.
Figure 1 Flow chart of the method used to detect strain-, species-, or genus-specific proteins. Proteins were designated as not unique if (a) The protein is part of a merged entry (containing identical proteins from different organisms) and one of the source organisms is considered non-self; (b) The query protein hits a non-self subject with E < 0.001; (c) The non-self best hit, when itself used as a query (in a reverse BLAST), retrieves the initial query as an organism-specific best hit (that is, they are reciprocal best hits, and thus potential orthologs); and (d) The non-self best hit fails to retrieve the initial query as its potential ortholog — but does indeed retrieve the initial query — and manual inspection reveals homology. Sequences that could not retrieve themselves with an expect value of 1e-14 or better in the forward BLAST using the BLOSUM62 matrix were retested using the PAM30 matrix.
Certain sequences failed to retrieve themselves using the BLOSUM62 matrix. It is known that the PAM matrices may increase the information content in the alignment and hence optimize the alignment score [15]. The PAM matrices were evaluated to identify the optimal word size, gap existence cost and gap extension cost values to be used. BLAST results using the different PAM matrices and different parameters were manually checked to identify which sequences should be analyzed using a different matrix (data not shown). We found that sequences that could not retrieve themselves with an expect value of 1e-14 or better in the forward BLAST using the BLOSUM62 matrix (usually short sequences <30 aa) should be queried again using the PAM30 matrix (word size = 2, gap existence cost = 9 and gap extension cost = 1) to improve the ability to retrieve homologs. The inclusion of the reverse BLAST step and optimized search parameters both yield an analysis method that is skewed toward tagging proteins as not unique whenever possible.
The information generated for each organism at each taxonomic level (genus, species, or strain) is stored in a flat-file. Data for each protein is given on a single machine-parsable line, and is tracked using a unique protein identifier to allow integration with other PIR resources. The line contains the protein ID, an indication of whether homologs were found (+ for yes, ! for no), evidence for homology or lack thereof, identity and source organism for any core protein(s) found, and an indication of whether the query protein is part of the core. The results can be queried and browsed through the CUPID web interface. Additional analysis of the data is facilitated by links to PIR protein resources [9] and other bioinformatics servers.
Data Update
Unique proteins for common food- and water-borne pathogens will be re-determined every six months using pre-computed BLAST (generated quarterly). De novo re-computation is necessary due to the constant increase in (and possible change to) protein sequences. We will add selected NIAID category pathogens with each semi-annual update cycle. In addition, we provide a mechanism to request additional proteome analyses. Users may select from a list of organisms, accessible from the web interface, that is populated by proteomes with stable sequence annotation in the UniProt Knowledgebase (UniProtKB) [16]. Analysis time will depend on several factors, including proteome size and evolutionary distance from other proteomes. The user will be notified upon completion of the analysis. Results from user-selected proteomes will be made available from the CUPID web site. Such analyses will be stamped with the date of the last computation, but will not be part of the regular update cycle.
Usage and utility
Retrieving strain-, species-, or genus-specific proteins
Unique proteins are retrieved by clicking an organism listed on the CUPID homepage (Figure 2), followed by clicking on the taxonomic level of interest. By default, the initial retrieval returns only core unique proteins (Figure 3). To retrieve just the unique proteins, one should select the taxonomic level of interest and "Unique" from the pull-down menus. For example, selecting "Genus" level "Unique" proteins for E. coli O157:H7 will display all proteins in E. coli O157:H7 that may or may not be present in other Escherichia, but are not present in any other genus. The "Unique" column will either have a plus sign (+), be left blank, or have a question mark (?). A plus sign means that the protein is unique by "no external hits", by no reverse BLAST hit, or by curator judgment. Blank means that the protein is not unique, and a question mark means that the relationship is of unclear status (that is, it cannot be unambiguously determined by computational means alone, and human judgment is required). The core status column indicates if the given protein has homologs in all proteomes considered self, while the core hits column lists the best hit protein from each of the self proteomes. The user can save the retrieved list in tab-delimited format or save the protein sequences in fasta format.
Figure 2 CUPID homepage. The number of core unique proteins encoded by each genome at the strain, species, and genus levels are indicated, as are the organisms considered for the core at the species and genus levels (boxes).
Figure 3 Example output from CUPID. Helicobacter pylori 26695 was selected to retrieve proteins encoded only in bacteria of the Helicobacter genus. The results display protein accessions, name, length, family classification, uniqueness status, evidence for uniqueness, core status, other core proteins, and additional links. Accession links (first column) connect the protein to (i) the protein entry report, (ii) iProClass [24] entry report with rich links to over 90 molecular databases, and (iii) the UniProtKB report. Additional links (last column) are provided to (i) the EMBL DNA sequence (derived from the UniProtKB cross-reference), (ii) a pre-filled form for tBLASTn for protein search against translated NCBI DNA databases (with Entrez post-processing set to filter out self hits) and (iii) signal sequence and transmembrane domain prediction using Phobius [25].
Retrieving the set of proteins specific to two or more organisms
Proteins unique to two or more organisms can be retrieved from the species- or genus-specific proteins results page. This is done, in effect, by redefining the proteomes to be considered for the core. For example, once the results for E. coli O157:H7 are displayed, one can display the proteins unique to E. coli O157:H7 and E. coli O157:H7 EDL933 by checking the box for "Core Hits must *only* be from" and the box for Tax ID 155864. This will retrieve proteins that are conserved in both E. coli O157:H7 and E. coli O157:H7 EDL933, but are not found anywhere else. Alternatively, selecting "Show Genus level All proteins" with the same checkboxes selected as above will display all the proteins conserved between E. coli O157:H7 and E. coli O157:H7 EDL933 strains – not just the unique ones – but will exclude those found in other Escherichia proteomes.
Retrieving all the proteins from a particular organism
All the proteins from an organism can be retrieved from any results page by selecting "Show XX level All proteins" (where XX can be Genus, Species, or Strain).
Determining the closest relative
The genus, species, or strain that is most closely related to the selected organism based on the best BLAST hits of its entire proteome can be identified by clicking on the TaxID for the top non-self hits on the results page. For example, from the result page of genus-specific proteins of E. coli O157:H7, the top non-Escherichia hits are to Shigella flexneri, and from the result page of strain-specific proteins for E. coli O157:H7 the top non-Escherichia coli O157:H7 hits are to E. coli O157:H7 EDL933 (not shown).
Description of selected results
We identified the unique proteins in thirty organisms (predominantly food- or water-borne bacterial pathogens and their close relatives). The results for ten common water and wastewater strains of Escherichia, Salmonella and Helicobacter are described below. E. coli is the primary indicator organism for biologically contaminated food and water. Several strains of E. coli, such as the laboratory strain K12 [17], are completely harmless, while E. coli CFT073 is an extra-intestinal uropathogenic bacterium and the O157:H7 strains are enterohemorrhagic pathogens. S. enterica Typhi is a human pathogen that causes enteric typhoid fever. The Ty2 strain of S. enterica was isolated before the advent of antibiotics and has no plasmids, whereas the CT18 strain harbors a multiple-drug-resistance plasmid and a cryptic plasmid. S. typhimurium LT2 is a strain of S. enterica (synonym: S. enterica subsp. enterica serovar Typhimurium strain LT2). H. pylori is associated with peptic ulcers and several types of gastric cancer. H. hepaticus causes chronic hepatitis and is also a recognized carcinogen. A summary of computer-assisted manual analyses of these organisms is shown in Table 1.
Table 1 Strain-, species-, and genus-specific core unique proteins from selected water- and wastewater-borne pathogens and pathogen-related organisms.
Organism Strain-specific proteins Species-specific proteins Genus-specific proteins
Escherichia coli CFT073 3 14 14
Escherichia coli K12 35 15 15
Escherichia coli O157:H7 117 12 12
Escherichia coli O157:H7 EDL933 73 12 12
Salmonella enterica subsp. enterica serovar Typhi Ty2 1 7 101
Salmonella enterica subsp. enterica serovar Typhi str. CT18 38 7 101
Salmonella typhimurium LT2 37 62 106
Helicobacter hepaticus ATCC 51449 236 250 5
Helicobacter pylori 26695 53 110 15
Helicobacter pylori J99 17 107 11
The number of unique proteins in each of four E. coli strains (K12, CFT073, O157:H7 and O157:H7 EDL933) did not show any apparent correlation with the pathogenic nature of the organism. For example, E. coli CFT073 (pathogenic) has only three unique proteins compared to 35 in K12 (non-pathogenic) and 117 and 73 in O157:H7 (pathogenic) and O157:H7 EDL933 (pathogenic) strains, respectively. High numbers of unique sequences in some organisms could be due to the unavailability of a closely related non-self genome or, as is likely in this example, unique plasmids that may be present in the query genome.
The number of genus/species-specific proteins for each E. coli is not identical as one might first expect; rather, they range from 12 to 15. This phenomenon can also be seen in other organisms where there are multiple genomes from the same species or genus. Three explanations are possible: i) each strain has different numbers of paralogs; ii) a frameshift produces two annotated open reading frames in one strain where another has one; iii) the comprehensive dataset used contains largely duplicate but non-identical entries.
The number of unique proteins for several E. coli strains drops as one compares strain level to genus level (except for the CFT073 strain), a trend that is also apparent for the Helicobacter strains. However, the opposite is true for the Salmonella. The progression is not necessarily linear or consistent. For example, the number of unique proteins from the CT18 strain of Salmonella enterica first drops from 38 at the strain level to 7 at the species level before again rising to 101 at the genus level.
A majority of the unique proteins detected in this study lacked meaningful functional annotation (i.e., were annotated simply as "hypothetical protein") (Table 2). The available annotation for the remainder indicates a preponderance of proteins with some relationship to pathogenesis or virulence (22%), or derived from phages (25%). In addition, a combination of annotation and subcellular localization prediction using PSORT-B [18] indicates that 26% of the proteins are external to the cell (cell wall attached or secreted).
Table 2 Summary of annotations for genus-specific unique proteins from selected water- and wastewater-borne pathogens and pathogen-related organisms.
Organism Unique Virulencea External Phage Unknown Misc.
Escherichia coli CFT073 182 19 28 0 134 1
Escherichia coli K12 262 0 8 0 254 0
Escherichia coli O157:H7 259 9 19 8 213 10
Escherichia coli O157:H7 EDL933 187 18 12 21 114 22
Salmonella enterica subsp. enterica serovar Typhi Ty2 244 19 24 0 198 3
Salmonella enterica subsp. enterica serovar Typhi str. CT18 307 4 3 19 268 13
Salmonella typhimurium LT2 233 0 2 43 181 7
Helicobacter hepaticus ATCC 51449 253 13 6 18 171 45
Helicobacter pylori 26695 199 9 16 5 159 10
Helicobacter pylori J99 663 11 1 2 637 12
TOTAL 2789 102 119 116 2329 123
aIncludes proteins derived from pathogenicity islands.
Discussion
It is evident from this study that a major reason that few unique proteins are found in some cases is the presence of sequence data for closely-related organisms and, by extension, the peculiarities of taxonomic designations. Therefore, only a general trend in the number of strain-, species-, or genus-specific proteins can be established. Strains with closely related sequenced strains tend to have relatively few unique proteins at that level while the converse is true for those without close relatives (compare Helicobacter pylori strains with Helicobacter hepaticus). The trend also holds true at the genus level (compare Escherichia or Helicobacter genus-specific proteins with Salmonella). We note that a proteome from a closely-related genus is represented in the protein database for both Escherichia (Shigella) and Helicobacter (Campylobacter). Shigella is so similar to E. coli that there are recommendations to consider them different species within the same genus [19], while Helicobacter pylori was once Campylobacter pylori.
The number of apparently unique genes encoded in specific organisms can depend on the definition of "unique" and the parameters and underlying database used to identify these proteins. For example, CUPID identifies 110 proteins unique to H. pylori 26695 at the species level, whereas the ORFanage database lists 260 proteins. The ORFanage database considers the protein HP0052 to be unique. BLAST with HP0052 (UniProt ID: O24893_HELPY) retrieves several non-Helicobacter proteins (Moraxella nonliquefaciens - E = 9e-22, Ehrlichia canis - E = 3e-18, Mycoplasma mycoides - E = 2e-15). None of these non-self hits have a complete genome sequence deposited in the database, and therefore were not considered during the construction of the ORFanage. In addition, other lists of unique proteins consider a protein to be unique if it did not produce a BLAST hit with the E value cut-off set to 10-3 [20,21], and do not further assess cases that have lower similarity. However, the contribution of low-similarity homologs is significant. For example, 950 proteins from the organisms presented in Table 1 that were considered not unique at the genus level were considered so because of a reciprocal best hit. Overall, we found that about 22% of proteins were removed from the list of unique proteins based on the reciprocal best hit criterion. In general, many sequences are considered to be singleton ORFs by other studies but not considered unique in this study because: a) a limited protein set was used for BLAST comparison as opposed to a comprehensive database; b) the definition of self is different (strain differences may not be considered); c) only unidirectional BLAST hits (without reverse BLAST) are considered to find homologs; and d) the cutoff used to define homolog is less conservative. Using a thorough computational method and a comprehensive database leads to a more conservative estimate of the number of unique proteins.
Despite the conservative approach used here, one must be mindful of certain pitfalls in deriving lists of unique proteins. First, a protein might be labeled as unique only because homologs from other organisms were missed upon submission of the sequence, or because of some other conceptual translation problem. In all cases, the short list of potential unique proteins should be further screened computationally at the DNA level using tBLASTn. Running tBLASTn using a protein of interest will make sure that the gene is indeed unique, at least with respect to the current pool of submitted sequences (to aid in this task, we have included a link to NCBI's tBLASTn web page). Second, many of the proteins identified here are remnants of prophage proteins [22]. The leading role of bacteriophages in shaping the E. coli O157:H7 genome is evident by the presence of 24 prophages and prophage-like elements, and several genes that have been laterally transferred [23]. At the strain level, labeling of such proteins as unique may be more a reflection of a gap in whole-genome sequence information than of true specificity. Accordingly, discrimination between individual strains (isolates) may require laboratory comparison methods such as pulsed-field gel electrophoresis or whole-cell fatty acid analysis. In contrast, identifying species- or genus-specific proteins can be done with confidence when multiple representatives have been sequenced. In such cases, conservation within multiple strains of a species (for example) gives confidence in the "reality" of the uniqueness because that status has been conserved over time (core unique proteins).
Precise identification of pathogens is important so that adequate action can be taken to either eliminate or reduce the threat of infection. One use of the CUPID system is to help identify diagnostic targets specific to a particular clade of these pathogens. The unique proteins form a short list of diagnostic targets to be validated in the laboratory. Proteins predicted to be external to the bacterial cell – possibly involved in host interactions and virulence – may be used to develop protein-based detection systems. In addition, it should be possible to use the DNA encoding these proteins as the basis for diagnostics. However, it is important to note that protein-identified DNA probes must be verified as to their uniqueness at the DNA level.
Conclusion
The salient features of CUPID are: a) provides sets of proteins unique to a strain, species, and genus level; b) includes a check for additional homologs based on reciprocal hits; c) uses different parameters for short sequences; d) provides the identity of the nearest non-self neighbor; and e) allows retrieval of unique, core, and core unique proteins at different taxonomic levels.
Availability and requirements
CUPID is freely accessible from the PIR website at .
List of abbreviations
CUPID – Core and Unique Protein Identification system
PIR – Protein Information Resource
Authors' contributions
RM conceived, designed and coordinated the study, developed a general outline for the algorithm and drafted the manuscript. DN developed the specific algorithm and was responsible for software design and implementation, and participated in the writing of the manuscript. CW and SM participated in the design and evaluation of the study and manuscript writing. RT and DN developed the web interface. All authors read and approved the final manuscript.
Acknowledgements
Development of CUPID was supported by grant AWD4220707 from the District of Columbia Water and Sewer Authority. The bioinformatics infrastructure on which CUPID is based was supported in part by grant U01-HG02712 from the National Institutes of Health. We would like to thank Drs. Winona Barker and Hongzhan Huang for reviewing the manuscript and providing useful comments.
==== Refs
Bernal A Ear U Kyrpides N Genomes OnLine Database (GOLD): a monitor of genome projects world-wide Nucleic Acids Res 2001 29 126 127 11125068 10.1093/nar/29.1.126
Eisen JA Fraser CM Phylogenomics: intersection of evolution and genomics Science 2003 300 1706 1707 12805538 10.1126/science.1086292
Tatusov RL Koonin EV Lipman DJ A genomic perspective on protein families Science 1997 278 631 637 9381173 10.1126/science.278.5338.631
Siew N Fischer D Analysis of singleton ORFans in fully sequenced microbial genomes Proteins 2003 53 241 251 14517975 10.1002/prot.10423
Siew N Azaria Y Fischer D The ORFanage: an ORFan database Nucleic Acids Res 2004 32 Database issue D281 3 14681413 10.1093/nar/gkh116
Li JB Zhang M Dutcher SK Stormo GD Procom: a web-based tool to compare multiple eukaryotic proteomes Bioinformatics 2005 21 1693 1694 15564299 10.1093/bioinformatics/bti161
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Miller W Lipman DJ Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694 10.1093/nar/25.17.3389
Gish W WU-BLASTP http://blast.wustl.edu 1996
Wu CH Yeh LS Huang H Arminski L Castro-Alvear J Chen Y Hu Z Kourtesis P Ledley RS Suzek BE Vinayaka CR Zhang J Barker WC The Protein Information Resource Nucleic Acids Res 2003 31 345 347 12520019 10.1093/nar/gkg040
Mazumder R Iyer LM Vasudevan S Aravind L Detection of novel members, structure-function analysis and evolutionary classification of the 2H phosphoesterase superfamily Nucleic Acids Res 2002 30 5229 5243 12466548 10.1093/nar/gkf645
Notredame C Higgins DG Heringa J T-Coffee: A novel method for fast and accurate multiple sequence alignment J Mol Biol 2000 302 205 217 10964570 10.1006/jmbi.2000.4042
Eddy SR Profile hidden Markov models Bioinformatics 1998 14 755 763 9918945 10.1093/bioinformatics/14.9.755
Zafar N Mazumder R Seto D CoreGenes: a computational tool for identifying and cataloging "core" genes in a set of small genomes BMC Bioinformatics 2002 3 12 11972896 10.1186/1471-2105-3-12
Mazumder R Kolaskar A Seto D GeneOrder: comparing the order of genes in small genomes Bioinformatics 2001 17 162 166 11238072 10.1093/bioinformatics/17.2.162
Altschul SF A protein alignment scoring system sensitive at all evolutionary distances J Mol Evol 1993 36 290 300 8483166 10.1007/BF00160485
Bairoch A Apweiler R Wu CH Barker WC Boeckmann B Ferro S Gasteiger E Huang H Lopez R Magrane M Martin MJ Natale DA O'Donovan C Redaschi N Yeh LS The Universal Protein Resource (UniProt) Nucleic Acids Res 2005 33 Database Issue D154 9 15608167
Blattner FR Plunkett G Bloch CA Perna NT Burland V Riley M Collado-Vides J Glasner JD Rode CK Mayhew GF Gregor J Davis NW Kirkpatrick HA Goeden MA Rose DJ Mau B Shao Y The complete genome sequence of Escherichia coli K-12 Science 1997 277 1453 1474 9278503 10.1126/science.277.5331.1453
Gardy JL Spencer C Wang K Ester M Tusnady GE Simon I Hua S deFays K Lambert C Nakai K Brinkman FS PSORT-B: Improving protein subcellular localization prediction for Gram-negative bacteria Nucleic Acids Res 2003 31 3613 3617 12824378 10.1093/nar/gkg602
Pupo GM Lan R Reeves PR Multiple independent origins of Shigella clones of Escherichia coli and convergent evolution of many of their characteristics Proc Natl Acad Sci U S A 2000 97 10567 10572 10954745 10.1073/pnas.180094797
Janssen PJ Audit B Ouzounis CA Strain-specific genes of Helicobacter pylori: distribution, function and dynamics Nucleic Acids Res 2001 29 4395 4404 11691927 10.1093/nar/29.21.4395
Charlebois RL Clarke GD Beiko RG St Jean A Characterization of species-specific genes using a flexible, web-based querying system FEMS Microbiol Lett 2003 225 213 220 12951244 10.1016/S0378-1097(03)00512-3
Krylov VN Role of horizontal gene transfer by bacteriophages in the origin of pathogenic bacteria Genetika 2003 39 595 620 12838609
Hayashi T Makino K Ohnishi M Kurokawa K Ishii K Yokoyama K Han CG Ohtsubo E Nakayama K Murata T Tanaka M Tobe T Iida T Takami H Honda T Sasakawa C Ogasawara N Yasunaga T Kuhara S Shiba T Hattori M Shinagawa H Complete genome sequence of enterohemorrhagic Escherichia coli O157:H7 and genomic comparison with a laboratory strain K-12 DNA Res 2001 8 11 22 11258796 10.1093/dnares/8.1.11
Wu CH Huang H Nikolskaya A Hu Z Barker WC The iProClass integrated database for protein functional analysis Comput Biol Chem 2004 28 87 96 15022647 10.1016/j.compbiolchem.2003.10.003
Kall L Krogh A Sonnhammer EL A combined transmembrane topology and signal peptide prediction method J Mol Biol 2004 338 1027 1036 15111065 10.1016/j.jmb.2004.03.016
|
16305751
|
PMC1310627
|
CC BY
|
2021-01-04 16:27:46
|
no
|
BMC Bioinformatics. 2005 Nov 23; 6:279
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-279
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1421625963510.1186/1471-2407-5-142Research ArticleClinical and pathologic factors associated with survival in young adult patients with fibrolamellar hepatocarcinoma Moreno-Luna Laura E [email protected] Oscar [email protected]ía-Leiva Jorge [email protected]ínez Braulio [email protected] Aldo [email protected] Misael [email protected]ón-Rodríguez Eucario [email protected] Department of Gastroenterology, Instituto Nacional de Ciencias Médicas y Nutrición "Salvador Zubirán" (INCMNSZ), Mexico City, Mexico2 Department of Medical Oncology, Instituto Nacional de Cancerología (INCan), Mexico City, Mexico3 Universidad Nacional Autonoma de Mexico (UNAM), Mexico City, Mexico4 Department of Pathology, Instituto Nacional de Ciencias Médicas y Nutrición "Salvador Zubirán" (INCMNSZ), Mexico City, Mexico5 Department Hemato-Oncology, Instituto Nacional de Ciencias Médicas y Nutrición "Salvador Zubirán" (INCMNSZ), Mexico City, Mexico2005 31 10 2005 5 142 142 25 4 2005 31 10 2005 Copyright © 2005 Moreno-Luna et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Fibrolamellar Carcinoma (FLC), a subtype of hepatocellular carcinoma (HCC), is a rare primary hepatic malignancy. Several aspects of the clinic features and epidemiology of FLC remain unclear because most of the literature on FLC consists of case reports and small cases series with limited information on factors that affect survival.
Methods
We did a retrospective analysis of the clinical and histological characteristics of FLC. We also determined the rate of cellular proliferation in biopsies of these tumors. We assessed whether these variables were associated with survival.
Results
We found 15 patients with FLC out of 174 patients with HCC (8.6%). Between patients with these neoplasms, we found statistically significant survival, age at onset, level of alpha fetoprotein, and an earlier stage of the disease. The 1, 3 and 5 year survival in patients with FLC was of 66, 40 and 26% respectively. The factors associated with a higher survival in patients with FLC were age more than 23 years, feasibility of surgical resection, free surgical borders, absence of thrombosis or invasion to hepatic vessels and the absence of alterations in liver enzymes. The size of the tumor, gender, cellular proliferation and atypia did not affect the prognosis.
Conclusion
We concluded that FLC patients diagnosed before 23 years of age have worse prognosis than those diagnosed after age 23. Other factors associated with worse prognosis in this study are: lack of surgical treatment, presence of positive surgical margins, vascular invasion, and altered hepatic enzymes.
==== Body
Background
Hepatocellular carcinoma (HCC) is the most frequent primary neoplasm of the liver. Its incidence is increasing in western countries mainly due to the growing Hepatitis C Virus epidemics. Though potentially curable with surgery, very few patients are candidates for resection because of tumor extension and poor liver function at diagnosis. The prognosis is poor, with a 1-year survival rate of 22.6% [1].
Fibrolamellar carcinoma (FLC) was described initially in 1958 by Edmonson, as a subtype of HCC [2]. It is characterized by eosinophilic polygonal cells and wide bands of fibrotic tissue settled in parallel laminas surrounded by a clustering of tumor cells [3-6]. It has not been linked with hepatic viruses, alcohol, estrogen use, or other risk factors traditionally associated with HCC. It occurs more frequently in children and adults less than 35 years old. It also exhibits a slower clinical course than the more common HCC. Controversy exists whether FLC has a better prognosis than HCC. Small retrospective series have shown a better survival after resection compared to HCC [4,7-10]. Nevertheless, a recent study of 10 patients with FLC did not report a difference in survival compared to 36 patients with classic HCC [11].
The differences in the prognosis of these patients are probably a consequence of a non cirrhotic liver and the expanding nature tumor growth permits a high rate of surgical resection. In contrast to classic HCC, which has well defined prognostic factors such as Alfa fetoprotein levels, performance status, liver function, and the severity of cirrhosis [12-14], no such predictive standards have been elucidated for FLC. The aim of this study was to describe the clinical, radiographic and histopathologic characteristics of FLC and analyze the factors associated with survival.
Methods
Medical records of patients with the diagnosis of HCC were reviewed in the Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán from 1990 to 2003. Patients with FLC were then identified from biopsy results. Patients with incomplete medical records or lacking biopsy results were excluded from the study.
Biopsies of patients with FLC were reviewed to confirm the diagnosis; some samples where dyed again to determine histopathologic characteristics such as atypia and mitosis index. Additional samples were analyzed by immunohistochemistry with the avidin-biotin-peroxidase complex and counterstained with hematoxylin; they were incubated overnight at 4°C either with rabbit anti-human proliferation cell nuclear antigen as a marker of cellular synthesis phase. A pathologist of our institution, blinded to the results of survival, counted the number of cell nuclei positive for nuclear antigen found per microscopy field at × 40 magnification in 10 different fields [15].
The variables of survival, alpha fetoprotein, mitotic index and cellular proliferation index were expressed as mean ± standard error. The SPSS program version 10 was used for the statistic analysis. Statistic significance was determined as P < 0.05 with two-sided test. The Student's t test and the chi square or Fisher's exact test was used to compare the continuous variables with normal distribution and nominal variables of patients with classic HCC and FLC respectively. Kaplan-Meier curves and log-rank was used for the survival analysis. The multivariate analysis was done using Cox regression modelling.
Results
We reviewed 174 medical records of patients with diagnosis of HCC and found 15 patients with FLC,(8.6%). Seventy percent of the FLC patients were females compared to 37% of the classic HCC patients (p = 0.045). The average age was lower in patients with FLC compared to patients with HCC (23 ± 2.6 years with a range of 17–45 years vs 58 ± 1.5 years with a range of 22–87 years respectively p = 0.001). No patient with FLC had a previous history of blood transfusions or a positive serology for hepatitis. In contrast, 75% of patients with HCC had positive hepatitis B and/or C virus serology (p = 0.0001). Thirty percent of the patients with FLC reported use of oral contraceptives, with a range of use of 6–60 months, versus 15% of the patients with classic HCC (p = 0.1). Most common signs and symptoms in patients with FLC were: right upper quadrant pain 83%, hepatomegaly 75%, and weight loss 67%. In laboratory screening we found mean serum albumin of 3.05 ± 0.2 vs 2.9 ± 0.1 g/dL in patients with classic HCC (p = 0.8). The alpha fetoprotein levels were significantly lower in patients with FLC compared to patients with classic HCC (81.5 ± vs 355.7 ± 65 ng/L, p = 0.003). The patients with FLC and classic HCC presented stage I 27% vs 6% (p = 0.026), stage II 7% vs 16% (p = 0.46), stage III 40 vs 67% (p = 0.042) and stage IV 27% vs 10% (p = 0.1) respectively. The mean survival time of patients with FLC was 30 ± 6.4 months vs 10.6 ± 3 of the classic HCC (p = 0.007) (Figure 1).
In FLC, the most frequent radiographic studies done for the diagnosis were CT scan in 92% of the patients, then MRI in 67% and USG in 58%. The most frequent characteristics found on CT scan were: solitary tumor (22% of the patients), lymphadenopathy (14%), linear hypodense images corresponding to central scar (13%), peripheral reinforcement (12%). The MRI characteristics were central scar (40%), defined contour (18%), more than one tumor (12%), and multilobular contour (12%).
Histopathologic characteristics of FLC biopsies included absence of cirrhosis, presence of fibrotic septum and granular cytoplasm. Furthermore, 92% of the patients demonstrated polygonal cells. Also, 75% of the patients had mild atypia with the remainder showing moderate atypia. In addition, 42% of the biopsies demonstrated pseudoinclusions; 42% had prominent nucleolus, while 33% showed hepatic steatosis (Figures 2A and 2B). The mitotic index average was 2± 1. In all tumors we found a high index of cellular proliferation, mean of 90 ± 5. (Figure 2C).
Surgical resection was performed in 80% of the FLC patients. The most frequent procedures were exploratory laparotomy with biopsy (40% of the patients), lobectomy (20%) and segmentectomy (20%). 60% had free surgical margins and 43% percent received chemotherapy. The overall survival of patients with FLC at 6, 12, 24, 36 and 60 months was 66, 66, 53, 40 and 26% (Figure 3. The factors associated with a better survival were age 23 years or more (median of the age) (8 ± 2 vs 65 ± 19 months; p = 0.0132) (Figure 4), resectability (absence of multiple tumors in more than one lobe, absence of involvement of a major branch of portal or hepatic veins and adequate hepatic function) (60 ± 10 vs 5 ± 2 months; p = 0.001) (Figure 5), negative surgical borders for tumor (65 ± 4 vs 20 ± 5 months; p = 0.06) (Figure 6), absence of thrombosis or invasion to the hepatic vein (60 ± 10 vs 5 ± 2 months; p = 0.006) (Figure 7) and normal liver enzymes (ALT and AST) (65 ± 21 vs 26 ± 11 months; p = 0.04) (Figure 8).
Factors unrelated to prognosis were gender (p = 0.646), presence of palpable mass (p = 0.84), hepatomegaly (p = 0.68), weight loss (p = 0.64), ascites (p = 0.57), fever (p = 0.18), jaundice (p = 0.34), hemoglobin value (p = 0.52), normal alkaline phosphatase (p = 0.68), hypoalbuminemia (0.9), increase in alpha fetoprotein (p = 0.67), level of atypia (p = 0.31), mitotic index (p = 0.9), index of cellular proliferation (p = 0.8) and TNM stage (p = 0.69). Multivariate analysis showed only age greater than 23 years (p = 0.027) and absence of multiple tumors in more than one lobe (p = 0.043) were associated with better survival.
Discussion
Fibrolamellar Hepatocarcinoma is a rare primary hepatic malignancy constituting approximately 1% of all cases of primary liver cancer. A recent retrospective study, the Surveillance, Epidemiology and End Results (SEER) program diagnosed only 68 cases of FLC out 7,896 cases of HCC (0.86%); nevertheless, FLC made up 13.4% of all HCC tumors in people below the age of 40 years [16]. In our study, we found a much higher frequency of 8.6% (15 of 174 patients of all ages). Possibly the number of patients with classic HCC was underestimated in our study as we excluded all patients without biopsies. Almost all patients lacking biopsy were those with advanced cirrhosis, a characteristic of patients with classic HCC rather than FLC. Our study showed a clear absence of association of FLC with Hepatitis virus and the absence of elevation in alpha fetoprotein compared to classic HCC [17].
Another interesting issue was the higher frequency of female gender in the group of FLC compared to classic HCC (60 vs 37%). This observation was also reported in the SEER study where the authors found a higher proportion of females (51.5 vs 26.3%) [16]. Some cases of FLC have been reported during pregnancy [18-20]. In our study, 56% of the female patients with FLC had oral contraceptive use. This was not significantly different from classic HCC patients, but we cannot exclude a hormonal influence in the physiopathology.
We found a significantly lower age at diagnosis in patients with FLC, similar to results reported in the SEER study; the age for FLC at diagnosis was 39 ± 20 vs 65 ± 13. A greater proportion of patients with FLC were classified as having localized disease (stage I) compared to patients with HCC. Similar results were described by the SEER study, where localized disease at diagnosis was seen in 41.2% of FLC patients vs 30.9% of HCC patients. One of the main radiological features found in our study was the presence of central scar. A retrospective study analyzed different radiological features in this kind of tumors and found central scar 71% on CT scans and 82% on MRI in patients with FLC [21].
A limitation of the SEER study was the absence of analysis of factors that modified survival. We found important differences in survival related to age; younger patients (less than 23 years) had a worse prognosis than those who were older. Another associated factor was the feasibility of surgical resection (multiple tumors in more than one lobe) and the presence of tumor in, and thrombosis of, liver vessels independent of the tumor size. We found that histological characteristics such as atypia and cellular proliferation index did not influence prognosis.
The marked differences in epidemiology and clinical course indicate that FLC is distinct from HCC. We found that patients with FLC had higher survival rates than patients with HCC. These results are consistent with the published literature. Some authors report a higher survival rate in patients with FLC after surgical resection, compared to patients with HCC in absence of cirrhosis [9]. In the SEER study they found that the 5 year survival rate was 31.8% for FLC compared with 6.8% for HCC, adjusting for differences in age, gender, race, stage of disease, receipt of resection or transplant, and time of diagnosis. FLC was independently associated with a 46% reduction in risk of mortality [16]. Conversely, another study found no difference in survival following surgical resection between patients with FLC and those with HCC. A multicenter trial of postresection chemotherapy (Pediatric Oncology Study Group Study 8945/Children's Cancer Study Group 8881) compared the survival of 10 children with FLC and 36 with HCC, without finding differences according to histology.
Conclusion
We concluded that FLC patients diagnosed before 23 years of age have worse prognosis than those diagnosed after age 23. Other factors associated with worse prognosis in this study are: lack of surgical treatment, presence of positive surgical margins, vascular invasion, and altered hepatic enzymes.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
JGL and AT reviewed medical records and participated in the design BM reviewed the pathological material and helped to draft the manuscript. MU and ELR participated in the design of the study and coordination. LML participated in the design of the study and performed the statistical analysis. OA conceived of the study, and participated in its design and coordination, performed the statistical analysis and helped to draft the manuscript. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
The authors thank Ali Zirakzadek M.D. for technical assistance.
Figures and Tables
Figure 1 Survival analysis of patients with HCC (n = 159) compared with FLC (n = 15) p = 0.007.
Figure 2 Fibrolamellar Hepatocarcinoma (400#215;), hematoxylin-eosin with big polygonal cells, and abundant eosinophil granular cytoplasm (A). Masson trichromic (100×) with cordons of hepatocytes separated with dense collagen septum (B). Immunohistochemistry for PCNA that shows high index of cellular proliferation of the nucleus (C)
Figure 3 Global survival of patients with FLC at 6, 12, 24, 36 and 60 months was 66, 66, 53, 40 and 26% (median survival of 30 ± 6 months).
Figure 4 Survival analysis of prognostic factors of patients with FLC. Age = 23 years (n = 7, median survival of 8 ± 2 months) compared to age > 23 years (n = 8, median survival of 65 ± 19 months), p = 0.0132.
Figure 5 Survival analysis of prognostic factors of patients with FLC. Patients who underwent surgical treatment (n = 10, median survival of 60 ± 10 months) vs patients with medical treatment only (n = 5, median survival of 5 ± 2 months), p = 0.0011.
Figure 6 Survival analysis of prognostic factors of patients with FLC. Patients who underwent surgical treatment with positive (n = 5, median survival of 20 ± 5 months) and negative (n = 5, median survival of 65 ± 4 months) surgical borders, p= 0.06.
Figure 7 Survival analysis of prognostic factors of patients with FLC. Patients with (n = 9, median survival of 5 ± 2 months) and without (n = 6, median survival of 60 ± 10 months) vascular invasion, p = 0.006.
Figure 8 Survival analysis of prognostic factors of patients with FLC. Patients with (n = 10, median survival of 26 ± 11 months) and without (n = 5, median survival of 65 ± 21 months) altered liver enzymes, p = 0.04.
Table 1 Association of global survival in patients with FLC to clinical and pathological characteristics.
Variable Survival (months) P (log rank)
Age
≤ 23 years 8 ± 2 0.013
> 23 years 65 ± 19
Gender
male 26 ± 6 0.646
female 42 ± 27
Surgical
resection 60 ± 10 0.011
No resection 5 ± 2
Surgical border
Positive 20 ± 5 0.06
Negative 65 ± 4
Fever
Present 67 ± 14 0.188
Absent 26 ± 7
Jaundice
Present 65 ± 19 0.345
Absent 26 ± 13
Palpable Mass
Present 42 ± 33
Absent 26 ± 14 0.84
Weight loss
Present 26 ± 6 0.646
Absent 42 ± 27
Ascites
Present 4 ± 32 0.578
Absent 42 ± 2
Hepatomegaly
Present 42 ± 20 0.681
Absent 20 ± 14
Vascular Invasion
Present 5 ± 2 .0061
Absent 60 ± 10
Atypia
low 26 ± 13 0.319
Moderate 65 ± 9
Hemoglobin
≤ 12 26 ± 19 0.527
>12 42 ± 28
Liver enzymes
abnormal 26 ± 11 0.04
normal 65 ± 21
Alkaline phosphatase
abnormal 65 ± 44 0.682
normal 42 ± 17
Albumin
abnormal 5 ± 30 0.9
normal 42 ± 19
Alfa fetoprotein
abnormal 20 ± 21 0.672
normal 42 ± 21
TNM
Stage I and II 65 ± 14 0.693
III 60 ± 12
IV 8 ± 8
==== Refs
EL-Serag HB Mason AC Key C Trends in survival of patients with hepatocellular carcinoma between 1977 and 1996 in United States Hepatology 2001 33 62 5 11124821 10.1053/jhep.2001.21041
Edmondson HA Differential diagnosis of tumors and tumor-like lesions of the liver in infancy and childhood Am J Dis Child 1956 91 168 86
Ruffin MT 4th Fibrolamellar Hepatoma Am J Gastroenterology 1990 85 577 81
Farhi DC Shikes RH Silverberg SG Ultrastructure of Fibrolamellar Oncocytic Hepatoma Cancer 1982 50 702 9 6284337
Wong LK Link DP Frey CF Ruebner RH Tesluk H Pimstone NR Fibrolamellar Hepatocarcinoma: Radiology, Management and Pathology AJR Am J Roentgenol 1982 139 172 5 6283868
Teitelbaum DH Tuttle S Carey LC Clausen KP Fibrolamellar Carcinoma of the Liver. Review of three cases and the presentation of a characteristic set of tumor markers defining this tumor Ann Surg 1985 202 36 41 2409935
Berman HA Burnham JA Sheahan DG Fibrolamellar carcinoma of the liver: an immunohistochemical study of nineteen cases and a review of the Literature Hum Pathol 1988 19 784 93 2456977
Craig JR Peters RL Edmondson HA Omata M Fibrolamellar carcinoma of the liver: a tumor of adolescents and young adults with distinctive clinico-pathologic features Cancer 1980 46 372 9 6248194
Lack EE Neave C Vawter GF Hepatocellular carcinoma. Review of 32 cases in childhood and adolescence Cancer 1983 52 1510 5 6311396
Bralet MP Regimbeau JM Pineau P Dubois S Loas G Degos F Valla D Belghiti J Degott C Terris B Hepatocellular carcinoma occurring in nonfibrotic liver: epidemiologic and histopathologic analysis of 80 French cases Hepatology 2000 32 200 4 10915724 10.1053/jhep.2000.9033
Katzenstein HM Krailo MD Malogolowkin MH Ortega JA Qu W Douglass EC Feusner JH Reynolds M Quinn JJ Newman K Finegold MJ Haas JE Sensel MG Castleberry RP Bowman LC Fibrolamellar Hepatocellular Carcinoma in Children and Adolescents Cancer 2003 97 2006 12 12673731 10.1002/cncr.11292
Izumi R Shimizu K Kiriyama M Hashimoto T Urade M Yagi M Mizukami Y Nonomura A Miyazaki I Alpha-fetoprotein production by hepatocellular carcinoma is prognostic of poor patient survival J Surg Oncol 1992 49 151 5 1372375
Yamashita Y Takahashi M Koga Y Saito R Nanakawa S Hatanaka Y Sato N Nakashima K Urata J Yoshizumi K Prognostic factors in the treatment of hepatocellular carcinoma with transcatheter arterial embolization and arterial infusion Cancer 1991 67 385 91 1845943
Nakakura EK Choti MA Management of hepatocellular carcinoma Oncology (Huntingt) 2000 14 1085 102
Arrieta O Garcia E Guevara P Garcia-Navarrete R Ondarza R Rembao D Sotelo J Hepatocyte growth factor is associated with poor prognosis of malignant gliomas and is a predictor for recurrence of meningioma Cancer 2002 94 3210 8 12115353 10.1002/cncr.10594
El-Serag HB Davila JA Is fibrolamellar carcinoma different from hepatocellular carcinoma? A US population-based study Hepatology 2004 39 798 803 14999699 10.1002/hep.20096
Hernandez-Castillo E Mondragon-Sanchez R Garduno-Lopez AL Gomez-Gomez E Ruiz-Molina JM Onate-Ocana LF Hepatocellular carcinoma in youth: a comparative analysis with hepatocarcinoma in adults Rev Gastroenterol Mex 2003 68 94 9 15127644
Dahan MH Kastell P Fibrolamellar hepatic carcinoma with a presentation similar to that of septic pregnancy. A case report J Reprod Med 2002 47 47 9 11838311
Gemer O Segal S Zohav E Pregnancy in a patient with fibrolamellar hepatocellular carcinoma Arch Gynecol Obstet 1994 255 211 2 7695368 10.1007/s004040050054
Kroll D Mazor M Zirkin H Schulman H Glezerman M Fibrolamellar carcinoma of the liver in pregnancy. A case report J Reprod Med 1991 36 823 7 1662721
Ichikawa T Federle MP Grazioli L Madariaga J Nalesnik M Marsh W Fibrolamellar hepatocellular carcinoma: imaging and pathologic findings in 31 recent cases Radiology 1999 213 352 61 10551212
|
16259635
|
PMC1310628
|
CC BY
|
2021-01-04 16:03:05
|
no
|
BMC Cancer. 2005 Oct 31; 5:142
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-142
|
oa_comm
|
==== Front
BMC Public HealthBMC Public Health1471-2458BioMed Central London 1471-2458-5-1211628865910.1186/1471-2458-5-121Research ArticleResponse of religious groups to HIV/AIDS as a sexually transmitted infection in Trinidad Genrich Gillian L [email protected] Brader A [email protected] Fulbright Fellowship Program for U.S. Students, Port of Spain, Trinidad, West Indies2 NCID/DVRD/IDPA, Centers for Disease Control and Prevention, Atlanta, GA, 30333, USA3 Center for Medical Sciences Education, Faculty of Medical Sciences, UWI, St. Augustine, Trinidad, West Indies2005 16 11 2005 5 121 121 17 12 2004 16 11 2005 Copyright © 2005 Genrich and Brathwaite; licensee BioMed Central Ltd.2005Genrich and Brathwaite; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
HIV/AIDS-related stigma and discrimination are significant determinants of HIV transmission in the Caribbean island nation of Trinidad and Tobago (T&T), where the adult HIV/AIDS prevalence is 2.5%. T&T is a spiritually-aware society and over 104 religious groups are represented. This religious diversity creates a complex social environment for the transmission of a sexually transmitted infection like HIV/AIDS. Religious leaders are esteemed in T&T's society and may use their position and frequent interactions with the public to promote HIV/AIDS awareness, fight stigma and discrimination, and exercise compassion for people living with HIV/AIDS (PWHA). Some religious groups have initiated HIV/AIDS education programs within their membership, but previous studies suggest that HIV/AIDS remains a stigmatized infection in many religious organizations. The present study investigates how the perception of HIV/AIDS as a sexually transmitted infection impacts religious representatives' incentives to respond to HIV/AIDS in their congregations and communities. In correlation, the study explores how the experiences of PWHA in religious gatherings impact healing and coping with HIV/AIDS.
Methods
Between November 2002 and April 2003, in-depth interviews were conducted with 11 religious representatives from 10 Christian, Hindu and Muslim denominations. The majority of respondents were leaders of religious services, while two were active congregation members. Religious groups were selected based upon the methods of Brathwaite. Briefly, 26 religious groups with the largest followings according to 2000 census data were identified in Trinidad and Tobago. From this original list, 10 religious groups in Northwest Trinidad were selected to comprise a representative sample of the island's main denominations. In-depth interviews with PWHA were conducted during the same study period, 2002–2003. Four individuals were selected from a care and support group located in Port of Spain based upon their perceived willingness to discuss religious affiliation and describe how living with a terminal infection has affected their spiritual lives. The interviewer, a United States Fulbright Scholar, explained the nature and purpose of the study to all participants. Relevant ethical procedures associated with the collection of interview data were adopted: interviews were conducted in a non-coercive manner and confidentiality was assured. All participants provided verbal consent, and agreed to be interviewed without financial or other incentive. Ethics approval was granted on behalf of the Caribbean Conference of Churches Ethics Committee. Interview questions followed a guideline, and employed an open-ended format to facilitate discussion. All interviews were recorded and transcribed by the interviewer.
Results
Religious representatives' opinions were grouped into the following categories: rationale for the spread of HIV/AIDS, abstinence, condom use, sexuality and homosexuality, compassion, experiences with PWHA, recommendations and current approach to addressing HIV/AIDS in congregations. Religious representatives expressed a measure of acceptance of HIV/AIDS and overwhelmingly upheld compassion for PWHA. Some statements, however, suggested that HIV/AIDS stigma pervades Trinidad's religious organizations. For many representatives, HIV/AIDS was associated with a promiscuous lifestyle and/or homosexuality. Representatives had varying levels of interaction with PWHA, but personal experiences were positively associated with current involvement in HIV/AIDS initiatives. All 4 PWHA interviewed identified themselves as belonging to Christian denominations. Three out of the 4 PWHA described discriminatory experiences with pastors or congregation members during gatherings for religious services. Nonetheless, PWHA expressed an important role for faith and religion in coping with HIV.
Conclusion
Religious groups in Trinidad are being challenged to promote a clear and consistent response to the HIV/AIDS epidemic; a response that may reflect personal experiences and respect religious doctrine in the context of sex and sexuality. The study suggests that (1) religious leaders could improve their role in the fight against HIV/AIDS with education and sensitization-specifically aimed at dismantling the myths about HIV transmission, and the stereotyping of susceptible sub-populations, and (2) a consultative dialogue between PWHAs and religious leaders is pivotal to a successful faith-based HIV intervention in Trinidad.
==== Body
Background
The HIV/AIDS epidemic in the Caribbean region is fuelled by stigma and discrimination, which are the most significant determinants of HIV infection and death from AIDS-related complications [1]. Trinidad and Tobago (T&T), an island nation situated off the coast of Venezuela and home to 1.3 million citizens, [2] shares with region members an incidence of HIV/AIDS second only to sub-Saharan Africa [3]. Since the first case of HIV was identified in Trinidad in 1983 in a homosexual male, [4] the prevalence of HIV/AIDS in adults (15–49 years old) has grown to 2.5%, with half of all new infections occurring in young people between 15–24 years [5]. HIV is primarily transmitted through unprotected sexual intercourse, [6] and is fueled by multiple sexual partnerships, substance abuse, and migration and gender inequalities [7]. The Trinidad and Tobago Ministry of Health suspects the infection is underreported and the actual number of cases is twice as high [5].
Stigma and discrimination create barriers to HIV testing and treatment, care and support networks for people living with and affected by HIV/AIDS [8,9]. In Trinidad, few cases of HIV infection are diagnosed early in the course of infection and the average time from HIV diagnosis to death is only 13 months [4]. Of all confirmed AIDS cases reported in 1999, 75% were identified as HIV positive within the same year as the AIDS diagnosis [4]. The statistics suggest that individuals may fear testing, perhaps as a result of the stigmatization of HIV throughout Trinidad and Tobago's society.
T&T is a spiritually-aware society where the majority of the population belongs to at least 1 of over 104 religious groups [10] that coexist peacefully, often participating in public celebrations of other faiths [11]. When the last census was taken, approximately 30% of the population was Roman Catholic, 24% Hindu, 11% Anglican, 6% Muslim, 3% Presbyterianism and 26% "other" [12]. Religious groups are geographically evenly distributed throughout the nation; active amongst the poorest communities and within areas of high HIV/AIDS prevalence [13]. The pervasiveness of diverse religious ideologies thus creates a complex environment for the transmission of a sexually transmitted infection.
Religious leaders are esteemed, frequently exchange with the public and maintain an influential role in policy-making in Trinidad [11]. They may use their position to promote HIV/AIDS awareness, fight stigma and discrimination in communities, and exercise compassion to facilitate healing for people living with HIV/AIDS (PWHA). Some religious groups are involved in such efforts. In 2001, the Caribbean Conference of Churches (CCC), the Regional Ecumenical Organization of the Caribbean, brought 120 church leaders and church workers from across the region together in a consultation on "Human Sexuality and HIV/AIDS in the Caribbean – A Theological Approach" [14]. The consultation raised awareness about the discrimination, fear, rejection, poverty and pain that PWHA may face in Trinidad's society.
Yet there are barriers to more active and widespread involvement in HIV/AIDS initiatives among religious groups in Trinidad. Debate over condom sales, for example, has hindered collaboration with public health organizations [15] and religious groups have not been optimally integrated into the HIV/AIDS care and support network. Further, there are few local research studies that explore religious leaders' incentives to promote and gain involvement in faith-based HIV/AIDS initiatives. One investigation of the potential to inspire a faith-based response to HIV/AIDS in T&T [13] found that HIV/AIDS-related stigma and discrimination inhibited active involvement. The present research expands upon previous work by sampling both diverse religious groups in Trinidad, and individuals who are living with HIV/AIDS. Accordingly, this study investigates how the perception of HIV/AIDS as a sexually transmitted infection impacts religious leaders' incentives to become involved in HIV/AIDS initiatives, and how the experiences of PWHA in religious gatherings have impacted their healing and coping with HIV/AIDS.
Methods
All relevant ethical procedures associated with collection of interview data were adopted, and the methods approved on behalf of the Caribbean Conference of Churches Ethics Committee. The interviewer explained the nature and purpose of the study to the participants and conducted the interviews in a non-coercive manner, with the assurance of confidentiality. All respondents provided verbal consent, and agreed to be interviewed without financial or other incentive.
Selection of religious representatives
Between November 2002 through April 2003, in-depth interviews were conducted with 11 consenting representatives from 10 religious organizations in Trinidad, including: Anglican, Open Bible, Pentecostal, Salvation Army, Unity School of Christianity, Seventh Day Adventists, Hindu, Jamaat al Muslimeen, Nation of Islam, and Roman Catholic (2 representatives). Selection of religious groups was based upon the methodology of Brathwaite [13]. Briefly, 26 religious groups with the largest followings according to 2000 census data were identified in Trinidad and Tobago. Groups were categorized into Christian, Hindu, Muslim, and 'native' (Bahai, Orisha). From this original list, 10 representative groups from Christian, Hindu and Muslim denominations with large and small congregations, and located in Northwest Trinidad were selected. Potential participants were contacted by phone, the research plan described, and interviews arranged with willing participants. Religious respondents were generally leaders of religious services (pastors, priests, pundits), while 2 were active congregation members. The majority of interviews with Christian representatives were conducted in offices adjacent to churches. Other interviews were held in an agreed location such as the interviewee's home. Interviews ranged from 1–2.5 hours. Interview questions followed a guideline (see Figure 1), but employed an open-ended question format to invite discussion of sensitive HIV/AIDS-related themes. Those themes were previously identified in a survey of 26 religious organizations, [13] and included: rationale for the spread of HIV/AIDS, abstinence, condom use, sexuality and homosexuality, compassion, experiences with PWHA, and recommendations and current approach to addressing HIV/AIDS in congregations.
Figure 1 Interview Questions for Religious Representatives.
People living with HIV/AIDS (PWHA)
In-depth interviews with PWHA were conducted during the same study period, 2002–2003. Four PWHA (P1–P4) were selected from an established HIV/AIDS support group in Port of Spain, Trinidad, based upon their perceived willingness to discuss sensitive issues and religious affiliation. Interview questions were open-ended, but followed a structured guideline (see Figure 2).
Figure 2 Belief Systems Questionnaire for People Living with HIV/AIDS.
The interviewer
The interviewer was a United States Fulbright Scholar who had studied HIV/AIDS and other public health issues, and intended to use the research on religious groups to stimulate faith-based contributions to ongoing HIV/AIDS initiatives in Trinidad. The interviewer was associated with the HIV/AIDS support group from which the 4 PWHA were selected, and had spent several hours each week interacting with the participants prior to the interviews. All interviews were audio recorded and transcribed by the interviewer.
Results
Verbatim statements with religious representatives were extracted from raw interview text and were categorized according to the following themes: rationale for the spread of HIV/AIDS, issues related to abstinence, condom use, sexuality, homosexuality, personal compassion and personal experiences with PWHAs. Similarly, the responses of PWHA were grouped into three main themes: response to HIV diagnosis, experiences in church, and current attitude toward religion.
Introduction to religious groups
Anglican Church (AC)
In keeping with "a more collegial approach to ministry," the Anglican community was engaged in educational programs, such as peer counseling, to sensitize and educate the clergy on issues surrounding HIV/AIDS. According to the AC representative, "AIDS is one of our chief priorities: our thesis that the faith based communities must be part of the solution than a problem by being negative and adversarial versus pastoral...."
Open Bible
The respondent felt that because HIV/AIDS prevalence in T&T had not reached critical proportions like in Africa, it was not an urgent problem. There was no written policy on HIV/AIDS because "we have not had a lot of cases to deal with. I guess if we had that amount we would. We have treated every case on an individual basis." It was felt that if the problem were addressed on a national level, churches would necessarily unite to develop a joint initiative.
Pentecostal
The representative upheld compassion for all individuals, recognizing the dignity of every human life. The organization would assist PWHA in order to help to facilitate conversion and lifestyle changes that would lead to an improved contribution to the community. HIV/AIDS has caused a devastation that requires communities and individuals to reexamine and change their behavior.
Roman Catholic (RCI, and RCII)
RCI believed the degradation of the traditional family unit, the erosion of traditional cultural values and undefined rites of passage have contributed to the spread of HIV/AIDS. Religious instruction in the Church has failed to meet the present day needs and concerns of the congregation. RCII felt that HIV/AIDS must evoke a discussion of sexuality and of how individuals could prepare themselves to make informed, meaningful and healthy choices that reflect who they are and not simply what they want. To date, the Roman Catholic Church has very effectively delivered home health care to individuals suffering from AIDS-related complications through ministries such as Caritas.
Salvation Army (SA)
The Salvation Army was founded in 1878 by William Boothe, a Methodist minister in England, who wanted to help the poor of England receive food, shelter and clothing. Thus, the organization is based on a philosophy of outreach, serving community needs, helping the sick and the poor. SA "does not condone sin...but we love the sinner," and would work with individuals to build a better lifestyle. SA advocated chastity, but condom use was preferred to acquiring HIV for individuals who choose to engage in sexual intercourse before marriage.
Seventh Day Adventist (SDA)
There was a written policy on HIV/AIDS, but the interviewee was not familiar with the text in the policies. The interviewee was disappointed in the rate at which the SDA organization had responded to the HIV/AIDS epidemic in Trinidad and acknowledged that there was the need for further education of clergymen within the church. Increasing education in society and among the clergy was the most critical component of reducing the spread of the virus.
Unity School
Unity was a small congregation of mostly women, who were described as traditional and conformist. HIV/AIDS did not affect the congregation and the representative did not have personal relationships or experiences with PWHA. The organization was not involved in outreach. Unity is based upon the philosophy that "human beings create their experiences by the activity of their thinking." Therefore prayer is "creative thinking that heightens the connection with God."
Hindu
The Hindu representative felt that HIV/AIDS occurred primarily among homosexuals and did not pose a significant problem for the Hindu organization. Furthermore, it was assumed that Hindus were less likely to acquire HIV due to high social and spiritual obligations to obey religious doctrine. According to the interviewee, HIV/AIDS was a medical problem. Although prayers and mantras are effective treatments and cures for disease, the organization was less concerned about bodily ailments than it is about eternal life. The interviewee felt that individuals living with HIV/AIDS in the Hindu community may feel discriminated against and ostracized because disease is an "unhygienic situation;" individuals living with HIV/AIDS are unclean and would be expected to stay away from organized worship. Compassion is inherent in Hinduism, but the religion does not provide the opportunity for confession and reconciliation.
Nation of Islam (NOI)
NOI supported a theory that HIV was man-made in a United States laboratory in a plan to control population growth. The NOI representative was skeptical of scientific literature and research on the efficacy of condoms and boldly supported abstinence as the only effective prevention of HIV and other STDs. NOI valued the individual, and is particularly protective of women. The organization fulfills the obligation to help anyone in need.
Jamaat al Muslimeen
This representative was highly active in HIV/AIDS education for the neediest urban communities. Jamaat al Muslimeen traveled on foot to reach individuals and families who demonstrated serious health risks, and provided education and condoms. However, the representative's activism, particularly in condom distribution, was not supported by the Jamaat al Muslimeen organization. According to the interviewee, the organization believed HIV/AIDS was a "sin from God."
Rationale for the spread of HIV/AIDS
Religious representatives described cultural, spiritual, and social factors that contributed to the spread of HIV/AIDS. RCI believed that new opportunities for HIV transmission were born in the fragmentation of traditional family structure and the erosion of the influence of religious doctrine in society. According to the Pentecostal representative, "if you follow the pattern that is established here in the Word, that you are not going to get into trouble, or there will be no involvement in sexual accounts...like HIV...or any of the other kinds...." The Open Bible representative "really believe that this started with homosexuality. They can't find a cure for it...what you sow, you reap."
The "Carnival mentality," was also felt to fuel the epidemic. Group representatives generally described how Carnival encouraged individuals to abandon their moral framework. The Hindu representative did not feel that Hindus would engage in this behavior, however.
Pentecostal: "It's what people are making of the Carnival mentality. If people would enjoy themselves and see Carnival as functional in terms of your socioeconomic...we will not have this kind of [situation]... [People] let their standards down...to do what they would not normally do. And this to me is prostituting Carnival...To go around and have sex, as if, you know, you have no control."
Open Bible: "Carnival is one of the greatest contributors to AIDS...there is no limits, there is no restrictions. People just let down all there guards that they will have had all year...And only afterwards they realize they make mistakes...too late."
Hindu: "You find too that around Carnival time...this is highlighted most of all, and you would find that many Hindus would not be seen in the streets taking part in Carnival. You may see [East] Indians, people of Indian descent, many of them would have already crossed over to other religions. And so the guard is already dropped...promiscuity, and licentiousness...."
Abstinence
There was little variability among religious groups in response to the organization's position on abstinence.
Pentecostal: "We are not going to tolerate at all any sexual activity outside of the bonds of marriage. We go in accordance with what the Word says that marriage is honorable in all, and the bed undefiled-the only time the bed is undefiled is when there is marriage."
Salvation Army: "Sex before marriage is not acceptable in the Christian Church."
SDA: "I believe that we are aware of the reality, of not everybody will abstain, but we would emphasize abstinence, from a doctrinal point of view, and because we believe that it is the safest... Our thing of abstinence is not strictly about avoiding HIV; abstinence is also about...pregnancy, it's about...sexual relationships, where you have amount of responsibility...If it's with somebody you are very interested in...not just a one night stand."
NOI: "...the sole objective is to promote abstinence-not the fact that you can get an STD. I don't relate to that at all. And we don't go around, you know, promoting condom use, we promote abstinence because there is a much greater value attached to the individual, particularly the female that if she was sensitized and made aware of that and I think that would be much more effective for her than the condom."
Condom use
Within Christian denominations and across religious groups, positions on condom use differed, ranging from an acceptance of condoms in lieu of the "reality" of HIV/AIDS, to a general contempt for their use as a substitute for self-control.
SDA: "From my personal point of view, you cannot promote the use of condoms, because in promoting the use of condoms, what you're actually doing is telling the person you cannot control your sexual urge. So because you cannot control your sexual urge, here is something to use when the urge comes...And you find that as a country growing, so many young people are contracting HIV in spite of the availability of condoms."
RCII: "What I was very clear about myself was that...we are not talking about condoms as contraceptives when you talk HIV. That the church's condemnation of condoms is about contraception; you are talking about a contra-abortive, which is people dying."
Pentecostal: "We don't think that it is fair or right to distribute condoms to... like young people who are not married. You are not supposed to be actively involved with sex outside of marriage, so any person or persons outside of marriage, we feel, should not really have the use of any condom. Condoms within the marriage we feel, should, in our view, be a matter for the people involved."
Salvation Army: "Sex before marriage is not acceptable in the Christian Church, but being practical in this day and age, better use a condom than get a disease."
Open Bible: "...teaching the people the importance of not being promiscuous...providing condoms...that doesn't fix the problem...the problem is the individual...so start from small...there is a whole lifestyle that starts from a little child, and you have to start there."
NOI: "I read one interesting quotation from a medical professional that has stayed with me. And it says that the AIDS virus passes through...the membrane of a condom like a golfball passes through a basketball hoop...."
The Jamaat al Muslimeen organization forbids condom use. However, the interviewee's position was in marked contrast to that held by the organization. The Jamaat al Muslimeen representative carried condoms on foot to the poorest of Trinidad's communities in opposition to the philosophy of the organization:
Jamaat al Muslimeen: "And what I do also is that I distribute condoms. And a lot of people that I have spoken to in the underdeveloped community, most of them use condoms sometimes, and sometimes they don't use condoms... we try to educate them towards the disease...."
Sexuality and homosexuality
Adherence to religious doctrine and the fear of shame are thought of as protective factors that insulate the Hindu group from a promiscuous lifestyle. The Unity group felt that Trinidad and Tobago was "probably the most promiscuous little country in the world." Roman Catholic representatives were concerned about the growing number of young children voluntarily initiating sexual activity, and believed that sex among the youth was becoming a defining characteristic of the culture. The RCII representative explained that gender equality provides enormous potential for mutuality, and needs to be explored by society.
RCII: "Sex is a powerful potent force in human society.... For me the one thing that is difficult in Caribbean society that is distinct from some more traditional societies, is that the kind of rituals of initiation which have allowed people to claim manhood, womanhood without become sexually active in the open sense, those rituals don't exist.... Sexuality is extremely fragile...we talk about it simply as something that we do when in fact it is something that you are...."
According to many representatives, education on sexuality in the context of religious doctrine was unnecessary, because religious tenets sufficiently define appropriate behavior: if one upheld the teachings in the sacred texts, HIV/AIDS would not be transmitted or acquired. The opinion on homosexuality was generally uniform across religious denominations, although personal attitudes varied in their degree of outrage; some groups called it "abominable," or "sickening". Some interviewees, however, expressed the potential for homosexuals to be converted, and adopt an acceptable form of behavior.
Open Bible: "We are not against homosexuals but we don't promote homosexuality...we strongly disagree with it. We believe that God never intended for people to live in a homosexual relationship and so certainly we don't in any form or fashion, entertain it...."
Unity: "AIDS is not new. I don't think AIDS comes from any homosexual behavior. I think people are capable of loving."
SDA: "From my personal observation I don't think that homosexuality is something that our ministers have a lot to do with...they have not done a lot of interacting."
Hindu: "In all religions, sex is looked at as very sacred. Whereas you would not find the Hindu woman covered all over like the Muslim, they ought to be quite protected, and I think from the woman folk point of view, it is even less a threat. You would find that homosexuals mainly from the male contingency, and...he would be out there in the world, and he may encounter certain situations, and he may get into...this promiscuous activity and may become homosexual...."
Compassion and religious representatives' experiences with PWHA
Personal interactions between group representatives and PWHA varied widely. Some representatives had buried individuals who had died from AIDS-related complications. The Jamaat al Muslimeen representative personally contacted individuals and communities in Port of Spain affected by HIV/AIDS to provide them with condoms, support and education. In contrast, other representatives, including Open Bible, Unity, and Hindu, reported little to no personal interaction with PWHA. Nonetheless, religious groups unanimously supported compassion for people living with and affected by HIV/AIDS. The expression of compassion was often associated with conversion and a desire to "help the person change."
RCII: "The point of breakthrough was to equate HIV positive persons with lepers in the Gospel story, so Jesus came for the lepers, he's come for them. Which is so horrible...but it was a way for the churches to open up to it...
"There was a conversion process involved in many people... from AIDS as punishment from God to AIDS as a sad event in human history which now demands a response from those who say they believe in the name of God, but that that response must always be compassion...."
Anglican: "Last month I buried at least two persons with AIDS. A twenty-five year old male, thirty-eight year old female-she sold drugs and she also sold herself-her body. A sad, sad, sad moment...AIDS is not an academic thing here...it is very concrete...."
Pentecostal: "So we feel we need to relate with them treat with them as members like anyone else and we ought to show that the same kind of love, the same kind of respect...and we do that."
The Pentecostal representative also articulated how myths about HIV transmission were dispelled in personal interactions with PWHA:
Pentecostal: "Many of the myths are...HIV can be spread with the use of utensils-many believe in those myths even within the church...The fallacies...or the myths have been put to rest for us...because of the experiences we have had dealing with people who have had HIV...."
Salvation Army: "We actually had an HIV/AIDS person who was very close to us in our formal apartment...he was bold enough to tell us he had the problem, and so we helped him a great lot...we knew that he needed extra food...he was rejected from his own house, and we had to help him get settled...so we sort of have an idea...it is a problem that needs help. If you come to me and say I have HIV/AIDS, I am not going to say that you did something wrong... We have reached out into many areas of social work because we have a heart for people."
SDA: "The churches responsibility is to show compassion. And not to check and find out how this person managed to get it. I think most Adventist churches are moving away from the medical fact...."
Hindu: "This religion is based upon compassion, if you do not have compassion, you are lacking in one of the major ingredients to be a Hindu. So we do not go about branding anybody, saying okay, you are a sinner. It is more understanding, it is also a recognition of the unhygienic situation that arises."
Jamaat al Muslimeen: "I have a twenty-one year old girl, she's HIV positive...I was even trying to get her to get public assistance because she and the two children are HIV positive. And I really thought that she was trying to turn around her life. But she just kept going with men without condoms."
According to some religious representatives, coping with HIV/AIDS was different than coping with other terminal diseases that are not sexually transmitted, such as cancer: people living with HIV/AIDS come to church requesting confession, whereas people fighting cancer want to be healed. The RCII represented expressed the contrast this way:
RCII: "What she was showing was a different form from someone who has cancer. But the fear that I saw in her was different from the fear I saw in cancer patients. It was fear coupled with guilt, and of course she came asking me to pray plenty... you hear the AIDS people telling you that there's a certain anger with themselves."
Recommendations and approach to addressing HIV/AIDS in congregations
Religious organizations differed on how to confront the HIV/AIDS epidemic in Trinidad, if at all. Some representatives, like RCII, began their involvement in HIV/AIDS care and support networks early in the epidemic, while others remained insulated.
RCII: "But what they did very well was to train a number of people would could go into homes and provide home care for people who had nobody else to care for them...More recently, the schools have been involved in education and the Catholic Church is in the process of providing, producing a video about HIV."
Open Bible: "I don't think there is a need to have something structured in place. I guess in Trinidad it is still a very private matter...I don't think that this can be addressed by any one local church, I think this is something that is more a national issue...and if it has to be dealt with then churches have to get together to deal with it because of the scope of it."
Anglican: "If you don't arrest this AIDS thing...what will happen is that so much money will be spent on AIDS and people infected with the virus, you have little or no money to spend on cancer, and diabetes, and all the other things."
Pentecostal: "Obey the word, abstain from sex, and avoid HIV."
SDA: "We have been more or less targeting young people. A lot of workshops going on as well. We run a home care training workshop, we also do a sensitizing program for our young people...we try to bring across in as many of our programs as we can...activities that relate to HIV and AIDS, so that we are educating our young people. In terms of our ministerial staf...there are courses that they must necessarily do in the program... there is a health course that ministers must do while they are there in their training, and that of course exposes them to the myths and realities of HIV and AIDS...."
Hindu: "When I started thinking more deeply, I thought to myself, this is not so much of a Hindu perspective, or a Christian perspective, or a religious perspective, but a medical issue...I think that the avenue provided for help, is one of a spiritual environment. But it is not through religious bodies."
Religion in the lives of PWHA
Both Roman Catholic representatives provided a rationale for seeking God to facilitate coping with an HIV diagnosis:
RCI: "Well, they're in a hopeless situation! They're beyond human help and so the next thing you turn toward God."
RCII: "Because this is fundamentally a religious society. If you say you don't believe in God, doors close in your face. If you say you believe in God you stand a much better chance of getting help from certain quarters. Plus the psychology of HIV, you come face to face with your own mortality. Turning to God is probably the most natural thing to do...Once you enter into a mode where you think you're dying you go through a whole process of anger, of bargaining."
Four PWHA described their experiences in church and their spiritual journey subsequent to receiving an HIV diagnosis. One respondent agreed with religious representative, RC1, in that HIV was a "crisis" situation. Managing feelings of guilt were an important part of the initial coping process. When 1 of the interviewees was diagnosed with HIV, she felt she was being punished for committing a sin, and her pastor confirmed those feelings. All 4 individuals described discriminatory experiences by clergy or congregation members, but for 3 of the 4 PWHA, negative experiences did not affect attendance nor diminish a spiritual journey. HIV diagnosis generally inspired a desire to explore spirituality.
The first interviewee, P1, went to church irregularly growing up. When she was diagnosed with HIV in her early twenties, the church provided peace and solitude during her "crisis." However, she stopped attending when she suspected her pastor ostracized PWHA.
Response to HIV diagnosis
P1: "With HIV that I went. Before I never really had any kind of crisis. If a boyfriend and I split up, that wasn't a crisis; they had other guys out there. It was really when I found out about my status...I would go to church and people would see me crying...crying down the place...just tears. Letting my heart pour out and talking to God."
Experiences in church
P1: "I went to the church for solitude to get some sense of peace some kind of understanding as to why this is happening to me...we decided that I could speak to the pastor's wife. And she sat down with me and she said, you know she had a son who also died from it. I mean nobody would have expected a pastor's son to get HIV because they're not supposed to be living a promiscuous life...She comforted me and she told me you know I'm welcomed in this church at anytime.
I came and I told [my friend] about it. She knew the guy who died, her son. His family was not nice to him. His mother his father was not nice to him. For her to be giving me another story. It was unbelievable to think that they were not nice... I think that is one of the reasons too that I did not go back."
Current attitude towards religion
P1: "I kind of gave up on myself and I started going to a lot of parties. So when Sunday came, I would go party Saturday night and when Sunday came I can't get up in the morning to go to church. And that kind of threw me off from going, from attending that. But I want to start back going."
P2 grew up attending church daily with his mom, who was Roman Catholic. Later in life he experimented with drugs and sex. Near-death experiences were the impetus to reach out to a spiritual counselor, and he later became active in community outreach programs.
Response to HIV diagnosis
P2: "I was thinking about dying all the time. The addicts felt sorry for me.
But I don't feel like dying anymore. It is only by the grace of God that I am not depressed. What I care about doing the right thing, taking my medication and learning more about it. I would like to carry the message to schools ...I don't feel any less than anybody because I know that Jesus Christ loves me and I trust him."
Experiences in church
P2: "I talked to the pastor and a few of the deacons high in the Church and when I was sick they didn't come by. I felt they didn't respond as they should have. Last Saturday my pastor shook my hand."
Current attitude towards religion
P2: "God is my armor, my weapon has made me overcome a lot of hurdles and a lot of hang-ups...it has played a very important part...even with my addiction my spirituality has changed a lot. I choose God."
P3 was active in an SDA church, but before her HIV diagnosis was a member of an Islamic organization.
Response to HIV diagnosis
P3: "It happened in 1995...and I couldn't believe that I was HIV positive, because at that time I was not educated about living healthy with HIV. So I thought to myself that it would be the end of the world for me.
...I decided, listen if I have to die I need to make peace with my God, whoever the creator is...I also had a friend, who was a Christian...and he said to me Christ could help you, Christ can heal you...I started reading the Bible, with a longing in my heart to find out...if the creator was hearing me or not. I used to pray like five times a day because my body was diminishing, my hair was falling off, I had sores all over my body...things just started getting clearer and clearer to me, and my eyes were just opened, slowly but continually being opened to what is real, what is the reality of life."
Experiences in church
P3: "I was impressed to let the church know of the power of God, because I know it was a miracle, and I wanted them to know that the God that they serve is still in the business of doing miracles...But they did not respond to it very well, therefore I was faced with plenty stigma, and word got around...and I just saw everybody start whispering...Now I'm faced with a reality...if I get involved with anybody, everybody...scorn me, and any young man within the church comes around me, bet your bottom dollar someone's going to tell them, she's HIV positive."
Current attitude towards religion
P3: "It's all well and fine that governments are looking into HIV care and treatment, but why isn't the religious sector taking part in a more meaningful way? Presently I am trying to get the Adventist, we already we have an AIDS ministry, but they are not involved in a holistic view in terms of educating even their own people, far less the general public.
The church has to play a very important role in the fight against HIV and AIDS... The reason that I think that the virus has mushroomed, is because of moral standards...moral standards have gone down."
P4 regularly attended and was active in a Pentecostal church. The HIV diagnosis was confirmed at 18 years old.
Response to HIV diagnosis
P4: "I thought I was being punished for my sins. That's what he said too. I am being punished. And I caused it on myself, and on and on and on he went. But I don't think that is true...The first camp I went to after I was diagnosed, I talked to a social worker. She used to really encourage me. The first time I went to her I was really depressed, and asked her if she thought God was punishing me for my sins. She said, is God evil? What about all the children born with HIV, are they being punished? Bad things happen to everybody."
Experiences in church
P4: "I talk to my pastor. But I come to find out that he was not so trustworthy...I went to my pastor and I was talking to him about it and I find out that he was telling everybody...He said they needed to know so that they could pray, so the church could pray. But he just said that so he could tell everybody...Once the board knows-the board is family member leaders. I think everybody probably knew or was told by someone else. I still see him. I'll say hi. That was about a year and a half ago.
The last time I went to church I was talking to this little boy, and this woman told me not to touch him, if you touch him you will give him your germs. The boy was six months old. I learned a lot."
Current attitude towards religion
P4: "Despite all the real bad things, I believe in my religion. I guess in every organization you'll find good people and bad people...I'm always really love my religion, love church and love God. To myself I feel comfortable, so I don't think people should influence my relationship with God, I'm really, really, really trying to work on it."
Conclusion
Christian, Hindu and Muslim religious representatives differed in their attitudes and opinions on the following themes: rationale for the spread of the HIV/AIDS epidemic in Trinidad, sexuality and homosexuality, condom use, processes for healing, and the impetus to become involved in faith-based initiatives. Levels of awareness about the prevalence of HIV/AIDS in Trinidad, susceptible sub-populations, and knowledge of the mechanisms of HIV transmission also varied. Religious representatives isolated subgroups who were believed to be particularly susceptible to HIV infection, and in so doing, implicitly articulated HIV/AIDS stigma in different ways. For example, the Hindu representative believed that HIV infection was generally limited to homosexuals and promiscuous non-Hindus. For the SDA and Pentecostal representatives, HIV was synonymous with a promiscuous lifestyle and the transgression of abstinence.
Representatives were generally in agreement in their advocacy for the sanctity of marriage prior to engaging in sexual intercourse. In correlation, the SDA representative felt that condoms facilitated the transgression of abstinence and the degradation of self-discipline. The NOI representative felt that condoms aided and abetted promiscuity in society, but went further in claiming that condoms were not even an effective barrier against sexually transmitted infections. In contrast, the "reality" of HIV/AIDS for the SA representative led to a more accepting attitude toward condom use. The Jamaat al Muslimeen and RCII representatives had personal experiences with PWHA, and it was their understanding that condoms were life-saving tools. Thus, the present research also suggested that personal experiences and interactions among religious representatives and PWHA dispelled myths surrounding HIV/AIDS transmission, and sensitized individuals to the HIV/AIDS "reality." The SDA interviewee believed that personal experiences with PWHA were critical to the HIV/AIDS sensitization process, and for dismantling myths about transmission.
The full potential for religious groups to contribute to HIV/AIDS awareness efforts is currently untapped. While the majority of representatives admit that HIV/AIDS is a serious problem that is affecting the country and the world, there was wide disparity in the impetus for implementing a faith-based initiative targeting HIV/AIDS-related issues. The Anglican interviewee supported a proactive initiative and HIV/AIDS was among the Church's 5 priorities; however for the Pentecostal, SDA, Open Bible and Hindu groups, HIV/AIDS was not a priority that needed immediate attention and warranted discussion among congregations. According to the Pentecostal representative, HIV could simply be avoided by adhering to the behavioral conduct outlined by Christian tenets. Despite the Jamaat al Muslimeen representative's personal efforts in raising awareness about HIV/AIDS in rural communities, she felt that the "mix" of HIV/AIDS initiatives and faith-based communities invited stigmatization of PWHA.
Whether it was called "divine purpose" according to the Hindu representative, "openness to the transcendent," by the RCII interviewee, or the "God conscious part" by the Open Bible representative, it was agreed that humankind are inherently spiritual beings; and that Trinidad is indeed a spiritually-aware society. For the 4 participants living with HIV/AIDS, all of whom identified themselves as Christians, an HIV diagnosis inspired an exploration of spirituality, and led to a deeper connection with God-despite experiences of isolation and discrimination in church. One PWHA was identified as HIV positive by her pastor during a worship service, so that he could exemplify deserving consequences of sexual behavior. The attitude of this pastor seems to reflect the opinion of the Open Bible and Pentecostal representatives: by abstaining from sex one avoids HIV. Other PWHA were also discriminated against by members of the congregation. Nonetheless, for 3 of the 4 individuals, negative and discriminatory experiences did not affect attendance in church nor attenuate a spiritual journey.
Religious representatives were generally willing to participate in a care-giving capacity for people with HIV/AIDS because these efforts were built into their existing mission. Religious groups in the past have publicly acknowledged a responsibility and desire to be involved in care-giving for people living with and affected by HIV/AIDS. For example, organizations such as the Roman Catholic-sponsored Caritas have been successful in home health care for PWHA and their families. At the 1998 Youth Summit, religious representatives formulated "support resolutions," recognizing the need for their involvement in communities through the provision of information and counseling services for adolescents in society [16]. A 2001 report revealed that religious groups desired to improve their capacity to contribute to care and support [13].
Regional conferences such as those led by the CCC indicate that religious groups are beginning to mobilize in confronting the HIV/AIDS epidemic. Since the completion of this study, in 2005, the CCC hosted another consultation with Faith Based Organizations to develop an HIV/AIDS policy and action plan, entitled, "Guidelines for Caribbean Faith-Based Organizations in Developing Policies and Action Plans to deal with HIV/AIDS" [17]. The document is part of a collaborative effort, "Building a Faith Based Response to HIV/AIDS in the Caribbean" to enhance the response of faith-based organizations to Trinidad's HIV/AIDS epidemic. Furthermore, the highest level of government supports the crucial role religious groups may play in mitigating the impact of HIV/AIDS stigma and discrimination. National HIV/AIDS prevention efforts involving faith-based organizations are mandated by the 5 year HIV/AIDS National Strategic Plan (NSP), whose goals include the provision of necessary support within a holistic framework for those persons infected and affected by HIV/AIDS. These goals are currently undertaken by the newly-formed National AIDS Coordinating Committee [18].
Nonetheless, several religious representatives agreed that the pace of efforts on behalf of religious organizations has been too slow. The present research raises the following question: if HIV were not sexually transmitted, would religious organizations respond to the epidemic in the same way? The HIV/AIDS epidemic in Trinidad urgently calls upon religious groups to provide a clear and consistent response to issues of sex and sexuality that resonates with personal experience and respects religious doctrine. However, the present research highlights inconsistent attitudes and opinions on the moral and spiritual issues surrounding HIV/AIDS; such inconsistencies may serve as a barrier to a united faith-based initiative in Trinidad. Religious groups across all faiths and denominations are challenged to recognize that human beings are sexual beings; herein lays the dilemma for religious groups. A faith-based intervention must understand the complexity of preserving the central tenets of organized religion while embodying compassion for individuals as sexual beings.
This study was limited by a small sample size and the geographic location of Northwest Trinidad; this may affect the generalization of results throughout Trinidad and Tobago. On average, only one representative from each religious organization was interviewed, and opinions expressed did not necessarily reflect those of the religious organization as a whole. Nonetheless, HIV/AIDS-related stigma and discrimination will continue to fester throughout Trinidad and Tobago until all the republic's religious leaders, esteemed in the public eye, possess accurate information about HIV transmission, which may then be conveyed to congregations; and until religious leaders are sensitized to the experiences of PWHA. Prior to involvement in community-based education, care and HIV/AIDS support initiatives, religious leaders must possess compassion that is reinforced by personal experiences with PWHA. PWHA are receptive to faith-based counseling and support provided by religious leaders and congregation members. Thus, a consultative dialogue between PWHA and religious leaders is pivotal to a successful faith-based HIV/AIDS initiative in Trinidad.
Abbreviations
AC: Anglican Church
AIDS: Acquired Immuno-Deficiency Syndrome
CAREC: Caribbean Epidemiology Center
HIV: Human Immunodeficiency Virus
NOI: Nation of Islam
PWHA: People Living with HIV/AIDS
RC: Roman Catholic (group representatives, RCI and RCII)
SA: Salvation Army
SDA: Seventh Day Adventist
STI: Sexually Transmitted Infection
Competing interests
The findings and conclusions in this report are those of Dr. Brathwaite and Ms. Genrich and do not necessarily represent the views of the funding agency, the Institute for International Education and Fulbright Fellowship Program, nor the organizations where the authors currently work, the University of the West Indies and the Centers for Disease Control and Prevention, respectively. There were no competing interests, financial or otherwise, in the present investigation, and no incentive on behalf of the Institute for International Education to obtain the results found. No fees or funding were obtained from any organization that could gain or lose from the publication of the manuscript, and no stocks or shares are held in any organization that stood to gain or lose financially from publication.
Authors' contributions
Dr. Brathwaite contributed to the design of the research method. Ms. Genrich carried out the in-depth interviews with the participants, religious representatives and individuals living with HIV/AIDS. Both Dr. Brathwaite and Ms. Genrich collaborated on the final analysis and manuscript preparation. Both authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
Acknowledgements
With deep appreciation we thank the Fulbright Fellowship Program and Institute for International Education for making this important investigation possible. The authors are sincerely grateful to all participants for willing to openly share their opinions and experiences. Special thanks to Dr. Dean Knolly Clarke for his insight and support.
==== Refs
The role of religious leaders in reducing stigma and discrimination related to HIV/AIDS 2003 ICASA Satellite Session, Kenyatta International Conference Centre: A report of the round table discussion
Epidemiologic Fact Sheets on HIV/AIDS and Sexually Transmitted Infections, 2000 Update
The Caribbean Regional Strategic Plan of Action for HIV/AIDS. Caribbean Task Force on HIV/AIDS Draft 2000
Bartholomew C Raju CC Jankey N The acquired immune deficiency syndrome in Trinidad: A report on two cases West Indian Med J 1983 32 177 80 6636722
Ministry of Health (MOH) Health Sector Reform Programme of Trinidad & Tobago, © 2000 Ministry of Health & HSRP. The Government of Trinidad & Tobago, All rights reserved 2005
AIDS Cases by Exposure Category and Gender (T&T): Reported Cases of AIDS by Major Exposure Category and Gender 2001
UNAIDS Report on the Global AIDS epidemic, Executive Summary AND UNAIDS Fact Sheet HIV/AIDS in the Caribbean 2004
Richards Peter Discrimination Marks HIV Treatment in Trinidad and Tobago Inter Press Service 2002
Ransome D Prevention and care: Trinidad's twin messages AIDS Watch 1990 5 7 12282759
Murray E Religions in Trinidad and Tobago 1996 Murray Enterprises, Trinidad and Tobago
Trinidad and Tobago International Religious Freedom Report 2003 Released by the Bureau of Democracy, Human Rights and Labor
Statistics at a glance, a publication of the Central Statistical Office, CSO, of the Ministry of Finance, Government of Trinidad and Tobago 2001
Brathwaite Brader A Inventory and assessment of religious groups in Trinidad and Tobago and their response to AIDS/HIV Unpublished 2001 CAREC/GTZ
Caribbean Conference of Churches (CCC) Caribbean Conference of Churches spearheads a regional faith-based response to HIV/AIDS 2005
Richards Peter Health-Trinidad and Tobago: religious groups oppose condom sales 2001
Fournillier J A report on the first phase of the study, Inventory and Assessment of Religious Groups and HIV/AIDS in Trinidad and Tobago A project of NAIDS Program of the Ministry of Health 2000
Guidelines for Caribbean faith-based organizations in developing policies and action plans to deal with HIV/AIDS A project of the Canadian International Development Agency (CIDA) ESAC Programme, implemented by the Caribbean Conference of Churches (CCC) 2004
Alexis-Thomas Caroline Personal communication 2004 correspondence
|
16288659
|
PMC1310629
|
CC BY
|
2021-01-04 16:28:59
|
no
|
BMC Public Health. 2005 Nov 16; 5:121
|
utf-8
|
BMC Public Health
| 2,005 |
10.1186/1471-2458-5-121
|
oa_comm
|
==== Front
BMC GenomicsBMC Genomics1471-2164BioMed Central London 1471-2164-6-1631629723910.1186/1471-2164-6-163Research ArticleAn acquisition account of genomic islands based on genome signature comparisons van Passel MWJ [email protected] A [email protected] HH [email protected] ACM [email protected] Kampen AHC [email protected] der Ende A [email protected] Academic Medical Center, Department of Medical Microbiology, Amsterdam, the Netherlands2 Academic Medical Center, Clinical Epidemiology and Biostatistics, Amsterdam, the Netherlands3 Bioinformatics Laboratory, Amsterdam, the Netherlands2005 18 11 2005 6 163 163 13 6 2005 18 11 2005 Copyright © 2005 van Passel et al; licensee BioMed Central Ltd.2005van Passel et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Recent analyses of prokaryotic genome sequences have demonstrated the important force horizontal gene transfer constitutes in genome evolution. Horizontally acquired sequences are detectable by, among others, their dinucleotide composition (genome signature) dissimilarity with the host genome. Genomic islands (GIs) comprise important and interesting horizontally transferred sequences, but information about acquisition events or relatedness between GIs is scarce. In Vibrio vulnificus CMCP6, 10 and 11 GIs have previously been identified in the sequenced chromosomes I and II, respectively. We assessed the compositional similarity and putative acquisition account of these GIs using the genome signature. For this analysis we developed a new algorithm, available as a web application.
Results
Of 21 GIs, VvI-1 and VvI-10 of chromosome I have similar genome signatures, and while artificially divided due to a linear annotation, they are adjacent on the circular chromosome and therefore comprise one GI. Similarly, GIs VvI-3 and VvI-4 of chromosome I together with the region between these two islands are compositionally similar, suggesting that they form one GI (making a total of 19 GIs in chromosome I + chromosome II). Cluster analysis assigned the 19 GIs to 11 different branches above our conservative threshold. This suggests a limited number of compositionally similar donors or intragenomic dispersion of ancestral acquisitions. Furthermore, 2 GIs of chromosome II cluster with chromosome I, while none of the 19 GIs group with chromosome II, suggesting an unidirectional dispersal of large anomalous gene clusters from chromosome I to chromosome II.
Conclusion
From the results, we infer 10 compositionally dissimilar donors for 19 GIs in the V. vulnificus CMCP6 genome, including chromosome I donating to chromosome II. This suggests multiple transfer events from individual donor types or from donors with similar genome signatures. Applied to other prokaryotes, this approach may elucidate the acquisition account in their genome sequences, and facilitate donor identification of GIs.
==== Body
Background
From the many different prokaryotic genomes that have been sequenced in the last decade, it has been concluded that horizontal gene transfer (HGT) contributed significantly to the shape and size of microbial genomes [1]. Initially, HGT was regarded as an oddity in microbial genetics, with only a few antibiotic resistance genes in circulation. Currently, the estimates of putatively horizontally acquired DNA range from 0.5% in the endocellular symbiont Buchnera sp. APS genome up to 25% in the euryarchaeal Methanosarcina acetivorans genome, with an average of 14% in 116 prokaryotic genomes [2].
This genomic patchwork is clearly visible in the amount of genomic islands (GIs) detected in microbial genomes [3]. As more genomes of environmental strains are being sequenced, a variety of GIs providing diverse metabolic capacities are discovered in these non-pathogenic strains, emphasising that lateral genetic transfer is not limited to virulence traits [4]. Previous work on the functional categories of putative transferred genes resulted in the complexity theory, which contemplates the negative correlation between the probability of efficient acquisition and participation of the transferred genes in complex interactions [5]. Recently, it was claimed that transferred genes are biased towards functional categories associated with the cell surface, pathogenicity and DNA binding genes, although the proportion of putative genes with unknown functions remains very high in acquired sequences [2]. Dobrindt and co-workers explain acquisition efficiency mainly in terms of fitness increase [4]. These findings imply that diverse and interesting capacities are being exchanged between micro organisms.
In addition to their occasional location between mobile elements, such as phage sequences and insertion sequences suggesting a heterologous origin [6-8], GIs have been found to be compositionally different from their host with regard to codon usage and GC content as well as dinucleotide composition. Both chromosomes of V. vulnificus CMCP6 contain a large number of GIs as identified by Garcia-Vallve and co-workers. In their Horizontal Gene Transfer Database (HGT-DB) 10 and 11 large (>10 kbp) putatively horizontally acquired gene clusters are described for V. vulnificus CMCP6 chromosome I and chromosome II, respectively. These putative GIs have been identified parametrically using the GC-content, the codon usage and the amino acid usage [9].
In this study, we assessed the relatedness and acquisition account of GIs present on chromosomes I and II of V. vulnificus CMCP6, using the genome signature as a measure of similarity between these islands. For this analysis we modified our previously described application δρ-web, which allows dinucleotide composition dissimilarity comparisons between an input sequence and a representative genome sequence [10,11]. The newly developed algorithm, Compare_Islands, allows comparisons between the genome signatures of GIs with each other and that of a selectable genome sequence, and enables a sequence composition similarity grouping, in which the δ* value variations of the different fragments are adjusted for sequence length.
Results
Developing the algorithm Compare_ Islands
A web based application calculating the genomic dissimilarity values between diverse input sequences is offered at our website also featuring δρ-web ([10],. Genomic islands (with a sequence length of 10 kbp and up [12]) may be compared in this application against other large input sequences. The output comprises a matrix with the number of input sequences and the genome sequence with which the user compares the GIs (in our case V. vulnificus chromosome I). The δ* values between these islands are subsequently adjusted for size-dependent signature variation. For these settings a hierarchic clustering is carried out in R.
Calculating the genomic dissimilarity between islands on chromosome I of V. vulnificus CMCP6
Initially, we assessed the composition dissimilarities among the 10 chromosome I GIs and the Vibrio chromosome I (table 1, Fig. 1A). Except for VvI-8, all of these GIs display a high genomic dissimilarity with chromosome I of V. vulnificus. Cross-referring these islands with the alternative GI detection tool IslandPath [13] verifies that VvI-8 is not considered anomalous in GC percentage or dinucleotide composition, but is dissimilar in codon usage. Anomalous loci VvI-3, VvI-4, and VvI-5 display a high GC percentage compared to the genomic values, whereas all other GIs have a lower GC percentage (table 1, Fig. 1A).
Table 1 The specifics of all large putative horizontally acquired gene clusters from chromosome I of V. vulnificus CMCP6, including their δ* (against V. vulnificus chromosome I) and GC composition values compared with chromosome I of V. vulnificus CMCP6. Vv5%, Vv10% and Vv10% are non-anomalous genomic fragments used as a reference clade. Ct1 and Ct2 represent Chlamydia trachomatis genomic fragments and are used as outgroups. The three V. vulnificus CMCP6 islands identified by Zhang and Zhang [22] are included in the last column. The presence of the V. vulnificus CMCP6 putative GIs in V. vulnificus YJ016 is verified by Web-ACT analysis [15].
Locus Coordinates Size (bp) Presence in V. vulnificus YJ016 chromosome I (fraction bp similar) GI characteristics δ* (× 1000) genomic fragments with a lower δ* GC% genomic fragments with a lower GC% Islands identified by [22].
VvI-1 6115–17532 11417 Partially present (3': 2325/11417) Preceeded by a transposase (VV10005) 59.4 89.6% 41.3% 5.57%
VvI-2 355728–393737 38010 Partially present (3': 2343/38010) Bordered by phage integrase (VV10372) 87.0 100% 37.8% 1.16% VVGI-2
VvI-3 1094281–1109572 15292 Entirely present - 42.7 81.3% 49.7% 95.3%
inter 1109572–1122005 12433 Entirely present - 48.4 82.9% 49.6% 95.1%
VvI-4 1122005–1138423 16419 Entirely present - 40.4 80.9% 50.7% 99.5%
VvI-5 1749663–1764864 15202 Entirely present - 46.9 85.6% 50.5% 98.6%
VvI-6 2017768–2042744 24976 Partially present(3': 50/24976) Putative integrase (VV12048) 74.4 98.5% 40.3% 2.29%
VvI-7 2437730–2603335 165606 Highly dispersed Superintegron integrase (VV12401), plasmid stabilisation protein encoding genes (VV12410)) 57.8 100% 41.1% 5.26% VVGI-1
VvI-8 2649661–2664017 14357 Entirely present - 27.0 41.7% 42.8% 8.33%
VvI-9 3033569–3043967 10399 Largely present (3': 6834/10399) Preceded by a transposase (VV12969) 51.5 82.5% 42.1% 7.94%
VvI-10 3260213–3279905 19692 Largely absent (few limited blocks of identity) Putative transposase (VV13182) 64.6 95.8% 39.8% 1.81% VVG1-3
VvI-7a 2437730–2519809 82080 Similar as VvI-7 See VVI-7 62.4 100% 40.8% 2.56% Similar as VvI-7
VvI-7b 2519810–2603335 83526 Similar as VvI-7 See VVI-7 53.1 100% 41.4% 2.56% Similar as VvI-7
VvI-5% 1155000–1170000 15001 Entirely present - 14.0 6.0% 47.3% 59.6%
VvI-10% 570000–585000 15001 Entirely present - 17.2 10.6% 46.4% 54.1%
VvI-25% 270000–285000 15001 Entirely present - 21.2 24.3% 46.5% 38.5%
Ct1 270000–285000# 15001 NR& NR 167 NR 40% NR
Ct2 305000–320000 15001 NR NR 169 NR 39% NR
ˆ) Coordinates from the V. vulnificus YJ016 chromosome I (note; the intergenic space beteen VVI0525-VV10526 in V. vulnificus YJ016 chromosome 1 was not included)
#) Coordinates for the C. trachomatis sequences are related to their respective genome sequence (Accession number AE001273).
&) NR denotes not relevant
Figure 1 Overview of the two V. vulnificus chromosomes. Schematic representation of the δ* values and GC content of large putative horizontally transferred gene clusters in A) chromosome I and B) chromosome II of V. vulnificus CMCP6 using a window size of 5 kbp (x-axis represents chromosome position). Red depicts the low GC content GIs, while blue depicts the high GC content GIs. In green, a large ribosomal protein gene cluster is depicted (Rib). The horizontal dashed red line represents the average δ* value and GC percentage, respectively.
As the imprint of the global signature is locally pervasive on the scale ranging from 50 kbp down to 125 bp [14], compensating for the genomic dissimilarity variation allows us to adjust the genomic dissimilarity for different variations with sequence length among the 10 GIs (for unadjusted δ* values between the GIs from V. vulnificus CMCP6 chromosome I see additional file 1). A hierarchic clustering analysis was carried out with normalised δ* values to assess the compositional relatedness of the GIs (Fig. 2). As a reference clade of compositionally similar fragments, three 15 kbp fragments of regions outside the genomic islands of chromosome I with δ* values lower than that of 5%, 10% and 25% of all chromosomal fragments of 15 kbp, respectively, were included in this analysis to indicate clade cut-off values (see material and methods). In addition, two 15 kbp fragments (Ct1 and Ct2) as well as the complete genome sequence of Chlamydia trachomatis were included as out-groups (table 1).
Figure 2 Hierarchic clustering with complete linkage of the V. vulnificus GIs (as described in table 1) based on the genome signature. Three non-anomalous genomic fragments (indicated with Vv5%, Vv10% and Vv25%) represent the conservative V. vulnificus (VvI) genomic variability, and this clade forms the cut-off value for the different clades (with the red dotted line; clades are indicated with black boxes). The Chlamydia clade consists of two genomic fragments (Ct1 and Ct2) and the genome sequence of C. trachomatis. VvII represents V. vulnificus chromosome II.
The hierarchical clustering analysis showed that the two C. trachomatis fragments grouped together with the C. trachomatis genome sequence, and apart from all other fragments, as expected. VvI-8, with the lowest genomic dissimilarity compared to the genome sequence, grouped together with the 3 chromosomal fragments VvI-5%, VvI-10% and VvI-25%, and the chromosome I of V. vulnificus (VvI).
Using the branching point of the three non-anomalous reference fragments with the complete chromosome as a cut-off limit, two distinct clades are distinguished, formed by VvI-1, VvI-6, VvI-9 and VvI-10 (clade I) and by VvI-3 and VvI-4 (clade II), respectively. The remaining VvI-2, VvI-5 and VvI-7 are singleton GIs. The normalised δ* between VvI-3 and VvI-4 is very low, and together with the sequence proximity in the genome we suggest that these regions might actually be part of one larger anomalous gene cluster. Supportive to this notion is the very low composition dissimilarities between the region between these islands (labelled "VvI-inter" in the tables and figures) and GIs VvI-3 and VvI-4, all three fragments of comparable length (table 1). Hence, the three fragments group together in the hierarchic cluster analysis (Fig. 2).
VvI-1 groups together with VvI-6, VvI-9 and VvI-10. However, VvI-10 and VvI-1 are located at start and the end of the annotation of chromosome I. As this chromosome is circular, these islands are in fact adjacent, which can also be seen on the graphical output of IslandPath [13] at . The low δ* scores between these two regions suggest a similar dinucleotide composition and therefore these regions actually form one island.
We tested compositional similarity consistency by splitting the superintegron VvI-7 in two parts (VvI-7a and VvI-7b), after which the clustering analysis is repeated. The topology of the tree remains intact, and the two superintegron parts are clustered together below our threshold (see additional file 1).
Comparison of chromosome I of V. vulnificus CMCP6 with that of YJ016 by Web-ACT [15] showed that the reference sequences (Vv5%, Vv10% and Vv25%) of chromosome I of CMCP6 as well as the compositionally non-anomalous GI VvI-8 are all present in YJ016. In addition, all GIs comprising clade II are also present in YJ016 (table 1). In contrast, all clade I GIs (VvI-1, VvI-6, VvI-10 and (to a lesser extent) VvI-9) as well as VvI-2 and VvI-7 GIs are (largely) absent in YJ016. These results indicate that in contrast to V. vulnificus YJ016, V. vulnificus CMCP6 gained GIs belonging to one cluster, consistent with the notion of a single acquisition event of these GIs, or exclusive exposition to a specific donor.
Calculating the genomic dissimilarity between GIs from both chromosomes of V. vulnificus CMCP6
Interestingly, the cluster analysis in figure 2 shows deep branching between chromosome I (VvI) and chromosome II (VvII), indicating substantial dissimilarity between these chromosomes. To assess the relationship between the GIs on chromosome I and on chromosome II we next included 11 putatively horizontally acquired gene clusters with length >10 kbp of chromosome II, as identified by the HGT-DB [9] in the analysis. First, the composition dissimilarities between the 11 chromosome II GIs and the Vibrio chromosome II was assessed. All GIs from chromosome II are anomalous in δ*, GC content or both (table 2, Fig. 1B).
Table 2 The specifics of all large putative horizontally acquired gene clusters from chromosome II of V. vulnificus CMCP6, including their δ* (against V. vulnificus chromosome I) and GC composition values compared with chromosome II of V. vulnificus CMCP6. Vv5%, Vv10% and Vv10% are non-anomalous genomic fragments used as a reference clade. Ct1 and Ct2 represent Chlamydia trachomatis genomic fragments and are used as outgroups. Zhang and Zhang [22] did not test V. vulnificus CMCP6 chromosome II.
Locus Coordinates Size (bp) Present in V. vulnificus YJ016 chromosome II (fraction bp similar) GI characteristics δ* (× 1000) Genomic fragments with a lower δ* % GC% Genomic fragments with a lower GC%
VvII-a 89575–104013 14438 Entirely present tRNA synthetase 35.5 74.8% 41.8% 1.6%
VvII-b 302441–313330 10889 Entirely present - 40.8 75.1% 42.5% 4.1%
VvII-c 452124–462927 10803 Largely present (7512/10803) Transposase (VV20421) 56.9 94.1% 44.5% 11.8%
VvII-d 541308–554178 12870 Entirely present - 61.7 96.5% 52.6% 100%
VvII-e 715669–749860 34191 Entirely absent Transposases (VV20693 and VV20695) 71.6 100% 42.9% 1.9%
VvII-f 1064764–1077776 13012 Entirely present - 54.6 93.6% 51.9% 100%
VvII-g 1083544–1106468 22924 Entirely present - 38.1 87.5% 51.7% 100%
VvII-h 1227530–1239793 12263 Entirely present - 44.8 82.7% 50.6% 96.0%
VvII-i 1420351–1433728 13377 Entirely present - 47.6 90.5% 50.8% 96.4%
VvII-j 1446375–1462593 16218 Entirely present - 60.2 95.6% 51.8% 100%
VvII-k 1724928–1739885 14957 Entirely present - 62.0 95.1% 42.7% 4.9%
VvII-5% 105000–120000 15001 Entirely present - 11.9 4.9% 46.1% 27.9%
VvII-10% 495000–510000 15001 Entirely present - 15.1 10.7% 47.2% 52.5%
VvII-25% 960000–975000 15001 Entirely present - 18.8 19.7% 48.1% 66.4%
Next, clustering analysis of the GIs of chromosome I and chromosome II was performed, in which the islands that were found to be compositionally very similar in figure 2 and in close proximity (VvI-1 and VvI-10, and VvI-3 and Vv-4), were taken as single entries, designated VvI-101 and VvI-3inter4, respectively. Addition of the GIs of chromosome II did not alter the overall topology of the clustering of the GIs of chromosome I (Fig. 3). As expected, the reference sequences group with the chromosomes from which they have been taken, while the Chlamydia chromosome and chromosomal fragments remain an outgroup. Cluster analysis yields 11 different branches above our conservative threshold for the 19 distinct GIs and the two chromosomes. We find that two GIs (VVII-a and VVII-b) from chromosome II are clustered with chromosome I, whereas one chromosome II GI (VvII-c) clusters with clade I and four chromosome II GIs (VvII-f, VvII-g, VvII-h and VvII-i) with clade II. In addition, two GIs (VvII-e and VvII-k) from the second chromosome are singletons, which branch just above our conservative threshold. Finally, the remaining two GIs (VvII-d and VvII-j) of chromosome II form a new clade III (Fig. 3).
Figure 3 Hierarchic clustering with complete linkage of the V. vulnificus GIs from both chromosomes (as described in tables 1 and 2) based on the genome signature. For both chromosomes three non-anomalous genomic fragments are included, which represent the conservative V. vulnificus (VvI and VvII) genomic variability. VvI and VvII represent V. vulnificus chromosome I and II, respectively. VvI-3inter4 and VvI-101 represent the concatenated islands of VvI-3, VvI-inter and VvI-4 and of VvI10 and VvI-1 respectively.
Discussion
Dinucleotide composition comparisons between different GIs may identify loci potentially originating from a compositionally similar donor. Identifying potential donors of acquired sequences facilitates the study of gene flow in the biosphere and the identification of the acquisition account may help understand how horizontal gene transfer influences genome evolution. We developed the algorithm, Compare_Islands, allowing comparisons between the genome signatures of GIs with each other and that of a selectable genome sequence, and enables a sequence composition similarity grouping. Robustness of the clustering methods was demonstrated by the clustering of two chromosomal fragments of C. trachomatis with its complete genome. In addition, normal chromosomal I and II fragments group with the complete chromosome I and chromosome II, respectively.
In the present study of V. vulnificus CMCP6, we found that some previously identified GIs are compositionally similar to each other, suggesting that they were derived from one donor or (compositionally) similar donors. For three clades (clade I clade II and clade III, Fig. 3), this implies either multiple transfer events from one donor or a single acquisition event followed by dispersion of the acquired fragment into multiple regions of the host genome thereafter. It should be noted however that a "clade" of two or more GIs does not necessarily imply evolutionary related donors of these GIs, as unrelated but compositionally similar donor types cannot be excluded.
There is some uncertainty as to what dissimilarity levels would be expected when comparing two islands, assuming that they come from donors with a similar genomic signature (ρ*), because the statistical fluctuations of ρ* can differ between candidate donors. In any event it is clear that some islands are substantially more similar to each other than they are to the host genome (e.g. VvI-3 and VvI-4) and could originate from closely related donors, while others (such as VvI-3 and VvI-6) are too different from each other to support the hypothesis of a recent common origin.
In this study we calculated the genomic dissimilarity scores of previously annotated putative GIs against V. vulnificus chromosome I, II and each other, whereas the chromosome contains many (more) acquired sequences with different signatures. Dissimilarity scores between the GIs and the genome would be more pronounced if the genome was purged of the acquired sequences. More pronounced dissimilarity scores would result in a more distinct cladification in the hierarchic clustering.
In addition, it is known that the discriminating ability of oligonucleotide composition comparisons is increased when longer motifs are used [16,17]. Therefore, if a higher resolution is considered necessary in order to compare different GIs, tetranucleotide or even longer oligonucleotide composition values may be of help. However, as the pervasive properties over large sequences has not been assessed per se for larger oligonucleotide motifs, the genome signature remains the most appropriate parameter for this sort of compositional analyses.
Of the V. vulnificus CMCP6 GIs, we propose that VvI-8 has not been horizontally transferred from a compositionally different donor, based on a low genomic dissimilarity with its respective host chromosome, a similar GC content compared to the chromosomal value and the absence of mobility elements (such as transposases or insertion elements) in or around this locus. Alternatively, VvI-8 might be acquired from a compositionally similar donor (e.g. a related Vibrio species) or it may have been acquired an evolutionary long time ago, resulting in a highly ameliorated fragment [18]. While high δ* values between GIs and the host genome sequence are indicative for acquisition from non-related donors via horizontal transfer, low δ* values cannot exclude recent acquisition events from compositionally similar donors, such as lateral gene transfer between related species [19]. It is of importance to note that parametric analyses can only indicate potential acquired regions by compositionally discordance. More elaborate strain analyses should subsequently provide further evidence of actual acquisition events, as has been done recently in V. cholerae and V. Vulnificus YJ016 with regard to GIs [20].
Various parameters have been described, such as the codon usage and the amino acid bias, that enable the identification of anomalous DNA in sequenced genomes (see for an extensive assessment [21]). Although improvements have been made in increasing the resolution obtained by individual parameters [14,22], a single parameter might not find all anomalous regions, and a combination of approaches obviously is preferred, as was already previously suggested [1]. An advantage of genome signature analyses is their applicability to identify anomalous DNA regions containing large stretches of noncoding DNA or small putative genes. In contrast, codon usage disregards the information in non-coding sequences and may not be feasible for very small open reading frames (<300 bp) such as ORFans [9].
From the GIs in V. vulnificus CMCP6, we propose that those in clade I, comprising the compositionally similar VvI-101, VvI-9 and VvII-c were acquired from one donor-species. This may either have been a single acquisition event followed by intra- and interchromosomal dispersal, or a series of acquisition events. VvI-10 and VvI-1, previously annotated as separate GIs, may be considered one GI for their actual proximity and compositional similarity (making a total of 19 GIs in V. vulnificus CMCP6). This emphasizes that linear analyses of circular genomes should be considered with care [22].
Similar to clade I, the GIs forming clade II, VvI-3 VvI-4, VvI-5, VvII-f, VvII-g, VvII-h and VvII-i may have been acquired successively from a compositionally similar donor or may have been dispersed upon a single acquisition event. Our results indicate that VvI-3 and VvI-4 are in fact part of one anomalous gene cluster, as the inter-island sequence displays a low dissimilarity with both VvI-3 and VvI-4, and a similarly ordered island is present in the related V. vulnificus YJ016 chromosome I. This single island has most likely been acquired in a single step, as recombination adjacent to islands originating from HGT events is considered unlikely [23,24].
Previously, the results of Chen and co-workers indicated that interchromosomal exchange had taken place between the two chromosomes in the various Vibrio genome sequences [25], and in Vibrio cholerae it was suggested that the second replicon itself may have been acquired horizontally [26]. Our results of a separate clustering of the chromosomes, as well as the clustering of GIs located on chromosome II with chromosome I support these findings. Figure 3 shows that the Vibrio chromosome I clade also contains GIs of chromosome II and it is appealing to speculate that VvII-a and VvII-b originally come from from chromosome I. In contrast, the Vibrio chromosome II clade does not contain any GIs of chromosome I or chromosome II, which may suggest that interchromosomal transfer of large anomalous gene clusters in V. vulnificus CMCP6 was unidirectional.
Concluding, our application Compare_Islands enables genome composition analyses with selectable window sizes and compositional comparisons between large sequences such as genomic islands. This allows an appraisal of the acquisition account of the large number of available prokaryotic genomes. In the case of V. vulnificus CMCP6, we propose a maximum number of 10 compositionally different donors for 19 distinct GIs. These results suggests that V. vulnificus accepted DNA from (compositionally) different sources, from some sources it accepted more DNA than from others, and a unidirectional flux of GIs from chromosome I to chromosome II is proposed.
Methods
The strategy is based on the dinucleotide relative abundance values or genome signature (ρ*XY). As published previously by Karlin and Burge, each genome has its own typical dinucleotide relative abundance values, which are conserved between related species, as they are thought to result from the DNA repair and replication machinery [27]. Although the genome signature is found to be constant in 50 kbp windows [8], smaller windows can be used to identify anomalous sequences [28]. This is done by calculating the average dinucleotide relative abundance difference in a size dependent manner between the input sequence and a (closely related) representative genome sequence. This approach has been described previously and turned into a web application [10].
The comparisons between different large input sequences, such as GIs from the same strain, can be performed similarly. It may identify islands with similar genome signature values, indicating a compositionally similar donor or (dispersion effects of) the same acquisition event. On the other hand it may identify GIs with such different composition values that a compositionally similar donor is unlikely. The theory behind this approach is based on the results from Jernigan and Baran, which state that dinucleotide composition differences between sequences of different sizes can be calculated as the imprint of the global signature is locally pervasive on all scales [14]. Jernigan and Baran suggested, as a first approximation, that the dissimilarity between a genomic sequence and its host genome be corrected for the length of the sequence: log(delta*_norm) = log(delta*) -0.5 log(length). Empirical studies by Jernigan and Baran showed that the coefficient to log(length) for most organisms vary between -0.5 and -0.35. However, since the (hypothetical) donor organism is unknown and because the signature of the genomic islands could easily show a different statistical behaviour from that of genomic DNA, we will assume a coefficient of -0.5. Therefore, the normalized δ* is the raw δ* divided by the square root of the product of the inverted lengths of the two islands.
In order to set a conservative level of relatedness, we include 3 chromosomal fragments of 15 kbp with low δ* values compared to the complete chromosome; i.e. 3 different fragments with δ* values lower than that of 5%, 10% and 25% of all chromosomal fragments of 15 kbp, respectively. The highest branching point in the clustering of these three fragments compared to the genome sequence is considered a conservative cut-off value for a clade, as this clade indicates relatively related fragments. Hierarchic clustering is carried out in R [29].
The Compare_Islands application is available at and includes user guidelines [11].
Sequences
Ten and 11 large putatively horizontally acquired gene clusters with length between 10 kbp and 166 kbp from V. vulnificus CMCP6 chromosome I and chromosome II, respectively (as identified by Garcia-Valve and colleagues and presented in the Horizontal Gene Transfer Database (HGT-DB [9])) were obtained via the Position Search/Segment Retrieval tool [30] using the coordinates from the HGT-DB (for chromosome I, see table 1, for chromosome II see table 2). Acquired sequences shorter than 10 kbp were ignored in this analysis as GIs are described to vary between 10–200 kbp [7]. Three of these putative GIs on chromosome I correspond to three previously identified islands by Zhang and Zhang [22]; for example, the large VvI-7 (166 kbp, table 1), is similar to a large super integron identified in the related strain V. vulnificus YJ016 [25], albeit highly dispersed. The 10 GIs of chromosome I were numbered VvI-1 to VvI-10 according to their position in the annotation of the chromosome, while the 11 GIs of chromosome II were assigned by VvII-a to VvII-k.
Authors' contributions
MWJvP, AB and AvdE devised the study and wrote the article, ACML and AHCvK set up the computational algorithms and HHT supplied the statistical background and calculations.
Supplementary Material
Additional File 1
This table shows the genomic dissimilarity (δ*, × 1000) values between all GIs of chromosome I of V. vulnificus CMCP6. In green, all low δ* values are indicated, which are clustered together in figures 1 and 2.
Click here for file
Acknowledgements
We would like to thank the two anonymous reviewers of this paper for their helpful suggestions.
==== Refs
Koonin EV Makarova KS Aravind L Horizontal gene transfer in prokaryotes: quantification and classification Annu Rev Microbiol 2001 55 709 742 11544372 10.1146/annurev.micro.55.1.709
Nakamura Y Itoh T Matsuda H Gojobori T Biased biological functions of horizontally transferred genes in prokaryotic genomes Nat Genet 2004 36 760 766 15208628 10.1038/ng1381
Mantri Y Williams KP Islander: a database of integrative islands in prokaryotic genomes, the associated integrases and their DNA site specificities Nucleic Acids Res 2004 32 D55 58 14681358 10.1093/nar/gkh059
Dobrindt U Hochhut B Hentschel U Hacker J Genomic islands in pathogenic and environmental microorganisms Nat Rev Microbiol 2004 2 414 424 15100694 10.1038/nrmicro884
Jain R Rivera MC Lake JA Horizontal gene transfer among genomes: the complexity hypothesis Proc Natl Acad Sci USA 1999 96 3801 3806 10097118 10.1073/pnas.96.7.3801
Blum G Ott M Lischewski A Ritter A Imrich H Tschape H Hacker J Excision of large DNA regions termed pathogenicity islands from tRNA-specific loci in the chromosome of an Escherichia coli wild-type pathogen Infect Immun 1994 62 606 614 7507897
Hacker J Kaper JB Pathogenicity islands and the evolution of microbes Annu Rev Microbiol 2000 54 641 679 11018140 10.1146/annurev.micro.54.1.641
Karlin S Detecting anomalous gene clusters and pathogenicity islands in diverse bacterial genomes Trends Microbiol 2001 9 335 343 11435108 10.1016/S0966-842X(01)02079-0
Garcia-Vallve S Guzman E Montero MA Romeu A HGT-DB: a database of putative horizontally transferred genes in prokaryotic complete genomes Nucleic Acids Res 2003 31 187 189 12519978 10.1093/nar/gkg004
van Passel MW Luyf AC van Kampen AH Bart A van der Ende A {delta}{rho}-Web, an online tool to assess composition similarity of individual nucleic acid sequences Bioinformatics 2005 21 3053 3055 15855249 10.1093/bioinformatics/bti460
The delta-rho-web and Compare_Islands tools
Hentschel U Hacker J Pathogenicity islands: the tip of the iceberg Microbes Infect 2001 3 545 548 11418328 10.1016/S1286-4579(01)01410-1
Hsiao W Wan I Jones SJ Brinkman FS IslandPath: aiding detection of genomic islands in prokaryotes Bioinformatics 2003 19 418 420 12584130 10.1093/bioinformatics/btg004
Jernigan RW Baran RH Pervasive properties of the genomic signature BMC Genomics 2002 3 23 12171605 10.1186/1471-2164-3-23
Abbott JC Aanensen DM Rutherford K Butcher S Spratt BG WebACT – an online companion for the Artemis Comparison Tool Bioinformatics 2005 16076890
Sandberg R Winberg G Branden CI Kaske A Ernberg I Coster J Capturing whole-genome characteristics in short sequences using a naive Bayesian classifier Genome Res 2001 11 1404 1409 11483581 10.1101/gr.186401
Pride DT Meinersmann RJ Wassenaar TM Blaser MJ Evolutionary implications of microbial genome tetranucleotide frequency biases Genome Res 2003 13 145 158 12566393 10.1101/gr.335003
Lawrence JG Ochman H Amelioration of bacterial genomes: rates of change and exchange J Mol Evol 1997 44 383 397 9089078
Linz B Schenker M Zhu P Achtman M Frequent interspecific genetic exchange between commensal Neisseriae and Neisseria meningitidis Mol Microbiol 2000 36 1049 1058 10844690 10.1046/j.1365-2958.2000.01932.x
O'Shea YA Finnan S Reen FJ Morrissey JP O'Gara F Boyd EF The Vibrio seventh pandemic island-II is a 26.9 kb genomic island present in Vibrio cholerae El Tor and O139 serogroup isolates that shows homology to a 43.4 kb genomic island in V. vulnificus Microbiology 2004 150 4053 4063 15583158 10.1099/mic.0.27172-0
Kahler CM Blum E Miller YK Ryan D Popovic T Stephens DS exl, an exchangeable genetic island in Neisseria meningitidis Infect Immun 2001 69 1687 1696 11179344 10.1128/IAI.69.3.1687-1696.2001
Zhang R Zhang CT A systematic method to identify genomic islands and its applications in analyzing the genomes of Corynebacterium glutamicum and Vibrio vulnificus CMCP6 chromosome I Bioinformatics 2004 20 612 622 15033867 10.1093/bioinformatics/btg453
Lawrence JG Gene transfer in bacteria: speciation without species? Theor Popul Biol 2002 61 449 460 12167364 10.1006/tpbi.2002.1587
Vetsigian K Goldenfeld N From The Cover: Global divergence of microbial genome sequences mediated by propagating fronts Proc Natl Acad Sci USA 2005 102 7332 7337 15878987 10.1073/pnas.0502757102
Chen CY Wu KM Chang YC Chang CH Tsai HC Liao TL Liu YM Chen HJ Shen AB Li JC Su TL Shao CP Lee CT Hor LI Tsai SF Comparative genome analysis of Vibrio vulnificus, a marine pathogen Genome Res 2003 13 2577 2587 14656965 10.1101/gr.1295503
Heidelberg JF Eisen JA Nelson WC Clayton RA Gwinn ML Dodson RJ Haft DH Hickey EK Peterson JD Umayam L Gill SR Nelson KE Read TD Tettelin H Richardson D Ermolaeva MD Vamathevan J Bass S Qin H Dragoi I Sellers P McDonald L Utterback T Fleishmann RD Nierman WC White O DNA sequence of both chromosomes of the cholera pathogen Vibrio cholerae Nature 2000 406 477 483 10952301 10.1038/35020000
Karlin S Burge C Dinucleotide relative abundance extremes: a genomic signature Trends Genet 1995 11 283 290 7482779 10.1016/S0168-9525(00)89076-9
van Passel MW Bart A Waaijer RJ Luyf AC van Kampen AH van der Ende A An in vitro strategy for the selective isolation of anomalous DNA from prokaryotic genomes Nucleic Acids Res 2004 32 e114 15304543 10.1093/nar/gnh115
The R project for statistical computing
The Institute for Genome Research
|
16297239
|
PMC1310630
|
CC BY
|
2021-01-04 16:32:47
|
no
|
BMC Genomics. 2005 Nov 18; 6:163
|
utf-8
|
BMC Genomics
| 2,005 |
10.1186/1471-2164-6-163
|
oa_comm
|
==== Front
BMC Evol BiolBMC Evolutionary Biology1471-2148BioMed Central London 1471-2148-5-671629318710.1186/1471-2148-5-67Research ArticleThe Enhancer of split and Achaete-Scute complexes of Drosophilids derived from simple ur-complexes preserved in mosquito and honeybee Schlatter Rebekka [email protected] Dieter [email protected] Universität Hohenheim, Institut für Genetik, Garbenstr. 30, 70599 Stuttgart, GERMANY2005 17 11 2005 5 67 67 15 7 2005 17 11 2005 Copyright © 2005 Schlatter and Maier; licensee BioMed Central Ltd.2005Schlatter and Maier; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In Drosophila melanogaster the Enhancer of split-Complex [E(spl)-C] consists of seven highly related genes encoding basic helix-loop-helix (bHLH) repressors and intermingled, four genes that belong to the Bearded (Brd) family. Both gene classes are targets of the Notch signalling pathway. The Achaete-Scute-Complex [AS-C] comprises four genes encoding bHLH activators. The question arose how these complexes evolved with regard to gene number in the evolution of insects concentrating on Diptera and the Hymenoptera Apis mellifera.
Results
In Drosophilids both gene complexes are highly conserved, spanning roughly 40 million years of evolution. However, in species more diverged like Anopheles or Apis we find dramatic differences. Here, the E(spl)-C consists of one bHLH (mβ) and one Brd family member (mα) in a head to head arrangement. Interestingly in Apis but not in Anopheles, there are two more E(spl) bHLH like genes within 250 kb, which may reflect duplication events in the honeybee that occurred independently of that in Diptera. The AS-C may have arisen from a single sc/l'sc like gene which is well conserved in Apis and Anopheles and a second ase like gene that is highly diverged, however, located within 50 kb.
Conclusion
E(spl)-C and AS-C presumably evolved by gene duplication to the nowadays complex composition in Drosophilids in order to govern the accurate expression patterns typical for these highly evolved insects. The ancestral ur-complexes, however, consisted most likely of just two genes: E(spl)-C contains one bHLH member of mβ type and one Brd family member of mα type and AS-C contains one sc/l'sc and a highly diverged ase like gene.
==== Body
Background
The Notch pathway is one of the best studied cell to cell communication systems in the animal kingdom. It is highly conserved and used from worm to man. This pathway is needed whenever cell decisions are influenced by cell-cell communication, and also during proliferation or pathway crosstalk [1]. Using Drosophila melanogaster as a model system, the Notch pathway was intensely studied over many years. The best defined process governed by Notch is called "lateral inhibition": cells of a given fate are singled out from an equivalence group of the same fate, whereas the differentiation of the other cells is suppressed by the Notch signal. This happens for example during neurogenesis, where neuroblasts are selected from proneural clusters; they keep neural fate, whereas the surrounding cells eventually differentiate as epidermoblasts. The name giving transmembrane Notch-receptor interacts physically with the extracellular domain of the transmembrane ligands Delta or Serrate of the signalling cell. After this activation, the intracellular domain of Notch is cleaved and travels into the nucleus where it transcriptionally activates together with Suppressor of Hairless [Su(H)] genes of the Enhancer of split complex [E(spl)-C]. E(spl) gene products in turn repress the activity of proneural genes encoded for example by the Achaete-Scute-Complex [AS-C]. As consequence these cells stay undifferentiated to become epidermoblasts later on, whereas the signalling cell enters into the programmed neural cell fate. In the focus of our studies are these two complexes, E(spl)-C and AS-C, since in D. melanogaster they are composed of several genes with complex expression patterns and specific yet partly redundant functions.
Enhancer of split was originally identified by genetic means as enhancer of the duplicated bristle phenotype found in the recessive Notch allele split [2]. In order to identify the responsible gene the Enhancer of split gene region has been cloned. In this region 13 transcription units are located and named m1 to m10 and mα to mδ. It was a surprise that seven of these genes encode structurally related proteins characterized by a basic and a helix-loop-helix domain (bHLH), a further alpha-helix forming 'orange domain' and a stereotypic terminus with the amino acids tryptophane, arginine, proline, tryptophane (WRPW). Later it was shown that this motif serves as binding site for the global co-repressor Groucho (Gro, transcription unit m9/10), which is encoded by a gene localised next to the bHLH gene cluster [3-10]. The bHLH genes m3, m5, m7, m8, mβ, mγ and mδ (see Fig. 1; [8]) are all transcriptional targets of Notch: they encode the effector proteins of the Notch signal at least in the process of lateral inhibition [1,11-13]. Apart from the seven bHLH genes and the neighbouring gro locus, the E(spl)-C comprises five further genes. Four genes mα, m2, m4 and m6 share structural similarity with the Bearded (Brd)-gene family and are themselves transcriptional targets of Notch, whereas m1 is completely unrelated and encodes a putative protease inhibitor [14,15]. Larger deficiencies encompassing several of E(spl) transcripts cause a severe neural hyperplasia, whereas loss of activity of single genes do not, suggesting redundancy of these seven bHLH genes [7,10,16-18]. However, remarkable differences were observed between the respective expression patterns in the embryo as well as in postembryonic tissues, arguing against complete redundancy [8,12,14,15,19-22]. Consistently, a high conservation of the entire complex was observed in the rather distantly related fly species Drosophila hydei [23]. The question, however, remains whether gene number and structure of the complex is conserved during longer terms of evolution. For example, in vertebrates an E(spl)-C like in D. melanogaster does not exist. Here the Notch target genes have been classified as HES (hairy/Enhancer of split) and HER/HESR (hairy/Enhancer of split related) genes, because the D. melanogaster segmentation gene hairy encodes a bHLH protein with orange domain and WRPW motif that is as similar to the vertebrate HES genes as are the E(spl) bHLH genes [24,25]. The vertebrate genes are not clustered in a complex. Apparently, in the course of evolution rearrangements occurred between these Notch target genes.
Figure 1 Conservation of the E(spl)-C in Drosophilids. A) The E(spl)-C is highly conserved in Drosophilids with regard to gene number and transcript orientation (arrows). The size of the complex is also almost the same; in D. pseudoobscura (D. pseu) it is only slightly larger than in D. melanogaster (D. mel). The smallest seems to be the D. virilis (D. vir) complex, however, the virilis sequence was not completed at the time. The best identity at protein level is found between the Gro orthologs (purple) followed by the bHLH proteins (red). Interestingly the proteins of the centrally located bHLH genes Mγ, Mβ and M3 (framed blue) are best conserved. Higher identities are seen between the melanogaster and pseudoobscura orthologs than between the ones of D. melanogaster and D. virilis, with the exception of M7 (blue circle). The worst conserved member of the complex is M1. (Numbers give % identity between the proteins). B) Alignment of the Mγ and (C) M7 orthologs. The bHLH (purple) and orange domains are the best conserved parts of the orthologs. The M7 sequences labelled with the black box are unexpectedly better conserved in D. virilis than in D. pseudoobscura compared with D. melanogaster. Identical residues are marked in blue; red shows highly related and yellow similar residues.
The AS-C in D. melanogaster comprises four genes, achaete (ac), scute (sc), lethal of scute (l'sc) and asense (ase) that all encode transcriptional activators of the bHLH class. They determine proneural fate and are thus required for the development of the central and peripheral nervous system [13,26-29]. These genes have been also studied in vertebrates. In the mouse there are three achaete-scute family members abbreviated ASH for Ac-Sc-Homolog [30]: MASH-1 and XASH-3/CASH-4 are two members that are involved in the development of the nervous system [30]. Both complexes are therefore good candidates to look for the changes that occurred during insect evolution.
In this work, we studied the evolution of the E(spl)-C and AS-C by making use of the recent advances in the genome projects of the Diptera D. pseudoobscura, D. virilis, Anopheles gambiae and the Hymenoptera Apis mellifera. The estimated distances are nearly 30 million years (Myr) separating D. melanogaster and D. pseudoobscura which both belong to the Sophophora subgenus and around 40 Myr separating the Sophophora from the Drosophila subgenus where D. virilis belongs to [31]. The distance between these modern dipterans and the more ancient ones like the Culicidae A. gambiae is estimated 200–250 Myr and that between Diptera and Hymenoptera like the honeybee A. mellifera 250–300 Myr [32]. Our study shows that both gene complexes are highly conserved in Drosophilids with regard to overall size, gene number and structural similarity of the encoded proteins. In contrast, more ancient dipterans like the mosquito and similarly also the honeybee have much simpler gene complex structures: the E(spl)-C consists of just one mβ like bHLH gene and one mα like Brd-type gene. However, in Apis two more bHLH/WRPW coding genes are found within about 200 kb and may reflect an enlargement of the E(spl)-C in this species. The AS-C consists of only one sc/l'sc like member and one further ase-like gene that is, however, widely diverged. These data suggest that the evolution of modern Drosophilids included an enlargement of these complexes, notably a multiplication of the genes in the E(spl)-C that seem to have subsequently specified their roles in Notch signalling pathway. Their strict conservation in Drosophilids argues for a diversification presumably driven by their highly specified expression patterns and regulatory activities.
Results
The Enhancer of split complex in Drosophilids
The Enhancer of split complex [E(spl)-C] consists of 13 transcription units (Fig. 1): seven genes (m3, m5, m7, m8, mβ, mγ, mδ) encode highly related basic helix-loop-helix proteins, four have been grouped to the Brd-family (m2, m4, m6, mα); m1 which encodes a serine protease inhibitor and l(3) groucho (gro) which encodes a co-repressor of the E(spl) bHLH protein family. Although the E(spl) bHLH proteins are partly redundant and, therefore, a loss or addition of genes could be without consequences, the complex is highly conserved in all studied Drosophilids with respect to gene order and number, transcription orientation and overall size. As expected, the evolution rate of the orthologs is different. The best conservation is found between the Gro orthologs (more than 96% identity, Fig. 1), whereas M1 displays the highest evolutionary rate (less than 61% identity between D. melanogaster and D. virilis; Fig. 1).
The bHLH proteins of the E(spl)-C
In the Drosophilids, all E(spl) bHLH genes are without intron, and the proteins contain the typical bHLH and orange domains and end with the WRPW motif. The best conserved bHLH ortholog is Mγ followed by M3, Mβ, M5, Mδ and M7/8 (Fig. 1). However, the evolutionary rate varies quite strongly. Comparing D. melanogaster with D. virilis, the highest identity score is found for Mγ with ~92% (Fig. 1A,B) and the lowest for M8 with ~81%. An even more striking difference in the identity scores is observed when comparing the D. melanogaster and D. pseudoobscura orthologs M3 and M7 (93% versus 73%; Fig. 1A,C). The low degree of conservation of the M7 proteins is rather surprising. It is based on one hand on peculiar size variations: 206 residues in D. pseudoobscura, 186 in D. melanogaster and 197 in D. virilis (Fig. 1C). On the other hand, the amino acid composition of the bHLH and orange domains is much better conserved between D. melanogaster and D. virilis than between D. melanogaster and D. pseudoobscura. This is different from all the other bHLH orthologs: the bHLH domains of M3 and Mβ are identical in all three species and also the other ones are extremely similar. Only one conservative change is detected in the Mγ bHLH domain of D. melanogaster compared with D. virilis, and just two in M5. The bHLH domains of the M8 and Mδ proteins contain also single non-conservative replacements, apart from a few conservative changes. However, D. pseudoobscura M7 shows an unusual high number of changes – six replacements and five conservative changes – within the bHLH domain if compared to the melanogaster ortholog (Fig. 1C).
The Bearded-protein family in the E(spl)-C
In general, the Brd-type proteins evolve faster than the bHLH proteins: M4 and Mα are the best conserved members with ~82% identity between virilis and melanogaster and are thus within the range of the fastest evolving m7/8 bHLH coding genes (Fig. 1). The overall structure of M4 and Mα orthologs is quite similar in all studied Drosophila species. The so-called 'bearded'-domain is completely identical in the Mα orthologs, M4 has only a few gaps. However, the predicted D. virilis m4 gene has an extended open reading frame of novel 132 residues at the 5' end, whereas the remaining 156 residues are conserved. The melanogaster m4 5' region reveals similarity at the DNA level, however, has no open reading frame. Therefore it remains questionable whether the larger open reading frame in D. virilis is indeed translated. The other two Brd-family members, M2 and M6, are much less conserved: Only approximately 68% identity is found between the respective orthologs of D. melanogaster and D. virilis. Despite this little conservation, the typical Brd protein domains can still be recognized. The most prominent is the predicted basic amphipathic α-helix domain in the N-terminal protein region [14].
The m1 gene in the E(spl)-C
The m1 gene encodes a protein that has the signatures of serine protease inhibitors [15]. Despite a low degree of conservation which ranges between 60 and 70% identity between the M1 orthologs (Fig. 1A), the structurally important cysteines residues are conserved in number and spacing [15]. Notably, the orthologous genes of D. pseudoobscura and D. virilis have a significantly longer open reading frame at the 5'end that extends the proteins for approximately 50 residues to 203 in the case of D. virilis. The extended protein parts share ~68% similarities within the first 30 residues between the two orthologs. Furthermore, the first nine residues have only one conservative exchange arguing for its translation in vivo. In melanogaster all three reading frames at the 5' end are blocked by several stop codons, excluding a likewise 5' extension. However, at the DNA level there are identities of 69% to the virilis ortholog and 79% to the pseudoobscura ortholog which could be also interpreted as conserved regulatory sequence.
The E(spl)-C in Anopheles gambiae
Albeit Anopheles belongs to the dipteran flies it does not contain an E(spl)-C that matches that of Drosophilids (Fig. 2A). Only a single transcription unit with respectable conservation that contains one intron was annotated in the genome project (ENSANGG00000017601; see Tab. 1). However, the predicted coding sequence does not end with a WRPW motif as expected for E(spl) bHLH proteins. By searching through the genomic sequence, we propose a different gene structure, where the transcript extends into the intron that maintains an open reading frame and ends with a WRPW motif. In this case, the E(spl) homolog of mosquito would be without intron and shares highest similarity to the Mβ/Mγ pair of D. melanogaster (80.3/75.7% similarity and 69.6/67.4% identity, respectively). Based on the similarity, we propose that it corresponds to D.m.mβ (Fig. 2B) and named it therefore A.g.mβ. Moreover, there is a single Brd-like gene in close proximity (~8 kb) of A.g.mβ (Fig. 2A). This gene encodes a protein most similar to the D. melanogaster Mα protein with almost 60% identity (Fig. 3A); therefore we named it A.g.mα. It shares all features of the Brd-family proteins described earlier [14], including the amphipathic α-helix domain in the N-terminal part (Fig. 3B).
Figure 2 The E(spl)-C in mosquito A. gambiae. (A) The E(spl)-C in the mosquito is composed of two putative genes, A.g.mβ and A.g.mα. Approximately 100 kb away a second bHLH coding gene was detected, however, the analysis predicted a close relationship to Deadpan and Hairy. Since a good fitting hairy ortholog is elsewhere in the mosquito genome, the gene is most likely a dpn ortholog. B) Alignment between A.g.Mβ and D.m.Mβ shows good conservation within the bHLH and the orange domains (marked) as well as the WRPW motif at the C-terminus. Identical residues are marked in blue; red shows highly related and yellow similar residues.
Figure 3 Conservation of the Brd-family member Mα. A) Alignment of the presumptive Mα proteins of D. melanogaster, A. mellifera and A. gambiae. Although the alignment reveals not much identity (blue), the postulated features that typify Brd-proteins are present [14]. Red, highly related and yellow, similar residues. B) Mα contains a typcial amphipathic α-helix with high concentration of lysine (K) and arginine (R) residues on one side of the wheel (red).
Table 1 Identity matrix of E(spl) bHLH proteins in D. melanogaster (% similarity/identity)
Mβ Mδ M8 M7 M5 M3 Ø
Mγ 75/68 68/59 69/57 75/67 72/61 74/64 72/63
Mβ 67/56 70/61 75/68 71/60 80/71 73/64
Mδ 64/50 67/59 65/50 67/53 66/55
M8 67/59 80/73 63/50 69/58
M7 67/61 70/61 70/63
M5 66/57 70/60
M3 70/59
Ø = average similarity/identity (%) of one bHLH member if compared to all others
Approximately 100 kb from the 3' end of A.g.mβ, we detected another sequence that might encode an E(spl) bHLH type protein (Fig. 2A; ENSANGG00000017548). However, the presumptive gene product is more highly related to D. melanogaster Deadpan (Dpn; 57% identity) than to E(spl) bHLH Mβ (52% identity). The conservation extends beyond the amino acid sequence: we find the same intron/exon structure in this Anopheles gene as in the deadpan and hairy genes from Drosophila (see also Fig. 8a). We believe that this Anopheles protein corresponds to Dpn (A.g.Dpn) rather than to Hairy, since it shares little more than 47% identity with D. melanogaster Hairy. The best hit with the D. melanogaster Hairy protein is found on the second chromosome in Anopheles (A.g.h; 72% identity at protein level). Other genes of the Drosophila E(spl)-C were not detected nearby: maybe they are not conserved enough to be discovered like e.g. m6 or they are located at totally different positions in the genome like gro.
The E(spl)-C in Apis mellifera
Like in the mosquito, there is no extended E(spl)-C in the honeybee. In fact, we find a similar structure of one E(spl) bHLH type and one Brd-type gene that share highest homology to Mβ and to Mα, respectively (Fig. 4A). The relative transcription orientation (head to head) is the same, suggesting that mβ and mα represent the ur-complex (compare Figs. 1A, 2A, 4A). However, the situation in honeybee is more complicated in several respects.
Figure 4 The E(spl)-C in Apis mellifera. A) A genomic region spanning about 250 kb (GroupUn.159; numbers are correspondingly) contains 3 presumptive E(spl) bHLH genes and one mα related gene. The gene at position ~250 kb encodes a bHLH protein with best overall identity to D.m.Mβ. This gene is disrupted by an intron inside the bHLH domain (dash and red arrowhead in B). Close by, at position 220 kb a mα related coding region is found. Based on high similarity to D.m.Mα and its close neighbourhood to A.m.mβ, we named it A.m.mα. A second intronless E(spl) like bHLH coding gene is located at position ~90 kb. We name it A.m.mγ since the encoded protein shows best overall identity to the D.m.Mγ. Approximately 50 kb away we detected sequences encoding an E(spl) like bHLH domain (A.m.Mβ') and a long open reading frame ending with a WRPW motif. The Ensembl honeybee database annotated a gene with five introns spanning 12 kb within this region but missed the respective motifs (GENSCAN00000025907). We propose a different gene structure; see Figure C for details. B) Alignment of the three putative honeybee bHLH proteins (A.m.Mβ', A.m.Mβ, A.m.Mγ) with D. melanogaster Mβ and Mγ proteins is shown. Identical residues are marked in blue; red shows highly related and yellow similar residues. Intron positions are marked with a triangle above and a dash in the respective A. mellifera sequence. C) Structure of the 12 kb GENSCAN00000025907 region. Black shows the Ensembl gene annotation, and blue the new ENSAPMG00000016895 annotation. Purple highlights a second exon that encodes part of the bHLH domain, the open reading frame extends into the predicted intron of GENSCAN00000025907 and terminates with WRPW.
For example, the A.m.mα homolog is predicted to consist of five exons that code for a protein with 402 residues. This is considerably larger than the 138 residues of D.m.Mα. Again, we propose a different gene structure based on the analysis of the genomic DNA, where the translation extends into the first postulated intron and ends shortly afterwards. The encoded protein than consists of only 168 amino acids and terminates with residues that are very similar to the Mα Drosophila homolog (M Q V A) (Fig. 3A). Moreover, it shows the typical amphipathic α-helix domain in the N-terminal part (Fig. 3B). The other motifs are also conserved, the extreme C-terminus is, however, only similar (Fig. 3A). This protein shows also high similarity to Drosophila Twin of m4 (Tom) that might slightly exceed that to D.m.Mα depending on the parameters used. As in mosquito, no other Brd-like protein was found in the honeybee database such that a Brd-complex seems non-existent, unlike in Drosophilids [33].
The Apis database annotates a single intron for the A.m.mβ gene that conforms to the GT AG rule. This is in contrast to all the Drosophila E(spl) bHLH genes that are intronless. Moreover, there are several possible start sites and it remains unclear which one is used. In D. melanogaster, there are other bHLH-WRPW encoding genes that contain introns, like deadpan (dpn) or Side. However, the respective protein sequences do not align well with A.m.Mβ, and the Apis genome contains predicted orthologs to these genes at other locations (see below). The A.m.Mβ protein is also highly similar to melanogaster Mγ with 72.3 versus 77.7% similarity and 65 versus 67.5% identity compared to D.m.Mβ (Fig. 4B). These differences are extremely small. In fact, the Apis database proposes this gene as mγ based on the similarity within the bHLH domain. However, we used for our comparison the entire protein sequences and calculated identity overall. Moreover, based on the E(spl)-C structure in Anopheles, we favour the hypothesis that mβ is the ancestral gene.
Within the ~250 kb contig (GroupUn. 159), there are two further stretches that might encode E(spl) type bHLH proteins. One is located about 150 kb apart. This gene contains no introns and the encoded protein shows homology to both Mβ and Mγ. In this case, the similarity seems slightly higher to Mγ than to Mβ (72.2 vs. 69.3% similarity and 66 vs. 62.7% identity). Therefore, we call the gene A.m.mγ (Fig. 4A,B). Both FlyBase and BlastN give a higher score to D. melanogaster Mβ than to Mγ. Presumably, both use similar paradigms based on an alignment of only the best conserved sequences, whereas we used less stringent parameters (see Methods) to allow an alignment of the complete sequences. As to be expected, the two E(spl) honeybee proteins A.m.Mβ and A.m.Mγ are highly related to each other with a similarity of 73% and an identity of 67%. Interestingly, this numbers are very similar to those from a likewise comparison of the D. melanogaster Mβ and Mγ homologs (Table 1).
Another 50 kb further up at position 50 (Fig. 4A), we found a short alignment to the E(spl) bHLH domain. However, there was no predicted gene, nor a start codon, nor a WRPW motif. Nearby at position ~46 kb there is an open reading frame of 133 residues split by one intron that belongs to a predicted database gene consisting of five exons (GENSCAN00000025907; black in Fig. 4C). However, lacking any similarity to known genes of Apis or other species, this gene remained without functional prediction in the Apis database. In agreement, our searches in the FlyBase did not detect any similar sequences. However, we predict an E(spl) bHLH-type protein encoded by this gene region: extension of the open reading frame into the adjacent intron ends in WRPW (Fig. 4C, blue exons). The bHLH encoding sequences (purple in Fig. 4C) are located within the second predicted intron of the putative gene shown in black. There are respective exon/intron boundary consensus sequences to allow for a single transcript that contains the bHLH domain, an orange domain and ends with the WRPW motif (Fig. 4C). This third E(spl) bHLH gene would comprise five exons. In fact, in the update May 2005 of the Ensembl honeybee database, the Ensembl automatic analysis pipeline predicts a very similar protein, however, with a different N-terminus and slightly smaller third and fourth exons. This gene than consists of four exons which would be similar to D. melanogaster hairy and dpn that contain two introns. However, the encoded protein is most similar to the E(spl) protein Mβ, so we call the gene A.m.mβ'. Since Apis hairy and dpn are found elsewhere in the genome (see below), we propose that the E(spl)-C in Apis mellifera consists of the ur-complex plus two further E(spl) bHLH genes most closely related to mγ and mβ. No other genes of the Drosophila E(spl)-C are present in that of the honeybee. We find a highly conserved Groucho ortholog, however, at a completely different position in the genome.
Conservation of other Hairy/E(spl)-like proteins known from Drosophila
In total, 12 genes are known in D. melanogaster to encode Hairy/E(spl)-like proteins, i.e. bHLH proteins that also have the orange domain and a WRPW-type Gro-binding motif (see Table 2). Apart from the seven E(spl) bHLH proteins, these include Hairy, Deadpan, Side, Hey and Her [34]. Moreover, there is similarity to Stich1/Sticky which has a bHLH and an orange domain but not the typical Gro-binding motif [35]. Since the number of E(spl) bHLH genes is not conserved in honeybee and mosquito, it was interesting to ask whether all the other genes were present. We searched the Ensembl database with the respective D. melanogaster protein sequences and found orthologs of all genes except of Her in both species (see Table 2). However, most of the predictions are incomplete. We know from D. melanogaster that these genes contain introns, which complicates the search for potential coding sequences within genomic DNA. Thus, our protein sequence predictions are uncertain. With the sole exception of Dpn, all the proteins are better conserved between Drosophila and Anopheles than between Drosophila and Apis, confirming the evolutionary relationship. The best conserved proteins are Hey and Hairy. The Hey orthologs are 76% identical between Drosophila and Anopheles and 66% between Drosophila and Apis and the Hairy orthologs between 72% and 65%, respectively. Less conservation is found for Side, Dpn and Stich1 (62/57 Side, 57/59 Dpn and 60/57 Stich1; % identity comparing fly with mosquito and honeybee, respectively). All proteins share the bHLH and orange domains. The WRPW motif of Hairy, Dpn and Side as well as the YRPW motif of Hey is present in the orthologs.
Table 2 Gene annotation used by the respective databases
D. melanogaster D. pseudoobscura
(contig 4374 Contig 4847) D. virilis Apis mellifera Anopheles gambiae
D.m. mδ (CG8328) D.p. mδ (178 652-178 089) D.v. mδ - -
D.m. mγ (CG8333) D.p. mγ (175 594-174 953) D.v. mγ A.m. mγ (ENSAPMG0000004887) -
D.m. mβ (CG14548) D.p. mβ (169 335–169 934) D.v. mβ A.m. mβ (ENSAPMG0000004881) A.g. mβ (ENSANGG00000017601)
D.m. mα (CG8337) D.p. mα (163 786-163 358) D.v. mα A.m. mα (GENSCAN0000001764) A.g. mα (SNA00000011401)
D.m. m1 (CG8342) D.p. m1 (158 207–158 734) D.v. m1 - -
D.m. m2 (CG6104) D.p. m2 (157 012–157 407) D.v. m2 - -
D.m. m3 (CG8346) D.p. m3 (151 792-151 136) D.v. m3 - -
D.m. m4 (CG6099) D.p. m4 (149 049–149 516) D.v. m4 - -
D.m. m5 (CG6096) D.p. m5 (142 321–142 881) D.v. m5 - -
D.m. m6 (CG8354) D.p. m6 (137 883-137 656) D.v. m6 - -
D.m. m7 (CG8361) D.p. m7 (134 008-133 391) D.v. m7 - -
D.m. m8 (CG8365) D.p. m8 (130 185-129 628) D.v. m8 - -
D.m. gro (CG8384) D.p.gro (148 473-117 489) D.v. gro - -
A.m. mβ' (ENSAPMG00000016895)
D.m. stich1 (CG17100) not analysed not analysed A.m. stich1 (ENSAPMG00000005857) A.g. stich1 (ENSANGG00000016365)
D.m. side (CG10446) not analysed not analysed A.m. side (ENSAPMG0000000088) A.g. side (ENSANGG00000014329)
D.m. dpn (CG8704) not analysed not analysed A.m. dpn (ENSAPMG00000004551) A.g. dpn (ENSANGG00000017548)
D.m. Hey (CG11194) not analysed not analysed A.m. Hey (ENSAPMG0000000726) A.g. Hey (ENSANGG00000021744)
D.m. Her (CG5927) not analysed not analysed - -
D.m. h (CG6494) not analysed not analysed A.m. h (ENSAPMG00000004545) A.g. h (ENSANGG00000018369)
D.m. ac (CG3796) not analysed D.v. ac - -
D.m. sc (CG3827) not analysed D.v. sc - -
D.m. l'sc (CG3839) not analysed D.v. l'sc A.m. ash (ENSAPMG00000003261) A.g. ash (ENSANGG00000010650(Q95VY6)
D.m. ase (CG3258) not analysed D.v. ase A.m. ase (ENSAPMG00000003265) A.g. ase (ENSANGG00000015341)
D.m. da (CG5102) not analysed not analysed A.m. da (ENSAPMP00000005673) A.g. da (ENSANGEST00000361691/SNAP000000012539)
D.m. Ocho (CG5138) not analysed not analysed - -
D.m. Tom (CG5185) not analysed not analysed - -
D.m. Brd (CG3096) not analysed not analysed - -
(-, not found)
The Achaete-Scute complex in Drosophilids
The Achaete-Scute complex (AS-C) is well conserved in D. virilis: all four genes, achaete (ac), lethal of scute (l'sc), scute (sc) and asense (ase) are found in the same order and orientation on the X-chromosome (Fig. 5A). Like in D. melanogaster, the genes are without introns. All proteins share the typical bHLH motif of the AS-C proteins and this domain reveals the lowest evolutionary rate. However, compared with the bHLH proteins of the E(spl)-C the bHLH proteins of the AS-C evolve faster. The complex can be separated into two clusters that are distinguished by their rate of conservation. On one hand, L'sc and Sc are well conserved with an identity between D. melanogaster and D. virilis of more than 75% and on the other hand Ac and Ase with an identity of less than 69% (Fig. 5A). Note that the highest divergence that was found between these two species in the E(spl)-C was for M8 with still almost 81% identity.
Figure 5 The AS-C of D. virilis. A) The AS-C is highly conserved between D. melanogaster and D. virilis concerning gene number, transcript orientation and overall size. At the protein level (% identity) the best conservation is found between the Sc and L'sc orthologs. However, the conservation rate is lower compared with the E(spl) bHLH proteins (see Fig. 1). The pcl gene, although only 50.4% identical, is found between l'sc and ase. Because of gaps in the genomic virilis sequence, a scheme of the D. melanogaster complex is shown. B) Alignment of the Ase protein orthologs of D. melanogaster and D. virilis. Note the extension of the D.v.Ase protein by repetitive amino acid stretches composed of poly N, poly Q and poly A. The best conservation is found within the bHLH domain (purple). Identical residues are marked in blue; red shows highly related and yellow similar residues.
Of the four AS-C gene members in D. melanogaster, ase stands out because it is much larger than the other three. In D. virilis, the size increase is even more striking: D.v.Ase is predicted to comprise 619 residues, whereas D.m.Ase is only 486 residues in length (Fig. 5B). This extension of more than 20% additional residues is caused by multiple insertions of repetitive sequences that code for poly-glutamine (Q), poly-alanine (A) and poly-asparagine (N) stretches (Fig. 5B). Like in D. melanogaster the unrelated gene pepsinogen-like (pcl) is located between l'sc and ase (Fig. 5A).
The AS-C in Anopheles gambiae
In the mosquito, we find only two potential achaete-scute like genes that are in close neighbourhood of less than 30 kb. Interestingly, like in Drosophila they are located on the X-chromosome. One of them encodes a protein that is very similar to L'sc and Sc proteins not only within the bHLH domain but also at the C-terminus (Fig. 6A). This is not unexpected because the C-terminus is involved in transcriptional activation as well binding of E(spl) bHLH proteins [36]. The BestFit program gives a higher score to L'sc (64% identity) than to Sc (57% identity). However, closer inspection reveals that some protein regions are more similar to D.m.Sc and others more to D.m.L'sc (Fig. 6A) suggesting common ancestry for this gene pair. The Anopheles data base predicts an intron, which however retains the open reading frame. In fact, Wülbeck and Simpson [37] cloned and sequenced the respective A.g.ash cDNA and showed that it is intronless. We detected three conservative amino acid exchanges between the published A.g.Ash protein sequence and that obtained from translating the database genomic DNA, namely at position 8 (M-L), position 189 (T-S) and position 311 (Q-H).
Figure 6 The AS-C in A. gambiae. A) Alignment of the achaete-scute homologous protein of A. gambiae (A.g.Ash; [36]) with melanogaster L'sc and Sc proteins. Note the high conservation of the bHLH domain (purple) and the very C-terminus. Comparison over the entire length gives a higher identity score to D.m.L'sc, however, the alignment shows also regions that are more similar to D.m.Sc. B) Alignment of the neighbouring bHLH gene product from A. gambiae with D. melanogaster Ase. The alignment is shown to A.g.Ase/i (database predicted version without intron) and A.g.Ase (second intron translated). The second predicted intron comprises almost 2.5 kb and ends with an exon translated into five residues (CSPTH; in A.g.Ase/i). Translation into this intron leads to A.g.Ase that is similar in size and in its terminus to D.m.Ase. Highest conservation is found in the bHLH domain (purple). Identical residues are marked in blue; red shows highly related and yellow similar residues.
The second presumptive gene has two predicted introns. The derived amino acid sequence shares between 54% and 57% identity with all four D. melanogaster AS-C proteins. By searching the D. melanogaster genome with the predicted protein sequence, the best hit was found to Ase protein (Fig. 6B). Accordingly, the Anopheles database defined it as ase homolog and so we named it A.g.ase. We note, however, that a precise appointment is difficult based on the lack of a significant similarity at the C-terminus which normally allows the distinction between the AS-C members.
The AS-C in honeybee
Like in the mosquito, the honeybee genome encodes only two AS-C like proteins. The two transcription units are the predicted Ensembl genes ENSAPMG00000003261 and ENSAPMG00000003265 (Table 2). They are located in the scaffold group 10.3 and are approximately 40 kb apart from each other. The former encodes a protein that is highly similar to the L'sc/Sc protein pair, so we named it A.m.ash (Fig. 7A). In contrast to the Drosophila AS-C genes, A.m.ash is predicted to contain a single intron which however, retains an open reading frame. Therefore, like in mosquito the encoded protein could be significantly larger. We thus aligned the protein sequences of A.m.Ash with and without translation of the predicted intron with D.m.L'sc (Fig. 7A). Since the open reading frame of the presumptive intron is translated primarily into serine residues, no alignment with the D.m.L'sc protein was possible for this part, supporting the intron prediction. We note that there are two more exon/intron boundary consensus sequences within the predicted intron (arrows in Fig. 7A). If these were used instead of the ones predicted, the intron would be somewhat smaller and A.m.Ash accordingly 25 amino acids larger. As shown in Fig. 7A, the resultant protein would be more similar in size to melanogaster L'sc protein and moreover, share additional similarities in this part of the protein. The Apis database does not provide a start for A.m.Ash, which we could deduce however from the alignment with the D. melanogaster protein.
Figure 7 The AS-C in A. mellifera. A) Comparison of D.m.L'sc with the predicted A.m.Ash protein. Two forms were compared, without intron sequence, A.m.Ash/i, and with translated intron, A.m.Ash. Arrows mark additional splice consensus sites. B) Within 40 kb of A.m.ash, there is a second potential gene encoding a widely diverged bHLH protein. Two different programs were used for gene prediction that gives A.m.Ase/GS (Chris Burge's Genscan program) and A.m.Ase/GW (GeneWise model); both predicted proteins were aligned with D.m.Ase. Decent conservation is only found in the putative bHLH domains (purple). Identical residues are marked in blue; red shows highly related and yellow similar residues.
Figure 8 Comparison of A.m.Mβ' with Dpn, Hairy and Her proteins. A) Comparison of A.m.Mβ' with A.m.Dpn, A.m.H, D.m.Dpn and D.m.H. The A.m.Mβ' protein belongs to the E(spl)-C, however, has several introns (triangle on top, vertical dashes within sequence). Two of the introns are within the bHLH domain at similar position as in D. melanogaster hairy or dpn genes. The bHLH domain is indicated in purple. B) Alignment of D. melanogaster er Hwith A m.Mβ' protein. Her has only one intron, however, at very similar position as the second intron of A.m.Mβ'. Introns are marked with triangles.
Interpretation of the second gene is much more difficult. The annotation by Ensembl automatic pipeline using GeneWise model based on either protein or aligned EST's resulted in a coding sequence within a single exon that lacked both start and stop codons, however, aligned well with AS-C bHLH domains. We thus extended our studies into the surrounding genomic DNA, where we detected one further open reading frame. The deduced amino acid sequence matched well with the N-terminal part of D. melanogaster Ac and Ase proteins. However, the C-terminus did not align convincingly (Fig. 7B/GW). Another gene prediction using Chris Burge's Genscan program [38] gave a transcript of 8 exons spanning over 8 kb of genomic DNA. The translation gave a larger protein again without start methionine that contained the single exon predicted by the GenWise model (Fig. 7B/GS). Again there was very little similarity in the C-terminal part of this and Drosophila AS-C proteins (Fig 7B). Because Genscan could not predict the protein start, we propose a combination of both models with the N-terminus as shown in the A.m.Ase/GW sequence and the C-terminus as in A.m.Ase/GS that might, however, end shorter than shown in Fig. 7B.
The high divergence from the Drosophila AS-C proteins renders precise predictions very difficult. In fact, under standard conditions like FlyBase BlastN only parts of the bHLH domain can be identified. Comparison with the D. melanogaster AS-C proteins gave minimally different scores, with the highest score found with D.m.Ac followed by D.m.Ase, dependent on the parameters. For example, the Apis database finds best scores with melanogaster Sc and L'sc. We named this gene A.m.ase by its similarity notably in the N-terminal part and based on the arrangement of the mosquito AS-C.
Conservation of predicted regulatory elements in E(spl)-C and AS-C
Neurogenesis in Drosophila is subjected to various levels of regulation. As described in the introduction, E(spl) bHLH proteins repress proneural gene activity. Negative regulation is brought about by repression of transcription as well as at the protein level [36,39-42]. A further mode of regulation involves RNA:RNA duplexes [43]. These are formed by small sequence stretches located in the 3' untranslated region (3' UTR) of the mRNAs of proneural AS-C genes and different members of the E(spl)-C. For example, l'sc mRNA contains so-called proneural boxes (PB-box) (GGAAGAC) which bind to the GY-boxes (GUCUUCC) of E(spl) m4 RNA [43]. We searched the genomic sequences adjacent to the coding sequences for respective regulatory elements and found them in D. virilis: there are two PB boxes in D.v.l'sc and GY-boxes in the 3'UTR of the predicted virilis E(spl) genes m3, m4, m5 and mγ just as in D. melanogaster and meanwhile published by Lai et al. [33]. Moreover, it has been shown that E(spl) genes are direct targets of the Notch signal involving the DNA binding protein Su(H) [11,13,15,19]. Accordingly, there are potential Su(H) binding sites (C/TGTGA/GGA) in all D. melanogaster E(spl) genes including m2 and m6 [14,15], which are also present in the respective virilis orthologs [44]. However, the predicted binding sites for proneural bHLH activators (E box: GCAGGTG) [14] are less well conserved during evolution. Whereas the D. virilis m2 and m4 orthologs contain such a regulatory element, we found no sequence fitting the E-box consensus in either the D.v.m6 or D.v.mα sequence.
In mosquito and honeybee, regulatory elements of RNA:RNA duplex-type were not detected. None of the two AS-C genes of either Apis or Anopheles contained PB-box like sequences in the 3' UTR, albeit the highly diverged sequence and gene structure of A.g.ase does not allow to definitely exclude their presence. Since there is no predicted m4 ortholog in Anopheles and Apis, we looked at the 3' end of the mα gene as the single Brd-family member. However, in none of these gene sequences did we find GY type boxes like in Drosophila. We note that some of the predicted gene structures are still incomplete and a search for small sequence stretches is notoriously difficult if one allows for variations. Therefore, there is the formal possibility that we have missed these sites.
Discussion
The Enhancer of split complex
Extensive genome analyses in the recent years revealed that there are not many examples of large gene complexes that are widely conserved. Prominent examples are the HOX (homeobox) complexes, which contain homeotic genes in Drosophila. HOX complexes are well conserved in metazoans despite some variations in gene number. HOX-genes encode regulatory proteins with specific individual functions and mutations affect different aspects of the body plan [45]. Not surprisingly, it is almost only the homeodomain, which serves as sequence-specific DNA binding motif that is conserved amongst different species [46]. In contrast, similarity amongst bHLH proteins encoded by the E(spl)-C extends over the entire length, even within the same species indicating rather recent duplication events. The D. melanogaster proteins M8/M5 and Mβ/M3 are most similar with over 70% identity, whereas Mδ is the most diverged. However, Mδ still shares at least 50% identity with other E(spl) bHLH protein members (see pair wise comparison in the identity matrix of Mδ with M8, M5 and M3; Table 1). More interesting is the analysis of the overall similarity amongst these proteins. Here, any one of the proteins is compared with the other six and the result is averaged. Clearly, Mβ (73/64%, similarity/identity) closely followed by Mγ (72/63%) is most similar to all others, whereas Mδ (66/55%) shows the lowest values (Table 1). One interpretation might be that the different bHLH genes evolved by duplication out of mβ or mγ. Remarkably, these two bHLH proteins besides M3 are the best conserved in the three Drosophila species (Fig. 1A). We would like to postulate that these are the most ancient proteins with the most general function and, therefore, the highest selection pressure. This hypothesis is supported by the finding that mβ has the most general expression pattern from which the others can be derived by a decrease of gene activity [19]. The conspicuous conservation of M3 might hint to an important function during egg development as this gene is expressed also maternally [17,22]. The high degree of conservation of all E(spl) bHLH orthologous proteins in Drosophilids, which is clearly higher than the similarity within this protein family in D. melanogaster, indicates specific and non-redundant roles during development (see also [23]). Some of these functions have been identified in the past [19,39]. It is conceivable that regulatory sequences were not duplicated or evolved more rapidly so that we now find highly dynamic expression patterns of these genes.
The ancestral E(spl)-C is composed of mβ and mα
As outlined above, mβ appears to be the ancestral bHLH gene of the E(spl)-C in Drosophilids based on its great similarity with all the other bHLH proteins. This assumption is strongly supported by the sequence conservation of the E(spl) bHLH proteins in A. gambiae and A. mellifera. The single E(spl) bHLH protein encoded by the mosquito genome has the highest identity to Mβ. The genome of honeybee contains three prospective genes that encode proteins most highly related to E(spl) D.m.Mβ and D.m.Mγ. All three are clustered within a single sequence contig, albeit they span a large segment of about 250 kb, whereas the whole E(spl)-C in D. melanogaster comprises roughly 50 kb. Despite the fact that two of these genes possess introns just within the bHLH domain and at positions close to the ones found in the D. melanogaster genes dpn, hairy or Her (Figs. 4, 8), the amino acid sequence similarity classifies them clearly as E(spl) bHLH proteins. A comparison of Anopheles and Apis proteins reveals, that the presumptive Mβ homologs have highest similarity (83%) and identity (76%), whereas the protein that we classified as A.m.Mγ is just 70% similar and 66% identical to A.g.Mβ.
In Drosophilids, mα is located close to mβ and is transcribed in the opposite direction (head to head; Fig. 1A). This arrangement is likewise found in Anopheles and Apis (Figs. 2A, 4A). Notably, A.m.mα is next to A.m.mβ, whereas the two Apis A.m.mγ and A.m.mβ' genes are much further apart (Fig. 4). We find this arrangement to be very ancient. In the beetle Tribolium, which on the tree of evolution is found even more deeply rooted (~300 Myr to Dipterans [32]), two similar genes coding for Mβ-like proteins (~65% and ~67% identity to D.m.Mβ) are found and one is within ~18 kb to a gene coding for an Mα-like protein (~52% identity to D.m.Mα) (unpublished data derived from the Tribolium database). We postulate that the ur-complex consisted of these two ancestral genes, mα and mβ. It is intriguing that they belong to the two different classes of Notch-responsive genes in the E(spl)-C, the bHLH and the Brd-class. In the fly, mα and bHLH genes are similarly expressed [14,15,22]. It is not unlikely that they share common regulatory elements that could explain their co-segregation in the process of evolution.
What about the third bHLH coding gene found in Apis, A.m.mβ'? This gene may have derived by duplication of A.m.mβ or of A.m.mγ. It is peculiar that this protein is more similar to D. melanogaster Mβ protein than to either A.m.Mβ or A.g.Mβ (see Table 3). In contrast, A.m.Mγ is more similar to A.m.Mβ than to A.m.Mβ'. Furthermore this gene has three introns; one of them is larger than 4 kb reminiscent of Drosophila hairy or dpn intron sizes. We think that it is unlikely that A.m.mβ' encodes one of the other Hairy/E(spl)-type proteins since respective orthologs were found in the A. mellifera genome with the exception of Her, which seems also absent from the mosquito genome. However, there are similarities between the A.m.Mβ' and D. melanogaster Her proteins, including one intron which is at a similar position (Fig. 8B). Although highly speculative, one might conclude that the Drosophila Her gene originally derived from an ancient E(spl) bHLH type gene. However, this speculation has to be proved or disproved by further investigations. The fact that the positions of the introns of Drosophila dpn and hairy are identical and the introns in A.m.mγ and A.m.mβ' are at very similar positions (Fig. 8) supports the notion of a common ancestry of these genes.
Table 3 Comparison between Apis mellifera, D. melanogaster and Anopheles gambia (% similarity/identity)
A.m.Mβ A.m.Mγ A.m.Mβ'1 A.m.Mβ'2 D.m. mβ D.m.Mγ
D.m.Mγ 72/65 72/66 71/61 71/62
D.m.Mβ 77/67 69/63 73/62 76/66
A.g.Mβ 83/76 70/66 67/57 66/57 80/70 76/67
A.m.Mγ 73/67 69/57 68/54
A.m.Mβ 69/60 64/56
A.m.Mβ'1: own prediction
A.m.Mβ'2: database prediction
Penalties: bestfit: Gap weight: 1, length weight: 1, max. penalized length: 30
The Achaete-Scute complex
Genes related to achaete or scute have been identified in a large number of species, from hydra [47] to mouse [30], and so we expect these also in the different insects. The AS-C was most intensely studied in various species of Schizophora flies, apart from Drosophila [28,37,48-53]. The number of genes varies between one and four, however, is not strictly correlated with the position in the phylogenetic tree. For example, AS-C of Calliphora vicina contains three genes, whereas other dipteran flies like Drosophila contain four. Two genes are found in the branchiopod crustacean Triops longicaudatus [54] and only one in hydra [47]. In Dipteran flies the expression patterns of the proneural genes are largely varied. This is regulated by positional information through the Iroquois Complex and pannier and in addition by a transcriptional feed-back loop involving AS-C proteins. Eventually, neural precursors are selected by the repressive activity of E(spl) bHLH proteins [55,56]. Thereby, location and number of the large bristles on the notum is precisely controlled. The mosquito is covered with rows of large sensory bristle, where number and position varies between individuals [57]. This is in accordance with the fact that there is only one scute-like gene, A.g.ash that is expressed all over the presumptive notum in a modular pattern [37]. Recently it was shown that the Anopheles A.g.ash gene can mimic the endogenous Drosophila genes and that overexpression leads to many ectopic bristles [37].
Albeit the bristle pattern on the notum of different Drosophilids varies slightly, bristle number and position is highly stereotyped [58]. Therefore, it is not surprising to find the AS-C highly conserved within Drosophilids. Yet, the rate of change came unexpected and is quite remarkable outside of the bHLH domain. Compared to E(spl) bHLH proteins, those encoded by AS-C have a rather low degree of similarity, most notably Ac. In fact, the big flesh fly Calliphora vicina, which like Drosophila belongs to the Schizophora, is totally lacking the ac gene and is covered with bristles [51]. In agreement, we were unable to find ac in Anopheles or Apis, arguing for rapid evolution. The best conservation rate is found in Sc and L'sc suggesting high evolutionary pressure and maybe common ancestry. Not only the bHLH domain, but also two small stretches outside (aa 203; SPTPS in D. melanogaster L'sc) and also the C-terminus are of high similarity, the latter found identical in Calliphora [51]. Presumably these protein domains are of functional importance. Indeed, the C-terminus acts as transcriptional activation domain and is also used to recruit E(spl) bHLH proteins [36]. Although the alignments of the respective genes of honeybee and mosquito to sc and l'sc are very similar, the tendency goes to a closer relationship to l'sc. However, we propose that this gene pair arose by duplication in the course of Drosophilid evolution, such that we may be looking at a common ancestor in the other two species.
The rate of conservation is very limited for the Ase homologs. Decent conservation is found within the bHLH domain, and moreover, a further well-conserved box is present (NGxQYxRIPGTNTxQxL; x are differences between A. gambiae and D. melanogaster). This sequence is likewise detected in the Ase protein of C. vicina, which however shares many more similarities with D.m.Ase [51]. In Apis, there is no such conservation outside of the bHLH domain, which itself is highly diverged. The overall degree of conservation is so poor that further statements about the relationship are difficult. We argue that this gene represents A.m.ase by its close proximity to A.m.ash, although other interpretations are similarly possible. An analysis of its expression pattern in honeybee may help to solve these questions.
Conclusion
We aimed towards an understanding of the evolution of E(spl)-C and AS-C complexes which in D. melanogaster comprise genes of apparent redundant functions. Our analysis covered insect species that belong to the orders Hymenoptera (honeybee) and Diptera and there to the suborders Nematocera (mosquito) and Brachycera (three species of the genus Drosophila) and thus spans an about 300 Myr window of evolution. We find that both E(spl)-C and AS-C expanded rather recently as they are only present in their nowadays complex structures in Drosophilids. In Apis and in Anopheles, we find very similar arrangements indicative of an ancient ur-complex. The E(spl)-C seems to have evolved from two genes, one HES-like and one Brd-like gene that are arranged in a head to head orientation. Both types of genes are responsive to Notch signalling in Drosophila. Our data suggest that the most ancient genes are E(spl) bHLH mβ and E(spl) mα from which the other E(spl)-C genes derived by duplication and subsequent change. Moreover, an E(spl) ur-complex is likewise detected in Tribolium castaneum that belongs to the order Coleoptera. In Drosophila the complex also gained unrelated genes like m1 and gro. The latter is highly conserved, however, located at different genomic positions. Whereas in Anopheles the ur-complex seems to exist in its original form, two additional mβ-like bHLH genes are found in the Apis genome that possess introns. These introns are at similar positions as the introns of two other HES-like genes, dpn and h which themselves are highly conserved in the three insect species, arguing for a common evolutionary history. Presumably, the introns are evolutionarily ancient as they are also found in the C. elegans E(spl)/h like gene lin-22. The AS-C seems to originate from a single sc/l'sc like bHLH gene and a second largely diverged bHLH gene that shares similarity with Drosophila ase. The high degree of variation in the latter makes it difficult to conclusively decide on the original arrangement of this gene complex.
Methods
Databases
Drosophila melanogaster gene and protein sequences were accessed in FlyBase [31]. The D. pseudoobscura database is found at the Human Genome Sequencing Center [59]. The genome of D. virilis is sequenced by Agencourt Bioscience Corporation and can be downloaded [60] or searched [61]. The Apis mellifera and Anopheles gambiae databases can be accessed with the Ensembl genome browser [62]. The honeybee genomic sequence is also available at the Human Genome Sequencing Center [59] and the EST sequences were obtained from the honeybee brain EST project [63]. Table 2 lists accession numbers of genes, contig numbers and positions of genes therein as well as accession numbers to gene predictions. Due to updates of the databases, gene annotation and names may be different now than reported here.
Sequence analysis
Sequences were downloaded and further studied applying the HUSAR programs of the Deutsche Krebsforschungszentrum [64,65]. Genomic DNA was translated with MAP. BESTFIT and GAP programs were used for alignments and calculation of similarity and identity scores. DOMAIN SWEEP was applied to define bHLH and orange domains, respectively. The amphipathic α-helices were predicted and drawn using WHEEL.
D. pseudoobscura sequence acquisition
The annotation of the D. pseudoobscura database is well advanced. TBlastN searches starting from FlyBase [31] or using D. melanogaster genes allow easy access to both E(spl)-C and AS-C gene complexes. The database also gives information on transcript length, structure and orientation.
D. virilis sequence acquisition
Searches were done with BlastN. However, the D. virilis database allows this scan only with DNA against DNA. We therefore searched first for a characteristic part of a D. melanogaster gene (i.e. the bHLH domain) using lowest percent identity option (75% identity) to find the respective virilis ortholog. We then used the FIND program to identify the respective contig containing these sequences within the downloaded genomic sequence. Subsequently, the entire contig was translated in all six possible reading frames and manually analysed for E(spl)-C and AS-C genes and proteins. The complete E(spl)-C was located in one contig. To identify the whole sequence of the AS-C, three overlapping contigs had to be investigated. However, the contigs covering the AS-C still contained many large unsequenced or uncertain stretches. Therefore, the exact size of the complex could not be defined. The sequence gaps do not affect the coding sequences of the studied genes.
A. gambiae and A. mellifera sequence acquisition
The other databases were screened with D. melanogaster protein sequences for members of E(spl)-C and AS-C against genomic DNA. After detecting convincing similarities, the surrounding genomic DNA was downloaded for further studies with the HUSAR programs. Most of the genes that we describe here have also been annotated as transcribed regions. However, the majority of the database predictions were either incomplete or inconsistent. For our predictions, we carefully analysed genomic DNA and possible coding sequences. All six reading frames were searched for bHLH domain sequences and in the case of the E(spl)-C bHLH proteins also for WRPW motifs. Predicted introns were scanned for open reading frames and their borders reinvestigated. Exon/intron boundaries were defined by obeying to the GT AG rule. Afterwards it was analysed whether the newly predicted introns affected the open reading frame.
Classification of gene homology
Gene homology and orthology was classified based on sequence identity with the respective D. melanogaster protein. In the studied Drosophila species all analysed genes have been identified in a 1:1 ratio and could therefore be classified as orthologs. The degree of similarity and identity between two related proteins was determined with the BestFit program [64]. Comparison of proteins from the three Drosophila species was done under the pre-configured standard conditions of the program (Gap weight: 8, length weight: 2). A TBlastN search in the Apis and Anopheles databases always returned several sequences, and we analysed up to 15 of the best hits. If the hit was in a region without predicted gene, we translated the respective sequence and analysed the open reading frames manually. They were analysed for expected protein motifs, like bHLH, orange or WRPW domain. The predicted protein sequences were then used for a BlastP search in FlyBase. In case of a 1:1 relationship, the genes were classified as orthologs. Since there are several E(spl)-C bHLH and AS-C bHLH genes in Drosophila, a 1:1 allocation was not possible, therefore, we classified them as homolog (see results). Protein sequences derived from the Apis or Anopheles genome projects are so diverged, that standard conditions only align the best conserved domains. We therefore changed the conditions to Gap weight: 1, length weight: 1, max. penalized length: 30. These relaxed conditions had little influence on either alignment of well-conserved sequences, the similarity or identity values. For example, the D.m.Mβ and D.v.Mβ orthologs share 89% identity under stringent conditions versus 93% identity under relaxed conditions. The reduced stringency however, allowed an alignment of the entire protein sequence also of the diverged proteins with the consequence that the identity values increased considerably compared to standard conditions. For example an alignment of A.g.Ash (371 residues) with D.m.L'sc (257 residues) under standard conditions aligns only the residues A.g.Ash 50–233 with D.m.L'sc 40–218 with an identity of 46%. Under relaxed conditions the whole protein sequences align with 64% identity.
Authors' contributions
RS has contributed substantially to data acquisition and participated in sequence alignments. DM designed the study, acquired and analysed the data and drafted the manuscript. Both authors have read and approved the final manuscript.
Acknowledgements
We are indebted to the different Genome Projects for free and easy access to genomic data resources, the FlyBase, the Human Genome Sequencing Center, the Agencourt Bioscience Corporation, the Ensembl Genome Browser and the Honeybee Brain EST Project at the University of Illinois. We thank A. Preiss for constant support and AP, A.C. Nagel and an anonymous reviewer for critically reading the manuscript.
==== Refs
Artavanis-Tsakonas S Rand MR Lake RJ Notch signaling: cell fate control and signal integration in development Science 1999 284 770 776 10221902 10.1126/science.284.5415.770
Welshons WJ Dosage sensitive interactions with split mutations in the presence of an Enhancer of split Drosophila Inf Service 1956 30 157 158
Dawson SR Turner DL Weintraub H Parkhurst SM Specificity for the hairy/Enhancer of split basic helix-loop-helix (bHLH) proteins maps outside the bHLH domain and suggests two separable modes of transcriptional repression Mol Cell Biol 1995 15 6923 6931 8524259
Delidakis C Artavanis-Tsakonas S The Enhancer of split locus of Drosophila encodes seven independent helix-loop-helix proteins Proc Natl Acad Sci USA 1992 89 8731 8735 1528887
Hartley D Preiss A Artavanis-Tsakonas S A deduced gene product from the Drosophila neurogenic locus, Enhancer of split, shows homology to mammalian G-protein β subunit Cell 1988 55 785 795 3142687 10.1016/0092-8674(88)90134-1
Klämbt C Knust E Tietze K Campos-Ortega JA Closely related transcripts encoded by the neurogenic gene complex Enhancer of split of Drosophila melanogaster EMBO J 1989 8 203 210 2540957
Knust E Tietze K Campos-Ortega JA Molecular analysis of the neurogenic locus Enhancer of split of Drosophila melanogaster EMBO J 1987 6 4113 4123 16453817
Knust E Schrons H Grawe F Campos-Ortega JA Seven genes of the Enhancer of split complex of Drosophila melanogaster encode Helix-loop-helix proteins Genetics 1992 132 505 518 1427040
Paroush Z Finley RL Kidd T Wainwright SM Ingham PW Brent R Ish-Horowitz D groucho is required for Drosophila neurogenesis, segmentation, and sex determination and interacts directly with hairy-related bHLH proteins Cell 1994 79 805 815 8001118 10.1016/0092-8674(94)90070-1
Preiss A Hartley DA Artavanis-Tsakonas S The molecular genetics of Enhancer of split, a gene required for embryonic neural development in Drosophila EMBO J 1988 7 3917 3927 3145200
Eastman DS Slee R Skoufos E Bangalore L Bray S Delidakis C Synergy between Suppressor of Hairless and Notch in regulation of Enhancer of split mγ and mδ expression Mol Cell Biol 1997 17 5620 5628 9271437
Jennings B Preiss A Delidakis C Bray S The Notch signaling pathway is required for Enhancer of split bHLH protein expression during neurogenesis in the Drosophila embryo Development 1994 120 3537 3548 7821220
Jennings B de Celis J Delidakis C Preiss A Bray S Role of Notch and achaete-scute complex in the expression of Enhancer of split bHLH proteins Development 1995 121 3745 3752
Lai EC Bodner R Posakony JW The Enhancer of split Complex of Drosophila includes four Notch-regulated members of the Bearded family Development 2000 127 3441 3455 10903170
Wurmbach E Wech I Preiss A The Enhancer of split complex of Drosophila melanogaster harbors three classes of Notch responsive genes Mech Dev 1999 80 171 180 10072784 10.1016/S0925-4773(98)00212-3
Delidakis C Preiss A Hartley DA Artavanis-Tsakonas S Two genetically and molecularly distinct functions involved in early neurogenesis reside within the Enhancer of split locus of Drosophila melanogaster Genetics 1991 129 803 823 1752423
Nagel AC Maier D Krauss S Mezger M Preiss A Neurogenic phenotypes induced by RNA interference with bHLH genes of the Enhancer of split complex of Drosophila melanogaster Genesis 2004 39 105 114 15170696 10.1002/gene.20033
Schrons H Knust E Campos-Ortega JA The Enhancer of split complex and adjacent genes in the 96F region of Drosophila melanogaster are required for segregation of neural and epidermal progenitor cells Genetics 1992 132 481 503 1427039
Cooper MT Tyler DM Furriols M Chalkiadaki A Delidakis C Bray S Spatially restricted factors cooperate with Notch in the regulation of Enhancer of split genes Dev Biol 2000 221 390 403 10790334 10.1006/dbio.2000.9691
de Celis JF de Celis J Ligoxygakis P Preiss A Delidakis C Bray S Functional relationships between Notch, Su(H) and the bHLH genes of the E(spl) complex: the E(spl) genes mediate only a subset of Notch activities during imaginal development Development 1996 122 2719 2728 8787746
Tata F Hartley DA Inhibition of cell fate in Drosophila by Enhancer of split genes Mech Dev 1995 51 305 315 7547476 10.1016/0925-4773(95)00377-0
Wech I Bray S Delidakis C Preiss A Distinct expression patterns of different Enhancer of split bHLH genes during Drosophila embryogenesis Dev Genes Evol 1999 209 370 375 10370119 10.1007/s004270050266
Maier D Marte BM Schäfer W Yu Y Preiss A Drosophila evolution challenges postulated redundancy in the E(spl) gene complex Proc Natl Acad Sci USA 1993 90 5464 5468 8516287
Fisher AL Caudy M Groucho proteins: transcriptional corepressors for specific subsets of DNA-binding transcription factors in vertebrates and invertebrates Genes Dev 1998 12 1931 1940 9649497
Fisher A Caudy M The function of hairy-related bHLH repressor proteins in cell fate decisions Bioassays 1998 20 298 306 10.1002/(SICI)1521-1878(199804)20:4<298::AID-BIES6>3.0.CO;2-M
Artavanis-Tsakonas S Simpson P Choosing a cell fate: a view from the Notch locus Trends Genet 1991 7 403 8 1668193 10.1016/0168-9525(91)90220-K
Campuzano S Modolell J Patterning of the Drosophila nervous system: the achaete-scute gene complex Trends Genet 1992 8 202 208 1496555
Cubas P de Celis J Campuzano S Modolell J Proneural clusters of achaete-scute expression and the generation of sensory organs in the Drosophila imaginal wing disc Genes Dev 1991 5 996 1008 2044965
Heitzler P Bourois M Ruel L Cateret C Simpson P Genes of the Enhancer of split and achaete-scute complexes are required for a regulatory loop between Notch and Delta during lateral signalling in Drosophila Development 1996 122 161 171 8565827
Bertrand N Castro D Guillemot F Proneural genes and the specification of neural cell fates Nature Rev Neurosci 2002 3 517 530 12094208 10.1038/nrn874
Flybase, A Database of the Drosophila Genom
Gerhart J Kirschner M Arthropod phylogeny Cells, embryos, and evolution 1997 Blackwell Science. Malden 305
Lai EC Tam B Rubin GM Pervasive regulation of Drosophila Notch target genes by GY-box, Brd-box, and K-box-class microRNAs Genes Dev 2005 19 1067 1080 15833912 10.1101/gad.1291905
Moore AD Barbel S Jan LY Jan YN A genomewide survey of basic helix-loop-helix factors in Drosophila Proc Natl Acad Sci USA 2000 97 10436 10441 10973473 10.1073/pnas.170301897
Prokopenko SN He Y Lu Y Bellen HJ Mutations affecting the development of the peripheral nervous system in Drosophila: A molecular screen for novel proteins Genetics 2000 156 1691 1715 11102367
Giagtzoglou N Koumbanakis KA Fullard J Zarifi I Delidakis C Role of the Sc C-terminus in transcription activation and E(spl) repressor recruitment J Biol Chem 2005 280 1299 1305 15507447 10.1074/jbc.M408949200
Wülbeck C Simpson P The expression of pannier and achaete-scute homologues in a mosquito suggests an ancient role of pannier as a selector gene in the regulation of the dorsal body plan Development 2002 129 3861 3871 12135924
Burge C Karlin S Prediction of complete gene structures in human genomic DNA J Mol Biol 1997 268 78 94 9149143 10.1006/jmbi.1997.0951
Giagtzoglou N Alifragis P Koumbanakis KA Delidakis C Two modes of recruitment of E(spl) repressors onto target genes Development 2003 130 259 270 12466194 10.1242/dev.00206
Gibert J-M Simpson P Evolution of cis-regulation of the proneural genes Int J Dev Biol 2003 47 643 651 14756340
Massari ME Murre C Helix-Loop-Helix proteins: regulators of transcription in eucaryotic organisms Mol Cell Biol 2000 20 429 440 10611221 10.1128/MCB.20.2.429-440.2000
Oellers N Dehio M Knust E bHLH proteins encoded by the Enhancer of split complex of Drosophila negatively interfere with transcriptional activation mediated by proneural genes Mol Gen Genet 1994 244 465 473 8078474 10.1007/BF00583897
Lai EC Posakony JW Regulation of Drosophila neurogenesis by RNA:RNA duplexes? Cell 1998 93 1103 1104 9657143 10.1016/S0092-8674(00)81454-3
Castro B Barolo S Bailey AM Posakony JM Lateral inhibition in proneural clusters: cis-regulatory logic and default repression by Suppressor of Hairless Development 2005 132 3333 3344 15975935 10.1242/dev.01920
Carroll SB Grenier JK Weatherbee SD From DNA to diversity: molecular genetics and the evolution of animal design 2001 46 Blackwell Science. Malden 148 153
McGinnis W Garber RL Wirz J Kuroiwa A Gehring WJ A homologous protein-coding sequence in Drosophila homeotic genes and its conservation in other metazoans Cell 1984 37 403 408 6327065 10.1016/0092-8674(84)90370-2
Grens A Mason E Marsh JL Bode HR Evolutionary conservation of a cell fate specification gene: the Hydra achaete scute homolog has proneural activity in Drosophila Development 1995 121 4027 4035 8575303
Alonso MC Cabrera CV The achaete-scute gene complex of Drosophila melanogaster comprises four homologous genes EMBO J 1988 7 2585 2591 2461300
Ghysen A Dambly-Chaudiere C From DNA to form: the achaete-scute complex Genes Dev 1988 2 495 501 3290049
Gonzalez F Romani S Cubas P Modolell J Campuzano S Molecular analysis of the asense gene, a member of the achaete-scute complex of Drosophila melanogaster, and its novel role in optic lobe development EMBO J 1989 8 3553 3562 2510998
Pistillo D Skaer N Simpson P scute expression in Calliphora vicina reveals an ancestral pattern of longitudinal stripes on the thorax of higher Diptera Development 2002 129 563 572 11830558
Skeath JB Carroll SB Regulation of the achaete-scute gene expression and sensory organ pattern formation in the Drosophila wing Genes Dev 1991 5 984 995 2044964
Villares R Cabrera CV The achaete-scute gene complex of D. melanogaster: conserved domains in a subset of genes required for neurogenesis and their homology to myc Cell 1987 50 415 424 3111716 10.1016/0092-8674(87)90495-8
Wheeler SR Skeath JB The identification and expression of achaete-scute genes in the branchiopod crustacean Triops longicaudatus Gene Expr Patterns 2005 5 695 700 15939382 10.1016/j.modgep.2005.02.005
Garcia-Garcia MJ Ramain P Simpson P Modolell J Different contributions of pannier and wingless to the patterning of the dorsal mesothorax of Drosophila Development 1999 126 3523 3532 10409499
Gomez-Skarmeta JL Diez del Corral R de la Calle-Mustienes E Ferre-Marco D Modolell J Araucan and caupolican, two members of the novel iroquois complex, encode homeoproteins that control proneural and vein-forming genes Cell 1996 85 95 105 8620542 10.1016/S0092-8674(00)81085-5
Simpson P Woehl R Usui K The development and evolution of bristle patterns in Diptera Development 1999 126 1349 1364 10068629
Bächli G Bural H Taxonomische Merkmale von Drosophila-Arten Diptera Drosophilidae 1985 Schweizerische Entomologische Gesellschaft, Zürich: ETH-Zentrum 20 30
Human Genome Sequencing Center
D. virilis (Drosophila virilis) Genome Browser Gateway
NCBI, Trace Archive database Mega BLAST search
Ensembl genome browser
Honeybee brain EST project
HUSAR – Sequence Analysis Package
Senger M Flores T Glatting K-H Ernst P Hotz-Wagenblatt A Suhai S W2H: WWW interface to the GCG sequence analysis package Bioinformatics 1998 14 452 457 9682058 10.1093/bioinformatics/14.5.452
|
16293187
|
PMC1310631
|
CC BY
|
2021-01-04 16:29:16
|
no
|
BMC Evol Biol. 2005 Nov 17; 5:67
|
utf-8
|
BMC Evol Biol
| 2,005 |
10.1186/1471-2148-5-67
|
oa_comm
|
==== Front
BMC CancerBMC Cancer1471-2407BioMed Central London 1471-2407-5-1441626289910.1186/1471-2407-5-144Research ArticleThe bone marrow aspirate and biopsy in the diagnosis of unsuspected nonhematologic malignancy: A clinical study of 19 cases Ozkalemkas Fahir [email protected] Rıdvan [email protected] Vildan [email protected] Tulay [email protected] Ulku [email protected] Hulya [email protected] Ender [email protected] Turkkan [email protected] Omer [email protected] Ahmet [email protected] Division of Hematology, Department of Internal Medicine Uludag University School of Medicine, Uludag University Hospital, Bursa, Turkey2 Department of Pathology, Uludag University School of Medicine, Uludag University Hospital, Bursa, Turkey3 Division of Medical Oncology, Department of Internal Medicine Uludag University School of Medicine, Uludag University Hospital, Bursa, Turkey2005 1 11 2005 5 144 144 18 5 2005 1 11 2005 Copyright © 2005 Ozkalemkas et al; licensee BioMed Central Ltd.2005Ozkalemkas et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
Although bone marrow metastases can be found commonly in some malignant tumors, diagnosing a nonhematologic malignancy from marrow is not a usual event.
Methods
To underscore the value of bone marrow aspiration and biopsy as a short cut in establishing a diagnosis for disseminated tumors, we reviewed 19 patients with nonhematologic malignancies who initially had diagnosis from bone marrow.
Results
The main indications for bone marrow examination were microangiopathic hemolytic anemia (MAHA), leukoerythroblastosis (LEB) and unexplained cytopenias. Bone marrow aspiration was not diagnostic due to dry tap or inadequate material in 6 cases. Biopsy results were parallel to the cytological ones in all cases except one; however a meticulous second examination of the biopsy confirmed the cytologic diagnosis in this patient too. The most common histologic subtype was adenocarcinoma, and after all the clinical and laboratory evaluations, the primary focus was disclosed definitively in ten patients (5 stomach, 3 prostate, 1 lung, 1 muscle) and probably in four patients (3 gastrointestinal tract, 1 lung). All work up failed in five patients and these cases were classified as tumor of unknown origin (TUO).
Conclusion
Our series showed that anemia, thrombocytopenia, elevated red cell distribution width (RDW) and hypoproteinemia formed a uniform tetrad in patients with disseminated tumors that were diagnosed via bone marrow examination. The prognosis of patients was very poor and survivals were only a few days or weeks (except for 4 patients whose survivals were longer). We concluded that MAHA, LEB and unexplained cytopenias are strong indicators of the necessity of bone marrow examination. Because of the very short survival of many patients, all investigational procedures should be judged in view of their rationality, and should be focused on treatable primary tumors.
==== Body
Background
Diagnosis and management of many hematologic diseases depends on examination of the bone marrow, which usually involves two separate specimens: a cytologic and a histologic preparation. While cytologic preparation of bone marrow, obtained by aspiration, allows excellent visualization of cell morphology, the second one, usually obtained with a Jamshidi needle, allows optimal evaluation of cellularity, fibrosis or infiltrative disease. In addition to hematologic malignancies, bone marrow examination has been increasingly useful in documenting metastatic involvement of tumors. During the past four decades, prospective evaluation of bone marrow aspirates and biopsy specimens has come into widespread use for accurate staging of many malignant diseases. Recognition of metastasis in random biopsies presents challenges to hematologists and pathologists when diagnosing the primary focus [1,2]
Although marrow metastases can be found commonly in some tumors, especially when newer sensitive methods are applied for the detection of tumor cells, diagnosing a nonhematologic malignancy from marrow is a rare event. We reviewed 19 patients with nonhematologic malignancies who were diagnosed initially from bone marrow, to underscore the value of bone marrow aspiration and biopsy as a short cut in establishing a diagnosis for disseminated tumors. Additionally we reported the details of the management and survival of the cases to offer a practical suggestion about work up of these patients to find a primary focus.
Methods
This study is based on our retrospective analysis of 19 patients with solid tumors whose diagnoses were made from bone marrow seen at the Department of Hematology, Uludag University Medical Faculty, over a period of 9 years. Patients with non-Hodgkin's lymphomas and Hodgkin's diseases were not included in this study and patients with known neoplastic disease were also excluded.
The standard technique was employed in obtaining the samples from posterior iliac crest using a Jamshidi needle (Regular/Adult, 11-gauge). All of the trephine biopsies were performed unilaterally, because clinically none of the diseases that focally involve the bone marrow were included in the differential diagnosis prior to the biopsy procedure. Length of the biopsy cores ranged between 1.2 cm and 2.2 cm (mean 1.7 cm). Trephine biopsies were fixed in 10% neutral buffered formaline for at least 24 hours, and then decalcified overnight in a decalcifying solution which is a mixture of 8% HCl and 10% formic acid at equal amounts of volume. Following the automated tissue processing, biopsies were embedded in parafin blocks, and 0.3 micrometer sections were cut. In some cases the sternum was sampled using a Rosenthal needle for aspiration. Touch preparations were done if the aspiration resulted in a "dry tap" or if aspiration material was considered to be technically inadequate for evaluation, or if it was hemodilute. Bone marrow aspiration, touch preparations and peripheral blood smears which were obtained at the same time by biopsy were stained by May Grunwald-Giemsa. Hematoxylene-eosin, Giemsa and reticulin stains were routinely performed in biopsy sections. If the nonhematologic properties of the tumor could be identified with routine stains, and if the primary of the metastatic tumor could easily be diagnosed morphologically with routine hematoxylene-eosin stains, as in the cases of signet ring cell carcinoma; no immunohistochemical study was held. The pathologist's approach to definitive diagnosis of the patient with metastasis of unknown primary effectively followed a few sequential steps: First of all we tried to determine the cell line of differentiation e.g. carcinoma, lymphoma, melanoma, sarcoma, or germ cell, with the help of morphological findings and if needed, immunohistochemical stains. Our panel of antibodies contained pancytokeratin, HMB45, Leukocyte Common Antigen (LCA), Vimentin and Placental Alkalen Phosphatase (PLAP). If it was an epithelial tumor we tried to determine the cytokeratin (CK) type or types of distribution in the tumor cells, since some subsets of cytokeratins are uniqe to certain tumor types. Our panel contained AE1/AE3, CAM5.2, CK7, CK20, and 34 beta E12. In our further studies we tried to determine if there was expression of supplemental antigens of epithelial or germ cell derivation, that was Carcinoembryonic Antigen (CEA), Epithelial membrane Antigen (EMA) or PLAP. Last step was to determine if there was expression of cell-specific structures or receptors that are unique identifiers of cell types, for example neuroendocrine granules, peptide hormones, thyroglobulin, Prostate Specific Antigen (PSA), Gross Cystic Disease Fluid Protein (GCDFP) or Thyroid Transcription Factor-1 (TTF-1). Our panel of antibodies contained Synaptophysin, Chromogranin, Neuron Specific Enolase (NSE), Thyroglobuline, PSA, GCDFP, and TTF-1. Cases with sarcomatous properties were immunostained with Desmin, Smooth Muscle Actin, S100, Vimentin and Myoglobulin. After the confirmation of original peripheral blood and bone marrow cytological findings and histopathological diagnoses by a senior hematologist and an expert pathologist the patient charts were reviewed. Cases with a history of malignancy at the time of presentation were excluded.
Patient characteristics were recorded in each case, including: presenting symptoms, onset of symptoms, physical examination findings, peripheral blood counts, peripheral blood morphology, diagnostic evaluation, management, and survival.
Results
Between 1995 and 2004, bone marrow metastasis was diagnosed in 19 samples among 5420 bone marrow aspirations and 856 bone marrow trephine biopsies. The ages of the patients were between 25 and 83 years (median age: 56), eight of them were female.
The main indications for bone marrow examination were microangiopathic hemolytic anemia (MAHA), leukoeryhtroblastosis (LEB) and peripheral cytopenias. Constitutional symptoms and pain were the most prominent presenting symptoms of the patients. Clinical findings were highly variable according to the underlying disease.
In the laboratory, the most common findings were anemia and thrombocytopenia, which were found in all patients. White blood cell counts (WBC) were between 3.3 × 109/l and 17.1 × 109/l (median: 8.1 × 109/l), red cell distribution width (RDW) was increased in all cases, whereas mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), reticulocyte percentage (RET) and erythrocyte sedimentation rate (ESR) were highly variable. Coagulation tests were carried out in 17 patients and in 8 of them at least one anomaly was detected at presentation; additionally, in two patients who had normal test results initially subsequent tests were found abnormal. All patients had hypoproteinemia; the second most common anomaly was elevated serum lactate dehydrogenase (LDH) level, which was found normal in only one patient (patient # 13) in biochemical analyses; some hepatic and renal parameters were abnormal in some patients but none of them was a constant finding.
In 6 cases bone marrow aspiration was not diagnostic due to dry tap or inadequate material. Histopathological examinations confirmed the nonlymphohematopoietic cell infiltration in first evaluation except one (patient # 15). However, a meticulous second evaluation of biopsy confirmed the cytologic diagnosis in this patient too. The most common histologic subtype was adenocarcinoma. After the all clinical and laboratory evaluations, the primary focus was disclosed in ten patients definitively (5 stomach, 3 prostate, 1 lung, 1 muscle) and in four patients probably (3 gastrointestinal tract, 1 lung). In five patients all work up failed and these cases were classified as tumor of unknown origin (TUO).
The prognosis of patients was very poor and survivals were only a few days or weeks except for 4 patients whose survivals were longer (patients # 9, 10, 17, 18). To avoid missing the individual features, detailed tables were prepared: Clinical and cytopathologic characteristics are shown in Table 1, clinical courses and survivals are shown in Table 2, and some hematological and biochemical features are shown in Table 3.
Table 1 Clinical and cytopathological characteristics of patients
No Age/sex Presenting symptoms/Onset of symptoms/Presence of constitutional symptoms Performance status*/Physical findings Cytological examination of peripheral blood Cytological examination of bone marrow Pathological examination of bone marrow
1 50/M Back and chest pain/3 weeks/WL, F, NS 3/Pallor, subicterus MAHA, LEB Dilute, not optimal, not diagnostic Adenocarcinoma with signet ring cell features (Suggestion: primary focus should be searched in GI tract)
2 40/M Abdominal pain, failure to passage gas and stool by rectum, hematemesis/2 weeks/NS 3/Ecchymoses, tenderness in epigastrium and lower-right quadrant MAHA, LEB Dilute, not optimal, not diagnostic Adenocarcinoma
3 41/F Lumbar and extremity pain, lack of appetite, nausea/4 weeks/WL, F 3/Pain with deep palpation of whole abdomen LEB, MAHA Dilute, not optimal, not diagnostic Adenocarcinoma
4 48/F Lack of appetite, fatigue, nausea, vomiting, fever/2 months/F, WL 3/A few ecchymoses LEB Dry tap; touch preparation is not optimal for evaluation Indifferentiated carcinoma (only CK positive strongly)
5 49/F Dyspepsia, weakness/1 month/F 3/Pallor, multiple ecchymoses, axillary single microLAP LEB Dense foreign cell infiltration forming groups (adenocarcinoma) Adenocarcinoma
6 71/M Lumbar and leg pain, somnolence/20 days/- 4/Pallor, impaired consciousness, dysorientation, dyscooperation, agitation LEB Dry tap; imprint: highly dense atypical somewhat large round or oval cells infiltration in clusters Atypical epithelial cells in clusters (round cells infiltration) (Suggestion: Primary focus should be investigated in lungs)
7 63/M Cough, dysphagia, abdominal swelling, weakness, prominent loss in weight/WL, F, NS/1 month 2/Scleral icterus, 5 cm hepatomegaly, melana in rectal digital palpation LEB Epithelioid cells in solid clusters (small cell carcinoma infiltration) Small cell carcinoma
8 57/F Back, lumbar and leg pain, weakness, lack of appetite/3 months/WL 3/Scleral icter, left axillary 2 cm LAP MAHA, LEB Infiltration with signet ring cells Metastatic carcinoma (signet ring cell adenocarcinoma)
9 45/M Lumbar and leg pain, prominent loss in weight, generalized body pain/2 months/WL 3/Pallor; LEB Infiltration with atypical large epithelioid cells Metastatic carcinoma (Suggestion: Primary tumor should be investigated in prostate)
10 25/F Weakness, hip pain/6 months/ WL 2/Pallor, right inguinal 2 cm LAP, 1 cm hepatomegaly, 2 cm splenomegaly Rare blastic cells Infiltration with blastic cells with vacuolated cytoplasm (MPO negative; flow: B and T cell markers negative) Higly dense atypical cells in alveolar structure-actin, desmin and vimentin positive, LCA and CK negative-(metastatic alveolar rhabdomyosarcoma)
11 35/M Nausea, vomiting, prominent loss of weight, fatigue (alcohol, hashish and heroin dependence)/1 month/F, WL 4/Pallor, cachexia MAHA Adenocarcinoma cell forming groups Metastatic adenocarcinoma
12 83/F Prominent loss in weight, nausea, vomiting; backpain/6 months/WL 3/Pallor; multipl ecchymoses MAHA, LEB Infiltration with signet ring cell carcinoma cells Metastatic adenocarcinoma (signet ring cell carcinoma metastasis)
13 61/F Headache, sore throat, abdominal pain, constipation, nausea, vomiting (hematemesis and melana history), weakness/2 months/- 2/Pallor and scleral icterus, pain with palpation of right hypochondrium LEB Dry tap in first 2 attempts and very dilute without particle in 3rd attempt. No atypical cells; imprint: technically inadequate Metastatic carcinoma (CK positive atypical epithelioid cell infiltration, some of them in signet ring cell shape)
14 75/M Multiple ecchymoses on body and on extremities, purpura on lower extremities; hematuria/2 days/- 1/Ecchymoses and purpura; 1.5 cm supraclavicular LAP Only minimal shift to left; no erythroblast or poikilocytosis Non-hematopoetic cell infiltration forming groups and some with mucineous character Metastatic adenocarcinoma (CK+ cells forming glandular and tubular structures
15 73/M Confusion, adynamia/20 days/WL 4/Hypotension, hypothermia, dehydration, scleral icterus, pallor, cachexia, 2 cm hepatomegaly Slight shift to left, toxic granulation, slight poikilocytosis, no erythroblast, no fragmentation Dilute and not optimal but there are nonhematopoietic cells in small groups like adenocarcinoma cells Nondiagnostic in first report but adenocarcinoma metastasis reported after meticulous examination of new further sections of blocks
16 75/F Dyspnea, abdominal swelling/20 days/- 4/Rales bilaterally, 2 cm hepatomegaly, pretibial edema, petechia in lower extremities LEB, no poikilositosis or fragmentation Atypical non-haematopoetic cells (adenocarcinoma metastasis) ND
17 68/M Weakness, lack of appetite, prominent loss of weight, lumbar pain/3 months/WL 1/Enlargement and nodulation in prostate in digital examination LEB, slight poikilocytosis, no fragmentation Infiltration with adenocarcinoma cells showing acinar and tubular structures Metastatic carcinoma compatible with prostate carcinoma (PSA +)
18 56/M Pain in hips and legs, weakness/2 months/F, WL, SN 2/Pallor, cachexia, multiple microLAPs in servical, axillary and inguinal regions, a few petechiae and eccyhymoses LEB, MAHA Infiltration with atypical epithelioid cells forming papillar and acinar structures (adenocarcinoma metastasis) CK + PAS- adenocarcinoma metastasis (Suggestion: Primary focus should be investigated in prostate)
19 45/M Neck and hip pain, abdominal pain, weakness/ 1 month/WL 3/restriction in physical activity LEB, MAHA Infiltration with adenocarcinoma cells Metastatic adenocarcinoma; CK+ epithelioid cells, some of them mucinous and in shape of signet cell (Suggestion: Primary focus should be investigated in stomach)
*According to WHO/ECOG; WL: weight loss; F: fever; NS: night sweats; LAP: lymphadenopathy; MAHA: microangiopathic hemolytic anemia; LEB: leukoerythroblastosis; MPO: myeloperoxidase; CK: Cytokeratin LCA: leukocyte common antigen; ND: not done
Table 2 Evaluation, clinical course and survival of patients
Evaluation for primary focus Final decision for primary focus Clinical course/Management Survival
1 Chest XR, abdominopelvic USG and transrectal USG: N; CEA and CA15.3 are high slightly and Ca19.9 is high more than 10 folds, among CEA, α FTP, PAP, CA-125, CA19.9, CA15.3 GI tractus General condition deteriorated rapidly; stupor and coma developed/Supportive care; 2 courses of TPE 6 days
2 Chest XR, abdominopelvic USG and CT: N; gastroscopy: malign ulcus Stomach DIC, subdural hematoma developed/Multiple erythrocyte, platelet and plasma transfusions; 2 courses of TPE 11 days
3 Abdominopelvic USG and CT: thickness on the antrum wall, gastrohepatic and portahepatic microLAPs; only α-FTP is in normal limits, among the CEA, α-FTP, CA 125, CA15.5; CA19.9 is high more than 10 folds; gastroscopy: infiltration in corpus and antrum (linitis plastica) Stomach DIC diagnosed at admission. Hematemesis and epistaxis developed later/Despite full transfusion support; died due to intracerebral bleeding 20 days
4 Chest XR, mammography: N; abdominopelvic USG and CT: N except minimal free pelvic fluid; upper GI tract endoscopy: unremarkable TUO Fever resisted despite AB; general condition deteriorated gradually, generalized seizures developed without abnormal cranial CT finding; hypoxemia developed due to secretions/Supportive care only 12 days
5 Chest XR: N; abdominopelvic CT: not optimal; suspected thickness on the stomach wall, suspected metastatic lesions in columna vertabralis; upper GI tract endoscopy: malign ulcus in cardia; CA125 is high slightly and CA-19-9 is high approximately 6 times, among CEA, α-FTP, Ca 125, CA 19-9, CA 15-3 Stomach AB resistant fever (FUO) and GI tract bleeding developed later/Despite full transfusion support her general condition deteriorated rapidly. Died in MODS picture 37 days
6 Cranial CT: N; thorax, abdomen and pelvis MR: multiple mediastinal LAPs in conglemeration with suspected parenchimal infiltration, benign prostate hyperthrophy; lumbar MR: multiple pathologic signal in backbone and degenerative alterations; Skeleton scintigraphy: multiple thoracal and lumbar uptake (degeneration, metastasis, trauma?) Lung? His consciousness impaired progressively; refractory fever and hypotension developed 5 days
7 Thorax, abdominal CT: subcarinal LAPs, a mass in right hilus, eosophagial compression, pulmonary artery and pericardium invasion, a hipodens lesion in 1 cm diameter in liver (USG in terminal period: multiple lesions compatible with metastasis); bronchoscopy: inoperable bronchial carcinoma; biopsy: small cell carcinoma); upper GI tract endoscopy: N. CEA, α-FTP, PSA, freePSA, CA125, CA 19.9 all: N Lung Pneumonia and atrial fibrillation developed/A course of CT (Etoposide+Cisplatine) was given. Died duo to CHF 47 days
8 Axillary node FNAB: benign; cranial MR: compatible with bone metastasis and leptomeningeal carcinomatosis; thorax CT: only bone metastasis; abdominopelvic CT: 3 mm hipodens lesion in liver (metastasis?), backbone metastasis; transvaginal USG: N; bone scintigraphy: multiple metastasis; mammography: N; whole spine MR: generalized sclerotic and lytic lesions; upper GI tract endoscopy: erythemateous gastritis; only CA 125 is high 2 folds, among CEA, α-FTP, CA 125, CA 19-9, CA15-3, BHCG TUO Performance status deteriorated gradually. GI tract bleeding developed/She refused colonoscopy and other supportive therapies and was discharged in very bad condition 38+ days
9 Chest XR: N; transrectal USG: prostate carcinoma?; prostate biopsy: adenocarcinoma; skelatal XR survey: multiple sclerotic metastasis and compression fracture in L3; bone scintigraphy: generalized metastatic involvement; tumor markers: PSA and free PSA are very high Prostate After his work up bisphosphonates therapy was initiated and was fallowed as outpatient; cranial metastasis developed later; despite progressive complaints he refused admission 7+ months
10 Abdominopelvic CT: a solid mass in the pelvis originated probably right gluteal muscle and homogeneous hepatosplenomegaly; biopsy from the mass: rhabdomyosarcoma; skelatal XR survey: lytic lesions in only pelvic bones and proximal femur; bone scintigraphy: pathologic uptake in bilateral knee, pelvic area and, 5. and 7. ribs Muscle VAC/IE (Vincristine, Adriamycine, Cyclophosphomide, Ifosfamide, etoposide) therapy resulted in partial response; died because of progression later 7 months
11 Chest XR: N; abdominopelvic US: N except homogen minimal hepatomegaly GI system? His general condition deteriorated rapidly; GI tract bleeding and subdural hematoma developed, Died because of herniation/Supportive care only 10 days
12 Abdominopelvic CT: normal except suspected rigidity in stomach wall; gastroscopy: malign ulcus in junction of corpus and fundus; biopsy: Adenocarcinoma; Mammography: N; Backbone XR: loss of height in Th11 and Th12 Stomach One course 5FU+FA was given; died as out patient 1 month
13 Nasopharynx biopsy: N; pleural fluid cytology: negative, biopsy: nonspecific chronic pleuritis; mammography:N; bone scintigraphy: multiple uptake; colonoscopy: N; gastroscopy: N; CEA and CA15.3: N, CA19.9 and CA125: very high GI system Transfusion support. Lost to follow up 45+ days
14 Chest XR and abdominopelvic USG: N TUO Nothing. Out of follow in 2nd week 14 days
15 Chest XR: nondiagnostic; CEA, α-FTP, PSA, free PSA, CA15.3, Ca19.9, Ca125 all: N TUO Despite vigorous transfusion support and antibiotics his vital functions deteriorated progressively and died in MODS 4 days
16 Chest XR: nondiagnostic; previous available tests: thorax CT: linear atelectasis and minimal right pleural fluid, 1–2 cm multiple mediastinal LAPs;.abdominal CT: homogeneous hepatomegaly and multiple cysts in 1.5 cm diameter in head of pancreas TUO She died because of hypertensive crises and CHF after admission 1 day
17 Pelvic and transrectal USG: Prostatic hypertrophy; prostate biopsy: Adenocarcinoma; thoracolumbar MR and bone scintigraphy: multiple bone metastasis in backbone Prostate Flutamide (antiandrogen) and Goserelin asetat (LH-LR analogue) were given. Paraparesis and paraplegia unresponsive to RT developed and died because of progressive disease and CHF 15 months
18 Neck and thorax CT: N; abdominopelvic CT: paraaortic 1.5 cm LAPs and heterogen prostatic hypertrophy, prostate biopsy: Adenocarcinoma; pelvis XR: multiple sclerotic lesions; bone scintigraphy; multiple + focuses in whole skeleton: PAP and PAS: very high; CEA, AFP, CA19.9: N Prostate Gaserolin asetat+ Bikalutamid (LH-RH analogues) were given; (he was in a good condition when writing) 3+ months
19 Bone scintigraphy: multiple + uptake; gastroscopy: malign ulcus; biopsy: signet cell carcinoma (antrum); CEA, CA 19.9: very high, Ca125: high, AFP, PSA, F-PSA:N; thoracal MR: loss of height in Th8; toraks CT: frosted glass appearance in lower and middle zones, minimal pleural effusion bilaterally; abdominopelvic CT:N; skelatal XR: Lumbar and pelvic sclerotic lesions Stomach Supportive care and palliative RT were given; he died because of progressive disease 51 days
XR: direct radiography; CEA: carcinoembryonic antigen; α FTP: alpha fetoprotein; PAP: prostate specific antigen; BHCG: Beta human chorionic gonodotropin; USG: ultrasonography; CT: computed tomography; MR: magnetic resonance; Th: thoracal; LAP: lymphadenopathy; GI: gastrointestinal; DIC: disseminated intravascular coagulation; N: normal; FNAB: fine needle aspiration biopsy; TUO: tumor of unknown origin; TPE: therapeutic plasma exchange; AB: antibiotic; FUO: Febris of unknown origin; MODS: multiple organ deficiency syndrome; CHF: congestive hearth failure; RT: radiotherapy;
Table 3 Hematologic parameters of peripheral blood and biochemistry of serum of patients*
No WBC (× 109/l) Hb (mg/dl) MCV (fl) MCH (pg) RDW PLT (× 109/l) RET (%) ESR (mm/1 h) Coagulation tests Total Protein/Albumin LDH Total/Direct Bil AST ALT ALP Urea Creatinine
1 8.1 9.9 94.5 33.3 17.0 32 2.5 17 PT and FDP: high 6.3/3.4 508 3.1/1.4 49 38 525 70 0.8
2 6.8 7.0 93.2 28.7 15.6 19 3.0 20 Ne initially and all abnormal later 5.6/3.0 1042 1.5/0.5 157 33 365 56 1.0
3 11.6 6.4 79.7 27.0 16.1 20 1.0 36 PT aPTT and FDP: abnormal 5.0/2.1 776 2.9/1.1 67 57 587 46 0.4
4 13.1 6.8 74.1 25.2 18.4 41 0.8 78 N 5.7/2.6 1669 0.9/0.5 104 48 282 148 2.1
5 15.8 6.6 82.0 26.3 16.7 18 3.5 34 N initially but abnormal later 5.8/2.7 1016 2.4/1.1 61 11 661 84 0.8
6 10.6 10.6 73.8 24.4 15.3 12 0.2 107 N 6.2/2.8 1255 0.5/0.3 52 22 240 89 1.2
7 4.2 9.3 85.7 29.2 15.2 21 1.4 80 N 5.8/3.6 1438 1.1/0.5 78 54 155 68 1.0
8 6.5 8.1 89.3 30.9 16.5 72 4.7 44 N 5.7/2.9 880 2.3/0.9 30 17 1548 33 0.5
9 6.4 9.5 61.9 20.5 16.3 45 1.0 65 N 6.4/3.4 665 0.5/0.4 41 14 433 43 0.3
10 7.7 5.3 79.3 27.5 15.8 11 0.0 140 N 5.7/2.2 1612 0.5/0.2 26 4 71 15 0.6
11 10.3 4.0 90.0 29.9 16.8 32 1.8 5 D-Dimer and PT: high 6.6/4.0 1899 1.3/0.7 78 22 262 48 0.8
12 11.5 7.0 84.7 26.4 17.0 37 3.2 30 PT, aPTT and D-Dimer: high 6.7/4.6 904 1.8/0.7 130 78 402 63 1.2
13 6.5 6.4 90.7 28.6 17.3 103 2.8 40 N 6.6/3.5 258 4.4/1.4 11 10 1097 26 0.5
14 9.1 12.6 87.9 30.0 14.6 48 0.8 10 ND 6.2/3.0 630 1.1/0.4 41 25 144 29 0.9
15 3.3 3.3 89.3 30.2 17.7 24 0.4 95 PT, aPTT and D-Dimer: high 5.4/2.6 777 1.8/1.6 43 18 82 161 1.6
16 16.2 8.7 92.1 30.3 17.8 23 ND 85 ND 5.9/2.3 708 3.3/1.6 41 18 156 114 2.3
17 37.1 6.6 80.5 26.2 21.7 76 3.6 94 PT and D-Dimer high, aPTT: N 7.3/3.3 1127 1.2/0.5 70 48 1225 23 1.0
18 37.4 8.5 91.8 28.9 16.0 18 2.2 30 PT, aPTT and D-Dimer: high 6.1/3.9 393 1.7/0.6 27 36 1507 17 0.5
19 17.1 8.4 91.0 31.5 16.9 54 4.2 47 PT high D-Dimer upper limit, aPTT:N 6.2/3.8 1290 0.7/0.3 68 75 638 37 0.5
*Normal ranges of serum chemical parameters: Total protein: 6.4–8.8 g/dl; Albumin3.0–5.5 g/dl; LDH: 190–380 U/l; Total bilirubin: 0.2–1.1 mg/dl; Direct bilirubin:0.0–0.4 mg/dl; AST: 0–40 U/L; ALT: 0–43 U/L; ALP: 27–147 U/L; Urea: 15–50 mg/dl; Creatinine: 0.5–1.6 mg/dl; PT: prothrombin time; aPTT: active partial thromboplastin time; FDP: fibrin degradation products ND: not done; N: normal
Discussion
Although bone marrow metastases are frequently encountered in patients with disseminated solid tumors, it is not a common event as a presenting sign. It stems from two possible reasons: i) in general, bone marrow is seldom the sole site of systemic involvement by malignant disease and ii) bone marrow examination in evaluating patients with suspected malignancy has a limited value unless supported by some other clinical finding, such as a leukoerythroblastic reaction [1,3,4]. Although there are a number of correlates that may be useful, there is no single specific finding that is an indicator of marrow infiltration. Furthermore, in most patients with neoplastic infiltration of the marrow, the peripheral blood findings do not differ significantly from those without marrow involvement. Our series showed that MAHA, LEB and unexplained cytopenias are strong indicators of the necessity of bone marrow examination. When we reviewed routine hematologic and biochemical parameters we found that only four were present in all patients: anemia, thrombocytopenia, elevated RDW and hypoproteinemia. However, our results should be interpreted with caution, as the current study was not designed to determine the predictive parameters of marrow metastases.
LEB is the term used to describe the combination of nucleated red cells (erythroblasts) and immature myeloid precursors (e.g. myelocytes and myeloblasts) in the peripheral blood film. The mechanism of leukoerythroblastic reactions is not defined. The invasion of metastatic cancer cells may cause the early release of some cytokines, leading to the development of a myelophthisic blood picture even before the marrow is completely replaced. This is a possible explanation for some instances of leukoerythroblastic changes in the blood of the patients with tumors in whom marrow metastases are not documented by histologic examination. Although metastatic foci can be found in a high percentage of some carcinomas, the development of frank LEB occurs much less frequently, so its absence should not exclude marrow involvement. If there is LEB in a case of suspected malignancy, bone marrow examination should be considered. Our series confirmed this judgment, because fifteen of the patients were presented with LEB. On the other hand despite clear bone marrow involvement in 4 patients there were no erythroid and myeloid precursors in their peripheral blood films (patients # 10, 11, 14, 15). LEB was reported at different rates in different series consisting of cancer patients with bone marrow metastases: 10/27 [5]; 19/25 [6]; 5/63 [7]; 26/73 [2]. Our relatively high LEB ratio (15/19) is not comparable with other series because they consisted of cancer patients who were diagnosed before bone marrow examination. Our high LEB ratio, when considered along with the bad outcomes, may reflect the late and advanced cases. In recent years there is a scarcity of publications on the association of myelophthisis with cancer. As mentioned by the authors in a recent report [8], early diagnosis and more effective therapies are possible explanations for this decrease.
MAHA or thrombotic microangiopathy (TM) describes the association of hemolytic anemia with red cell fragmentation caused by microangiopathy mechanically. Cancer related thrombotic microangiopathy (CR-TM) is a rare and severe complication; it usually occurs in the late or terminal stage of cancer with a short-term life-threatening prognosis [9,10]. CR-TM shares certain clinical similarities with thrombotic thrombocytopenic purpura such as neurological and renal impairment; also both are characterized by circulating platelet aggregates containing ultra large multimers of Von Willebrand factor (VWF). A recent report showed no VWF cleaving protease deficiency in a patient with metastatic adenocarcinoma of the colon and microangiopathic hemolysis, who is refractory to plasma exchange [11]. We do not have any data about protease activity but, because of an almost identical clinical picture with TTP as presentation, we started plasma exchange immediately in patient # 1 and # 2 but therapy failed as expected. Hemolysis, in our CR-TM patients, was so severe that several units of transfusion per day were required to maintain a safe hemoglobin level. Reticulocytosis is expected but this finding could not be a reliable indicator of hemolysis, as seen in our patients, because of marrow infiltration or chronic disease. In our series, seven out of eight MAHA patients showed LEB and six showed abnormality in some coagulation tests suggesting disseminated intravascular coagulation. When we reviewed our eight MAHA cases there were two TUO and one prostate carcinoma. All definitive stomach carcinomas and a probable gastrointestinal carcinoma were in the MAHA group. The survivals of our MAHA patients were between 11 and 51 days except for one (patient # 18; prostate carcinoma). This suggests that once microangiopathic hemolysis is seen in a patient with disseminated carcinoma, the overall prognosis is poor, especially in stomach adenocarcinoma presented with MAHA.
"Dry tap" is a term used to describe failure to obtain bone marrow on attempted marrow aspirations. Extensive marrow fibrosis and hypercellularity have been proposed as mechanisms to account for the inability to withdraw marrow by aspiration [12,13]. Because it can be attributed to faulty technique it should only be used retrospectively after review of the biopsy. We have two dry taps (patients # : 4, 6). If the definition of dry tap includes cases in which material is obtained but no, or inadequate marrow cells found in films we have four additional cases (patients # 1, 2, 3, 13). We concluded that bone marrow biopsy is certainly indicated whenever aspiration results in an insufficient material especially in the presence of LEB, MAHA, cytopenias and an elevated serum LDH.
Classically, marrow biopsy is unequivocally the best method of detecting lymphomatous involvement because in approximately one-third of cases, the aspirate is unremarkable and the biopsy shows tumor. In solid tumors other than those of lymphomatous origin the data in the literature comparing the relative value of bone marrow aspiration and biopsy in detecting marrow involvement are not so conclusive. [2,14]. Naturally trephine biopsies have a definitive advantage over aspirates in cases of dry tap. Additionally histologic sections may allow classification of the type of tumor cells, and this is of particular value in the investigation of a patient with a malignancy of unknown primary site. Aspirations were diagnostic in 12 patients in our series. The other attempts yielded no, or inadequate material. Cytological diagnoses could not be confirmed in one patient because of the patient's objection to biopsy procedure. Only in one instance was cytology superior to biopsy in first examination (a retrospective examination of biopsy confirmed the cytological finding in this patient). Although biopsies have some advantages, tumor cells occasionally can be seen in aspirate preparations when biopsy sections are normal as mentioned in literature, these two procedures should therefore be regarded as complementary.
The management of a patient, whose solid malignancy is disclosed from bone marrow, depends on his/her primary tumor. The pathologist can assist the clinician by thoroughly examining the histologic specimen. In some cases an immunoperoxidase staining for organ-specific antigen examination might be sufficient to bring out of the primary focus. As an example, an immunoperoxidase stain for prostatic acid phosphatase or prostate-specific antigen may be helpful in establishing a diagnosis of metastatic prostate cancer. On the opposite side of the spectrum in some cases the pathologist could not comment on the primary site because of a poorly differentiated tumor. Maximum effort should be made to minimize the target. In these tumors as a first step leukocyte common antigen may be used to differentiate lymphohematopoietic neoplasm from other cancers. Patient # 7 and 10, who were reported recently elsewhere separately [15,16], are good examples for bringing out the primary focus by extra stainings. The examination of patients with bone marrow metastases of unknown origin should focus on detecting treatable primary tumors [17]. This work up may result in certain improvements in survivals of patients with prostate carcinoma. Bone marrow metastases commonly arise from lung, breast and prostate cancers; therefore, in case of no clinical sign a reasonable work up should include a chest roentgenogram, a prostate examination with serum prostate specific acid phosphotase (PAP) in men; a breast examination and mammography in women; computed tomography scans also should be performed in suspected cases. Serum tumor markers are not so useful in many cases except PAP; considering their less specificity it is no surprising. Because the expected survival of many patients is quite short, laboratory and imaging studies should be considered in view of their rationality.
Conclusion
In conclusion, MAHA, LEB and unexplained cytopenias are strong indicators of the necessity of bone marrow examination. Anemia, thrombocytopenia, elevated RDW and hypoproteinemia form a uniform tetrad in patients with disseminated tumors that are diagnosed via bone marrow aspiration and biopsy. Because of the very short survival of many patients, all investigational procedures should be focused on treatable primary tumors.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
FO, RA and VO participated in the conception and design of study, acquisition of data, analysis and interpretation of data. TO, UO, HO and EK participated in acquisition of data and drafting the article. TE, OY and AT participated in revising it critically for important intellectual content. All authors read and approved the final manuscript.
Pre-publication history
The pre-publication history for this paper can be accessed here:
==== Refs
Papac RJ Bone marrow metastases Cancer 1994 74 2403 2413 7922993
Mohanty SK Dash S Bone marrow metastasis in solid tumors Indian J Pathol Microbiol 2003 46 613 616 15025356
Burkhardt R Frisch B Bartl R Kettner G Schlag R Hill W Detection of haematologic and nonhaematologic cancer by bone biopsy Cancer Detect Prev 1981 4 619 627 7349829
Sar R Aydogdu I Ozen S Sevinc A Buyukberber S Metastatic bone marrow tumours: a report of six cases and review of the literature Haematologia (Budap) 2001 31 215 223 11855783 10.1163/15685590152763755
Bezwoda WR Lewis D Livini N Bone marrow involvement in anaplastic small cell lung cancer. Diagnosis, hematologic features, and prognostic implications Cancer 1986 58 1762 1765 3019512
Tritz DB Doll DC Ringenberg QS Anderson S Madsen R Perrt MC Yarbro JW Bone marrow involvement in small cell lung cancer. Clinical significance and correlation with routine laboratory variables Cancer 1989 63 763 766 2536586
Campling B Quirt I DeBoer G Feld R Shepherd FA Evans WK Is bone marrow examination in small cell lung cancer really necessary? Ann Intern Med 1986 105 508 512 3019199
Makoni SN Labor DA Clinical spectrum of myelophthisis in cancer patients Am J Hematol 2004 76 92 93 15114608 10.1002/ajh.20046
Lin YC Chang HK Sun CF Shih LY Microangiopathic hemolytic anemia as an initial presentation of metastatic cancer of unknown primary origin South Med J 1995 88 683 687 7777893
Nordstrom B Strang P Microangiopathic hemolytic anemias (MAHA) in cancer. A case report and review Anticancer Res 1993 13 1845 1849 8267391
Forman RB Benkel SA Novik Y Tsai HM Presence of ADAMTS13 activity in a patient with metastatic cancer and thrombotic microangiopathy Acta Haematol 2003 109 150 152 12714826 10.1159/000069291
Humpries JE Dry tap bone marrow aspiration: Clinical significance Am J Hematol 1990 35 247 250 2239919
Hyun BH Bone marrow examination: Adventures in diagnostic hematology Yonsei Med J 1986 27 100 105 3751124
Atac B Lawrence C Goldberg SN Metastatic tumor: The complementary role of the marrow aspirate and biopsy Am J Med Sci 1991 302 211 213 1928232
Ali R Ozkalemkas F Ozcelik T Ozan U Ozkocaman V Tunali A Small cell lung cancer presenting as acute leukaemia Cytopathology 2005 16 262 263 16181315 10.1111/j.1365-2303.2005.00233.x
Ali R Ozkalemkas F Ozan U Ozcelik T Ozkocaman V Filiz G Manavoglu O Tunali A Rhabdomyosarcoma of the perianal region presenting as acute leukemia Ann Hematol 2004 83 729 730 15300408 10.1007/s00277-004-0912-5
Ringenberg QS Doll DC Yarbro JW Perry MC Tumors of unknown origin in the bone marrow Arch Intern Med 1986 146 2027 2028 3767548 10.1001/archinte.146.10.2027
|
16262899
|
PMC1310632
|
CC BY
|
2021-01-04 16:39:12
|
no
|
BMC Cancer. 2005 Nov 1; 5:144
|
utf-8
|
BMC Cancer
| 2,005 |
10.1186/1471-2407-5-144
|
oa_comm
|
==== Front
BMC BioinformaticsBMC Bioinformatics1471-2105BioMed Central London 1471-2105-6-2701628392310.1186/1471-2105-6-270Research ArticleCommensurate distances and similar motifs in genetic congruence and protein interaction networks in yeast Ye Ping [email protected] Brian D [email protected] Forrest A [email protected] Joel S [email protected] The Department of Biomedical Engineering, The Johns Hopkins University, Baltimore, MD 21218, USA2 The High-Throughput Biology Center, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA3 The McKusick-Nathans Institute of Genetic Medicine, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA4 The Department of Molecular Biology and Genetics, The Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA2005 9 11 2005 6 270 270 4 4 2005 9 11 2005 Copyright © 2005 Ye et al; licensee BioMed Central Ltd.2005Ye et al; licensee BioMed Central Ltd.This is an Open Access article distributed under the terms of the Creative Commons Attribution License (), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Background
In a genetic interaction, the phenotype of a double mutant differs from the combined phenotypes of the underlying single mutants. When the single mutants have no growth defect, but the double mutant is lethal or exhibits slow growth, the interaction is termed synthetic lethality or synthetic fitness. These genetic interactions reveal gene redundancy and compensating pathways. Recently available large-scale data sets of genetic interactions and protein interactions in Saccharomyces cerevisiae provide a unique opportunity to elucidate the topological structure of biological pathways and how genes function in these pathways.
Results
We have defined congruent genes as pairs of genes with similar sets of genetic interaction partners and constructed a genetic congruence network by linking congruent genes. By comparing path lengths in three types of networks (genetic interaction, genetic congruence, and protein interaction), we discovered that high genetic congruence not only exhibits correlation with direct protein interaction linkage but also exhibits commensurate distance with the protein interaction network. However, consistent distances were not observed between genetic and protein interaction networks. We also demonstrated that congruence and protein networks are enriched with motifs that indicate network transitivity, while the genetic network has both transitive (triangle) and intransitive (square) types of motifs. These results suggest that robustness of yeast cells to gene deletions is due in part to two complementary pathways (square motif) or three complementary pathways, any two of which are required for viability (triangle motif).
Conclusion
Genetic congruence is superior to genetic interaction in prediction of protein interactions and function associations. Genetically interacting pairs usually belong to parallel compensatory pathways, which can generate transitive motifs (any two of three pathways needed) or intransitive motifs (either of two pathways needed).
==== Body
Background
A powerful tool to dissect the genetic buffering contributing to robustness of an organism is to identify gene pairs whose individual mutants are viable, but whose double mutants are lethal or exhibit reduced fitness [1,2]. These are particular types of genetic interactions, which more generally indicate that the phenotype of a double mutant differs from that expected from the phenotypes of the single mutants. Other types of genetic interaction include epistasis (an anticipated combined effect is not observed) and suppression (a defect is rectified by a second mutation). For convenience, we use genetic interaction henceforth to refer specifically to synthetic lethal and synthetic fitness genetic interactions.
Genetic interaction partners have been described as acting either in parallel compensating pathways, or in the same essential process [2]. Through revealing gene redundancy and compensating pathways, genetic interaction has contributed to the understanding of gene functions as well as the networks and pathways in which gene products participate [3-6]. It is also highly relevant to understanding genetic instability and variation occurring in various human diseases [2].
While a genetic interaction indicates that genes have compensating function, it does not necessarily indicate that the gene products work in the same pathway, for example as indicated by biochemical, physical interactions between proteins. Protein interactions can indicate correct network topology by linking proteins within the same biological pathway. The recent availability of high-throughput genetic interaction screens [3-6] and protein interaction screens [7-10] for the model organism Saccharomyces cerevisiae (budding yeast) provides a unique opportunity to investigate the genetic interaction network and protein interaction network both individually and jointly. Genetic interactions often reflect functional relationships that reach far beyond local protein interactions. Protein interaction data from high-throughput approaches are known to include false positive as well as physiologically relevant observations. It is critical to understand the correlations between genetic and protein interactions, as information derived from these two types of networks can provide complementary views for developing our understanding of how genes function in specific biological pathways, and how failures of these pathways lead to pathologic conditions that are relevant to the occurrence and progression of human diseases.
Graph theoretic approaches have been applied to study global properties of protein interaction networks and genetic interaction networks in yeast [6,11-22]. A few global network analyses also directly compared the genetic and protein interaction maps. It has been suggested that the current genetic interaction network is at least four times denser than the protein interaction network; genetic interactions are significantly more abundant between physically interacting proteins and the number of common genetic neighbors between two genes correlates with a known protein-protein interaction [6]. Other studies show that highly connected hubs in the protein network have a higher probability to genetically interact with each other [23], that the two-hop physical-genetic interaction is the top predictor of genetic interactions [24], and that probabilistic network models favor between-pathway explanations over within-pathway explanations for synthetic lethal genetic interactions [22].
Here, we present a global and local network investigation of the connections among genetic interaction, genetic congruence, and protein interaction networks for yeast, focusing on quantitative comparison of path length and motifs. Our results demonstrate that the genetic congruence network inferred from direct genetic interactions largely overlaps with the protein interaction network, with corresponding distances and motifs, while the genetic interaction network does not. This finding indicates that genetic congruence provides evidence for physical interaction and protein complex membership, as well as similar gene functions. The genetic congruence network we have defined can function as a mini-map to reveal network properties before the entire genetic interaction map is completed in yeast.
Results
Network overview
The genetic interactions used in this study are taken from published experiments detecting cell growth defects through screening a deletion of interest (query gene) against ~5000 viable yeast single-deletion strains (target genes) [3-6]. As only ~150 queries have been reported, the current network covers ~2% of the entire map. Therefore, many observations will be re-assessed after completion of the map. Specifically, the entire observed genetic network is expected to be symmetric when query and target genes are reversed. To account for the symmetric property of the entire genetic network, we have constructed both an asymmetric genetic network that includes all currently available high-throughput genetic interactions and the symmetric genetic network that covers interactions only between genes that have been used as queries (Fig. 1A). The graph of the symmetric genetic network is shown in Fig. 1B. Each node represents a gene, and each edge represents the genetic interaction between two connected genes. The edges are considered undirected, and we do not distinguish between edges that were detected in one or both directions. High connectivity in the symmetric genetic network (Fig. 1B) reflects that query genes were selected based on related functionality [6].
Figure 1 Schematic illustration of genetic and congruence networks. A. Asymmetric and symmetric genetic networks are represented in matrix form; filled squares represent observed genetic interactions. The symmetric network includes only genes used as queries. B. The symmetric genetic interaction network contains 126 genes. C. A congruence network was calculated from the symmetric genetic interaction network using a threshold congruence score of 6.
Previous analysis has suggested that shared genetic interaction partners correlate with physical interactions [6]. Quantitative measures for partner sharing in physical interaction networks has been defined as Mutual Clustering Coefficients (MCC) [14]. Here we use the negative Log10 of the P-value of the hypergeometric MCC as a quantitative measure of neighbor sharing in the genetic interaction network, and for convenience term it the congruence score [25]. Higher scores indicate that two genes share more genetic interaction partners than expected by chance. A genetic congruence network is then derived from introducing non-directed edges between congruent genes, using the congruence score to provide an edge weight (Fig. 1C). Asymmetric and symmetric congruence networks have been constructed from the corresponding genetic networks, respectively. A P-value of 0.01 for shared genetic interaction partners after correcting for multiple testing corresponds to a congruence score of 8 for the congruence network derived from the asymmetric genetic interactions and a congruence score of 6 for the network derived from the symmetric genetic interactions.
The protein interaction network we used is derived from ~45,000 protein-protein interactions compiled from the large-scale yeast two-hybrid and mass spectrometry analyses [7-10]. Each interaction has been assigned with a confidence score that corresponds to the network edge weight. The confidence score represents the probability that two proteins interact with each other [12].
The size and global topological measures for genetic, congruence, and protein networks are provided (Table 1). The average degree is the number of edges per node, and the clustering coefficient measures the interconnectivity around a node. Interestingly, average degrees nearly halve but clustering coefficients double from genetic networks to congruence networks. The values for the protein network are in between those for genetic and congruence networks. These suggest that the congruence network tends to be highly clustered. We quantitatively demonstrate with the following results on path lengths and local motifs that the inferred congruence links from shared patterns of genetic interactions have greater relevance to protein interactions than underlying direct genetic interactions.
Table 1 Standard global topological measures describing network structure. Detailed analyses on path lengths and local motifs are described in Fig. 2 and 3.
Asymmetric genetic network Symmetric genetic network Asymmetric congruence network Threshold = 8 Symmetric congruence network Threshold = 6 Protein network Threshold = 0.5
No. of nodes 1004 111 122 61 3208
No. of edges 3799 813 267 146 13038
Average degree 7.6 14.6 4.4 4.8 8.1
Average clustering coefficient 0.10 0.37 0.73 0.84 0.45
Network distances
Conventional analysis shows genetically interacting genes encode proteins in the same complex more often than would be expected by chance [6]. Because physical associations and genetic interactions each report on functional similarity, we might naively expect that physical and genetic links should be correlated. However, it has also been recognized that the number of genetic interaction pairs having direct physical interaction is a small fraction of the total number of genetic interaction pairs (~1%) [6]. Therefore, given currently known genetic and protein interactions and their overlap, the majority of genetic interactions do not connect physical partners.
To quantitatively study the global relationships between genetic and physical interactions, we calculated the shortest path length for any two genes in the genetic interaction network and the shortest path length for corresponding gene products in the protein interaction network, and then compared these two path lengths. Our results reveal that most protein pairs are distributed 3–4 links apart in the protein interaction network, regardless of whether there is a genetic interaction between the gene pair (Fig. 2A). This indicates that characteristic path lengths in genetic and physical interaction networks are incommensurate. Results are similar using symmetric and asymmetric genetic networks. These observations support the concept that genetically interacting pairs usually have no direct physical interactions. If we define pathways by the context of physical interactions and assume genes with physical interactions function in a single pathway and without physical interactions act in parallel pathways, then our results suggest that genetically interacting genes are more likely to belong to parallel compensating pathways. Other groups have used similar reasoning to identify components of pairs of complementary pathways from joint analysis of physical and genetic interactions [22].
Figure 2 Path length comparison for genetic, congruence, and protein networks. A. There is little correlation between short paths in the genetic interaction network and short paths in the protein interaction network. B. Protein interaction confidence increases with congruence score. C. The path length in the protein network decreases monotonically with the congruence score. D. High-scoring paths in the congruence network are correlated with short distances in the protein interaction network, indicating that these networks are commensurate. Results are displayed for the observed and randomized networks. Error bars indicate one standard error. The random value if present is comparable to the observed value (P-value > 0.05).
We asked whether the other view of genetic interactions, i.e. genetic congruence, might yield improved concordance with physical interactions. We first explored the relationship between pair-wise genetic congruence versus direct physical interaction. High-throughput physical interaction data sets are known to include many false-positives, which can confound analysis. Confidence scores have been developed to reflect the probability that a physical interaction is a true-positive [12]. We observed that protein interaction confidence increases with the congruence score (Fig. 2B). Above the congruence score of 8 and 6, which corresponds to the network P-value of 0.01 for the asymmetric and symmetric networks respectively, all protein pairs exhibit high confidence interactions with confidence score greater than ~0.8. This implies that genetic congruence acts as an indication of high-confidence protein interactions. It is notable that information from a purely genetic experiment correlates well with information from a purely proteomic experiment. We also used receiver operating characteristic (ROC) curves to assess the relationship of congruence scores and physical interactions. ROC curves for asymmetric and symmetric congruence scores both climb rapidly away from the origin with high true positive rates and low false positive rates [see additional file 1, supp. fig. S2]. According to the area under the curve, the congruence score from the symmetric network performs better than the score from the asymmetric network, but at the cost of making fewer predictions. This is in agreement with the result from Fig. 2B that congruence scores of the symmetric network predict higher confidence physical interactions as compared with those of asymmetric network. The reason for the differences may be due to biased selection of query genes, as the symmetric network only contains query genes and all query genes were selected from a few related biological processes [6].
We further investigated the pair-wise congruence in the context of the protein interaction network. Our results show that the shortest path of physical interactions between congruent pairs decreases from ~3.6 links to 1 link (direct physical interaction) with increasing of congruence score (Fig. 2C). The path length transition begins when the congruence score increases beyond 8 and 6 for asymmetric and symmetric congruence networks, respectively. Once the score reaches 21 and 20 for asymmetric and symmetric networks, the congruent gene encoded proteins coincide with known direct physical interactions (4 pairs with congruence score = 21 in the asymmetric network and 1 pair with congruence score = 20 in the symmetric network).
Finally, to explore the connection between the congruence network and the protein network, we computed the highest score path for any two genes in the congruence network. Edge weights are in the range of 0 and 1 generated by applying a sigmoid function to the congruence scores (see Methods). The higher the path score, the higher probability two genes share similar genetic interaction partners. When comparing the highest path score in the congruence network with the shortest path length in the protein interaction network, we observed that the physical distance decreases monotonically from the average path length ~3.6 links to 1 (direct physical interaction) as the highest path score increases in both asymmetric and symmetric congruence networks (Fig. 2D). Therefore, transitive genetic congruence is commensurate with physical distances, which is similar to direct genetic congruence (Fig. 2C).
Network motifs
Network motifs represent significantly recurrent patterns of simple interactions in complex networks [17]. Comparison of local structures in the network can help reveal the connections among superficially unrelated biological or social networks [18]. Additionally, the local structure of the network contributes to the understanding of global organization of the network [16]. To contrast the local structure of three types of networks, we counted the abundance of non-directed triads and tetrads in genetic, congruence, and protein networks. The random networks used to detect tetrads were generated to preserve the same triad counts as the real network [18].
More significantly, we can determine network transitivity through the observation of whether a transitive or intransitive motif is enriched or depleted in the network. Transitivity is a common network property that interactions of A-B and B-C imply elevated probability of interaction of A-C. We developed a characteristic, termed the motif transitivity score (MTS), as a quantification of the motif transitivity [see Methods and additional file 1, supp. table S1]. The positive values indicate transitive motifs while the negatives represent intransitive motifs. The network transitivity has been quantified by the clustering coefficient before [26,27], which is closely related to the motif transitivity score defined here. We have found good agreement between motif enrichments (Fig. 3A) and average clustering coefficients (Table 1), i.e. congruence and protein networks are more clustered compared with the genetic network.
Figure 3 Motif characterization for genetic, congruence, and protein networks. A. Both transitive and intransitive motifs are enriched in the genetic network, tetrad4 and tetrad6 for the asymmetric network and triad2, tetrad1 and tetrad4 for the symmetric network. Only transitive motifs are enriched in congruence and protein networks, triad2 and tetrad6 for symmetric and asymmetric congruence networks, triad2, tetrad3, and tetrad6 for the protein network. Motif enrichment criteria are as defined in [17] (see Methods). B. The connections between triangle and square motifs in the symmetric genetic network. Three types of relationships exist between triangles and squares and the percentage of each scenario is labeled. The red numbers indicate individual pathways.
When using the asymmetric genetic and congruence networks for comparison with the physical network, the pattern of enriched motifs (the relative motif ratio) is significantly correlated for congruence and protein interaction networks (Pearson correlation coefficient R = 0.76, P-value = 0.03), and these are anti-correlated with the enriched motifs for direct genetic interactions (R = -0.66, P-value = 0.08; R = -0.69, P-value = 0.06, respectively) (Fig. 3A). This is consistent with the above global distance analysis, supporting significant overlap between congruence and protein networks.
The enriched motifs in asymmetric congruence and protein networks are all transitive, including triad2 (triangle motif) and tetrad6. The triangle motif is the most significantly enriched motif, suggesting the transitive property of congruence interactions and physical interactions. This result is in agreement with our observation in the previous section that the transitive congruence is correlated with short physical distance (Fig. 2D). The asymmetric genetic interaction network, however, consists of both intransitive motif tetrad4 (square motif) and transitive motif tetrad6, with the square motif as the most enriched structure.
The detection of intransitive motifs in the asymmetric genetic network may be due to the artifact that the interactions have not yet been tested. It does not necessarily mean that these interactions do not exist. To overcome this bias, we repeated motif-finding procedure using the symmetric genetic network and corresponding congruence network (Fig. 3A). The pattern of enriched motifs is still significantly correlated for symmetric genetic congruence and protein interaction networks (Pearson correlation coefficient R = 0.73, P-value = 0.04), but these are insignificantly correlated with those for the symmetric genetic network (R = 0.29, P-value = 0.49; R = 0.10, P-value = 0.82, respectively). The enriched motifs in the symmetric congruence network remain the same as for the asymmetric congruence network, i.e. all transitive motifs, triad2 (triangle motif) and tetrad6.
A final concern is that the transitive motifs arise from the mathematical process of generating congruence scores: if genes A and B share synthetic lethal partners, and B and C share partners, then A and C may have an increased probability of sharing partners. To address the question, we followed the following protocol [see additional file 1, supp. fig. S3]: (1) Randomize the genetic interaction network. (2) Calculate congruence scores for gene pairs in the randomized network. (3) Set a threshold and calculate motif enrichment for the random congruence network. We repeated this process 100 times for both the symmetric and the asymmetric genetic interaction networks. The typical extreme value for the maximum congruence score observed was 5 for the symmetric network and 6 for the asymmetric. Thus, applying the same cutoff for congruence scores as in the actual network, 6 for symmetric and 8 for symmetric, typically rejects all the congruence edges in the randomized network. We reduced the thresholds to retain the same number of congruence edges as in the actual network, with mean values of 1.8 (symmetric) and 3.2 (asymmetric) over the 100 randomizations. The average clustering coefficient is significantly smaller in the random networks than the actual network: 0.23 vs. 0.84 (random vs. actual symmetric, P-value 10-402), and 0.12 vs. 0.73 (random vs. actual asymmetric, P-value 10-933). Although the transitive motif triad2 (triangle) is enriched in the random congruence network relative to a fully random network, the motif count is far below that observed in the actual congruence network [see additional file 1, supp. table S3]. Other patterns of motif enrichment are quite different: tetrad4 (square motif, intransitive) is enriched in the random congruence network and depleted in the actual network, and tetrad6 (4-clique, transitive) is enriched in the actual network but not in the random network [see additional file 1, supp. fig. S4]. The transitive motifs in the congruence network are therefore enriched significantly beyond what would be expected based solely on the method of defining congruence edges.
Both transitive and intransitive motifs are still detected in the symmetric genetic interaction network. However, the types are different from those in the asymmetric genetic network. The transitive triangle motif becomes the most enriched structure in the symmetric genetic network, in agreement with a previous study that genetic interaction partners of a gene have an increased likelihood to interact with each other [24]. One source of the triangle motif could be the requirement for any two of three pathways for viability. Notably, the square motif is still highly enriched in the symmetric genetic network despite the abundance of the triangles, indicating that the square motif will remain enriched when the complete genetic interaction network is determined.
The view from recent studies indicates that high clustering is a generic feature of biological networks, as exemplified by protein interaction and protein domain networks [13]. However, we find that the genetic interaction network has both transitive and intransitive motifs. The coexistence of triangle and square motifs in the genetic network suggests two scenarios for genetic interactions between pathway components. In one scenario, genetic interactions between two pathways generate a square motif. Each edge crosses between the pathways, and genes at opposite corners are in the same pathway. In the second scenario, any two of three pathways are required for viability. Genetic interactions cross between all three pathways, generating the triangle motif.
To further answer the question whether the enriched triangles and squares overlap with each other or are excluded from each other, we compared the members of triangle and square motifs in the symmetric genetic network (Fig. 3B). Results show that one-node sharing is the dominant scenario (76%) for triangles and squares. Assuming three pathways for the triangle motif and two pathways for the square motif, the one-node sharing case defines four parallel pathways with one shared by the square and triangle. Two-node sharing accounts for 22% of total possibilities, and suggests three parallel pathways with two shared by the triangle and square. Only 2% of total cases are the complete overlap of the triangle and square, which is in an agreement with our observation that tetrad5 is not an enriched motif in the symmetric genetic network (Fig. 3A).
Because the completed genetic interaction map will necessarily be symmetric (except for false-positives or false-negatives), the enriched motifs in the symmetric genetic network are more relevant than the enriched motifs in the asymmetric genetic network.
Biological relevance
Correct interpretation of the relationship between genetic and protein interactions enables interesting biological predictions. As we have demonstrated in previous sections, genetic congruence and protein networks are similarly organized with corresponding distances and motifs. Then, we would expect that two genes closer in the congruence network have higher tendency to physically interact with each other, reside within one protein complex, and involve in similar biological process.
To validate this prediction generally, we plotted protein complex membership versus the distance in the genetic network and the path score in the congruence network (Fig. 4A). The probability of co-residence in a protein complex increases with the congruence path score, and scores greater than 0.7 indicate same protein complex membership. On the other hand, gene products binned by distance in the genetic interaction network have uniformly low probability of protein complex co-residence.
Figure 4 Congruence network but not genetic network predicts protein complex membership and functional association. A. Short distance in congruence network implies protein complex membership. B. Close distance in congruence network suggests similar function. GO [28] hierarchy depth is normalized to the range of 0 and 1 by [depth-min(depths)]/[max(depths)-min(depths)], where depths are calculated for each GO category, biological process and molecular function. As distance results are similar for symmetric and asymmetric networks, we only present those for the symmetric network.
Physical interactions usually suggest functional association. Accordingly, we asked whether congruence also indicates functional connection besides physical connection. As an initial validation, we found that genes close in the congruence network share similar functional annotations recorded in the database of Gene Ontology (GO)[28], i.e. biological process and molecular function (Fig. 4B). Moreover, the functional similarity is consistently higher for gene pairs based on path score in the congruence network than based on distance in the genetic network.
An example of congruence coinciding with protein interaction and function association is the prefoldin complex, which includes PAC10, GIM3, GIM4, GIM5, and YKE2. These five genes are clustered in the congruence network and the average path score between any two members of this complex is 0.51 (Fig. 1C). They are all chaperone proteins forming a complex, which promotes efficient protein folding [29,30].
Discussion
We have demonstrated that high genetic congruence implies high probability of a physical interaction and short distance in the physical interaction network. Short distances in the congruence network (measured by a high path score), but not in the genetic network, are commensurate with distances in the protein network. To account for false-positives in the high throughput protein interaction datasets, parallel analyses were performed using a protein network with edges weighted according to interaction confidence [see additional file 1, supp. fig. S5], and the results were similar to those obtained from the un-weighted protein network. A guide to the figures showing path length comparisons among genetic, congruence, and protein networks is provided [see additional file 1, supp. table S4]. Local structure indicates similar transitive motif enrichment in congruence and protein networks, while the genetic network significantly consists of transitive as well as intransitive motifs. Both global distance analysis and local motif analysis demonstrate that the genetic congruence network possesses similar network transitivity to the protein network.
The similarity between congruence and protein networks and the dissimilarity between genetic and protein networks have yielded three interesting conclusions with biological significance. First, we have demonstrated that significant genetic congruence correlates strongly with protein complex membership and functional association. Second, genetically interacting pairs usually belong to compensatory pathways without direct physical interactions. Finally, the coexistence of triangles and squares in the genetic network indicates that robustness may be due to two pathways that compensate each other (squares), or three pathways any two of which are needed (triangles).
While the protein interaction and genetic congruence networks exhibit a high degree of similarity, we do not expect them to be identical because they are based on distinctly different experimental measures. The protein interaction network is based on protein binding constants in cellular extracts under selective precipitation conditions [7,8] or within cells through over-expression of tested proteins [9,10]. The congruence network is based on growth defects exhibited by cells lacking a pair of gene products cultured under standard conditions[6]. Thus, high congruence may not necessarily indicate a physical interaction. The concordance we observed between congruence and protein interaction network structures provides strong support for the argument that they both faithfully reflect biologically relevant network relationships.
The conclusions drawn from our study are limited by the current coverage of genetic and protein networks. This is especially true for the genetic network, which is at low coverage. Moreover, the current genetic network is biased by query gene selection. The ~150 query genes all have relative large numbers of interaction partners and related functionality[6]. As the coverage and symmetric property are increased, we expect that the average degree and average clustering coefficient will decline. Network distance results are robust in response to changes in genetic network symmetry and protein network edge weight. The symmetric genetic network has been used for motif counting and the relative motif ratio is insensitive to network size[18]. Therefore, we believe that our conclusions on network distances and motifs should continue to hold as the entire genetic interaction network is mapped.
Conclusion
In summary, we have demonstrated that genetic congruence is superior to genetic interaction in predicting protein interactions and within-pathway functional associations. In contrast, genetic interaction pairs usually act in parallel compensatory pathways. Motif study indicates that genetic interactions bear both transitive and intransitive characters. Consideration of the symmetric property of a complete genetic interaction network is crucial to determination of motif enrichment for the genetic network.
Methods
Genetic interaction networks
The genetic interaction dataset is derived from a recent high throughput study in budding yeast [6]. The interaction is detected by cell growth defect through introducing a deletion of interest (query gene) into all viable yeast single-deletion strains (target gene). Interactions derived from 6 essential query genes, including MYO2, SCC1, CDC2, CDC7, CDC42, and CDC45 were removed in our study because phenotypes exhibited by conditional alleles of essential genes may include loss of function, unregulated function, and gain of function, while null alleles of nonessential genes are by definition solely loss of function mutations. Results and conclusions do not change, however, when these 6 essential genes are included in the analysis.
We constructed two types of genetic networks. The asymmetric genetic network includes currently available high throughput genetic interactions, i.e. 3799 genetic interactions between 126 non-essential query genes and 982 target genes. The symmetric genetic network only contains interactions between query genes, i.e. 813 genetic interactions between 108 non-essential query genes and 104 target genes that have been used as queries.
Randomization of genetic interactions
Genetic interactions from the high throughput study [6] were reported as an interaction between the query gene and the target gene. A randomized network was generated by keeping the query gene list unchanged, randomly matching one of the target genes according to the probability of each target gene shown in the interaction list with replacement. Duplicate query-target pairs and self-interaction pairs, which are not possible in the experimental networks, were rejected during randomization. Results depict the average over 1000 randomizations.
Genetic congruence networks
The congruence score was defined as -log10 [hypergeometric P - value (x ≥ kobs], and hypergeometric P−value (x≥kobs)=∑x=kmin(m,n)C(m,x)C(t−m,n−x)/C(t,n)
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@6AB8@, where two target genes having m and n genetic interaction partners share x partners from a list of t query genes, and C(j,k) is the binomial coefficient j!/k!(j-k)! [25]. Related measures have been used to analyze protein interaction networks to predict protein-protein interactions [14]. The congruence score is calculated for every target gene pair in the symmetric and asymmetric genetic networks. The symmetric and asymmetric congruence networks are derived from the corresponding genetic networks, respectively. The distribution of network size over different congruence scores is provided [see additional file 1, supp. fig. S1]. The congruence score of 8 (122 nodes with 267 edges) for asymmetric congruence network corresponds to the network P-value of 0.01 after correction for multiple testing of per-link P-value 0.01/9822 = 10-8. Similarly, the congruence score of 6 (61 nodes with 146 edges) is the cutoff value for the symmetric congruence network.
Protein interaction network
We used 47,783 protein-protein interactions with confidence scores [12] compiled from the large-scale two-hybrid data sets of protein-protein interactions [9,10] and mass spectrometry analysis of protein complexes [7,8]. The distribution of network size over different confidence scores is provided [see additional file 1, supp. fig. S1].
Network distances
The shortest path distance was counted for any two nodes in the un-weighted genetic interaction and protein interaction networks. The shortest path length is the sum of lengths of individual linkage.
The SEEDY algorithm [31] was used to compute highest score path distance for the weighted genetic congruence and protein interaction networks. The highest score path is the path with the maximal value of the product of edge weights. Disconnected components are ignored for both shortest path and highest score path calculations.
The edge weight for the protein network is the confidence score (in the range of 0 and 1) [12]. The edge weight for the genetic congruence network is derived from a sigmoid function w=e(s−a)/b1+e(s−a)/b
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=vr0dc8meaabaqaciGacaGaaeqabaqabeGadaaakeaacqWG3bWDcqGH9aqpdaWcaaqaaiabdwgaLnaaCaaaleqabaGaeiikaGIaem4CamNaeyOeI0IaemyyaeMaeiykaKIaei4la8IaemOyaigaaaGcbaGaeGymaeJaey4kaSIaemyzau2aaWbaaSqabeaacqGGOaakcqWGZbWCcqGHsislcqWGHbqycqGGPaqkcqGGVaWlcqWGIbGyaaaaaaaa@432F@ (in the range of 0 and 1), where s is the congruence score, a and b are parameters. The rationale of introducing the above sigmoid function is derived from the probability distribution of Pr(true positive|s) = Pr(protein interaction|s) as genes sharing genetic interaction partners usually exhibit physical association [6]. The parameters a = 15.9 and b = 1.6 are the best-fit values for the sigmoid function to form a smoothed interpolation of Pr(protein interaction|s) for the asymmetric congruence network [see additional file 1, supp. fig. S6]. Results were not sensitive to the choice of parameter values [see additional file 1, supp. fig. S7]. Similarly, a = 17.7 and b = 3.4 are the best-fit values for the symmetric congruence network.
Network motifs
We used the mfinder1.1 – network motifs detection tool to count non-directed triad and tetrad motifs in genetic interaction, genetic congruence, and protein interaction networks. Both symmetric and asymmetric genetic networks were used for motif searching. Motifs were also counted for the symmetric congruence network with cutoff value of 6, the asymmetric congruence network with cutoff value of 8, and the protein network with confidence score greater than 0.5 [12]. Motif results are insensitive to the threshold values for congruence and protein networks [see additional file 1, supp. table S2]. The Metropolis algorithm was used to conserve the number of triads in random networks for tetrad motif counting. The relative motif ratio (RMR) was calculated to represent the abundance of each motif relative to random networks in which each node has the same number of edges as the corresponding node in the real network. The formula for RMR is defined as RMR=Δi/(∑Δi2)1/2, Δi=Nreali−<Nrandi>Nreali+<Nrandi>+ε, and ε=4
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7985@. The criteria taken for enriched motifs are NrealZscore > 2, Nreal/Nrand > 1.1, Uniqueness ≥ 4 where Uniqueness is the number of times a motif appears in the network with completely disjoint groups of nodes [17,18].
To quantify the motif transitivity, we give the definition of motif transitivity score (MTS) as MTS=3×number of 'Δ' - number of 'V'3×number of 'Δ' + number of 'V'
MathType@MTEF@5@5@+=feaafiart1ev1aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacH8akY=wiFfYdH8Gipec8Eeeu0xXdbba9frFj0=OqFfea0dXdd9vqai=hGuQ8kuc9pgc9s8qqaq=dirpe0xb9q8qiLsFr0=vr0=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@7A19@, where 'Δ' is a group of 3 vertices each of which is connected to the other two, and 'V' is a group of 3 vertices only one of which is connected to the other two. The 'Δ' and 'V' are mutually exclusive subgroups in the MTS calculation. The factor of 3 accounts for the fact that each 'Δ' is equivalent to three 'V'. This formula quantifies the motif transitivity in the range from -1 to 1, and is insensitive to the motif size. The MTS is 1 for a fully connected motif, and is -1 for a motif without the triangle. The values of MTS for triads and tetrads are listed [see additional file 1, supp. table S1].
Authors' contributions
PY designed and performed the study and drafted the manuscript. JSB helped conceive the study and guided the work. BDP and FAS helped contribute ideas discussed in the manuscript. All authors read and approved the final manuscript.
Supplementary Material
Additional File 1
contains supplemental figures and tables.
Click here for file
Acknowledgements
JSB acknowledges support from NIGMS, NCRR, and the Whitaker Foundation. FAS and BDP acknowledge support from the NHGRI. BDP acknowledges support from an NIH/NIGMS training grant. The authors acknowledge the reviewers for helpful suggestions that improved the manuscript and motivated the motif enrichment calculations for congruence networks obtained from randomized genetic interaction networks.
==== Refs
Hartman JL Garvik B Hartwell L Principles for the buffering of genetic variation Science 2001 291 1001 1004 11232561 10.1126/science.291.5506.1001
Tucker CL Fields S Lethal combinations Nat Genet 2003 35 204 205 14593402 10.1038/ng1103-204
Ooi SL Shoemaker DD Boeke JD DNA helicase gene interaction network defined using synthetic lethality analyzed by microarray Nat Genet 2003 35 277 286 14566339 10.1038/ng1258
Pan X Yuan DS Xiang D Wang X Sookhai-Mahadeo S Bader JS Hieter P Spencer F Boeke JD A robust toolkit for functional profiling of the yeast genome Mol Cell 2004 16 487 496 15525520 10.1016/j.molcel.2004.09.035
Tong AH Evangelista M Parsons AB Xu H Bader GD Page N Robinson M Raghibizadeh S Hogue CW Bussey H Andrews B Tyers M Boone C Systematic genetic analysis with ordered arrays of yeast deletion mutants Science 2001 294 2364 2368 11743205 10.1126/science.1065810
Tong AH Lesage G Bader GD Ding H Xu H Xin X Young J Berriz GF Brost RL Chang M Chen Y Cheng X Chua G Friesen H Goldberg DS Haynes J Humphries C He G Hussein S Ke L Krogan N Li Z Levinson JN Lu H Menard P Munyana C Parsons AB Ryan O Tonikian R Roberts T Sdicu AM Shapiro J Sheikh B Suter B Wong SL Zhang LV Zhu H Burd CG Munro S Sander C Rine J Greenblatt J Peter M Bretscher A Bell G Roth FP Brown GW Andrews B Bussey H Boone C Global mapping of the yeast genetic interaction network Science 2004 303 808 813 14764870 10.1126/science.1091317
Gavin AC Bosche M Krause R Grandi P Marzioch M Bauer A Schultz J Rick JM Michon AM Cruciat CM Remor M Hofert C Schelder M Brajenovic M Ruffner H Merino A Klein K Hudak M Dickson D Rudi T Gnau V Bauch A Bastuck S Huhse B Leutwein C Heurtier MA Copley RR Edelmann A Querfurth E Rybin V Drewes G Raida M Bouwmeester T Bork P Seraphin B Kuster B Neubauer G Superti-Furga G Functional organization of the yeast proteome by systematic analysis of protein complexes Nature 2002 415 141 147 11805826 10.1038/415141a
Ho Y Gruhler A Heilbut A Bader GD Moore L Adams SL Millar A Taylor P Bennett K Boutilier K Yang L Wolting C Donaldson I Schandorff S Shewnarane J Vo M Taggart J Goudreault M Muskat B Alfarano C Dewar D Lin Z Michalickova K Willems AR Sassi H Nielsen PA Rasmussen KJ Andersen JR Johansen LE Hansen LH Jespersen H Podtelejnikov A Nielsen E Crawford J Poulsen V Sorensen BD Matthiesen J Hendrickson RC Gleeson F Pawson T Moran MF Durocher D Mann M Hogue CW Figeys D Tyers M Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry Nature 2002 415 180 183 11805837 10.1038/415180a
Ito T Chiba T Ozawa R Yoshida M Hattori M Sakaki Y A comprehensive two-hybrid analysis to explore the yeast protein interactome Proc Natl Acad Sci U S A 2001 98 4569 4574 11283351 10.1073/pnas.061034498
Uetz P Giot L Cagney G Mansfield TA Judson RS Knight JR Lockshon D Narayan V Srinivasan M Pochart P Qureshi-Emili A Li Y Godwin B Conover D Kalbfleisch T Vijayadamodar G Yang M Johnston M Fields S Rothberg JM A comprehensive analysis of protein-protein interactions in Saccharomyces cerevisiae Nature 2000 403 623 627 10688190 10.1038/35001009
Bader GD Hogue CW Analyzing yeast protein-protein interaction data obtained from different sources Nat Biotechnol 2002 20 991 997 12355115 10.1038/nbt1002-991
Bader JS Chaudhuri A Rothberg JM Chant J Gaining confidence in high-throughput protein interaction networks Nat Biotechnol 2004 22 78 85 14704708 10.1038/nbt924
Barabasi AL Oltvai ZN Network biology: understanding the cell's functional organization Nat Rev Genet 2004 5 101 113 14735121 10.1038/nrg1272
Goldberg DS Roth FP Assessing experimentally derived interactions in a small world Proc Natl Acad Sci U S A 2003 100 4372 4376 12676999 10.1073/pnas.0735871100
Maslov S Sneppen K Specificity and stability in topology of protein networks Science 2002 296 910 913 11988575 10.1126/science.1065103
Vazquez A Dobrin R Sergi D Eckmann JP Oltvai ZN Barabasi AL The topological relationship between the large-scale attributes and local interaction patterns of complex networks Proc Natl Acad Sci U S A 2004 101 17940 17945 15598746 10.1073/pnas.0406024101
Milo R Shen-Orr S Itzkovitz S Kashtan N Chklovskii D Alon U Network motifs: simple building blocks of complex networks Science 2002 298 824 827 12399590 10.1126/science.298.5594.824
Milo R Itzkovitz S Kashtan N Levitt R Shen-Orr S Ayzenshtat I Sheffer M Alon U Superfamilies of evolved and designed networks Science 2004 303 1538 1542 15001784 10.1126/science.1089167
Lin N Wu B Jansen R Gerstein M Zhao H Information assessment on predicting protein-protein interactions BMC Bioinformatics 2004 5 154 15491499 10.1186/1471-2105-5-154
Jansen R Yu H Greenbaum D Kluger Y Krogan NJ Chung S Emili A Snyder M Greenblatt JF Gerstein M A Bayesian networks approach for predicting protein-protein interactions from genomic data Science 2003 302 449 453 14564010 10.1126/science.1087361
Middendorf M Ziv E Adams C Hom J Koytcheff R Levovitz C Woods G Chen L Wiggins C Discriminative topological features reveal biological network mechanisms BMC Bioinformatics 2004 5 181 15555081 10.1186/1471-2105-5-181
Kelley R Ideker T Systematic interpretation of genetic interactions using protein networks Nat Biotechnol 2005 23 561 566 15877074 10.1038/nbt1096
Ozier O Amin N Ideker T Global architecture of genetic interactions on the protein network Nat Biotechnol 2003 21 490 491 12721566 10.1038/nbt0503-490
Wong SL Zhang LV Tong AH Li Z Goldberg DS King OD Lesage G Vidal M Andrews B Bussey H Boone C Roth FP Combining biological networks to predict genetic interactions Proc Natl Acad Sci U S A 2004 101 15682 15687 15496468 10.1073/pnas.0406614101
Ye P Peyser BD Pan X Boeke JD Spencer FA Bader JS Gene function prediction from congruent synthetic lethal interactions in yeast Molecular Systems Biology 2005 In press
Newman ME Strogatz SH Watts DJ Random graphs with arbitrary degree distributions and their applications Phys Rev E Stat Nonlin Soft Matter Phys 2001 64 26118
Watts DJ Strogatz SH Collective dynamics of 'small-world' networks Nature 1998 393 440 442 9623998 10.1038/30918
Ashburner M Ball CA Blake JA Botstein D Butler H Cherry JM Davis AP Dolinski K Dwight SS Eppig JT Harris MA Hill DP Issel-Tarver L Kasarskis A Lewis S Matese JC Richardson JE Ringwald M Rubin GM Sherlock G Gene ontology: tool for the unification of biology. The Gene Ontology Consortium Nat Genet 2000 25 25 29 10802651 10.1038/75556
Geissler S Siegers K Schiebel E A novel protein complex promoting formation of functional alpha- and gamma-tubulin Embo J 1998 17 952 966 9463374 10.1093/emboj/17.4.952
Vainberg IE Lewis SA Rommelaere H Ampe C Vandekerckhove J Klein HL Cowan NJ Prefoldin, a chaperone that delivers unfolded proteins to cytosolic chaperonin Cell 1998 93 863 873 9630229 10.1016/S0092-8674(00)81446-4
Bader JS Greedily building protein networks with confidence Bioinformatics 2003 19 1869 1874 14555618 10.1093/bioinformatics/btg358
|
16283923
|
PMC1310633
|
CC BY
|
2021-01-04 16:27:47
|
no
|
BMC Bioinformatics. 2005 Nov 9; 6:270
|
utf-8
|
BMC Bioinformatics
| 2,005 |
10.1186/1471-2105-6-270
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633216010.1371/journal.pbio.0040001Research ArticleAnimal BehaviorBotanyEcologyPlant ScienceFunctional Diversity of Plant–Pollinator Interaction Webs Enhances the Persistence of Plant Communities Pollinator Diversity and Ecosystem SustainabilityFontaine Colin [email protected]
1
2
Dajoz Isabelle
1
2
Meriguet Jacques
1
2
Loreau Michel
2
3
1UMR 7618: Biogéochimie et Ecologie des Milieux Continentaux (BIOEMCO), Ecole Normale Supérieure, Paris, France2UMR 7625: Fonctionnement et Évolution des Systèmes Écologiques, Ecole Normale Supérieure, Paris, France3Department of Biology, McGill University, Montréal, Québec, CanadaWaser Nick Academic EditorUniversity of California at RiversideUnited States of America1 2006 13 12 2005 13 12 2005 4 1 e123 5 2005 11 10 2005 Copyright: © 2006 Fontaine et al.2006This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Diverse Pollination Networks Key to Ecosystem Sustainability
Pollination is exclusively or mainly animal mediated for 70% to 90% of angiosperm species. Thus, pollinators provide an essential ecosystem service to humankind. However, the impact of human-induced biodiversity loss on the functioning of plant–pollinator interactions has not been tested experimentally. To understand how plant communities respond to diversity changes in their pollinating fauna, we manipulated the functional diversity of both plants and pollinators under natural conditions. Increasing the functional diversity of both plants and pollinators led to the recruitment of more diverse plant communities. After two years the plant communities pollinated by the most functionally diverse pollinator assemblage contained about 50% more plant species than did plant communities pollinated by less-diverse pollinator assemblages. Moreover, the positive effect of functional diversity was explained by a complementarity between functional groups of pollinators and plants. Thus, the functional diversity of pollination networks may be critical to ecosystem sustainability.
Manipulating plant and pollinator communities provides experimental evidence that the persistence of a plant community can be affected by a loss of diversity among its pollinating fauna.
==== Body
Introduction
Understanding the consequences of biodiversity loss for ecosystem functioning and services is currently a major aim of ecology [1,2]. Animal-mediated pollination is one of the essential ecosystem services provided to humankind [3,4]. The negative impact of pollinator decline on the reproductive success of flowering plants has been documented at the species level [5–7], but little information is available at the community level [8]. Increasing the scale of study to the community level is essential to account for potential competitive or facilitative effects among species that belong to the plant–pollinator network. Such effects, which are often linked to diversity [9,10], are known to have large influences on ecological processes such as community productivity and stability [11,12].
Experimental evidence for diversity effects on the functioning of terrestrial ecosystems is mainly available for plants. As primary producers, plants play a central role in the flow of energy within ecosystems [13,14]. Animal-pollinated angiosperms represent up to 70% of plant species in numerous communities and ecosystems [15]. Mutualistic interactions between animals and plants form several intricate interaction webs [16]. Recent analysis of plant–pollinator and plant–frugivore interaction webs demonstrates that these contain a continuum from fully specialist to fully generalist species [17,18]. However, these networks are structured in a nested way [19,20], with specialists mainly interacting with generalists. Such a pattern might have important consequences for ecosystem functioning, because it might confer resilience to perturbations such as the extinction of species [21] if, for example, generalist pollinators buffer the loss of specialist pollinators [18,22–24]. Furthermore, this hypothesis does not take into account the dynamical properties of these networks. In a plant–pollinator community, variations in species diversity at different trophic levels may lead to an adaptation of interaction strengths [25], which may in turn affect the total effectiveness of pollination. We conclude that more information is urgently needed concerning the impacts of biodiversity loss on multispecies and multitrophic interactions.
To experimentally test the effect of functional diversity on the functioning and persistence of plant–pollinator communities, we defined functional groups of plants and pollinators based on morphological traits. For plants, two functional groups with three species each were defined according to accessibility of floral rewards (pollen and nectar; see Figure 1). The first group (group 1) included Matricaria officinalis, Erodium cicutarium, and Raphanus raphanistrum, which have easily accessible floral rewards and will be called “open flowers.” The second group (group 2), called “tubular flowers,” included Mimulus guttatus, Medicago sativa, and Lotus corniculatus, all of which present floral rewards hidden at the bottom of a tubular corolla. For pollinators, two functional groups were defined according to mouthparts length (Figure 1). The first group (group A) included three species of syrphid flies (Diptera) with short mouthparts: Saephoria sp., Episyrphus balteatus, and Eristalis tenax. The second group (group B) included three species of bumble bees with longer mouthparts: Bombus terrestris, B, pascuorum, and B, lapidarius. Note that in this case a functional trait (long mouthparts) and a phylogenetic group are confounded. Preliminary observations showed that these six insect species contribute up to 70% of all pollinating visits to flowers in our study area in France. Constructing a plant–pollinator network with these four functional groups leads to a nested structure with specialists interacting with generalists (Figure 1, third column). In principle, syrphid flies cannot efficiently pollinate tubular flowers because their mouthparts are too short.
Figure 1 Experimental Pollination Web
Summary of the characteristics upon which functional groups of pollinators (left) and plants (right) were based. In the middle, the arrows linking insect heads to flower types show the theoretical pollination network when all functional groups are present.
At the beginning of spring 2003, we set up 36 4-m2 caged experimental plant communities. There were three plant treatments following a “substitutive” design [26]. Two of them contained one of the two plant functional groups alone (group 1 or 2), whereas the third contained both plant functional groups in combination (group 3). We applied three different pollination treatments to each plant treatment, by introducing each pollinator functional group alone (group A or B), or both groups together (group C). This full factorial design led to nine experimental treatments, which were replicated four times each, making a total of 36 experimental units. The pollination treatments were applied in two consecutive years (June–July 2003 and 2004). We controlled for the total number of pollinator visits received by each plot during the two pollination seasons (1,000 visits in 2003 and 1,200 visits in 2004) to allow an unbiased comparison of pollination efficiency among the various experimental treatments.
In August and September 2003, we counted the number of fruits on each plant in every plot. We also counted the number of seeds per fruit on five collected fruits per plant. Lastly, in April 2004 and 2005, we measured both the number of plant species present at the seedling stage (recruitment richness) and the total number of seedlings (recruitment density) to determine the effects of the experimental treatments on the natural recruitment of the next plant generation.
Results
Effects on Plant Reproductive Success
The reproductive success of the two plant functional groups after the first season is analysed in Table 1. There was a significant effect of pollination treatment on the number of fruits per plant (Table 1, left; standardized means ± standard error [SE]: syrphid −0.278 ± 0.061, bumble bee 0.221 ± 0.065, and both 0.063 ± 0.068). Orthogonal contrasts on pollination treatment indicate that the identity of the pollinator guild (syrphid [A] versus bumble bee [B]) had a significant effect. There was a higher fruit production in bumble bee–pollinated communities than in those pollinated by syrphids. Moreover, the breakdown of the interaction of pollination and plant treatments into the orthogonal contrasts A1 versus B1 and A2 versus B2 indicates that the two plant functional groups responded differently to the identity of the pollinator functional group. Tubular 3 flowers (group 2) produced significantly fewer fruits in the syrphid treatment, whereas open flowers (group 1) produced the same amount of fruits whatever the identity of the pollinator functional group (Figure 2A). This supports our hypothesis that bumble bees were able to pollinate both plant functional groups whereas syrphids could only efficiently pollinate open flowers. Although the functional diversity of plant or pollinator treatment alone had no significant effect, fruit production tended to increase with both plant and pollinator functional diversity (contrast [A1 + A2 + B1 + B2] versus C3; Figure 2B).
Figure 2 Effects of Pollinator Identity and Diversity on Plant Reproductive Success
The left panels show the effects of pollinator guild identity (S indicates syrphid flies, B indicates bumble bees) on the reproductive success of the two plant guilds (open circle indicates open-flowers [group 1], closed circle indicates tubular-flowers [group 2]). Reproductive success was measured by (A) the standardized number of fruits per plant and (B) the standardized number of seeds per fruit. The right panels show the effects of the functional diversity of pollination treatments (triangle), plant treatment (inverted triangle) and both (diamond) on the standardized numbers of fruits per plant (C) and seeds per fruit (D). Lines connecting symbols indicate significant effects (solid indicates p < 0.001, dashed indicates p < 0.08). Error bars represent one standard error. See Table 1 for statistical analysis.
Table 1 Analysis of Plant Reproductive Success
With respect to seed set per fruit, the interaction between plant and pollination treatment was marginally significant (Table 1, right). As with fruit production, the contrasts A1 versus B1 and A2 versus B2 indicate that the two plant functional groups responded differently to pollinator functional group identity. The pattern, however, was different: Open flowers produced significantly fewer seeds per fruit in the bumble bee treatment than in the syrphid treatment (Figure 2C). This means that bumblebees were less-efficient pollinators than syrphids for open flowers. This could be due to the higher rate of geitonogamous visits (i.e., consecutive visits to different flowers of the same plant, resulting in self-fertilization) by bumblebees. Indeed, preliminary observations using a similar experimental design showed that bumble bees perform a higher percentage of geitonogamous visits than do syrphids (I. Dajoz, unpublished data). Finally, the mean number of seeds per fruit in the plant communities tended to increase with functional diversity of pollination treatments (contrast [A + B] versus C; Figure 2D).
Effects on Natural Recruitment
We analysed the long-term effects of our pollination treatments on the natural recruitment of our experimental plant treatments after the first and second pollination seasons. The results are presented in Table 2. There was a significant effect of year on recruitment richness with a higher richness after the second pollination season (mean ± SE: 1.916 ± 0.075 in 2004, and 2.291 ± 0.0856 in 2005). Among the possible causes was a severe drought in 2003 [27], which likely affected both plant and insect populations. Such a drought did not occur in 2004. This difference in climate between years may account for a large part of the year effect.
Table 2 Analysis of Plant Recruitment Richness and Density
Recruitment richness was significantly different among plant treatments, with fewer species recruiting in tubular communities (Figure 3). This is very likely due to two perennial species (whereas all species are annuals in the other group) which may have different reproductive traits and create differences in competitive intensity among the plant treatments. There was a significant effect of pollination treatment, with a higher recruitment richness when both groups of pollinators were present (means ± SE: syrphid 1.854 ± 0.973, bumble bee 2.052 ± 0.826, and both 2.406 ± 1.062). However, as suggested by the significant interaction between plant and pollination treatments, the pattern was more complex (Figure 3A). In fact, pollination treatments had no effect on recruitment richness in open-flower plant treatment (Figure 3A, left). In the tubular-flower plant treatment, recruitment in the syrphid fly treatment tended to be lower than in the other pollination treatments (Figure 3A, centre). But the positive effect of pollinator functional diversity was obvious in the plant treatment that contained both plant functional groups (Figure 3A, right). In the mixed plant treatment, recruitment richness under the most functionally diverse pollination treatment was substantially above that in the two other treatments.
Figure 3 Effects of Pollination Treatments on Plant Recruitment
Effects of pollination by syrphid flies (S), bumble bees (B), or both (S + B) on (A) recruitment richness (mean number of plant species present as seedlings in a quadrat) and (B) recruitment density (mean number of plant individuals present as seedlings in a quadrat) in the various plant treatments. Error bars represent one standard error. Lower-case letters indicate statistically significant differences among pollination treatments within a plant treatment (Bonferroni-adjusted t-test, p < 0.05).
Considering recruitment density, there was also a significant effect of year, with a higher density after the second pollination season (mean ± SE: 26.784 ± 2.324 in 2004 and 31.319 ± 1.937 in 2005), and a significant effect of plant treatment, with fewer individuals recruiting in tubular communities (Figure 3B, centre). These year and plant-treatment effects can be explained in the same way as for recruitment richness (see above). There was also a significant effect of pollination treatment, with a lower recruitment density when plant communities were pollinated by syrphid flies alone (means ± SE: syrphids: 24.104 ± 20.464, bumble bees: 34.364 ± 32.781, and both 28.688 ± 21.459). This is congruent with our results on the number of fruits produced per plant (see Table 1, contrast A versus B). As for recruitment richness, there was a significant interaction between plant and pollination treatments (Figure 3B). In the open-flower plant treatment, recruitment density was not significantly different among pollination treatments (Figure 3B, left). But in the tubular-flower plant treatment, recruitment density was significantly higher in the bumble bee treatment than in the other pollination treatments (Figure 3B, centre). Finally, in the mixed plant treatment, the same pattern as for recruitment richness was observed: There was a higher density in the mixed pollination treatment than in single-guild pollination treatments (Figure 3B, right).
Note that these results on natural recruitment are not an artefact caused by sampling small quadrats in heterogeneous experimental plots since the same patterns were observed when data from all quadrats in a plot were pooled.
Pollination Visitation Web in the Mixed Plant Treatment
To explain the strong effect of pollinator functional diversity on the persistence of mixed plant communities, we carried out a log-linear analysis on the visitation rate of each insect species in a given pollination treatment, for the six plant species of the mixed plant treatment. Data from the year 2003 are illustrated in Figure 4, and the results of the analysis on both years are presented in Table 3. In the second year, there was a significant effect of plant functional group identity: Tubular flowers received a higher number of visits than did open flowers (mean visitation frequency ± SE: for open flowers 0.236 ± 0.097 and for tubular flowers 0.763 ± 0.097). This is very likely due to the two well-established perennial species, which produced a more attractive floral display during the second year of the experiment. For the two years of the experiment, there was a significant interaction between plant functional group and pollinator functional group. This indicates that the two pollinator functional groups were specialised on different plant functional groups (mean visitation frequency ± SE on open flowers and tubular flowers, respectively: in 2003, for bumble bees 0.128 ± 0.058 and 0.433 ± 0.075; for syrphids 0.327 ± 0.043 and 0.113 ± 0.052; in 2004, for bumble bees 0.01 ± 0.005 and 0.58 ± 0.075; for syrphids 0.23 ± 0.055 and 0.18 ± 0.087). Syrphids mainly visited open flowers whereas bumble bees preferentially visited tubular flowers (Figure 4). Even though bumble bees can pollinate open flowers quite efficiently when this is the only plant functional group present (as shown by the reproductive success, recruitment diversity, and recruitment density of the open-flower plant treatment in the bumble bee treatment, Figures 2 and 3), they focus on the tubular-flower group in the mixed plant treatment. In the mixed pollination treatment, the match between plant and pollinator functional groups leads to a more homogenous distribution of pollinator visits among plant groups than in the other pollination treatments. Ultimately, this significantly increases the reproductive success of plants, most likely through the homogenisation of pollinator visits and the minimization of inefficient pollinator visits.
Figure 4 Visitation Web in the Communities with Both Plant Types
Distribution of pollinator visits for the year 2003, among the six plant species in the plant treatment containing the two plant functional groups, (A) for the mixed pollination treatment (S + B) and (B) for the single functional group pollination treatments (S or B). The length of the side of the black squares shows the proportion of visits by a given pollinator species on each plant species. Lower-case letters represent plant species: a, Ma. officinalis; b, E. cicutarium; c, R. raphanistrum; d, Mi.guttatus; e, Me. sativa; f, L. corniculatus. Numbers represent pollinator species: 1, Saephoria sp.; 2, Ep. balteatus; 3, Er. tenax; 4, B. terrestris; 5, B. pascuorum; 6, B. lapidarius.
Table 3 Analysis of Visitation Rates
Discussion
Previous studies on the diversity of plant–pollinator interaction webs were either descriptive [16], carried out on a single plant species [6,7,28–30], or based on simulation [21] and theoretical approaches [22,31]. To our knowledge, this is the first experimental evidence that the persistence of a plant community can be affected by a loss of diversity of its pollinating fauna. Of course, our experimental communities differed from natural ones in several respects. Among other things, the interaction networks we studied were much simpler than those occurring in nature; in particular, they contained fewer species in each trophic level. But such simplifications from natural situations are often necessary to carry out controlled experiments.
In plant communities that contained only open flowers, plants produced fewer seeds per fruit in the bumblebee treatment than in the syrphid treatment (Figure 2C), but this was compensated by a sufficiently high fruit production, leading to a richness and density of natural recruitment that was similar to the other pollination treatments (Figure 3A and 3B left). Thus, in these communities, all pollination treatments were equally effective in the long term.
In plant communities that contained only tubular flowers, syrphids were inefficient pollinators; fruit production was very low (Figure 2A) and insufficient to allow a good natural recruitment. Bumble bees were the most effective pollinators (Figure 3A and 3C, centre). Note that in the bumble bee treatment, the very high value of average recruitment density was due to three measurements in two replicates, in which only M. guttatus seedlings were recorded at a very high density (more than 150 seedlings per quadrat). To test the effect of these outliers, we removed them and repeated our analysis. The same significant effects were observed, except for the effect of pollination treatment, which became marginally significant (p = 0.0645). The new mean number of seedlings per quadrat for this experimental treatment was 32.17 ± 4.55 (SE), which is still slightly above the value for the pollination treatment with both pollinator groups. For plant communities that contained only tubular flowers, recruitment richness in the two pollination treatments that contained bumblebees was similar.
These results are in agreement with our theoretical pollination network presented in Figure 1. In our experimental system, syrphids can be considered as specialist pollinators since they efficiently pollinate only open flowers. Bumble bees were potentially generalists as they induced an important fruit production of the two plant types and a good recruitment in the open- and tubular-flower plant treatments. Our results on the reproductive success and recruitment of single-guild plant treatments indicate that there are strong functional group identity effects since our plant functional groups responded differently to our pollinator functional groups.
However, the functional diversity of both the plant and pollination treatments was also important. Plant reproductive success tended to increase with pollinator functional diversity when the number of seeds per fruit was considered, and with both plant and pollinator functional diversity when the number of fruits per plant was considered (Figures 2B and 2D). Although recruitment in single-guild plant treatments was mainly affected by the identity of functional groups, the effect of functional diversity was dramatic in the mixed plant treatment. Natural recruitment of plant communities visited by mixed pollinator guilds was largely above that in other pollination treatments.
Pollination by syrphids alone allowed the reproduction of open flowers but not tubular flowers, as expected from the specialisation of syrphids. More surprisingly, however, bumble bees failed to be efficient generalist pollinators. Most of their visits occurred on tubular flowers (Figure 4), resulting in a relatively poor recruitment of open flowers. The only pollination treatment that achieved a high recruitment of both open and tubular flowers when they were mixed, was the one containing the two insect functional groups (Figure 3, right). When syrphids and bumble bees simultaneously pollinated mixed plant communities, they each focused on their target plant functional group, leading to more efficient visits and a better distribution of visits among plant functional groups (Figure 4). Ultimately, it was the pollination treatment with both pollinator functional groups that produced the highest richness and density of natural recruitment. Consequently, since most natural plant communities contain both open and tubular flowers, pollinator functional diversity should strongly enhance the persistence of these communities.
Although our experimental system differed from natural communities, and information about the reciprocal effects of the functional diversity of plant communities on the diversity of pollinator communities would be useful, our study indicates that the functional diversity of plant–pollinator interaction webs may be critical for the persistence and functioning of ecosystems and should be carefully monitored and protected. The loss of pollinator functional diversity is likely to trigger plant population decline or extinctions [4], which in turn are likely to affect the structure and composition of natural plant communities and the productivity of many agroecosytems that rely on insect pollination [8]. Ultimately, higher trophic levels may be affected since the diversity and biomass of consumers depend on primary production. Our results strongly suggest that the functional diversity of complex interaction webs plays a crucial role in the sustainability of ecosystems.
Materials and Methods
Experimental plant communities
At the beginning of spring 2003, plant communities were set up in a meadow that remained almost undisturbed for 10 years at the Station Biologique de Foljuif, France, 80 km southwest of Paris. Prior to the establishment of the communities, soil was sterilized by injecting 120 °C steam (30 min) to destroy the seed bank and soil pathogens. In each of the 36 4-m2 plots, a total of 30 adult plants were planted on a grid, spaced 25 cm from each other, to minimize competition and homogenise spatial distribution. Thus, plant density was the same in all experimental plots. We selected a moderate density to maintain within- and among-species competition to a low level, and to allow enough space for future recruitment in the plots. Each of these plant communities was enclosed in a 2-m–high nylon mesh cage in order to eliminate natural pollinator visitation.
Pollination rounds
During the flowering seasons (June–July 2003 and 2004), pollinators were captured around the study area and introduced into the cages. The relative abundance of pollinator species in the various pollination treatments reflects their natural abundances. From preliminary observations, we had noticed that, in order to have no more than three insects active at the same moment in a 4-m2 plot, it was necessary to put about eight syrphid flies, or six bumble bees, or a mixture of six syrphids and four bumble bees in each pollination cage. Each pollination round in a given plot included 200 visits in the year 2003 and 300 in the year 2004. In total, each plot received either four (in 2004) or five (in 2003) pollination rounds, leading to a total of 1,000 visits per plot in 2003 and 1,200 in 2004.
Pollination activity
Bumble bees needed approximately 30 min after introduction in the cages to calm down and start to pollinate. In the pollination treatment with both pollinator guilds, we then introduced syrphids, which started to pollinate immediately. Mean visitation time was not significantly different between insects in the cages and in nature. This was true both for bumble bees (mean visitation time in cages: 3.25 ± 0.92 s, mean visitation time in nature: 2.91 ± 1.33 s, t = 1.51, df = 96, p = 0.133) and for syrphids (mean visitation time in cages: 40.21 ± 8.89 s, mean visitation time in nature: 35.38 ± 14.75 s, t = 0.77, df = 12, p = 0.45).
Measurement of reproductive success
One month after the first pollination treatments, we counted the total number of fruits on each plant, except for M. guttatus and M. officinalis in which fruits cannot be counted without collecting them. We randomly took five fruits per plant of each species to estimate the number of seeds per fruit.
Measurement of recruitment richness and density
Recruitment richness and density were estimated during the second (April 2004) and third (April 2005) year of the experiment by counting the number of seedlings of each species in four 1,600-cm2 quadrats in each plot.
Statistical analysis
Statistical analyses were performed using SAS 8.2 software.
For the analysis of plant reproductive success, we log-transformed the data to ensure normality. We standardized the data by species using the formula: x − μ/σ (where μ = the mean and σ = the standard deviation of number of fruits or number of seeds per fruit for a given plant species) in order to make the data comparable among the various species and functional groups. We used a mixed analysis of variance (ANOVA) model (SAS proc mixed), in which the fixed effects were plant treatment, pollination treatment, and their interaction term. To investigate the effects of the various plant and pollination treatments, we subdivided a priori each main effect into two components using orthogonal contrasts. The first contrast tested the effect of the identity of the plant or pollinator functional group, i.e. one group versus the other. The second tested the effect of the functional diversity of the plant or pollination treatment, i.e. single-guild versus mixed-guild plant or pollination treatments. Similarly, we subdivided the interaction into three orthogonal contrasts testing the effects of pollinator functional group identity on each plant guild, and the effect of the functional diversity of both plant and pollination treatments. See Table 1 for the construction of the contrasts.
For the analysis of plant recruitment, we used a repeated measure ANOVA model (SAS proc mixed). The fixed effects were pollination treatment, plant treatment, year, and all the interaction terms. The repeated effect was year, and the subject effect was replicate. For recruitment density, data were log transformed.
For each year of the experiment, the visitation rate of pollinators on each plant species in the communities with both plant functional groups was analysed using a mixed log-linear model (glimix macro, SAS). We subdivided the pollination treatment into two effects: pollinator functional diversity (one or two pollinator functional groups) and identity of the pollinator functional group (bumble bees or syrphids). The model included pollinator species nested within identity of pollinator functional groups, plant species nested within identity of plant functional group, identity of pollinator functional groups, identity of plant functional groups, pollinator functional diversity, and all interaction terms. The replicate was a random effect.
We thank Carine Collin, Romain Gallet, Jean-Francois le Galliard, Jacques Gignoux, Andy Gonzalez, Gérard Lacroix, Gaelle Lahoreau, Louis Lambrecht, Manuel Massot, Naoise Nunan, Virginie Tavernier, and Elisa Thebault for useful discussions; and Marco Banchi, Yves Bas, Mathilde Baude, Alix Boulouis, Marion Decoust, Patricia Genet, Alexandra Kabadajic, Mohsen Kayal, Fanny Marlin, and Emilie Patural for great help in the field and in the lab. We also thank Andy Gonzalez, Andy Hector, Marcel van der Heijden, Claire Kremen, Jane Memmott, Nick Waser, and three anonymous reviewers for constructive and useful comments on the manuscript. We acknowledge the financial support of the Quantitative Ecology Coordinated Incentive Action (ACI Ecologie Quantitative) of the Ministry of Research (France).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. CF and ID designed the experiment. CF, ID, and JM performed the experiment. CF analysed the data. CF, ID, and ML conceived the work and wrote the paper.
Citation: Fontaine C, Dajoz I, Meriguet J, Loreau M (2006) Functional diversity of plant–pollinator interaction webs enhances the persistence of plant communities. PLoS Biol 4(1): e1.
Abbreviations
Asyrphid group
ANOVAanalysis of variance
Bbumble bee group
Ccombined (syrphid and bumble bee) group
dfdegrees of freedom
group 1open flower
group 2tubular flower
group 3combined (open and tubular) flower
SEstandard error
==== Refs
References
Loreau M Naeem S Inchausti P Bengtsson J Grime JP Biodiversity and ecosystem functioning: Current knowledge and future challenges Science 2001 294 804 808 11679658
Naeem S Thompson LJ Lawler SP Lawton JH Woodfin RM Declining biodiversity can alter the performance of ecosystems Nature 1994 368 734 737
Costanza R Arge R Groot R Farber S Grasso M The value of the world's ecosystem services and natural capital Nature 1997 387 253 260
Kearns CA Inouye DW Waser NM Endangered mutualisms: The conservation of plant-pollinator interactions Annu Rev Ecol Syst 1998 29 83 112
Kremen C Williams NM Thorp RW Crop pollination from native bees at risk from agricultural intensification Proc Natl Acad Sci U S A 2002 99 16812 16816 12486221
Klein AM Steffan-Dewenter I Tscharntke T Fruit set of highland coffee increases with the diversity of pollinating bees Proc R Soc Lond B Biol Sci 2003 270 955 961
Herrera CM Components of pollinator ‘quality': Comparative analysis of a diverse insect assemblage Oikos 1987 50 79 90
Tepedino VJ The importance of bees and other pollinators in maintaining floral species composition Great Basin Nat Mem 1979 3 139 150
Van der Heijden MGA Klironomos JN Ursic M Moutoglis P Streitwolf-Engel R Mycorrhizal fungal diversity determines plant biodiversity, ecosystem variability and productivity Nature 1998 396 69 72
Cardinale BJ Palmer MA Collins SL Species diversity enhances ecosystem functioning through interspecific facilitation Nature 2002 415 426 429 11807553
Callaway RM Brooker RW Choler P Kikvidze Z Lortie CJ Positive interactions among alpine plants increase with stress Nature 2002 417 844 848 12075350
Thebault E Loreau M Food-web constraints on biodiversity-ecosystem functioning relationships Proc Natl Acad Sci U S A 2003 100 14949 14954 14638942
Hector A Schmid B Beierkuhnlein C Caldeira MC Diemer M Plant diversity and productivity experiments in European grasslands Science 1999 286 1123 1127 10550043
Tilman D Reich PB Knops J Wedin D Mielke T Diversity and productivity in a long-term grassland experiment Science 2001 294 843 845 11679667
Axelrod DI Tax S The evolution of flowering plants Evolution after Darwin. Volume 1, The evolution of life 1960 Chicago University of Chicago Press 227 305
Memmott J The structure of a plant-pollinator food web Ecol Lett 1999 2 276 280
Olesen JM Jordano P Geographic patterns in plant-pollinator mutualistic networks Ecology 2002 83 2416 2424
Jordano P Bascompte J Olesen JM Invariant properties in coevolutionary networks of plant-animal interactions Ecol Lett 2003 6 69 81
Bascompte J Jordano P Melian CJ Olesen JM The nested assembly of plant-animal mutualistic networks Proc Natl Acad Sci U S A 2003 100 9383 9387 12881488
Vázquez DP Aizen MA Asymmetric specialization: A pervasive feature of plant-pollinator interactions Ecology 2004 85 1251 1257
Memmott J Waser NM Price MV Tolerance of pollination networks to species extinctions Proc R Soc Lond B Biol Sci 2004 271 2605 2611
Ashworth L Aguilar R Galetto L Aizen MA Why do pollination generalist and specialist plant species show similar reproductive susceptibility to habitat fragmentation? J Ecol 2004 92 717 719
Aizen MA Feinsinger P Forest fragmentation, pollination, and plant reproduction in a chaco dry forest, Argentina Ecology 1994 75 330 351
Vázquez DP Simberloff D Ecological specialization and susceptibility to disturbance: Conjectures and refutations Am Nat 2002 159 606 623 18707385
Kondoh M Foraging adaptation and the relationship between food-web complexity and stability Science 2003 299 1388 1391 12610303
Austin MP Fresco LFM Nicholls AO Groves RH Kaye PE Competition and relative yield: Estimation and interpretation at different densities and under various nutrient concentrations using Silybum marianum and Cirsium vulgare
J Ecol 1998 76 157 171
Ciais P Reichstein M Viovy N Granier A Ogee J Europe-wide reduction in primary productivity caused by the heat and drought in 2003 Nature 2005 437 529 533 16177786
Motten AF Campbell DR Alexander DE Miller HL Pollination effectiveness of specialist and generalist visitors to a North Carolina population of Claytonia virginica
Ecology 1981 62 1278 1287
Thostesen AM Olesen JM Pollen removal and deposition by specialist and generalist bumblebees in Aconitum septentrionale
Oikos 1996 77 77 84
Kremen C Williams NM Bugg RL Fay JP Thorp RW The area requirements of an ecosystem service: Crop pollination by native bee communities in California Ecol Lett 2004 7 1109 1119
Vázquez DP Aizen MA Null model analyses of specialization in plant-pollinator interactions Ecology 2003 84 2493 2501
|
16332160
|
PMC1310649
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 13; 4(1):e1
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040001
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633604310.1371/journal.pbio.0040003Research ArticleBotanyPathologyEcologyGenetics/Genomics/Gene TherapyInfectious DiseasesMicrobiologyMolecular Biology/Structural BiologyPlant ScienceVirologyGastroenterology/HepatologyRNA Viral Community in Human Feces: Prevalence of Plant Pathogenic Viruses Abundant Plant Viruses in Human FecesZhang Tao
1
Breitbart Mya
2
Lee Wah Heng
1
Run Jin-Quan
1
Wei Chia Lin
1
Soh Shirlena Wee Ling
1
Hibberd Martin L
1
Liu Edison T
1
Rohwer Forest
2
Ruan Yijun [email protected]
1
1Genome Institute of Singapore, Singapore2Department of Biology, San Diego State University, San Diego, California, United States of AmericaDangl Jeffrey Academic EditorUniversity of North CarolinaUnited States of America1 2006 20 12 2005 20 12 2005 4 1 e310 8 2005 25 10 2005 Copyright: © 2006 Zhang et al.2006This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Thriving Community of Pathogenic Plant Viruses Found in the Human Gut
The human gut is known to be a reservoir of a wide variety of microbes, including viruses. Many RNA viruses are known to be associated with gastroenteritis; however, the enteric RNA viral community present in healthy humans has not been described. Here, we present a comparative metagenomic analysis of the RNA viruses found in three fecal samples from two healthy human individuals. For this study, uncultured viruses were concentrated by tangential flow filtration, and viral RNA was extracted and cloned into shotgun viral cDNA libraries for sequencing analysis. The vast majority of the 36,769 viral sequences obtained were similar to plant pathogenic RNA viruses. The most abundant fecal virus in this study was pepper mild mottle virus (PMMV), which was found in high concentrations—up to 109 virions per gram of dry weight fecal matter. PMMV was also detected in 12 (66.7%) of 18 fecal samples collected from healthy individuals on two continents, indicating that this plant virus is prevalent in the human population. A number of pepper-based foods tested positive for PMMV, suggesting dietary origins for this virus. Intriguingly, the fecal PMMV was infectious to host plants, suggesting that humans might act as a vehicle for the dissemination of certain plant viruses.
A comparative metagenomic analysis of RNA viruses in the human gut identifies the vast majority as plant pathogens.
==== Body
Introduction
The human gastrointestinal tract is the natural habitat for a large microbial community including species from the kingdoms Archaea, Bacteria, and Eukarya [1]. It is estimated that the human gastrointestinal (GI) flora contains 1014 microorganisms. Most of these microbes are symbiotic to the human host and beneficial to food digestion [2,3]. The GI microbiota also contains enteric viruses, including a variety of bacteriophages and a number of known human viruses and uncharacterized viruses. Bacteriophages can influence food digestion by regulating microbial communities in the human GI tract through lytic and lysogenic replication [4]. Bacteriophages may also contribute to human health by controlling invading pathogens [5]. In addition to bacteriophages, the other well-studied human enteric viruses are the viral pathogens associated with gastroenteritis. They can infect the human small intestine cells, causing damage to the epithelial lining and the absorptive villi, leading to the malabsorption of water and an electrolyte imbalance [6,7]. Many viral pathogens have been isolated from the feces of gastroenteritis patients, including rotavirus, astrovirus, calicivirus, hepatatis E virus, certain members of coronavirus and torovirus, and the enteric adenovirus (serotypes 40 and 41) [8–11]. Except for the adenoviruses, which contain DNA genomes, all the others are RNA viruses. Despite intensive studies, many causative agents of human gastroenteritis are still unknown.
Traditionally, discovery of viruses was dependent on culturing the viruses in host cells in order to propagate and isolate enough pure virions for characterization. However, it is generally known that the large majority of viruses, including enteric viruses, cannot be cultivated using standard techniques. For instance, some Norwalk viral agents causing gastroenteritis could not be grown in cell cultures [12,13]. The only way to obtain adequate virus particles for characterization of these viruses was to feed the volunteer human adults with stool filtrates derived from the disease outbreak [14]. Hence, culture-based methods are insufficient for large-scale characterization of the viral community in the human GI tract. In addition, viruses do not have ubiquitously conserved genetic elements such as rDNA that can be used as diversity and evolutionary distance markers [15].
Metagenomic analyses offer an opportunity to directly characterize mixed genomes of uncultured viruses. In combination with tangential flow filtration for viral particle isolation and concentration from large volume samples, metagenomic analyses have been employed to survey the DNA viruses in seawater [16] and from human feces [17]. Based on the analysis of about 500 sequences, it was found that the majority of DNA viruses in human feces were novel, and most of the recognizable sequences belonged to bacteriophages. Similar metagenomic approaches have also been applied, in large scale, to bacteria in seawater [18] and other environmental samples [19,20].
To date, very little information is available on the human enteric RNA viral flora, despite the fact that many RNA viruses are known etiologic agents of gastroenteritis. To expand our understanding of the RNA viral flora in the human GI tract, we conducted a comprehensive metagenomic analysis of uncultured RNA viruses isolated from the feces of healthy humans. Surprisingly, we discovered that some plant RNA viruses were highly abundant in human feces.
Results
Identification of RNA Viruses in Human Feces
Three fecal samples from two healthy adults living in San Diego were used for virus isolation. Samples 1 and 2 were from the same individual, with a gap of 6 mo between the sample collections. Sample 3 was from a second individual. We concentrated the feces-borne viral particles using tangential flow filtration as described previously [17]. We further treated the viral concentrates with DNase and RNase to eliminate potential contamination with free nucleic acids. Viral RNA was extracted from each sample and converted into cDNA, which was used to construct a shotgun library for sequencing analysis. From the three libraries (designated as Lib 1, Lib 2, and Lib 3 corresponding to samples 1, 2, and 3), we generated 10,576, 13,572, and 12,621 high-quality sequence reads, respectively (Table 1). Each sequence read represents an individual clone in the libraries.
Table 1 Clone Sequences in Three RNA Viral Shotgun Libraries
Of the total 36,769 sequences, 33,643 (91.5%) were similar to known sequences in the GenBank nr (nonredundant) database based on tBLASTx (e-value <0.001), and are therefore designated as “known” (Table 1). Of these 33,643 known sequences, 25,779 (76.6%) were most similar to viruses and 7,864 (23.4%) were most similar to Bacteria, Archaea, or Eukarya. Of the nonviral known sequences, 6,828 (86.8%) were similar to ribosomal RNA. The presence of ribosomal RNA-like sequences in these libraries may reflect the copurification of some ribosomes with the virions from the fecal extracts during tangential flow filtration.
Based on the best hit of tBLASTx analysis, the virus-like sequences were assigned to 42 viral species, including two animal viruses, 35 plant viruses, one yeast virus, and four bacteriophages (Table 2). The number of sequences related to yeast viruses and bacteriophages in these libraries were insignificant, with only one and seven clones matching them, respectively. Surprisingly, of the virus-like sequences, 25,040 (97.1%) were homologous to plant viruses, while less than 735 (2.9%) resembled animal viruses. This is distinct from the previously reported DNA fecal viral library, in which most virus-like sequences were similar to phages [17]. Furthermore, no RNA phage-like sequences were observed in this study, which may reflect the rare occurrence of RNA phages in human feces [21]. Notably, very few DNA virus-like sequences (eight clones) were found in these libraries, indicating that DNA contamination was essentially eliminated during our isolation procedure.
Table 2 Viruses Identified in Three RNA Viral Libraries
The most abundant virus in the fecal RNA viral libraries was pepper mild mottle virus (PMMV). PMMV is a well-characterized plant pathogenic virus that infects all species in the genus Capsicum, including a wide variety of sweet and hot peppers [22,23]. A total of 21,838 sequences were mapped across the entire genome of PMMV (6,357 bp). The levels of sequence homology to the PMMV reference genome sequence were highly variable, suggesting that multiple variants of PMMV might be present in these samples. In addition, 24 out of the 35 plant viruses detected in these fecal samples were known pathogens of consumable crops including fruits, vegetables, tobaccos, and cereals (Table 3), suggesting that foods contaminated with plant viruses might have contributed to the RNA viral flora in human feces.
Table 3 Plant Viruses Identified in Human Feces and Their Known Host Plants
The most common animal virus (total of 735 clones) identified in the RNA viral libraries was a picobirnavirus (PBV, 733 clones). PBV belongs to a recently identified family of RNA viruses with bisegmented genomes, which are present in the stools of humans and animals [24,25]. In humans, PBV has been found in both healthy individuals and diarrhea patients [26]. The other animal virus identified (Moloney murine leukemia virus) was only represented by two clones. In summary, we found that the human feces-borne RNA viral community is dominated by viruses similar to plant RNA viruses, exemplified by the abundance of PMMV.
The Diversity and Dynamics of the RNA Viral Community in Human Feces
Of the 42 viral species identified, 32 were present in only one of the libraries. Among the common viral species, four were present in all three libraries, and six were found in only two libraries. Hence, the overlap between these libraries was quite small in terms of the types of viral species (see Table 2).
Although there were 27, 24, and ten viral species identified in Lib 1, 2, and 3, respectively, the majority of the clones in these libraries were represented by only a few dominant viral species (see Table 2). For instance, in Lib 1, PMMV, oat blue dwarf virus, and grapevine asteroid mosaic-associated virus were the top three species represented by 80%, 13.6%, and 3.0% of the 9,961 virus-like clones, respectively. In Lib 2, PMMV (75.7%), PBV (8%), maize chlorotic mottle virus (MCMV, 7.7%), oat chlorotic stunt virus (OCSV, 2.9%), panicum mosaic virus (PanMV, 1.8%), and tobacco mosaic virus (1.3%) were the six major viral species of the 7,807 virus-like clones. In Lib 3, PMMV was the only dominant viral species, accounting for 99.4% of the 8,011 virus-like sequences. It is interesting to note that, except for PMMV, which was predominant in all three libraries, the other major species found in each of the three libraries were quite different. For instance, oat blue dwarf virus and grapevine asteroid mosaic-associated virus were the second and third most abundant viruses in Lib 1, but were undetectable in Lib 2. Similarly, MCMV, OCSV, and PanMV were the second, third, and fourth most abundant viruses in Lib 2, but were not found in Lib 1. Since Lib 1 and Lib 2 were derived from feces collected from the same individual 6 mo apart, these differences most likely reflect fluctuations in the fecal RNA viral community of a single individual over time.
To further address the dynamics of human feces-borne viruses, we compared the PBV-like clone sequences from Lib 1 and Lib 2 (Figure 1). It is known that human PBV has two distinct genogroups, represented by two prototype strains, 4-GA-91 and 1-CHN-97 [26,27]. The members of the former genogroup are highly conserved, while the members of the latter are quite variable. The sequence identity between the two prototypes is 28% at the nucleotide level and 40% at the amino acids level. To facilitate the comparative analysis, we clustered the PBV-like sequences of each library into contigs (using minimum overlaps of 40 bp and over 95% identity) and aligned them to the partial genome sequences of the 4-GA-91 and 1-CHN-97 available in GenBank. The five contigs from 27 PBV-like sequences from Lib 1 were highly similar (over 90% identical at nucleotide level) to the conserved PBV strain 4-GA-91 (Figure 1A) but had no matches to the other prototype strain. In contrast, the 26 contigs from 701 PBV-like sequences from Lib 2 were only marginally homologous (46%–69% homology at the amino acid level) to the variable strain 1-CHN-97 (Figure 1B). It is important to note that the PBV populations in the two fecal samples (samples 1 and 2) from the same individual represented distinct genogroups, once again highlighting the dynamic nature of the human enteric RNA viral flora.
Figure 1 Different PBV Strains Found in Two Fecal Samples from the Same Individual
(A) The PBV-like sequence segments identified in Lib 1 were aligned to the partial genome sequence of PBV strain 4-GA-91 using BLASTn. The identities of nucleotide sequence between the contigs and the reference PBV sequence were 95%–99%.
(B) The PBV-like sequence segments in Lib 2 were too remote to both known PBV strains (4-GA-91 and 1-CHN-97) at the nucleotide level, but could be aligned to the PBV strain 1-CHN-97 using tBLASTx. The identity of amino acid sequences between the PBV-like sequence segments in Lib 2 and the reference PBV genome sequence were 46%–69%.
(C) Colored bars indicate the similarity level between library sequences with template sequences as measured by BLAST score.
Similarly, the other fecal viruses, including PMMV, identified in this study also appeared variable. Using a minimum overlap of 40 bp and more than 95% identity, we clustered all PMMV-like sequences and generated 376 (Lib 1), 359 (Lib 2), and 225 (Lib 3) unique PMMV genome sequence segments. The largest PMMV contig was 3,158 bp in length and contained 1,817 clones. Even though the feces-borne PMMV sequences gave over 200-fold coverage of the reference PMMV genome, a complete fecal PMMV genome could still not be assembled de novo. As shown in Figure 2, the PMMV-like sequences within each library had a wide range of BLAST scores to the PMMV reference genome, suggesting that the feces-borne PMMV sequences were very divergent and represented novel PMMV variants. Similar results were observed for other viral species (unpublished data).
Figure 2 Alignment of Assembled PMMV-Like Sequences from Three Shotgun Libraries with the Reference PMMV Genome Sequence
The PMMV-like viral genome sequence segments from Lib 1 (A), Lib 2 (B), and Lib 3 (C) were aligned with the reference PMMV genome sequence (6,357 bp). Colored bars (D) indicated the similarity level between library sequences with template sequences as measured by BLAST score.
We further analyzed the phylogenetic relationship of PMMV variations in these libraries. There are seven complete PMMV genomes and 14 individual PMMV coat protein (CP) gene sequences available in GenBank. To utilize the maximum number of reference sequences, we decided to use the CP gene for phylogenetic analysis together with the PMMV-like sequences identified in the three libraries. Since the first 101 bp of the 474-bp CP gene (from 5,685 to 6,158 bp of the PMMV genome) had the most sequences (44 unique assembled PMMV sequence segments) mapping to it, we chose this region for phylogenetic analysis. The majority of the 44 feces-borne PMMV sequences were grouped into five main phylogenetic clusters (Figure 3). The previously known PMMV CP gene sequences were found mostly in three clusters (clusters I, II, and III). Interestingly, most of the PMMV sequences found in Lib 1 were grouped in cluster IV, however, only one sequence from Lib 2 was grouped into this cluster, indicating that the PMMV strains were very different even in two fecal samples collected from the same individual, and implying that the human fecal PMMV population was dynamic.
Figure 3 Phylogenetic Tree of PMMV Sequences
A region of 101 bp in the PMMV CP gene was chosen for sequence comparison and phylogenetic analysis. A total of 44 assembled sequences (>50 bp) from the three fecal virus libraries were located within this region. These sequences were aligned with 21 known PMMV CP genes from GenBank using ClustalW with default parameter settings. In this phylogenetic tree of the PMMV CP gene, sequences from Lib 1 are highlighted in pink, Lib 2 in blue, and Lib 3 in yellow. GenBank accession numbers of known CP genes are shown unshaded.
Taken together, our sequence analysis clearly showed that the RNA viral flora in human feces is diverse and dynamic at both species and strain levels, although we cannot rule out the possibility that the heterogeneous nature of human feces and different sequencing coverage in these libraries may also contribute to the variations observed.
The Prevalence of PMMV in the Human Population
To confirm the presence of the viruses identified through shotgun library sequencing, we used RT-PCR to detect these viruses directly from the fecal samples used for library construction. We first detected the presence of PMMV in a fecal sample and validated that our RT-PCR detection is specific and reliable (Figure 4A and 4B). We then extended the test for the five most abundant viruses identified in Lib 2 (PMMV, MCMV, PBV, OCSV, and PanMV; see Table 2), and all were positive from the primary fecal sample 2 (Figure 4C). Furthermore, the relative quantities of the RT-PCR products for each of the viruses tested were in good correlation with the relative abundance reflected by the number of clones observed for each of the viruses in the shotgun library.
Figure 4 RT-PCR Detection of Fecally Borne RNA Viruses
(A) PMMV (lane 1) was detected by RT-PCR using PMMV specific primers in a fecal RNA extract. This PMMV band is PMMV primer-specific (lane 2) and dependent on reverse transcription (lane 3).
(B) The specificity of the RT-PCR reaction for detecting fecal PMMV was further assessed by the use of nonspecific PCR primers (lane 1 versus lane 2) and respiratory syncytial virus (RSV; American Type Culture Collection #VR-1401) as a nonspecific RNA template (lanes 1 and 2 versus lanes 3 and 4). The identities of RT-PCR products (PMMV in lane 1; RSV in lane 4) were confirmed by sequencing analysis.
(C) RNA viruses were directly detected by RT-PCR from the total RNA of fecal sample 2: PMMV (lane 1), MCMV (lane 2), PBV, segment 2 (lane 3), OCSV (lane 4), and PanMV (lane 5).
(D) Equal amounts of dry weight of food (meal samples for 2 d prior to fecal collection) and feces from three individuals were assayed by RT-PCR to compare the amounts of PMMV present. The estimated numbers of virions in 1 g of dry food and feces were 1.21 × 106 (lane 1), 2.3 × 107 (lane 2), 1.63 × 107 (lane 3), 3.64 × 109 (lane 4), 2.42 × 107 (lane 5), and 1.95 × 108 (lane 6) as determined by TaqMan RT-PCR.
(E) Fecal samples from six additional individuals from San Diego were analyzed for detection of PMMV. The positive control is shown in lane 7.
(F) Detection of PMMV in fecal samples of nine individuals from Singapore, including one infant (lane 9). Lane 10 is the positive control.
(G) Detection of PMMV from seven chili sauces from Singapore.
Since the plant viruses identified from the libraries were mostly food crop pathogens, we suspected that the plant viruses found in feces might have originated in food. We then used PMMV as the model to test the existence of plant viruses in food, and estimated the viral loads in food and feces by TaqMan quantitative RT-PCR assay. Three pairs of food-fecal samples from three healthy individuals were collected for total RNA extraction. For each sample pair, small aliquots were taken from all the foods consumed by the participants for 2 d prior to the time of feces collection. The quantity of RNA used for the RT-PCR assay was adjusted based on equal dry weight. As shown in Figure 4D, PMMV was found in both food and feces from all three individuals, and the virus copy number was in the range of 106–109 per gram of dry feces based on the TaqMan quantitative RT-PCR results. The PMMV viral load in feces was enriched 20-fold (Figure 4D; lane 2 versus 1), 224-fold (lane 4 versus 3), and 8-fold (lane 6 versus 5) in pairwise comparisons with food (estimated virus copy numbers are shown in Figure 4). Besides the possible experimental error in measurement, the differences in enrichment of PMMV in feces might reflect the different levels of reduction and condensation of mass from food to feces through absorption during digestion in different individuals and at different times.
To determine the prevalence of PMMV among human fecal samples, six more fecal samples were collected from different individuals in San Diego, California, United States for RT-PCR analysis. Three of these additional samples were positive for PMMV (Figure 4E; lanes 2, 4, and 5). Together with the previous three positive fecal samples tested (Figure 4D; lanes 2, 4, and 6), the overall positive rate was 6/9 (66.7%) in San Diego. To ascertain if PMMV is found only in North Americans, we tested nine human fecal samples (including one from a newborn) from Singapore, and six were positive for PMMV (Figure 4F; lanes 1, 2, and 5–8). Notably, the infant fecal sample was negative for PMMV (lane 9). Hence, the presence of PMMV in feces is not geographically restricted. Among the nine individuals from Singapore, six had had peppers in their recent meals. Interestingly, PMMV was detected in feces from five of them.
Since peppers were indicated as a potential source of PMMV, we collected a panel of pepper-containing foods from San Diego for RT-PCR testing. Out of 22 fresh and processed pepper samples tested, only three were positive, including bottled hot chili sauce and powdered chili (unpublished data). Interestingly, none of the fresh, healthy-looking peppers tested were positive for PMMV. To further investigate the presence of PMMV in chili sauces, seven randomly collected samples from food stalls in Singapore were tested; four of these chili sauces contained PMMV (Figure 4G; lanes 1 and 3–5). Together, these data demonstrate that food is a potential source of PMMV in human feces, and that intact PMMV nucleic acids can survive standard food processing.
We then assessed if the feces-borne PMMV viruses were viable and able to infect host plants. Two Hungarian wax pepper (Capsicum annuum) plants were inoculated with a small amount of PMMV-positive fecal suspension. Within 1 wk, all inoculated plant leaves developed typical symptoms of PMMV infection (Figure 5), while the five control plants remained healthy. Total RNA extracted from the infected and healthy leaves of the same plant were tested for PMMV (Figure 5B). PMMV could only be detected in the infected leaf (Figure 5B; lane 2), and this PCR diagnostic band was confirmed by sequencing. Thus, the PMMV shed from the human GI tract is still infectious to host plants.
Figure 5 Fecally Borne PMMV Is Infectious to a Pepper Plant
(A) Fecal supernatant containing PMMV was inoculated on to a leaf of a Capsicum plant (Day 0). After 7 d of inoculation, this leaf developed typical symptoms of viral infection (Day 7).
(B) RNA extracts from uninfected control leaves (lane 1) and PMMV-positive fecal supernatant challenged leaves (lane 2) were tested for PMMV by RT-PCR.
Discussion
The original intention of this study was to identify human enteric viruses or other human viruses through a metagenomic survey of uncultured RNA viruses in human feces. However, we found that animal viruses were uncommon in healthy human feces and were limited to PBVs. It might be possible that human viruses replicating in mucosal cells were undetectable in stools from healthy individuals but are abundantly shed into feces in inflammatory conditions when symptoms such as diarrhea develop [28]. Therefore, analysis of fecal samples from diarrhea patients would be the logical extension of this study to identify unknown pathogenic human enteric viruses.
Unexpectedly, we discovered a large and diverse community of plant RNA viruses exemplified by PMMV in human feces. PMMV was found in two-thirds of the individuals tested from two different continents (San Diego in North America and Singapore in Southeast Asia), suggesting that PMMV is prevalent in human populations. We also observed that the RNA viral community in human feces was dynamic. The representation of RNA viruses found in a single individual over time or in different individuals varied substantially. Considering that most of the feces-borne plant viruses identified in this study are known to be pathogenic to vegetable crops and that PMMV was detectable in processed pepper food (chili sauces), we speculate that foods might be the major source of fecally borne RNA viruses. Fluctuations of the RNA viral populations in human feces may be associated with the kinds of food consumed, methods of food preparation, and idiosyncratic conditions in the GI tract.
We further provide evidence that the fecal PMMV was viable and could cause infection to a host plant. PMMV viruses appeared to be stable through the human digestive system. This is consistent with the observation that members of Tobamovirus, including PMMV, are extremely robust viruses and can survive in challenging environments [29,30]. The data presented here suggest that humans (and likely other animals) may play a significant role as transmission vectors for certain plant viruses through their digestive tracts. The discovery of this previously unknown transmission route for plant viruses is potentially important for a better understanding of certain agricultural practices, because in traditional agriculture and in developing countries, human and animal refuse are often used as fertilizers for crops.
Since the PMMV was abundant in the fluid phase of fecal material by RT-PCR and library sequencing, it is conceivable that these viruses might disassociate from the plant tissues and therefore are readily accessible to host epithelial cells and microbial cells existing in the intestinal environment. Earlier studies showed that plant viral particles could be assembled in Escherichia coli cells [31,32], which leads to the speculation that some feces-borne plant viruses might be capable of interacting with microbes in human guts. In this study, the amount of fecal PMMV appeared to be significantly higher than the virus load found in food based on equal dry weight. This increase could be a result of the digestive reduction of food, or viral replication in the human gut. PMMV, like other members of Tobamovirus, is a plus-stranded RNA virus [33]. During replication in host cells, a complementary minus-strand RNA is synthesized from the genomic plus-stranded RNA template and serves as the template for genomic RNA. Hence, evidence of viral replication is the presence of the minus-strand RNA. We were unable to detect the minus strand of PMMV as an indication of replication using a strand-specific RT-PCR method [34,35]. Hence, the evidence for active replication of PMMV in human feces is currently lacking. Further research efforts are needed to determine whether and how plant viruses may interact with intestinal cells or microorganisms in the human GI tract.
This study demonstrates the existence of abundant and dynamic populations of live plant pathogenic viruses in human feces. It is already known that complex symbiotic interactions exist between viruses, plant hosts, and their insect vectors [36,37]. For the tomato yellow leaf curl virus, the lifespan of its whitefly vector, Bemisua tabaci, is reduced by one-fifth, and their fecundity is reduced by one-half during viral transmission [38], which suggests that viral replication may affect health of the vector. Given that nonpathogenic commensal intestinal bacteria can significantly alter the immune system of an organism [39], it is conceivable that viruses such as PMMV, replicating or not, may also have an effect. Nevertheless, the stability of PMMV, and perhaps other plant viruses, in the human GI tract may allow them to be used as a platform for oral vaccine development [40] to prevent diarrhea and other intestinal disorders.
Materials and Methods
Fecal samples
Three fecal samples from two healthy adults living in San Diego, California, United States were collected for RNA viral isolation and sequencing analysis. Samples 1 and 2 were from the same individual with a gap of 6 mo between sample collections. Sample 3 was from a second individual. The two human participants in this study had no history or recent symptoms of diarrhea or any known gastrointestinal illness, and had a reasonably balanced daily diet of meat, vegetables, and grains (bread, rice, or noodle).
Additional fecal samples were obtained later for RT-PCR detection of specific feces-borne viruses. Six fecal samples were collected from six healthy humans who worked together in the same building, but lived in different locations in the San Diego area. Subsequently, nine fecal samples were collected from nine individuals living in Singapore (including one infant). For the Singapore group, we undertook a comprehensive record and analysis of food intake prior to sample donation (Table 4). The food intake of participants was documented over a period of 3 d prior to fecal sample collection. Foods were grouped by their main nutritional composition, namely protein, carbohydrate, fat, dietary fiber, and calcium. Foods that contain added microorganisms were categorized as probiotics.
Table 4 Food Intake of Singapore Participants Three Days Prior to Donating Sample
Viral isolation
Viral isolation and concentration were performed as described previously [17]. Briefly, approximately 500 g of fresh fecal matter was resuspended in 5 l of SM buffer (0.1 M NaCl, 50 mM Tris-HCl [pH 7.4], 10 mM MgSO4), and this material was passed through a Nitex filter (pore size ~100 μm) followed by filtration through a 0.2-μm tangential flow filter (Amersham Biosciences, Little Chalfont, United Kingdom). The viral particles were then concentrated on a 100-kDa tangential flow filter (Amersham Biosciences). The viral concentrates were stained with SYBR Gold (Invitrogen, Carlsbad, California, United States) and visualized using epifluorescent microscopy to ensure the recovery of viruses and removal of microbial cells [41]. DNase I and RNase A were added to a final concentration of 1 unit per 100 μl of viral concentrate, and the sample was incubated at 37 °C for 3 h to remove any contamination by free nucleic acids. Viral RNA was extracted using the Qiagen RNA/DNA Midi Kit (Qiagen, Valencia, California, United States) according to the manufacturer's instructions, and was then treated with DNase to eliminate any DNA contamination.
Viral RNA shotgun library construction
Approximately 140–200 ng of purified viral RNA of each sample was utilized for cDNA synthesis using a random priming approach following the protocol from SuperScript Choice System (Invitrogen). Double-stranded cDNA was divided into two aliquots, which were digested separately with either TaqI (New England Biolabs, Beverly, Massachusetts, United States) or Csp6I (Fermentas, Burlington, Ontario, Canada) to generate short overlapping DNA fragments. TaqI- and Csp6I-digested cDNA fragments were ligated with TaqI and Csp6I adaptors (oligo sequences in Table 5) at 16 °C overnight, respectively. The TaqI and Csp6I adaptor-ligated DNA fragments were amplified by PCR using forward primers TaqI-F and Csp6I-F for 40 cycles, respectively. DNA fragments between 500 and 1,000 bp were fractionated and cloned into Zero Blunt TOPO PCR cloning vectors (Invitrogen) for sequencing.
Table 5 Oligonucleotide Sequences of Adaptors, Primers, and Probes Used in This Study
Sequence production and analysis
The viral cDNA clones were prepared as plasmid templates for sequencing by an ABI3730 DNA analyzer (Applied Biosystems, Foster City, California, United States). The raw sequence reads were subjected to base calling and vector/adaptor trimming using PHRED [42,43]. High-quality sequences for all three libraries were searched for homology against the nonredundant nucleotide database (nr) in GenBank using tBLASTx [44,45]. Sequences that had tBLASTx hits with an e-value less than 0.001 were considered known, and the top hit was assigned to the particular sequence for annotation; otherwise they were considered unknown. Based on the best tBLASTx hit, virus-like sequences were further classified based on hosts, such as plant, animal, or bacterial origins. Nonviral sequences were subjected to BLASTn against the European ribosomal RNA database (http://www.psb.ugent.be/rRNA/index.html). Sequences that had BLAST hits to this database with an e-value less than 0.001 were considered to be structural ribosomal RNA.
Sequence clustering
For each library, shotgun viral sequences were clustered into longer viral genome segments. We first used PHRAP [46] with highly stringent parameters (<qual_show 20>, <confirm_length 50>, <confirm_score 100>, <minscore 180>, and <maxgap 0>), and generated 2,553 unique viral genome sequence segments that included 257 contigs and 2,296 singletons from Lib 1, 4,958 (708 contigs and 4,250 singletons) from Lib 2, and 3,546 (405 contigs and 3,141 singletons) from Lib 3. The PHRAP-generated unique viral sequence segments were further clustered using the Sequencher program with the parameters of a minimum of 40 bp overlap and greater than 95% nucleotide identity, which resulted in 1,109 (297 contigs and 812 singletons), 1,746 (409 contigs and 1,337 singletons), and 1,184 (270 contigs and 914 singletons) unique viral sequence segments for Lib 1, 2, and 3, respectively.
Phylogenetic analysis of PMMV sequences
The PMMV CP gene was used for phylogenetic analysis. There are 21 known PMMV CP gene sequences available in GenBank. From the three libraries, 44 assembled fecal PMMV sequence segments were aligned within the first 101 bp of the 474 bp CP gene (from 5,685 to 6,158 bp of the PMMV genome). ClustalW (http://www.ebi.ac.uk/clustalw/) was used for phylogenetic comparison of these sequences, and TreeView (Win32) was used to generate the phylogram tree (http://taxonomy.zoology.gla.ac.uk/rod.html).
RT-PCR detection of fecally borne RNA viruses
Approximately 20 g of each fecal sample was diluted 1:1 with phosphate-buffered saline and homogenized for 1 min at medium speed in a Stomacher 400 (Seward, Thetford, Norfolk, UK) [47,48]. The samples were then centrifuged at 10,000 g for 5 min to sediment particulate matter, and the supernatant was subjected to nucleic acid extraction using the Qiagen viral RNA Mini kit. Approximately 100 ng of fecal RNA was used for first strand cDNA synthesis using virus-specific reverse primers (Table 5). The reverse transcription reaction was performed using SuperScript II RNase H-reverse transcriptase following the manufacturer's protocol (Invitrogen). The RT products (1/10 of total reaction) were used as templates for PCR reactions using virus-specific primer pairs (Table 5) and amplifying for 35 cycles. PCR products were analyzed by electrophoresis on agarose gels. The identities of all PCR products were confirmed by sequencing. All RT-PCR tests included RNA-less and RT-less samples as negative controls.
Food samples analyzed in this study were subjected to the same procedure for RNA extraction and RT-PCR detection of PMMV.
TaqMan assay for quantification of PMMV
To determine the copy number of PMMV in fecal and food samples, a TaqMan assay was used to quantify the viral load in these samples according to the manufacturer's recommendation (Applied Biosystems). Briefly, the assay was performed in a 20-μl reaction mixture containing 2 μl of fecal RNA, five units of MuLV reverse transcriptase, eight units of recombinant RNase inhibitor, 10 μl of 2× universal PCR Master Mix with no UNG (Applied Biosystems), 0.9 μM primers (PMMV-FP1 and PMMV-RP1; Table 5), and 0.2 μM probe (PMMV-Probe1; Table 5). The real-time RT-PCR reactions were carried out in an ABI Prism 7900HT Sequence Detection System (Applied Biosystems). The RT reaction was performed at 48 °C for 30 min, followed by 10 min at 95 °C for activation of DNA polymerase, and then 40 cycles of 15 s at 95 °C and 1 min at 60 °C. Negative controls (RT-less and RNA-less) and serial dilutions of positive control (a plasmid clone) were included in every PCR assay.
Infection of plants with fecally borne viruses
Two PMMV-positive fecal samples (10 g) were resuspended in 10 ml of SM buffer, centrifuged at 12,000 g for 15 min to pellet the bacteria and large debris, and the supernatants were mixed with ground glass. Two healthy Hungarian wax pepper (Capsicum annuum) plants were used for the test, and two leaves per plant were inoculated with 2 μl of the sample supernatant/ground glass mixture with gentle grinding followed by rinsing with sterile water. The inoculated leaves were then monitored visually over a period of 2 wk, and photos were taken with a digital camera to document viral infection symptoms. The leaves that developed viral symptoms were collected for RNA extraction and viral detection by RT-PCR and subsequent sequencing analysis. Negative controls of five plants were subjected to the same inoculation procedure, with the exception of the fecal supernatant addition.
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/) accession numbers of organisms discussed in this paper are PMMV genome (NC_003630) and PBV prototype strains 4-GA-91 (AF246948) and 1-CHN-97 (AF246939).
The authors would like to acknowledge the following colleagues: H. Thoreau, L. Lim, X. Liu, and the Sequencing Group of Genome Institute of Singapore (GIS) for technical support; L. Ling for the Singapore fecal sample collection; K. Srinivasan for help with plant virus classification; and F. Verhoef for assistance on some data analysis. This work is funded by the Agency for Science, Technology and Research (A*STAR), Singapore, and partially supported by National Science Foundation (United States) DEB-BE 04–21955 to FR. MB was funded by a Science to Achieve Results fellowship from the United States Environmental Protection Agency.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. FR and YR conceived and designed the experiments. TZ, MB, JR, CLW, SWLS, and FR performed the experiments. TZ, MB, WHL, SWLS, and YR analyzed the data. TZ and YR wrote the paper. MB, MLH, ETL, and FR helped revise the paper.
Citation: Zhang T, Breitbart M, Lee WH, Run J, Wei CL, et al. (2006) RNA viral community in human feces: Prevalence of plant pathogenic viruses. PLoS Biol 4(1): e3.
Abbreviations
CPcoat protein
GIgastrointestinal
MCMVmaize chlorotic mottle virus
OCSVoat chlorotic stunt virus
PanMVpanicum mosaic virus
PBVpicobirnavirus
PMMVpepper mild mottle virus
RTreverse transcription
==== Refs
References
Backhed F Ley RE Sonnenburg JL Peterson DA Gordon JI Host-bacterial mutualism in the human intestine Science 2005 307 1915 1920 15790844
Xu J Gordon JI Inaugural article: Honor thy symbionts Proc Natl Acad Sci U S A 2003 100 10452 10459 12923294
Nicholson JK Holmes E Wilson ID Gut microorganisms, mammalian metabolism and personalized health care Nat Rev Microbiol 2005 3 431 438 15821725
Weinbauer MG Ecology of prokaryotic viruses FEMS Microbiol Rev 2004 28 127 181 15109783
Gorski A Weber-Dabrowska B The potential role of endogenous bacteriophages in controlling invading pathogens Cell Mol Life Sci 2005 62 511 519 15747058
Glass RI Bresee J Jiang B Gentsch J Ando T Gastroenteritis viruses: an overview Novartis Found Symp 2001 238 5 19 19 25 discussion 11444035
Murray P Rosenthal K Pfaller M Medical microbiology, 5th ed 2005 Philadelphia Mosby/Elsevier 976
Middleton PJ Viruses that multiply in the gut and cause endemic and epidemic gastroenteritis Clin Diagn Virol 1996 6 93 101 15566895
Knipe D Howley P Griffin D Lamb R Martin M Fields virology, 4th ed 2001 Philadelphia Lippincott Williams and Wilkins 3,280 p
Wilhelmi I Roman E Sanchez-Fauquier A Viruses causing gastroenteritis Clin Microbiol Infect 2003 9 247 262 12667234
Clark B McKendrick M A review of viral gastroenteritis Curr Opin Infect Dis 2004 17 461 469 15353966
Dolin R Blacklow NR DuPont H Formal S Buscho RF Transmission of acute infectious nonbacterial gastroenteritis to volunteers by oral administration of stool filtrates J Infect Dis 1971 123 307 312 5111887
Dolin R Blacklow NR DuPont H Buscho RF Wyatt RG Biological properties of Norwalk agent of acute infectious nonbacterial gastroenteritis Proc Soc Exp Biol Med 1972 140 578 583 4624851
Xi JN Graham DY Wang KN Estes MK Norwalk virus genome cloning and characterization Science 1990 250 1580 1583 2177224
Rohwer F Edwards R The phage proteomic tree: A genome-based taxonomy for phage J Bacteriol 2002 184 4529 4535 12142423
Breitbart M Salamon P Andresen B Mahaffy JM Segall AM Genomic analysis of uncultured marine viral communities Proc Natl Acad Sci U S A 2002 99 14250 14255 12384570
Breitbart M Hewson I Felts B Mahaffy JM Nulton J Metagenomic analyses of an uncultured viral community from human feces J Bacteriol 2003 185 6220 6223 14526037
Venter JC Remington K Heidelberg JF Halpern AL Rusch D Environmental genome shotgun sequencing of the Sargasso sea Science 2004 304 66 74 15001713
Tyson GW Chapman J Hugenholtz P Allen EE Ram RJ Community structure and metabolism through reconstruction of microbial genomes from the environment Nature 2004 428 37 43 14961025
Tringe SG von Mering C Kobayashi A Salamov AA Chen K Comparative metagenomics of microbial communities Science 2005 308 554 557 15845853
Osawa S Furuse K Watanabe I Distribution of ribonucleic acid coliphages in animals Appl Environ Microbiol 1981 41 164 168 7224619
Wetter C Serological identification of four tobamoviruses infecting pepper Plant Disease 1984 68 597 599
Wetter C Conti M Pepper mild mottle virus. Descriptions of plant viruses, No. 330 1988 Warwick (United Kingdom) Association of Applied Biologists
Pereira HG Fialho AM Flewett TH Teixeira JM Andrade ZP Novel viruses in human faeces Lancet 1988 2 103 104 2898672
Chandra R Picobirnavirus, a novel group of undescribed viruses of mammals and birds: A minireview Acta Virol 1997 41 59 62 9199719
Banyai K Jakab F Reuter G Bene J Uj M Sequence heterogeneity among human picobirnaviruses detected in a gastroenteritis outbreak Arch Virol 2003 148 2281 2291 14648286
Rosen BI Fang ZY Glass RI Monroe SS Cloning of human picobirnavirus genomic segments and development of an RT-PCR detection assay Virology 2000 277 316 329 11080479
Griffin DW Donaldson KA Paul JH Rose JB Pathogenic human viruses in coastal waters Clin Microbiol Rev 2003 16 129 143 12525429
Castello JD Lakshman DK Tavantzis SM Rogers SO Bachand GD Detection of infectious tomato mosaic tobamovirus in fog and clouds Phytopathology 1995 85 1409 1412
Rogers SO Starmer WT Castello JD Recycling of pathogenic microbes through survival in ice Med Hypotheses 2004 63 773 777 15488645
Jagadish MN Huang D Ward CW Site-directed mutagenesis of a potyvirus coat protein and its assembly in Escherichia coli
J Gen Virol 1993 74 893 896 8492094
Hwang DJ Roberts IM Wilson TM Expression of tobacco mosaic virus coat protein and assembly of pseudovirus particles in Escherichia coli
Proc Natl Acad Sci U S A 1994 91 9067 9071 8090770
Lewandowski D Dawson W Granoff A Webster R Tobamoviruses Encyclopedia of virology, 2nd ed 1999 Philadelphia Elsevier 1780 1783
Craggs JK Ball JK Thomson BJ Irving WL Grabowska AM Development of a strand-specific RT-PCR based assay to detect the replicative form of hepatitis C virus RNA J Virol Methods 2001 94 111 120 11337045
Peyrefitte CN Pastorino B Bessaud M Tolou HJ Couissinier-Paris P Evidence for in vitro falsely-primed cDNAs that prevent specific detection of virus negative strand RNAs in dengue-infected cells: Improvement by tagged RT-PCR J Virol Methods 2003 113 19 28 14500123
Power AG Insect transmission of plant viruses: A constraint on virus variability Curr Opin Plant Biol 2000 3 336 340 10873852
Hogenhout SA Redinbaugh MG Ammar el D Plant and animal rhabdovirus host range: A bug's view Trends Microbiol 2003 11 264 271 12823943
Rubinstein G Czosnek H Long-term association of tomato yellow leaf curl virus with its whitefly vector Bemisia tabaci Effect on the insect transmission capacity, longevity and fecundity J Gen Virol 1997 78 2683 2689 9349491
Macpherson AJ Harris NL Interactions between commensal intestinal bacteria and the immune system Nat Rev Immunol 2004 4 478 485 15173836
Yusibov V Hooper DC Spitsin SV Fleysh N Kean RB Expression in plants and immunogenicity of plant virus-based experimental rabies vaccine Vaccine 2002 20 3155 3164 12163267
Noble RT Fuhrman JA Use of SYBR Green I for rapid epifluorescence counts of marine viruses and bacteria Aquat Microb Ecol 1998 14 113 118
Ewing B Green P Base-calling of automated sequencer traces using phred. II. Error probabilities Genome Res 1998 8 186 194 9521922
Ewing B Hillier L Wendl MC Green P Base-calling of automated sequencer traces using phred. I. Accuracy assessment Genome Res 1998 8 175 185 9521921
Altschul SF Gish W Miller W Myers EW Lipman DJ Basic local alignment search tool J Mol Biol 1990 215 403 410 2231712
Altschul SF Madden TL Schaffer AA Zhang J Zhang Z Gapped BLAST and PSI-BLAST: A new generation of protein database search programs Nucleic Acids Res 1997 25 3389 3402 9254694
Green P Phragment assembly program, version 0.99 [computer program] 1996 Available: http://bozeman.mbt.washington.edu/phrap.docs/phrap.html . Accessed 1 November 2005
Andrews WH Wilson CR Poelma PL Romero A Rude RA Usefulness of the stomacher in a microbiological regulatory laboratory Appl Environ Microbiol 1978 35 89 93 623476
Purvis U Sharpe AN Bergener DM Lachapelle G Milling M Comparison of bacterial counts obtained from naturally contaminated foods by means of stomacher and blender Can J Microbiol 1987 33 52 56 3552165
|
16336043
|
PMC1310650
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e3
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040003
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633604410.1371/journal.pbio.0040004Research ArticleCell BiologyImmunologyInfectious DiseasesMicrobiologyIdentification of Drosophila Gene Products Required for Phagocytosis of Candida
albicans
Phagocytosis of C. albicans by DrosophilaStroschein-Stevenson Shannon L
1
Foley Edan
2
O'Farrell Patrick H
2
Johnson Alexander D [email protected]
1
2
1Department of Microbiology and Immunology, University of California, San Francisco, California, United States of America2Department of Biochemistry and Biophysics, University of California, San Francisco, California, United States of AmericaSchneider David Academic EditorStanford UniversityUnited States of America1 2006 20 12 2005 20 12 2005 4 1 e415 8 2005 27 10 2005 Copyright: © 2006 Stroschein-Stevenson et al.2006This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Fly Genes That Help Devour a Fungal Parasite
Phagocytosis is a highly conserved aspect of innate immunity. We used Drosophila melanogaster S2 cells as a model system to study the phagocytosis of Candida albicans, the major fungal pathogen of humans, by screening an RNAi library representing 7,216 fly genes conserved among metazoans. After rescreening the initial genes identified and eliminating certain classes of housekeeping genes, we identified 184 genes required for efficient phagocytosis of C. albicans. Diverse biological processes are represented, with actin cytoskeleton regulation, vesicle transport, signaling, and transcriptional regulation being prominent. Secondary screens using Escherichia coli and latex beads revealed several genes specific for C. albicans phagocytosis. Characterization of one of those gene products, Macroglobulin complement related (Mcr), shows that it is secreted, that it binds specifically to the surface of C. albicans, and that it promotes its subsequent phagocytosis. Mcr is closely related to the four Drosophila thioester proteins (Teps), and we show that TepII is required for efficient phagocytosis of E. coli (but not C. albicans or Staphylococcus aureus) and that TepIII is required for the efficient phagocytosis of S. aureus (but not C. albicans or E. coli). Thus, this family of fly proteins distinguishes different pathogens for subsequent phagocytosis.
Mcr and the closely related Drosophila Tep proteins (proteins similar to mammalian secreted immune complement) bind to the surface of invading microbes and are required to promote the phagocytosis of specific pathogens.
==== Body
Introduction
Dedicated host phagocytic cells suppress microbial proliferation by engulfing pathogens and subsequently engaging additional host defenses through cytokine production and antigen presentation [1–3]. Pathogen recognition activates signaling pathways within the phagocytic cell that induce the rearrangement of the actin cytoskeleton and thereby lead to engulfment of the pathogen. For example, signaling activates Rho family GTPases, which regulate actin assembly during phagocytosis [4]. In addition to actin rearrangements, insertion of new membrane occurs at the site of phagosome assembly, contributing to the membrane protrusions elaborated during engulfment [5,6]. Although the morphological changes and cytoskeletal rearrangements that occur during phagocytosis have been well described, there have been, until recently, few attempts to comprehensively identify the cellular components that are required for phagocytosis, and these have used bacterial species [7–10]. With this in mind, we have undertaken a systematic approach to identify host genes required for the efficient phagocytosis of a fungal pathogen, Candida albicans.
C. albicans is a common commensal fungal organism found in the gastrointestinal tract and other tissues of more than 50% of healthy adults [11]. Extremes of age, injury, antibiotic use, and a compromised immune response predispose individuals to the development of mucosal or life-threatening systemic infections. C. albicans is the now the fourth most common organism detected in systemic infections [12], and mortality approaches 35% [13]. The predisposition of neutropenic and HIV+ patients with decreased CD4+ T cells to C. albicans infections suggests that both innate immunity and acquired cell-mediated immunity are involved in mediating host resistance to C. albicans infections [3,14].
The genetically tractable fruit fly, Drosophila melanogaster, is a well-established system for studying conserved components of innate immunity [1,2,15]. For example, studies in Drosophila were instrumental in revealing the significance of Toll signaling in the innate immune response [16,17]. In addition, Drosophila has been successfully used to study the interaction of several human pathogens—including Listeria monocytogenes, Plasmodia, Mycobacterium marinum, and C. albicans—with conserved features of the innate immune system [18–22].
Drosophila plasmatocytes are macrophage-like cells and the predominant of three distinct hemocyte types: they phagocytose cell debris and invading microbes. Phagocytosis appears important for full immunity against pathogens, as blocking phagocytosis in imd mutant flies sensitized them to infection with Escherichia coli [1]. The Drosophila S2 cell line is believed to be derived from embryonic plasmatocytes and shares many properties with plasmatocytes, including robust phagocytosis [23]. The advent of RNA interference (RNAi) and the availability of genomic RNAi libraries open new possibilities for the exploration of phagocytosis in this model system. RNAi in S2 cells has been used to systematically study phagocytosis of E. coli, L. monocytogenes, and Mycobacterium fortuitum [7,9,10], and RNAi in vivo in Anopheles gambiae to study phagocytosis of E. coli and Staphylococcus aureus [8]. In this paper, we describe a high-throughput assay to study the phagocytosis of C. albicans by the S2 cell line. Using this RNAi-based screen, we have explored the roles of 7,216 evolutionarily conserved genes in the phagocytosis of C. albicans by S2 cells. From this screen, we identified 184 genes required for the efficient phagocytosis of C. albicans. In many cases, we identified multiple subunits of known protein complexes or multiple components of known biochemical pathways. In secondary assays, we distinguished genes specifically required for the phagocytosis of C. albicans from those required generally for phagocytosis. We performed additional experiments with one such Candida-specific component, Macroglobulin complement related (Mcr), a protein highly conserved in metazoans (CG7586). We show that Mcr is required for the efficient phagocytosis of C. albicans, but not that of E. coli, S. aureus, or latex beads. We also demonstrate that Mcr is secreted into the media of S2 cells and binds the cell surface of C. albicans, presumably leading to recognition and phagocytosis by S2 cells. We also show that Mcr exhibits specificity in its recognition of the cell surface of C. albicans. Mcr binds more to wild-type C. albicans than to a C. albicans mutant (Δefg1/Δefg1) or to the common laboratory strain of Saccharomyces cerevisiae (S288c), both of which are poorly phagocytosed. Mcr has four close relatives in the Drosophila genome, thioester proteins (Tep), known as TepI, TepII, TepIII, and TepIV (TepI: CG18096; TepII: CG7052; TepIII: CG7068; TepIV: CG10363). None of the four Teps is required for efficient phagocytosis of C. albicans; however, we show that TepII is required for efficient phagocytosis of the gram-negative bacteria E. coli and TepIII for efficient phagocytosis of the gram-positive bacteria S. aureus. These findings show that different members of this conserved group of five proteins show specificity for different pathogens and support studies conducted in mosquito that suggested that these proteins act as a primitive complement system targeting pathogens for immune destruction [8,24,25].
Results
Phagocytosis of C. albicans by Drosophila S2 Cells
D. melanogaster is rapidly emerging as a model system to study numerous human pathogens including C. albicans [7,9,18–20,22]. To investigate whether the hemocyte-derived fly S2 cell line efficiently phagocytoses C. albicans, we co-incubated green fluorescent protein (GFP)–expressing C. albicans with S2 cells, fixed the samples, and monitored adherence and phagocytosis of C. albicans (Figure 1A). Within a few minutes of mixing, many S2 cells were bound by C. albicans, and some S2 cells exhibited actin-rich cytoplasmic protrusions forming toward the C. albicans (arrows in panel i in Figure 1A). By 30 min, evidence of C. albicans phagocytosis was clear, with the actin cytoskeleton forming pseudopodia engulfing the C. albicans (arrowheads in panel ii in Figure 1A). After 1 h of co-incubation, many S2 cells had engulfed at least one C. albicans cell (panel iii in Figure 1A). Engulfment of C. albicans led to changes in S2 cell morphology, including condensation and displacement of the DNA. To quantify the phagocytosis, we scored the percentage of S2 cells that had internalized one or more C. albicans cells (Figure 1B). Under these conditions, approximately 50% of S2 cells phagocytosed at least one C. albicans within 3 h. Phagocytosis did not critically depend on the C. albicans cells being alive, as heat-killed C. albicans cells were phagocytosed in significant amounts (Figure 1B). The difference between live and heat-killed C. albicans is probably attributable to the fact that live C. albicans were actively dividing over the timecourse, whereas the numbers of heat-killed C. albicans obviously did not increase. In contrast to C. albicans, a common laboratory strain of S. cerevisiae (S288c) was poorly phagocytosed, indicating that there may be a specific mechanism of recognition of C. albicans not shared by all fungi. We detected only a low level of adherence of S. cerevisiae to the surface of S2 cells, indicating the reduced phagocytosis is likely due, at least in part, to a difference in recognition of S. cerevisiae versus that of C. albicans (unpublished data).
Figure 1 Phagocytosis of C. albicans by Drosophila S2 Cells
(A) The Drosophila hemocyte-like S2 cell line phagocytoses C. albicans. S2 cells were co-incubated with GFP expressing C. albicans for the indicated times. Cells were fixed, and the filamentous actin of S2 cells was stained with rhodamine phalloidin and the S2 cell DNA with Hoechst 33258.
(B) Quantification of phagocytosis of C. albicans and S. cerevisiae by S2 cells. S2 cells and the indicated fungal strain were co-incubated for various times, and the percentage of S2 cells that had phagocytosed one or more C. albicans was quantified by counting 50–100 S2 cells. The maximum time shown is 3 h, as the levels of phagocytosis did not significantly increase after this timepoint. Results are the average of four experiments, and the error bars indicate the standard deviation. As described in Materials and Methods, the 3-h results were evaluated for statistical significance using the t-test, assuming unequal variance. As indicated by the asterisks, the values for heat-killed C. albicans and S. cerevisiae were statistically different from that of live C. albicans, with a confidence level p < 0.01.
Identification of Genes Required for Phagocytosis of C. albicans
As phagocytosis is such a critical element of metazoan immune responses to invading microorganisms, we used a library of 7,216 double-stranded RNAs (dsRNAs) representing most of the phylogenetically conserved genes of D. melanogaster to identify cellular components responsible for phagocytosis of C. albicans. In an initial visual screen, S2 cells treated with dsRNA for 4 d were mixed with GFP-expressing C. albicans. Cells were lightly fixed without detergent to prevent permeabilization of the plasma membrane and stained with a polyclonal antibody against whole-cell lysates of C. albicans and a Cy3 conjugated secondary antibody. Under these conditions, C. albicans cells in the media fluoresce green with a red outline, while those phagocytosed by S2 cells appear only green (Figure 2A). Wells were visually scored for a reduction in the number of phagocytosed cells, visible as green-only cells (Figure 2A). The effect of each of the 7,216 dsRNAs was independently scored by two investigators. The small numbers of discrepancies in the initial pass were resolved through additional screening. In addition, wells with reduced phagocytosis were further investigated to rule out a general cytotoxic effect for a decrease in phagocytosis, as described in Materials and Methods. In all cases, the scoring was carried out without knowledge of the genes represented by the dsRNA.
Figure 2 Identification of Genes Required for Phagocytosis of C. albicans
(A) High-throughput assay for phagocytosis. GFP-expressing C. albicans (green) were co-incubated with S2 cells to allow phagocytosis. Cells were lightly fixed, and non-phagocytosed C. albicans were secondarily labeled with a rabbit anti–C. albicans antibody and Cy3-labeled anti-rabbit antibody (red). S2 cell DNA (blue) was labeled with Hoechst 33258. Left panels, wild-type S2 cells; right panels: S2 cells treated with RNAi against actin (Act5C).
(B) One hundred eighty-four dsRNAs decreased phagocytosis of C. albicans. The 184 genes were categorized and plotted in a pie graph with the number of genes in each class indicated.
(C–E) Secondary screens used to further test the RNAi-treated S2 cells specificity of phagocytosis of C. albicans (C), E. coli (D), or latex beads (E). (C) Phagocytosis of C. albicans. GFP-expressing C. albicans (green) were co-incubated with S2 cells to allow phagocytosis. Cells were lightly fixed, and non-phagocytosed C. albicans were secondarily labeled with an anti–C. albicans antibody and Cy3-labeled anti-rabbit antibody (red). S2 cell DNA was labeled with Hoechst 33258 to mark the position of the S2 cell. The level of phagocytosis was quantified by counting the percentage of S2 cells that had phagocytosed one or more C. albicans. (D) Drosophila S2 cells phagocytose E. coli. GFP-expressing E. coli (green) were co-incubated with S2 cells to allow phagocytosis in a similar assay as (C). Cells were lightly fixed, and an anti–E. coli antibody was used to label non-phagocytosed E. coli (red). The level of phagocytosis was quantified by counting the percentage of S2 cells that had phagocytosed one or more E. coli. (E) Drosophila S2 cells phagocytose 2-μm latex beads. Yellow-green fluorescent latex beads were co-incubated with S2 cells to allow phagocytosis. The S2 cell filamentous actin cytoskeleton was labeled with rhodamine phalloidin (red) and the DNA with Hoechst 33258 (blue). The level of phagocytosis was quantified by counting the percentage of S2 cells that had phagocytosed one or more latex beads.
(F) One hundred eighty-four dsRNAs disrupt the phagocytosis of C. albicans by S2 cells. The genes required for phagocytosis of C. albicans are listed along with the effect on phagocytosis of E. coli and latex beads. The color-based scale is given below and corresponds to the percentage of S2 cells that phagocytosed one or more C. albicans, E. coli, or latex beads. Genes were categorized based upon function as in (B). Mean values for phagocytosis by wild-type, untreated S2 cells were: C. albicans 52%, E. coli 56%, and latex beads 51%.
From this screen, we initially identified 401 dsRNAs that significantly decreased S2 cell phagocytosis of C. albicans. A significant number of these genes represented different subunits of large protein assemblies that carry out general aspects of gene expression and protein turnover in the cell. For example, we identified 45 different ribosomal subunits, 31 RNA processing enzymes, 15 general transcription proteins, and 30 proteasome subunits (Table S1). The high frequency of dsRNAs in these categories is not due to lack of specificity of the approach. Indeed, the fact that we identified so many of the ribosome and proteasome subunits indicates a high degree of internal consistency. We reasoned that genes involved in these generic processes are likely to influence phagocytosis indirectly and hence have excluded them from further analysis. We note that another screen for phagocytosis of intracellular bacteria obtained and excluded a large class of ribosome and proteasome genes [7].
We resynthesized and retested the remaining 280 dsRNAs for reduction of C. albicans phagocytosis. For this analysis, we quantified the efficiency of phagocytosis and rejected dsRNAs whose effects failed to meet a specified criterion. The levels of C. albicans phagocytosis were quantified as the percentage of S2 cells that had phagocytosed one or more C. albicans, based on a sample size of roughly 100 S2 cells for each dsRNA. We defined a significant decrease in phagocytosis as anything less than 1.5 standard deviations below the mean for untreated cells (52%). Thus, dsRNAs that reduced the percentage of S2 cells phagocytosing C. albicans to 44% or below were scored as statistically significant as explained in Materials and Methods. One hundred eighty-four of the dsRNAs passed this additional test, which was more rigorous than our initial screen (Figure 2B); thus, 96 genes, many of which were probably false positives, were eliminated in the secondary screen. This new set of 184 genes includes 52 genes known from prior studies [4,6,7,9,26–28] to function in phagocytosis and 132 genes whose roles in phagocytosis have not been previously described. Of the genes previously known to function in phagocytosis, Rac1, Rac2, Cdc42, other regulators of actin dynamics, actin itself, all five CopI vesicle proteins, Syx5a (t-Snare), Snap, PI3K, and InaC (Protein Kinase C) were identified in the screen, thereby confirming the validity of the methodologies (Figure 2 and Table S2). Because the dsRNA collection represented conserved genes, many of the biochemical functions of the genes not previously implicated in pathogenesis can be surmised. The 184 genes in the screen can be broken down into the following categories: (1) actin, actin-regulating, and actin-binding proteins (21 genes); (2) vesicle transport, including all five CopI vesicle coat proteins (Cop), several snare proteins, and several regulators of vesicle function (16 genes); (3) sequence-specific DNA-binding proteins that presumably regulate gene transcription (30 genes); (4) signaling components, seven of which have been previously implicated in phagocytosis (27 genes); (5) a set of genes annotated as being involved in immunity and defense against pathogens (eight genes); (6) a catchall category of miscellaneous functions, including protein degradation, metabolism, protein transport, and protein folding (57 genes); and (7) conserved genes having no known function (25 genes).
Agaisse et al. and Philips et al. [7,9] described genes required for the microbial entry and survival in S2 cells of two bacteria, M. fortuitum and L. monocytogenes, by S2 cells, and it is useful to compare the results of these screens with our results for C. albicans. Of the 184 genes identified in our screen, only 21 and 33 genes, respectively, are shared with these two other screens. Most of the overlapping genes encode actin regulatory proteins and vesicle transport proteins, suggesting these processes are particularly important for phagocytosis. There are several possible reasons why the overlap among the screens is not more extensive, and they are taken up in the Discussion.
Of the genes identified in our screen, we wished to distinguish those specifically required for phagocytosis of C. albicans from those with a more general role in phagocytosis. We therefore tested all 184 dsRNAs that impaired C. albicans phagocytosis for their effects on the phagocytosis of GFP-expressing E. coli and yellow-green fluorescently labeled latex beads (Figure 2C–2E). Wild-type S2 cells efficiently phagocytose E. coli (Figure 2D) and, to a much lesser extent, latex beads (Figure 2E). The phagocytosis of latex beads required much longer incubation times to detect significant amounts of phagocytosis, an observation that presumably reflects the lack of pathogen-specific signals on their surface. While most dsRNAs identified in the screen affected phagocytosis of all three challengers, a small number affected phagocytosis of C. albicans only. Other dsRNAs reduced phagocytosis of all three challengers but affected C. albicans to a much greater extent than that of E. coli or latex beads (Figure 2F; see Table S2). For the remainder of this study, we concentrate on one dsRNA that specifically impaired phagocytosis of C. albicans. This RNA corresponds to a gene annotated in Flybase as Drosophila Mcr.
Mcr-Dependent Phagocytosis of C. albicans
The Drosophila Mcr protein is a member of the α2Macroglobulin/complement family of proteins (Figure 3A). To confirm its selectivity for C. albicans, we quantified with a detailed timecourse the relative phagocytosis of C. albicans, E. coli, and latex beads by S2 cells treated with dsRNA against Mcr or, as a control, against SCAR dsRNA (Figure 3B–3E). SCAR is an actin-nucleating protein that is required for the phagocytosis of C. albicans, E. coli, and latex beads (see Table S2). RNAi against Mcr reduced phagocytosis of C. albicans, but not E. coli or latex beads when compared with untreated cells. In contrast, RNAi against SCAR strongly affected all three reactions (Figure 3B–3E). To confirm that the reduction in phagocytosis after treatment with Mcr dsRNA was indeed due to a reduction in the Mcr gene product, we designed a second dsRNA against the 3′ untranslated region (UTR) of Mcr. RNAi against either Mcr or the Mcr 3′ UTR similarly reduced phagocytosis of C. albicans, confirming that Mcr is indeed important for phagocytosis of C. albicans (Figure 3F).
Figure 3 Mcr-Dependent Phagocytosis of C. albicans
(A) Schematic representation of α2M-related proteins. Drosophila Mcr is compared with a close homolog in A. gambiae (Ag Mcr [Tep13]), TepI from both Drosophila and Anopheles and the human homologs CD109, α2M, and C3. Various conserved domains are colored as indicated in the gray box. Numbers correspond to amino acid position. The sequences of the conserved thioester domains are given below the schematic. Dm, Drosophila melanogaster; Ag, Anopheles gambiae; Hs, Homo sapiens.
(B) RNAi against SCAR reduces phagocytosis of C. albicans, E. coli, and latex beads (row 2). RNAi against Mcr significantly decreased phagocytosis of only C. albicans (row 3). Cells were stained as in Figure 4C–4E. Column 1, GFP expressing C. albicans—green, S2 cell DNA—blue, non-phagocytosed C. albicans—red; column 2, GFP-expressing E. coli—green, S2 cell DNA—blue, non-phagocytosed E. coli—red; column 3, latex beads—green, S2 cell DNA—blue, S2 cell actin cytoskeleton—red.
(C) dsRNA against both SCAR and Mcr decreases phagocytosis of C. albicans. S2 cells were treated with dsRNA against SCAR and Mcr as described in Materials and Methods and then co-incubated with C. albicans for the indicated times. The percentage of S2 cells phagocytosing one or more C. albicans was quantified and plotted. The 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
(D) Mcr dsRNA does not reduce phagocytosis of E. coli. S2 cells were treated with dsRNA against SCAR and Mcr and then co-incubated with E. coli for the indicated times. The percentage of S2 cells phagocytosing one or more E. coli was quantified and plotted. The 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
(E) Mcr dsRNA does not reduce phagocytosis of latex beads. S2 cells were treated with dsRNA against SCAR and Mcr and then co-incubated with green fluorescent latex beads for the indicated times. The percentage of S2 cells phagocytosing one or more latex beads was quantified and plotted. The 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
(F) An additional dsRNA against the 3′ UTR of Mcr was generated and tested for disruption of C. albicans phagocytosis. S2 cells were treated with RNAi directed against both the coding region and the 3′ UTR of Mcr and then co-incubated with C. albicans for the indicated times. The percentage of S2 cells phagocytosing one or more C. albicans was quantified and plotted. The 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
The Mcr/Tep Family of Proteins Determine Specificity of Pathogen Phagocytosis by Drosophila S2 Cells
Mcr is closely related to a family of four Teps in Drosophila [29]. Indeed, Mcr has been referred to in at least one publication as Tep6 [30]; however, it lacks the cysteine residue that forms the defining thioester of the Teps. TepI, TepII, TepIII, and TepIV were represented in our library of dsRNAs, yet Mcr was the only family member identified in the screen as being required for phagocytosis of C. albicans. This finding and characterization of mosquito Tep1 suggest that Mcr and the four Teps may be involved in the phagocytosis of specific classes of pathogens [8,25]. To test this idea, S2 cells treated with dsRNA against SCAR, Mcr, and each of the four Teps were tested for their ability to phagocytose three different pathogens (Figure 4). To represent a broad spectrum of pathogens, we used C. albicans, E. coli (a gram-negative bacterium), and S. aureus (a gram-positive bacterium). RNAi against SCAR reduced phagocytosis of all three pathogens (Figure 4A–4C). Of Mcr and the Teps, phagocytosis of C. albicans was only decreased by dsRNA against Mcr (Figure 4A), an outcome predicted from the results of the screen. Phagocytosis of E. coli was decreased only by TepII dsRNA (Figure 4B), and S. aureus by TepIII dsRNA (Figure 4C). These results indicate that Mcr and the four Teps constitute a family of proteins, the members of which provide specificity in the phagocytosis of different pathogens.
Figure 4 The Mcr/Tep Family of Proteins Determine Specificity of Pathogen Phagocytosis by Drosophila S2 Cells
(A) S2 cells were treated with dsRNA against SCAR, Mcr, or one of the Drosophila Teps and co-incubated with C. albicans. The percentage of S2 cells phagocytosing one or more C. albicans was quantified and plotted.
(B) The S2 cells treated above were also co-incubated with E. coli, and phagocytosis was quantified.
(C) The RNAi-treated S2 cells above were co-incubated with S. aureus, and phagocytosis was quantified. In all graphs, the 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
S2 Cells Secrete Mcr into the Culture Medium
It seems likely that Mcr is a secreted protein, since it contains a putative signal sequence and is related to mammalian α2M, to mammalian complement components, and to Anopheles aTep1, all of which are known to be secreted (see Figure 4). Cell lysates and conditioned media from S2 cells were analyzed by SDS-PAGE and Western blotting with a primary antibody raised against a peptide located near the amino terminus of Mcr (Figure 5A). Full-length Mcr (approximately 200 kD) was detected in both cell lysates and in conditioned media, indicating that Mcr is a secreted protein. To verify that this 200-kD species is Mcr, we analyzed its levels in cells and media that were treated with RNAi (Figure 5B). RNAi against SCAR did not significantly change the levels of Mcr in either cell lysate or conditioned medium, whereas RNAi against Mcr significantly reduced the levels of the 200-kD protein in both of them (Figure 5B). If Mcr secretion is relevant to phagocytosis, providing Mcr in conditioned media should alleviate the phagocytosis defect in S2 cells treated with RNAi against Mcr. At the start of the phagocytosis assay, untreated S2 cells, SCAR RNAi-treated S2 cells, and Mcr RNAi-treated S2 cells were diluted into either fresh media or conditioned media (from untreated S2 cells) and tested for phagocytosis of C. albicans (Figure 5C). SCAR RNAi and Mcr RNAi both severely suppressed phagocytosis in fresh media. The addition of conditioned media slightly increased phagocytosis by SCAR-treated cells but significantly increased phagocytosis by Mcr-treated cells to levels near those of untreated cells. Conditioned media from Mcr RNAi–treated cells only partially increased phagocytosis activity, indicating that conditioned media from wild-type S2 cells is required for a full rescue of phagocytosis (unpublished data). These results indicate that Mcr presence in the conditioned media is required to rescue the loss of phagocytosis by Mcr RNAi treatment (Figure 5C). These experiments also indicate that Mcr is efficiently synthesized and secreted by S2 cells prior to their exposure to C. albicans.
Figure 5 S2 Cells Secrete Mcr into the Culture Media
(A) Mcr is secreted into the culture media. Whole-cell lysates were prepared from S2 cells (lane 1) and compared to Schneider's medium with 2% FBS (lane 2) or Schneider's medium with 2% FBS collected from S2 cells (conditioned media, lane 3) by immunoblotting with an anti-Mcr antibody.
(B) RNAi against Mcr depletes Mcr protein from cell lysates and from the conditioned media. Cell lysates and conditioned media were collected from wild-type S2 cells or cells treated with RNAi against Mcr or SCAR and probed by immunoblotting with an anti-Mcr antibody.
(C) Conditioned media rescues the phagocytosis defect of Mcr RNAi-treated cells. Wild-type S2 cells or cells treated with RNAi against Mcr or SCAR were plated in new Schneider's medium with 10% FBS or conditioned media with 10% FBS from wild-type S2 cells and incubated with C. albicans for various times. The percentage of S2 cells that had phagocytosed one or more C. albicans was quantified and graphed. A t-test was used to test the statistical significance between wild-type cells in new media versus SCAR- or Mcr RNAi–treated cells in new media and wild-type cells in conditioned media versus SCAR- or Mcr RNAi–treated cells in conditioned media (see Materials and Methods). An asterisk indicates comparisons that showed statistically significant differences (p < 0.01). Mcr RNAi–treated cells in wild-type-conditioned media were not significantly different from wild-type cells in conditioned media.
(D) Mcr interacts with C. albicans cells. C. albicans was co-incubated either with new media containing 2% FBS or conditioned media containing 2% FBS from wild-type S2 cells for 2 h, washed, and analyzed by immunoblotting with anti-Mcr. Lane 1, S2 cell lysates; lane 2, new media; lane 3, conditioned media; lane 4, C. albicans incubated in new media; lane 5, C. albicans incubated in conditioned media.
Specific Binding of Mcr to the C. albicans Cell Surface
To test whether Mcr can directly bind to the surface of C. albicans cells, C. albicans cells were incubated with conditioned media from untreated S2 cells, precipitated, washed extensively, and analyzed by Western blotting using the Mcr antibody. As shown in Figure 5D, the Mcr in the conditioned media efficiently bound to C. albicans and remained bound during the extensive washing steps. The binding of Mcr to C. albicans is specific, as demonstrated by the following two experiments. First, the binding of Mcr shows a marked preference for C. albicans (strain Caf2–1) over the common S. cerevisiae lab strain S288C (Figure 6A). This observation parallels the more efficient phagocytosis of C. albicans Caf2–1 compared with S. cerevisiae S288c (see Figure 1B and Figure 6B). Second, the binding of Mcr shows a preference for wild-type C. albicans over a mutant of C. albicans deleted for the EFG1 gene, Δefg1/Δefg1 (see Figure 6A). EFG1 encodes a transcriptional regulator that regulates many genes, some of which affect properties of the cell wall [31,32]. Although we do not know the precise defects in the Δefg1/Δefg1 strain, the fact that Mcr binds more poorly to it than to a wild-type strain argues that Mcr binding must be specific for some feature on the C. albicans cell surface. As predicted from this idea, we found that the Δefg1/Δefg1 mutant is poorly phagocytosed by S2 cells (see Figure 6B). These experiments, taken together, support the idea that Mcr recognizes features of the C. albicans cell surface and that binding of Mcr to C. albicans results in its efficient phagocytosis by S2 cells.
Figure 6 Specific Binding of Mcr to the C. albicans Cell Surface
(A) Wild-type C. albicans, S. cerevisiae, or Δefg1/Δefg1 mutant C. albicans were co-incubated with conditioned media containing 2% FBS from wild-type S2 cells for 2 h, washed, and analyzed by immunoblotting with anti-Mcr antibody. Lane 1, conditioned media; lane 2, new media; lane 3, Mcr bound to wild-type C. albicans; lane 4, Mcr bound to S. cerevisiae; lane 5, Mcr bound to Δefg1/Δefg1 mutant C. albicans.
(B) Quantification of phagocytosis of C. albicans wild-type and mutant strains and S. cerevisiae by S2 cells. S2 cells and the indicated fungal strain were co-incubated for various times, and the percentage of S2 cells that had phagocytosed one or more C. albicans was quantified by counting 50–100 S2 cells. Results are the average of four experiments, and the error bars indicate the standard deviation. The 3.5-h timepoints were analyzed using a t-test assuming unequal variance. Those values that differ significantly from untreated cells (p < 0.01) are indicated by asterisks.
Discussion
Phagocytosis of invading pathogens is a critical component of metazoan innate immune systems. In this study, we investigate the phagocytosis of C. albicans, the most prevalent fungal pathogen of humans, using Drosophila as a model host organism. Drosophila has been well established as a model system for analyzing human microbial pathogens [18–22].
Identification of Genes Important for Phagocytosis of C. albicans by Fly S2 Cells
Although the morphological features and cytoskeletal rearrangements underlying phagocytosis are well described, there have been few comprehensive attempts to identify the core requirements for phagocytosis. None of these have investigated phagocytosis of a fungal pathogen [7–10,33]. Using a dsRNA library representing 7,216 Drosophila genes conserved in other metazoans, we carried out an RNAi-based screen to identify genes required for efficient phagocytosis of C. albicans by S2 cells. Following rescreening and after eliminating genes encoding ribosomal subunits, proteasome subunits, general transcription factors, mRNA processing enzymes, and other components involved in general aspects of gene expression and protein turnover, we identified 184 genes required for the efficient phagocytosis of C. albicans by S2 cells. Among genes well known to function in phagocytosis, we identified actin itself, many of its regulators including SCAR, and multiple components of the Arp2/3 complex [4]. Phagocytosis involves extensive rearrangements of the cytoskeleton, and the identification of these genes in the screen was strongly predicted from prior work [4,7,9]. We also identified all five Cops in the CopI vesicle coat and several SNARE proteins. CopI vesicles are thought to be indirectly required for pseudopod extension because they are required for the maintenance of a pool of VAMP3 (vSNARE)–containing endomembrane vesicles. These vesicles are thought to be needed for insertion into the plasma membrane during pseudopod formation and phagocytosis [6]. The extent of the increase in surface area is dramatized by the large increase in size of S2 cells when phagocytosing several Candida (see Figure 1A). We also identified a number of previously implicated signaling components, including PI3K, which is required for both insertion of exocytic membranes in the plasma membrane [27,28] and phagosome maturation [26].
It is important to note that although our screen identified the majority of genes previously implicated in phagocytosis it did not identify all of them. For example, although three subunits of the Arp2/3 complex were identified in the screen, four were not, even though their corresponding dsRNAs were present in our library. There are several possible explanations for this type of failure, including poor RNAi, potentially lethal effects of certain RNAs, and the timing of RNAi treatment. For example, in the case of profilin, the timing of RNAi treatment was critical for detecting its role in cytokinesis [34]. Thus, genes present in the library but not identified in our screen can be rigorously excluded from having a role in phagocytosis of C. albicans only through additional experimentation. In any case, the fact that the screen, which was carried out blindly with respect to the identities of the dsRNAs, identified many known components of phagocytosis confirms the validity of the approach. It also implicates the large number of remaining genes as having important roles in this process. All 184 genes identified in the screen are summarized in Table S2. Taken as a whole, these genes represent a number of different cellular processes, including cytoskeletal rearrangements (21 genes), vesicle transport (16 genes), signal transduction (27 genes), and transcriptional regulation (30 genes) (see Results and Tables S1 and S2). We also identified three looser categories of genes: a set of genes encoding diverse functions in metabolism, protein turnover, and transport (57 genes); a set of genes annotated in Flybase as implicated in defense (eight genes); and a set of genes of unknown function, despite their being conserved among metazoans (25 genes).
Giot et al. [35] described a partial interaction map of Drosophila proteins, based largely on two-hybrid experiments. If the genes identified by our screen are superimposed on this map (keeping in mind that, because the interaction map is incomplete, many genes are simply not represented in the map), they form a highly interactive network that covers only a portion of the total map space (Figure 7). Many of the newly identified components can be seen to interact directly or indirectly with a known component of phagocytosis. From this analysis, many testable predictions can be made regarding the roles of specific gene products identified in the phagocytosis screen. For example, an unstudied transmembrane protein with predicted Ser/Thr kinase activity (CG5790) interacts with βCop and therefore may be directly involved in vesicle transport during phagocytosis or may be critically regulated by vesicle transport. A second example covers the target of rapamycin (TOR) kinase signaling pathway. Although Drosophila TOR itself was not represented in our library, our RNAi screen identified many pathway members, including PI3K, SNF1A (AMPK), S6K, TSC1, Gigas (TSC2), InaC (PKC), and MTS (PP2A) (see Figure 7). This pathway includes two TOR complexes, one responding to growth factors, nutrients, and energy (ATP) and leading to changes in cell size and number and the other regulating actin organization by an unknown mechanism [36]. Our results suggest that phagocytosis is closely connected to the TOR pathway and could impinge upon this unknown mechanism. Many additional insights and predictions regarding the role of individual gene products in phagocytosis can be gleaned from this type of analysis. We view the list of 184 genes implicated in phagocytosis as a resource that will stimulate future approaches and discoveries.
Figure 7 Genes Identified in the Screen as Being Required for Phagocytosis of C. albicans Were Superimposed onto a Drosophila Genomic Yeast Two-Hybrid Interaction Map
Interactions are displayed, and several pathways are outlined. Dark-blue circles indicate genes identified as reduced phagocytosis of C. albicans, with light-blue circles indicating genes present in the two-hybrid map but not identified in the phagocytosis screen. Several functional groups are circled as indicated. This diagram represents only a portion of the complete two-hybrid map [35], indicating that the genes identified in the phagocytosis screen affect a limited number of cellular processes.
In evaluating the results of any RNAi screen, it is of interest to compare them with analogous screens. Two 2005 analyses examined the ability of an S2 cell line to support intracellular infection by M. fortuitum and L. monocytogenes [7,9]. Of the 184 genes identified in our screen, 21 genes were shared with the M. fortuitum screen and 33 with the L. monocytogenes screen. The overlapping genes encoded key components of actin dynamics and vesicle transport; as described above, both processes are well known to have critical roles in the uptake of pathogens and inert particles, and perhaps these two groups of genes define the core processes of phagocytosis. The lack of a more extensive overlap might be due to certain technical differences in the screens (different RNAi libraries, different screening protocols, or different significance thresholds, etc.). However, we believe that much of the difference is due to the nature of the pathogen investigated. M. fortuitum and L. monocytogenes are intracellular bacterial pathogens, whereas C. albicans is a fungal pathogen that is not believed to proliferate intracellularly. Indeed, genes whose role we have characterized in the most detail (Mcr and Tep genes) were not uncovered in these other screens, and we failed to detect the CR gene that was the focus of one of these reports [9]. These results suggest that important features of the recognition or phagocytosis process may be relevant to some pathogens, but not others. Reciprocally, it should be emphasized that not all of the genes scored are likely to be directly involved in phagocytosis. For example, as noted by Ramet [10], the Serpent transcription factor, which was detected as required for phagocytosis in the screen for E. coli phagocytosis and our screen, is required for differentiation of hematopoietic cells and appears to be defective in phagocytosis because of a general change in cellular phenotype. By directing our attention to genes that are pathogen specific, we hope to favor the identification of genes directly involved in the recognition process. We are particularly interested in the eight gene products identified in our screen and annotated in Flybase as being involved in the defense response, as they represent a variety of different types of proteins, none of which had previously been shown to be required for phagocytosis, and none of which were identified in the M. fortuitum and L. monocytogenes screens [7,9]. Several of these genes have established roles in immunity, including an IκB homolog, Cactus, two peroxidases involved in reactive oxygen metabolism, and Cyp33, which is expressed in T cells and regulates gene expression. In other cases, the annotation of these genes was based simply upon the expression pattern of homologous genes in other organisms. For example, the mammalian homolog of CG4615 is expressed in macrophages but not in monocytes. Two of these eight genes, Mcr and Cyp33, appear to be specifically required for the phagocytosis of C. albicans when compared with E. coli.
Mcr and Related Proteins
An original aim of our work was to understand how a particular pathogen, C. albicans, is efficiently phagocytosed by fly S2 cells. As described, we tested which of the 184 genes identified in this study were specific for C. albicans and which were required more generally for phagocytosis. To this end, we rescreened the 184 dsRNAs for effects on the phagocytosis of E. coli and latex beads. Although C. albicans and E. coli are taken up much more rapidly than beads, phagocytosis of beads could be reproducibly measured over longer incubation times. Most of the 184 dsRNAs inhibited phagocytosis generally, although several had a much greater effect on C. albicans (see Figure 2). We chose one of these C. albicans–specific genes, Mcr, for further analysis.
Mcr is a member of the α2 macroglobulin/complement family (see Figure 3A), which includes at least 11 family members in humans. Members include secreted protease inhibitors (α2M, PZP) and components of complement (C3, C4A, C4B, and C5). Several other family members (CD109, CPAMD8, Ovostatin 1, Ovostatin 2) have fewer well-characterized functions. The prototypical member, α2 macroglobulin (α2M), binds to secreted proteases, including those of pathogens, leading to their uptake and inactivation by host cells. Human α2M also interacts with cytokines to regulate their distribution and activity [37]. The complement cascade is an ancient response to pathogens that triggers opsonization of a pathogen, formation of a membrane-attack complex, and in vertebrates the activation of the adaptive immune system [38]. Historically, the complement cascade was thought to reside only in vertebrates; however, studies suggest its presence in lower eukaryotes, including ascidians and sea urchins [39], and studies in mosquito documented immune functions of the complement-related Teps [8,24,25].
Drosophila encodes five proteins that are closely related to the human α2M/complement family of proteins. These are Mcr, the component identified in our screen, and also TepI, TepII, TepIII, and TepIV. The human protein most closely related to the Mcr-Tep family is CD109, a GPI-anchored protein whose function has not been fully explored [40,41]. Teps are expressed in Drosophila larvae and adult flies upon infection by E. coli [29,42]. The function of the Teps is probably best understood in the mosquito A. gambiae. A. gambiae Tep1 (aTep1) is required for phagocytosis of E. coli by a mosquito cell line [25]. aTep1 is secreted, proteolytically processed, and a fragment adheres to E. coli through formation of a thioester bond. During infections of adult mosquitoes by Plasmodium berghei parasites, aTep1 binds the surface of the parasite. RNAi knockdown of aTep1 allows more parasite oocysts to survive in the midgut of the mosquito, suggesting that aTep1 plays a critical role in killing malarial parasites [24]. This aTep1-dependent killing of malarial parasites occurs in a compartment devoid of hemocytes and does not involve phagocytosis. The findings suggest that aTep1 may target a microbe to multiple immune defense pathways [24]. The Tep1 thioester bond is presumed to form at a conserved thioester domain (GCGEQN) that is found in all four Drosophila Teps, aTep1, human CD109, and other members of the human α2M/complement class of proteins. Mcr and a close relative in A. gambiae lack the critical cysteine through which these covalent bonds are formed.
In this study, we show that Mcr is secreted by S2 cells and binds tightly to C. albicans in the absence of S2 cells. The defect in phagocytosis caused by Mcr RNAi can be reversed through the addition of conditioned media from normal S2 cells, suggesting that secreted Mcr may be the active Mcr required for efficient phagocytosis of C. albicans. Unlike aTep1 or complement in mammals [25,38,39], we did not detect any evidence of proteolytic processing of Mcr, suggesting that the full-length protein is the active form. This may be related to the fact that Mcr lacks the critical cysteine residue present in the Teps and presumably does not form thioester linkages. Mcr appears specific for the phagocytosis of C. albicans, as its reduction by RNAi had little or no effect on phagocytosis of E. coli or S. aureus. Moreover, Mcr binding exhibits specific recognition for C. albicans compared to even closely related fungi. Thus, Mcr binding shows a marked preference for C. albicans over the common S. cerevisiae lab strain S288c. In addition, Mcr binds significantly more to wild-type C. albicans than to a C. albicans mutant (Δefg1/Δefg1) that has altered cell-wall properties. Both S. cerevisiae and the C. albicans Δefg1/Δefg1 mutant are poorly phagocytosed by S2 cells, further supporting the idea that Mcr plays a critical role in recognizing wild-type C. albicans and promoting its subsequent phagocytosis.
Given that Mcr appears specific for C. albicans phagocytosis, we also investigated the possible roles of the four closely related Drosophila Teps. RNAi directed against TepII specifically reduced phagocytosis of E. coli, a gram-negative bacterium, and RNAi directed against TepIII specifically reduced phagocytosis of S. aureus, a gram-positive bacterium. None of the Tep reductions had any effect on phagocytosis of C. albicans. Thus this family of five closely related proteins collectively functions to promote the phagocytosis of a diverse set of pathogens, with individual family members showing specificity for certain classes of pathogens.
Unbiased screens for the genes required for specific pathogen recognition should give a broad view of the mechanisms targeting innate immune responses. Our studies implicate Mcr as one important contributor to the recognition of C. albicans. This result is supported by other findings suggesting that the related Teps have a similar role. The Teps have been shown to be related to complement, and in the mosquito A. gambiae, Tep1 is required for the phagocytosis of E. coli by mosquito 5.1* cells and killing of the malaria parasite in vivo [24,25]. Our analysis of the different Drosophila Teps show that they too are specialized in the recognition of different pathogens. But what of the other genes showing C. albicans–specific effects? Much remains to be done to determine whether recognition is combinatorial with different genes contributing to different branches of the recognition, or whether, like the complement system of mammals, recognition by Mcr is complex, involving multiple components in the recognition and the signaling.
Materials and Methods
Strains and plasmids
The C. albicans CAF2–1 strain (URA3/ura3::λimm434) was used for most experiments [43]. GFP–C. albicans expresses GFP under the control of the ADH1 promoter [44]. The deletion strain Δefg1 (Δura3::λimm434/Δura3::λimm434 Δefg1::hisG-URA3-hisG/Δefg1::hisG) has been described previously [45]. The S. cerevisiae wild-type strain used was MATα S288c [46]. GFP–E. coli (DH5α) expresses GFP under the control of the bacterial ribosomal promoter [47,48]. Shirley Lowe (University of California, San Francisco, United States) kindly provided S. aureus.
Cell culture
Drosophila S2 cells were cultured in Schneider's medium (Invitrogen, Carlsbad, California, United States) supplemented with 10% fetal bovine serum (FBS), penicillin, and streptomycin (pen/strep).
RNAi
The dsRNA library used in this screen has been described previously [48]. S2 cells were plated into 96-well plates at a density of 50,000 cells per well in a culture volume of 150 μl per well. dsRNA was added to a final concentration of 10 μg/ml, and the cells were incubated for four days at 25 °C to allow depletion of the corresponding gene product.
Phagocytosis screen
Primary screen: 1 × 105 dsRNA-treated S2 cells were plated in 96-well plastic tissue culture plates in 150 μl of Schneider's medium with 10% FBS and pen/strep. FITC-labeled (VWR, West Chester, Pennsylvania, United States, 5mg/ml), GFP-expressing C. albicans was added to each well containing S2 cells at a density of 2 × 105
C. albicans per well and incubated for 2 h at 25 °C. S2 and C. albicans mixtures were transferred to glass-bottom, Concanavalin A–coated 96-well microplates (Greiner Bio-One, Longwood, Florida, United States) and incubated for 1 h. Cells were fixed and processed as described below. Wells were visually screened for an apparent decrease in phagocytosed C. albicans (green-only) cells. Both SLS and EF independently screened all wells without knowledge of the identity of the dsRNAs. The few discrepancies between SLS and EF screens were resolved with further analysis. In wells with fewer phagocytosed C. albicans, further examination of the S2 cells was performed to rule out cytotoxic effects. dsRNAs were eliminated from further analysis if there were no remaining S2 cells. In wells with fewer S2 cells than normal, the well was examined further for phagocytosis by the remaining S2 cells. In many of these wells, the remaining S2 cells still phagocytosed C. albicans normally, and those were not scored as having a phagocytosis defect.
Secondary screens with C. albicans: the 280 positive dsRNAs were resynthesized and rescreened with more rigorous standards by visually quantifying the number of S2 cells phagocytosing. Briefly, 1 × 105 dsRNA-treated S2 cells were plated in 96-well plastic tissue culture plates in 150 μl of Schneider's medium with 10% FBS and pen/strep. FITC-labeled (VWR, 5mg/ml), GFP-expressing C. albicans was added to each well containing S2 cells at a density of 5 × 105 per well and incubated for 2 h at 25 °C. S2 and C. albicans mixtures were transferred to glass-bottom, Concanavalin A–coated 96-well microplates (Greiner) and incubated for 1 h. Phagocytosis assays were performed using similar conditions for DH5α E. coli expressing GFP (5μl of an overnight saturated culture per well, phagocytosis for 2 h at 25 °C), and yellow-green fluorescently labeled 2-μm latex beads (Sigma, St. Louis, Missouri, United States, 2 × 106 beads per well, phagocytosis for 20 h at 25 °C). Cells were fixed and processed as described below. While the overall efficiency of phagocytosis varied between experiments, phagocytosis by wild-type S2 cells was highly consistent within each experiment. For this reason, all secondary screen phagocytosis assays were completed in 1 d with the same batch of S2 cells. S2 cells were counted and scored for having phagocytosed one or more C. albicans, E. coli, or latex beads. As described in the Results, we used a significance threshold of 1.5 standard deviations below the mean for these secondary screens. A t-test was used to compare the percentage of phagocytosis of six wild-type wells to a value of 44% for the dsRNA-treated wells. The significance values (C. albicans p = 0.012, E. coli p = 0.001, latex beads p = 0.005) indicated that these criteria are reasonably stringent, and 44% represents a statistically significant threshold.
Phagocytosis assays
dsRNA-treated S2 cells (1 × 105) were plated in 96-well plastic tissue culture plates in 150 μl of Schneider's medium with 10% FBS and pen/strep. GFP-expressing C. albicans was added to each well containing S2 cells at a density of 5 × 105
C. albicans per well and incubated for various times at 25 °C. S2 and C. albicans mixtures were transferred to glass-bottom, Concanavalin A–coated, 96-well microplates (Greiner) and incubated for 1 h. Phagocytosis assays were performed using similar conditions for DH5α E. coli expressing GFP (5 μl of an overnight saturated culture per well), S. aureus (30 μl of a FITC-labeled overnight culture per well), or yellow-green fluorescently labeled 2-μm latex beads (Sigma), 1 × 106 beads per well. Cells were fixed and processed as described below. S2 cells were counted and scored for having phagocytosed one or more C. albicans or other pathogen. All graphs represent the mean ± the standard deviation of at least four counted samples. The difference between wild-type and RNAi treatments was statistically analyzed by t-test assuming unequal variances. A p-value < 0.01 was considered significant.
Immunofluorescence and microscopy
After co-incubation, the cell culture media was aspirated and allowed to dry for 2 min. Cells were fixed with 1% formaldehyde in PBS for 5 min, washed with 1× PBS, and blocked with 5% FBS in PBS for 2–4 h. The cell surfaces of C. albicans, E.coli, or S. aureus were detected with primary antibodies raised against whole C. albicans (Biodesign, Saco, Maine, United States, Cat# B65411R), whole E. coli (Biodesign, Cat# B47711G), or whole S. aureus (Biodesign, Cat# B65881R). Primary antibodies were visualized with Cy3-conjugated goat anti-rabbit or rabbit anti-goat antibodies (Jackson ImmunoResearch, West Grove, Pennsylvania, United States). In each case, pathogens were considered phagocytosed if they were not visualized with their corresponding antibody. DNA was visualized with Hoechst 33258, and filamentous actin was detected with rhodamine-coupled phalloidin (both from Molecular Probes, Eugene, Oregon, United States). Immunofluorescent images were taken with a Zeiss Axiovert 200M microscope using AxioVision software (Carl Zeiss, Oberkochen, Germany). Images were processed using Zeiss AxioVision 3D Deconvolution software, and figures were assembled with Adobe Photoshop and Illustrator (Adobe Systems, San Jose, California, United States).
Western blotting
S2 cells were harvested as described [48], and conditioned media were collected from cells grown for 48 h. Cell lysate and conditioned media were separated by SDS-PAGE and analyzed by Western blotting. Mcr was detected by using a polyclonal rabbit antibody generated against a peptide in the amino half of Mcr (CGQTNPSDRPPYRTDSGS) (Bethyl Laboratories, Montgomery, Texas, United States). To detect Mcr interactions with C. albicans, 5 × 107
C. albicans, mutant strain or S. cerevisiae, were incubated in 3 ml of new Schneider's medium with 2% FBS, or in conditioned media with 2% FBS, for 2 h. C. albicans cells were washed one time with media and one time with wash buffer (0.05% NP-40, 120 mM NaCl, 50 mM Hepes [pH 7.5], and 5 mM EDTA) and loaded on SDS-PAGE gel with SDS-sample buffer and analyzed by Western blotting with the Mcr antibody.
Supporting Information
Table S1 Genes Not Further Followed Up with Secondary Assays
dsRNAs that decrease the phagocytosis of C. albicans but were not followed up in further studies, including genes involved in general transcription, translation, RNA processing, and the proteasome.
(207 KB DOC).
Click here for additional data file.
Table S2 Measurement of the Percentage of S2 Cells Phagocytosing after Treatment with dsRNA against Individual Genes
dsRNAs that decrease the phagocytosis of C. albicans. Column 3, closest human homolog; column 4, function of Drosophila protein if known, function is based upon mammalian or other homologs if Drosophila protein function unknown; column 5, genes that have previously been implicated in phagocytosis; columns 6–8, the percentage of S2 cells that have phagocytosed one or more C. albicans (C.a.), E. coli (E.c.), or latex beads (beads). 50–100 S2 cells were counted for each condition.
(679 KB DOC).
Click here for additional data file.
Accession Numbers
The FlyBase (http://flybase.bio.indiana.edu/search/) accession numbers for the genes and gene products discussed in this paper are Mcr (FBgn0020240), TepI (FBgn0041183), TepII (FBgn0041182), TepIII (FBgn0041181), and TepIV (FBgn0041180). The NCBI Entrez (http://www.ncbi.nlm.nih.gov/gquery/gquery.fcgi) accession number for A. gambiae Ag Mcr (Tep13) is EAA12257.2.
The dsRNA library used in this screen was produced by Ben Eaton, EF, Nico Stuurman, Graeme Davis, PHO, and Ron Vale at the University of California, San Francisco (UCSF). We thank Shirley Lowe of the UCSF microbiology teaching labs for supplying the S. aureus strain. We are grateful to Roland Bainton, Anthony DeFranco, Yuh Nung Jan, and Lewis Lanier for comments on the manuscript. This work was supported in part by grants from the National Institutes of Health to ADJ (RO1 AI49187) and to PHO (RO1 AI60102), a Jane Coffin Childs postdoctoral research grant to SLS, and a Damon Runyon Cancer Research Foundation postdoctoral grant to EF. This paper represents a collaboration between the O'Farrell and Johnson laboratories, to which each lab contributed equally.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. SLS and EF conceived, designed, and performed the experiments. SLS, EF, PHO, and ADJ analyzed the data and wrote the paper.
Citation: Stroschein-Stevenson SL, Foley E, O'Farrell PH, Johnson AD (2006) Identification of Drosophila gene products required for phagocytosis of Candida albicans. PLoS Biol 4(1): e4.
Abbreviations
Copcoat protein
dsRNAdouble-stranded RNA
GFPgreen fluorescent protein
McrMacroglobulin complement related
RNA interferenceRNAi
Tepthioester protein
TORtarget of rapamycin
UTRuntranslated region
==== Refs
References
Elrod-Erickson M Mishra S Schneider D Interactions between the cellular and humoral immune responses in Drosophila
Curr Biol 2000 10 781 784 10898983
Hoffmann JA Kafatos FC Janeway CA Ezekowitz RA Phylogenetic perspectives in innate immunity Science 1999 284 1313 1318 10334979
Romani L Immunity to fungal infections Nat Rev Immunol 2004 4 1 23 14661066
Chimini G Chavrier P Function of Rho family proteins in actin dynamics during phagocytosis and engulfment Nat Cell Biol 2000 2 E191 E196 11025683
Bajno L Peng XR Schreiber AD Moore HP Trimble WS Focal exocytosis of VAMP3-containing vesicles at sites of phagosome formation J Cell Biol 2000 149 697 706 10791982
Hackam DJ Botelho RJ Sjolin C Rotstein OD Robinson JM Indirect role for COPI in the completion of FCgamma receptor-mediated phagocytosis J Biol Chem 2001 276 18200 18208 11279223
Agaisse H Burrack LS Philips J Rubin EJ Perrimon N Genome-wide RNAi screen for host factors required for intracellular bacterial infection Science 2005 309 1248 1251 16020693
Moita LF Wang-Sattler R Michel K Zimmermann T Blandin S In vivo identification of novel regulators and conserved pathways of phagocytosis in A. gambiae
Immunity 2005 23 65 73 16039580
Philips JA Rubin EJ Perrimon N
Drosophila RNAi screen reveals CD36 family member required for mycobacterial infection Science 2005 309 1251 1253 16020694
Ramet M Manfruelli P Pearson A Mathey-Prevot B Ezekowitz RA Functional genomic analysis of phagocytosis and identification of a Drosophila receptor for E. coli
Nature 2002 416 644 648 11912489
Bernhardt H Knoke M Mycological aspects of gastrointestinal microflora Scand J Gastroenterol Suppl 1997 222 102 106 9145460
Edmond MB Wallace SE McClish DK Pfaller MA Jones RN Nosocomial bloodstream infections in United States hospitals: A three-year analysis Clin Infect Dis 1999 29 239 244 10476719
Calderone RA Gow NAR Calderone RA Host recognition by Candida species Candida and candidiasis 2002 Washington (DC) ASM Press 67 86
Romani L Immunity to Candida albicans : Th1, Th2 cells and beyond Curr Opin Microbiol 1999 2 363 367 10458979
Hoffmann JA The immune response of Drosophila
Nature 2003 426 33 38 14603309
Ip YT Reach M Engstrom Y Kadalayil L Cai H Dif, a dorsal-related gene that mediates an immune response in Drosophila
Cell 1993 75 753 763 8242747
Lemaitre B Nicolas E Michaut L Reichhart JM Hoffmann JA The dorsoventral regulatory gene cassette spatzle/Toll/cactus controls the potent antifungal response in Drosophila adults Cell 1996 86 973 983 8808632
Alarco AM Marcil A Chen J Suter B Thomas D Immune-deficient Drosophila melanogaster : A model for the innate immune response to human fungal pathogens J Immunol 2004 172 5622 5628 15100306
Cheng LW Portnoy DA
Drosophila S2 cells: An alternative infection model for Listeria monocytogenes
Cell Microbiol 2003 5 875 885 14641173
Dionne MS Ghori N Schneider DS
Drosophila melanogaster is a genetically tractable model host for Mycobacterium marinum
Infect Immun 2003 71 3540 3550 12761139
Mansfield BE Dionne MS Schneider DS Freitag NE Exploration of host-pathogen interactions using Listeria monocytogenes and Drosophila melanogaster
Cell Microbiol 2003 5 901 911 14641175
Schneider D Shahabuddin M Malaria parasite development in a Drosophila model Science 2000 288 2376 2379 10875925
Lavine MD Strand MR Insect hemocytes and their role in immunity Insect Biochem Mol Biol 2002 32 1295 1309 12225920
Blandin S Shiao SH Moita LF Janse CJ Waters AP Complement-like protein TEP1 is a determinant of vectorial capacity in the malaria vector Anopheles gambiae
Cell 2004 116 661 670 15006349
Levashina EA Moita LF Blandin S Vriend G Lagueux M Conserved role of a complement-like protein in phagocytosis revealed by dsRNA knockout in cultured cells of the mosquito, Anopheles gambiae
Cell 2001 104 709 718 11257225
Botelho RJ Scott CC Grinstein S Phosphoinositide involvement in phagocytosis and phagosome maturation Curr Top Microbiol Immunol 2004 282 1 30 14594212
Cox D Tseng CC Bjekic G Greenberg S A requirement for phosphatidylinositol 3-kinase in pseudopod extension J Biol Chem 1999 274 1240 1247 9880492
Cox D Greenberg S Phagocytic signaling strategies: Fc(gamma)receptor-mediated phagocytosis as a model system Semin Immunol 2001 13 339 345 11708889
Lagueux M Perrodou E Levashina EA Capovilla M Hoffmann JA Constitutive expression of a complement-like protein in toll and JAK gain-of-function mutants of Drosophila
Proc Natl Acad Sci U S A 2000 97 11427 11432 11027343
Blandin S Levashina EA Thioester-containing proteins and insect immunity Mol Immunol 2004 40 903 908 14698229
Lane S Birse C Zhou S Matson R Liu H DNA array studies demonstrate convergent regulation of virulence factors by Cph1, Cph2, and Efg1 in Candida albicans
J Biol Chem 2001 276 48988 48996 11595734
Sohn K Urban C Brunner H Rupp S EFG1 is a major regulator of cell wall dynamics in Candida albicans as revealed by DNA microarrays Mol Microbiol 2003 47 89 102 12492856
Pearson AM Baksa K Ramet M Protas M McKee M Identification of cytoskeletal regulatory proteins required for efficient phagocytosis in Drosophila
Microbes Infect 2003 5 815 824 12919849
Echard A Hickson GR Foley E O'Farrell PH Terminal cytokinesis events uncovered after an RNAi screen Curr Biol 2004 14 1685 1693 15380073
Giot L Bader JS Brouwer C Chaudhuri A Kuang B A protein interaction map of Drosophila melanogaster
Science 2003 302 1727 1736 14605208
Inoki K Ouyang H Li Y Guan KL Signaling by target of rapamycin proteins in cell growth control Microbiol Mol Biol Rev 2005 69 79 100 15755954
Borth W Alpha 2-macroglobulin, a multifunctional binding protein with targeting characteristics FASEB J 1992 6 3345 3353 1281457
Gasque P Complement: A unique innate immune sensor for danger signals Mol Immunol 2004 41 1089 1098 15476920
Fujita T Matsushita M Endo Y The lectin-complement pathway—Its role in innate immunity and evolution Immunol Rev 2004 198 185 202 15199963
Lin M Sutherland DR Horsfall W Totty N Yeo E Cell surface antigen CD109 is a novel member of the alpha(2) macroglobulin/C3, C4, C5 family of thioester-containing proteins Blood 2002 99 1683 1691 11861284
Solomon KR Sharma P Chan M Morrison PT Finberg RW CD109 represents a novel branch of the alpha2-macroglobulin/complement gene family Gene 2004 327 171 183 14980714
De Gregorio E Spellman PT Rubin GM Lemaitre B Genome-wide analysis of the Drosophila immune response by using oligonucleotide microarrays Proc Natl Acad Sci U S A 2001 98 12590 12595 11606746
Fonzi WA Irwin MY Isogenic strain construction and gene mapping in Candida albicans
Genetics 1993 134 717 728 8349105
Hull CM Johnson AD Identification of a mating type-like locus in the asexual pathogenic yeast Candida albicans
Science 1999 285 1271 1275 10455055
Braun BR Johnson AD TUP1, CPH1 and EFG1 make independent contributions to filamentation in Candida albicans
Genetics 2000 155 57 67 10790384
Green SR Johnson AD Promoter-dependent roles for the Srb10 cyclin-dependent kinase and the Hda1 deacetylase in Tup1-mediated repression in Saccharomyces cerevisiae
Mol Biol Cell 2004 15 4191 4202 15240822
Basset A Khush RS Braun A Gardan L Boccard F The phytopathogenic bacteria Erwinia carotovora infects Drosophila and activates an immune response Proc Natl Acad Sci USA 2000 97 3376 3381 10725405
Foley E O'Farrell PH Functional dissection of an innate immune response by a genome-wide RNAi screen PLoS Biol 2004 2 e203 10.1371/journal.pbio.0020203 15221030
|
16336044
|
PMC1310651
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e4
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040004
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633604510.1371/journal.pbio.0040005Research ArticleBioinformatics/Computational BiologyEvolutionGenetics/Genomics/Gene TherapyMolecular Biology/Structural BiologyPlant ScienceHorizontal Transfer of a Plant Transposon Horizontal Transfer of a Plant TransposonDiao Xianmin
1
¤Freeling Michael
1
Lisch Damon [email protected]
1
1Department of Plant and Microbial Biology, University of California at Berkeley, Berkeley, California, United States of AmericaMartienssen Robert Academic EditorCold Spring Harbor Laboratory, United States of America1 2006 20 12 2005 20 12 2005 4 1 e519 4 2005 31 10 2005 Copyright: © 2006 Diao et al.2006This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Jumping Genes Cross Plant Species Boundaries
The majority of well-documented cases of horizontal transfer between higher eukaryotes involve the movement of transposable elements between animals. Surprisingly, although plant genomes often contain vast numbers of these mobile genetic elements, no evidence of horizontal transfer of a nuclear-encoded transposon between plant species has been detected to date. The most mutagenic known plant transposable element system is the Mutator system in maize. Mu-like elements (MULEs) are widespread among plants, and previous analysis has suggested that the distribution of various subgroups of MULEs is patchy, consistent with horizontal transfer. We have sequenced portions of MULE transposons from a number of species of the genus Setaria and compared them to each other and to publicly available databases. A subset of these elements is remarkably similar to a small family of MULEs in rice. A comparison of noncoding and synonymous sequences revealed that the observed similarity is not due to selection at the amino acid level. Given the amount of time separating Setaria and rice, the degree of similarity between these elements excludes the possibility of simple vertical transmission of this class of MULEs. This is the first well-documented example of horizontal transfer of any nuclear-encoded genes between higher plants.
Sequencing and analysis of MULE transposons and their surrounding genomic regions from closely related grass species and rice provides evidence of horizontal transfer in plants.
==== Body
Introduction
Horizontal transfer can be defined as the process by which genes can move between reproductively isolated species. It is far more frequent than was once thought, particularly among prokaryotes [1], to the extent that it has complicated analysis of bacterial evolution [2]. Like many transfers between bacteria, most well-documented transfers detected to date between higher eukaryotes involve movement of transposable elements [3]. The best-documented examples involve the transfer of P [4] and mariner elements [5] between animal species. In contrast, no case of horizontal transfer of a nuclear-encoded plant gene has been reported. Recently, systematic analysis of mitochondrial genes and group I mitochondrial introns has revealed that they have been subject to frequent horizontal transfer between plant species [6–8], suggesting that there is no fundamental barrier to gene flow between reproductively isolated plant species. Interestingly, despite the vast numbers of transposable elements found in most plant nuclear genomes and the demonstrated propensity of these elements to move between animal species, no evidence for horizontal transfer of nuclear-encoded plant transposons has been detected. This is in part due to the lack of comprehensive surveys of these elements in a large number of plants. However, extensive phylogenetic analysis of haT [9,10] and mariner [11] transposons in a variety of plant species has been consistent with vertical, rather than horizontal, transmission of these elements in plants. The same appears to be the case for
CACTA elements [12] and the Ty1-copia group of retrotransposons [13] as well. The case of mariner elements is particularly surprising given that mariner elements in metazoans are particularly prone to horizontal transfer, to the extent that such transfer has been hypothesized to be an integral part of the success of this class of elements [14].
MuDR is an unusually active and mutagenic maize class II (DNA intermediate) transposon. It is the autonomous member of a family of transposable elements in maize, all of which share the same 200–base pair (bp) terminal inverted repeats, but each of which contains a unique internal sequence. MuDR encodes two genes: mudrA, the putative transposase, and mudrB, a helper gene; these are translated into two proteins: MURA and MURB, respectively [15]. Mu-like elements (MULEs) containing mudrA homologs and long terminal inverted repeats are widespread among both monocot and dicot angiosperms [16–20]. Any given species can have multiple, distinct subfamilies of MULEs [17].
Previous analysis of DNA gel blots and the sequences of a portion of MULEs from a number of grass species revealed a distinct lack of congruence between species phylogenies and that of their MULEs [17], consistent with horizontal transfer. However, given the presence of multiple MULE subfamilies in most grass genomes, a lack of congruence between host and MULE phylogenies would be expected if particular MULE subfamilies were lost from some plant lineages but not from others. Thus, although suggestive, incongruence between transposon and host species phylogenies cannot be expected to conclusively demonstrate horizontal transfer. Our criteria for unequivocal evidence of horizontal transfer of MULEs require that sequences from two species be too similar in noncoding regions to be reasonably expected to have diverged at the same time as their respective host species. Based on such evidence, we report horizontal transfer of a nuclear-encoded transposable MULE between two Gramineae (grass) lineages: the Panicoid and the Bambosoid subfamilies. Extensive phylogenetic and paleontologic evidence suggests that these subfamilies diverged between 30 and 60 million years ago [21–23]. The degree of similarity in noncoding and synonymous sequence between MULEs in these groups is inconsistent with this divergence time.
Results
We have isolated portions of MULEs from several species of Setaria (millets of the tribe Panicoideae) and compared them with those in publicly available databases and with sequences we have previously obtained [17]. Some of these sequences are unexpectedly similar to a small group of rice MULEs. The observed degree of similarity in noncoding and synonymous sites is much higher than one would predict given the time separating Setaria and rice.
A total of 27 sequences from eight species of the genus Setaria were obtained. All 27 sequences encoded potential products with similarity to the MURA protein from maize (data not shown). Nineteen nonredundant sequences were subjected to more detailed analysis. Phylogenetic analysis revealed that all of the sequences fell into previously defined class II or class III groups of the MuDR family of MULEs (Figure 1). These classes of elements are phylogenetically distinct from the other known functional MULEs such as Jittery in maize [17] and AtMu1 in Arabidopsis [24]. As has been previously observed with other species, a single species of Setaria may have elements from more than one class of MULEs, consistent with previous observations that multiple, paralogous families of MULEs can exist within a single genome [15].
Figure 1 Phylogenetic Analysis of Various mudrA-Homologous Sequences from Grasses Representative of Several Major Subfamilies of Grasses
Species designations are as follows: Zm, Zea mays (MuDR is mudrA); Zl, Zea luxurians; Sv, Setaria viridis; Sa, S. anceps; Sf, S. faberi; Ss, S. sphacelata; Sg, S. glauca; Sp, Setaria palmifolia; Sb, Sorghum bicolor; Cl, Coix lacryma; Os, O. sativa; Pv, P. virigatum; Vz, Vetevaria zizanoides; Mr, Muhlenberia rigens; Mm, Muhlenbergia macroura; So, Saccharum officinarum; Sk, Shibataea kumasaca; Sn, Sinarundinaria nitida; As, Avena sativa; Ca, Calamagrostis acutifolia; Ta, Triticum aestivum, Hv, Hordeum vulgare; Am, Ammophila arenaria; Fr, Festuca rubra; Bm, Briza maxima. Numbers represent individual clones from each species, or the last three digits of the accession number if the sequence was obtained from NCBI. Colored blocks indicate major subfamilies: green, Panicoideae; blue, Chloridoideae; yellow, Pooideae; red, Bambusoideae. Branch lengths are proportional to distance, as indicated by the scale bar. Bootstrap support is as indicated for each branch.
MULE sequences from Setaria italica (foxtail millet, not shown) and the closely related Setaria faberi (giant foxtail, “sf” in Figure 1) are strikingly similar (approximately 90% identical over 585 bp) to an element present on Chromosome 5 of the nuclear genome of Oryza sativa ssp. japonica. This element is designated Os493. A nearly identical element (98.9% nucleotide identity) is also present in O. sativa ssp. indica. Based on the observation that the sequences flanking the transposon insertion in both of these O. sativa ssp. are the same, we conclude that this element is present at the same position in each of these subspecies. Os493 has similarity to MURA from maize from amino acid 37 to amino acid 730 of the 823–amino acid maize protein. Os493 is flanked by the 9-bp direct repeats typical of Mu transposon insertions. A second, truncated element, Os072, is located on Chromosome 1 in japonica and is also present at the same position in the indica genome. Counting indels as single events, this element is 96% identical to Os493. Like the full-length elements, this element is flanked by the same 9-bp flanking duplications in both subspecies. A final truncated element on Chromosome 5, Os408, is present in the japonica genome but is absent in sequences available from the indica genome. This element is 94% identical to Os493. These are the only rice elements with significant (>80% nucleotide identity over 585 bp) similarity to the Setaria elements. blastn (Nucleotide-Nucleotide Basic Local Alignment Search Tool [BLAST]; National Center for Biotechnology Information [NCBI], Bethesda, Maryland, United States) searches using the full-length element from rice reveal that all other significant blast hits (e value <10−4) to Os493 that are not members of the genus Oryza are to sequences from maize, sorghum, or sugarcane, all of which, like Setaria, are Panicoid grasses of the tribe Andropogoneae. To confirm the presence of this class of elements in our rice sample, we used primers specific to Os493 to amplify several of them. One product was identical over 794 bp to Os072, a close relative of Os493. A second product was 98% identical to Os493 (data not shown).
To understand more fully the overall distribution of this class of MULEs within the grasses, Southern blot hybridization was employed using a portion of the mudrA homolog from one S. faberi element (designated Sf3) as a probe to a wide variety of species representing the major subfamilies of the grasses (Figure 2). Hybridization and washing of the probe were performed at high stringency (see Materials and Methods); we have empirically determined that this degree of stringency will not detect sequences less than 87% identical to the probe (see below). The results revealed a strikingly discontinuous distribution of this class of elements. As expected based on our sequencing and database searches, it is present in a subset of the Panicoideae, with the strongest hybridization occurring with DNA from Setaria sp., although some species, such as Setaria sphacelata, hybridized poorly to the probe. Note that the probe did not detect any MULEs in Setaria anceps, which has an element that is 87% identical to the probe, thus defining the limit of detection for this probe. At this level of stringency, we did not detect signal from any species of the subfamilies Chloridoideae or Pooideae and only weak signal from Panicoid grasses such as sorghum and maize. As expected, DNA from rice hybridized well to the probe, and the size of the BamHI fragment is that expected for Os072 (the predicted size of the Os493 fragment would be too large to transfer onto this blot). Two other Oryzoids, Ehrharta erecta and Zizania latifolia, did not hybridize. Interestingly, three species of bamboo indigenous to East Asia (Shibateae kumasaca, Fargesia nitida, and Phyllostachys vivax) hybridized well to the probe, but three species of bamboo indigenous to Central America (Chusquea montana, Chusquea quila, and Otatea acuminata) did not. Primers designed from the Setaria element were used to amplify sequences from Shibateae kumasaca and Fargesia nitida; sequencing of these products confirmed that they are highly similar to both the rice and the Setaria sequences. Indeed, each was 93% identical over 641 bp to the Setaria Sf3 element and 95% identical to rice Os493. Despite the higher similarity of these sequences to those in Setaria, we focused our analysis on a comparison between rice and Setaria because of the availability of the fully sequenced rice genome.
Figure 2 DNA Gel Blots of Representative Species from All Major Subfamilies of the Grasses
All blots were probed with a fragment of MULE sf3 from S. faberi. Subfamilies (i.e., Pooideae) are as indicated.
(A) 1, Shibataea kumasaca; 2, Ehrharta erecta; 3, Oryza sativa indica; 4, Oryza sativa japonica; 5, Zizania latifolia; 6, Nardus stricta; 7, Diarrhena japonica; 8, Brachypodium sylvaticum; 9, Phalaris aquatica; 10, Hordeum vulgare; 11, Triticum aestivum; 12, Chasmanthium latifolium; 13, Andropogon gerardii; 14, Sorghum bicolor; 15, Tripsacum dactyloides; 16, S. faberi; 17, S. italica; 18, Zea mays; 19, Muhlenbergia rigens; 20, Eleusine indica; 21, Boutelova curtipendula; 22, Cortaderia jubata.
(B) 1. Zizania latifolia; 2, Shibataea kumasaca; 3, Chusquea montana; 4, Chusquea quila; 5, Sinarundinaria nitida; 6, Phyllostachys vivax; 7, Otatea acuminata; 8, Zeigotes sp.; 9, Brachypodium sylvaticum; 10, Zizania latifolium; 11, O. sativa (japonica); 12, S. faberi; 13, S. italica.
(C) 1, S. italica; 2, S. italica; 3, S. viridis; 4, S. faberi; 5, S. glauca; 6, S. anceps; 7, S. sphacelata; 8, O. sativa; 9, Zea mays.
To extend our analysis to include both coding and noncoding sequences, we obtained most of the complete sequence of a single MULE from S. faberi. Using PCR primers designed from the rice MULE terminal inverted repeats, it was possible to amplify and sequence a complete copy of a Setaria element with the exception of the 20 bp on both ends that were used as primers. This nearly complete element is designated Sf4. Sf4 differs from the Setaria elements obtained in the first round of amplifications, but it retains significant similarity to Os493 (see Figure 1). The presence of frameshift and/or stop codons in both the Setaria and rice MULEs (Figure 3) suggests that neither element is currently functional. However, both sequences contain long terminal inverted repeats (approximately 176 bp) and clear nucleotide similarity to the mudrA gene of MuDR. A region near the 5′ terminal inverted repeats (nucleotides 222–560 in Os493) has a number of deletions or insertions in one of the two elements, including a 41-bp deletion in the Os493 element. A second region near the 3′ terminal inverted repeats (nucleotides 3350–3568 in Os493) has reduced homology (only 58% identity). Not counting these two regions and counting indels as single events, overall nucleotide identity of the two MULEs is 88% over 3.8 kb. This is a degree of similarity comparable to that of a well-conserved host gene. For example the waxy gene from S. italica shares 88% identity over 1.6 kb of open reading frame with waxy from rice (Table 1).
Figure 3 Schematic of the MULE Elements from Rice (Os493) and Setaria faberi (Sf4)
Black blocks represent regions deleted in one or the other sequence. Grey blocks represent putative introns, which are numbered in the Setaria and rice elements with Roman numerals I–III. The second intron is collinear with the third intron in mudrA from maize. The mudrA introns are numbered 1–3. Shaded blocks on the ends of the elements indicate the terminal inverted repeats. For comparison, the 5' end of MuDR from maize (which includes the mudrA gene) is also included. Note that only the third intron of the mudrA gene and the third intron in the rice and Setaria is present in all three elements. The position of the RF2 and RR2 PCR primers are as indicated on the MuDR element. Stops and frameshifts in the Setaria and rice element (assuming introns are spliced) are at the positions indicated. Dotted lines connecting MuDR to Sf4 indicate regions of similarity.
Table 1 Comparisons of Coding and Noncoding Sequence Similarities
Due to the long period of time separating Setaria and rice (roughly 50 million years), high sequence similarity over more than 3 kb can only be due to selection or horizontal transfer; the species are distantly related such that long sequences not under selection are no longer recognizably homologous [25]. Further, conservation of noncoding sequences is far lower in plants than it is in animals, despite very similar neutral mutation rates [26].
In order to directly compare genes from Setaria and rice, we have examined several nuclear-encoded genes from these two species. Only a few such genes have been sequenced from the genus Setaria, mostly from S. italica. We have obtained the complete sequence of two nuclear-encoded cDNAs from S. italica that have clear homologs in rice: acetyl-coenzyme carboxylase (acc1) and oxidoreductase (o. reductase) (Table 1). The total cDNA coding sequence available was 7,765 bp (Table 1). The total untranslated upstream and downstream sequence (sequences immediately adjacent to the start and stop codons) from these genes was 1,203 bp (data not shown). When comparing rice with S. italica, these genes show an overall average identity of 76% in coding sequences (Table 1) and only 40% in the untranslated sequences after alignment (data not shown). The corrected frequency of nonsynonymous substitutions for these genes ranges from 0.41 to 1.17, and the corrected frequency of synonymous substitutions ranges from 0.05 to 0.16. Importantly, the ratio of synonymous to nonsynonymous substitutions per site (dS/dN) for these genes was 7.2 and 8.0, consistent with selection for function at the amino acid level (Table 1). The dissimilarity of the untranslated sequences immediately upstream and downstream of the coding sequences as well as the divergence of synonymous sites is consistent with the absence of selection on these sequences during the roughly 50 million years since the divergence of the Setaria and rice lineages. Very little genomic sequence is available for any Setaria. However, the complete genomic copy of S. italica waxy gene is available. The exons of this gene were 88% identical to the rice waxy gene over 1,584 bp. In contrast, the introns shared only 52% identity over 1,243 bp (Table 1), which is comparable to the frequency of substitutions in UTR and synonymous sites in this gene and the cDNA sequences (data not shown).
Unlike other genes in rice and Setaria, the high degree of homology of the MULEs is not the result of selection at the level of amino acid sequence. This can be inferred from analysis of both introns and synonymous sites.
Because the horizontal transfer hypothesis rests heavily on the analysis of the MULE introns, we will describe them in some detail. The MULEs in rice and Setaria contain three predicted short introns at the same locations in each element (Figure 4; see also Figure 3). The first predicted intron represents an insertion relative to mudrA in maize, which lacks this sequence. The intron is 82 bp long in the rice element and 78 bp long in the Setaria element. Successful splicing of this putative intron would maintain the reading frame of the rice element and precisely remove the insertion from both MULEs (Figure 4A). Left unspliced, the intron in both MULEs would result in the introduction of two stop codons and would alter the reading frame of the putative transcript of the rice element. Although no homologous cDNAs are available, an exon trap database, in which genomic sequences are expressed to examine splicing patterns, reveals that this intron can be precisely spliced from Os493.
Figure 4 A Comparison of Coding and Noncoding Sequences from Various MULEs
(A) An alignment of the nucleotide sequence of a portion of the first exon, the first intron, and a portion of the second exon from the rice, Setaria, and sorghum MULEs. Identical nucleotides are displayed as periods; differences are as indicated.
(B) An alignment of the 3' end of exon 2, intron 2, exon 3, intron 3, and the 5' end of exon 4 from mudrA homologs from rice (Os493), S. faberi (Sf4), and a related element from maize (Zm890). In both panels, amino acid translations, shaded for similarity to Os493, are portrayed below the nucleotide alignments. MURA sequence is provided for comparison.
The second intron is in the same position as is the third intron of mudrA (see Figure 2). This intron is 71 bp in both the Setaria and rice elements. Unspliced, this intron would introduce a frameshift and several stop codons to both the rice and the Setaria elements. Successful splicing of this intron in the Setaria and rice MULEs would result in expression of a short additional region of conservation between the proteins encoded by these elements and MURA (Figure 4B). This region includes a previously identified nuclear localization signal [27]. The blastn searches identify a maize EST (AW067488) with high similarity to Os493 and Sf4 (78% and 77% nucleotide identity, respectively, in this region). It is much more similar to Os493 and Sf4 than it is to MuDR and is presumably related by vertical transmission to the MULE in Setaria. A nearly identical (99% nucleotide identity) genomic version of this cDNA, designated Zm890, is also available. A comparison of the genomic to cDNA sequence reveals that this intron in this maize element is spliced precisely at sites predicted for the second intron of Os493. A comparison of the maize cDNA to the genomic version of the sequence also reveals the presence of a third intron in this region that is also predicted to be spliced from Os493 and Sf4. This intron is 74 bp in the rice (Os493), Setaria (Sf4), and maize (Zm890) MULEs. Unspliced, this intron would introduce a frameshift and at least one stop codon in all three elements. The mudrA gene from MuDR in maize lacks this intron.
When we compare the Setaria and rice MULEs, we find that the degree of conservation is as high in these three putative introns (89% identity over the 227-bp total in the three introns) as it is in the rest of the transposon (Figures 4 and 5). For comparison, we can use a portion of the similar element in maize (Zm890) for which we have both genomic and cDNA sequences. Although the exon sequences are quite similar to the rice and Setaria MULEs, the intron sequences retain only limited similarity (Figures 4 and 5). Overall, in the region depicted in Figure 4B that compares the maize and Setaria elements, which includes 348 bp of exon sequence and 145 bp of intron sequence, exon sequences are 92% identical (319 of 348) in comparing the rice and the Setaria elements. Together, the last two introns are 88% identical (128 of 145). In contrast, when the maize MULE Zm890 is compared to Os493, although the exon sequences are 78% identical (270 of 348), the intron sequences are only 58% identical (85 of 145), consistent with selection on exon but not intron sequences.
Figure 5 Bar Graph Displaying Percent Similarity of the Setaria (Sf4), Maize (Zm890), and Sorghum (Sb662) Elements with the Rice Element (Os493)
The region analyzed includes the last portion of exon 1, intron 1, portions of exon 2, intron 2, exon 3, intron 3, and the first portion of exon 4. Although percent similarity for exon 1 and intron 1 between Sf4 and Os493 is shown, the corresponding region in Zm890 is not available, nor are sequences for the last two introns of Sb662. The percent figure is the percent identity of each element in the specified region to Os493 in that region. Portions in which no sequence was available are given a 0% value. T Exon and T Intron refer to the sum of exons and introns, respectively, that were compared.
A similar situation can be observed when comparing the first intron to a portion of a sorghum element (Sb632) available in the database (CW076632) that also has an inserted sequence at this position (see Figure 4A). While flanking exon sequences from this element are 73% identical to Os493 (147 of 201 nucleotides), this intron is only 46% identical to that of Os493 (38 of 82 nucleotides). In contrast, this intron is 90% identical (74 of 82 nucleotides) when comparing Os493 with Sf4.
Thus, in each case where homologous sequence is available, selection appears to favor sequence similarity in exons but not in introns, except when Os493 and Sf4 are compared. The difference between the expected similarity of the introns in these transposons based on the waxy gene data (52% of 227 = 118) and the observed similarity (89% of 227 = 202) is highly significant (p < 0.001 using a χ2 test).
A common measure of selection is the ratio dS/dN [28]. When waxy sequences from Setaria and rice are compared, that proportion is 9.82. This is a typical result for a reasonably well conserved host gene from these two species since selection operates much more efficiently on nonsynonymous base substitutions. In contrast, the putative coding regions from rice and Setaria MULEs have a dS/dN ratio of only 1.82, consistent with very weak selection (Table 1). This is further supported by the observation that the dS/dN ratio for the comparison between the Os493 and Zm890, two sequences that we hypothesize diverged at the same time as their hosts, is 3.96, consistent with selection on the nonsynonymous sites in this element. Similarly, when a well-conserved portion of exon Sb632 is compared to the same region of Os493 (nucleotides 1546–1917), there is clear evidence of selection at the amino acid level, with a dS/dN ratio of 7.5. For comparison, when Os493 and Sf4 are compared in the same region, the dS/dN ratio is only 2.1 (data not shown). These data strongly suggest that although MULEs can be subject to selection at the amino acid level (for example, when comparing Sb632 and Os493), the high degree of similarity between the elements in Setaria and rice is not due to selection at this level.
Codon bias is a possible source of selective constraint on synonymous nucleotides. A common measure of codon bias, the effective number of codons (Nc), can vary from 21 (only one codon used per amino acid) to 61 (equal use of all possible codons) [29]. The Setaria and rice sequences each have an Nc value of 48, consistent with only relatively moderate codon bias. We also observe that other pairs of mudrA homologs compared to date radically diverge at synonymous sites [17], suggesting that selection on MULEs can operate to maintain amino acid similarity but does not prevent silent substitutions. Together, these data suggest that codon bias is unlikely to explain the high degree of similarity between the Setaria and rice MULEs.
One possible explanation for the low sequence divergence of the rice and Setaria elements is that they are found in regions with a reduced mutation frequency. In order to test this hypothesis, genes flanking the Os493 insertion in rice were compared with orthologous sequences in maize. Maize was used as a proxy for Setaria because it is equally distant from rice as is Setaria, and, in contrast to Setaria, there are a very large number of genomic and cDNA sequences available in maize, making the identification of true orthologs relatively straightforward. If this region were subject to a reduced mutation frequency, we would expect to observe a low degree of variation when comparing intron or silent site sequences from the genes in rice with orthologs in maize.
The closest genes in rice with recognizable orthologs in other grasses were 144 kb 5′ of the Os493 element and 67 and 79 kb 3′ of this element. Although there were some closer putative ORFs, the lack of close homology to any known cDNA or genomic sequence made them poor candidates for comparison. Although each of these genes has at least one paralog in rice, the sequences we identified in maize were more similar to these genes than they were to their paralogs, indicating that the maize sequences are the true orthologs. The intron/exon boundaries of these genes were well supported by cDNA data as well as by protein homology.
Comparisons between the rice genes in the region surrounding Os493 and their orthologs in maize revealed a similar frequency of base substitutions in introns and synonymous sites as was observed in our other comparisons (see Table 1). Overall, exon sequences were 85% identical (2,297 of 2,704); introns were only 50% identical (891 of 1,799) (Table 1). dS, a measure of the synonymous substitution frequency, was similar to that observed in other comparisons, ranging from 0.54 to 0.87. These data demonstrate that the region into which the Os493 element is inserted does not exhibit a markedly reduced frequency of mutations.
In summary, the data are consistent with horizontal transfer of a MULE between two species of grass at some point well after the divergence of the Bambosoid and the Panicoid subfamilies. The degree of similarity of these two elements in noncoding sequences makes it highly unlikely that these two sequences diverged at the same time as their hosts.
Discussion
Our conclusion that a MULE was horizontally transferred between the Setaria and rice lineages leans heavily on evidence that these lineages diverged roughly 50 million years ago. Fortunately, the phylogeny of the grasses has been extensively documented, using a variety of criteria, including morphology, chloroplast restriction sites, both nuclear and mitochondrial genes, and overall gene and chromosome order [23]. Paleontologic evidence suggests that the overall age of the grasses is between 55 and 70 million years [30]. The minimum divergence time for the subfamily Bambosoideae and the rest of the grass subfamilies has been estimated to be between 35 and 60 million years [21,22,31], and the genus Setaria has been placed with high confidence in the subfamily Panicoideae [32]. Our analysis of three pairs of homologous genes from Setaria and rice is consistent with an early divergence for Setaria and rice; sequences immediately upstream and downstream of the coding sequences are radically diverged, as are intron sequences in the waxy gene (see Table 1). Further, corrected frequencies of synonymous substitution (dS) range from 0.41 to 1.17. The nuclear gene encoding plastid acetyl-CoA carboxylase (acc1) is particularly revealing because it has been used extensively in phylogenetic analysis of various grass subgroups [33] (and references therein). Not surprisingly, Setaria acc1 is much more similar to a portion of another Panicoid grass (Panicum virgatum) than it is to rice acc1 (95% versus 79% identity, respectively, over 696 bp).
Although the Setaria and rice MULEs are 89% identical, they have a dS/dN value of only 1.82, and their introns are as similar as the rest of the transposon (see Table 1; Figure 5). Thus, in contrast to host genes, the degree of similarity between synonymous sites and between introns when comparing the Setaria and rice MULEs is indeed an anomaly when comparing genes from these two species.
Although it is a formal possibility that the introns identified in the rice and Setaria MULEs are differentially spliced, resulting in conservation of intron sequences, we find this unlikely. In order to maintain the correct open reading frame and to avoid the appearance of multiple stop codons, each of the three introns in the Setaria and rice MULEs must be spliced. Further, we observe that homologs of Sf4 and Os493, Zm890 and Sb632, maintain similarity in the exons but lose similarity in the introns, suggesting that these introns are not normally subject to selection in this class of MULEs. Together, these data strongly suggest that all three putative introns are historically functional, suggesting that conservation within these introns when comparing the Setaria and rice MULEs is probably not due to differential splicing.
Our analysis of genes near the MULE Os493 insertion in rice demonstrates that this region does not exhibit a particularly low frequency of base substitutions; the introns in these genes exhibit little homology to those of their maize orthologs. Of course, it is still a formal possibility that the region immediately adjacent to the MULE exhibits a particularly low mutation frequency. However, the absence of sequences similar to the regions immediately adjacent to Os493 in any database suggests that these sequences are not well conserved. Further, it is worth noting that both the Setaria and the rice elements are located at a number of different chromosomal positions. Since all of these elements are quite similar to each other in noncoding and silent sites, they would all have to be located in regions with reduced mutation frequencies, which does not seem likely.
Overall, our data clearly indicate that the degree of similarity we observe between the Setaria and rice MULEs is not due to selection, nor does it appear that either MULEs in general or the regions in which these particular MULEs are found have a particularly low mutation rate; the only reasonable alternative explanation is horizontal transfer.
The mechanism of transfer between ancestors of the Bambusoid and the Panicoid subfamilies is a matter of pure speculation. Based on the divergence of the MULE sequences available and the presence of this class of elements in both rice and some bamboos, the transfer probably occurred several million years ago, although there may well be a Setaria MULE that we have not yet identified that is even more similar to the rice elements than the ones we have identified. The presence of the same MULE at the same chromosomal position in both the indica and japonica varieties demonstrates that the transfer happened before their divergence, which has been dated at roughly 1 million years ago [34]. The presence of very similar MULEs in a number of bamboo species suggests that the transfer may have occurred prior to the divergence of rice and bamboo. However, the rice and bamboo sequences are remarkably similar (95% identity between species that diverged at least 36 million years ago [31]). Further, the pattern of hybridization appears to have more to do with physical, rather than phylogenic, proximity [35]. Two Old World members of the subtribe Bambusinae (Shibataea kumasaca and Phyllostachys vivax) hybridize well to the probe, but so does the Old World bamboo Sinaruninaria nitada, a member of the subtribe Arundinariinae. Indeed, sequence analysis reveals that the MULEs from Shibateae kumasaca and Sinaruninaria nitada, members of different subtribes, are 99% identical (715 of 722). However, none of the three New World bamboos hybridize to the probe, including the two Chusques and Otatea acuminata, all of which are members of the subtribe Arundinariinae. Together, these data suggest that there likely were additional horizontal transfer events between these bamboo lineages as well.
The ancestors of O. sativa [36] and S. italica and S. faberi [37] all arose in southeast Asia, suggesting (although by no means proving) physical proximity at the time of transfer, and S. faberi is now a common weed next to rice fields in central China (personal observation). Both rice [38] and Setaria [39] are obligate self-fertilizers, but both can outcross at a low frequency. There is no evidence, however, that these rather distant subfamilies can intercross.
The possibility of horizontal transfer of genetic material between plants has been a subject of considerable interest to those concerned about the escape of transgenes from an intended species to another species. Often, the concern is related to movement of genes from plants to associated microbial species [40], which has in fact been observed to occur under laboratory conditions [41], but movement between plant species is also a valid concern given recent evidence for frequent transfer of mitochondrial genes between plant species [6,8]. Certainly, there is ample evidence for gene transfer between cultivated crops and their wild relatives via hybridization [42], and even very wide crosses (albeit with heroic efforts) are possible between quite distantly related plants [42]. Our data suggest that in addition to mitochondrial encoded genes, nuclear-encoded plant transposons have been transferred as well. Given the phylogenetic distances involved, we suggest this particular transfer was mediated via a vector of some kind.
It is worth noting that there are several features of transposon behavior that make them particularly prone to horizontal transfer. Transposable elements have the capacity to insert themselves into the chromosomes of possible vectors and, subsequently, into host chromosomes. Subsequent to transfer, they can spread rapidly throughout a given species, as is evidenced by the rapid spread of P elements in Drosophila melanogaster [43]. Thus, it is not surprising that many examples of horizontal transfer of transposons have been identified. However, it is also worth noting that the elements in both Setaria and rice appear to be inactive, as are the vast majority of all transposable elements in most species. This is likely related to the capacity of genomes to recognize and epigenetically inactivate foreign DNA from a variety of sources [44]. The problem of dealing with invasive DNA is, then, hardly a new one, and it is one that most organisms appear to be quite competent to deal with; indeed, maintaining transgene activity in the face of endogenous mechanisms evolved to silence invading DNA is a major problem for genetic engineers [45]. Without the intrinsic invasive properties of transposable elements, it is likely that other genes, such as transgenes, could effectively invade a new species only to the extent that they provide clear selective benefits to the recipient species.
What is perhaps most surprising about these results is that they are not more common. Plants are far more likely to undergo interspecific crosses than are animals [42], and, unlike animals, plants do not sequester their germ line. A vector-mediated transformation event that occurs in the vegetative or floral meristems can potentially be transmitted to the next generation. Thus, it is remarkable that these results represent the first well-documented case of horizontal transfer of nuclear genes between plants, particularly given the observation that plant mitochondrial genes appear to be particularly prone to horizontal transfer. Given the vast quantity of sequence data now available for a variety of plant species, the time would seem ripe for a comprehensive search for horizontally transferred plant genes.
Materials and Methods
Biological materials
Setaria samples were obtained from a variety of sources. Collection sites and accession numbers are available on request. Samples from all other species are as described in Lisch et al. [17].
Amplification and cloning
PCR amplification of the conserved portion of the mudrA transposase in a number of Setaria sp. was performed using PCR primers RF2:
CTTAGTGTAAACTCAACTGC and RR2:
GGCTTGCCAGTGTGTTGCCA . These primers are anchored in two well-conserved portions of the mudrA gene and span the region of nucleotides 1751–2369 of the complete MuDR element from maize [17]. Amplification products were cloned into the TA vector (Invitrogen, Carlsbad, California, United States), and multiple independent clones from each species were sequenced. In each case, both strands were sequenced. Each species gave rise to multiple products with few repeats, suggesting that only a subset of amplified products were sequenced from any given species. A total of 19 sequences from eight species of Setaria were subjected to detailed analysis. Clones from several species were strikingly similar to sequences in rice. One clone from S. faberi was found to be 90% identical over 585 bp to a portion of rice Chromosome 5. The terminal inverted repeats of that sequence were identified using blastn. The borders of the insertion were confirmed by identification of the host direct repeat sequence, a short sequence of DNA that is duplicated upon insertion. This complete element was designated Os943. A single primer specific to the first 20 bp of Os493 in rice (
GAGAAAATTGCAATTATAGG) was synthesized and used to amplify a similar complete element from S. faberi. One product of 3.85 kb was cloned into a TA vector (Invitrogen), and both strands of the complete sequence of a single clone were obtained using a series of overlapping sequencing primers. Products from each overlapping sequencing reaction were assembled using SeqMan (DNAStar, Madison, Wisconsin, United States). Although the sequence of this nearly complete element was different from that obtained in the first round of amplification (using primers RF2 and RR2), it retained a high degree of similarity to Os493. All samples were subjected to 35 rounds of amplification, gel isolated, and purified using the QIAquick Gel Extraction Kit (Qiagen, Valencia, California, United States) and subcloned using the TOPO TA Cloning kit (Invitrogen). The University of California, Berkeley sequencing facility performed sequencing using an Applied Biosystems sequencer (Foster City, California, United States).
Sequence analysis
Sequences were aligned to each other using ClustalW (http://www.ebi.ac.uk/clustalw/index.html) [46] or using blastn if the sequences were highly similar. The rice and Setaria sequences were also compared to MURA using tblastn (NCBI). This alignment was used to determine the correct reading frame for both the rice and Setaria sequences, which were then modified to maintain that reading frame for subsequent analysis of codon usage and bias. Putative introns were identified using SplicePredictor (http://bioinformatics.iastate.edu/cgi-bin/sp.cgi) and GeneSeqer (http://www.maizegdb.org/geneseqer.php) as well as comparisons to available cDNA and gene trap sequences and to MURA. These methods use a Bayesian model to predict the placement of intron sequences based on a large database of empirically determined splice sites in plants [47]. The last two putative introns (nucleotides 2809–2878 and 2996–3069 in Os493) gave high scores (p > 0.9) for donor and acceptor sites using SplicePredictor and GeneSequer and were missing in a spliced maize cDNA (AW067488). The first intron (1122–1203) also gave a high score using SplicePredictor and GeneSeqer and, when spliced, restored continuity with the maize MURA protein sequence. The fact that the inserted sequence precisely corresponds to predicted mRNA splice sites strongly suggests that this sequence is in fact an intron. Two other high-scoring putative introns (547–625 and 1978–2056) were excluded from the analysis because they lacked both cDNA and protein homology support.
A portion of recognizable open reading frame (2,394 bp) from MULEs Os493 and Sf4 was examined for differences in synonymous and nonsynonymous changes using the SNAP (Synonymous/Nonsynonymous Analysis Program) tool in the HIV Sequence Database (http://www.hiv.lanl.gov) [48] after manual correction of frameshifts or small deletions. This program uses the algorithm devised by Nei and Gojobori [28]. Codon bias as determined by the Nc value was computed using CodonW (http://bioweb.pasteur.fr/seqanal/interfaces/codonw.html) [29]. Multiple sequence alignments used to generate the phylogenetic tree in Figure 1 were performed using the ClustalW server available at European Bioinformatics Institute (http://www.ebi.ac.uk/clustalw/) with default parameters. The nucleotide sequences used corresponded to a 583-bp portion of MuDR (nucleotides 1171–2349). Included in the phylogenetic analysis are sequences previously obtained at this laboratory [17] as well as the best blastn hits to each of the sequences obtained in this study. In order to comprehensively search the database, each available sequence was blasted against all available databases, and the top two hits were then added to the dataset. This process was then repeated. Close duplicates were excluded. The phylogenetic tree in Figure 1 was generated based on the distance method, using PAUP* Version 4.0b8 (Sinauer Associates, Sunderland, Massachusetts, United States) using BioNJ with the Kimura 2 parameter distance correction and with 1,000 bootstraps. Branches with less than 50% bootstrap support were collapsed.
Identification of rice homologs of S. italica cDNAs
Complete cDNA sequences from S. italica were obtained from NCBI. These sequences were used as queries to databases including all publicly available rice sequences, including the complete draft of the indica subspecies. The most significant blast hits were retrieved and aligned with the Setaria sequences using ClustalW. Coding versus noncoding sequences were determined using annotation available for each gene or by using blastx to compare with known protein sequences. To determine similarity of introns, noncoding upstream and downstream UTRs, the sequences of each gene (or portion of gene) pair were aligned using ClustalW and the percent identity calculated.
Identification of genes near the Os493 and their orthologs in maize
The Os493 insertion was located using the Gramene database (http://www.gramene.org/, which allows visualization of large contiguous regions of the rice genome, along with associated significant BLAST hits to a variety of other databases, including cDNA and genomic information from rice, maize, and other grasses. The region in rice flanking Os493 can be visualized at (http://www.gramene.org/japonica/contigview?highlight=&chr=5&vc_start=2800000&vc_end=3065000&x=44&y=11). The NCBI accession numbers for the corresponding proteins are as indicated in Table 1. Genes in rice were chosen for comparison with maize if the following criteria were met: (1) the rice gene had multiple hits to rice cDNA sequences (supporting theoretical intron/exon boundaries), (2) the rice gene matched a large number of maize and other genomic and cDNA sequences, (3) the gene was not a transposon, and (4) the hit to maize was better than it was to any rice orthologous sequence. When possible, maize genomic matches were combined using DNAStar, SeqMan program (Lasergene) in order to obtain as much contiguous maize sequence as possible. After assembly, exon and intron sequences from each rice gene were independently aligned with the orthologous maize exons and introns using either blastn (for exons, which had a high degree of similarity) or ClustalW (http://www.ebi.ac.uk/clustalw/index.html) (for introns, which could not be aligned using blastn). A summary of the results of these alignments is presented in Table 1.
DNA gel blotting
DNA extraction and gel blotting were performed as described in Lisch et al. [17]. The DNA was digested with BamHI. Washes were performed at 65° C in 0.2× SSPE and 0.2% SDS for a total of 1 hr. The probe used was a 585-bp fragment of a mudrA-homologous sequence obtained from S. faberi (Sf3).
Supporting Information
Accession Numbers
The GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession numbers for rice sequences in which the MULEs described in this paper are found are Os493 in O. sativa ssp. japonica (AC093493, nucleotides 152975–156800) and O. sativa ssp. indica (
AAAA01000154, nucleotides 41163–45085), a spliced version of Os493 (CL970142), Os072 (AP001072, nucleotides 86139–89558), Os650 (
AAAA01004650, nucleotides 3332–6752), and Os408 (AP003408, nucleotides 40419–41521). The accession numbers for portions of other MULES described in this paper are Zm890 (CG270890) and Sb632 (CW076632). The accession number for the complete MULE from S. faberi designated Sf4 that was sequenced by this laboratory is DQ287976.
The accession numbers for the genes and gene products discussed in this paper are MULE EST similar to Zm890 (AW067488), waxy from S. italica (AB089141), waxy from O. sativa (AF515483), acc1 from O. sativa (NM_196052), acc1 (acetyl-coenzyme A carboxylase) from S. italica (AY219174), acc1 from O. sativa (NM_196052), acc1 from P. virgatum (AF342959), oxidoreductase from S. italica (AY266141), oxidoreductase from O. sativa (AK067136), waxy (granule-bound starch synthase) from S. italica (AB089141), and waxy from O. sativa (AF515483). The accession numbers for proteins encoded by genes flanking Os493 in O. sativa are as indicated in Table 1.
The authors thank George Theodoris, Richard Slotkin, and Margaret Woodhouse for critical reading of the manuscript; the University of California, Berkeley Botanical Garden for providing plant material for this study; and Zoya Akulova-Bakulova for collection and technical assistance. This work was funded by grants from the Novartis Foundation to University of California, Berkeley, the National Science Foundation (MCB 0112346 and DBI 0321726) to DL and MF, and a grant from Natural Science Foundation of China (program code 30370766 to XD).
Competing interests. The authors have declared that no competing interests exist.
Author contributions. DL conceived and designed the experiments. XD performed the experiments. MF and DL analyzed the data and contributed reagents/materials/analysis tools. DL wrote the paper.
¤ Current address: Institute of Millet Crops, Hebei Academy of Agricultural and Forestry Sciences, Shijiazhuang, China
Citation: Diao X, Freeling M, Lisch D (2006) Horizontal transfer of a plant transposon. PLoS Biol 4(1): e5.
Abbreviations
BLASTBasic Local Alignment Search Tool
bpbase pair
MULE
Mu-like element
NCBINational Center for Biotechnology Information
==== Refs
References
Jain R Rivera MC Moore JE Lake JA Horizontal gene transfer in microbial genome evolution Theor Popul Biol 2002 61 489 495 12167368
Doolittle WF Phylogenetic classification and the universal tree Science 1999 284 2124 2129 10381871
Syvanen M Kado C Horizontal gene transfer 2002 New York Academic Press 445
Silva JC Kidwell MG Horizontal transfer and selection in the evolution of P elements Mol Biol Evol 2000 17 1542 1557 11018160
Robertson HM Soto-Adames FN Walden KO Avancini RMP Lampe DJ Syvanen M Kado CI The mariner transposons of animals: Horizontally jumping genes Horizontal gene transfer 1998 New York Chapman and Hill 268 284
Bergthorsson U Adams KL Thomason B Palmer JD Widespread horizontal transfer of mitochondrial genes in flowering plants Nature 2003 424 197 201 12853958
Bergthorsson U Richardson AO Young GJ Goertzen LR Palmer JD Massive horizontal transfer of mitochondrial genes from diverse land plant donors to the basal angiosperm Amborella Proc Natl Acad Sci U S A 2004 101 17747 17752 15598737
Cho Y Palmer J Multiple acquisitions via horizontal transfer of a group I intron in the mitochondrial cox1 gene during evolution of the Araceae family Mol Biol Evol 1999 16 1155 1165 10486971
Kempken F Windhofer F The hAT family: A versatile transposon group common to plants, fungi, animals, and man Chromosoma 2001 110 1 9 11398971
Rubin E Lithwick G Levy AA Structure and evolution of the hAT transposon superfamily Genetics 2001 158 949 957 11454746
Feschotte C Wessler SR
Mariner -like transposases are widespread and diverse in flowering plants Proc Natl Acad Sci U S A 2002 99 280 285 11756687
Wicker T Guyot R Yahiaoui N Keller B
CACTA transposons in Triticeae. A diverse family of high-copy repetitive elements
Plant Physiol 2003 132 52 63 12746511
Matsuoka Y Tsunewaki K Evolutionary dynamics of Ty1-copia group retrotransposons in grass shown by reverse transcriptase domain analysis Mol Biol Evol 1999 16 208 217 10028288
Hartl DL Lohe AR Lozovskaya ER Modern thoughts on an ancyent marinere: Function, evolution, regulation Annu Rev Genet 1997 31 337 358 9442899
Lisch D Mutator transposons Trends Plant Sci 2002 7 498 504 12417150
Yu Z Wright SI Bureau TE Mutator-like elements in Arabidopsis thaliana Structure, diversity and evolution Genetics 2000 156 2019 2031 11102392
Lisch DR Freeling M Langham RJ Choy MY Mutator transposase is widespread in the grasses Plant Physiol (Rockville) 2001 125 1293 1303 11244110
Rossi M Araujo PG Van Sluys MA Survey of transposable elements in sugarcane expressed sequence tags (ESTs) Genet Mol Biol 2001 24 147 154
Mao L Wood TC Yu YS Budiman MA Tomkins J Rice transposable elements: A survey of 73,000 sequence-tagged-connectors Genome Res 2000 10 982 990 10899147
Turcotte K Srinivasan S Bureau T Survey of transposable elements from rice genomic sequences Plant J 2001 22 169 179
Stebbins GL Coevolution of grasses and herbivores Ann Mo Bot Gard 1981 68 75 86
Wolfe KH Gouy ML Yang YW Sharp PM Li WH Date of the monocot dicot divergence estimated from chloroplast DNA-sequence data Proc Natl Acad Sci U S A 1989 86 6201 6205 2762323
Kellogg EA Evolutionary history of the grasses Plant Physiol 2001 125 1198 1205 11244101
Singer T Yordan C Martienssen R Robertson's Mutator transposons in A. thaliana are regulated by the chromatin-remodeling gene decrease in DNA methylation (DDM1) Genes Dev 2001 15 591 602 11238379
Kaplinsky NJ Braun DM Penterman J Goff SA Freeling M Utility and distribution of conserved noncoding sequences in the grasses Proc Natl Acad Sci U S A 2002 99 6147 6151 11972021
Lockton S Gaut BS Plant conserved non-coding sequences and paralogue evolution Trends Genet 2005 21 60 65 15680516
Ono A Kim SH Walbot V Subcellular localization of MURA and MURB proteins encoded by the maize MuDR transposon Plant Mol Biol 2002 50 599 611 12374294
Nei M Gojobori T Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions Mol Biol Evol 1986 3 418 426 3444411
Wright F The ‘effective number of codons' used in a gene Gene 1990 87 23 29 2110097
Jacobs BF Kingston JD Jacobs LL The origin of grass-dominated ecosystems Ann Mo Bot Gard 1999 86 590 643
Guo YL Ge S Molecular phylogeny of Oryzeae (Poaceaf) based on DNA sequences from chloroplast, mitochondrial, and nuclear genomes Am J Bot 2005 92 1548 1558 21646172
Giussani LM Cota-Sánchez JH Zuloaga FO Kellogg EA A molecular phylogeny of the grass subfamily Panicoideae (Poaceae) shows multiple origins of C 4 photosynthesis Am J Bot 2001 88 1993 2012 21669633
Huang SX Su XJ Haselkorn R Gornicki P Evolution of switchgrass (Panicum virgatum L ) based on sequences of the nuclear gene encoding plastid acetyl-CoA carboxylase Plant Sci 2003 164 43 49
Bennetzen JL Comparative sequence analysis of plant nuclear genomes: Microcolinearity and its many exceptions Plant Cell 2000 12 1021 1029 10899971
Clayton WD Renvoize SA Coode MJE
Genera graminum
Grasses of the world 1986 London Her Majesty's Stationary Office 389 p
Khush GS Origin, dispersal, cultivation and variation of rice Plant Mol Biol 1997 35 25 34 9291957
Rominger JM Taxonomy of Setaria (Gramineae) in North America Biol Monogr 1962 29 1 118
Allard RW Principles of plant breeding 1960 New York John Wiley 485
Li HW Li CH Pao WK Cytological and genetical studies of the interspecific cross of the cultivated foxtail millet, Setaria italica and the green foxtail Setaria viridis
J Am Soc Agronomy 1945 37 32 54
Nielsen KM Bones AM Smalla K van Elsas JD Horizontal gene transfer from transgenic plants to terrestrial bacteria—A rare event? FEMS Microbiol Rev 1998 22 79 103 9729765
Tepfer D Garcia-Gonzales R Mansouri H Seruga M Message B Homology-dependent DNA transfer from plants to a soil bacterium under laboratory conditions: Implications in evolution and horizontal gene transfer Transgenic Res 2003 12 425 437 12885164
Ellstrand NC Current knowledge of gene flow in plants: Implications for transgene flow Phil Trans R Soc Lond Ser B-Biol Sci 2003 358 1163 1170 12831483
Anxolabehere D Kidwell M Periquet G Molecular characteristics of diverse populations are consistent with the hypothesis of a recent invasion of Drosophila melanogaster by mobile P elements Mol Biol Evol 1988 5 252 269 2838720
Wassenegger M Gene silencing-based disease resistance Transgenic Res 2002 11 639 653 12509140
Lessard PA Kulaveerasingam H York GM Strong A Sinskey AJ Manipulating gene expression for the metabolic engineering of plants Metab Eng 2002 4 67 79 11800576
Thompson JD Higgins DG Gibson TJ Clustal-W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice Nucleic Acids Res 1994 22 4673 4680 7984417
Usuka J Zhu W Brendel V Optimal spliced alignment of homologous cDNA to a genomic DNA template Bioinformatics 2000 16 203 211 10869013
Korber B Rodrigo AG Jr Learn GH HIV signature and sequence variation analysis Computational analysis of HIV molecular sequences 2002 Dordrecht, the Netherlands Kluwer Academic Publishers 55 74
|
16336045
|
PMC1310652
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e5
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040005
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpbioplosbiolPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 1633604610.1371/journal.pbio.0040010Research ArticleDevelopmentEts1-Related Protein Is a Key Regulator of Vasculogenesis in Zebrafish Zebrafish Etsrp Regulates VasculogenesisSumanas Saulius
1
Lin Shuo [email protected]
1
1Department of Molecular, Cell, and Developmental Biology, University of California, Los Angeles, United States of AmericaStemple Derek Academic EditorWellcome Trust Sanger InstituteUnited Kingdom1 2006 20 12 2005 20 12 2005 4 1 e1008 7 2005 1 11 2005 Copyright: © 2006 Sumanas and Lin.2006This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
A Master Switch? estrp Directs Blood Vessel Development in Zebrafish
The "Ets" Factor: Vessel Formation in Zebrafish-The Missing Link?
During embryonic development, multiple signaling pathways control specification, migration, and differentiation of the vascular endothelial cell precursors, angioblasts. No single gene responsible for the commitment of mesenchymal cells to the angioblast cell fate has been identified as yet. Here we report characterization and functional studies of Etsrp, a novel zebrafish ETS domain protein. etsrp embryonic expression is only restricted to vascular endothelial cells and their earliest precursors. Morpholino knockdown of Etsrp protein function resulted in the complete absence of circulation in zebrafish embryos. Angioblasts in etsrp–morpholino-injected embryos (morphants) failed to undergo migration and differentiation and did not coalesce into functional blood vessels. Expression of all vascular endothelial molecular markers tested was severely reduced in etsrp morphants, whereas hematopoietic markers were not affected. Overexpression of etsrp RNA caused multiple cell types to express vascular endothelial markers. etsrp RNA restored expression of vascular markers in cloche mutants, defective in hematopoietic and endothelial cell formation, arguing that etsrp functions downstream of cloche in angioblast formation. etsrp gene function was also required for endothelial marker induction by the vascular endothelial growth factor (vegf) and stem cell leukemia (scl/tal1). These results demonstrate that Etsrp is necessary and sufficient for the initiation of vasculogenesis.
The authors report on a novel protein, Ets1 related protein (Etsrp), which is both necessary and sufficient for the development of the endothelial cells that line blood vessels (angioblasts).
==== Body
Introduction
Vasculogenesis, or formation of vascular endothelial cells de novo, begins very early after the initiation of gastrulation in a vertebrate embryo. In a mammalian embryo, vasculogenesis starts with the formation of the blood islands in the yolk sac and angioblast precursors in the head mesenchyme [1]. In the zebrafish, the first angioblasts arise from the lateral plate mesoderm, migrate to the trunk midline between ten- and 15-somite stages in response to hedgehog signaling, and coalesce to form the primary axial vessels of the trunk, the dorsal aorta, and the cardinal vein [2–4]. During subsequent angiogenesis, the axial vessels sprout to form secondary vessels in the trunk region of a zebrafish embryo [5]. In the zebrafish, as well as the mammalian yolk sac, there is a close association between the primitive hematopoietic cells and the developing endothelium, suggestive of a common precursor, the hemangioblast [6]. Zebrafish cloche mutants lack nearly all blood and endothelial cells suggesting that the hemangioblast lineage has been affected [7,8]. A basic helix-loop-helix transcription factor Scl/tal1 is expressed in hematopoietic and endothelial cells in both mouse and zebrafish suggesting a possible role in specification of the hemangioblast [9–12]. Recent knockdown analysis has demonstrated its essential function in blood and dorsal aorta formation in zebrafish embryos [13,14].
Members of the Ets family of transcription factors play multiple roles during vasculogenesis, angiogenesis, and hematopoiesis [15]. They share a conserved DNA-binding domain of 85 amino acids which folds into a winged helix-turn-helix motif. Ets1, the founding member of the family, is expressed in the embryonic endothelial cells as well as putative hemangioblasts, lymphoid precursors, and myeloid hematopoietic cells [16–19]. The expression of Ets1 transcripts is associated in vivo with the activation of endothelial cells and the induction of angiogenesis. In vitro, Ets1 is expressed by proliferating and migrating endothelial cells but not after these cells have reached confluence [20,21]. In endothelial cells in vitro, Ets1 has been shown to regulate expression of the two vascular endothelial growth factor (Vegf) receptors Flt1 and Flk1 [22,23], the endothelial cell-specific adhesion protein VE-cadherin [24], the vascular-specific tyrosine kinases Tie1 and Tie2 [25], and numerous other target genes [15]. Ets1 antisense oligonucleotides inhibited angiogenesis in vitro [26,27]. However, no defects in vascular development have been found in the mouse embryos, nearly completely deficient in Ets1 function [28,29]. Therefore, the function of Ets1 in vascular development in vivo, if any, remains unclear.
A different Ets family member Fli1 has been implicated in mouse megakaryopoiesis, hemostasis, and maintenance of vascular integrity [30,31]. Zebrafish fli1 homolog is expressed during early somitogenesis in the putative hemangioblast cells, while later being restricted to vascular endothelial cells [32].
Recently, an endothelial-cell-specific Ets1-related zebrafish protein Etsrp has been identified [33]. In the current study, we demonstrate a critical role of Etsrp during vasculogenesis. In the absence of Etsrp, angioblast precursors fail to migrate, differentiate, and coalesce into functional vessels. Furthermore, ectopic expression of etsrp is sufficient to induce endothelial markers in a variety of different cell types throughout the embryo. These results suggest that Etsrp functions as a critical regulatory gene, directing etsrp-expressing cells to adopt vascular endothelial cell fate during embryonic development.
Results
Etsrp Encodes a Novel Ets1-Related Protein
A novel Ets1-related protein Etsrp has been identified in a recent microarray study of the zebrafish cloche mutants [33]. We performed detailed analysis of the protein sequence and its developmental expression pattern. Etsrp-predicted protein sequence contains 366 amino acids and displays 29% identity (37% similarity) to the human Ets1 protein and 26% identity (34% similarity) to the human Ets2 (Figure 1A) (unpublished data). Although Etsrp and hEts1 proteins are 88% similar within the ETS domain region (Figure 1A), there is little similarity between the two proteins within the rest of the sequence, therefore Etsrp may not be a functional ortholog of the Ets1 protein. Supporting this idea, a recently published protein sequence of the zebrafish, Ets1 homolog, was found to be highly similar to the Ets1 subfamily throughout most of the protein sequence (Figure 1A) [34]. No other homology domains were found outside the ETS domain of Etsrp. We were unable to find an apparent Etsrp ortholog in other species. As evident from the homology tree, Etsrp is the most closely related to Ets1 and Ets2 subfamilies (Figure 1B). We performed synteny analysis of human, mouse, and zebrafish ets1 or etsrp chromosomal regions. Human and mouse ets1 genes are positioned next to fli1 genes, which encode related ETS domain proteins (Figure 1C). ets1 and fli1 genes are transcribed in opposite directions and are likely to have originated from the same ancestral precursor via gene duplication. Zebrafish etsrp is also positioned in the opposite direction and next to a fli1-related gene fli1b [34]. This analysis suggests that etsrp is evolutionarily related to the mammalian ets1 genes. As the zebrafish genome has undergone an additional duplication [35], etsrp and ets1 may have diverged from each other and acquired separate functions. Alternatively, it is possible that ets1 duplicated prior to the divergence of teleosts and tetrapods, and etsrp was subsequently lost in the tetrapod lineage.
Figure 1 Sequence Analysis of the Zebrafish Ets1-Related Protein Etsrp
(A) Shows alignment of Etsrp and its closest human and zebrafish homologs Ets1 proteins. Etsrp and hEts1 share 88% homology within the ETS DNA-binding domain (underlined in red), and a very limited homology within the rest of the sequence. Identical and similar amino acids are shaded in grey. (B) The homology tree of Etsrp and its closest human, mouse, and zebrafish homologs. Length of horizontal branches is directly proportional to the evolutionary distance between the proteins. Zebrafish Ets1 and Ets2 protein sequences have been predicted using the available EST sequences TC282499 and TC270146 (http://www.tigr.org). GeneWorks 2.5 has been used to build the alignment and the homology tree. (C) Chromosomal location of the zebrafish etsrp, mouse, and human ets1 genes. In all cases, they are positioned next to a fli1 homolog.
Etsrp Expression Pattern Analysis
We analyzed the expression pattern of etsrp RNA in early embryos using in situ hybridization. No expression was observed prior to the one-somite stage (unpublished data). From the two-somite stage onward, etsrp RNA was localized to two stripes of cells within the lateral mesoderm (five-somite embryo, Figure 2A). Two distinct expression domains, in the anterior and posterior parts of an embryo, were apparent. The anterior stripes merged at the prechordal plate. The etsrp-expressing cells are likely endothelial cell precursors based on the expression of the gene at later stages. At the 15-somite stage, expression in two bilateral stripes of cells within the lateral mesoderm in the anterior and middle/posterior parts of an embryo, was apparent (Figure 2B–2D). In addition, a stripe of etsrp-expressing cells was noted at the midline and extended through the trunk and posterior parts of an embryo. The middle stripe likely represents the endothelial precursor cells that have already migrated to the future intermediate cell mass (ICM) region. By 26 hours-post-fertilization (hpf), etsrp was expressed in vascular endothelial cells in the embryo marking main axial, head, and intersegmental vessels (Figure 2E). In addition, a group of etsrp-expressing cells was located in the intermediate mesoderm region (Figure 2E and 2F). These cells which by 36 hpf were located close to the pronephros (Figure 2G), and appeared to migrate posteriorly, may represent endothelial cell precursors of pronephric vessels and/or gut vessels. By 36 hpf etsrp expression had mostly disappeared from the axial vessels and was prominent in a subset of head vessels, aortic arches, the cardinal vein plexus region, posterior intersomitic vessels, and the dorsal longitudinal anastomotic vessel (Figure 2G). At 52 hpf, etsrp expression was observed in a subset of head vessels, the common cardinal vein, blood vessels of the pectoral fin, the cardinal vein plexus region, and weakly in posterior intersegmental vessels and the dorsal vessel (Figure 2H).
Figure 2 Expression Pattern of etsrp as Analyzed by In Situ Hybridization
Anterior is to the left except as noted. (A) five-somite stage. etsrp is expressed within lateral mesoderm in two distinct expression domains, in the anterior and posterior parts of an embryo. (B) 15-somite stage, lateral view. (C) dorsal view. (D) transverse section. etsrp is expressed in two bilateral stripes of presumptive angioblasts within the lateral mesoderm in the anterior and the trunk and posterior parts of an embryo (arrows, D). In addition, a stripe of etsrp-expressing cells is apparent at the midline and extends through the middle and posterior parts of an embryo (arrowhead, C and D). (E) 26 hpf stage. etsrp is expressed in vascular endothelial cells of the axial, head, and intersomitic vessels. Note a group of etsrp-expressing cells bilaterally located in the intermediate mesoderm (arrowhead). (F) Transverse section through the trunk region of a 30 hpf embryo. Arrowhead shows one of the etsrp-expressing cells located within the lateral/intermediate mesoderm. Expression of etsrp in the axial vessels is weak at this stage and not apparent in this section. nt, neural tube; n, notochord; y, yolk. (G) 36 hpf stage. etsrp is expressed in a subset of head vessels, the aortic arches (aa), the cardinal vein plexus (pl) region, posterior intersomitic vessels, and the dorsal longitudinal anastomotic vessel (dlav). Note a group of etsrp-expressing cells located in the endodermal region (arrowhead). (H) 52 hpf stage. etsrp expression is observed in a subset of head vessels, common cardinal vein, pectoral fin bud blood vessels (fb), cardinal vein plexus region, and weakly in posterior intersegmental vessels and the dorsal vessel.
Morpholino Knockdown of Etsrp Function Results in Loss of Circulation
We used antisense morpholino (MO) oligonucleotides [36,37] to knockdown the function of Etsrp. Etsrp–MO-injected embryos (morphants) showed no apparent circulation. While red blood cells were actively circulating in wild-type (wt) embryos at 34 hpf, in etsrp morphants they stayed at their formation site, the ICM region (Figure 3A and 3B). This observation was confirmed by the o-dianisidine staining of heme in the red blood cells (Figure 3C–3E). We performed microangiography analysis by injecting fluorescein-labeled high molecular weight dextran into the sinus venosus of the 2-day-old embryos. Lower doses of etsrp MOs caused partial loss of circulation in the intersegmental and axial vessels, particularly, in the posterior part of the embryo (Figure 3F and 3G). Head circulation was not affected at this dose (Figure 3G). Higher doses of etsrp MOs resulted in the complete loss of functional blood vessels (Figure 3H and 3I, Table 1). Other than the described defects, etsrp morphants appeared morphologically normal (Figure 3B). At 2–3 days-post-fertilization (dpf), etsrp morphants developed pericardial edema and eventually became necrotic and died (unpublished data). Injection of two different etsrp-specific translation-blocking MOs resulted in similar circulatory defects, demonstrating specificity of the observed phenotype (Table 1). To confirm that etsrp MOs were specifically inhibiting synthesis of Etsrp protein, we generated etsrp–green fluorescent protein (GFP) modified bacterial artificial chromosome (BAC). BAC construct of approximately 250 kilobases containing etsrp gene was modified using recA-mediated homologous recombination to replace etsrp coding sequence with GFP [38]. Microinjection of etsrp-GFP BAC resulted in the mosaic GFP expression (Figure S1A). Etsrp MO2, directed against the 5′ UTR of etsrp, completely inhibited this transient GFP expression (Figure S1B). Control injection of a 5-base mismatch MO had no effect on GFP expression (Figure S1C). These results confirm that etsrp MOs specifically inhibit etsrp gene function. Microinjection of a low dose of etsrp MO into flk1-GFP transgenic embryos [39] resulted in partial loss of flk1 expression in intersegmental vessels (Figure 3J–3L). High doses of etsrp MOs resulted in nearly complete loss of flk1 expression in axial vessels and strong downregulation of its expression in head vessels (Figure 3M).
Figure 3 MO Knockdown of Etsrp Protein Function Disrupts Blood Vessel Formation in the Zebrafish Embryos
(A,B) Morphological analysis of live etsrp morphants at 34 hpf. (A) Uninjected control embryo. (B) 5 ng of etsrp MO1-injected embryo. Notice that red blood cells are scattered throughout the circulatory system in the control uninjected embryo while they accumulate at their formation site within the intermediate cell mass (arrow, B) in the etsrp morphant. (C–E) o-dianisidine staining of heme in the red blood cells of uninjected control (C), 5 ng of etsrp MO1-injected (D) and 5 ng of etsrp MO2-injected (E) embryos at 34 hpf. While many circulating blood cells are apparent within the common cardinal vein before entering the heart in the control embryo (arrow, C), they stay at their formation site within the ICM region in etsrp morphants (arrows). (F–I) Microangiography analysis of the circulatory system by injecting fluorescein-labeled dextran into the sinus venosus of etsrp morphants at 55 hpf. (F) Control uninjected, (G) 1 ng of etsrp MO1-injected (H) 2.5 ng of etsrp MO1-injected (I) 5 ng of etsrp MO1-injected embryos. Note that the embryo in (G) has lost circulation in the posterior vessels, the embryo in (H) has lost circulation in most vessels, and the embryo in (I) has no circulation at all. (J–M) Analysis of blood vessels in live flk1-GFP transgenic embryos at 26 hpf. (J) Control uninjected, (K) 1 ng of etsrp MO1-injected, (L) 2.5 ng of etsrp MO1-injected, (M) 15 ng of etsrp MO1+MO2 (1:1) mix-injected embryos. Note the gaps in formation of intersegmental vessels in (K) (arrowhead), the missing (arrowhead) and abnormally branched, (arrow) intersegmental vessels in (L), and the nearly completely eliminated flk1 expression from axial vessels (arrow) in (M).
Table 1 Dose Dependence of Etsrp MO Phenotype
Vascular Endothelial Markers Are Lost in Etsrp Morphants
We performed analysis of different molecular markers in etsrp morphants using high doses of etsrp MOs. Kinase insert domain receptor (kdr, flk1) [40], cadherin 5 (cdh5, VE-cadherin) [41], adrenomedullin receptor (admr) [42], dual specificity phosphatase 5 (dusp5) [42], and C1qR-like (crl) [42] genes are expressed in all vascular endothelial cells while fms-related tyrosine kinase 4 (flt4) [40] is restricted to the venous vessels in early zebrafish embryos. Expression of all of these markers in axial and intersegmental vessels was nearly completely abolished in etsrp morphants at 24 hpf (Figure 4A–4L). Head vessel expression, however, was reduced but still apparent even at high MO doses (Figure 4A–4D) (unpublished data). We were unable to use even higher MO doses to attempt to completely eliminate Etsrp function due to mild toxic effects observed at doses above 15 ng (unpublished data).
Figure 4 Molecular Analysis of Vascular Endothelial and Hematopoietic Markers in etsrp Morphants
(A, C, E, G, I, K, M, O, Q, S) Uninjected control embryo; (D, H, J, L, P, R, T) 8 ng etsrp MO2-injected embryo; (B, F, N) 12 ng etsrp MO1+MO2 (1:1) mix-injected embryos for the maximum knockdown. All embryos are at 24 hpf, except as noted. Scale bar, 0.2 mm. (A,B) flk1 expression; (C,D) admr expression; (E,F) cdh5 expression; (G,H) dusp5 expression; (I,J) flt4 expression; (K,L) crl expression. Note that the vascular expression of the markers in (A–L) is almost absent in etsrp morphants. (M,N) fli1 expression; (O,P) etsrp expression. Note the more intense etsrp expression and reduced fli1 expression in angioblasts which remain dispersed and fail to coalesce into blood vessels in etsrp morphants. Inset, (P) DIC image of scattered angioblasts in etsrp morphants. (Q,R) scl expression, 22 hpf. Note that scl expression appears unaffected at this stage except for more intense staining in a subset of head vessels in etsrp morphants (arrowhead). (S,T) gata1 expression, 22 somites. Note that no significant difference in hematopoietic gata1 expression is observed between the control embryos and etsrp morphants.
Expression of friend leukemia integration 1 (fli1) [40] was downregulated but still apparent in etsrp morphants (Figure 4M and 4N). Interestingly, expression of etsrp itself was upregulated in etsrp morphants, suggesting the presence of a negative autoregulatory loop (Figure 4O and 4P). However, the number of etsrp-expressing angioblasts was reduced in etsrp morphants. These angioblasts remained scattered within the lateral mesoderm failing to coalesce into functional blood vessels (Figure 4P). Formation of hematopoietic precursors appeared unaffected as evident from scl and gata1 expression (Figure 4Q–4T).
We analyzed formation of hematopoietic and vascular progenitors during early somitogenesis stages. A transcription factor scl is expressed in both hematopoietic and vascular endothelial cell precursors [12]. Expression of scl in etsrp morphants was reduced in the anterior domain and absent from the trunk domain while its posterior expression was not affected at the six- and ten-somite stages (Figure 5A–5D). fli1 is expressed in the putative hemangioblast cells while later being restricted to vascular endothelial cells [32]. Anterior expression of fli1 was absent in etsrp morphants while its tail expression was not affected at the six-somite stage (Figure 5E and 5G). At the ten-somite stage, fli1 expression in etsrp morphants was absent from the anterior and trunk domains (Figure 5F and 5H). As described earlier, angioblasts migrate from the lateral mesoderm towards the midline during somitogenesis [2,3]. etsrp-expressing angioblasts were present within the lateral mesoderm but failed to migrate towards the midline in etsrp morphants (Figure 5I–5L). Also, no scl- or fli1-expressing migrating angioblasts were observed in etsrp morphants (unpublished data). etsrp expression level within the angioblasts was strongly increased in etsrp morphants suggesting the presence of negative autoregulation. All of the etsrp-expression domains, including the trunk domain, were present in etsrp morphants (Figure 5I–5L) (unpublished data).
Figure 5 Molecular Analysis of Early Vasculogenesis in etsrp Morphants
(A, B, E, F, I, K) Uninjected control embryo; (C, D, G, H, J, L) 8–10 ng etsrp MO2-injected embryo. Anterior is to the left in all panels. (A–H) Embryos were flat mounted with their yolk removed. (A–D) scl expression; six-somite (A,C) and ten-somite (B,D) stages. Note that the anterior domain of scl expression (arrows) is reduced and the trunk domain (arrowheads) is missing in etsrp morphants. (E–H) fli1 expression; six-somite (E,G) and ten-somite (F,H) stages. Note that the anterior domain of fli1 expression (arrows) is missing in the etsrp morphants, while the posterior domain is not affected. Also note that the trunk domain of fli1 expression (arrowheads, F,H) is missing at the ten-somite stage in etsrp morphants. (I–L) Etsrp knockdown blocks angioblast migration towards the midline as assayed by etsrp expression at the 16-somite (I,J) and 20-somite (K,L) stages. (I,K) Uninjected control embryo; (J,L) 7.5 ng etsrp MO2-injected embryo. Dorsal view, anterior is to the left. Note that the midline stripe of angioblasts (arrows) is missing in etsrp morphants. Also notice more intense etsrp expression in pre-migratory angioblasts (arrowheads) in etsrp morphants as compared to control embryos.
Etsrp is Sufficient for the Vascular Endothelial Marker Induction
We analyzed if etsrp mRNA was sufficient for induction of the vascular endothelial markers. Injection of synthetic etsrp mRNA into zebrafish embryos at the 1–16 cell stage resulted in strong ectopic induction of vascular endothelial markers flk1, scl, fli1, and cdh5 during somitogenesis (Figure 6A–6F) (unpublished data). Large patches of intense ectopic flk1 expression were observed within different germ layers including dorsal, lateral, and ventral mesoderm, endoderm and neuroectoderm (Figure 6E and 6F) (unpublished data). No induction of hematopoietic marker gata1 was observed (Figure 6G and 6H). These results indicate that etsrp is sufficient to induce vascular endothelial gene expression in a variety of cell types. Furthermore, etsrp specifically induces vascular markers without affecting closely related hematopoiesis.
Figure 6 etsrp RNA Overexpression Induces Ectopic Expression of Vascular Endothelial Markers
Dorsal view, anterior to the left in all panels except for (E,F) which are lateral views. (A, C, E, and G) Control uninjected embryo; (B, D, F, and H) 100 pg of etsrp RNA-injected embryo. (A,B) scl expression at the eight-somite stage; (C,D) flk1 expression at the nine-somite stage. Note the strong ectopic induction of scl and fli1 upon overexpression of etsrp RNA. (E,F) Live flk1-GFP embryo at the 14-somite stage; fluorescent and transmitted light images were overlayed. Note the very strong ectopic induction of GFP expression in different tissues including neuroectoderm (arrow, F) upon etsrp RNA overexpression. Fluorescence in the control uninjected flk1-GFP embryo in (E) is not detectable under the same exposure. (G,H) gata1 expression at the 16-somite stage. Note that gata1 expression is not affected upon etsrp overexpression. (I–L) etsrp RNA induces flk1 expression in clo mutant embryos as analyzed using flk1 probe at the ten- to 12-somite stages. (I) wt (or clo+/−) embryo, (J) wt (or clo+/−) embryo injected with 100 pg of etsrp RNA, (K) clo−/− embryo, (L) clo−/−embryo injected with 100 pg of etsrp RNA. Note that in a clo+/− (or wt) embryo etsrp RNA induces ectopic flk1 (arrow, J) in addition to the endogenous flk1 expression (arrowheads, J) while clo-/− etsrp RNA-injected embryo shows only ectopic flk1 (arrows, L).
We tested if etsrp was sufficient for vascular induction in clo−/− mutant embryos. etsrp mRNA was injected into the progeny from clo+/− carriers, and the embryos were analyzed for flk1 expression at the ten- to 12-somite stages. As expected, 25% (23 out of 92) of the uninjected progeny from clo+/− carriers showed no flk1 expression. Among etsrp mRNA injected embryos (n = 110), 29% displayed normal flk1 expression pattern (with minor distortions in some embryos), 45% showed both endogenous and ectopic flk1 expression, 19% showed only ectopic flk1 expression, and 6% showed no detectable flk1 expression (Figure 6I–6L). The last two groups apparently represent clo−/− homozygous mutants. These results show that the induction of flk1 by etsrp is independent of clo function, suggesting that etsrp functions downstream of clo. etsrp mRNA did not fully restore the endogenous pattern of flk1 expression in clo−/− mutants most likely because it was expressed ubiquitously and not localized to vascular progenitors.
Etsrp Function Is Required for flk1 Induction by Vegf and Scl
We analyzed the epistatic relationship between vegf and etsrp. vegf overexpression has been reported to induce strong expression of vascular markers such as flk1 [43] (Figure 7A and 7B). Co-injection of vegf mRNA together with etsrp MO resulted in the loss of flk1 expression, similar to etsrp MO phenotype, indicating that etsrp function is required for flk1 induction by vegf signaling (Figure 7C and 7D). Downregulation of vegf expression resulted in the loss of etsrp expression in the intersegmental vessels (Figure 7E and 7F). etsrp expression in dorsal aorta was also not apparent while cardinal vein seemed expanded. This is consistent with the previous report of vegf involvement in regulating arterial fate in zebrafish [44]. Expression of etsrp was not affected in vegf morphants during mid-somitogenesis stages (unpublished data).
Figure 7 etsrp Is Required for vegf and scl Signaling
(A–D) Etsrp is required for Vegf signaling as assayed for flk1 expression at 26 hpf. (A) Control uninjected embryo, (B) vegf RNA-injected embryo, (C) 7.5 ng of etsrp MO2-injected embryo, (D) vegf RNA- and etsrp MO2-co-injected embryo. Note that vegf RNA induces strong flk1 expression in (B) while vegf RNA and etsrp MO co-injection results in loss of flk1 expression in (D), similar to the etsrp morphant phenotype in (C). (E,F) Etsrp expression analysis in Vegf morphants at 26 hpf. (E) Control uninjected embryo; (F) 10.5 ng of vegf MO-injected embryo. Note that vegf morphants have lost etsrp expression in the intersegmental vessels (arrowhead, E). (G–J) Scl knockdown affects gata1 but not etsrp expression at the 15-somite stage. Dorsal view, anterior is to the left. (G,I) Control uninjected embryo; (H,J) 10 ng scl UTR-MO-injected embryo. (G,H) gata1 expression; (I,J) etsrp expression. (K–N) Etsrp is required for scl signaling in clo mutants as analyzed for flk1 expression at the 15-somite stage. (K) Control uninjected embryo; (L) 7.5 ng etsrp MO2-injected embryo; (M) scl RNA-injected embryo; (N) scl RNA- and etsrp MO2-co-injected embryo. Note that scl RNA causes ectopic flk1 expression in (M) which is lost upon knockdown of Etsrp in (N).
Scl has been implicated in the hematopoietic and vascular development in zebrafish [13,14]. Downregulation of Scl function using two different MOs had no significant effect on the etsrp expression in angioblasts while the hematopoietic expression of gata1 was completely eliminated (Figure 7G–7J). Overexpression of scl mRNA has been reported to cause induction of vascular markers [45]. We tested if etsrp MO would block this induction by co-injecting it together with scl mRNA. The embryos were analyzed for flk1 expression at the 15-somite stage (Figure 7K–7N). As expected, no or only extremely weak flk1 expression was observed in etsrp–MO-injected embryos (Figure 7L). Overexpression of scl mRNA caused strong induction of flk1 (Figure 7M). Embryos, co-injected with scl mRNA and etsrp MO showed no or very weak flk1 expression (Figure 7N). These results demonstrate that etsrp function is required for flk1 induction by scl.
Discussion
In this study, we report characterization and functional analysis of a novel zebrafish Ets1-related protein, Etsrp. Knockdown of Etsrp resulted in the complete absence of functional blood vessels and downregulation of all endothelial-specific markers analyzed. In zebrafish, hedgehog signaling is necessary for migration of angioblasts from the lateral mesoderm toward the midline, where they subsequently differentiate within the ICM region [4]. Such migration is not apparent in the anterior region which gives rise to the anterior vessels including head vessels. Interestingly, vasculogenesis defects in etsrp morphants were more pronounced in the posterior region of the embryo which was particularly evident at lower downregulation levels. At high MO doses, endothelial cells were nearly completely absent from axial and intersegmental vessels. However, head expression of endothelial markers was not completely eliminated. It is possible that the remaining amount of Etsrp protein activity is sufficient for the head vasculogenesis. Upregulation of etsrp RNA was particularly strong in the head region in etsrp morphants, which may partially compensate for the MO-mediated translation inhibition. Alternatively, etsrp function may be redundant in the vasculogenesis of head vessels.
Among the vascular-specific genes analyzed, expression of only fli1 and scl was not globally downregulated in the early embryos. The early fli1 expression, which overlaps with the hematopoietic factor gata2 expression, is unaffected in the clo mutants and scl morphants and has been suggested to mark the common blood and vascular precursors, hemangioblasts [13,32]. fli1 is later expressed specifically in the vascular endothelial cell precursors, and this expression was reduced but not completely eliminated in etsrp morphants. fli1 expression may not be directly regulated by etsrp, therefore undifferentiated angioblasts retain some of fli1 expression in etsrp morphants. scl is expressed in both hematopoietic and vascular progenitors throughout most of the early development. Although we did not see global downregulation of scl expression in etsrp morphants, it is likely that scl expression is lost in angioblasts but retained in hematopoietic precursors. Interestingly, expression of both fli1 and scl genes was lost from the anterior and trunk domains of the lateral mesoderm. Anterior domain contains angioblast and myeloid progenitors, which appear to be absent from the etsrp morphants. The difference between the trunk and posterior lateral mesoderm has not been previously analyzed in great detail. Possibly, the trunk region contains mostly angioblasts, while the tail region contains both angioblasts and hematopoietic precursors.
Overexpression of etsrp induced strong ectopic expression of vascular markers in multiple cell types, including even tissues that normally commit to a very different fate such as neuroectoderm. Furthermore, overexpression of etsrp specifically initiated vascular development without affecting related hematopoiesis, which argues that the observed effects are not caused by an early ventralization of the whole embryo. These results indicate that etsrp is sufficient to initiate vasculogenesis. As a contrast, overexpression of other regulators of vasculogenesis such as vegf and scl induced expression of vascular markers only within the lateral or somitic mesoderm [14,43,45].
Overexpression of vegf and scl did not have an effect on vascular development in the absence of etsrp function, which suggests that etsrp is an essential mediator of vegf and scl signaling, at least in the vascular induction. Loss of vegf or scl did not affect early expression of etsrp within angioblasts. Overexpression of etsrp caused strong scl induction, indicating that etsrp plays an important role in controlling scl expression, at least within angioblasts. This is also supported by the loss of scl expression in the anterior and trunk regions in etsrp morphants. Our data suggest that etsrp is necessary for scl expression within angioblasts, and both genes are then required for induction of multiple vascular endothelial genes.
The current study shows for the first time that a single gene can be necessary and sufficient for initiating vascular development in vertebrates. Our results demonstrate that Etsrp acts as a very early regulator of vasculogenesis. These findings will greatly advance our understanding of vascular development and the general mechanisms of cell fate specification.
Materials and Methods
Microinjection of MOs
Two etsrp-specific MOs (MO1,
TTGGTACATTTCCATATCTTAAAGT and MO2,
CACTGAGTCCTTATTTCACTATATC; Gene Tools, Inc.) were used to inhibit the function of Etsrp protein. For the dose response curve, 1–6 nl of 1 mg/ml MO solution in the Danieau buffer supplemented with 15 mM Tris-Cl (pH 7.5) was injected into one- to two-cell-stage embryos [36]. For phenotypic and marker analysis, 3–4nl of 2.5 mg/ml MO2 solution was injected. A five-base MO2 mismatch (
CAGTGAGACCTTAATTCAGTATAAC) was used as a control MO. A previously described Vegf-A-1 MO (kindly donated by S.C. Ekker) was used to inhibit Vegf function [46]. Scl translation-blocking UTR-MO (
GCTCGGATTTCAGTTTTTCCATCAT) and previously described splice-blocking MO [13] were used to inhibit the function of Scl protein. Microinjections were performed as described [47].
BAC modification and microinjection
CHORI-211 BAC library from Children's Hospital Oakland Research Institute, Oakland, California, United States was used to identify etsrp-containing–BAC of approximately 250 kilobases. Replacement of Etsrp coding sequence with GFP was essentially performed as described [38]. Purified BAC DNA (100–150 pg) was injected into blastomere at the one-cell stage.
RNA overexpression and epistasis experiments
Etsrp overexpression construct was generated by subcloning the open reading frame of etsrp into the SpeI site of pT3TS [47]. To synthesize mRNA, Etsrp-T3TS was linearized with XbaI and transcribed using T3 mMessage mMachine Kit (Ambion, Austin, Texas, United States). At the two- to 16-cell stage for overexpression studies, 75–15 pg of etsrp mRNA was injected into zebrafish embryos. For epistasis studies, approximately 50 pg of VEGF121 and VEGF165 mRNA 1:1 mixture was injected [43]. Approximately 300 pg of scl mRNA was injected for epistasis studies [12].
In situ hybridization
In situ hybridization was performed as described [48]. The following probes were used: etsrp [42], flk1 [40], fli1 [40], flt4 [40], scl [12], gata1 [49], admr [42], cdh5 [42], crl [42].
Analysis of etsrp knockdown phenotype
o-dianisidine heme staining was performed as described [49]. Microangiography was performed as described [46].
Zebrafish strains
Majority of etsrp knockdown analysis was performed in the wild-type zebrafish from Scientific Hatcheries (Huntington Beach, California, United States). As confirmed by sequencing, no polymorphisms in the etsrp MO-binding sites were detected in this wild-type strain. In addition, flk1-GFP transgenic zebrafish line was used [39]. clom39 line was used in etsrp overexpression experiments [7].
Supporting Information
Figure S1 Etsrp MO Blocks GFP Expression in the Embryos Injected with etsrp–GFP-Modified BAC Construct
Mosaic panels have been generated by taking individual images of randomly chosen embryos from each batch. Embryos are at the 90% epiboly stage. (A) Following microinjection, etsrp-GFP BAC is transiently expressed in a mosaic pattern. Note that the transient BAC expression commonly does not recapitulate the endogenous expression pattern. (B) Co-injection of etsrp-GFP BAC and 10 ng of etsrp MO2 completely eliminated GFP fluorescence. (C) Co-injection of etsrp-GFP BAC and 10 ng of 5-base MO2 mismatch had no effect on GFP fluorescence.
(2.5 MB TIF).
Click here for additional data file.
Accession Numbers
GenBank (http://www.ncbi.nlm.nih.gov/Genbank) accession number for etsrp is DQ021472.
Research was supported by grants from the National Institutes of Health (R01 DK54508 to SL and T32 HL069766 to SS). We thank S. Palencia-Desai and H. Jiang for their help with experiments.
Competing interests. The authors have declared that no competing interests exist.
Author contributions. SL and SS conceived and designed the experiments. SS performed the experiments. SL and SS analyzed the data. SS wrote the paper.
Citation: Sumanas S, Lin S (2006) Ets1-related protein is a key regulator of vasculogenesis in zebrafish. PLoS Biol 4(1): e10.
Abbreviations
BACbacterial artificial chromosome
dpfdays-post-fertilization
GFPgreen fluorescent protein
hpfhours-post-fertilization
ICMintermediate cell mass
MOmorpholino
Vegfvascular endothelial growth factor
wtwild-type
==== Refs
References
Risau W Flamme I Vasculogenesis Annu Rev Cell Dev Biol 1995 11 73 91 8689573
Zhong TP Childs S Leu JP Fishman MC Gridlock signaling pathway fashions the first embryonic artery Nature 2001 414 216 220 11700560
Fouquet B Weinstein BM Serluca FC Fishman MC Vessel patterning in the embryo of the zebrafish: Guidance by notochord Dev Biol 1997 183 37 48 9119113
Gering M Patient R Hedgehog signaling is required for adult blood stem cell formation in zebrafish embryos Dev Cell 2005 8 389 400 15737934
Isogai S Lawson ND Torrealday S Horiguchi M Weinstein BM Angiogenic network formation in the developing vertebrate trunk Development 2003 130 5281 5290 12954720
Robb L Elefanty AG The hemangioblast—an elusive cell captured in culture Bioessays 1998 20 611 614 9780835
Stainier DY Weinstein BM Detrich HW Zon LI Fishman MC Cloche, an early acting zebrafish gene, is required by both the endothelial and hematopoietic lineages Development 1995 121 3141 3150 7588049
Parker L Stainier DY Cell-autonomous and non-autonomous requirements for the zebrafish gene cloche in hematopoiesis Development 1999 126 2643 2651 10331976
Begley CG Aplan PD Denning SM Haynes BF Waldmann TA The gene SCL is expressed during early hematopoiesis and encodes a differentiation-related DNA-binding motif Proc Natl Acad Sci U S A 1989 86 10128 10132 2602361
Kallianpur AR Jordan JE Brandt SJ The SCL/TAL-1 gene is expressed in progenitors of both the hematopoietic and vascular systems during embryogenesis Blood 1994 83 1200 1208 8118024
Gering M Rodaway AR Gottgens B Patient RK Green AR The SCL gene specifies haemangioblast development from early mesoderm Embo J 1998 17 4029 4045 9670018
Liao EC Paw BH Oates AC Pratt SJ Postlethwait JH SCL/Tal-1 transcription factor acts downstream of cloche to specify hematopoietic and vascular progenitors in zebrafish Genes Dev 1998 12 621 626 9499398
Patterson LJ Gering M Patient R Scl is required for dorsal aorta as well as blood formation in zebrafish embryos Blood 2005 105 3502 3511 15644413
Dooley KA Davidson AJ Zon LI Zebrafish scl functions independently in hematopoietic and endothelial development Dev Biol 2005 277 522 536 15617691
Lelievre E Lionneton F Soncin F Vandenbunder B The Ets family contains transcriptional activators and repressors involved in angiogenesis Int J Biochem Cell Biol 2001 33 391 407 11312108
Vandenbunder B Pardanaud L Jaffredo T Mirabel MA Stehelin D Complementary patterns of expression of c-ets 1, c-myb and c-myc in the blood-forming system of the chick embryo Development 1989 107 265 274 2483681
Pardanaud L Dieterlen-Lievre F Expression of C-ETS1 in early chick embryo mesoderm: relationship to the hemangioblastic lineage Cell Adhes Commun 1993 1 151 160 8081877
Queva C Leprince D Stehelin D Vandenbunder B p54c-ets-1 and p68c-ets-1, the two transcription factors encoded by the c-ets-1 locus, are differentially expressed during the development of the chick embryo Oncogene 1993 8 2511 2520 8361762
Anderson MK Hernandez-Hoyos G Diamond RA Rothenberg EV Precise developmental regulation of Ets family transcription factors during specification and commitment to the T cell lineage Development 1999 126 3131 3148 10375504
Tanaka K Oda N Iwasaka C Abe M Sato Y Induction of Ets-1 in endothelial cells during reendothelialization after denuding injury J Cell Physiol 1998 176 235 244 9648911
Wernert N Raes MB Lassalle P Dehouck MP Gosselin B c-ets1 proto-oncogene is a transcription factor expressed in endothelial cells during tumor vascularization and other forms of angiogenesis in humans Am J Pathol 1992 140 119 127 1370594
Wakiya K Begue A Stehelin D Shibuya M A cAMP response element and an Ets motif are involved in the transcriptional regulation of flt-1 tyrosine kinase (vascular endothelial growth factor receptor 1) gene J Biol Chem 1996 271 30823 30828 8940064
Kappel A Ronicke V Damert A Flamme I Risau W Identification of vascular endothelial growth factor (VEGF) receptor-2 (Flk-1) promoter/enhancer sequences sufficient for angioblast and endothelial cell-specific transcription in transgenic mice Blood 1999 93 4284 4292 10361126
Lelievre E Mattot V Huber P Vandenbunder B Soncin F ETS1 lowers capillary endothelial cell density at confluence and induces the expression of VE-cadherin Oncogene 2000 19 2438 2446 10828886
Iljin K Dube A Kontusaari S Korhonen J Lahtinen I Role of ets factors in the activity and endothelial cell specificity of the mouse Tie gene promoter Faseb J 1999 13 377 386 9973326
Chen Z Fisher RJ Riggs CW Rhim JS Lautenberger JA Inhibition of vascular endothelial growth factor-induced endothelial cell migration by ETS1 antisense oligonucleotides Cancer Res 1997 57 2013 2019 9157999
Wernert N Stanjek A Kiriakidis S Hugel A Jha HC Inhibition of angiogenesis in vivo by ets-1 antisense oligonucleotides-inhibition of Ets-1 transcription factor expression by the antibiotic fumagillin Angew Chem Int Ed Engl 1999 38 3228 3231 10556911
Barton K Muthusamy N Fischer C Ting CN Walunas TL The Ets-1 transcription factor is required for the development of natural killer cells in mice Immunity 1998 9 555 563 9806641
Wang D John SA Clements JL Percy DH Barton KP Ets-1 deficiency leads to altered B cell differentiation, hyper-responsiveness to TLR9 and autoimmune disease Int Immunol 2005 17 1179 1191 16051621
Hart A Melet F Grossfeld P Chien K Jones C Fli-1 is required for murine vascular and megakaryocytic development and is hemizygously deleted in patients with thrombocytopenia Immunity 2000 13 167 177 10981960
Spyropoulos DD Pharr PN Lavenburg KR Jackers P Papas TS Hemorrhage, impaired hematopoiesis, and lethality in mouse embryos carrying a targeted disruption of the Fli1 transcription factor Mol Cell Biol 2000 20 5643 5652 10891501
Brown LA Rodaway AR Schilling TF Jowett T Ingham PW Insights into early vasculogenesis revealed by expression of the ETS-domain transcription factor Fli-1 in wild-type and mutant zebrafish embryos Mech Dev 2000 90 237 252 10640707
Sumanas S Jorniak T Lin S Identification of novel vascular endothelial-specific genes by the microarray analysis of the zebrafish cloche mutants Blood 2005 106 534 541 15802528
Zhu H Traver D Davidson AJ Dibiase A Thisse C Regulation of the lmo2 promoter during hematopoietic and vascular development in zebrafish Dev Biol 2005 281 256 269 15893977
Hoegg S Brinkmann H Taylor JS Meyer A Phylogenetic timing of the fish-specific genome duplication correlates with the diversification of teleost fish J Mol Evol 2004 59 190 203 15486693
Nasevicius A Ekker SC Effective targeted gene knockdown in zebrafish Nat Genet 2000 26 216 220 11017081
Sumanas S Larson JD Morpholino phosphorodiamidate oligonucleotides in zebrafish: A recipe for functional genomics? Brief Funct Genomic Proteomic 2002 1 239 256 15239891
Gong S Yang XW Li C Heintz N Highly efficient modification of bacterial artificial chromosomes (BACs) using novel shuttle vectors containing the R6Kgamma origin of replication Genome Res 2002 12 1992 1998 12466304
Cross LM Cook MA Lin S Chen JN Rubinstein AL Rapid analysis of angiogenesis drugs in a live fluorescent zebrafish assay Arterioscler Thromb Vasc Biol 2003 23 911 912 12740225
Thompson MA Ransom DG Pratt SJ MacLennan H Kieran MW The cloche and spadetail genes differentially affect hematopoiesis and vasculogenesis Dev Biol 1998 197 248 269 9630750
Larson JD Wadman SA Chen E Kerley L Clark KJ Expression of VE-cadherin in zebrafish embryos: A new tool to evaluate vascular development Dev Dyn 2004 231 204 213 15305301
Sumanas S Jorniak T Lin S Identification of novel vascular endothelial-specific genes by the microarray analysis of the zebrafish cloche mutants Blood 2005 106 534 541 15802528
Liang D Chang JR Chin AJ Smith A Kelly C The role of vascular endothelial growth factor (VEGF) in vasculogenesis, angiogenesis, and hematopoiesis in zebrafish development Mech Dev 2001 108 29 43 11578859
Lawson ND Vogel AM Weinstein BM Sonic hedgehog and vascular endothelial growth factor act upstream of the Notch pathway during arterial endothelial differentiation Dev Cell 2002 3 127 136 12110173
Gering M Yamada Y Rabbitts TH Patient RK Lmo2 and Scl/Tal1 convert non-axial mesoderm into haemangioblasts which differentiate into endothelial cells in the absence of Gata1 Development 2003 130 6187 6199 14602685
Nasevicius A Larson J Ekker SC Distinct requirements for zebrafish angiogenesis revealed by a VEGF-A morphant Yeast 2000 17 294 301 11119306
Hyatt TM Ekker SC Vectors and techniques for ectopic gene expression in zebrafish Methods Cell Biol 1999 59 117 126 9891358
Jowett T Analysis of protein and gene expression Methods Cell Biol 1999 59 63 85 9891356
Detrich HW Kieran MW Chan FY Barone LM Yee K Intraembryonic hematopoietic cell migration during vertebrate development Proc Natl Acad Sci U S A 1995 92 10713 10717 7479870
|
16336046
|
PMC1310653
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e10
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040010
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040012SynopsisBotanyPlant SciencePlantsDiverse Pollination Networks Key to Ecosystem Sustainability Synopsis1 2006 13 12 2005 13 12 2005 4 1 e12Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Functional Diversity of Plant-Pollinator Interaction Webs Enhances the Persistence of Plant Communities
==== Body
As animal extinctions continue at the rate of one every 16 years, it's unclear how declining biodiversity will disturb ecosystem dynamics. Of special concern are the pollinators, essential players in the reproductive biology of plants, the earth's primary producers. Millions of years of evolutionary coadaptations lie behind the perfect pairing of pollinator proboscis anatomy with plant flower structure, as well as the mechanisms plants use to attract reproductive assistants to their food rewards. Agave plants emit musky aromas that attract lesser long-nosed bats to nectar stores within their flowers, for example. As the bats travel from flower to flower, pollen collects and then falls from their fur, facilitating cross-pollination.
These mutually beneficial relationships are sometimes so specialized that the loss of one species threatens the existence of the other, raising troubling questions about the likely consequences of declining diversity in pollination networks. In a new study, Colin Fontaine et al. tackled this question by experimentally manipulating plant and pollinator interactions under natural conditions. The authors found strong functional relationships between different pollinators and plant communities, with the highest plant community sustainability associated with the most diverse group of pollinators. These findings suggest that loss of biodiversity in pollination networks may threaten the persistence of plant communities.
For their study, the authors chose plants with easy and harder access to food rewards—three open-flower and three tubular-flower species—and insects with short and longer mouthparts—three syrphid fly and three bumblebee species. In the spring of 2003, Fontaine et al. set up 36 plant communities in nylon-mesh enclosures in a meadow 80 kilometers (about 50 miles) southwest of Paris, after sterilizing the soil to destroy seeds and pathogens. They planted 30 adult plants in each plot at the same density, and then captured and released local pollinators into the cages during the flowering season (June–July 2003, 2004). To test all the possible plant–pollinator combinations, the authors set up three plant treatments (open flowers, tubular flowers, and both flower types), then applied three pollination treatments (flies, bees, and both insects) to each plant treatment.
A month after the first pollination treatments, the authors tallied all the fruit on each plant, then randomly selected five fruits per plant (excepting one plant species from each group that would have required harvesting the fruit) to estimate seed production per plant. During the seedling season, they totaled the plants and the seedlings to measure plant population and reproductive success. Pollinator identity determined fruit production, with bee-pollinated plants most productive, and the different plant groups responded differently to the two pollinator groups. As expected, short-mouthed syrphid flies pollinated only open flowers, while bees pollinated both plant types. As a result, tubular flowers produced far fewer fruits with syrphid pollinators while open-flower fruit production remained the same regardless of pollinator. Fruit production increased along with both plant and pollinator diversity. Seed production was a bit more complicated. Though bee-pollinated open flowers produced fewer seeds per plant than those pollinated by syrphids, higher fruit production compensated by producing more seedlings. Fruit production increased with pollinator diversity.
Loss of biodiversity in pollination networks may threaten the persistence of plant communities
As for long-term effects on plant reproductive capacity and success, tubular plant communities had fewer plants at the seedling stage than openflowered plants (and even fewer when pollinated by syrphids). The plant species number and total plant number increased when both pollinator groups were present, and were highest with maximum plant and pollinator diversity. Seedling production showed a similar pattern: mixed plant communities treated with both pollinators yielded the most seeds.
What happened? Not surprisingly, the pollinators stuck to their preferred plant: syrphids visited mostly open flowers, and bees visited mostly tubular flowers. Bees can pollinate open flowers but prefer tubular flowers when they have the choice, suggesting that bees may not fill a void left by a different pollinator. The presence of both pollinators allowed more appropriate pairings between insects and flowers—each performing a complementary role—leading to increased pollination efficiency and plant reproductive success.
While the study offers an admittedly pared down view of pollination networks, it demonstrates the value of studying the functional effects of pollination networks in the field. These results show that losing a species affects plant–pollinator communities, and that such losses may ultimately trigger further reductions in biodiversity, possibly reverberating through the food chain. With as many as 70% of plant species dependent on animal pollinators and at least 82 mammalian pollinator species and 103 bird pollinator species considered threatened or extinct, this is sobering news. —Liza Gross
|
0
|
PMC1310654
|
CC BY
|
2021-01-05 08:21:47
|
no
|
PLoS Biol. 2006 Jan 13; 4(1):e12
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040012
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040015SynopsisBotanyEcologyInfectious DiseasesMicrobiologyPlant ScienceVirusesHomo (Human)PlantsThriving Community of Pathogenic Plant Viruses Found in the Human Gut Synopsis1 2006 20 12 2005 20 12 2005 4 1 e15Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
RNA Viral Community in Human Feces: Prevalence of Plant Pathogenic Viruses
==== Body
What some people won't do for science. In the early 1980s, Barry Marshall was convinced that Helicobacter pylori, not stress, caused stomach ulcers and inflammation. To convince the skeptics, he drank a straight broth of H. pylori, and promptly developed severe gastric distress. In October, he won the Nobel Prize for Physiology or Medicine (along with Robin Warren) for his efforts.
Since Marshall's experiment, it's become clear that a wide variety of bacteria, viruses, and parasites cause gastroenteritis, and that many other uncharacterized pathogens likely cause stomach flu as well. With some 100 trillion microorganisms calling the human gut home, pathogens typically represent a slim (though often raucous) minority. Bacteria, by far the most abundant resident, mostly aid digestion. Viruses in the gut come in good (bacteriophages can aid digestion and control invasive pathogens) and bad (RNA viruses) varieties. Many viral gastroenteritis pathogens were identified by analyzing patients' fecal samples, but the viruses wouldn't grow in test tubes, forcing scientists to look for other ways to generate enough viruses for study. Amazingly, they found volunteers to ingest infected stool filtrates.
Thankfully, the tools of metagenomics offer a better way. In a new study, Tao Zhang, Yijun Ruan, and their colleagues used biomolecule filtration methods and genomics to capture and characterize the global community of RNA viruses living in the human gut. The authors set out looking for pathogenic enteric viruses, but found very few human viruses. Instead, they found a “large and diverse community” of plant RNA viruses. The most abundant of these viruses infects vegetable crops, a finding that suggests that humans likely contract the virus from food—and, most surprisingly, that humans may transmit the virus to plants.
The human gut appears to harbor infectious strains of the pepper mild mottle virus (seen infecting peppers, above), suggesting that humans may serve as vector for certain plant viruses. Image: UF/IFAS Pest Alert Web site/Pamela Roberts
For their study, Zhang et al. collected three fecal samples from two healthy adults in Southern California (one volunteer supplied a second sample six months later), used filtration methods to isolate the viral particles, and then constructed three libraries (one corresponding to each sample) of nearly 37,000 sequences for analysis. Of more than 33,000 known sequences, about 75% resembled viruses, sorted into 42 different species. (The rest resembled sequences from Bacteria, Archaea, and Eukarya domains.) Only about 3% of the viral-like sequences resembled animal viruses, while 97% resembled plant viruses. Twenty-four out of 35 plant viruses identified infect commercial crops, including fruit, vegetables, and tobacco.
The most abundant plant virus, called pepper mild mottle virus (PMMV), infects a wide variety of sweet and hot peppers. When the authors compared the genetic sequences of the various PMMV strains, and then compared the sequence variation of a specific PMMV gene from all the strains, they found significant divergence at both the species and gene levels. This variability, the authors explain, suggests that PMMV exists as a diverse, dynamic population in the human gut.
So how did PMMV enter the gut? Zhang et al. enlisted three more volunteers, and tested all the food they ate two days prior to providing a fecal sample. PMMV was found in their food and at exponentially higher levels in their feces. A random test of pepper-based foods in Southern California found evidence of PMMV in three of 22 samples (none of the healthy-looking peppers tested positive). To control for local effects, the authors expanded their study to Singapore, and found PMMV in human volunteers as well as in some of the pepper-based offerings collected from food stalls. Humans not only carry this plant pathogen, we help transmit it: when Zhang et al. inoculated Hungarian wax peppers with human-borne PMMV, every one of the plants developed a PMMV infection. Since humans can transmit the virus, other animals probably can, too—suggesting that farmers with chronically infected crops might want to test any animal manure they use.
Though it's unclear how the plant viruses manage to persist inside the human gut—do they co-opt intestinal cells or use bacteria to reproduce?—the stability of PMMV suggests that it might serve as a target to help deliver vaccines or treatments aimed at intestinal disorders. What about the enteric viruses Zhang et al. originally sought? Human viruses may escape detection in healthy individuals, the authors explain, pointing to the next logical step: soliciting donations from patients with symptomatic gastroenteritis. Using the same global approach outlined here, the authors may well identify a host of other viral causes of gastric disorders. —Liza Gross
|
0
|
PMC1310655
|
CC BY
|
2021-01-05 08:28:16
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e15
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040015
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040017SynopsisDevelopmentDanio (zebrafish)Teleost FishesVertebratesAnimalsEukaryotesA Master Switch? estrp Directs Blood Vessel Development in Zebrafish Synopsis1 2006 20 12 2005 20 12 2005 4 1 e17Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Ets1-Related Protein Is a Key Regulator of Vasculogenesis in Zebrafish
==== Body
As the primary distributor of nutrients and oxygen to cells throughout the vertebrate body, the circulatory system provides life support to the body's tissues and organs. Consequently, the cellular and developmental signals that control development of this critical organ system, composed of the heart, the blood vessels, and the blood cells they contain, are the subject of intense study. Medical researchers also find it valuable to know the signals that control blood vessel development because solid tumor masses depend on infiltrating blood vessels to grow.
Circulatory system development proceeds similarly in all vertebrates, with the heart and aorta beginning to form early on (about the point at which recognizable structures like the head begin to form). Specialized cells called angioblasts create blood vessels that bud off from the aorta and branch out to form the rest of the animal's vasculature. With its transparent embryos that develop outside the mother, the zebrafish is a favored model organism for circulatory system development.
Zebrafish sporting a mutation called cloche (a French word referencing the mutant animal's bell-shaped heart) lack both blood vessels and blood cells. This discovery produced an important insight: the loss of two different cell types with one mutation suggests that both types of cells may arise from a common precursor at some point in early development. But until now, no single protein has been identified that differentiates the hemocytes that are destined to become blood cells from the angioblasts destined to become blood vessels. In a new paper, Saulius Sumanas and Shuo Lin describe a gene called etsrp that is specifically required for blood vessel development.
The authors had originally discovered etsrp in a screen they conducted to find genes whose expression is altered by the cloche mutation. To learn what exactly etsrp might be doing during development, the authors first looked for clues in its sequence by comparing it to the sequence of similar known proteins, the Ets family of transcription factors. The chromosomal location of etsrp suggested it had arisen as a result of a genetic duplication of the founding member of this family, the ets1 gene (thus its name, which stands for “Ets1-related protein”). It's been known for some time that Ets family transcription factors play many roles in the development of the circulatory system by binding DNA and controlling the expression of genes critical to the development of circulatory system cell lineages, suggesting that etsrp may also be involved.
With this information in hand, the authors set out to investigate where and when etsrp is expressed during zebrafish development by looking for the presence of etsrp mRNA in developing embryos. They found etsrp mRNA expressed early in development in tissues that eventually give rise to blood vessels; later on, they saw etsrp mRNA expressed in more mature blood vessel structures throughout the animal. These findings suggested that etsrp might be involved in the designation of these structures during development. To see whether etsrp is required for blood vessel development, Sumanas and Lin disrupted the expression of etsrp in developing embryos. Interestingly, loss of etsrp resulted in the absence of blood vessels in the developing animal, even though the animals were able to make blood cells normally; the embryos' blood cells remained clumped at their formation site next to the yolk extension, rather than entering circulation. It appeared there just weren't any blood vessels around for the blood cells to move through. In support of this conclusion, Sumanas and Lin could not detect evidence that any body cells expressed the surface proteins normally associated with blood vessel identity. Therefore, expression of these markers appears to be dependent on the presence of etsrp.
The transparent zebrafish embryo helped researchers identify a putative master gene controlling blood vessel development
If the expression of blood vessel–specific markers is dependent on etsrp, is it also true that etsrp is sufficient for expression of these markers? The authors found that artificially causing the overexpression of etsrp in early embryos resulted in the inappropriate expression of blood vessel–associated proteins in cells that normally do not express them. Finally, since cloche mutants also lack expression of blood vessel–specific markers (in addition to lacking blood cell–specific markers), the authors wondered whether artificially expressing etsrp in cloche mutants could restore the expression of markers associated with the development of blood vessels. Indeed, they found that etsrp could restore the expression of the blood vessel–specific marker, flk1, in cloche mutant embryos. Taken together, these data indicate that etsrp is both necessary and sufficient for blood vessel development in the zebrafish, lending important insights into our understanding of circulatory system development. —Caitlin Sedwick
|
0
|
PMC1310656
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e17
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040017
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040018SynopsisCell BiologyDrosophilaInsectsFly Genes That Help Devour a Fungal Parasite Synopsis1 2006 20 12 2005 20 12 2005 4 1 e18Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Identification of Drosophila Gene Products Required for Phagocytosis of Candida albicans
==== Body
Faced with a world full of potentially harmful bacteria, fungi, and viruses, multicellular organisms have evolved efficient innate mechanisms to combat infection soon after a pathogenic exposure. Many aspects of this innate response are shared by organisms as diverse as flies and humans, pointing to a common and ancient origin for these innate defenses. While humans and other vertebrates possess adaptive immunity—which can recognize billions of different pathogens and mount better responses with each exposure to a pathogen—flies and other invertebrates must rely only on innate immunity.
As early as 1882, Élie Metchnikoff discovered that cells in starfish and water fleas ingested and destroyed microbes as part of the immune response. (The Russian zoologist and microbiologist would receive the Nobel Prize for his discovery.) Only recently have scientists realized how many different organisms use this strategy, called phagocytosis, to protect themselves. When a pathogen is detected, signaling pathways within the phagocytic cell reconfigure the cytoskeleton and send out pseudopodia protrusions that engulf the pathogen. While these morphological and structural changes have been well described, few studies have outlined the cellular components driving this process. In a new study by Shannon Stroschein-Stevenson et al., the labs of Patrick O'Farrell and Alexander Johnson turned to the fruit fly, Drosophila melanogaster—a well-established model organism for both genetics and innate immunity—to explore the phagocytosis of the fungus Candida albicans, a widespread human pathogen.
Drosophila cells engulf the human pathogen Candia albicans, tagged with green fluorescent protein
Using the S2 cell line, which is derived from flies and shares many properties with the plasmatocytes that perform phagocytosis in the fly, the authors used RNA interference (RNAi) to conduct a global search for genes related to phagocytosis. They identified several genes specifically dedicated to dispatching C. albicans, then focused on one gene whose function in the fly was unknown. The gene, called Macroglobulin complement related (Mcr), is closely related to human proteins that activate what's known as the complement cascade, an ancient mechanism that flags pathogens for subsequent recognition by phagocytic cells. The authors show that Mcr is closely related to four fly proteins (members of the thioester protein [Tep] family), and that these proteins act on different pathogens, functioning as part of a “primitive complement system” that targets specific pathogens.
Stroschein-Stevenson et al. first mixed fluorescently tagged C. albicans with the S2 cells and observed the S2 cells ingest the fungal cells, confirming that the cell line can phagocytose the fungus. Next, they screened the cells for phagocytosis defects using RNAi; RNAi disrupts the function of a target gene by using short lengths of double-stranded RNA (dsRNA) with complementary sequences to that gene to force the degradation of the gene's messenger RNA and prevent its translation into protein. The authors used a library of over 7,000 dsRNAs corresponding to most of the fly's conserved genes to screen for phagocytosis gene candidates. After treating the S2 cells with dsRNAs, they mixed the cells with the fluorescently tagged fungus to visualize the effects on phagocytosis.
The initial screen flagged some 400 dsRNAs that decreased fungal phagocytosis. A few more screens excluded dsRNAs that were likely either false positives or indirectly involved, leaving 184 dsRNAs that impaired phagocytosis. The screen identified many known phagocytosis-related genes that had been picked up in earlier screens (for example, the cytoskeletal protein actin and its various regulators, which help form the active membrane protrusions required for phagocytosis), but it also turned up many other genes not previously implicated in the process.
To distinguish genes specific to C. albicans from those with a broader role in phagocytosis, the authors repeated the screen with two different “challengers,” Escherichia coli and latex beads. Most of the 184 dsRNAs impaired phagocytosis of all three targets, but only a few specifically disrupted C. albicans phagocytosis, including the dsRNA for Mcr. Additional experiments confirmed Mcr's specificity for C. albicans. Since Mcr is closely related to a family of four fly Tep proteins, the authors tested whether each protein is required for the phagocytosis of three distinct pathogens: C. albicans, E. coli, and Staphylococcus aureus. Again, impaired phagocytosis of C. albicans occurred only when Mcr was disabled. (Adding Mcr proteins to the S2 growing medium restored the cells' ability to phagocytose C. albicans.) E. coli phagocytosis was reduced only with TepII disruption, and S. aureus phagocytosis slowed only with TepIII silencing.
The authors go on to show that S2 cells secrete Mcr, which then binds to C. albicans (but not to a closely related fungus) and promotes phagocytosis. Altogether, these results show that the fly's innate immune system dispatches specialized proteins to destroy specific pathogens. The 184 genes identified in this screen should prove a valuable resource for investigations into the cellular components of an ancient defensive strategy. And with the visual screen described here, even researchers without the latest automated equipment can get started right away. —Liza Gross
|
0
|
PMC1310657
|
CC BY
|
2021-01-05 08:21:46
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e18
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040018
|
oa_comm
|
==== Front
PLoS BiolPLoS BiolpmedplosmedPLoS Biology1544-91731545-7885Public Library of Science San Francisco, USA 10.1371/journal.pbio.0040035SynopsisBioinformatics/Computational BiologyEvolutionMolecular Biology/Structural BiologyPlant SciencePlantsJumping Genes Cross Plant Species Boundaries Synopsis1 2006 20 12 2005 20 12 2005 4 1 e35Copyright: © 2005 Public Library of Science.2005This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are properly credited.
Horizontal Transfer of a Plant Transposon
==== Body
In the early 1950s, legendary plant geneticist Barbara McClintock found the first evidence that genetic material can jump from one place to another within the genome. The variegated kernels of her maize plants, she determined, resulted from mobile elements that had inserted themselves into pigment-coding genes, changing their expression. McClintock's mobile elements, or transposons, moved over generations within a single species. More recently, another form of genetic mobility has been discovered—genetic information can sometimes be transferred between species, a process called horizontal gene transfer. While horizontal genetic transfer occurs most commonly in bacteria, it has been detected in animals as well. Most transfers between higher animals involve the movement of transposons. Horizontal transfer can also occur between the mitochondrial DNA of different plant species. Until now, however, no one had found evidence for horizontal transfer in the nuclear DNA of plants.
In a new report, Xianmin Diao, Michael Freeling, and Damon Lisch studied the genomes of millet and rice, two distantly related grasses that diverged 30–60 million years ago. While the two grasses show significant genetic divergence from accumulating millions of years of mutations, they carry some transposon-related DNA segments that are surprisingly similar. The authors conclude that these sequences were transferred horizontally between the two plants long after they went their separate ways.
Transposons of the class identified by Diao et al. typically consist of a variable length of DNA that codes for one or more enzymes flanked by repeating sequences called terminal inverted repeats (TIRs). These repeats can bind to each other to form a “lollipop” that is easily excised from the DNA strand, carrying the rest of the transposon along with it. Plant genomes are rife with transposons, many of which are relatively passive. Transposons from the “Mutator” family in maize, however, are especially active, frequently causing mutations as they insert themselves into new positions in the genome. They perform this jump with assistance from the two proteins they code for, a transposase and a helper gene.
DNA from many species of plants contains several families of cousins of the Mutator transposons. These “Mutator-like elements,” or MULEs, code for a protein similar to the transposase, as well as the TIR sequences. Diao et al. identified 19 distinct MULEs in the DNA of various species of millet (genus Setaria), and compared these with the rice genome sequence, which was published in 2002. They compared the sequence similarity of these MULEs to that of other proteins that are also conserved in the same species for which sequences are available. Strikingly, they observed much higher sequence similarity between the MULEs from millet and rice than is typical for transposons. The greater similarity of the MULE DNA is easily explained if it jumped somehow, horizontally, between the species, but there could be alternative explanations. The match could have arisen without horizontal transfer, for example, if the MULE DNA had been under positive selection, as typically happens for protein-coding genes that confer some survival or reproductive benefit. In such cases, natural selection tends to preserve the integrity of these sequences.
A genome-wide analysis of millet (above) and rice revealed the first clear evidence of horizontal gene transfer in plants
To test for signs of selection, the researchers looked at regions of the MULE DNA that don't appear to code for protein. The similarity between these noncoding regions in millet and rice MULEs was just as high as for the coding regions, even though selection probably doesn't influence them. Even within the coding sections, “synonymous” mutations—which don't change the protein sequence and so are not prone to selection—showed few differences between these elements.
Another explanation for the low divergence of the rice and millet MULE sequences could be that they occur within a genomic region that, for whatever reason, experienced lower than average mutation rates. If this were the case, sequences adjacent to the elements should also show reduced variation. The authors tested this alternative hypothesis with the help of maize, which has more genomic sequence available than millet, by comparing genes flanking MULE regions in rice with evolutionarily conserved sequences in maize. The sequences did not show the similar degree of reduced variation predicted for below-average mutation rates.
Since neither selection nor low mutation frequency can explain the similar DNA between the grasses, the authors conclude, a transposon must have carried it between millet and rice long after these species diverged. Interestingly, the authors also found similar sequences in bamboo, raising the question of how common horizontal transfer may be between plant species. Given that plant mitochondrial genes appear “particularly prone to horizontal transfer,” the authors note, “it is remarkable that these results represent the first well-documented case of horizontal transfer of nuclear genes between plants.” But as researchers begin to explore the growing databases of plant genomic sequences, they can determine whether this finding constitutes an anomaly—or points to a significant force in plant genome evolution. —Don Monroe
|
0
|
PMC1310658
|
CC BY
|
2021-01-05 08:28:16
|
no
|
PLoS Biol. 2006 Jan 20; 4(1):e35
|
utf-8
|
PLoS Biol
| 2,005 |
10.1371/journal.pbio.0040035
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8037ehp0113-00147316263498Commentaries & ReviewsHypertension and Exposure to Noise near Airports (HYENA): Study Design and Noise Exposure Assessment Jarup Lars 1Dudley Marie-Louise 1Babisch Wolfgang 2Houthuijs Danny 3Swart Wim 3Pershagen Göran 4Bluhm Gösta 4Katsouyanni Klea 5Velonakis Manolis 6Cadum Ennio 7Vigna-Taglianti Federica 7 for the HYENA Consortium1 Department of Epidemiology and Public Health, Imperial College London, London, United Kingdom2 Department of Environmental Hygiene, Federal Environmental Agency, Berlin, Germany3 National Institute for Public Health and the Environment, Bilthoven, the Netherlands4 Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden5 Department of Hygiene and Epidemiology, Medical School6 Laboratory of Prevention, Nurses School, University of Athens, Athens, Greece7 Environmental Epidemiologic Unit, Regional Agency for Environmental Protection, Piedmont Region, Grugliasco, ItalyAddress correspondence to L. Jarup, Imperial College London, Department of Epidemiology and Public Health, Norfolk Pl., London, W2 1PG, UK. Telephone: 44-20-7594-3337. Fax: 44-20-7594-3196. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 13 7 2005 113 11 1473 1478 23 2 2005 13 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. An increasing number of people live near airports with considerable noise and air pollution. The Hypertension and Exposure to Noise near Airports (HYENA) project aims to assess the impact of airport-related noise exposure on blood pressure (BP) and cardiovascular disease using a cross-sectional study design. We selected 6,000 persons (45–70 years of age) who had lived at least 5 years near one of six major European airports. We used modeled aircraft noise contours, aiming to maximize exposure contrast. Automated BP instruments are used to reduce observer error. We designed a standardized questionnaire to collect data on annoyance, noise disturbance, and major confounders. Cortisol in saliva was collected in a subsample of the study population (n = 500) stratified by noise exposure level. To investigate short-term noise effects on BP and possible effects on nighttime BP dipping, we measured 24-hr BP and assessed continuous night noise in another sub-sample (n = 200). To ensure comparability between countries, we used common noise models to assess individual noise exposure, with a resolution of 1 dB(A). Modifiers of individual exposure, such as the orientation of living and bedroom toward roads, window-opening habits, and sound insulation, were assessed by the questionnaire. For four airports, we estimated exposure to air pollution to explore modifying effects of air pollution on cardiovascular disease. The project assesses exposure to traffic-related air pollutants, primarily using data from another project funded by the European Union (APMoSPHERE, Air Pollution Modelling for Support to Policy on Health and Environmental Risks in Europe).
air pollutionaircraftblood pressurehypertensionnoiseroad traffic
==== Body
An increasing number of people live in the vicinity of major airports and experience considerable noise and air pollution. Raised blood pressure (BP) is a major risk factor for coronary heart disease and the major risk factor for stroke (Whitworth 2003). Environmental noise is a significant problem in Europe, and it is estimated that roughly 20% of the European Union’s population (close to 80 million people) are exposed to noise levels that are considered unacceptable (European Commission 1996).
Few investigators have studied health effects associated with exposure to aircraft noise. Cardiovascular effects due to noise exposure have been studied to some extent, but no clear exposure–response relations are currently known (Babisch 2000), although a recent German study showed an excess risk of myocardial infarction related to traffic noise, but only in men (Babisch et al. 2005). An early European study showed higher treatment rates for “heart trouble” and hypertension among residents close to a major airport than among people living farther away (Knipschild 1977), and a later review found that hypertension was more prevalent among individuals living close to airports (Vacheron 1992). However, results are equivocal both with respect to BP increases (Babisch et al. 1990; Lercher et al. 2000; Pulles et al. 1990) and the prevalence of hypertension (Bluhm et al. 2001; Eiff and Neus 1980; Herbold et al. 1989; Knipschild and Sallé 1979; Maschke 2003). A recent cross-sectional study indicated an exposure–response relation between residence distance from a Swedish airport and hypertension (Rosenlund et al. 2001). Similar results were found in a community sample around a military airbase on Okinawa, Japan, and in a cross-sectional survey around Schiphol airport in Amsterdam, the Netherlands (Franssen et al. 2004; Matsui et al. 2004).
In 1999, the World Health Organization stated that the overall evidence available at the time suggested a weak association between long-term noise exposure and BP elevation or hypertension, and that cardiovascular effects are associated with long-term exposure to A-weighted average sound pressure levels [dB(A)] (LAeq,24hr) throughout the day/night in the range of 65–70 dB(A) (Berglund et al. 1999). However, a recent German study suggested that traffic noise at lower levels might increase the risk of myocardial infarction and high BP, finding an increased odds ratio for medical treatment of hypertension in subjects with an exposure during the day/night of > 60/50 dB(A) compared with subjects with an exposure < 60/50 dB(A) (Maschke 2003). Recent studies suggest that nighttime exposure might be particularly relevant for health (Babisch et al. 1999; Health Council of the Netherlands 2004; Lercher and Kofler 1993; Maschke 2003).
Stress hormones are useful indicators to study mechanisms and interactions between noise and health outcomes such as BP (Babisch et al. 2001). The cortisol level is a good indicator of stress (Wust et al. 2000). Salivary cortisol correlates well with free levels of cortisol in serum, and correctly collected saliva samples have the advantage of being stable for long periods at room temperature (Hofman 2001), which facilitates their use in multicenter studies.
Community noise studies have traditionally considered only noise from a single specific source such as aircraft or road traffic. However, recent studies suggest that aircraft noise might be more annoying than road traffic noise (Miedema and Oudshoorn 2001), but the extent to which the findings from individual studies can be extrapolated to other environments is at present unclear. It is not meaningful to use a total integrated estimate of noise in studies of nonauditory effects, because sound energy is not the only factor that causes stress reactions. Attitudes toward noise and the activities disturbed by it may modify the effect of noise quite considerably, as well as objective characteristics such as time pattern and sound frequency distribution. However, in studies of effects of community noise, nuisance from other noise sources should be addressed as independent (or interacting) factors (as is commonly done in environmental epidemiologic studies) (Babisch 2002; Babisch et al. 2003; Pulles et al. 1990).
The literature relating hypertension to air pollution is sparse, and two recently published studies show contradictory results (Ibald-Mulli et al. 2004; Zanobetti et al. 2004), but there is a wealth of literature on air-pollution–related cardiovascular effects, particularly associated with short-term changes in particulate air pollution levels (Dockery 2001). A recent study showed increased cardiopulmonary mortality associated with living near major roads (Hoek et al. 2002). However, although mortality was associated with living near roads, there was less consistency in the relation with ambient air pollution concentrations. In spite of this, no attempts were made to adjust for road-traffic–related noise. Thus, it may be important to assess ambient air pollution exposure as a possible confounder/effect modifier of the association between community noise and cardiovascular risk.
The aim of the Hypertension and Exposure to Noise near Airports (HYENA) project is to assess the impacts on cardiovascular health (primarily reflected by high BP) of noise generated by aircraft and road traffic near six major European airports (Athens, Greece; Milano/ Malpensa, Italy; Amsterdam/Schiphol, the Netherlands; Stockholm/Arlanda, Sweden; Berlin/Tegel, Germany; and London/ Heathrow, UK). The project will identify and quantify noise exposure in individuals, relating exposure to prevalence of high BP.
The project will study a subsample of subjects in more detail, recording 24-hr BP (every 15 min) and continuous night noise measurements, to assess the short-term effects of aircraft noise during nighttime and its effects on BP nighttime dipping, which is an established risk factor for cardiovascular disease (Lee et al. 2005; Ohkubo et al. 2002).
The project will evaluate the modifying effects of traffic-related air pollution [nitrogen dioxide, particulate matter with aerodynamic diameter ≤10 μm (PM10)] on noise-associated cardiovascular risk factors and cardiovascular disease (high BP, ischemic heart disease). Standardized methods for assessing exposure and effect are used, and exposure–response relationships between environmental noise exposure and health outcomes will be calculated. The project will analyze acute BP changes related to short-term aircraft noise exposure (nighttime in particular).
Materials and Methods
Study population and study area.
A total of 6,000 persons (men and women, 45–70 years of age) who have lived at least 5 years in the vicinity of the study airports have been selected using existing noise contours around the airports, aiming to maximize exposure contrast. The selection of the study areas was based on existing data on aircraft noise and road traffic noise levels. Recent aircraft noise contours were available for Milano/Malpensa, Berlin/Tegel, Stockholm/Arlanda, London/Heathrow, and Amsterdam/Schiphol. There was limited information for the new Athens airport, but predicted noise contours have been calculated in the planning process. The selection process created exposure contrast to aircraft noise and road traffic noise within countries, ensuring that sufficient numbers of inhabitants in the appropriate age range had expected exposures > 60 dB(A) and < 50 dB(A). The preferred noise distributions, serving as selection guidelines in the HYENA study, are shown in Table 1.
Road traffic noise may be difficult to predict because it is a ground-based source with a complex propagation path from source to receptor. For the initial selection process of the study population, we used local noise data to obtain road traffic exposure classification of locations and populations. If such data were unavailable, two simplified methods derived from more complex models were applied. The first is based on a U.K. method of calculating noise from roads (Great Britain Department of Transport 1988) where exposure depends only on distance to source, speed, and traffic flow. The second is an adaptation of a Dutch method (RMV 2002), which, in addition to the U.K. method, takes into account the numbers of cars, small- and medium-size trucks, and large trucks.
An example of the results of the initial selection in the Netherlands is given in Table 2. Based on the distribution of aircraft noise and road traffic noise in approximately 1 million homes surrounding Schiphol, the numbers of homes available for selection in the various exposure categories are shown in Table 2, which indicates that in the Netherlands it is not possible to comply with the current study demand for 5% of the population to be exposed to levels ≥70 dB(A). However, it is possible to select some additional homes < 70 dB(A), so the requirement for 20% of the population to be exposed to ≥65 dB(A) can be met. The Dutch HYENA study area with the annual average aircraft noise data for 2001 at grid level [day-evening-night noise level (Lden)] and a close-up showing administrative boundaries and residential address coordinates are shown in Figure 1.
In four study areas (Malpensa, Arlanda, Heathrow, and Athens airports), 24-hr BP and continuous nighttime noise will be measured in a subsample (n = 100) of highly exposed people (> 65 dB) and referents (n = 100; < 50 dB).
Health outcomes.
Blood pressure.
The protocol for assessing BP in HYENA is partly based on protocols previously developed and used in the large multicenter projects INTERSALT (International Study of Electrolyte Excretion and Blood Pressure) and INTERMAP (International Study of Macro-and Micronutrients and Blood Pressure) (Elliott and Stamler 1988).
Until recently the Rz (random zero) sphygmomanometer has been the standard for measuring BP in population studies, but the procedure is prone to errors (Staessen et al. 2000). Automated instruments reduce observer error and can print out measurements, ensuring that recorded data are accurate (O’Brien 2000). These techniques are now well established in clinical research and are increasingly used in occupational and environmental medicine (Staessen et al. 2000). The HYENA study protocol requires that only validated instruments be used. The procedures required for the validation have been thoroughly standardized (O’Brien et al. 2002).
Specially trained staff members (nurse or equivalent) measure BP at three occasions during a single visit. The visits are distributed over the day as far as feasible, to account for diurnal variations in BP. Information about activities of potential influence for BP levels during the assessment day is included in the questionnaire. In a subsample (n = 200), 24-hr BP measurements are carried out every 15 min, using validated instruments (Mobilograph; Numed Cardiac Diagnostics, Sheffield, UK).
Stress hormones.
A subsample of the study population (n = 500) has been identified, stratified by noise level. The participant gathers saliva in a test tube the day before the interview at three occasions; 30 min after awakening, immediately before lunch, and before going to bed. The tubes are collected on the day of the interview. The samples are frozen and kept at a temperature of at least −20°C until analysis. The laboratory determines saliva cortisol using a radioimmunoassay technique and collaborates with a European network for comparisons of saliva cortisol concentrations in different populations.
Confounders and effect modifiers.
A standardized questionnaire was designed, including validated questions on annoyance and noise disturbance from sources other than air and road traffic (e.g., neighbors) (Fields et al. 2001; Guski et al. 1999). The questionnaire also collects data on major well-known confounders (e.g., dietary habits, smoking, and other lifestyle factors as well as occupational noise exposure) using validated questions from previous studies.
Noise exposure assessment.
The aim of the noise exposure assessment is to determine individual exposure to aircraft and road traffic noise for each participant, using noise models and noise maps, calculating noise load at grid level. The noise model will calculate required noise levels with a resolution of 1 dB(A). Grid size depends on noise source (the best calculation for aircraft is 250 × 250 m, and for road traffic 10 × 10 m).
Since the start of the HYENA project, much progress has been made in the European Union on the harmonization of noise indicators, calculations, and mapping. Directive 2002/49/EC (European Commission 2002) obliges E.U. members to produce noise maps before 2007 for their larger agglomerations (> 250,000 inhabitants). Therefore, road traffic noise maps will be available for some study areas in the HYENA project. The directive selected Lden and LAeq 2300–0700 hr (Lnight) as the common noise indicators, and methods for noise modeling are described. Current exposure to aircraft and road traffic noise will be calculated for separate periods of the day [LAeq 0700–1900 hr (Lday), LAeq 1900–2300 hr (Levening), Lnight], which enables the calculation of combinations of the three indicators, including LAeq,24hr. These time periods are default values that can be modified according to differences between countries (Directive 2002/49/EC); the day–evening–night level (Lden) is defined by the formula given in Equation 1:
We selected the year 2002 as reference for assessment of current levels of aircraft and road traffic noise. It is assumed that the noise levels in this year are representative of the 5-year time period preceding the health status assessment. If not, teams are allowed to choose another representative year for this period.
Noise exposure will be assessed for each participant by linking home addresses to modeled aircraft and road traffic noise levels. Therefore, knowledge about the exact place of residence of the participants is needed. Address coordinates will be entered in a geographic information system. The difference between noise load at the facade of the sleeping room and that of the living room will be evaluated for road traffic noise. For aircraft noise, this difference is negligible.
Modifiers of individual exposure, such as the orientation of living and bedroom toward roads, window-opening habits, and sound insulation, are assessed during the home visits, using questionnaires and visual inspection. Based on this information, subjects with the same noise level at the front of the house will be stratified according to room orientation. In urban situations, where streets commonly are the predominant noise sources, attenuation of sound levels could be expected between the front and the back of a detached house of (at least) 10 dB(A), and (at least) 20 dB(A) for a terraced house. These attenuation levels will be adapted to each local situation.
Historical exposure to aircraft and road traffic noise over the past 5–10 years will be assessed if feasible. Modeling will not be performed for the years before 2000, but existing noise data will be used for the historical exposure assessment, when available. As a consequence, historical data might be available in local noise indicators only obtained with local noise models. To ensure cross-country comparability, these data will be converted to the noise indicators Lden, Lnight, and LAeq,24hr. For the conversion, we will model noise levels for the reference year (default 2002) using local models as well as the standard models used in HYENA. Based on a local comparison of model results, historical data will be adjusted to the results of the standard models. Cumulative exposure estimates will be computed by combining the individual (historical) residence noise data with the number of years spent at each address.
To facilitate comparability between the HYENA countries, a common noise model will be used to assess current exposure to aircraft noise. Internationally, the integrated noise model (INM) is at present the most accepted model and will serve as the standard model for assessment of current exposure. For the assessment of aircraft noise, information about airport, aircraft fleet, and runway use is essential, and data on flight tracks and flight procedures need to be collected, as well as meteorologic data and, if applicable, terrain elevation. Flight tracks will be based on radar tracks, representing the actual flown flight tracks. Flight profiles can be based on nominal (International Civil Aviation Organization/INM) profiles. The minimum output of the noise model will be the required noise levels with a resolution of 1 dB(A) on a grid with a minimum resolution of 250 × 250 m. Where feasible, the model estimates will be validated by noise measurements.
Road traffic noise assessment is less uniform among countries. Therefore, locally used models are more tailored to each local situation, emphasizing the collection of good quality data. The European Commission’s Working Group Assessment of Exposure to Noise recently published The Good Practice Guide for Strategic Noise Mapping (GPG) (European Commission WG-AEN 2003) to facilitate the production and comparability of road noise mapping. The GPG contains a tool kit that enables evaluation of the relevant input data with regards to validity and accuracy. HYENA countries will use the tool kits included in the GPG.
Countries preferring to use another noise model could use the European interim model on road noise, the French national computation method (CERTU/CSTB/LCPC/SETRA 1997). However, because this model is detailed and complex, HYENA countries may also consider the use of the less complicated models such as the German model RLS-90 (Schade 2004). The GPG recommends that the noise model should at least be able to calculate first-order reflections. If necessary, adjustment for reflections could be applied to any (grid-based) noise level to determine noise exposure for people living in dwellings. Although an adjustment is a compromise and will cause some inaccuracies, the GPG states that inaccuracies due to data deficiencies are likely to be much more significant.
For the study of short-term effects on BP, noise exposure will be assessed indoors during the night, using a type 1 noise meter (CESVA SC310; CESVA Instruments, SL, Barcelona, Spain), which records noise levels eight times per second, complemented by noise recording, using mp3 players, to allow characterization of the noise source (e.g., aircraft, traffic, indoor).
Air pollution exposure assessment.
For four airports (Athens, Malpensa, Schiphol, and Heathrow), exposure to air pollution from both aircraft and road traffic will be assessed to explore possible confounding and interactive effects of air pollution on cardiovascular disease. For each of these airports, exposures to traffic-related air pollutants (NO2, PM10) will be assessed, using results from another E.U.-funded project (Air Pollution Modelling for Support to Policy on Health and Environmental Risks in Europe; APMoSPHERE). Additional modeling will be performed at some airports. Modeled estimates will be validated by monitored data.
In APMoSPHERE, 1-km-resolution emission maps of PM1 0, sulfur dioxide, nitrous oxides (NOx), and carbon monoxide, by source sector, are being developed for all E.U. countries, using data on source distribution and activity, together with other proxies (e.g., population density) to redistribute national emission totals to local levels. Emissions from most source sectors are modeled as point or area sources, or some combination of the two. Data on point sources are derived from the 2001 European Pollutant Emission Register database (European Commission 2001).
Road transport represents a major source of emission for the pollutants relevant for HYENA (PM10, NOx). Thus, specific attention needs to be given to estimation of emissions from road transport. Initially, total emissions for each pollutant by vehicle type are obtained, as well as data on the distribution of vehicle types by road type. These are used to subdivide the total emissions by vehicle type into three categories: highways (motorways) and urban and rural roads. Separate calculations are made for urban and rural roads, and the population weights thus computed are used to redistribute the urban and rural emissions across the road network. Maps are integrated to provide a total emission from all vehicle types across all roads.
For airports, data on location and activity are available for each E.U. country. Emissions are estimated on the basis of source area and the level of airport activity. Detailed area data are not available for all airports; therefore, three size classes are applied where these data are lacking: large (25 km2) for major international airports, such as Heathrow; medium (9 km2) for smaller international airports; and small (1 km2) for local and regional airports. Only emissions from aircraft ground movements, take-off, and landing are considered. However, for the airports where detailed modeling will be undertaken (e.g., Heathrow), ground traffic will be included, and we will establish how much this will add to air pollution concentrations. Total national emissions within this sector are thus disaggregated to specific grid squares using activity-weighted areas.
For each participant in the HYENA air pollution substudy (n = 4,000), exposure estimates will be made by modeling mean annual and 95th percentile concentrations outdoors at the place of residence. Cumulative air pollution exposure estimates will be computed for each HYENA study participant by integrating place of residence, time spent at each address, and exposure data.
Study power and statistical analysis.
To detect a mean difference of 3 mmHg systolic BP (140 vs. 143 mmHg, SD = 20 mmHg) between groups in this cross-sectional study, approximately 700 persons are needed per group to achieve 80% power. Similarly, a mean difference of 2 mmHg diastolic BP (85 vs. 87 mmHg, SD = 12 mmHg) can be detected with approximately 600 persons per group. Thus, our study has sufficient power to detect small differences in BP potentially resulting from noise exposure.
We will analyze BP in relation to noise levels in each country and then pool the results in the final analyses using a fixed-effects model. We will analyze the impact of work-related noise by stratifying on occupational noise exposure assessed by the questionnaire. Individuals using antihypertensive drugs will be analyzed separately. Certain indoor environment characteristics (e.g., double glazing) will modify outdoor exposure and will be taken into account by stratified analysis.
Discussion
Previous studies that have indicated possible associations with airport-related noise and cardiovascular effects most often focused on one airport with relatively small study populations, as noted in an editorial accompanying the study on the Swedish airport (Rosenlund et al. 2001). The editorial suggested that a larger multicenter study was needed to test the hypothesis that airport-generated noise may give rise to raised BP (Pattenden 2001). The HYENA study has taken these suggestions into account as well as combining a large study size with a thorough exposure assessment to achieve maximum resolution in the analysis.
Selection of study areas.
The selection of study areas involved a number of deliberations to minimize confounding and ensure comparability between countries. Where feasible, we avoided areas where substantial changes in noise exposure had occurred (or would occur) in previous (or coming) years, and we chose areas with suitable noise data, such as road traffic intensity and speed. We aimed to select areas with low migration and to avoid areas with sound insulation programs, where feasible. Finally, we aimed to avoid areas with other sources of noise exposure (e.g., rail, industry) and to choose areas with similar socioeconomic status within countries. Nevertheless, some differences in socioeconomic status between areas are unavoidable and will be controlled for in the analyses.
Noise modeling.
Although the INM model is the most widely used, some countries may prefer to use their own model, such as the U.K. ANCON (Aircraft Noise Contour model) (Ollerhead 1992). However, the models are very similar, the computation methods being essentially identical, using industry-supplied data to relate aircraft thrust and height with noise emission for individual aircraft types. At London airports, more than one-third of movements are made by aircraft with no matching INM noise power distance data available (Jopson et al. 2002).
The study areas vary among HYENA countries, from urban areas with highly dense populations (Germany and United Kingdom) to mixed (Sweden and the Netherlands) and more rural areas (Greece and Italy). Therefore, there is a marked difference in the complexity of the modeling of road traffic noise, where the more urban areas necessitate a more detailed approach because of complex factors such as attenuation and reflection.
Before the adoption of Directive 2002/49/EC, almost every country in the European Union had its own national noise model. The adaptation of the noise directive will facilitate noise exposure assessment in the HYENA study and contribute to achieving comparable estimates between countries. We are aware that the simplified methods for road traffic noise modeling ignore important modifying factors such as type of road surface, barriers, acceleration near crossing, and surface reflection, but we believe this not to be a major problem for the selection of eligible study participants.
The quality of the noise modeling together with the use of the GPG should fulfill the requirements of the HYENA project. However, for home addresses with relatively low road traffic noise exposure, the noise exposure assessment may become inaccurate, because of deviations in the input data despite the use of the GPG. For example, traffic intensity might be so low that small deviations from the flows could result in large effects in the calculated noise load. Because input data are critical for the modeling of road traffic noise, HYENA countries will aim for the highest possible accuracy, when selecting the data collection methods according to the tool kits in the GPG.
Air pollution.
Airports generate both air pollution and noise. Aircraft movements are a major source of noise, whereas much of the air pollution (NO2, PM10) is associated with road transport related to airport activities. As a result, the geographic patterns of noise and air pollution near the airport tend to differ, with noise pollution following the main flight paths at take-off and landing, and air pollution often showing a more regional distribution, determined by the network of feeder roads. These spatial differences in the two pollutants provide the opportunity to examine their independent and interactive effects.
Exposure misclassification.
As in all epidemiologic studies, the HYENA study will suffer from a degree of exposure misclassification due to uncertainties in modeling noise levels from transport. However, a doubling of traffic will account for an increase of only 3 dB(A). Given an exposure range of about 45–75 dB(A), the impact on the analyses will probably not be substantial. Nevertheless, it is clear that the rather complex modeling involved will give rise to uncertainty, which we will assess by performing sensitivity analyses in the final data analysis.
Blood pressure measurements.
Single measurements or multiple readings taken by an observer at one or even several times through the day may reflect a subject’s true BP only to a minor extent (Staessen et al. 2000). Nevertheless, this method has commonly been used in environmental epidemiology. Ambulatory BP monitoring makes it possible to record the BP throughout the whole day in patients engaged in their normal activities and provides a reliable estimate of their BP over a 24-hr period. However, although 24-hr monitoring is preferable, the costs are usually prohibitive when the study population is large.
Stress hormones.
Three aspects of saliva cortisol assessments are crucial in relation to reactions to long-lasting stressors such as aircraft noise. First, as long as subjects have retained ability to up-regulate cortisol, levels may become elevated. This may be particularly relevant during the early morning hours. Second, when the life situation has been disturbed for a long time, the ability to down-regulate cortisol may be inhibited. This is particularly relevant during the late hours when cortisol excretion in normal subjects is much lower than during the early hours immediately after awakening. Finally, it is believed that subjects who have suffered from severe stressors for a long time may have exhausted the ability of the cortisol system to regulate in the normal way. In such cases, levels become abnormally low and show very little variation. Saliva cortisol measurements may show high and low levels, and high as well as low variability during exposure to long-lasting stressors. The exhausted group, however, is small in most studies of normal populations. Therefore, long-lasting stressor exposure may result in elevated cortisol levels, and perhaps lowered ability to decrease levels at night before bedtime.
Saliva sampling has the advantage above collection of blood specimens in that it is easy and cheap to administer. The study subjects can easily be instructed to collect samples themselves. Hence, many samples can be collected, and this makes it possible to study circadian disturbances in cortisol regulation.
Short-term effects of night noise and BP dipping.
Numerous laboratory studies have investigated sleep disturbances by acute noise; the effects on BP have been rarely assessed, and only a few field studies have been published (Babisch 2002). We will study short-term effects of noise on BP in 200 subjects, primarily to explore a possible link between short- and long-term effects through the inhibition of nighttime BP dipping, which is an established risk factor for cardiovascular disease (Lee et al. 2005; Ohkubo et al. 2002). We will compare day and night average BP levels and examine the association of nighttime BP to the average noise levels in the bedroom, assessing effects of aircraft noise and other noise sources separately.
Conclusion
To our knowledge, the HYENA study is the first large multicenter study designed to assess the effects of exposure to aircraft and road traffic noise on BP and cardiovascular disease, as well as possible modifying effects of exposure to air pollution. Study airports are located in several different European countries offering a wide range of noise exposure, ensuring exposure contrast and cross-country variations (e.g., due to cultural and climate differences), which will be addressed in the analysis. Final results can be expected in 2007.
The project is funded by a grant from the European Commission (Directorate General Research) Quality of Life and Management of Living Resources, Key Action 4 Environment and Health (grant QLRT-2001-02501). The Ministry of Housing, Spatial Planning and the Environment; the Ministry of Transport, Public Works and Water Management; and the Ministry of Health, Welfare and Sports cosponsor the activities in the Netherlands.
Figure 1 Study area with aircraft noise data at grid level (Lden, 2001) and a close-up showing administrative boundaries and residential address coordinates.
Table 1 Preferred distributions for aircraft noise and (in parentheses) for road traffic noise.
Cumulative fraction Noise level [dB(A)]
0.05 of the population exposed above 70
0.20 (0.25) of the population exposed above 65
0.35 (0.40) of the population exposed above 60
0.45 of the population exposed above 55
0.15 of the population for aircraft and road traffic noise exposed below 51
Table 2 Air and road traffic noise levels at the facade of houses in an area of 55 × 55 km near Schiphol Airport in Amsterdam.
Lden aircraft noise [dB(A)]
LAeq07–23 road noise [dB(A)] ≤ 50 51–55 56–60 61–65 66–70 ≥ 71 Total Percent
≤ 45 126,000 19,200 3,750 550 40 0 149,540 12.71
46–50 204,000 31,900 4,800 650 20 0 241,370 20.52
51–55 253,000 44,700 5,800 1,450 200 20 305,170 25.94
56–60 330,000 22,000 2,200 1,600 70 10 355,880 30.25
61–65 98,000 9,100 750 400 40 10 108,300 9.21
66–70 11,000 1,600 150 100 30 10 12,890 1.10
≥ 71 2,800 250 30 100 0 0 3,180 0.27
Total 1,024,800 128,750 17,480 4,850 400 50 1,176,330 100.00
Percent 87.12 10.95 1.49 0.41 0.03 0.00 100.00
LAeq07–23, A-weighted average sound pressure level [dB(A)], day and evening.
==== Refs
References
Babisch W 2000 Traffic noise and cardiovascular disease: epidemiological review and synthesis Noise Health 8 9 32 12689458
Babisch W 2002 The noise/stress concept, risk assessment and research needs Noise Health 4 1 11 12537836
Babisch W Beule B Schust M Kersten N Ising H 2005 Traffic noise and risk of myocardial infarction Epidemiology 16 33 40 15613943
Babisch W Fromme H Beyer A Ising H 2001 Increased catecholamine levels in urine in subjects exposed to road traffic noise: the role of stress hormones in noise research Environ Int 26 475 481 11485215
Babisch W Ising H Gallacher JEJ 2003 Health status as a potential effect modifier of the relation between noise annoyance and incidence of ischaemic heart disease Occup Environ Med 60 739 745 14504361
Babisch W Ising H Gallacher JEJ Elwood PC Sweetnam PM Yarnell JWG 1990 Traffic noise, work noise and cardiovascular risk factors: the Caerphilly and Speedwell collaborative heart disease studies Environ Int 16 425 435
Babisch W Ising H Gallacher JEJ Sweetnam PM Elwood PC 1999 Traffic noise and cardiovascular risk: the Caerphilly and Speedwell studies, third phase—10-year follow up Arch Environ Health 54 210 216 10444043
Berglund B Lindvall T Schwela DH eds. 1999. Guidelines for Community Noise. Geneva:World Health Organization. Available: http://whqlibdoc.who.int/hq/1999/a68672.pdf [accessed 23 February 2005].
Bluhm G Nordling E Berglind N 2001. Increased prevalence of hypertension in a population exposed to road traffic noise. In: Internoise 2001: Proceedings of the 2001 International Congress and Exhibition on Noise Control Engineering (Boone R, ed). Maastricht, the Netherlands:Nederlands Akoestisch Genootschap, 1563–1566.
CERTU/CSTB/LCPC/SETRA 1997. Bruit des infrastructures routières: Méthode de calcul incluant les effets météorologiques, version expérimentale, NMPB-routes-96. Lyon, France:Centre d’études sur les réseaux, les transports, l’urbanisme et les constructions publiques.
Dockery DW 2001 Epidemiologic evidence of cardiovascular effects of particulate air pollution Environ Health Perspect 109 483 486 11544151
Eiff AWv Neus H 1980 Verkehrslärm und Hypertonie-Risiko. 1. Mitteilung: Verkehrslärm und Hypertonie-Risiko [In German] Munch Med Wochenschr 122 893 896
Elliott P Stamler R 1988 Manual of operations for “INTER-SALT” an International Cooperative Study on the Relation of Sodium and Potassium to Blood Pressure Control Clin Trials 9 1S 118S 3396367
European Commission 1996. Green Paper: Future Noise Policy. Brussels:European Commission.
European Commission 2001. EPER (European Pollutant Emission Register). Available: http://eper.cec.eu.int/eper/ [accessed 5 September 2005].
European Commission 2002. Directive 2002/49/EC of the European Parliament and of the Council of 25 June 2002 relating to the assessment and management of environmental noise. Official Journal of the European Communities L189 of 18.7.2002, 12–25.
European Commission WG-AEN (Working Group Assessment of Exposure to Noise) 2003. Good Practice Guide for Strategic Noise Mapping and the Production of Associated Data on Noise Exposure Position Paper. Version 1. Available: http://europa.eu.int/comm/environment/noise/pdf/best_practice_guide.pdf [accessed 5 September 2005].
Fields JM De Jong RG Gjestland T Flindell IH Job RFS Kurra S 2001 Standardized general-purpose noise reaction questions for community noise surveys: research and recommendation J Sound Vib 242 641 679
Franssen EAM van Wiechen CMAG Nagelkerke NJD Lebret E 2004 Aircraft noise around a large international airport and its impact on general health and medication use Occup Environ Med 61 405 413 15090660
Great Britain Department of Transport 1988. Calculation of Road Traffic Noise. London:Her Majesty’s Stationery Office.
Guski R Felscher-Suhr U Schuemer R 1999 The concept of noise annoyance: how international experts see it J Sound Vib 223 513 527
Health Council of the Netherlands 2004. The Influence of Nighttime Noise on Sleep and Health. Report No. 2004/14E. The Hague:Gezondheidsraad.
Herbold M Hense HW Keil U 1989 Effects of road traffic noise on prevalence of hypertension in men: results of the Luebeck Blood Pressure Study Soz Praventivmed 34 19 23 2785319
Hoek G Brunekreef B Goldbohm S Fischer P van den Brandt PA 2002 Association between mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study Lancet 360 1203 1209 12401246
Hofman LF 2001 Human saliva as a diagnostic specimen J Nutr 131 1621S 1625S 11340128
Ibald-Mulli A Timonen KL Peters A Heinrich J Wolke G Lanki T 2004 Effects of particulate air pollution on blood pressure and heart rate in subjects with cardiovascular disease: a multicenter approach Environ Health Perspect 112 369 377 14998755
Jopson I Rhodes D Havelock P 2002. Aircraft noise model validation—how accurate do we need to be? Paper presented at Institute of Acoustics conference Action on Environmental Noise; 13 November 2002; Stratford upon Avon, UK. Available: http://www.caa.co.uk/docs/68/dap_ercd_1102_modelaccuracy.pdf [accessed 5 September 2005].
Knipschild P 1977 Medical effects of aircraft noise: community cardiovascular survey Int Arch Occup Environ Health 40 185 190 591083
Knipschild P Sallé H 1979 Road traffic noise and cardiovascular disease Int Arch Occup Environ Health 44 55 59 521164
Lee KW Blann AD Lip GY 2005 High pulse pressure and nondipping circardian blood pressure in patients with coronary artery disease: relationship to thrombogenesis and endothelial damage/dysfunction Am J Hypertens 18 104 115 15691624
Lercher P Kofler W 1993. Adaptive behavior to road traffic noise blood pressure and cholesterol. In: Noise and Man ’93, Proceedings of the 6th International Congress on Noise as a Public Health Problem, Nice 1993 (Vallet M, ed). Arcueil, France:Institut National de Recherche sur les Transports et leur Sécurité, 465–468.
Lercher P Widmann U Kofler W 2000. Transportation noise and blood pressure: the importance of modifying factors. In: Proceedings of the 29th International Congress and Exhibition on Noise Control Engineering (Cassereau D, ed). InterNoise 2000, Vol 4. Nice, France:Sociéte Francaise d’Acoustique, 2071–2075.
Maschke C 2003. Epidemiological research on stress caused by traffic noise and its effects on high blood pressure and psychic disturbances. In: Proceedings of the 8th International Congress on Noise as a Public Health Problem (Jong RD, Houtgast T, Franssen EAM, Hofman W, eds). Shiedam, the Netherlands:Foundation ICBEN, 96–101.
Matsui T Uehara T Miyakita T Hitamatsu K Osada Y Yamamoto T 2004 The Okinawa study: effects of chronic aircraft noise on blood pressure and some other physiological indices J Sound Vib 277 469 470
Miedema HM Oudshoorn CG 2001 Annoyance from transportation noise: relationships with exposure metrics DNL and DENL and their confidence intervals Environ Health Perspect 109 409 416 11335190
O’Brien E 2000 Replacing the mercury sphygmomanometer requires clinicians to demand better automated devices BMJ 320 815 816 10731152
O’Brien E Pickering T Asmar R Parati G Staessen J 2002 International protocol for validation of blood pressure measuring devices in adults. Working Group on Blood Pressure Monitoring of the European Society of Hypertension. International Protocol for validation of blood pressure measuring devices in adults Blood Press Monit 7 3 17 12040236
Ohkubo T Hozawa A Yamaguchi J Kikuya M Ohmori K Michimata M 2002 Prognostic significance of the nocturnal decline in blood pressure in individuals with and without high 24-h blood pressure: the Ohasama study J Hypertens 20 2183 2189 12409956
Ollerhead JB 1992. The CAA Aircraft Noise Contour Model: ANCON Version 1. DORA Repeort 9120. London:Civil Aviation Authority.
Pattenden S 2001 Air traffic noise and hypertension in Stockholm County [Editorial] Occup Environ Med 58 761 11706140
Pulles MPJ Biesiot W Stewart R 1990 Adverse effects on environmental noise ion health: an interdisciplinary approach Environ Int 16 437 445
RMW (Reken- en Meetvoorschrift Wegverkeerslawaai) 2002. Ministry of Housing, Spatial Planning and the Environment [in Dutch]. Staatscourant 62:27 maart 2002, the Netherlands.
Rosenlund M Berglind N Pershagen G Järup L Bluhm G 2001 Increased prevalence of hypertension in a population exposed to aircraft noise Occup Environ Med 58 769 773 11706142
Schade LR 2004 Genauigkeit und Eindeutigkeit: Ein Vergleich des französischen Berechnungsverfahrens für Geräuschimmissionen durch Strassenverkehr mit seinem deutschen Pendant [In German] Z Lärmbekämpfung 51 39 47
Staessen JA O’Brien ET Thijs L Fagard RH 2000 Modern approaches to blood pressure measurement Occup Environ Med 57 510 520 10896957
Vacheron A 1992 Cardiovascular effects of noise [in French] Bull Acad Natl Med 176 387 392 1504863
Whitworth JA 2003 World Health Organization (WHO)/ International Society of Hypertension (ISH) statement on management of hypertension J Hypertens 21 1983 1992 14597836
Wust S Federenko I Hellhammer DH Kirschbaum C 2000 Genetic factors, perceived chronic stress, and the free cortisol response to awakening Psychoneuroendocrinology 25 707 720 10938450
Zanobetti A Canner MJ Stone PH Schwartz J Sher D Eagan-Bengston E 2004 Ambient air pollution and blood pressure in cardiac rehabilitation patients Circulation 110 2184 2189 15466639
|
16263498
|
PMC1310905
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 13; 113(11):1473-1478
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8037
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8076ehp0113-00147916263499Commentaries & ReviewsGenetic Factors That Might Lead to Different Responses in Individuals Exposed to Perchlorate Scinicariello Franco Murray H. Edward Smith Lester Wilbur Sharon Fowler Bruce A. Division of Toxicology, Agency for Toxic Substances and Disease Registry, Centers for Disease Control and Prevention, Atlanta, Georgia, USAAddress correspondence to F. Scinicariello, Division of Toxicology, ATSDR, CDC, 4770 Buford Highway, MS: F-32, Atlanta, GA 30341 USA. Telephone: (770) 488-3331. Fax: (770) 488-4178. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 29 6 2005 113 11 1479 1484 2 3 2005 29 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Perchlorate has been detected in groundwater in many parts of the United States, and recent detection in vegetable and dairy food products indicates that contamination by perchlorate is more widespread than previously thought. Perchlorate is a competitive inhibitor of the sodium iodide symporter, the thyroid cell–surface protein responsible for transporting iodide from the plasma into the thyroid. An estimated 4.3% of the U.S. population is subclinically hypothyroid, and 6.9% of pregnant women may have low iodine intake. Congenital hypothyroidism affects 1 in 3,000 to 1 in 4,000 infants, and 15% of these cases have been attributed to genetic defects. Our objective in this review is to identify genetic biomarkers that would help define subpopulations sensitive to environmental perchlorate exposure. We review the literature to identify genetic defects involved in the iodination process of the thyroid hormone synthesis, particularly defects in iodide transport from circulation into the thyroid cell, defects in iodide transport from the thyroid cell to the follicular lumen (Pendred syndrome), and defects of iodide organification. Furthermore, we summarize relevant studies of perchlorate in humans. Because of perchlorate inhibition of iodide uptake, it is biologically plausible that chronic ingestion of perchlorate through contaminated sources may cause some degree of iodine discharge in populations that are genetically susceptible to defects in the iodination process of the thyroid hormone synthesis, thus deteriorating their conditions. We conclude that future studies linking human disease and environmental perchlorate exposure should consider the genetic makeup of the participants, actual perchlorate exposure levels, and individual iodine intake/excretion levels.
genetic susceptibilityhypothyroidismmutationsNISPendred syndromependrinperchloratethyroid glandTPO
==== Body
Sequencing of the human genome has brought new emphasis and increased interest in gene–environment interactions and is becoming relevant in defining public health policies. For many years, people’s susceptibility to xenobiotics have been known to differ significantly. Now, several techniques are available to identify and characterize the genetic correlates of interindividual variability. The goal of environmental genomics is to help investigators understand how genetic variability influences individual responses to environmental factors on the basis of the assumption that high-risk genotypes accumulate more damage and therefore are at greater risk of developing exposure-related diseases. Thus, genomics information may lead to development of predictive biomarkers that identify potentially sensitive populations and earlier prediction of adverse outcomes, ultimately resulting in better intervention strategies (Kelada et al. 2003).
Public Health and Perchlorate
The advent and use of new, highly sensitive detection techniques have identified contamination of groundwater by perchlorate in many parts of the United States, primarily in association with industries involved in rocket, explosives, and fireworks manufacturing and propellant handling. Concentrations measured in most public water supplies are < 50 μg/L; however, levels as high as several hundred microliters per liter have been reported in some drinking water wells in certain communities (Motzer 2001). The recent detection of perchlorate in vegetable and dairy food products (Kirk et al. 2003; Smith et al. 2001) indicates that contamination by perchlorate is more widespread than previously thought.
Perchlorate is the dissociated anion of perchlorate salts such as potassium perchlorate, sodium perchlorate, and ammonium perchlorate and is extremely water soluble and environmentally stable. Therefore, the perchlorate ion is identical whether derived from potassium, sodium, or ammonium salts. Potassium perchlorate was used primarily as a pharmaceutical agent to treat hyperthyroidism. It is now used mainly in flares and automobile airbags, although it is still used for diagnostic purposes and for treatment of hyperthyroidism. Sodium perchlorate is used in the manufacture of slurry explosives. Ammonium perchlorate is widely used as a component of propellants for rockets, missiles, and fireworks (Soldin et al. 2001).
Perchlorate was first detected in high concentrations by monitoring wells in California during the early 1990s. When initially detected in California, the Region 9 Office of the U.S. Environmental Protection Agency developed a provisional reference dose for perchlorate of 0.0001 mg/kg/day in 1992. The reference dose later was revised to 0.0005 mg/kg/day in 1995. These values were based on a dated acute exposure study that showed that single doses of potassium perchlorate caused release of iodide from thyroids of patients with Graves disease, an autoimmune condition that results in hyperthyroidism (Stanbury and Wyngaarden 1952). In 2005 the reference dose was changed again to 0.0007 mg/kg/day.
Mode of Action of Perchlorate in Humans
Perchlorate is a competitive inhibitor of the sodium iodide symporter (NIS), the thyroid cell–surface protein responsible for transporting iodide from the plasma into the thyroid. Therefore, it prevents further synthesis of the thyroid hormone (TH). It has no effect on the iodination process itself; rather, it displaces iodide by competitive uptake at the NIS. Perchlorate is concentrated by the thyroid tissue in a manner similar to iodide, but it is not significantly metabolized either in the gland or peripherally (Wolff 1998). Eskandari et al. (1997) disputed the notion that perchlorate is translocated via NIS into the cell and that perchlorate acts on NIS as a blocker, not as a substrate. Therefore, it is possible that perchlorate may cross the thyrocyte membrane by diffusion. In rats and humans, perchlorate appears to be eliminated rapidly, primarily in urine (>90%), virtually unchanged (Anbar et al. 1959; Eichler and Hackenthal 1962).
Several other inorganic anions such as thiocyanate and nitrate that are present in dietary and environmental sources have goitrogenic effects (Greer et al. 1966). Similar to perchlorate, they both competitively inhibit iodide uptake at NIS. Several studies have been conducted to determine the relative effects of perchlorate, thiocyanate, and nitrate on radioactive iodine uptake (RAIU) inhibition. Studies in rats showed that perchlorate was approximately 10 times more potent than thiocyanate and about 300 times more potent than nitrate in inhibiting RAIU in the thyroid. Furthermore, thiocyanate was slightly more potent than iodide (Wyngaarden et al. 1953). Tonacchera et al. (2004) demonstrated, in Chinese hamster ovary (CHO) cell lines stably transfected with the human NIS gene, that the relative potency of perchlorate on RAIU inhibition was 15, 30, and 240 times that of thiocyanate, iodide, and nitrate, respectively. The inhibiting effects when the cell lines where exposed to a mixture of perchlorate, thiocyanate, and nitrate were simply additive.
Thyroid Hormone Synthesis
Thyroid hormone plays a key role in the growth and differentiation of many organs. It is especially important for development of the central nervous system during the prenatal and postnatal periods (reviewed by Zoeller et al. 2002). A severe shortage of TH for several weeks after birth results in serious mental and motor handicaps. During pregnancy the mother provides substantial amounts of TH to the fetus (Vulsma et al. 1989), so the delay in cerebral development caused by congenital hypothyroidism (CH) results mainly from postnatal TH deficiency. The risk for mental retardation and the difficulty in recognizing the disease were reasons for introducing neonatal mass screening programs. Therefore, the most serious effects of perchlorate might occur during the first trimester when the brain is forming and developing and TH supply is totally dependent on maternal supply of iodine and of thyroxine (T4) and triiodothyronine (T3)
To understand the potential impact of perchlorate on a gene–environment interaction model, we need to consider T3 and T4 in a proper biosynthesis context. TH synthesis and secretion are exquisitely regulated negative-feedback systems that involve the hypothalamus, pituitary, and thyroid glands. The hypothalamus secretes thyrotropin-releasing hormone (TRH), a tripeptide (pyroGlu-His-Pro) synthesized in the paraventricular nucleus of the hypothalamus. The TRH, transported by axons, binds to TRH receptors in the pituitary thyrotropes, a subpopulation of pituitary cells that secrete thyroid-stimulating hormone (TSH). TRH stimulation leads to release and synthesis of new TSH in thyrotropes. The TSH binds to the TSH receptor in the thyroid gland cells. TSH is the primary regulator of TH release and secretion. Both TRH and TSH secretion are negatively regulated by THs: when T4 reaches an adequate circulating level, the hypothalamus and pituitary reduce their output of TRH and TSH; they increase their output of TRH and TSH when the circulating blood level of T4 is low. A number of thyroid genes, including NIS, thyroglobulin (Tg), and thyroid peroxidase (TPO), are stimulated by TSH and promote the synthesis of TH (Zoeller 2003).
Iodine is critical to thyroid gland function and TH synthesis and secretion. The first step in thyroidal iodine metabolism is the cellular uptake of iodide from the extracellular fluid. The thyroidal iodine uptake is tightly regulated by the NIS, an intrinsic plasma membrane protein in the thyroid follicular cells (Dohán et al. 2003). From the follicular cell, the iodide moves across the apical membrane, transported by pendrin protein (Yoshida et al. 2002). The iodide is then delivered to the cell–colloid interface, where it is oxidized by TPO and bound to tyrosyl residues in the Tg. This iodination of specific tyrosines on Tg yields monoiodinated and diiodinated residues [monoiodotyrosines (MITs) and diiodotyrosines (DITs)] that are enzymatically coupled to form T4 and T3. The iodinated Tg containing MIT, DIT, T4, and T3 then is stored as an extracellular storage polypeptide in the colloid within the lumen of thyroid follicular cells.
Perchlorate does not undergo metabolism, but genetic defects of its target, that is, the NIS, may lead to low iodine uptake in the thyroid gland, thus depressing production of THs. In this scenario, exposure to perchlorate may further reduce the already low iodide uptake and decrease production of THs. The combined effects of perchlorate with a genetic decrease in THs would hence delineate a population at risk for decreased thyroid function.
We reviewed published data to identify genetic factors that might lead to different responses in people exposed to perchlorate in the environment. Because perchlorate inhibits iodide uptake, we focused on the genetic defects causing CH involving the iodination process of the THs, particularly a) defects in iodide transport from circulation into the thyroid cell; b) defects in iodide transport from the thyroid cell to the follicular lumen, often combined with inner ear deafness [Pendred syndrome (PDS)]; and c) defects of iodide organification.
A positive perchlorate discharge test is used as a diagnostic tool in most of these medical conditions. A positive diagnosis can be obtained by administering 1 g potassium perchlorate 2 hr after a tracer dose of 131I. In normal individuals, radioiodide accumulation in the thyroid gland ends after the administration of potassium perchlorate, but there is little loss of the thyroidal radioactivity previously accumulated in the gland. Instead, potassium perchlorate causes almost complete discharge of the unbound fraction of thyroid iodide in individuals with defects of iodide organification and with PDS. Therefore, these people could have different responses to environmental perchlorate exposure than normal individuals.
Defects in iodide transport from circulation into the thyroid cell.
The NIS is the plasma membrane glycoprotein that mediates active iodide uptake into the thyroid follicular cells. This process is the crucial first step in TH biosynthesis. NIS couples the inward transport of sodium, which occurs in favor of its electrochemical gradient, to the simultaneous inward translocation of iodide against its electrochemical gradient. Two sodium ions per iodide ion are translocated into the cells (Dai et al. 1996; Eskandari et al. 1997). The sodium gradient that drives iodide uptake is maintained by the Na+/K+ ATPase.
Congenital iodide transport deficit (ITD) is an infrequent autosomic recessive condition characterized by inability of the thyroid gland to maintain a concentration gradient of iodide between the plasma and the thyroid follicular cell, resulting in hypothyroidism, diffuse or nodular goiter, and little or no uptake of radioiodine. The disorder has been linked to a defect of the NIS. In the absence of a functional NIS molecule, iodide has no access to the thyroid follicular cells, resulting in decreased TH biosynthesis and higher circulating levels of TSH, which in turn stimulates the morphologic and biochemical changes in the thyroid that result in development of goiter (De La Vieja et al. 2000).
The gene coding for human NIS has been mapped to chromosome 9p12-13.2. It has 15 exons and coding for a glycoprotein of 643 amino acids. NIS is a protein with 13 putative transmembrane domains, an extracellular amino terminus, and an intracellular carboxyl terminus (De La Vieja et al. 2000). About 58 cases of ITD from 33 families have been reported worldwide. Thirty of 31 cases from 21 families were studied at the molecular level and had several homozygous or compound heterozygous mutations of the perchlorate-sensitive NIS gene. Eleven mutations have been identified: V59E, G93R, Q267E, C272X, T354P, G395R, frameshift 515X, Y531X, G543E, ΔM142-Q323, and ΔA439-P443 (Fujiwara et al. 1997, 1998, 2000; Kosugi et al. 1998a, 1998b, 1999, 2002; Matsuda and Kosugi 1997; Pohlenz et al. 1997, 1998; Tonacchera et al. 2003). The single substitution in codon 354 converting from ACA (Thr) to CCA (Pro) was the most common mutation detected in 10 patients with homozygous mutations, and in four patients with compound heterozygous mutation (Fujiwara et al. 1997, 1998; Kosugi et al. 1998a, 1998b; Matsuda and Kosugi 1997). All were Japanese, suggesting that the mutant NIS T354P is more common in Japan. However, the frequency of this gene in the Japanese population is unknown because only 185 healthy people, representing only 370 alleles, have been genotyped.
The frequency of mutations in the NIS gene in the population is not known. Heterozygous persons do not express the phenotype; therefore, NIS gene defects can be detected only when both alleles are affected. People with homozygous mutations that cause partial loss of function may not be detected when, under conditions of high iodide intake, full preservation of iodide concentrating function is not required to achieve normal hormone synthesis. Therefore, impairment of thyroidal iodide concentration requires not only mutations in both NIS alleles but also defects that cause virtually complete loss of function.
The therapeutic treatment of ITD patients consists of l-T4 administration. Some patients also are supplemented with potassium iodide, thus underscoring the degree of functional loss of the mutated NIS. In these persons, perchlorate intake from contaminated sources could further reduce the functional activity of the mutated NIS in concentrating iodide in the thyroid.
Defects in iodide transport from the thyroid cell to the follicular lumen, often combined with inner ear deafness (PDS).
PDS, an autosomal recessive disorder characterized by deafness and goiter, is the most common cause of syndromic deafness, accounting for up to 10% of all hereditary hearing loss (Fraser 1965; Nilsson et al. 1964). A phenotypic heterogeneity exists among affected persons, and thyroid dysfunction is particularly variable. At least 50% of affected persons have normal circulating levels of TH, whereas others develop clinical hypothyroidism (Reardon et al. 1999). Most affected persons demonstrate impaired iodide organification, as determined by a positive perchlorate discharge test. Hearing loss in PDS is prelingual and, in at least 80% of patients, is associated with structural defects of the inner ear, including a dilatation of the vestibular aqueduct and the Mondini defect of the cochlea (Johnsen et al. 1989). The PDS gene (SLC26A4) has been linked to chromosomal region 7q31 and contains an open reading frame of 2,343 bp encompassing 21 exons (Coyle et al. 1996; Sheffield et al. 1996). The predicted gene product pendrin is a highly hydrophobic 780 amino acid protein that transports chloride and iodide and mediates the exchange of chloride and formate. In the thyroid gland, a disorder in the function of pendrin may cause diminished iodide transport over the apical membrane that results in iodide remaining in the thyrocyte and a consequent decrease of organification of iodide. As a result, iodide accumulates in the cytoplasm and is discharged if thiocyanate or perchlorate is given (perchlorate discharge test). A decrease in the amount of radiolabeled iodide over the thyroid of > 10% is considered positive. At least 85 independent SLC26A4 gene mutations have been characterized as causing PDS and nonsyndromic deafness, in some cases confirmed by a normal perchlorate discharge test (Adato et al. 2000; Blons et al. 2004; Bogazzi et al. 2000, 2004; Campbell et al. 2001; Coucke et al. 1999; Coyle et al. 1998; Everett et al. 1997; Fugazzola et al. 2000; Kopp et al. 1999; Li et al. 1998; Lopez-Bigas et al. 2002; Namba et al. 2001; Park et al. 2003; Prasad et al. 2004; Reardon et al. 2000; Scott et al. 2000; Tekin et al. 2003; Tsukamoto et al. 2003; Usami et al. 1999; Van Hauwe et al. 1998; Yong et al. 2001). Although these mutations are distributed throughout the coding sequence, having been identified in 19 of the 21 exons, the spectrum of mutations appears to show geographic differences. In Caucasian patients, the L236P, T416P, and IVS8+1G > A mutations account for nearly half of all SLC26A4 mutant alleles, whereas in Japanese patients, these mutations are rare (Campbell et al. 2001; Tsukamoto et al. 2003). By contrast, H723R and ISV7–2A > G are the prevalent alleles accounting for most observed SLC26A4 mutations in Korean and Japanese studies (Park et al. 2003; Tsukamoto et al. 2003). Some researchers have suggested that the frequency of these mutations could represent a founder effect rather than mutational hot spots.
A disorder in the function of pendrin will cause a diminished iodide transport over the apical membrane, which causes iodide to remain in the thyrocyte. Intake of perchlorate from a contaminated source may cause discharge of iodide from the thyrocyte, further exacerbating the organification defect, with resulting decrease of TH synthesis.
Moreover, at present, it is not known whether perchlorate will affect the function of the normal pendrin protein to transport iodide. Molecular studies addressing whether perchlorate may act on iodide transport through inhibition of the pendrin protein in a fashion similar to the NIS are needed and welcomed.
Defects in iodide organification.
Iodide organification is the process by which iodine is oxidized and bound to thyrosine residue in Tg. Thyroid iodide organification disorder represents a group of defects characterized by discharge of substantial percentage of labeled iodide from the thyroid after administration of perchlorate (perchlorate discharge test) or thiocyanate. This discharge indicates a defect in converting accumulated iodide to organically bound iodine. The discharge may be partial or complete, thus defining partial or total defects. Partial iodide organification defects (PIODs) are characterized by release of < 50% of the accumulated radioiodine. Total iodide organification defects (TIODs) are characterized by release of > 90% of the accumulated radioiodine.
Iodination of the tyrosine residue is catalyzed by the membrane-bound thyroperoxidase (TPO). However, the oxidation of iodine requires hydrogen peroxide synthesized outside the thyroid follicular cell at the apical border catalyzed by the thyroid complex. Recently, two proteins of this complex, DUOX1 (also known as THOX1) and DUOX2 (also known as THOX2), have been identified (De Deken et al. 2000; Dupuy et al. 1999). The DUOX1 and DUOX2 genes are co-localized on the 15q15.3 chromosome and code for proteins of 1,551 and 1,548 amino acids, respectively. The DUOX1 and DUOX2 structure includes seven transmembrane-spanning domains, three NADPH- and one FAD-binding site, and 2EF-hand motifs. During the past three decades, few cases of thyroidal hydrogen peroxide have been described, but the molecular bases of these defects have just recently been investigated. Moreno et al. (2002) reported mutations in the DUOX2 gene, resulting in premature stop codon, in four CH patients with unexplained iodide organification defects. One patient with permanent CH and TIOD carried a homozygous substitution, whereas three patients with temporary CH and PIOD carried heterozygous mutations that cause premature termination signal.
Lack of or insufficient activity of the DUOX2 protein diminishes hydrogen peroxide production, resulting in decreased activity of TPO and accumulation of iodide in the thyrocyte. Intake of environmental perchlorate, which inhibits iodine inflow, also may cause discharge of unbound iodine, further deteriorating the iodine organification process.
Under oxidative conditions, TPO catalyzes the coupling of iodotyrosines to iodothyronine residue in Tg. Thyroperoxidase is a glycosylated hemoprotein encoded by the TPO gene located on chromosome 2p25. The gene contains 17 exons coding for a protein of 933 amino acids. The protein has a transmembrane helix with a large extracellular N-terminal part containing a heme group. TPO defects are believed to be among the most frequent causes of abnormalities in thyroid iodide organification defect causing goitrous CH. TPO activity is not detectable in thyroid tissue of patients with TIOD. Absence of TPO activity implicates the inability to iodinate tyrosine residue in Tg and to couple these residues to form THs, mainly T4 and some T3 and rT3 (reverse T3) Inactivating mutations in both TPO alleles have been found in patients with CH caused by TIOD. With use of a variety of molecular techniques for mutation deletion, 36 mutations have now been defined for TPO. These include frameshift mutations caused by nucleotide insertion or deletion, as well as missense, nonsense, and splice site mutations (Abramowicz et al. 1992; Ambrugger et al. 2001; Bakker et al. 2000; Bikker et al. 1994, 1995, 1997; Kotani et al. 2001; Nascimento et al. 2003; Niu et al. 2002; Pannain et al. 1999; Rivolta et al. 2003; Santos et al. 1999; Umeki et al. 2002, 2004; Wu et al. 2002). The first reported mutation was a homozygous GGCC insertion in exon 8 of the TPO gene. The resulting frameshift generates a stop codon in exon 9, which results in a grossly truncated protein with no expected activity (Abramowicz et al. 1992). In a Dutch study of 45 patients from 40 families with CH caused by TIOD, the GGCC insertion in exon 8 at nucleotide position 1287 was the most common mutation found (Bakker et al. 2000). It was detected in 36% of the investigated TPO alleles and in 51% of the families investigated either in a homozygous or a compound heterozygous fashion. In this study, mutations in both TPO alleles were found in 29 families: for 13 families in a homozygous fashion and for 16 families in a compound heterozygous fashion. A total of 16 different mutations were found, including 8 novel mutations: 6 frame-shift mutations, 6 missense mutations, 3 splice site mutations, and 1 nonsense mutation. Most of these mutations occurred in exon 8, 9, or 10, which encode for the active part of the enzyme involved in the heme binding. In one patient with classic TIOD, a homozygous deletion in exon 14 appeared to have resulted from partial maternal isodisomy of the short arm of chromosome 2 carrying the defective TPO gene (Bakker et al. 2001). In some patients alternative splicing would generate a partially active form of the enzyme. In others an early termination signal would prevent translation of the fully active protein (Abramowicz et al. 1992; Bikker et al. 1994, 1995; Mangklabruks et al. 1991; Santos et al. 1999). Umeki et al. (2002) described 2 novel mutations in the TPO gene, R665W and G771R, in exons 11 and 13, respectively. The former was found in the patient’s father (heterozygous) and the latter in her mother, also heterozygous. No TPO activity was detectable with cells transfected with mutated mRNAs. Moreover, the mutated TPO proteins showed abnormal cellular localization, exhibiting immunofluorescence only in the intracellular structure. Therefore, the loss of apical membrane localization of the mutated TPO was the main cause for the iodide organification defect.
PIODs also can be caused by disorders in TPO. In an investigation of TPO mutations in five families with PIOD, Nascimento et al. (2003) found a compound heterozygous mutation in three patients from one family inherited from both heterozygous parents. In the other four families, they found only heterozygous TPO mutations or polymorphisms, suggesting the translated protein could be partially inactive. Recently, PIOD caused by TPO gene was diagnosed in three siblings (Kotani et al. 2003). The three siblings with goiter and latent-to-mild hypothyroidism had a compound heterozygous mutation for a missense mutation (G1687T) and a deletion in exon 10 (1808-13del), resulting in a produced protein with two deleted amino acids ΔD574-L4575. From the expression studies, the mutated ΔD574-L4575–TPO synthesized THs to some extent (Kotani et al. 2003).
A common feature of patients with thyroid organification disorders syndrome is the discharge of iodine from the thyroid after administration of perchlorate. The level of perchlorate administrated in the diagnostic test is higher than the reported level of contaminated sources. However, it is biologically plausible that cumulative ingestion of perchlorate through a contaminated source may cause some degree of iodine discharge from thyrocytes. In populations with partial activity of the TPO enzyme, exposure to high enough levels of environmental perchlorate could cause unbound iodide discharge; therefore, less iodine will be available for biosynthesis of THs, thus further deteriorating their conditions.
Relevant Studies of Perchlorate in Humans
Many studies have attempted to provide useful information on the dose–response relation of perchlorate-related health effects. Several ecologic studies have compared thyroid function in newborns using T4 and TSH screening data in infants born to mothers in areas with different perchlorate exposure. However, these studies yielded contradictory results. Brechner et al. (2000) found higher TSH in newborns in Yuma, Arizona, which has high perchlorate exposure, than in Flagstaff, Arizona, which has lower exposure. However, whether perchlorate exposure caused the observed TSH effect cannot be addressed because of the lack of direct perchlorate measurement in the study. By contrast, F.X. Li et al. (2000) and Z. Li et al. (2000) found no association in Nevada newborns between low T4 and TSH levels and perchlorate exposure. A limitation of these studies is that the investigators did not collect data on individual exposure to perchlorate and on iodine intake levels. In a population-based ecologic study using California Newborn Screening Program data, Schwartz (2001) claimed to identify a significant dose–response association between perchlorate exposure and T4, and an association of perchlorate exposure and being a presumptive positive for CH. These data contrast with a previous ecologic analysis (Lamm and Doemland 1999) that found no increase of CH incidence in California and Nevada counties with perchlorate levels of 4–16 μg/L in drinking water supplies.
Crump et al. (2000) conducted a study in three proximate cities in northern Chile that had different concentrations of perchlorate in tap water, involving 162 school-age children and 9,784 newborns. These authors found no alteration of thyroid function or incidence of CH in Taltal, Chile, where the tap water contained 100–120 μg/L perchlorate, compared with two other regions of Chile with low or no perchlorate in the water. However, the data also showed high levels of urine iodine, indicating that iodine intake in the population was very high, possibly overcoming the inhibitory effect of perchlorate on thyroid function.
To establish the dose response in humans for the perchlorate inhibition of thyroidal iodide uptake and the short-term effects on circulating TH, Greer et al. (2002) gave perchlorate in drinking water at 0.007, 0.02, 0.1, or 0.5 mg/kg per day to 37 male and female volunteers for 14 days. In 24 participants 8-and 24-hr measurements of thyroidal 123I uptake (RAIU) were performed before exposure, on exposure days 2 and 14, and 15 days postexposure. Results from the study indicated a true no-effect level of perchlorate of 5.2 or 6.4 μg/kg/day for RAIU. Considering that a 70 kg adult drinks 2 L of water per day, this dose would be ingested if the drinking water contained 182–224 μg/L. In addition, the dose of 0.5 mg/kg/day taken for 14 days did not produce changes in circulating levels of T4 or TSH, suggesting that short-term consumption of perchlorate levels of 17.5 mg/L in drinking water would not affect circulating levels of THs. The authors suggested that this failure of perchlorate to influence circulating levels of TH resulted from the storage capacity of the normal adult thyroid gland, which contains unreleased stored hormones lasting for several months. However, as pointed out by Zoeller (2003), the case may be different for a late gestation fetus or neonate, where the estimated intrathyroidal amount of hormone stored is less than that required for 1 day (Van den Hove et al. 1999; Vulsma et al. 1989). Thus, the concentration of perchlorate sufficient to reduce thyroidal iodine uptake in a fetus or neonate may be sufficient to produce a significant decrement in circulating levels of TH. The fetal thyroid gland obtains iodide for its own TH synthesis from the maternal circulation through the placenta. Placental transfer of perchlorate has been reported in guinea pig (Postel 1957). In human, whether perchlorate crosses from the mother to the fetus during pregnancy is not known. However, this placental transfer could be biologically plausible because expression of the NIS has been reported in human placenta (Bidart et al. 2000). Moreover, perchlorate may concentrate in milk because the NIS protein is induced in lactating breast tissue by prolactin (Tazebay et al. 2000). Perchlorate might decrease iodide uptake into milk, thus reducing the sole source of iodine to the infant. Differently from adults, who most likely can recover from transient hypothyroidism without permanent health consequences, a short period of TH insufficiency may produce permanent neurologic deficits in children (Van Vliet 1999). The study of no-effect level (Greer et al. 2002) was conducted in healthy adults with normal iodine intake, and it is debatable whether 14 days is sufficient time to illustrate perchlorate effect on humans. This no-effect level most likely would be lower in populations with genetic defects causing CH and in populations with lower iodine uptake. The Third National Health and Nutrition Examination Survey (NHANES III), conducted during 1988–1994, found that the percentages of males and females with urinary iodine concentrations < 5 μg/dL were substantially higher in every age category than in the 1971–1974 survey (Hollowell et al. 1998). In pregnant women, these percentages were 6.9% in NHANES III and 1.0% in NHANES I (Hollowell et al. 1998). The overall decline in the last few decades raises concern that a fairly large number of people in the United States may lack adequate iodine intake.
Conclusions
Exposure to perchlorate, which inhibits iodine uptake, has the biologic potential to cause hypothyroidism and, in pregnant women, severely damage the fetus and the newborn. NHANES III data suggest that 4.3% of the U.S. population may be subclinically hypothyroid (Hollowell et al. 2002). CH affects about 1 in 3,000 to 1 in 4,000 infants and in about 15% of cases may result from a defect of thyroid hormonogenesis, mostly inherited in an autosomal recessive fashion (Vulsma and de Vijlder 2000). Such defects may result from abnormalities in several steps involved in TH synthesis. Our literature review identified possible homozygous or compound heterozygous mutations of genes involved in thyroid iodine synthesis that cause hypothyroidism that could be used to define a potential susceptible population to perchlorate exposure. In a Mendelian fashion, the number of carriers of heterozygous mutated gene causing CH would be higher than the number of the reported CH cases. Given the logical connection between perchlorate, diminished iodine uptake, hypothyroidism, and thyroid-related health effects, people exhibiting heterozygous or homozygous genetic mutations in genes involved in the TH synthesis, especially in a milieu of low iodine uptake, can reasonably be expected to be more susceptible than people who show no genetic variability to the effects of perchlorate. Several studies based on T4 and TSH screening data in infants born to mothers in areas with different perchlorate exposure mostly have found no increase in hypothyroidism incidence. However, these studies lacked estimates of individual perchlorate exposure, as well as estimates of individual iodine uptake. The only study that included iodine values showed no significant association between perchlorate and hypothyroidism. However, it showed high urinary iodide, suggesting the high iodine uptake could easily have upset the inhibition factor of the perchlorate. We conclude that future epidemiologic and population-based studies as well as no-effect studies concerning the link between human disease and environmental perchlorate exposure should consider among their variables the genetic makeup of the participants, actual perchlorate exposure levels, and individual iodine uptake and excretion levels.
We thank O. Harris for critical comments.
This project was supported under a cooperative agreement from the Centers for Disease Control and Prevention through the Association of Teachers of Preventive Medicine (ATPM). F.S. is a recipient of an ATPM Career Development Award.
==== Refs
References
Abramowicz MJ Targovnik H Varela V Cochaud P Krawiec L Pisarev MA 1992 Identification of a mutation in the coding sequence of the human thyroid peroxidase gene causing congenital goiter J Clin Invest 90 1200 1204 1401057
Adato A Raskin L Petit C Bonne-Tamir B 2000 Deafness heterogeneity in a Druze isolate from the Middle East: novel OTOF and PDS mutations, low prevalence of GJB2 35delG mutation and indication for a new DFNB locus Eur J Hum Genet 8 437 442 10878664
Ambrugger P Stoeva I Biebermann H Torresani T Leitner C Gruters A 2001 Novel mutations of the thyroid peroxidase gene in patients with permanent congenital hypothyroidism Eur J Endocrinol 145 19 24 11415848
Anbar M Guttmann S Lweitus Z 1959 The mode of action of perchlorate ions on the iodine uptake of the thyroid gland Int J Appl Radiat Isot 7 87 96 13793274
Bakker B Bikker H Hennekam RC Lommen EJ Schipper MG Vulsma T 2001 Maternal isodisomy for chromosome 2p causing severe congenital hypothyroidism J Clin Endocrinol Metab 86 1164 1168 11238503
Bakker B Bikker H Vulsma T de Randamie JS Wiedijk BM de Vijlder JJM 2000 Two decades of screening for congenital hypothyroidism in the Netherlands: TPO gene mutations in total iodide organification defects (an update) J Clin Endocrinol Metab 85 3708 3712 11061528
Bidart JM Lacroix L Evain-Brion D Caillou B Lazar V Frydman R 2000 Expression of Na+/I- symporter and Pendred syndrome genes in trophoblast cells J Clin Endocrinol Metab 85 4367 4372 11095481
Bikker H Baas F de Vijlder JJM 1997 Molecular analysis of mutated thyroid peroxidase detected in patients with total iodide organification defects J Clin Endocrinol Metab 82 649 653 9024270
Bikker H Den Hartog MT Baas F Gons MH Vulsma T de Vijlder JJM 1994 A 20 base pair duplication in the human thyroid peroxidase gene results in a total iodide organification defect and congenital hypothyroidism J Clin Endocrinol Metab 79 248 252 8027236
Bikker H Vulsma T Baas F de Vijlder JJM 1995 Identification of five novel inactivating mutation in the human thyroid peroxidase gene by denaturing gradient gel electrophoresis Hum Mutat 6 9 16 7550241
Blons H Feldmann D Duval V Messaz O Denoyelle F Loundon N 2004 Screening of SLC26A4 (PDS) gene in Pendred’s syndrome: a large spectrum of mutations in France and phenotypic heterogeneity Clin Genet 66 333 340 15355436
Bogazzi F Raggi F Ultimieri F Campomori A Cosci C Berrettini S 2000 A novel mutation in the pendrin gene associated with Pendred’s syndrome Clin Endocrinol 52 279 285
Bogazzi F Russo D Raggi F Ultimieri F Berrettini S Forli F 2004 Mutations in the SLC26A4 (pendrin) gene in patients with sensorineural deafness and enlarged vestibular aqueduct J Endocrinol Invest 27 430 435 15279074
Brechner R Parkhurst G Humble W Brown M Herman W 2000 Ammonium perchlorate contamination of Colorado River drinking water is associated with abnormal thyroid function in newborns in Arizona J Occup Environ Med 42 777 782 10953814
Campbell C Cucci RA Prasad S Green GE Edeal JB Galer CE 2001 Pendred syndrome, DFNB4, and PDS/SLC26A4 : identification of eight novel mutations and possible genotype–phenotype correlations Hum Mutat 17 403 411 11317356
Coucke PJ Van Hauwe P Everett LA Demirhan O Kabakkaya Y Dietrich NL 1999 Identification of two different mutations in the PDS gene in an inbred family with Pendred syndrome J Med Genet 36 475 477 10874637
Coyle B Coffey R Armour JA Gausden E Hochberg Z Grossman A 1996 Pendred syndrome (goitre and sensorineural hearing loss) maps to chromosome 7 in the region containing the nonsyndromic deafness gene DFNB4 Nat Genet 12 421 423 8630497
Coyle B Reardon W Herbrick JA Tsui LC Gausden E Lee J 1998 Molecular analysis of the PDS gene in Pendred syndrome Hum Mol Genet 7 1105 1112 9618167
Crump C Michaud P Tellez R Reyes C Gonzalez G Montgomery EL 2000 Does perchlorate in drinking water affect thyroid function in newborns or school-age children? J Occup Environ Med 42 603 612 10874653
Dai G Levy O Carrasco N 1996 Cloning and characterization of the thyroid iodide transporter Nature 379 458 460 8559252
De Deken X Wang D Many MC Costagliola S Libert F Vassart G 2000 Cloning of two human thyroid cDNAs encoding new members of the NADPH oxidase family J Biol Chem 275 23227 23233 10806195
De la Vieja A Dohán O Levy O Carrasco N 2000 Molecular analysis of the sodium/iodide symporter: impact on thyroid and extrathyroid pathophysiology Physiol Rev 80 1083 1095 10893432
Dohán O De la Vieja A Paroder V Riedel C Artani M Reed M 2003 The sodium/iodide symporter (NIS): characterization, regulation and medical significance Endocr Rev 24 48 77 12588808
Dupuy C Ohayon R Valent A Noel-Hudson MS Deme D Virion A 1999 Purification of a novel flavoprotein involved in the thyroid NADPH oxidase. Cloning of the porcine and human cDNAs J Biol Chem 274 37265 37269 10601291
Eichler O Hackenthal E 1962 [Secretion and metabolism of perchlorate measured with 36 ClO4 ] Naunyn Schmiedebergs Arch Exp Pathol Pharmakol 243 554 565
Eskandari S Loo DD Dai G Levy O Wright EM Carrasco N 1997 Thyroid Na+ /I symporter. Mechanism, stoichiometry, and specificity J Biol Chem 272 27230 27238 9341168
Everett LA Glaser B Beck JC Idol JR Buchs A Heyman M 1997 Pendred syndrome is caused by mutations in a putative sulphate transporter gene (PDS) Nat Genet 17 411 422 9398842
Fraser GR 1965 Association of congenital deafness with goiter (Pendred’s syndrome) Ann Hum Genet 28 201 249 14304636
Fugazzola L Mannavola D Cerutti N Maghnie M Pagella F Bianchi P 2000 Molecular analysis of the Pendred’s syndrome gene and magnetic resonance imaging studies of the inner ear are essential for the diagnosis of true Pendred’s syndrome J Clin Endocrinol Metab 85 2469 2475 10902795
Fujiwara H Tatsumi K Miki K Harada T Miyai K Taki S 1997 Congenital hypothyroidism caused by a mutation in the Na+ /I– symporter Nature Gen 16 124 125
Fujiwara H Tatsumi K Miki K Harada T Okada S Nose O 1998 Recurrent T354P mutation of the Na+ /I symporter in patients with iodide transport defect J Clin Endocrinol Metab 83 2940 2943 9709973
Fujiwara H Tatsumi K Tanaka S Kimura M Nose O Amino N 2000 A novel V59E missense mutation in the sodium iodide symporter gene in a family with iodide transport defect Thyroid 10 471 474 10907989
Greer MA Goodman G Pleus RC Greer SE 2002 Health effects assessment for environmental perchlorate contamination: the dose-response for inhibition of thyroidal radioiodine uptake in humans Environ Health Perspect 110 927 937 12204829
Greer MA Stott AK Milne KA 1966 Effects of thiocyanate, perchlorate and other anions on thyroidal iodine metabolism Endocrinology 79 237 247 4288428
Hollowell JG Staehling NW Flanders WD Hannon WH Gunter EW Spencer CA 2002 Serum TSH, T(4), and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III) J Clin Endocrinol Metab 87 489 499 11836274
Hollowell JG Staehling NW Hannon WH Flanders DW Gunter EW Maberly GF 1998 Iodine nutrition in the United States. Trends and public health implications: iodine excretion data from National Health and Nutrition Examination Surveys I and III (1971–1974 and 1988–1994) J Clin Endocrinol Metab 83 3401 3408 9768638
Johnsen T Videbaek H Olsen KP 1989 CT-scanning of the cochlea in Pendred’s syndrome Clin Otolaryngol 14 389 393 2582633
Kelada SN Eaton DL Wang SS Rothman NR Khoury MJ 2003 The role of genetic polymorphisms in environmental health Environ Health Perspect 111 1055 1064 12826477
Kirk AB Smith EE Tian K Anderson TA Dasgupta PK 2003 Perchlorate in milk Environ Sci Technol 37 4979 4981 14620826
Kopp P Arseven OK Sabacan L Kotlar T Dupuis J Cavaliere H 1999 A novel mutation in the sodium/iodide symporter gene in the largest family with iodide transport defect J Clin Endocrinol Metab 84 336 341 9920104
Kosugi S Bhayana S Dean HJ 1999 A novel mutation in the sodium/iodide symporter gene in the largest family with iodide transport defect J Clin Endocrinol Metab 84 3248 3253 10487695
Kosugi S Inoue S Matsuda A Jhiang SM 1998b Novel, missense and loss-of-function mutations in the sodium/iodide symporter gene causing iodide transport defect in three Japanese patients J Clin Endocrinol Metab 83 3373 3376 9745458
Kosugi S Okamoto H Tamada A Sanchez-Franco F 2002 A novel peculiar mutation in the sodium/iodide symporter gene in Spanish siblings with iodide transport defect J Clin Endocrinol Metab 87 3830 3836 12161518
Kosugi S Sato Y Matsuda A Ohyama Y Fujieda K Inomata H 1998a High prevalence of T354P sodium/iodide symporter gene in Spanish siblings with iodide transport defect J Clin Endocrinol Metab 83 4123 4129 9814502
Kotani T Umeki K Kawano J Suganuma T Hishinuma A Ieiri T 2003 Partial iodide organification defect caused by a novel mutation of the thyroid peroxidase gene in three siblings Clin Endocrinol 59 198 206
Kotani T Umeki K Yamamoto I Ohtaki S Adachi M Tachibana K 2001 Iodide organification defects resulting from cosegregation of mutated and null thyroid peroxidase alleles Mol Cell Endocrinol 182 61 68 11500239
Lamm SH Doemland M 1999 Has perchlorate in drinking water increased the rate of congenital hypothyroidism? J Occup Environ Med 41 409 411 10337612
Li FX Byrd DM Deyhle GM Sesser DE Skeels MR Katkowsky SR 2000 Neonatal thyroid stimulating hormone level and perchlorate in drinking water Teratology 62 429 431 11091365
Li XC Everett LA Lalwani AK Desmukh D Friedman TB Green ED 1998 A mutation in PDS causes non-syndromic recessive deafness Nat Genet 18 215 217 9500541
Li Z Li FX Byrd D Deyhle GM Sesser DE Skeels MR 2000 Neonatal thyroxine level and perchlorate in drinking water J Occup Environ Med 42 200 205 10693082
Lopez-Bigas N Melchionda S de Cid R Grifa A Zelante L Govea N 2002 Identification of five new mutations of PDS/SLC26A4 in Mediterranean families with hearing impairment Hum Mutat 20 77 78 12112665
Mangklabruks A Billerbeck AE Wajchenberg B Knobel M Cox NJ DeGroot LJ 1991 Genetic linkage studies of thyroid peroxidase (TPO ) gene in families with TPO deficiency J Clin Endocrinol Metab 72 471 476 1671388
Matsuda A Kosugi S 1997 A homozygous missense mutation of the sodium/iodide symporter gene causing iodide transport defect J Clin Endocrinol Metab 82 3966 3971 9398697
Moreno JC Bikker H Kempers MJ van Trotsenburg AS Baas F de Vijlder JJ 2002 Inactivating mutations in the gene for thyroid oxidase 2 (THOX2 ) and congenital hypothyroidism N Engl J Med 347 95 102 12110737
Motzer WE 2001 Perchlorate: problems, detection, and solutions Environ Forensics 2 301 311
Namba A Abe S Shinkawa H Kimberling WJ Usami SI 2001 Genetic features of hearing loss associated with ear anomalies: PDS and EYA1 mutation analysis J Hum Genet 46 518 521 11558900
Nascimento AC Guedes DR Santos CS Knobel M Rubio IG Medeiros-Neto G 2003 Thyroperoxidase gene mutations in congenital goitrous hypothyroidism with total and partial iodide organification defect Thyroid 13 1145 1151 14751036
Nilsson LR Borgfors N Gamstorp I Holst HE Liden G 1964 Nonendemic goitre and deafness Acta Paediatr 53 117 131 14128171
Niu DM Hwang B Chu YK Liao CJ Wang PL Lin CY 2002 High prevalence of a novel mutation (2268 insT) of the thyroid peroxidase gene in Taiwanese patients with total iodide organification defect and evidence for a founder effect J Clin Endocrinol Metab 87 4208 4212 12213873
Pannain S Weiss RE Jackson CE Dian D Beck JC Sheffield VC 1999 Two different mutations in the thyroid peroxidase gene of a large inbred Amish kindred: power and limits of homozygosity mapping J Clin Endocrinol Metab 84 1061 1071 10084596
Park HJ Shaukat S Liu XZ Hahn SH Naz S Ghosh M 2003 Origins and frequencies of SLC26A4 (PDS ) mutations in east and south Asians: global implications for the epidemiology of deafness J Med Genet 40 242 248 12676893
Pohlenz J Medeiros-Neto G Gross JL Silveira SP Knobel M Refetoff S 1997 Hypothyroidism in a Brazilian kindred due to iodide trapping defect caused by a homozygous mutation in the sodium/iodide symporter gene Biochem Biophys Res Commun 240 488 491 9388506
Pohlenz J Rosenthal IM Weiss RE Jhiang SM Burant C Refetoff S 1998 Congenital hypothyroidism due to mutations in the sodium/iodide symporter. Identification of a nonsense mutation producing a downstream cryptic 3’ splice site J Clin Invest 101 1028 1035 9486973
Postel S 1957 Placental transfer of perchlorate and triiodothyronine in the guinea pig Endocrinology 60 53 66 13384383
Prasad S Kolln KA Cucci RA Trembath RC Van Camp G Smith RJ 2004 Pendred syndrome and DFNB4-mutation screening of SLC26A4 by denaturing high-performance liquid chromatography and the identification of eleven novel mutations Am J Med Genet 124A 1 9 14679580
Reardon W Coffey R Chowdhury T Grossman A Jan H Britton K 1999 Prevalence, age of onset, and natural history of thyroid disease in Pendred syndrome J Med Genet 36 595 598 10465108
Reardon W O’Mahoney CF Trembath R Jan H Phelps PD 2000 Enlarged vestibular aqueduct: a radiological marker of Pendred syndrome, and mutation of the PDS gene Q J Med 93 99 104
Rivolta CM Esperante SA Gruneiro-Papendieck L Chiesa A Moya CM Domene S 2003 Five novel inactivating mutations in the thyroid peroxidase gene responsible for congenital goiter and iodide organification defect Hum Mutat 22 259 12938097
Santos CL Bikker H Rego KGM Nascimento AC Tambascia M De Vijlder JJM 1999 A novel mutation in the TPO gene in goitrous hyperthyroid patients with iodide organification defect Clin Endocrinol 51 165 172
Schwartz J 2001. Gestational Exposure to Perchlorate Is Associated with Measures of Decreased Thyroid Function in a Population of California Neonates [Master’s thesis]. Berkeley, CA:University of California Berkeley.
Scott DA Wang R Kreman TM 2000 Functional differences of the PDS gene product are associated with phenotypic variation in patients with Pendred syndrome and non-syndromic hearing loss (DFNB4) Hum Mol Genet 9 1709 1715 10861298
Sheffield VC Kraiem Z Beck JC Nishimura D Stone EM Salameh M 1996 Pendred syndrome maps to chromosome 7q21–34 and is caused by an intrinsic defect in thyroid iodine organification Nat Genet 12 424 426 8630498
Smith PN Theodorakis CW Anderson TA Kendall RJ 2001 Preliminary assessment of perchlorate in ecological receptors at the Longhorn Army Ammunition Plant (LHAAP), Karnack, Texas Ecotoxicology 10 305 313 11556118
Soldin OP Braverman LE Lamm SH 2001 Perchlorate clinical pharmacology and human health: a review Ther Drug Monit 23 316 331 11477312
Stanbury JB Wyngaarden JB 1952 Effect of perchlorate on the human thyroid gland Metabolism 1 533 539 13002073
Tazebay UH Wapnir IL Levy O Dohan O Zuckier LS Zhao QH 2000 The mammary gland iodide transporter is expressed during lactation and in breast cancer Nat Med 6 871 878 10932223
Tekin M Akcayoz D Comak E Bogoclu G Duman T Fitoz S 2003 Screening the SLC26A4 gene in probands with deafness and goiter (Pendred syndrome) ascertained from a large group of students of the schools for the deaf in Turkey Clin Genet 64 371 374 12974744
Tonacchera M Agretti P De Marco G Elisei R Perri A Ambrogini E 2003 Congenital hypothyroidism due to a new deletion in the sodium/iodide symporter protein Clin Endocrinol 59 500 506
Tonacchera M Pinchera A Dimida A Ferrarini E Agretti P Vitti P 2004 Relative potencies and additivity of perchlorate, thiocyanate, nitrate, and iodide on the inhibition of radioactive iodide uptake by the human sodium iodide symporter Thyroid 12 1012 1019 15650353
Tsukamoto K Suzuki H Harada D Namba A Abe S Usami S 2003 Distribution and frequencies of PDS (SLC26A4 ) mutations in Pendred syndrome and nonsyndromic hearing loss associated with enlarged vestibular aqueduct: a unique spectrum of mutations in Japanese Eur J Hum Genet 11 916 922 14508505
Umeki K Kotani T Kawano J Suganuma T Yamamoto I Aratake Y 2002 Two novel missense mutations in the thyroid peroxidase gene, R665W and G771R, result in a localization defect and cause congenital hypothyroidism Eur J Endocrinol 146 491 498 11916616
Umeki K Yamamoto I Yukizane S Kotani T 2004 Congenital hypothyroidism caused by a unique thyroid peroxidase allele containing two mutations, C1708T and C2737T J Pediatr Endocrinol Metab 17 231 234 15055360
Usami S Abe S Weston MD Shinkawa H Van Camp G Kimberling WJ 1999 Non-syndromic hearing loss associated with enlarged vestibular aqueduct is caused by PDS mutations Hum Genet 104 188 192 10190331
van den Hove MF Beckers C Devlieger H de Zegher F De Nayer P 1999 Hormone synthesis and storage in the thyroid of human preterm and term newborns: effect of thyroxine treatment Biochimie 81 563 570 10403191
Van Hauwe P Everett LA Coucke P Scott DA Kraft ML Ris-Stalpers C 1998 Two frequent missense mutations in Pendred syndrome Hum Mol Genet 7 1099 1104 9618166
van Vliet G 1999 Neonatal hypothyroidism: treatment and outcome Thyroid 9 79 84 10037081
Vulsma T de Vijlder JIM 2000. Thyroid disease in newborns, infants and children. In: Oxford Textbook of Endocrinology and Diabetes (Wass JAH Shalet SM, eds). Oxford, UK:Oxford University Press, 532–544.
Vulsma T Gons MH de Vijlder JJ 1989 Maternal-fetal transfer of thyroxine in congenital hypothyroidism due to a total organification defect or thyroid agenesis N Engl J Med 321 13 16 2733742
Wolff J 1998 Perchlorate and the thyroid gland Pharmacol Rev 50 89 105 9549759
Wu JY Shu SG Yang CF Yang CF Lee CC Tsai FJ 2002 Mutation analysis of thyroid peroxidase gene in Chinese patients with total iodide organification defect: identification of five novel mutations J Endocrinol 172 627 635 11874711
Wyngaarden JB Stanbury JB Rapp B 1953 The effects of iodide, perchlorate, thiocyanate and nitrate administration upon the iodide concentrating mechanism of the rat thyroid Endocrinology 52 568 574 13060263
Yong AM Goh SS Zhao Y Eng PH Koh LK Khoo DH 2001 Two Chinese families with Pendred’s syndrome: radiological imaging of the ear and molecular analysis of the pendrin gene J Clin Endocrinol Metab 86 3907 3911 11502831
Yoshida A Taniguchi S Hisatome I Royaux IE Green ED Kohn LD 2002 Pendrin is an iodide-specific apical porter responsible for iodide efflux from thyroid cells J Clin Endocrinol Metab 87 3356 3361 12107249
Zoeller RT 2003 Challenges confronting risk analysis of potential thyroid toxicants Risk Anal 23 143 162 12635729
Zoeller RT Dowling AL Herzig CT Iannacone EA Gauger KJ Bansal R 2002 Thyroid hormone, brain development, and the environment Environ Health Perspect 110 suppl 3 355 361 12060829
|
16263499
|
PMC1310906
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 29; 113(11):1479-1484
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8076
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7562ehp0113-00148516263500ResearchPersonal Exposure to Ultrafine Particles and Oxidative DNA Damage Vinzents Peter S. 1Møller Peter 1Sørensen Mette 1Knudsen Lisbeth E. 1Hertel Ole 2Jensen Finn Palmgren 2Schibye Bente 3Loft Steffen 11 Department of Occupational and Environmental Health, University of Copenhagen, Copenhagen, Denmark2 National Environmental Research Institute, Copenhagen, Denmark3 National Institute of Occupational Health, Copenhagen, DenmarkAddress correspondence to P.S. Vinzents, Department of Occupational and Environmental Health, University of Copenhagen, Øster Farimagsgade 5, opg. B, 2. Sal, Postbox 2099, DK-1014, Copenhagen N, Denmark. Telephone: 45-35-32-76-55. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 31 5 2005 113 11 1485 1490 8 12 2004 31 5 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Exposure to ultrafine particles (UFPs) from vehicle exhaust has been related to risk of cardiovascular and pulmonary disease and cancer, even though exposure assessment is difficult. We studied personal exposure in terms of number concentrations of UFPs in the breathing zone, using portable instruments in six 18-hr periods in 15 healthy nonsmoking subjects. Exposure contrasts of outdoor pollution were achieved by bicycling in traffic for 5 days and in the laboratory for 1 day. Oxidative DNA damage was assessed as strand breaks and oxidized purines in mononuclear cells isolated from venous blood the morning after exposure measurement. Cumulated outdoor and cumulated indoor exposures to UFPs each were independent significant predictors of the level of purine oxidation in DNA but not of strand breaks. Ambient air concentrations of particulate matter with an aero-dynamic diameter of ≤10 μm (PM10), nitrous oxide, nitrogen dioxide, carbon monoxide, and/or number concentration of UFPs at urban background or busy street monitoring stations was not a significant predictor of DNA damage, although personal UFP exposure was correlated with urban background concentrations of CO and NO2, particularly during bicycling in traffic. The results indicate that biologic effects of UFPs occur at modest exposure, such as that occurring in traffic, which supports the relationship of UFPs and the adverse health effects of air pollution.
comet assayexposureoxidative DNA damagepersonaltrafficultrafine particles
==== Body
Epidemiologic studies have associated exposure to ambient air particulate matter (PM) with pulmonary and cardiovascular diseases and cancer (Brunekreef and Holgate 2002; Pope et al. 2002). To date, the majority of studies have dealt with the relationship between health outcomes and the ambient levels of PM10 and PM2.5, which are the mass of particles with a aerodynamic diameters ≤10 and 2.5 μm, respectively. Recently, however, interest has focused on the ultrafine particle (UFP) fraction with a diameter ≤100 nm, which are abundant in numbers but contribute little to particle mass. Mechanistically, UFPs are important because of the adverse health effects caused by their high alveolar deposition fraction, large surface area, chemical composition, ability to induce inflammation, and potential to translocate to the circulation (Donaldson et al. 2001; Donaldson and Tran 2002; Nemmar et al. 2002, 2004; Schins et al. 2004). Vehicle emissions, particularly those related to diesel engines, are a major source of ambient UFPs, which penetrate to the indoor environment (Franck et al. 2003; Levy et al. 2002).
A few epidemiologic studies have associated daily changes in number concentrations of UFPs measured at a single urban background monitoring station with daily cardiovascular and pulmonary mortality as well as lung function or use of medicine among patients with asthma (Ibald-Mulli et al. 2002; Penttinen et al. 2001; Peters et al. 1997; von Klot et al. 2002; Wichmann et al. 2000). However, the relationship between number concentrations in urban background and personal exposure to UFPs is not known, and direct links between ambient UFPs and health effects have not been established. Because people spend around 90% of their time indoors (Jenkins et al. 1992), it is widely recognized that a significant proportion of personal exposure to particles occurs in the indoor environment. Indoor UFPs consist of a combination of ambient particles that readily penetrate buildings and infiltrate indoor air (Franck et al. 2003; Levy et al. 2002; Long et al. 2001a; Ozkaynak et al. 1996) and nonambient particles generated indoors during the daily activities of home occupants. Major indoor sources of UFPs include smoking, cooking, candle burning, and other combustion-related processes as well as chemical reactions between, for example, terpenes and ozone (Abt et al. 2000; Dennekamp et al. 2001; Levy et al. 2000; Long et al. 2000; Ozkaynak et al. 1996).
Personal monitors can be used to measure individual exposure. By means of biomarkers based on putative mechanisms of action, exposure can be related to biologic effects, allowing substantiation of causal relationships and identification of relevant sources and exposure scenarios. The mechanisms of action of adverse health effects of PM are based on experimental studies thought to involve induction of inflammation and oxidative stress (Donaldson et al. 2001; Donaldson and Tran 2002; Knaapen et al. 2004; Schins et al. 2004). The generation of oxidative stress may involve radicals and soluble transition metals on the surface of UFPs and activation of production of reactive oxygen species in macrophages, granulocytes, and target cells as well as redox cycling of quinone metabolites of polyaromatic hydrocarbons. In this context UFPs appear more potent than fine or coarse particles per unit mass (Brown et al. 2000, 2001). Experimental studies in vivo and in vitro point to DNA oxidation as an important target of UFPs and fine-fraction PM (Brown et al. 2000, 2001; Dybdahl et al. 2003; Knaapen et al. 2004; Risom et al. 2003a; Schins et al. 2002). Recently, we have shown significant relationships between individual exposure to PM2.5, assessed as mass collected on filters over 48 hr, and biomarkers of oxidative damage to DNA bases in terms of 8-oxodeoxyguanosine (8-oxodG), proteins, and lipids among healthy subjects (Sørensen et al. 2003a, 2003b, 2003c). However, this exposure measurement cannot discriminate between indoor and out-door exposure, and ambient PM2.5 mass is influenced by long-range transport of nitrate-and sulfate-based fine particles (Ruuskanen et al. 2001).
Because UFPs are ubiquitous, even in indoor environments, exposure is unavoidable, and only levels of exposure can be compared. In the present cross-over study, time-resolved personal exposure to traffic-and indoor-related UFPs was assessed by portable equipment and related to oxidative DNA damage in mononuclear blood cells on 6 different days in 15 subjects after low-intensity bicycling exercise in traffic or indoors. Measurements with outdoor bicycling were repeated on 5 days in order to have variation in outdoor exposure for each individual due to differences in traffic density and meteorologic conditions. The control of outdoor exposure and the wide gradient for each subject allowed study of dose–response relationship and comparison of the contribution of outdoor exposure and indoor exposure. We also assessed personal exposure and DNA damage in relation to ambient concentrations of air pollutants measured at two curbside monitoring stations on busy streets and at one urban background station.
Materials and Methods
Personal monitoring.
Fifteen healthy non-smoking subjects, 10 males (25.3 ± 3.5 mean years of age, ± SD) and 5 females (25.4 ± 1.5 years) participated in the study after giving informed consent. The local ethics committee approved the study. In a cross-over design with subjects serving as their own control, personal exposure to UFPs was measured for 18 hr on weekdays six times for each person in the period from March through June 2003. Two subjects were studied simultaneously on each occasion. Condensation particle counters (TSI 3007; TSI, St. Paul, MN, USA) with continuous measurement of the number concentrations of UFPs (10–100 nm) were carried in backpacks with the inlet tube placed in the breathing zone. The instruments were equipped with external batteries, and the subjects were trained to supply them with 2-propanol every 8 hr. The instruments count particles optically after they have grown in size in an atmosphere saturated with 2-propanol, which must be supplied at 8-hr intervals. Time series of 1 min average concentrations were logged during each day. For practical reasons the sampling was interrupted during the night. Two data sets were lost because of technical errors. Exposure was referred to as number concentration of UFPs per milliliter. Cumulated exposure was defined as average concentration multiplied by time with minute as time unit; that is, the unit of cumulated exposure was minutes × UFPs per milliliter (for convenience, tables and figures display units of 106 minutes × UFPs/mL). The particle counters were validated by showing strong correlations in measurements (both instruments: r > 0.999, n = 13) when compared with a TSI 3010 stationary particle counter (TSI) on aerosols of NaCl in 10–20 nm size ranges from 20 to 200 nm and the regression lines going through the origin. Comparison of the two employed TSI 3007 instruments showed a constant difference in counting efficiency of 8.9%, which was corrected for. With this correction the two instruments also gave identical results when carried by different subjects that bicycled the exposure route together (data not shown).
Five of the 6 days of exposure measurement included bicycling in central Copenhagen on a 20-km predefined route during morning and/or afternoon rush hours. The mean bicycling time was 93 ± 15 min. This allowed study of dose–response relationships associated with variations in outdoor exposure for each individual due to differences in traffic density and meteorologic conditions. One exposure measurement day included the same workload at the same intensity on an ergometer bicycle in a room at the Panum Institute (Copenhagen, Denmark) with air intake away from streets and minimal number concentrations of UFPs.
The relationship between heart rate and workload was established for each subject from an ergometer bicycle test, and the average workloads during traffic bicycling on each of the 5 days were calculated from the average heart rates measured during traffic bicycling. Increased pulmonary ventilation will increase the deposition possibility of UFPs dependent on the breathing pattern (Daigle et al. 2003). A conservative estimate is achieved by assuming proportionality between increased pulmonary ventilation and increased deposition (D) of UFPs. Because pulmonary ventilation during moderate dynamic exercise increases linearly with work intensity (P), the increased UFP deposition during traffic bicycling compared with UFP deposition during rest or light exercise (P = 60 W ) can be found as:
Individual values of increased pulmonary deposition of UFPs during traffic bicycling were estimated, and cumulated personal traffic exposure was adjusted. The average estimated increase in deposition was a factor of 1.43 ± 0.37) (n = 67). The subjects kept a diary for recording periods of bicycling, other outdoor activities, and indoor time and activities, including exposure to cooking fumes, burning candles, and environmental tobacco smoke. The subjects were asked to keep the latter exposures at the lowest possible level. The distribution of time spent on outdoor and indoor activities is shown in Table 1.
Determination of oxidative DNA damage.
In the morning after each exposure measurement day, mononuclear blood cells were isolated from venous blood samples in Vacutainer CPT tubes with sodium heparin tubes (Becton Dickinson and Company, Rutherford, NJ, USA) and centrifuged at 1,650 × g for 20 min at room temperature. The cell layer was obtained and washed in cold RPMI medium from Gibco (Grand Island, NY, USA) and centrifuged at 400 × g for 15 min at 4°C. Most of the supernatant was removed, and the pellet was resuspended in cold preservation medium with volume-percent as follows: 40% RPMI, 50% fetal bovine serum (Gibco), and 10% dimethyl sulfoxide (AppliChem, Darmstadt, Germany). The samples were stored at −80°C for later analysis. Oxidative DNA damage was determined by single-cell gel electrophoresis (comet assay) as strand breaks and base damage in terms of sites sensitive to formamidopyrimidine glycosylase (FPG), which cleaves DNA at sites of oxidized purines and mainly detects 8-oxodG (Collins et al. 1997; Sørensen et al. 2003d). Briefly, cells were thawed on ice, embedded in 0.75% low-melting-point agarose (Sigma, Copenhagen, Denmark) on Gelbond films (BioWhittaker Molecular Applications, Rockland, ME, USA), and lysed for a minimum of 1 hr at 4°C (2.5 M NaCl; 0.1 M EDTA; 10 mM Tris, base pH 10; 1% Triton X-100). The gel-embedded nuclei were washed 3 × 5 min in cold buffer (40 mM HEPES, 0.1 M KCl, 0.5 mM EDTA, 0.2 mg/mL bovine serum albumin, pH 8) to remove the lysis solution. The FPG-sensitive sites were detected by incubation of the agarose-embedded nuclei with 1 μg/mL FPG protein (kindly provided by A. Collins, University of Olso, Oslo, Sweden) for 45 min at 37°C. The nuclei were subsequently treated in alkaline solution (300 mM NaOH, 1 mM EDTA, pH > 13) for 40 min and electrophoresed in the same solution at 4°C for 20 min at 25 V and 300 mA. The level of DNA damage in each sample was scored in 100 nuclei according to a five-class system (range of score is 0–400). The net level of FPG-sensitive sites was obtained as the difference in score between samples incubated with FPG protein and buffer. This score was translated into lesions per 106 base pairs (bp) by means of a calibration curve based on induction of strand breaks by X ray, which has a known yield [European Standards Committee on Oxidative DNA Damage (ESCODD) 2003; Møller et al. 2004a]. We have used a conversion factor of 0.056 Gy equivalents per score, or 0.011 lesions/106 bp per score (assuming that a human diploid cell contains 4 × 1012 Da DNA, corresponding to 6 × 109 bp). All samples from a subject were coded and analyzed simultaneously in duplicate, minimizing effects of interassay variation. The method has been validated in interlaboratory trials and is believed to be free from artifactual oxidation problems (ESCODD 2003).
Fixed station monitoring.
Ambient concentrations of air pollutants were measured on all exposure days at two curbside busy street stations along the bicycling route and at one urban background station on a rooftop at 20 m height approximately 500 m from the start and end of the bicycling route. Ambient air concentrations of nitric oxide, nitrogen dioxide, carbon monoxide, and PM10 were measured continuously with a 1-hr time resolution by standard methods at all stations under the Danish Air Quality Monitoring Programme (Kemp and Palmgren 2004). The instruments used for PM10 measurements were the tapered element oscillating microbalance (series 1400a ambient particulate monitor; Rupprecht & Patashnick, East Greenbush, NY, USA). The instrument was approved by the U.S. Environmental Protection Agency for PM10 ambient particulate monitoring. The mass measurements were performed at 50°C to stabilize the water content of the particles, but at the same time other volatile compounds, for example, ammonium nitrate and organic volatiles, will be lost. One street station also measured size-fractionated number concentrations of UFPs by a scanning mobility particle sizer (Palmgren et al. 2003). Temperature, relative humidity, and wind speed were recorded at the urban background station.
Statistical analysis.
Statistical analysis of DNA damage was carried out by means of mixed-effects models, which allow both random and fixed effects. The subject level was a random factor, and cumulated exposure to UFPs occurring during bicycling, remaining time outdoors and indoors, and monitoring station values were tested as potential predictor variables with fixed effects. The effect of bicycling indoors or outdoors on total exposure to UFPs and DNA damage was also assessed by two-factorial analysis of variance, including subject as factor. The DNA damage and personal exposure variables were cubic root transformed before analysis to achieve normal distributions. Similarly, in another analysis the relationship between personal log-transformed exposure occurring outdoors during bicycling and other activities, or indoors, and 24-hr average monitoring station log-transformed measurements was analyzed by linear mixed-effects models with subject level as random factor. SPSS (version 11.0; SPSS Inc., Chicago, IL, USA) was used for analysis.
Results
Typical 18-hr personal exposure profiles are shown in Figure 1. Peak concentration of indoor UFPs usually coincided with presence of indoor sources such as cooking, burning candles, or environmental tobacco smoke recorded in the subjects’ diaries. The exposure during bicycling in traffic was significantly inversely correlated with air temperature and wind speed as well as directly correlated with the measured concentrations of ambient pollutants at both background and street monitoring stations (Table 2). Weaker but significant correlations were found between indoor UFP exposure and air temperature (inverse) and concentrations of NO2 (background station) and CO (background station and street station) and between UFP exposure during other outdoor activities and air temperature and CO concentrations (Table 2).
In linear mixed-effects models with subjects as a random factor, background monitoring station measurements of ambient temperature and CO concentration, and ambient temperature and NO2 concentration at one of the street stations were the only significant predictors of UFP exposure during bicycling in traffic (R2 = 0.60 and R2 = 0.74, respectively). In contrast, air temperature was the only significant predictor of UFP exposure during other outdoor activities (R2 = 0.09), and background concentration of CO was the only significant predictor of indoor UFP exposure (R2 = 0.11).
Bicycling in traffic increased the cumulated exposure to UFPs significantly, although indoor exposure contributed more because of the much longer time spent indoors (Table 3).
After bicycling in traffic the level of oxidative DNA base damage in terms of FPG-sensitive sites was increased 4-fold (p < 0.001) compared with the level measured after bicycling indoors, but there was no effect on DNA strand breaks (Table 3, Figure 2). The level of FPG-sensitive sites (per 106 bp) was significantly predicted by the personal cumulated exposure to UFPs with independent contributions from outdoor and indoor observation periods. The regression coefficients of the mixed-effects model of level of DNA damage, including both outdoor and indoor exposures, with subjects as random factor, were estimated as 1.50 × 10−3 [95% confidence interval (CI), 0.59 × 10−3 to 2.42 × 10−3; p = 0.002] for cumulated outdoor exposure and 1.07 × 10−3 (95% CI, 0.37 × 10−3 to 1.77 × 10−3; p = 0.003) for cumulated indoor exposure.
The level of DNA damage and the cumulated exposure were cubic root transformed before the mixed-effects model analysis. The model explained 50.3% (R2) of the variation, and the residuals were randomly and normally distributed as confirmed by nonparametric tests (Runs test and Kolmogorov-Smirnov test). The regression coefficient should in principle describe the dose–response relationship, although they are not easy to interpret in absolute numbers because of the cubic root transformations. The levels of DNA damage were not significantly associated with any 24-hr average concentration of ambient air pollutants measured at a monitoring station (Pearson’s r < 0.303).
Discussion
In this study oxidative DNA base damage in circulating mononuclear blood cells was associated with personal exposure to UFPs, and short-term higher intensity exposure in traffic was associated with elevated levels of damage. Cumulated outdoor and indoor exposures contributed independently to the association, which showed clear dose–response relationships. The level of damage was not associated with ambient concentrations of air pollutants at a monitoring station, although the concentrations of several of these were associated with personal UFP exposure during bicycling, in particular.
Oxidative DNA damage is mutagenic and carcinogenic per se and may be considered a biomarker of oxidative stress, which is also thought to be involved in cardiovascular and pulmonary disease due to UFPs (Brown et al. 2001; Donaldson et al. 2001; Li et al. 2003; Schins et al. 2004). After indoor bicycling the level of DNA damage was very low and at a level corresponding to well-nourished healthy volunteers with minimum exposures (Møller and Loft 2004). This low level could be assessed with good precision by an X-ray–calibrated visual scoring system, which we find more sensitive than computer-based image analysis (Møller et al. 2004a). The increase in FPG-sensitive sites in DNA of median 0.06 per 106 bp in circulating mononuclear cells after outdoor bicycling would require a radiation dose of approximately 0.14 Gy to induce, assuming a yield of 0.43 FPG sites per 106 bp/Gy, as found in mice in vivo (Risom et al. 2003b). However, radiation induces many types of DNA damage, and this comparison cannot be used for risk characterization. We have previously found a significant association between oxidative DNA base damage, without changes in strand breaks, and personal exposure to PM in terms of PM2.5 measured as mass over 48 hr in young healthy subjects in Copenhagen (Sørensen et al. 2003b). In that study DNA damage was assessed at the end of the monitoring period, similar to the design in the present study. The lack of measurable effects of PM on DNA strand breaks may be due to the very rapid repair by ligases, whereas guanine oxidation is repaired relatively slowly by base excision followed by strand nicking, insertion of nucleotide(s) in the gap, and rejoining by ligases (Hoeijmakers 2001; Risom et al. 2003b). Indeed, DNA base oxidation has been found to be much more sensitive than strand breaks to environmental factors, including several types of air pollution, smoking, and antioxidant intervention (Avobge et al. 2005; Møller and Loft 2002, 2004; Møller et al. 2004b; Sørensen et al. 2003d). In a mouse study the level of oxidized guanine in lung DNA was increased, whereas strand breaks were unchanged 1 and 24 hr after inhalation of diesel exhaust particles (Risom et al. 2003a).
Similar to our findings for DNA base oxidation in the present and a previous study (Sørensen et al. 2003b), we have also found significant associations between personal exposure to black smoke, measured as reflectance of material collected on PM2.5 filters, and oxidation of plasma proteins, and a similar association between the mass of the filter material and lipid peroxidation in plasma, although the latter was significant only among women (Sørensen et al. 2003c). However, the cumulated exposure measurement in the previous studies did not allow assessment of effects of UFPs and distinction between outdoor and indoor sources (Sørensen et al. 2003b, 2003c). Staying outdoors in traffic, particularly during bicycling, provided higher intensity of exposure for limited periods of time, whereas staying indoors provided prolonged periods of generally low-intensity exposure, although with some activity-related peaks. Vehicle exhaust is the main source of outdoor UFPs, which can penetrate indoors where additional sources include environmental tobacco smoke, cooking, burning of candles, and chemical reactions (Abt et al. 2000; Dennekamp et al. 2001; Levy et al. 2000; Long et al. 2000; Ozkaynak et al. 1996). The parameter estimate of the mixed-effects model describing the level of DNA damage in relation to exposure to UFPs was nominally larger for outdoor than for indoor exposure. This could suggest larger potency of the outdoor UFPs, compared with indoor UFPs, possibly by a factor of 3 considering the cubic root transformations. The personal UFP monitors we used would also measure liquid droplets in the 10–100 nm size range, which could be particularly abundant during, for example, cooking and could have limited toxicologic potential. However, the 95% CIs had considerable overlap, and no firm conclusion can be drawn. Moreover, the particles we measured in numbers indoors or outdoors could not be characterized in other aspects that could have indicated causal components. Nevertheless, diesel exhaust particles have consistently been shown to induce 8-oxodG in experimental animals and in vitro (Brown et al. 2000, 2001; Dybdahl et al. 2003; Knaapen et al. 2004; Risom et al. 2003a; Schins et al. 2002). Moreover, UFPs can be translocated to the circulation upon inhalation and may interact directly with circulating mononuclear cells, possibly explaining the DNA base oxidation found in the present study (Donaldson et al. 2001; Donaldson and Tran 2002; Nemmar et al. 2002, 2004; Schins et al. 2004; Semmler et al. 2004). The toxicity of indoor particles has only been assessed for PM2. and coarse particles with respect to inflammatory potential in vitro, and the potential for inducing DNA damage is unknown, and indoor UFPs have yet to be investigated (Long et al. 2001b; Monn and Becker 1999; Roponen et al. 2003). Other studies with exposure assessment based on residence or occupation in urban areas also point to an association between ambient air pollution and oxidative DNA damage, for example, in nasal biopsies and leukocytes of subjects in Mexico City or in urine from bus drivers in Copenhagen (Calderon-Garciduenas et al. 1996, 1999; Fortoul et al. 2003, 2004; Loft et al. 1999)
Our subjects performed modest exercise in terms of bicycling at moderate speed. This increases internal exposure to UFPs by increasing both ventilation and probably lung deposition, as shown recently (Daigle et al. 2003). We took into account the increased ventilation in our exposure assessment by calculations based on the increases in heart rate at fixed workloads. Without this correction outdoor UFPs would have appeared even more potent with respect to induction of DNA base damage. We did not take into account a possible increase in the fractional deposition during outdoor bicycling caused by a change in the breathing pattern. This may also explain the possible higher potency of outdoor UFPs.
Personal exposure to UFPs when bicycling in traffic was inversely related to temperature and wind speed, which is consistent with increased formation through condensation of gases at low temperatures and dispersion by wind. Ambient concentration of UFPs and CO measured at street stations were the strongest predictors of outdoor personal UFP exposure during bicycling, which is consistent with traffic as the major source (Palmgren et al. 2003). Similarly, CO was the strongest predictor measured in urban background. The UFP exposure during other outdoor activities and indoor exposure were less strongly associated with 24-hr monitoring station measurements, and none of the monitoring station measurements was significantly associated with the level of oxidative DNA damage. Compared with direct measurement of personal exposure, monitoring stations measurements are poorer predictors of both exposure and biologic effects. Nevertheless, the significant association between CO concentrations in urban background and personal exposure to indoor UFPs supports that traffic emissions have some contribution to indoor UFPs.
This study design, including direct measurement of personal exposure and traffic-related contrasts, has proved promising in demonstrating association between UFPs and biologic effects in terms of oxidative DNA base damage. The results support the importance of UFPs in causing health effects related to generation of oxidative stress by air pollutants. Moreover, concern about the health effects of even small high-intensity exposures of UFPs in ambient air may be relevant.
The study was supported by the Danish Environmental Protection Agency and the Research Centre for Environmental Health under the Danish Ministry of the Interior and Health.
Figure 1 Examples of personal UFP exposure profiles on a laboratory (A) and traffic (B) bicycling day. Indoor and outdoor periods and activities are marked.
Figure 2 Relationship between oxidative DNA base damage as FPG lesions in mononuclear blood cells on the morning after exposure and exposure to UFPs during 5 days of bicycling in traffic (open circles) and 1 day of bicycling in the laboratory (solid circles) in 15 healthy subjects. One data point at (x, y) = (12 × 106, 0.62) is omitted from the figure to limit the scale. Indoor and outdoor exposures to UFPs were significant independent predictors of the FPG lesions in a mixed-effects model (R2 = 0.503).
Table 1 Distribution of time (min) as mean ± SD spent in traffic, outdoors, and indoors on six occasions in each of 15 healthy subjects.
Bicycling (days) Time bicycling on designated route Time bicycling elsewhere Time outdoors not bicycling Time indoors
In traffic (n = 74) 93 ± 15 7 ± 21 62 ± 66 751 ± 65
Indoors (n = 14) — 22 ± 21 59 ± 59 837 ± 62
Table 2 Geometric means (GM) and geometric SDs (GSD) of air pollutants concentrations, and partial correlation (subject controlled) between meteorologic conditions and ambient log-transformed concentrations of air pollutants measured as 24-hr averages at monitoring stations against personal exposure to UFPs for 15 subjects, each measured on five or six occasions.
Measure GM (GSD) na Bicycling on exposure route (5 occasions) Other outdoor activities (6 occasions) Indoors (6 occasions)
UFPs (personal exposure)
GM (GSD) na — 32.4b (1.49) 74 19.6b (1.78) 84 13.4b (1.96) 89
Correlations
Background station
Temperature — −0.619* −0.300* −0.320*
Wind speed — −0.516* −0.145 −0.132
NOx 13.4c (1.61) 73 0.439* 0.207 0.259
NO2 11.3c (1.52) 73 0.454* 0.237 0.293*
CO 273c (1.35) 73 0.651* 0.317* 0.371*
PM10 16.9c (1.53) 75 0.290 0.126 0.193
Street station 1
UFPs 30.4b (1.38) 75 0.493* 0.179 0.255
NOx 72.4c (1.44) 72 0.486* 0.193 0.105
NO2 32.1c (1.31) 72 0.394* 0.147 0.118
Street station 2
NOx 51.7c (1.76) 74 0.444* 0.228 0.226
NO2 24.2c (1.49) 74 0.415* 0.207 0.266
CO 788c (1.52) 74 0.556* 0.289* 0.311*
PM10 23.5c (1.48) 75 0.428* 0.198 0.249
NOx, nitrogen oxide.
a GM (GSD) number of measurements.
b Data are expressed in units of 103 UFPs/mL.
c Data are expressed as μg/m3.
* Significant correlations at the 0.01% level (two-tailed).
Table 3 Median and interquartile range of cumulated exposure to UFPs and oxidative DNA damage as FPG lesions and strand breaks (SB) in 15 subjects bicycling in traffic or indoors, on six occasions.
Cumulated exposure to UFPs (106 min × UFPs/mL)
DNA damage (per 106 bp)
Bicycling (days) Traffic bicycling Remaining time outdoors Time indoors FPG SB
In traffic (n = 74) 3.01a (2.25–4.44) 1.54a (0.68–3.28) 10.5a (5.86–16.7) 0.08b (0.04–0.12) 0.06 (0.03–0.11)
Indoors (n = 14) — 1.42 (0.52–2.41) 9.20 (6.15–13.1) 0.02 (0–0.04) 0.06 (0.02–0.12)
a Total UFP exposure (sum) increased compared with day with indoor bicycling (p = 0.004).
b DNA damage increased compared with day with indoor bicycling (p = 0.0003).
==== Refs
References
Abt E Suh HH Catalano PJ Koutrakis P 2000 Relative contribution of outdoor and indoor particle sources to indoor concentrations Environ Sci Technol 34 3579 3587
Avobge P Ayi-Fanou L Autrup H Loft S Fayomi B Sanni A 2005 Ultrafine particulate matter and high-level benzene urban air pollution in relation to oxidative DNA damage Carcinogenesis 26 613 620 15591089
Brown DM Stone V Findlay P MacNee W Donaldson K 2000 Increased inflammation and intracellular calcium caused by ultrafine carbon black is independent of transition metals or other soluble components Occup Environ Med 57 685 691 10984341
Brown DM Wilson MR MacNee W Stone V Donaldson K 2001 Size-dependent proinflammatory effects of ultrafine poly-styrene particles: a role for surface area and oxidative stress in the enhanced activity of ultrafines Toxicol Appl Pharmacol 175 191 199 11559017
Brunekreef B Holgate ST 2002 Air pollution and health Lancet 360 1233 1242 12401268
Calderon-Garciduenas L Osnaya-Brizuela N Ramirez-Martinez L Villarreal-Calderon A 1996 DNA strand breaks in human nasal respiratory epithelium are induced upon exposure to urban pollution Environ Health Perspect 104 160 168 8820583
Calderon-Garciduenas L Wen-Wang L Zhang YJ Rodriguez-Alcaraz A Osnaya N Villarreal-Calderon A 1999 8-Hydroxy-2′-deoxyguanosine, a major mutagenic oxidative DNA lesion, and DNA strand breaks in nasal respiratory epithelium of children exposed to urban pollution Environ Health Perspect 107 469 474 10339447
Collins AR Dobson VL Dusinska M Kennedy G Stetina R 1997 The comet assay: what can it really tell us? Mutat Res 375 183 193 9202728
Daigle CC Chalupa DC Gibb FR Morrow PE Oberdorster G Utell MJ 2003 Ultrafine particle deposition in humans during rest and exercise Inhal Toxicol 15 539 552 12692730
Dennekamp M Howarth S Dick CAJ Cherrie JW Donaldson K Seaton A 2001 Ultrafine particles and nitrogen oxides generated by gas and electric cooking Occup Environ Med 58 511 516 11452045
Donaldson K Stone V Seaton A MacNee W 2001 Ambient particle inhalation and the cardiovascular system: potential mechanisms Environ Health Perspect 109 suppl 4 523 527 11544157
Donaldson K Tran CL 2002 Inflammation caused by particles and fibers Inhal Toxicol 14 5 27 12122558
Dybdahl M Risom L Møller P Autrup H Wallin H Vogel U 2003 DNA adduct formation and oxidative stress in colon and liver of Big Blue® rats after dietary exposure to diesel particles Carcinogenesis 24 1759 1766 12919963
ESCODD (European Standards Committee on Oxidative DNA) 2003 Measurement of DNA oxidation in human cells by chromatographic and enzymic methods Free Radic Biol Med 34 1089 1099 12684094
Fortoul TI Valverde M Lopez MC Bizarro P Lopez I Sanchez I 2003 Single-cell gel electrophoresis assay of nasal epithelium and leukocytes from asthmatic and nonasthmatic subjects in Mexico City Arch Environ Health 58 348 352 14992309
Fortoul TI Valverde M Lopez MC Vila-Costa MR Vila-Casado MC Mussali-Galante P 2004 Genotoxic differences by sex in nasal epithelium and blood leukocytes in subjects residing in a highly polluted area Environ Res 94 243 248 15016590
Franck U Herbarth O Wehner B Wiedensohler A Manjarrez M 2003 How do the indoor size distributions of airborne submicron and ultrafine particles in the absence of significant indoor sources depend on outdoor distributions? Indoor Air 13 174 181 12756011
Hoeijmakers JH 2001 Genome maintenance mechanisms for preventing cancer Nature 411 366 374 11357144
Ibald-Mulli A Wichmann HE Kreyling W Peters A 2002 Epidemiological evidence on health effects of ultrafine particles J Aerosol Med 15 189 201 12184869
Jenkins PL Phillips TJ Mulberg EJ Hui SP 1992 Activity patterns of Californians: use and proximity to indoor pollutant sources Atmos Environ 26A 2141 2148
Kemp K Palmgren F 2004. The Danish Air Quality Monitoring Programme. Annual Summary for 2003. NERI Technical Report No. 497. Roskilde, Denmark:National Environmental Research Institute. Available: http://technicalreports.DMU.dk [accessed 29 August 2005].
Knaapen AM Borm PJ Albrecht C Schins RP 2004 Inhaled particles and lung cancer. Part A: Mechanisms Int J Cancer 109 799 809 15027112
Levy JI Dumyahn T Spengler JD 2002 Particulate matter and polycyclic aromatic hydrocarbon concentrations in indoor and outdoor microenvironments in Boston, Massachusetts J Expo Anal Environ Epidemiol 12 104 114 11965527
Levy JI Houseman EA Ryan L Richardson D Spengler JD 2000 Particle concentrations in urban microenvironments Environ Health Perspect 108 1051 1057 11102296
Li N Hao MQ Phalen RF Hinds WC Nel AE 2003 Particulate air pollutants and asthma—a paradigm for the role of oxidative stress in PM-induced adverse health effects Clin Immunol 109 250 265 14697739
Loft S Poulsen HE Vistisen K Knudsen LE 1999 Increased urinary excretion of 8-oxo-2′-deoxyguanosine, a biomarker of oxidative DNA damage, in urban bus drivers Mutat Res 441 11 19 10224318
Long CM Suh HH Catalano PJ Koutrakis P 2001a Using time-and size-resolved particulate data to quantify indoor penetration and deposition behavior Environ Sci Technol 35 2089 2099 11393992
Long CM Suh HH Kobzik L Catalano PJ Ning YY Koutrakis P 2001b A pilot investigation of the relative toxicity of indoor and outdoor fine particles: in vitro effects of endotoxin and other particulate properties Environ Health Perspect 109 1019 1026 11689347
Long CM Suh HH Koutrakis P 2000 Characterization of indoor particle sources using continuous mass and size monitors J Air Waste Manag Assoc 50 1236 1250 10939216
Møller P Friis G Risom L Plesner G Kjærsgaard J Vinzents P 2004a Intra-laboratory comet assay sample scoring exercise for determination of formamidopyrimidine DNA glycosylase sites in human mononuclear blood cell DNA Free Radic Res 11 1207 1214 15621698
Møller P Loft S 2004 Interventions with antioxidants and nutrients in relation to oxidative DNA damage and repair Mutat Res 551 79 89 15225583
Møller P Loft S 2002 Oxidative DNA damage in human white blood cells in dietary antioxidant intervention studies Am J Clin Nutr 76 303 310 12144999
Møller P Viscovich M Lykkesfeldt J Loft S Jensen A Poulsen HE 2004b Vitamin C supplementation decreases oxidative DNA damage in mononuclear blood cells of smokers Eur J Nutr 43 267 274 15309445
Monn C Becker S 1999 Cytotoxicity and induction of proinflammatory cytokines from human monocytes exposed to fine (PM2.5 ) and coarse particles (PM10–2.5 ) in outdoor and indoor air Toxicol Appl Pharmacol 155 245 252 10079210
Nemmar A Hoet PH Vanquickenborne B Dinsdale D Thomeer M Hoylaerts MF 2002 Passage of inhaled particles into the blood circulation in humans Circulation 105 411 414 11815420
Nemmar A Hoylaerts MF Hoet PH Nemery B 2004 Possible mechanisms of the cardiovascular effects of inhaled particles: systemic translocation and prothrombotic effects Toxicol Lett 149 243 253 15093270
Ozkaynak H Xue J Spengler J Wallace L Pellizzari E Jenkins P 1996 Personal exposure to airborne particles and metals: results from the Particle TEAM study in Riverside, California J Expo Anal Environ Epidemiol 6 57 78 8777374
Palmgren F Wahlin P Kildeso J Afshari A Fogh CL 2003 Characterisation of particle emissions from the driving car fleet and the contribution to ambient and indoor particle concentrations Phys Chem Earth 28 327 334
Penttinen P Timonen KL Tiittanen P Mirme A Ruuskanen J Pekkanen J 2001 Ultrafine particles in urban air and respiratory health among adult asthmatics Eur Respir J 17 428 435 11405521
Peters A Wichmann HE Tuch T Heinrich J Heyder J 1997 Respiratory effects are associated with the number of ultra-fine particles Am J Respir Crit Care Med 155 1376 1383 9105082
Pope CA III Burnett RT Thun MJ Calle EE Krewski D Ito K 2002 Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution JAMA 287 1132 1141 11879110
Risom L Dybdahl M Bornholdt J Vogel U Wallin H Møller P 2003a Oxidative DNA damage and defence gene expression in the mouse lung after short-term exposure to diesel exhaust particles by inhalation Carcinogenesis 24 1847 1852 12919962
Risom L Møller P Vogel U Kristjansen PEG Loft S 2003b X-ray-induced oxidative stress: DNA damage and gene expression of HO-1 , ERCC1 and OGG1 in mouse lung Free Radic Res 37 957 966 14670003
Roponen M Toivola M Alm S Nevalainen A Jussila J Hirvonen MR 2003 Inflammatory and cytotoxic potential of the airborne particle material assessed by nasal lavage and cell exposure methods Inhal Toxicol 15 23 38 12476358
Ruuskanen J Tuch T Ten Brink H Peters A Khlystov A Mirme A 2001 Concentrations of ultrafine, fine and PM2.5 particles in three European cities Atmos Environ 35 3729 3738
Schins RP Duffin R Hohr D Knaapen AM Shi T Weishaupt C 2002 Surface modification of quartz inhibits toxicity, particle uptake, and oxidative DNA damage in human lung epithelial cells Chem Res Toxicol 15 1166 1173 12230410
Schins RP Lightbody JH Borm PJ Shi T Donaldson K Stone V 2004 Inflammatory effects of coarse and fine particulate matter in relation to chemical and biological constituents Toxicol Appl Pharmacol 195 1 11 14962500
Semmler M Seitz J Erbe F Mayer P Heyder J Oberdorster G 2004 Long-term clearance kinetics of inhaled ultrafine insoluble iridium particles from the rat lung, including transient translocation into secondary organs Inhal Toxicol 16 453 459 15204761
Sørensen M Autrup H Hertel O Wallin H Knudsen LE Loft S 2003b Personal exposure to PM2.5 in an urban environ ment and biomarkers of genotoxicity Cancer Epidemiol Biomarkers Prev 12 191 196 12646506
Sørensen M Autrup H Moller P Hertel O Jensen SS Vinzents P 2003a Linking exposure to environmental pollutants with biological effects Mutat Res 544 255 271 14644327
Sørensen M Dragsted LO Hertel O Knudsen LE Loft S 2003c Personal PM2.5 exposure and markers of oxidative stress in blood Environ Health Perspect 111 161 166 12573899
Sørensen M Skov H Autrup H Hertel O Loft S 2003d Urban benzene exposure and oxidative DNA damage Sci Total Environ 309 69 80 12798093
von Klot S Wolke G Tuch T Heinrich J Dockery DW Schwartz J 2002 Increased asthma medication use in association with ambient fine and ultrafine particles Eur Respir J 20 691 702 12358349
Wichmann HE Spix C Tuch T Wolke G Peters A Heinrich J 2000 Daily mortality and fine and ultrafine particles in Erfurt, Germany. Part I: Role of particle number and particle mass Res Rep Health Eff Inst 98 5 86 11918089
|
16263500
|
PMC1310907
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 31; 113(11):1485-1490
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7562
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8010ehp0113-00149116263501ResearchDistribution of Brevetoxin (PbTx-3) in Mouse Plasma: Association with High-Density Lipoproteins Woofter Ricky T. Spiess Page C. Ramsdell John S. Marine Biotoxins Program, Center for Coastal Environmental Health and Biomolecular Research, National Oceanic and Atmospheric Administration–National Ocean Service, Charleston, South Carolina, USAAddress correspondence to: J.S. Ramsdell, Coastal Research Branch, Center for Coastal Environmental Health and Biomolecular Research, NOAA-National Ocean Service, 219 Fort Johnson Rd., Charleston, SC 29412. Telephone: (843) 762-8510. Fax: (843) 762-8700. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 23 6 2005 113 11 1491 1496 11 2 2005 23 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. We investigated the brevetoxin congener PbTx-3 to determine its distribution among carrier proteins, including albumin and blood lipoproteins. Using a radiolabeled brevetoxin tracer (PbTx-3), we found that 39% of the radiolabel remained associated with components in mouse plasma after > 15 kDa cutoff dialysis. Of this portion, only 6.8% was bound to serum albumin. We also examined the binding of brevetoxin to various lipoprotein fractions. Plasma, either spiked with PbTx-3 or from mice treated for 30 min with PbTx-3, was fractionated into different-sized lipoproteins by iodixanol gradient ultracentrifugation. Each fraction was then characterized and quantified by agarose gel electrophoresis and brevetoxin radioimmunoassay, respectively. In both the in vitro and in vivo experiments, the majority of brevetoxin immunoreactivity was restricted to only those gradient fractions that contained high-density lipoproteins (HDLs). Independent confirmation of brevetoxin binding to HDLs was provided by high molecular weight (100 kDa cutoff) dialysis of [3H]PbTx-3 from lipoprotein fractions as well as a scintillation proximity assay using [3H]PbTx-3 and purified human HDLs. This information on the association of brevetoxins with HDLs provides a new foundation for understanding the process by which the toxin is delivered to and removed from tissues and may permit more effective therapeutic measures to treat intoxication from brevetoxins and the related ciguatoxins.
bloodbrevetoxinciguatoxinHDLlipoproteinplasma
==== Body
Red tides have been documented on the Gulf Coast of Florida as early as 1530 (Taylor 1917). They occur almost annually and often persist for many months (Woodcock 1948). Red tides are caused by the dinoflagellate Karenia brevis (formerly Gymnodinium breve and Ptychodiscus brevis), which produces a series of polycyclic ethers called brevetoxins (PbTx) (Davis 1948; Lin et al. 1981; Martin and Chatterjee 1969; Poli et al. 1986 ). These events are responsible for fish, waterfowl, and marine mammal mortalities (Landsberg, 2002; Landsberg and Steidinger 1998) as well as for human intoxication. Aerosol forms of the toxin are produced by wind and wave action and move inshore, causing transient respiratory irritation in people that inhale the toxin (Pierce 1990). Humans can also experience the more severe symptoms of neurotoxic shellfish poisoning (NSP) as a result of consuming shellfish that have accumulated brevetoxin (McFarren et al. 1965).
The action of brevetoxins on its pharmacologic target, the voltage-sensitive sodium channel, leads to activation at normal resting potential and inhibition of inactivation (Huang et al. 1984; Sheridan and Adler, 1989; Westerfield et al. 1977). However, before reaching its target tissue, brevetoxin must be absorbed into the general blood supply. Administration of [3H]PbTx-3 orally to rats leads to accumulation of toxin in the liver, stomach, and intestines (Cattet and Geraci 1992). Fat-soluble substances, such as brevetoxins, coalesce into fat droplets in the stomach and are emulsified into mixed micelles by the action of bile acids secreted into the intestine. The mixed micelles are likely absorbed via phagocytosis by absorptive cells of the intestinal villi. Toxin then enters lymph vessels that drain to the subclavian vein to enter the liver via the portal circulation. In the liver, fat-soluble toxicants are in part detoxified by a two-step process that conjugates a polar entity that is released back into the intestine, where it may be reabsorbed and eventually eliminated in urine.
Several toxicokinetic studies treating rats with [3H]-PbTx3 have shown that blood retains detectable levels of the radioligand (Benson et al. 1999; Cattet and Geraci 1993; Poli et al. 1990b). More recently, studies treating mice, rats, or fish and measuring toxin using receptor assay and radioimmunoassay have quantified blood brevetoxin and determined that substantial levels (20–30 nM) are retained in all three species (Fairey et al. 2001; Radwan et al. 2005; Woofter et al. 2003, 2005). These values are approximately one order of magnitude higher than the concentration of toxin necessary to activate voltage-gated sodium channels in nerve, heart, or muscle (Bottein Dechraoui and Ramsdell 2003). This indicates that a significant fraction of brevetoxin is bound to elements in blood, reducing its availability to be biologically active at the sodium channel. Brevetoxin, like other lipophilic agents, is likely to partition to cellular elements in blood, and brevetoxin immunoreactivity has been reported in tissue lymphocytes and macrophages of manatees poisoned by brevetoxins (Bossart et al. 1998). Additionally, brevetoxins may associate with specialized transport proteins in the plasma.
Blood is technically a tissue containing a fluid matrix known as the plasma. Plasma is a protein-rich solution containing a diverse number of proteins that serve a variety of functions, including transport, immune response, and tissue-to-tissue signaling (Anderson and Anderson 2002). A primary function of plasma is the distribution of insoluble substances through the use of carrier proteins. Albumin is the most abundant plasma protein and serves as a low-affinity, high-capacity binding protein for steroid hormones and delivers these insoluble substances to target cells in the peripheral circulation (Kragh-Hansen 1981). Plasma also includes more highly specialized binding proteins. Perhaps best known are the high-affinity binding proteins for sex steroids, thyroid hormones, and corticosteroids. In addition, the plasma contains a unique lipoprotein transport system for the precursor of these steroid hormones, cholesterol. Lipoproteins, protein-coated fatty acid/cholesterol complexes, distribute cholesterol to tissues and remove cholesterol from the plasma.
Plasma carrier proteins are believed to have evolved, at least in part, to serve a functional role to bind and transport nonendogenous hydrophobic substances (Baker 2002). We examined the binding of brevetoxin to plasma carrier proteins in mice, looking especially at lipoproteins, and evaluated the role of plasma carrier proteins in the distribution of brevetoxins to target tissues and the elimination of brevetoxins from the organism.
Materials and Methods
Brevetoxin mouse plasma spike.
Mouse plasma in EDTA was obtained from Harlan Bioproducts (Indianapolis, IN). We spiked a 9.991-mL sample of mouse plasma with 9 μL of 100 μg/mL PbTx-3 to give a concentration of approximately 100 ng/mL PbTx-3 in the mouse plasma. This solution was then covered, vortexed, and stored at 6°C for 2 hr. We added 2 mL of 60% iodixanol {5,5′-[(2-hydroxy-1-3 propanediyl)-bis(acetylamino)] bis [N,N′-bis(2,3-dihydroxypropyl-2,4,6-tri-iodo-1,3-benzenecarboxamide)]} to the 10 mL spiked mouse plasma and allowed it to sit for 2 hr at approximately 4°C. All iodixanol solutions were kept wrapped in aluminum foil and stored at 4°C. In two ultracentrifuge tubes, 9% iodixanol solution (5 mL) was underlayed by 5 mL of the spiked mouse plasma in iodixanol. We carefully pipetted 5 mL of HEPES buffered saline on top.
Brevetoxin mouse exposure.
We obtained 20 female ICR mice, 18–20 g, from Harlan Sprague Dawley (Indianapolis, IN). The mice were kept for 24 hr with food and water given ad libitum. We injected 10 of the mice intraperitoneally (ip) with a maximally tolerable dose (310 μg/kg) of PbTx-3 (Calbiochem, La Jolla, CA) in 3.1% methanol in phosphate-buffered saline (PBS). The other 10 mice were injected ip with 3.1% methanol in PBS. The brevetoxin congener (PbTx-3 was chosen for these experiments because it is less reactive to metabolism in rodents than its precursor (PbTx-2) (Radwan et al. 2005). We based the 30-min time point for blood collection on previous toxicokinetic experiments to assure measurable levels of toxin in fractionated plasma by radioimmunoassay (RIA) (Woofter et al. 2003). After a 30-min exposure, the mice were euthanized with carbon dioxide and exsanguinated via cardiac puncture to the left ventricle with a heparinized 1-cc syringe. The blood was collected in two BD Vacutainer CPT tubes (BD Vacutainer Systems, Franklin Lakes, NJ). The tubes were centrifuged for 20 min at 2,100 relative centrifugal force. We then extracted the plasma by pipette and transferred it to culture tubes. Mice were treated in accordance with the Guidelines for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources 1996), and all possible efforts were made to reduce animal suffering and to minimize the number of animals used.
Gradient ultracentrifugation.
Two ultracentrifuge tubes containing 5 mL of 9% iodixanol solution were underlayed with 4 mL extracted mouse plasma in 1 mL iodixanol. Five milliliters of HEPES buffered saline was carefully pipetted on top. Samples were centrifuged at 16°C in a Sorvall swinging bucket rotor (Kendro Laboratory Products, Asheville, NC) for 51 hr at 160,000 × g (acceleration and deceleration 8). The tubes were extracted without shaking after ultracentrifugation. We punctured the ultracentrifuge tubes 3 cm from the bottom of the tube and collected 0.5-mL fractions. Fractions from the spiked mouse plasma tubes were collected in Eppendorf tubes, and fractions from the injected mice were collected in glass centrifuge tubes.
Agarose gel electrophoresis lipoprotein analysis.
We used a HYDRAGEL LIPO + Lp(a) K20 gel electrophoresis kit (SEBIA, Norcross, GA) for fraction characterization. Stock buffer (75 mL) was diluted in distilled water to a volume of 1 L. We made the Sudan black staining solution by adding 80 mL pure ethanol, 1 mL Sudan black stock solution, and 70 mL distilled water. This was allowed to stir gently for 30 min before use. A wash solution was prepared by diluting 16 mL of the wash stock solution in distilled water to a volume of 1 L. We used a destaining solution of 45% ethanol in water.
We filled the HYDRAGEL K20 applicator wells with 10 μL of lipoprotein fraction; each applicator was loaded within 2 min. The applicator was lowered onto the applicator carrier holding the gel and was allowed to rest on the gel for 7.5 min before being removed and discarded. We placed the gel into a SEBIA K20 electrophoresis chamber and allowed it to run for 90 min at 50 V. The gels were removed from the chamber and placed in an 80°C oven for 6–10 min or until dry. After being allowed to cool to room temperature, we placed each gel in a gel holder and immersed it in 50 mL Sudan black staining solution for 15 min. Each gel was then placed in 50 mL destaining solution for 5 min and then immersed in 50 mL wash solution for 1 min. The gels were dried at 80°C for 6–10 min and visually characterized. We stored gels for possible later use at approximately 4°C.
Brevetoxin radioimmunoassay.
We performed radioimmunoassays for detecting brevetoxin using a sheep antisera prepared against a PbTx-2-fetuin conjugate (Garthwaite et al. 2001; Woofter et al. 2001). RIAs were run in 12 × 75 borosilicate glass tubes in PBS containing 137 mM NaCl, 8 mM Na2HPO4, 1.5 mM KH2PO4, and 2.7 mM KCl (all from Sigma Chemical Company, St. Louis, MO), and 0.01% Emulphor-EL 620 (GAF, New York). The assay tubes consisted of PbTx-3 standard or ultracentrifuged lipoprotein fraction, anti-PbTx antiserum (1:4,000), and [3H]PbTx-3 (0.4 nM), in PBS (final assay volume of 500 μL). [3H]-PbTx-3 (21 Ci/mmol; 98.7% radiochemical purity by HPLC) was produced by sodium borohydride reduction of PbTx-2 by contract with Amersham Biosciences (Buckinghamshire, UK). An undetermined amount of [3H]-PbTx-9 double reaction product is likely present in the above preparation. The seven PbTx-3 standards ranged from 0.01 ng/mL to 1,000 ng/mL. We allowed the PbTx-3 standards and lipoprotein fractions to preincubate in buffer at room temperature with the anti-PbTx-3 antibody for 1 hr before the [3H]PbTx-3 tracer was added. The tubes were placed on a shaker (Titramax 100; Heidolph Instruments, Cinnaminson, NJ) and incubated 1 hr. We added Sac-Cel (Alpco Diagnostics, Windham, NH) to the assay tubes and filtered the bound antibody onto 25-mm glass fiber filters. Each assay tube was then rinsed with PBS (3 × 2 mL) using a 48-sample, Semi-Auto Harvester (Brandel, Gaithersburg, MD). We placed the filters in 5.0-mL Scintiverse (Fisher, Suwanee, GA) and counted the radioactivity on a Tri-Carb 3100TR Liquid Scintilation Counter (Packard-PerkinElmer, Wellesley, MA).
Direct sandwich mouse albumin ELISA.
We used mouse albumin ELISA (enzyme-linked immunosorbent assay) quantitation kits (Bethyl Laboratories, Inc., Montgomery, TX) to quantify the albumin in each lipoprotein fraction. The plate was coated with a goat anti-mouse albumin antibody buffer. After a 60-min incubation, the plate was washed and then blocked with the postcoat solution. After incubation with the blocking postcoat solution for 30 min, we added the standard curve (7.8–10,000 ng/mL plus blank) and samples (lipoprotein fractions) to the plate and incubated it for 60 min. The samples were then washed off the plate and the horseradish peroxidase (HRP)-conjugated detection antibody (diluted 1:80,000) was added and incubated in the plate for 60 min. After washing the plate, we added TMB (3,5,3′,5′-tetramethylbenzidine), which reacts with the HRP to form a blue end product. We stopped the TMB reaction by adding 2 M H2SO4, which turned the blue product yellow. We then read the plate on a FluoStar plate reader (BMG Labtechnologies, Durham, NC) at 450 nm.
[3H]PbTx-3–plasma albumin binding study.
Purified mouse albumin (100 mg) was solubilized in 1X PBS containing 137 mM NaCl, 8 mM Na2HPO4, 1.5 mM KH2PO4, and 2.7 mM KCl (all from Sigma Chemical Company). Each 300-μL 15-kDa molecular weight Spectra/Pro Float-A-Lyzer dialysis tube (Spectrum Laboratories Inc., Rancho Dominguez, CA) contained 282.5 μL of each albumin solution and 17.5 μL of a brevetoxin solution. The final concentration in each tube was 19, 38, and 76 mg/mL albumin in PBS, and 42.5 mg/mL albumin in mouse reference serum (Bethyl Laboratories, Inc.) with 29.8 ng/mL PbTx-3/[3H]PbTx-3 (32,000 cpm). The dialysis tubes were prepared and dialyzed in 1X PBS for a total of 22 hr with one buffer change. We counted the bound [3H]PbTx-3 in the dialysis tubes on a 1211 RackBeta Liquid Scintilation Counter (Wallac-PerkinElmer, Wellesley, MA).
High-density lipoprotein SPA assay.
We modified high-density lipoprotein scintillation proximity assay (SPA) kits (Amersham Biosciences, Piscataway, NJ), using [3H]PbTx-3 as the radioligand, for direct brevetoxin binding to HDLs. Each experiment was composed of Btest, NSB (nonspecific binding), and B0 groups. Every well contained 10 μL [3H]PbTx-3 (43 nM, 53,000 dpm) tracer and 10 μL SPA beads (poly-l-lysine coated YSi beads) with a final volume of 60 μL. Each Btest well contained 30 μL PbTx-3 (50 μM), 10 μL HDL (0.5 mg/mL) in addition to the tracer and beads. The NSB group contained 40 μL PBS and no HDL, whereas the B0 group contained 30 μL PBS and 10 μL HDL.
To determine percent inhibition, we used the following equation:
Data analysis.
We determined all concentrations and EC50 (median effective concentration) values using Prism Graph Pad 4.0 (GraphPad Software, Inc., San Diego, CA).
Results
We first determined whether mouse plasma has high molecular-binding components for brevetoxins. Mouse plasma and standard concentrations of albumin were spiked with [3H]PbTx-3, and unbound tracer was removed by 15-kDa molecular weight cutoff dialysis. Mouse albumin (20–80 mg/mL) retained a small fraction (< 6.8%) of the radiolabel after dialysis (Table 1). In contrast, mouse reference plasma containing 43 mg/mL albumin retained substantially more (39%) of the radiolabel after dialysis. These results indicated that the majority of brevetoxin binding in plasma is to component other than albumin.
We next examined if brevetoxin bound to lipoproteins in plasma. Mouse plasma was spiked with PbTx-3 and iodixanol gradient ultracentrifugation performed to separate lipoprotein particles based on density. The relative abundance of HDL and LDL/VLDL in the iodixanol gradient fractions was determined by agarose gel electrophoresis. We collected 26 0.5-mL fractions between the platelets and chylomicron bands. Fractions 1 and 2 had exclusively HDL, and fractions 3–20 showed decreasing HDL levels and increasing LDL/VLDL levels; fractions 21–26 had exclusively LDL/VLDL. Analysis of each fraction by radioimmunoassay indicated that the most dense lipoprotein fractions, those containing only HDLs, contained the greatest amount (85%) of brevetoxin immunoreactivity within the lipoprotein fractions (Figure 1).
We then examined the lipoprotein fractions to determine whether they may have been contaminated with concentrated levels of albumin. To determine if the albumin in these fractions contributed significantly to the binding of brevetoxin, each fraction was analyzed for the presence of mouse albumin using ELISA. Low but detectable albumin was present in some of the fractions, and distribution between fractions paralleled the distribution of brevetoxin (Figure 2). The total amount of albumin in the first two fractions (HDL) was 1.3 mg, a value much lower than the amount present in plasma (1.3 mg/4 mL vs. 40 mg/mL). Hence, although there was some contamination of HDL fractions with albumin, it was far less than could be accountable for brevetoxin binding.
An independent approach was also taken to assure that the brevetoxin binding to the HDL fractions was not the result of other plasma carrier proteins. Each lipoprotein fraction, preincubated with [3H]PbTx-3, was dialyzed against a high molecular weight (100 kDa) cutoff membrane. The two homogenous HDL fractions once again showed the majority of binding of the lipoprotein fractions (Figure 3). This dialysis also permitted analysis of the platelet and chylomicron fractions for brevetoxin binding, and approximately 25% of the binding was to platelets and 2.5% was to the chylomicron fractions.
We confirmed our results using an HDL SPA. After incubating 6.5 pmol [3H]PbTx-3 with 5 μg purified human HDL under non-disturbed equilibrium conditions, 0.069 ± 0.015 fmol [3H]PbTx-3 bound per microgram of HDL. The binding of brevetoxin to HDL was reversible and saturable, as it was inhibited by 40% ± 7% in the presence of 50 μM unlabeled PbTx-3.
The final experiment was to expose mice in vivo to PbTx-3 to compare the distribution of brevetoxin in lipoproteins from our in vitro results to an in vivo exposure. Mice were treated ip with 310 μg/kg PbTx-3 for 30 min and their plasma was pooled. The plasma lipoproteins were separated and analyzed by RIA as described for the experiment presented in Figure 1. High levels of brevetoxin were found in the first two fractions containing exclusively HDLs, with decreasing levels of brevetoxin paralleling the decreasing levels of HDLs in fractions 3–11 (Figure 4). Substantial brevetoxin immunoreactivity was also associated with the platelet fraction.
Discussion
Binding to plasma/albumin.
Previous studies from our laboratory have found that brevetoxin achieves levels in whole blood more than a magnitude higher than its effective intrinsic concentration (i.e., the concentration required to bind voltage-gated sodium channels in nerve, muscle, or heart) (Woofter et al. 2003). Brevetoxin blood values remain between 25 and 30 nM for the first 12 hr in mice, whereas brevetoxin is effective at approximately 1–5 nM at site 5 of the voltage-gated sodium channel. This suggests that the majority of brevetoxin in blood may not be immediately biologically available and may be bound to cellular elements in blood or the fluid matrix (i.e., plasma). Because plasma contains specialized carrier proteins that may serve to transport brevetoxin to tissues, our investigation has focused on brevetoxin distribution in plasma. Dialysis of [3H]PbTx-3 spiked plasma revealed that 39% of the radiolabel was retained by plasma fractions of > 15 15kDa molecular weight. Of this, only < 6.8% of the binding was accounted for by binding to albumin (43 mg/mL) under these conditions. Albumin is a common binding protein for lipophilic compounds and has a well-characterized hydrophobic binding pocket. Whether albumin represents a greater percentage of binding under true equilibrium conditions in plasma cannot be resolved from these experiments; however, other factors in plasma provide nearly five times more binding capacity under the conditions used for this initial experiment.
Binding to high-density lipoproteins.
Lipoproteins are predominantly recognized as plasma carrier particles for cholesterol and triglycerides. They are a heterogeneous class of protein–lipid aggregates traditionally classified by particle density and subsequently by apoprotein composition. We found that brevetoxin added to mouse plasma localized to HDL fractions after being purified by iodixanol gradient ultracentrifugation and characterized by agarose gel electrophoresis mobility. Because of some contamination of albumin from the adjacent platelet/fibrin fractions, we also confirmed using high molecular weight dialysis of [3H]PbTx-3 that the tracer was associated with plasma components > 100 kDa molecular weight. Additionally, we confirmed binding of [3H]PbTx-3 to human HDL under non-disturbed equilibrium SPA.
HDLs are the highest density class of lipoproteins. They are assembled in the interstitial space from aggregation of free phospholipids, cholesterol, and apo A1 proteins as discoidal “nascent” HDL particles. These particles then serve as a substrate for lecithin:cholesterol acyltransferase, leading to the esterification of cholesterol. The cholesterol ester localizes as an inner core and forces the HDL particles to form spheres, known as HDL3. The maturation of HDL3 to HDL2 results from the transfer of phospholipids, free cholesterol, and apoproteins, released from the lipolyzed VLDLs, to the HDL particle (Eisenberg 1984). These HDL2 particles are taken up by the liver and release cholesterol by a process known as reverse cholesterol transport. Hence, each class of HDL is composed of a phospholipid, cholesterol, and lipoprotein A1 outer membrane that successively accumulates triglycerides and cholesterol esters in an expanding inner core. The process by which HDLs accumulate cholesterol, first described by Bailey (1965), is the initial step (cholesterol efflux) of reverse cholesterol transport (Glomset 1973). This process is still not fully resolved; however, at least three separate mechanisms play a role in the uptake of cholesterol by HDL. The first mechanism is passive diffusion of cholesterol, which is a simple equilibrium of cholesterol between cellular and lipoprotein compartments. This process is not restricted to HDL, but is also operative for LDL and plasma-binding proteins such as albumin (Zhao and Marcel 1996), yet it is considered to be particularly well-suited to mature HDL2 because of the high cholesterol ester content. The other cholesterol-uptake mechanisms involve scavenger receptor B1 (SRB1) and the ATP-binding cassette subfamily, A member 1 (ABCA1) transporter.
Binding to low-density lipoproteins.
Our data indicate that brevetoxin binds predominantly to the HDL class of lipoproteins in mouse plasma. However, this probably does not reflect the situation in other species, including humans, because mice, unlike normolipidemic humans, have small amounts of VLDLs and LDLs (as low as 20% vs. 70–80%, respectively) in the total lipoprotein pool (Grass et al. 1995). The low levels of VLDLs and LDLs present in mouse plasma may have reduced our ability to detect more substantial brevetoxin binding to these lipoprotein classes. Studies with several different classes of hydrophobic drugs, including cyclosporine A, amphotericin B, and nystatin, indicated that binding occurs to HDLs as well as to VLDLs and LDLs under in vitro conditions (Brajburg et al. 1984; Lemaire and Tillement, 1982; Wasan and Cassidy 1998). The mechanism for brevetoxin uptake under these conditions is also consistent with the passive diffusion of cholesterol. Certain drugs show a preference for HDL among the lipoproteins; however, this may be due to the high protein:lipid ratio of HDL (Cassidy et al. 1998; Kennedy and Wasan 1999; Wasan et al. 1997). Further studies will be needed with brevetoxins to determine the degree of selectivity for the toxin to the different lipoprotein classes, with particular reference to lipoprotein profiles representative of human plasma.
Functional significance of brevetoxin binding to HDL.
PbTx-3 administered systemically to mice distributed to HDL fractions, indicating that brevetoxin binding to HDL may have physiologic significance. Under in vivo conditions, passive diffusion mechanisms likely play an important role in the uptake of brevetoxins to HDL. However, brevetoxins may also utilize two other mechanisms for cholesterol efflux. The first is SRB1-facilitated aqueous diffusion of cholesterol (Acton et al. 1996), a process that appears to involve the reorganization of membrane lipid domains that transfers cholesterol to various classes of lipoproteins (Rhainds and Brissette 2004). The second is the ABCA1 transporter, originally identified as a genetic defect in Tangier disease (Lawn et al. 1999). Regardless of the mechanism of brevetoxin uptake, hepatocytes would then promote transformation, conjugation, and reabsorption of brevetoxin into the bile. Excretion of a more polar metabolite would favor intestinal reabsorption and excretion via the kidneys into the urine. Recent studies have indicated that PbTx-2, the nonreduced precursor of PbTx-3, is particularly sensitive to transformation and conjugation in rats and is eliminated rapidly in the urine (Radwan et al. 2005) and hence would be a candidate to use this pathway.
Basis for susceptibility and strategies for therapeutics.
The distribution of brevetoxins to lipoproteins may also indicate a basis for susceptibility in human populations, based on blood cholesterol-containing lipids. Indeed, studies with other compounds that bind to lipoproteins in plasma have shown that dyslipidemia alters drug distribution and enhances toxicity (Wasan and Cassidy 1998). In this regard, the distribution of brevetoxins by ip toxin exposure is not likely to reflect distribution of toxin by oral exposure, as was demonstrated for chlorpromazine and imipramine (Bickel 1975). After oral exposure, it is likely that absorption of brevetoxins across the intestinal mucosa will lead to their incorporation into chylomicrons. This may provide an important immediate distribution mechanism into the bloodstream and peripheral cells and then uptake by HDL or possibly LDL by one of the above-mentioned mechanisms of cholesterol efflux. Accordingly, lipoprotein profile or genetic variation in several of the mechanisms for reverse cholesterol transport (SRB1, ABCA1) may be expected to alter brevetoxin toxicokinetics and hence sensitivity to toxicity. Likewise, each of these pathways may provide a means for potential therapeutic intervention of brevetoxin in neurotoxic shellfish poisoning and the more prolonged effects of related polyether ciguatoxin for ciguatera fish poisoning. Indeed, the plasma levels of these toxins remain elevated (Bottein Dechraoui et al. 2005; Radwan et al. 2005; Woofter et al. 2003) and hence are continuously undergoing enterohepatic recirculation and intestinal reabsorption. In this light, another potential target for therapeutics is the enterohepatic recirculation pathway. Through the binding of bile salts and reduced intestinal reabsorption, bile acid sequestrants effectively disrupt this pathway and have been reported to reduce the symptoms of ochratoxin (Kerkadi et al. 1998, 1999). Accordingly, targeting both pathways for reverse cholesterol transfer and enterohepatic recirculation may ultimately lead to the development of therapies to mitigate the symptoms of neurotoxic shellfish poisoning and ciguatera fish poisoning.
The National Ocean Service (NOS) does not approve, recommend, or endorse any proprietary product or material in this publication. No reference shall be made to NOS, or to this publication furnished by NOS, in any advertising or sales promotion which would indicate or imply that NOS approves, recommends, or endorses any proprietary product or proprietary material mentioned herein or which has as its purpose any intent to cause directly or indirectly the advertised product to be used or purchased because of NOS publication.
Figure 1 RIA determination of brevetoxin in lipoprotein fractions of spiked mouse plasma. Mouse plasma was spiked in vitro with PbTx-3 (100 ng/mL), and lipoprotein fractions were collected after iodixanol density gradient ultracentrifugation. RIA results are given as PbTx-3 equivalents (ng)/fraction and are mean ± SD from a single experiment.
Figure 2 Determination of albumin in lipoprotein fractions of mouse plasma using direct sandwich ELISA. Results are shown as albumin (ng)/fraction and are mean ± SD from a single experiment.
Figure 3 Determination of [3H]PbTx-3 association with plasma components with molecular weight > 100 kDa. ref, reference. Results are given as the percentage of PbTx-3 bound.
Figure 4 RIA determination of brevetoxin in lipoprotein fractions from mice exposed, in vivo, to PbTx-3. Mouse plasma was collected and pooled after being injected (ip) with PbTx-3 or vehicle. This plasma was then fractionated after iodixanol density gradient ultracentrifugation. The RIA results shown here are given as PbTx-3 equivalents (ng)/fraction and are mean values ± SD from a single experiment.
Table 1 Association of brevetoxin to mouse albumin and plasma.
Sample Albumin (mg/mL) PbTx-3 (% bound)
Mouse albumin 76 5.0
Mouse albumin 38 6.8
Mouse albumin 19 2.9
Reference plasma 43 38.7
Mouse plasma and different concentrations of mouse albumin were spiked with [3H]PbTx-3, and bound PbTx-3 was separated by 15 kDa molecular weight cutoff dialysis. Values are from individual sample dialysis bags that were analyzed in quintuplicate, with SEM < 5%.
==== Refs
References
Acton S Rigotti A Landschulz KT Xu S Hobbs HH Krieger M 1996 Identification of scavenger receptor SR-BI as a high density lipoprotein receptor Science 271 518 520 8560269
Anderson NL Anderson NG 2002 The human plasma proteome: history, character, and diagnostic prospects Mol Cell Proteomics 1 845 867 12488461
Bailey JM 1965 Lipid metabolism in cultured cells. IV. Serum alpha globulins and cellular cholesterol exchange Exp Cell Res 37 175 182 14263309
Baker ME 2002 Beyond carrier proteins: albumin, steroid hormones and the origin of vertebrates J Endocrinol 175 121 127 12379496
Benson JM Thischler DL Baden DG 1999 Uptake, tissue distribution, and excretion of brevetoxin-3 administered to rats by intratracheal instillation J Toxicol Environ Health 56 345 355
Bickel MH 1975 Binding of chlorpromazine and imipramine to red cells, albumin, lipoproteins and other blood components Pharm Pharmacol 27 733 738
Bossart GD Baden DG Ewing RY Roberts B Wright SD 1998 Brevetoxicosis in manatees (Trichechus manatus latirostris ) from the 1996 epizootic: gross, histologic, and immunohistochemical features Toxicol Pathol 26 276 282 9547868
Bottein Dechraoui MY Ramsdell JS 2003 Type B brevetoxins show tissue selectivity for voltage-gated sodium channels: Comparison of brain, skeletal muscle, and cardiac sodium channels Toxicon 41 919 927 12782093
Bottein Dechraoui MY Wang Z Turquet J Chinain M Darius T Cruchet P 2005 Biomonitoring of ciguatoxin exposure in mice using blood collection cards Toxicon 46 243 251 15979117
Brajburg J Elberg S Bolard J Medoff G 1984 Interaction of plasma proteins and lipoproteins with amphotericin B J Infect Dis 149 986 992 6376657
Cassidy SM Strobel FW Wasan KM 1998 Plasma lipoprotein distribution of liposomal nystatin is influenced by protein content of high-density lipoproteins Antimicrob Agents Chemother 42 1878 1888 9687378
Cattet M Geraci JR 1993 Distribution and elimination of ingested brevetoxin (PbTx-3) in rats Toxicon 31 1483 1486 8310449
Davis CC 1948 Gymnodinium breve : a cause of discolored water and animal mortality in the Gulf of Mexico Bot Gaz 109 358 360
Eisenberg S 1984 High density lipoprotein metabolism J Lipid Res 25 1017 1058 6392459
Fairey ER Shuart NG Busman M Moeller PD Ramsdell JS 2001 Biomonitoring brevetoxin exposure in mammals using blood collection cards Environ Health Perspect 109 717 720 11485871
Glomset JA 1973 The metabolic role of lecithin: cholesterol acyltransferase: perspectives from pathology Adv Lipid Res 11 1 65
Grass D Saini U Felkner RH Wallace RE Lago WJ Young SG 1995 Transgenic mice expressing both human apolipoprotein B and human CETP have a lipoprotein cholesterol distribution similar to that of normolipidemic humans J Lipid Res 36 1082 1091 7658156
Huang JM Wu CH Baden DG 1984 Depolarizing action of a red-tide dinoflagellate brevetoxin on axonal membranes J Pharmacol Exp Ther 229 615 621 6325670
Institute of Laboratory Animal Resources 1996. Guide for the Care and Use of Laboratory Animals. 7th ed. Washington, DC:National Academy Press.
Kennedy AL Wasan KM 1999 Preferential distribution of amphotericin B lipid complex into human HDL3 is a consequence of high density lipoprotein coat lipid content J Pharm Sci 8 1149 1155 10564063
Kerkadi A Barriault C Marquardt RR Frohlich AA Yousef IM Zhu XX 1999 Cholestyramine protection against ochratoxin A toxicity: role of ochratoxin A sorption by the resin and bile acid enterohepatic circulation J Food Prot 62 1461 1465 10606152
Kerkadi A Barriault C Tuchweber B Frohlich AA Marquardt RR Bouchard G 1998 Dietary cholestyramine reduces ochratoxin A-induced nephrotoxicity in the rat by decreasing plasma levels and enhancing fecal excretion of the toxin J Toxicol Environ Health 53 231 250
Kragh-Hansen U 1981 Molecular aspects of the ligand binding to serum albumin Pharmacol Rev 33 17 53 7027277
Landsberg JH 2002 The effects of harmful algal blooms on aquatic organisms Rev Fish Sci 10 113 390
Landsberg JH Steidinger KA 1997. A historical review of red tide events caused by Gymnodinium breve as related to mass mortalities of the endangered manatee (Trichechus manatus latirostris) in Florida, USA. In: Harmful Microalgae, the Proceedings of the Eighth International Conference on Harmful Algae, 25–29 June 1997, Vigo, Spain (Reguera B, Blanco J, Fernandez M, Wyatt T, eds). Santiago de Compostela, Spain:Xunta de Galicia and IOC of UNESCO Publishers, 97–100.
Lawn RM Wade DP Garvin MR Wang X Schwartz K Porter JG 1999 The Tangier disease gene product ABC1 controls the cellular apolipoprotein-mediated lipid removal pathway J Clin Invest 104 R25 31 10525055
Lemaire M Tillement JP 1982 Role of lipoproteins and erythrocytes in the in vitro binding and distribution of cyclosporin A in the blood J Pharm Pharmacol 34 715 718 6129301
Lin Y Risk M Ray SM Van Engen D Clardy J Golik J 1981 Isolation and structure of brevetoxin B from the “red tide” dinoflagellate Ptychodiscus brevis (Gymnodinium breve ) J Am Chem Soc 103 6773 6775
Martin DF Chatterjee AB 1969 Isolation and characterization of a toxin from the Florida red tide organism Nature 221 59 5782610
McFarren EF Tanabe H Silva FJ Wilson WB Campbell JE Lewis KH 1965 The occurrence of a ciguatera-like poison in oysters, clams, and Gymnodinium breve cultures Toxicon 3 111 123 5867066
Pierce RH Henry MS Proffitt LS Hasbrouck PA 1990. Red tide toxin (brevetoxin) enrichment in marine aerosol. In: Toxic Marine Phytoplankton (Graneli E, Sunstron B, Edler L, Anderson DM, eds). Amsterdam:Elsevier Scientific Publishing, 397–402.
Poli MA Mende TJ Baden DG 1986 Brevetoxins, unique activators of voltage-sensitive sodium channels, bind to specific sites in rat brain synaptosomes Mol Pharmacol 30 129 135 2426567
Poli MA Templeton CB Pace JG Hines HB 1990a. Detection, metabolism, and patholophysiology of brevetoxins. In: Marine Toxins: Origin, Structure, and Molecular Pharmacology (Hall S, ed). Washington, DC:American Chemical Society, 176–191.
Poli MA Templeton CB Thompson WL Hewetson JF 1990b Distribution and elimination of brevetoxin PbTx-3 in rats Toxicon 28 903 910 2080516
Radwan F Wang Z Ramsdell JS 2005 Identification of a rapid detoxification mechanism of brevetoxin in rats Toxicol Sci 85 839 846 15746006
Rhainds D Brissette L 2004 The role of scavenger receptor class B type I (SR-BI) in lipid trafficking. Defining the rules for lipid traders Int J Biochem Cell Biol 36 39 77 14592533
Sheridan RE Adler M 1989 The actions of a red tide toxin from Ptychodiscus brevis on single sodium channels in mammalian neuroblastoma cells FEBS Lett 247 448 452 2541024
Taylor HF 1917. Mortality of fishes on the west coast of Florida. Report of the U.S. Commissioner of Fisheries, Appendix III. Bureau of Fisheries Document 848. Washington, DC:Government Printing Office.
Wasan KM Cassidy SM 1998 Role of plasma lipoproteins in modifying the biological activity of hydrophobic drugs J Pharm Sci 87 411 424 9548892
Wasan KM Pritchard PH Ramaswamy M Wong W Donnachie EM Brunner LJ 1997 Differences in lipoprotein lipid concentration and composition modify the plasma distribution of cyclosporine Pharm Res 1613 1620 9434283
Westerfield M Moore JW Kim YS Padilla GM 1977 How Gymnodinium breve red tide toxin(s) produces repetitive firing in squid axons Am J Physiol 232 C23 29 556889
Woodcock AH 1948 Note concerning human respiratory irritation associated with high concentrations of plankton and mass mortality of marine organisms J Mar Res 7 56 62
Woofter R Dechraoui MY Garthwaite I Towers NR Gordon CJ Cordova J 2003 Measurement of brevetoxin levels by radioimmunoassay of blood collection cards after acute, long-term, and low-dose exposure in mice Environ Health Perspect 111 1595 1600 14527838
Woofter RT Brendtro K Ramsdell JS 2005 Uptake and elimination of brevetoxin in blood of striped mullet (Mugil cephalus ) after aqueous exposure to Karenia brevis Environ Health Perspect 113 11 16 15626641
Zhao Y Marcel YL 1996 Serum albumin is a significant intermediate in cholesterol transfer between cells and lipoproteins Biochemistry 35 7174 7180 8679545
|
16263501
|
PMC1310908
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 23; 113(11):1491-1496
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8010
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7585ehp0113-00149716263502ResearchProximity to Pollution Sources and Risk of Amphibian Limb Malformation Taylor Brynn 123Skelly David 1Demarchis Livia K. 1Slade Martin D. 2Galusha Deron 2Rabinowitz Peter M. 21 School of Forestry and Environmental Studies, Yale University, New Haven, Connecticut, USA2 Occupational and Environmental Medicine Program and 3 Department of Epidemiology and Public Health, Yale University School of Medicine, New Haven, Connecticut, USAAddress correspondence to P.M. Rabinowitz, Yale University School of Medicine, Occupational and Environmental Medicine Program, 135 College St., New Haven, CT 06510 USA. Telephone: (203) 785-5885. Fax: (203) 785-7391. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 11 7 2005 113 11 1497 1501 15 9 2004 11 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The cause of limb deformities in wild amphibian populations remains unclear, even though the apparent increase in prevalence of this condition may have implications for human health. Few studies have simultaneously assessed the effect of multiple exposures on the risk of limb deformities. In a cross-sectional survey of 5,264 hylid and ranid metamorphs in 42 Vermont wetlands, we assessed independent risk factors for nontraumatic limb malformation. The rate of nontraumatic limb malformation varied by location from 0 to 10.2%. Analysis of a subsample did not demonstrate any evidence of infection with the parasite Ribeiroia. We used geographic information system (GIS) land-use/land-cover data to validate field observations of land use in the proximity of study wetlands. In a multiple logistic regression model that included land use as well as developmental stage, genus, and water-quality measures, proximity to agricultural land use was associated with an increased risk of limb malformation (odds ratio = 2.26; 95% confidence interval, 1.42–3.58; p < 0.001). The overall discriminant power of the statistical model was high (C = 0.79). These findings from one of the largest systematic surveys to date provide support for the role of chemical toxicants in the development of amphibian limb malformation and demonstrate the value of an epidemiologic approach to this problem.
agricultureamphibiananimal sentinelmalformationteratogenwater pollution
==== Body
In 1995, amphibians with severely malformed limbs were discovered in Minnesota (Blaustein and Johnson 2003). The next year, amphibians with truncated and missing limbs were found at several sites stretching along 120 miles of Vermont’s Lake Champlain shoreline (Levey et al. 2003). Since then, elevated rates of developmental abnormalities have been found in most U.S. states and Canada, raising concern that amphibians are serving as animal sentinels of human environmental health hazards (Burkhart et al. 1998, 2000; Daszak et al. 2001; van der Schalie et al. 1999).
The exact etiology of amphibian limb abnormalities (including missing, extra, and malformed limbs) remains unclear. The major suspects have been ultraviolet B (UV-B) radiation, trauma, parasitic trematode infestation, and xenobiotic pesticides or chemicals (Ouellet 2000). Although many researchers have searched for single causative agents, some have stated that the etiology is likely multifactorial (Linder et al. 2001).
In laboratory settings, some evidence has supported each of the etiologic hypotheses. Controlled exposure to UV radiation has produced amphibian dysmorphogenesis (Ankley et al. 1998, 2002) and truncated limbs (Levey et al. 2003; Meteyer et al. 2000; Ouellet et al. 1997). These studies, however, have generally failed to reproduce the spectrum of abnormalities occurring in wild populations. The abnormalities that result from experimental UV exposures are most often bilateral and symmetrical (Ankley et al. 1998, 2002), whereas abnormalities found in the wild are not (Levey et al. 2003; Meteyer et al. 2000; Ouellet et al 1997). Furthermore, the levels of UV exposure used in the laboratory experiments may not reflect real-world exposures.
Another primary etiologic hypothesis of limb abnormality is trauma related to predation. Although predation can account for some forms of abnormality in the wild, the hypothesis, more generally, has not been supported as the primary cause of increased amphibian abnormality rates in recent years (Blaustein and Johnson 2003; Levey et al. 2003). Although predation should always be considered in this field of research, it is likely that deformities caused by predation have a different etiologic pathway and therefore, in research, should be distinguished from the other hypotheses of dysmorphogenesis.
Macroparasite infection is the third etiologic hypothesis of amphibian limb abnormality. The parasite hypothesis continues to receive substantial attention and is perhaps the most thoroughly explored of all the hypotheses (Blaustein and Johnson 2003). The most commonly explored parasite is the trematode Ribeiroia ondatrae. Infestation with Ribeiroia is associated with limb abnormalities in some amphibian species (Johnson et al. 2001, 2002). Laboratory experiments suggest that the mechanism of Ribeiroia-induced abnormalities may involve mechanical disturbance of growing limb cells or interference with a retinoid-sensitive signaling pathway (Johnson and Sutherland 2003).
The role of exposure to potential chemical teratogens has also been investigated. Candidate toxicants have included nonpolar organics, metals (Burkhart et al. 1998; Linder et al. 2001; Stocum 2000), herbicides, pesticides, and other components of agricultural runoff (Bishop et al. 1999; Ouellet et al. 1997), sewage (Linder et al. 2001; Ouellet 2000), and pharmaceuticals (Mizgireuv et al. 1984). Such chemical agents could directly affect amphibians’ development or act indirectly by increasing amphibian susceptibility to other environmental stressors such as infectious disease, predation, and UV-B light. Investigations into chemically mediated limb abnormalities have used two major approaches. First, amphibian eggs and larvae have been exposed to water and sediment collected from field sites with high abnormality rates (Fort et al. 1999, 2001). Second, the specific suspected amphibian teratogens have been tested in the laboratory using toxicologic assays (Degitz et al. 2000, 2003; LaClair et al. 1998). Together, these studies have shown that exposure to field-collected water and sediment can result in limb abnormalities and that a number of chemicals can have severe teratogenic effects on amphibians related to dose or concentration (Burkhart et al. 1998). However, there are inconsistencies between laboratory and field results and no single causative chemical has been identified.
The chemical teratogen hypothesis has particular relevance to human health risk. If waterborne chemical toxicants are involved in amphibian malformations, there is potential for shared exposure with human populations through dermal contact, ingestion, and inhalation routes.
Despite the fact that multiple hypotheses have emerged to explain the phenomenon of amphibian limb abnormalities, few studies to date have made use of research techniques allowing examination of multiple factors. For example, most studies of chemical causes did not also collect data regarding infectious agents, and few of the studies have used multivariable statistical techniques to model the relative effects of different factors.
We performed a cross-sectional study of risk factors for deformities in a large systematic sample of amphibians. The large sample size allowed for the creation of a multivariate model to test the association between amphibian limb malformation and a number of independent risk factors. The types of land use adjacent to wetlands where amphibians were surveyed served as a proxy measure for possible sources of water pollution. We hypothesized that, after adjusting for confounding factors, amphibians in wetlands adjacent to agriculture, septic systems, or lawns would be at greater risk for amphibian hind limb malformation due to the waterborne chemical runoff associated with such land uses.
Materials and Methods
Study sample.
Between 24 May and 28 August 2002, amphibian specimens consisting of two species of hylids (Hyla versicolor and Pseudacris crucifer) and four species of ranids (Rana pipiens, Rana catesbeiana, Rana clamitans, and Rana sylvatica) were collected from 42 wetlands in the Lake Champlain Basin of Vermont. We selected wetlands in a representative fashion along an urbanization gradient ranging from relatively undisturbed forest habitat (Green Mountain National Forest), to rural communities, to neighborhoods in Burlington. To collect specimens at each site, standardized collection methodology consisted of pipe samplers (steel pipe, 35-cm diameter, 0.91 m long) and dip nets (46 cm × 23 cm with 1-mm mesh). An attempt was made to collect up to 300 specimens from each site (a maximum of 100 during each of three visits). After capture, specimens were first anesthetized in a solution of tricaine methanesulfonate and then placed in 70% ethanol, according to protocols for the study that were approved by the Yale University Animal Care and Use Committee. Each specimen was later examined in the laboratory in order to determine genus and species, Gosner stage, and presence of malformation. A subsample of specimens was examined by dissection for the presence of parasite metacercariae, including Ribeiroia.
Exposure assessment.
For each of the 42 wetlands, on-site field observations were used to determine whether agriculture or lawns were located proximate to the wetland. The presence of agriculture and lawns were then scored as yes/no responses for each site. In addition, we independently estimated land use/land cover within 200 m of each amphibian sample site, which was calculated using geographic information system (GIS) data. The latitude and longitude coordinates of each site were used with land-use/land-cover GIS layers to map the land use within 200 m of each site. The Raster land-cover data set from Vermont GIS (U.S. Geological Survey 1999) was brought into ArcMap (Environmental Systems Research Institute, Redlands, CA) and was converted to vector data (polygons) to allow clipping of the features for each buffer zone. These polygons were then clipped using the geoprocessing command by a 200 m buffer surrounding the GPS (global positioning satellite) coordinates representing the study site’s general area. These clipped land-use/land-cover areas were then summarized for each location by their classification codes to calculate land-cover percentages represented in the data. For this analysis, the 21 possible land-use categories were used to determine the presence of agriculture or forest. We compared the GIS measures for agriculture and forest with field observations of agriculture and lawns. There was a strong correlation between GIS coded agriculture land-use and field observations of agriculture and lawn near the study site (p < 0.0001 for both). Both observed agriculture and lawns were negatively correlated with GIS land-use estimates for forest cover (p < 0.0001 for both).
Several quantitative measurements were made at each site, including pH, conductivity, dissolved oxygen, temperature, total nitrogen, and total phosphorus for the water samples. If more than one measurement took place at an individual wetland, the measurements were averaged together for that site.
As an additional investigation into the possible role of parasitic infection, a representative sample of snails from the study wetlands was investigated for evidence of Ribeiroia infection.
Case definition.
Each individual specimen was examined by a trained technician and classified according to a case definition adapted from a well-established amphibian limb abnormality classification scheme (Meteyer et al. 2000). To be classified as a “case” of limb malformation, a specimen had to exhibit one of the following: missing or reduced hind limb elements, complete but malformed hind limb, and/or duplicated hind limb elements or segments.
For the analysis of limb malformations, we included only those specimens with Gosner stage of ≥26 (Gosner 1960). At this stage, hind limbs are visible, and it is possible to assess whether gross deformities are present.
Of the original 5,983 specimens, this led to the exclusion of 684 with stage < 26, and 27 with missing stage, leaving 5,272. An additional 8 cases were excluded because of unknown length, leaving 5,264 in the final study sample. For our case definition, we excluded trauma deformities (abnormalities that appeared to be a result of trauma, including presence of open wounds, edema, scarring, bone fractures, etc.) because of the differing etiologies of trauma and nontrauma abnormalities.
Statistical analysis.
All analyses were performed using SAS software (version 8.02; SAS Institute Inc., Cary, NC).
Simple frequencies were calculated for categorical variables, including Gosner stage, genus, water-quality measures, presence of pollution sources, and presence of limb malformation. Means and SDs were calculated for continuous variables.
To determine bivariate associations between exposures and rates of malformation, we performed simple logistic regression. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for each individual factor. An OR calculates the odds of a risk factor being present in the affected cases compared with the odds that it is present in the non-affected individuals. A multivariate logistic regression model was developed to determine the independent predictors of malformation. To determine which variables to include in the multivariate model, we examined collinearity between independent variables using correlation coefficients. Variables with correlations above levels at which multicollinearity would be considered likely (r > 0.40) were excluded from the multivariable analyses. We found that a number of variables had significant collinearity, including nitrogen and phosphorus (r = 0.67) and pH and dissolved oxygen (r = 0.79). Therefore, neither phosphorus nor pH was included in the final multivariate model.
The multivariate logistic regression analysis used a backward selection process to eliminate nonsignificant variables from the model (criteria for elimination from the model set at p > 0.05). ORs, 95% CIs, and p-values for association between individual risk factors and limb malformation were determined for variables remaining in the final model. The overall discriminant power of the model was assessed using a C-statistic.
Results
Characteristics of study sample.
Table 1 shows the characteristics of 5,264 specimens meeting eligibility criteria (specimens with Gosner stage of ≥26). The average Gosner stage was 36 (field stage 4: toes 3–5 separated). Overall, 83 specimens (1.6%) showed evidence of non-traumatic limb malformation. The study-site–specific rate of malformation ranged from 0 to 10.2%. The lowest rates were found in the wetlands located in the Green Mountain National Forest. Table 1 also shows the relative prevalence of subtypes of nontraumatic malformation. The most common type was malformed limb or element (68 of 83), followed by missing limb or element (25 of 83). Some individuals had more than one type of abnormality.
Examination of a subsample of individual specimens (n = 40) revealed no evidence of Ribeiroia infection. Similarly, representative samples of host snails from the sample wetlands did not demonstrate evidence of Ribeiroia infection (Skelly D, unpublished data).
Exposure assessment.
Table 2 summarizes the exposure characteristics of the 42 wetland study sites. More than 40% of the wetlands had some degree of agriculture nearby; this ranged from pastureland to cropland to intensive dairy farming. Lawns were present near > 35% of wetlands. The wide range of values for water-quality measures reflects, in part, variations in runoff sources; the wide variability of conductivity reflects proximity to roads treated with salt in winter. The highest nitrogen readings were found in a pond located downhill from a dairy barn.
Risk factor analysis.
Table 3 shows the simple and multiple logistic regression measures of association between different risk factors and nontraumatic amphibian limb malformation. In the simple (bivariate) analyses, Gosner stage, proximity to agriculture, proximity to lawns, and dissolved oxygen each showed a positive association with malformation.
In the multivariate logistic regression model, Gosner stage remained highly significantly associated with malformation (OR = 1.18; p < 0.0001). In other words, the risk of malformation increased 18% for each increase in Gosner stage. Although malformation rates were slightly higher in ranids than in hylids, the effect of genus was not significant in the multivariate model. Furthermore, analyzing risk factors for malformation in ranid species alone or hylid species alone produced results similar to those of our combined model (data not shown). In the multivariate model, proximity to agriculture remained a highly significant predictor of malformation risk (OR = 2.26; 95% CI, 1.42–3.58; p < 0.001). None of the other independent variables tested showed a significant association in the multivariate model. The discrimination and fit of the logistic regression model were good, with a C-statistic = 0.79 and a p-value for the Hosmer-Lemeshow goodness-of-fit test of 0.10.
Discussion
The results of this study, based on one of the largest systematic sampling of limb deformities in wild amphibian populations to date, suggest that the composition of landscapes surrounding wetlands affects rates of limb malformation. In particular, proximity to human-associated land uses, including agriculture and lawns, is associated with an increased risk. This positive association persists even after adjusting for the effect of developmental stage and variation in water-quality measures such as nitrogen and pH. Proximity to agriculture was associated with a more than doubling of the risk of limb malformation.
Major strengths of this study include its sample size and systematic design. Although other studies have assessed agricultural land use in relation to amphibian population size in specific geographic areas (Davidson et al. 2001, 2002; Ray et al. 2002), few studies have assigned exposure variables based on observed pollution sources in a systematic manner in order to assess malformation rates. Several studies have discussed land use and land-use change as drivers of the abnormality and decline phenomena (Collins and Storfer 2003; McCallum and Trauth 2003) but have not gone on to analyze outcomes with respect to these factors.
Unlike many previous studies, this study uses a human epidemiologic methodology to assess an ecologic problem (with possible relevance to human health). Evaluation of multiple stressors will require continued development of effective study methodologies and a movement away from the single-agent hypothesis testing of the past (Rabinowitz et al. 1999). To our knowledge, this is one of the first studies to use multivariate techniques (which require a large sample size) to evaluate the relative importance of multiple factors while adjusting for possible confounding. This reduces the possibility that an observed association between a risk factor (e.g., agriculture) and the outcome of interest (limb malformation) is due in fact to a confounding risk factor that is linked with both the exposure and the outcome in such a way that spurious etiologic associations are inferred.
What is the possible biologic basis of the observation that proximity to agricultural land use is associated with risk for amphibian limb malformation? We argue that exposure to sources of anthropogenic pollution is a likely explanation. Without identification of specific compounds, any specific hypothesis remains speculative. However, agricultural runoff may include a variety of chemicals, including pesticides, animal wastes, and fertilizers, that could threaten water quality. One study of an agricultural watershed found evidence of 38 pesticides in water samples, including 30 herbicides, 4 fungicides, 3 insecticides, and 1 metabolite of one of the herbicides (Kreuger 1998). Water-borne toxicants in agricultural runoff could directly affect development, either singly or in chemical mixtures (Burkhart et al. 2000). Agricultural runoff has been associated with impaired hatching success (De et al. 2002). Other lipophilic chemicals in runoff may also play a role—recent studies found that exposing Rana pipiens tadpoles to extracted compounds from semipermeable membrane devices in freshwater ponds could cause malformation in the presence of UV radiation (Bridges et al. 2004).
There are a number of biologic mechanisms by which chemicals in agricultural runoff could cause deformities. A recent study found an association between agricultural runoff including the herbicides atrazine, deethyl-atrazine, simazine, metolachlor, dimethenamide, chlopyralide, dicamba, and bentazone and plasma retinoid levels in Rana catesbiana (Berube et al. 2005); retinoids appear to function in signaling pathways for limb development (Stocum 2000). It is also possible that herbicides or other organic compounds could directly affect the expression of genes that help determine limb development. Because thyroid hormone has been found to protect against the development of abnormalities in the experimental setting (Fort et al. 2001), chemically induced alterations of thyroid function could potentially increase the malformation rate.
Another potential mechanism for the effect of agricultural chemicals on amphibian malformation could be through effects on either parasite population density or on host immune function leading to increased rate of parasite infection (Christin et al. 2003). Our investigation considered the possibility that parasitic infection with Ribeiroia could have played a role in the etiology of limb deformities at the study sites. However, our analysis to date does not support the parasitic infection hypothesis. Examination of individual amphibians as well as representative samples of host snails from the study sites did not reveal evidence that either this parasite or trematode infection in general was playing a role in the causation of limb malformation.
Some studies have suggested that nutrient loads such as phosphorus and nitrogen, present in some agricultural runoff, could contribute to developmental abnormality risk by contributing to eutrophication (Johnson and Chase 2004). In this sample, we had the opportunity to adjust for the effect of measured nutrients in sampled water and found that these nutrients did not remain significant predictors of malformation in the multivariate model.
There are several possible reasons why the overall rate of nontraumatic hind limb abnormality in this study sample (1.6%) was lower than figures reported for amphibian limb abnormalities in some other surveys (Burkhart et al. 2000). First, there was variation in rates of malformations between study sites, with rates in excess of 10% at some sites. The systematic nature of the present survey was designed to reduce selection bias, and most study sites were consequently selected without prior knowledge of the site-specific rate of amphibian limb abnormalities. As a result, in some locations, no abnormalities were detected. Second, we used a more restrictive case definition than many other surveys, taking care to exclude trauma-induced limb deformities, thereby lowering the overall abnormality rate. Third, rates of abnormalities fluctuate from year to year in individual wetlands, and some of the ponds sampled had experienced abnormality rates as high as 30% in previous years (Levey et al. 2003). It therefore seems likely that the variability in malformation rates between ponds was related to environmental factors.
The study relied on a limited number of measurements of water quality, as well as visual observations of land use as an indicator of potential pollution sources. Therefore, the observed associations between runoff sources, measured water quality, and malformation rate must be viewed as preliminary. Determining exposure at the level of the group rather than the individual risks the “ecologic fallacy” of assigning potential causation to one factor when in fact another, unmeasured factor is actually responsible. Although the rate of malformation in ponds near agriculture was increased, it could be that this difference was due to other factors that were not assessed. Further studies involving direct measurements of toxicants in water, and case–control comparisons of affected and unaffected individuals need to be carried out in order to identify candidate toxicants or chemical mixtures that could be investigated in the laboratory setting.
Water-quality measures are dependent on multiple factors and vary over a season and year to year depending on runoff sources, rainfall, temperature, and other factors. By using a limited number of water-quality measures taken during a single year, it is possible that some exposure misclassification occurred, with water conditions during sensitive periods of amphibian development varying from the measurements at the time of sampling. However, such misclassification, in epidemiologic studies of cause and effect, tends to be nondifferential and bias the results toward observing no effect (Checkoway and Eisen 2005). Therefore, the fact that we found strong associations between a number of exposures and the rate of malformation indicates that such associations are real and significant. The good discrimination and fit of the multivariate logistic regression model further support the strength of the observed associations.
The human health relevance of these findings remains to be determined. If a chemical toxicant or mixture could be identified that is capable of causing limb malformations in wild amphibians, several additional lines of investigation could shed light on whether a risk to humans exists. First, it would be useful to know what human health outcome would be analogous to amphibian limb deformities; this could be explored if gene sequence homology between humans and amphibians was present for target genes affected by the toxicants. Second, it would be important to establish a dose–response relationship and determine whether environmental exposures to such a toxicant would be capable of causing human health effects. Finally, epidemiologic studies including both amphibian and human populations could determine whether rates of malformation in amphibians show a correlation with a suspected human health outcome. Although to date none of these connections have been established, they will be worthy of further exploration if confirmatory studies support the role of toxicant chemicals in the etiology of amphibian developmental abnormalities.
We thank the field and laboratory research staff: S. Bolden, N. Cothran, B. Fellman, K. Freidenberg, N. Freidenfelds, and N. Rabinowitz. M.R. Cullen provided assistance with study design and analysis. M. Hines of the U.S. Geological Survey assisted with the geographic information system coding of land use. R. Levey provided critical logistical support and advice.
This study was funded in part by the National Institutes of Health/National Science Foundation Ecology of Infectious Diseases Program (National Institute of Environmental Health Sciences grant 5R01ES011067).
Table 1 Characteristics of the study population (n = 5,264).
Malformation typea
Species No. (%) Gosner stage (mean ± SD) Malformation [n (%)] Missing Malformed Extra
Hyla versicolor 235 (4.5) 36.0 ± 6.4 4 (1.7) 1 3 0
Pseudacris crucifer 895 (17.0) 34.8 ± 6.2 6 (0.7) 3 3 0
Total hylids 1,130 (21.5) 35.0 ± 6.2 10 (0.9) 4 6 0
Rana catesbeiana 319 (6.1) 37.7 ± 6.7 5 (1.6) 2 3 0
Rana clamitans 1,176 (22.3) 40.3 ± 6.2 32 (2.7) 9 28 2
Rana pipiens 1,702 (32.3) 35.8 ± 8.0 33 (1.9) 10 28 0
Rana sylvatica 937 (17.8) 31.7 ± 5.2 3 (0.3) 0 3 0
Total ranids 4,134 (78.5) 36.3 ± 7.5 73 (1.8) 21 62 2
Total 5,264 (100) 36.0 ± 7.2 83 (1.6) 25 68 2
a Some individuals exhibited more than one type of malformed limb or element.
Table 2 Exposure assessment for study sites (n = 42).
Characteristic Exposure
Proximity to pollution sources [n (%)]
Agriculture nearby or adjacent 17 (40.5)
Lawn nearby or adjacent 15 (35.7)
Water-quality measures
pH 7.6 ± 0.9 (5.7–9.7)
Conductivity (μS) 303.3 ± 307.8 (10–1,350)
Dissolved oxygen (mg/L) 6.5 ± 3.4 (1.2–18.3)
Temperature (°C) 21.2 ± 3.6 (12.8–29.1)
Total nitrogen (mg/L) 0.8 ± 0.8 (0.1–5.4)
Total phosphorus (mg/L) 0.2 ± 0.4 (0.004–2.2)
μS, micro-Siemens. Values shown are mean ± SD (range) except where indicated.
Table 3 Risk factors for nontraumatic limb malformation.
Bivariate analysis
Multivariate modela
Risk factor OR (95% CI) p-Value OR (95% CI) p-Value
Characteristic
Gosner stage 1.20 (1.14–1.26) < 0.0001 1.18 (1.13–1.24) < 0.0001
Genus (Rana vs. Hyla) 2.01 (1.04–3.91) 0.04 — —
Proximity to pollution sources
Agriculture (nearby or adjacent) 3.08 (1.96–4.85) < 0.0001 2.26 (1.42–3.58) < 0.001
Septic system or lawn adjacent 2.06 (1.34–3.19) < 0.001 — —
Water-quality measures
Conductivity (μS) 1.00 (1.00–1.00) 0.46 — —
Dissolved oxygen (mg/L) 1.14 (1.06–1.23) < 0.001 — —
Temperature (°C) 1.05 (0.98–1.13) 0.14 — —
Total nitrogen (mg/L) 1.22 (0.91–1.63) 0.18 — —
μS, micro-Siemens.
a Certain variables were removed from the model during backward selection modeling (—; exclusion criteria, p > 0.05).
==== Refs
References
Ankley GT Diamond SA Tietge JE Holcombe GW Jensen KM Defoe DL 2002 Assessment of the risk of solar ultraviolet radiation to amphibians. I. Dose-dependent induction of hindlimb malformations in the northern leopard frog (Rana pipiens ) Environ Sci Technol 36 2853 2858 12144258
Ankley GT Tietge JE DeFoe DL Jensen KM Holcombe GW Durhan EJ 1998 Effects of ultraviolet light and methoprene on survival and development of Rana pipiens Environ Toxicol Chem 17 2530 2542
Berube V Boily MH DeBlois C Dassylva N Spear PA 2005 Plasma retinoid profile in bullfrogs, Rana catesbeiana , in relation to agricultural intensity of sub-watersheds in the Yamaska River drainage basin, Quebec, Canada Aquat Toxicol 71 109 120 15642636
Bishop CA Mahony NA Struger J Ng P Pettit K 1999 Anuran development, density and diversity in relation to agricultural activity in the Holland River watershed, Ontario, Canada (1990–1992) Environ Monit Assess 57 21 43
Blaustein AR Johnson PT 2003 Explaining frog deformities Sci Am 288 2 60 65 12561459
Bridges C Little E Gardiner D Petty J Huckins J 2004 Assessing the toxicity and teratogenicity of pond water in north-central Minnesota to amphibians Environ Sci Pollut Res 11 233 239
Burkhart JG Ankley G Bell H Carpenter H Fort D Gardiner D 2000 Strategies for assessing the implications of malformed frogs for environmental health Environ Health Perspect 108 83 90 10620528
Burkhart JG Helgen JC Fort DJ Gallagher K Bowers D Propst TL 1998 Induction of mortality and malformation in Xenopus laevis embryos by water sources associated with field frog deformities Environ Health Perspect 106 841 848 9831545
Checkoway H Eisen EA 2005. Epidemiology. In: Textbook of Clinical Occupational and Environmental Medicine (Rosenstock L, Cullen MR, Brodkin CA, Redlich CA, eds). Philadelphia:Elsevier Saunders, 119–137.
Christin MS Gendron AD Brousseau P Menard L Marcogliese DJ Cyr D 2003 Effects of agricultural pesticides on the immune system of Rana pipiens and on its resistance to parasitic infection Environ Toxicol Chem 22 1127 1133 12729224
Collins JP Storfer A 2003 Global amphibian declines: sorting the hypotheses Divers Distrib 9 89 98
Daszak P Cunningham AA Hyatt AD 2001 Anthropogenic environmental change and the emergence of infectious diseases in wildlife Acta Trop 78 103 116 11230820
Davidson C Shaffer HB Jennings MR 2001 Declines of the California red-legged frog: climate, UV-B, habitat, and pesticides hypotheses Ecol Appl 11 464 479
Davidson C Shaffer HB Jennings MR 2002 Spatial tests of the pesticide drift, habitat destruction, UV-B, and climate-change hypotheses for California amphibian declines Conserv Biol 16 1588 1601
De S Pettit KE Bishop CA Cheng KM Elliott JE 2002 Effects of agricultural runoff on native amphibians in the Lower Fraser River Valley, British Columbia, Canada Environ Toxicol Chem 21 353 360 11833805
Degitz SJ Holcombe GW Kosian PA Tietge JE Durhan EJ Ankley GT 2003 Comparing the effects of stage and duration of retinoic acid exposure on amphibian limb development: chronic exposure results in mortality, not limb malformations Toxicol Sci 74 139 146 12773776
Degitz SJ Kosian PA Makynen EA Jensen KM Ankley GA 2000 Stage- and species-specific developmental toxicity of all-trans retinoic acid in four native North American ranids and Xenopus laevis Toxicol Sci 57 264 274 11006356
Fort DJ Propst TL Stover EL Helgen JC Levey RB Gallagher K 1999 Effects of pond water, sediment, and sediment extracts from Minnesota and Vermont, USA, on early development and metamorphosis of Xenopus Environ Toxicol Chem 18 2305 2315
Fort DJ Rogers RL Paul RR Miller MF Clark P Stover EL 2001 Effects of pond water, sediment and sediment extract samples from New Hampshire, USA on early Xenopus development and metamorphosis: comparison to native species J Appl Toxicol 21 199 209 11404831
Gosner KL 1960 A simplified table for staging anuran embryos and larvae Herpetology 16 183 190
Johnson P Chase JM 2004 Parasites in the food web: linking amphibian malformations and aquatic eutrophication Ecol Lett 7 521 526
Johnson P Sutherland D 2003 Amphibian deformities and Ribeiroia infection: an emerging helmithiasis Trends Parasitol 19 332 335 12901930
Johnson PTJ Lunde KB Haight RW Bowerman J Blaustein AR 2001 Ribeiroia ondatrae (Trematoda: Digenea) infection induces severe limb malformations in western toads (Bufo boreas ) Can J Zool 79 370 379
Johnson PTJ Lunde KB Thurman EM Ritchie EG Wray SN Sutherland DR 2002 Parasite (Ribeiroia ondatra e) infection linked to amphibian malformations in the western United States Ecol Monogr 72 151 168
Kreuger J 1998 Pesticides in stream water within an agricultural catchment in southern Sweden, 1990–1996 Sci Total Environ 216 227 251 9646531
LaClair JJ Bantle JA Dumont J 1998 Photoproducts and metabolites of a common insect growth regulator produce developmental deformities in Xenopus Environ Sci Technol 32 1453 1461
Levey R Shambaugh N Fort D Andrews J 2003. Investigations into the Causes of Amphibian Malformations in the Lake Champlain Basin of New England. Waterbury, VT:Vermont Department of Environmental Conservation.
Linder G Krest SK Sparling D eds. 2001. Amphibian Decline: An Integrated Analysis of Multiple Stressor Effects. Pensacola, FL:Society of Environmental Toxicology and Chemistry (SETAC).
McCallum ML Trauth SE 2003 A forty-three year museum study of northern cricket frog (Acris crepitans ) abnormalities in Arkansas: upward trends and distributions J Wildl Dis 39 522 528 14567212
Meteyer CU Loeffler IK Fallon JF Converse KA Green E Helgen JC 2000 Hind limb malformations in free-living northern leopard frogs (Rana pipiens ) from Maine, Minnesota, and Vermont suggest multiple etiologies Teratology 62 151 171 10935979
Mizgireuv I Flax N Borkin LJ Khudoley VV 1984 Dysplastic lesions and abnormalities in amphibians associated with environmental conditions Neoplasma 31 175 181 6717687
Ouellet M 2000. Amphibian deformities: current state of knowledge. In: Ecotoxicology of Amphibians and Reptiles (Sparling DW, Linder G, Bishop CA, eds). Pensacola, FL:Society of Environmental Toxicology and Chemistry (SETAC), 617–646.
Ouellet M Bonin J Rodrigue J DesGranges J Lair S 1997 Hindlimb deformities (ectromelia, ectrodactyly) in free living anurans from agricultural habitats J Wildl Dis 33 95 104 9027696
Rabinowitz PM Cullen M Lake H 1999 Wildlife as sentinels for human health hazards: a review of study designs J Environ Med 1 217 223
Ray N Lehmann A Joly P 2002 Modeling spatial distribution of amphibian populations: a GIS approach based on habitat matrix permeability Biodivers Conserv 11 2143 2165
Stocum DL 2000 Frog limb deformities: an “eco-devo” riddle wrapped in multiple hypotheses surrounded by insufficient data Teratology 62 147 150 10935978
U.S. Geological Survey 1999. LandLandcov_NLCDVT: Vermont Land Cover Data Set 1992. U.S. Geological Survey. Vermont Center for Geographic Information. Available: http://www.vcgi.org/ [accessed 10 February 2004].
van der Schalie WH Gardner HS Bantle JA De Rosa CT Finch RA Reif JS 1999 Animals as sentinels of human health hazards of environmental chemicals Environ Health Perspect 107 309 315 10090711
|
16263502
|
PMC1310909
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 11; 113(11):1497-1501
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7585
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7655ehp0113-00150216263503ResearchFraming Scientific Analyses for Risk Management of Environmental Hazards by Communities: Case Studies with Seafood Safety Issues Judd Nancy L. 123Drew Christina H. 12Acharya Chetana 13Mitchell Todd A. 5Donatuto Jamie L. 56Burns Gary W. 7Burbacher Thomas M. 13Faustman Elaine M. 123Marine Resources for Future Generations 41 Department of Environmental and Occupational Health Services,2 Institute for Risk Analysis and Risk Communication, and3 Center for Ecogenetics and Environmental Health, University of Washington, Seattle, Washington, USA4 Korean Women’s Association (Tacoma, Washington, USA), Indochinese Cultural and Service Center (Tacoma, Washington, USA), Tacoma–Pierce County Health Department (Tacoma, Washington, USA), Citizens for a Healthy Bay (Tacoma, Washington, USA), Washington Department of Fish and Wildlife (Olympia, Washington, USA), and Washington Department of Health (Tumwater, Washington, USA)5 Swinomish Indian Tribal Community, Water Resources Program, La Conner, Washington, USA6 Department of Resource Management, University of British Columbia, Vancouver, British Columbia, Canada7 Shoalwater Bay Indian Tribe Environmental Laboratory, Tokeland, Washington, USAAddress correspondence to E.M. Faustman, Institute for Risk Analysis and Risk Communication, 4225 Roosevelt Way NE #100, Seattle, WA 98105-6099, USA. Telephone: (206) 616-4299. Fax: (206) 616-4875. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 27 6 2005 113 11 1502 1508 12 10 2004 27 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Risk management provides a context for addressing environmental health hazards. Critical to this approach is the identification of key opportunities for participation. We applied a framework based on the National Research Council’s (NRC) analytic–deliberative risk management dialogue model that illustrates two main iterative processes: informing and framing. The informing process involves conveying information from analyses of risk issues, often scientific, to all parties so they can participate in deliberation. In the framing process, ideas and concerns from stakeholder deliberations help determine what and how scientific analyses will be carried out. There are few activities through which affected parties can convey their ideas from deliberative processes for framing scientific analyses. The absence of participation results in one-way communication. The analytic–deliberative dialogue, as envisioned by the NRC and promoted by the National Institute of Environmental Health Sciences (NIEHS), underscores the importance of two-way communication. In this article we present case studies of three groups—an Asian and Pacific Islander community coalition and two Native American Tribes—active in framing scientific analyses of health risks related to contaminated seafood. Contacts with these organizations were established or enhanced through a regional NIEHS town meeting. The reasons for concern, participation, approaches, and funding sources were different for each group. Benefits from their activities include increased community involvement and ownership, better focusing of analytical processes, and improved accuracy and appropriateness of risk management. These examples present a spectrum of options for increasing community involvement in framing analyses and highlight the need for increased support of such activities.
Asian and Pacific Islanderscase studiescommunitiescommunity-based participatory researchframingrisk managementseafoodtribal nations
==== Body
Risk management provides a context for addressing environmental health hazards. Critical to this approach is the identification of key opportunities for participation. Ideally, affected parties are involved early and throughout the decision process through continuing dialogue. The reality, however, often falls short of this ideal. The involvement of affected parties is commonly limited to community members’ being informed of the results collected and assessed by scientific experts and decision makers. Increasingly, communities are provided the opportunity to comment on documents or studies that are presented to them in near-final form, but rarely is community input used to frame and provide context at the outset of the studies themselves. Here we explore some of the opportunities and challenges of broader community participation within the theoretical structure of the risk-management paradigm. We begin by presenting a model of the analytic–deliberative risk-management framework, with an emphasis on framing activities in this structure, and then present three examples that illustrate community approaches to framing exercises. Using examples of activities by communities with concerns about seafood safety, we explore a range of options for increasing community involvement in shaping the scientific approaches used in risk management.
Our examples come primarily from established connections between University of Washington researchers and community partners. These connections originated or were developed more fully through the National Institute of Environmental Health Sciences (NIEHS) Center for Ecogenetics and Environmental Health’s town meeting, “Voices for Healthy Environments, Healthy Communities,” held in Seattle, Washington in September 2000. The NIEHS Center for Ecogenetics and Environmental Health (CEEH) researchers and staff interacted with > 300 participants, representing > 40 community groups, tribal nations, legislators, and agencies, in challenging discussions of race, poverty, and pollution. This was one of 16 town meetings supported by NIEHS across the country as part of NIEHS’s commitment to developing a research agenda responsive to community needs (O’Fallon et al. 2003). The case studies presented here provide lessons for expanding community participation in designing environmental health risk research questions (framing) under various circumstances including the rationale for community action, differences in resources, and involvement of scientific experts.
Community Involvement in Analytic–Deliberative Risk-Management Dialogue
Involving affected parties at all major phases of the risk decision process is an important component of nearly all risk-management paradigms (National Research Council 1996, 2000a). Affected parties should be allowed to express their own needs and help shape objectives for risk management. However, involvement is challenging. One barrier to effective participation is not involving affected parties early enough in the process. This is often seen when fish contamination problems are addressed and when fish advisories are issued (Jardine 2003). Other barriers include information failing to reach communities, the lack of awareness of some environmental health issues, and the varying degrees of scientific understanding. Experience, skills, scientific training, and local knowledge and values can vary considerably among participants working on a decision process, and agencies may not have access to important local knowledge, may not understand what affected parties care about, or may not be aware of behaviors that affect exposure to contamination. Researchers need help with all these issues to appropriately address risk concerns.
The National Academy of Sciences (NAS) has repeatedly called for early, active, continuous, and transparent community involvement in risk-influenced activities (National Research Council 1996, 2000a, 2000b). Understanding Risk: Informing Decisions in a Democratic Society (National Research Council 1996) offers a detailed framework for improving complex decision processes. It describes an analytic–deliberative process, in which theories, results, and scientific analyses inform the deliberative processes used to discuss and determine the appropriate course of action. At the same time, the deliberative processes frame the scientific analyses. During the many decision phases, the participants (public officials, scientists, and interested/affected parties) interact and participate in the analysis and deliberation.
To facilitate our implementation of the NAS framework, we adapted the original NAS framework to specifically highlight the interplay among the analytic–deliberative processes (Figure 1; Drew et al. 2003; National Research Council 1996). The trio of participants (affected parties, technical specialists, and decision makers) is fundamental to the process, and each group should participate in all phases. Moreover, individuals may participate as members of more than one group, depending on training, experience, and their role in the decision process. Little attention has been paid to the information needs inherent to the analytic–deliberative process (Drew et al. 2004). Generally, more attention has been given to the informing aspects than to the framing aspects, and more tools have been developed to support the analytic aspects of the processes than the deliberative aspects. As a consequence, participation in the framing process, especially by affected parties, is often limited.
Sometimes involvement activities are too focused on one-way information flow: from those who are making decisions (such as government agencies) to those who are being informed. Most involvement paradigms call for two-way information flow, but they offer few specific recommendations for facilitating this, particularly for increasing participation in designing research questions (Drew et al., in press). Various public participation models and tools offer opportunities to inform, consult, involve, collaborate with, and empower affected/interested parties [International Association for Public Participation (IAP2) 2000; Renn et al. 1995].
Community-based participatory research (CBPR) provides a tool for expanding community involvement in research projects and potentially for increasing participation by affected community members (O’Fallon and Dearry 2002). The NIEHS defines CBPR as a methodology that promotes active community involvement in the processes that shape research and intervention strategies and that promotes involvement in the conduct of research studies. The CBPR approach is designed to apply more generally to environmental health issues of concern, to ensure meaningful involvement by community members.
These CBPR principles of early and active community engagement also apply to increasing community involvement in all aspects of the analytic–deliberative risk dialogue. An advantage of considering environmental health issues in a risk context is that risk-management science is directed toward providing information for decision making and dealing with uncertainties (Faustman and Omenn 2001; Morgan and Henrion 1990). The principles of CBPR can be achieved more easily when the analytic–deliberative approach is applied in its ideal form (i.e., when all interested and affected parties are involved in informing and framing processes). Using example case studies, we discuss options for moving beyond processes that simply inform affected communities to processes that involve communities in framing relevant scientific questions.
Informing
Informing makes information from analytic processes (often scientific or research) accessible to all parties, so community members may more fully participate in deliberative (risk-management) discussions. In the context of fish contamination issues, community involvement is often limited to informing activities. There is a growing literature describing, evaluating, and improving these activities, most related to the issuance of fish advisories (Burger et al. 2003; Connelly and Knuth 1998; Jardine 2003; Knuth et al. 2003; Shubat et al. 1996). A common theme from many of these studies is the need for two-way communication and earlier involvement by communities. The analytic–deliberative dialogue (Figure 1) is an iterative process and can be flexible as new information becomes available and new participants join the process.
Issuing fish advisories is often not an iterative process, however. Once advisories are issued, the public is wary of reusing a fishing resource that once was declared unsuitable (Jardine 2003). The issuance of advisories tends to be a top-down process, as decisions about acceptable risks and alternatives are often made without including affected parties. Such top-down processes may not be appropriate for all consumption and cultural groups [Columbia River Inter-Tribal Fish Commission (CRITFC) 1994; Sechena et al. 1999; Shubat et al. 1996; Suquamish Tribe 2000; Toy et al. 1996]. This is reflected in the increasing number of fish advisory evaluations calling for early involvement (Jardine 2003). In more advanced models, information flow is two-way but is still limited to informing activities such as risk communication about fish contamination (Burger et al. 2003; Jardine 2003; Knuth et al. 2003). This might include community partners developing fish advisories (informing) without being involved in the scientific analysis used to shape the advisory (framing) (Figure 1). Without real meaningful involvement during the framing steps, informing processes will not be as significant to affected communities.
Framing
Framing allows concerns that arise through deliberative processes to shape analyses (Figure 1). This presents the potential for major expansion of community involvement in the risk-management process. In particular, there are many such possibilities in risk management of contaminated fishing resources.
As noted previously, efforts in framing activities have been limited (Jardine 2003; Knuth et al. 2003). Reasons for this can include a lack of communication among community groups, technical specialists, and decision makers, leading to nontransparent decision processes (Drew et al. 2004, in press). In other words, how do researchers and decision makers select a research agenda or a decision process after environmental hazards or issues are recognized? Another reason affected parties are not involved in framing research more often is that there are limited funds dedicated to support involvement up front.
There are several benefits of expanding participation in framing research questions for all parties involved in the analytic–deliberative dialogue. Community participation may result in the design of more effective analyses (Bierle 2002; Drew et al., in press; Israel et al. 2001). This participation may also promote research addressing community needs, community acceptance of the processes, understanding of environmental health risks, and informed behavior changes (Jardine 2003). Moreover, community involvement in framing may increase overall dialogue and thereby improve informing processes essential to risk management.
In our experience, more effort has been focused on informing than on framing risk questions and risk management activities (Drew et al. 2004, in press; Judd et al. 2003b; Polifka and Faustman 2002). Our objective here is to report several community framing activities that have shaped how analytical processes (research) will be carried out to assess the safety of fish consumption. By exploring similarities and differences across the three examples, we hope to present a range of framing approaches that may also be appropriate for other groups.
Case Studies of Communities Involved in Framing
We have had the privilege of collaborating with several dynamic communities that are proactively addressing their environmental health concerns. Here we highlight their efforts in framing aspects of the analytic–deliberative risk-management process. Common themes across these examples, including challenges and benefits, are explored using a case-study approach (Yin 1994). These descriptive case studies document collaboration between university researchers and community, tribal, and agency partners. The three case studies describe interactions with Marine Resources for Future Generations, the Swinomish Indian Tribal Community, and the Shoalwater Bay Indian Tribe. These interactions have been through participation on advisory boards, and the importance of relationship building has been key. All three groups are located in Washington State, and the importance of fish and seafood in each is high. Recent seafood consumption surveys indicate that average tribal and Asian and Pacific Islander (API) community members consume 3–10 times the amount of fish and shellfish of average U.S. consumers [Sechena et al. 1999; Suquamish Tribe 2000; Toy et al. 1996; U.S. Environmental Protection Agency (EPA) 1997a]. High-end tribal consumers may eat 20 times the amount of average U.S. consumers (Suquamish Tribe 2000; U.S. EPA 1997a). In addition, sources and types of fish and shellfish consumed differ from community to community (Judd et al. 2004). Traditional diets and reliance on subsistence fishing/harvesting contribute to the higher consumption rates of tribal and API community members.
Each of these groups has concerns about specific contaminants (e.g., polychlorinated biphenyls, biotoxins, pesticides, and methyl-mercury) in seafood they eat regularly. Our previous studies indicated that the specific collection, preparation, and consumption practices of tribes and API communities may place them at greater exposure to some contaminants. Additionally, our studies have shown that monitoring practices by some regulatory agencies may not be sufficient to evaluate or protect these vulnerable groups from the potential health risks (Judd et al. 2003a,b).
Each community has its own story of how their efforts to address potential health risks from consuming contaminated seafood began and how they eventually became active in framing activities.
Marine Resources for Future Generations.
The Marine Resources for Future Generations (MRFFG) program began in 1997 in Pierce County, Washington. The initial mission of the group was to ensure the safe and wise use of seafood resources and compliance with state regulations, such as licensing and appropriate shellfish collection, by API communities in the county. The group includes two social service organizations: the Korean Women’s Association (KWA), which serves the Korean, Samoan, and Filipino communities, and the Indochinese Cultural and Service Center (ICSC), which serves the Vietnamese, Cambodian, and Laotian communities. Government agencies and nongovernmental partners provide support and educational resources and make MRFFG a strong coalition. The connection with the University of Washington (UW) was made at the NIEHS town meeting during a seafood safety break-out session, and UW staff has since attended the monthly meetings, provided technical advice, and assisted in MRFFG projects.
For many API communities, seafood is an important part of both nutrition and cultural traditions, making seafood safety a very pressing matter. The Washington Department of Fish and Wildlife (WDFW) was concerned that their usual methods of education (multilingual brochures and signs) were not reaching many API community members. The MRFFG group began when KWA and ICSC joined with WDFW to address illegal harvesting issues, including shellfish collection from closed and contaminated beaches. Other partners soon joined, and over the years the group’s efforts have expanded to include many other issues including non–point-source pollution, mercury in fish, and invasive species. An initial condition for participating agencies and organizations is a long-term commitment, not just a pilot project effort. This has been key to the success of the group that has held monthly meetings for the last 7 years, even as grant support has waxed and waned.
Early on, MRFFG’s educational outreach found that the sources of seafood sold at local markets were unknown. This was an important issue for the group to address. As community members began to understand that some beaches were not safe for harvesting shellfish, they wanted to know the source of the seafood they purchased in markets. At the same time, the group was concerned that education about local contamination had led people to believe that seafood from anywhere else (besides local contaminated beaches) would be cleaner. MRFFG launched its own effort to investigate the sources of local seafood. This project is an excellent example of community-driven framing of problems in the risk-management process because these efforts focused on developing and pursuing scientific questions to better understand potential health risks.
The main goal of the project was to talk with local vendors and determine the sources of their seafood. If the seafood was local, they wanted to know specifically which beach it was from and whether it was legally harvested, as well as the sources of imported seafood. Another goal of the project was to provide education about the health importance of regulations for collection and sale of seafood to vendors. The businesses were all within API communities, and MRFFG wanted to support these businesses by providing them with information to help ensure community health, which would ultimately also benefit retailers. MRFFG’s multilingual youth administered the surveys in a nonthreatening manner, collecting information, not enforcing regulations.
Fourteen youth participated and visited 10 stores in Tacoma and Seattle, serving mostly Korean, Vietnamese, Cambodian, Samoan, and Filipino community members. Results indicated that the stores were importing from overseas most of the fish they sold, and this choice was driven by both customer and owner preferences. Seventy percent were aware of health dangers related to seafood, but at least 20% of the stores had no awareness of health dangers associated with shellfish contamination or illegal harvesting. MRFFG concluded that they needed to increase awareness of seafood safety issues to ensure community health. This process began with providing literature from the WDFW and the Washington Department of Health. Thus, this framing and analysis project fed into an informing process in an iterative way and expanded community involvement.
MRFFG continues educational efforts with local shopkeepers to ensure the safety of the seafood they sell. They have also begun investigation and education efforts with stores about the environmental dangers of importing invasive species. These community-driven efforts have promoted community health while encouraging community businesses. Outside groups, even those fluent in Asian languages, could not have performed this investigation and education process as effectively as the youth because the business owners might have perceived a threat (in the form of an enforcement action), and they might not have provided information.
Funding for MRFFG projects, such as this one, have come from a variety of sources, including U.S. EPA headquarters and Region 10, regional foundations, the Puget Sound Water Quality Action Team, the Russell Family Foundation, and several MRFFG member organizations. The group has also successfully obtained funds through competitive processes geared primarily toward community organizations. Despite funding being an annual uncertainty, MRFFG has effectively leveraged their resources to address community seafood safety concerns. The group’s longevity rests in the continued commitment of its members that extends beyond the funding period of one grant or project. The efforts of MRFFG also demonstrate that community groups with limited resources can engage in framing activities that empower them to make more educated decisions about managing environmental health risks.
Shoalwater Bay Indian Tribe.
The Shoalwater Bay Indian Tribe is concerned about the potential impacts of environmental quality on their health for several reasons. The Shoalwater Bay Indian Reservation (SBIR) is located on Willapa Bay, in the most isolated rural area of northern Pacific County in Washington State. The tribal community includes just 237 people (Shukovsky 2002). Fish and seafood are major dietary components for the Shoalwater; these resources have very important traditional and spiritual roles in tribal communities (CRITFC 1994; Suquamish Tribe 2000; Toy et al. 1996). Although a small tribe, the Shoalwater must deal with a large variety of environmental issues. One of the biggest of these is the widespread commercial use of pesticides on lands surrounding the reservation. Diazinon has been sprayed over the nearby cranberry bogs to kill fire worms, which destroy the plants. Railroad ties, heavily treated with a fungicide to prevent rotting, are situated throughout the bogs. The pesticide carbaryl is applied to the many oyster beds around Shoalwater Bay (and Willapa Bay, a larger connected body of water) in an effort to retard ghost shrimp populations. The tideflats are also sprayed routinely with glyphosate to control Spartina, a destructive weed. Other environmental concerns include the presence of fecal coliform and marine biotoxins. Harmful algal blooms that release biotoxins, such as saxitoxin and domoic acid, have led to several recent beach closures for shellfish harvesting [Commission on Asian Pacific American Affairs (CAPAA) 2004; Washington Department of Health (WADOH) 2004]. Additionally, many septic systems on or adjacent to the reservation are failing (Laundry Alternative 2004; Puget Sound Action Team 2004). All of these factors may affect shellfish quality.
In the mid-1990s, the U.S. EPA conducted several environmental assessments (water, air and soil quality) in the region (U.S. EPA 1997b). These investigations, made in response to a high prenatal and neonatal mortality rate within the Shoalwater Tribal community, have been limited in scope. The final report recommended further testing at additional sample sites to provide more complete information (U.S. EPA 1997b).
In September 2000, Shoalwater leaders attended the CEEH’s town meeting and voiced their concerns to NIEHS Director Kenneth Olden. As a result of this meeting, the NIEHS provided support to enable the CEEH’s Community Outreach and Education Program (COEP) and the Shoalwater’s Environmental Division to work together. This effort represents one of many projects implemented by Shoalwater’s Environmental Division, most of which are administered and managed internally. Their new on-site environmental laboratory has increased the ability of the tribe to engage in many framing and analytical activities independently to address their environmental health risk concerns. Additionally, to holistically address health concerns on the reservation, the Shoalwater constructed a new health clinic and have developed intensive prenatal care and well-baby programs.
The Shoalwater, in collaboration with COEP and the Institute for Risk Analysis and Risk Communication (IRARC), has used NIEHS support to engage in framing tribal environmental concerns. The Shoalwater developed a seafood consumption survey tool and a shellfish quality management plan. Both the shellfish plan and the survey tool were included in a proposal submitted to the Administration for Native Americans (ANA) that has since received funding. The ANA project described monitoring subsistence food species that are consumed by tribal members for environmental contaminants. This approach was favored by most tribal members, who were surveyed using a pilot seafood-consumption survey tool. The results will be used to create a prioritized list of the species to be tested for contaminants. The results of these tests will be incorporated into the tribal management plan to assess the shellfish quality in Willapa Bay. The Shoalwater Tribe is also awaiting response on other research proposals submitted to U.S. EPA and NIEHS. These include studies to look at seafood contamination in the context of other dietary risk factors and, when funded, will use technical contacts at the University of Washington.
The Shoalwater have faced many difficulties, but they have maximized their resources to address their concerns. Proposal development can be a daunting task, particularly for communities with many competing priorities and limited technical, material, and human resources. The Shoalwater Tribe has successfully developed competitive proposals that will enable them to more fully frame and analyze their environmental health risk concerns.
Swinomish Indian Tribal Community.
The Swinomish Tribe’s research project, Bioaccumulative Toxics in Native American Shellfish (BTNAS), is another example of a tribal community framing their own questions. The Swinomish Reservation is located on the shores of central Puget Sound and is home to 1,000 Native Americans, of whom 700 are enrolled Swinomish members. Swinomish Tribe members are concerned about environmental contamination threatening their traditional use of resources, particularly shellfish. There are numerous potential sources of contamination within a mile radius of the reservation, including petrochemical and industrial facilities, landfills, municipal sewer outfalls, two marinas, two boatyards, log storage facilities, and agricultural land treated with pesticides and fertilizers. The Swinomish Tribe has initiated investigation into the potential contamination of water, sediments, and shellfish. The purpose of the project is to ensure safety and promote continuation of healthy, traditional lifestyles and/or to begin proactively addressing cleanup and mitigation of contaminated sources. The Swinomish Tribe requested that a screening study of contaminants be performed in Padilla and Fidalgo Bays by the Washington State Department of Ecology. The initial study indicated the presence of numerous persistent pollutants, including arsenic and polychlorinated dibenzofurans (PCDFs) (Johnson et al. 1997). Later studies indicated the need for additional sampling to understand the magnitude and the health implications of the contamination (Johnson 1999, 2000).
Shellfish contamination represents one of a number of threats to the Swinomish maintaining their traditional lifestyle. It is extremely important to the Swinomish that the effort to investigate the contamination and potential health risks be performed by the Swinomish Tribe. The Swinomish have significant internal resources, including several environmental scientists with advanced degrees, an on-site chemistry lab, and an ongoing shellfish monitoring program funded by the U.S. EPA and the Bureau of Indian Affairs (BIA), primarily for paralytic shellfish monitoring. Moreover, this is an issue of sovereignty. The Swinomish Tribe prefers to control how such a study is conducted to ensure that it addresses (frames) the Swinomish Tribe’s environmental health concerns and that the information gathered is used and interpreted by the Swinomish Tribe.
In summer of 2000, the Swinomish Planning Office and their intern (funded by the Environmental Careers Organization) developed the BTNAS proposal. Although the Swinomish Tribe possessed the infrastructure required to develop an in-depth environmental sampling, analysis, risk management, and education plan with a significant cultural component, they were unfamiliar with the complexities of a federal grant application. The Swinomish sought help with this technical challenge at CEEH. Additionally, at the NIEHS town meeting, the Swinomish Tribe submitted their concerns related to the difficulties of the grant proposal procedure for communities unfamiliar with the federal funding process. Providing feedback to agencies that clearly have a mandate and desire for community-based research should make it easier for communities with the capacity to receive grants directly.
With final approval from the Swinomish Tribe’s governing council, the grant was submitted and received favorably by the NIEHS, but was not funded. It was, however, recommended to the U.S. EPA, and in 2002, the Swinomish Tribe was awarded the largest-ever U.S. EPA research grant to a tribal nation. The Swinomish Tribal Planning Office had the core staff and resources to take on a project of this magnitude, in addition to many other ongoing water quality projects. The project necessitated hiring new staff for the many new responsibilities and activities. Currently, IRARC and COEP researchers act as advisors to the BTNAS project and have assisted and/or acted as principal investigators for subsequent grant applications. So far the BTNAS project has collected two seasons of field samples, and sample analysis is in progress. The planning office staff has been annually updating the Swinomish General Council on the progress of the BTNAS project. The Swinomish Annual Report and the free monthly tribal newsletter, Keeyoks, provide information to tribal members about BTNAS project developments. Additionally, the Swinomish environmental education program works in the public schools, providing outreach and education on local environmental health issues.
More recently, the Swinomish organized a meeting of environmental scientists from several nearby tribes to discuss common concerns, upcoming funding opportunities, and approaches for sharing resources. This meeting was significant in that it was organized by the tribes, for the tribes. The BTNAS project has also been presented at several scientific meetings.
The BTNAS project is another good example of a community framing their own environmental health questions. To pursue the specific questions of the Swinomish Tribe about the condition of the local environment and safe consumption of shellfish, a technical approach is needed. The Swinomish Tribe has the resources to develop a plan, obtain funding, and pursue these questions. Because the Swinomish Tribe developed the plan, it addresses their needs while maintaining tribal sovereignty through tribal control of research activities, findings, and interpretation. The Swinomish Tribe Planning Office is in an optimal position to inform the tribal community about the project and incorporate community feedback for framing future activities. The ongoing activities illustrate how the Swinomish Tribe is using information from this research to evaluate their risks from shellfish exposure.
Summary of Community Experiences with Framing Activities
A challenge for researchers is determining how to work with communities to understand how their questions are framed and how to incorporate this process in their research programs. This challenge has been identified in previous work, such as involving communities in risk-management processes related to cleanup and transportation of nuclear waste (Drew et al. 2003, in press). In that case and the case examples presented here, the challenges are unique to each situation and require significant time investments and resources for the communities and the collaborators. This has also been identified through numerous CBPR projects (O’Fallon and Dearry 2001, 2002; O’Fallon et al. 2003; Seifer 2000; Thompson et al. 2001). The examples of efforts by MRFFG, the Swinomish Tribe, and the Shoalwater Bay Indian Tribe illustrate a range of opportunities for communities in framing activities. Each community had different issues and approaches, including who was involved, how the effort was financed, and the types of outcomes. The various outcomes are summarized in Table 1, and the many common themes that the groups shared are described in Appendix 1.
The MRFFG project presents a grass-roots approach to addressing community problems. After embarking on an educational effort (informing) to reduce community exposures to contaminants in locally collected shellfish, the group recognized the importance of assuring the safety of seafood at local markets. This work grew out of their original mission, which had not included investigatory work. However, as the group framed the question of local markets’ sources of seafood, they found that they lacked information. Undaunted, they took the initiative and pursued the information themselves (Table 1). This was done primarily by leveraging limited funds from government and private sources. MRFFG drew on expertise and support from all its members: community youth, elders, county and federal agencies, and nonprofit and academic partners. This example demonstrates that groups that do not typically perform scientific investigations can perform framing activities and that framing and analysis can be done with limited resources if the group has a strong commitment to addressing the question. By internally carrying out the study, the community has ownership of the activity and can better facilitate community education and dialogue about the results. Developing and pursuing these questions internally fosters community interest, support, and positive action to address problems.
The Shoalwater Bay Indian Tribe’s effort to develop a proposal to investigate contaminated shellfish represents a very different approach that began with support from government agencies (NIEHS) and collaboration between their own scientists and outside scientific experts. The Tribe engages in many research efforts to ensure a healthy community. In this particular example, the Tribe investigated potential shellfish contamination in collaboration with outside researchers (Table 1). This preliminary investigation, itself a framing exercise, was used in several subsequent research proposals, some that have been funded and some that are pending. Thus, the community was able to obtain support, both financial and technical, specific for its framing efforts. This has led to the development of successful research proposals specific to the Tribe’s questions and concerns.
The Swinomish Tribe has had an ongoing shellfish-monitoring program, but this was not adequate to address concerns about bioaccumulative toxicants in shellfish. The Swinomish Tribe obtained funding and is currently engaged in research including iterative framing of questions about shellfish contamination (Table 1). It has been paramount for the Swinomish to have tribal autonomy over the scientific questions asked, project execution, data collection, and data interpretation. The information collected will be used to evaluate current and future risks from shellfish exposures. The Tribe received some help from academic researchers with the grant application process, in addition to technical and outreach expertise.
Thus, the examples presented here demonstrate a range of possibilities in terms of the questions asked, the way they were formulated and pursued, how experts were involved, and how they were funded. Some projects leveraged limited funds from a variety of sources to pursue their concerns, and some obtained resources specifically for framing questions, which they then used for research and/or in developing more complete proposals. Tribes are in a unique situation with regard to applying for funding in that they, as sovereign nations, often have more developed infrastructures than many community groups. They are also eligible for some tribal-funding sources (e.g., BIA and U.S. EPA) that cannot be pursued by other communities.
Despite many differences in their problems and approaches, many common themes emerged from the experiences of these communities (Appendix 1). Some common benefits of framing that are shared across the groups include research that better meets community needs and increased community ownership. These examples also show how framing can help build internal knowledge and capacity. For all the groups, environmental issues are among many competing issues, and the process of framing may be outside the usual scope of the group’s activities. Finally, trust and connections beyond the community may also be needed, and the process of framing may develop as many or more questions than it addresses. These commonalities highlight benefits and challenges that may be part of framing by other communities and can be helpful in determining the potential utility of the process and in anticipating some of its difficulties.
Conclusion
This article has presented three case studies of successful community action in framing scientific analyses of environmental health risks. We used the NRC’s analytic–deliberative process to think about the different components needed for CBPR in a risk context. The analytic–deliberative process prompts us to pay special attention to roles communities can play in both framing research questions and in informing and educating all parties involved in the risk process. Framing is an integral part of the analytic–deliberative risk process and can open important opportunities for two-way dialogue and communication among researchers and community/tribal partners. Few accounts in the literature have shown how this happens and why. We have presented three case studies related to seafood safety that illustrate how the framing process can work. The efforts of these case study groups and their partners have opened opportunities and empowered them to address their environmental health risk concerns.
The case studies present a range of possibilities for communities to be involved in framing activities. These projects span different environmental problems with communities using a variety of approaches, including how (or if) outside experts were involved and how the effort was funded. There were elements of framing and informing in each of the examples, demonstrating the interconnectedness and importance of both. Many common themes from their experiences emerged, including how framing helped in capacity building, how they balanced competing concerns, and how the communities benefited. However, given the pressure to deliver maximum production for grant dollars spent, there is little incentive for researchers and agency staff to engage in activities that are not mandated, may not be recognized as results, and are likely to be time and resource intensive. Increasing community and tribal participation through framing and CBPR requires significant investments of time and resources by all the collaborators. Given the value of this broader involvement, funding agencies should recognize, encourage, and even mandate community involvement to specifically frame and address environmental health risk issues. This research direction will ultimately lead to more relevant and realistic environmental health risk management solutions.
The Pacific Northwest Center for Human Health and Oceans is funded by the National Institute of Environmental Health Sciences (NIEHS; P50ES012762) and by the National Science Foundation (OCE-0434087); the Center for Eco-genetics and Environmental Health is funded by the NIEHS (P30ES07033) and the U.S. Environmental Protection Agency STAR (R-829467).
Appendix 1. Common Themes from Framing Exercises
Benefits
Community framing better addresses local environental health risk concerns
Community framing leads to better communication of research processes and findings
Greater incentive for community framing if community and tribes are involved in data analysis
While outside resources are useful, it is important that the community maintains ownership and that framing is driven by local needs Capacity building
Application processes to obtain funding are challenging for groups not traditionally involved in grant writing
Internal community/tribal technical resources may be limiting; framing processes may help build organizational capacity
Long-term commitment by all partners is required for framing processes to succeed Competing issues
Environmental health risks are among many competing community priorities
Framing processes (e.g., developing environmental risk questions and/or proposals and actions to pursue them) may be outside the scope of general activities of many organizations Other
Importance of trust with collaborating partners (regarding commitments to partnerships, tribal sovereignty, or enforcement issues)
Interconnections within small communities facilitates communication and participation
Framing is an expanding process and often leads to more questions
Figure 1 Model of the analytic–deliberative risk process adapted from Drew et al. (2003) and the National Research Council (1996).
Table 1 Summary of case study framing activities and outcomes
Group Issue Framing activity Outcome
MRFFG Concern about the sources of seafood and seafood safety at community stores Community youth interviewed local merchants Better characterization of seafood sources and improved understanding of potential exposure and risks from these
Shoalwater Bay Indian Tribe Concern about local shellfish contamination’s impact on community health Developed an assessment plan and submitted a grant proposal to fund research Obtained funding to sample shellfish for contaminants and to perform seafood consumption surveys
Swinomish Indian Tribal Community Concern about local shellfish contamination’s potential effect on current and future resource use Expanded existing infrastructure for shellfish monitoring to include bioaccumulative biotoxicants Data collected is being used to evaluate current and future risk from shellfish exposures
==== Refs
References
Bierle TC 2002 The quality of stakeholder based decisions Risk Anal 22 4 739 749 12224747
Burger J McDermott MH Chess C Bochenek E Perez-Lugo M Pflugh KK 2003 Evaluating risk communication about fish consumption advisories: Efficacy of a brochure versus a classroom lesson in Spanish and English Risk Anal 23 4 791 803 12926571
CAPAA 2004. Shellfish harvesting alert for marine biotoxin. Olympia, WA:Commission on Asian Pacific American Affairs. Available: http://www.capaa.wa.gov/doh.html [accessed 10 May 2004].
Connelly NA Knuth BA 1998 Evaluating risk communication: Examining target audience perceptions about four presentation formats for fish consumption health advisory information Risk Anal 18 5 649 659 9853397
CRITFC 1994. A Fish Consumption Survey of the Umatilla, Nez Perce, Yakama, and Warm Springs Tribes of the Columbia River Basin. Technical report 94-3. Portland, OR:Columbia River Inter-Tribal Fish Commission.
Drew CH Grace D Silbernagel S Hemmings E Smith A Griffith WC 2003 Nuclear waste transportation: case studies of identifying stakeholder risk information needs Environ Health Perspect 111 263 272 12611653
Drew CH Kern M Martin T Blazek M Power M Faustman EM In press. The Hanford openness workshops: Fostering open and transparent decision making at the department of energy. In: Research on Social Problems in Public Policy.
Drew CH Nyerges T Leschine T 2004 Promoting transparency of long term environmental decisions: the Hanford decision mapping system pilot project Risk Anal 24 6 1641 1664 15660618
Faustman EM Omenn G 2001. Risk assessment. In: Cassarett and Doull’s Toxicology (Klaassen C, ed). 6th ed. San Francisco, CA:McGraw-Hill, 83–104.
IAP2 2000. The IAP2 public participation spectrum. Denver, CO:International Association for Public Participation. Available: http://www.iap2.org [accessed 11 February 2002].
Israel B Lichtenstein R Lantz P McGranaghan R Allen A Guzman RJ 2001 The Detroit community-academic urban research center: development implementation and evaluation J Public Health Manag Pract 7 5 1 19 11680026
Jardine CG 2003 Development of a public participation and communication protocol for establishing fish consumption advisories Risk Anal 23 3 461 471 12836839
Johnson A 1999. Investigation of Chemical Contamination at Whitmarsh Landfill and Padilla Bay Lagoon. Olympia, WA:Washington State Department of Ecology.
Johnson A 2000. Sediment Quality on the West Side of Inner Fidalgo Bay. Olympia, WA:Washington State Department of Ecology.
Johnson A Serdar D Davis D 1997. Survey for Petroleum and Other Chemical Contaminants in the Sediments of Fidalgo Bay. Olympia, WA:Washington Department of Ecology.
Judd NL Griffith WC Faustman EM 2004 Consideration of cultural and lifestyle factors in defining susceptible populations for environmental disease Toxicology 198 1–3 121 133 15138036
Judd NL Griffith WC Kalman DA Faustman EM 2003a Assessment of PCB congener analytical methods: do they meet risk assessment needs? Arch Environ Contam Toxicol 44 1 132 139 12434228
Judd NL O’Neill SM Kalman DA 2003b Are seafood PCB data sufficient to assess health risk for high seafood consumption groups? Hum Ecol Risk Assess 9 3 691 707
Knuth BA Connelly NA Sheeshka J Patterson J 2003 Weighing health benefit and health risk information when consuming sport-caught fish Risk Anal 23 6 1185 1197 14641893
Laundry Alternative 2004. Pollution threatens shellfish in 12 counties. Middletown Springs, VT:Laundry Alternative Inc. Available: http://www.laundry-alternative.com/Septic_Systems/failing_septic_wa1.htm [accessed 10 May 2004].
Morgan GM Henrion M 1990. Uncertainty: A Guide to Dealing with Uncertainty in Quantitative Risk and Policy Analysis. New York:Cambridge University Press.
National Research Council 1996. Understanding Risk: Informing Decisions in a Democratic Society. Washington, DC: National Academy Press.
National Research Council 2000a. Long-Term Institutional Management of U.S. Department of Energy Legacy Waste Sites. Washington, DC:National Academy Press.
National Research Council 2000b. A Risk Management Strategy for PCB-contaminated Sediments. Washington, DC:National Academy Press.
O’Fallon LR Dearry A 2001 Commitment of the National Institute of Environmental Health Sciences to community-based participatory research for rural health Environ Health Perspect 109 suppl 3 469 473 11427398
O’Fallon LR Dearry A 2002 Community-based participatory research as a tool to advance environmental health sciences Environ Health Perspect 110 suppl 2 155 159 11929724
O’Fallon LR Wolfle GM Brown D Dearry A Olden K 2003 Strategies for setting a national research agenda that is responsive to community needs Environ Health Perspect 111 1855 1860 14644657
Polifka JE Faustman EM 2002 Developmental toxicity: web resources for evaluating risk in humans Toxicology 173 1–2 35 65 11955683
Puget Sound Action Team 2004. Gov. Gary Locke awards grant for shellfish protection. Olympia, WA:Puget Sound Action Team. Available: http://www.psat.wa.gov/News/releases/GO_05_03.htm [accessed 10 May 2004].
Renn O Webler T Wiedemann P 1995. Fairness and Competence in Citizen Participation: Evaluating Models for Environmental Discourse. Dordrecht, the Netherlands:Kluwer Academic Publishers.
Sechena R Nakano C Liao S Polissar NL Lorenzana R Truong S 1999. Asian and Pacific Islander seafood consumption study. EPA 910/R-99-003. Seattle, WA:EPA Office of Environmental Assessment.
Seifer SD 2000 Engaging colleges and universities as partners in healthy communities initiatives Public Health Rep 115 2–3 234 237 10968760
Shubat PJ Raatz KA Olson RA 1996 Fish consumption advisories and outreach programs for southeast Asian immigrants Toxicol Ind Health 12 3–4 427 434 8843559
Shukovsky P 2002. Tribe’s future surviving after 10-year plague of lost children. Seattle Post-Intelligencer (Seattle, WA), June 19.
Suquamish Tribe 2000. Fish Consumption Survey of the Suquamish Indian Tribe of the Port Madison Reservation, Puget Sound Region. Suquamish, WA:Suquamish Tribe.
Thompson B Coronado G Puschel K Allen E 2001 Identifying constituents to participate in a project to control pesticide exposure in children of farmworkers Environ Health Perspect 109 suppl 3 443 448 11427394
Toy KA Polissar NL Liao S Mittelstaedt GD 1996. A Fish Consumption Study of the Tulalip and Squaxin Island Tribes of the Puget Sound Region. Marysville, WA:Tulalip Tribes, Department of Environment.
U.S. EPA 1997a. Exposure Factors Handbook, Vol. 2. Food Ingestion Factors. Washington, DC:U.S. Environmental Protection Agency.
U.S. EPA 1997b. The Shoalwater Bay Reservation: A Limited Environmental Assessment. 910?r-96-013. Washington, DC:U.S. Environmental Protection Agency.
WADOH 2004. Recreational shellfish closures due to biotoxins or pollution. Olympia, WA:Washington Department of Health. Available: http://ww4.doh.wa.gov/scripts/esrimap.dll?name=bioview&Cmd=Map&Step=1 [accessed 10 May 2004].
Yin RK 1994. Case Study Research: Design and Methods. 2nd ed. Thousand Oaks, CA:Sage.
|
16263503
|
PMC1310910
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 27; 113(11):1502-1508
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7655
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7927ehp0113-00150916263504ResearchAssociations of Uric Acid with Polymorphisms in the δ-Aminolevulinic Acid Dehydratase, Vitamin D Receptor, and Nitric Oxide Synthase Genes in Korean Lead Workers Weaver Virginia M. 12Schwartz Brian S. 123Jaar Bernard G. 23Ahn Kyu-Dong 4Todd Andrew C. 5Lee Sung-Soo 4Kelsey Karl T. 6Silbergeld Ellen K. 7Lustberg Mark E. 8*Parsons Patrick J. 9Wen Jiayu 1Lee Byung-Kook 41 Division of Occupational and Environmental Health, Department of Environmental Health Sciences, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA2 Department of Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA3 Department of Epidemiology, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA4 Institute of Industrial Medicine, SoonChunHyang University, Asan, South Korea5 Department of Community and Preventive Medicine, Mount Sinai School of Medicine, New York, New York, USA6 Department of Cancer Cell Biology, Harvard School of Public Health, Boston, Massachusetts, USA7 Division of Environmental Health Engineering, Department of Environmental Health Sciences, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA8 Department of Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, Maryland, USA9 Lead Poisoning/Trace Elements Laboratory, Wadsworth Center, New York State Department of Health, Albany, New York, USAAddress correspondence to B.-K. Lee, Institute of Industrial Medicine, SoonChunHyang University, 646 Eupnae-Ri, Shinchang-Myun, Asan-Si, Choongnam, 336-745 South Korea. Telephone: 82-41-530-1760. Fax: 82-41-530-1778. E-mail:
[email protected]* Current address: Department of Medicine, York Hospital, York, PA.
The authors declare they have no competing financial interests.
11 2005 27 6 2005 113 11 1509 1515 14 1 2005 27 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Recent research suggests that uric acid may be nephrotoxic at lower levels than previously recognized and that it may be one mechanism for lead-related nephrotoxicity. Therefore, in understanding mechanisms for lead-related nephrotoxicity, it would be of value to determine whether genetic polymorphisms that are associated with renal outcomes in lead workers and/or modify associations between lead dose and renal function are also associated with uric acid and/or modify associations between lead dose and uric acid. We analyzed data on three such genetic polymorphisms: δ-aminolevulinic acid dehydratase (ALAD), endothelial nitric oxide synthase (eNOS), and the vitamin D receptor (VDR). Mean (± SD) tibia, blood, and dimercaptosuccinic acid–chelatable lead levels were 37.2 ± 40.4 μg/g bone mineral, 32.0± 15.0 g/dL, and 0.77± 0.86 μg/mg creatinine, respectively, in 798 current and former lead workers. Participants with the eNOS Asp allele had lower mean serum uric acid compared with those with the Glu/Glu genotype. Among older workers (age ≥ median of 40.6 years), ALAD genotype modified associations between lead dose and uric acid levels. Higher lead dose was significantly associated with higher uric acid in workers with the ALAD1-1 genotype; associations were in the opposite direction in participants with the variant ALAD1-2 genotype. In contrast, higher tibia lead was associated with higher uric acid in those with the variant VDR B allele; however, modification was dependent on participants with the bb genotype and high tibia lead levels. We conclude that genetic polymorphisms may modify uric acid mediation of lead-related adverse renal effects.
δ-aminolevulinic acid dehydrataseendothelial nitric oxide synthasegenetic susceptibility factorskidney functionlead exposureuric acidvitamin D receptor
==== Body
Increasing evidence suggests that lead exposure may contribute to decreased renal function (Kim et al. 1996; Lin et al. 1999, 2003; Muntner et al. 2003; Payton et al. 1994; Staessen et al. 1992; Yu et al. 2004) and increased uric acid levels (Lin et al. 2002, 2001; Shadick et al. 2000) at lower doses than previously recognized, particularly in susceptible populations. Similarly, an increasing body of literature suggests that uric acid may be nephrotoxic at lower levels than currently appreciated (Johnson et al. 2003). In previous analyses in the same population of lead workers reported here, we found associations between lead dose and increased uric acid in older workers and evidence that uric acid may mediate some lead-related adverse renal effects (Weaver et al. 2005). Therefore, to understand the mechanisms of lead-related nephrotoxicity, it would be helpful to determine whether genetic polymorphisms that are associated with renal outcomes in lead workers and/or modify associations between lead dose and renal function are also associated with uric acid and/or modify associations between lead dose and uric acid. In this study we evaluated three such genetic polymorphisms: δ-aminolevulinic acid dehydratase (ALAD), the BsmI polymorphism of the vitamin D receptor (VDR) gene, and the Glu298Asp polymorphism of the endothelial nitric oxide synthase (eNOS) gene. We examined associations of these polymorphisms and their effect modification on associations of three lead dose biomarkers in models of uric acid in a cross-sectional analysis of data from the first of three evaluations of 798 current and former Korean lead workers in a longitudinal study of the adverse health effects of inorganic lead exposure.
Materials and Methods
Study design and population.
We performed a cross-sectional analysis of first evaluation data from 798 current and former lead workers enrolled in a longitudinal study of the adverse health effects of inorganic lead exposure. All participants provided written, informed consent. The study protocol was approved by institutional review boards at the SoonChunHyang University and the Johns Hopkins University Bloomberg School of Public Health. Participation in the study was voluntary, and workers were paid approximately $30 for their time and effort. As previously described (Schwartz et al. 2001; Weaver et al. 2003a), workers were recruited from 26 different facilities including lead battery, lead oxide, lead crystal, and radiator manufactures and secondary lead smelting. No medical exclusionary criteria (e.g., blood pressure, renal disease) were applied. Study participants were not currently occupationally exposed to other known renal toxicants.
Data collection.
Data collection was completed either at the Institute of Industrial Medicine of the SoonChunHyang University in Chonan or at the study plants, using previously reported methods (Schwartz et al. 2001; Weaver et al. 2003a). Data and biologic specimens collected included a standardized questionnaire on demographics and medical and occupational history; blood pressure; height and weight measurement; a blood specimen [for blood lead, blood urea nitrogen (BUN), serum creatinine, uric acid, and genotyping]; a spot urine sample [for N-acetyl-β-d-glucosaminidase (NAG), retinol-binding protein (RBP), and creatinine]; and tibia lead concentration. A 4-hr urine collection after oral administration of 10 mg/kg dimercaptosuccinic acid (DMSA) was also obtained to measure DMSA-chelatable lead and creatinine clearance (787 participants completed this collection).
Laboratory methods.
We measured lead biomarkers and renal outcomes using previously reported methods (Schwartz et al. 2001; Weaver et al. 2003a). In brief, blood lead was measured (Fernandez 1975) with a Zeeman background-corrected atomic absorption spectrophotometer (model 8100; Hitachi Ltd. Instruments, Tokyo, Japan) at the Institute of Industrial Medicine, a certified reference laboratory for lead in Korea. Tibia lead was assessed via a 30-min measurement of the left midtibia diaphysis using 109Cd to fluoresce the K-shell X-rays of lead. The lead X-rays were recorded with a radiation detector and then quantified and compared with calibration data to estimate the concentration of lead in bone (Todd and Chettle 1994; Todd and McNeill 1993). All point estimates, including negative values, were retained in the statistical analyses, to minimize bias and to avoid censoring of data (Kim et al. 1995). Urine lead levels in the 4-hr collection were measured at the Wadsworth Center of the New York State Department of Health (Albany, NY, USA) by electrothermal atomic absorption spectrometry with Zeeman background correction (model 4100ZL; Perkin-Elmer, Norwalk, CT, USA) (Parsons and Slavin 1999).
We measured BUN, serum creatinine, and uric acid using an automatic chemical analyzer (model TBA 40FR biochemical analyzer; Toshiba, Tokyo, Japan). Urine creatinine was measured with the Sigma kit (Sigma Chemical Company, St. Louis, MO, USA). Measured creatinine clearance was defined as [(urinary creatinine in milligrams per deciliter × urine volume in milliliters) ÷ serum creatinine in milligrams per deciliter] ÷ collection time in minutes. Calculated creatinine clearance was obtained from the Cockcroft-Gault equation (Cockcroft and Gault 1976). We measured NAG using the PPR NAG test kit (PPR Diagnostics, Ltd., London, UK), and RBP was measured using a modification of the method of Topping et al. (1986).
We isolated DNA for genotyping from whole blood samples using the QIAamp blood kit (QIAGEN, Hilden, Germany). The ALAD polymorphism assayed (δ-aminolevulinic acid dehydratase NCBI accession no. AY319481) results in two alleles: ALAD1 and the variant, ALAD2, which has a G-to-C transversion at codon 177. The protocol for ALAD genotyping involved an initial amplification that generated a 916 bp fragment (Schwartz et al. 1995; Wetmur et al. 1991; Ziemsen et al. 1986). A second amplification, using a pair of nested primers (kindly provided by J. Wetmur), generated an 887 bp fragment. The amplified fragment was cleaved at the diagnostic Msp1 site on the ALAD2 allele. The Glu298Asp polymorphism of the eNOS gene (endothelial nitric oxidase synthase; NCBI accession no. X76307) involves a G-to-T transversion at nucleotide position 894 within exon 7, which results in substitution of aspartic acid for glutamic acid at codon 298; the variant allele is referred to as the Asp or T allele. Genotype was determined by a modification of the assay of Hibi et al. (1998) as previously described (Weaver et al. 2003b). The VDR BsmI polymorphic site in intron 8 [vitamin D (1,25-dihydroxyvitamin D3) receptor; NCBI accession no. AY342401] consists of the common allele, denoted b, and a variant, denoted B, in which the BsmI restriction enzyme site is absent. Amplification used primers originating in exon 7 and intron 8 as previously published (Schwartz et al. 2000b).
Statistical analysis.
The primary goals of our analysis were to examine associations of ALAD, VDR, and eNOS genotype with uric acid in lead workers while controlling for other covariates and to evaluate whether ALAD, VDR, and eNOS genotypes modified associations between the lead dose biomarkers (tibia lead, blood lead, DMSA-chelatable lead) and uric acid while controlling for other covariates. Statistical analysis was completed using SAS software, version 8.2 (SAS Institute, Cary, NC, USA).
Initially, we examined variable distributions. The distributions of NAG and RBP showed departures from normality and were thus ln-transformed; the adequacy of this transformation was subsequently confirmed by examination of residuals from the final regression models. Linear regression modeling with a dichotomous genotype variable was used to compare uric acid by genotype, while controlling for the same covariates used in the interaction models. For ALAD, participants with the ALAD1-2 genotype were compared with the reference group of participants with ALAD1-1 genotype. Because of small numbers, all analyses combined homozygous and heterozygous variant genotype carriers for VDR (BB and Bb) and eNOS (Glu/Asp and Asp/Asp). Linear regression modeling with genotype and lead variable cross-product terms was used to evaluate effect modification by genotype on associations between lead biomarkers and uric acid. Models with tibia lead were repeated after removal of data from participants with the common allele whose tibia lead levels were higher than the range in those with the variant allele. Covariate selection for final regression models used a priori variables [age, sex, body mass index (BMI; defined as weight in kilograms divided by the square of height in meters)] in modeling that included other biologically relevant variables in separate models, as previously described (Weaver et al. 2005). Covariates retained included age, sex, BMI, categorical alcohol consumption (current, former, never), and, when noted in specific models, systolic blood pressure and serum creatinine. Models are presented with and without these last two variables because their interrelatedness with lead dose and uric acid suggests that neither modeling approach is entirely adequate (Weaver et al. 2005). Continuous independent variables were centered at the mean.
Because effect modification by age on associations between lead biomarkers and uric acid was observed in this population (Weaver et al. 2005), we dichotomized the population by median age and repeated both types of models described above (main effect and interaction) to evaluate associations of genotype in models of uric acid and effect modification by genotype on associations between lead biomarkers and uric acid in different age groups. Last, a model that included cross-products of tibia lead and combined ALAD/VDR genotype subsets was evaluated in participants in the older half of the population to determine if the genotype association previously reported in this population (lead workers with the ALAD1-1 genotype were statistically less likely to have the VDR bb genotype; Schwartz et al. 2000a) was a factor in the effect modification observed in this study. Only two participants in the older age group had both variant alleles (ALAD1-2/VDR Bb or BB), which precluded meaningful analysis of this group; this subset was removed from the model. The final model assessed differences in the slopes of the relations between tibia lead and uric acid in the group of participants with the ALAD1-2/VDR bb genotypes and the ALAD1-1/VDR Bb or BB genotypes, in separate cross-product terms, compared with the reference group of participants with the ALAD1-1 and VDR bb genotypes.
As in previous analyses (Weaver et al. 2003a), we evaluated models for linear regression assumptions and the presence of outlying points using added variable plots (Weisberg 1985), which are graphical summaries of the relation between y and a particular x adjusted for the other covariates. For each plot, two lines were overlaid: the regression line, and a line determined by a scatter plot smoothing method (lowess) that calculates a locally weighted least squares estimate for each point in the scatter plot (Cleveland 1979). This allows an examination of the data for outliers that are overly influential, as evidenced by inconsistency between the lowess and regression lines. When warranted, models were repeated without outliers. Models were also assessed for collinearity by examining variance inflation factors and conditional indices.
Results
Selected demographics, exposure, and health outcome measures.
A total of 79 (9.9%) participants were heterozygous for the ALAD2 allele, and none was homozygous (Table 1). For VDR, 85 (10.7%) were genotype Bb, and 4 (0.5%) were BB. For eNOS, 114 (14.4%) participants were genotype Glu/Asp, and 6 (0.7%) were Asp/Asp. Mean (± SD) and frequencies for selected demographic, exposure, and outcome variables are presented by genotype in Table 1. Mean uric acid was normal in all genotype subsets. Medians for selected measures are presented by genotype in groups dichotomized by median age (Table 2).
ALAD.
Mean uric acid did not differ by ALAD genotype in the total population or in either age group (dichotomized by median age of 40.6 years). Among all participants, ALAD genotype modified the association between blood lead and uric acid. β-Coefficients were 0.0047 (p = 0.10) for blood lead in the reference group of those with the ALAD1-1 genotype and −0.0158 (p = 0.05) for the cross-product term of ALAD1-2 genotype and blood lead, respectively (thus, the slope of the relation between blood lead and uric acid in participants with the ALAD1-2 genotype was −0.0111). This effect was confined to workers at or above the median age (Table 3, panels 1 and 2, blood lead models). The inverse nature of these associations by genotype, from the blood lead model in panel 2 of Table 3, is shown graphically in Figure 1. Borderline effect modification on the association between tibia lead and uric acid was also observed in older participants (Table 3, panel 2, tibia lead models). Consistent with known mechanisms for the hyperuricemic effects of lead (Weaver et al. 2005), β-coefficients decreased after additional adjustment for blood pressure and renal function (Table 3, panel 3). Results were similar when calculated creatinine clearance was added to these models instead of serum creatinine (data not shown). Added variable plots (Figure 2) of partial residuals of the tibia lead model (Table 3, panel 2) suggested different tibia lead level ranges by genotype. In order to compare differences over a similar range, data from 43 participants in the ALAD1-1 group with tibia lead levels > 89 μg Pb/g bone mineral (levels above the crude range in participants with the ALAD1-2 genotype) were removed. A positive association between tibia lead and uric acid in those with the ALAD1-1 genotype (Table 3, truncated tibia lead models, panel 2) was then observed.
VDR.
Similar to ALAD, VDR genotype was not associated with uric acid levels, and effect modification was confined to older lead workers (Table 4, tibia lead models, panels 1 and 2). However, in contrast to ALAD, tibia lead was positively associated with uric acid in those participants with the variant VDR B allele, and effect modification on associations between blood lead and uric acid was not observed (Table 4). Modeling that included cross-products of tibia lead with combined genotype subsets compared with a reference group of participants with both the ALAD1-1 and VDR bb genotypes revealed that the opposite direction associations between tibia lead and uric acid in participants with the VDR B allele compared with those with the ALAD1-2 allele were not simply due to the genotype associations previously reported in this population (lead workers with the ALAD1-1 genotype were statistically less likely to have the VDR bb genotype; Schwartz et al. 2000a) (Table 5). Similar to ALAD, the range of tibia lead levels differed by genotype (Figure 3). In contrast, when data from 27 participants with the VDR bb genotype with tibia lead levels > 103 μg Pb/g bone mineral (above the range in participants with VDR Bb or BB genotypes) were removed in order to compare differences over a similar range, effect modification was no longer significant (Table 4, truncated tibia lead models, panel 2).
eNOS.
In contrast to ALAD and VDR, the eNOS genotype was associated with uric acid levels. Mean serum uric acid was lower in participants with the eNOS Asp allele compared with those with the Glu/Glu genotype (β= −0.1913; SE β= 0.0932; p = 0.04; adjusted for age, sex, BMI, serum creatinine, systolic blood pressure, and blood lead). Additional modeling in two groups, dichotomized by median age, revealed that this association was confined to participants in the older half of the population. No effect modification by eNOS genotype on associations of lead biomarkers and uric acid was observed.
Discussion
We evaluated whether polymorphisms in three genes (ALAD, VDR, and eNOS) were associated with uric acid or modified relations of lead biomarkers with uric acid in a cross-sectional analysis of data from the first evaluation in a longitudinal study of 798 current and former Korean lead workers. After adjustment, participants with the eNOS Asp allele had lower mean uric acid. Effect modification by ALAD on associations between lead dose and uric acid was observed in participants whose ages were in the older half of the age range. Higher lead dose was associated with higher uric acid in workers with the ALAD1-1 genotype; associations were in the opposite direction in participants with the ALAD1-2 genotype. VDR genotype modified the association of tibia lead and uric acid, also in the older half of the population. However, in contrast to ALAD, higher tibia lead was associated with higher uric acid in those with the variant B allele, but this effect was dependent on data from participants with the bb genotype and high tibia lead levels.
The ALAD enzyme is a principal lead-binding protein. The isozymes in those with the variant ALAD2 allele bind a greater proportion of blood lead than does the isozyme in individuals with the ALAD1-1 genotype (Bergdahl et al. 1997). In the population that is the focus of this report, mean blood lead was higher in participants with the ALAD1-2 genotype compared with those with the ALAD1-1 genotype (Schwartz et al. 2000a). However, neither tibia nor DMSA-chelatable lead levels differed significantly. Other studies have also reported that participants with the ALAD2 allele have higher blood lead levels than do those with the ALAD1-1 genotype, although studies in populations with mean blood lead levels < 10 μg/dL have generally not reported a difference (Hu et al. 2001; Kamel et al. 2003; Kelada et al. 2001). The limited data on associations of ALAD genotype with bone lead levels reveal no difference in some studies (similar to results in our population) (Fleming et al. 1998; Smith et al. 1995). Lower cortical (tibia) and/or trabecular (patella, calcaneus) bone lead levels in those with the ALAD2 allele have been reported in others (Hu et al. 2001; Kamel et al. 2003). Other toxicokinetic differences have also been reported in participants with the ALAD2 allele (Fleming et al. 1998; Hu et al. 2001; Schwartz et al. 1997; Smith et al. 1995). Overall, these data suggest that tighter binding of lead by the isozymes of the ALAD2 allele decreases lead sequestration in bone. Therefore, the impact on toxicity would depend on whether enzyme bound lead is bioavailable and more toxic than lead that is stored in bone and subsequently released. As a result, toxicity may differ by target organ.
Two studies have examined the impact of ALAD genotype on serum uric acid levels in lead exposed populations. In a study of 691 construction workers, whose mean blood lead was 7.8 μg/dL, Smith et al. (1995) found borderline (p = 0.09) higher mean uric acid after adjustment for age, alcohol ingestion, and blood lead in the 96 participants with the ALAD2 allele compared with those with the ALAD1-1 genotype. Effect modification was not evaluated. Wu et al. (2003) found no difference in uric acid by genotype in 709 participants in the Normative Aging Study (mean blood lead = 6.2 μg/dL). However, effect modification by genotype on associations of patella and tibia lead with uric acid of borderline significance (p < 0.1) was observed; β-coefficients (and slopes) were greater in those with the variant allele. This difference was significant in participants whose patella lead levels were > 15 μg/g bone mineral (p = 0.04).
Synthesizing our data with these studies is complicated by the inverse associations we observed. Similar inverse relations between blood and DMSA-chelatable lead and renal outcomes in participants with the ALAD2 allele were previously reported in this population (Weaver et al. 2003b). Because uric acid is filtered at the glomerulus, levels in serum are also influenced by renal function. Therefore, the associations between lead dose and uric acid may be due to the same process causing inverse associations between lead dose and renal function in those with the ALAD2 allele. If this process represents lead-related hyperfiltration, the associations between lead dose and uric acid may become positive as this population ages and eventually, as in the Normative Aging Study, be stronger in those with the ALAD2 allele than in those with the ALAD1-1 genotype. However, several steps are involved in the renal handling of uric acid, including a secretion step in the proximal tubule, which is also thought to be adversely affected by lead. The fact that effect modification by ALAD genotype on associations between tibia lead and uric acid persists after adjustment for renal function suggests that one or more of the other renal handling processes for uric acid are involved. Analysis of our longitudinal data will be very helpful in understanding these complex relations.
The potential for uric acid–related nephrotoxicity must be considered in these associations, as well. When previously published models of effect modification by ALAD on associations between lead dose and renal function (Weaver et al. 2003b) were also controlled for uric acid, the effect modification observed remained statistically significant (data not shown), indicating that this effect is not mediated solely through uric acid. However, the current analyses do provide further evidence, in addition to our recently published results (Weaver et al. 2005), that lead dose increases uric acid in these workers, even after control for variables that are both confounders and mediators, such as blood pressure and renal function.
Polymorphisms of the gene encoding for the VDR are of interest in lead-exposed populations because of the importance of the receptor for calcium absorption and bone mineralization. These pathways affect lead absorption from the gastrointestinal tract and may affect lead storage and/or release from bone (Onalaja and Claudio 2000). In the lead workers studied in this report, participants with the B allele were found to have significantly higher blood, tibia, and DMSA-chelatable lead than did those with the bb genotype (Schwartz et al. 2000a). The effect modification of VDR genotype on the association of tibia lead and uric acid in participants who are in the older half of the population may reflect toxicodynamic differences. However, conclusions must be tempered by the fact that this effect is dependent on, at most, 27 participants with tibia lead levels > 103 μg Pb/g bone mineral who reduce the β estimate of the relation between tibia lead and uric acid in participants with the bb genotype. Further, no associations between VDR genotype and renal outcomes were observed, nor was consistent effect modification by VDR on associations between lead measures and renal outcomes present in this population (Weaver et al. 2003b). Interestingly, the shape of the lowess lines for the relations between tibia lead and uric acid in workers with the common genotypes (ALAD1-1 and VDR bb; Figures 2 and 3) suggests differences at higher tibia lead levels that could reflect survivor bias. We plan to explore this further in longitudinal analyses.
eNOS catalyzes the transformation of l-arginine to the vasodilator, nitric oxide (NO). Animal models of kidney disease indicate that administration of l-arginine results in decreased renal damage that is thought to be mediated via increased NO (Klahr 2001). The functional significance of the Glu298Asp polymorphism remains uncertain. Some authors have reported decreased NO with the variant Asp allele (Noiri et al. 2002; Persu et al. 2002; Sofowora et al. 2001). However, others have reported no difference in various functional measures (Golser et al. 2003; Hoffmann et al. 2005; Li et al. 2004; Moon et al. 2002). If the Asp allele is ultimately shown to decrease NO, the variant may confer additional risk in lead exposure that also decreases NO (Vaziri 2002). In addition, this eNOS polymorphism may modify lead toxicokinetics. A recent study of data from the third evaluation of the lead workers in this report (n = 652) found that eNOS genotype modified the relation between patella lead and age; workers with the Asp allele had higher increases in patella lead with increasing age than did lead workers who were homozygous for the Glu allele (Theppeang et al. 2004). In the population that is the focus of this report, we found inconsistent associations of eNOS genotype with renal function (higher BUN and measured creatinine clearance in those with the Asp allele) (Weaver et al. 2003b). Effect modification was also noted; longer lead job duration was associated with higher serum creatinine and lower calculated creatinine clearance in participants with the Asp allele and with higher calculated creatinine clearance in participants with the Glu/Glu genotype (Weaver et al. 2003b). The lower uric acid observed in those with the Asp allele may indicate that, if this allele does confer increased renal risk in lead exposed populations, uric acid does not contribute further to that risk.
Our data suggest that all three genetic polymorphisms may affect uric acid in these lead workers. However, effect modification by ALAD genotype on associations between lead dose and uric acid was most consistent with the observed effect modification by this genotype on associations between lead dose and renal function. This work also suggests that effect modification by genotype is present only in the older half of the population. However, our ability to explore effect modification by genotype in different age groups is limited by the small number of participants with the variant alleles in these age-stratified models. As a result, even after removal of outliers, most data points in the variant groups are influential. Longitudinal data analysis will be useful, particularly in understanding the inverse associations seen with ALAD.
We thank Y.-B. Kim, B.-K. Jang, and G.-S. Lee for assistance with data collection in South Korea and J. Wetmur of the Mount Sinai School of Medicine for providing the nested primers for ALAD genotyping.
This research was supported by the following grants: ES07198 (B.S.S.), 2 ES07198 (V.M.W.), ES00002 (K.T.K.), and National Research Service Award F30-ES05922-02 (M.E.L.) from the National Institute of Environmental Health Sciences; KRF-2000-00545 (B.-K.L.) from the Korea Research Foundation; ATPM TS288-14/14 (E.K.S.) from the Centers for Disease Control and Prevention; and the Richard Ross Clinician Scientist Award (B.G.J.) from the Johns Hopkins University School of Medicine.
Figure 1 Plot of model assessing effect modification by ALAD genotype on the association of blood lead and uric acid in Korean lead workers whose ages are ≥ 40.6 years (Table 3, panel 2). Regression lines, generated using mean values of covariates in the model (age, sex, BMI, and alcohol use), are overlaid on crude data. The solid regression line represents the adjusted relation between blood lead and uric acid in older participants with the ALAD1-1 genotype (circles); the dashed regression line represents the adjusted relation between blood lead and uric acid in older participants with the ALAD1-2 genotype (stars).
Figure 2 Added variable plot of the model assessing effect modification by ALAD genotype on the association between tibia lead and uric acid in Korean lead workers ≥ 40.6 years of age (Table 3, panel 2). For each plot, the regression line (dashed line) and the lowess line (solid line) of the partial residual data points, adjusted for age, sex, BMI, and alcohol use, are overlaid. For ease of interpretation, axes have been scaled so that the plotted residuals are centered around mean uric acid and tibia lead, rather than around zero.
Figure 3 Added variable plot of the model assessing effect modification by VDR genotype on the association between tibia lead and uric acid in Korean lead workers ≥ 40.6 years of age (median) (Table 4, panel 2). For each plot, the regression line (dashed line) and the lowess line (solid line) of the partial residual data points, adjusted for age, sex, BMI, and alcohol use, are overlaid. For ease of interpretation, axes have been scaled so that the plotted residuals are centered around mean uric acid and tibia lead, rather than around zero.
Table 1 Selected demographic, exposure, and outcome variables by ALAD, eNOS, and VDR genotype in 798 Korean lead workers.a
ALAD eNOS VDR
Characteristic 1–1 1–2 Glu/Glu Asp/Glu or Asp/Asp bb Bb or BB
No. (%) 716 (90.1) 79 (9.9) 673 (84.9) 120 (15.1) 709 (88.8) 89 (11.2)
Sex n (%)
Male 569 (79.5) 62 (78.5) 537 (79.8) 93 (77.5) 572 (80.7) 62 (69.7)
Female 147 (20.5) 17 (21.5) 136 (20.2) 27 (22.5) 137 (19.3) 27 (30.3)
Alcohol use, n (%)
No previous alcohol 207 (29.0) 24 (30.4) 201 (30.0) 30 (25.0) 208 (29.4) 23 (25.8)
Current use 466 (65.2) 49 (62.0) 429 (63.8) 84 (70.0) 456 (64.4) 62 (69.7)
Past use 42 (5.9) 6 (7.6) 42 (6.3) 6 (5.0) 44 (6.2) 4 (4.5)
BMI (kg/m2) 23.1 ± 3.0 22.3 ± 2.6 23.1 ± 3.0 22.7 ± 2.8 22.9 ± 2.9 23.9 ± 3.4
Age (years) 40.5 ± 10.2 40.1 ± 9.7 40.3 ± 10.1 41.1 ± 10.1 40.2 ± 10.1 42.7 ± 10.3
Systolic blood pressure (mm Hg) 123.4 ± 16.5 122.3 ± 14.5 123.4 ± 16.5 122.7 ± 15.5 122.6 ± 15.6 129.1 ± 20.6
Blood lead (μg/dL) 31.7 ± 14.9 34.2 ± 15.9 32.0 ± 15.1 31.2 ± 14.6 31.6 ± 14.8 34.8 ± 16.1
Tibia lead (μg Pb/g bone mineral) 37.5 ± 40.6 31.4 ± 29.5 37.5 ± 41.6 35.8 ± 34.0 37.1 ± 41.2 38.1 ± 33.5
DMSA-chelatable lead (μg Pb/mg creatinine) 0.76 ± 0.82 0.84 ± 1.19 0.78 ± 0.90 0.72 ± 0.62 0.75 ± 0.87 0.93 ± 0.76
Uric acid (mg/dL) 4.8 ± 1.2 4.6 ± 1.1 4.9 ± 1.2 4.6 ± 1.1 4.8 ± 1.2 4.7 ± 1.1
Serum creatinine (mg/dL) 0.90 ± 0.15 0.86 ± 0.12 0.90 ± 0.16 0.89 ± 0.13 0.90 ± 0.16 0.89 ± 0.13
ALAD, eNOS, and VDR genotyping were completed on 795, 793, and 798 lead workers, respectively. Modified from Weaver et al. (2003b).
a Values are mean ± SD unless otherwise noted.
Table 2 Medians of selected demographic, exposure, and outcome variables in Korean lead workers by ALAD and VDR genotype in two groups dichotomized by median age (40.6 years).
ALAD VDR
Age < 40.6 years
Age ≥ 40.6 years
Age < 40.6 years
Age ≥ 40.6 years
Characteristic 1–1 1–2 1–1 1–2 bb Bb or BB bb Bb or BB
No.a 355 42 361 37 360 38 349 51
Age (years) 32.8 33.3 48.6 49.4 32.7 33.7 48.3 49.7
Lead job duration (years) 3.9 4.1 9.7 9.5 3.8 4.2 9.8 8.3
Blood lead (μg/dL) 28.8 31.9 30.4 34.3 29.5 29.4 30.4 35.5
Tibia lead (μg Pb/g bone mineral) 20.9 22.1 30.7 25.6 20.9 25.4 29.4 35.1
DMSA-chelatable lead (μg Pb/mg creatinine) 0.39 0.54 0.64 0.64 0.39 0.60 0.62 0.82
Uric acid (mg/dL) 5.1 4.9 4.4 4.1 5.1 5.2 4.4 4.2
BUN (mg/dL) 13.9 12.4 14.4 13.7 13.7 13.1 14.4 14.0
Serum creatinine (mg/dL) 0.93 0.89 0.88 0.81 0.92 0.91 0.87 0.86
Measured creatinine clearance (mL/min) 121.8 119.0 101.4 108.9 121.5 118.1 103.3 100.6
Calculated creatinine clearance (mL/min) 102.9 102.0 83.4 85.4 102.8 101.6 83.6 83.1
a Actual value, not median.
Table 3 Linear regression models of uric acid, evaluating effect modification by ALAD genotype on associations of blood and tibia lead in two groups of lead workers, dichotomized by median age (40.6 years).
Panel 1: age < 40.6 years
Panel 2: age ≥ 40.6 years
Panel 3: age ≥ 40.6 years
Variable β-Coefficient SE β p-Value β-Coefficient SE β p-Value β-Coefficient SE β p-Value
Blood lead models
Intercept 4.4070 0.1776 < 0.01 4.7507 0.1108 < 0.01 4.5906 0.1078 < 0.01
Age (years) −0.0348 0.0084 < 0.01 −0.0123 0.0098 0.21 −0.0188 0.0098 0.06
Systolic blood pressure (mm Hg) — — — — — — 0.0046 0.0027 0.09
Serum creatinine (mg/dL) — — — — — — 2.4921 0.3799 < 0.01
ALAD1-2 −0.2619 0.1591 0.10 −0.0292 0.1776 0.87 0.0870 0.1693 0.61
Blood lead (μg/dL)a −0.0043 0.0040 0.27 0.0127 0.0040 < 0.01 0.0089 0.0038 0.02
Blood lead × ALAD1-2b −0.0161 0.0134 0.23 −0.0212 0.0102 0.04 −0.0143 0.0098 0.14
Tibia lead models
Intercept 4.3928 0.1746 < 0.01 4.7280 0.1075 < 0.01 4.6027 0.1074 < 0.01
Age (years) −0.0335 0.0083 < 0.01 −0.0062 0.0096 0.52 −0.0157 0.0098 0.11
Systolic blood pressure (mm Hg) — — — — — — 0.0059 0.0027 0.03
Serum creatinine (mg/dL) — — — — — — 2.0767 0.4380 < 0.01
ALAD1-2 −0.3553 0.2330 0.13 −0.1993 0.2013 0.32 −0.1180 0.1955 0.55
Tibia lead (μg Pb/g bone mineral)a −0.0044 0.0020 0.03 0.0009 0.0013 0.48 0.0002 0.0013 0.89
Tibia lead × ALAD1-2b −0.0047 0.0103 0.65 −0.0151 0.0079 0.06 −0.0138 0.0077 0.07
Truncated tibia lead modelsc
Intercept 4.8218 0.1159 < 0.01 4.6499 0.1159 < 0.01
Age (years) −0.0074 0.0103 0.48 −0.0160 0.0103 0.12
Systolic blood pressure (mm Hg) — — — 0.0058 0.0028 0.04
Serum creatinine (mg/dL) — — — 2.3661 0.4620 < 0.01
ALAD1-2 −0.2763 0.2110 0.19 −0.1737 0.2033 0.39
Tibia lead (μg Pb/g bone mineral)a 0.0079 0.0027 < 0.01 0.0063 0.0026 0.02
Tibia lead × ALAD1-2b −0.0214 0.0084 0.01 −0.0193 0.0080 0.02
—, model shown was not adjusted for that covariate. Panels 1 and 2 display results in the younger and older groups, respectively. Panel 3 shows results in the older group after additional control for systolic blood pressure and serum creatinine. Models were also adjusted for sex, BMI, and alcohol use.
a Reference category of homozygotes for the common gene allele (ALAD1-1).
b p-Values for the cross-product terms reflect the statistical significance of the difference between the slopes of the regression line for the variant gene group and the regression line for the reference gene group; slopes in the variant gene group are obtained by adding the β-coefficient of the cross-product term to the β-coefficient for the reference category [i.e., the slope of the relation between blood lead and uric acid in those with ALAD1-2 genotype is −0.0085 in panel 2 (0.0127 + −0.0212)].
c Tibia lead levels > 89 μg Pb/g bone mineral were removed from models.
Table 4 Linear regression models of uric acid, evaluating effect modification by VDR genotype on associations of blood and tibia lead in two groups of lead workers, dichotomized by median age (40.6 years).
Panel 1: age < 40.6 years
Panel 2: age ≥ 40.6 years
Panel 3: age ≥ 40.6 years
β-Coefficient SE β p-Value β-Coefficient SE β p-Value β-Coefficient SE β p-Value
Blood lead models
Intercept 4.3906 0.1754 < 0.01 4.7529 0.1104 < 0.01 4.6087 0.1074 < 0.01
Age (years) −0.0341 0.0083 < 0.01 −0.0118 0.0099 0.23 −0.0190 0.0099 0.06
Systolic blood pressure (mm Hg) — — — — — — 0.0056 0.0027 0.04
Serum creatinine (mg/dL) — — — — — — 2.1009 0.3302 < 0.01
VDR Bb or BB 0.0073 0.1654 0.97 −0.0447 0.1565 0.78 −0.1201 0.1502 0.42
Blood lead (μg/dL)a −0.0056 0.0040 0.17 0.0111 0.0040 < 0.01 0.0087 0.0038 0.02
Blood lead × VDR Bb or BBb 0.0018 0.0109 0.87 −0.0041 0.0093 0.66 −0.0026 0.0089 0.77
Tibia lead models
Intercept 4.3660 0.1756 < 0.01 4.7474 0.1070 < 0.01 4.6267 0.1064 < 0.01
Age (years) −0.0338 0.0083 < 0.01 −0.0027 0.0097 0.78 −0.0122 0.0098 0.22
Systolic blood pressure (mm Hg) — — — — — — 0.0061 0.0027 0.02
Serum creatinine (mg/dL) — — — — — — 2.1603 0.4382 < 0.01
VDR Bb or BB −0.0599 0.2093 0.78 0.0842 0.1515 0.58 −0.0140 0.1484 0.93
Tibia lead (μg Pb/g bone mineral)a −0.0038 0.0021 0.07 −0.0008 0.0014 0.57 0.0015 0.0014 0.27
Tibia lead × VDR Bb or BBb −0.0033 0.0083 0.69 0.0138 0.0062 0.03 0.0142 0.0060 0.02
Truncated tibia lead modelsc
Intercept 4.7765 0.1110 < 0.01 4.6361 0.1108 < 0.01
Age (years) −0.0057 0.0101 0.57 −0.0142 0.0101 0.16
Systolic blood pressure (mm Hg) — — — 0.0061 0.0028 0.03
Serum creatinine (mg/dL) — — — 2.2604 0.4534 < 0.01
VDR Bb or BB 0.0502 0.1555 0.75 −0.0608 0.1522 0.69
Tibia lead (μg Pb/g bone mineral)a 0.0036 0.0023 0.12 0.0026 0.0022 0.26
Tibia lead × VDR Bb or BBb 0.0100 0.0064 0.12 0.0106 0.0062 0.09
—, model shown was not adjusted for that covariate. Panels 1 and 2 display results in the younger and older groups, respectively. Panel 3 shows results in the older group after additional control for systolic blood pressure and serum creatinine. Models were also adjusted for sex, BMI, and alcohol use.
a Reference category of homozygotes for the common gene allele (VDR bb).
b p-Values for the cross-product terms reflect the statistical significance of the difference between the slopes of the regression line for the variant gene group and the regression line for the reference gene group; slopes in the variant gene group are obtained by adding the β-coefficient of the cross-product term to the β-coefficient for the reference category [i.e., the slope of the relation between tibia lead and uric acid in those with VDR Bb or BB genotypes is 0.0130 in panel 2 (−0.0008 + 0.0138)].
c Tibia lead levels > 103 μg Pb/g bone mineral were removed from models.
Table 5 Linear regression model evaluating effect modification by combined ALAD and VDR genotypes on the association between tibia lead and uric acid, in lead workers ≥ 40.6 years of age (median).
Blood lead models β-Coefficient SE β p-Value
Intercept 4.6078 0.1086 < 0.01
Age (years) −0.0133 0.0099 0.18
Systolic blood pressure (mm Hg) 0.0058 0.0027 0.03
Serum creatinine (mg/dL) 2.1352 0.4420 < 0.01
ALAD1-2 and VDR bb −0.1086 0.2067 0.60
ALAD1-1 and VDR Bb or BB 0.0111 0.1533 0.94
Tibia lead (μg Pb/g bone mineral)a −0.0008 0.0014 0.59
Tibia lead × ALAD1-2 and VDR bbb −0.0122 0.0079 0.12
Tibia lead × ALAD1-1 and VDR Bb or BBb 0.0140 0.0061 0.02
Models were also adjusted for sex, BMI, and alcohol use.
a Reference category for homozygotes for both common gene alleles (ALAD1-1 and VDR bb).
b p-Values for the cross-product terms reflect the statistical significance of the difference between the slopes of the regression line for the variant gene groups and the regression line for the reference gene group. Slopes in the variant gene groups are obtained by adding the
β-coefficient of the cross-product term to the β-coefficient for the reference category [i.e., the slope of the relation between tibia lead and uric acid in those with both the ALAD1-1 and VDR Bb or BB genotypes is 0.0132 (−0.0008 + 0.0140)].
==== Refs
References
Bergdahl IA Grubb A Schütz A Desnick RJ Wetmur JG Sassa S 1997 Lead binding to δ-aminolevulinic acid dehydratase (ALAD) in human erythrocytes Pharmacol Toxicol 81 153 158 9353844
Cleveland WS 1979 Robust locally weighted regression and smoothing scatterplots J Am Stat Assoc 74 829 836
Cockcroft DW Gault MH 1976 Prediction of creatinine clearance from serum creatinine Nephron 16 31 41 1244564
Fernandez FJ 1975 Micromethod for lead determination in whole blood by atomic absorption, with use of the graphite furnace Clin Chem 21 558 561 1116290
Fleming DEB Chettle DR Wetmur JG Desnick RJ Robin J-P Boulay D 1998 Effect of the δ-aminolevulinate dehydratase polymorphism on the accumulation of lead in bone and blood in lead smelter workers Environ Res 77 49 61 9593628
Golser R Gorren AC Mayer B Schmidt K 2003 Functional characterization of Glu298Asp mutant human endothelial nitric oxide synthase purified from a yeast expression system Nitric Oxide 8 7 14 12586536
Hibi K Ishigami T Tamura K Mizushima S Nyui N Fujita T 1998 Endothelial nitric oxide synthase gene polymorphism and acute myocardial infarction Hypertension 32 521 526 9740620
Hoffmann IS Tavares-Mordwinkin R Castejon AM Alfieri AB Cubeddu LX 2005 Endothelial nitric oxide synthase polymorphism, nitric oxide production, salt sensitivity and cardiovascular risk factors in Hispanics J Hum Hypertens 3 233 240 15565175
Hu H Wu M-T Cheng Y Sparrow D Weiss S Kelsey K 2001 The δ-aminolevulinic acid dehydratase (ALAD) polymorphism and bone and blood lead levels in community-exposed men: the Normative Aging Study Environ Health Perspect 109 827 832 11564619
Johnson RJ Kang DH Feig D Kivlighn S Kanellis J Watanabe S 2003 Is there a pathogenetic role for uric acid in hypertension and cardiovascular and renal disease? Hypertension 41 1183 1190 12707287
Kamel F Umbach DM Lehman TA Park LP Munsat TL Shefner JM 2003 Amyotrophic lateral sclerosis, lead, and genetic susceptibility: polymorphisms in the delta-aminolevulinic acid dehydratase and vitamin D receptor genes Environ Health Perspect 111 1335 1339 12896855
Kelada SN Shelton E Kaufmann RB Khoury MJ 2001 δ-Aminolevulinic acid dehydratase genotype and lead toxicity: a HuGE review Am J Epidemiol 154 1 13 11427399
Kim R Aro A Rotnitzky A Amarasiriwardena C Hu H 1995 K X-ray fluorescence measurements of bone lead concentration: the analysis of low-level data Phys Med Biol 40 1475 1485 8532760
Kim R Rotnitzky A Sparrow D Weiss S Wager C Hu H 1996 A longitudinal study of low-level lead exposure and impairment of renal function JAMA 275 1177 1181 8609685
Klahr S 2001 The role of nitric oxide in hypertension and renal disease progression Nephrol Dial Transplant 16 60 62 11369823
Li R Lyn D Lapu-Bula R Oduwole A Igho-Pemu P Lankford B 2004 Relation of endothelial nitric oxide synthase gene to plasma nitric oxide level, endothelial function, and blood pressure in African Americans Am J Hypertens 17 560 567 15233974
Lin JL Ho HH Yu CC 1999 Chelation therapy for patients with elevated body lead burden and progressive renal insufficiency: a randomized, controlled trial Ann Intern Med 130 7 13 9890856
Lin JL Lin-Tan DT Hsu KH Yu CC 2003 Environmental lead exposure and progression of chronic renal diseases in patients without diabetes N Engl J Med 348 277 286 12540640
Lin JL Tan DT Ho HH Yu CC 2002 Environmental lead exposure and urate excretion in the general population Am J Med 113 563 568 12459402
Lin JL Yu CC Lin-Tan DT Ho HH 2001 Lead chelation therapy and urate excretion in patients with chronic renal diseases and gout Kidney Int 60 266 271 11422760
Moon J Yoon S Kim E Shin C Jo SA Jo I 2002 Lack of evidence for contribution of Glu298Asp (G894T) polymorphism of endothelial nitric oxide synthase gene to plasma nitric oxide levels Thromb Res 15 129 134 12431478
Muntner P He J Vupputuri S Coresh J Batuman V 2003 Blood lead and chronic kidney disease in the general United States population: results from NHANES III Kidney Int 63 1044 1050 12631086
Noiri E Satoh H Taguchi J Brodsky SV Nakao A Ogawa Y 2002 Association of eNOS Glu298Asp polymorphism with end-stage renal disease Hypertension 40 535 540 12364359
Onalaja AO Claudio L 2000 Genetic susceptibility to lead poisoning Environ Health Perspect 108 suppl 1 23 28 10698721
Parsons PJ Slavin W 1999 Electrothermal atomization atomic absorption spectrometry for the determination of lead in urine: results of an interlaboratory study Spectrochim Acta B 54 853 864
Payton M Hu H Sparrow D Weiss S 1994 Low-level lead exposure and renal function in the Normative Aging Study Am J Epidemiol 140 821 829 7977292
Persu A Stoenoiu MS Messiaen T Davila S Robino C El-Khattabi O 2002 Modifier effect of ENOS in autosomal dominant polycystic kidney disease Hum Mol Gen 11 229 241 11823442
Schwartz BS Lee BK Lee GS Stewart WF Lee SS Hwang KY 2001 Associations of blood lead, dimercaptosuccinic acid-chelatable lead, and tibia lead with neurobehavioral test scores in South Korean lead workers Am J Epidemiol 153 453 464 11226977
Schwartz BS Lee B-K Lee G-S Stewart WF Simon D Kelsey K 2000a Associations of blood lead, dimercaptosuccinic acid-chelatable lead, and tibia lead with polymorphisms in the vitamin D receptor and δ-aminolevulinic acid dehydratase genes Environ Health Perspect 108 949 954 11049814
Schwartz BS Lee B-K Stewart W Ahn K-D Springer K Kelsey K 1995 Associations of δ-aminolevulinic acid dehydratase genotype with plant, exposure duration, and blood lead and zinc protoporphyrin levels in Korean lead workers Am J Epidemiol 142 738 745 7572945
Schwartz BS Lee B-K Stewart W Sithisarankul P Strickland PT Ahn K-D 1997 δ-Aminolevulinic acid dehydratase genotype modifies 4-hour urinary lead excretion after oral administration of dimercaptosuccinic acid Occup Environ Med 54 241 246 9166129
Schwartz BS Stewart WF Kelsey KT Simon D Park S Links JM 2000b Associations of tibial lead levels with BsmI polymorphisms in the vitamin D receptor in former organolead manufacturing workers Environ Health Perspect 108 199 203 10706524
Shadick NA Kim R Weiss S Liang MH Sparrow D Hu H 2000 Effect of low level lead exposure on hyperuricemia and gout among middle aged and elderly men: the Normative Aging Study J Rheumatol 27 1708 1712 10914856
Smith CM Wang X Hu H Kelsey KT 1995 A polymorphism in the δ-aminolevulinic acid dehydratase gene may modify the pharmacokinetics and toxicity of lead Environ Health Perspect 103 248 253 7768225
Sofowora G Dishy V Xie H-G Imamura H Nishimi Y Morales CR 2001 In-vivo effects of Glu298Asp endothelial nitric oxide synthase polymorphism Pharmacogenetics 11 809 814 11740345
Staessen JA Lauwerys RR Buchet JP Bulpitt CJ Rondia D Vanrenterghem Y 1992 Impairment of renal function with increasing blood lead concentrations in the general population N Engl J Med 327 151 156 1608406
Theppeang K Schwartz BS Lee BK Lustberg ME Silbergeld EK Kelsey KT 2004 Associations of patella lead with polymorphisms in the vitamin D receptor, delta-aminolevulinic acid dehydratase and endothelial nitric oxide synthase genes J Occup Environ Med 46 528 537 15213514
Todd AC Chettle DR 1994 In vivo X-ray fluorescence of lead in bone: review and current issues Environ Health Perspect 102 172 177 8033846
Todd AC McNeill FE 1993. In vivo measurements of lead in bone using a 109Cd “spot” source. In: Human Body Composition (Ellis KJ, Eastman JD, eds). New York:Plenum Press, 299–302.
Topping MD Forster HW Dolman C Luczynska CM Bernard AM 1986 Measurement of urinary retinol-binding protein by enzyme-linked immunosorbent assay, and its application to detection of tubular proteinuria Clin Chem 32 1863 1866 3530532
Vaziri ND 2002 Pathogenesis of lead-induced hypertension: role of oxidative stress J Hypertens 20 S15 S20
Weaver VM Jaar BG Schwartz BS Todd AC Ahn K-D Lee S-S 2005 Associations among lead dose biomarkers, uric acid, and renal function in Korean lead workers Environ Health Perspect 113 36 42 10.1289/EHP.7317 15626645
Weaver VM Lee B-K Ahn K-D Lee G-S Todd AC Stewart WF 2003a Associations of lead biomarkers with renal function in Korean lead workers Occup Environ Med 60 551 562 12883015
Weaver VM Schwartz BS Ahn K-D Stewart WF Kelsey KT Todd AC 2003b Associations of renal function with polymorphisms in the δ-aminolevulinic acid dehydratase, vitamin D receptor, and nitric oxide synthase genes in Korean lead workers Environ Health Perspect 111 1613 1619 10.1289/ehp.6116 [Online 12 June 2003]. 14527840
Weisberg S 1985. Applied Linear Regression. New York:John Wiley & Sons.
Wetmur JG Lehnert G Desnick RJ 1991 The δ-aminolevulinate dehydratase polymorphism: higher blood lead levels in lead workers and environmentally exposed children with the 1-2 and 2-2 isozymes Environ Res 56 109 119 1769358
Wu M-T Kelsey K Schwartz J Sparrow D Weiss S Hu H 2003 A δ-aminolevulinic acid dehydratase (ALAD) polymorphism may modify the relationship of low-level lead exposure to uricemia and renal function: the Normative Aging Study Environ Health Perspect 111 335 340 12611663
Yu CC Lin JL Lin-Tan DT 2004 Environmental exposure to lead and progression of chronic renal diseases: a four-year prospective longitudinal study J Am Soc Nephrol 15 1016 1022 15034104
Ziemsen B Angerer J Lehnert G Benkmann HG Goedde HW 1986 Polymorphism of delta-aminolevulinic acid dehydratase in lead-exposed workers Int Arch Occup Environ Health 58 245 247 3770964
|
16263504
|
PMC1310911
|
CC0
|
2021-01-04 23:41:31
|
no
|
Environ Health Perspect. 2005 Nov 27; 113(11):1509-1515
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7927
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7632ehp0113-00151616263505ResearchWithin-Home versus Between-Home Variability of House Dust Endotoxin in a Birth Cohort Abraham Joseph H. 123Gold Diane R. 12Dockery Douglas W. 1Ryan Louise 1Park Ju-Hyeong 4Milton Donald K. 1251 Harvard School of Public Health, Boston, Massachusetts, USA2 Channing Laboratory, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, USA3 Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA4 Division of Respiratory Disease Studies, National Institute for Occupational Safety and Health, Morgantown, West Virginia, USA5 University of Massachusetts Lowell, Lowell, Massachusetts, USAAddress correspondence to D. Milton, Department of Work Environment, School of Health and Environment, University of Massachusetts–Lowell, 1 University Ave., Lowell, MA 01854 USA. Telephone: (978) 934-4850. Fax: (978) 452-5711. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 5 7 2005 113 11 1516 1521 1 10 2004 5 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Endotoxin exposure has been proposed as an environmental determinant of allergen responses in children. To better understand the implications of using a single measurement of house dust endotoxin to characterize exposure in the first year of life, we evaluated room-specific within-home and between-home variability in dust endotoxin obtained from 470 households in Boston, Massachusetts. Homes were sampled up to two times over 5–11 months. We analyzed 1,287 dust samples from the kitchen, family room, and baby’s bedroom for endotoxin. We fit a mixed-effects model to estimate mean levels and the variation of endotoxin between homes, between rooms, and between sampling times. Endotoxin ranged from 2 to 1,945 units per milligram of dust. Levels were highest during summer and lowest in the winter. Mean endotoxin levels varied significantly from room to room. Cross-sectionally, endotoxin was moderately correlated between family room and bedroom floor (r = 0.30), between family room and kitchen (r = 0.32), and between kitchen and bedroom (r = 0.42). Adjusting for season, the correlation of endotoxin levels within homes over time was 0.65 for both the bedroom and kitchen and 0.54 for the family room. The temporal within-home variance of endotoxin was lowest for bedroom floor samples and highest for kitchen samples. Between-home variance was lowest in the family room and highest for kitchen samples. Adjusting for season, within-home variation was less than between-home variation for all three rooms. These results suggest that room-to-room and home-to-home differences in endotoxin influence the total variability more than factors affecting endotoxin levels within a room over time.
dust endotoxinendotoxinintraclass correlationvariance components
==== Body
Endotoxin is biologically active lipopolysaccharide, a component of the outer cell membrane of gram-negative bacteria. Endotoxin has potent proinflammatory effects that have been well characterized at high doses among adults but are less well understood in home settings and among infants and children (Liu 2002; Reed and Milton 2001). Data suggesting that early-life infections and exposure to a farming lifestyle decrease the risk of childhood allergic disease have led to the hypothesis that early-life household endotoxin exposure may be protective against asthma and allergy (Braun-Fahrlander et al. 2002; Celedon et al. 2002; Martinez and Holt 1999; von Mutius et al. 2000). To test this hypothesis, epidemiologists have begun to measure endotoxin levels in house dust samples in population-based studies (Braun-Fahrlander et al. 2002; Gehring et al. 2001a, 2001b; Park et al. 2001; Rizzo et al. 1997; von Mutius et al. 2000).
In principle, the goal of endotoxin assessment for use in studies of endotoxin and chronic disease onset is to estimate subjects’ exposure, appropriately integrated over time and space. However, the appropriate sampling strategy is not well defined, and practical limitations often dictate actual sampling protocols. Commonly, endotoxin sampling would include collection of dust at only one point in time from one or at most a few rooms. Repeated endotoxin sampling within the time period of interest is seldom attempted. For optimal classification of chronic exposure, however, the relationship between point exposure measurements and temporal and spatial averages is of particular relevance. Nevertheless, few data are available on the relationship of endotoxin measured at a specific time and place in the home to endotoxin measured in other rooms within the home and at other times.
In the setting of linear regression, the within- to between-subject variance ratio is an estimate of the signal-to-noise ratio and has been used to assess the misclassification bias that occurs when using an imperfectly measured or surrogate exposure. A large within- to between-subject variance ratio indicates that a single exposure sample will provide a less precise estimate of chronic exposure. Park et al. (2000) applied a variance components model to estimate within- and between-home variances in endotoxin measurements in monthly samples over 12 months in a convenience sample of 20 Boston homes. Within-home variations in endotoxin levels were greater than between-home variations, except for endotoxin sampled in bed dust of adult participants. If generally true, the findings of Park et al. (2000) indicate that comparisons of exposure between homes based on a single assessment of endotoxin levels in sites other than the bed are not particularly useful—even for assessing average exposure over 1 year. Sites other than the bed may be relevant for endotoxin exposure, particularly for infants and toddlers in the United States, who often have relatively little dust in their plastic-covered bed mattresses and who spend a great deal of time in other rooms and crawling on the floor. Two recent reports of variability within and between homes in larger epidemiologic studies in Germany suggest that over 1 year, single measurements may be sufficient to distinguish exposure between homes, but that more measurements are needed to assess long-term average exposure (Heinrich et al. 2003; Topp et al. 2003).
In this report, we used a variance components analysis to reexamine the utility of endotoxin measurements in dust collected from different rooms in distinguishing average exposure during the first months to 1 year of life using a sample of 470 homes of children in Boston, Massachusetts. We assessed the correlation of endotoxin sampled in one room with levels in other rooms, and the correlation of a single endotoxin measurement with measurements of endotoxin in the same room 5–11 months later. We estimated room-specific within- to between-home variance ratios and explored the implications of these variance estimates for epidemiologic studies of dust endotoxin and health outcomes.
Materials and Methods
Cohort.
The Epidemiology of Home Allergens and Asthma study is a longitudinal birth cohort study of environmental predictors of development of allergy and asthma among children born to a parent or parents with a history of allergy and/or asthma (Gold et al. 1999). The study is investigating the relationship between indoor allergen exposure and the development of allergy and asthma in early childhood. Between September 1994 and June 1996, women who had given birth at the Brigham and Women’s Hospital in Boston were asked if she or the baby’s father had a history of allergy, hay fever, or asthma. Women answering affirmatively were asked to complete a screening questionnaire. Inclusion criteria included history of allergy, hay fever, and/or asthma in at least one parent, maternal age ≥18 years, English or Spanish speaking, residence in the greater Boston area, and no plans to move in the next year. Infants were excluded for premature birth (< 36 weeks), birth with major congenital or teratologic abnormalities, or admission to the neonatal intensive care unit. Of the 1,405 women who completed the screening questionnaire, 499 mothers (505 children) met inclusion and exclusion criteria and agreed to participate.
Dust collection and endotoxin assessment.
Within the first 3 months of the index child’s birth, an initial exposure assessment was conducted on the 499 homes of participants (Chew et al. 1998). An exposure assessment was conducted approximately 6 months later in a subset of homes. House dust was collected on a 19 × 90 mm cellulose extraction thimble using a modified Eureka Mighty-Mite vacuum cleaner (Eureka Co., Bloomington, IL). Separate dust samples were collected from the kitchen floor, family room, and the floor of the infant’s bedroom. In the kitchen, the floor under cabinets, around the refrigerator, and under the sink were vacuumed for 5 min. In the family room, the seat cushion, arms, and back of the chair most often occupied by the primary caregiver were vacuumed for 2.5 min. Two square meters of the floor surrounding this chair was also vacuumed for 2.5 min. In the bedroom, 2 m2 of floor surrounding the baby’s crib was vacuumed for 5 min. Collected dust was immediately placed in airtight bags. Initial sampling of dust to be used for endotoxin analysis was conducted between November 1994 and October 1996. The second dust sampling was conducted in a subset of homes between June 1995 and October 1996. Homes were selected for repeat sampling if the initial sampling was conducted during winter months. In the laboratory, dust samples were sifted using a 425-μm mesh sieve to remove large debris (e.g., breakfast cereal) and provide a more uniformly mixed, fine dust sample for partition into aliquots for several assays. The fine dust was then weighed and aliquoted for future analysis. Dust samples were stored desiccated at −20°C until extraction. Samples were analyzed for allergen and fungi and additionally analyzed for endotoxin only if there was > 200 mg of dust in the sieved sample. Endotoxin levels were not determined for 29 (6%) of the 499 participating homes. Up to six samples (three rooms with up to two samples) were possible per home. In the 470 homes with at least one endotoxin sample, we collected a mean of 2.7 and median of three samples per home.
The endotoxin activity of dust samples was determined with the kinetic Limulus assay with the resistant-parallel-line estimation (KLARE) method (Milton et al. 1992, 1997). Limulus amebocyte lysate was supplied by BioWhittaker (Walkersville, MD), and control standard endotoxin was obtained from Associates of Cape Cod (East Falmouth, MA). Endotoxin measurements were adjusted for lot-to-lot variation in Limulus amebocyte lysate sensitivity to house dust endotoxin [lot 6L016C used for assay of 42% of the samples was used as the standard lot, and nine additional lots each used for 2–11% of samples were adjusted using data from previously described lot-to-lot comparison assays (Milton et al. 1997)]. Control standard potency was determined for each combination of lysate and standard with reference to the reference standard endotoxins EC5 or EC6 [U.S. Pharmacopoeia, Inc., Rockville, MD; 1 ng EC5 and EC6 = 10 endotoxin units (EU)] available at the time the assays were performed, by simultaneous assay of the control with the reference or with a control traceable to assay with the reference. Results are reported as EU per milligram of dust sampled. The median coefficient of variation of the assay of house dust samples, 23%, was previously reported (Milton et al. 1997). None of the samples was below the limit of detection.
Statistical analysis.
We used SAS version 8.2 for all statistical analyses (SAS Institute Inc., Cary, NC) and assessed the normality of endotoxin distributions using the Shapiro-Wilk normality test. The dust endotoxin data were log-transformed to normalize the distribution of residuals in the mixed-effects models. We compared means in a mixed-effects model to account for correlation of samples within the same home. The correlation of endotoxin measured in dust sampled from different rooms in a home was assessed using Pearson correlation coefficients (with room-specific averages for rooms with replicate endotoxin observations) and using a mixed-effects model.
We fit a mixed-effects model of log endotoxin levels as a function of room and season adjusting for the correlation of repeated measurements within the same home. Inclusion of a random room effect within homes and declaration of a repeated-measures structure allowed us to characterize variation within and between homes and over time (Hamlett et al. 2003; Lyles et al. 1997; Rappaport 1991; Rappaport et al. 1995; Symanski et al. 1996). We obtained parameter estimates using restricted maximum likelihood (Diggle 2002). More precisely, the general mixed-effects model is described by the expression
where Yij is the jth repeated observation of log-transformed endotoxin for home i. The terms
are fixed covariates associated with the jth repeated measure on the ith home. The residual variance, ɛ ij, is modeled to include an appropriate correlation structure between endotoxin observations. For each room, the model estimates within-home and between-home variance, σ2w and σ2b, respectively. We then calculated within- to between-home variance ratios and intrahome correlation coefficients for endotoxin sampled from the floor of the subject’s bedroom, the family room, and the kitchen dust samples. The within- to between-home variance ratio characterizes the degree to which a single observation of endotoxin is representative of chronic exposure. The intrahome correlation coefficient characterizes the reproducibility (stability) of repeated endotoxin measurements over time (Rosner 1995). To estimate 95% confidence intervals (CIs), standard errors of within- to between-home variance ratios, and intrahome correlation coefficients were estimated using the delta method, using asymptotic variance and covariance estimates of the room-specific within-home and between-home variances estimated by the mixed effects model.
Definition of categorical variables.
The season of dust sampling was categorized as winter (November through March), spring (April and May), summer (June through August), or fall (September and October), to match Boston’s climate (Chew et al. 1999). Presence of a pet dog was categorized as no dog versus one or more dogs. The type of house occupied by the family was grouped into a) single-family or two-family dwellings or b) homes in apartment buildings with three or more units.
Results
A total of 1,287 endotoxin measurements were taken from 470 of the 499 participating homes. The initial home assessment included 320 bedroom, 401 family room, and 245 kitchen dust samples that were assayed for endotoxin activity. In the follow-up home assessment, endotoxin was measured in 102 bedroom, 147 family room, and 72 kitchen dust samples. In all, 82 (17%) homes had only one endotoxin measurement, 147 (31%) had two measurements, 127 (27%) had three measurements, 61 (13%) had four measurements, 32 (7%) had five measurements, and 21 (4%) had all six measurements. Repeated endotoxin activity measurements were available for 90 bedrooms, 125 family rooms, and 55 kitchens (180, 250, and 110 samples, respectively; Table 1).
Distribution of endotoxin levels in house dust.
Endotoxin ranged from 2 to 1,945 EU/mg of dust, with a geometric mean (GM) of 82 EU/mg of dust, a median of 81 EU/mg, and a geometric standard deviation of 2.1 (Table 2). Endotoxin was lowest in the bedroom floor samples, intermediate in family room samples, and highest in the kitchen floor samples, and similar in rooms with and without repeat samples. Endotoxin levels were highest during the summer and lowest in winter.
Adjusting for season, home assessment (initial or follow-up), presence of a dog in the home, housing type, and the correlation between observations made in the same home, we found that GM endotoxin varied significantly according to the room in which dust was sampled (p < 0.001) (Table 3). Endotoxin levels also varied by season in this model, with highest levels in the summer compared with fall (p = 0.002), winter (p < 0.001), and spring (p = 0.054). Endotoxin did not differ significantly between the initial and repeated samples (p = 0.494) in the multivariate model, after adjusting for season. As previously reported, GM endotoxin was higher (p < 0.001) in the 75 (16%) homes with dogs relative to homes without dogs, and higher in the 356 (76%) one-and two-family homes compared with homes in multiunit apartment buildings (p = 0.004).
Correlation of endotoxin levels in house dust.
Cross-sectional correlations between room-specific endotoxin levels were low to moderate (Table 4). The mixed-model correlations were similar to the Pearson correlation coefficients. Relative to the cross-sectional comparison of endotoxin from different rooms within homes, repeated room-specific endotoxin levels (5–11 months apart) were more highly correlated for bedroom floor (r = 0.65; 95% CI, 0.56–0.75), kitchen floor (r = 0.65; 95% CI, 0.53–0.76), and family room (r = 0.54; 95% CI, 0.44–0.63). Thus, the temporal correlation of endotoxin levels measured over the 5- to 11-month time was greater than the spatial correlation in those measurements (Table 4). Correlation coefficients estimated without adjustment for season were consistently lower, relative to those estimated while adjusting for fixed effects of season (Table 5).
We divided measurements of endotoxin sampled at two different times into quartiles of the time interval between samples. We did not find any consistent decrease in the correlations moving from shorter to longer time spans between sampling. In fact, for samples taken from the baby’s bedroom floor, the correlation between endotoxin sampled at two points in time increased with increasing time between sampling. However, the sample sizes for each time interval are small and the correlations are correspondingly less stable.
Endotoxin variance components: variation within and between homes.
We found that season-adjusted within-home variability was lowest for endotoxin in dust sampled from the baby’s bedroom floor, higher in family room samples, and highest for endotoxin in kitchen floor dust (Table 5). Season-adjusted between-home variability was lowest in dust sampled from the family room, higher for bedroom samples, and again highest for kitchen dust endotoxin (Table 5). In models adjusting for season, the within-home variance was less than the between-home variance for all three rooms, suggesting that factors affecting endotoxin levels within a home over time influence the total variability less than factors contributing to home-to-home differences in endotoxin. Within- to between-home variance ratios were lowest for bedroom floor and kitchen floor endotoxin and somewhat higher for family room samples, an observation explained by the comparatively small between-home variation in family room endotoxin (Table 5). These findings suggest that the determinants of endotoxin levels over time and from home-to-home are room specific.
We divided homes into those without dogs and those with at least one dog (Table 6), but found no consistent differences in variance components by presence of a dog. Unexpectedly, we found that between-home variation in endotoxin was uniformly higher in homes without dogs relative to homes with dogs. However, this did not result in correspondingly uniform changes in the correlation of endotoxin levels over time.
A similar subgroup analysis was conducted for housing type (Table 6). We did not find consistent differences in the variance components comparing single- or two-family houses with apartments in buildings with three or more units. Notably, for kitchen endotoxin in multiunit buildings, we observed a within-home variance 3.7 times that of the between-home variance, whereas the within-home variance was smaller than the between-home variance for the other rooms sampled.
Discussion
We assessed the distributions of, correlations between, and components of variation in endotoxin levels in dust sampled from the homes of subjects participating in an ongoing birth cohort study. In the homes studied, dust endotoxin levels were correlated over 5–11 months (range of intrahome correlations, 0.54–0.65), and slightly less correlated across rooms within homes (range of cross-sectional, room-to-room correlations, 0.30–0.42). Within-home to between-home variance ratios were below one for all samples: 0.53 and 0.54 for bedroom and kitchen dust endotoxin, respectively, and 0.85 for family room endotoxin. Thus, single endotoxin measurements are a reasonable proxy for average exposure during the first few months to 1 year of life and capable of distinguishing among children in metropolitan Boston with regard to endotoxin exposure in early life.
Park et al. (2000) analyzed the variance components of endotoxin in dust collected in a one-year longitudinal study of a convenience sample of 20 homes of students, faculty, and university staff in Boston. This report builds on that work by Park and colleagues by characterizing variability in dust endotoxin using a much larger and more representative sample of homes in the metropolitan Boston area with at most two measurements per room in different seasons. Gereda et al. (2000) made repeated measurements of house dust endotoxin on 11 homes, 6 months apart, finding no significant differences in endotoxin of dust samples from the two assessments. They did not report the replicate data in their limited number of homes. Heinrich et al. (2003) reported repeated measurements of endotoxin in homes over a 1-year period. They found that endotoxin measurements expressed as units per area were more consistent and better able to distinguish between homes than were measurements expressed as units per gram. However, both methods gave higher between- than within-home variances and suggested that single measurements could be used as proxies for average exposure during the year of sampling. In our study, the protocol for collection of kitchen and family room dust samples did not use standard areas because they targeted furnishings or certain architectural features of the rooms, and therefore precluded estimation and analysis of units per area.
Endotoxin levels in a given room were only moderately correlated with those from other rooms in the same home, suggesting that an endotoxin sample from a single room may not indicate endotoxin in other rooms or the house as a whole. In an epidemiologic study, dust sampling in several rooms, plus a determination of the relative time spent by the subject in each room may provide a better estimate of household endotoxin exposure at a point in time. Room-to-room correlations between endotoxin levels within homes were assessed using two approaches: Pearson correlations for simple cross-sectional analysis, using the mean level for each room if repeated measures were available; and a mixed-effects model (Heederik et al. 1991; Rosner 1995). The cross-sectional correlation coefficients are estimates of the degree to which room-specific endotoxin is indicative of endotoxin levels in other rooms within the home. The mixed-effects model estimates correlations using all of the data accounting for the correlation in repeated endotoxin levels and thereby decreasing the uncertainty of these estimates. In this study, the conventional Pearson correlation coefficients were for the most part qualitatively similar to the correlation coefficients estimated from the mixed model.
In our primary analysis, the largest between-home variance component was observed for kitchen dust endotoxin, followed by that for bedroom floor samples. The between-home variance of family room endotoxin was comparatively lower. However, in multiunit buildings, the between-home variation was much lower for kitchen floor samples, compared with the between-home variance for bedroom and family room endotoxin.
Within-home variances were highest for kitchen floor endotoxin. Room-specific differences in the within-home variance component were most dramatic for endotoxin sampled from kitchens in multiunit buildings. The large within-home variation in endotoxin observed for kitchen dust samples may be due to water, food products, and vegetable matter being present to varying degrees in homes and over time.
The reproducibility of repeated endotoxin measurements in dust from the kitchen and bedroom floors, as indicated by intrahome correlation coefficients of 0.65 for both, was greater than reported by Park et al. (2000) The intrahome correlation for endotoxin sampled from the family room (r = 0.54) was lower than observed for the other rooms in this analysis but higher than those reported by Park et al. (2000) for kitchen and bedroom floor dust. Park and colleagues did not sample family room dust. Possibly, rooms with more usage and foot traffic have more variability in endotoxin levels over time. The moderate temporal stability of endotoxin levels observed in this assessment suggests that a single exposure assessment provides a reasonable, although not an optimal indication of endotoxin levels over time.
Within-home variability in endotoxin was less than between-home variability for all three rooms, suggesting that factors affecting endotoxin levels within a home over time influence the total variability less than factors contributing to home-to-home differences in endotoxin. We observed smaller within-home variances, larger between-home variances, and correspondingly smaller within- to between-home variance ratios than those observed by Park et al. (2000). In contrast to Park et al. (2000), we sampled dust in a far larger number of homes that were likely more representative of metropolitan Boston area households (e.g., Park and colleagues did not include homes with dogs) and thus also expected to have a larger between-home variance. The present study was limited, however, by having fewer repeated measurements and insufficient repeated bed dust and air samples for analysis. Another limitation was that we could not compute endotoxin loading per unit area because dust was collected from family room furnishings as well as floors and around the perimeters of the kitchen.
The ratio of within- to between-home variance may be used to better interpret reports of associations between endotoxin levels and disease outcomes and inform endotoxin exposure assessment strategies for future studies. In an optimal study of chronic exposure to house dust endotoxin, all variability would be observed between homes and endotoxin levels would not vary over time in the same sampling area. In that case, the within-home to between-home variance ratio would be zero, and provided there are no other sources of bias, a single endotoxin measurement would provide an unbiased estimate of the effects of chronic exposure on an outcome. In practice, a single measurement of endotoxin taken in one room of a home is likely to be an imperfect surrogate for chronic endotoxin exposure in that home. If the within-home variance is nonzero, the observed room-specific endotoxin level will deviate from the true room-specific mean level. If we assume that the observed measure is an imperfect measure of the true mean endotoxin for a room but that the error in measurement is uncorrelated with the true endotoxin level, the association between single samples of endotoxin in homes and health effects is likely to be attenuated relative to the true effect of chronic exposure (Heederik et al. 1991; Zeger et al. 2000).
The relationship between the effect estimate obtained using an observed, misclassified exposure and the true effect estimate has been derived in the univariate setting with one exposure variable and no covariables. The attenuation of the effect estimate is given by
where β* is the observed linear effect estimate, β is the true effect estimate, σ2w is the within-home variance, σ2b is the between-home variance, and n is the number of repeated samples per sampling unit. The magnitude of attenuation increases as the within-home to between-home variance ratio increases. Because the magnitude of misclassification depends not on the value of either within- or between-home variance but on the ratio of the two, there are several theoretical approaches to reduce or avoid the bias of the exposure–outcome relationship. Namely, one could maximize variability of endotoxin across subjects, thereby increasing σ2b, or sample endotoxin repeatedly to better estimate true room-specific mean levels within homes.
Applying this theory to our findings, epidemiologic studies using a single house-dust endotoxin observation as an index of chronic exposure may underestimate the effect of endotoxin on an outcome, given that such an effect exists and no other bias or misclassification is present. In the present study, if dust from the three rooms were equally good proxies for actual exposure, using family room endotoxin as the exposure measure, which has the highest within-home to between-home variance ratio, would result in the largest degree of attenuation of effects, relative to using endotoxin from the other rooms.
Variance components provide a statistical basis for sampling but should not be the only determinant of a home sampling strategy. To properly assess exposure, one must consider other determinants of exposure, including where the subjects spend their time. We found that at the time of the first dust sampling, 64% of the children were reported to spend most of their time in the family room, whereas 12 and 6% reported spending most of their time in the kitchen and bedroom, respectively. Eighty-five percent of participants classified the child’s time spent in the family room as more than in other rooms. Thus, use of family room dust samples may provide a better indicator of exposure compared with using only bedroom or only kitchen dust samples.
The true window in which endotoxin exposure may act to modify allergen sensitivity is not known. There is experimental evidence that endotoxin effects are both time and dose dependent (Eisenbarth et al. 2002; Tulic et al. 2000; Wan et al. 2000). It is possible that exposures in a specific perinatal period may be protective of allergic disease development, whereas similar exposures occurring at less relevant periods or at different doses may be innocuous or even promote allergic disease. Thus, the timing of endotoxin exposure sampling with respect to the development of the child may be more important in defining risk than the season in which the sample was collected.
The initial motivation for conducting the repeated-measures dust sampling was to assess the effects of season on indoor allergens focusing on homes initially sampled during winter months. Thus, the second home assessment was conducted in a complementary season relative to the first home visit. The first home assessments were conducted during all seasons, although the samples from homes selected for repeated sampling were collected during the winter and spring months. In contrast, the repeated measurements all were taken from dust collected during the summer and fall months. Because of seasonal variability in endotoxin levels, GM endotoxin was higher for the second home sampling compared with the first assessment. The variability and correlation in endotoxin over time were assessed using both models that adjusted for season of sampling and those that did not. Adjusting for a fixed-effect of season decreased the variability observed within homes and increases variation between homes. As a result, correlations increased and the ratio of within- to between-home variance decreased after controlling for season of sampling.
The endotoxin levels we observed (maximum < 2,000 EU/mg dust) are comparable with those seen in studies of house dust endotoxin in other urban settings (Gereda et al. 2000; Park et al. 2000; Rizzo et al. 1997) but lower than in studies including rural or farm homes (Braun-Fahrlander et al. 2002; von Mutius et al. 2000). We observed lowest endotoxin levels in dust sampled during the winter months and highest levels in dust sampled during the summer. Park et al. (2000) found similar seasonal patterns for outdoor samples of endotoxin but suggested that endotoxin samples from indoor house dust may not follow this pattern consistently. Our finding of seasonal variability is consistent with the findings of Rizzo et al. (1997) in a case–control study of endotoxin and asthma in children 6–16 years of age living in São Paulo, Brazil, who reported endotoxin levels to be generally lower in the winter months and higher in summer months. Study-to-study comparisons of endotoxin are often limited by interlab differences in endotoxin assay protocols. However, the samples described here were assayed by the same laboratory using the same protocols and Limulus lysates as the data reported by Park et al. (2000). Because this sample included only urban and suburban homes, our results may generalize only to other metropolitan regions of developed countries in temperate climates. Although we observed variation in endotoxin levels within and between homes in our study, the degree of heterogeneity is likely small relative to industrial or agricultural settings, where sources of endotoxin exist in particular locations and not in others. Similarly, one might expect a larger degree of between-home variability, and perhaps also different patterns of variability in endotoxin if we included both nonfarm and farm households, as has been done in Europe (Braun-Fahrlander et al. 2002).
Our sampling design was not balanced with respect to season, but this poses no problem for estimation of the variance components using the mixed-effects model. The precision of the temporal variance component estimates (within-home variance) was limited by the fact that we sampled endotoxin at most two times from a given room. On the other hand, this sample included a large number of homes compared with previous studies.
All else being equal, bedroom and kitchen floor samples provided slightly more stable estimates of endotoxin over time. Within-home variation in endotoxin levels was smaller than between-home variation for the three sampling locations. The correlation over time and the ratio of within-home to between-home variance observed in this study support the use of a single endotoxin measurement as a marker for chronic endotoxin exposure in association studies.
We thank D. Sredl for assistance with data management, the study research assistants, and especially the participants of the Home Allergens and Asthma Study.
This study was supported by the National Institute of Environmental Health Sciences (NIEHS) grant R01 ES-07036 and NIEHS Center grant 2P30ES00002. J.H.A. received training support from National Institutes of Health/National Heart, Lung, and Blood Institute grant HL07427-23.
Table 1 Summary of sample sizes for endotoxin in house dust samples.a
No. of endotoxin samples collected
Assessment Total Bedroom floor Family room Kitchen floor
Initial 966 320 401 245
Follow-up 321 102 147 72
Combined 1,287 422 548 317
No. of repeated samples 540 180 250 110
a Dust sampling was conducted according to a standardized protocol. Not all homes had sufficient dust collected to assay for endotoxin. In the home with endotoxin observations, the total amount of dust was not associated with endotoxin levels (data not shown).
Table 2 Summary of the distribution of house dust endotoxin levels (EU/mg dust) for selected covariates.
Percentile
No.a GMb GSD Minimum 25th 50th 75th Maximum
All samples 1,287 82 2.1 2 52 81 127 1,945
Bedroom floor
Total 422 67 2.0 2 44 67 103 761
Single sample 242 66 2.1 2 43 66 102 761
Repeated samples 180 70 1.8 16 48 70 103 629
Family room
Total 548 83 2.0 2 53 83 123 1,945
Single sample 298 82 2.1 2 53 83 129 713
Repeated samples 250 83 2.0 14 57 83 119 1,945
Kitchen floor
Total 317 105 2.2 4 62 110 173 1,201
Single samples 207 101 2.3 4 62 107 166 1,201
Repeated samples 110 112 2.1 12 63 112 191 852
Home assessment
Initial 966 79 2.1 2 49 77 126 1,201
Follow-up 321 92 2.0 4 59 88 131 1,945
Season
Summer 458 97 1.9 4 65 97 138 761
Fall 246 83 2.1 9 54 80 120 1,945
Winter 428 69 2.2 2 42 65 110 1,201
Spring 155 79 2.1 9 48 77 135 580
Dogsc
No 1,058 78 2.1 2 49 76 119 1,945
Yes 229 106 2.1 17 68 101 166 956
Housing typed
Single- or two-family 1,001 86 2.0 9 56 86 131 1,249
Multiunit building 286 68 2.4 2 41 66 110 1,945
GSD, geometric standard deviation.
a No. of endotoxin samples collected.
b GMs are unadjusted.
c Presence of a dog in the home was categorized as none versus one or more.
d Housing type was dichotomized as being a one- or two-family home versus part of a multiunit building.
Table 3 Fixed-effects results from mixed-effects model.a
Fixed-effect variable Percent change from reference level 95% CI (%) p-Valueb
Sample
Bedroom floor 82 76–89 < 0.001
Family room —
Kitchen floor 124 112–137 < 0.001
Home assessment
Initial —
Follow-up 96 86–107 0.494
Season
Summer —
Fall 84 75–93 0.002
Winter 69 61–77 < 0.001
Spring 86 73–100 0.054
Dog in home
No —
Yes 131 116–147 < 0.001
House type
Single- or two-family home —
Multiunit apartment 83 73–94 0.004
a Includes fixed effects for room being sampled, home assessment, season, pet dog, and house type. The model provides estimates of the relative change in mean endotoxin for each covariable, independent of the other fixed-effects variables in the model, accounting for the correlation between endotoxin levels measured in the same home. The reference group is endotoxin sampled from the family room during the summer, in single/two-family homes with no dogs. GM endotoxin in the reference group was 98.3 EU/mg.
b Wald test.
Table 4 Correlation of endotoxin levels between rooms (off-diagonal) and within rooms over time (diagonal).a
Bedroom floor Family room Kitchen floor
Bedroom floor 0.65 0.30 0.42
— n = 299 n = 185
Family room 0.33 0.54 0.32
n = 299 — n = 233
Kitchen floor 0.41 0.27 0.65
n = 185 n = 233 —
a Room-specific intrahome correlation coefficients derived from the within- and between-home variance components are presented on the diagonal. Pearson correlation coefficients are below the diagonal and correlation coefficients derived from the variance components are above the diagonal. The mixed-effects model included indicators for fixed effects of season. If replicate samples were available, the average was used to calculate Pearson correlation coefficients.
Table 5 Within-home variance (σ2w) and between-home variance (σ2b), the σ2w:σ2b ratio, and correlations within rooms over time for endotoxin in dust sampled from the bedroom floor, family room, and kitchen floor.a
Sample, model σ2w σ2b σ2w:σ 2b(95% CI) Correlation over time (95% CI)
Bedroom floor
a 0.038 0.056 0.69 (0.30–1.08) 0.59 (0.48–0.71)
b 0.031 0.058 0.53 (0.26–0.80) 0.65 (0.56–0.75)
Family room
a 0.050 0.043 1.15 (0.53–1.77) 0.46 (0.36–0.57)
b 0.042 0.049 0.85 (0.45–1.26) 0.54 (0.44–0.63)
Kitchen floor
a 0.049 0.079 0.62 (0.2–0.99) 0.62 (0.50–0.73)
b 0.045 0.083 0.54 (0.23–0.86) 0.65 (0.53–0.76)
a Variance components for model a were estimated using a mixed-effects model with a random effect for the room being sampled within homes and a fixed room effect. Model b was additionally adjusted for indicators of season.
Table 6 Within-home variance (σ2w) and between-home (σ2b) variance, the σ2w:σ2b ratio, and correlations within rooms over time for endotoxin in dust sampled from the bedroom floor, family room, and kitchen floor, by pet dog(s) and housing type.a
Sample homes Observations (n) σ2w σ2b σ2w:σ 2b (95% CI) Correlation over time (95% CI)
Bedroom floor
No dog 353 0.029 0.058 0.50 (0.22–0.78) 0.67 (0.56–0.77)
Dog(s) 69 0.042 0.051 0.82 (−0.52–2.16) 0.55 (0.20–0.89)
One/two-family 325 0.033 0.043 0.76 (0.28–1.24) 0.57 (0.44–0.70)
Multiunit 97 0.024 0.083 0.29 (−0.02–0.61) 0.77 (0.60–0.95)
Family room
No dog 459 0.045 0.049 0.91 (0.43–1.40) 0.52 (0.42–0.63)
Dog(s) 89 0.021 0.045 0.47 (0.03–0.92) 0.68 (0.51–0.85)
One/two-family 421 0.036 0.038 0.94 (0.40–1.48) 0.52 (0.40–0.63)
Multiunit 127 0.060 0.080 0.74 (−0.02–1.51) 0.57 (0.36–0.79)
Kitchen floor
No dog 246 0.043 0.082 0.53 (0.13–0.92) 0.65 (0.51–0.80)
Dog(s) 71 0.047 0.068 0.68 (−0.09–1.45) 0.59 (0.37–0.82)
One/two-family 253 0.043 0.079 0.55 (0.18–0.91) 0.65 (0.51–0.78)
Multiunit 64 0.111 0.030 3.72 (−8.10–15.54) 0.21 (−0.16–0.58)
a Variance components were estimated using a mixed-effects model with a random effect for sampling room within homes, a fixed room effect, and a fixed season effect.
==== Refs
References
Braun-Fahrlander C Riedler J Herz U Eder W Waser M Grize L 2002 Environmental exposure to endotoxin and its relation to asthma in school-age children N Engl J Med 347 12 869 877 12239255
Celedon JC Litonjua AA Ryan L Weiss ST Gold DR 2002 Day care attendance, respiratory tract illnesses, wheezing, asthma, and total serum IgE level in early childhood Arch Pediatr Adolesc Med 156 3 241 245 11876667
Chew GL Burge HA Dockery DW Muilenberg ML Weiss ST Gold DR 1998 Limitations of a home characteristics questionnaire as a predictor of indoor allergen levels Am J Respir Crit Care Med 157 5 pt 1 1536 1541 9603135
Chew GL Higgins KM Gold DR Muilenberg ML Burge HA 1999 Monthly measurements of indoor allergens and the influence of housing type in a northeastern US city Allergy 54 10 1058 1066 10536884
Diggle P 2002. Analysis of Longitudinal Data. New York:Oxford University Press.
Eisenbarth SC Piggott DA Huleatt JW Visintin I Herrick CA Bottomly K 2002 Lipopolysaccharide-enhanced, toll-like receptor 4-dependent T helper cell type 2 responses to inhaled antigen J Exp Med 196 12 1645 1651 12486107
Gehring U Bolte G Borte M Bischof W Fahlbusch B Wichmann HE 2001a Exposure to endotoxin decreases the risk of atopic eczema in infancy: a cohort study J Allergy Clin Immunol 108 5 847 854 11692114
Gehring U Heinrich J Jacob B Richter K Fahlbusch B Schlenvoigt G 2001b Respiratory symptoms in relation to indoor exposure to mite and cat allergens and endotoxins. Indoor Factors and Genetics in Asthma (INGA) Study Group Eur Respir J 18 3 555 563 11589355
Gereda JE Leung DYM Thatayatikom A Streib JE Price MR Klinnert MD 2000 Relation between house-dust endotoxin exposure, type 1 T-cell development, and allergen sensitisation in infants at high risk of asthma Lancet 355 9216 1680 1683 10905243
Gold DR Burge HA Carey V Milton DK Platts-Mills T Weiss S 1999 Predictors of repeated wheeze in the first year of life: the relative roles of cockroach, birth weight, acute lower respiratory illness, and maternal smoking Am J Resp Crit Care Med 160 1 227 236 10390405
Hamlett A Ryan L Serrano-Trespalacios P Wolfinger R 2003 Mixed models for assessing correlation in the presence of replication J Air Waste Manag Assoc 53 4 442 450 12708508
Heederik D Boleij JSM Kromhout H Smid T 1991 Use and analysis of exposure monitoring data in occupational epidemiological study in the Dutch animal food industry Appl Occup Environ Hyg 6 6 458 464
Heinrich J Holscher B Douwes J Richter K Koch A Bischof W 2003 Reproducibility of allergen, endotoxin and fungi measurements in the indoor environment J Expo Anal Environ Epidemiol 13 2 152 160 12679795
Liu AH 2002 Endotoxin exposure in allergy and asthma: reconciling a paradox J Allergy Clin Immunol 109 3 379 392 11897980
Lyles RH Kupper LL Rappaport SM 1997 A lognormal distribution-based exposure assessment method for unbalanced data Ann Occup Hyg 41 1 63 76 9072951
Martinez FD Holt PG 1999 Role of microbial burden in aetiology of allergy and asthma Lancet 354 suppl 2 SII12 SII15 10507253
Milton DK Feldman HA Neuberg DS Bruckner RJ Greaves IA 1992 Environmental endotoxin measurement: the kinetic Limulus assay with resistant-parallel-line estimation Environ Res 57 212 230 1568442
Milton DK Johnson DK Park J-H 1997 Environmental endotoxin measurement: interference and sources of variation in the Limulus assay of house dust Am Ind Hyg Asso J 58 861 867
Park J-H Gold DR Spiegelman DL Burge HA Milton DK 2001 House dust endotoxin and repeated wheeze in the first year of life Am J Respir Crit Care Med 163 322 328 11179100
Park J-H Spiegelman DL Burge HA Gold DR Chew GL Milton DK 2000 Longitudinal study of dust and airborne endotoxin in the home Environ Health Perspect 108 1023 1028 11102291
Rappaport SM 1991 Assessment of long-term exposures to toxic substances in air Ann Occup Hyg 35 1 61 121 2035954
Rappaport SM Lyles RH Kupper LL 1995 An exposure-assessments strategy accounting for within- and between-worker sources of variability Ann Occup Hyg 39 4 469 495 7661513
Reed CE Milton DK 2001 Endotoxin stimulated innate immunity: a contributing factor for asthma J Allergy Clin Immunol 108 2 157 166 11496229
Rizzo MC Naspitz CK Fernandez-Caldas E Lockey RF Mimica I Sole D 1997 Endotoxin exposure and symptoms in asthmatic children Pediatr Allergy Immunol 8 3 121 126 9532251
Rosner B 1995. Fundamentals of Biostatistics. Belmont, CA:Duxbury Press.
Symanski E Kupper LL Kromhout H Rappaport SM 1996 An investigation of systematic changes in occupational exposure Am Ind Hyg Assoc J 57 8 724 735 8765201
Topp R Wimmer K Fahlbusch B Bischof W Richter K Wichmann HE 2003 Repeated measurements of allergens and endotoxin in settled house dust over a time period of 6 years Clin Exp Allergy 33 12 1659 1666 14656352
Tulic MK Wale JL Holt PG Sly PD 2000 Modification of the inflammatory response to allergen challenge after exposure to bacterial lipopolysaccharide Am J Respir Cell Mol Biol 22 5 604 612 10783133
von Mutius E Braun-Fahrlander C Schierl R Riedler J Ehlermann S Maisch S 2000 Exposure to endotoxin or other bacterial components might protect against the development of atopy Clin Exp Allergy 30 9 1230 1234 10971468
Wan GH Li CS Lin RH 2000 Airborne endotoxin exposure and the development of airway antigen-specific allergic responses Clin Exp Allergy 30 3 426 432 10691903
Zeger SL Thomas D Dominici F Samet JM Schwartz J Dockery D 2000 Exposure measurement error in time-series studies of air pollution: concepts and consequences Environ Health Perspect 108 419 426 10811568
|
16263505
|
PMC1310912
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 5; 113(11):1516-1521
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7632
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7983ehp0113-00152216263506ResearchSome Environmental Contaminants Influence Motor and Feeding Behaviors in the Ornate Wrasse (Thalassoma pavo) via Distinct Cerebral Histamine Receptor Subtypes Giusi Giuseppina 1Facciolo Rosa Maria 1Alò Raffaella 1Carelli Antonio 1Madeo Maria 1Brandmayr Pietro 2Canonaco Marcello 11 Comparative Neuroanatomy Laboratory, and2 Zoocenoses Laboratory, Ecology Department, University of Calabria, Cosenza, ItalyAddress correspondence to M. Canonaco, Comparative Neuroanatomy Laboratory, Ecology Department, University of Calabria, 87030 Rende (CS), Italy. Telephone: 0039-984-492974. Fax: 0039-984-492986. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 14 7 2005 113 11 1522 1529 1 2 2005 14 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Common environmental contaminants such as heavy metals and pesticides pose serious risks to behavioral and neuroendocrine functions of many aquatic organisms. In the present study, we show that the heavy metal cadmium and the pesticide endosulfan produce such effects through an interaction of specific cerebral histamine receptor subtypes in the teleost ornate wrasse (Thalassoma pavo). Treatment of this teleost with toxic cadmium levels for 1 week was sufficient to induce abnormal swimming movements, whereas reduced feeding behaviors were provoked predominantly by elevated endosulfan concentrations. In the brain, these environmental contaminants caused neuronal degeneration in cerebral targets such as the mesencephalon and hypothalamus, damage that appeared to correlate with altered binding levels of the three major histamine receptors (subtypes 1, 2, and 3). Although cadmium accounted for reduced binding activity of all three subtypes in most brain regions, it was subtype 2 that seemed to be its main target, as shown by a very great (p < 0.001) down-regulation in mesencephalic areas such as the stratum griseum central layer. Conversely, endosulfan provided very great and great (p < 0.01) up-regulating effects of subtype 3 and 1 levels, respectively, in preoptic-hypothalamic areas such as the medial part of the lateral tuberal nucleus, and in the suprachiasmatic nucleus. These results suggest that the neurotoxicant-dependent abnormal motor and feeding behaviors may well be tightly linked to binding activities of distinct histamine subtypes in localized brain regions of the Thalassoma pavo.
cadmiumdiencephalonendosulfanhistamine receptor subtypemesencephalonteleost fish
==== Body
A number of neurotoxic environmental contaminants, recognized as endocrine disruptors, have aroused much interest in the field of neuroendocrinology (Pillai et al. 2003). In particular, polycyclic aromatic hydrocarbons, polychlorinated biphenyls, and many heavy metals such as arsenic, cobalt, and mercury in the organic form appear to impair growth, development, and sociosexual behavior in vertebrates (Beauvais et al. 2001; Bisson and Hontela 2002). Cadmium is one of the heavy metals that pose an increasing health threat to ecologic communities and humans (Harvey et al. 1999). Because of widespread industrial applications such as the use of alloys for metal coatings and nickel-cadmium batteries as well as the burning of fossil fuels, urban traffic, and waste incineration, this pollutant is emitted into the atmosphere (Liao and Freedman 1998). It is taken up readily by humans and other mammals not only via inhalation but also via the food web (Waalkes et al. 1992). By binding to cysteine residues or generating reactive oxygen species (Risso-de Faverney et al. 2001), Cd has been shown to influence genomic and postgenomic processes in liver, kidney, lung, and brain (Minami et al. 2001). Similar alterations appear to be linked with neuronal dysfunctions in the hypothalamic–pituitary–testicular axis and inadequate neurosecretory activities of pituitary cells (Lafuente et al. 2001). Interestingly, these same degenerative processes in the preferential brain region, that is, the olfactory bulb, appear to corroborate evidence that foraging and aggression controlled by this region constitute vital behaviors in fish (Griffiths and Armstrong 2000; Scherer et al. 1997).
The persistence and accumulation of the insecticide endosulfan (6,7,8,9,10,10-hexachloro-1,5,5a,6,9,9a-hexahydro-6,9-methano-2,4,3-benzodioxathiepin-3-oxide) in human tissues are of concern (Martinez et al. 1997). Its resistance to biologic degradation and its low water solubility not only favor binding to soil particles and persistence within surface waters but also prompt bioconcentrations resulting in up to 600 times ambient water concentration in some species (Miller and Llados 1999). The purpose of this insecticide was to protect economically important crops such as tobacco and cotton; little attention was paid to neurologic risks to humans or other animals in terrestrial and aquatic ecosystems. Recent studies show that similar pesticides alter aggressive and reproductive behaviors in teleost fish (Carlson et al. 1998; Roex et al. 2001) through the interference of cerebral neuromediating systems (Clark 1997) such as the histaminergic system. This system is actively involved in blocking chemical-dependent stressful conditions such as writhing and consequently modifies responses to cold and immobilization states (Ferretti et al. 1998; Ito 2000).
The histamine (HA) receptor complex is one of the main neurosignaling systems that, in addition to “allergic” and anti-inflammatory processes, has been recognized for its role in many neurologic functions such as attention, arousal, cognition, movement, and feeding in mammalian species (Lin et al. 1996). Structurally, the histaminergic neuronal fibers, originating from the hypothalamus, are projected extensively throughout the central nervous system and promote their actions via three distinct receptor subtypes denoted HnR (H1R, H2R, H3R) (Hill et al. 1997). All but H3R subtypes are postsynaptically located and are coupled positively to adenylyl cyclase and phospholipase C, whereas H3R has been isolated at both the presynaptic and postsynaptic levels (Brown et al. 2001). Recently, Spieler et al. (1999) observed that inhibition of the H1R site is linked to the improvement of appetitive reversal learning and memory tasks in the goldfish Carassius auratus—a relationship that was further supported by the detection, via the application of their specific and selective antagonists, of H3R (Peitsaro et al. 2000) and H1R (Choich et al. 2004) in brain areas of the zebrafish Danio rerio and the tilapia Oreochromis niloticus, respectively. Moreover, identification of these subtypes in other vertebrates such as amphibians and reptiles (Brodin et al. 1990; Inagaki et al. 1991) is consistent with their highly conserved profile throughout vertebrate phylogeny (Kaslin and Panula 2001). On the basis of the above data, the intention of our study was to establish whether Cd and endosulfan neurotoxicologic effects on behavioral activities could be controlled through cerebral histaminergic neuronal mechanisms in the ornate wrasse. In fish, adaptations to environmental variables, including chemical–physical water properties such as temperature, photoperiod, ionic balance, and pollutants (Huang et al. 2004), are involved in the acceleration of the diandric-protogynic physiologic state, which makes fish valuable models for the investigation of neurologic adaptation mechanisms (Bass and Grober 2001; Larson et al. 2003). Furthermore, the host-cleaning symbiosis of the ornate wrasse Thalassoma pavo not only is critical for the stability of ecosystems (Zander and Sötje 2002) but also improves environmental conditions for commercially valuable fish and thus makes the ornate wrasse an important ecologic bio-marker species.
Materials and Methods
Animals.
We collected 32 young female ornate wrasses (body weight, 20–25 g; length, ~ 16–18 cm) from the Tyrrhenian Sea; they were acclimated to laboratory conditions for 1 week. During this period, fish were maintained in flow-through tanks containing 150 L of seawater (19–21°C, pH 7.8) under a 12/12-hr light/dark photoperiod and fed daily (10 g/kg body weight) with commercial food (Morubel, Milan, Italy) corresponding to 2% of biomass in the tank. Animal maintenance and experimental procedures were in accordance with the Guide for Use and Care of Laboratory Animals (European Communities Council Directive 1986), and efforts were made to minimize animal suffering and reduce number of specimens used.
Experimental treatments and behavioral observations.
In the first part of the study, fish (n = 12) were exposed for 1 week to one of the two concentrations of Cd (CdCl2 · 2H2O; Sigma, Milan, Italy): a sublethal concentration (2.26 mg/L; n = 6) or a calculated maximum acceptable toxicant (MAT) concentration (11.32 mg/L; n = 6), which were both less than the 96-hr LC50 (lethal median concentration) value of 28.68 mg/L (Giusi et al. 2004) and the 96-hr LC50 value of 20.12 mg/L obtained in the white sea bass, Lates calcarifer (Thophon et al. 2003). This contaminant was prepared by dissolving CdCl2 · 2H2O in appropriate volumes of seawater. Other fish (n = 12) were exposed for the same length of time to two concentrations of endosulfan (Sigma): a sublethal concentration (0.2 μg/L; n = 6) and a calculated MAT concentration (1.3 μg/L; n = 6) that were both less than the 96-hr LC50 value of 3.30 μg/L (Giusi et al. 2004), and the 96-hr LC50 value varying between 1.4 and 1.5 μg/L for freshwater fish (Johnson and Finley 1980), sublethal and MAT concentrations that fall within the ranges reported in surface waters (0.039–0.205 μg/L) and after runoff water events (0.01–1.3 μg/L) from agricultural areas, respectively (Naqvi and Vaishnavi 1993). The pesticide was dissolved in seawater. We compared both treatment groups with controls (n = 8) consisting of fish maintained under identical conditions except that only vehicle was added to the tanks. During the entire experiment, accumulation of wastes and pathogens was avoided by replacing the tanks with fresh seawater every day. To reduce to a minimum the stressful conditions of this operation, we rapidly transferred the fish with a small fishing net to the tanks containing fresh seawater and either one of the two contaminants in an effort to achieve the intended nominal concentrations. Throughout the behavioral sessions, we checked the feeding habits of the ornate wrasse to ensure that the fish ingested water containing the neurotoxicant, although the uptake of these compounds primarily depends on their passage through the gill system (Thophon et al. 2003).
We assessed the behavior and mortality for all fish that received either Cd or endosulfan in four 1-hr sessions each day for 1 week. The motor and feeding behaviors that we analyzed included hyperactive movements consisting in either swimming toward the surface, swimming in the same direction, or “bumping” into each other or against the tank; assuming a “relaxed” position or simply being inactive; and hyper-ventilation, which is defined as the number of times that the operculum opens and closes in a 1-hr observation session. We also recorded feeding frequency and quantity (milligrams) of food ingested during each observation session. The quantity of food ingested was determined after the residual food recovered at the bottom of the tank was dried and weighed. The above motor activities and feeding behaviors of both treatment and control groups, estimated as mean activity per 24 hr ± SE, were recorded with a digital video camera (model TR 7000 E; Sony, Tokyo, Japan) and elaborated at a personal computer (Microsoft Windows XP; Microsoft Corp., Redmond, WA)) using EthoLog software (version 2.2.5; Visual Basic, São Paulo, Brazil) for behavioral analyses.
Amino cupric silver staining.
To establish whether abnormal behavioral activities were related to neuronal damage, fish treated in the same manner as in the behavioral study with sublethal (n = 3) and MAT (n = 3) concentrations of Cd and endosulfan were decapitated and their brains quickly removed (within 30 sec) and stored at −40°C according to common cryostat procedures for unfixed brains (Canonaco et al. 1997). Brains were mounted on a freezing stage of a sliding cryostat (Microm-HM505E; Zeiss, Wallford, Germany), and a serial set of representative coronal sections (30 μm) was selected at an interval of 240 μm for amino cupric silver staining procedures according to previously published methods (de Olmos et al. 1994); we adapted the exposition time (25 min) to neutral red for the different brain sections of our fish species. This method, which has been used mainly for mammalian brain studies, proved to be appropriate for cerebral neuronal fields that have undergone degeneration processes. The brain sections were rinsed with distilled H2O, placed into dishes containing the preimpregnating solution (silver nitrate [AgNO3], distilled H2O, D,L-alanine, copper nitrate [Cu(NO3)2], cadmium nitrate [Cd(NO3)2], lanthanum nitrate [La(NO3)2], neutral red, pyridine triethanolamine, isopropanol), heated in a microwave oven (45–50°C) for 50 min, and cooled at room temperature for 3 hr. The sections were then rinsed in distilled H2O, and after a quick rinse in acetone they were placed in an impregnating solution AgNO3, distilled H2O, ethanol, acetone, lithium hydroxide (LiOH), ammonium hydroxide (NH4OH)] for 50 min, followed by a 25-min fixation in a reducer solution (formalin, citric acid monohydrate, ethanol, distilled H2O) at a temperature range of 32–35°C. These sections were left overnight in distilled H2O, and the next day they were placed in a first bleaching solution [potassium ferricyanide in potassium chlorate solution, lactic acid] for 60 sec at room temperature. Afterward, they were bleached in a second bleaching solution (potassium permanganate, sulfuric acid) for 60 sec and rinsed in distilled H2O. For the stabilization phase, sections were transferred in sodium thiosulfate solution and rinsed again in distilled H2O. Finally, they were immersed in a rapid fixer solution for 5 min and counterstained with 0.5% neutral red solution (Carlo Erba, Milan, Italy) for 25 min, dehydrated in ethanol (50–100%) and xylene, and mounted with DPX (p-xylene-bis[N-pyridinium bromide]; Sigma) for observations with a bright-field Dialux EB 20 microscope (Leitz, Stuttgart, Germany). The effects of both neurotoxicants on the argyrophilic reaction at the different brain levels were compared with controls that consisted of fish maintained under conditions identical to those of the two treatment groups except that only vehicle was added to the tanks. Because the same negative results were obtained at all brain levels, only two representative posterior areas were illustrated and compared with the different brain areas of the two treatment groups.
Effects of Cd and endosulfan on H1R–H3R.
The neurotoxic actions of Cd and endosulfan were also correlated with the type of distribution pattern of the HnR neuronal system. Fish treated with the sublethal (n = 4) and MAT (n = 4) concentrations of Cd and endosulfan along with their control (n = 6) were used. The brains were removed and quickly frozen for storage at −40°C, and brain sections (14 μm thick) were thawed, dried at room temperature, and then handled according to in vitro binding studies for mammals (Ryu et al. 1996) that were adapted for fish brain sections (Peitsaro et al. 2000). Briefly, we incubated sections in 150 mM sodium potassium phosphate buffer (Sigma) 2 mM MgCl2 and 100 μM dithiothreitol pH 7.4 (Roche Diagnostic, Milan, Italy) containing different concentrations (0.5–20 nM) of [3H]-N-α-methyl-HA (NAMH; PerkinElmer Life Sciences, Boston, MA, USA). Some sections were incubated with 10 nM [3H]-NAMH using a wipe assay procedure. This concentration displayed the greatest affinity for HnR in the presence of the different values (1 μM–1 nM) of the following specific HA antagonists (Sigma): H1R antagonist pyrilamine, H2R antagonist cimetidine, and H3R antagonist thioperamide. Other sections were incubated with 10 nM [3H]-NAMH plus 500 μM of their corresponding antagonist for nonspecific binding values that proved to be similar to that of the background; subsequently an autoradiographic film (Hyperfilm; Amersham, Piscataway, NJ, USA) was apposed to dried sections and to slides containing plastic standards.
After an exposure period of 6 weeks (25°C), we developed the autoradiographic films according to previous methods (Canonaco et al. 1997) and we evaluated the different H1R–H3R binding densities, expressed in femtamole per milligram wet tissue weight, with a Panasonic Telecamera (objective lens FD; 50 mm, 1:3.5; Canon, Milan, Italy) attached to a Macintosh computer-assisted image analyzer system running Scion-Image 2.0 (National Institutes of Health Image, Bethesda, MD, USA). We stained labeled sections with cresyl violet acetate to identify the diencephalic, mesencephalic, and telencephalic brain regions, using the perch fish atlas (Cerdá-Reverter et al. 2001a, 2001b).
Statistical analysis.
For the receptor binding study, Scatchard analyses of saturation binding data, which were fitted by a one-site and/or two-site model [based on the significance of extrasum squares using a LIGAND program (Munson and Rodbard 1980)] supplied relative affinity states and maximal receptor binding densities. To compare behavioral observations and histaminergic receptor binding data, we compared the treatment groups using a one-way analysis of variance (ANOVA) when there was a significant p-value ≤ 0.05, according to the Neuman-Keuls multiple-range post hoc test.
Results
Behavioral analysis.
Treatment of the ornate wrasse with Cd and endosulfan accounted for a net differentiation in the type of behavior responses. The MAT concentration of both stressors—11.32 mg/L and 1.3 μg/L, respectively—induced stereotype motor behaviors during the entire experimental session. Fish treated with Cd at 11.32 mg/L exhibited greater (p < 0.001; Figure 1A) hyperactive swimming activities such as moving in only a vertical direction and/or “bumping” against each other or against the glass tanks, in contrast to controls, which were often inactive and spent most of their time along the bottom of the tank. Fish treated with a concentration of 2.26 mg/L Cd displayed only moderate stereotype behaviors (Figure 1B), including hyperactive movements that consisted of swimming mainly in a vertical direction toward the surface of the water, whereas controls exhibited more random movements. Conversely, endosulfan caused a significant increase of some hyperactive movements (p < 0.05; Figure 1A,B) such as swimming in a vertical direction, whereas “bumping” type of swimming behaviors occurred in a less significant manner. This pesticide markedly reduced feeding, even at the lower concentration (0.2 μg/L). With both concentrations of endosulfan tested, feeding behavior was irregular, and overall, treated animals ate less food than did controls (Figure 1C,D). The MAT concentrations of both contaminants caused an excessive production of mucus on the operculum surface and, after 24 hr, hyperventilation became increasingly more severe up to the end of the study (Figure 1E).
Analysis of amino cupric silver–stained tissue.
From the amino cupric silver staining analysis, it was possible to correlate these abnormal behaviors with evident neurodegeneration processes in telencephalic and mesencephalic regions. In particular, a MAT concentration of Cd supplied damaged external pyramidal neuron, as exhibited by a typically argyrophilic dark neuronal perikarya and often by a shrunken and folded appearance compared with little or no damage in controls (Figure 2d,h). This feature was limited mainly to the medial dorsal part of the telencephalon, subdivision 2 (Dm2; Figure 2a), and the pyramidal layer of the mesencephalic stratum griseum central (SGC; Figure 2b), which showed consistent dark axonal processes. The effects of Cd seemed to extend to other areas of the brain, namely, the anterior part of the nucleus glomerulosus (NGa; Figure 2c) of the diencephalic pretectal region that is involved, via mesencephalic circuits, with the regulation of visual motor functions in teleosts (Kaslin and Panula 2001). With endosulfan, substantial neurodegeneration was present in ventral telencephalic regions such as the entopeduncular nucleus (e; Figure 2e) plus the diencephalic suprachiasmatic nucleus (NSC; Figure 2f) and the medial part of lateral tuberal nucleus (NLTm; Figure 2g). In these brain regions endosulfan produced an altered pattern of neurons defined as an “interrupted string of pearls” as noted with degeneration of interneurons of mammals (Siegel et al. 1999).
Effects of Cd and endosulfan on H1R–H3R.
When the regional distribution of HA receptors was determined in the presence of Cd and endosulfan, we observed a peculiar pattern of histaminergic expressing neurons in the same above brain regions of Thalassoma pavo. Such a relationship was based on a similar optimal [3H]-NAMH binding constant (Figure 3) in both treated and control fish with respect to that of rodents (unpublished data). Overall, the highest (> 140 < 200 fmol/mg wet tissue weight) HA binding densities were shown to be typical of rostral areas such as the preoptic nucleus (NPO) as well as the torus longitudinalis (TLo) and SGC of midbrain regions, whereas lower (> 70 < 110 fmol/mg wet tissue weight) binding densities were reported for the central nucleus of the ventral telencephalon and molecular stratum of the cerebellum. Application of the selective HA receptor antagonists enabled us to demonstrate that it was the diencephalic region that proved to be a preferential target of the major distribution differences of all subtypes (H1R–H3R), as displayed by notable displacement capacities of these subtypes in the preoptic area (Figure 4), as well as high H1R and H2R levels in areas such as NPO (45%) and in the nucleus of the saccus vasculosus (NSV; 43%), respectively (Figure 5). The subtype H3R was predominantly higher in some regions and especially in Dm2 (45%) of the telencephalon and in TLo (44%) of the mesencephalon.
It is noteworthy that fish treated with a MAT Cd dose showed a down-regulating effect of H2R–expressing neurons, as displayed by the low binding densities in some midbrain regions of the representative autoradiograms (Figure 6). Of all the regions examined, SGC (−115%; p < 0.001) and NGa (−90%) of the mesencephalon (Figure 7A) seemed to contain the greatest down-regulating activities of H2R-producing neurons, whereas a moderate (p < 0.05) reduction was evident in the habenular nucleus (NH; −45%). A similar reduction that appeared to be also maintained for H1R-producing neurons and precisely a very great and great (p < 0.01) reduction of H1R levels in TLo (−105%) and the central posterior thalamic nucleus (CP; −65%), respectively, whereas a moderate up-regulating activity was instead detected for NLTm (+38%; Figure 7B). The subtype H3R did not appear to be a major target of Cd (Figure 7C) aside from the moderately higher levels (+40%) obtained in the external cellular layer (ECL) of the olfactory bulb.
The effects of endosulfan appeared instead to be preferentially directed toward H3R-producing neurons, as shown by the greater binding densities in the representative autoradiograms of midbrain regions (Figure 6b). The diencephalic region (Figure 8A) provided very great up-regulating effects, especially in the NLTm (+110%) and the nucleus of the posterior hypothalamic recess (NRP; +78%). By contrast, greatly decreased levels were detected in another hypothalamic area, that is, the ventromedial thalamic nucleus (VM; −70%). Moreover, the H1R-producing neurons of this brain region were a preferred target for endosulfan effects (Figure 8B), as indicated by the greatly increased levels in the NSC (+68%) plus moderately higher levels in the NSV (+40%). Conversely, the other subtype (H2R) did not prove to be a preferred target of this pesticide (Figure 8C) despite the moderately higher H2R levels in NPO (+60%).
Discussion
We describe here for the first time neurotoxicologic effects of the heavy metal Cd and the insecticide endosulfan that are responsible for abnormal motor and feeding behaviors through histaminergic mechanisms in the ornate wrasse. A first abnormal behavior consisted of stereotype motor activities such as swimming in a constant direction or “bumping” against each other and/or against the glass tank, especially when the fish were treated with a MAT Cd concentration. These abnormal behaviors, as reported previously for Thalassoma pavo observed under field conditions (Giusi et al. 2004) and in a wide variety of fauna ranging from terrestrial vertebrates such as rodents (Lafuente et al. 2001) to aquatic species such as amphibians (James et al. 2004) and Chordata Ascidaecea (Bellas et al. 2001), should not be surprising because of wide distribution of this heavy metal in the different ecosystems. This condition appears mainly in aquatic communities because Cd readily accumulates in the different tissues after uptake via the calcium transport pathway of gill’s chloride cells (Wood 2001), above all in the olfactory structures that are considered to be its preferential target (Tallkvist et al. 2002). In this context the interference of such sensory communicating structures in the ornate wrasse may offset normal responses to olfaction-mediated stimuli such as migration and physical contact with other fish, which is in accordance with the irregular responses to alarm signals as well as modification of aggressive social relationships that have been reported in rainbow trout treated with toxic Cd doses (Scott et al. 2003).
In pesticides both sublethal and MAT concentrations of endosulfan caused abnormal feeding behaviors, whereas altered swimming movements were less evident than in Cd-treated animals. As a consequence, the consumption of food at an asynchronous rhythm and at different time intervals is in good agreement with other pesticides, accounting for feeding difficulties via neuronal functional hindrances in the goldfish (Bretaud et al. 2000). Similar difficulties obtained even under sublethal concentrations tends to suggest that sensorimotor threshold activities are susceptible to this contaminant, as shown by “startled” motor behaviors being tightly associated with the olfactory-dependent neuromediation of optomotor responses such as predation, foraging, and orientation toward food odor (Pan and Dutta 1998). These olfactory-dependent responses appear to be determining elements for feeding behaviors throughout the various biologic phases of the fish, as demonstrated by both young and adult Japanese killifish being neither attracted to nor able to consume food after receiving similar endosulfan doses (Gormley and Teather 2003).
When the toxicologic actions of both environmental contaminants were assessed at the structural level of the brain, notable neuro-degenerative events were observed, as shown by the diffused amino cupric silver staining of neurons in the different brain regions. With this method it was possible not only to immediately and clearly detect the precise location of neuronal trauma (Siegel et al. 1999) but also to distinguish between somata and axonal damage in some diencephalic, mesencephalic, and telencephalic sites of Thalassoma pavo. Of the brain regions exposed to MAT Cd concentrations, the telencephalic Dm2 displayed the greatest axonal fiber damage and interstitial edema. This condition fits nicely with the infiltration properties of the heavy metal in mammalian telencephalic regions such as the hippocampus (Mendez-Armenta et al. 2001), which is involved in analogous functions such as learning, spatial memory, and motor behaviors that are controlled by Dm2 in fish (Portavella et al. 2004). Even SGC and NGa of mesencephalic and pretectal areas, respectively, which are related to the modulation of multisensorial inputs (visual, acoustic, and electroreceptive signals), supplied perturbed dendritic spine formation and deformed soma in a fashion similar to that of mesencephalic trigeminal neurons of rodents exposed to Cd (Yoshida 2001). The effects of endosulfan were instead involved mainly with axonal deformations of interneurons in diencephalic areas such as NSC and preoptic area of the hypothalamus, an effect that tends to overlap cellular alterations and interstitial infiltration events induced by the pesticide carbofuran in teleosts (Ram et al. 2001). Ram et al. (2001) also showed that such a contaminant was responsible for a reduction in number and size of neurons and consequently altered neurotransmission functions in this same brain region.
Interestingly, the neuronal alterations provoked by both environmental contaminants in the present study seemed to coincide with changes of the histaminergic transcriptional activities in some telencephalic and mesencephalic regions and in the anterior and posterior areas of the hypothalamus. The hypothalamic area is considered to be a key production site of HA not only in mammals (Pillot et al. 2002) but also in amphibians (Airaksinen and Panula 1990) and in some fish species such as the zebrafish (Kaslin and Panula 2001). Because toxicologic effects of Cd and endosulfan occur in distinct and localized brain regions seems to support strongly a behavior-linked relationship of these neurotoxins, as demonstrated by Cd being preferentially directed toward the motor-controlling cerebral regions and endosulfan being involved predominantly on endocrine-dependent activities of hypothalamic areas. The effects of Cd exposure on pretectal and tegmental areas are characterized primarily by a down-regulatory activity of H2R-expressing neurons, whereas a similar activity of H1R-expressing neurons was detected for TLo and CP. Moreover, on the basis of the low levels of H2R occurring not only in key motor telencephalic areas but also in mesencephalic and cerebellar regions of the mormyrid electric fish (Han et al. 2000) and of other vertebrates (Minami et al. 2001), it appears that a down-regulation of this subtype might represent an important condition of the histaminergic inhibitory effects on locomotor behaviors (Santos et al. 2003). The inhibitory effects may be accomplished by the regulation of parameters such as swimming velocity, location of objects, and overall vestibular activities that are controlled by these same brain regions (Meek 1990; Xue et al. 2003).
Conversely, endosulfan appeared predominantly to promote enhanced levels of H3R-expressing neurons mainly in hypothalamic areas such as NLTm and NRP as well as of H1R-expressing neurons in NSC. This relationship appears to be strengthened by decreased swimming and feeding behaviors obtained immediately (after 2 hr) in the Chinook salmon when treated with the organophosphate pesticide diazinon (Scholz et al. 2000). The finding that the diencephalic region is a major target of pesticide toxic effects should not be surprising because polychlorinated biphenyls interfere with other hypothalamic activities, including the regulation of body temperature and the activities of the hypothalamic–pituitary–gonadal circuits, with severe consequences on reproductive and hormone-releasing activities (Bloomquist 2003; Cooper et al. 2000). It is noteworthy that high levels of H3R-expressing neurons have been correlated with a reduction of food intake through the suppression of appetite and energy expenditure in the same hypothalamic areas (Takahashi et al. 2002). In addition, the high levels of H1R-expressing neurons in other hypothalamic sites of the ornate wrasse plus the inhibition of these subtypes accounting for improved feeding habits in the goldfish (Spieler et al. 1999) appear to be consistent with an important inhibitory role of H1R and H3R, at least in hypothalamic nuclei of this teleost.
In conclusion, these results provide direct evidence that the toxicologic risks of endosulfan and Cd on the motor and feeding behavior of Thalassoma pavo, as shown by evident morphologic neuronal damages and distinct HnR-expressing patterns, appear to be very strongly correlated with histaminergic neurosignaling mechanisms. Although most research to date has mostly considered the physiologic risks of the environmental toxicants, here we show that the abnormal behaviors could be linked to specific HA subtype interactions operating in some cerebral regions, at least in the ornate wrasse. Consequently, the motor activities appear to be tightly linked to Cd via variations of mainly H2R-expressing neurons in the mesencephalic and telencephalic regions, whereas modified feeding behaviors induced by endosulfan seem to be related to the differences of H1R- and H3R-expressing neurons mainly in hypothalamic areas. We are still at the beginning of this research, but molecular neuronal interests directed toward the role of environmental disruptors on aquatic organisms could provide further insights regarding not only the behavioral hazards of these contaminants but also neurotoxic mechanisms operating during the entire development cycle of fish, with the intent of minimizing ecologic and commercial risks of this very important class of vertebrates.
This study was supported partially by the contract sponsor MEMO-BIOMAR research program and COFIN (Cofinancial Projects) of MIUR (Italian University Research Ministry).
Figure 1 Assessment of effects of Cd and endosulfan on some motor activities of Thalassoma pavo: (A) hyperactive movements, (B) movements toward surface, (C) feeding frequency, (D) quantity (mg) of food ingested, and (E ) hyperventilation activity. For these behaviors a sublethal concentration of endosulfan and Cd as well as a MAT concentration of the two contaminants, respectively, were compared with controls. Values (means of activities/24 hr ± SE) of motor activities and feeding behaviors were estimated daily during four 1-hr observations for 1 week, as described in “Materials and Methods.” The behavioral data were analyzed by one-way ANOVA followed where necessary by post hoc Neuman-Keuls multiple-range test.
*p < 0.05; ***p < 0.001.
Figure 2 Photomicrographs showing the amino cupric silver staining pattern in rostral (i), middle (j), and posterior (k) brain areas of the Thalassoma pavo, treated with a MAT concentration of (a–c) Cd or (e–g) endosulfan. The effects of Cd (n = 6; arrows) were mostly observed in telencephalic and in mesencephalic areas such as Dm2 (a) and in the piriform SGC neurons of the optic tectum (b), respectively, and in the pretectal NGa (c), compared with control (n = 8); controls gave comparable results at all brain levels for both neurotoxicants as described in “Materials and Methods,” and so these same controls (d, h) were also used for the effects of endosulfan. In the case of endosulfan (n = 6), the major effects (arrows) were detected in the interneurons of the entopeduncular nucleus (e) and in the preoptic NSC area (f) and NLTm (g) of the hypothalamic lobe.
Abbreviations: CP, central posterior thalamic nucleus; Dc2, central part of dorsal telencephalon, subdivision 2; Dl, lateral part of the dorsal telencephalon; Dm2–Dm4, medial part of the dorsal telencephalon, subdivisions 2–4; DP, dorsal posterior thalamic nucleus; E, entopeduncular nucleus; NAP, anterior periventricular nucleus; NAT, anterior tuberal nucleus; NGa, anterior part of the nucleus glomerulosus; NH, habenular nucleus; NLTm, medial part of lateral tuberal nucleus; NLTv, ventral part of lateral tuberal nucleus; NPGm, medial preglomerular nucleus; NPO, preoptic nucleus; NPP, posterior periventricular nucleus; NSC, suprachiasmatic nucleus; NT, nucleus taenia; OT, optic tectum; POA, preoptic area; PSp, parvocellular superficial pretectal nucleus; SCO, subcommissural organ; TLo, torus longitudinalis; VCe, cerebellum valvula; VM, ventromedial thalamic nucleus; Vot, ventral optic tract; Vp, postcommissural nucleus of the ventral telencephalon.
Figure 3 (A) A saturation curve of [3H]-NAMH binding (fmol/mg wet tissue weight ± SE), using wipe assays, was determined for the preoptic area of the Thalassoma pavo treated with MAT concentrations of Cd and endosulfan and compared with controls as described in “Materials and Methods.” (B) From the linear Scatchard plot, the negative slope was calculated to provide the mean dissociation constant (nM), whereas the intercept of the curve at the abscissa provided the maximal number of binding sites. Evaluation of saturation-binding study supplied similar results in three separate trials.
Figure 4 Displacement curves of [3H]-NAMH (% of total binding) in preoptic area of the Thalassoma pavo (n = 6) were generated in the presence of different concentrations (1 μM to 1 nm) of cold NAMH and of selective HA antagonists thioperamide, pyrilamine, and cimetidine as described in “Materials and Methods.” Each point represents mean ± SE of three separate tests.
Figure 5 Percentage binding levels (of total) ± SE of H1R, H2R, and H3R sites in diencephalic (A) and extra-diencephalic (B) regions of the Thalassoma pavo (n = 6) were determined in the presence of their respective selective antagonists as described in “Materials and Methods.”
Abbreviations: Dm2, medial part of the dorsal telencephalon, subdivision 2; NPO, preoptic nucleus; NRP, nucleus of the posterior hypothalamic recess; NSC, suprachiasmatic nucleus; NSV, nucleus of the saccus vasculosus; SGC, stratum griseum central; TLo, torus longitudinalis; VM, ventromedial thalamic nucleus.
Figure 6 Representative binding autoradiograms displaying distinct receptor densities (black line) of H2R in the posterior regions of the Thalassoma pavo treated with a MAT concentration of Cd (a; n = 4) and of H3R in the same brain regions of animals that, instead, received a MAT concentration of endosulfan (b; n = 4), with respect to their corresponding (e, f) controls (n = 6). Binding pattern of these two subtypes appeared to be highly specific as shown by very similar background levels reported for [3H]-NAMH in presence of a 500× concentration of the selective antagonists cimetidine (c) and thioperamide (d), respectively, as described in “Materials and Methods.”
Abbreviations: NAT, anterior tuberal nucleus; NGa, anterior part of the nucleus glomerulosus; NLTm, medial part of lateral tuberal nucleus; OT, optic tectum; TLo, torus longitudinalis; VM, ventromedial thalamic nucleus.
Figure 7 The effects of both sublethal and MAT concentrations of Cd on H2R (A), H1R (B), and H3R (C) with respect to their controls (n = 6) were expressed as a percentage binding level ± SE in the different brain regions of the Thalassoma pavo, as described in “Materials and Methods.” The levels were compared using one-way ANOVA followed where necessary by a post hoc Neuman-Keuls multiple-range test.
*p < 0.05; **p < 0.01; ***p < 0.001.
Abbreviations: CP, central posterior thalamic nucleus; E, entopeduncular nucleus; ECL, external cellular layer of olfactory bulb; NGa, anterior part of the nucleus glomerulosus; NH, habenular nucleus; NLTm, medial part of lateral tuberal nucleus; NPO, preoptic nucleus; NSC, suprachiasmatic nucleus; NSV, nucleus of the saccus vasculosus; SGC, stratum griseum central; TLo, torus longitudinalis; VM, ventromedial thalamic nucleus.
Figure 8 The effects of both sublethal and MAT concentrations of endosulfan on H3R (A), H1R (B), and H2R (C ) with respect to their controls (n = 6) were expressed as a percentage binding level ± SE in the different brain regions of the Thalassoma pavo, as described in “Materials and Methods.” The levels were compared using one-way ANOVA followed where necessary by a post hoc Neuman-Keuls multiple-range test.
*p < 0.05; **p < 0.01; ***p < 0.001. Abbreviations: CP, central posterior thalamic nucleus; Dm2, medial part of the dorsal telencephalon, subdivision 2; NH, habenular nucleus; NLTm, medial part of lateral tuberal nucleus; NPO, preoptic nucleus; NRP, nucleus of the posterior hypothalamic recess; NSC, suprachiasmatic nucleus; NSV, nucleus of the saccus vasculosus; SGC, stratum griseum central; TLo, torus longitudinalis; VM, ventromedial thalamic nucleus.
==== Refs
References
Airaksinen MS Panula P 1990 Comparative neuroanatomy of the hystaminergic system in the brain of the frog Xenopus laevis J Comp Neurol 292 412 423 2341609
Bass AH Grober MS 2001 Social and neural modulation of sexual plasticity in teleost fish Brain Behav Evol 57 293 300 11641565
Beauvais SL Jones SB Parris JT Brewer SK Little EE 2001 Cholinergic and behavioral neurotoxicity of carbaryl and cadmium to larval rainbow trout (Oncorhynchus mykiss ) Ecotoxicol Environ Saf 49 84 90 11386719
Bellas J Vasquez E Beiras R 2001 Toxicity of Hg, Cu and Cr on early developmental stages of Ciona intestinalis (Chordata, Ascidiacea) with potential application marine water quality assessment Water Res 35 12 2905 12 11471690
Bisson M Hontela A 2002 Cytotoxic and endocrine-disrupting potential of atrazine, diazinon, endosulfan, and mancozeb in adrenocortical steroidogenic cells of rainbow trout exposed in vitro Toxicol Appl Pharmacol 180 110 117 11969378
Bloomquist JR 2003 Chloride channels as tools for developing selective insecticides Arch Insect Biochem Physiol 54 145 156 14635176
Bretaud S Toutant JP Saglio P 2000 Effects of carbofuran, diuron and nicosulfuron on acetylcholinesterase activity in goldfish (Carassius auratus ) Ecotoxicol Environ Saf 47 117 124 11023689
Brodin L Hökfelt T Grillner S Panula P 1990 Distribution of histaminergic neurons in the brain of the lamprey Lampetra fluviatis as revealed by histamine-immunoreactivity J Comp Neurol 292 435 442 2341611
Brown RE Stevens RS Haas HL 2001 The physiology of brain histamine Progr Neurobiol 63 637 642 11164999
Canonaco M Tavolaro R Facciolo RM 1997 Dimorphic distribution of the two main GABAA binding sites in cortical and limbic areas of a rodent living in natural environmental conditions J Comp Neurol 380 423 434 9087523
Carlson RW Bradbury SP Drummond RA Hammermesiter DE 1998 Neurological effects on startle responses and escape from predation by medaka exposed to organic chemicals Aquat Toxicol 43 51 68
Cerdá-Reverter JM Zanuy S Munoz-Cueto JA 2001a Cytoarchitectonic study of the brain of a perciform species, the sea bass Dicentrarchus labrax . I. The telencephalon J Morphol 247 217 228 11223929
Cerdá-Reverter JM Zanuy S Munoz-Cueto JA 2001b Cytoarchitectonic study of the brain of a perciform species, the sea bass Dicentrarchus labrax . II: The diencephalon J Morphol 247 229 251 11223930
Choich JA El-Nabawi A Silbergeld EK 2004 Evidence of histamine receptors in fish brain using an in vivo [14 C]2-deoxy-glucose autoradiographic method and an in vitro receptor-binding autoradiographic method Environ Res 94 86 93 14643290
Clark JM 1997 Insecticides as a tool in probing vital receptors and enzymes in excitable membranes Pest Biochem Physiol 57 235 254
Cooper RL Stoker TE Tyrey L Goldman JM McElroy WK 2000 Atrazine disrupts the hypothalamic control of pituitary-ovarian function Toxicol Sci 53 297 307 10696778
de Olmos JS Beltramino CA De Lorenzo SDO 1994 Use of amino-cupric silver technique for the detection of early and semiacute neuronal degeneration caused by neurotoxicants, hypoxia, and physical trauma Neurotoxicol Teratol 16 6 545 561 7532272
European Communities Council Directive 1986 Guide for use and care of laboratory animals Off J Eur Commun L358 1 29
Ferretti C Blengio M Ghi P Adage T Portaleone P Ricci Gamalero S 1998 Hypothalamic histamine release in normal and stressed rats is affected by sex and aging Pharmacol Biochem Behav 59 225 260
Giusi G Madeo M Alò R Carelli A Facciolo RM Canonaco M 2004 Effects of environmental stressors on neuronal histaminergic and behavioral activities of the teleost Thalassoma pavo Horm Behav 46 106
Gormley KL Teather KL 2003 Developmental, behavioural, and reproductive effects experience by Japanese medaka (Oryzias latipes ) in response to short-term exposure to endosulfan Ecotoxicol Environ Saf 54 330 338 12651189
Griffiths SW Armstrong JD 2000 Differential responses of kin and nonkin salmon to patterns of water flow: does recirculation influence aggression? Anim Behav 59 1019 1023 10860529
Han VZ Grant K Bell CC 2000 Rapid activation of GABAergic interneurons and possible calcium independent release in the mormyrid electrosensory lobe J Neurophysiol 83 3 1592 1604 10712482
Harvey PW Rush KC Cockburn A 1999. Endocrine and hormonal toxicology: an integrated mechanistic and target systems approach. In: Endocrine and Hormonal Toxicology (Harvey PW, Rush KC, Cockburn A, eds). Chichester, UK:Wiley, 1–11.
Hill SJ Ganellin CR Timmermann H Schwartz JC Shankley NP Young JM 1997 International union of pharmacology. XIII: Classification of histamine receptors Pharmacol Rev 49 253 278 9311023
Huang YS Wang LH Tsai CL 2004 Photoperiod affects the development of central neurotransmitter systems of tilapia, Oreochromis mossambicus Neurosci Lett 355 201 204 14732466
Inagaki N Panula P Yamatodani A Wada H 1991 Organization of the histaminergic system in the brain of the teleost Trachurus trachurus J Comp Neurol 310 94 102 1939731
Ito C 2000 The role of brain histamine in acute and chronic stresses Biomed Pharmacother 54 263 267 10917464
James SM Little EE Semlitsch RD 2004 The effect of soil composition and hydration on the bioavailability and toxicity of cadmium to hibernating juvenile American toads Environ Pollut 132 523 532 15325468
Johnson WW Finley MT 1980. Handbook of Acute Toxicity of Chemicals to Fish and Aquatic Invertebrates. Resource 137. Washington, DC:U.S. Department of Interior, Fish and Wildlife Service.
Kaslin J Panula P 2001 Comparative anatomy of the histaminergic and other aminergic system in zebrafish (Danio rerio ) J Comp Neurol 440 342 377 11745628
Lafuente A Fenàndez-Rey E Seara R Pérez-Lorenzo M Esquifino AI 2001 Alternate cadmium exposure differentially affects amino acid metabolism within the hypothalamus, median eminence, striatum and prefrontal cortex of male rats Neurochem Internat 39 187 192
Larson ET Norris DO Summers CH 2003 Monoaminergic changes associated with socially induced sex reversal in the saddleback wrasse Neuroscience 119 251 263 12763086
Liao VH Freedman JH 1998 Cadmium-regulated genes from the nematode Caenorhabditis elegans . Identification and cloning of new cadmium-responsive genes by differential display J Biol Chem 273 31962 31970 9822667
Lin JS Hou Y Sakai K Jouvet M 1996 Histaminergic descending inputs to the mesopontine tegmentum and their role in the control of cortical activation and wakefulness in the cat J Neurosci 16 1523 1537 8778302
Martinez MP Angulo R Pozo R Jodral M 1997 Organochlorine pesticides in pasteurized milk and associated health risks Food Chem Toxicol 35 621 624 9225021
Meek J 1990. Tectal morphology: connections, neurones and synapses. In: The Visual System of Fish (Douglas R, Djamgoz M, eds). London:Chapman and Hall, 239–277.
Mendez-Armenta M Barroso-Moguel R Villeda-Hernandez J Nava-Ruiz C Rios C 2001 Histopathological alterations in the brain regions of rats after prenatal combined treatment with cadmium and dexamethasone Toxicology 161 3 189 199 11297807
Miller LL Llados R 1999. Toxicological Profile for Endosulfan. Draft Report 4.3. Atlanta, GA:U.S. Agency for Toxic Substance and Disease Registry, 162–163, 169.
Minami A Takeda A Nishibaba D Takefuta S Oku N 2001 Cadmium toxicity in sinaptic neurotransmission in the brain Brain Res 894 336 339 11251212
Munson PJ Rodbard D 1980 LIGAND: a versatile computerize approach for characterization of ligand-binding systems Anal Biochem 107 220 239 6254391
Naqvi SM Vaishnavi C 1993 Bioaccumulation potential and toxicity of endosulfan insecticide to not-target animals Comp Biochem Physiol C 105 347 361 7900959
Pan G Dutta HM 1998 The inhibition of brain acetyl-cholinesterase activity of juvenile largemouth bass Micropterus salmoides by sublethal concentrations of diazinon Environ Res A 79 133 137
Peitsaro N Anichtchik OV Panula P 2000 Identification of a histamine H3 -like receptor in the zebrafish (Danio rerio ) brain J Neurochem 75 718 724 10899947
Pillai A Priya L Gupta S 2003 Effects of combined exposure to lead and cadmium on the hypothalamic-pituitary axis function in proestrous rats Food Chem Toxicol 41 379 384 12504170
Pillot C Heron A Cochois V Tardivel-Lacombe J Ligneau X Schwartz JC 2002 A detailed mapping of the histamine H3 receptor and its gene transcript in rat brain Neuroscience 114 1 173 193 12207964
Portavella M Torres B Salas C 2004 Avoidance response in goldfish: emotional and temporal involvement of medial and lateral telencephalic pallium J Neurosci 24 9 2335 2342 14999085
Ram RN Singh IJ Singh DV 2001 Carbofuran induced impairment in the hypothalamo-neurohypophyseal-gonadal complex in the teleost, Channa punctalus (Bloch) J Environ Biol 22 3 193 200 12017260
Risso-de Faverney C Devaux A Lafaurie M Girard JP Bailly B Rahmani R 2001 Cadmium induced apoptosis and genotoxicity in rainbow trout hepatocytes through generation of reactive oxygene species Aquat Toxicol 53 65 76 11254947
Roex EWM Giovannangelo M van Gestel CAM 2001 Reproductive impairment in the zebrafish Danio rerio , upon chronic exposure to 1,2,3-trichlorobenzene Ecotoxicol Environ Saf 48 196 201 11161695
Ryu JH Yanai K Zhao XL Watanabe T 1996 The effect of dopamine D1 receptor stimulation on the up-regulation of histamine H3 -receptors following destruction of the ascending dopaminergic neurones Br J Pharmacol 118 585 592 8762081
Santos NR Huston JP Brandão ML 2003 Blockade of histamine H2 receptors of the periaqueductal gray and inferior colliculus induces fear-like behaviors Pharmacol Biochem Behav 75 25 33 12759110
Scherer E McNicol RE Evans RE 1997 Impairment of lake trout foraging by chronic exposure to cadmium: a black-box experiment Aquat Toxicol 37 1 7
Scholz NL Truelove NK French BL Berejikian BA Quinn TP Casillas E 2000 Diazinon disrupt antipredator and homing behaviours in Chinook salmon (Oncorhynchus tshawytscha ) Can J Fish Aquat Sci 57 1911 1918
Scott GR Sloman KA Rouleau C Wood CM 2003 Cadmium disrupts behavioural and physiological responses to alarm substance in juvenile rainbow trout (Oncorhyncus mykiss ) J Exp Biol 206 1779 1790 12727999
Siegel JM Nienhuis R Gulyiani S Ouyang S Wu MF Mignot E 1999 Neuronal degeneration in canine narcolepsy J Neurosci 19 1 248 257 9870955
Spieler RE Nelson CA Huston JP Mattioli R 1999 Post-trial administration of H1 histamine receptor blocker improves appetitive reversal learning and memory in goldfish, Carassius auratus Neurosci Lett 277 5 8 10643884
Takahashi K Suwa H Ishikawa T Kotani H 2002 Targeted disruption of H3 receptors results in changes in brain histamine tone leading to an obese phenotype J Clin Invest 110 12 1791 1800 12488429
Tallkvist J Persson E Henriksson J Tiälve H 2002 Cadmium-metallothionein interactions in the olfactory pathways of rats and pikes Toxicol Sci 67 108 113 11961223
Thophon S Kruatrachue M Upatham ES Pokethitiyook S Jaritkhuan S 2003 Histopathological alterations of white seabass Lates calcarifer , in acute and subchronic cadmium exposure Environ Pollut 121 307 320 12685759
Waalkes MP Coogan TP Barter RA 1992 Toxicological principles of metal carcinogenesis with special emphasis on cadmium Crit Rev Toxicol 22 175 201 1388705
Wood CM 2001. Toxic responses of the gill. In: Target Organ Toxicity in Marine and Freshwater Teleosts (Schlenk D, Benson WH, eds). London:Taylor & Francis, 1–87.
Xue HG Yamamoto N Yang CY Kerem G Yoshimoto M Imura K 2003 Fiber connections of the torus longitudinalis and optic tectum in holocentrid teleosts J Comp Neurol 462 194 212 12794743
Yoshida S 2001 Re-evaluation of acute neurotoxic effects of Cd2+ on mesencephalic trigeminal neurons of the adult rat Brain Res 892 1 102 110 11172754
Zander CD Sötje I 2002 Seasonal and geographical differences in cleaner fish activity in the Mediterranean Sea Helgol Mar Res 55 232 241
|
16263506
|
PMC1310913
|
CC0
|
2021-01-04 23:41:31
|
no
|
Environ Health Perspect. 2005 Nov 14; 113(11):1522-1529
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7983
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8083ehp0113-00153016263507ResearchPersonal Care Product Use Predicts Urinary Concentrations of Some Phthalate Monoesters Duty Susan M. 12Ackerman Robin M. 12Calafat Antonia M. 3Hauser Russ 141 Department of Environmental Health, Occupational Health Program, Harvard School of Public Health, Boston, Massachusetts, USA2 Department of Nursing, School for Health Studies, Simmons College, Boston, Massachusetts, USA3 National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA4 Vincent Memorial Obstetrics and Gynecology Service, Andrology Laboratory and In Vitro Fertilization Unit, Massachusetts General Hospital, Boston, Massachusetts, USAAddress correspondence to R. Hauser, Environmental Health Department, Occupational Health Program, Building 1, Room 1405, 665 Huntington Ave., Boston, MA 02115-9957 USA. Telephone: (617) 432-3326. Fax: (617) 432-0219. E-mail:
[email protected] 2005 18 7 2005 113 11 1530 1535 3 3 2005 18 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Phthalates are multifunctional chemicals used in a variety of applications, including personal care products. The present study explored the relationship between patterns of personal care product use and urinary levels of several phthalate metabolites. Subjects include 406 men who participated in an ongoing semen quality study at the Massachusetts General Hospital Andrology Laboratory between January 2000 and February 2003. A nurse-administered questionnaire was used to determine use of personal care products, including cologne, aftershave, lotions, hair products, and deodorants. Phthalate monoester concentrations were measured in a single spot urine sample by isotope dilution–high-performance liquid chromatography coupled to tandem mass spectrometry. Men who used cologne or aftershave within 48 hr before urine collection had higher median levels of monoethyl phthalate (MEP) (265 and 266 ng/mL, respectively) than those who did not use cologne or aftershave (108 and 133 ng/mL, respectively). For each additional type of product used, MEP increased 33% (95% confidence interval, 14–53%). The use of lotion was associated with lower urinary levels of monobutyl phthalate (MBP) (14.9 ng/mL), monobenzyl phthalate (MBzP) (6.1 ng/mL), and mono(2-ethylhexyl) phthalate (MEHP) (4.4 ng/mL) compared with men who did not use lotion (MBP, 16.8 ng/mL; MBzP, 8.6 ng/mL; MEHP, 7.2 ng/mL). The identification of personal care products as contributors to phthalate body burden is an important step in exposure characterization. Further work in this area is needed to identify other predictors of phthalate exposure.
environmentpersonal care productsphthalatesurinary metabolites
==== Body
Phthalates are used industrially as plasticizers and solvents and as stabilizers for colors and fragrances. They are found in personal care products, medications, paints, adhesives, and medical equipment made with polyvinyl chloride plastics [Agency for Toxic Substances and Disease Registry (ATSDR) 1995, 2001, 2003]. Diethyl phthalate (DEP), di(2-ethylhexyl) phthalate (DEHP), butylbenzyl phthalate (BBzP), and di-n-butyl phthalate (DBP) are used in personal care products (Houlihan et al. 2002; Koo and Lee 2004). The potential effects of phthalates on human health are not well characterized. There is a paucity of existing data describing phthalate-associated human health outcomes, although animal studies have found testicular toxicity associated with phthalate exposure (Li et al. 1998; Parks et al. 2000).
Two studies provide preliminary evidence of associations between urinary concentrations of monoethyl phthalate (MEP), a metabolite of DEP, and DNA damage in human sperm (Duty et al. 2003a), as well as relationships of monobutyl phthalate (MBP) and monobenzyl phthalate (MBzP) phthalate, metabolites of DBP and BBzP, respectively, with decreased sperm motility (Duty et al. 2003b). In a recent epidemiologic study prenatal exposure to MEP, MBP, MBzP, and monoisobutyl phthalate was associated with shortened anogenital distance (AGD) in male infants (Swan et al. 2005). In rodent studies AGD is a sensitive measure of prenatal antiandrogen exposure.
Despite the recent public and scientific interest on the potential human health effects of phthalates, routes of human exposure to phthalates have not been adequately characterized. Potential routes include dietary ingestion of phthalate-containing foods, inhalation of indoor and outdoor air, and dermal exposure through the use of personal care products that contain phthalates. As far as we know, the proportional contribution of phthalate-containing personal care products to total body burden has not been studied. Houlihan et al. (2002) quantified phthalate levels in 72 personal care products obtained at a supermarket in the United States, including hair gel/hair spray, body lotion, fragrances, and deodorant. DEP was detected in 71% of these products, DBP in 8%, BBzP in 6%, and DEHP in 4% of the products tested (Houlihan et al. 2002). In a recent study (Koo and Lee 2004), high-performance liquid chromatography (HPLC) was used to quantify the levels of the same four phthalates in 102 hair sprays, perfumes, deodorants, and nail polishes purchased at retail stores in Seoul, Korea. DBP was detected in 19 of the 21 nail polishes and in 11 of the 42 perfumes; DEP was detected in 24 of the 42 perfumes and 2 of the 8 deodorants.
The assertion that phthalates are absorbed into the circulation through human skin is physiologically plausible and is supported by a limited number of human and animal studies (ATSDR 1995, 2001, 2003). The stratum corneum of the epidermis regulates transdermal absorption, and uptake is achieved through passive diffusion (Howard et al. 2001). Water-soluble substances penetrate hydrolyzed keratin, whereas lipid-soluble substances such as phthalates, especially DEP and other low-molecular-weight phthalates, can dissolve into lipid materials between keratin filaments. After penetration of the epidermis, diffusion into the dermal and subcutaneous layers is generally uninhibited because of the nonselective and porous aqueous mediums in these layers. Substances can then enter the systemic circulation through venous and lymphatic capillaries. With increased hydration, rates of absorption of more hydrophilic compounds can be increased 3–5 times more than usual. Epidermal permeability also varies greatly between species (Howard et al. 2001).
In one study dermal doses of DEHP were administered to human volunteers over a 24-hr period, and approximately 1.8% of the total dose was absorbed (Wester et al. 1998). Another experiment involved the topical application of DBP to human volunteers. The authors determined that 68 mg would be absorbed in 1 hr if the skin surface of the whole body were saturated with the chemical (Hagedorn-Leweke and Lippold 1995). In another study human breast skin was exposed in vitro to 14C-DEP, and average absorption under conditions of occlusion was 3.9% compared with 4.8% without occlusion at 72 hr. However, this was much slower and less complete compared with absorption through rat skin (Mint et al. 1994).
In vitro and animal experiments have also indicated that phthalates are absorbed percutaneously (Barber et al. 1992; Deisinger et al. 1998; Elsisi et al. 1989; Melnick et al. 1987; Mint et al. 1994; Ng et al. 1992; Scott et al. 1987). However, the mechanism explaining differential rates of uptake is not agreed upon. Scott et al. (1987) attributed the phthalate-specific rates of absorption to varying degrees of lipophilicity. Elsisi et al. (1989) observed that the lengths of the alkyl chains were inversely associated with the relative rates of absorption; except for dimethyl phthalate, DEP has the shortest alkyl chain (ATSDR 1995).
Although diester and monoester phthalates have short biologic half-lives of approximately 6–12 hr and do not accumulate (ATSDR 1995, 2001, 2003), the frequent application of personal care products may result in semi-steady-state levels, making it possible to estimate typical phthalate body burden from a single urine sample (Hauser et al. 2004; Hoppin et al. 2002). After exposure, diester phthalates, which may be found in personal care products, are metabolized to monoester metabolites, the suspected toxic agents (Li et al. 1998). For this reason and to avoid contamination, monoester phthalate metabolites rather than the parent diesters are commonly measured (Blount et al. 2000).
Our objective in the present study was to determine whether the use of personal care products predicted urinary levels of phthalate monoesters, and to identify subject characteristics that predicted phthalate levels.
Materials and Methods
Design and setting.
This study was approved by the Human Subject Committees at the Harvard School of Public Health, Massachusetts General Hospital (MGH), and Simmons College. All subjects signed an informed consent. Subjects were participants in an ongoing study on phthalates and male reproductive health. They were recruited between January 2000 and February 2003 from the Andrology Laboratory at MGH. Males between 20 and 54 years of age who were partners of subfertile couples were eligible; those who have had a vasectomy were excluded. Approximately 65% of eligible men agreed to participate. The most frequently cited reason for not participating was lack of time. A total of 406 men were recruited.
Personal care product use assessment.
A trained research nurse administered a brief questionnaire to each subject at the time of his visit to the MGH andrology clinic for semen and urine sample collection. Information was obtained on personal care product use, smoking status, age, height, weight, race, and use of medications. Participants were specifically asked whether they had used hair gel/hair spray, lotion, aftershave, cologne, or deodorant in the 48 hr before the collection of the urine sample. They were also asked to record the time they last used the products within the 48-hr period.
Urinary phthalate monoester measurement.
A single spot urine sample was collected from each participant in a sterile plastic specimen cup (which was prescreened for phthalates) on the same day that the questionnaire was administered. The analytical approach has been described in detail (Blount et al. 2000) and adapted to both enable the detection of additional monoesters and improve efficiency of the analysis (Silva et al. 2003). Briefly, measurement of monoester metabolites, namely, MEP, MBP, mono(2-ethylhexyl) phthalate (MEHP), MBzP, and monomethyl phthalate (MMP), entailed enzymatic deconjugation of the phthalates from their glucuronidated form, solid-phase extraction, HPLC separation, and tandem mass spectrometry detection. The limits of detection (LODs) were approximately 1 ng/mL. One method blank, two quality control samples (human urine spiked with phthalate monoesters), and two sets of standards were analyzed along with every 21 unknown urine samples. Analysts at the Centers for Disease Control and Prevention (CDC) were blind to all information concerning subjects. To control for urinary dilution, urinary concentrations were adjusted according to specific gravity. Specific gravity was measured using a handheld refractometer (National Instrument Company Inc., Baltimore, MD). The following formula was used to adjust phthalate concentrations by specific gravity: Pc = P[(1.024 − 1)/SG − 1], where Pc represents specific gravity–corrected phthalate concentration (ng/mL), P is the measured phthalate concentration (ng/mL), and SG is the specific gravity of the sample. Specific gravity–adjusted monoester phthalate levels were used as continuous outcome variables in statistical models.
Statistical methods.
All analyses were performed using SAS software (version 8.1; SAS Institute Inc., Cary, NC). The use of each personal care product was categorized into a dichotomous variable (yes/no use in the 48 hr before the urine sample collection).
Because the phthalate monoester levels were not normally distributed, nonparametric tests were used to assess univariate associations between personal care product use and urinary phthalate levels. Multiple linear regression was used to explore the relationship between each of the five personal care products and each of the five log-transformed monoester phthalate concentrations. In addition, a six-level sum variable was created, representing the number of different types of products used by a participant in the past 48 hr; possible values for this variable were 0, 1, 2, 3, 4, or 5. To determine if a dose–response relationship existed between urinary phthalate levels and the number of types of personal care products used, a trend test was performed using sum variable as an ordinal variable. For urinary phthalate concentrations that were below the LOD, a value equal to half the LOD was imputed (except when quantification was given) as follows: MEP, 0.605 ng/mL; MBzP, 0.235 ng/mL; MBP, 0.47 ng/mL; MEHP, 0.435 ng/mL; and MMP, 0.355 ng/mL.
After evaluating appropriateness using quadratic terms, we modeled age and body mass index (BMI; kilogram per square meter) as continuous independent variables. Smoking status was categorized as current smoker and current nonsmoker (includes ex-smokers and never smokers). Race was coded as African American, Hispanic, and other race, with Caucasian as the reference group. On the basis of biologic plausibility and statistical factors (i.e., change in parameter estimate), we included age, BMI, race, and smoking variables in all models as potential confounders.
To explore the relationship between time of product use and urinary levels of the phthalates, we regressed log-phthalate levels on the time between product use and urine sample collection (referred to as TIMEDIF). TIMEDIF was categorized into four intervals: product use 0–3 hr before urine sample collection (TIME0–3); > 3 but ≤6 hr (TIME3–6); > 6 but ≤8 hr (TIME6–8), and > 8 hr (TIME9). Approximately 75–85% of subject’s product use was within 8 hr of urine collection, and therefore we used TIMEDIF > 8 hr as the reference category.
Results
Subject demographics.
Of the 406 men recruited for an ongoing semen quality study, 37 did not provide urine samples. Of the remaining 369, specific-gravity analyses were not available for 32, leaving 338 for primary analysis. Additionally, one urine sample was missing MMP concentrations. The study population was composed largely of white (n = 275, 82%), nonsmoking men (n = 304, 91%) (Table 1). There were 19 African-American men, 18 Hispanic men, and 24 men of other race/ethnicity.
Personal care product use.
Eleven men (3%) did not provide complete product use information (Table 1). Most men reported use of deodorant (89%), whereas fewer men reported using hair gel (37%), lotion (34%), cologne (29%), and aftershave (13%). Nine men (2.7%) did not use any of the personal care products listed on the questionnaire, 114 (33.7%) used only one type of product, 119 (35.3%) used two types of products, 71 (21%) used three types of products, 22 (6.5%) used four different types of products, and only 3 (0.9%) of the men used five or more different types of products within 48 hr of urine collection. The percentage of African-American (59%) and Hispanic (53%) men who reported using cologne within 48 hr of urine collection was higher the percentage of Caucasian men (25%) or men of other races (25%). Additionally, African-American men (65%) were more likely than Hispanic (44%), Caucasian (30%), or men of other races (43%) to use lotion. No other associations were seen between any other personal care products and race. Interestingly, men who used aftershave were almost twice as likely (18.5%) to also use cologne as non-aftershave users (9.8%) (chi squared p = 0.03). There were no consistent relationships among any of the other products used.
Urinary phthalate monoesters.
There was a wide distribution of both specific gravity–adjusted (Table 2) and -unadjusted phthalate monoester levels (Table 3). Five phthalate monoesters were detected in 75–100% of subjects. MEP was the most prevalent (100%), followed by MBP (95%) and MBzP (90%). MEHP and MMP were both found in about 75% of subjects. Phthalate metabolite concentrations are presented both adjusted for specific gravity and unadjusted for comparison with other studies. The highest geometric mean levels were found for MEP (179 ng/mL), followed by MBP (16.6 ng/mL), MBzP (7.1 ng/mL), MEHP (6.6 ng/mL), and last, MMP (4.5 ng/mL).
Covariate relationships.
Race and cigarette smoking status were predictors of MEP and MBP levels (Table 4). We found significantly higher median MEP levels among African-American men (506 ng/mL) and Hispanic men (395 ng/mL) compared with Caucasian men (140 ng/mL) and those men categorized as other race (125 ng/mL). Median MBP levels in Caucasian men (15.3 ng/mL) were also lower than among African-American men (32.7 ng/mL) and Hispanic men (29.1 ng/mL), and among men identified as other race (26.5 ng/mL). Median MEP levels in current smokers (250 ng/mL) were significantly higher than among nonsmokers (143 ng/mL) (Table 4). BMI was weakly, although positively, correlated with MEP (Spearman correlation coefficient of 0.1, p < 0.05). Age was not associated with any of the five phthalate concentrations. Wilcoxon rank-sum tests showed positive associations between the sum variable for product use and African-American men and men of other races compared with Caucasians. BMI was positively associated with those identified as other race.
Product use and urinary phthalate relationship.
In the univariate analyses, median MEP levels were higher among cologne users (265 ng/mL) compared with those who did not use cologne (108 ng/mL). Likewise, men who used aftershave had higher median MEP levels (266 ng/mL) than men who did not (133 ng/mL). Fragranced products such as cologne and aftershave contain relatively higher DEP levels than other personal care products. Figure 1, created on a subset of men who used cologne plus additional products, depicts the rise in MEP levels with specific combinations of personal care product use.
Median MBP was lower among men who had used deodorant (16.3 ng/mL) compared with those who did not use deodorant (22.5 ng/mL). The use of lotion was associated with lower median levels of MBP (14.9 ng/mL), MBzP (6.1 ng/mL), and MEHP (4.4 ng/mL) compared with men who did not use lotion (MBP, 16.8 ng/mL; MBzP, 8.6 ng/mL; MEHP, 7.2 ng/mL) (Table 5).
Men of Hispanic, Caucasian, and other races who used cologne had considerably higher median MEP levels (981, 444, and 178 ng/mL, respectively) than non-cologne users of similar race (138, 102, and 116 ng/mL; p = 0.09, < 0.001, and 0.13, respectively). Interestingly, African-American men who used cologne had 30% lower median MEP levels compared with non-cologne users (371 ng/mL vs. 508 ng/mL), although the differences were not statistically significant. Hispanic and Caucasian men had substantially higher MEP levels if they used aftershave (1076 and 220 ng/mL) than if they did not (138 and 126 ng/mL; p = 0.08 and 0.03, respectively). African-American men who used aftershave had 33% lower MEP levels compared with non-aftershave users (340 ng/mL vs. 508 ng/mL), although the differences were not statistically significant. No other race/product associations were observed.
Multiple linear regression.
In multiple linear regression models, after adjusting for race, smoking status, BMI, and age, urinary levels of MEP were 2.57 times higher among men who had used cologne and 1.71 times higher among aftershave users compared with men who did not report the use of these products (p < 0.0001 and 0.02, respectively) (Table 6). There was also a dose–response relationship between urinary phthalate MEP levels and the number of types of personal care products used. For every additional type of product used, MEP concentrations increased 33% (95% confidence interval, 14–53%; trend test p = 0.0002) (Figure 2). The use of deodorant was associated with 30% lower MBP levels (p = 0.08). MBP, MBzP, and MEHP levels were 31% (p = 0.004), 34% (p = 0.003), and 34% (p = 0.003) lower, respectively, among men who had used lotion within the past 48 hr before urine collection compared with men who had not.
Time of product use.
In secondary analyses, we explored the relationship between time of product use and urinary levels of phthalate monoesters. Statistical power was limited in these secondary analyses as a result of small sample sizes, generally fewer than 15 subjects for each of the four TIME strata. The analyses were performed only among users of each specific product. Cologne use at TIME0–3, TIME3–6, and TIME6–8 compared with cologne use at TIME9 was associated with an increase in MEP of 1.7-fold (p = 0.17), 2.8-fold (p < 0.01), and 1.1-fold (p = 0.75), respectively. No consistent time trends were observed for the other phthalates and cologne use. Aftershave was inconsistently associated with a 2- to 3-fold increase in MEP levels—3.0-fold increase at TIME0–3 (p = 0.15), 2.0-fold increase at TIME3–6 (p = 0.25), and 2.6-fold increase at TIME6–8 (p = 0.11)—compared with aftershave use at TIME9. No time trends were observed for the other phthalates and aftershave use. For lotion use at TIME0–3, TIME3–6, and TIME6–8, MBP concentration increased 1.9-fold (p = 0.03), 1.2-fold (p = 0.55), and 1.2-fold (p = 0.52) compared with lotion use at TIME9. No significant time relationships were found between lotion use and any other phthalate or between deodorant or hair gel use and any of the phthalates.
Discussion
In the present study, men who used cologne and/or aftershave within the 48-hr period before the collection of the urine sample had higher urinary levels of MEP. This is not unexpected because previous studies have demonstrated that DEP, the parent compound of MEP, is an ingredient in many colognes, deodorants, and fragranced products (Houlihan et al. 2002; Koo and Lee 2004) and that percutaneous absorption of DEP occurs (Api 2001; ATSDR 1995; Mint et al. 1994; Scott et al. 1987). More striking is the steepness of the dose–response relationship between the number of product types used in the 48 hr before urine collection and urinary MEP levels. DEP was found in 71% of the personal care products tested in one study (Houlihan et al. 2002), whereas DEHP, DBP, and BBzP were found in fewer than 10% of products. In another study, DEP was found in 57% of the perfumes and 25% of the deodorants surveyed; DBP, DEHP, and BBzP were not detected in any of the deodorants and in fewer than 27% of the perfumes (Koo and Lee 2004). Therefore, it is plausible that MEP would have a strong relationship with multiple product use, whereas the other phthalate monoesters would not.
Interestingly, the use of body lotion was associated with lower levels of MBP, MBzP, and MEHP. The reason for this relationship is not known, although several hypotheses are plausible. It is possible that other ingredients in body lotion may act as a barrier to the absorption of DBP, BBzP, and DEHP. It is also feasible that men who use lotion use fewer other personal care products. However, chi-squared tests did not show significant inverse relationships between the use of body lotion and other products (data not shown). An alternative explanation is that the urinary levels of these monoesters reflect exposure to their parent phthalates other than by use of personal care products. Percutaneous absorption after dermal exposure is expected to be lower for DBP, BBzP, and DEHP than for DEP.
The quantities of phthalates present in different brands of deodorant, aftershave, hair gel/hair spray, lotion, and cologne are known to be quite variable (Houlihan et al. 2002; Koo and Lee 2004). In the present study, because information on the use of specific brand name products was not gathered, the analysis was performed by category of product. This approach is likely to introduce bias toward the null because not all products within a given category contain phthalates and those that do contain phthalates do so at variable concentrations. Because the participants in this study are all male, it is unclear whether the findings of this study may be generalizable to women, who may use different types and combinations of personal care products.
It is unclear why current smokers had higher levels of MEP. The results, however, were unstable because the sample size was small: only 31 men (9%) were current smokers. One potential explanation is that smoking may alter the toxicokinetics of DEP. Although DBP, unlike DEP, is listed as an ingredient in the filters of Phillip Morris cigarettes (Phillip Morris 2004), MBP was not found to be related to current smoking status.
Racial differences in urinary levels of MEP and MBP were consistent with previous data from the National Health and Nutrition Examination Survey (NHANES) 1999–2000 that have shown African Americans and Hispanics have higher urinary levels of MEP and MBP than do Caucasians (CDC 2003; Silva et al. 2004). In our study we explored the MEP and race associations for use of specific personal care products. The higher MEP levels for Hispanic than for Caucasian men appeared related to differentially higher cologne and aftershave use. Interestingly, the higher urinary MEP levels in African-American than in Caucasian men did not appear to be related to higher cologne and/or aftershave use. Therefore, the use of other products not identified in this study, different sources of exposure to DEP, or differential toxicokinetics may be driving the high MEP levels among African-American men. After accounting for race, age, and smoking status in the statistical models, MEP levels were still significantly higher among cologne and aftershave users; African-American race remained an independent predictor of MEP levels. However, it is important to note that only 18 African Americans participated in the study, and these findings may be related to chance because of the small numbers. Further study on racial/ethnic differences is warranted.
In an earlier study on the relationship between demographic characteristics and urinary phthalate levels among a nonrepresentative subset of 289 participants of NHANES III, MBP, MBzP, and MEHP were higher in individuals of low socioeconomic status (Koo et al. 2002). Urban residence was also significantly associated with higher MEHP and MBP levels. Socioeconomic status and area of residence were not controlled for in the present study, and these factors could potentially account for some of the differences measured between the racial groups. Finally, it is also possible that higher personal care product use or the selection of certain types of products among racial groups may contribute to differences in urinary phthalate levels.
The time elapsed between product use and urine sample collection influenced the relationship between cologne use and MEP concentrations. MEP was 2.7-fold higher when cologne was used between 3 and 6 hr before urine collection compared with when it was used 8 hr or more before urine collection. Therefore, to best assess the relationship of cologne use on urinary MEP levels, we suggest that urine collection should occur 3–6 hr after cologne use.
When time of lotion use was not accounted for in the analysis, there was an inverse association between urinary levels of MBP and lotion use. However, in analyses in which time of use was explored, MBP concentrations were significantly higher within the first 3 hr after lotion use compared with lotion use 8 hr or more before. The lotion use MBP relationship may require a larger data set to determine how use correlates with MBP levels in urine samples collected at variable times after applying lotion.
Although aftershave use between 0 and 8 hr before urine collection was associated with 2- to 3-fold higher MEP concentration compared with aftershave use more than 8 hr before urine collection, each strata had fewer than 10 subjects, and the reference group had only 15. This could explain why the aftershave–time of use relationships did not reach statistical significance.
To put these findings into perspective, a comparison with previous work is offered. The interquartile difference (443 ng/mL) in MEP, associated with increased DNA damage in sperm (Duty et al. 2003b), was approximately 2- to 3-fold higher than the difference in levels of MEP observed between men who did versus those who did not use cologne (312 ng/mL) or aftershave (131 ng/mL), respectively. MBP and MBzP, found in our previous study to be associated with decreased sperm motility and concentration (Duty et al. 2003b), were not found to be associated with aftershave or cologne use.
Conclusions
Cologne and aftershave use were associated with significantly higher urinary MEP levels after controlling for age, BMI, smoking, and race. Additionally, a dose–response relationship was found between the number of different types of personal care products used and MEP urinary concentrations. Interestingly, lotion was inversely associated with most phthalate levels. Secondary analysis revealed that, for cologne, product use 3 to 6 hr before urine collection was most predictive of urinary MEP concentration. However, for lotion, product use in the 3 hr before urine collection was most predictive for MBP concentration.
The identification of personal care products as contributors to phthalate body burden is an important step in exposure characterization. Additionally, the results of this study suggest that the time that products are used in relation to the time that the urinary samples are collected should be documented. This will help reduce random measurement error in statistical analysis. Further work is needed to identify additional predictors of phthalate exposure.
We thank J. Rico, J. Frelich, L. Godfrey-Bailey, L. Pothier, A. Trisini, R. Dadd, M. Silva, and J. Reidy.
We acknowledge the National Institute of Environmental Health Sciences grant ES09718, ES00002. National Institutes of Health training grant T32 ES07069 supported R.M.A. The authors declare they have no competing financial interests.
Figure 1 Specific-gravity–adjusted urinary MEP concentration according to combinations of product types used. Data points represent medians; error bars represent 25th and 75th percentiles.
Figure 2 Specific-gravity–adjusted urinary MEP concentrations according to number of product types used. Data points represent medians; error bars represent 25th and 75th percentiles.
Table 1 Characteristics of study subjects (n = 338).
Characteristic Value
Age [median (25%, 75%)] 35.0 (32.0, 39.1)
BMI [median (25%, 75%)] 27.5 (25.0, 30.6)
Racea [n (%)]
White 275 (82)
Black/African American 18 (5)
Hispanic 19 (6)
Other 24 (7)
Smokingb [n (%)]
Current smoker 31 (9)
Nonsmoker (ex- and never smoker) 304 (91)
Use of personal care products [n (%)]
Lotionc 110 (34)
Hair gel/hair sprayd 121 (37)
Aftershavee 42 (13)
Deodorantf 299 (89)
Cologneg 94 (29)
a Race data missing for 2 men.
b Smoking data missing for 3 men.
c Lotion use data missing for 11 men.
d Hair gel/hair spray data missing for 7 men.
e Aftershave data missing for 8 men.
f Deodorant data missing for 1 man.
g Cologne data missing for 8 men.
Table 2 Distribution of specific gravity–adjusted urinary levels of phthalate monoesters: percentiles and summary statistics (ng/mL).
Percentile
Phthalate n 5th 25th 50th 75th 95th Mean ± SD Geometric mean
MEP 338 24.5 58.2 154 503 2,030 490 ± 979 179
MEHP 338 < LOD 2.4 6.3 19.1 116.1 27.6 ± 69.1 6.6
MBP 338 3.1 10.3 16.5 30.6 68.2 76.2 ± 798.4 16.6
MBzP 338 < LOD 4.0 7.7 14.1 39.7 14.0 ± 34.6 7.1
MMP 337 < LOD 2.1 4.8 11.4 32.1 10.8 ± 22.8 4.5
LODs (ng/mL): MEP, 1.21; MBzP, 0.47; MBP, 0.94; MEHP, 0.87; MMP, 0.71.
Table 3 Distribution of unadjusted urinary levels of phthalate monoesters: percentiles and summary statistics (ng/mL).
Percentile
Phthalate n 5th 25th 50th 75th 95th Mean ± SD Geometric mean
MEP 338 17.0 48.5 145 457 1,953 485 ± 1,008 164
MEHP 338 < LOD 1.9 5.2 18.4 134.6 25.6 ± 60.1 6.0
MBP 338 2.2 7.8 14.5 31.7 75.1 85.6 ± 932.8 14.9
MBzP 338 < LOD 2.9 6.8 14.1 41.3 13.90 ± 32.4 6.0
MMP 337 < LOD 1.7 4.5 10.1 31.3 11.0 ± 31.6 4.1
LODs (ng/mL): MEP, 1.21; MBzP, 0.47; MBP, 0.94; MEHP, 0.87; MMP, 0.71.
Table 4 Median (25th and 75th percentiles) urinary levels of phthalate monoesters (ng/mL)a by race and smoking status.
MEP MEHP MBP MBzP MMP
Race
Black 506* (294, 1,134) 7.4 (3.9, 8.7) 32.7* (18.1, 42.5) 10.7 (6.8, 21.4) 6.0 (2.0, 12.0)
Hispanic 395* (83.3, 1,076) 5.6 (3.3, 20.7) 29.1* (17.3, 42.4) 12.2 (3.5, 19.5) 5.2 (1.9, 11.9)
Other 125 (40.3, 218) 7.1 (1.7, 10.8) 26.5* (7.1, 38.4) 6.4 (2.3, 11.9) 4.6 (2.0, 11.3)
White 140 (56.6, 469) 6.2 (2.3, 20.7) 15.3 (9.9, 26.9) 7.4 (4.0, 14.2) 7.3 (3.3, 11.3)
Smoking
Yes 250* (96.5, 826) 5.0 (1.8, 12.6) 20.9 (10.8, 46.9) 8.0 (4.3, 17.4) 8.5 (4.6, 18.2)
No 144 (57.5, 465) 6.4 (2.5, 19.6) 16.3 (10.3, 30.0) 7.7 (4.0, 14.2) 4.5 (2.1, 10.4)
a Specific gravity–adjusted phthalate levels.
*Univariate regression analysis p ≤0.05; reference group for race is whites.
Table 5 Median (25th and 75th percentiles) urinary levels of phthalate monoesters (ng/mL)a by personal care products used 48 hr before urine sample collection.
MEP MEHP MBP MBzP MMP
Lotion
Yes 136 (60.0, 438) 4.6* (1.6, 11.5) 15.7* (8.5, 28.1) 6.1* (3.0, 11.5) 4.2 (2.0, 10.4)
No 160 (56.6, 528) 7.2 (2.7, 20.4) 16.8 (10.3, 31.7) 8.6 (4.6, 15.1) 5.1 (2.4, 12.1)
Cologne
Yes 422* (155, 1076) 5.3 (1.8, 16.9) 18.6 (12.2, 33.6) 10.5 (4.6, 16.7) 4.7 (2.6, 13.2)
No 110 (48.0, 293) 6.6 (2.5, 19.1) 15.4 (9.6, 29.2) 6.8 (3.9, 13.8) 4.9 (2.1, 10.3)
Deodorant
Yes 165 (60.4, 534) 6.1 (2.4, 20.7) 16.3 (10.3, 29.2) 7.7 (4.0, 14.2) 4.7 (2.0, 11.3)
No 91.4 (36.4, 323) 6.3 (2.2, 13.7) 22.5 (10.8, 39.0) 7.5 (4.9, 14.0) 6.9 (2.6, 12.6)
Aftershave
Yes 266* (123.2, 625) 6.1 (2.7, 13.8) 17.9 (10.9, 33.6) 7.9 (4.2, 16.2) 4.8 (2.4, 8.6)
No 135 (54.5, 477) 6.3 (2.3, 18.9) 16.3 (10.2, 30.2) 7.6 (3.9, 14.1) 4.8 (2.1, 12.0)
Hair gel/spray
Yes 182 (57.8, 547) 7.7 (2.6, 21.6) 16.0 (8.9, 24.5) 8.0 (4.2, 14.2) 4.9 (2.0, 11.4)
No 139 (58.2, 464) 5.9 (2.1, 14.6) 16.7 (11.1, 31.7) 7.6 (4.0, 14.2) 4.8 (2.3, 11.7)
a Specific-gravity–adjusted phthalate levels.
*p ≤0.05 in multivariate linear regression models adjusted for age, BMI, race, and smoking.
Table 6 Multiplicative factorsa (95% confidence interval) for a change in urinary phthalate monoester levelb associated with use of personal care products within the past 48 hr (n = 323).
Product type MEP MEHP MBP MBzP MMP
Lotion 0.97 (0.70–1.33) 0.66 (0.44–0.99) 0.69 (0.53–0.88) 0.66 (0.50–0.87) 0.92 (0.66–1.29)
Cologne 2.57 (1.88–3.53) 0.96 (0.63–1.46) 1.06 (0.82–1.38) 1.16 (0.88–1.54) 1.19 (0.84–1.67)
Deodorant 1.24 (0.75–2.05) 1.23 (0.65–2.34) 0.70 (0.47–1.04) 0.95 (0.62–1.47) 0.94 (0.55–1.59)
Aftershave 1.71 (1.10–2.64) 0.93 (0.53–1.64) 1.00 (0.70–1.43) 1.16 (0.80–1.69) 0.97 (0.61–1.55)
Hair gel 1.15 (0.85–1.57) 1.23 (0.83–1.81) 0.92 (0.72–1.17) 0.95 (0.73–1.24) 0.99 (0.72–1.35)
a All models are adjusted for age, BMI, smoking, and race. Multiplicative factors represent multiplicative changes in phthalate levels associated with use of specific personal care products within the past 48 hr after back-transformation of phthalate concentrations: 1.0, no change in urinary phthalate level; < 1.0, multiplicative decrease in phthalate level; > 1.0, multiplicative increase in phthalate level.
b In all models, log transformations of specific gravity–adjusted phthalate concentrations were used.
==== Refs
References
Api AM 2001 Toxicological profile of diethyl phthalate: a vehicle for fragrance and cosmetic ingredients Food Chem Toxicol 39 97 108 11267702
ATSDR 1995. Toxicological Profile for Diethyl Phthalate. Atlanta, GA:Agency for Toxic Substances and Disease Registry. Available: http://www.atsdr.cdc.gov/toxprofiles/tp73.html [accessed 23 May 2003].
ATSDR 2001. Toxicological Profile for Di-n-butyl Phthalate. Atlanta, GA:Agency for Toxic Substances and Disease Registry. Available: http://www.atsdr.cdc.gov/toxprofiles/tp135.html [accessed 23 May 2003].
ATSDR 2003. Toxicological Profile for Di(2-Ethylhexyl) Phthalate. Atlanta, GA:Agency for Toxic Substances and Disease Registry. Available: http://www.atsdr.cdc.gov/toxprofiles/tp9.html [accessed 23 May 2003].
Barber ED Teetsel NM Kolberg KF Guest D 1992 A comparative study of the rates of in vitro percutaneous absorption of eight chemicals using rat and human skin Fundam Appl Toxicol 19 4 493 497 1426706
Blount BC Milgram KE Silva NJ Malek NA Reidy JA Needham LL 2000 Quantitative detection of eight phthalate metabolites in human urine using HPLC-APCI-MS/MS Anal Chem 72 17 4127 4134 10994974
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:Centers for Disease Control and Prevention. Available: http://www.cdc.gov/exposurereport/2nd/phthalates.htm [accessed 1 August 2003].
Deisinger PJ Perry LG Guest D 1998 In vivo percutaneous absorption of DEHP from DEHP-plasticized polyvinyl chloride film in male Fischer 344 rats Food Chem Toxicol 36 6 521 527 9674960
Duty SM Silva MJ Barr DB Brock JW Ryan L Chen Z 2003b Phthalate exposure and human semen parameters Epidemiology 14 3 269 276 12859026
Duty SM Singh NP Silva MJ Barr DB Brock JW Ryan L 2003a The relationship between environmental exposures to phthalates and DNA damage in human sperm using the neutral comet assay Environ Health Perspect 111 1164 1169 12842768
Elsisi AE Carter DE Sipes IG 1989 Dermal absorption of phthalate diesters in rats Fundam Appl Toxicol 12 70 77 2925020
Hagedorn-Leweke U Lippold BC 1995 Absorption of sun-screens and other compounds through human skin in vivo : derivation of a method to predict maximum fluxes Pharm Res 12 9 1354 1360 8570535
Hauser R Meeker J Park S Silva M Calafat A 2004 Temporal Variability of urinary phthalate levels in men of reproductive age Environ Health Perspect 112 1734 1740 15579421
Hoppin JA Brock JW Davis BJ Baird DD 2002 Reproducibility of urinary phthalate metabolites in first morning urine samples Environ Health Perspect 110 515 518 12003755
Houlihan J Brody C Schwan B 2002. Not Too Pretty: Phthalates, Beauty Products and the FDA. Environmental Working Group, Coming Clean, and Health Care Without Harm. Available: http://www.nottoopretty.org/images/NotTooPretty_final.pdf [accessed 3 September 2003].
Howard J Page N Perkins M 2001. Toxicology Tutor II: Toxicokinetics, Absorption: Dermal Route. Bethesda, MD:Division of Specialized Information Services, National Library of Medicine, National Institutes of Health. Available: http://www.sis.nlm.nih.gov/enviro/ToxTutor/Tox2/a24.htm [accessed 7 November 2003].
Koo HJ Lee BM 2004 Estimated exposure to phthalates in cosmetics and risk assessment J Toxicol Environ Health 67 1901 1914
Koo J-W Parham F Kohn MC Masten SA Brock JW Needham LL 2002 The association between biomarker-based exposure estimates for phthalates and demographic factors in a human reference population Environ Health Perspect 110 405 410 11940459
Li LH Jester WF Orth JM 1998 Effects of relatively low levels of mono-(2-ethylhexyl) phthalate on cocultured Sertoli cells and gonocytes from neonatal rats Toxicol Appl Pharmacol 153 2 258 265 9878596
Melnick RL Morrissey RE Tomaszewski KE 1987 Studies by the National Toxicology Program on di(2-ethylhexyl)phthalate Toxicol Ind Health 3 2 99 118 3617073
Mint A Hotchkiss SAM Caldwell J 1994 Percutaneous absorption of diethyl phthalate through rat and human skin in vitro Toxicol In Vitro 8 251 256 20692913
Ng KM ChuI Bronaugh RL Franklin CA Somers DA 1992 Percutaneous absorption and metabolism of pyrene, benzo[a]pyrene, and di(2-ethylhexyl) phthalate: comparison of in vitro and in vivo results in the hairless guinea pig Toxicol Appl Pharmacol 115 2 216 223 1641855
Parks LG Ostby JS Lambright CR Abbott BD Klinefelter GR Barlow NJ 2000 The plasticizer diethylhexyl phthalate induces malformations by decreasing fetal testosterone synthesis during sexual differentiation in the male rat Toxicol Sci 58 2 339 349 11099646
Phillip Morris 2004. Ingredients in Cigarettes. Richmond, VA: Phillip Morris. Available: http://www.philipmorrisusa.com/product_facts/ingredients/non_tobacco_ingredients.asp [accessed 23 November 2000].
Scott RC Dugard PH Ramsey JD Rhodes C 1987 In vitro absorption of some o -phthalate diesters through human and rat skin Environ Health Perspect 74 223 227 3691429
Silva MJ Barr DB Reidy JA Malek NA Hodge CC Caudill SP 2004 Urinary levels of seven phthalate metabolites in the US population from the National Health and Nutrition Examination Survey (NHANES) 1999–2000 Environ Health Perspect 112 331 338 14998749
Silva MJ Malek NA Hodge CC Reidy JA Kato K Barr DB 2003 Improved quantitative detection of 11 urinary phthalate metabolites in humans using liquid chromatography-atmospheric pressure chemical ionization tandem mass spectrometry J Chromatogr B 789 2 393 404
Swan SH Main KM Liu F Stewart SL Kruse RL Calafat AM 2005. Decrease in anogenital distance among male infants with prenatal phthalate exposure. Environ Health Perspect 10.1289/EHP.8100.
Wester RC Melendres J Sedik L Maibach H Riviere JE 1998 Percutaneous absorption of salicylic acid, theophylline, 2,4-dimethylamine, diethyl hexyl phthalic acid, and p -aminobenzoic acid in the isolated perfused porcine skin flap compared to man in vivo Toxicol Appl Pharmacol 151 1 159 165 9705899
|
16263507
|
PMC1310914
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 18; 113(11):1530-1535
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8083
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8094ehp0113-00153616263508ResearchInduction of Proinflammatory Cytokines and C-Reactive Protein in Human Macrophage Cell Line U937 Exposed to Air Pollution Particulates Vogel Christoph Franz Adam 1Sciullo Eric 1Wong Pat 1Kuzmicky Paul 1Kado Norman 12Matsumura Fumio 131 Department of Environmental Toxicology, University of California, Davis, California, USA2 California Environmental Protection Agency, Air Resources Board, Sacramento, California, USA3 Center for Environmental Health Sciences, University of California, Davis, California, USAAddress correspondence to C. Vogel, Department of Environmental Toxicology, University of California, Davis, CA 95616,USA. Telephone: (530) 752-1337. Fax: (530) 752-5300. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 21 7 2005 113 11 1536 1541 7 3 2005 21 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Exposure to particulate matter air pollution causes inflammatory responses and is associated with the progression of atherosclerosis and increased cardiovascular mortality. Macrophages play a key role in atherogenesis by releasing proinflammatory cytokines and forming foam cells in subendothelial lesions. The present study quantified the inflammatory response in a human macrophage cell line (U937) after exposure to an ambient particulate sample from urban dust (UDP) and a diesel exhaust particulate (DEP). The effect of native UDP and DEP was compared with their corresponding organic extracts (OE-UDP/OE-DEP) and stripped particles (sUDP/sDEP) to clarify their respective roles. Exposure to OE-UDP, OE-DEP, UDP, DEP, and 2,3,7,8-tetrachlorodibenzo-p-dioxin led to a greater increase of interleukin (IL)-8, tumor necrosis factor-α, and cyclooxygenase-2 mRNA expression than did the stripped particles, whereas sUDP, sDEP, UDP, and DEP led to a greater production of C-reactive protein and IL-6 mRNA. The particles and the organic extract-induced expression of cyclooxygenase-2 and cytochrome P450 (CYP)1a1 was significantly suppressed by co-treatment with an aryl hydrocarbon receptor (AhR) antagonist, indicating that these effects are mainly mediated by the organic components, which can activate the AhR and CYP1a1. In contrast, the induction of C-reactive protein and IL-6 seems to be a particle-related effect that is AhR independent. The inflammatory response induced by particulate matter was associated with a subsequent increase of cholesterol accumulation, a hallmark of foam cells. Together, these data illustrate the interaction between particulate matter and the inflammatory response as well as the formation of cholesterol-accumulating foam cells, which are early markers of cardiovascular disease.
aryl hydrocarbon receptorcyclooxygenase-2C-reactive proteincytokinesfoam cellsinterleukin-8inflammationmacrophagesparticles2,3,7,8-tetrachlorodibenzo-p-dioxin
==== Body
Numerous epidemiologic studies have observed that exposure to particulate matter (PM) air pollution, which occurs in many urban and industrial environments, is associated with an increase of cardiovascular diseases and mortality (Brook et al. 2003; Morris 2001). Although the exact components of PM and the exact mechanism leading to cardiovascular disease and cardiopulmonary disease mortality from exposure to PM are still unknown, several studies have shown that systemic inflammation may be a key step in these pathological processes through inflammatory mediators (Salvi et al. 1999) such as cyclooxygenase-2 (COX-2), interleukin (IL)-1β, and tumor necrosis factor-α(TNFα), which are among the most important mediators of the inflammatory response (Ross 1999) in the development of atherosclerotic vascular disease (Libby 2002). A positive correlation of C-reactive protein (CRP) and coronary artery disease, which could be explained by the atherogenic effects of chronic inflammation, is well known (Beck et al. 1999; Mendall et al. 1996). Recently, an association between minor but chronic elevation of serum CRP levels and future major cardiovascular events has been shown (Yeh 2004). Elevated levels of proinflammatory cytokines and CRP play a significant role in the genesis of atherosclerosis and in plaque instability (Libby 2002). CRP activates complement through binding to the Fcγ receptor and enhancing phagocytosis of low-density lipoprotein, leading to the formation of foam cells (Zwaka et al. 2001), thus directly contributing to the development of atherosclerosis. Despite the epidemiologic evidence, experimental verification of the causal relationship among air pollution, CRP, and cardiovascular disease is still limited.
In vitro studies show increased levels of proinflammatory cytokines including TNFα, IL-1β, IL-6, and IL-8, which have been described in various cell types after exposure to PM (Mathiesen et al. 2004; Monn et al. 2003; Monn and Becker 1999; van Eeden et al. 2001). Acute exposure to diesel exhaust also increased IL-8 production in human airways (Salvi et al. 2000). However, elevated serum levels of CRP, the classic acute-phase protein, has only been found to be associated with exposure to an elevated concentration of PM in humans (Kim et al. 2005; Peters et al. 2001; Pope et al. 2004a; Seaton et al. 1999). Other harmful effects described by these authors involved the triggering of acute vasoconstriction and the development of atherosclerosis. A few animal models have shown the harmful effects of inhalation of air pollutants on cardiovascular functions (Campen et al. 2003; Gordon et al. 1998), as well as on the etiology of atherosclerosis (Suwa et al. 2002). Suwa et al. (2002) showed that exposure of rabbits to PM10 (PM with areodynamic diameter ≤10 μm) causes progression of atherosclerotic lesions, and a number of alveolar macrophages phagocytosed PM10. Direct effects of PM may occur via components that are able to cross the pulmonary epithelium into the circulation, such as gases, ultrafine particles (Nemmar et al. 2002), and soluble co-pollutants (e.g., polycyclic aromatic hydrocarbons and transition metals).
To clarify the contribution of each component of PM in the induction of the inflammatory response, we systematically compared the effects induced by PM derived from different sources such as diesel exhaust particulates (DEP) and urban dust particulates (UDP) with those induced by their organic extracts OE-DEP/OE-UDP and the fine particles or coarse fraction, represented by their stripped particles sDEP/sUDP and the ultrafine particles carbon black (CB).
The present study provides evidence that the organic components of the native particles DEP and UDP play a major role in mediating the increase of the inflammatory cytokines TNFα, IL-8, and COX-2. We also demonstrate, for the first time, an increased expression of CRP in macrophages induced by the particles that is mediated by the particulate composition rather than their organic components.
Materials and Methods
Reagents.
National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1649, an atmospheric particulate material collected in an urban area, and a diesel exhaust particulate sample, NIST SRM 2975, were purchased from NIST (Gaithersburg, MD). CB 95 nm in diameter (FR103) were provided by Degussa (Frankfurt, Germany). We prepared stock solutions of particles by suspending them in autoclaved distilled water and by ultrasonication for 2 min at maximum power (100 W). Particles were used at 2.5, 10, or 40 μg/cm2, equivalent to 12.5, 50, or 200 μg/mL. Concentrations are preferentially expressed in micrograms per square centimeter because particles rapidly sediment onto the cell layer. UDP and DEP were extracted by dichloromethane in a soxhlet apparatus. After sonication the extract was filtered (0.45 μm Acrodisc) and concentrated to 1 mL by TurboVap and stored in precleaned amber vials. The extract obtained was dried and then redissolved in dimethylsulfoxide. We used the OE-DEP and OE-UDP at concentrations corresponding to the amount of particles at 10 μg/cm2. 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD, > 99% purity) was originally obtained from Dow Chemicals Co. (Midland, MI). Dimethylsulfoxide and phorbol-12-myristate-13-acetate (TPA) were obtained from Aldrich Chemical Co. (St. Louis, MO). Other molecular biological reagents were purchased from Qiagen (Valencia, CA) and Roche (Indianapolis, IN).
Cell culture and differentiation.
We obtained human U937 monocytic cells from the American Tissue Culture Collection (Manassas, VA) and maintained them in RPMI 1640 medium containing 10% fetal bovine serum (Gemini, Woodland, CA), 100 U penicillin, and 100 μg/mL streptomycin supplemented with 4.5 g/L glucose, 1 mM sodium pyruvate, and 10 mM HEPES. Cell culture was maintained at a cell concentration between 2 × 105 and 2 × 106 cells/mL. For differentiation into macrophages, U937 cells were treated with TPA (5 μg/mL) and allowed to adhere for 48 hr in a 5% CO2 tissue culture incubator at 37°C, after which they were fed with TPA-free medium.
Cell viability assay.
To assess the effect of particles on viability of U937 macrophages, we used the trypan blue exclusion test (McAteer and Davis 1994). A 10-μL portion of re-suspended cell pellet was placed in 190 μL PBS with 200 μL trypan blue (0.5% dilution in 0.85% NaCl) added. After 5 min we loaded 10 μL of the cell suspension into a hemocytometer and determined the proportion of nonviable to viable cells.
Cellular cholesterol and protein determinations.
We extracted free and esterified cholesterol (total cholesterol) directly from macrophage monolayers in situ in the cell culture dish. After the indicated time of treatment, each PBS-washed monolayer was scraped off in 400 μL RI PA buffer and incubated for 30 min on ice. Unsoluble material was removed by centrifugation at 12,000 × g for 20 min at 4°C and aliquots were used for protein determination according to Bradford (1976). We determined the amount of free and esterified cholesterol (total cholesterol) using a colorimetric method (Roche) in the presence of cholesterol oxidase and cholesterol esterase and then measured the absorbance at 405 nm.
Quantitative real-time reverse transcription–PCR.
We isolated total RNA from U937 cells using a high-pure RNA isolation kit (Roche) and carried out cDNA synthesis as previously described (Vogel et al. 2004b). Quantitative detection of β-actin and differentially expressed genes was performed with a LightCycler Instrument (Roche Diagnostics, Mannheim, Germany) using the QuantiTect SYBR Green PCR Kit (Qiagen) according to the manufacturer’s instructions. DNA-free total RNA (1.0 μg) was reverse-transcribed using 4 U Omniscript reverse transcriptase (RT; Qiagen) and 1 μg oligo(dT)15 in a final volume of 40 μL. The primers for each gene (Table 1) were designed on the basis of the respective cDNA or mRNA sequences using OLIGO primer analysis software provided by Steve Rozen and Whitehead Institute/MIT Center for Genome Research (Rosen and Skaletsky 2000), so that the targets were 100–200 bp in length. PCR amplification was carried out in a total volume of 20 μL, containing 2 μL cDNA, 10 μL 2 × QuantiTect SYBR Green PCR Master Mix, and 0.2 μM of each primer. The PCR cycling conditions were 95°C for 15 min followed by 40 cycles of 94°C for 15 sec, 60°C for 20 sec, and 72°C for 10 sec. We performed detection of the fluorescent product at the end of the 72°C extension period. We ran negative controls concomitantly to confirm that the samples were not cross-contaminated. A sample with DNase- and RNase-free water instead of RNA was concomitantly examined for each of the reaction units described above. To confirm the amplification specificity, we subjected the PCR products to melting curve analysis. We performed all PCR assays in triplicate. The intra-assay variability was < 7%. For quantification we analyzed data with the LightCycler analysis software. The variables were examined for one-sided Student’s t test. The results are given as the mean ± the SDs of the mean.
Antibodies and Western blotting.
Monoclonal anti-human CRP antibody was purchased from Sigma Chemical Co. Rabbit polyclonal anti-human actin and a horseradish peroxidase-conjugated secondary antibody were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). We separated whole-cell proteins on a 10% SDS–polyacrylamide gel and blotted them onto a PVDF (polyvinylidene fluoride) membrane (Immuno-Blot; BioRad, Hercules, CA). The antigen-antibody complexes were visualized using the chemoluminescence substrate SuperSignal West Pico (Pierce, Rockford, IL) as recommended by the manufacturer. For quantitative analysis, we quantified respective bands using a ChemiImager 4400 (Alpha Innotech Corp., San Leandro, CA).
Statistical analysis.
All experiments were repeated a minimum of 3 times and results were expressed as mean ± SD. We determined statistical differences using Student’s t test and for the analysis of the significance between pairs of mean values, we used the Bonferroni post-hoc test.
Results
Dose-dependent effect of DEP and UDP on inflammatory factors and CYP1a1 expression.
To address the dose-dependent effect of the particles, we studied the mRNA expression of inflammatory factors and CYP1a1 24 hr after treatment with various concentrations of DEP and UDP. As shown in Table 2, treatment of U937 macrophages with the native particles of DEP or UDP in the range of 2.5, 10, or 40 μg/cm2 cell culture area led to dose-dependent mRNA induction of COX-2, TNFα, IL-6, IL-8, CRP, C/EBPβ (CCAAT/enhancer binding protein), and CYP1a1.
Except for COX-2, the UDP-induced expression on these genes was more prominent than the effect of DEP. All parameters were significantly induced by UDP at the low concentration of 2.5 μg/cm2. In contrast, COX-2, IL-6, C/EBPβ, and CRP were significantly increased only at 10 or 40 μg/cm2 DEP. The most conspicuous effect of UDP was found in the case of IL-8 (31.6-fold), whereas DEP showed the strongest effect on CRP expression (19.7-fold). To estimate the toxic potency, we compared the effects of DEP and UDP with various concentrations of TCDD, which has been shown to be an efficient inducer of inflammatory factors and foam-cell formation in U937 macrophages. Except for IL-6, which was downregulated by TCDD, we observed concentration-dependent increases of COX-2, TNFα, and IL-8, showing a 5.1-, 2.8-, and 2.5-fold increase, respectively, at the lowest concentration of TCDD tested (0.1 nM). CRP was also significantly increased at higher concentrations of TCDD (1 and 10 nM), which correlated with the induction of C/EBPβ.
Evaluation of cytotoxicity of DEP and UDP.
Cytotoxic effects on U937 macrophages were measured for DEP, UDP, sDEP, sUDP, as well as for their corresponding extracts OE-DEP and OE-UDP. The viability of cells cultured in medium alone with 100 μL PBS added served as control. After incubating the cells with particles or extracts for 24 hr at 37°C, we determined cytotoxicity by trypan blue exclusion test. Cell death in the unexposed U937 macrophages was 8% (Figure 1). In U937 macrophages treated with 2.5 or 10 μg/cm2 DEP, UDP, sDEP, sUDP, OE-DEP, or OE-UDP, no significant effect on cell viability compared to control cells was found (data not shown). However, at the highest dose of 40 μg/cm2, treatment with DEP as well as with UDP led to a significant increase of the number of dead cells by 7 and 10%, respectively (Figure 1).
Effect of organic compounds in DEP- and UDP-induced expression of proinflammatory marker genes COX-2, TNFα, and IL-8.
Macrophages involved in atherosclerotic lesions are a primary source of inflammatory cytokines. Cytokines can contribute to initiation and progression of atherosclerotic lesions by triggering multiple cellular functions such as leukocyte recruitment and synthesis/degradation of extracellular matrix. Therefore, we tested the expression of selected biomarkers of the proinflammatory response after treatment with DEP and UDP. To determine which component of DEP and UDP is the most critical one for this response, the DEP-and UDP-induced mRNA expressions were compared with those induced by sDEP, sUDP, and their organic components OE-DEP and OE-UDP. Both types of particles were used at a concentration of 10 μg/cm2, and the final concentration of the organic extracts was equivalent to 10 μg/cm2 of the particles. Treatment for 24 hr with DEP and UDP significantly induced a 7.5- and 10-fold increase of COX-2, a 4- and 7-fold increase of TNFα, and 8- and 25-fold increase of IL-8, respectively (Figure 2). The increases of COX-2, TNFα, and IL-8 mRNA levels by the organic compounds from the diesel particulates (OE-DEP) were more than 2-fold higher than the effect of the native particle DEP (Figure 2A), whereas the organic compounds of the urban dust particulates (OE-UDP) led to increases of these cytokines comparable to the native particle UDP (Figure 2B). For the stripped particles, sDEP, the effects on COX-2, TNFα, and IL-8 mRNA expression remained 3-fold lower than that induced by DEP, and about 8-fold lower than that induced by their organic compounds, suggesting the role of organic compounds in these responses (Figure 2A). The increase of COX-2, IL-8, and TNFαmRNA after treatment with sUDP was about 2-fold less pronounced than the effects of UDP or the organic extract OE-UDP (Figure 2B). Treatment with CB (10 μg/cm2) had no significant effect on the mRNA expression of COX-2, TNFα, or IL-8 (Figure 2A).
Role of the particles on DEP- and UDP-induced expression of CRP and IL-6.
CRP participates in the systemic response to inflammation and is an important cardiovascular risk factor. Elevated concentrations of PM have been shown to be associated with increases of serum CRP levels in men (Peters et al. 2001; Pope et al. 2004b). In the present study we observed a 6- and 13-fold increase of CRP mRNA expression after exposure to 10 μg/cm2 DEP or 10 μg/cm2 UDP for 24 hr, respectively (Figure 3). Treatment with DEP or sDEP resulted in similar induction rates of CRP mRNA (6-fold), but OE-DEP had a significantly lesser effect (2.5-fold) on CRP mRNA expression than DEP or sDEP (Figure 3A). sUDP led to significantly higher induction (22-fold) of CRP mRNA than did UDP or the organic compounds OE-UDP (3-fold) (Figure 3B).
To test whether elevated mRNA levels correspond with increased accumulation of the CRP protein, we performed Western blots. In whole-cell protein of U937 macrophages, we observed 3- to 4-fold higher protein levels of CRP in DEP-, UDP-, sDEP-, and sUDP-treated cells compared to control, OE-DEP–, or OE-UDP–treated cells (Figure 4).
IL-6 has been shown to be elevated after exposure to particles in macrophages in numerous studies (e.g., Monn and Becker 1999) and is the most potent inflammatory cytokine for the induction of human CRP. We found that after 24 hr, DEP, UDP, and their corresponding stripped particles significantly increased the mRNA level of IL-6 by 5- and 7-fold, respectively (Figure 3). However, similar to the results found for CRP, the organic compounds OE-DEP or OE-UDP did not significantly induce mRNA expression of IL-6 (Figure 3). Treatment with CB (10 μg/cm2) had no significant effect on the mRNA expression of IL-6. Only the CRP mRNA level was slightly elevated after exposure to CB; however, the effect was not statistically significant (Figure 3A).
The fact that treatment with TCDD, OE-DEP, or OE-UDP did not increase IL-6 and led to only a moderate increase of CRP mRNA expression compared to sDEP, sUDP, DEP, or UDP suggests that particles rather than the co-pollutants mediate the increase of CRP and IL-6.
Effect of various inhibitors on DEP- and UDP-mediated CRP and COX-2 induction.
CRP induction by DEP, UDP, sDEP, or sUDP was blocked by about 75% after pretreatment for 15 min with 100 μg/mL aggregated IgG, which blocks binding to the Fcγ receptor. Preincubation for 15 min with 100 nM wortmannin, which inhibits Fcγ receptor–dependent ingestion and activation, blocked the induction of CRP mediated by DEP, UDP, sDEP, and sUDP by about 50% (Figure 5). Neither aggregated IgG nor wortmannin led to a significant inhibition of DEP- or UDP-mediated COX-2 induction (Figure 6). Conversely, the aryl hydrocarbon hydroxylase (AhR) antagonist luteolin had no significant effect on the particle-induced expression of CRP (Figure 5). However, the induction of COX-2 mediated by DEP/UDP, as well as their organic extracts, was significantly suppressed (50%) by luteolin (Figure 6).
DEP- and UDP-induced CYP1a1 mRNA level.
The CYP1a1 mRNA level increased 40-fold in DEP-treated (10 μg/cm2) cells, whereas the organic extract OE-DEP led to a markedly higher increase of 170-fold in comparison to control cells. Exposure to 10 μg/cm2 UDP- or OE-UDP led to about 100-fold elevated levels of CYP1a1 mRNA after 24 hr of treatment. The stripped particles sUDP had still significant effects and increased CYP1a1 mRNA levels by 20-fold (Figure 7).
To test the role of the AhR in OE-DEP–and OE-UDP–mediated increase of CYP1a1, we co-treated cells with the AhR antagonist luteolin (10 μM) and OE-DEP or OE-UDP for 24 hr. Co-treatment with luteolin (10 μM) inhibited OE-DEP– and OE-UDP–mediated CYP1a1 induction by about 50%, which indicates the involvement of the AhR in this process. As a positive control, cells were treated with 10 nM TCDD for 24 hr and co-treated with luteolin. TCDD led to a 110-fold increase of CYP1a1 mRNA level, which was inhibited by 95% after co-treatment with luteolin (Figure 7).
Stimulation of cholesterol accumulation in U937 macrophages by UDP and DEP.
Foam cells are primarily macrophages laden with cholesterol ester-rich cytoplasmic lipid inclusions. To quantify total amount of cholesterol in U937 macrophages, we used a colorimetric method in the presence of cholesterol oxidase and cholesterol esterase. Exposure for 5 days to 10 μg/cm2 DEP as well as to UDP stimulated the accumulation of cholesterol by 2-and 2.5-fold, respectively (Figure 8). Both organic extracts OE-DEP or OE-UDP increased the amount of cholesterol by about 2.3-fold above control. The stripped particles sDEP and sUDP did not significantly increase the amount of cholesterol in U937 macrophages compared to control cells (Figure 8). Results from cholesterol assay were consistent with findings from Oil Red O staining (data not shown).
Discussion
There is strong evidence from epidemiologic and animal studies that exposure to air pollution particulates play a role in the development of cardiovascular diseases such as atherosclerosis and heart diseases (Pope et al. 2004; Suwa et al. 2002). DEP and UDP, which are the most important components of PM2.5 (PM with aerodynamic diameter ≤2.5 μm) and PM10, respectively, in many urban areas, have been suspected. The results presented in this study show that diesel particles as well as urban dust cause the induction of several proinflammatory factors such as COX-2, TNFα, C/EBPβ, IL-6, and IL-8 in human macrophages. Exposure to these particles also results in a significant elevation of CRP mRNA and protein levels in U937 macrophages. We used U937-derived macrophages because these cells are frequently used to develop foam cells after treatment with modified low density protein (Martens et al. 1998) and as described earlier by exposure to environmental toxicants like TCDD (Vogel et al. 2004a). Concomitant with induction of inflammatory factors, the accumulation of total cholesterol was significantly increased in DEP-or UDP-treated macrophages. Cholesterol accumulation in macrophages is a hallmark of foam-cell formation indicating early lesion of atherosclerosis (Linton and Fazio 2003). Thus, the particle-mediated inflammatory response and subsequent formation of foam cells may contribute directly to the progression of atherosclerosis and other cardiovascular events.
Several studies have shown that the toxicity of PM might be linked to the generation of reactive oxygen species (ROS) in the lungs (Tao et al. 2003), which can be detected by their electron spin resonance signals (Kadiiska et al. 1997). ROS might also play a role in promoting a state of systemic inflammation. In the current study we performed experiments to estimate the signaling mechanisms of the proinflammatory response induced by DEP or UDP using human monocyte-derived macrophages. Organic compounds like polyaromatic hydrocarbons (PAH) adsorbed on UDP, and especially DEP, which induce CYP1a1 gene expression (Figure 7) seemed to be mainly involved in the response of inflammatory factors such as COX-2, IL-8, and TNFα. Besides COX-2 and TNFα, we observed, for the first time, a dose-dependent increase of IL-8 mRNA level in cells treated with the AhR ligand TCDD. Results of this study show that the organic extract of OE-DEP is significantly more effective at inducing COX-2, IL-8, and TNFα than its native particle DEP, whereas OE-UDP led to a similar increase compared to its native particle UDP. These results suggest that the organic co-pollutants are highly adsorbed by DEP and thus less bioavailable compared to UDP. The stripped particles of both diesel and urban dust had significantly less effect on the induction of COX-2, IL-8, and TNFα.
To analyze the contribution of AhR-activating compounds, we co-treated cells with the AhR antagonist luteolin and the particles or their organic extracts. Our results (Figures 5–7) clearly show that luteolin is more effective in suppressing COX-2 or CYP1a1 than CRP in all samples. In turn, both IgG and wortmannin were much better inhibitors of CRP than COX-2. These results indicate that the solvent extraction procedure could effectively separate the AhR agonists (i.e., luteolin inhibited components) from the particles. Furthermore, the stripped particles showed properties different from the solvent extracts in that affected cells showed high mRNA expression of CRP by treatment with stripped or native particles but not with organic extracts (Figure 3).
We have previously shown that dioxin-type chemicals are powerful inducers of inflammation in U937 macrophages (Vogel et al. 2004a), and therefore it is logical to explain the action of solvent extracts to activate COX-2. However, the finding that stripped particles from air pollutants selectively activate CRP is new. It is known that insoluble, fine particles have the property to affect Fcγ receptor activity on the macrophage membrane and thereby trigger the process of phagocytosis and uptake of low-density lipoprotein (Kleinman et al. 2003; Luo et al. 2005; Ohtsuka etal. 2000; Swanson and Hoppe 2004; Zwaka etal. 2001). The notion that the stripped particles are also acting through the Fcγ receptor is supported by our observation that aggregated IgG, the specific ligand for Fcγ receptor, is effective in suppressing the same cell response indicates that the phagocytosis of those particles is the key event accompanying the Fcγ receptor stimulating action of sDEP and sDEP, since phosphatidylinositol 3-kinase, which is sensitive to wortmannin, is a crucial factor mediating phagocytosis of macrophages (Song et al. 2004). Activation of Fcγ receptor by aggregated IgG might also trigger inhibitory signaling pathways that suppress the effects of particles on the expression of CRP. The effect of the ultrafine CB on CRP expression was rather small, which indicates that not only the chemical components of the PM but also other factors such as surface properties and shape affect toxicity.
One interesting aspect is that the timing of particle-mediated induction of CRP was correlated with elevated levels of IL-6 mRNA, which can mediate transcriptional activation of CRP. Both CRP and IL-6 induction by PM were blocked by aggregated IgG or wortmannin, which inhibits Fcγ receptor–dependent ingestion and activation. Neither aggregated IgG nor wortmannin led to a significant inhibition of DEP-or UDP-induced COX-2, TNFα, or IL-8. The close relationship between IL-6 and CRP has been pointed out by many scientists (e.g., Monton and Torres 1998), but in most cases CRP production is carried out in liver as a result of stimulation by circulating IL-6. The fact that IL-6 acts as an autocrine factor to stimulate CRP production in macrophages is not well known; however, the increased synthesis and secretion of CRP, IL-6, and soluble IL-6 receptor by macrophage-derived foam cells in the arterial intima has been demonstrated (Ballou and Lozanski 1992; Jones et al. 1999; Libby 2002; Lusis 2000; Ross 1999). Recent studies also show that binding of C/EBPβ is critical for induction of CRP expression (Agrawal et al. 2003), which could explain the increase of CRP by TCDD, as TCDD induces the expression of C/EBPb (Vogel et al. 2004a; Vogel and Matsumura 2003) but not IL-6 (Table 2). According to Du Clos and Mold (2004), CRP acts through Fcγ receptor to play important roles in infection, inflammation, and autoimmune diseases. This phenomenon of action of stripped particles deserves close attention in the future.
In conclusion, we have shown that air pollution particles have two major classes of toxic components. One is the dioxin-type AhR agonist, which is extractable by solvents, and the other type is the stripped particle, which elicits a different pattern of mRNA activation from that induced by dioxin-type chemicals. These findings may contribute to a better understanding of the differential toxicity of various constituents and sources of PM, including their chemical/biological components alone or in combination with PM. Further research is necessary to fully elucidate this mechanism of differential toxicity. Other relevant sources of PM should be collected and tested, such as those from the combustion of alternative fuels, and evaluated for their potential contribution as risk factors for cardiovascular disease in both urban and rural environments.
This study was supported by grants ESO5233 and ESO05707 from the National Institute of Environmental Health Sciences and by grant IRG 95-125-07 from the American Cancer Society.
Figure 1 Cytotoxicity of U937 macrophages exposed to DEP or UDP. Percent cytotoxicity was assessed by trypan blue exclusion test. Cells were exposed to DEP or UDP at various concentrations (2.5, 10, or 40 μg/cm2). Control cells (Ctrl) received 100 μL PBS only. Error bars represent mean ± SD of three independent experiments.
*Significantly different from control cells (p < 0.05)
Figure 2 Effect of DEP (A), UDP (B), their corresponding stripped particles, or their organic extracts preparation on TNFα, COX-2, and IL-8 mRNA expression in U937 macrophages. Increased mRNA levels of TNFα, COX-2, and IL-8 are shown. (A) U937 macrophages were treated for 24 hr with 10 μg/cm2 DEP. To examine the effect of the stripped particles and the organic components of these particle samples, cells were treated with equivalent amounts of the corresponding sDEP or OE-DEP. As a control, cells were treated with 10 μg/cm2 CB. (B) U937 macrophages were treated for 24 hr with 10 μg/cm2 UDP. To examine the effect of the stripped particles and the organic components of these particle samples, cells were treated with equivalent amounts of the corresponding sUDP or OE-UDP. Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly increased compared to control cells (p < 0.05). **Significantly lower than in native particle-treated cells (p < 0.05). #Significantly increased compared to native particles-treated cells (p < 0.05)
Figure 3 Effect of DEP (A), UDP (B), their corresponding stripped particles, or their organic extracts preparation on IL-6 and CRP mRNA expression inflammatory mediators in U937 macrophages. Increased mRNA levels of IL-6 and CRP are shown. (A) U937 macrophages were treated for 24 hr with 10 μg/cm2 DEP. To examine the effect of the stripped particles and the organic components of these particle samples, we treated cells with equivalent amounts of the corresponding sDEP or OE-DEP. As a control, cells were treated with 10 μg/cm2 CB. (B) U937 macrophages were treated for 24 hr with 10 μg/cm2 UDP. To examine the effect of the stripped particles and the organic components of these particle samples, cells were treated with equivalent amounts of the corresponding sUDP or OE-UDP. Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly increased compared to control cells (p < 0.05). **Significantly lower than in native particle-treated cells (p < 0.05). #Significantly increased compared to native particle-treated cells (p < 0.05)
Figure 4 Increased intracellular protein level of CCRP. (A) The levels of CRP in whole-cell lysates from U937 macrophages 48 hr after treatment with 10 μg/cm2 DEP, UDP, sDEP, sUDP, OE-DEP, OE-UDP, or 1% PBS as vehicle control (C) were determined by Western blot analysis. Equivalent amounts of whole-cell lysates (100 μg protein) were loaded in each lane on 10% SDS-polyacrylamide gels and analyzed by immunoblotting using a CRP- or actin-(housekeeping protein as control) specific antibody. (B) Densitometric evaluation of CRP protein band intensities normalized to actin protein band intensities. Mean values of two independent experiments are shown.
Figure 5 Effect of various inhibitors on DEP- or UDP-induced mRNA expression of CRP. U937 macrophages were pretreated for 15 min with 100 μg/mL aggregated human IgG, 10 μM Lut, or 0.1 μM wortmannin (Wort). Cells were then treated with 10 μg/cm2 DEP, UDP, or sDEP, sUDP for 24 hr. The mRNA expression was analyzed by real-time reverse transcription-PCR and results were normalized to β-actin and given as fold increase compared to the mRNA level in control cell (= 1). Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly increased compared to control cells (p < 0.05). **Significantly lower than in cells treated with native particles, stripped particles, or their organic extracts (p < 0.05)
Figure 6 Effect of various inhibitors on DEP- or UDP-induced mRNA expression of COX-2. U937 macrophages were pretreated for 15 min with 100 μg/mL aggregated human IgG, 10 μM luteolin (Lut), or 0.1 μM wortmannin (Wort). Cells were then treated with 10 μg/cm2 DEP, UDP or OE-DEP, OE-UDP for 24 hr. The mRNA expression was analyzed by real-time RT-PCR and results were normalized to β-actin and given as fold increase compared to the mRNA level in control cell (= 1). Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly increased compared to control cells (p < 0.05). **Significantly lower than in cells treated with native particles, stripped particles, or their organic extracts (p < 0.05)
Figure 7 Effect of the AhR-antagonist luteolin on DEP-, UDP-, or TCDD-induced CYP1a1 expression. U937 macrophages were treated with 10 μg/cm2 DEP, UDP, sDEP, sUDP, OE-DEP, OE-UDP, or 10 nM TCDD. To antagonize AhR binding and activation, cells were co-treated with 10 μM luteolin (Lut) and OE-DEP, OE-UDP, or TCDD. After 24 hr of treatment, CYP1a1 mRNA expression was analyzed by real-time RT-PCR and results were normalized to β-actin and given as fold increase compared to the mRNA level in control cell (= 1). Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly increased compared to control cells (p < 0.05). **Significantly lower than in cells treated with native particles, their organic extracts, or TCDD (p < 0.05).
Figure 8 Cholesterol accumulated in U937 macrophages. Cells were treated for 5 days with 10 μg/cm2 DEP, UDP, sDEP, sUDP, or corresponding amounts OE-DEP, OE-UDP. Vehicle control cells received 1% PBS. Total cholesterol was determined using a colorimetric method in the presence of cholesterol oxidase and cholesterol esterase. Values are given as mean ± SD of triplicates of three independent experiments.
*Significantly different from vehicle control (p < 0.05)
Table 1 Primer for quantitative real-time PCR analyses.
Gene Forward primer (5′–3′) Reverse primer (5′–3′)
β-actin GGACTTCGAGCAAGAGATGG AGCACTGTGTTGGCGTACAG
C/EBPβ GACAAGCACAGCGACGAGTA AGCTGCTCCACCTTCTTCTG
COX-2 TGAAACCCACTCCAAACACA GAGAAGGCTTCCCAGCTTTT
CRP ATACACTGTGGGGGCAGAAG CCGCCAAGATAGATGGTGTT
CYP1a1 TAGACACTGATCTGGCTGCAG GGGAAGGCTCCATCAGCATC
IL-6 GAACTCCTTCTCCACAAGCG TTTTCTGCCAGTGCCTCTTT
IL-8 CTGCGCCAACACAGAAATTA ATTGCATCTGGCAACCCTAC
TNFα CAGAGGGAAGAGTTCCCCAG CCTTGGTCTGGTAGGAGACG
Table 2 Dose-dependent effect of DEP and UDP on COX-2, TNFα, IL-6, IL-8, C/EBPβ, CRP, and CYP1a1 mRNA expression compared to the dose-dependent effect of TCDD.
DEP (μg/cm2)
UDP (μg/cm2)
TCDD (nM)
Gene 0.1 1.0 10.0 2.5 10 40 2.5 10 40
COX-2 1.6 ± 0.5 (ns) 5.9 ± 1.1 19.5 ± 2.4 3.4 ± 0.4 6.8 ± 1.2 12.2 ± 2.1 5.1 ± 1.0 22.8 ± 2.1 43.4 ± 5.1
TNFα 1.8 ± 0.2 3.8 ± 0.6 8.5 ± 1.2 2.2 ± 0.2 6.9 ± 1.6 16.1 ± 1.1 2.8 ± 0.3 6.8 ± 1.1 7.8 ± 1.5
IL-6 1.3 ± 0.4 (ns) 3.5 ± 0.9 4.2 ± 0.9 2.0 ± 0.3 4.8 ± 1.2 6.5 ± 0.8 1.1 ± 0.4 (ns) 0.7 ± 0.5 (ns) 0.5 ± 0.2*
IL-8 2.3 ± 0.5 7.5 ± 1.2 11.6 ± 0.7 2.8 ± 0.4 13.5 ± 2.1 31.6 ± 3.1 2.5 ± 0.2 14.5 ± 1.4 19.7 ± 2.2
C/EBPβ 1.5 ± 0.4 (ns) 2.2 ± 0.6 3.7 ± 0.8 1.8 ± 0.2 2.5 ± 0.3 4.8 ± 1.0 1.6 ± 1.1 (ns) 2.1 ± 0.3 3.4 ± 0.6
CRP 1.4 ± 1.0 (ns) 6.1 ± 1.1 19.7 ± 3.1 2.1 ± 0.4 13.4 ± 1.1 21.8 ± 2.1 1.2 ± 0.7 (ns) 2.3 ± 0.4 3.8 ± 0.5
CYP1a1 8.6 ± 0.8 37.9 ± 2.1 68.5 ± 4.0 28.5 ± 3.1 110.5 ± 11.2 137.0 ± 9.8 18.8 ± 2.1 120.5 ± 11.0 250.8 ± 22.5
ns, not significant. U937 macrophages were treated with 2.5, 10, or 40 μg/cm2 DEP or UDP. As a positive control, cells were treated with 0.1, 1.0, or 10 nM TCDD for 24 hr and mRNA was analyzed by real-time RT-PCR. Control cells received 1% PBS or 0.1% dimethylsulfoxide. Results are normalized to β-actin and given as fold increase of the mRNA levels in treated cells versus controls (= 1). Values are given as mean ± SD of triplicates of three independent experiments. All values significantly increased compared to control cells (p < 0.05) unless otherwise noted.
* Significantly lower than in control cells (p < 0.05)
==== Refs
References
Agrawal A Cha-Molstad H Samols D Kushner I 2003 Overexpressed nuclear factor-kappaB can participate in endogenous C-reactive protein induction, and enhances the effects of C/EBPbeta and signal transducer and activator of transcription-3 Immunology 108 539 547 12667216
Ballou SP Lozanski G 1992 Induction of inflammatory cytokine release from cultured human monocytes by C-reactive protein Cytokine 4 361 368 1420997
Beck MO Silveiro SP Friedman R Clausell N Gross JL 1999 Asymptomatic coronary artery disease is associated with cardiac autonomic neuropathy and diabetic nephropathy in type 2 diabetic patients Diabetes Care 22 1745 1747 10526747
Bradford MM 1976 A rapid and sensitive method for the quantitation of microgram quantities of protein utiilizing the principle of protein-dye binding Anal Biochem 72 248 254 942051
Brook RD Brook JR Rajagopalan S 2003 Air pollution: the “heart” of the problem Curr Hypertens Res 5 32 39
Campen MJ McDonald JD Gigliotti AP Seilkop SK Reed MD Benson JM 2003 Cardiovascular effects of inhaled diesel exhaust in spontaneously hypertensive rats Cardiovasc Toxicol 3 353 361 14734832
Du Clos TW Mold C 2004 C-reactive protein: an activator of innate immunity and a modulator of adaptive immunity Immunol Res 30 261 278 15531769
Gordon T Nadziejko C Schlesinger R Chen LC 1998 Pulmonary and cardiovascular effects of acute exposure to concentrated ambient particulate matter in rats Toxicol Lett 96–97 285 288
Jones SA Novick D Horiuchi S Yamamoto N Szalai AJ Fuller GM 1999 C-reactive protein: a physiological activator of interleukin 6 receptor shedding J Exp Med 189 599 604 9927522
Kadiiska MB Mason RP Dreher KL Costa DL Ghio AJ 1997 In vivo evidence of free radical formation in the rat lung after exposure to an emission source air pollution particle Chem Res Toxicol 10 1104 1108 9348432
Kim JY Chen JC Boyce PD Christiani D 2005 Exposure to welding fumes is associated with acute systemic inflammatory responses Occup Environ Med 62 157 163 15723880
Kleinman MT Hyde DM Bufalino C Basbaum C Bhalla DK Mautz WJ 2003 Toxicity of chemical components of fine particles inhaled by aged rats: effects of concentration J Air Waste Manage Assoc 53 1080 1087
Libby P 2002 Inflammation in atherosclerosis Nature 420 868 874 12490960
Linton MF Fazio S 2003 Macrophages, inflammation, and atherosclerosis Int J Obes Relat Metab Disord 27 suppl 3 S35 S40 14704742
Luo Y Tucker SC Casadevall A 2005 Fc- and complement-receptor activation stimulates cell cycle progression of macrophage cells from g1 to s J Immunol 174 7226 7233 15905568
Lusis AJ 2000 Atherosclerosis Nature 407 233 241 11001066
Martens JS Reiner NE Herrera-Velit P Steinbrecher UP 1998 Phosphatidylinositol 3-kinase is involved in the induction of macrophage growth by oxidized low density lipoprotein J Biol Chem 273 4915 4920 9478935
Mathiesen M Pedersen EK Bjorseth O Egeberg KW Syversen T 2004 Heating of indoor dust causes reduction in its ability to stimulate release of IL-8 and TNFα in vitro compared to non-heated dust Indoor Air 14 226 234 15217476
McAteer Davis JM 1994. Basic cell culture: a practical approach, basic cell culture and the maintenance of cell lines. In: Basic Cell Culture: A Practical Approach (Davis JM, ed). New York:Oxford University Press.
Mendall MA Patel P Ballam L Strachan D Northfield TC 1996 C reactive protein and its relation to cardiovascular risk factors: a population based cross sectional study BMJ 312 1061 1065 8616412
Monn C Becker S 1999 Cytotoxicity and induction of proinflammatory cytokines from human monocytes exposed to fine (PM2.5 ) and coarse particles (PM10–2.5 ) in outdoor and indoor air Toxicol Appl Pharmacol 155 245 252 10079210
Monn C Naef R Koller T 2003 Reactions of macrophages exposed to particles <10 μm Environ Res 91 35 44 12550086
Monton C Torres A 1998 Lung inflammatory response in pneumonia Monaldi Arch Chest Dis 53 56 63 9632909
Morris RD 2001 Airborne particulates and hospital admission for cardiovascular disease: a quantitative review of the evidence Environ Health Perspect 109 495 500 11544153
Nemmar A Hoet PH Vanquickenborne B Dinsdale D Thomeer M Hoylaerts MF 2002 Passage of inhaled particles into the blood circulation in humans Circulation 105 411 414 11815420
Ohtsuka Y Clarke RW Mitzner W Brunson K Jakab GJ Kleeberger SR 2000 Interstrain variation in murine susceptibility to inhaled acid-coated particles Am J Physiol Lung Cell Mol Physiol 278 469 476
Peters A Frohlich M Doring A Immervoll T Wichmann HE Hutchinson WL 2001 Particulate air pollution is associated with an acute phase response in men; results from the MONICA-Augsburg Study Eur Heart J 22 1198 1204 11440492
Pope CA III Burnett RT Thurston GD Thun MJ Calle EE Krewski D 2004a Cardiovascular mortality and long-term exposure to particulate air pollution: epidemiological evidence of general pathophysiological pathways of disease Circulation 109 71 77 14676145
Pope CA III Hansen ML Long RW Nielsen KR Eatough NL Wilson WE 2004b Ambient particulate air pollution, heart rate variability, and blood markers of inflammation in a panel of elderly subjects Environ Health Perspect 112 339 344 14998750
Rosen S Skaletsky HJ 2000. Primer3 on the WWW for general users and for biologist programmers. In: Bioinformatics Methods and Protocols: Methods in Molecular Biology (Krawetz S, Misener S, eds). Totowa, NJ:Humana Press, 365–386.
Ross R 1999 Atherosclerosis-an inflammatory disease N Engl J Med 340 115 126 9887164
Salvi S Blomberg A Rudell B Kelly F Sandstrom T Holgate ST 1999 Acute inflammatory responses in the airways and peripheral blood after short-term exposure to diesel exhaust in healthy human volunteers Am J Respir Crit Care Med 159 702 709 10051240
Salvi SS Nordenhall C Blomberg A Rudell B Pourazar J Kelly FJ 2000 Acute exposure to diesel exhaust increases IL-8 and GRO-alpha production in healthy human airways Am J Respir Crit Care Med 161 550 557 10673199
Seaton A Soutar A Crawford V Elton R McNerlan S Cherrie J 1999 Particulate air pollution and the blood Thorax 54 1027 1032 10525563
Song X Tanaka S Cox D Lee SC 2004 Fcgamma receptor signaling in primary human microglia: differential roles of PI-3K and Ras/ERK MAPK pathways in phagocytosis and chemokine induction J Leukoc Biol 75 1147 1155 14982949
Suwa T Hogg JC Quinlan KB Ohgami A Vincent R van Eeden SF 2002 Particulate air pollution induces progression of atherosclerosis J Am Coll Cardiol 39 935 942 11897432
Swanson JA Hoppe AD 2004 The coordination of signaling during Fc receptor-mediated phagocytosis J Leukoc Biol 76 1093 1103 15466916
Tao F Gonzalez-Flecha B Kobzik L 2003 Reactive oxygen species in pulmonary inflammation by ambient particulates Free Rad Biol Med 35 327 340 12899936
van Eeden SF Tan WC Suwa T Mukae H Terashima T Fujii T 2001 Cytokines involved in the systemic inflammatory response induced by exposure to particulate matter air pollutants (PM10 ) Am J Respir Crit Care Med 164 826 830 11549540
Vogel CF Sciullo E Matsumura F 2004a Activation of inflammatory mediators and potential role of Ah-receptor ligands in foam cell formation Cardiovasc Toxicol 4 363 374 15531779
Vogel CF Sciullo E Park S Liedtke C Trautwein C Matsumura F 2004b Dioxin increases C/EBPbeta transcription by activating cAMP/protein kinase A J Biol Chem 279 8886 8894 14684744
Vogel CF Matsumura F 2003 Interaction of 2,3,7,8-tetrachlorodibenzo-p -dioxin (TCDD) with induced adipocyte differentiation in mouse embryonic fibroblasts (MEFs) involves tyrosine kinase c-Src Biochem Pharmacol 66 1231 1244 14505802
Yeh ET 2004 CRP as a mediator of disease Circulation 109 suppl 1 II11 II14 15173057
Zwaka TP Hombach V Torzewski J 2001 C-reactive protein-mediated low density lipoprotein uptake by macrophages: implications for atherosclerosis Circulation 103 1194 1197 11238260
|
16263508
|
PMC1310915
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 21; 113(11):1536-1541
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8094
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8132ehp0113-00154216263509ResearchHealth Effects of a Mixture of Indoor Air Volatile Organics, Their Ozone Oxidation Products, and Stress Fiedler Nancy 12Laumbach Robert 12Kelly-McNeil Kathie 12Lioy Paul 12Fan Zhi-Hua 12Zhang Junfeng 123Ottenweller John 4Ohman-Strickland Pamela 3Kipen Howard 121 Department of Environmental and Occupational Medicine, University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School, Piscataway, New Jersey, USA2 Environmental and Occupational Health Sciences Institute, Piscataway, New Jersey, USA3 University of Medicine and Dentistry of New Jersey–School of Public Health, Piscataway, New Jersey, USA4 Veterans Affairs Medical Center, East Orange, New Jersey, USAAddress correspondence to N. Fiedler, Environmental and Occupational Health Sciences Institute, UMDNJ-Robert Wood Johnson Medical School, 170 Frelinghuysen Rd., Room 210, Piscataway, NJ 08854 USA. Telephone: (732) 445-0123 Ext. 625. Fax: (732) 445-0130. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 21 7 2005 113 11 1542 1548 21 3 2005 21 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In our present study we tested the health effects among women of controlled exposures to volatile organic compounds (VOCs), with and without ozone (O3), and psychological stress. Each subject was exposed to the following three conditions at 1-week intervals (within-subject factor): VOCs (26 mg/m3), VOCs + O3 (26 mg/m3 + 40 ppb), and ambient air with a 1-min spike of VOCs (2.5 mg/m3). As a between-subjects factor, half the subjects were randomly assigned to perform a stressor. Subjects were 130 healthy women (mean age, 27.2 years; mean education, 15.2 years). Health effects measured before, during, and after each 140-min exposure included symptoms, neurobehavioral performance, salivary cortisol, and lung function. Mixing VOCs with O3 was shown to produce irritating compounds including aldehydes, hydrogen peroxide, organic acids, secondary organic aerosols, and ultrafine particles (particulate matter with aerodynamic diameter < 0.1 μm). Exposure to VOCs with and without O3 did not result in significant subjective or objective health effects. Psychological stress significantly increased salivary cortisol and symptoms of anxiety regardless of exposure condition. Neither lung function nor neurobehavioral performance was compromised by exposure to VOCs or VOCs + O3. Although numerous epidemiologic studies suggest that symptoms are significantly increased among workers in buildings with poor ventilation and mixtures of VOCs, our acute exposure study was not consistent with these epidemiologic findings. Stress appears to be a more significant factor than chemical exposures in affecting some of the health end points measured in our present study.
building-related illnesslung functionneurobehavioralozonestresssymptomsvolatile organic compounds
==== Body
Between 800,000 and 1.2 million buildings in the United States may be associated with building-related illnesses, and thus, between 30 and 70 million workers are exposed to potentially unhealthy working conditions (Kreiss 1990; Woods 1989). Mixtures of volatile organic compounds (VOCs) and ozone (O3) are prominent pollutants in indoor environments (Fan et al. 2003). In some cases, VOCs measured in office buildings are associated with complaints of mucosal irritation and non-specific symptoms such as headache (Hodgson et al. 1991; Ten Brinke et al. 1998). Furthermore, healthy men and women intentionally exposed to similar mixtures of VOCs report increased symptoms of eye, nose, and throat irritation and reduced air quality ratings relative to clean air conditions (Hudnell et al. 1992; Prah et al. 1998). The number of symptoms reported in controlled exposure studies, however, are relatively few and of mild intensity compared with the ongoing complaints of office workers (Apter et al. 1994; Mendell and Smith 1990; Nordstrom et al. 1994; Zweers et al. 1992). Thus, some investigators suggest that when O3 reacts with VOCs in building environments, secondary products including ultrafine particles (particulate matter with aerodynamic diameter < 0.1 μm)may mediate the more substantial effects found in offices (Bako-Biro et al. 2004; Fan et al. 2003; Rohr et al. 2002; Weschler and Shields 2000; Wolkoff and Nielsen 2001). Our study assesses a selected suite of subjective and objective markers in response to the following exposure conditions: VOCs, VOCs + O3, and ambient air with a 1-min spike of VOCs [masked clean air (MCA)]. We hypothesized that exposure to VOCs or VOCs + O3 would result in greater symptom severity, compromised neurobehavioral performance, reduced lung function, and increased salivary cortisol relative to the MCA exposure (hypothesis 1: exposure main effect).
Gender and psychological stress also contribute to health complaints in buildings (Crawford and Bolas 1996; Hodgson 1995; Mendell 1993). For example, Bachmann and Myers (1995) found gender and psychological symptoms to be significant predictors of symptoms in two problem and one nonproblem building. Temperature, uncomfortable humidity, and reported odors, however, were also associated with symptoms in the buildings investigated. Women consistently report the highest prevalence of symptoms (Skov et al. 1989; Stenberg and Wall 1995), although external psychological stress (work load and control) is also associated with complaints (Norback et al. 1990; Ryan and Morrow 1992). Therefore, we chose to include only women in our study and to expose subjects to chemical mixtures with and without psychological stress. We hypothesized that subjects would report significantly greater symptom severity and would show a greater cortisol response when exposed to VOCs or VOCs + O3 with psychological stress compared with these exposure conditions without stress or to the MCA condition with or without psychological stress (hypothesis 2: exposure × stress interaction).
In summary, indoor environmental quality is affected by numerous factors, including biological, chemical, and particulate pollutants; temperature and humidity; quality of the heating, ventilation, and air conditioning system; noise; light; and odor (Mendell and Heath 2005). Our present study assessed the interaction of chemical pollutants and psychological stress on subjective (i.e., symptoms) and objective (i.e., cortisol, lung function, neurobehavioral performance) indicators of health effects, while holding temperature, humidity, noise, and light constant. Furthermore, our study added an untested exposure dimension created by combining VOCs with O3, shown in previous studies to produce a suite of irritating gas and condensed-phase products (Fan et al. 2003).
Materials and Methods
Subjects
One hundred thirty healthy, nonsmoking women, who were on average 27.2 years of age (SD = 8.0) with 15.2 (SD = 1.9) years of education, were recruited via advertisements in local newspapers. The ethnic composition of the sample was as follows: Caucasian, 56% (n = 73); black, 10% (n = 13); Hispanic, 8% (n = 10); Asian, 20% (n = 26); other, 6% (n = 8). Subjects completed a medical history, physical examination, and standard clinical blood chemistry to rule out previous significant occupational exposure to chemicals and the following health conditions: neurologic disease or brain injury, stroke or cardiovascular disease, serious pulmonary disease including asthma, liver or kidney disease, serious gastrointestinal disorders, known endocrine disease, pregnancy or lactation, and major psychiatric conditions, including psychoses, bipolar disorder, alcoholism or drug abuse, and multiple chemical sensitivity with significant illness behavior or disability. Three hundred forty-one individuals were screened for the study. Fifty-four were excluded at the telephone screening because of medical conditions, 74 never came in for their appointed physical examination, 40 declined to participate after the physical examination, 19 were excluded for medical conditions discovered at the physical examination, 4 dropped out after their first exposure, and 6 dropped out after their second exposure (n = 197). Fourteen of the 144 subjects who completed the study were pilot subjects whose data were not included in the analysis.
Dependent Measures
Symptom questionnaire.
Subjects rated each symptom on a ratio scale from 0 (barely detectable/no sensation) to 100 (strongest imaginable) (Green et al. 1996). We chose symptoms based on previous literature assessing the health effects of indoor air mixtures (Molhave et al. 1986; Prah et al. 1998). These symptoms included the cognitive and physical effects expected of VOC mixtures (Hudnell et al. 1992; Molhave et al. 1986), anxiety symptoms associated with the odor of exposure, eye irritation, upper respiratory and lower respiratory symptoms associated with O3 and the secondary products generated from the reactions of O3 with VOCs (Fan et al. 2003), and somatic symptoms not typically associated with VOC mixtures (Appendix 1) (Dalton et al. 1997).
Neurobehavioral.
This computerized divided-attention test of cognitive performance, performance on-line (POL) (Mills et al. 1996) offered five different levels of complexity. The test was validated in alcohol dosing trials and was developed explicitly for use in repeated-measures studies of alcohol and drug effects. POL included a central task in which the subject was presented with two lanes of traffic, divided by a double yellow line. Four conditions of “headlights” and “tail lights” appeared on any one trial. The subject was instructed to press the space bar only when a “safe” condition (i.e., left lane, white headlights, and right lane, red tail lights) existed. The peripheral task required the subject to respond with one of four arrow keys (up, down, left, right) in the direction of the critical stimulus (red octagon among other shapes). Task difficulty increased by increasing the number of distracting stimuli in the peripheral display to a random assortment of different colored circles, squares, and triangles. For the divided-attention display, the subject responded to both central and peripheral critical stimuli. After practice, 10 trials of 45 displays were presented at the most complex level. A composite performance score, composed of seven component scores, including hits, misses, false positives, response latency, and responses to targets at varying levels of visual angle, was the performance variable measured (Badiani et al. 1995).
Cortisol assays.
An extensive literature documents the significant (r ≥0.90) association between salivary and plasma cortisol (Kirschbaum and Hellhammer 1989, 1994). Salivary flow, which may be affected by anxiety, is not documented to affect the concentration of salivary cortisol (Dirks et al. 1988; Kahn et al. 1988), although circadian rhythm affects cortisol production (Walker et al. 1984). Thus, we tested subjects at the same time of day to control for the well-known circadian effect on cortisol production. Kirschbaum and Hellhammer (1989, 1994) also reported that, although a highly significant correlation is shown between salivary and plasma cortisol, absolute values vary significantly. However, in our study, we evaluated relative change scores rather than absolute values. Because hormonal fluctuations also affect cortisol levels, we measured salivary estradiol at baseline before each exposure session to account for any effects of ovulation on salivary cortisol.
Consistent with Kirschbaum and Helhammer (1989), we collected samples using the Salivette (Sarstedt Inc., Rommelsdorf, Germany) method. We asked the subject to chew on a cotton swab for 60 sec, then place the swab into a Salivette holder and affix the cap. The samples were centrifuged at 3,000 rpm for 10 min, which produced 0.5–1 mL of saliva. The saliva was frozen. Samples were analyzed, blind to exposure condition and subject characteristics, in one of the co-investigators’ laboratories (J.O.). Samples were run in duplicate, and equal numbers of samples from each group were run in the same assays.
Pulmonary function test.
With the subjects in standing position with nose clips attached, spirometry was performed using a Multispiro SX spirometry system (Creative Biomedics, Inc., San Clemente, CA) that was calibrated daily. We used the highest value for each parameter from any one of three reproducible tracings before exposure for comparison with the highest value for each parameter postexposure. Forced vital capacity (FVC), forced expiratory volume in 1 sec (FEV1), and forced expiratory flow 5–75% (FEF25–75) were parameters of interest.
Independent Variables
Public speaking task stressor.
After a 4-min silent preparation period, subjects constructed and delivered a 4-min speech on the following controversial scenarios: a) construct a presentation on the causes of gray hair from a Reader’s Digest article; b) present a position on whether homosexual men should be allowed in the military and be given special civil rights protection; c) defend herself against a shoplifting charge in front of three authority figures (al’Absi et al. 1997). In a direct comparison al’Absi et al. (1997) demonstrated that these three different topics caused a similar level of stress (cortisol response). We randomized the order of the three scenarios across subjects such that each subject delivered a speech on a different topic in each of her three exposure conditions. After the preparation period, subjects were told by the experimenter to stand, face the video camera, and begin giving the speech. To enhance the stressfulness of the procedure, we told each subject that the research technician was evaluating the speech as it was given and that three staff members would evaluate a videotape of the speech later; the subject was promised an additional stipend of $10 if she performed well. After completion of the three experimental sessions, all subjects received the added stipend regardless of performance.
Controlled environment facility and exposure generation.
The Environmental and Occupational Health Sciences Institute Controlled Environment Facility (CEF) is a stainless steel room 2.2 m × high × 4.1 m wide × 2.7 m deep with a total volume of 25 m3. The air supply is treated in a series of conditioning processes, which include air cooling/heating, humidification/dehumidification, and filtration through carbon and high-efficiency particulate air (HEPA) filters. All parameter controls are computer interfaced to maintain constant conditions in the CEF. To simulate a typically ventilated office building and to allow sufficient time for the formation of O3–VOC reaction products, the air flow rate through the facility was controlled at 1.8 ± 0.2 air changes/hr for all the exposures conducted in this study. The air supply enters the facility through two diffusers in the ceiling and exits through the perforated stainless steel floor to the exhaust vents. Small brushless fans (to prevent unwanted particle generation from brush degradation) were used in the CEF to ensure that the air was well mixed. During exposure sessions, the relative humidity and temperature were controlled to a range of between 24 and 49% and 73–82°F (23–28°C), respectively. A Teflon partition separated the subjects’ work stations, which were equipped with a computer and typing stand on a stainless steel table. This configuration allowed exposure of two subjects at a time. The technician communicated separately with each subject through headphones and could view subjects at all times through a two-way window. For subject safety, an electrocardiogram (ECG) signal was collected using disposable electrodes and sampled at 996 Hz through the Flexcomp Biomonitoring system 1.51 B (Thought Technology Ltd., Montreal, Quebec, Canada). This allowed continuous monitoring of heart rate during the exposure period.
The composition and relative weight of the VOC mixture used were similar to those in previous indoor air studies (Kjaergaard et al. 1991; Molhave et al. 1986; Otto et al. 1992) and are reported in Table 1 (Fan et al. 2003). d-Limonene, the most frequently identified terpene in indoor air, was also added to the mixture. A flask containing the liquid mixture of these 23 compounds was heated to 250°C by a hot plate, flash evaporated, and injected into the clean air stream that delivered the VOC mixture into the CEF. Constant concentrations of chemical compounds can be maintained in the CEF by injection of the chemicals continuously into the air supply, which flows through the CEF without recirculation. The desired concentration of VOCs in the indoor environment was achieved by adjusting the flow rate of the delivering air using a mass flow controller.
Total VOC concentration in the air of the indoor environment was approximately 26 mg/m3, with the concentration of each compound below the threshold limit value recommended for occupational exposure by the American Conference of Governmental Industrial Hygienists (1992). MCA was generated by introducing a 1-min spike of the VOC mixture at approximately 10% of the exposure concentration for the VOC exposure condition. The maximum total VOC concentration in the CEF during the MCA condition was approximately 2.5 mg/m3. Given the air exchange rate used in the experiments, it took about 90 min for the masking concentration to decay to 0.25 mg/m3.
The background O3 concentration is < 3 ppb in the CEF. O3 generated in situ by an O3 generator was delivered into the CEF and was maintained at a steady-state concentration of 40 ppb. This was consistent with the intention to examine the effect of products of O3–VOC reactions rather than O3 itself. Table 2 compares the concentrations of byproducts generated in the VOC + O3 condition relative to the VOC condition. Exposure generation and characterization were described in detail in a previous publication (Fan et al. 2003).
Procedures
Before receiving a complete physical examination, including a medical history review and routine blood chemistries, subjects who met inclusion criteria gave written informed consent according to the University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School Institutional Review Board. To reduce practice and novelty effects, subjects were shown the CEF and trained to perform the following procedures: symptom questionnaire, speech and neurobehavioral tasks, salivary sampling for cortisol and estradiol, nasal lavage, and spirometry. Subjects were randomly assigned to the order of exposure conditions, half with and half without the stressor. Subjects were told during informed consent that they may be asked to perform a public speaking task. They were not told whether they would be performing the task before any exposure session. Thus, subjects were blind to the stress condition. Neither the subject nor the research technicians were told exposure conditions.
Each experimental session was 3 hr in duration and occurred in the morning. Subjects were asked not to use caffeine or alcohol on the day before and on the day of the testing session. Subjects also could not have an active upper respiratory illness (either infection or allergy) or use medication for allergies or other respiratory conditions for 1-week before each exposure session. These conditions were queried by a nurse before participation on each day of exposure, and subjects were rescheduled if necessary.
Two subjects were tested at the same time. On the day of the first exposure session, a pregnancy test was given. Under typical room air conditions, a spirogram was performed in our clinical center before and after each exposure session. Subjects also completed the symptom questionnaire, practiced the POL, and performed baseline nasal lavage and salivary sampling for estradiol analysis in our clinical center.
We escorted subjects to the CEF, where they were seated at a table, had ECG sensors attached, and were given a pair of headphones used to communicate instructions for administration of questionnaires and tasks by the research technician. We gave each subject a loose-leaf binder with dividers to separate each of the questionnaires and tasks to be completed. Baseline measures before exposure were collected while the subject sat quietly with filtered room air, after which the exposure was administered for 140 min (Figure 1). During the initial 20 min of the exposure, subjects read quietly. We administered a 10-min vigilance task to maintain a consistent level of alertness across subjects before introduction of the psychological stressor (Jennings et al. 1992). In this simple color detection task, we instructed subjects to count bars that appeared periodically on a computer screen. After completion of the task, we asked subjects the number of bars counted. Subjects then completed either the public speaking task or simple arithmetic. While subjects performed the simple arithmetic task, they heard classical music through their headphones to screen out the other subject’s voice. After this period, all subjects typed a standard text for 10 min. A different text was used for each experimental session. As indicated in Figure 1, subjects completed questionnaires, collected salivary samples for cortisol analysis, and performed the neurobehavioral task during clean air baseline and the exposure period. Immediately after exposure we escorted subjects to our clinical center, where they performed postexposure spirometry, nasal lavage, the neurobehavioral task, and completed a questionnaires (Figure 1).
Statistical Analysis
Symptoms.
We analyzed the effects of exposure, stress, and time on both presence/absence of symptoms and symptom severity. For presence/absence, if a subject reported any symptom at all, then a “yes” was recorded; otherwise, a “no” was recorded. For presence/absence, a hierarchical logistic regression (Davidian and Giltinan 1995; McCulloch and Searle 2000) modeled the log-odds of no symptoms being reported for assessment at minute 15 (baseline) through minute 185 (after removal from exposure facility). For symptom severity, data were analyzed using a hierarchical Poisson regression model. In both cases generalized estimating equations were used that accounted for correlations between repeated measurements on the same individual (Liang and Zeger 1986). Tests of the exposure effects were conducted using type 3 score tests (Liang and Zeger 1986) of the interaction between exposure and time. Time was entered into the model as a categorical variable. We used contrasts to test whether individual changes in symptoms from baseline (minute 15) to each subsequent time point differed between exposures. The mean odds of reporting symptoms or the mean severity of symptoms at baseline was assumed to be the same for all three exposures. This analysis was first completed for the total symptoms and then for each classification of symptoms: VOC physical, VOC cognitive, eye irritation, anxiety, upper respiratory, lower respiratory, and somatic control symptoms. Results were based on the 130 subjects who received all three exposures, including the MCA condition. Uncorrected α-values are reported, with the αlevel after Bonferroni correction noted for each group of multiple comparisons.
Cortisol analyses.
Cortisol is normally elevated during ovulation, potentially attenuating a woman’s cortisol response to a stressor at that time. Therefore, cortisol data were analyzed both with and without data from any specific exposure session in which the subject’s estradiol level was ≥5 ng/mL. Based on this cutoff, 13 of the 130 (10%) subjects were ovulating during their MCA exposure, 13 of 130 (10%) were ovulating during the VOC exposure, and 15 of 130 (11.5%) were ovulating during the VOC + O3 exposure. On comparison of the cortisol data analyses with and without ovulation sessions, no differences in outcome were noted. Therefore, sessions in which subjects were ovulating were included in the final analyses of cortisol.
Given the continuous response, SAS PROC MIXED (SAS Institute Inc. 1999) was used to analyze the fixed-factor effects. The cortisol response was right skewed and therefore, for the linear models, was transformed using a square root transformation to better satisfy the normality assumption required for the mixed linear model.
Neurobehavioral performance and lung function.
A mixed linear model was used to test the effect of exposure on the composite score from the neurobehavioral task and on indicators of lung function.
Results
Hypothesis 1: Exposure Main Effect
Symptoms.
After controlling for baseline symptoms (minute 15), the overall test of the exposure main effect on total symptom severity did not confirm hypothesis 1 (Figure 2). However, marginal effects for presence/absence of the subscales of VOC physical symptoms (chi squared = 18.01; df = 10; p < 0.05) and severity of lower respiratory symptoms (chi squared = 16.92; df = 10; p < 0.08) were observed; no effects were observed for the other categories of symptoms.
Neurobehavioral performance, salivary cortisol, and lung function.
There was no significant main effect of exposure on neurobehavioral performance (F = 1.00, df = 6, 387; p = 0.4273), salivary cortisol (F = 0.40; df = 6, 316; p = 0.888), or lung function (data not shown for neurobehavioral or cortisol measures; see Table 3 for lung function).
Hypothesis 2: Exposure × Stressor Interaction
Symptoms.
Hypothesis 2 was not confirmed for presence/absence or for total symptom severity. However, regardless of exposure condition, subjects who were in the stress condition reported significantly greater severity on the anxiety sub-scale (chi squared = 22.73; df = 5; p < 0.0004) than those who were not. Specifically, relative to baseline either before (minute 15) or after exposure onset (minutes 60 and 90), symptoms of anxiety were significantly more severe after the stressor at minutes 110 and 125 (Figure 3). No other symptom subscale was significantly affected by stress or by the exposure × stressor interaction.
Salivary cortisol.
The hypothesized interaction effect of exposure × stress on cortisol was not significant, but the main effect of stress on cortisol was significant (F = 4.90; df = 3, 347; p = 0.0024). The significance of the main effect of stress was due to changes in cortisol levels from before the stressor (minute 90) to after the stressor (minute 125) (t = 4.00; p < 0.0001). When examining changes from minutes 110 to 125, on average, the cortisol levels decreased (from 0.178 to 0.154) for the no-stressor condition and increased slightly (from 0.140 to 0.147) for thestressor condition (t = 3.63; p = 0.0003) (Figure 4).
Discussion
The most striking result of our study was the lack of significant subjective or objective health effects from exposure to mixtures of VOCs both with and without O3, despite significant differences in the chemical composition of the air for the three conditions (Table 2). Numerous epidemiologic studies suggest that symptoms are significantly increased among workers in buildings with poor ventilation and mixtures of VOCs (Mendell 1993; Mendell et al. 2002; Sieber et al. 1996), many of which likely contain low levels of O3. Our present controlled acute exposure to similar chemical mixtures in young women, however, failed to support these epidemiologic findings. Relative to ambient air masked with a pulse of VOCs, neither the VOC nor the VOC + O3 exposures caused significantly increased symptom reports, changes in cortisol, reduced neurobehavioral performance, or changes in lung function. Subjects were more likely to report some VOC physical symptoms in the VOC and VOC + O3 conditions relative to the MCA condition. However, these effects were of marginal significance and became nonsignificant with appropriate correction for multiple comparisons. Conversely, although stress did not exacerbate exposure effects, stress significantly increased symptoms of anxiety. The effect of stress was further validated by the significant difference in cortisol for those who received the stressor relative to subjects who did not.
There was a marginal (p < 0.08) increase in severity, but not incidence, of lower respiratory symptoms with the VOC and VOC + O3 exposure. However, there were no significant changes in the lung function parameters attributable either to the VOC or VOC + O3 exposure. This is consistent with work showing increased lower respiratory symptoms at 50 mg/m3 but not at 25 mg/m3 of VOCs alone, but no change in spirometry or increase in inflammatory mediators from induced sputum at either VOC concentration, again without O3 (Pappas et al. 2000). Although rodent bioassays had indicated that relatively higher concentrations of limonene–O3 oxidation products were irritating to the respiratory tract, at our exposure concentrations we did not see an effect on lung function and only a marginal effect on symptoms (Rohr et al. 2002; Wilkins et al. 2001). Indicators of nasal inflammation were also negative for our exposures (Laumbach et al., in press). In contrast to our present findings, Kleno and Wolkoff (2004) reported increases in eye blink frequency among a small number of subjects with eye exposure only to limonene oxidation products and nitrate radicals relative to clean air. Although we did not measure eye blink frequency, symptoms of eye irritation were not significantly greater in the VOC + O3 condition.
Our present findings regarding neurobehavioral performance are consistent with those of Otto et al. (1992), who reported no changes in neurobehavioral performance among subjects exposed to a similar mixture of 23 VOCs relative to clean air. However, the research group from the Technical University of Denmark reported several studies in which subjects were exposed to off-gassing from a 20-year-old carpet and showed reductions in productivity on tasks that simulate office work (typing, calculations) (Wargocki et al. 1999, 2000a). Although the total VOCs were of similar concentrations between the exposure conditions with and without the carpet (~ 2.34 ppm), the composition of the chemical mixture differed substantially between the two conditions, and likely contained unidentified compounds associated with emissions from the carpet. In contrast to the negative findings in our present study, Wargocki et al. (1999, 2000a) also reported more symptoms (headache) and slower typing speed in response to the condition with the old carpet (polluted condition). Because the total VOC concentration (2.35 ppm) was less than in our present study, symptom and performance differences could be ascribed to differences in the chemical composition, to unspecified biological components, or to differences in task demands. Neither acetone nor acetic acid was present in the VOC mixture used in our present study (only small amounts were produced from the O3 and VOC reactions), but such a specific chemical effect seems unlikely. Wargocki et al. (1999, 2000b) also suggested that simulated work tasks (typing, calculations) performed over a longer period of time (265 min) than in our present study were more sensitive to the effects of poor air quality than are standard neurobehavioral tasks. Thus, although the POL was given for approximately 20 min on two occasions during exposure, this did not require the sustained effort needed to perform continuous typing for 47 min on two occasions as in the Wargocki et al. (1999, 2000b) studies. However, the use of an old carpet as an exposure was ultimately not comparable with the present specific VOC mixture.
Although some increased symptoms were observed in previous controlled exposures using similar “indoor air” mixtures (Hudnell et al. 1992; Prah et al. 1998), actually only a few out of many symptoms assessed in those studies were significantly increased. A careful examination of those symptoms exacerbated by exposure reveals some consistency with our present findings. Prah et al. (1998) reported that relative to clean air, mixtures of VOCs increased ratings of nasal irritation, odor intensity, and air quality but not health (cough, sore throat) and cognitive symptoms (memory loss, dizziness). Similarly comparing responses of a VOC mixture with clean air, Hudnell et al. (1992) reported significantly reduced air quality ratings and increased odor level, but they also reported increased symptoms of headache, eye irritation, drowsiness, and throat irritation. However, Hudnell et al. (1992) did not control for multiple statistical comparisons among the individual tests of 22 symptoms (p-value set at < 0.05). Furthermore, no previous indoor air study has “masked” the clean air condition to control for the effects of odor on symptoms. Several studies suggest that when subjects rate air contaminated with various combinations of limonene, O3, and office products (paper), they report dissatisfaction with air quality (Knudsen et al. 2002; Tamas et al., in press). However, these studies did not measure health symptoms.
To further clarify our findings, we conducted power calculations to estimate the size of the effect that would be necessary to detect a change between symptoms reported at baseline (minute 15) and either minute 60 or 125. Based on 130 subjects receiving all three exposures, we calculated the minimum actual symptom difference between the VOC and VOC + O3 exposures and the MCA exposure that would be needed to attain a power of 90%. In these power calculations, we assumed that the size of the exposure effect, relative to MCA, was the same for VOCs and VOCs + O3. To simplify the calculations, we used changes from baseline as the response variable in a repeated-measures analysis of variance. We calculated the variance components for each response and time point from the existing data. To account for multiple testing, the Bonferroni corrected significance level was used corresponding to the level used for testing in our present study.
The results indicate that in all cases, minimum average increases of between 0.6 and 3.0 points in the average score for each category of symptoms would have to exist in order to detect a difference with a power of 90% (Table 4). With symptom severity scored on a scale of 0 to 100, this indicates that very small changes were detectable with the sample size used in our study.
We performed similar power calculations for the main effect of exposure on neurobehavioral performance, cortisol, and lung function. The minimum detectable difference in neurobehavioral performance between the VOC, VOC + O3, and MCA conditions for a power of 90% was calculated at 1.30 using a 0.05 significance level. Because of a within-session learning effect, the MCA exposure resulted in a mean increase from baseline (baseline mean = 32.72; SD = 9.1) of 2.10 points on the neurobehavioral task. With 130 subjects we could detect a relative reduction of 1.30 points resulting in 0.80 point improvement for the VOC or VOC + O3 exposure conditions (i.e., 2.10 – 1.30 = 0.80). For cortisol, the minimum detectable difference in changes from baseline needed for a power of 90% was 0.066 μg/dL using a 0.05 significance level. The MCA condition resulted in a mean decrease from baseline (the average baseline = 0.2627; SD = 0.1831) of 0.1148 μg/dL. Thus, a decrease of no more than 0.0488 μg/dL in the VOC or VOC + O3 conditions was needed to detect a difference. Finally, for lung function, the minimum detectable differences in changes from baseline are 109.7 mL, 122.8 mL, and 0.917 L/sec, respectively, for changes in FEV1, FVC, and FEF25–75. These were calculated with a significance level of p = 0.0167 (0.05/3). Thus, for all dependent measures and the present sample size, relatively small changes were needed to detect a difference between exposure conditions.
Conclusions
In conclusion, relatively brief, one-time exposures to mixtures of VOCs or VOCs and their oxidation products, at the upper bound of typical indoor concentration range, did not appear to cause significant acute changes in symptoms, neurobehavioral performance, or lung function in healthy women. In contrast, the psychological stressor was effective in producing increased autonomic arousal, as indicated by salivary cortisol, and in causing increased symptoms of anxiety, both well-documented effects of psychological stress. However, stress and exposure were neither synergistic nor additive in their effects on symptoms or neurobehavioral performance. The effect of stress, however, was isolated to symptoms of anxiety and did not generalize to other more typical symptoms associated with poor indoor air such as nasal irritation or headache. Although the irritation potency of complex and variable mixtures of VOCs and their oxidation products in buildings is difficult to predict, reported air concentrations of VOCs in buildings with poor indoor air quality are typically an order of magnitude lower than the VOC concentrations used in this study. Thus, our results suggest that the VOC concentration alone may not be the most salient factor to account for acute health complaints. Our present results support the conclusion that 3-hr exposures to VOCs or to the reaction products of VOCs and O3 at concentrations typically found in nonindustrial buildings are unlikely to be a significant cause of acute health complaints or effects for most occupants.
Several caveats need to be considered for our present study. This study included the largest number of subjects to date and intentionally selected only women for study because of their hypothesized vulnerability to report indoor air quality symptoms. However, the extent to which our results apply to indoor air problems experienced in a work environment with many other demands and chronic exposures is problematic. Thus, the lack of health effects observed may simply be a function of the necessarily acute exposure paradigm with healthy young subjects. Conversely, the exposure concentrations were quantitatively higher than those documented in most buildings with indoor air complaints. Furthermore, work demands were modeled in our study through use of a known stressor as well as requirements for computerized neurobehavioral tasks, and the former was successful in causing autonomic arousal and symptoms of anxiety. Another attribute of our study was that all exposure conditions were conducted at a relatively high air exchange rate (~ 1.8 air exchanges/hr), whereas health complaints associated with indoor air quality often occur in buildings with poor ventilation (i.e., air exchange rates were an order of magnitude lower than the air exchange rate used in our study). Overall, this study suggested that for a 2-hr time period, psychological stress may be a more potent factor than ambient chemical mixtures in the complaints attributed to poor indoor environments.
Funding for our research was supported by a grant from the National Institute for Occupational Safety and Health (R01 OH03691-01) and by National Institute of Environmental Health Sciences Center grant ES05022.
Appendix 1. Symptom list
VOC physical
Headache
Fatigue
Lightheaded
Drowsy
Nausea
VOC cognitive
Difficulty concentrating
Disoriented/confused
Dizzy
Eye irritation
Eye irritation (burning, dryness, or itching)
Runny/watery eyes
Anxiety
Feel jittery in body
Feel nervous
Heart palpitations
Feel tense
Worried
Upper respiratory
Sneeze
Nasal congestion
Choking
Throat irritation (burning or dryness)
Nose irritation, dryness, or itching
Lower respiratory
Short of breath
Wheezy
Chest tightening
Chest pain
Coughing
Somatic control
Skin irritation or dryness
Stomachache
Numbness/tingling
Ear ringing
Leg cramps
Back pain
Sweating
Body aches
Figure 1 Time line for experimental procedure. Abbreviations: Cort1, cortisol collection at minute 15; Cort2, cortisol collection at minute 90; Cort3, cortisol collection at minute 110; Cort 4, cortisol collection at minute 130; Neuro1, neurobehavioral task (POL) at minute 15; Neuro2, Neurobehavioral task (POL) at minute 60; Neuro3, neurobehavioral task (POL) at minute 140; Neuro4, neurobehavioral task (POL) at 5 min postexposure; Sym1, symptom questionnaire at baseline; Sym2, symptom questionnaire at minute 15; Sym3, symptom questionnaire at minute 60; Sym4, symptom questionnaire at minute 90; Sym5, symptom questionnaire at minute 110; Sym6, symptom questionnaire at minute 130; Sym7, symptom questionnaire at 5 min postexposure.
Figure 2 Total mean symptom severity at each time point across exposures.
Figure 3 Mean anxiety symptom severity at each time point across exposures stress versus no stress. Bonferroni correction: p < 0.003. Time point comparison for anxiety symptom sensitivity is as follows: change from minute 15 (baseline) to minute 110 (poststress), χ2 = 15.17 (p < 0.0001); change from minute 60 (after exposure onset) to minute 110 (poststress), χ 2 = 12.52 (p < 0.0004); change from minute 90 (prestress) to minute 110 (poststress), χ 2 = 30.16 (p < 0.0001); change from minute 110 to minute 125, χ2 = 22.29 (p < 0.0001); change from minute 110 to minute 185 (poststress), χ 2 = 11.64 (p < 0.0006).
Figure 4 Cortisol over time for stress versus no stress regardless of exposure.
Table 1 The mixture of VOCs.
No. Compound Relative weight Concentration (mg/m3)
1 n-Butylacetate 10 8.25
2 p-Xylene 10 8.25
3 n-Butanol 1 0.825
4 n-Decane 1 0.825
5 1-Decene 1 0.825
6 1,1-Dichloroethane 1 0.825
7 d-Limonene 1 0.825
8 Ethylbenzene 1 0.825
9 Ethoxyethylacetate 1 0.825
10 n-Hexanal 1 0.825
11 n-Hexane 1 0.825
12 n-Nonane 1 0.825
13 α -Pinene 1 0.825
14 2-Butanone 0.1 0.083
15 Cyclohexane 0.1 0.083
16 3-Methyl-2-butanone 0.1 0.083
17 4-Methyl-2-pentanone 0.1 0.083
18 n-Pentanal 0.1 0.083
19 Isopropanol 0.1 0.083
20 n-Propylbenzene 0.1 0.083
21 1,2,4-Trimethylbenzene 0.1 0.083
22 n-Undecane 0.1 0.083
23 1-Octene 0.01 0.008
Total 26.330
Table 2 Summary of products observed during different exposure conditions.
Exposure conditions Ultrafine particle number concentration (particles/cm3)a Particle mass concentration (μg/m3)b Formaldehyde (μg/m3)b p-Tolualdehyde (μg/m3)b Glyoxal (μg/m3)b Hydrogen peroxide (ppb)a,c
MCA NAd < 5 7 ND ND NAd
VOCs (23 VOCs) 2,500 3–7 13 ND ND 0.3
VOCs + O3 (O3 + 23 VOCs) 46,000 140 40 6.2 4.6 1.9
ND, not detected.
a Two-hour average concentration.
b Four-hour average concentration.
c Includes hydrogen peroxide and organic hydroperoxides.
d Not measured but expected to be the same or lower than the 23 VOC-only condition.
Table 3 Spirometry changes after exposures to MCA, VOCs, and VOCs + O3 (mean difference ± SD; n = 130).
MCA VOCs VOCs ± O3 p-Value
Change in FEV1 (mL) –50.46 ± 174.02 6.15 ± 176.56 –25.30 ± 210.33 0.05
Change in FVC (mL) –71.54 ± 189.22 –42.62 ± 184.40 –60.38 ± 237.11 0.52
Change in FEF25–75 (L/sec) –0.00 ± 0.41 0.51 ± 3.63 0.09 ± 0.53 0.12
Difference is the post-minus preexposure score. Bonferroni correction: p < 0.02.
Table 4 Effect sizes for symptom severity based on 130 subjects.
VOC
Respiratory
Change (from 15 min) Total symptoma General Cognitive Eye Anxiety Upper Lower Somatic
60 min 0.99 2.69 2.48 2.07 1.43 1.42 0.74 1.04
125 min 1.15 2.98 2.69 2.15 1.85 1.49 0.64 1.13
a Calculated at a significance level of 0.05. The effect sizes for all remaining subcategories of symptoms were calculated at a significance level of 0.05/7 = 0.0071.
==== Refs
References
al’Absi M Bongard S Buchanan T Pincomb GW Licinio J Lovallo WR 1997 Cardiovascular and neuroendocrine adjustment to public speaking and mental arithmetic stressors Psychophysiology 34 266 275 9175441
American Conference of Governmental Industrial Hygienists 1992. 1992–1993 Threshold Limit Values for Chemical Substances and Physical Agents and Biological Exposure Indices. Cincinnati, OH:American Conference of Governmental Industrial Hygienists.
Apter A Bracker A Hodgson M Sidman J Leung W-Y 1994 Epidemiology of the sick building syndrome J Allergy Clin Immunol 94 277 288 8077580
Bachmann MO Myers JE 1995 Influences on sick building syndrome symptoms in three buildings Soc Sci Med 40 245 251 7899936
Badiani A Browman KE Robinson TE 1995 Influence of novel versus home environments on sensitization to the psychomotor stimulant effects of cocaine and amphetamine Brain Res 674 291 298 7796109
Bako-Biro Z Wargocki P Weschler CJ Fanger PO 2004 Effects of pollution from personal computers on perceived air quality, SBS symptoms and productivity in officers Indoor Air 14 178 187 15104785
Crawford JO Bolas SM 1996 Sick building syndrome, work factors and occupational stress Scan J Work Environ Health 22 243 250
Dalton P Wysocki CJ Brody MJ Lawley HJ 1997 Perceived odor, irritation, and health symptoms following short-term exposure to acetone Am J Ind Med 31 558 569 9099358
Davidian M Giltinan DM 1995. Nonlinear Models for Repeated Measurement Data. New York:Chapman & Hall.
Dirks H Colic D Haeckel R 1988. Die Bestimmung von cortisol im Speichel und ihre klinische Anwendung [in German]. In: Speicheldiagnostik (Haeckel R, ed). Darmstadt, Germany: GIT-Verlag, 86–93.
Fan Z Lioy P Weschler C Fiedler N Kipen H Zhang J 2003 Ozone-initiated reactions with mixtures of volatile organic compounds under simulated indoor conditions Environ Sci Technol 37 1811 1821 12775052
Green HG Dalton P Cowart B Shaffer G Rankin K Higgins J 1996 Evaluating the labeled magnitude scale for measuring sensations of taste and smell Chem Senses 21 323 334 8670711
Hodgson M 1995 The sick-building syndrome Occup Med State Art Rev 10 167 175
Hodgson MJ Frohliger J Permar E Tidwell C Traven ND Olenchock SA 1991 Symptoms and microenvironmental measures in nonproblem buildings J Occup Med 33 527 533 2037908
Hudnell HK Otto DA House DE Molhave L 1992 Exposure of humans to a volatile organic mixture. II: Sensory Arch Environ Health 47 31 38 1540000
Jennings JR Kamarck T Stewart C Eddy M Johnson P 1992 Alternate cardiovascular baseline assessment techniques: vanilla or resting baseline Psychophysiology 29 742 751 1461961
Kahn J-P Rubinow DR Davis CL Kling M Post RM 1988 Salivary cortisol: a practical method for evaluation of adrenal function Biol Psychiatry 23 335 349 3257706
Kirschbaum C Hellhammer DH 1989 Salivary cortisol in psychobiological research: an overview Neuropsychobiology 22 150 169 2485862
Kirschbaum C Hellhammer DH 1994 Salivary cortisol in psychoneuroendocrine research: recent developments and applications Psychoneuroendocrinology 19 313 333 8047637
Kjaergaard SK Molhave L Pedersen OF 1991 Human reactions to a mixture of indoor air volatile organic compounds Atmos Environ 25 1417 1426
Kleno J Wolkoff P 2004 Changes in eye blink frequency as a measure of trigeminal stimulation by exposure to limonene oxidation products, isoprene oxidation products and nitrate radicals Int Arch Occup Environ Health 77 235 243 15007652
Knudsen H Afshari A Ekberg L Lundgren B 2002. Impact of ventilation rate, ozone and limonene on perceived air quality in offices. In: Indoor Air 2002: Proceedings of the 9th International Conference on Indoor Air Quality and Climate, June 30–July 5, Monterey, CA (Levin H, Bendy G, Cordell J (ed). Vol II. Santa Cruz, CA:The International Academy of Indoor Air Sciences, 285–290.
Kreiss K 1990 The sick building syndrome: where is the epidemiologic basis Am J Publ Health 80 1172 1173
Laumbach R Fiedler N Gardner C Laskin DL Fan Z-H Zhang J In press. Nasal effects of a mixture of volatile organic compounds and their ozone oxidation products. J Occup Environ Med.
Liang KY Zeger LL 1986 Longitudinal data analysis using generalized linear models Biometrika 73 13 22
McCulloch CE Searle SR 2000. Generalized, Linear, and Mixed Models. New York:John Wiley & Sons.
Mendell M Smith A 1990 Consistent pattern of elevated symptom rates in air-conditioned office buildings: a re-analysis of epidemiologic studies Am J Public Health 80 1193 1199 2400029
Mendell MJ 1993 Non-specific symptoms in office workers: a review and summary of the epidemiologic literature Indoor Air 3 227 236
Mendell MJ Fisk WJ Kreiss K Levin H Alexander D Cain WS 2002 Improving the health of workers in indoor environments: priority research needs for a national occupational research agenda Am J Public Health 92 1430 1440 12197969
Mendell MJ Heath GA 2005 Do indoor pollutants and thermal conditions in schools influence student performance? A critical review of the literature Indoor Air 15 27 52 15660567
Mills KC Parkman KM Spruill SE 1996 A PC-based software test for measuring alcohol and drug effects in human subjects Alcohol Clin Exp Res 20 1582 1591 8986207
Molhave L Back B Pedersen OF 1986 Human reactions to low concentrations of volatile organic compounds Environ Int 12 167 175
Norback D Ingegerd M Widstrom J 1990 Indoor air quality and personal factors related to the sick building syndrome Scan J Work Environ Health 16 121 128
Nordstrom K Norback D Akselsson R 1994 Effect of air humidification on the sick building syndrome and perceived indoor air quality in hospitals: a four month longitudinal study Occup Environ Med 51 683 688 8000493
Otto D Hudnell HK House D Molhave L Counts W 1992 Exposure of humans to a volatile organic mixture, behavioral assessment Arch Environ Health 47 23 30 1539999
Pappas GP Herbert RJ Henderson W Koenig J Stover B Barnhart S 2000 The respiratory effects of volatile organic compounds Int J Occup Environ Health 6 1 8 10637531
Prah JD Case MW Goldstein GM 1998 1998 equivalence of sensory responses to single and mixed volatile organic compounds at equimolar concentrations Environ Health Perspect 106 739 744 9799190
Rohr AC Wilkins CK Clausen PA Hammer M Nielsen GD Wolkoff P 2002 Upper airway and pulmonary effects of oxidation products of (+)-alpha-pinene, d-limonene, and isoprene in BALB/c mice Inhal Toxicol 14 663 684 12122569
Ryan CM Morrow KA 1992 Dysfunctional buildings or dysfunctional people: an examination of the sick building syndrome and allied disorders J Consult Clin Psychol 60 220 224 1592951
SAS Institute Inc 1999. SAS/STAT User’s Guide, Version 8. 1999. Cary, NC:SAS Institute Inc.
Sieber WK Petersen MR Stayner L 1996 The NIOSH indoor environmental evaluation experience: part three—associations between environmental factors and self-reported health outcomes Appl Occup Environ Hyg 11 1387 1392
Skov P Valbjorn O Pedersen BV Danish Indoor Climate Study Group 1989 Influence of personal characteristics, job-related factors and psychosocial factors on the sick building syndrome Scan J Work Environ Health 15 286 295
Stenberg B Wall S 1995 Why do women report “sick building symptoms” more often than men Soc Sci Med 40 491 502 7725123
Tamas G Weschler CJ Toftum J Fanger PO In press. The influence of ozone-limonene reactions on perceived air quality. Indoor Air.
Ten Brinke J Selvin S Hodgson AT Fisk WJ Mendell CP Koshland CP 1998 Development of new volatile organic compound exposure metrics and their relationship to “sick building syndrome” symptoms Indoor Air 8 140 152
Walker RF Joyce BG Dyas J 1984. Immunoassays of steroids in saliva. In: Salivary Cortisol. 1: Monitoring Changes in Normal Adrenal Activity (Read GF, Riad-Fahmy D, Walker RF, eds). Cardiff, Wales:Alpha Omega, 308–316.
Wargocki P Wyon DP Baik YK Clausen G Fanger PO 1999 Perceived air quality, sick building syndrome (SBS) symptoms, and productivity in an office with two different pollution loads Indoor Air 9 165 179 10439554
Wargocki P Wyon DP Fanger P 2000a. Productivity is affected by the air quality in offices. In: Proceedings of Healthy Buildings 2000, 6–10 August 2000, Helsinki, Finland (Sappanen O, Sateri J, eds). Vol 1. Oy, Finland:SIY Indoor Air Information, 635–640.
Wargocki P Wyon DP Sundell J Clausen G Fanger P 2000b The effects of outdoor supply rate in an office on perceived air quality, sick building syndrome (SBS) symptoms, and productivity Indoor Air 10 222 236 11089327
Weschler CJ Shields HC 2000 The influence of ventilation on reactions among indoor pollutants: modeling and experimental observations Indoor Air 10 92 100 11980107
Wilkins CK Clausen PA Wolkoff P Larsen ST Hammer M Larsen K 2001 Formation of strong airway irritants in mixtures of isoprene/ozone and isoprene/ozone/nitrogen dioxide Environ Health Perspect 109 937 941 11673123
Wolkoff P Nielsen GD 2001 Organic compounds in indoor air—their relevance for perceived indoor air quality? Atmos Environ 35 4407 4417
Woods JE 1989 Cost avoidance and productivity in owning and operating buildings Occup Med State Art Rev 4 753 770
Zweers T Preller L Brunekreef B Boleij JSM 1992 Health and indoor climate complaints of 7043 office workers in 61 buildings in the Netherlands Indoor Air 2 127 136
|
16263509
|
PMC1310916
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 21; 113(11):1542-1548
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8132
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8195ehp0113-00154916263510ResearchThyroid-Hormone–Disrupting Chemicals: Evidence for Dose-Dependent Additivity or Synergism Crofton Kevin M. 1Craft Elena S. 1*Hedge Joan M. 1Gennings Chris 2Simmons Jane E. 3Carchman Richard A. 2Carter W. Hans Jr2DeVito Michael J. 31 Neurotoxicology and 3 Experimental Toxicology Divisions, National Health and Environmental Effects Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA2 Solveritas LLC, Richmond, Virginia, USAAddress correspondence to K.M. Crofton, Neurotoxicology Division, MD-105-04, National Health and Environmental Effects Laboratory, U.S. EPA, Research Triangle Park, NC 27711 USA. Telephone: (919) 541-2672. Fax: (919) 541-4849. E-mail:
[email protected]* Current address: Nicholas School of the Environment and Earth Sciences, Duke University, Durham, NC 27708 USA.
C.G., R.A.C., and W.H.C. were supported by U.S. EPA contract RFQ-RT-03-00298.
C.G., R.A.C., and W.H.C. have a financial interest in Solveritas LLC. All other authors declare they have no competing financial interests.
11 2005 21 7 2005 113 11 1549 1554 11 4 2005 21 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Endocrine disruption from environmental contaminants has been linked to a broad spectrum of adverse outcomes. One concern about endocrine-disrupting xenobiotics is the potential for additive or synergistic (i.e., greater-than-additive) effects of mixtures. A short-term dosing model to examine the effects of environmental mixtures on thyroid homeostasis has been developed. Prototypic thyroid-disrupting chemicals (TDCs) such as dioxins, polychlorinated biphenyls (PCBs), and poly-brominated diphenyl ethers have been shown to alter thyroid hormone homeostasis in this model primarily by up-regulating hepatic catabolism of thyroid hormones via at least two mechanisms. Our present effort tested the hypothesis that a mixture of TDCs will affect serum total thyroxine (T4) concentrations in a dose-additive manner. Young female Long-Evans rats were dosed via gavage with 18 different polyyhalogenated aromatic hydrocarbons [2 dioxins, 4 dibenzofurans, and 12 PCBs, including dioxin-like and non-dioxin-like PCBs] for 4 consecutive days. Serum total T4 was measured via radioimmunoassay in samples collected 24 hr after the last dose. Extensive dose–response functions (based on seven to nine doses per chemical) were determined for individual chemicals. A mixture was custom synthesized with the ratio of chemicals based on environmental concentrations. Serial dilutions of this mixture ranged from approximately background levels to 100-fold greater than background human daily intakes. Six serial dilutions of the mixture were tested in the same 4-day assay. Doses of individual chemicals that were associated with a 30% TH decrease from control (ED30), as well as predicted mixture outcomes were calculated using a flexible single-chemical-required method applicable to chemicals with differing dose thresholds and maximum-effect asymptotes. The single-chemical data were modeled without and with the mixture data to determine, respectively, the expected mixture response (the additivity model) and the experimentally observed mixture response (the empirical model). A likelihood-ratio test revealed statistically significant departure from dose additivity. There was no deviation from additivity at the lowest doses of the mixture, but there was a greater-than-additive effect at the three highest mixtures doses. At high doses the additivity model underpredicted the empirical effects by 2- to 3-fold. These are the first results to suggest dose-dependent additivity and synergism in TDCs that may act via different mechanisms in a complex mixture. The results imply that cumulative risk approaches be considered when assessing the risk of exposure to chemical mixtures that contain TDCs.
additivitycumulative riskpolyhalogenated aromatic hydrocarbonssynergismthyroid hormone disruptors
==== Body
Thyroid-disrupting chemicals (TDCs) are xenobiotics that alter the structure or function of the thyroid gland, alter regulatory enzymes associated with thyroid hormone (TH) homeostasis, or change circulating or tissue concentrations of THs. TDCs include a wide range of chemical structures. Chemicals such as perchlorate inhibit the uptake of iodide into the thyroid gland, with subsequent decrease in iodine-based TH synthesis (Wolff 1998). Other chemicals (e.g., thionamides, amitrole, and ethylenethiourea) decrease TH synthesis by inhibition of thyroid peroxidase (Capen 1997, 1998; Hill et al. 1998; Hurley 1998; McClain 1995). Many classes of xenobiotics alter TH levels by altering catabolic pathways. Polyhalogenated aromatic hydrocarbons (PHAHs) represent one such class of chemicals that induce uridine diphosphoglucuronosyl transferases (UGTs). UGTs glucuronidate THs, and induction of these enzymes increases the elimination of THs (Hill et al. 1998; Hood and Klaassen 2000a; McClain et al. 1989; Oppenheimer et al. 1968).
A major uncertainty regarding the endocrine-disrupting ability of environmental xenobiotics is the potential for additive or synergistic (i.e., greater-than-additive) effects of exposure to mixtures (Daston et al. 2003; International Programme on Chemical Safety 2002). Solving the problem of predicting the effects of chemical mixtures is a daunting task. There are limited studies in the peer-reviewed literature that examine mixtures of TDCs (Desaulniers et al. 2003; Teuschler et al. 2002; Wade et al. 2002). Desaulniers et al. (2003) found that use of 2,3,7,8-tetra-chlorodibenzo-p-dioxin (TCDD) toxic equivalents predicted the additive effects of a mixture of coplanar polychlorinated biphenyls (PCBs), polychlorinated dibenzo-p-dioxins, and polychlorinated dibenzofurans on circulating thyroxine (T4) concentrations in neonatal rats. Wade et al. (2002) subchronically exposed adult male rats to a complex mixture of 16 organochlorines, lead, and cadmium. Effects on thyroid histopathology and hormone concentrations were underpredicted based on assumptions of additivity using published health advisories [e.g., reference doses (RfDs), acceptable daily intakes (ADIs)]. These previous efforts investigated the effects of mixtures without concurrent experimental characterization of the effects of the individual chemicals. This type of approach is useful on a case-by-case basis but does not help answer global issues in the arena of mixtures risk assessments (LeBlanc and Olmstead 2004).
Our present study tested the hypothesis that a mixture of 18 PHAHs acts in a dose-additive manner. The hypothesis was tested using a flexible single-chemical-required (FSCR) method of analysis (Gennings et al. 2004). This model assumes that the effects of the mixture will be predicted by the constraint of Berenbaum’s definition of additivity (Berenbaum 1985). In addition this model allows the calculation of the predicted mixture outcome for chemicals with differing dose thresholds and maximum asymptotes (Gennings et al. 2002, 2004). A short-term oral exposure model (Craft et al. 2002) was used to estimate the impact of 18 PHAHs, both alone and as dilutions of an 18-chemical mixture, on serum T4 concentrations. This exposure paradigm allowed for an economic approach to deriving extensive dose–response information (seven to nine doses per chemical) for 18 individual chemicals. Doses associated with a 30% TH decrease from control (ED30 estimates) were calculated for each chemical, rather than ED50 estimates because some chemicals had asymptotic responses at a 50% decrease. We then tested a mixture of these 18 PHAHs in which the chemical ratios were based on a rough average of concentrations found in breast milk, fish, and other food sources of human exposure (Giesy et al. 1994; Larsen et al. 1994; Patterson et al. 1994; Schecter et al. 1994a, 1994b). Concentrations of the individual chemicals in the undiluted mixture were at least an order of magnitude lower than those found to have significant biologic activity (with the exception of PCB-126, where there was an ~ 16% decrease in T4 at the dose found in the highest concentration of the mixture). Last, the exposures ranged from approximately background human body burdens to body burdens similar to some highly exposed populations (DeVito et al. 1995; Liem et al. 2000; Longnecker et al. 2003; Lorber 2002). We used this approach to decrease uncertainty in low-dose extrapolation in mixture testing (Feron and Groten 2002).
Materials and Methods
Chemicals.
All individual PHAHs were obtained from Accustandard Corporation (New Haven, CT) or Radian Corporation (Austin TX) at purities > 99%. Non-coplanar PCBs were custom synthesized for 99.9% purity. The following chemicals were tested: 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD); 1,2,3,7,8-pentachlorodibenzo-p-dioxin (PCDD); 2,3,7,8-tetrachlorodibenzofuran (TCDF); 1,2,3,7,8-pentachlorodibenzofuran (1-PCDF); 2,3,4,7,8-pentachlorodibenzofuran (4-PCDF); 1,2,3,4,6,7,8,9-octachlorodibenzo-furan (OCDF); 2,4,4′-trichlorobiphenyl (PCB-28); 2,2′,5,5′-tetrachlorobiphenyl (PCB-52); 3,3′,4,4′-tetrachlorobiphenyl (PCB-77); 2,2′,4,5,5′-pentachlorobiphenyl (PCB-101); 2,3,3′,4,4′-pentachlorobiphenyl (PCB-105); 2,3′,4,4′,5-pentachlorobiphenyl (PCB-118); 3,3′,4,4′,5-pentachlorobiphenyl (PCB-126); 2,2′,3,4,4′,5′-hexachlorobiphenyl (PCB-138); 2,2′,4,4′,5,5′-hexachlorobiphenyl (PCB-153); 2,3,3′,4,4′,5-hexachlorobiphenyl (PCB-156); 3,3′,4,4′,5,5′-hexachlorobiphenyl (PCB-169); and 2,2′,3,4,4′,5,5′-heptachloro-biphenyl (PCB-180). The dose ranges and number of dose groups are provided in Table 1. The PHAH mixture was custom synthesized by Cambridge Isotope Laboratories (Andover, MA) and delivered to the U.S. Environmental Protection Agency (EPA) in corn oil at 3 times the highest tested concentration. Concentrations of individual chemicals in the mixture (Table 2) were verified by chromatographic/mass spectrophotometric analyses (Accustandard Corp., New Haven, CT ). The concentrations used in our present study were determined analytically and varied only slightly from the target concentrations. The ratio of chemicals in the mixture was based on the ratios of PHAHs found in breast milk, fish, and other sources of human exposure (Giesy et al. 1994; Larsen et al. 1994; Patterson et al. 1994; Schecter et al. 1994a, 1994b). All dosing solutions were prepared by dilution of the stock solutions with corn oil (Sigma Chemical Co., St. Louis, MO).
Animals and dosing.
We obtained female Long Evans rats (23 days of age) from Charles River Laboratory (Raleigh, NC) and allowed them to acclimate for a minimum of 4 days in an animal facility approved by the Association for Accreditation of Laboratory Animal Care before being treated. Two animals were housed per plastic cage (45 cm × 24 cm × 20 cm), with heat-treated pine shaving bedding. They were maintained at 21 ± 2°C with 50 ± 10% humidity on a 12/12-hr light/dark cycle (lights on 0600–1800 hr). Feed (Purina Rodent Chow 5001, Barnes Supply Co., Durham, NC) and tap water were provided ad libitum. All animal procedures were approved by the U.S. EPA Institutional Animal Care and Use Committee.
We dosed rats by oral gavage for 4 consecutive days with each individual chemical to establish dose–effect functions. This paradigm was previously shown to result in dose-related decreases in T4 concentrations after exposure to PCB-126 and PCB-153 (Craft et al. 2002). Dose ranges and numbers of dose groups are shown in Table 1. We spaced doses at one-half and one-third log units with the aim to have two to three doses with no measurable response, three to four doses closely spaced around the no-effect level, and two to three doses on the descending portion of the dose response. We used two to three separate blocks of animals (separate groups of animals ordered, dosed, and sampled at different dates) to map the dose response for each chemical. Blocks were used to enable testing of large dose ranges. Numbers of animals per group were not similar for each block and ranged from 4 at some of the highest concentrations to 14 in some low-dose and control groups. Control animals (n = 8–14) were dosed with the vehicle (1.0 mL/kg corn oil) only. The mixture study was conducted with serial dilutions of the mixture (n = 12/dose). We recorded body weights daily and adjusted dosing volumes daily. Rats were semirandomly assigned to treatment groups by counterbalancing for body weights. On the day following the last dose, animals were randomly sacrificed by decapitation (no anesthesia) between 0800 and 1000 hr. Trunk blood was harvested and allowed to clot on ice for 45–90 min. Serum was obtained by centrifugation of clotted blood at 2,500 rpm at 4°C for 20 min and stored at −80°C until analysis.
T4 assays.
We measured serum total T4 by standard radioimmunoassay assay kits (Diagnostic Products Corporation, Los Angeles, CA) and analyzed all samples in duplicate. Intraassay and interassay coefficients of variance for all assays were below 10%. Control group means ranged from 41.7 to 55.6 μg/dL, with an average coefficient of variation of 15.3. All data values were standardized to percentage of control for each chemical [(experimental value/control mean) × 100].
Statistical modeling.
The definition of additivity (i.e., zero interaction) used is given by Berenbaum (1985) and can be related to the isobologram for a combination of chemicals (Loewe 1953; Loewe and Muischnek 1926) through the interaction index. That is, in a combination of c (here, c = 18) chemicals, let Eirepresent the concentration or dose of the ith component alone that yields a fixed response (i.e., ED30), y0, and let xirepresent the concentration/dose of the ith component in combination with the c agents that yields the same response. According to this definition, if the substances interact in an additive fashion, then
If the left side of Equation 1, termed the interaction index, is < 1, then a greater-than-additive interaction (e.g., synergism) can be claimed at the combination of interest. If the left side of Equation 1 is > 1, then a less-than-additive interaction (e.g., antagonism) can be claimed at the combination.
The 18 chemicals were combined according to a specified mixing ratio (Table 2) and evaluated experimentally. The mixing ratio was selected based on the ratios found in breast milk, fish, and other sources of human exposure, as described above. The mixing ratio is denoted in terms of the proportion, ai, of each chemical in the mixture (Table 2) such that
The FSCR approach of Gennings et al. (2004) allows for different threshold parameters and range parameters for each chemical and fixed-ratio mixture. The empirical mixture data were modeled (termed the empirical model) using a nonlinear exponential model of the form
where αi+ γi= 100, x is the dose of the ith chemical (i = 1, . . . ,18) or the mixture dose for the mixture ray (i = 19), αiis an unknown parameter defined by the maximum effect for the i th chemical or mixture, βiis an unknown parameter defined by the slope for the i th chemical or mixture, and δiis an unknown parameter defined by the threshold along the i th ray for single chemicals (in terms of dose) or mixtures (in terms of total dose of the mixture). Preliminary inspection of the data provided evidence that the variance of the observed response increased with the mean [i.e., Var(Y ) = τμ for unknown dispersion parameter τ]. Unknown parameters were estimated using the method of maximum quasi likelihood (McCullagh and Nelder 1989). The model in Equation 3 was used to estimate the ED30 (i.e., the dose associated with a 30% decrease from control) for each single chemical. The delta method (Agresti 2002) was used to estimate large sample 95% confidence intervals on the ED30 estimates.
We estimated the mixture dose–response curve (Figure 1, empirical model) using the model in Equation 3 for the specified mixture. The dose–response curve for the mixture assuming additivity (Figure 1, additivity model) was estimated using only the single-chemical dose–response model parameters and the single-chemical data (Equation 3), then predicting along the mixture ray with the constraint of additivity given in Equation 1.
To determine whether there was a statistically significant deviation from additivity, we used a quasi-likelihood ratio test to compare the empirical mixture model to the restricted additivity model based on an F-distribution. The restricted additivity model (Gennings et al. 2004) included only the single-chemical dose–response model parameters but used both the single-chemical and mixture data. We used this restricted model to predict the mean responses for the mixture data using the constraint of additivity given in Equation 1. In addition we compared the predicted responses from the mixture data under the hypothesis of additivity (following the methods in Gennings et al. 2002) to the observed sample means using an F-test (df = 6, 1,305).
Results
We noted no visible signs of toxicity after the short-term PHAH treatments. There were no treatment-related effects on body weight gain. Summary statistics for the single-chemical curve fits, including ED30 estimates and 95% confidence limits, are shown in Table 3. Summary statistics for the individual chemical dose groups can be found in Supplemental Data Table 1 (http://ehp.niehs.nih.gov/docs/2005/8195/supplemental.pdf).
Data modeling with the FSCR method provided maximum effect parameters (asymptotes) and dose threshold parameters for each chemical and for the mixture. Note that the dose–response model for OCDF was reduced to background (100%) because the slope parameter (β) was not significant (p = 0.84) and the maximum effect parameter was not different from 100%. The estimates for the maximum effect parameters (α) for the single chemicals clustered into three groups, with maximums at 14, 31, and 50% of control (Table 4). The data were also modeled with a single-chemical-required (SCR) approach that requires similar asymptotes (Casey et al. 2004). This model proved to be inadequate with significant lack of fit because of clearly different asymptotes (p < 0.0001). The FSCR model proved more appropriate as it allowed for three different asymptotic levels, dose thresholds, and no overall lack of fit (p > 0.05).
Table 5 presents summary statistics for the mixture data. The data reveal a mixture-dose–dependent decrease in T4 concentrations that produced a maximal decrease of approximately 50%. The experimental mixture data and the fits of the empirical and additivity models to the mixture data are shown in Figure 1. Comparison of the dose–response curve for the mixture under the hypothesis of additivity (Figure 1, dashed line) to the fit to the empirical data (Figure 1, solid line) illustrate the dose-dependent nature of the nonadditive effects of the mixture (Figure 1). A quasi-likelihood ratio test rejected the hypothesis of additivity (p < 0.001). Table 6 lists the results of an overall (6 df) test of additivity where the null hypothesis is that the true mean is equivalent to the predicted mean from the FSCR model. From this test, there is statistically significant evidence of departure from additivity (p < 0.001). Associated with this test are the six comparisons at each mixture-dose group (Table 6). The T4 mean for the mixtures at the three highest mixture-dose groups (667, 1,335, and 2,002 μg/kg/day) were each significantly different from that predicted under additivity. Because the sample means are below that predicted under additivity, there is evidence of an interaction (greater than additivity) at these dose groups. The difference between the additivity model and empirical data (Figure 1) at the three highest mixture doses (i.e., area of maximal difference) was approximately 15% in terms of T4 concentration or 2.5-fold on a microgram per kilogram per day dose basis. There was no evidence for significant departure from additivity at the three lowest doses of the mixture (Table 6).
Discussion
The present study tested the hypothesis that a mixture of TDCs affect T4 concentrations in a dose-additive manner. We designed the mixture so that highest mixture-dose levels of the individual chemicals were at or below their no observed effect levels. The FSCR additivity model analyses demonstrate cumulative effects of low doses of the mixture and synergistic cumulative effects of the highest dosages of the mixture. These data advocate consideration of cumulative risk approaches when assessing the risk of exposures to chemical mixtures that contain TDCs.
The single-chemical and mixture data were modeled successfully using the FSCR model. Results demonstrate a very wide range of effective doses of PHAHs that decrease TH concentrations. These findings confirm previous work demonstrating that short-term exposure to TCDD (Craft et al. 2002), and some individual PCB congeners for example, PCB congeners 47, 95, 101, and 153 (Craft et al. 2002; Khan et al. 2002; Saeed and Hansen 1997) cause hypothyroxinemia in the rat. Our present work expands these findings by providing dose–response data and relative potencies for 2 dioxins, 4 furans, and 12 PCB congeners. OCDF was not effective at the doses used in this animal dose model. This was expected because of the limited absorption of this fully chlorinated dibenzofuran (Birnbaum and Couture 1988; DeVito et al. 1998). In addition the ED30 estimates provide a basis for establishing relative potency values for these chemicals.
Analyses of the mixtures data demonstrated a dose-dependent synergy. The additivity model underestimated the actual toxic effect of the mixture at the three highest doses tested (Figure 1). Effects of the three lowest doses of the mixture were not significantly different than that predicted by the additivity model. These conclusions are based on the use of the FSCR method (Gennings et al. 2004). These data were also analyzed using an SCR method (Gennings et al. 2002). Although the SCR model provided significant evidence of a greater-than-additive effect (data not shown), this model was not appropriate for use with these data because of significant lack of fit to the data. The SCR model assumes a similar asymptote for all single chemicals and the mixture, a condition not satisfied in the present data set. Use of the FSCR model allowed for multiple asymptotic levels and dose thresholds and resulted in a model with no overall lack of fit.
Three conclusions are apparent from these data. The first is that exposure to the 18 chemical mixture results in dose-dependent greater-than-additive effects on T4 concentrations at the highest mixture doses. This conclusion is supported by the FSCR analysis. The second conclusion is that although the greater-than-additive effects are statistically significant, the magnitude of underestimation of the experimental data (Figure 1) by the additivity model (Figure 1) is not large. On a dose basis, the underestimation is about 2.5-fold for the three highest doses of the mixture (Figure 1). This suggests that, even in the high mixture-dose region, the effects of this mixture are predicted by additivity with a fair degree of accuracy. The third conclusion is that departure from additivity was not detected in the low-dose region. Although this suggests that dose additivity predicts effects on T4 at low exposures, it is tempered by a presumed low statistical power to detect differences in this area of the dose response.
A significant finding in the present experiment is that the mixture actually caused decreases in T4 concentrations. This occurred even though the individual chemical concentrations in the mixture were below effective doses. For example, at the second highest mixture dose there was a 38% decrease in T4. The individual dose of PCB-153 at this mixture dose was approximately 254 μg/kg/day. The lowest effective dose of PCB-153 administered alone is much greater than 2,000 μg/kg/day. This relationship was similar for all the chemicals in the mixture with one exception, PCB-126. The dose of PCB-126 in the highest dose of the mixture caused about a 16% decrease in T4. These data clearly demonstrate the principle that simple mathematical addition of effects (i.e., effect addition) of individual chemicals will not predict the effects of these TDCs in a mixture.
The biologic reasons for the greater-than-additive effect of this mixture are currently unknown. Risk assessment approaches to additivity assume, where data are lacking otherwise, that chemicals with similar modes of action act in a dose-additive fashion (U.S. EPA 1986, 2000). Although all the chemicals used here decrease circulating T4 concentrations, they may do so via a number of different mechanisms. One postulated mechanism for the reduction in T4 concentrations is the up-regulation of hepatic UGT isoforms that glucuronidate T4, leading to biliary elimination (Capen 1997; DeVito et al. 1999; Hill et al. 1998; McClain et al. 1989). Evidence suggests that UGT1A1 and UGT1A6 are responsible for T4 glucuronidation in the rat (Vansell and Klaassen 2002; Visser et al. 1993). These UGT isoforms are induced by aryl hydrocarbon receptor (AhR), constitutive androstane receptor (CAR), and pregnane-X receptor (PXR) agonists. The dioxins, furans, and coplanar PCBs (e.g., PCB-77, PCB-126) all activate AhR (Wilson and Safe 1998), whereas the more non-coplanar PCBs (e.g., PCB-52, PCB-138, PCB-153) act via CAR/PXR pathways (Connor et al. 1995; Tabb et al. 2004). Some of the chemicals tested (e.g., PCB-105, PCB-118) are agonists for AhR, PXR, and CAR. Activation of these UGTs through the different nuclear receptors may play a role in the synergistic effects. Differential regulation of microsomal enzymes that glucuronidate T4 versus T3 (triiodothyronine) may also be responsible (Hood and Klaassen 2000a). There are a number of other postulated mechanisms for altering circulating and tissue levels of THs. Hydroxylated metabolites of PCBs displace T4 from transthyretin, a major serum transport protein in rats (Brouwer et al. 1998). This mechanism has been hypothesized to decrease bound T4, resulting in greater uptake, catabolism, and elimination of T4 (Van den Berg et al. 1991). PCBs also alter deiodinases and therefore iodination of THs (Hood and Klaassen 2000b; Morse et al. 1993). There is some evidence that PCBs increase uptake of T4 into the liver (Martin 2002), possibly by altering thyroid transporters (Guo et al. 2002). In addition, Khan and Hansen (2003) and colleagues have demonstrated decreased pituitary sensitivity to thyroid-stimulating hormone by two PCB congeners. Therefore, the synergistic effect may be the result of activation of multiple pathways by the mixture, with the measured effect, T4, a common downstream end point for these pathways.
The curve fits to the individual chemical data revealed three levels of maximum efficacy (Table 4). Because of the limited number of chemicals, it is difficult to quantitatively describe the structure–activity relationship for maximal T4 decreases. In addition the dose–response determinations were not designed to allow prediction of the asymptotic efficacy but instead aimed to characterize the low end of the dose–response functions. In some cases the maximal efficacy was driven by the highest dose tested, which did not demonstrate a clear maximal effect (e.g., PCB-28, PCB-52, PCB-169). The data do support, with a number of exceptions, a rough separation of chemicals into the more dioxin-like chemicals at the 50% point, and mono- and di-ortho substituted chemicals having an asymptote at 14%. A likely explanation for the different efficacies is that the PHAHs act through a variety a mechanisms, as discussed above, and the interaction of these mechanisms differentially affects T4 levels.
The significance of these findings for environmental exposures is tempered by some uncertainty. In our present study we used a weanling animal model with a short (i.e., 4-day) exposure duration. Short exposure durations, coupled with differences in half-lives of the chemicals in the mixture that vary from a few weeks to many months (Van den Berg et al. 1994), yield potential pharmacokinetic differences that may confound extrapolation of these results. Pharmacokinetic differences between short-term and steady-state exposures may also include differences in saturation of induction and metabolite generation. Thus, extrapolation of our present findings to chronic exposures should be moderated by these uncertainties.
Extrapolation of our present work in rats to humans is tempered by the uncertainty in how the mode(s) of action of the TDCs may differ between species. Current hypotheses on the mechanisms by which PHAHs decrease T4 include up-regulation of hepatic UGTs and sulfotransferases, direct effects on the thyroid gland, and displacement of T4 from serum transport proteins (Brouwer et al. 1998). Cross-species extrapolation of these mechanisms is difficult (Crofton 2004). In addition one must consider the degree of TH disruption that will lead to adverse outcomes. Small decreases (~25%) in maternal T4 during the early fetal period will lead to adverse neurofunctional outcomes (i.e., IQ scores) in humans (Haddow et al. 2002; Morreale de Escobar et al. 2000). Limited data in animals suggest that T4 decreases need to exceed 50% before adverse outcomes can be detected (Crofton 2004).
A limited number of studies have examined the effects of complex mixtures of endocrine-disrupting chemicals (EDCs) (Desaulniers et al. 2003; Tinwell and Ashby 2004; Wade et al. 2002). Desaulniers et al. (2003) examined the effects of a mixture of 16 coplanar PCBs, PCDDs, and PCDFs on T4 concentrations in neonatal rats. Decreases in T4 were associated with dioxin equivalents using the toxic equivalency factor methodology (Desaulniers et al. 2003; Van den Berg et al. 1998). Consistent with our present findings, Wade et al. (2002) found that effects on thyroid histopathology and hormones were underpredicted based on additivity of published health advisories (e.g., RfDs and ADIs). Evaluation of different models for determining the effects of a mixture of seven EDCs on uterotrophic responses led to a conclusion that the most expedient method is to bioassay the mixture rather than test individual chemicals (LeBlanc and Olmstead 2004; Tinwell and Ashby 2004). These studies lack, either by study design or statistical approach, the ability to test for additivity. The present work expands the previous work by applying a rigorous statistical analysis to test for additivity.
Conclusions
The present work demonstrates that the cumulative effect of a mixture of TDCs is predicted by additivity at low doses and synergy at high doses. These data suggest that low doses of heterogeneous TDCs that alter thyroid homeostasis should be considered together when calculating the risk of exposures to mixtures. Future work should endeavor to expand these conclusions to low-dose chronic exposures and broaden testing of mixtures to include chemicals from diverse classes of thyroid disruptors such as TH synthesis inhibitors.
Supplementary Material
Supplemental Data Table 1 Supplemental Data Table 1 lists the chemicals tested, doses (μg/kg/day), group mean serum total thyroxine (T4) concentrations expressed as percentage of control, standard deviations, and group sample sizes. These data are available on the EHP website (http://ehp.niehs.nih.gov/docs/2005/8195/supplemental.pdf). Raw data files can be obtained by contacting the corresponding author.
We thank T. Grim at Cambridge Isotope Laboratories for help with the custom synthesis of the mixture. This work would not have been possible without the invaluable technical assistance of T. Zhou and D. Ross. G. We acknowledge G. LeBlanc and S. Padilla for commenting on a previous version of the manuscript.
This manuscript has been reviewed by the National Health and Environmental Effects Research Laboratory, U.S. EPA, and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Figure 1 Predicted effects on serum total T4 from the single-chemical data in an additivity model and empirical effects of the PHAH mixture. Results demonstrate a significant deviation from additivity at the three highest mixture doses. The effects of the lower mixture doses were not significantly different than that predicted by additivity. The inset illustrates the same data plotted as log-dose.
Table 1 Chemicals tested, dose ranges, and number of doses, for individual chemicals.
Chemical Dose range (μg/kg/day) No. of dosesa
TCDD 0.0001–10 10
PCDD 0.003–10 10
TCDF 0.3–100 7
1-PCDF 0.03–100 7
4-PCDF 0.03–90 9
OCDF 0.1–300 8
PCB-28 100–90,000 9
PCB-52 100–90,000 9
PCB-77 100–30,000 8
PCB-101 50–30,000 9
PCB-105 90–90,000 8
PCB-118 10–10,000 9
PCB-126 0.001–100 10
PCB-138 100–90,000 9
PCB-153 100–90,000 9
PCB-156 10–10,000 8
PCB-169 1–1,000 8
PCB-180 100–90,000 8
a Includes a control group.
Table 2 Chemical composition of the mixture.
Chemical Concentrationa (μg/mL) Ratio (TCDD) Ratio (total mass)
TCDD 0.013 1.0 0.000007
PCDD 0.013 1.0 0.000007
TCDF 0.019 1.4 0.000010
1-PCDF 0.006 0.4 0.000003
4-PCDF 0.026 1.9 0.000013
OCDF 0.065 4.6 0.000032
PCB-28 78.600 5,605.3 0.039237
PCB-52 155.200 11,074.7 0.077523
PCB-77 2.000 141.1 0.000988
PCB-101 153.800 10,973.4 0.076814
PCB-105 76.700 5,468.9 0.038282
PCB-118 381.100 27,186.0 0.190302
PCB-126 0.610 43.1 0.000302
PCB-138 380.900 27,168.7 0.190181
PCB-153 382.200 27,265.9 0.190861
PCB-156 13.100 934.4 0.006541
PCB-169 0.400 28.1 0.000197
PCB-180 377.900 26,957.1 0.188700
a Chemical concentration in the highest dose of the mixture administered.
Table 3 Summary statistics for individual chemicals, and the ED30 and 95% confidence intervals for each chemical.
Chemical ED30 (μg/kg/day) 95% Confidence interval αAsymptote estimatea
TCDD 0.15 (0.08, 0.22) 50
PCDD 1.51 (1.10, 1.92) 31
TCDF 4.65 (1.90, 7.40) 50
1-PCDF 15.6 (10.17, 21.01) 50
4-PCDF 27.5 (17.05, 38.01) 50
OCDF — — —
PCB-28 76,103.0 (50,142, 102,064) 50
PCB-52 33,025.0 (20,958, 45,092) 50
PCB-77 852.0 (655, 1,049) 31
PCB-101 4,833.0 (3,819, 5,847) 31
PCB-105 1,031.0 (861, 1,200) 14
PCB-118 1,289.0 (1,103, 1,475) 14
PCB-126 1.33 (0.77, 1.88) 50
PCB-138 8,001.0 (6,692, 9,310) 14
PCB-153 12,696.0 (10,659, 14,732) 14
PCB-156 760.0 (629, 891) 14
PCB-169 227.0 (167, 286) 31
PCB-180 30,541.0 (23,122, 37,960) 31
a Percentage of control (e.g., α= 14 represents an 86% decrease in T4 concentration relative to the control mean).
Table 4 Parameter estimates from the FSCR model.
Parameter Estimate SE p-Value
α1,2,6,12,13,16,17 50.26 1.45 < 0.001
α3,10,11,14,15 30.74 1.95 < 0.001
α4,5,7,8,9 14.33 1.03 < 0.001
αmix 42.29 14.78 0.004
β1 (1-PCDF) −0.0608 0.0164 < 0.001
β2 (4-PCDF) −0.0378 0.0142 0.008
β3 (PCB-101) −0.000119 0.000020 < 0.001
β4 (PCB-105) −0.000833 0.000171 < 0.001
β5 (PCB-118) −0.000696 0.000109 < 0.001
β6 (PCB-126) −0.719000 0.225 0.001
β7 (PCB-138) −0.000054 0.000006 < 0.001
β8 (PCB-153) −0.000034 0.000003 < 0.001
β9 (PCB-156) −0.000567 0.000061 < 0.001
β10 (PCB-169) −0.002616 0.000549 < 0.001
β11 (PCB-180) −0.000021 0.000005 < 0.001
β12 (PCB-28) −0.000012 0.000003 < 0.001
β13 (PCB-52) −0.000028 0.000006 < 0.001
β14 (PCB-77) −0.000666 0.000102 < 0.001
β15 (PCDD) −0.374900 0.0639 < 0.001
β16 (TCDD) −6.505000 2.414 0.007
β17 (TCDF) −0.331000 0.274 0.228
θmix −0.000872 0.000467 0.062
δ1 0.389 1.500 0.795
δ2 3.094 3.927 0.431
δ3 76.410 398.449 0.848
δ4 513.1 147.9 0.001
δ5 669.5 147.5 < 0.001
δ6 0.043 0.130 0.742
δ10 9.812 25.104 0.696
δ11 3,227.3 5,581.7 0.563
δ16 0.004 0.020 0.855
δ17 1.859 0.977 0.057
δmix 49.5 74.3 0.506
Table 5 Effects of PHAH mixture on serum T4 concentrations.
Mixture dose (μg/kg/day) Test solution (% stock) T4 mean (% control ± SD) Sample size
0 — 100.0 ± 11.8 12
20.0 0.33 106.8 ± 22.7 12
66.7 1.1 98.5 ± 21.7 12
200.3 3.3 94.6 ± 22.9 12
667.5 11 73.7 ± 14.6 12
1,335.1 22 62.9 ± 10.8 12
2,002.6 33 52.2 ± 15.1 12
Table 6 Test results for the hypothesis that mean T4 values for the mixture dose are equal to those predicted under the additivity model.
Statistical test Mixture dose (μg/kg/day) Statistic p-Value
Overall F-test (df 6, 1,305) — 3.43 0.002
Individual F-tests (df 1, 1,305) 20.0 1.65 0.200
66.7 0.03 0.862
200.3 0.21 0.647
667.5 8.76 0.003*
1,335.1 7.91 0.005*
2,002.6 10.10 0.002*
*Dose groups where the mean T4 response is significantly different (p > 0.05) from that predicted under additivity.
==== Refs
References
Agresti A 2002. Categorical Data Analysis. 2nd ed. New York:John Wiley & Sons.
Berenbaum MC 1985 The expected effect of a combination of agents: the general solution J Theor Biol 114 3 413 431 4021503
Birnbaum LS Couture LA 1988 Disposition of octa-chlorodibenzo-p -dioxin (OCDD) in male rats Toxicol Appl Pharmacol 93 1 22 30 3353999
Brouwer A Morse DC Lans MC Schuur AG Murk AJ Klasson-Wehler E 1998 Interactions of persistent environmental organohalogens with the thyroid hormone system: mechanisms and possible consequences for animal and human health Toxicol Ind Health 14 1–2 59 84 9460170
Capen C 1998 Correlation of mechanistic data and histopathology in the evaluation of selected toxic endpoints of the endocrine system Toxicol Lett 102–103 405 409
Capen CC 1997 Mechanistic data and risk assessment of selected toxic end points of the thyroid gland Toxicol Pathol 25 1 39 48 9061850
Casey M Gennings C Carter JH Jr Moser VC Simmons JE 2004 Detecting interaction(s) and assessing the impact of component subsets in a chemical mixture using fixed-ratio mixture ray designs J Agric Biol Environ Stat 9 3 339 361
Connor K Safe S Jefcoate CR Larsen M 1995 Structure-dependent induction of CYP2B by polychlorinated biphenyl congeners in female Sprague-Dawley rats Biochem Pharmacol 50 11 1913 1920 8615872
Craft ES DeVito MJ Crofton KM 2002 Comparative responsiveness of hypothyroxinemia and hepatic enzyme induction in Long-Evans rats versus C57BL/6J mice exposed to TCDD-like and phenobarbital-like polychlorinated biphenyl congeners Toxicol Sci 68 2 372 380 12151633
Crofton KM 2004 Developmental disruption of thyroid hormone: correlations with hearing dysfunction in rats Risk Anal 24 6 1665 1671 15660619
Daston GP Cook JC Kavlock RJ 2003 Uncertainties for endocrine disrupters: our view on progress Toxicol Sci 74 2 245 252 12730617
Desaulniers D Leingartner K Musicki B Yagminas A Xiao GH Cole J 2003 Effects of postnatal exposure to mixtures of non-ortho-PCBs, PCDDs, and PCDFs in prepubertal female rats Toxicol Sci 75 2 468 480 12883090
DeVito M Biegel L Brouwer A Brown S Brucker-Davis F Cheek AO 1999 Screening methods for thyroid hormone disruptors Environ Health Perspect 107 407 415 10210697
DeVito MJ Birnbaum LS Farland WH Gasiewicz TA 1995 Comparisons of estimated human body burdens of dioxinlike chemicals and TCDD body burdens in experimentally exposed animals Environ Health Perspect 103 820 831 7498094
DeVito MJ Ross DG Dupuy AE Jr Ferrario J McDaniel D Birnbaum LS 1998 Dose-response relationships for disposition and hepatic sequestration of polyhalogenated dibenzo-p -dioxins, dibenzofurans, and biphenyls following subchronic treatment in mice Toxicol Sci 46 2 223 234 10048125
Feron VJ Groten JP 2002 Toxicological evaluation of chemical mixtures Food Chem Toxicol 40 6 825 839 11983277
Gennings C Carter WH Jr Campain JA Bae D Yang RSH 2002 Statitical analysis of interactive cytotoxicity in human epidermal keratinocytes following exposure to a mixture of four metals J Agric Biol Environ Sci 7 1 58 73
Gennings C Carter WH Jr Carney EW Charles GD Gollapudi BB Carchman RA 2004 A novel flexible approach for evaluating fixed ratio mixtures of full and partial agonists Toxicol Sci 80 1 134 150 15084752
Giesy JP Verbrugge DA Othout RA Bowerman WW Mora MA Jones PD 1994 Contaminants in fishes from Great Lakes-influenced sections and above dams of three Michigan rivers. I: Concentrations of organo chlorine insecticides, polychlorinated biphenyls, dioxin equivalents, and mercury Arch Environ Contam Toxicol 27 2 202 212 8060164
Guo GL Choudhuri S Klaassen CD 2002 Induction profile of rat organic anion transporting polypeptide 2 (oatp2) by prototypical drug-metabolizing enzyme inducers that activate gene expression through ligand-activated transcription factor pathways J Pharmacol Exp Ther 300 1 206 212 11752118
Haddow JE Palomaki GE Williams J 2002 Thyroid-stimulating-hormone concentrations and risk of hypothyroidism Lancet 360 9350 2081 2082 12504442
Hill RN Crisp TM Hurley PM Rosenthal SL Singh DV 1998 Risk assessment of thyroid follicular cell tumors Environ Health Perspect 106 447 457 9681971
Hood A Klaassen CD 2000a Differential effects of microsomal enzyme inducers on in vitro thyroxine (T(4)) and triiodothyronine (T(3)) glucuronidation Toxicol Sci 55 1 78 84 10788562
Hood A Klaassen CD 2000b Effects of microsomal enzyme inducers on outer-ring deiodinase activity toward thyroid hormones in various rat tissues Toxicol Appl Pharmacol 163 3 240 248 10702363
Hurley PM 1998 Mode of carcinogenic action of pesticides inducing thyroid follicular cell tumors in rodents Environ Health Perspect 106 437 445 9681970
International Programme on Chemical Saftety 2002. Global Assessment of the State-of-the-Science of Endocrine Disruptors. WHO/PCS/EDC/02.2 (Damstra T, Barlow S, Bergman A, Kavlock R, Van Der Kraak G, eds). Geneva: World Health Organization.
Khan MA Hansen LG 2003 Ortho -substituted polychlorinated biphenyl (PCB) congeners (95 or 101) decrease pituitary response to thyrotropin releasing hormone Toxicol Lett 144 2 173 182 12927361
Khan MA Lichtensteiger CA Faroon O Mumtaz M Schaeffer DJ Hansen LG 2002 The hypothalamo-pituitary-thyroid (HPT) axis: a target of nonpersistent ortho-substituted PCB congeners Toxicol Sci 65 1 52 61 11752685
Larsen BR Turrio-Baldassarri L Nilsson T Iacovella N Di Domenico A Montagna M 1994 Toxic PCB congeners and organochlorine pesticides in Italian human milk Ecotoxicol Environ Saf 28 1 1 13 7523063
LeBlanc GA Olmstead AW 2004 Evaluating the toxicity of chemical mixtures [Letter] Environ Health Perspect 112 A729 A730 15345361
Liem AK Furst P Rappe C 2000 Exposure of populations to dioxins and related compounds Food Addit Contam 17 4 241 259 10912239
Loewe S 1953 The problem of synergism and antagonism of combined drugs Arzneimittelforschung 3 6 285 290 13081480
Loewe S Muischnek H 1926 Effect of combinations: mathematical basis of problem Naunyn Schmiedebergs Arch Exp Pathol Pharmakol 114 313 326
Longnecker MP Wolff MS Gladen BC Brock JW Grandjean P Jacobson JL 2003 Comparison of polychlorinated biphenyl levels across studies of human neurodevelopment Environ Health Perspect 111 65 70 12515680
Lorber M 2002 A pharmacokinetic model for estimating exposure of Americans to dioxin-like compounds in the past, present, and future Sci Total Environ 288 1–2 81 95 12013550
Martin LA 2002. Differential effects of polychlorinated biphenyl (PCB) mixtures and congeners on the disposition of thyroxine (T4) in rats [PhD Dissertation]. Piscataway, NJ:Rutgers University and Robert Wood Johnson Medical School.
McClain RM 1995 Mechanistic considerations for the relevance of animal data on thyroid neoplasia to human risk assessment Mutat Res 333 1–2 131 142 8538620
McClain RM Levin AA Posch R Downing JC 1989 The effect of phenobarbital on the metabolism and excretion of thyroxine in rats Toxicol Appl Pharmacol 99 2 216 228 2734788
McCullagh P Nelder JA 1989. Generalized Linear Models. 2nd ed. London:Chapman & Hall.
Morreale de Escobar G Obregon MJ Escobar del Rey F 2000 Is neuropsychological development related to maternal hypothyroidism or to maternal hypothyroxinemia? J Clin Endocrinol Metab 85 11 3975 3987 11095417
Morse DC Groen D Veerman M van Amerongen CJ Koeter HB Smits van Prooije AE 1993 Interference of polychlorinated biphenyls in hepatic and brain thyroid hormone metabolism in fetal and neonatal rats Toxicol Appl Pharmacol 122 1 27 33 8378931
Oppenheimer JH Bernstein G Surks MI 1968 Increased thyroxine turnover and thyroidal function after stimulation of hepatocellular binding of thyroxine by phenobarbital J Clin Invest 47 6 1399 1406 5653217
Patterson DG Jr Todd GD Turner WE Maggio V Alexander LR Needham LL 1994 Levels of non-ortho -substituted (coplanar), mono- and di-ortho -substituted polychlorinated biphenyls, dibenzo-p -dioxins, and dibenzofurans in human serum and adipose tissue Environ Health Perspect 102 suppl 1 195 204 8187709
Saeed A Hansen LG 1997 Morphometric changes in the prepubertal female rat thyroid gland following acute exposure to 2,2′,4,4′-tetrachlorobiphenyl and Aroclor 1242 J Toxicol Environ Health 51 5 503 513 9233383
Schecter A Stanley J Boggess K Masuda Y Mes J Wolff M 1994a Polychlorinated biphenyl levels in the tissues of exposed and nonexposed humans Environ Health Perspect 102 suppl 1 149 158 8187704
Schecter A Startin J Wright C Kelly M Papke O Lis A 1994b Congener-specific levels of dioxins and dibenzo-furans in U.S. food and estimated daily dioxin toxic equivalent intake Environ Health Perspect 102 962 966 9738211
Tabb MM Kholodovych V Grun F Zhou C Welsh WJ Blumberg B 2004 Highly chlorinated PCBs inhibit the human xenobiotic response mediated by the steroid and xenobiotic receptor (SXR) Environ Health Perspect 112 163 169 14754570
Teuschler L Klaunig J Carney E Chambers J Conolly R Gennings C 2002 Support of science-based decisions concerning the evaluation of the toxicology of mixtures: a new beginning Regul Toxicol Pharmacol 36 1 34 39 12383716
Tinwell H Ashby J 2004 Sensitivity of the immature rat uterotrophic assay to mixtures of estrogens Environ Health Perspect 112 575 582 15064164
U.S. EPA 1986 Guidelines for the health risk assessment of chemical mixtures Fed Reg 51 34014 34025
U.S. EPA 2000. Supplementary Guidance for Conducting Health Risk Assessment of Chemical Mixtures. Washington, DC:U.S. Environmental Protection Agency.
Van den Berg KJ van Raaij JA Bragt PC Notten WR 1991 Interactions of halogenated industrial chemicals with transthyretin and effects on thyroid hormone levels in vivo Arch Toxicol 65 1 15 19 2043046
Van den Berg M Birnbaum L Bosveld AT Brunstrom B Cook P Feeley M 1998 Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife Environ Health Perspect 106 775 792 9831538
Van den Berg M De Jongh J Poiger H Olson JR 1994 The toxicokinetics and metabolism of polychlorinated dibenzo-p -dioxins (PCDDs) and dibenzofurans (PCDFs) and their relevance for toxicity Crit Rev Toxicol 24 1 1 74 8172651
Vansell NR Klaassen CD 2002 Increase in rat liver UDP-glucuronosyltransferase mRNA by microsomal enzyme inducers that enhance thyroid hormone glucuronidation Drug Metab Dispos 30 3 240 246 11854140
Visser TJ Kaptein E van Raaij JA Joe CT Ebner T Burchell B 1993 Multiple UDP-glucuronyltransferases for the glucuronidation of thyroid hormone with preference for 3,3′,5′-triiodothyronine (reverse T3) FEBS Lett 315 1 65 68 8416813
Wade MG Parent S Finnson KW Foster W Younglai E McMahon A 2002 Thyroid toxicity due to subchronic exposure to a complex mixture of 16 organochlorines, lead, and cadmium Toxicol Sci 67 2 207 218 12011480
Wilson CL Safe S 1998 Mechanisms of ligand-induced aryl hydrocarbon receptor-mediated biochemical and toxic responses Toxicol Pathol 26 5 657 671 9789953
Wolff J 1998 Perchlorate and the thyroid gland Pharmacol Rev 50 1 89 105 9549759
|
16263510
|
PMC1310917
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 21; 113(11):1549-1554
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8195
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8006ehp0113-00155516263511ResearchUltrafine Particles Cross Cellular Membranes by Nonphagocytic Mechanisms in Lungs and in Cultured Cells Geiser Marianne 1Rothen-Rutishauser Barbara 1Kapp Nadine 1Schürch Samuel 12Kreyling Wolfgang 3Schulz Holger 3Semmler Manuela 3Hof Vinzenz Im 4Heyder Joachim 3Gehr Peter 11 Institute for Anatomy, University of Bern, Bern, Switzerland2 Department of Physiology and Biophysics, Faculty of Medicine, The University of Calgary, Calgary, Alberta, Canada3 GSF-National Research Center for Environment and Health, Institute for Inhalation Biology, Neuherberg/Munich, Germany4 Institute of Pathophysiology, University of Bern, Bern, SwitzerlandAddress correspondence to M. Geiser, Institute of Anatomy, University of Bern, Baltzerstrasse 2, CH-3000 Bern 9, Switzerland. Telephone: 41-31-631-8475. Fax: 41-31-631-3807. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 26 5 2005 113 11 1555 1560 9 2 2005 26 5 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. High concentrations of airborne particles have been associated with increased pulmonary and cardiovascular mortality, with indications of a specific toxicologic role for ultrafine particles (UFPs; particles < 0.1 μm). Within hours after the respiratory system is exposed to UFPs, the UFPs may appear in many compartments of the body, including the liver, heart, and nervous system. To date, the mechanisms by which UFPs penetrate boundary membranes and the distribution of UFPs within tissue compartments of their primary and secondary target organs are largely unknown. We combined different experimental approaches to study the distribution of UFPs in lungs and their uptake by cells. In the in vivo experiments, rats inhaled an ultrafine titanium dioxide aerosol of 22 nm count median diameter. The intrapulmonary distribution of particles was analyzed 1 hr or 24 hr after the end of exposure, using energy-filtering transmission electron microscopy for elemental microanalysis of individual particles. In an in vitro study, we exposed pulmonary macrophages and red blood cells to fluorescent polystyrene microspheres (1, 0.2, and 0.078 μm) and assessed particle uptake by confocal laser scanning microscopy. Inhaled ultrafine titanium dioxide particles were found on the luminal side of airways and alveoli, in all major lung tissue compartments and cells, and within capillaries. Particle uptake in vitro into cells did not occur by any of the expected endocytic processes, but rather by diffusion or adhesive interactions. Particles within cells are not membrane bound and hence have direct access to intracellular proteins, organelles, and DNA, which may greatly enhance their toxic potential.
aerosolerythrocyteslungsmacrophagesmicroscopynanoparticlesratssurfactant
==== Body
High concentrations of airborne particles have been associated with increased pulmonary and cardiovascular mortality, with indications of a specific toxicologic role for ultrafine particles (UFPs; particles with diameters < 0.1 μm) (Peters et al. 1997). UFPs may induce inflammatory and prothrombotic responses, promoting atherosclerosis, thrombogenesis, and the occurrence of other cardiovascular events (Schulz et al. 2005). Human data suggest that inhaled UFPs influence lung physiology (Pietropaoli et al. 2004). UPFs may also affect the autonomic nervous system or act directly on cells in various organs and induce mutations (Harder et al. 2005; Samet et al. 2004). After exposure of the respiratory system to UFPs, the UFPs may appear within hours in many compartments of the body, including the liver, heart, and nervous system (Brown et al. 2002; Kreyling et al. 2002; Oberdörster et al. 2004).
UFPs are formed by gas-to-particle conversion or by incomplete fuel combustion. Despite considerable efforts to reduce air pollution, the environmental burden by UFPs may have increased rather than decreased over time (Kreyling et al. 2003). Moreover, the fast-growing nanotechnology industry generates new UFPs daily, which may become aerosolized at some stage and may present additional health risks. UFPs possess increased toxicity compared with larger particles composed of the same materials (Ferin et al. 1992). Their environmental burden is characterized by high number concentrations but low mass concentrations. Thus, a relatively large surface area per unit mass facilitates adsorption of various organic compounds from the ambient air and enhances interaction with biological molecules within the organism.
Deposition of UFPs in the respiratory system is caused by diffusional displacement. Depending on particle size, deposition occurs efficiently in the nose, the conducting airways, and the alveoli. Although particles with diameters > 1 μm usually remain on the epithelial surface upon their deposition (Gehr et al. 1990; Geiser et al. 2003; Schürch et al. 1990) and are subjected to clearance by cough, mucociliary transport, and/or phagocytosis by macrophages, UFPs seem to penetrate the boundary membranes of the lungs rapidly—a unique feature for insoluble particles (Brown et al. 2002; Kreyling et al. 2002; Oberdörster et al. 2002). In addition, transport across the olfactory epithelium and accumulation in the brain were reported for various UFP types (Oberdörster 2004). In vitro experiments revealed penetration of UFPs into mitochondria of macrophages and epithelial cells that was associated with oxidative stress and mitochondrial damage (Li et al. 2003).
Because everyone on earth inevitably inhales thousands to millions of UFPs with each breath, it is important to assess health risks by UFP air pollution. The costs of actions to be taken to reduce ambient aerosol particles are high and will affect the economy greatly, presenting an urgent need to clarify the fate of inhaled UFPs. To date, the mechanisms by which UFPs penetrate boundary membranes and the distribution of UFPs within tissue compartments of their primary and secondary target organs are largely unknown.
This study is the first to investigate the distribution of inhaled UFPs within lungs at the individual particle level and combines different experimental approaches—an in vivo inhalation study in rats and an in vitro cell exposure study on pulmonary macrophages and red blood cells (RBCs).
In the in vivo experiments, rats inhaled an ultrafine titanium dioxide aerosol of 22 nm count median diameter (CMD) during 1 hr, resulting in a deposition of 4–5 μg TiO2 per animal. The intrapulmonary distribution of deposited particles was analyzed immediately or 24 hr after the end of exposure, using energy-filtering transmission electron microscopy (EFTEM) to allow elemental microanalysis of individual particles (Kapp et al. 2004).
In the in vitro study, we exposed cultured porcine pulmonary macrophages and human RBCs to fluorescent polystyrene microspheres with diameters of 1, 0.2, and 0.078 μm and assessed particle uptake by confocal laser scanning microscopy (CLSM).
Materials and Methods
Animals.
The animal experiments were conducted under federal guidelines for the use and care of laboratory animals (German Animal Protection Law) and were approved by the District of Upper Bavaria (Approval No. 211-2531-108/99) and by the GSF Institutional Animal Care and Use Committee, as well as in accordance with the Swiss Federal Act on Animal Protection and the Swiss Animal Protection Ordinance. Ten young, adult, male WKY/NCrl BR rats [body weight (bw) 250 ± 10 g (mean ± SD; Charles River, Sulzfeld, Germany] were housed under standard conditions, with access to food and water ad libitum, in a room controlled for humidity (55% relative humidity) and temperature (22°C) and with lighting on a 12-hr day/night cycle. Animals were anesthetized by intramuscular injection of a mixture of medetomidine (Domitor; 150 μg/100 g bw; Pfizer GmbH, Karlsruhe, Germany), midazolam (Dormicum; 0.2 mg/100 g bw; Hoffmann-La Roche AG, Grenzach-Wyhlen, Germany), and Fentanyl (0.5 μg/100 g bw; Janssen-Cilag GmbH, Neuss, Germany) for inhalation and lung fixation and anticoagulated by intra-peritoneal injection of 2,000 IU heparin (Heparin-Natrium-ratiopharm, ratiopharm GmbH, Ulm/Donautal, Germany) for lung fixation (Kapp et al. 2004). Anesthesia was antagonized by subcutaneous injection of atipamezole (Antisedan; Pfizer GmbH), flumazenil (Anexate, Hoffmann-La Roche AG), and naloxone (Narcanti, Janssen Animal Health, Neuss, Germany).
Aerosols and inhalation.
Titanium is suitable for EFTEM because it does not interfere with the heavy metals used for tissue preparation (Kapp et al. 2004). TiO2 is inert and nonpathogenic (Templeton 1994), and ultrafine aerosols thereof remain as insoluble particles (Donaldson et al. 2002). Generally, commercially available TiO2 particles (Degussa, Düsseldorf, Germany) show a positive Zeta potential of 20–30 mV as measured by Malvern Zetasizer 3000 HS (Malvern Instruments Ltd., Malvern, Worcestershire, UK) The generation and inhalation of the TiO2 aerosol used in this study has been described in detail by Kapp et al. (2004). Briefly, ultrafine TiO2 aerosols were generated with a Palas spark generator (Palas GmbH, Karlsruhe, Germany) in a pure argon plus 0.1% oxygen stream. The aerosol was quasi-neutralized by a radioactive 85Kr source, diluted, and conditioned for inhalation. Particle size distribution and number concentration were monitored continuously by a differential electrical mobility particle sizer and a condensation particle counter. An aerosol of 22 nm CMD (geometric SD of 1.7) was produced. The mean number concentration was 7.3 × 106 particles/cm3 (SD 0.5 × 106 particles/cm3), corresponding to a mass concentration of 0.11 mg/m3. The 22 nm particles measured as aerosol particles are already agglomerates of smaller primary structures formed immediately after spark ignition and condensation. The estimated size of the primary structures was 4 nm as derived from the measured specific surface area of 330 m2/g of the UFPs produced in this study and the TiO2 bulk density of 4.2 g/cm3.
Each anesthetized rat was placed in an airtight plethysmograph box. The animals inhaled the aerosol in pairs (one each for the 1-hr and 24-hr examinations) for 1 hr via an endotracheal tube by negative-pressure ventilation (−1.5 Pa) at a breathing frequency of 45/min, resulting in a minute volume of about 200 cm3/min (Kreyling et al. 2002).
From breathing and aerosol parameters, the deposited amount of TiO2 was calculated to be 4–5 μg in each rat. After aerosol exposure, anesthesia of one of the paired rats was antagonized as described above, and the animal was returned to its cage for 24 hr.
Lung fixation and tissue preparation.
Lungs were fixed either 1 hr or 24 hr after the aerosol inhalation by sequential intravascular perfusion with buffered 2.5% glutaraldehyde (Agar Scientific Ltd., Plano GmbH, Wetzlar, Germany), 1.0% osmium tetroxide (Simec, Zofingen, Switzerland), and 0.5% uranyl acetate (Fluka Chemie GmbH, Sigma-Aldrich, Buchs, Switzerland). Lungs were then subjected to systematic tissue sampling, dehydration in a graded series of ethanol, and embedding in Epon (Fluka) (Im Hof et al. 1989; Kapp et al. 2004). Ultrathin (≤50 nm) sections were cut from five to eight tissue blocks per animal, mounted onto uncoated 600-mesh copper grids, and stained with uranyl acetate and lead citrate (Ultrostain, Leica, Glattbrugg, Switzerland).
Particle localization and elemental micro-analysis in situ.
On ultrathin sections, 12 fields, corresponding to an area of 1,820 μm2 each, were systematically sub-sampled and investigated for the presence and localization of TiO2 particles in a LEO 912 transmission electron microscope (LEO, Oberkochen, Germany) equipped with an in-column energy filter allowing energy dispersion for element specific contrast. TiO2 particles were identified by parallel electron energy-loss spectroscopy (parallel-EELS), electron spectroscopic imaging, and image-EELS (Kapp et al. 2004). For elemental micro-analysis, we used the L 2 , 3 edge of Ti at 456 eV energy loss. We obtained bright-field and structure-sensitive micrographs (recorded at 250 eV), as well as element-specific contrast for TiO2, by digital acquisition.
Cell culture experiments.
Porcine lung macrophages (provided by K. McCullough and H. Gerber, Institute for Virus Diseases and Immune Prophylaxis, Mittelhäusern, Switzerland) were cultured at 106 cells/mL in two-chamber slides (VWR International AG, Dietikon, Switzerland) for 24 hr at 37°C and 5% CO2 in RPMI 1640 medium (containing 25 mM Hepes; LabForce AG, Nunningen, Switzerland) with 10% fetal bovine serum (LabForce), 1% l-glutamine (LabForce), and 1% penicillin/streptomycin (Gibco BRL, Invitrogen AG, Basel, Switzerland). Human RBCs, always freshly isolated from the same donor, were cultured at 8 × 106 cells/mL and for 6–24 hr, as described above.
For CLSM, fluorescent polystyrene microspheres with diameters of 1, 0.2, and 0.078 μm (Fluoresbrite plain yellow green; Polysciences, Chemie Brunschwig AG, Basel, Switzerland) were added to the cells at 1010 particles/mL in supplement-free RPMI 1640 medium for 4 hr. To inhibit phagocytic uptake, cells were pretreated with cytochalasin D (cytD, 10 μg/mL; Fluka) for 30 min and during incubation with particles (Thiele et al. 2001). Macrophages were then washed in phosphate-buffered saline (PBS), fixed in 3% paraformaldehyde/PBS, treated with 0.1 M glycine/PBS, permeabilized with 0.2% Triton X-100/PBS, and stained for F-actin with rhodamine-phalloidin (1:100; Molecular Probes, VWR International AG, Lucerne, Switzerland) for 60 min at 37°C. RBCs were fixed in 2.5% glutaraldehyde/PBS, resulting in a red autofluorescence. Preparations were mounted in PBS:glycerol (2:1) containing 170 mg/mL Mowiol 4-88 (Calbiochem, VWR International AG).
For TEM analysis, RBCs were incubated with 0.025 μm gold particles (ANAWA Trading SA, Wangen, Switzerland) at 6.6 × 1010 particles/mL. RBCs were fixed with buffered 2.5% glutaraldehyde, 1% osmium tetroxide, and 0.5% uranyl acetate, and prepared for TEM as described above for lung tissue.
Microscopic analysis of cultured cells.
We used a MicroRadiance system from Bio-Rad (Hemel Hempstead, UK) combined with an inverted Nikon microscope (Eclipse TE3000; lasers: GHe/Ne 543 nm and Ar 488 nm) (Nikon, Küsnacht, Switzerland). Optical sections with a voxel dimension of 50 × 50 × 200 nm were taken with a 100× /1.4 plan-apochromate objective. We performed image processing and visualization using IMARIS software (Bitplane AG, Zurich, Switzerland). We applied a deconvolution algorithm using Huygens2 software (Scientific Volume Imaging B.V., Hilversum, Netherlands) to increase axial and lateral resolutions and to decrease noise (Rothen-Rutishauser et al. 1998). Experiments were performed in triplicate or quadruplicate, and 30–50 cells were scanned by CLSM for each data point. We examined ultrathin sections of RBCs incubated with gold particles in a Philips 300 TEM at 60 kV (Philips, Zurich, Switzerland).
Results
Distribution of inhaled UFPs in lungs.
We found TiO2 particles on the luminal side of airways and alveoli as well as within each tissue compartment of the lung (Figures 1 and 2). Particles were localized within epithelial and endothelial cells, within fibroblasts and between collagen fibrils in the connective tissue, within blood capillaries, and even within RBCs (Figure 1). Intracellular particles were localized most often in the cytoplasm and rarely within the nucleus. Particles within cells were not membrane bound. On average, 79.3 ± 7.6% (mean ± SD) of the particles were found on the luminal side of airways and alveoli, 4.6 ± 2.5% were within epithelial or endothelial cells, 4.8 ± 4.5% within the connective tissue, and 11.3 ± 3.9% within the capillaries. The relative distributions of particles among different lung compartments at 1 hr and at 24 hr after inhalation were not different from each other and correlated with the volume fractions of the respective compartments (Figure 2).
Uptake of UFPs by macrophages and RBCs.
We studied the uptake of fine (1–0.2 μm) and ultrafine (< 0.1 μm) fluorescent microspheres in cultured macrophages as a model for phagocytic cells and in human RBCs as a model for nonphagocytic cells. Additionally, we treated macrophage cultures with cytD to block phagocytosis. Cells were prepared for CLSM, and a deconvolution algorithm was applied to increase lateral as well as axial resolution. We found particles of all sizes within macrophages (Figure 3A), however, with different percentages of cells involved in particle uptake. On average, 77 ± 15% (mean ± SD) of the macrophages contained UFPs, 21 ± 11% contained 0.2 μm particles, and 56 ± 30% contained 1 μm particles (Figure 3B). CytD did not inhibit the uptake of ultrafine and 0.2 μm particles but blocked the uptake of 1 μm particles by these cells (Figure 3B). We found ultrafine and 0.2 μm, but not 1 μm, particles in RBCs (Figure 4A). TEM analysis of RBCs incubated with 0.025 μm gold particles showed that intracellular particles were not membrane bound (Figure 4B).
Discussion
The ultrastructural analyses of lung tissue demonstrated that 1 hr after aerosol inhalation, 24% of ultrafine TiO2 particles, on average, were located within and beyond the epithelial barrier (i.e., in the main lung tissue compartments, in the cytoplasm, and in the nucleus of the cells). These results confirm data from human studies, where the inhalation of ultrafine carbon particles affected pulmonary diffusing capacity (Pietropaoli et al. 2004), suggesting that particles in the interstitium have physiologic effects.
This study also provides evidence for some particle translocation into the micro-vasculature. In previous studies on iridium particles, we found minute fractions of particles translocated into secondary target organs (Kreyling et al. 2002; Semmler et al. 2004b). Different particle materials—iridium versus TiO2—may have resulted in different particle translocation patterns, and we do not know the exact composition and structure of the ultrafine particle surface, which is likely to influence particle translocation. The present study focuses on particle distribution within the primary target organ, the lung, and the results do not determine which fraction of particles may have escaped the lung micro-vasculature to be systemically circulated.
Particles found within cells were not membrane bound, indicating a nonendocytic uptake. In addition, the overall distribution pattern of the particles in the lungs (i.e., the percentages of particles in the different lung compartments) at 1 hr and at 24 hr after particle inhalation was the same and was correlated with the volume densities of the corresponding lung compartments, implying that ultrafine TiO2 particles can move between tissue compartments without restraint. Our results are in contrast with those obtained by Stearns et al. (2001), who studied the uptake of ultrafine TiO2 particles in vitro in the A549 epithelial cell line. They found that membrane-bound vesicles contained mostly large aggregates of TiO2. Sometimes vesicles with clusters consisting of as few as two to three particle profiles were observed. In a pilot study in vitro with porcine macrophages, we obtained similar results (data not shown). Whether these clusters contained few particles only or were sectioned through the top of larger ones is not known. However, because ultrafine TiO2 aggregate very quickly in polar liquids such as cell culture medium and because particle concentration was fairly high, as seen from micrographs, it is very likely that particles aggregated within the cell culture medium and that cells engulfed these large clusters by an endocytic pathway. Particle agglomeration on the lung epithelium during the 1-hr inhalation, however, is very unlikely because of the size of the inner lung surface and the number of deposited UFPs.
The fact that 80% of the retained TiO2 particles were still on the luminal side of the epithelium even 24 hr after inhalation is surprising and in contrast to previous studies on the lavageability of ultrafine iridium particles, where only 20% of the particles could be lavaged from the epithelial surfaces 24 hr after inhalation (Kreyling et al. 2002). The low lavageability of ultrafine iridium particles in contrast to high lavageability of 80% of 0.5–10-μm particles (Oberdörster et al. 2002) was interpreted as either higher adhesion of UFPs to epithelial membranes or epithelial uptake and penetration of UFPs into the interstitium. The large fraction of TiO2 particles on the luminal side of the epithelium in the present study strongly supports higher adhesion of UFPs to epithelial structures. However, it must also be considered that proteins are very likely to bind rapidly to the particles, which then may affect the further metabolic fate of the particles in terms of their adhesion, residence time on the epithelium or uptake, and even penetration through the epithelium. Along this line, the difference of the two particle materials and surfaces may have led to binding of those proteins, which then may have mediated major uptake and penetration of iridium particles into and through the epithelium. We already have first evidence that ultrafine commercial TiO2 particles bind more readily to other proteins in the lung-lining fluid than carbonaceous and amorphous silica particles (Semmler et al. 2004a).
Microscopic analyses of phagocytic and nonphagocytic cells incubated with different particle types showed that macrophages take up fine and ultrafine polystyrene microspheres and that treatment with cytD inhibits the uptake of 1.0-μm particles, but not uptake of the smaller particles, by these cells. Ultrafine polystyrene and gold particles also entered RBCs and were not membrane bound.
The mechanisms of intracellular uptake of macromolecules, particles, and even cells are subsumed as endocytosis. Material to be ingested is progressively enclosed by the plasma membrane, which eventually detaches to form an endocytic vesicle. Phagocytosis and pinocytosis are distinguished by the size of endocytic vesicles formed. Phagocytosis, a receptor-mediated, actin-based process, is characteristic for neutrophils, macrophages, and dendritic cells. It is the main mechanism for the clearance of insoluble 1- to 3-μm particles from the alveoli. Pinocytosis involves the ingestion of fluid and solutes via vesicles of about 100 nm in diameter. There are at least four basic mechanisms, most of which can be demonstrated in lungs and involve specific receptor–ligand interactions: a) macro-pinocytosis, b) clathrin-mediated, actin-based endocytosis, c) caveolae-mediated endo- or transcytosis, and d) clathrin- and caveolae-independent endocytosis (Conner and Schmid 2003). Kruth et al. (1999) described an additional endocytic process, patocytosis, in which hydrophobic polystyrene particles < 0.5 μm are transported through induced plasma membrane channels into an extensive labyrinth of interconnected membrane-bound compartments. None of these endocytic pathways, all of which include vesicle formation, is likely to account for the translocation of UFPs in our study, as intracellularly localized particles were not membrane bound. Moreover, because RBCs contained UFPs and cytD treatment of macrophages did not prevent UFP translocation into these cells, particle uptake by any actin-based mechanism can also be excluded.
Transport via pores, as suggested for lung–blood substance exchange (Conhaim et al. 1988; Hermans and Bernard 1999), is another potential mechanism for UFP translocation. TiO2 particles may diffuse through such pores. A transport mechanism by diffusion is consistent with the observed spatial distribution of UFPs in our inhalation study. Thus far, signal-mediated transport via pores has been demonstrated only for ultrafine gold particles of up to 39 nm in diameter, through the nuclear pore complex in Xenopus oocytes, where transport velocities depended on particle size (Panté and Kann 2002).
Passive uptake (not triggered by receptor–ligand interactions) may also occur by electrostatic, Van der Waals, or steric interactions, subsumed under “adhesive interactions” (Rimai et al. 2000). Rimai et al. showed that 8-μm glass particles were approximately 90% engulfed by a polystyrene substrate, compared with 22-μm particles, which were only 30% engulfed. However, the influence of particle size on their engulfment was not clarified in these publications.
Several concepts for the nonspecific engulfment of particles through interfacial structures (including cell membranes) have been suggested. A thermodynamic model using the “wettability criterion” was successful in predicting passive particle uptake, although it did not take into account the elastic properties of the cell membrane (Chen et al. 1997). In another thermodynamic analysis combined with a molecular dynamics simulation, Bresme and Quirke (1999) showed that line tension influences the wetting behavior of nanoparticles at liquid–vapor and liquid–liquid interfaces. These authors found negative line tension values for particles of a few nanometers in diameter, but positive values for those an order of magnitude larger. A negative line tension favors the initial wetting of a spherical particle after its approach to an interface.
Shanahan (1990) studied the engulfment of solid-surface heterogeneities equivalent to particles in the nanometer range by unbalanced capillary forces (free energy perturbations). Thermal capillary waves cause fluid droplets to coalesce with a fluid substrate by film drainage at the interface, breakage of the film, and intrusion of the particle into the bulk phase (Aarts et al. 2004). Thus, thermal capillary fluctuations may enhance particle transport through cell membranes.
Experimental results demonstrate consistently greater immersion of smaller particles than larger ones into a liquid substrate covered by surfactant film in vitro as well as in situ in airways. These results support the concept that line tension plays a significant role in particle displacement (Geiser et al. 2000; Schürch et al. 1999).
It remains to be determined which chemical and physical properties of membranes and particles are responsible for the translocation of UFPs in vivo. Interestingly, we did not see any difference in particle uptake in vitro with respect to differing surface charges or surface chemistry when we used three different particle types: a metal, a metal oxide, and a synthetic polymeric material (data not shown). However, these particles were added to the cells in suspension and did not approach the cells from the air or require passage through a surfactant film first, as in the in vivo inhalation experiments. In these in vivo experiments, electrostatic interactions are likely important for particle deposition and subsequent retention.
In summary, UFPs of various materials can cross any cellular membrane, but neither endocytosis, which is based on vesicle formation, nor any actin-based mechanisms are likely to account for UFP translocation into the cell. Our results from the inhalation experiments with TiO2 particles point to a transport mechanism that includes adhesive interactions or, in terms of thermodynamics, interfacial and line tension effects. In addition, particle diffusion and uptake promoted by thermal capillary waves might play a role in particle transport through membranes. After the deposition of nanometer-size particles, their further fate may be largely independent from particle surface chemistry and charge.
Consequently, it is a possible fate of inhaled ambient UFPs that they are transferred from the lungs to most other organs. In a first analysis of ultrathin sections from hearts of the same rats, we found ultrafine TiO2 particles in the connective tissue, that is, within fibroblasts (data not shown). There may be no means on the cellular level to prevent, influence, or direct their uptake. Moreover, the toxic potential of UFPs is greatly enhanced by their free location and movement within cells, which promote interactions with intracellular proteins and organelles and even the nuclear DNA.
Potential health implications of our findings are related not only to ambient UFPs but also to engineered “nanoscaled particles,” which may be released into our environment during their production, transport, and aging, or during waste disposal (The Royal Society 2004). Routes of exposure to nanomaterials include oral, cutaneous, and inhalative uptake, the latter being addressed in the present study. Although the number of particles translocated into the cells may vary substantially (Nanosafe 2004) according to their physicochemical properties, the data of the present study strongly suggest that adverse health outcomes associated with the uncontrolled presence of nanoscale particles in tissues require further attention.
We thank S. Frank, B. Haenni, B. Kupferschmid, and B. Tschirren for excellent technical assistance and L.M. Cruz-Orive for his help with the lung sampling design.
This study was supported by the Swiss National Science Foundation; the Swiss Agency for the Environment, Forest, and Landscape; and the Silva Casa Foundation.
Figure 1 EFTEM micrographs of particles (arrows) in the lung parenchyma. Abbreviations: AL, alveolar lumen; C, collagen fibril; CL, capillary lumen; EC, erythrocyte; EN, capillary endothelial cell; EOS, eosinophil granulocyte; EP, epithelium; F, fibroblast; FC, fibroblast cytoplasm; FN, fibroblast nucleus. (A) Particle (diameter, 85 nm) in the connective tissue between Cs. (B) Particle (diameter, 41 nm) in the FC near its FN. (C) Particle (diameter, 81 nm) in the cytoplasm of an EN. (D) Particle (diameter, 41 nm) within an EC in the CL. (E) Particle (diameter, 50 nm) within the FN. Bars = 500 nm.
Figure 2 Relative distribution of particles localized in the different lung compartments at 1 hr and 24 hr after inhalation. Volume densities (Vv) for lung tissue compartments from Burri et al. (1973), Pinkerton et al. (1992), and Tschanz et al. (1995, 2003).
Figure 3 (A) CLSM micrographs of fluorescent polystyrene spheres (1, 0.2, and 0.078 μm) taken up by macrophages in the absence (−) or presence (+) of cytD (bar = 2 μm); F-actin is shown in red, and particles are green. The xy and xz projections allow clear differentiation between internalized (arrows) and extracellular (arrowheads) particles; projections are indicated by open arrowheads. (B) Particle uptake by macrophages in the absence and presence of cytD. Data are expressed as mean ± SD of three to four experiments scanning 30–50 cells each by CLSM.
Figure 4 (A) CLSM micrographs of fluorescent polystyrene spheres taken up by RBCs. Autofluorescence of the cells is shown in red, and particles are green (bar = 2 μm). The xy and xz projections allow clear differentiation between internalized (arrows) and extracellular (arrowheads) particles; open arrowheads mark the position of the projections. (B, C) TEM micrographs showing uptake of 0.025 μm gold particles by RBCs; the particles are not membrane bound (arrows). (B) Bar = 1 μm. (C) Higher magnification of portion of (B); bar = 0.2 μm.
==== Refs
References
Aarts DG Schmidt M Lekkerkerker HN 2004 Direct visual observation of thermal capillary waves Science 304 847 850 15131300
Bresme F Quirke N 1999 Nanoparticulates at liquid/liquid interfaces Phys Chem Chem Phys 1 2149 2155
Brown JS Zeman KL Bennett WD 2002 Ultrafine particle deposition and clearance in the healthy and obstructed lung Am J Respir Crit Care Med 166 1240 1247 12403694
Burri HP Dbaly J Weibel EW 1973 The postnatal growth of the rat lung. I. Morphometry Anat Rec 178 711 730 4592625
Chen H Langer R Edwards DA 1997 A film tension theory of phagocytosis J Colloid Interface Sci 190 118 133 9241149
Conhaim RL Eaton A Staub NC Heath TD 1988 Equivalent pore estimate for the alveolar-airway barrier in isolated dog lung J Appl Physiol 64 1134 1142 2452819
Conner SD Schmid SL 2003 Regulated portals of entry into the cell Nature 422 36 44 12621425
Donaldson K Brown D Clouter A Duffin R MacNee W Renwick L 2002 The pulmonary toxicology of ultrafine particles J Aerosol Med 15 213 220 12184871
Ferin J Oberdörster G Penney DP 1992 Pulmonary retention of ultrafine and fine particles in rats Am J Respir Cell Mol Biol 6 535 542 1581076
Gehr P Schürch S Berthiaume Y Im Hof V Geiser M 1990 Particle retention in airways by surfactant J Aerosol Med 3 27 43
Geiser M Schürch S Gehr P 2003 Influence of surface chemistry and topography of particles on their immersion into the lung’s surface-lining layer J Appl Physiol 94 1793 1801 12547838
Geiser M Schürch S Im Hof V Gehr P 2000. Structural and interfacial aspects of particle retention. In: Particle-Lung Interaction, Vol 143 (Gehr P and Heyder J, eds). New York:Marcel Dekker Inc., 291–322.
Harder V Gilmour P Lentner B Karg E Takenaka S Ziesenis A 2005 Cardiovascular responses in unrestrained WKY rats to inhaled ultrafine carbon particles Inhal Toxicol 17 29 42 15764481
Hermans C Bernard A 1999 Lung epithelium-specific proteins: characteristics and potential applications as markers Am J Respir Crit Care Med 159 646 678 9927386
Im Hof V Scheuch G Geiser M Gebhard J Gehr P Heyder J 1989 Techniques for the determination of particle deposition in lungs of hamsters J Aerosol Med 2 247 259
Kapp N Kreyling W Im Hof V Schulz H Gehr P Geiser M 2004 Electron energy loss spectroscopy for analysis of inhaled ultrafine particles in rat lungs Microsc Res Tech 63 298 305 15170760
Kreyling WG Semmler M Erbe F Mayer P Takenaka S Schulz H 2002 Translocation of ultrafine insoluble iridium particles from lung epithelium to extrapulmonary organs is size dependent but very low J Toxicol Environ Health 65 1513 1530
Kreyling WG Tuch T Peters A Pitz M Heinrich J Stolzel M 2003 Diverging long-term trends in ambient urban particle mass and number concentrations associated with emission changes caused by the German unification Atmos Environ 37 3841 3848
Kruth HS Chang J Ifrim I Zhang WY 1999 Characterization of patocytosis: endocytosis into macrophage surface-connected compartments Eur J Cell Biol 78 91 99 10099931
Li N Sioutas C Cho A Schmitz D Misra C Sempf J 2003 Ultrafine particulate pollutants induce oxidative stress and mitochondrial damage Environ Health Perspect 111 455 460 12676598
Luther W 2004. Industrial Applications of Nanomaterials –Chances and Risks. 2004. Technological Analysis, Vol 54. Düsseldorf, Germany:VDI Technologiezentrum GmbH.
Oberdörster G 2004. Kinetics of inhaled ultrafine particles in the organism. In: Effects of Air Contaminants on the Respiratory Tract – Interpretations from Molecules to Meta Analysis (Heinrich U, ed). Stuttgart, Germany:Fraunhofer IRB Verlag, 121–143.
Oberdörster G Sharp Z Atudorei V Elder A Gelein R Kreyling W 2004 Translocation of inhaled ultrafine particles to the brain Inhal Toxicol 16 437 445 15204759
Oberdörster G Sharp Z Atudorei V Elder A Gelein R Lunts A 2002 Extrapulmonary translocation of ultrafine carbon particles following whole-body inhalation exposure of rats J Toxicol Environ Health A 65 1531 1543 12396867
Panté N Kann M 2002 Nuclear pore complex is able to transport macromolecules with diameters of ~39 nm Mol Biol Cell 13 425 434 11854401
Peters A Wichmann HE Tuch T Heinrich J Heyder J 1997 Respiratory effects are associated with the number of ultrafine particles Am J Respir Crit Care Med 155 1376 1383 9105082
Pietropaoli AP Frampton MW Hyde RW Morrow PE Oberdörster G Cox C 2004 Pulmonary function, diffusing capacity, and inflammation in healthy and asthmatic subjects exposed to ultrafine particles Inhal Toxicol 16 suppl 1 59 72 15204794
Pinkerton KE Gehr P Crapo JD 1992. Architecture and cellular composition of the air-blood barrier. In: Comparative Biology of the Normal Lung, Vol I. Treatise on Pulmonary Toxicology (Parent RA, ed). Boca Raton, FL:CRC Press Inc., 121–144.
Rimai DS Quesnel DJ Busnaia AA 2000 The adhesion of dry particles in the nanometer to micrometer –size range Colloids Surf A Physicochem Eng Aspects 165 3 10
Rothen-Rutishauser BM Messerli MJ van der Voort H Günthert M Wunderli-Allenspach H 1998 Deconvolution combined with digital colocalization analysis to study the spatial distribution of tight and adherens junction proteins J Comp Assist Microsc 10 103 111
Samet JM DeMarini DM Malling HV 2004 Biomedicine. Do airborne particles induce heritable mutations? Science 304 971 972 15143266
Schulz H Harder V Ibald-Mulli A Khandoga A Koenig W Krombach F 2005 Cardiovascular effects of fine and ultrafine particles J Aerosol Med 18 1 24 15741770
Schürch S Gehr P Im Hof V Geiser M Green F 1990 Surfactant displaces particles toward the epithelium in airways and alveoli Respir Physiol 80 17 32 2367749
Schürch S Geiser M Lee MM Gehr P 1999 Particles at the airway interfaces of the lung Colloids Surf B Biointerfaces 15 339 353
Semmler M Regula J Oberdörster G Heyder J Kreyling WG 2004a Lung-lining fluid proteins bind to ultrafine insoluble particles: a potential way for particles to pass air-blood barrier of the lung? [Abstract] Eur Respir J 24 suppl 48 100s
Semmler M Seitz J Erbe F Mayer P Heyder J Oberdörster G 2004b Long-term clearance kinetics of inhaled ultrafine insoluble iridium particles from the rat lung including transient translocation into secondary organs Inhal Toxicol 16 453 459 15204761
Shanahan MER 1990 Capillary movement of nearly axisymmetric sessile drops J Phys D Appl Phys 23 321 327
Stearns RC Paulauskis JD Godleski JJ 2001 Endocytosis of ultrafine particles by A549 cells Am J Respir Cell Mol Biol 24 108 115 11159043
Templeton DM 1994. Titanium. In: Handbook on Metals in Clinical and Analytical Chemistry (Seiler HG, Sigel A, Sigel H, eds). New York:Marcel Dekker, Inc., 627–630.
The Royal Society and The Academy of Engineering 2004. Nanoscience and Nanotechnologies: Opportunities and Uncertainties. Available; http://www.nanotec.org.uk/finalReport.htm [accessed 13 September 2005].
Thiele L Rothen-Rutishauser B Jilek S Wunderli-Allenspach H Merkle HP Walter E 2001 Evaluation of particle uptake in human blood monocyte-derived cells in vitro. Does phagocytosis activity of dendritic cells measure up with macrophages? J Control Release 76 59 71 11532313
Tschanz SA Damke BM Burri HP 1995 Influence of post-natally administrated glucocorticoids on rat lung growth Biol Neonate 68 229 245 8580214
Tschanz SA Makanya AN Hänni B Burri HP 2003 Effects of neonatal high-dose short-term glucocorticoid treatment on the lung: a morphologic and morphometric study in the rat Pediatr Res 53 72 80 12508084
|
16263511
|
PMC1310918
|
CC0
|
2021-01-04 23:41:31
|
no
|
Environ Health Perspect. 2005 Nov 26; 113(11):1555-1560
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8006
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7868ehp0113-00156116263512ResearchConsistent Pulmonary and Systemic Responses from Inhalation of Fine Concentrated Ambient Particles: Roles of Rat Strains Used and Physicochemical Properties Kodavanti Urmila P. 1Schladweiler Mette C. 1Ledbetter Allen D. 1McGee John K. 1Walsh Leon 1Gilmour Peter S. 2Highfill Jerry W. 1Davies David 1Pinkerton Kent E. 3Richards Judy H. 1Crissman Kay 1Andrews Debora 1Costa Daniel L. 11 Pulmonary Toxicology Branch, Experimental Toxicology Division, National Health and Environmental Effects Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina, USA2 Center for Environmental Medicine, Asthma, and Lung Biology, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA3 Center for Health and the Environment, University of California at Davis, Davis, California, USAAddress correspondence to U.P. Kodavanti, Pulmonary Toxicology Branch, MD: B143-01, Experimental Toxicology Division, National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 USA. Telephone: (919) 541-4963. Fax: (919) 541-0026. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 27 6 2005 113 11 1561 1568 16 12 2004 27 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Several studies have reported health effects of concentrated ambient particles (CAP) in rodents and humans; however, toxicity end points in rodents have provided inconsistent results. In 2000 we conducted six 1-day exposure studies where spontaneously hypertensive (SH) rats were exposed to filtered air or CAPs (≤ 2.5 μm, 1,138–1,765 μg/m3) for 4 hr (analyzed 1–3 hr afterward). In seven 2-day exposure studies in 2001, SH and Wistar Kyoto (WKY) rats were exposed to filtered air or CAP (≤ 2.5 μm, 144–2,758 μg/m3) for 4 hr/day × 2 days (analyzed 1 day afterward). Despite consistent and high CAP concentrations in the 1-day exposure studies, no biologic effects were noted. The exposure concentrations varied among the seven 2-day exposure studies. Except in the first study when CAP concentration was highest, lavageable total cells and macrophages decreased and neutrophils increased in WKY rats. SH rats demonstrated a consistent increase of lavage fluid γ -glutamyltransferase activity and plasma fibrinogen. Inspiratory and expiratory times increased in SH but not in WKY rats. Significant correlations were found between CAP mass (microgram per cubic meter) and sulfate, organic carbon, or zinc. No biologic effects correlated with CAP mass. Despite low chamber mass in the last six of seven 2-day exposure studies, the levels of zinc, copper, and aluminum were enriched severalfold, and organic carbon was increased to some extent when expressed per milligram of CAP. Biologic effects were evident in those six studies. These studies demonstrate a pattern of rat strain–specific pulmonary and systemic effects that are not linked to high mass but appear to be dependent on CAP chemical composition.
concentrated ambient particlesfibrinogenγ-glutamyltransferasehypertensive ratslung inflammationmacrophagesneutrophilsWistar Kyoto rats
==== Body
A number of studies have recently reported pulmonary and cardiovascular health effects of concentrated ambient particles (CAP) in animals and humans (Clarke et al. 1999, 2000; Ghio et al. 2000; Gordon et al. 2000; Huang et al. 2003; Kodavanti et al. 2000a; Saldiva et al. 2002; Smith et al. 2003). Real-time ambient particle concentrators are designed to concentrate ambient particulate matter (PM) (Gordon et al. 1999; Sioutas et al. 1995), allowing toxicologic studies to be conducted at higher than ambient concentrations. This approach allows mechanistic characterization of PM effects and identification of causative constituents. One of the limitations of this approach, however, is that the physicochemical properties of CAP and ambient PM may differ because the overall enrichment of CAP may depend on the size of incoming particles.
A review of animal toxicologic studies involving CAP exposures has shown inconsistencies, and generally subtle health effects. Moreover, there is a lack of correlation between health end points and PM mass or its causative components (Clarke et al. 2000; Gordon et al. 2000; Kodavanti et al. 2000a; Saldiva et al. 2002; Smith et al. 2003). However, in clinical studies, CAP effects as determined by analysis of blood/plasma markers and constriction of arteries have been readily apparent (Ghio et al. 2003; Huang et al. 2003; Urch et al. 2004). These differences could be due to greater human sensitivity as well as reduced variability because the subject exposed to CAP most often serves as his or her own control for measured biologic variables. Repeated sampling is technically difficult with most end points in animals.
Although the animal studies using model particulate samples have provided links between specific constituents and biomarkers of health effects (Kodavanti et al. 1998), they often have been criticized for being irrelevant to human scenarios because of the exposure methods, dosages, or unrealistic composition of particles. CAP exposure studies, in contrast, are difficult to reproduce because of the varying dynamics of ambient PM concentration and composition. Therefore, without marked and reproducible biologic effects of ambient PM, it has been difficult to investigate mechanisms and causality. CAP studies performed using identical exposure protocols and novel biomarkers generally need to be evaluated for consistencies in concentration or composition-related trends.
On the basis of studies conducted using combustion and ambient particles, metals have been shown to comprise a class of significant causative constituents of PM health effects (Dye et al. 1999; Kodavanti et al. 1998). Metals have been ubiquitously detected in ambient PM, albeit at very low concentrations (Harrison and Yin 2000; Kodavanti et al. 2000a; Saldiva et al. 2002; Smith et al. 2003). The major metallic components of ambient PM at different locations include iron, silicon, aluminum, copper, and zinc. Recently, ambient organics have also been implicated in adverse health effects (Carvalho-Oliveira et al. 2005). The specific health effects of each of these components and their potential interactions in the environment and at the cellular level are not fully understood.
In this article, we describe multiple exposure studies using spontaneously hypertensive (SH) and Wistar Kyoto (WKY) rats. These studies were designed to identify consistency in the pattern of biologic response to correlate with CAP mass and composition, and potential susceptibility factors. We hypothesized that SH and WKY rats respond differentially to CAP and that selected biologic indicators reflect these differences in a strain-specific manner. We also explored whether CAP effects on biologic end points related better to mass or to putative causative constituents.
Materials and Methods
Animals.
Healthy, male, 10- to 12-week-old, normotensive WKY and SH (SHR/NCrlBR) rats (derived from WKY rats by segregation of the hypertensive trait and inbreeding) were purchased from Charles River Laboratories (Raleigh, NC). All rats were maintained in an isolated animal room in an animal facility approved by the Association for Assessment and Accreditation of Laboratory Animal Care [maintained at 21 ± 1°C, 50 ± 5% relative humidity (RH), 12-hr light/dark cycle] for 1–2 weeks of quarantine and nonexposure periods. All animals received standard (5001) Purina rat chow (Dyets, Inc., Bethlehem, PA) and water ad libitum except during CAP exposure periods of 4–5 hr. The U.S. Environmental Protection Agency (EPA) Animal Care and Use Committee approved the protocol for the use of rats in inhalation studies.
CAP exposure.
Six repeat studies of 4 hr/day, 1-day exposures were conducted between 17 October and 16 November 2000, and seven repeat studies of 4 hr/day, 2-day exposures were conducted between 27 August and 24 October 2001. Rats in each study (SH rats for 1-day exposure studies, and SH and WKY rats for 2-day exposure studies) were computer randomized into two groups after sorting them from low weight to high weight to ensure similar means and distribution. One group was exposed to clean air, and the other to CAP using the U.S. EPA fine-mode CAP exposure system (1-day exposure studies; n = 5–9 rats per group, and 2-day exposure studies; n = 4–5 rats per group) (Sioutas et al. 1995). The series of four virtual impactors produced an empirical ambient fine-mode (≤ 2.5 μm) particle concentration enhancement of 40–60 times ambient levels in the exposure chamber. Animals were exposed generally between 0830 and 1330 hr for a total period of 4 hr during each exposure day [see Supplemental Material for details (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)].
Outside environmental conditions were continuously monitored using a weather station (Weather Monitor II; Davis Instruments, Haywood, CA) sited within 150 ft of the CAP system inlet. Ambient temperature, RH, dew point, wind speed, wind direction, and barometric pressure were recorded at 30-min intervals during each exposure. Control and exposure chamber temperature and RH were measured continuously (Omega RH-411 temperature and RH Thermo hygrometers; Omega Engineering, Stamford, CT).
A superimposed map of daily wind direction and speed for all 2001 exposure days was prepared from individual maps obtained from the World Wide Web–based Real-Time Environmental Applications and Display System (READY). This system has been developed for accessing and displaying meteorologic data on the National Oceanic and Atmospheric Administration (NOAA) Air Resources Laboratory web server (NOAA 2004).
CAP organic and elemental analysis.
Samples for analysis of CAP mass concentration were collected on preweighed Teflon filters for the duration of each exposure. Postexposure, filters were weighed and concentrations determined by sample mass/sample flow volume (μg/m3). Ambient levels of total suspended particulate and fine-mode (≤ 2.5 μm) particles were measured gravimetrically using Teflon filters (2.0 μm, 37 mm, and 47 mm Teflo, R2PJ037, and R2PJ047; Pall Corp., East Hills, NY). These filters sequester particles > 0.3-μm in size with 99.7% efficiency [more details in Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)].
Organic and elemental carbon contents of CAP collected on quartz filters were determined using National Institute of Occupational Health (NIOSH) thermal-optical method 5040 (Sunset Laboratory, Tigard, OR). To determine soluble ion and elemental content, we extracted each Teflon filter in distilled water for 1 hr and centrifuged the extracts at 17,000 × g for 30 min. We removed two aliquots of each supernatant and analyzed the first aliquot as is for sulfate and nitrate content using ion chromatography (McGee et al. 2003). We acidified the second aliquot to a pH < 2.0 using concentrated nitric acid to keep soluble metal salts in soluble form. We then analyzed the acidified extracts for elemental content using inductively coupled plasma–mass spectroscopy (ICP-MS) (McGee et al. 2003).
Whole-body plethysmograph data acquisition and analysis.
We employed a barometric whole-body plethysmograph system (Buxco Electronics Inc., Sharon, CT) to obtain data on pulmonary ventilation for 2-day exposure studies. This methodology permitted monitoring of a number of ventilatory parameters, including breathing frequency (f ), tidal volume (Tv), minute ventilation (MV), peak expiratory flow (PEF), peak inspiratory flow (PIF), inspiratory time (Ti), expiratory time (Te), pause (PAU), and enhanced pause (PENH) (Tankersley et al. 1997). Unrestrained, free-moving animals were placed in individual chambers and allowed 1 min to settle down followed by 5 min of monitoring. We analyzed the data from each animal before the first CAP or air exposure and after the last exposure.
Necropsy and sample collection.
Necropsies were performed within 3 hr after exposure in six 1-day exposure studies except for the fourth study where necropsies were performed 18–20 hr postexposure. All necropsies were performed 18–20 hr after the second exposure in 2-day exposure studies. At designated time points rats were weighed and anesthetized with sodium pentobarbital (50–100 mg/kg, intraperitoneally). Blood was collected from the abdominal aorta, directly into blood collection tubes containing EDTA (for complete blood counts) or citrate as anticoagulants (for plasma protein analysis). The trachea was cannulated and the left lung was tied. The bronchoalveolar lavage was performed as described previously (Kodavanti et al. 2000b).
Blood chemistry and cytology.
Plasma fibrinogen and complete blood counts were performed by the University of North Carolina Memorial Hospital Core Facility (Chapel Hill). Each blood sample containing citrate anticoagulant was centrifuged at 4,500 rpm for 10 min at 4°C. Plasma was then analyzed for fibrinogen as described by Gardner et al. (2000). The complete blood counts were performed on a Technicon H-2 hematology analyzer (Bayer Corp., Tarrytown, NY) using Bayer Technicon reagents. Angiotensin-converting enzyme (ACE) activity was measured in citrated plasma using reagents and controls from Sigma Diagnostics (St. Louis, MO). C-Reactive protein (CRP) was measured in citrated plasma using an SPQ II kit that contained its own calibrations and controls (Diasorin Inc., Stillwater, MN). The ACE activity and CRP assays were adapted for use on the Cobas Fara II clinical analyzer (Hoffmann-La Roche, Branchburg, NJ).
BALF analysis for determining lung injury.
We used one aliquot of whole bronchoalveolar lavage fluid (BALF) to determine total cell counts (Coulter counter; Coulter Inc., Miami, FL); a second aliquot was centrifuged (Shandon 3 Cytospin; Shandon, Pittsburgh, PA) to prepare cell differential slides. We dried the slides at room temperature and stained them with LeukoStat (Fisher Scientific Co., Pittsburgh, PA). Macrophages and neutrophils were counted using light microscopy (> 200 cells counted per slide). We centrifuged the remaining BALF at 1,500 × g to remove cells, and analyzed the supernatant fluid for markers of lung injury. Total protein was analyzed using Coomassie Plus Protein Assay Kit (Pierce, Rockford, IL). We analyzed BALF albumin using a commercially available kit (Diasorin). Lactate dehydrogenase (LDH) activity (U/L) was determined using Kit 228 from Sigma Chemical Co. (St. Louis, MO). We measured N-acetyl glucosaminidase (NAG) activity (U/L) using a kit and standards from Roche Diagnostics (Indianapolis, IN). γ -Glutamyltransferase (GGT) activity was measured using a kit from Thermo Trace Ltd. (Melbourne, Australia). These assays were adapted for use on the Cobas Fara II (Hoffmann-La Roche) clinical analyzer. We mixed an aliquot of BALF with an equal volume of 6% perchloric acid and vortexed. After standing on ice for 10 min, it was centrifuged (14,000 × g) for 10 min (4°C), and then supernatants stored at –80°C. We determined total glutathione and ascorbic acid as described previously (Kodavanti et al. 2000b).
Determination of cytokines in BALF using enzyme-linked immunosorbent assay.
WKY rat BALF samples from all 2-day exposure studies were analyzed for interleukin-6 (IL-6), tumor necrosis factor-α, and macrophage inflammatory protein-2 using a sandwich enzyme-linked immunosorbent assay (ELISA) technique. Rat-specific cytokine assay kits were obtained from Biosource International (Camarillo, CA) and were used in performing ELISA. We ran each sample in triplicate. Sample optical density values were measured at 450 nm wavelength using microtiter plate reader (SpectraMax Pro 340PC; Molecular Devices, Sunnyvale, CA). This system uses SoftMax Pro (version 2.6.1; Molecular Devices) software to run the plate reader and analyze the data.
Statistics.
For the data analysis of the 1-day exposure studies, we assumed a homogeneous variance for blood lymphocytes, platelets, hematocrit, and hemoglobin. Ranks were used to determine significance levels for all other biologic parameters where heterogeneity of variance was apparent. To determine significance levels, we used a crossed-design analysis of variance (ANOVA) with an interaction term.
Because of the small number of rats used per group in each of the 2-day exposure studies, the variance was heterogeneous, and therefore, we performed no statistical testing to determine study-to-study differences. We pooled the data into responses related to rat strain and exposure group and performed statistical analysis to determine if there were an interaction between exposure and seven studies. If there was no interaction, then we performed further ANOVA to determine if the strains responded differently for each variable. If there was no interaction, then we tested strain and exposure differences using ranks in ANOVA. Each strain was used separately in an ANOVA to determine significant differences between air and CAP exposures.
For the 2-day studies, we used averages of 2- to 4-hr exposures for mass and composition in all correlation analyses. We determined correlations for ambient air chemistry values using the Pearson correlation coefficient and also the Spearman method. Spearman correlations tend to reduce the influence of very large or very small measurements.
Results
Composition of exposure atmospheres.
The atmospheric conditions, before PM is concentrated, can influence its chemical composition during formation. All 1- and 2-day exposures occurred between 0830 and 1330 hr in an attempt to minimize this factor. The daytime high atmospheric temperatures were in 70–79°F except for the last exposure day (52°F), with the difference between daytime high and nighttime low temperatures being 17–28°F during 1-day exposure studies. In 2-day exposure studies, the daytime high temperatures reached 80–89°F, and the difference between daytime high and nighttime low temperatures was 11–30°F. Atmospheric humidity varied from 100% to < 50%. The mean temperature of control and CAP chambers varied within 2.5% and RH within 25% during all exposures.
Elemental and organic composition of CAP.
Particle size range did not change significantly between 1- or 2-day exposure studies except for a slightly larger size during the 2-day exposure studies (Table 1). During 1-day exposure studies, generally high CAP concentrations were achieved with only a small variation between exposure days (Table 1, studies 1–6). This hindered the ability to establish relationships between components and mass, or within components. During seven 2-day exposure studies, low CAP concentrations were achieved except for the first study, and the exposure concentrations varied markedly between studies (Table 1, studies A–G).
The metal analysis included only water-soluble elements. Of many elements measured (aluminum, arsenic, barium, beryllium, cadmium, cobalt, copper, lead, manganese, nickel, silver, titanium, and zinc), the most abundant were aluminum, zinc, and copper (Table 1). Silica, sodium, and iron, which are also likely among the most abundantly detected metals, were not measured. The sulfate concentration was lowest when the lowest CAP concentrations were achieved (Table 1, 2-day exposure studies A–G). When the highest chamber concentration of CAP was achieved, the sulfate concentrations were predominant (Table 1, study A).
The components that accounted for > 50% of CAP mass in all samples were sulfate, organic carbon, and elemental carbon (Table 1, 2-day exposure studies A–G). Organic carbon concentrations were 10–20 times higher than elemental carbon, considering samples from both years, and were significantly associated with CAP mass (Figure 1A). However, we observe no linearity between elemental carbon and CAP mass, as determined for the 2-day exposure studies (Figure 1B). The regression analysis of the samples from 2-day exposure studies indicated correlations between sulfate and CAP mass (Figure 1C). Significant correlations also existed between zinc and CAP mass (Figure 1D). Using CAP composition data of 2-day exposure studies, these four variables, organic carbon, elemental carbon, sulfate, and zinc, compared with mass concentrations produced correlations of 0.95, 0.18, 0.94, and 0.64, respectively (Figure 1). As expected, nitrate concentrations were low compared with sulfate. CAP nitrate concentrations varied between 37 μg/m3 (27 October 2000) and as low as 4 μg/m3 (2–3 October 2001). However, there appeared to be no linear relationship between nitrate levels and the mass (Table 1).
When individual CAP components were compared, it was apparent that organic and elemental carbon correlated poorly (r = 0.02) (Figure 2A). However, organic carbon mass correlated with sulfate and zinc but not with aluminum (data not shown). Zinc levels correlated well with elemental carbon (Figure 2B) and sulfate (Figure 2C), demonstrating correlations of 0.72 and 0.82. Correlations also existed between lead and elemental carbon (Figure 2D). When levels of metals in CAP were calculated per given CAP mass (microgram per milligram), it was apparent that aluminum, copper, and zinc were enriched severalfold on the PM during 2-day exposure studies B–G (Table 2). In these same studies organic carbon was also enriched to a smaller degree when expressed per milligram of CAP (Table 2).
To determine the potential direction of ambient particle migration and their source, we evaluated backward weather trajectories (Figure 3). The wind direction during 10 and 11 October 2001 was from the east, primarily coming from Atlantic Ocean; during this time, there was the least amount of particles and sulfate concentrated in the chamber (Table 1, 2-day exposure studies A–G). There was no other noticeable consistency in wind pattern, speed, or direction that could be linked to specific particle composition and biologic effects.
Baseline rat strain differences in biologic end points.
The mean values of all seven combined 2-day exposure studies for WKY and SH rats of control and CAP groups are given in Tables 3 and 4, and in Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf). SH and WKY rats differ markedly in several breathing parameters (Table 3). As we have shown previously (Kodavanti et al. 2000b, 2002), baseline values of several pulmonary markers also differ between WKY and SH rats. The levels of neutrophils, protein, and albumin were higher, but antioxidants such as glutathione, ascorbate, and uric acid were lower in BALF of SH rats than in WKY rats (Table 4). All the hematologic values assayed were higher in SH rats [Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)], which is consistent with hypertensive disease (Bianchi et al. 1986).
CAP-induced changes in breathing parameters.
Net body weight gains after CAP exposure did not change in any of the strains or exposure groups (data not shown). We measured breathing parameters for the 2-day exposure studies in each rat before and after CAP exposures to allow for paired analysis (Figure 4A,B). Although overall patterns of changes induced by CAP were consistent in six of seven studies, no individual study group reached significance with regard to changes in breathing parameters. The data for all air and all CAP groups were therefore combined for statistical analysis. Significant differences were noted in a number of breathing parameters by this maneuver (Table 3). The paired analysis included the percentage difference from baseline for all control and CAP groups for each rat strain. SH rats exposed to CAP demonstrated a significant overall increase in inspiratory time (Ti) and expiratory time (Te) relative to those of air controls (Figure 4B). These parameters were not significantly different in WKY rats (Figure 4A).
Biochemical and inflammatory indicators of pulmonary injury and blood/plasma markers.
None of the biochemical and inflammatory parameters analyzed in the six 1-day exposure studies showed a consistent change. Five studies showed a marginal increase in plasma fibrinogen, which was not significant (data not shown).
Although statistically not significant in each study, many parameters showed consistent CAP effects in a rat strain–specific manner during 2-day exposure studies. Combining data for all seven studies for those parameters demonstrated significant CAP effects [Table 4 and Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)]. Total lavageable cells decreased in all studies except for the first one (study A) where the highest CAP concentration occurred (Figure 5A). This decrease in total cells was observed only in WKY rats and was associated with a decrease in lavageable macrophages (Figure 5B). The other effect we observed in WKY rats was the increase in neutrophils (Figure 5C) in all except for the first study when CAP concentrations were high. Thus, for the WKY rats, total cell count, alveolar macrophages, and neutrophils (%) were significantly different (p = 0.01, 0.0001, and 0.004, respectively).
SH but not WKY rats demonstrated a consistent increase in GGT activity in all studies except for the first one [Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)]. GGT is a membrane-bound enzyme thought to indicate cell membrane integrity (Ernst et al. 2002). The plasma fibrinogen levels were higher in five of seven exposures in SH rats [Supplemental Material (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf)], but surprisingly, during the first and the last exposure studies when the CAP concentrations were highest, the plasma fibrinogen levels did not increase. This effect was less consistent in WKY rats but was still statistically significant when all groups were combined.
Biologic responses in WKY or SH rats were correlated with exposure variables such as CAP mass, organic and inorganic carbon, sulfate, and other major elemental constituents (microgram per cubic meter) using Spearman’s correlation. However, none of the variables showed significant correlation. Nevertheless, it should be noted that when biologic responses were correlated with metals such as aluminum, copper, and zinc normalized per unit mass of CAP (microgram per milligram), zinc correlated best with plasma fibrinogen in SH rats (p = 0.0023) (Figure 6). Correlations also existed between other metals and GGT levels but were less remarkable with neutrophils and total cell changes.
Discussion
Because CAP composition is likely to vary significantly on different days during changing seasons, studies using identical protocols repeated over an extended time period provide an opportunity to identify modifying factors of biologic response. Here, we report that seven 2-day exposure studies (4 hr/day) conducted between 24 August and 27 October 2001, caused rat strain–specific and relatively consistent effects in a selected group of biologic variables. However, the changes in these variables were not correlated with CAP mass. The concentrations of aluminum, copper, and zinc were enriched severalfold per given mass of CAP (microgram per milligram) in studies that showed biologic response. Organic carbon was also enriched to a small degree in those studies but not sulfate, suggesting that physicochemical properties of CAP were different during those seven replicative studies. We hypothesize that CAP effects may be revealed only when unique CAP composition is formed (metal–organic enriched). However, direct mechanistic linking of biologic responses to a given constituent remains a challenge because most routinely used biologic variables in animals show small effects, and generally there is limited data availability.
There are potential explanations for the lack of effect of CAP on health end points in 1-day exposure studies. The exposures occurred only for 1 day (4 hr), and the responses were determined immediately after the exposure except in one replicate when necropsy was done 1 day postexposure. The higher and more variable baseline levels of neutrophils in SH rats may also have obscured any modest CAP effects, and the time to necropsy may have been too short to induce the migration of neutrophils in the lung.
Although neutrophilic inflammation and decreases in number of lavageable macrophages were apparent in WKY rats in most of 2-day exposure studies, no correlation existed with mass concentration. On the contrary, at the highest CAP levels no increase in neutrophils was apparent, suggesting that within a given range of concentrations, PM mass may not be the primary determinant of biologic response in the rat. The lack of correlations between mass and biologic end points has been noted in other studies (Kodavanti et al. 2000a; Saldiva et al. 2002; Smith et al. 2003). Potential interactions between constituents of CAP in the air or at the cellular level may significantly alter the toxicity of constituents.
Variation in CAP composition is governed by the atmospheric transformation, transport of emissions from downwind power and industrial plants, vehicular emissions, domestic combustion activities, forest fires, and a range of natural sources. In the 2-day exposure studies we observed a significant correlation between CAP mass and several of its components (expressed as microgram per cubic meter), such as sulfate, organic carbon, and zinc, but not other metals such as aluminum, copper, and lead. The significant correlation between elemental carbon and zinc points to the contribution by vehicular emissions. Also, the air shed in Research Triangle Park, North Carolina, is likely to have a significant sulfate contribution from regional transport over the nearby highway.
Given the significant differences in CAP mass and the health effects during 2001, we predicted that the weather trajectories would provide insight into the potential sources of CAP components. The composition of particles coming from the east (Atlantic Ocean) during 2 exposure days was different from that of particles migrating from the northwest and southwest. High sulfate content of CAP during southwest and northwest winds, and reduction thereof, during 2 days when wind was coming from the east, was potentially linked to contribution by industrial activities.
A significant discrepancy was noted in gravimetric measurement of CAP mass and total mass of components. This discrepancy could be due to inaccuracy in measurement of organic carbon as a mass of organic content. It has been suggested that the discrepancy in unaccountable PM mass stems, from the imprecision of organic carbon measurements, and from organic carbon concentration not representing total organic mass (Andrews et al. 2000). To obtain more accurate estimates of the most accountable CAP mass, Turpine et al. (2000) suggested that organic carbon concentrations be multiplied by a factor of 1.4. It is an estimate of the average molecular weight of organic mass per gram of organic carbon in atmospheric particle samples (Turpine et al. 2000). However, this multiplicative factor can vary depending on the source of ambient PM.
In search of CAP compositional factors that might have caused biologic effects, we noted that the water-soluble elements such as aluminum, copper and zinc, when expressed per milligram of CAP mass, were enriched severalfold in 2-day exposure studies when positive biologic responses were evident. The plasma fibrinogen changes in SH rats better correlated with the levels of metals (per mg CAP), especially water-soluble zinc. This relationship has also been recently demonstrated in humans exposed to CAP from the nearby town of Chapel Hill (Huang et al. 2003). The levels of organic carbon per milligram of CAP also were slightly enriched and were correlated with changes in fibrinogen. These data suggest that the overall physicochemical makeup rather than particle mass may be important, especially considering water-soluble metallic constituents and organics. We can hypothesize that metal and organic-enriched PM may cause greater tissue damage than particles having low overall metal–organic mass because concentration of these causative components may reach high at the site of their deposition in the microenvironment of airways.
Because other CAP studies in the literature measured total metals using X-ray fluorescence (Saldiva et al. 2002; Smith et al. 2003), it is not possible to compare levels of major elemental CAP components such as aluminum, iron, silica, copper, and zinc because our studies focused on the water-soluble elements using ICP-MS (Kodavanti et al. 2000a). We can hypothesize that readily water-soluble metals induce acute effects, whereas less soluble metals on PM collected within phagolysosomes can slowly leach off and produce long-term health effects. Thus, our correlations of water-soluble components with health effects possibly reflect an acute and direct action of metals. However, the role of organics cannot be excluded because organics constitute a bigger portion of CAP mass relative to elements, and levels of organics were correlated with biologic end points. Determination of individual organic species is critical in identifying their roles in health effects. Also, because sulfate is one more major CAP constituent, its interaction with metals and organics and overall effect on physicochemistry can play a crucial role in biologic activity of particles.
Small but consistent CAP-related changes in various breathing parameters were noted across all seven studies in SH rats; however, these changes were not reflected as an enhanced pause (PENH) increase, believed to be associated with altered airway function. It is not clear whether the strain differences in CAP response are due to differences in the basal values for breathing parameters in these two strains. It is important to note that changes in breathing parameters in SH rats were not associated with CAP-induced neutrophilic inflammation because these later changes were not consistent in SH rats. It remains to be investigated whether the increases in inspiratory and expiratory times were related to CAP-induced autonomic stimulation.
The purpose of measuring a variety of biologic end points was partly to ascertain which biomarker best reflects fluctuations in CAP mass and components. We had hypothesized that SH rats would demonstrate specific susceptibility to the cardiovascular effects, whereas WKY rats would demonstrate a more apparent neutrophilic response because of their consistent low baseline values compared with those of SH rats (Kodavanti et al. 2002). As hypothesized, WKY but not SH rats demonstrated a clear difference in neutrophilic inflammation between air and CAP groups. Also, the increase in neutrophils in WKY rats was associated with a marked and consistent decrease in total lavageable macrophages. A possible explanation for this difference is that macrophages in WKY rats might have been activated after CAP exposure and thus difficult to remove by lavage. The alveolar macrophages in SH rats, however, may reside in an activated state because of underlying disease and thus cannot be further activated by CAP. Macrophages and the lung epithelial cells in SH rats more readily express Toll-like receptors than WKY rats (Gilmour et al. 2004).
The membrane-bound enzyme GGT is involved in transport of amino acids across cell membranes and has been used as an early marker of precancerous lesions and of glutathione synthesis activity (Ernst et al. 2002). The detection of this enzyme in BALF might indicate loss of cell membrane integrity (Ernst et al. 2002). GGT levels increased only in SH rats in 2-day exposure studies. However, LDH and NAG, enzyme indicators of cytotoxicity, were not increased in SH rats.
These multiple CAP studies conducted over 2 years have provided some insight into the role of CAP composition in relation to biologic effects and the sensitivity of two rat strains. It is clear that CAP mass concentrations were not driving the response in our experimental settings; rather, it appeared that the water-soluble metals and organic enrichment of particles might be more critical in eliciting acute health effects, especially the association of plasma fibrinogen increase with zinc and organic carbon. The WKY rat may be an appropriate animal model for CAP effects on inflammation, whereas SH rats may represent a better model to study systemic effects of CAP.
Supplementary Material
Supplemental Material Supplemental Material available online (http://ehp.niehs.nih.gov/docs/2005/7868/supp.pdf).
We thank D. Doerfler (U.S. EPA) and J. Boere, (National Institute for Public Health and the Environment, Bilthoven, the Netherlands) for statistical analysis of the data. E. Lappi of the U.S. EPA provided technical help during exposures. D. Smith of Sunset Laboratory (Tigard, Oregon, USA) determined organic and elemental carbon content of particulates. F. Weber (Research Triangle Institute, Research Triangle Park, North Carolina, USA) performed inductively coupled plasma–mass spectroscopy and ion chromatography analysis of particulate extracts. J. Hovel (Computer Sciences Corporation, Sterling, Virginia) is acknowledged for preparing the superimposed map of weather trajectories. We also thank L. Birnbaum, J. Samet, and W. Russo (U.S. EPA) for their critical review of the manuscript.
The research described in this article has been reviewed by the National Health and Environmental Effects Research Laboratory, U.S. EPA, and approved for publication. Approval does not signify that the contents necessarily reflect the views and the policies of the agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
Figure 1 Correlations between CAP mass concentration in chambers and the levels of potential causative constituents (2-day exposure studies). The x-axis represents the mass values obtained during first and second day for each study (except that data for first day in study B are not available). The correlations with CAP mass are (A) organic carbon, r = 0.95, p = 0.0001; (B) elemental carbon, r = 0.18, p = 0.54; (C) sulfate, r = 0.94, p = 0.0001; (D) zinc, r = 0.64, p = 0.014.
Figure 2 Interrelationships between concentrations of major constituents of CAP during 2-day exposure studies. A linear regression model was applied to determine if changes in the levels of major components were interrelated. The x- and y-axes depict values for different components during first and second day for each study (except that first-day data in study B are not analyzed). The correlations within components are (A) organic carbon versus elemental carbon, r = 0.02, p = 0.95; (B) elemental carbon versus zinc, r = 0.72, p = 0.006; (C) sulfate versus zinc, r = 0.55, p = 0.042; (D) elemental carbon versus lead, r = 0.82, p = 0.0006.
Figure 3 A map of superimposed weather trajectories for all 2-day CAP exposure studies during 2001. Individual backward weather trajectories were obtained from NOAA (2004). The lines pointing to Raleigh, North Carolina, show migration of air for 24 hr before 0900 hr on each exposure day. The distance between two color-matching markings on each line represents the distance traveled by air for 6 hr.
Figure 4 Effect of CAP exposure on breathing parameters in WKY (A) and SH (B) rats. Abbreviations: f, breathing frequency; Mv, minute volume; PAU, enhanced pause; PEF, peak expiratory flow; PENH, enhanced pause; PIF, peak inspiratory flow; Te, expiratory time; Ti, inspiratory time; TV, tidal volume. The y-axis represents percent change from preexposure values. Animal responses from all seven 2-day exposure studies were combined for WKY–air, WKY– CAP, SH–air, and SH– CAP groups, and percent change values represent mean of 28–35 animals/group.
Figure 5 Inflammatory cells in BALF from WKY and SH rats after filtered air or CAP exposure in relation to mass concentrations of CAP. (A) Total cells. (B) Alveolar macrophages (AM). (C) Neutrophils. A–G on the x-axis denote individual 2-day exposure studies. The line graphs indicate mean mass concentrations of CAP in chambers during each study, with the response variable plotted on the right y-axis. Values represent mean ± SE of four to five animals per group during each study.
Figure 6 Correlation of zinc levels in the CAP (microgram per milligram) and CAP-induced plasma fibrinogen change over air control for 2-day exposure studies. Changes in plasma fibrinogen (CAP group – air group) were plotted against zinc levels of CAP as determined per milligram of mass during each day of exposure.
Table 1 Chamber CAP concentrations and its organic and leachable elemental composition during 1- and 2-day exposure studies.
Study CAP exposure date Chamber concentration (μg/m3) Particle size (μm; mean ± SD)a Sulfate (μg/m3) EC (μg/m3) OC (μg/m3) NO3 (μg/m3) Al (ng/m3) Ba (ng/m3) Cu (ng/m3) Pb (ng/m3) Mn (ng/m3) Ni (ng/m3) Zn (ng/m3)
1 17 Oct 2000 1,765 1.09 ± 1.31 583 22.6 359 29.6 1,578 229 948 300 153 45 1,398
2 18 Oct 2000 1,748 1.12 ± 1.28 883 16.9 306 10.1 711 154 236 181 85 26 683
3 20 Oct 2000 1,504 1.07 ± 1.35 463 19.0 288 7.8 1,383 628 5,045b 243 229 22 1,306
4 25 Oct 2000 1,138 1.16 ± 1.40 235 17.5 278 22.2 2,728 456 781 190 154 118 860
5 27 Oct 2000 1,388 1.19 ± 1.36 363 15.0 284 37.2 2,782 349 290 286 164 36 780
6 16 Nov 2000 1,176 1.09 ± 1.40 132 32.8 419 10.6 2,767 599 367 229 537 41 1,852
A 27–28 Aug 2001 2,758 1.27 ± 1.42 823 15.5 581 15.0 1,082 208 386 207 154 32.3 1,259
Bc 11–12 Sep 2001 604 1.44 ± 1.48 188 10.3 169 5.7 699 212 514 134 118 7.9 704
C 18–19 Sep 2001 994 1.36 ± 1.50 264 22.2 320 7.4 1,463 324 573 318 161 38.5 1,295
D 2–3 Oct 2001 685 1.37 ± 1.49 181 23.2 225 4.0 1,542 351 846 224 188 13.3 1,283
E 10–11 Oct 2001 144 1.48 ± 1.42 10 8.1 ND 9.7 896 143 604 24 87 8.0 351
F 17–18 Oct 2001 199 1.39 ± 1.45 58 7.0 116 4.6 1,361 184 586 59 96 18.7 545
G 23–24 Oct 2001 1,129 1.44 ± 1.37 215 13.7 382 12.6 1,383 379 530 116 177 33.0 969
Abbreviations: EC, elemental carbon; ND, not determined; OC, organic carbon. Each study is listed in the order that it was performed (1–6 for 1-day exposure studies and A–G for 2-day exposure studies).
a Particle size represents aerodynamic diameter based on number count per cubic centimeter measured as dN/dlogDp (mean diameter weighted by number/logarythmic interval of particle diameter). A minimum of one sample was collected per hour of exposure. Particle size data were averaged first over the course of each exposure for 1-day exposure studies, and then over 2 days for 2-day exposure studies, except for the exposures of 11 September where only 1 data point existed.
b Note that this value for copper obtained was very high for no explainable technical reason.
c Data collected on 11 September are for exposures < 1 hr.
Table 2 Levels of organic and major elemental constituents per given mass of CAP during 1-and 2-day exposure studies (microgram/milligram).
Study/daya Exposure year OC EC Sulfate Al Cu Pb Zn
A 2000 203 12.8 330 2.7 0.5 0.17 0.8
B 2000 175 9.7 505 0.8 0.1 0.1 0.4
C 2000 191 12.6 308 3.0 NA 0.16 0.9
D 2000 244 15.4 207 11.6 0.7 0.17 0.8
E 2000 205 10.8 262 7.7 0.2 0.21 0.6
F 2000 356 27.9 112 21.0 0.3 0.19 1.6
A1 2001 199 5.5 256 1.3 0.1 0.09 0.6
A2 2001 244 5.8 345 1.4 0.2 0.06 0.3
B2 2001 279 17.0 311 3.7 0.9 0.22 1.2
C1 2001 334 33.6 232 8.4 0.7 0.39 1.4
C2 2001 315 15.2 287 4.1 0.5 0.28 1.2
D1 2001 308 40.0 255 10.1 0.9 0.38 2.1
D2 2001 273 28.5 273 4.7 1.5 0.28 1.7
E1 2001 NA NA 57 111.5 4.8 0.17 2.4
E2 2001 NA 58.0 89 68.5 3.5 0.17 2.5
F1 2001 NA 56.1 373 24.8 3.8 0.41 3.6
F2 2001 434 24.3 247 22.8 2.5 0.24 2.3
G1 2001 351 14.6 179 10.0 0.7 0.11 0.8
G2 2001 323 9.4 203 3.2 0.2 0.1 0.9
Abbreviations: EC, elemental carbon; NA, not analyzed; OC, organic carbon.
a Numerals 1 and 2 refer to the first and second day of exposure, respectively. Note that during B2–F2 exposure days, levels of aluminum, copper, and zinc per given mass increased severalfold when compared with other exposure days.
Table 3 Combined mean values (± SE) obtained for breathing parameters in SH and WKY rats after exposure to filtered air or CAP for all 2-day exposure studies.
Test Expression unit WKY/air WKY/CAP SH/air SH/CAP
Frequency (f ) Breaths/min 266 ± 8 239 ± 11 290 ± 11 279 ± 8**
Tidal volume (Tv) mL 1.20 ± 0.04 1.30 ± 0.04 1.40 ± 0.05* 1.46 ± 0.06*
Minute volume (Mv) mL/min 265 ± 8 247 ± 11 341 ± 10* 342 ± 10*
Peak expiratory flow (PEF) mL/sec 12.2 ± 0.4 12.0 ± 0.5 16.9 ± 0.5* 17.5 ± 0.7*
Peak inspiratory flow (PIF) mL/sec 16.6 ± 0.4 16.1 ± 0.6 22.7 ± 1.0* 22.6 ± 1.5*
Expiratory time (Te) Sec 0.19 ± 0.01 0.24 ± 0.02 0.20 ± 0.02 0.21 ± 0.02
Inspiratory time (Ti) Sec 0.12 ± 0.00 0.13 ± 0.01 0.10 ± 0.00 0.11 ± 0.00
Pause (PAU) Nondimensional 0.62 ± 0.01 0.65 ± 0.02 0.64 ± 0.03 0.68 ± 0.03
Enhanced pause (PENH) Nondimensional 0.45 ± 0.01 0.50 ± 0.02 0.50 ± 0.03 0.54 ± 0.03
These data represent the absolute value of breathing parameters obtained after air or CAP exposure. Because the data in this table do not compare the net difference between baseline values obtained before and after air/CAP exposure, CAP-related differences may not be as apparent as shown in Figure 4. The number of observations/rats for WKY filtered air or CAP groups were 28, and for SH, 35.
* Significant strain effect at p ≤ 0.05.
** Significant CAP effect at p ≤ 0.05.
Table 4 Combined mean values (± SE) obtained for pulmonary injury/inflammatory markers analyzed in BALF of WKY and SH rats after exposure to filtered air or CAP for all 2-day exposure studies.
Parameter Expression unit WKY/FA WKY/CAP SH/FA SH/CAP
Total protein μg/mL 184 ± 34 148 ± 8 285 ± 19* 291 ± 17*
Albumin μg/mL 27.0 ± 7.8 19.9 ± 1.8 54.0 ± 4.3 54.1 ± 4.2
LDH activity U/L 32.8 ± 1.6 29.9 ± 2.1 18.1 ± 1.6* 16.8 ± 0.5*
GGT activity U/L 2.68 ± 0.21 2.92 ± 0.22 2.64 ± 0.15 3.55 ± 0.15**
NAG activity U/L 3.20 ± 0.20 3.10 ± 0.20 2.24 ± 0.11* 2.54 ± 0.11*
Glutathionea ng/mL 0.80 ± 0.09 1.11 ± 0.10 0.59 ± 0.04 0.65 ± 0.04
Ascorbate ng/mL 738 ± 45 818 ± 51 373 ± 17* 382 ± 18*
Uric acid ng/mL 72.3 ± 8.4 93.6 ± 9.1 38.2 ± 5.1* 43.4 ± 5.9*
Total cells Cells/mL × 105 1.22 ± 0.17 0.77 ± 0.06** 1.01 ± 0.05 1.04 ± 0.07
Macrophages Cells/mL × 105 1.12 ± 0.12 0.65 ± 0.06** 0.85 ± 0.05 0.87 ± 0.07
Neutrophils Cells/mL × 105 0.08 ± 0.01 0.11 ± 0.01** 0.15 ± 0.01* 0.17 ± 0.01*
IL-6b pg/mL 17.2 ± 2.1 19.8 ± 1.4 ND ND
TNF-α pg/mL 11.2 ± 1.2 14.0 ± 1.3 ND ND
MIP-2 pg/mL 354 ± 16 371 ± 16 ND ND
Abbreviations: FA, filtered air; MIP-2, macrophage inflammatory protein-2; ND, not determined; TNF-α, tumor necrosis factor-α. The number of observations/rats for WKY filtered air or CAP groups were 28, and for SH, 35.
a Note that the levels of total glutathione are near the lower detectable levels and therefore variable.
b Also IL-6 was below the detection limit in BALF (detection limits: IL-6, 62.5 pg/mL; MIP-2, 80 pg/mL; TNF-α, 9.37 pg/mL).
* Significant strain effect at p ≤ 0.05.
** Significant CAP effect at p ≤ 0.05.
==== Refs
References
Andrews E Saxena P Musarra S Hildemann LM Koutrakis P McMurry PH 2000 Concentration and composition of atmospheric aerosols from the 1995 SEAVS experiment and a review of the closure between chemical and gravimetric measurements J Air Waste Manage Assoc 50 648 664
Bianchi G Ferrari P Cusi D Salardi S Guidi E Niutta E 1986 Genetic and experimental hypertension in the animal model-similarities and dissimilarities to the development of human hypertension J Cardiovasc Pharmacol 8 S64 S70 2427888
Carvalho-Oliveira R Pozo RM Lobo DJ Lichtenfels AJ Martins-Junior HA Bustilho JO 2005 Diesel emissions significantly influence composition and mutagenicity of ambient particles: a case study in Sao Paulo, Brazil Environ Res 98 1 7 15721877
Clarke RW Catalano PJ Koutrakis P Krishna Murthy GG Sioutas C Paulauskis SJ 1999 Urban air particulate inhalation alters pulmonary function and induces pulmonary inflammation in a rat model of chronic bronchitis Inhal Toxicol 11 637 656 10477440
Clarke RW Coull B Reinisch U Catalano P Killingsworth CR Koutrakis P 2000 Inhaled concentrated ambient particles are associated with hematological and bronchoalveolar lavage changes in canines Environ Health Perspect 108 1179 1187 11133399
Dye JA Adler KB Richards JH Dreher KL 1999 Role of soluble metals in oil fly ash-induced airway epithelial injury and cytokine gene expression Am J Physiol 277 L498 L510 10484457
Ernst H Rittinghausen S Bartsch W Creutzenberg O Dasenbrock C Gorlitz BD 2002 Pulmonary inflammation in rats after intratracheal instillation of quartz, amorphous SiO2 , carbon black, and coal dust and the influence of poly-2-vinylpyridine-N -oxide (PVNO) Exp Toxicol Pathol 54 109 126 12211632
Gardner SY Lehmann JR Costa DL 2000 Oil fly ash-induced elevation of plasma fibrinogen levels in rats Toxicol Sci 56 175 180 10869466
Ghio AJ Hall A Bassett MA Cascio WE Devlin RB 2003 Exposure to concentrated ambient air particles alter hematologic indices in humans Inhal Toxicol 15 1465 1478 14648359
Ghio AJ Kim C Devlin RB 2000 Concentrated ambient air particles induce mild pulmonary inflammation in healthy human volunteers Am J Respir Crit Care Med 162 981 988 10988117
Gilmour PS Schladweiler MC Richards JH Ledbetter AD Kodavanti UP 2004 Hypertensive rats are susceptible to TLR4-mediated signaling following exposure to combustion source particulate matter Inhal Toxicol 16 suppl 1 5 18 15204789
Gordon T Gerber H Fang CP Chen LC 1999 A centrifugal particle concentrator for use in inhalation toxicology Inhal Toxicol 11 71 87 10380160
Gordon T Nadziejko C Chen LC Schlesinger R 2000 Effects of concentrated ambient particles in rats and hamsters: an exploratory study Res Rep Health Eff Inst 93 5 34 10897487
Harrison RM Yin J 2000 Particulate matter in the atmosphere: which particle properties are important for its effects on health? Sci Total Environ 249 85 101 10813449
Huang YC Ghio AJ Stonehuerner J McGee J Carter JD Grambow SC 2003 The role of soluble components in ambient fine particles-induced changes in human lungs and blood Inhal Toxicol 15 327 342 12635002
Kodavanti UP Hauser R Christiani DC Meng ZH McGee J Ledbetter A 1998 Pulmonary responses to oil fly ash particles in the rat differ by virtue of their specific soluble metals Toxicol Sci 43 204 212 9710962
Kodavanti UP Mebane R Ledbetter A Krantz T McGee J Jackson MC 2000a Variable pulmonary responses from exposure to concentrated ambient air particles in a rat model of bronchitis Toxicol Sci 54 441 451 10774827
Kodavanti UP Schladweiler MC Ledbetter AD Hauser R Christiani DC McGee J 2002 Temporal association between pulmonary and systemic effects of particulate matter in healthy and cardiovascular compromised rats J Toxicol Environ Health A 65 1545 1569 12396868
Kodavanti UP Schladweiler MC Ledbetter AD Watkinson WP Campen MJ Winsett DW 2000b The spontaneously hypertensive rat as a model of human cardiovascular disease: evidence of exacerbated cardiopulmonary injury and oxidative stress from inhaled emission particulate matter Toxicol Appl Pharmacol 164 250 263 10799335
McGee JK Chen LC Cohen MD Chee GR Prophete CM Haykal-Coates N 2003 Chemical analysis of World Trade Center fine particulate matter for use in toxicologic assessment Environ Health Perspect 111 972 980 12782501
NOAA 2004. Air Resources Laboratory. Washington,DC:National Oceanic and Atmospheric Administration. Available: http://www.arl.noaa.gov/ready/[accessed 1 October 2004].
Saldiva PH Clarke RW Coull BA Stearns RC Lawrence J Murthy GG 2002 Lung inflammation induced by concentrated ambient particles is related to particle composition Am J Respir Crit Care Med 165 1610 1617 12070061
Sioutas C Koutrakis P Burton RM 1995 A technique to expose animals to concentrated fine ambient aerosols Environ Health Perspect 103 172 177 7737066
Smith KR Kim S Recendez JJ Teague SV Menache MG Grubbs DE 2003 Airborne particles of the California central valley alter the lungs of healthy adult rats Environ Health Perspect 111 902 908 12782490
Tankersley CG Fitzgerald RS Levitt RC Mitzner WA Ewart SL Kleebergeer SR 1997 Genetic control of differential baseline breathing pattern J Appl Physiol 82 874 881 9074977
Turpine BJ Saxena P Andrews E 2000 Measuring and simulating particulate organics in the atmosphere: problems and prospects Atmos Environ 34 2983 3013
Urch B Brook JR Wasserstein D Brook RD Rajagopalan S Corey P 2004 Relative contribution of PM2.5 chemical constituents to acute arterial vasoconstriction in humans Inhal Toxicol 16 345 352 15204750
|
16263512
|
PMC1310919
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 27; 113(11):1561-1568
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7868
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8038ehp0113-00156916263513ResearchDo Organohalogen Contaminants Contribute to Histopathology in Liver from East Greenland Polar Bears (Ursus maritimus)? Sonne Christian 12Dietz Rune 1Leifsson Pall S. 3Born Erik W. 4Letcher Robert J. 5Kirkegaard Maja 1Muir Derek C. G. 6Riget Frank F. 1Hyldstrup Lars 71 Department of Arctic Environment, National Environmental Research Institute, Roskilde, Denmark2 Department of Veterinary Basic Sciences, and3 Department of Veterinary Pathobiology, Royal Veterinary and Agricultural University, Frederiksberg, Denmark4 Greenland Institute of Natural Resources, Nuuk, Greenland5 National Wildlife Research Centre, Canadian Wildlife Service, Environment Canada, Carleton University, Ottawa, Ontario, Canada6 National Water Research Institute, Environment Canada, Burlington, Ontario, Canada7 University Hospital of Hvidovre, Hvidovre, DenmarkAddress correspondence to C. Sonne, Arctic Wildlife Research Veterinarian, National Environmental Research Institute, Department of Arctic Environment, Frederiksborgvej 399, P.O. Box 358, DK-4000 Roskilde, Denmark. Telephone: 45-46-30-19-54. Fax: 45-46-30-19-14. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 5 7 2005 113 11 1569 1574 24 2 2005 5 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. In East Greenland polar bears (Ursus maritimus), anthropogenic organohalogen compounds (OHCs) (e.g., polychlorinated biphenyls, dichlorodiphenyltrichloroethane, and polybrominated diphenyl ethers) contributed to renal lesions and are believed to reduce bone mineral density. Because OHCs are also hepatotoxic, we investigated liver histology of 32 subadult, 24 adult female, and 23 adult male East Greenland polar bears sampled during 1999–2002. Light microscopic changes consisted of nuclear displacement from the normal central cytoplasmic location in parenchymal cells, mononuclear cell infiltrations (mainly portally and as lipid granulomas), mild bile duct proliferation accompanied by fibrosis, and fat accumulation in hepatocytes and pluripotent Ito cells. Lipid accumulation in Ito cells and bile duct hyperplasia accompanied by portal fibrosis were correlated to age, whereas no changes were associated with either sex or season (summer vs. winter). For adult females, hepatocytic intracellular fat increased significantly with concentrations of the sum of hexachlorocyclohexanes, as was the case for lipid granulomas and hexachlorobenzene in adult males. Based on these relationships and the nature of the chronic inflammation, we suggest that these findings were caused by aging and long-term exposure to OHCs. Therefore, these changes may be used as biomarkers for OHC exposure in wildlife and humans. To our knowledge, this is the first time liver histology has been evaluated in relation to OHC concentrations in a mammalian wildlife species, and the information is important to future polar bear conservation strategies and health assessments of humans relying on OHC-contaminated food resources.
bile duct proliferationchlordanesdichlorodiphenyltrichloroethanedieldrinEast GreenlandHCBhexacyclohexanesIto cellslipid granulomaslivermononuclear cell infiltrationspolar bearpolybrominated diphenyl etherspolychlorinated biphenylsportal fibrosis∑DDT∑HCH∑PBDE∑PCBUrsus maritimus
==== Body
In rats and mink, several acute studies of polychlorinated biphenyls (PCBs) have associated these compounds with hepatotoxicity (Bergman et al. 1992; Bruckner et al. 1974; Chu et al. 1994; Jonsson et al. 1981; Kelly 1993; Kimbrough et al. 1971; MacLachlan and Cullen 1995; Parkinson 1996). Specifically in the liver, acute organohalogen compound (OHC) toxicity is mediated through sub-cellular toxicity, leading to impaired ATP, protein synthesis, and other changes (Kelly 1993; Parkinson 1996), and chronic exposure also may affect endocrine homeostasis via up-regulation of cytochrome P450 isozymes (e.g., CYP1A and CYP1B) (Boon et al. 1992; Lin et al. 2003; van Duursen et al. 2003; Wong et al. 1992).
In marine wildlife, chronic exposure to organohalogen compounds [OHCs; e.g., PCBs, dichlorodiphenyltrichloroethane (DDT), and polybrominated diphenyl ethers (PBDEs)] has been associated with toxic effects on several organ systems (Bergman 1999; Bergman and Olsson 1985; Bergman et al. 2001; Schumacher et al. 1993). However, histologic liver changes associated with high environmental levels of OHCs in wildlife have been investigated only in birds, such as cormorants (Phalacrocorax carbo) (Fabczak et al. 2000), and fish, such as common bream (Abramis brama) (Koponen et al. 2001), but never in marine or terrestrial mammals.
Polar bears are the most OHC-contaminated species in the Arctic, and those from East Greenland and Svalbard (Norway) carry the most contaminants because of their reliance on OHC-polluted blubber, mainly from ringed seal (Phoca hispida) and bearded seal (Erignathus barbatus), contaminanted by OHCs originating from lower-latitude airborne pollution [Arctic Monitoring and Assessment Programme (AMAP) 2004; de March et al. 1998; Ramsay and Stirling 1988]. At Svalbard, recent studies of PCBs and organochlorine (OC) pesticides in polar bears have indicated negative associations with plasma testosterone (males), progesterone (females), cortisol (both sexes), retinol (both sexes), and thyroxine hormone (both sexes) (Braathen et al. 2004; Haave et al. 2003; Oskam et al. 2003, 2004; Skaare et al. 2001). Additionally, high levels of PCBs/OC pesticides were associated with low levels of IgG in the Svalbard bears, suggesting possible immunotoxic effects (Bernhoft et al. 2000; Lie et al. 2004, 2005). In East Greenland polar bears, OHCs are believed to reduce bone mineral density (BMD) and to be a cofactor in the development of renal lesions and splenic changes (Kirkegaard et al. 2005; Sonne et al. 2004, in press). To determine if OHCs are also a cofactor in hepatotoxicity, liver tissue histology was examined in 79 East Greenland polar bears sampled during the subsistence hunt from 1999 to 2002, and liver histology was compared with individual OHC adipose tissue levels in 65 of the bears. These new results are intended to fill part of the existing knowledge gap in understanding the significance, nature, and effects of chronic environmental OHC exposure.
Materials and Methods
Sampling.
All polar bear samples were collected from January through September by local subsistence hunters in the Scoresby Sound area in central East Greenland (69°00′N to 74°00′N) during 1999–2002. A tissue subsample from the periphery of a randomly chosen liver lobe was collected from 79 individuals and fixed in a phosphate-buffered formaldehyde/alcohol solution (3.5% formaldehyde, 86% ethanol, and 10.5% H2O), which prevented freeze damage. In addition, sternal subcutaneous adipose tissue was sampled from 65 of the individuals for OHC analyses and stored in separate poly-ethylene plastic bags until arrival at the laboratory in Roskilde, where they were transferred into rinsed [acetone (Supra solv. 1.00012), n-hexane (Unisolv 1.04369) both from Merck, KGaA, Darmstadt, Germany] glass containers, and covered with aluminum foil in between the sample and the plastic lid. All samples were taken < 12 hr postmortem and preserved frozen during the hunt and later kept at −20°C before preparation and examination at the veterinary pathology laboratory in Copenhagen, Denmark (histology); GLIER, Windsor, Ontario, Canada (organochlorines); and NWRI, Burlington, Ontario, Canada (PBDEs).
Age estimation.
The age determination was carried out by counting the cementum growth layer groups of the lower left incisor (I3) after decalcification, thin sectioning (14 μm), and staining (toluidine blue) using the method described by Dietz et al. (1991) and Hensel and Sorensen (1980). When necessary, the individuals were categorized as adult males (≥ 6 years of age), adult females (≥ 5 years of age), and subadults (those remaining) (Rosing-Asvid et al. 2002). In the evaluation of sex difference in the prevalence of histologic liver changes, bears were categorized as old at ≥ 15 years of age based on Derocher and Stirling (1994).
Histology.
The liver tissue was trimmed, processed conventionally, embedded in paraffin, sectioned at about 4 μm, and stained with hematoxylin (aluminum-hematein) and eosin (H&E) and periodic acid-Schiff for routine diagnostics; Van Gieson and Masson Trichrome to detect fibrous tissue (collagen); Best’s carmine to demonstrate glycogen storage; Sudan III to detect lipid (frozen tissue); and Perls’ Prussian blue reaction and Schmorl technique for detecting hemosiderin and lipofuscin pigments, respectively (Bancroft and Stevens 1996; Lyon et al. 1991).
We evaluated six histologic changes and grouped them semiquantitatively as follows:
Portal mononuclear cell infiltrations: absent, unifocally, multifocally, or diffuse
Random mononuclear cell infiltrations: absent, unifocally, multifocally, or diffuse
Lipid granulomas: average number in five fields at 10 × magnification
Hepatocytic intracellular fat: absent, foamy, multifocal macrovesiculary, or diffuse macrovesiculary
Visible Ito cells: average number in five fields at 20 × magnification
Mild multifocal bile duct hyperplasia accompanied by portal fibrosis: absent or present.
For each histologic change, the degree of change was measured as follows:
Portal mononuclear cell infiltrations: mild (unifocally), moderate (multifocally), severe (diffuse)
Random cell infiltrations: mild (< 1), moderate (1–3), severe (> 3)
Lipid granulomas: mild (< 1), moderate (1 to < 2), severe (2–5)
Hepatocytic intracellular fat: mild (foamy), moderate (multifocal macrovesiculary), severe (diffuse macrovesiculary)
Ito cells: mild (< 10), moderate (10 to < 50), severe (50–200).
Analyses of OHCs.
Polar bear subcutaneous adipose tissue samples (n = 65) were analyzed for PCBs, DDTs, chlordanes (CHLs), dieldrin, hexacyclohexanes (HCHs), and hexachlorobenzene (HCB) according to Dietz et al. (2004) and Sandala et al. (2004) at the Great Lakes Institute for Environmental Research (University of Windsor, Windsor, Ontario, Canada). An external standard quantification approach used for PCBs and OC pesticides in the subcutaneous adipose tissues was based on peak area of the gas chromatography-electron capture detection response, which is described in detail by Luross et al. (2002).
Briefly, ∑PCB is the sum of the concentrations of the 51 individual or coeluting PCB congeners (if detected), given by International Union of Pure and Applied Chemistry (IUPAC) number: 31/28, 52, 49, 44, 42, 64/71, 74, 70, 66/95, 60, 101/84, 99, 97, 87, 110, 151, 149, 118, 146, 153, 105, 141, 179, 138, 158, 129/178, 182/187, 183, 128, 174, 177, 171/202/156, 200, 172, 180, 170/190, 201, 203/196, 195, 194, and 206. ∑DDT is the sum of 4,4′-DDT, 4,4′-dichlorodiphenyl-dichloroethane (DDD), and 4,4′-dichlorodiphenyldichloroethylene (DDE). ∑HCH is the sum of the α-, β-, and γ-hexachlorocyclohexane. ∑CHL is the sum of oxychlordane, trans-chlordane, cis-chlordane, trans-nonachlor, cis-nonachlor, and heptachlor epoxide. OHC fractions were subsequently sent to the National Water Research Institute for determination of brominated diphenyl ether (PBDE) flame retardants. PBDEs (n = 65) were determined by electron capture negative ion (low resolution) mass spectroscopy using an external standard. Briefly, ∑PBDE is the sum of the concentrations of the 35 individual or coeluting congeners (if detected), given by IUPAC number: 10, 7, 11, 8, 12/13, 15, 30, 32, 28/33, 35, 37, 75, 71, 66, 47, 49, 77, 100, 119, 99, 116, 85, 155/126, 105, 154, 153, 140, 138, 166, 183, 181, and 190. Gas chromatographic conditions for the PBDEs were as described by Luross et al. (2002).
Statistics.
The statistical analyses were performed with SAS statistical software (version 8, and Enterprise Guide, version 1; SAS Institute, Cary, NC, USA); the level of significance was set at p ≤ 0.05, and levels of significance at 0.05 < p ≤ 0.10 were considered a trend. The OHC data were log-transformed (base e) before the analyses in order to meet the assumption of normality and homogeneity of the variance.
For each specific histologic liver change, we performed a one-way analysis of variance (ANOVA) to test for differences in mean age between individuals with and without that specific histologic liver change (Table 1). In the case of hepatocytic lipid, we compared foamy cytoplasm with macrovesicular lipid. Furthermore, we tested whether there was a relationship between sex or season (summer, 1 June through 30 September; winter, 1 October through 31 May), and histologic liver changes using a chi-square test. In the case of Ito cells and bile duct hyperplasia accompanied by portal fibrosis, we performed the chi-square test within subadult, adult, and old bears to determine age dependency. The chi-square test was also used to test the relationship between Ito cells and fatty granulomas.
We then performed a one-way ANOVA to test for differences in mean concentrations of each group of OHCs (PCBs, DDTs, CHLs, dieldrin, HCHs, HCB, and PBDEs) between subadults, adult females, and adult males (Table 2). The results were then evaluated from Tukey’s post hoc test. In order to test the relationship between concentrations of OHCs and age, we used a linear regression model for subadults, adult females, and adult males.
Finally, we tested the relationship between the concentrations of each group of OHCs (PCBs, DDTs, CHLs, dieldrin, HCHs, HCB, and PBDEs, respectively) and each histologic liver change (absent vs. present) by an analysis of covariance (Table 3). This was conducted for each of the three age/sex groups using OHC concentration as the dependent variable, age as the covariable, and histologic liver change as the class variable, including their first-order interaction links (age × histologic liver change). The statistical analyses were performed separately on subadults, adult females, and adult males in cases of CHLs, dieldrin, HCHs, and HCB, because the age relationships and/or concentrations differed among these three age/sex groups. In the case of lipid granulomas, the relationship to OHCs was analyzed based on the presence or absence of Ito cells. After a successive reduction of non-significant interactions, judged from the type III sum of squares (p ≤ 0.05), the significance of each of the remaining factors was evaluated from the final model least-square mean.
Results
We studied a total of 79 free-ranging East Greenland polar bears (24 subadults, 24 adult females, 22 adult males, 4 old females, and 5 old males), collected from 1999 through 2002 (Table 1). No background data describing the general liver histology of free-ranging polar bears were available in the scientific literature. The morphology of the liver tissue was similar to other carnivorous species; however, interlobular fibrous septa were lacking as in other ursid species (Frappier 1998; Heier et al. 2003, in press; Kelly 1993; Leighton et al. 1988; MacLachlan and Cullen 1995; Prunescu et al. 2003). Kupffer cells, located in the space of Disse, tested positive for hemosiderin (iron pigments) (Lyon et al. 1991), and hepatocytes tested positive for deposits compatible with glycogen (Bancroft and Stevens 1996). In all individuals, parenchymal cells exhibited nuclear displacement toward the cell membrane (Figure 1) (Sato et al. 2001).
Mononuclear cell infiltrations and lipid granulomas.
We found portal mononuclear cell infiltrations (lymphocytes, macrophages, and neutrophils), as described by Kelly (1993) and MacLachlan and Cullen (1995), in 18% of the animals and multifocally mononuclear cell infiltrations in 12% of the bears examined (Table 1, Figure 1). Additionally, we detected lipid granulomas, also described by these authors, in 76% of the animals. None of these three cell infiltration types was related to age, sex, or season (all, p > 0.05) (Table 1). Finally, we found a trend of livers with visible Ito cells showing a larger frequency of fatty granulomas, compared with livers without visible Ito cells (p < 0.06).
In addition, we found one case of unifocal necrosis and a single case of fibrin exudation, described by Kelly (1993) and MacLachlan and Cullen (1995), but we did not investigate the significance further.
Lipids.
All animals showed hepatocytic microvesicular lipid accumulation (foamy cytoplasm), and 84% showed sharply demarcated macrovesicular lipid vacuoles in mainly periacinar (zones 2–3) hepatocytes (Table 1, Figure 2). In addition, we found non-parenchymal lipid vacuoles of diverging size and numbers in centroacinary Ito cells—located in the narrow space of Disse, between hepatocytes—mainly periacinary (zones 2–3) (Table 1, Figure 2) (Kelly 1993; Leighton et al. 1988; MacLachlan and Cullen 1995; Senoo et al. 1999, 2001). Intrahepatocytic lipid accumulation was not related to age (p > 0.05), whereas Ito cell lipid accumulation was highly related to age (p < 0.01) (Table 1). None of the lipid changes was related to sex or season (summer vs. winter) (Table 1).
Bile duct proliferation and portal fibrosis.
Mild bile duct proliferation accompanied by portal fibrosis was found in 8% of the animals (Table 1, Figure 3). These changes were associated with age (both, p < 0.01); no relationships were found to sex or season (Table 1).
OHCs and histologic changes.
Levels of ∑PCB, ∑CHL, ∑DDT, dieldrin, ∑HCH, HCB, and ∑PBDE in 65 of the examined polar bears are presented in Table 2. ∑CHL, ∑PCB, ∑DDT, dieldrin, ∑HCH, and ∑PBDE did not differ significantly among age/sex groups, but HCB was higher in subadults when compared with adult males (p ≤ 0.05) (Table 2). We found a significant negative relationship between age and HCHs, HCB, and dieldrin (all, p < 0.05) for adult females, and between age and ∑CHL and dieldrin in adult males (both, p < 0.01) (Table 2). Further information about age and sex variation of OHCs in the present East Greenland polar bears has been published by Dietz et al. (2004) and Sandala et al. (2004).
The statistical analyses were performed separately on subadults, adult females, and adult males in cases of ∑CHL, dieldrin, ∑HCH, and HCB because concentrations and/or age relationships differed between the three groups of individuals (Table 2). We tested whether the concentrations of each OHC group differed between the degree of histologic liver changes (absent vs. present); for adult females we found a significant relationship between ∑HCH and hepatocytic macrovesicular lipids (vacuoles), and for adult males we found a significant relationship between HCB and lipid granulomas (both, p < 0.05) (Table 3).
Discussion
We found nuclear displacement toward the cell membrane in all individuals. In studies of polar bears from Svalbard, Sato et al. (2001) revealed the same findings. It has been proposed that this displacement is related to the high vitamin A accumulation (natural storage) in Ito cell cytoplasmic lipid droplets and hepatocytes, accumulated through the extensive feeding on blubber from ringed seal and bearded seal (Käkelä et al. 1997; Ramsay and Stirling 1988). In general, such a displacement is associated with hepatitis, carcinomas, hyperplasia (adenomatous), or regeneration (Sato et al. 2001). However, such changes were not found in the Svalbard study (Sato et al. 2001), and in only two cases were unifocal hepatitis and regeneration found in the present study. We could not evaluate whether there was a relation between nuclear displacement and OHCs or hepatocytic lipid accumulation because we found the displacement in nearly all individuals. Therefore, we hypothesize that displacement may be a natural phenomenon in free-ranging polar bears, probably related to vitamin A intake and/or a result of lipid/OHCs accumulation (cytoskeletal displacement).
Mononuclear cell infiltrations and lipid granulomas.
Mononuclear cell infiltrates—accompanied by fibrosis—is a reaction to local depositioning of microorganisms and/or injury of local blood vessels from, for example, toxic compounds (Kelly 1993; MacLachlan and Cullen 1995). These cell infiltrates are therefore a nonspecific inflammatory reaction that can be linked to even minor tissue damage (Kelly 1993; MacLachlan and Cullen 1995). The fact that liver tissue, rich in visible Ito cells, had a higher number of lipid granulomas indicates that microorganisms (originating from the blood supply) play a role in the random multifocal necrosis (rupture of Ito cells) observed (Kelly 1993; MacLachlan and Cullen 1995). However, if the lipophilic toxic OHCs accumulate in the lipid rich Ito cells, we hypothesize that OHCs may play a role in the burst of Ito cells, as well.
Lipids.
In the present study, we found macrovesicular lipid in periacinar hepatocytes. Because polar bears are hyperphagic from April to July, they build up their fat deposits during this period (Messier et al. 1992; Ramsay and Stirling 1988), and a seasonal pattern in Ito cell numbers may be expected as was the case for the fatty tissue lipid percentage (Dietz et al. 2004). Intrahepatocytic accumulated lipid vacuoles showed a zonary pattern similar to that found in individuals exposed to toxic substances, which produce a characteristic periacinar injury due to the low oxygen gradient (hypoxia and high concentrations of, for example, cytochrome P450). This could sensitize the liver parenchyma in this zone to metabolic disorders, resulting in lipid accumulation (Kelly 1993; MacLachlan and Cullen 1995; Parkinson 1996).
We also found lipid accumulation in periacinary Ito cells. In polar bears, the Ito cells are one of the major accumulation and storage sites for lipophilic vitamin A (Leighton et al. 1988; Senoo et al. 1999, 2001) and probably also lipophilic OHCs, as mentioned above. As for hepatocytic lipid accumulation, we did not find a seasonal pattern in the number of Ito cells, but we did find that the number of Ito cells is related to age. If the Ito cell number reflects the vitamin A exposure through marine prey species, mainly ringed seal and bearded seal (Ramsay and Stirling 1988), young bears would have lower numbers of Ito cells because they do not start eating prey rich in vitamin A until they are weaned at approximately 2 years of age (Derocher and Stirling 1994). This may then explain the age difference in the number of Ito cells in the liver.
Bile duct proliferation and portal fibrosis.
Bile duct proliferation has been associated with toxic injury, parasitism, or periductular fibrosis in terrestrial animals (Kelly 1993; MacLachlan and Cullen 1995) and is therefore a non-specific reaction to chronic extrinsic and/or environmental factors. Specifically in arctic mammals, bile duct proliferations have been reported in arctic beluga whale (Delphinapterus leucas), but the pathogenesis of this could not be determined (Woshner et al. 2002).
Age-related portal fibrosis, due to chronic infections (cholangitis and biliary obstruction), is a common nonspecific histologic diagnosis in mammals (Kelly 1993; MacLachlan and Cullen 1995), and it has been reported in the Romanian brown bear (Ursus arctos) (Prunescu et al. 2003) and arctic beluga whale (Woshner et al. 2002). Prunescu et al. (2003) showed seasonal liver fibrosis (highest in spring) of the hepatic venous system, possibly due to pre-hibernation physiologic adaptations. Our findings were not in agreement with such a seasonal fibrosis pattern, however, because portal fibrosis was present with bile duct proliferations in all individuals.
Liver changes and OHCs.
To our knowledge, liver histology in relation to environmental levels of OHCs has been studied only in birds, such as cormorants (Fabczak et al. 2000), and fish, such as common bream (Koponen et al. 2001), but never in marine or terrestrial mammals. Therefore, it is difficult to evaluate the relationship between liver histology and chronic exposure to environmental levels of OHCs in the East Greenland polar bear because basic knowledge in this field is extremely sparse.
Mononuclear cell infiltrates (lymphocytes and neutrophils) randomly distributed (lipid granulomas) or portally (around triads) have been associated with subacute PCB exposure in mink (Mustela vison) (Bergman et al. 1992). We found the same pattern in polar bears, which supports the hypothesis that OHCs could be a cofactor in the liver changes of the East Greenland polar bears in the present study. However, this could also be a result of microorganisms. Although the results from the laboratory studies are nonspecific reactions, parallels to our results are obvious.
Hepatotoxic substances (e.g., copper, pyrrolizidine alkaloids, carbon tetrachloride, and phytotoxins) usually produce a periacinar zone 2–3 injury due to the low oxygen gradient (hypoxia) and high concentrations of, for example, cytochrome P450 isozymes (activation of reactive metabolites) of this zone (Kelly 1993; MacLachlan and Cullen 1995; Parkinson 1996). We found such a zonary appearance in hepatocytic accumulation in the polar bears in the present study. Abnormal amounts of fat are known to be accumulated in the liver during high lipid ingestion, starvation, abnormal hepatocytic function, excessive dietary intake of carbohydrates, and decreased synthesis of -apoproteins (lipoproteins) (Kelly 1993; MacLachlan and Cullen 1995; Parkinson 1996). Hence, the large content of lipids in polar bear livers could be a function of hyperphagia and starvation due to seasonal changes in food resources, as discussed above, although we did not find a seasonal pattern. However, acute toxic investigations of PCBs, DDTs, and dieldrin in laboratory rats have shown to induce high lipid accumulation—probably due to decreased production of lipoproteins through impaired ATP synthesis and protein synthesis—in periacinary hepatocytes (accumulated as foamy cytoplasm or large vacuoles) (Bergman et al. 1992; Bruckner et al. 1974; Kelly 1993; Kimbrough et al. 1971, 1972; MacLachlan and Cullen 1995; Parkinson 1996). Therefore, OHCs may be a cofactor in the development of lipid accumulation in the present study, although significant differences in OHC concentrations were not found.
The signs of chronic inflammation, also in relation to Glisson’s triads (bile duct proliferation accompanied by portal fibrosis), as well as the hepatocytic lipid accumulation, could possibly indicate long-term exposure to liver toxic substances (OHCs) in the East Greenland polar bear, as well. However, other than the OHC considerations and age, liver histology in free-ranging Atlantic bottlenose dolphin (Tursiops truncatus) (Rawson et al. 1993) and arctic beluga whale (Woshner et al. 2002), in relation to mercury exposure, have shown changes similar to those in the present study. The East Greenland polar bears in the present study have also accumulated considerable amounts of mercury in the liver tissue (2.13–13.4 μg/g wet weight) (Dietz et al. 1990, 2000), which are in the range of adverse toxic effect levels for terrestrial mammals (Thompson 1996).
Conclusions
In the present study, we found the following histologic changes in liver tissue from 79 East Greenland polar bears: nuclear displacement, mononuclear cell infiltrations, mild bile duct proliferation accompanied by portal fibrosis, and fat accumulation. Two of the changes (Ito cells and bile duct hyperplasia accompanied by portal fibrosis) were related to age, whereas none were related to sex or season. The signs and type of chronic inflammation, and the zonary lipid accumulation in hepatocytes, may indicate chronic exposure to environmental levels of OHCs. In addition, we found significant relationships for ∑HCH and hepatocytic lipid accumulation in adult females and between HCB and lipid granulomas in adult males. We therefore suggest that the histologic changes were a result of aging and long-term exposure to OHCs, but other environmental factors, such as microorganisms and mercury, cannot be excluded.
Correction
The range of mercury in liver tissue of East Greenland polar bears was incorrect in the original manuscript published online but has been corrected here. The authors also found additional information that was not included in their original manuscript: Hori et al. [Hori S, Obana H, Kashimoto T, Otake T, Nishimura H, Ikegami N, et al. 1982. Effect of polychlorinated biphenyls and polychlorinated quaterphenyls in cynomolgus monkey (Macaca fascicularis). Toxicology 24(2):123–139] found an association between bile duct proliferation and PCB exposure, and also reported that mononuclear cell infiltrates were associated with subacute PCB exposure in cynomolgus monkeys (Macaca fascicularis). Also, the authors would like to state that it is impossible to evaluate whether liver changes and possible demineralization of the skeletal system (Sonne et al. 2004) and renal lesions (Sonne et al., in press) have an impact on the health status of each individual polar bear.
We thank H. Tuborg, B. Sandell, J. Brønlund, and local hunters for organizing sampling in East Greenland, and E. Heier for sharing digital images.
Financial support was provided by the Danish Cooperation for Environment in the Arctic, the Commission for Scientific Research in Greenland, and the Canada Research Chairs Program.
Figure 1 Liver tissue stained with H&E showing portal mononuclear cell infiltration in a 3.5-year-old (subadult) female (A; 10×), random mononuclear cell infiltration in a 20-year-old female (B; 20×), and lipid granulomas in a 16-year-old female (C; 40×) in liver tissue stained with H&E. Note the abnormal localization of the hepatocytic nuclei in (C). Bars = 50 μm.
Figure 2 Lipid accumulation in liver tissue stained with H&E. (A) Zone 2–3 hepatocytic macrovesicular lipid (vacuoles; 2.5×) in a 4-year-old (subadult) female; inset, taken from (A; 10×). (B) Ito cell lipid accumulation in a 20-year-old female; 10×. Bars = 25 μm.
Figure 3 Mild bile duct proliferation accompanied by portal fibrosis (H&E; 20×). Bar = 50 μm.
Table 1 Prevalence of histologic liver changes in relation to age, sex, and season in 79 East Greenland polar bears sampled during 1999–2002.
Degree of change [% (n)]
Histologic liver change Absent Mild Moderate Severe Age [(F )p] Sex [ (F )p] Season [(F )p]
Portal mononuclear cell infiltrations 82 (65) 8 (6) 8 (6) 2 (2) NS NS NS
Random mononuclear cell infiltrations 87 (69) 11 (9) 1 (1) 0 (0) NS NS NS
Hepatocytic intracellular fat 0 (0) 16 (13) 24 (19) 60 (47) NS NS NS
Lipid granulomas 24 (19) 35 (28) 32 (25) 9 (7) NS NS NS
Lipid accumulation in Ito cells 25 (20) 18 (14) 24 (19) 33 (26) (8)* NS NS
Mild bile duct hyperplasia with fibrosis 92 (73) 8 (6) 0 (0) 0 (0) (11)* NS NS
NS, not significant. Hepatic changes are divided into degrees of change (absent, mild, moderate, and severe); see “Materials and Methods” for criteria.
* Individuals with histologic liver changes were significantly older (mean age) than individuals without histologic liver changes (p < 0.01).
Table 2 OHC concentrations (mean ± SD, ng/g lipid weight) in subcutaneous adipose tissue of 65 East Greenland polar bears investigated for histologic liver changes during 1999–2001.
OHCs Subadults (n = 27) Adult females (n = 21) Adult males (n = 17)
∑PCB 6,130 ± 3,290 5,303 ± 2,157 7,081 ± 3,197
∑DDT 468 ± 240 380 ± 206 476 ± 259
∑CHL 1,518 ± 1,009 1,349 ± 559 1,016 ± 576*
Dieldrin 215 ± 114 179 ± 59** 172 ± 93#
∑HCH 184 ± 73 182 ± 155## 217 ± 144
HCB 114 ± 103 75 ± 68† 51 ± 32††
∑PBDE 57 ± 32 59 ± 36 51 ± 32
* Significant negative relationship with age (p < 0.01; R2 = 0.51).
** Significantly negative relationship with age (p ≤ 0.05; R2 = 0.26).
# Significant negative relationship with age (p < 0.01; R2 = 0.45).
## Significantly negative relationship with age (p ≤ 0.05; R2 = 0.25).
† Significantly negative relationship with age (p ≤ 0.05; R2 = 0.2).
†† Significantly lower compared with subadults (p ≤ 0.05).
Table 3 Significant results from analyses of relationships between histologic liver changes and OHCs in adult female and male East Greenland polar bears, 1999–2001.
Age/sex group Histologic liver change OHCs (n, F, R2)p
Adult females Hepatocytic intracellular fat ∑HCH (17, 8.5, 0.42)*
Adult males Lipid granulomas HCB (21, 9.8, 0.52)**
* Significantly higher OHC level (least square mean) in individuals with mild/moderate changes than in individuals without changes (p ≤ 0.05).
** Significantly higher OHC level (least-square mean) in individuals with mild/moderate changes than in individuals without changes (p < 0.01).
==== Refs
References
AMAP 2004. AMAP Assessment 2002: Persistent Organic Pollutants in the Arctic. Oslo:Arctic Monitoring and Assessment Programme.
Bancroft JD Stevens A 1996. Theory and Practice of Histological Techniques. New York:Churchill Livingstone.
Bergman A 1999 Health condition of the Baltic grey seal (Halichoerus grypus ) during two decades. Gynaecological health improvement but increased prevalence of colonic ulcers Apmis 107 3 270 282 10223299
Bergman A Bäcklin BM Järpild B Grimelius L Wilander E 1992 Influence of commercial polychlorinated biphenyls and fractions there of on liver histology in female mink (Mustela vison ) Ambio 21 8 591 595
Bergman A Bergstrand A Bignert A 2001 Renal lesions in Baltic grey seals (Halichoerus grypus ) and ringed seals (Phoca hispida botnica ) Ambio 30 7 397 409 11795214
Bergman A Olsson M 1985 Pathology of baltic grey seal and ringed seal females with special reference to adrenocortical hyperplasia: is environmental pollution the cause of a widely distributed disease syndrome? Finnish Game Res 44 47 62
Bernhoft A Skaare JU Wiig Ø Derocher AE Larsen HJS 2000 Possible immunotoxic effects of organochlorines in polar bears (Ursus maritimus ) at Svalbard J Toxicol Environ Health A 57 7 561 574 10777247
Boon JP Everaarts JM Hillebrand MTJ Eggens ML Pijnenburg J Goksoyr A 1992 Changes in levels of hepatic biotrans-formation enzymes and hemoglobin levels in female plaice (Pleuronectes platessa ) after oral-administration of technical polychlorinated biphenyl mixture (Clophen A40) Sci Total Environ 114 113 133 1594918
Braathen M Derocher AE Wiig Ø Sørmo EG Lie E Skaare JU 2004 Relationships between PCBs and thyroid hormones and retinol in female and male polar bears Environ Health Perspect 112 826 833 15175168
Bruckner JV Khanna KL Cornish HH 1974 Effect of prolonged ingestion of polychlorinated biphenyls on the rat Food Cosmet Toxicol 12 323 330 4140828
Chu I Villeneuve DC Yagminas A LeCavalier P Poon R Feeley M 1994 Subchronic toxicity of 3,3′,4,4′,5-pentachlorobiphenyl in the rat. Clinical, biochemical, hematological and histopathological changes Fundam Appl Toxicol 22 457 468 8050640
de March BGE de Wit CA Muir DCG Braune B Gregor DJ Norstrom RJ 1998. Persistent organic pollutants. In: AMAP Assessment Report: Arctic Pollution Issues. Oslo:Arctic Monitoring and Assessment Programme, 183–372.
Derocher AE Stirling I 1994 Age-specific reproductive-performance of female polar bears (Ursus maritimus ) Can J Zool 234 4 527 536
Dietz R Heide-Jørgensen MP Härkönen T Teilmann J Valentin N 1991 Age determination of european harbour seal (Phoca vitulina L .) Sarsia 76 17 21
Dietz R Overgaard Nielsen C Munk Hansen M Hansen CT 1990 Organic mercury in Greenland birds and mammals Sci Total Environ 95 41 51 2402624
Dietz R Riget F Born EW 2000 Geographical differences of zinc, cadmium, mercury and selenium in polar bears (Ursus maritimus ) from Greenland Sci Total Environ 245 25 48 10682354
Dietz R Riget FF Sonne-Hansen C Letcher RJ Born EW Muir DCG 2004 Seasonal and temporal trends in polychlorinated biphenyls and organochlorine pesticides in East Greenland polar bears (Ursus maritimus ), 1990–2001 Sci Total Environ 331 107 124 15325144
Fabczak J Szarek J Andrzejewska A Smoczynski SS 2000 The PCB level and ultrastructural pattern of the liver of cormorants Med Weter 56 12 788 792
Frappier BL 1998. Digestive system. In: Textbook of Veterinary Histology (Dellmann HD, Eurell JA, eds). Baltimore, MD:Lippincott Williams & Wilkins, 164–202.
Haave M Ropstad E Derocher AE Lie E Dahl E Wiig Ø 2003 Polychlorinated biphenyls and reproductive hormones in female polar bears at Svalbard Environ Health Perspect 111 431 436 12676595
Heier A Grone A Vollm J Kubber-Heiss A Bacciarini LN 2003 Immunohistochemical study of retinol-binding protein in livers of polar bears (Thalarctos maritimus ) Vet Pathol 40 2 196 202 12637760
Heier A Sonne C Gröne A Leifsson PS Dietz R Born EW In press. Liver RBP immunohistochemistry in free-ranging polar bears. An immunohistochemical study of retinol-binding protein (RBP) in livers of polar bears (Ursus maritimus) from East Greenland. Zoo Wild Med.
Hensel RJ Sorensen FE 1980 Age determination of live polar bears Int Conf Bear Res Manage 4 93 100
Jonsson HT Walker EM Greene WB Hughson MD Hennigar GR 1981 Effects of prolonged exposure to dietary DDT and PCB on rat liver morphology Arch Environ Contam Toxicol 10 171 183 7224669
Käkelä R Hyvärinen H Käkelä A 1997 Vitamins A1 (retinol), A2 (3,4-didehydroretinol) and E (γ-tocopherol) in the liver and blubber of lacustrine and marine ringed seals (Phoca hispida sp.) Comp Biochem Physiol 116B 1 27 33
Kelly WR 1993. The liver and biliary system. In: Pathology of Domestic Animals (Jubb KVF, Kennedy PC, Palmer N, eds). San Diego:Academic Press, 319–406.
Kimbrough RD Gaines TB Linder RE 1971 The ultrastructure of livers of rats fed DDT and dieldrin Arch Environ Health 22 460 467 5545497
Kimbrough RD Linder RE Gaines TB 1972 Morphological changes in livers of rats fed polychlorinated biphenyls. Light microscopy and ultrastructure Arch Environ Health 25 354 364 4346789
Kirkegaard M Sonne C Leifsson PS Dietz R Born EW Letcher RJ Muir DCG 2005 Histology of selected immunological organs in polar bear (Ursus maritimus) from East Greenland in relation to levels of organohalogens Sci Total Environ 341 1–3 119 132 15833246
Koponen K Myers MS Ritola O Huuskonen SE Seppa-Lindstrom P 2001 Histopathology of feral fish from a PCB-contaminated freshwater lake Ambio 30 3 122 126 11436658
Leighton FA Cattet M Norstrom R Trudeau S 1988 A cellular basis for high-levels of vitamin-A in livers of polar bears (Ursus maritimus )—the Ito cell Can J Zool 66 2 480 482
Lie E Larsen HJS Larsen S Johansen GM Derocher AE Lunn NJ 2004 Does high organochlorine (OC) exposure impair the resistance to infection in polar bears (Ursus maritimus )? Part I: Effect of OCs on the humoral immunity J Toxicol Environ Health A 67 555 582 15129552
Lie E Larsen HJS Larsen S Johansen GM Derocher AE Lunn NJ 2005 Does high organochlorine (OC) exposure impair the resistance to infection in polar bears (Ursus maritimus )? Part II: Possible effects of OCs on mitogen-and antigen-induced lymphocyte proliferation J Toxicol Environ Health A 68 457 484 15799246
Lin PP Hu SW Chang TH 2003 Correlation between gene expression of aryl hydrocarbon receptor (AhR), hydrocarbon receptor nuclear translocator (Arnt), cytochromes P4501A1 (CYP1A1) and 1B1 (CYP1B1), and inducibility of CYP1A1 and CYP1B1 in human lymphocytes Toxicol Sci 71 1 20 26 12520072
Luross JM Alaee M Sergeant DB Cannon CM Whittle DM Solomon SR 2002 Spatial distribution of polybrominated diphenyl ethers and polybrominated biphenyls in lake trout from the Laurentian Great Lakes Chemosphere 46 665 672 11999789
Lyon H Andersen AP Hasselager E Høyer PE Møller M Prentø 1991. Theory and Strategy in Histochemistry. Berlin:Springer Verlag.
MacLachlan NJ Cullen JM 1995. Liver, biliary system and exocrine pancreas. In: Thomsons Special Veterinary Pathology (Carlton WW, Donald McGavin M, eds). St. Louis, MO:Mosby Year Book, 81–115.
Messier F Taylor MK Ramsay MA 1992 Seasonal activity patterns of female polar bears (Ursus maritimus ) in the Canadian Arctic as revealed by satellite telemetry J Zool (Lond) 992 226 219 229
Oskam IC Ropstad E Dahl E Lie E Derocher AE Wiig Ø 2003 Organochlorines affect the major androgenic hormone, testosterone, in male polar bears (Ursus maritimus ) at Svalbard J Toxicol Environ Health A 66 22 2119 2139 14710596
Oskam IC Ropstad E Dahl E Lie E Derocher AE Wiig Ø 2004 Organochlorines affect the steroid hormone cortisol in polar bears (Ursus maritimus ) at Svalbard, Norway J Toxicol Environ Health A 67 959 977 15205038
Parkinson A 1996. Biotransformation of xenobiotics. In: Casarett and Doull’s Toxicology—the Basic Science of Poisons (Klaassen CD, ed). New York:McGraw-Hill, 113–186.
Prunescu CC Serban-Parau N Brock JH Vaughan DM Prunescu P 2003 Liver and kidney structure and iron content in romanian brown bears (Ursus arctos ) before and after hibernation Comp Biochem Phys A 134 21 26
Ramsay MA Stirling I 1988 Reproductive biology and ecology of female polar bears (Ursus maritimus ) J Zool (Lond) 214 601 634
Rawson AJ Patton GW Hofmann S Pietra GG Johns L 1993 Liver abnormalities associated with chronic mercury accumulation in stranded Atlantic bottlenose dolphins Ecotoxicol Environ Saf 25 41 47 7682917
Rosing-Asvid A Born EW Kingsley MCS 2002 Age at sexual maturity of males and timing of the mating season of polar bears (Ursus maritimus ) in Greenland Polar Biol 25 878 883
Sandala GM Sonne-Hansen C Dietz R Muir DCG Valters K Bennett ER 2004 Methyl sulfone and hydroxylated PCB metabolites in adipose and whole blood of polar bear (Ursus maritimus ) from Scoresby Sound, Greenland Sci Total Environ 331 125 141 15325145
Sato M Miura M Kojima N Higashi N Imai K Sato T 2001 Nuclear deviation in hepatic parenchymal cells on sinusoidal surfaces in arctic animals Cell Struct Funct 26 71 77 11482455
Schumacher U Zahler S Horny HP Heidemann G Skirnisson K Welsch U 1993 Histological investigations on the thyroid glands of marine mammals (Phoca vitulina , Phocoena phocoena ) and the possible implications of marine pollution J Wildlife Dis 29 1 103 108
Senoo H Imai K Higashi N Wake K Kojima N Miura M 2001. Transport and hepatic storage of vitamin A in arctic animals. In: Cells of the Hepatic Sinusoid, Vol 8 (Wiss E, Knook DL, de Zanger R, Fraaser R, eds). Leiden, the Netherlands:Kupffer Cell Foundation, 207–209.
Senoo H Imai K Wake K Wold H Moskaug J Kojima N 1999. Vitamin A-storing system in mammals and birds in an arctic area—a study in the Svalbard Archipelago. In: Cells of the Hepatic Sinusoid, Vol 7 (Wiss E, Knook DL, de Zanger R, Fraaser R, eds). Leiden, the Netherlands:Kupffer Cell Foundation, 34–35.
Skaare JU Bernhoft A Wiig Ø Norum KR Haug E Eide DM 2001 Relationship between plasma levels of organochlorines, retinol and thyroid hormones from polar bears (Ursus maritimus ) at Svalbard J Toxicol Environ Health A 62 227 241 11245393
Sonne C Dietz R Born EW Riget FF Kirkegaard M Hyldstrup L 2004 Is bone mineral composition disrupted by organochlorines in East Greenland polar bears (Ursus maritimus )? Environ Health Perspect 112 1711 1716 15579418
Sonne C Dietz R Leifsson PS Born EW Kirkegaard M Letcher RJ In press. Are organohalogen contaminants a cofactor in the development of renal lesions in East Greenland Polar Bears (Ursus maritimus)? Environ Toxicol Chem.
Thompson DR 1996. Mercury in birds and terrestrial mammals. In: Environmental Contaminants in Wildlife. Interpreting Tissue Concentrations (Beyer WM, Heinz GH, Redmon-Norwood AW, eds). New York:CRC Press, 341–356.
van Duursen MBM Sanderson JT van der Bruggen M van der Linden J van den Berg M 2003 Effects of several dioxin-like compounds on estrogen metabolism in the malignant MCF-7 and nontumorigenic MCF-10A human mammary epithelial cell lines Toxicol Appl Pharmacol 190 3 241 250 12902195
Wong S Fournier M Coderre D Banska W Krzystyniak K 1992. Environmental Immunotoxicology. In: Animal Biomarkers as Pollution Indicators (Peakall DB, eds). London:Chapman and Hall, 8–189.
Woshner VM O’Hara TM Eurell JA Wallig MA Bratton GR Suydam RS 2002 Distribution of inorganic mercury in liver and kidney of beluga and bowhead whales through autometallographic development of light microscopic tissue sections Toxicol Pathol 30 2 209 215 11950164
|
16263513
|
PMC1310920
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 5; 113(11):1569-1574
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8038
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8091ehp0113-00157516263514ResearchPM Source Apportionment for Short-Term Cardiac Function Changes in ApoE−/− Mice Lippmann Morton 1Hwang Jiang-Shiang 2Maciejczyk Polina 1Chen Lung-Chi 11 New York University School of Medicine, Tuxedo, New York, USA2 Institute of Statistical Science, Academia Sinica, Taipei, TaiwanAddress correspondence to M. Lippmann, New York University School of Medicine, 57 Old Forge Rd., Tuxedo, NY 10987 USA. Telephone: (845) 731-3558. Fax: (845) 351-5472. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 5 7 2005 113 11 1575 1579 28 2 2005 5 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Daily rates of cardiovascular mortality and morbidity are have been associated with daily variations in fine particulate matter (aerodynamic diameter ≤2.5 μm, PM2.5), but little is known about the influences of the individual source-related PM2.5 categories or the temporal lags for the effects. We investigated heart rate (HR) and HR variability (HRV) data collected during a 5-month study involving 6 hr/day, 5 day/week exposures of normal (C57) mice and a murine model for atherosclerotic disease (ApoE−/−) in Sterling Forest (Tuxedo, New York, USA). The mice were exposed to concentrated ambient particles (PM2.5 concentrated 10-fold, producing an average of 113 μg/m3). Daily 6-hr PM2.5 air samples were analyzed by X-ray fluorescence, permitting attribution to major PM source categories [secondary sulfate (SS), resuspended soil (RS), residual oil (RO) combustion, and other, largely due to motor vehicle traffic]. We examined associations between these PM2.5 components and both HR and HRV for three different daily time periods: during exposure, the afternoon after exposure, and late at night. For HR there were significant transient associations for RS during exposure, and for SS in the afternoon after exposure. For HRV, there were comparable associations with RO in the afternoon after exposure and for both SS and RS late at night. The biologic bases for these associations and their temporal lags are not known but may be related to the differential solubility of the biologically active PM components at the respiratory epithelia and their access to cells that release mediators that reach the cardiovascular system. Clearly, further research to elucidate the underlying processes is needed.
concentrated ambient particulate matterheart rateheart rate variabilitymotor vehicle pollutionPM2.5residual oilresuspended soilsecondary sulfatesource apportionment
==== Body
Many published studies have demonstrated statistically significant associations between ambient air fine particulate matter (aerodynamic diameter ≤2.5 μm, PM2.5) mass concentrations and short-term changes in heart rate (HR) and/or HR variability (HRV) in humans and laboratory animals (U.S. Environmental Protection Agency 2004). However, interpretation of these findings is complicated in that the effects may go in either direction and are observed on some days and not on others. These inconsistencies may be due to the limitations of PM2.5 mass as an index of exposure to the biologically active components of the ambient PM2.5. It is well known that the composition of ambient air PM2.5 has considerable temporal and spatial variability. Studies in human volunteers and laboratory animals have been limited in their power to identify the causal components because they require both the long-term collection of electrocardiographic (ECG) data and simultaneous availability of data on tracers or factors associated with PM2.5 composition.
In a recent study of the effects of sub-chronic (5–6 months of daily 6 hr) exposures of normal (C57) mice and a mouse model of atherosclerosis (apolipoprotein deficient, ApoE−/−, mice) to fine concentrated ambient particles (CAPs) in Tuxedo, New York, USA, at an average concentration of 113 μg/m3, we generated the kinds and amounts of data needed to address the issue raised in this article. The results of the overall subchronic study design and the results obtained for associations of PM2.5 mass concentration with progressive changes in HR and HRV and the changes in atherosclerotic plaque, gene expression, and brain cell distribution at the end of the study are described in a series of articles (Chen and Hwang 2005; Chen and Nadziejko 2005; Gunnison and Chen 2005; Hwang et al. 2005; Lippmann et al. 2005a
2005b; Maciejczyk et al. 2005; Veronesi et al. 2005). In another article (Maciejczyk and Chen 2005) describing the parallel study that went on simultaneously with the inhalation study, we exposed BEAS-2B cells (an airway epithelial cell line) in vitro to CAPs and reported that nuclear factor-kappa B (NFκB) expression was most closely associated with the residual oil (RO) component, which was, on average, 1.4% of the PM2.5 mass.
For this article, we used the 5 months of daily 6-hr source apportionments described in Maciejczyk and Chen (2005), the continuous HR data for exposure (weekday) days provided in Hwang et al. (2005), and the corresponding HRV data given in Chen and Hwang (2005) to determine the source-related PM2.5 components’ short-term associations with HR and HRV.
Materials and Methods
The methods used to generate the factors associated with specific major PM source categories were described by Maciejczyk and Chen (2005). Briefly, fine CAPs were collected from a rural area upwind of New York City for the 0900- to 1500-hr period on weekdays only, March through September 2003. Chemical composition data for CAPs were modeled using factor analysis with varimax orthogonal rotation to determine four particle source categories contributing significant amount of mass to CAPs at Sterling Forest (Tuxedo, New York). These source categories are regional secondary sulfate (SS) characterized by high sulfur, silicon, and organic carbon; resuspended soil (RS) characterized by high concentrations of calcium, iron, aluminum, and silicon; RO-fired power plants emissions of the Eastern United States identified by presence of vanadium, nickel, and selenium; and motor vehicle (MV) traffic and unknown other sources. To estimate the mass contributions of each individual source category, the CAP mass concentration was regressed against the factor scores. Using the method developed by Thurston and Spengler (1985), we determined that regional sulfate was the largest contributor to average mass (56.1%), followed by soil (11.7%). The RO combustion accounted for 1.4%, and the MV traffic and other sources category contributed 30.9%.
The methods used to process the voluminous HR and HRV data for the same period were described by Hwang et al. (2005) and Chen and Hwang (2005). Briefly, they used their recently developed nonparametric method (Nadziejko et al. 2004) to estimate the daily time periods that mean HR differed significantly between the CAPs and the air sham-exposed groups. CAP exposure most affected HR between 0130 and 0430 hr. With the response variables being the average HR, they adopted a two-stage modeling approach to obtain the estimates of chronic and acute effects on the changes of this variable. In the first stage, a time-varying model estimated daily crude effects. In the second stage the true mean of the estimated crude effects was modeled with a polynomial function of time for chronic effects, a linear term of daily CAP exposure concentrations for acute effects, and a random component for unknown noise. A Bayesian framework combined these two stages.
For the analyses of HRV, the times in milliseconds of occurrence of two consecutive R waves in the ECG waveform (RR) were calculated on a beat-to-beat basis. Because of limitation in data storage capacity, the RR intervals were recorded consecutively for 5 sec in every 15-min interval for all mice during 10–27 April 2003, and for ApoE−/− mice in the control group during 22 April through 20 July 2003. The rest of recordings were taken consecutively about 10 sec in every 5-min interval for the mice. There are about 34–64 and 100 RR intervals recorded in 15-and 5-min intervals, respectively. For the analysis, we decided to work on fluctuations of RR intervals on an every 15-min basis. To match the data in the 15-min recordings, we used only the first 60 RR intervals in the last of 3 consecutive 5-min intervals. The two HRV indices that we used were the standard deviations of the RR intervals (SDNNs) and the square root of the mean squared differences (RMSSD) of successive RR intervals in 5 sec. The nonparametric method identified the 0000- to 0500-hr period during which the two groups had the largest HRV differences within each day. To match the HR analyses of effects with the HRV changes, we used the same period (0130–0430 hr) for calculating mean log SDNN and log RMSSD to represent daily HRV responses for this period for each mouse. In the analysis of effects on HR, we also calculated daily responses for the 1100- to 1300-hr period during exposure for examining acute effects. However, because the number of normal RR intervals recorded during the exposure period was small because of interference from the perforated metal chamber, we instead used the 1600- to 1800-hr interval after exposure as an alternate for calculating daily HRV response. Daily changes in HR during this period, which were not reported in the previous study, were also calculated for this analysis.
To examine whether variations of concentrations in major sources are correlated with short-term changes of cardiac functions in exposed mice, we adopted the following approach:
Let Xijkd be the average cardiac function measurement for mouse j in the ith group at a given period on the dth day of the kth week, where
We have seen that daily cardiac function measurements changed over the 5 months. Such changes may be caused by the cumulative effects of aging, exposure, and other unknown environmental factors. To examine the association between exposed level and acute cardiac function change on exposure days, we generated baseline adjusted measurements for each mouse on the exposure days by subtracting averaged measurement on the previous weekend from each measurement on weekdays. Presumably, the daily series of these baseline-adjusted measurements Yijkd = Xijkd – (Xijk1 + Xijk2)/2 will have little cumulative effect. To see whether the idea worked or not, we explored the data. Figure 1 shows two series of daily averaged baseline adjusted measurements of HR at the 1100- to 1300-hr period for mice in the control and exposure groups. The exposure chamber effects reduced HR in both groups, which also corresponded to the quiescent period of mouse circadian rhythm during the daytime. The two series also share the same quadratic shape. Although it is not clear why this has happened, some common factors have strong effects on measurements of mice in both control and exposure groups. Instead of searching for a smooth curve for modeling the pattern caused by common factors, we can simply use the baseline-adjusted measurements of the nine mice in the control group to calculate an average for each exposure day. That is the darker curve plotted in Figure 1. If there is no exposure effect, the darker curve and lighter curve of averaged measurements for mice in the exposure group will not differ. In fact, the difference between two curves shown in Figure 2 indicates that CAP exposure had the effect of reducing HR. The difference series in the plot also show no trend over the 5 months, indicating that cumulative effects have been removed. Hence, we may construct a model to fit the baseline-adjusted measurements for examining whether the short-term cardiac function changes are related to exposure levels of the identified source factors F1, F2, F3, and F4. A linear model is given by
where ɛ ijkd is an autoregressive process of order one. If the estimate of β h differs significantly from zero, we may claim that the hth source factor is associated with the acute changes of HR and HRV.
Results
Associations between sources and short-term HR changes.
Using the source apportionment factors from Maciejczyk and Chen (2005), we have the following four source classes: SS, RS, RO, and MV. There were no significant associations between these four source categories and HR in the C57 normal mice at any of the three intervals. However, as shown in Table 1, there were highly significant associations between PM2.5 and the RS source factor and decreases in HR for the ApoE−/− mice during the daily CAP exposures but no associations with the other source factors. By contrast, Tables 2 and 3 indicate that there was no residual association of HR with PM2.5 or the RS factor later in the afternoon or late that night.
In the afternoon, there was a significant association between decreases in HR and the SS factor for the ApoE−/− mice that had not been present during exposure and did not persist into the nighttime period. It is also of some interest that the MV traffic and other source category was not significantly associated with HR during any of the three time periods.
For the C57 mice, there were no significant associations of HR with PM2.5 or any of its component source classes during any of the three daily time periods.
Associations between sources and short-term HRV changes.
It is unfortunate that there was too much signal noise during the exposures to permit reliable analyses of HRV changes during the hours of CAP exposure. We therefore cannot tell whether the transient effect of PM2.5 or its RS source component on HR was also present for HRV. For C57 mice, the only significant association was between the MV and other source factor and a decline in RMSSD during the afternoons after the exposures (p = 0.00; data not shown). For the ApoE−/− mice (Table 4), there were very strong associations of HRV with the RO source factor in the afternoon. These decreases in HRV did not persist at night (Table 5) and had not been seen for HR at any time period. Finally, there were strong associations between HRV during the nighttime hours and both the SS source category and the RS source category that were not seen for HR at the other intervals, or for HRV at the other time periods. However, it must be noted that although the SS source factor was associated with decreased HRV, the RS source category was associated with an increase in HRV. For PM2.5, there was a significant (p = 0.03) decrease in RMSSD and a nearly significant (p = 0.07) decrease in SDNN for the 0130- to 0430-hr interval but no such an association during the 1600- to 1800-hr period.
Discussion
Interpretation of the various significant (p < 0.05) associations between source factors and the HR and HRV variables in CAP-exposed mice at this time would be speculative at best, especially because three of the source factors showed some association at one interval or another, and the fourth (MV traffic and other category) showed a strong association (p = 0.00) with RMSSD in the afternoon after exposure in the C57 mice. The strongest associations for the ApoE−/− mice are summarized in Table 6.
For the evaluation of the changes on HR and HRV in the last column of Table 6, we have calculated the changes in the measured parameters over the interquartile range of concentrations as is commonly done in epidemiology. For HR, the changes are for exposures at the third quartile to the first quartile of the measured concentrations. The results show about 3–4 beats/min (bpm) changes. For HRV, the interquartile change is the ratio of RMSSDs between the third quartile and first quartile of the concentrations. The results show about 2–6% changes. These are relatively small changes, but they may have played some role in the progressive changes in HR that we observed during the course of the 5 months of exposure that were described by Hwang et al. (2005), and the changes in HRV that were reported by Chen and Hwang (2005).
It is also interesting that the reduction in HR during the daily exposures associated with PM2.5 (−4.1 bpm) may have been due entirely to the influence of the RS factor (−4.5 bpm) and that there was an increase in HR (+2.6 bpm) in the afternoons after the exposures in the same source factor. This appears to have been compensated by the decrease in HR in the afternoon after the exposures (−2.5 bpm) associated with the SS factor. Such a compensation would be consistent with the lack of any association of HR with PM2.5 in this interval.
The RO combustion factor, which did not have any significant association with HR, appears to have had the effect of increasing RMSSD by 6.2% during the afternoons after the exposure but not at the other intervals. The other observed statistically significant changes in RMSSD were associated with opposite effects during the late night period by the RS and SS source components, with the SS factor perhaps accounting for the significant association in the same direction for the association of RMSSD with PM2.5 during the same period.
It is also of interest that the effects reported here for HR and HRV were occurring at relatively low concentrations of outdoor PM2.5 and its component source-related factors. The average PM2.5 CAPs during the 6-hr exposures was only 113 μg/m3. Thus, the 24-hr average exposures were only 28.3 μg/m3 because the mice were breathing air that was filtered of the outdoor air components during the balance of the day. Outdoor PM2.5 does not have much diurnal variation, and it infiltrates indoors with a high degree of penetration. People are therefore exposed to concentrations of PM2.5 of ambient origin at near ambient concentrations for 24-hr each day. If indeed the ApoE−/−mouse is a good model for people with atherosclerosis, and if the HR and HRV responses to CAPs in these mice seen in this study are relevant to them, then such responses may be occurring in this human subpopulation at current ambient levels on many days each year.
There have been no previous reports that examined such responses at various periods during and after daily exposures. The only report of different ambient air PM source categories having different lagged effects was that of the PM Source Apportionment Workshop, in which human mortality effects were associated with different days of lag (Thurston et al. 2005).
Although there have been no previous reports of cardiac function effects that go in opposite directions after low-level environmental exposures, there have been such examples for other physiologic responses. In previous work in this laboratory, we reported in both humans (Leikauf et al. 1981) and rabbits (Schlesinger 1985) that short-term inhalation of a low concentration of submicrometer sulfuric acid aerosol increased the rate of mucociliary particle clearance from tracheobronchial airways, whereas a higher concentration (1 mg/m3) retarded such clearance. Similarly, the inhalation of the fresh smoke from two cigarettes accelerated tracheobronchial particle clearance in both humans and donkeys, whereas the smoke from 10 or more cigarettes slowed the particle clearance in donkeys (Lippmann et al. 1982). In another study in this laboratory, Schlesinger (1989) examined the effect of 14 days of sulfuric acid inhalation on particle clearance from the pulmonary region of rabbit lungs and found that low-level exposures accelerated such macrophage-mediated clearance, whereas higher levels of exposure retarded the clearance. In addition, subchronic inhalation exposures to both cigarette smoke and sulfuric acid produced persistent changes in particle clearance (Lippmann et al. 1987).
The fact that three different source factors showed some indication of a strong association with either HR or HRV in ApoE−/− mice in this study, with the SS source component having an effect in the opposite direction to that of the RS source component, illustrates the complexity facing researchers when designing studies to identify the causal factors for the PM-associated adverse health effects reported in the epidemiologic literature. It may well be that most, if not all, PM source categories have some, if various, effects on cardiac physiology, with various lag structures, and that some components mitigate the effects produced by other components. Also, we do not know at this time about the short-term effects, and their temporality, of inhaled PM2.5 on other organ systems. However, if most of the major components of PM2.5 produce some short-term biologic responses, then the commonly used integral measure of PM2.5 mass concentration, that is, 24-hr average PM2.5, may be serving as a reasonable integrating index for at least some of the short-term health risks. In any case, the results reported in this article provide us and others with additional factors to consider in the planning of our future laboratory and field studies of PM health effects.
As noted above, Maciejczyk and Chen (2005) reported that in vitro NFκB expression of BEAS-2B cells exposed to CAPs collected during the daily 6 hr in vivo exposures was significantly increased in association with the RO, but not with the other source categories of the CAPs. The NFκB expression is an index of cellular oxidative stress and the release from the cells of mediators affecting systemic inflammation. This mechanism for biologic response is consistent with short lag times between respiratory tract particle deposition and cardiac function changes. However, because the NFκB index of biologic response to CAP exposure provides no within-day temporality, it is not possible to make a direct comparison with the lagged HR and HRV responses reported in this article. The different lag structures of the responses reported in this article may be related to the solubility of the biologically active components in each source category.
We plan to pursue the issues raised by the results reported here in our future subchronic exposure studies in mice. In terms of comparable investigations in humans, a study would need access to a population that is being continuously monitored for cardiac function as well as time-resolved PM2.5 compositional data. The only study we are aware of to date looking for cardiac responses to ambient air PM was by Sullivan et al. (2005), in which they examined the relation between PM2.5 exposure (measured by nephthelometry) and the number of hours preceding the onset of myocardial infarction (MI). They found no significant associations between MI and the nephthelometry data. It is possible that nephthelometry measurements may not be representative of the active components of ambient PM mix or that the nephthelometry measurements correlate with outcomes other than MI.
Conclusions
The availability of data on HR and HRV over a 5-month period during subchronic exposures of mice to the regional anthropogenic CAPs at New York University’s Sterling Forest laboratory in Tuxedo, New York, and during the afternoon and nighttime periods after the daily exposures, as well as elemental composition data for each day’s exposure, enabled us to examine daily source apportionments of the major source categories during the exposures and their association with HR and HRV during each of the three time periods. The RS component was strongly associated with a transient decrease in HR during exposure, comparable with that of the whole PM2.5. The SS component was strongly associated with a transient HR decrease in the afternoon after the day’s exposure. The RO component was strongly associated with increases in HRV in the afternoon after the day’s exposure. The SS and RS components were strongly associated with HRV in the nighttime period, with decreased HRV for the SS component and increased HRV for the RS component. These effects were occurring after exposures at daily average PM2.5 concentrations occurring frequently in the United States and may be relevant to the subpopulation with atherosclerotic disease.
The biologic bases for these various associations and their temporal lags are not known at this time but may relate to the differential solubilities of the PM components at the respiratory epithelia and their access to cells that release mediators that reach the cardiovascular system. Further research that can elucidate the underlying processes is clearly needed.
This research was conducted as part of the New York University Center for Particulate Matter Health Effects Research supported by grant R827351 from the U.S. Environmental Protection Agency and is part of a basic environmental health science center program supported by the National Institute of Environmental Health Sciences (grant ES 00260).
Figure 1 Daily average measurements of HR (bpm) for CAP-exposed and air sham–exposed (control) ApoE−/− mice during the daily exposures (1100–1300 hr).
Figure 2 The difference (D) in HR (bpm) between CAP-exposed and air sham-exposed (control) ApoE−/− mice during the daily exposures (1100–1300 hr).
Table 1 HR parameter estimates for the 1100- to 1300-hr period for ApoE−/− mice.
PM source component Value SE t-Value Pr(> |t|)
SS −3.78E–02 2.41E–02 −1.55 0.12
RS −3.61E–01 1.40E–01 −2.57 0.01
RO −6.61E–01 7.23E–01 −0.91 0.36
MV 7.91E–02 2.44E–01 −0.32 0.75
PM2.5 −4.77E–02 1.33E–02 −3.59 0.00
Table 2 HR parameter estimates for the 1600- to 1800-hr period for ApoE−/− mice.
PM source component Value SE t-Value Pr(> |t|)
SS −3.63E–02 1.82E–02 −2.00 0.05
RS 2.09E–01 1.07E–01 1.96 0.05
RO 5.92E–01 5.79E–01 1.02 0.31
MV 2.36E–01 1.94E–01 1.22 0.22
PM2.5 6.55E–03 9.76E–03 0.67 0.50
Table 3 HR parameter estimates for the 0130- to 0430-hr period for ApoE−/− mice.
PM source component Value SE t-Value Pr(> |t|)
SS 3.83E–02 2.02E–02 1.89 0.06
RS −8.79E–02 1.18E–01 −0.74 0.46
RO −4.12E–01 6.23E–01 −0.66 0.51
MV −1.62E–01 2.08E–01 −0.78 0.44
PM2.5 −7.46E–03 1.10E–02 −0.68 0.50
Table 4 HRV parameter estimates for the 1600- to 1800-hr period for ApoE−/− mice.
Ln RMSSD (sec)
Ln SDNN (sec)
PM source component Value SE t-Value Pr(> |t|) Value SE t-Value Pr(> |t|)
SS 3.74E–04 2.81E–04 1.33 0.18 −4.40E–06 2.77E–04 −0.02 0.99
RS −2.20E–03 1.67E–03 −1.32 0.19 −1.90E–03 1.65E–03 −1.15 0.25
RO 2.64E–02 9.11E–03 2.89 0.00 2.62E–02 9.13E–03 2.87 0.00
MV −3.57E–03 2.98E–03 −1.20 0.23 −4.41E–03 2.95E–03 −1.49 0.14
PM2.5 1.42E–04 1.51E–04 0.94 0.35 −1.18E–04 1.48E–04 −0.80 0.42
Table 5 HRV parameter estimates for the 0130- to 0430-hr period for ApoE−/− mice.
Ln RMSSD (sec)
Ln SDNN (sec)
PM source component Value SE t-Value Pr(> |t|) Value SE t-Value Pr(> |t|)
SS −1.07E–03 2.44E–04 −4.38 0.00 −9.28E–04 2.36E–04 −3.94 0.00
RS 3.40E–03 1.43E–03 2.38 0.02 2.43E–03 1.38E–03 1.76 0.08
RO −3.92E–03 7.59E–03 −0.52 0.61 −1.44E–03 7.32E–03 −0.20 0.84
MV 3.29E–03 2.51E–03 1.31 0.19 3.22E–03 2.42E–03 1.33 0.18
PM2.5 −2.86E–04 1.32E–04 −2.16 0.03 −2.08E–04 1.28E–04 −1.63 0.10
Table 6 Short-term cardiac function changes associated with PM components with some significant p-values.
Concentration (μg/m3)
PM source component Time of day (hr) Affected variable Effect coefficient (× 10−3) p-Value Mean First quartile Third quartile Interquartile change
PM2.5 1100–1300 HR −47.67 0.00 113.0 55.21 141.48 −4.1 bpm
RS 1100–1300 HR 361.23 0.01 13.18 5.88 18.36 −4.5 bpm
RS 1600–1800 HR 209.46 0.05 13.18 5.88 18.36 2.6 bpm
SS 1600–1800 HR −36.30 0.05 63.41 25.08 79.20 −2.5 bpm
RO 1600–1800 RMSSD 26.37 0.00 1.53 0.01 2.30 6.2%
SS 130–430 RMSSD −1.07 0.00 63.41 25.08 79.20 −5.6%
RS 130–430 RMSSD 3.40 0.02 13.18 5.88 18.36 4.3%
PM2.5 130–430 RMSSD −0.29 0.03 113.0 55.21 141.48 −2.4%
==== Refs
References
Chen LC Hwang JS 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. IV: Characterization of acute and chronic effects of ambient air fine particulate matter exposures on heart rate variability Inhal Toxicol 17 209 216 15804938
Chen LC Nadziejko C 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. V: CAPs exacerbate aortic plaque development in hyperlipidemic mice Inhal Toxicol 17 217 224 15804939
Gunnison A Chen LC 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. VI: Measurement of gene expression in heart and lung tissue following exposure Inhal Toxicol 17 225 233 15804940
Hwang JS Nadziejko C Chen LC 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. III: Acute and chronic effects of CAPs on heart rate, heart rate variance, and body temperature Inhal Toxicol 17 199 207 15804937
Leikauf G Yeates DB Wales KA Spektor D Albert RE Lippmann M 1981 Effects of sulfuric acid aerosol on respiratory mechanics and mucociliary particle clearance in healthy nonsmoking adults Am Ind Hyg Assoc J 42 273 282 7234686
Lippmann M Gearhart JM Schlesinger RB 1987 Basis for a particle size-selective TLV for sulfuric acid aerosols Appl Ind Hyg 2 188 199
Lippmann M Gordon T Chen LC 2005a Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. I: Introduction, objectives and experimental plan Inhal Toxicol 17 177 187 15804935
Lippmann M Gordon T Chen LC 2005b Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice: IX. Integral assessment and human health implications of subchronic exposures of mice to CAPs Inhal Toxicol 17 255 261 15804943
Lippmann M Schlesinger RB Leikauf G Spektor D Albert RE 1982 Effects of sulphuric acid aerosols on respiratory tract airways Ann Occup Hyg 26 677 690 7181298
Maciejczyk P Chen LC 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. VIII: Source-related daily variations in in vitro responses to CAPs Inhal Toxicol 17 243 253 15804942
Maciejczyk P Zhong M Li Q Xiong J Nadziejko C Chen LC 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice: II. The design of a concentrated ambient particulate matter exposure system for biometric telemetry monitoring Inhal Toxicol 17 189 197 15804936
Nadziejko C Chen LC Nádas A Hwang JS 2004 The “fishing license” method for analyzing the time course of effects in repeated measurements Stat Med 23 1399 1411 15116349
Schlesinger RB 1985 Effects of inhaled acids on respiratory tract defense mechanisms Environ Health Perspect 63 25 38 3908089
Schlesinger RB 1989 Factors affecting the response of lung clearance systems to acid aerosols: role of exposure concentration, exposure time, and relative acidity Environ Health Perspect 79 121 126 2707191
Sullivan J Sheppard L Schreuder A Ishikawa N Siscovick D Kaufman J 2005 Relation between short-term fine-particulate matter exposure and onset of myocardial infarction Epidemiology 16 41 48 15613944
Thurston GD Ito K Mar T Christensen WF Eatough DJ Henry RC 2005. The Workshop on the Source Apportionment of PM Health Effects Inter-Comparison of Results and Implications. Environ Health Perspect 10.1289/ehp.7989.
Thurston GD Spengler JD 1985 A quantitative assessment of source contributions to inhalable particulate matter in metropolitan Boston, MA Atmos Environ 19 9 25
U.S. Environmental Protection Agency 2004. Air Quality Criteria for Particulate Matter. EPA/600/P-99/002aF and EPA/600/P-99/002bF. Research Triangle Park, NC:U.S. Environmental Protection Agency.
Veronesi B Makwana O Pooler M Weir L Chen LC 2005 Effects of subchronic exposures to concentrated ambient particles (CAPs) in mice. VII: Degeneration of dopaminergic neurons Inhal Toxicol 17 235 241 15804941
|
16263514
|
PMC1310921
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 5; 113(11):1575-1579
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8091
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8028ehp0113-00158016263515ResearchCellular and Hormonal Disruption of Fetal Testis Development in Sheep Reared on Pasture Treated with Sewage Sludge Paul Catriona 1Rhind Stewart M. 2Kyle Carol E. 2Scott Hayley 1McKinnell Chris 1Sharpe Richard M. 11 MRC Human Reproductive Sciences Unit, Centre for Reproductive Biology, Queen’s Medical Research Institute, University of Edinburgh, Edinburgh, United Kingdom2 Macaulay Institute, Craigiebuckler, Aberdeen, United KingdomAddress correspondence to R.M. Sharpe, MRC Human Reproductive Sciences Unit, Centre for Reproductive Biology, Queen’s Medical Research Institute, 49 Little France Crescent, Edinburgh EH16 4TJ, UK. Telephone: 44-131-242-6387. Fax: 44-131-242-6231. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 11 7 2005 113 11 1580 1587 17 2 2005 11 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The purpose of this study was to evaluate whether experimental exposure of pregnant sheep to a mixture of environmental chemicals added to pasture as sewage sludge (n = 9 treated animals) exerted effects on fetal testis development or function; application of sewage sludge was undertaken so as to maximize exposure of the ewes to its contents. Control ewes (n = 9) were reared on pasture treated with an equivalent amount of inorganic nitrogenous fertilizer. Treatment had no effect on body weight of ewes, but it reduced body weight by 12–15% in male (n = 12) and female (n = 8) fetuses on gestation day 110. In treated male fetuses (n = 11), testis weight was significantly reduced (32%), as were the numbers of Sertoli cells (34% reduction), Leydig cells (37% reduction), and gonocytes (44% reduction), compared with control fetuses (n = 8). Fetal blood levels of testosterone and inhibin A were also reduced (36% and 38%, respectively) in treated compared with control fetuses, whereas blood levels of luteinizing hormone and follicle-stimulating hormone were unchanged. Based on immunoexpression of anti-Müllerian hormone, cytochrome P450 side chain cleavage enzyme, and Leydig cell cytoplasmic volume, we conclude that the hormone changes in treated male fetuses probably result from the reduction in somatic cell numbers. This reduction could result from fetal growth restriction in male fetuses and/or from the lowered testosterone action; reduced immunoexpression of α-smooth muscle actin in peritubular cells and of androgen receptor in testes of treated animals supports the latter possibility. These findings indicate that exposure of the developing male sheep fetus to real-world mixtures of environmental chemicals can result in major attenuation of testicular development and hormonal function, which may have consequences in adulthood.
anti-Müllerian hormoneenvironmental chemicalsfollicle-stimulating hormoneFSHgonocyteinhibin-ALeydig cellLHperitubular myoid cellSertoli cellsewage sludgetestosterone
==== Body
There is growing concern that exposure of adults and, especially, of the developing fetus to environmental chemicals could have potentially detrimental effects on various aspects of health, particularly reproductive development. This concern stems from four areas of information. First, evidence for a range of reproductive disorders in various wildlife species, ranging from aquatic snails and fish to mammals, has established close associations with environmental chemical exposure [International Programme on Chemical Safety (IPCS) 2002; Toppari et al. 1996]. Second, the occurrence and possibly increasing prevalence of reproductive disorders in human males are thought to originate in fetal life (Sharpe and Skakkebaek 2003; Skakkebaek et al. 2001) and can be induced in animal models by fetal exposure to environmental chemicals (Fisher et al. 2003; Mylchreest et al. 1999; Parks et al. 2000). Third, our understanding is growing regarding the many pathways by which particular environmental chemicals—so-called endocrine disruptors (EDs)—can alter hormone production, metabolism, and/or action in animals or in isolated cells (IPCS 2002). Fourth, evidence shows that humans are exposed to a broad range of environmental chemicals [Centers for Disease Control and Prevention (CDC) 2003].
In contrast to some of the reproductive disorders in wildlife, in which an environmental chemical cause is clearly established (IPCS 2002), concerns for human reproductive health are based largely on extrapolation from laboratory animal studies in which exposure to supraenvironmental levels of test chemicals have been used (IPCS 2002; Sharpe and Irvine 2004). It can be argued that, because induction of adverse effects in such models only occurs at levels of exposure considerably greater than those experienced by humans, effects at real-world concentrations of the chemical in question are unlikely to occur. A counterargument is that, in normal life, humans (and wildlife) are exposed to a cocktail of environmental chemicals (CDC 2003) that may have additive or interactive hormonal effects (Rajapakse et al. 2002). In some areas of toxicology, understanding of the toxicologic effects of mixtures of environmental chemicals is relatively well developed and can be modeled (see Liao et al. 2002). In contrast, for EDs or other environmental chemicals with potential effects on the developing reproductive system, only a small number of studies in laboratory settings have been undertaken (Rajapakse et al. 2002; Tindall and Ashby 2004). To date, such studies have established only proof of principle (Rajapakse et al. 2002) and studies in vivo in laboratory animals are at a similarly early stage of development (Tindall and Ashby 2004). Moreover, such studies are confined to combinations of a small number of selected test chemicals rather than reflecting the multitude of environmental chemical exposures to which humans are subjected in the course of a normal day. Although there are major difficulties in establishing whether such exposures might contribute to human disorders, three developments have given them a new urgency: a) the evidence for increasing prevalence of human male reproductive disorders as a consequence of testicular maldevelopment in fetal life (Sharpe and Skakkebaek 2003; Skakkebaek et al. 2001); b) the induction of comparable effects in laboratory animals as a result of fetal exposure to certain phthalate esters (Fisher et al. 2003; Mylchreest et al. 1999; Parks et al. 2000); and c) recent evidence that associates human fetal exposure to phthalates with reduced production/action of androgens in the male fetus (Swan et al. 2005). Collectively, these studies highlight the potential vulnerability of the male fetus to endocrine disruption as a result of environmental chemical exposures.
One situation with potential fetal exposure to a naturally occurring mixture of environmental chemicals is in farm animals that are reared on pasture that has been fertilized with sewage sludge. The latter is broadly reflective of most human chemical exposures in that it contains outputs from domestic, agricultural, and industrial sources and contains a cocktail of man-made chemicals (phthalates, alklyphenolics, bisphenolics, polyaromatic hydrocarbons, organochlorine pesticides, etc.) as well as inorganic compounds and heavy metals such as lead and mercury (Smith 1995; Webber and Lesage 1989). Land application of sewage sludge is a technique widely used around the world and has grown in popularity due to tightening of regulations for disposal of such waste at sea and the pressures to recycle waste (Commission of the European Communities 1994; Swanson et al. 2004). In the present studies, we have investigated whether long-term exposure of ewes to the mix of chemicals present in sewage sludge affects development of the fetal testis when these animals become pregnant. Using a protocol that was designed to maximize potential exposure to the chemicals present in sewage sludge, we have evaluated the development of Sertoli and Leydig cells in terms of their numbers and hormonal function as well as germ cell numbers.
Materials and Methods
Animals, blood, and tissue collection.
Animals were maintained on pasture at the Macaulay Institute research station at Hartwood, Lanarkshire, Scotland. Animals were maintained at conventional stocking rates, according to the pasture height. They were inspected by a qualified shepherd on a daily basis, and routine animal care and vaccination procedures were conducted, as prescribed by best practice protocols.
From July 1997 until the end of July 1999, liquid, digested sewage sludge was applied twice annually to three 9-ha plots until five separate applications had been made. Thereafter, thermally dried sludge pellets were applied twice annually at similar rates because of changes in sludge production and spreading practices by the U.K. water authorities at the time. The composition of the sludge on a dry-matter basis was not altered. On each occasion, sludge was applied at a rate of 2.25 metric tons of dry matter per hectare to the whole of each plot, using either a pivot irrigation system (liquid sludge; estimated to cover > 95% of the surface) or a lime spreader (pelleted dried sludge), respectively. This rate of sludge application, which resulted in the application of about 225 kg nitrogen/ha/year, was consistent with normal management practice at the time, although recommendations for good practice in the United Kingdom have since been revised (SEDE and Arthur Andersen 2001). Our studies were designed to result in the maximum rate of contamination of the herbage and topsoil, and thus the maximum likely risk of exposure of grazing animals to the chemical constituents of sewage sludge through their food. Animals were not allowed to graze the pasture for a minimum of 3 weeks after sludge application, as prescribed by legislation (Great Britain Parliament 1989). Control ewes were maintained on similar pasture to which 225 kg of nitrogen/ha/year was applied using conventional, inorganic fertilizers. In the study flocks, treated and control groups consisted of three replicate groups of five breeding ewes in each of four age categories, so that at any one time there were 120 ewes in the study. All ewes were from similar genetic stock. Each year, ewes that were 6 years of age were slaughtered during pregnancy, and replacement animals were brought into the flocks to be bred for the first time. Stocking rates for ewes on control and sewage sludge-fertilized pastures were comparable. There was no supplementary feeding for either group.
Nine ewes maintained for the previous 5 years on conventionally fertilized pasture (controls) and nine ewes reared on sewage sludge-treated pasture were used for the present studies. All ewes were synchronized in estrus, using progestagen sponges, before mating to rams from the same genotype and source. Because estrus was synchronized, conception can be predicted to have occurred within a 48-hr window at the end of the first cycle after sponge removal. This was used as the basis for determining gestational age of approximately 110 days (GD110), when all animals were euthanized according to Schedule 1 protocols as defined by the U.K. Animals (Scientific Procedures) Act, 1986. Fetuses were collected from both sets of ewes, and all of the resulting males (n = 12 in both control and treated groups) were used for the studies detailed below. Maternal and fetal body weight and fetal testis weight were recorded at slaughter; fetal blood samples were also collected and serum isolated after centrifugation and stored at −20°C. Testes were fixed for 6 hr in Bouin’s fixative and then transferred to 70% ethanol until analysis. The testes were subsequently cut in half sagittally using a razor blade and processed in an automated Leica processor (Leica, Nussloch, Germany) before embedding in paraffin wax. Fetal body weight for female fetuses was also recorded.
Immunohistochemistry.
Sagittal 5 μm sections of each fetal testis were cut, floated onto slides, and dried at 50°C overnight. Slides were dewaxed in xylene, hydrated gradually through graded alcohols, and washed in water. Some antibodies [androgen receptor (AR) and Ki-67] required antigen retrieval by pressure cooking in 0.01 M citrate buffer (pH 6.0), after which sections were washed in Tris-buffered saline (TBS; 0.05 M Tris-HCl, pH 7.4, 0.85% NaCl) twice for 5 min. Endogenous peroxidase activity was blocked by immersion in 3% (vol/vol) hydrogen peroxide in methanol (BDH Laboratory Supplies, Poole, Dorset, UK) for 30 min, followed by rinsing in tap water and washing in TBS for 5 min. Nonspecific binding sites were blocked with the appropriate normal serum diluted 1:5 in TBS containing 5% bovine serum albumin (BSA; Sigma Chemical Co., Poole, Dorset, UK) applied for 30 min at room temperature. For anti-Müllerian hormone (AMH) and α-smooth muscle actin (SMA) immunostaining, normal rabbit serum was used; for all other antibodies, normal swine serum was used (Scottish Antibody Production Unit, Carluke, Scotland, UK). Tissues were then incubated with the primary antibody (Table 1) diluted in the appropriate blocking serum, and incubated overnight in a humidified chamber at 4°C.
After washing twice in TBS for 5 min, secondary antibodies were added to sections and incubated for 30 min at room temperature. A 1:500 dilution in the appropriate blocking serum of biotinylated rabbit anti-mouse IgG (DAKO, High Wycombe, UK) for SMA and Ki-67, biotinylated rabbit anti-goat IgG (DAKO) for AMH, and biotinylated swine anti-rabbit IgG (DAKO) for AR and P450 side chain cleavage enzyme (P450scc) was used. After two 5-min washes in TBS, the biotinylated antibody was linked to horseradish peroxidase (HRP) by 30 min incubation with avidin-biotin-HRP complex (ABC-HRP; DAKO) diluted in Tris-HCl (pH 7.4). This was followed by two 5-min washes in TBS. Antibody localization was determined by application of liquid diaminobenzidine substrate chromogen system (DAKO) for 1–5 min until the brown positive staining in control sections was optimal as determined by repeated microscopic examination. The color reaction was stopped by immersion in water. Sections were subsequently counterstained with hematoxylin, dehydrated in graded ethanols, cleared in Histoclear (VWR, Lutterworth, UK) and then xylene, and mounted using Pertex mounting medium (CellPath plc, Hemel Hempstead, UK). Testis sections from all control and treated animals were run for each antibody at the same time and under the same conditions to ensure comparability. Slides were then evaluated subjectively but systematically so that the intensity of immunostaining was scored for each animal on a scale from negative (−) through weakly (+) to strongly (++++) positive. Median values for immunostaining score were then determined for control (C) and treated (T) animals and compared. Immunostained sections were photographed using a Provis microscope (Olympus Optical, London, UK) fitted with a digital camera (Canon EOS 10D, Surrey, UK). Captured images were then transferred to a personal computer and compiled using Photoshop 7.0 (Adobe Systems Inc., San Jose, CA, USA).
Determination of gonocyte, Sertoli, and Leydig cell numbers per testis.
Stereologic analyses to determine cell numbers used AMH-immunostained slides for Sertoli cell and gonocyte counts and P450scc-immunostained slides for Leydig cell counts. For measurements, images were captured from an Olympus BH2 microscope using a video camera (Hitachi HV-C20, Tokyo, Japan) and were analyzed with Image Pro Plus software with a Stereology 5.0 plug-in (Media Cybernetics, Berkshire, UK). Slides were viewed using a 40× objective for counting and a 100× oil-immersion objective for nuclear volume measurement. The software was used to trace around each section, creating an area of interest (AOI). Forty fields, randomly selected by the program, within the AOI were then examined on one sagittal cross-section of the testis from each animal. A grid consisting of 432 evenly distributed points was superimposed over the digital image of each microscopic field studied. The numbers of intersections on the grid overlying the component of interest (Sertoli cell nuclei, Leydig cell nuclei, and cytoplasm or gonocyte nuclei) were then counted using a manual tag system. The ratio of the total count per testis to the total number of points possible (40 fields × 432 points per field) multiplied by testicular volume (cubic centimeters, derived from testis weight) was considered to be the absolute volume of the cellular component in question. These data were converted to absolute numbers of cells per testis by determining the average volume of the nucleus of each cell type. The mean nuclear volume of Leydig cells and gonocytes was measured using the stereology program, which calculates volume based on the assumption that all objects are spherical. The software takes the average of three measured diameters per nucleus. A minimum of 70 nuclei per cell type per testis were measured, and an average nuclear volume was then calculated. Because the Sertoli cell nucleus is not spherical, its volume was determined using a different approach (Mahood et al. 2005; Sharpe et al. 2000). An AOI was created by drawing around the Sertoli cell nucleus, within which the computer program then determined the average length of several diameters measured at 2° intervals that passed through the center of the nucleus. This was measured for a minimum of 70 Sertoli cell nuclei per testis, and mean nuclear volume was then determined.
Fetal ovine testes possess a fairly thick tunica (capsule) and contain a large central rete, neither of which contain Sertoli and Leydig cells or gonocytes. Because estimation of cell number per testis used testis weight as the measure of overall testis volume, this would lead to overestimation of cell numbers. To overcome this problem, the percentage of testis volume occupied by rete and capsule was measured using the Image Pro Plus software. One representative fetal testis from a sewage-sludge–exposed mother and one from a control mother was serially sectioned, and every 30th slide was immunostained for AMH, which does not stain the rete or the capsule and therefore clearly demarcates these structures. Three areas were then measured in the same way that the nuclear area of Sertoli cells was measured, described above: the rete, the testicular parenchyma (including rete), and the entire testis. The average percentage of testicular volume occupied by parenchyma was then calculated from these measurements, and this percentage used to correct the average volume of the testis in each animal before calculation of cell number per testis.
Determination of the Sertoli cell and gonocyte proliferation index.
The proliferation index of Sertoli cells was determined by counting the number of Ki-67–immunopositive and immunonegative Sertoli cell nuclei in 30 seminiferous cords per animal. The proliferation index of gonocytes was calculated in the same way. The Leydig cell proliferation index was not determined because Ki-67–immunopositive Leydig cells could not be reliably distinguished from other interstitial cells.
Hormone measurements.
Fetal serum levels of follicle-stimulating hormone (FSH) and luteinizing hormone (LH) were measured by radioimmunoassays that have been described and validated previously for sheep (McNeilly et al. 1986); the assay standards used were NIDDK-FSH-RP2 and NIH-LS18, and assay sensitivities were 0.1 and 0.2 ng/mL for FSH and LH, respectively. Serum levels of testosterone were measured using an enzyme-linked immunosorbent assay adapted from an earlier radioimmunoassay method (Corker and Davidson 1981), as described previously (Rivas et al. 2002); the limit of detection was 8 pg/mL. Serum levels of inhibin A, the main inhibin type produced by Sertoli cells in the male sheep (McNeilly et al. 2002), were measured using a two-site, enzyme-linked immunoassay that uses a capture antibody directed against amino acid sequence 82–114 of the human and ovine βA subunit and a C-specific biotinylated monoclonal antibody raised against a synthetic peptide that corresponds to amino acid sequence 1–32 of the human α-C subunit, as the detection antibody (Knight et al. 1998); the limit of detection was 20 pg/mL. For each hormone, all fetal blood samples were run in a single assay.
Statistical analysis.
We compared weights, cell numbers, and hormone levels in control and treated animals using Student’s t-test. Correlations between variables were assessed using the Pearson correlation test (Graphpad software, San Diego, CA, USA). In the present studies, four C and six T male fetuses were derived from multiple pregnancies and might therefore be viewed as not being truly independent samples. When values for such co-twins were meaned, and these values then reentered in statistical tests, it did reduce the level of significance for all parameters, but not so as to alter any of the main conclusions of the study. Mean values for all parameters for these co-twins also did not differ significantly from the remaining animals. Data are therefore presented with co-twins included as independent samples.
Results
Testes taken from male fetuses from ewes grazed on sewage-sludge–treated pastures are referred to as T testes, and those taken from male fetuses from ewes maintained on conventionally fertilized pasture are referred to as C testes. Two of the control ewes are known to have escaped into adjoining pastures in which a ram was present, between progestagen sponge removal and the prescribed mating date. Because the size of the four fetuses from these animals (F22, 23, 39, 40; Table 2) was entirely consistent with them having been conceived one cycle (17 days) earlier than were the remaining fetuses in the study, on the basis of known growth trajectories, it was concluded that they were probably mated one cycle earlier than the remaining animals. Similarly, the size of the T animal F47 (Table 2) indicated that it was probably conceived one cycle later than the remaining animals, based on fetal weight. Exclusion of these five animals from analyses did not materially affect the results or conclusions from these studies, although it did reduce the statistical power and significance of several of the findings and, in the instance of the most variable parameter (testosterone levels), it resulted in loss of statistical significance. For completeness, data pertaining to the five fetuses conceived at different times are indicated in all of the figures, and p-values are presented both with and without inclusion of data from these fetuses. Similarly, in the text below, where percent reductions caused by treatment are referred to, two values are quoted, the first excluding the five fetuses and the second including them.
Effects of treatment on maternal and fetal body weight and fetal testis weight.
Maternal exposure of ewes to pastures treated with sewage sludge (for the preceding 5 years, including the gestational period of 110 days) resulted in a 15–36% reduction in body weight and a 32–46% reduction in testis weight of male fetuses compared with those of C fetuses, although maternal body weight was similar in the two groups (Table 2). The mean (± SD) body weight of female fetuses from the same treated ewes (1,400 ± 183 g; n = 8) was also significantly reduced (p < 0.05) versus controls (1,593 ± 258 g; n = 15). Mean body weights were significantly higher (p < 0.05) in male C fetuses than in female C fetuses, but this sex difference was minimal in T animals. A significant positive correlation was observed between testis weight and body weight in both T (p < 0.005, r = 0.76) and C (p < 0.0001, r = 0.92) male fetuses, although this ratio was at least 20% lower in T than in C animals (Table 2).
Cell number per testis.
The number of Sertoli cells per T testis was 34–51% lower than in C testes at GD110 (p < 0.05; Figure 1A). However, average Sertoli cell nuclear volume remained relatively constant in the T (95.4 ± 8.8 μm3) and C (100 ± 10.5 μm3) testes.
The number of Leydig cells per testis was reduced by 37–46% in T compared with C testes (Figure 1B). The mean nuclear volume per Leydig cell for T testes (122 ± 14.7 μm3), however, was similar to that for controls (112 ± 14.5 μm3). The cytoplasmic volume of Leydig cells was comparable in T (91.4 ± 23.0 × 10−6 μm3) and C testes (79.6 ± 18.7 × 10−6 μm3) (Figure 1D), implying that the steroidogenic function of individual Leydig cells in T testes is not impaired.
As with Sertoli and Leydig cells, the total number of gonocytes per testis was significantly (p = 0.001) reduced (by 43–44%) in T compared with C testes (Figure 1C). Again, there was no significant difference in mean nuclear volume per gonocyte between T (294.7 ± 32.7 × 10−6 μm3) and C (310.1 ± 29.1 × 10−6 μm3) testes. Because multinucleated gonocytes have been shown to be induced in the fetal rat testis by in utero exposure to certain phthalates (Fisher et al. 2003; Mylchreest et al. 1999), we searched for these in C and T testes, but only occasional such cells (one to five per testis cross-section) were found in both groups.
Despite the differences in absolute cell number, described above, there was no significant difference between the T and C testes in the number of Sertoli cells (respectively, 180.8 × 106 cells/g and 193.5 × 106 cells/g), Leydig cells (273.2 × 106 cells/g and 279.9 × 106 cells/g), or gonocytes (20.0 × 106 cells/g and 21.2 × 106 cells/g) when expressed per gram of testis. There was also no significant difference in the ratios of each of the aforementioned cell types with each other in C and T testes (data not shown).
Changes in fetal plasma hormone levels.
There was no significant difference in plasma levels of either FSH or LH between T and C animals on GD110 (Figure 2A,B). However, plasma inhibin A concentrations were significantly decreased, by 38–46%, in T animals (p < 0.05; Figure 2C). Plasma testosterone levels showed a reduction of 50% in T compared with C testes (p < 0.05), and although a decrease in mean levels was still evident (36%) when excluding the five animals referred to above, it was no longer statistically significant (Figure 2D). We applied Pearson’s correlation test to determine the correlation between plasma levels of FSH and inhibin A and between LH and testosterone, but we found no significant correlation in either case.
Correlation between plasma hormone levels and cell number.
Because FSH can regulate Sertoli cell number and LH regulates testosterone secretion by Leydig cells, we explored the possible existence of significant correlations between these sets of parameters. No significant correlation between plasma FSH levels and Sertoli cell number or between LH and testosterone levels was found in either T or C animals (data not shown), implying that the lower Sertoli cell number and testosterone levels in T animals are not explained by altered gonadotropin levels.
Cell proliferation (Ki-67 immunostaining) in seminiferous cords.
We determined the proliferation index for Sertoli cells and gonocytes on the basis of the percentage of cells immunostained for Ki-67. There was no significant difference in the proliferation index in T versus C testes for either Sertoli cells (T, 18.6 ± 4.3%; C, 19.1 ± 3.1%; mean ± SD) or gonocytes (T, 31.4 ± 6.2%; C, 32.3 ± 7.2%).
Evaluation of testicular cell function using cell-specific and other markers.
Sertoli cells.
In order to determine whether or not there were any treatment effects on Sertoli cell function, immunoexpression of AMH was evaluated and was found to be of similar intensity in T and C testes (Figure 3A,B).
Leydig cells.
For Leydig cell function, we evaluated immunoexpression of two protein markers: P450scc and 3β-hydroxysteroid dehydrogenase (3β-HSD). No difference in the intensity of immunoexpression of P450scc was observed between T and C testes (Figure 3C,D). Immunoexpression of 3β-HSD was of a very low intensity in both T and C testes and so was not considered to be a good means of Leydig cell identification or of evaluating their function.
AR immunoexpression.
Nuclear immunoexpression of the AR was evident in interstitial and peritubular cells whereas Sertoli cells were largely immunonegative. There appeared to be less intense immunostaining for AR in interstitial and peritubular cells in T compared with C testes (Figure 3E,F). There was also strong immunostaining of AR in the cells of the rete epithelium in testes from both control and treated animals (data not shown).
Peritubular myoid cells.
Seminiferous cords appeared to be normally formed in T and C testes, on gross inspection. However, mean testosterone levels were lower in T than C testes, so immunoexpression of SMA in peritubular cells was evaluated because it is suggested to be androgen regulated in some species (Schlatt et al. 1993). Immunostaining intensity was generally lower in T than in C testes (Figure 3G,H).
Discussion
The principal aim of this study was to determine whether or not long-term experimental exposure of pregnant ewes to real-world cocktails of environmental chemicals (present in sewage sludge applied to the pasture) had any effect on testicular development in the male fetus. This was prompted by concerns about deteriorating human male reproductive health, in particular falling sperm counts (Swan et al 2000), and its possible relationship to environmental chemical exposures during fetal development (Sharpe and Irvine 2004; Sharpe and Skakkebaek 2003). Our results show that long-term exposure of breeding ewes to a mixture of chemicals added to pasture in sewage sludge (T animals), according to standard farming and European Union–recommended practices at the time the study was initiated (1997), resulted in major reductions (32–51%) in the numbers and hormonal function of the two principal somatic cell types of the fetal testis (Sertoli and Leydig cells) as well as a parallel reduction in the numbers of fetal germ cells. These changes were associated with growth restriction of male and female fetuses; body weight in the pregnant ewes was unaffected. The present study was restricted to late fetal life and therefore did not address whether the adverse testicular changes in T male fetuses have any permanent consequences. However, it might be anticipated that some aspects of the masculinization process in T male offspring could be attenuated (due to suppression of testosterone levels), as might sperm-producing capacity in adulthood (due to reduced Sertoli cell number). These remain to be explored, as does the more complex issue of which constituents of the treatment might have induced the adverse effects on testicular development. Under the present study conditions, sludge application has only minor effects on the soil concentrations of at least some of the more readily degraded EDs (Rhind et al. 2002); other studies indicate that the same is generally true of other, more persistent EDs (Smith 1995). It is therefore likely that the presently observed effects on the male fetus after exposure to sewage sludge reflects the effects of exposure to a mixture of environmental chemicals, and these may not be EDs as such.
In the present study we used established stereologic methods (Mahood et al. 2005; Sharpe et al. 2000) to quantify the total number of Sertoli cells, Leydig cells, and gonocytes per testis. There was a marked and consistent reduction (between 32 and 51%) in numbers of all three cell types in T compared with C fetal testes. The numerical changes in Sertoli and Leydig cells in treated animals were matched by parallel reductions in hormone production by these two cell types, as evidenced by the blood levels of inhibin A (secreted by Sertoli cells) and testosterone (secreted by Leydig cells), although the latter change was only statistically significant when all animals from the study were included. The similarity in degree of suppression of Sertoli and Leydig cell number, on the one hand, and in blood levels of their secreted hormones, on the other hand, suggests that the hormone-producing function of these cells is normal and that it is the reduction in cell number that accounts for the reduction in hormone levels in blood. Three other pieces of information reinforce this view. First, average Leydig cell cytoplasmic volume was not reduced in T animals, and steroidogenic function of the Leydig cells invariably goes hand-in-hand with their cytoplasmic volume (Ewing and Zirkin 1983). Second, immunoexpression of functional markers (AMH in Sertoli cells, P450scc in Leydig cells) did not reveal any obvious difference between T and C testes. Third, there was no change in the hormonal drive (LH, FSH) to the T testes, compared with C testes, that could account for either the reductions in cell number or reduced hormone production.
Several pieces of new information suggest that testosterone plays an important role in regulating Sertoli cell proliferation in fetal (Johnston et al. 2004) and early postnatal (Atanassova et al. 2005; Johnston et al. 2004; Ramaswamy et al. 2000) life, raising the possibility that treatment-induced suppression of testosterone levels in the present study could have contributed to the reduction in Sertoli cell number. The observation of a reduction in intensity of AR immunoexpression in testes of treated animals, which can occur when testosterone levels are low and which leads to destabilization of the AR protein (Bremner et al. 1994; Zhou et al. 1995), provides indirect support for this possibility, as does the reduction in SMA immunostaining in peritubular cells. The perinatal effect of androgens on Sertoli cell number is considered to be mediated via the peritubular myoid cells that express ARs (Atanassova et al. 2005; Sharpe 2005), whereas Sertoli cells do not express ARs in fetal life (Sharpe 2005), as confirmed for the sheep in the present study. If the reduction in Sertoli cell number in T testes in the present study results secondarily from reduced testosterone levels/action, the primary adverse change in treated animals could be in the proliferation and/or differentiation of the Leydig cells.
In the present studies, male T fetuses showed growth restriction compared with controls of the same gestational age, whereas growth restriction was somewhat less evident in female fetuses from the same pregnancies and the treated pregnant ewes themselves did not show any difference in body weight from control ewes. The latter observation makes it unlikely that the growth effects observed in the male fetuses result simply from generalized toxicity. The reduction in body weight in male T fetuses essentially resulted in obliteration of the normal sex difference in body weight—a change associated with reduced testosterone levels in the treated males.
It is well established for both sheep (Mitchell et al. 2002) and humans (Hindmarsh et al. 2002; Schwarzler et al. 2004) that male fetuses grow at a higher rate on average than do female fetuses, such that males are larger for most growth parameters than are females at most gestational ages, as well as at birth. Testosterone production by the male may contribute to this difference (de Zegher et al. 1999; Gill and Hosking 1995). Whether or not the treatment-induced growth restriction in the male fetuses contributed to the reduction in testicular parameters in the present study is uncertain (see Rhind et al. 2001). Experimental growth restriction in fetal sheep, as a result of underfeeding of the pregnant ewes, was shown in one study to result in a 20% reduction in Sertoli cell number at around birth (Bielli et al. 2002), whereas three other studies found no significant change in Sertoli cell number and/or testis weight (Da Silva et al. 2003; Rae et al. 2002a, 2002b). Furthermore, in one of these studies (Rae et al. 2002b) an increase, rather than a decrease, in blood testosterone levels in growth-restricted male fetuses was reported, in contrast to the present findings. We are not aware of any published experimental manipulation in sheep that results in such major reductions in fetal testis weight and testicular hormones as found in treated animals in the present study. Irrespective of the exact interrelationships between body weight, testicular growth, and testosterone production, the present findings may be consistent with observations in the human that show a robust relationship between fetal growth restriction and increased risk of reproductive developmental disorders such as cryptorchidism and hypospadias (Akre et al. 1999; Weidner et al. 1999), testicular cancer (Skakkebaek et al. 2001; Toppari et al. 1996), and, less consistently, reduced sperm counts (Cigognani et al. 2002; Francois et al. 1997; Olsen et al. 2000). It is also of relevance that all of these disorders can be associated with reduced production or action of testosterone by the fetal testis (Sharpe and Skakkebaek 2003; Skakkebaek et al. 2001). It will be important to explore whether or not such abnormalities occur also in treated sheep in follow-up studies to those presently reported.
The present study does not identify the cause(s) of growth restriction and impaired testis development in treated male fetuses other than to demonstrate its association with the experimental application of sewage sludge to the pasture. The sewage sludge contains a complex cocktail of chemicals that includes heavy metals, alkylphenolic compounds, phthalates, and other classes of EDs (Rhind et al. 2002; Smith 1995), and cases could be made for several of these individual compounds playing some part in the presently observed changes in treated animals (see IPCS 2002; Sharpe and Irvine 2004; Sweeney et al. 2000; Toppari et al. 1996). So far, no major difference has been observed in phthalate or alkylphenol tissue levels in T and C ewes (Rhind et al. 2005b) and/or their fetuses, although some differences in heavy metal exposure are evident (Rhind et al. 2005a; Wilkinson et al. 2001). It seems most likely that the presently observed effects in male fetuses stem from exposure of their mothers to a combination of chemicals that were present in the sewage sludge, but identifying what these may be is a difficult task.
It is not possible, from the information available, to gauge whether the present findings in sheep have direct relevance to humans in terms of their exposure to such chemicals, because the effective chemicals present in sewage sludge have not been identified. Nevertheless, because human waste is an important contributor to sewage sludge, it is not unreasonable to assume that humans are themselves exposed to many of its constituent chemicals, even if the actual levels of exposure may be much lower in the human than was the case in sheep from the present studies, in which sewage sludge was delivered onto the surface of the soil or herbage to maximize ewe exposure. The levels of exposure achieved experimentally in this study also probably exceed those in animals that are grazing on land fertilized by sewage sludge according to current recommendations in the United Kingdom. However, the importance of the present study is that it demonstrates that prolonged exposure of ewes to a cocktail of chemicals present in sewage sludge, with clear relevance to real-world chemical exposures, retards fetal growth and leads to major attenuation of testicular growth and parallel attenuation of hormone production, and that the effects observed in the fetal sheep appear relevant to concerns about human male reproductive disorders that stem from maldevelopment of the fetal testis (Skakkebaek et al. 2001).
We thank A. Esnal and G. Johnston for expert technical help. We also thank the staff of the Hartwood research station for their assisstance in the management of experimental animals.
This work was partly funded by the Scottish Executive Environment and Rural Affairs Department.
Figure 1 Numbers of Sertoli cells (A), Leydig cells (B), gonocytes (C), and Leydig cell cytoplasmic volume (D) in the testes of C and T fetuses on GD110. Each symbol corresponds to the cell number for an individual C or T animal (n = 12 per group); symbols that are circled indicate animals for which the gestational age may be different from GD110, and p-values in parentheses indicate the effect of removing these animals from statistical analysis. The horizontal line indicates the mean for each group.
Figure 2 Plasma levels of FSH (A), LH (B), inhibin A (C), and testosterone (D) in the testes of C and T fetuses on GD110. Each symbol corresponds to the cell number for an individual C or T animal (n = 12 per group); symbols that are circled indicate animals for which the gestational age may be different from GD110, and p-values in parentheses indicate the effect of removing these animals from statistical analysis. The horizontal line indicates the mean for each group.
Figure 3 Immunoexpression (brown staining) of AMH (A, B), P450ssc (C, D), AR (E, F), and SMA (G, H) in testicular sections from representative C and T animals. Note the lower expression of AR (E, F) in interstitial and peritubular cells (arrows) and the lower expression of SMA (G, H) in peritubular cells (arrows) of T animals, compared with C animals. Arrowheads in (G, H) indicate SMA staining in perivascular cells. Bar = 100 μm.
Table 1 Antibodies used for immunohistochemistry.
Antibody Type Source Dilution
AMH Goat IgG Santa Cruz Biotechnology (Santa Cruz, CA, USA) 1:1,000
AR Rabbit IgG Novocastra (Newcastle-Upon-Tyne, UK) 1:20
P450scc Rabbit IgG Chemicon (Hampshire, UK) 1:200
Ki-67 Rabbit IgG DAKO (High Wycombe, UK) 1:100
SMA Mouse monoclonal antibody Sigma (Poole, UK) 1:3,000
Table 2 Effects of sewage sludge exposure on maternal and fetal body weight and on fetal testis weight (at GD110).
Fetus
Fetus no. Treatment Ewe weight (kg) BW (g) TW (mg) TW:BW (%) Birth status
F31 Control 90 2,224 478 2.1 Triplet
F7 Control 75 1,706 273 1.6 Twin
F10 Control 75 1,575 347 2.2 Twin
F13 Control 79 1,613 307 1.9 Twin
F22a Control 84 2,733 366 1.3 Twin
F23a Control 84 3,545 600 1.7 Twin
F29 Control 92 1,347 172 1.3 Quad
F30 Control 92 1,478 221 1.5 Quad
F33 Control 90 2,127 336 1.6 Triplet
F34 Control 93 2,053 346 1.7 Single
F39a Control 81 3,059 505 1.7 Twin
F40a Control 81 3,389 641 1.9 Twin
Mean ± SD 84.7 ± 6.6 2,237 ± 768 383 ± 145 1.7 ± 0.3
Omitting F22, 23, 39, 40 84.9 ± 8.2 1,765 ± 326 310 ± 93 1.7 ± 0.3
F1 Treated 80 1,979 204 1.0 Triplet
F2 Treated 80 1,678 190 1.1 Triplet
F5 Treated 78 2,175 302 1.4 Single
F11 Treated 85 1,435 218 1.5 Twin
F16 Treated 95 1,871 269 1.4 Twin
F19 Treated 80 1,246 197 1.6 Triplet
F21 Treated 80 1,082 138 1.3 Triplet
F25 Treated 75 1,053 192 1.8 Triplet
F26 Treated 75 1,021 165 1.6 Triplet
F38 Treated 81 1,466 269 1.8 Twin
F43 Treated 88 1,471 189 1.3 Triplet
F47a Treated 66 789 152 1.9 Single
Mean ± SD 80.3 ± 7.9 1,438 ± 425 207 ± 50 1.5 ± 0.3
Omitting F47 81.9 ± 6.5 1,497 ± 391 212 ± 49 1.4 ± 0.3
Including all samples NS p = 0.0029 p = 0.0007 p = 0.037
Omitting F22, 23, 39, 40, 47 NS p = 0.06 p = 0.011 p = 0.0286
Abbreviations: BW, body weight; NS, not significant; TW, testis weight.
a Data for animals that may differ from the rest in terms of gestational age (see “Results”).
==== Refs
References
Akre O Lipworth L Cnattinguis S Sparen P Ekbom A 1999 Risk factor patterns for cryptorchidism and hypospadias Epidemiology 10 364 369 10401869
Atanassova NN Walker M McKinnell C Fisher JS Sharpe RM 2005 Evidence that androgens and oestrogens, as well as follicle-stimulating hormone, can alter Sertoli cell number in the neonatal rat J Endocrinol 184 107 117 15642788
Bielli A Perez R Pedrana G Milton JT Lopez A Blackberry MA 2002 Low maternal nutrition during pregnancy reduces the number of Sertoli cells in the newborn lamb Reprod Fertil Dev 14 333 337 12467358
Bremner WJ Millar MR Sharpe RM Saunders PTK 1994 Immunohistochemical localization of androgen receptors in the rat testis: evidence for stage-dependent expression and regulation by androgens Endocrinology 135 1227 1234 8070367
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. Publ. No. 02-0716. Atlanta, GA:Centers for Disease Control and Prevention. Available: http://www.cdc.gov/nceh/dls/report/ [accessed 12 May 2003].
Cigognani A Alessandroni R Pasinin A Pirazzoli P Cassio A Barbieri E 2002 Low birth weight for gestational age and subsequent male gonadal function J Pediatr 141 376 380 12219058
Commission of the European Communities 1994. Directorate Generale XI. Waste Management—Sewage Sludge. Part 1. Survey of Sludge Production, Treatment, Quality and Disposal in the European Union. European Commission. Brussels.
Corker CS Davidson DW 1981 Radioimmunoassay of testosterone in various biological fluids without chromatography J Steroid Biochem 9 319 323
Da Silva P Aitken RP Rhind SM Racey PA Wallace JM 2003 Effect of maternal overnutrition during pregnancy on pituitary gonadotrophin gene expression and gonadal morphology in female and male foetal sheep at day 103 of gestation Placenta 24 248 257 12566252
de Zegher F Devlieger H Eeckels R 1999 Fetal growth: boys before girls Horm Res 51 258 259 10559673
Ewing LL Zirkin B 1983 Leydig cell structure and steroidogenic function Recent Progr Horm Res 39 599 635 6415769
Fisher JS Macpherson S Marchetti N Sharpe RM 2003 Human ‘testicular dysgenesis syndrome’: a possible model based on in utero exposure of the rat to dibutyl phthalate Hum Reprod 18 1383 1394 12832361
Francois I de Zegher F Spiessens C D’Hooge T Vanderschueren D 1997 Low birth weight and subsequent male infertility Pediatr Res 42 899 901 9396576
Gill JW Hosking BJ 1995 Acute prenatal androgen treatment increases birth weights and growth rates in lambs J Anim Sci 73 2600 2608 8582849
Great Britain Parliament 1989. The Sludge (Use in Agriculture) Regulations. Statutory Instruments 1989, No. 1263. London:Her Majesty’s Stationary Office.
Hindmarsh PC Geary MP Rodeck CH Kingdom JC Cole TJ 2002 Intrauterine growth and its relationship to size and shape at birth Pediatr Res 52 263 268 12149505
IPCS (International Programme on Chemical Safety) 2002. Global Assessment of the State-of-the-Science of Endocrine Disruptors (Damstra T, Barlow S, Bergman A, Kavlock R, Van Der Kraak G, eds). Available: http://ehp.niehs.nih.gov/who/ [accessed 12 May 2003].
Johnston H Baker PJ Abel M Charlton HM Jackson G Fleming L 2004 Regulation of Sertoli cell number and activity by follicle-stimulating hormone and androgen during postnatal development in the mouse Endocrinology 145 318 329 14551232
Knight PG Feist SA Tannetta DS Bleach ECL Fowler PA O’Brien M 1998 Measurement of inhibin-A (αβA isoform) during the oestrous cycle, after manipulation of ovarian activity and during pregnancy in ewes J Reprod Fertil 113 159 166 9713389
Liao KH Dobrev ID Dennison JE Jr Andersen ME Reisfeld B Reardon KF 2002 Application of biologically based computer modelling to simple or complex mixtures Environ Health Perspect 110 suppl 6 957 963 12634125
Mahood IK McKinnell C Fisher JS Walker M Hallmark N Sharpe RM 2005 Abnormal Leydig cell aggregation in the fetal testis of rats exposed to di(n -butyl) phthalate and its possible role in testicular dysgenesis Endocrinology 146 613 623 15539561
McNeilly AS Jonassen JA Fraser HM 1986 Suppression of follicular development after chronic LHRH immunoneutralization in the ewe J Reprod Fertil 76 481 490 3080596
McNeilly AS Souza CJ Baird DT Swanston IA McVerry J Crawford J 2002 Production of inhibin A not B in rams: changes in plasma inhibin A during testis growth, and expression of inhibin/activin subunit mRNA and protein in adult testis Reproduction 123 827 835 12052237
Mitchell LM Silveira M Mylne MJ Matthews K Dingwall WS 2002 Seasonal differences in lamb birthweight do not arise from inherent differences in the oocyte and/or early embryo Reprod Fertil Dev 14 207 213 12219943
Mylchreest E Sar M Cattley R Foster PMD 1999 Disruption of androgen-regulated male reproductive development by di(n -butyl) phthalate during late gestation in rats is different from flutamide Toxicol Appl Pharmacol 156 81 95 10198273
Olsen J Bonde J-P Basso O Hjollund NHI Sorensen HT Abell A 2000 Birthweight and semen characteristics Int J Androl 23 230 235 10886426
Parks L Ostby J Lambright C Abbott B Klinefelter GD Barlow N 2000 The plasticizer diethylhexyl phthalate induces malformations by decreasing fetal testosterone synthesis during sexual differentiation in the male rat Toxicol Sci 58 339 349 11099646
Rae MT Kyle CE Miller DW Hammond AJ Brooks AN Rhind SM 2002a The effects of undernutrition, in utero, on reproductive function in adult male and female sheep Anim Reprod Sci 72 63 71 12106966
Rae MT Rhind SM Fowler PA Miller DW Kyle CE Brooks AN 2002b Effect of maternal undernutrition on fetal testicular steroidogenesis during the CNS androgen-responsive period in male sheep fetuses Reproduction 124 33 39 12090916
Rajapakse N Silva E Kortenkamp A 2002 Combining xeno-estrogens at levels below individual no-observed-effect concentrations dramatically enhances steroid hormone action Environ Health Perspect 110 919 921 12634120
Ramaswamy S Plant TM Marshall GR 2000 Pulsatile stimulation with recombinant single chain human luteinizing hormone elicits precocious Sertoli cell proliferation in the juvenile male rhesus monkey (Macaca mulatta ) Biol Reprod 63 82 88 10859245
Rhind SM Kyle CE Own J 2005a Accumulation of potentially toxic metals in the liver tissue of sheep grazed on sewage sludge-treated pastures Anim Sci 81 107 113
Rhind SM Kyle CE Telfer G Duff EI Smith A 2005b Alkyl phenols and diethyl hexyl phthalate (DEHP) in tissues of sheep grazing pastures fertilized with sewage sludge or inorganic fertilizer Environ Health Perspect 113 447 453 15811823
Rhind SM Rae MT Brooks AN 2001 Effects of nutrition and environmental factors on the fetal programming of the reproductive axis Reproduction 122 205 214 11467971
Rhind SM Smith A Kyle CE Telfer G Martin G Duff E 2002 Phthalate and alkyl phenol concentrations in soil following applications of inorganic fertiliser or sewage sludge to pasture and potential rates of ingestion by grazing ruminants J Environ Monit 4 142 148 11871695
Rivas A Fisher JS McKinnell C Atanassova N Sharpe RM 2002 Induction of reproductive tract developmental abnormalities in the male rat by lowering androgen production or action in combination with a low dose of diethylstilbestrol: evidence for importance of the androgen-estrogen balance Endocrinology 143 4797 4808 12446607
Schlatt S Weinbauer G Arslan M Nieschlag E 1993 Appearance of alpha-smooth muscle actin in peritubular cells of monkey testes is induced by androgens, modulated by follicle-stimulating hormone, and maintained after hormonal withdrawal J Androl 14 340 350 8288489
Schwarzler P Bland JM Holden D Campbell S Ville Y 2004 Sex-specific antenatal reference growth charts for uncomplicated singleton pregnancies at 15–40 weeks of gestation Ultrasound Obstet Gynecol 23 23 29 14970994
SEDE and Arthur Andersen 2001. Disposal and Recycling Routes for Sewage Sludge. Part 1. Sludge Use Acceptance Report. European Commission/Directorate Generale Environment. Luxembourg:European Communities.
Sharpe RM 2005. Sertoli cell endocrinology and signal transduction: androgen regulation. In: Sertoli Cell Biology (Griswold M, Skinner M, eds). San Diego:Academic Press, 199–216.
Sharpe RM Irvine DS 2004 How strong is the evidence of a link between environmental chemicals and adverse effects on human reproductive health? BMJ 328 447 451 14976101
Sharpe RM Skakkebaek NE 2003 Male reproductive disorders and the role of endocrine disruption: advances in understanding and identification of areas for future research Pure Appl Chem 75 2023 2038
Sharpe RM Walker M Millar MR Atanassova NN Morris K McKinnell C 2000 Effect of neonatal gonadotropin-releasing hormone antagonist administration on Sertoli cell number and testicular development in the marmoset: comparison with the rat Biol Reprod 62 1685 1693 10819772
Skakkebaek NE Rajpert-De Meyts E Main K 2001 Testicular dysgenesis syndrome: an increasingly common developmental disorder with environmental aspects Hum Reprod 16 972 978 11331648
Smith SR 1995. Organic pollutants. In: Agricultural Recycling of Sewage Sludge and the Environment. Wallingford, UK:CAB International, 207–236.
Swan SH Elkin EP Fenster L 2000 The question of declining sperm density revisited: an analysis of 101 studies published 1934–1996 Environ Health Perspect 108 961 966 11049816
Swan SH Main KM Liu F Stewart SL Kruse RL Calafat AM 2005 Decrease in anogenital distance among male infants with prenatal phthalate exposure Environ Health Perspect 113 1056 1061 16079079
Swanson RL Bortman ML O’Connor P Stanford HM 2004 Science, policy and the management of sewage materials. The New York City experience Marine Pollut Bull 49 679 687
Sweeney T Nicol L Riche JF Brooks AN 2000 Maternal exposure to octylphenol suppresses fetal follicle-stimulating hormone secretion, testis size and Sertoli cell number Endocrinology 141 2667 2673 10875272
Tindall H Ashby J 2004 Sensitivity of the immature rat uterotrophic assay to mixtures of estrogens Environ Health Perspect 112 575 582 15064164
Toppari J Larsen JC Christiansen P Giwercman A Grandjean P Guillette LJ Jr 1996 Male reproductive health and environmental xenoestrogens Environ Health Perspect 104 suppl 4 741 803 8880001
U.K. Animals (Scientific Procedures) Act 1986. Available: http://www.homeoffice.go.uk/docs/animallegislation.html
Webber MD Lesage S 1989 Organic contaminants in Canadian municipal sludges Waste Manage Res 7 63 82
Weidner IS Moller H Jensen TK Skakkebaek NE 1999 Risk factors for crytptorchidism and hypospadias J Urol 161 1606 1609 10210427
Wilkinson JM Hill J Livesey CT 2001 Accumulation of potentially toxic elements in the body tissues of sheep grazed on grassland given repeated applications of sewage sludge Anim Sci 72 179 190
Zhou ZX Lane MV Kemppainen JA French FS Wilson EM 1995 Specificity of ligand-dependent androgen receptor stabilization: receptor domain interactions influence ligand dissociation and receptor stability Mol Endocrinol 9 208 218 7776971
|
16263515
|
PMC1310922
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 11; 113(11):1580-1587
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8028
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7992ehp0113-00158816263516ResearchA Rapid, Physiologic Protocol for Testing Transcriptional Effects of Thyroid-Disrupting Agents in Premetamorphic Xenopus Tadpoles Turque Nathalie Palmier Karima Le Mével Sébastien Alliot Caroline Demeneix Barbara A. Unité Mixte de Recherche, Centre National de la Recherche Scientifique, Evolution des Régulations Endocriniennes, Department of Regulations, Development and Molecular Diversity, Museum National d’Histoire Naturelle, Paris, FranceAddress correspondence to B.A. Demeneix, UMR CNRS, Evolution des Régulations Endocriniennes, Department of Regulations, Development and Molecular Diversity, Museum National d’Histoire Naturelle, 7 Rue Cuvier, 75231 Paris Cedex 05, France. Telephone: 33 140 793 607. Fax: 33 140 793 618. E-mail:
[email protected]. and N.T. are inventors of “Transgenic Clawed Frog Embryos and Use Thereof as Detectors of Endocrine Disruptors in the Environment,” International patent no. WO 03102176. The remaining authors declare they have no competing financial interests.
11 2005 11 7 2005 113 11 1588 1593 3 2 2005 11 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Increasing numbers of substances present in the environment are postulated to have endocrine-disrupting effects on vertebrate populations. However, data on disruption of thyroid signaling are fragmentary, particularly at the molecular level. Thyroid hormone (TH; triiodothyronine, T3) acts principally by modulating transcription from target genes; thus, thyroid signaling is particularly amenable to analysis with a transcriptional assay. Also, T3 orchestrates amphibian metamorphosis, thereby providing an exceptional model for identifying thyroid-disrupting chemicals. We combined these two advantages to develop a method for following and quantifying the transcriptional action of T3 in Xenopus laevis tadpoles. This technology provides a means of assessing thyroid activity at the molecular level in a physiologically relevant situation. Moreover, translucent tadpoles are amenable to “on-line” imaging with fluorescent reporter constructs that facilitate in vivo measurement of transcriptional activity. We adapted transgenesis with TH-responsive elements coupled to either luciferase or green fluorescent protein to follow T3-dependent transcription in vivo. To reduce time of exposure and to synchronize responses, we optimized a physiologic pre-treatment protocol that induced competence to respond to T3 and thus to assess T3 effects and T3 disruption within 48 hr. This pretreatment protocol was based on a short (24 hr), weak (10−12 M) pulse of T3 that induced TH receptors, facilitating and synchronizing the transcriptional responses. This protocol was successfully applied to somatic and germinal transgenesis with both reporter systems. Finally, we show that the transcriptional assay allows detection of the thyroid-disrupting activity of environmentally relevant concentrations (10−8 M) of acetochlor, a persistent herbicide.
acetochlorendocrine disruptiongerminal transgenesisgreen fluorescent proteinmetamorphosissomatic gene transferthyroidtranscriptionXenopus laevis
==== Body
The thyroid hormone (TH) triiodothyronine (T3) is critical to vertebrate development and growth, playing vital roles during central nervous system (CNS) development and during organogenesis of heart, muscles, bones, and lungs. In both developing and mature organisms, numerous physiologic functions are regulated by T3 availability, including energy metabolism, thermogenesis, pituitary hormone production, and lipogenesis (Yen 2001). The most striking example of T3 action in vertebrates is anuran amphibian metamorphosis, one of the best-studied hormone-regulated developmental processes. Amphibian metamorphosis is totally dependent on T3 and is associated with dramatic morphologic and physiologic changes, including cell death, division, or differentiation (Dodd and Dodd 1976).
The total dependence of amphibian metamorphosis on TH has logically led to the suggestion that metamorphosis can be used to assess TH disruption. Accordingly, the Xenopus embryonic metamorphosis assay (XEMA) has been proposed by the Organization for Economic Cooperation and Development (OECD) Task Force on Endocrine Disrupters Testing and Assessment as an in vivo assay for identification of substances with potential to disrupt functions of the thyroid system. In this in vivo test, several morphologic and histologic parameters are used to define the potential of a chemical to perturb the thyroid axis (OECD 2004). Because the test covers most of the natural metamorphic process, it takes at least 4 weeks. In contrast, transcriptional responses to T3 are much more rapid, with changes being measurable within hours or days. Indeed, T3 actions are mainly mediated by their nuclear receptors (TH receptors, TRs), ligand-dependent transcription factors. In vertebrates, two genes encode TRs: TR-α and TR-β (Mangelsdorf et al. 1995). Generally, TRs form heterodimers with the 9-cis-retinoic acid receptor and interact with comodulator complexes, thereby repressing or activating transcription. Heterodimers bind to TH response elements (or T3 responsive element; TRE) in target genes.
We chose to take advantage of the speed of the transcriptional responses to TH and the ease of measuring reporter gene activities, such as luciferase (luc) and green fluorescent protein (GFP), using transgenic approaches. Moreover, both somatic (de Luze et al. 1993; Nakajima and Yaoita 2003; Ulisse et al. 1996) and germinal transgenesis (Coen et al. 2001; Huang and Brown 2000; Marsh-Armstrong et al. 1999; Oofusa et al. 2001) are now widely applied to Xenopus laevis for dissecting TH-dependent regulations during metamorphosis. The transgenic models we optimized here are based on the fundamental design of composite reporter gene constructs with a hormone-sensitive regulatory region upstream of a fluorescent protein cDNA. We started from the premise that following transcriptional responses in vivo is often marred by high variability. In the case of following TH responses in tadpoles, this variability could be due to variations in endogenous TR levels. Because TR-β is strongly inducible by T3 itself, we chose to prime tadpoles to respond to an ulterior T3 exposure with a short, weak pulse of T3 that was then fully rinsed out. This protocol produced rapid (48 hr), robust, and reproducible responses to TH agonists. Applying this protocol to germinally transgenic tadpoles, we were able to reveal the actions of the preemergent herbicide acetochlor through increased TH responses.
Materials and Methods
Plasmid constructs.
The –246 to +130 bp sequence of the TH/bZIP promoter (GenBank accession no. U37375; GenBank 2004) was amplified by polymerase chain reaction (PCR) from X. laevis genomic DNA using the primers 5′-CTGTTATATAGAGGCAGAGGG-3′ and 5′-CTATACCTGAATGGGCAGCAG-3′, and then cloned into pGEMt-easy vector (Promega, Lyon, France). A SacII-PstI–digested fragment was excised and cloned into pBluescript (Promega). A SacI-HindIII–digested fragment was excised and cloned into pGL2 basic vector (Promega), producing TH/bZIP-luc.
To obtain the TH/bZIP-eGFP (enhanced GFP) transgene, we proceeded in two steps. We cloned a SalI-ApaI eGFP cDNA and the SV40 polyA signal fragment into the TH/bZIP promoter containing pBluescript. A large SacI-ApaI fragment was excised and cloned into a pBluescript vector with two insulators. To this end, a 1,668-bp fragment of the lysozyme gene (GenBank accession no. X98408) from chicken genomic DNA (Stief et al. 1989) was amplified by PCR using the primer 5′-TGACTCGAGGGATCCATAATATAACTGTACC-3′ and 5′-TGAGGTACCAAGCTTAAAAGATTGAAGCAC-3′. One insulator copy was cloned into XhoI and KpnI sites of the pBluescript vector, and the second was cloned into the SmaI site. The complete vector with the two copies of insulators was linearized with EcoRV, and the large SacI-ApaI fragment corresponding to the eGFP cDNA and SV40 polyA signal was inserted.
The γ-crystallin promoter coupled to a RedFP (red fluorescent protein) plasmid was a gift from L. Zimmermann (Medical Research Council, London).
Animals and treatments.
We obtained sexually mature X. laevis frogs from d’Elevage de Xénope du Centre National de la Recherche Scientifique (Montpellier, France). Tadpoles were raised in dechlorinated and deiodinated tap water (1:2) and fed with nettle powder (Vallée, Chanzeaux, France). Tadpoles were staged according to Nieuwkoop and Faber (1956; NF staging). The care and treatment of animals used in this study were in accordance with institutional and national guidelines (Sciences et Médecine des Animaux de laboratoire à l'ENVL 2005).
T3, 3,5,3′-triiodothyroacetic acid (TRIAC), and acetochlor were purchased from Sigma (St. Quentin Fallavier, France).
Somatic gene transfer and germinal trans-genesis.
Somatic gene transfer in Xenopus muscle and brain was performed as described previously (de Luze et al. 1993; Ouatas et al. 1998; Trudeau et al. 2004).
Germinally transgenic tadpoles were produced by restriction enzyme-mediated integration nuclear transplantation according to Kroll and Amaya (1996), with the following modifications: sperm was purified by centrifugation on a two-layer discontinuous Percoll (Sigma) gradient before the permeabilization step, which was performed with digitonin (Sigma) instead of lysolecithin. Two plasmids were used: TH/bZIP-eGFP plasmid and a γ-crystallin promoter coupled to a RedFP plasmid, which is expressed only in the eye. This latter plasmid allows selection of transgenic F0 tadpoles during early development before the TH/bZIP driven green fluorescence appears in the tadpole body.
Imaging.
Images were captured using an Olympus fluorescent dissecting microscope equipped with an Olympus video camera DP50 (Olympus, Rungis, France). Before photographing, germinally transgenic tadpoles NF stage 52 were anaesthetized in 0.1% tricaine methanesulfonate (MS-222; Sigma) and the skull opened to expose the brain. All pictures were taken with the same parameters (32 × objective and 5-sec exposure time). Quantification was performed using ImageJ software (Rasband 1997). Data are expressed in relative units of fluorescence.
Luciferase activity.
Tadpoles were sacrificed by decapitation after anesthesia in 0.1% MS-222. Tissues were dissected, frozen in liquid nitrogen, and stored at –80°C until assayed according to the manufacturer’s instructions (Promega) as previously reported (de Luze et al. 1993). Luciferase activity is expressed as relative light units (RLU). Because in some experiments tadpoles vary in size, luciferase values were normalized against protein content. Protein was measured according to the manufacturer’s instructions (BioRad, Marnes-La-Coquette, France).
Statistical analysis of results.
In vivo gene transfer results are expressed as mean ± SE per group. Differences between means were analyzed by Student’s t-test or analysis of variance and the Tukey-Kramer test where appropriate. Differences were considered significant at p < 0.05. In many cases, typical experiments are shown, each experiment having been repeated at least twice (with n ≥ 8 tadpoles/experiment) and providing the same results.
RNA extraction and semiquantitative reverse transcriptase (RT)-PCR analysis.
Tadpole tails were harvested into RNAlater (Ambion, Huntingdon, United Kingdom) at 4°C. Total RNA was extracted using RNAble reagent (Eurobio, Les Ulis, France) following the manufacturer’s protocol. Reverse transcription was performed on 2 μg RNA, in 20 μL final volume. Primer hybridization on RNAs was done by mixing total RNAs with specific reverse primers (2 μM each): for TR-β, 5′-CTTTTCTATTCTCTCCACGCTAGC-3′; for the internal control Rpl8, 5′-GACGACCAGTACGACGA-3′ (Havis et al. 2003). Mixes were incubated (2 min, 65°C) and then cooled to room temperature. A 10-μL mix containing 4 μL reverse transcription buffer (Invitrogen, Cergy Pontoise, France), 1 μL dNTP (dATP, dTTP, dCTP, and dGTP, 25 mM each; Pharmacia, Saclay, France), 2 μL 0.1 M dithiothreitol (Invitrogen), and 0.5 μL reverse transcriptase (RT) SuperScript II (5 U/μL; Invitrogen) was added to hybrid primers/RNA before incubation (1 hr, 45°C). After reverse transcription, we used 2 μL of each cDNA sample and 0.1 μL (0.1 μCi) of [α-32P] dCTP for PCR in a final volume of 50 μL containing 25 μL PCR Master Mix (Abgene, Courtaboeuf, France), 2 μL of each primer (forward and reverse for the gene of interest and the internal control at 2 μM each; TR-β forward, 5′-ATAGTTAATGCGCCCGAGGGTGGA-3′; and the internal control, Rpl8 forward, 5′-AAAGAGAAACTGCTGGC-3′). The PCR reaction consisted of 22 cycles of 60 sec at 94°C, 1 min at 55°C, and 1 min at 72°C (Robocycler; Stratagene, Amsterdam, The Netherlands). Fifteen microliters of PCR products was resolved in 6% acrylamide-trisborate-EDTA buffer gels and autoradiographed.
Results
Treatment with a rapid, weak pulse of T3 induces competence to respond and synchronizes T3 transcriptional responses in somatic gene transfer.
Our overall aim in these experiments was to exploit the T3-dependent inducibility of the TR-β gene to prime tadpoles so as to have access to a model that could reveal rapid, robust, and reproducible TH transcriptional responses. This priming, or pretreatment procedure, was considered a prerequisite to using premetamorphic (NF stage 54) tadpoles in a reporter gene assay, because intragroup variability can be quite high (de Luze et al. 1993). Using the promoter of the TH/bZIP gene, which is a T3 target gene encoding a transcription factor (Furlow and Brown 1999), we compared the effects of a short pulse (24 hr) of 10−13 M or 10−12 M T3 for the pre-treatment against unpretreated, control tadpoles. After a 24-hr rinse period, both sets of tadpoles (pretreated and control) were used for somatic gene transfer in the caudal muscle and exposed to T3 (10−8 M) for 2 days. We found 2-fold increases in mean levels of TH/bZIP expression in caudal muscle of control tadpoles and those given a pretreatment pulse of T3 at 10−13 M (Figure 1B). However, these differences were not significant (p > 0.05). In contrast, in tadpoles transiently pretreated with 10−12 M T3, 2 days of exposure to 10−8 M T3 induced a nearly 4-fold increase in the transcriptional response of the TH/bZIP-luc construct. This increase was very significant (p < 0.01). Interestingly, the basal levels of transcription from the TH/bZIP promoter decreased as a function of pretreatment pulse concentration, most probably reflecting stronger repression of basal expression by the unliganded TR (Sachs 2004).
A weak pulse of T3 induces TR-β expression in caudal muscle of tadpoles within 6 hr.
To verify that the pretreatment protocol was indeed inducing TR-β, we followed TR expression in caudal muscle of pretreated tadpoles using semiquantitative RT-PCR. Figure 2 shows that there is a significant, 2.2-fold induction in TR-β expression within 6 hr of exposure of NF stage 54 tadpoles to 10−12 M T3. Given this finding, and the robust response produced by pretreatment, all the following experiments were performed on animals pretreated with 10−12 M T3 during 24 hr, followed with a rinse of 24 hr before exposure to TH agonists for 48 hr.
The somatic gene transfer method allows dose-dependent detection of TH agonists in brain and muscle.
We used somatic gene transfer with the pretreatment protocol to test transcriptional responses to other TH agonists. Figure 3 shows that pretreated tadpoles exposed to TRIAC (5 × 10−8 M) for 48 hr displayed a 3-fold increase in transcription from the TH/bZIP promoter (p < 0.05) compared with controls. Similar results were found with 3,5,3′,5′-tetraiodothyronine (T4; data not shown).
In order to test the sensitivity of the somatic gene transfer method and the eventual tissue specificity in transcriptional response to TH agonists, we compared responses in caudal muscle or the brain of pretreated tadpoles. As shown in Figure 4A, in caudal muscle, a 5 × 10−10 M T3 exposure did not significantly increase transcription, but 5 × 10−9 M and 5 × 10−8 M T3 induced significant responses of 3-fold (p < 0.01) and 5-fold (p < 0.05), respectively. Figure 4B shows results from similar experiments performed using somatic gene transfer in the tadpole brain. All the T3 concentrations used (10−10 M, 10−8 M, and 10−7 M) gave very significant transcriptional responses from the TH/bZIP-luc construct: 3-fold (p < 0.01), 5-fold (p < 0.001), and 5.5-fold (p < 0.001), respectively. Thus, the brain is a more sensitive TH target than is muscle, in terms of transcriptional responses from the TH/bZIP-luc construct.
The pretreatment protocol can be used on germinal transgenic tadpoles.
Because the TH/bZIP construct used in somatic gene transfer experiments had proven suitable for detecting of T3 agonists in muscle and brain, we used the same construct in germinal trans-genesis (Kroll and Amaya 1996). We prepared a vector with the TH/bZIP-GFP chimeric gene between two copies of the lysozyme chicken gene as insulators (Stief et al. 1989). As a means of verifying transgenesis, we injected a second plasmid containing the γ-crystallin promoter coupled to a RedFP. This plasmid is expressed early in the eye and allows selection of transgenic F0 tadpoles before the TH/bZIP-dependent green fluorescence appears in the tadpole body.
We first followed TH/bZIP-driven fluorescence during early development and metamorphosis, focusing on the brain and on limb buds. In limb buds, the transgene is barely expressed at NF stage 51, and expression remains weak until NF stages 61–62. During metamorphosis, the signal intensifies. The transgene is expressed in the CNS at NF stage 51, and the fluorescent signal increases gradually through each metamorphic stage (Figure 5A). However, TH/bZIP expression remains at low enough levels in the brain so as not to interfere with fluorescence induction by TH agonists.
A number of F0 germinally transgenic tadpoles were selected and bred to obtain F1 tadpoles. We next exposed F1 transgenic tadpoles to T3 (10−8 M) using the pretreatment protocol. We confirmed that a short (24 hr) pretreatment with a low concentration of T3 (10−12 M), followed by a rinse, permits a rapidly detectable fluorescent signal in limb buds (forelimb and hindlimb buds) (Figure 5B), that is, after 48-hr exposure to T3 (10−8 M). In contrast, if the pretreatment protocol is not used, 4 days of exposure to T3 (10−8 M) is required to obtain a significant induction (data not shown). Fluorescence was induced in the brain, the olfactory nerves, and the gills (Figure 5B). TRIAC (5 × 10−8 M) induced similar responses in the brain (data not shown) and limb buds (Figure 5B).
Acetochlor disruption of thyroid signaling can be assessed within 48 hr.
Because the pre-treatment protocol can be used with germinally transgenic tadpoles to reveal T3 effects, we applied it to NF stage 52 tadpoles to assess the potential thyroid-disrupting effects of the herbicide acetochlor. Pretreated tadpoles were exposed to 10−10 M T3 or to 10–10 M T3 plus 10−8 M acetochlor for 48 hr. As shown in Figure 6A, the 20% increase in fluorescence in brains of germinally transgenic tadpoles was amplified by addition of 10−8 M acetochlor (p < 0.001 vs. controls). Figure 6B shows examples of germinally transgenic tadpoles brains that were quantified using ImageJ software.
Discussion
A central part of this study included establishing a rapid and sensitive method for assaying TH agonist activity within a shorter time frame than the several weeks needed to record TH effects on morphologic changes. This objective required fulfillment of numerous criteria: low background with no interference from endogenous hormone, robust and statistically significant responses, dose dependence, and low threshold. To avoid interference from endogenous hormone, we used euthyroid tadpoles at stages of development where TH levels are naturally low (Leloup and Buscaglia 1977). We had previously found that, when using euthyroid tadpoles in the somatic gene transfer test, a minimum of 4 days was necessary for significant induction of T3-dependent transcription (de Luze et al. 1993). Moreover, variability of responses among individuals was high. We theorized that the delay to response and the range of response levels could be due to lack of competence to respond to TH, possibly due to insufficient TRs in the target tissue, caudal muscle. Indeed, many experimental data have shown that TRs are expressed only at low levels before premetamorphosis, after which TR-β is strongly induced by the T3 signal, the TR-β genes in X. laevis having complex promoters containing multiple positive TREs (Urnov and Wolffe 2001). To overcome this TR insufficiency, we chose to treat tadpoles with a brief, weak pulse of T3 to induce competence to respond to a later exposure to TH. The logic was that the short pulse of T3 should synchronize and harmonize the tadpole responses by up-regulating expression of TR-β, and possibly cofactors, thus facilitating TH responses. Further, given that the animals were then rinsed, it was expected that any hormone taken up would be degraded during the 24-hr rinse period, given that the half-life of T3 in most vertebrates is around 18–24 hr (Van Middlesworth 1974).
We established that a short priming or pre-treatment (24 hr) used with a weak concentration of T3 (10−12 M) induced an up-regulation of TR-β expression and synchronized responses. These low concentrations and short exposure times do not induce any major morphologic changes in the tadpoles. TR-β transcript levels were increased 2-fold within 6 hr by 10−12 M T3 (Figure 2). These data confirmed previous experiments on whole tail tissue (Havis et al. 2003) or on tail tissues undergoing apoptosis (Helbing et al. 2003). In both cases, the authors observed a significant increase of TR-β expression, however, with much higher T3 concentrations (10−8 M T3 and 10−7 M T3) than we used.
We next determined whether responses were physiologic. We show that this is the case, in that they are dose dependent in muscle and brain (Figure 4) and are sensitive to known TH agonists, for example, TRIAC (Figure 4). Interestingly, using the TH/bZIP construct we observed a lower threshold to TH agonists in the brain than in muscle. This sensitivity correlates with the main site of TH/bZIP expression, that is, the CNS. Indeed, this gene was first isolated from the diencephalon of X. laevis tadpoles (Denver et al. 1997), along with another 33 TH-regulated genes, including deiodinase and other metabolic enzymes. Some of these genes have also been isolated in neonatal mammals and chicks and could provide useful targets to be employed in screening approaches for thyroid disruption.
The pretreatment protocol permits rapid detection and quantification of TH agonist action in germinally transgenic tadpoles.
Somatic gene transfer is ideal for comparing responses of different constructs and for setting up physiologic protocols. There is no need to establish founders or to maintain frog lines, and a number of test situations can be compared simultaneously. However, once a construct and a protocol have been selected, then germinal gene transfer becomes a much more efficient method for scaling up procedures for screening purposes. Using germinal transgenesis has the advantages of several hundred tadpoles per brood and a homogeneous population in terms of transgene insertion site and consequent regulatory controls. Furthermore, germinally transgenic tadpoles can also be used in a long-term assay, such as the XEMA test, to study impacts of longer term exposure to chemicals. The GFP signal can provide information on tissue-specific and developmental stage-specific actions during metamorphic progress, information that cannot easily be gleaned from the XEMA test on wild-type tadpoles (OECD 2004). Because TH/bZIP is expressed in all target tissues during amphibian metamorphosis and is highly responsive to TH in somatic gene transfer, we used it in germinal transgenesis. The genomic PCR fragment used with the luciferase reporter gene in somatic gene transfer experiments was fused to the eGFP reporter gene and inserted into an insulator-containing plasmid. The use of insulators helps overcome the influence of insertional position effects on transcriptional response, ensuring more homogeneous basal levels of expression of the transgene. Moreover, insulators have been shown to protect transgenes from methylation and maintain transgene expression in descendants (Kirillov et al. 1996) and have been successfully used in mice (Ciana et al. 2001). We applied the pretreatment protocol to TH/bZIP germinally transgenic tadpoles. TH/bZIP promoter-driven eGFP expression was significantly induced in brain and limb buds after a 2-day exposure to T3 (10−8 M) or to TRIAC (5 × 10−8 M), whereas no fluorescence was seen in caudal muscle during either natural metamorphosis or induced metamorphosis. This observation could be explained by the fact that we used only 400 bp of the TH/bZIP promoter, a fragment that might not include the enhancers responsible for targeting muscle expression.
Applying the pretreatment protocol to germinally transgenic tadpoles allows rapid assessment of thyroid-potentiating effects of acetochlor.
Several current lines of research have revealed the widespread presence of hormonal pollutants in the environment. These disrupting substances of natural or synthetic origin interfere with hormone action affecting numerous functions, including homeostasis, reproduction, development, and behavior (Kavlock and Ankley 1996).
We tested the methodology described herein for assessing potential thyroid disruptors using the well-established thyroid disruptor acetochlor [2-chloro-N-(ethoxymethyl)-N-(2-ethyl-6-methylphenyl)acetamide]. This herbicide was introduced in 1994 in the midwestern United States (Kolpin et al. 1996). Acetochlor is persistent; surface water concentrations of acetochlor were found to be 0.2–4.5 nM at 1–3 months after application (Kolpin et al. 1996). Significant levels can still be detected in shallow groundwater 1 year after application (Crump et al. 2002). Acetochlor has been shown to alter thyroid axis functions in the rat (Wilson et al. 1996) and to alter the rate of metamorphosis in Rana pipiens as well as in X. laevis (Cheek et al. 1999; Crump et al. 2002). Using germinally transgenic tadpoles, we have shown that acetochlor amplifies the transcriptional response of the TH/bZIP promoter-driven eGFP reporter gene in the head region of pretreated tadpoles. This effect was observed with a weak, physiologic concentration of T3 (10−10 M), underlining the sensitivity of this in vivo method to assess actions of chemicals interfering with low physiologic amounts of T3. We also used somatic gene transfer to test the effects of acetochlor alone. However, we saw no effects on transcriptional response from TH/bZIP-luc (data not shown). Moreover, other authors using Xenopus (Crump et al. 2002) also saw no effect of 10−8 M acetochlor in the absence of T3 on endogenous TR-β and TH/bZIP expression in the tail. Similarly, using a Rana model, Cheek et al. (1999) showed that 7 days of exposure to acetochlor alone does not accelerate metamorphosis, even after a pretreatment with 10−9 M T3 during 3 days. These multiple observations indicate that acetochlor has no effect in the absence of T3. Moreover, concentrations of acetochlor in the environment are in the range of 10−9 M to 10−8 M, with the effect of acetochlor in presence of T3 occurring at 10−8 M (Cheek et al. 1999). Given that acetochlor persists in water, and that our results show that it can physiologically modify TH effects at environmentally relevant concentrations, our results bolster the concept that acetochlor contamination is a matter of acute environmental concern.
TH regulates a wide range of biologic processes during development and adult life. The fact that considerable numbers of compounds have the potential to interfere with different aspects of thyroid system function and TH action raises an urgent need for the development of an in vivo assay for detection of thyroid-axis–disrupting molecules. There is a long-standing debate in the field of endocrine disruption as to whether it is more important to reveal potential disrupting effects or to address the mechanisms of action underlying disruption. In the present study, we have opted to refine a test that will allow the detection of a wide range of disrupting chemicals rather than reveal mechanisms of action. Indeed, using the transcriptional response to natural ligand as an end point in an in vivo (vs. in vitro) context allows one to encompass a large range of potential interferences. For instance, if a chemical interferes with TH degradation, this should be picked up by a modification of the response to exogenous ligand. For example, the use of sodium perchlorate efficiently blocks metamorphosis by interfering with TH production. We have performed RT-PCR on such perchlorate-treated tadpoles and found decreased expression of TR-β (data not shown). This would be one example of a reduction of TH availability that is also played out at the level of a TRE-containing gene (the promoter of TR-β gene contains several functional TREs; Urnov and Wolffe 2001). Similarly, if a potential disruptor modulates receptor or comodulator availability, this, too, will be detected. Moreover, even if endogenous levels of TH are low in tadpoles at the stage used in these experiments, significant effects of disruptors on secretion and distribution of endogenous hormone will be detected. Thus, although the assay described does not address the eventual mechanisms of disruption, it will allow a broad spectrum of effects to be discerned.
A final point is that our model is both adaptable and flexible. The possibility exists of following two different hormone response systems simultaneously in the same tadpoles. This can be achieved by either using two separate plasmid constructs for transgenesis or using bicistronic plasmids (Fu et al. 2002). Furthermore, transgenesis in X. laevis takes advantage of the fact that the amphibian endocrine system has high similarity to that of other vertebrates and therefore offers the possibility of generalizing the approach to screen for other hormonal pollutants. This test is thus predictive for eventual hazards to both wildlife and human health.
In conclusion, we have developed a sensitive and rapid in vivo method to assess thyroid agonist activity, an approach that can potentially be combined with and applied to other hormonal axes.
We thank G. Benisti, J.-P. Chaumeil, and E. LeGoff for excellent animal care and technical support.
This work was supported by grants from the Centre National de la Recherche Scientifique, the Muséum National d’Histoire Naturelle, the Agence Nationale de la Valorisation, and the European Union (EU contract 506319) to CASCADE.
Figure 1 A significant transcriptional response induced by a 24-hr pretreatment, or priming, pulse of 10−12 M T3 within 48 hr in tadpoles later exposed to 10−8 M T3. NS, not significant. (A) A linear schema indicating the timing of the pretreatment/rinse/exposure protocol. Tadpoles were pretreated with or without T3 24 hr, rinsed in water, and fed during 24 hr before injection with 200 ng TH/bZIP-luc construct in the caudal muscle. (B) Measured TH/bZIP-luc transcription in pretreated tadpoles exposed or not exposed to T3 (10−8 M) for 48 hr. Values shown are mean ± SE (n = 12/group). In each case, the experiment was repeated three times, providing similar results.
**p < 0.01.
Figure 2 Induction of TR-β expression by a weak T3 pulse. To evaluate the effect of the pretreatment protocol on TR-β expression, tadpoles were exposed for 6 hr to 10−12 M T3. Total RNA was extracted from caudal muscles and used for RT-PCR analysis of TR-β expression. Rpl8 was used as the internal control. (A) Typical scan obtained after 22 cycles of PCR amplification. (B) Same results quantified by Phosphoimager scanning (Molecular Dynamics, Sunnyvale, CA, USA). Values shown are mean ± SE of five independent experiments expressed as multiples of induction, where 1 is equal to expression in the absence of T3 (untreated tadpole; Rpl8) as the control level. For each sample, densitometry readings were normalized against the value for Rpl8.
*p < 0.05.
Figure 3 Action of TH agonists assessed using the somatic gene transfer method and pretreatment. Tadpoles were pretreated 24 hr with 10−12 M T3 and then rinsed and fed during 24 hr before injection of 200 ng TH/bZIP-luc construct in the caudal skeletal muscle. TH/bZIP-luc transcription was measured in injected tadpoles exposed to 5 × 10−8 M T3 or 5 × 10−8 M TRIAC for 48 hr. Values shown are mean ± SE (n = 12/group). Each experiment was repeated three times, providing similar results.
*p < 0.05. ** < 0.01.
Figure 4 Dose dependency of TH effects on transcriptional responses. NS, not significant. (A) Transcriptional response in tadpoles pretreated for 24 hr with 10−12 M T3 and then rinsed and fed during 24 hr before injection of 200 ng TH/bZIP-luc construct in the caudal skeletal muscle. The TH/bZIP-luc transcription was measured in injected tadpoles exposed or not exposed to 5 × 10−8 M, 5 × 10−9 M, or 5 × 10−10 M T3 for 48 hr. (B) Transcriptional response in tadpoles injected with 500 ng TH/bZIP-luc construct in the brain, pretreated 24 hr with 10−12 M T3, and then rinsed and fed during 24 hr before exposure to 10−7 M, 10−8 M, or 10−10 M T3. TH/bZIP-luc transcription was measured after 48 hr. Values shown are mean ± SE (n = 12/group). Each experiment was repeated three times, providing similar results.
*p < 0.05. **p < 0.01. #p < 0.001.
Figure 5 (A–F) Transcription responses of the TRE-containing transgene to endogenous TH induced by natural metamorphosis in germinally transgenic F0 embryos at NF stages 51 (A–C) and 62 (D–F). (A, D) Brain. (B–F) Limb buds. TH/bZIP-eGFP is expressed first in the brain (A, yellow arrowhead) and then in other tissues, and persists throughout larval development (A, D). No fluorescence above background is present in limb buds at NF stage 51 (C, yellow arrow). The signal increases throughout larval development until metamorphosis is reached (NF stage 62), when it increases strongly (F). Bars = 1.6 mm (A); 0.5 mm (B, C); 1.8 mm (D); 1.7 mm (E, F).
(G–L) The pretreatment protocol significantly reduced time for response to T3 and to TH analogues in TH/bZIP-eGFP transgenic F0 tadpoles (G, H, head; I–L, hindlimb). Tadpoles were pretreated for 24 hr with 10−12 M T3 at NF stages 51–52, and then rinsed and fed during 24 hr before being exposed to 10−8 M T3 (J) or to 5 × 10−8 M TRIAC (L). Fluorescence in the CNS (H) and in hindlimb buds (J, L) was observed after 2 days of treatment. Yellow arrows indicate limb buds; arrowhead indicates the brain area. White arrows indicate crystallin-RFP expression in the eye. Bars = 2 mm (G, H); 0.8 mm (I–L).
Figure 6 The TH enhancing activity of the pesticide acetochlor revealed in the brains of germinally transgenic premetamorphic tadpoles. Abbreviations: a, anterior; NS, not significant; p, posterior. (A) NF stage 50–52 germinally transgenic tadpoles bearing a TH/bZIP-eGFP transgene were pretreated for 24 hr with 10−12 M T3 and then rinsed and fed for 24 hr.Fluorescence was measured in tadpole brains after 48 hr exposure to 10−10 M T3 or to 10−10 M T3 plus 10−8 M acetochlor. Values shown are mean ± SE of three experiments expressed as multiples of induction, where 1 = control expression in the absence of T3. Data were normalized and analyzed by Student’s t-test. (B) Representative photographs of strongly fluorescent tadpoles brain from each group (n = 15 tadpoles per group). Bars = 0.4 mm.
*p < 0.05. #p < 0.001.
==== Refs
References
Cheek AO Ide CF Bollinger JE Rider CV McLachlan JA 1999 Alteration of leopard frog (Rana pipiens ) metamorphosis by the herbivide acetochlor Arch Environ Contam Toxicol 37 70 77 10341044
Ciana P Di Luccio G Belcredito S Pollio G Vegeto E Tatangelo L 2001 Engineering of a mouse for the in vivo profiling of estrogen receptor activity Mol Endocrinol 15 1104 1113 11435611
Coen L du Pasquier D Le Mevel S Brown S Tata J Mazabraud A 2001 Xenopus Bcl-X(L) selectively protects Rohon-Beard neurons from metamorphic degeneration Proc Natl Acad Sci USA 98 7869 7874 11427732
Crump D Werry K Veldhoen N Van Aggelen G Helbing CC 2002 Exposure to the herbicide acetochlor alters thyroid hormone-dependent gene expression and metamorphosis in Xenopus laevis Environ Health Perspect 110 1199 1205 12460798
de Luze A Sachs L Demeneix B 1993 Thyroid hormone-dependent transcriptional regulation of exogenous genes transferred into Xenopus tadpole muscle in vivo Proc Natl Acad Sci USA 90 7322 7326 8346251
Denver RJ Pavgi S Shi Y-B 1997 Thyroid hormone-dependent gene expression program for Xenopus neural development J Biol Chem 272 8179 8188 9079635
Dodd MHI Dodd JM 1976. The biology of metamorphosis. In: Physiology of Amphibia (Lofts B, ed). New York:Academic Press, 467–599.
Fu L Buchholz D Shi Y-B 2002 Novel double promoter approach for identification of transgenic animals: a tool for in vivo analysis of gene function and development of gene-based therapies Mol Reprod Dev 62 470 476 12112579
Furlow JD Brown DD 1999 In vitro and in vivo analysis of the regulation of a transcription factor gene by thyroid hormone during Xenopus laevis metamorphosis Mol Endocrinol 13 2076 2089 10598583
GenBank 2004. Searching GenBank. Available: http://www.ncbi.nlm.nih.gov/Genbank/GenbankSearch.html [accessed 21 September 2005].
Havis E Sachs LM Demeneix B 2003 Metamorphic T3-response genes have specific co-regulator requirements EMBO Rep 4 883 888 12947412
Helbing CC Werry K Crump D Domanski D Veldhoen N Bailey CM 2003 Expression profiles of novel thyroid hormone-responsive genes and proteins in the tail of Xenopus laevis tadpoles undergoing precocious metamorphosis Mol Endocrinol 17 1395 1409 12690095
Huang H Brown DD 2000 Overexpression of Xenopus laevis growth hormone stimulates growth of tadpoles and frogs Proc Natl Acad Sci USA 97 190 194 10618393
Kavlock RJ Ankley GT 1996 A perspective on the risk assessment process for endocrine-disruptive effects on wildlife and human health Risk Anal 16 6 731 739 8972105
Kirillov A Kistler B Mostoslavsky R Cedar H Wirth T Bergman Y 1996 A role for nuclear NF-kappaB in B-cell-specific demethylation of the Igkappa locus Nat Genet 13 435 441 8696338
Kolpin KW Nations BK Thurman EM Goolsby DA 1996 Acetochlor in the hydrologic system in the midwestern United States, 1994 Environ Sci Technol 30 1459 1464
Kroll KL Amaya E 1996 Transgenic Xenopus embryos from sperm nuclear transplantations reveal FGF signaling requirements during gastrulation Development 122 3173 3183 8898230
Leloup J Buscaglia M 1977 La triiodothyronine: hormone de la métamorphose des amphibiens [in French] C R Acad Sci 284 2261 2263
Mangelsdorf DJ Thummel C Beato M Herrlich P Schutz G Umesono K 1995 The nuclear receptor superfamily: the second decade Cell 83 835 839 8521507
Marsh-Armstrong N Huang H Berry DL Brown DD 1999 Germ-line transmission of transgenes in Xenopus laevis Proc Natl Acad Sci USA 96 14389 14393 10588715
Nakajima K Yaoita Y 2003 Dual mechanisms governing muscle cell death in tadpole tail during amphibian metamorphosis Dev Dyn 227 246 255 12761852
Nieuwkoop PD Faber J 1956. Normal Table of Xenopus laevis (Daudin): A Systematical and Chronological Survey of the Development from the Fertilized Egg till the End of Metamorphosis. Amsterdam:Elsevier North Holland.
OECD 2004. Detailed Review Paper on Amphibian Metamorphosis Assay for the Detection of Thyroid Active Substances. Series on Testing and Assessment No. 46. Paris: Organisation for Economic Co-Operation and Development. Available: http://www.oecd.org/document/30/0,2340,en_2649_34377_1916638_1_1_1_1,00.html [accessed 11 October 2005].
Oofusa K Tooi O Kashiwagi A Kashiwagi K Kondo Y Watanabe Y 2001 Expression of thyroid hormone receptor β A gene assayed by transgenic Xenopus laevis carrying its promoter sequences Mol Cell Endocrinol 181 97 110 11476944
Ouatas T Le Mevel S Demeneix BA de Luze A 1998 T3-dependent physiological regulation of transcription in the Xenopus tadpole brain studied by polyethylenimine based in vivo gene transfer Int J Dev Biol 42 1159 1164 9879714
Rasband WS 1997. ImageJ. Bethesda, MD:National Institutes of Health. Available: http://rsb.info.nih.gov/ij/ [accessed 21 September 2005].
Sachs LM 2004 Corepressor requirement and thyroid hormone receptor function during Xenopus development Vitam Horm 68 209 230 15193456
Sciences et Médecine des Animaux de laboratoire à l'ENVL 2005. Historique de la législation française sur la protection animale. Available: http://www.vet-lyon.fr/ens/expa/guidelines/regl_protectionanimale.html [accessed 21 September 2005].
Stief A Winter DM Stratling WH Sippel AE 1989 A nuclear DNA attachment element mediates elevated and position-independent gene activity Nature 341 343 345 2797152
Trudeau VL Turque N Le Mével S Alliot C Gallant N Coen L 2004 Assessment of estrogenic endocrine disrupting chemical actions in the brain using in vivo somatic gene transfer Environ Health Perspect 113 329 334 15743723
Ulisse S Esslemont G Baker BS Krishna V Chatterjee K Tata JR 1996 Dominant-negative mutant thyroid hormone receptors prevent transcription from Xenopus thyroid hormone beta gene promoter in response to thyroid hormone in Xenopus tadpoles in vivo Proc Natl Acad Sci USA 93 1205 1209 8577741
Urnov FD Wolffe AP 2001 An array of positioned nucleosomes potentiates thyroid hormone receptor action in vivo J Biol Chem 276 19753 19761 11274156
Van Middlesworth L 1974. Metabolism and excretion of thyroid hormones. In: Handbook of Physiology, Section 7: Endocrinology, Vol III. Thyroid (Greep RO, Ashwood, EB, eds). Washington, DC:American Physiological Society, 215–231.
Wilson AGE Thake DC Heydens WE Brewster DW Hotz KJ 1996 Mode of action of thyroid tumor formation in the male Long-Evans rat administered high doses of alachlor Fundam Appl Toxicol 33 16 23 8812207
Yen PM 2001 Physiological and molecular basis of thyroid hormone action Physiol Rev 81 1097 1142 11427693
|
16263516
|
PMC1310923
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 11; 113(11):1588-1593
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7992
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7664ehp0113-00159416263517ResearchSystemic Effects of Arctic Pollutants in Beluga Whales Indicated by CYP1A1 Expression Wilson Joanna Y. 1Cooke Suzy R. 12Moore Michael J. 1Martineau Daniel 3Mikaelian Igor 3*Metner Donald A. 4Lockhart W. Lyle 4Stegeman John J. 11 Department of Biology, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA2 Department of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA3 Department of Pathology and Veterinary Microbiology, University of Montréal, Montréal, Québec, Canada4 Freshwater Institute, Winnipeg, Manitoba, CanadaAddress correspondence to J.J. Stegeman, Woods Hole Oceanographic Institution, Mail Stop #32, 45 Water St., Woods Hole, MA 02540 USA. Telephone: (508) 289-2320. Fax: (508) 457-2134. E-mail:
[email protected]* Current address: Hoffmann-Laroche, Nutley, New Jersey, USA.
The authors declare they have no competing financial interests.
11 2005 14 7 2005 113 11 1594 1599 15 10 2004 14 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Cytochrome P450 1A1 (CYP1A1) is induced by exposure to polycyclic aromatic hydrocarbons (PAHs) and planar halogenated aromatic hydrocarbons (PHAHs) such as non-ortho polychlorinated biphenyls (PCBs). In this study, we examined CYP1A1 protein expression immunohistochemically in multiple organs of beluga whales from two locations in the Arctic and from the St. Lawrence estuary. These beluga populations have some of the lowest (Arctic sites) and highest (St. Lawrence estuary) concentrations of PCBs in blubber of all cetaceans. Samples from these populations might be expected to have different contaminant-induced responses, reflecting their different exposure histories. The pattern and extent of CYP1A1 staining in whales from all three locations were similar to those seen in animal models in which CYP1A has been highly induced, indicating a high-level expression in these whales. CYP1A1 induction has been related to toxic effects of PHAHs or PAHs in some species. In St. Lawrence beluga, the high level of CYP1A1 expression coupled with high levels of contaminants (including CYP1A1 substrates, e.g., PAH procarcinogens potentially activated by CYP1A1) indicates that CYP1A1 could be involved in the development of neoplastic lesions seen in the St. Lawrence beluga population. The systemic high-level expression of CYP1A1 in Arctic beluga suggests that effects of PAHs or PHAHs may be expected in Arctic populations, as well. The high-level expression of CYP1A1 in the Arctic beluga suggests that this species is highly sensitive to CYP1A1 induction by aryl hydrocarbon receptor agonists.
Arcticbeluga whaleCYP1A1cytochrome P450 1A1immunohistochemistrySt. Lawrence estuary
==== Body
Chemical contaminants are ubiquitous in the world’s oceans. Specific health effects and overt disease in fish have been linked to high concentrations of contaminants in some coastal regions of North America (e.g., Murchelano and Wolke 1991; Myers et al. 1990). Atmospheric processes distribute many of these same contaminants to the polar regions (MacDonald et al. 2000), where they accumulate in the fatty tissue of top predators. In the Arctic, prominent contaminants fall into five broad categories: chlorinated industrial compounds including polychlorinated biphenyls (PCBs), organic pesticides, polycyclic aromatic hydrocarbons (PAHs), metals, and radionuclides (MacDonald et al. 2000). Persistent contaminants that accumulate in the lipid-rich blubber of whales include PCBs, DDT (dichlorodiphenyl-trichloroethane), and other chlorinated pesticides. Concentrations of these contaminants in Arctic ecosystems have declined somewhat over the past 20 years, yet they persist in marine mammal species, including the beluga (Muir et al. 1999). Although cetaceans in the Arctic have contaminant concentrations that are at least 10 times lower than the most highly contaminated cetaceans from other locations (Norstrom and Muir 1994), Arctic animals may yet be at risk for adverse health effects.
We obtained multiple internal organ tissue samples from beluga whales that stranded dead along the St. Lawrence estuary and during subsistence hunts in the Mackenzie Delta and Hudson Bay. The St. Lawrence beluga population was designated endangered in 1980 (Seargent 1986), a status continued under the Species at Risk Act in Canada (2002); thus, stranded animals represent the only means to obtain tissues for scientific study from this location. The Mackenzie Delta and Hudson Bay sites harbor two separate Arctic populations of beluga, both of which have levels of PCB and PAH contaminants an order of magnitude lower than those at the St. Lawrence estuary site. These populations, including the two Arctic populations, are geographically separated and represent separate stocks of this species. Beluga are located in offshore waters during winter months and in coastal regions during summer. They have a complex diet including species of fish and crustaceans, which is their primary route of exposure to environmental contaminants.
Among marine mammals, odontocetes (toothed whales), including beluga, may be at the greatest risk of contaminant effects because these animals are top predators that accumulate contaminants to a higher degree than do mysticetes (baleen whales) (O’Shea and Brownell 1994). Concentrations of PCBs as high as 300 μg/g (lipid weight basis) have been recorded in odontocete blubber (Ross et al. 2000). Planar halogenated aromatic hydrocarbons (PHAHs), the dioxin-like contaminants that include non-ortho and mono-ortho substituted PCBs, are of special concern because even at low doses they can affect development of the immune, nervous, and reproductive systems in animal models (Birnbaum and Tuomisto 2000).
Assessing health effects of contaminants in cetaceans is difficult because experimental exposures are precluded and fresh tissues are rarely available. Molecular changes associated with exposure to selected compounds can suggest whether systemic effects are likely. Cytochrome P450 1A (CYP1A) induction is elicited by PAHs and PHAHs via binding to the aryl hydrocarbon receptor (AHR) (Whitlock 1999). The mammalian CYP1A gene subfamily contains two members: CYP1A1 and CYP1A2. Typically in mammals, CYP1A2 expression is limited to liver, whereas CYP1A1 is more strongly inducible in extrahepatic organs as well as liver. CYP1A1 induction has been correlated to higher-order toxic effects, including thymic atrophy, weight loss, and lethal toxicity induced by PCB, polychlorinated dibenzodioxin, and polychlorinated dibenzofuran exposure in rodents (Safe 1987). Thus, systemic CYP1A1 expression can indicate a risk for toxic effects.
In an earlier study, CYP1A1 levels in liver of Arctic beluga were strongly correlated with the concentration of non-ortho and mono-ortho PCBs in blubber—compounds that are known to induce CYP1A1 through the AHR (White et al. 1994)—indicating that CYP1A1 is a good biomarker of exposure in beluga. In this study, we examined cellular location and relative levels of CYP1A1 expression in multiple organs of beluga whale from the St. Lawrence estuary and from two locations in the Arctic (Beaufort Sea and western Hudson Bay populations). Samples from these populations might be expected to have different contaminant-induced responses, reflecting their different exposure histories. Beluga from the St. Lawrence estuary population have high concentrations of PCBs, chlorinated pesticides (Muir et al. 1996), and metals (mercury, lead, and selenium) (Wagemann et al. 1990) in blubber. These animals also show highly elevated prevalences of overt pathologies linked to toxicants (Martineau et al. 1988); thus, we addressed the question of whether CYP1A1 might be expressed in multiple organs of beluga from several regions, because this would indicate whether biochemical effects in these animals might be occurring systemically and could indicate the sensitivity of this species to PHAH toxicity.
Materials and Methods
We obtained tissue samples from multiple organs of beluga whale from three separate populations. The public and officials of various government agencies reported beluga whales found dead, stranded on the St. Lawrence estuary shoreline. The carcasses were immediately transported by truck to the postmortem room of the Faculté de Médecine Vétérinaire of Université de Montréal, 500 km to the southwest, where pathologists assisted by veterinary students examined them upon arrival. Fourteen beluga were included in this study from the St. Lawrence estuary. Samples were additionally collected during subsistence hunts of the Mackenzie Delta (n = 15) and Hudson Bay (n = 12) beluga populations. The ages of the animals from each population included in this study are shown in Table 1. Standard body measurements, total body, organ, and blubber weights, and tooth counts were determined when possible. Organs sampled included adrenal gland, brain, bladder, colon, gonad (ovary and testis), heart, kidney, liver, lung, skin, and thyroid. The time to necropsy was < 12 hr postmortem for Arctic animals. For the St. Lawrence beluga, carcasses were recovered < 3 days after death judging by the extent of postmortem changes and accounting for the cold water temperature, which retards autolysis. Marked autolysis was seen in the livers of some of the St. Lawrence beluga. Age was determined in the Hudson Bay and St. Lawrence beluga by sectioning teeth longitudinally and counting dentine growth layers on sections using a binocular microscope, using the standard of two growth layers per year (Brodie 1982). We calculated age for the Mackenzie Delta beluga using length measurements and established age–length relationships (Doidge 1990).
The tissues were removed at necropsy, and small samples were fixed in 10% neutral buffered formalin, embedded in paraffin, and sectioned at 5 μm. We assessed CYP1A1 expression by immunohistochemical analysis with monoclonal antibody (Mab) 1-12-3, as previously described (Smolowitz et al. 1991). Mab 1-12-3 recognizes an epitope that in mammals is specific to CYP1A1 but not to CYP1A2 (Drahushuk et al. 1998), and use in Western blot shows a single band in beluga whale liver microsomes (White et al. 1994). We calculated a semiquantitative index (0–15) of CYP1A1 expression determined by immunohistochemistry by multiplying the intensity (0–5) and occurrence (0–3) of label for each cell type in a given organ. A linear relationship between this index and CYP1A protein content measured by immunoblot has been previously shown for expression in liver and for CYP1A1 induced in cells in culture (Hahn et al. 1993; Woodin et al. 1997). Serial sections were stained with the nonspecific antibody UPC-10 (Sigma-Aldrich Co., St. Louis, MO, USA) to control for any nonspecific staining. Replicate slides were stained with hematoxylin and eosin. Although tissue fixation times were not controlled between the samples, they were processed into paraffin blocks within 2 months for all organs except the adrenal and thyroid gland from the St. Lawrence beluga. Epitope recognition with this antibody was equivalent among scup liver samples held in formalin between 2 weeks and 5 months (Smolowitz R, Stegeman J, unpublished observations), a period encompassing the times that beluga tissues were in formalin.
Differences in CYP1A1 expression in the three populations of beluga were determined using analysis of variance and either the Tukey-Kramer or Sheffe’s model when there were unequal or equal numbers of samples, respectively. Samples were always divided between site and sex for statistical analyses.
Results and Discussion
Analysis of liver and extrahepatic organs showed patterns of CYP1A1 expression consistent with a strong induction of CYP1A1, based on what has been seen in mammalian and non-mammalian vertebrate models [Tables 1 and 2, and supplementary material (http://ehp.niehs.nih.gov/docs/2005/7664/supplement.pdf)]. CYP1A1 expression was seen in vascular endothelial cells in multiple organs of all individuals included in this study, including all lung (n = 33) and skin samples (n = 13) and nearly all bladder (16 of 18), testes (13 of 17), and adrenal (12 of 13) samples (Table 3). The expression of endothelial CYP1A1 in multiple organs examined from each individual whale indicates a systemic effect of contaminants in the Arctic beluga whale. The levels and patterns of CYP1A1 expression in selected organs are considered below.
CYP1A1 expression in liver.
CYP1A1 was highly expressed in hepatic parenchyma of Arctic beluga liver (Table 2). Typically in mammalian liver, CYP1A1 expression is localized to periportal parenchyma in untreated or slightly induced animals, and panlobular expression is seen only in animals in which CYP1A1 is strongly induced (Oinonen and Lindros 1998). High-level CYP1A1 expression that is panlobular, as in Figure 1, is fully consistent with CYP1A1 having been strongly induced in liver of Arctic beluga. Surprisingly, the levels of CYP1A1 expression in liver from the highly contaminated St. Lawrence beluga were significantly lower than those in the Arctic animals (Table 2), despite the greater exposure to inducing compounds: liver PCB concentrations in male beluga average 1,445 ng/g and 132 ng/g in the St. Lawrence and Arctic, respectively (Metcalfe et al. 1999). CYP1A1 expression may be suppressed in St. Lawrence beluga liver, potentially as a result of high-level contaminant exposure. CYP1A is suppressed in liver but not in other organs of fish experimentally exposed to high doses of non-ortho PCB congeners (e.g., PCB-126) (Schlezinger and Stegeman 2001). Unlike PCBs, high levels of PAHs are not known to suppress CYP1A1 expression.
The time from death to necropsy was longer for the St. Lawrence beluga than for the Arctic beluga, and histologic analyses do show autolysis in these liver samples. Although CYP1A1 expression in other organs was not significantly lower in the St. Lawrence beluga than in Arctic animals, the liver degrades at a faster rate. It is likely that differences in hepatic CYP1A1 expression simply reflect degradation and the difference between time of death and fixation for samples collected from subsistence hunts and strandings. Unfortunately, it is impossible to collect from these sites under identical conditions. There are no subsistence hunts of the St. Lawrence beluga population, and locating stranded animals in the Arctic is not feasible. Yet the pattern and levels of CYP1A1 expression in Arctic animals indicate a substantial induction in those beluga.
CYP1A1 expression in lung.
As indicated above, endothelial CYP1A1 levels (Table 3, Figure 2A,C,D) were high in lung. CYP1A1 expression was seen also in chondrocytes and bronchiolar epithelium but was not seen in type 1 or type 2 pneumocytes [see supplementary material (http://ehp.niehs.nih.gov/docs/2005/7664/supplement.pdf)]. The predominant environmental exposure route for CYP1A1 inducers is dietary, but a recent study in mice suggests that PCB uptake can be greater via inhalation than from diet (Casey et al. 1999). Thus, consideration of nondietary exposures such as inhalation may be warranted in regions where PAH and/or PCB exposure levels are likely to be high. Hormonal, histo-pathologic, and behavioral changes were seen in mice exposed to 0.9 μg/m3 Aroclor 1242 in the air (Casey et al. 1999), a concentration that is approximately 10,000-fold higher than Arctic atmospheric PCB concentrations (MacDonald et al. 2000). Atmospheric sources of PCBs in the Arctic could result in an estimated lung exposure of 1.3–67 ng/day in Arctic beluga (Table 4). Likewise, an inhaled PAH exposure could be expected to range from 5.6 to 363 ng/day in Arctic beluga (Table 4), although this exposure would be dominated by lower-molecular-weight PAHs such as fluorene and phenanthrene (MacDonald et al. 2000), which do not typically induce CYP1A (Bols et al. 1999). Given that, and considering that type 1 pneumocytes (the pulmonary cell type primarily involved in gas exchange in the lung) did not express CYP1A1 in beluga, it is more plausible to conclude that CYP1A1 induction in lung was solely the result of dietary exposure and that the contribution of inhaled contaminants was marginal.
CYP1A1 expression in bladder.
In bladder, CYP1A1 was highly expressed in both endothelium (Table 3, Figure 2B) and transitional epithelium forming the bladder mucosa (Table 2, Figure 3). CYP1A1 in transitional epithelium was most highly expressed in umbrella cells, the cells in direct contact with urine. A transitional cell carcinoma of the bladder has been found in a beluga from the highly contaminated St. Lawrence estuary (Martineau et al. 1985). In humans, CYP1A is involved in the activation of a variety of potential bladder carcinogens (Gonzalez and Gelboin 1994), is expressed in primary transitional cell tumors of the urinary bladder, and has been correlated with tumor grade (G1–G3) (Murray et al. 1995). CYP1A1 could be involved in the development of bladder tumor in the St. Lawrence beluga population.
The expression of CYP1A1 in bladder was as high in the Arctic beluga as in the St. Lawrence beluga. Considering that, in the transitional epithelium, the most highly induced cells were in direct contact with urine (Figure 3), the induction of CYP1A1 in bladder presumably was caused by contaminants excreted into the urine. Potential CYP1A1 inducers in urine include both PCBs and PAHs. PAHs are eliminated more rapidly and therefore accumulate to a much lower degree than do PCBs, yet higher levels of exposure would still result in significant CYP1A1 induction. In the St. Lawrence estuary, PAH exposures are higher and likely more important for CYP1A1 induction than in the Arctic: Sediment-associated PAHs are 500–4,500 ng/g (Martel et al. 1986) and 400–980 ng/g (MacDonald et al. 2000) in the St. Lawrence and Arctic, respectively. The relative contribution of PCBs and PAHs to urinary contaminants is unknown in the St. Lawrence beluga. In the Arctic, where contaminants are atmospherically derived, atmospheric PAHs are dominated by those that do not induce CYP1A1 (MacDonald et al. 2000); thus, PAH contributions to CYP1A1 induction may be minimal in Arctic beluga whales.
PCBs that are AHR agonists are highly correlated to CYP1A1 levels in liver of Arctic beluga (White et al. 1994), indicating that induction of CYP1A1 in bladder of Arctic beluga is likely to be related to PCB exposure. In mice exposed to PCBs in the diet, only 5% of oral dose appeared in urine, mainly as conjugates (Wehler et al. 1989). We calculated a total body burden based on PCB concentrations in the blubber, blubber weight, and total body weight (Table 5). Given this total body dose, we can estimate an upper limit on the maximal oral dose in Arctic beluga. The urinary PCB concentrations would be very small in the Arctic beluga, presumably much less than 0.05–0.1 mg/kg (5% of the upper limit on the maximal oral dose), because we would expect most urinary PCBs to be conjugated and not able to induce CYP1A1. These results suggest that the doses required for CYP1A1 induction in bladder are low and indicate that beluga are very sensitive in their responses to PHAH contaminants. Determining urinary concentrations of PAHs and PCBs would be important to confirm that such compounds are present in this organ and what concentrations are responsible for such high-level CYP1A1 expression.
CYP1A1 expression in testis.
Moderate levels of CYP1A1 expression were seen in the spermatogenic series in the testis (Table 1). Although this may have included some Sertoli cells, CYP1A1 expression appeared primarily in spermatogonia and spermatocytes. In studies with several other mammals, testicular microsomal preparations have shown very low or no CYP1A activity (Machala et al. 1998; Revel et al. 2001; Roman et al. 1998). Testicular CYP1A activity was not induced in rats (Roman et al. 1998), bulls (Machala et al. 1998), or mice (Revel et al. 2001) exposed to a variety of inducers, although the AHR and dimerization partner ARNT (aryl hydrocarbon receptor nuclear translocator), which are required for CYP1A induction, are present in testis (Roman et al. 1998). Immunohistochemical analyses of mouse testes showed CYP1A1 in interstitial cells only, and this was reportedly not induced by benzo[a]pyrene (Revel et al. 2001). CYP1A1 expression in the spermatogenic series is an unusual finding in a mammalian species. Considering that CYP1A1 is involved in the activation of procarcinogens and generation of reactive oxygen species, high-level CYP1A1 expression in the spermatogenic series could be significant for sperm function and gamete development.
Implications of CYP1A1 induction in beluga.
The high levels of CYP1A1 expression in the beluga whale from both the Arctic and the St. Lawrence estuary are consistent with high sensitivity of this species to CYP1A inducers. Interestingly, except for the liver, the level of expression was not markedly different between animals from the St. Lawrence estuary and those from the Arctic, despite significant differences in contaminant exposure and apparent tumor prevalence. No tumors were found in 50 Arctic beluga examined, whereas 21 tumors were found in 100 St. Lawrence beluga, resulting in an annual cancer rate of 163 per 100,000 animals calculated for the St. Lawrence estuary beluga (Martineau et al. 2002). However, the necropsies on the Arctic beluga were not as detailed as necropsies on the St. Lawrence animals, and the Arctic animals examined were much younger. Detailed necropsies will need to be performed on older Arctic animals to determine the prevalence of tumors in Arctic populations of beluga.
Liver CYP1A1 expression was previously shown to be highly correlated to mono-ortho and non-ortho PCBs, ligands for the AHR, in the blubber of Arctic beluga whale (White et al. 1994). The high levels and cellular patterns of CYP1A1 expression are similar to those seen in animal models exposed to high levels of inducers; other mammals do not show this broad pattern of induction unless exposed to high concentrations of contaminants. Thus, even lower doses of contaminants, like those seen in the Arctic animals, appear able to highly induce CYP1A1 in beluga. Beluga whales express levels of CYP1A1 in various organs that are similar to or greater than levels of expression in organs of other cetaceans for which such data are available (Wilson 2003), despite having lower levels of potential inducers in their tissues. These data support the idea that beluga are a more sensitive species, at least compared with other cetaceans. The sensitivity of beluga to CYP1A1 inducers is reflected also in the beluga AHR. The beluga AHR has been cloned, expressed, and shown to bind 2,3,7,8-tetrachlorodibenzo-p-dioxin (a prototypical inducer) with a similar binding affinity to that of the C57 strain mouse (Jensen and Hahn 2001). This strain of mouse is highly sensitive to PHAH toxicity (Shen et al. 1991), and the AHR properties suggest that beluga may be similarly sensitive to these contaminants. In the St. Lawrence estuary, only beluga, and not other resident cetacean species, have been found with tumors (De Guise et al. 1994), indicating that beluga may also be particularly susceptible to chemical carcinogenesis. The fact that beluga show a systemic response to CYP1A1 inducers, even at lower doses, indicates that other toxic effects elicited by AHR agonists may be expected, even in populations from the relatively uncontaminated Arctic.
Conclusions
Although Arctic cetaceans have contaminant concentrations that are among the lowest reported, these exposures still could be biologically significant. Despite such low exposures, Arctic beluga have a pattern and extent of CYP1A1 expression that is similar to those seen in animal models that are maximally induced. These data, supported by in vitro studies of the beluga AHR (Jensen and Hahn 2001), and the presence of tumors only in beluga but not other cetaceans resident in the St. Lawrence estuary (De Guise et al. 1994), suggest that beluga are highly sensitive to CYP1A1 inducer substrates. In the St. Lawrence estuary, CYP1A1 could be involved in the development of neoplastic lesions seen in this beluga population (Martineau et al. 1994). Because beluga have a systemic response to PHAH contaminants at low doses, toxic effects may be expected in Arctic populations.
Supplementary Material
Supplemental Material Supplemental Material is available online (http://ehp.niehs.nih.gov/docs/2005/7664/supplement.pdf).
We acknowledge those involved in sample collection, including the Tuktoyuktuk and Inuvik Hunters and Trappers Associations, R. Felix, F. Day, J. Voudrach, and S. Smith.
Funding was provided by the Woods Hole Oceanographic Institution (WHOI) Sea Grant program, National Oceanic and Atmospheric Administration Sea Grant NA86RG0075 R/B-162, a Postgraduate Scholarship B from the Natural Sciences and Engineering Research Council of Canada, and WHOI’s academic programs. WHOI contribution 11085.
Figure 1 CYP1A1 expression in liver from beluga whale. CYP1A1 is labeled pink to dark red; arrows indicate cells with labeling and the identical cell type without labeling. (A) CYP1A1 expression in hepatic parenchyma from an Arctic beluga. (B) Serial section from Arctic beluga shown in (A) stained using the nonspecific antibody UPC-10. (C) CYP1A1 expression in hepatic parenchyma from a St. Lawrence beluga. Magnification, 400×.
Figure 2 CYP1A1 expression in lung and bladder endothelium from beluga whale. CYP1A1 is labeled pink to dark red; arrows indicate cells with labeling and the identical cell type without labeling. (A) CYP1A1 expression in lung endothelium from an Arctic beluga. (B) CYP1A1 expression in bladder endothelium from an arteriole from an Arctic beluga. (C) CYP1A1 expression in lung endothelium from an Arctic beluga. (D) Serial section from Arctic beluga shown in (C) was labeled using the nonspecific antibody UPC-10. Magnification: A, C, D, 400×; B, 200×.
Figure 3 CYP1A1 expression in transitional epithelium from the bladder of Arctic beluga whale. CYP1A1 is labeled pink to dark red; arrows indicate cells with labeling and the identical cell type without labeling. (A) CYP1A1 expression in the transitional epithelium of bladder; labeling is most intense in the umbrella cells. (B) Serial section from beluga shown in (A) labeled using the nonspecific antibody UPC-10. Magnification, 400×.
Table 1 Sample summary.
Site No. Age (years) Published PCB concentrations (μg/g)a
Mackenzie Delta 12 Male > 9b 4.9
3 Female 4–5b —
Hudson Bay 9 Male 4.5–13 2.7
3 Female 7–17.5 —
St. Lawrence 7 Male Neonate to 26 78.9
7 Female 5–31.5 29.6
—, data not available.
a PCB concentrations were determined in blubber and are based on published data (Muir et al. 1996) and not from samples included in this study.
b Based on age–length relationships (Doidge 1990).
Table 2 CYP1A1 expression in epithelia of selected internal organs of beluga whale determined immunohistochemically.a
Site Sex Liver hepatic parenchyma Bladder transitional epithelium Testis spermatogenic seriesb
Mackenzie Delta Male 11.7 ± 2 (12) 8 ± 1.2 (7) 6.3 ± 1.5 (12)
Female 10 ± 2.8 (2) — —
Hudson Bay Male 12.4 ± 2.1 (7) — 3 ± 1.4 (2)
Female 10 ± 3.5 (3) 12 (1) —
St. Lawrence Male 0.3 ± 0.8 (8)* 6 ± 8.5 (6) 4.5 ± 1.2 (4)
Female 3.3 ± 3.2 (6)* 11.25 ± 2.9 (4) —
—, organ not available.
a CYP1A1 expression levels shown are means ± SD (n). CYP1A1 expression is on a scale of 0–15, based on occurrence and intensity of staining, as described in “Materials and Methods.”
b May include some Sertoli cells.
* Mean is significantly different than other sites at p < 0.05.
Table 3 Endothelial CYP1A1 expression in selected internal organs of beluga whale determined immunohistochemically.a
Site Sex Brain Bladder Gonad Kidney Liver Lung
Mackenzie Delta M 1.6 ± 1.4 (12) 8 ± 1.2 (7) 5 ± 2.6 (12) 6.5 ± 2.1* (12) 3.75 ± 3.1 (12) 9.1 ± 2.1 (12)
F 1.5 ± 2.1 (2) — 2.25 ± 3.2 (2) 9 ± 1.4* (2) 2 ± 2.8 (2) 8 ± 2.8 (2)
Hudson Bay M 1.7 ± 2.6 (6) — 1 ± 1.4 (2) 1.8 ± 2.3 (9) 2.8 ± 4.1 (7) 6.7 ± 3.7 (9)
F 0 ± 0 (2) 0 (1) — 0 ± 0 (3) 0.7 ± 1.2 (3) 5.5 ± 1.8 (3)
St. Lawrence M 2.3 ± 3.7 (8) 7.6 ± 6.1 (6) 2.7 ± 4.6 (4) 0 ± 0 (7) 0 ± 0 (8) 7 ± 1.4 (4)
F 3 ± 3.0 (5) 13.1 ± 3.8 (4) 1.25 ± 1.5 (4) 2.2 ± 2.8 (7) 1.6 ± 1.9 (6) 6.3 ± 3.5 (5)
Abbreviations: —, organ not available; F, female; M, male.
a CYP1A1 expression levels shown are means ± SD (n). CYP1A1 expression is on a scale of 0–15, based on occurrence and intensity of staining, as described in “Materials and Methods.”
* Mean is significantly different than other sites at p < 0.05.
Table 4 PCB and PAH dose to lungs of Arctic beluga whale via inhalation.
Parameter Range of value
Tidal volume (L)
Bottlenose dolphina 10
Gray whalea 62
Respiration rate (breaths/min)
Bottlenose dolphinb 2
Weddell seala 8
Volume inspiredc 0.29–7.1 × 105 L/day
Air concentrations (pg/m)d
PCBs 44–94.3
PAHs 194–508
Lung dose (ng/day)e
PCBs 1.3–67
PAHs 5.6–363
a Wartzok (2002).
b Cockcroft and Ross (1990).
c Tidal volume (L) × respiration rate (breaths/min) × 1,440 min/day.
d MacDonald et al. (2000).
e Air concentration (pg/m) × volume inspired (L/day) × 0.001 m3/L.
Table 5 Whole-body dose of PCBs in Arctic beluga whale in this study.
Parameter Range of value
Length 335–447 cm
Body weighta 474–995 kg
Blubber weightb 189.6–398 kg
Blubber contaminantsc 2.7–4.9 μg/g
Whole-body dosed 1.08–1.96 mg/kg
a Weight (kg) = 10−3.84 length (cm)2.58 (Doidge 1990).
b Body weight (kg) × percent body weight as blubber; blubber weight is 40% body weight in beluga whale (Seargent and Brodie 1969).
c Muir et al. (1996).
d Blubber contaminants (mg/kg) × blubber weight (kg) ÷ body weight (kg).
==== Refs
References
Birnbaum LS Tuomisto J 2000 Non-carcinogenic effects of TCDD in animals Food Add Contamin 17 275 288
Bols NC Schirmer K Joyce EM Dixon DG Greenberg BM Whyte JJ 1999 Ability of polycyclic aromatic hydrocarbons to induce 7-ethoxyresorufin O-deethylase activity in a trout liver cell line Ecotoxicol Environ Safe 44 118 128
Brodie PF 1982 The beluga (Delphinapterus leucas ): growth at age based on a captive specimen and a discussion of factors affecting natural mortality estimates Rep Int Whal Comm 32 445 447
Casey AC Berger DF Lombardo JP Hunt A Quimby F 1999 Aroclor 1242 inhalation and ingestion by Sprague-Dawley rats J Toxicol Environ Health A 56 311 342 10094245
Cockcroft VG Ross GJB 1990. The growth and behavioural development of a captive bottlenose dolphin calf. In: The Bottlenose Dolphin (Leatherwood S, Reeves R, eds). San Diego:Academic Press, 461–478.
De Guise S Lagace A Beland P 1994 Tumors in St. Lawrence beluga whales (Delphinapterus leucas ) Vet Pathol 31 444 449 7941233
Doidge DW 1990 Age-length and length-weight comparisons in the beluga, Delphinapterus leucas Can Bull Fish Aquat Sci 224 59 68
Drahushuk AT McGarrigle BP Larsen KE Stegeman JJ Olson JR 1998 Detection of CYP1A1 protein in human liver and induction by TCDD in precision-cut liver slices incubated in dynamic organ culture Carcinogenesis 19 1361 1368 9744530
Gonzalez FJ Gelboin HV 1994 Role of human cytochromes P450 in the metabolic activation of chemical carcinogens and toxins Drug Metab Rev 26 165 183 8082563
Hahn ME Lamb TM Schultz ME Smolowitz RM Stegeman JJ 1993 Cytochrome P4501A induction and inhibition by 3,3′,4,4′-tetrachlorobiphenyl in an Ah receptor-containing fish hepatoma cell line (PLHC-1) Aquat Toxicol 26 185 208
Jensen BA Hahn ME 2001 cDNA cloning and characterization of a high affinity aryl hydrocarbon receptor in a cetacean, the beluga, Delphinapterus leucas Toxicol Sci 64 41 56 11606800
MacDonald RW Barrie LA Bidleman TF Diamond ML Gregor DJ Semkin RG 2000 Contaminants in the Canadian Arctic: 5 years of progress in understanding sources, occurrence and pathways Sci Total Environ 254 93 234 10885446
Machala M Neca J Drabek P Ulrich R Sabatova V Nezveda K 1998 Effects of chronic exposure to PCBs on cytochrome P450 systems and steroidogenesis in liver and testis of bulls (Bos taurus ) Comp Biochem Physiol A Mol Integr Physiol 120 65 70 9773499
Martel L Gagnon MJ Masse R Leclerc A Tremblay L 1986 Polycyclic aromatic hydrocarbons in sediments from the Saguenay Fjord, Canada Bull Environ Contam Toxicol 37 133 140 3719138
Martineau D De Guise S Fournier M Shugart L Girard C Lagace A 1994 Pathology and toxicology of beluga whales from the St. Lawrence estuary, Quebec, Canada: past, present, future Sci Total Environ 154 201 215 7973607
Martineau D Lagace A Beland P Higgins R Armstrong D Shugart LR 1988 Pathology of stranded beluga whales (Delphinapterus leucas ) from the St. Lawrence estuary, Quebec, Canada J Comp Pathol 98 287 311 3134469
Martineau D Lagace A Masse R Morin M Beland P 1985 Transitional cell carcinoma of the urinary bladder in a beluga whale (Delphinapterus leucas ) Can Vet J 26 297 302 17422578
Martineau D Lemberger K Dallaire A Labelle P Lipscomb TP Michel P 2002 Cancer in wildlife, a case study: beluga from the St. Lawrence estuary, Québec, Canada Environ Health Perspect 110 285 292 11882480
Metcalfe C Metcalfe T Ray S Paterson G Koenig B 1999 Polychlorinated biphenyls and organochlorine compounds in brain, liver and muscle of beluga whales (Delphinapterus leucas ) from the Arctic and St. Lawrence estuary Mar Environ Res 47 1 15
Muir D Braune B DeMarch B Norstrom R Wagemann R Lockhart L 1999 Spatial and temporal trends and effects of contaminants in the Canadian Arctic marine ecosystem: a review Sci Total Environ 230 1 3 10466227
Muir DCG Ford CA Rosenberg B Norstrom RJ Simon M Beland P 1996 Persistent organochlorines in beluga whales (Delphinapterus leucas ) from the St. Lawrence River estuary. 1. Concentrations and patterns of specific PCBs, chlorinated pesticides and polychlorinated dibenzo-p -dioxins and dibenzofurans Environ Pollut 93 219 234 15091361
Murchelano RA Wolke RE 1991 Neoplasms and nonneoplastic liver lesions in winter flounder, Pseudopleuronectes americanus , from Boston Harbor, Massachusetts Environ Health Perspect 90 17 26 1646709
Murray GI Taylor VE McKay JA Weaver RJ Ewen SW Melvin WT 1995 Expression of xenobiotic metabolizing enzymes in tumours of the urinary bladder Int J Exp Pathol 76 271 276 7547441
Myers MS Landahl JT Krahn MM Johnson LL McCain BB 1990 Overview of studies on liver carcinogenesis in English sole from Puget Sound; evidence for a xenobiotic chemical etiology. I: Pathology and epizootiology Sci Total Environ 94 33 50 2360038
Norstrom RJ Muir DC 1994 Chlorinated hydrocarbon contaminants in Arctic marine mammals Sci Total Environ 154 107 128 7973601
Oinonen T Lindros KO 1998 Zonation of hepatic cytochrome P-450 expression and regulation Biochem J 329 17 35 9405271
O’Shea TJ Brownell RL Jr 1994 Organochlorine and metal contaminants in baleen whales: a review and evaluation of conservation implications Sci Total Environ 154 179 200 7973606
Revel A Raanani H Younglai E Xu J Han R Savouret JF 2001 Resveratrol, a natural aryl hydrocarbon receptor antagonist, protects sperm from DNA damage and apoptosis caused by benzo(a )pyrene Reprod Toxicol 15 479 486 11780955
Roman BL Pollenz RS Peterson RE 1998 Responsiveness of the adult male rat reproductive tract to 2,3,7,8-tetra-chlorodibenzo-p -dioxin exposure: Ah receptor and ARNT expression, CYP1A1 induction, and Ah receptor down-regulation Toxicol Appl Pharmacol 150 228 239 9653054
Ross PS Ellis GM Ikonomou MG Barrett-Lennard LG Addison RF 2000 High PCB concentrations in free-ranging pacific killer whales, Orcinus orca : effects of age, sex and dietary preference Mar Pollut Bull 40 504 515
Safe S 1987 Determination of 2,3,7,8-TCDD toxic equivalent factors (TEFs): support for the use of the in vitro AHH induction assay Chemosphere 16 791 802
Schlezinger JJ Stegeman JJ 2001 Induction and suppression of cytochrome P450 1A by 3,3′,4,4′,5- pentachlorobiphenyl and its relationship to oxidative stress in the marine fish scup (Stenotomus chrysops ) Aquat Toxicol 52 101 115 11164533
Seargent D 1986 Present status of white whales (Delphinapterus leucas ) in the St. Lawrence Eastern Canada Can Field Nat 113 61 81
Seargent DE Brodie PF 1969 Body size in white whales, Delphinapterus leucas J Fish Res Board Can 26 2561 2580
Shen ES Gutman SI Olson JR 1991 Comparison of 2,3,7,8-tetrachlorodibenzo-p -dioxin-mediated hepatotoxicity in C57BL/6J and DBA/2J mice J Toxicol Environ Health 32 367 381 2016751
Smolowitz RM Hahn ME Stegeman JJ 1991 Immunohistochemical localization of cytochrome P-4501A1 induced by 3,3′,4,4′-tetrachlorobiphenyl and by 2,3,7,8-tetra-chlorodibenzoafuran in liver and extrahepatic tissues of the teleost Stenotomus chrysops (scup) Drug Metab Dispos 19 113 123 1708328
Species at Risk Act 2002. Bill C-5 12 December 2002. Available: http://www.sararegistry.gc.ca [accessed 29 September 2005].
Wagemann R Stewart REA Beland P Desjardins C 1990 Heavy metals and selenium in tissues of beluga whales, Delphinapterus leucas , from the Canadian Arctic and the St. Lawrence estuary Can Bull Fish Aquat Sci 224 191 206
Wartzok D 2002. Breathing. In: The Encylcopedia of Marine Mammals (Perrin WF, Wursig B, Thewissen JGM, eds). San Diego, CA:Academic Press, 164–169.
Wehler EK Bergman A Brandt I Darnerud PO Wachtmeister CA 1989 3,3′,4,4′-Tetrachlorobiphenyl: excretion and tissue retention of hydroxylated metabolites in the mouse Drug Metab Dispos 17 441 448 2571487
White RD Hahn ME Lockhart WL Stegeman JJ 1994 Catalytic and immunochemical characterization of hepatic microsomal cytochromes P450 in beluga whale (Delphinapterus leucas ) Toxicol Appl Pharmacol 126 45 57 8184432
Whitlock JP Jr 1999 Induction of cytochrome P4501A1 Annu Rev Pharmacol Toxicol 39 103 125 10331078
Wilson JY 2003. Cytochrome P450 1A1 and Aromatase (CYP19) in Cetaceans: Enzyme Expression and Relationship to Contaminant Exposure [Ph.D. Thesis]. Woods Hole, MA:Massachusetts Institute of Technology and Woods Hole Oceanographic Institute.
Woodin BR Smolowitz RM Stegeman JJ 1997 Induction of cytochrome P4501A in the intertidal fish Anoplarchus purpurescens by Prudhoe Bay crude oil and environmental induction in fish from Prince William Sound Environ Sci Technol 31 1198 1205
|
16263517
|
PMC1310924
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 14; 113(11):1594-1599
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7664
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8209ehp0113-00160016263518ResearchOrganotins Disrupt the 11β-Hydroxysteroid Dehydrogenase Type 2–Dependent Local Inactivation of Glucocorticoids Atanasov Atanas G. 1Nashev Lyubomir G. 1Tam Steven 2Baker Michael E. 2Odermatt Alex 11 Department of Nephrology and Hypertension, Department of Clinical Research, University of Berne, Berne, Switzerland2 Department of Medicine, University of California, San Diego, La Jolla, CaliforniaAddress correspondence to A. Odermatt, Department of Nephrology and Hypertension, Department of Clinical Research, University of Berne, Freiburgstrasse 15, 3010 Berne, Switzerland. Telephone: 41-31-632-9438. Fax: 41-31-632-9444. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 14 7 2005 113 11 1600 1606 15 4 2005 14 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Organotins, important environmental pollutants widely used in agricultural and industrial applications, accumulate in the food chain and induce imposex in several marine species as well as neurotoxic and immunotoxic effects in higher animals. Reduced birth weight and thymus involution, observed upon exposure to organotins, can also be caused by excessive glucocorticoid levels. We now demonstrate that organotins efficiently inhibit 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), converting active 11β-hydroxyglucocorticoids into inactive 11-ketoglucocorticoids, but not 11β-HSD1, which catalyzes the reverse reaction. Di- and tributyltin as well as di- and triphenyltin inhibited recombinant and endogenous 11β-HSD2 in lysates and intact cells with IC50 values between 500 nM and 3 μM. Dithiothreitol protected 11β-HSD2 from organotin-dependent inhibition, indicating that organotins act by binding to one or more cysteines. Mutational analysis and 3-D structural modeling revealed several important interactions of cysteines in 11β-HSD2. Cys90, Cys228, and Cys264 were essential for enzymatic stability and catalytic activity, suggesting that disruption of such interactions by organotins leads to inhibition of 11β-HSD2. Enhanced glucocorticoid concentrations due to disruption of 11β-HSD2 function may contribute to the observed organotin-dependent toxicity in some glucocorticoid-sensitive tissues such as thymus and placenta.
cortisoldibutyltin11β-hydroxysteroid dehydrogenaseglucocorticoidinhibitionorganotintoxicitytributyltintriphenyltin
==== Body
Organotins belong to the most widely used organometallic compounds, with an estimated annual production of approximately 50,000 tons. Derivatives of dialkyltin compounds such as dibutyltin (DBT), diphenyltin (DPT), and dioctyltin (DOT) are used in industry as stabilizers in polyvinyl chloride (PVC) and as catalysts in various products, whereas trialkyltins, including tributyltin (TBT) and triphenyltin (TPT) are used in agriculture as fungicides and pesticides and as antifouling agents for large ships (Fent 1996). Organotins are ubiquitous environmental pollutants especially relevant for water ecosystems. Accumulation of these lipophilic compounds has been observed in various species of snails, mussels, and fish (Bhosle et al. 2004; Coelho et al. 2002), causing an increased incidence of sterility or imposex (imposition of male sex characters onto the female) (Evans and Nicholson 2000; Fent 2003).
The main sources of organotin intake for humans are seafood contaminated because of the exposure to antifouling agents (Takahashi et al. 1999), and drinking water contaminated because of the leaching from PVC water pipes (Sadiki and Williams 1999). Additional sources are indoor dust, and liquids stored in plastic containers, including various alcoholic beverages (Liu and Jiang 2002). In higher species, including mammals, organotins tend to accumulate in certain organs, namely liver, kidney, and brain (Fait et al. 1994). Organotins efficiently penetrate through the skin and easily cross the placenta and blood–brain barrier (Adeeko et al. 2003; Cooke et al. 2004; Hasan et al. 1984).
Comparison of the effects of various trialkyltins indicated that the compounds with short alkyl groups such as trimethyltin (TMT) and triethyltin were mainly neurotoxic, whereas organotins with alkyl chains of intermediate length (tripropyltin and TBT) were primarily immunotoxic (Snoeij et al. 1985). The higher trialkyltin homologs trihexyltin and trioctyltin were found to be only slightly toxic; however, further metabolism in vivo converted them to their dialkyltin forms, which are also highly immunotoxic (Penninks et al. 1985; Seinen and Willems 1976; Snoeij et al. 1988). A single oral dose of DOT, DBT, or TBT induces a dose-related reduction of the relative thymus weight in rats, and impaired cell-mediated immunity was observed after dietary exposure to TPT for several weeks (Krajnc et al. 1984; Seinen et al. 1977a, 1977b; Snoeij et al. 1988; Vos et al. 1984a, 1984b, 1990). Furthermore, exposure of pregnant rats to organotins causes reduced birth weight (Adeeko et al. 2003; Cooke et al. 2004; Crofton et al. 1989).
Reduced birth weight has also been observed with prolonged intrauterine glucocorticoid exposure (Benediktsson et al. 1993; Lindsay et al. 1996a, 1996b; Stewart et al. 1995). After such an insult, circulating cortisol levels remained elevated throughout adult life, indicating a permanently disturbed regulation of the hypothalamic–pituitary–adrenal axis, which leads to a higher susceptibility for cardiovascular and metabolic disorders including obesity, insulin resistance, and type II diabetes (Drake et al. 2005; Seckl et al. 2000). In the placenta the fetus is protected from the high maternal glucocorticoid concentration through the activity of 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2), which converts active 11β-hydroxyglucocorticoids (cortisol in human, corticosterone in rodents) into inactive 11-ketoglucocorticoids (cortisone in human, 11-dehydrocorticosterone in rodents) (reviewed in Stewart and Krozowski 1999). Impaired 11β-HSD2 activity, due to mutations or the presence of inhibitors such as glycyrrhetinic acid (GA), strongly correlates with reduced birth weight and metabolic complications in later life of the offspring (Drake et al. 2005; Lindsay et al. 1996b; Odermatt 2004; Seckl et al. 2000).
Moreover, exposure of rats to excessive levels of glucocorticoids causes thymus involution (Schuurman et al. 1992), a phenomenon also evident after exposure to organotins. Treatment of rats with high doses of the 11β-HSD inhibitor GA led to a significant elevation of systemic glucocorticoid levels accompanied by thymocyte apoptosis (Horigome et al. 1999).
Despite the fact that both exposure to excessive levels of organotins and glucocorticoids cause low birth weight and thymus involution in animal models, the impact of organotins on the control of the intracellular availability of glucocorticoids has not been studied. Therefore, we investigated the effect of various organotins on the activities of 11β-HSD1, converting inactive 11-keto-glucocorticoids to active 11β-hydroxyglucocorticoids, and of 11β-HSD2, catalyzing the opposite reaction. We also studied the mechanism of organotin-dependent inhibition of 11β-HSD2.
Materials and Methods
Chemicals and reagents.
We purchased [1,2,6,7-3H]-cortisol, [2,4,6,7-3H]-estrone, and [2,4,6,7-3H]-estradiol from Amersham Pharmacia (Piscataway, NJ, USA); [1,2,6,7-3H]-cortisone from American Radiolabeled Chemicals (St. Louis, MO, USA); cell culture media and supplements from Invitrogen (Carlsbad, CA, USA); and steroid hormones from Steraloids (Wilton, NH, USA). All other chemicals were obtained from Fluka AG (Buchs, Switzerland) and were of the highest grade available. Organotins were dissolved in dimethyl sulfoxide (DMSO) and stored as 20-mM stock solutions at −70°C. Human 11β-HSD1 and 11β-HSD2 expression constructs in pcDNA3 vector (Invitrogen) were described previously (Odermatt et al. 1999). Plasmids containing cDNA from human 17β-HSD1 or 17β-HSD2, kindly provided by Stefan Andersson, were recloned into pcDNA3 vector by PCR with primers at the 5′ end containing a HindIII restriction site (17β-HSD1) or a BamHI restriction site (17β-HSD2), a Kozak consensus sequence (Kozak 1989) and the initiation codon, and primers at the 3′ end containing a stop codon followed by an XbaI restriction site. All constructs were verified by sequencing.
Cell culture.
HEK-293 (human embryonic kidney) cells stably transfected with FLAG (Asp–Tyr–Lys–Asp–Asp–Asp–Asp–Lys)-tagged human 11β-HSD2 (Schweizer et al. 2003) were grown at 37°C under 5% carbon dioxide to 60–70% confluence in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal calf serum, 4.5 g/L glucose, 50 U/mL penicillin/streptomycin, and 2 mM glutamine. SW-620 (human colorectal adenocarcinoma) cells and JEG-3 (human choriocarcinoma) cells were cultured according to the recommendations of the supplier (American Type Culture Collection, Manassas, VA, USA). The recently described RCCD-2 aldosterone-sensitive rat cortical collecting duct cells (Djelidi et al. 2001) were cultured in DMEM/Ham’s F-12 (1:1), 14 mM NaHCO3, 2 mM glutamine, 10 U/mL penicillin/streptomycin, and 20 mM HEPES, pH 7.4.
Activity assays in cell lysates.
To measure 11β-HSD2 activity, stably transfected HEK-293 cells (Schweizer et al. 2003) were grown in 10-cm culture dishes to 90% confluence. Cells were rinsed 3 times with phosphate-buffered saline (PBS) and resuspended in 2 mL ice-cold buffer TS2 (100 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM MgCl2, 250 mM sucrose, 20 mM Tris-HCl, pH 7.4). Aliquots of the cell suspension were frozen at −20°C, retaining full enzymatic activity for at least 3 months. For determinationof 11β-HSD2 activity, aliquots were thawed, sonicated, and diluted 1:12 in buffer TS2 (4°C). We carried out reactions in a final volume of 20 μL containing 10 nCi [1,2,6,7-3H]-cortisol, 400-μM NAD+, and different concentrations of unlabeled cortisol. Final cortisol concentrations were 40 nM for measurements of inhibitors and ranged between 10 nM and 200 nM for determination of apparent Km values. Incubations were for 10 min at 37°C.
For determination of 11β-HSD1, 17β-HSD1, and 17β-HSD2 activity, HEK-293 cells transfected by the calcium-phosphate precipitation method were harvested 48 hr later, washed with PBS, and centrifuged for 3 min at 150 × g. Supernatants were removed, cell pellets quick-frozen in a dry ice ethanol bath, and stored at −70°C. For assaying 11β-HSD1, we dissolved pellets in TS2 buffer; for 17β-HSD1 or 17β-HSD2, we used a buffer containing 50 mM potassium phosphate, 20% glycerol, and 1 mM EDTA. We measured 11β-HSD1 oxoreductase activity as described recently (Atanasov et al. 2004), using radiolabeled cortisone as substrate. 17β-HSD1 and 17β-HSD2 activities were measured in the presence of radiolabeled estrone or estradiol and unlabeled steroid at final concentrations of 200 nM and 500 μM NADPH or NAD+, respectively.
Determination of 11β-HSD2 activity in intact cells and MTT cytotoxicity assay.
HEK-293 cells stably transfected with 11β-HSD2 (25,000 cells per well) were seeded 24 hr prior to the assay in poly-d-lysine coated 96-well Biocoat plates (Becton Dickinson, Basel, Switzerland). The medium was carefully removed, followed by the addition of 30 μL fresh medium, 10 μL medium containing various concentrations of organotins, and 10 μL radiolabeled cortisol. The reaction volume was 50 μL, with a final cortisol concentration of 40 nM. The cells were incubated for 2 hr at 37°C under 5% CO2. We stopped reactions by adding an excess of unlabeled cortisone and cortisol in methanol, and separated steroids using thin-layer chromatography, followed by scintillation counting. 11β-HSD2 activity in JEG-3, SW-620, and RCCD-2 cells was measured similarly by adjusting the cell density and reaction time to obtain a maximal conversion of cortisol between 15 and 25%.
To ensure that the observed inhibition of 11β-HSD2 activity was not due to cell death, we assessed cytotoxicity of organotin compounds parallel to the activity assay under identical conditions. The corresponding organotin compound was added to the cells, followed by incubation for 2 hr at 37°C under 5% CO2. Cells were washed with PBS and incubated in fresh medium containing 0.5 mg/mL 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). After conversion of MTT, we removed the medium and added 200 μL DMSO to the insoluble fraction. Conversion of MTT was kept below OD 0.9 (A570–A690).
Site-directed mutagenesis and analysis of mutant 11β-HSD2 enzymes.
Mutations were introduced into the C-terminally FLAG-tagged 11β-HSD2 cDNA in Bluescript vector by site-directed mutagenesis according to the Quick Change mutagenesis kit (Stratagene, Amsterdam, the Netherlands) (Odermatt et al. 1999). All constructs were verified by sequencing and recloned into pcDNA3 expression vector. Wild-type and mutant enzymes were expressed in HEK-293 cells, lysates were prepared, and proteins were separated by sodium dodecyl sulfate (SDS) gel electrophoresis. Proteins were transferred to nitrocellulose, and expression levels of 11β-HSD2 constructs were detected using mouse monoclonal antibody M2 raised against the FLAG epitope and visualized with a horseradish peroxidase conjugated anti-mouse antibody and enhanced chemiluminescence Western kit (Pierce, Rockford, IL, USA). After detection of 11β-HSD2 constructs, nitrocellulose membranes were stripped and incubated with rabbit polyclonal anti-actin IgG (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA) and horseradish peroxidase–conjugated goat anti-rabbit IgG to adjust for the amount of proteins loaded on the gel. The expression of mutant relative to wild-type enzyme was adjusted for calculation of kinetic parameters.
Three-dimensional modeling.
The 3-D model of human 11β-HSD2 from Arnold et al. (2003) was used with NAD+ extracted from protein data bank file 1AHI (Tanaka et al. 1996). Human 11β-HSD2 was then minimized for 100,000 iterations with Discover 3 (Accelrys Inc., San Diego, CA, USA) using the extensible and systematic force field (ESFF), with a distant dependent dielectric constant of 2, to model water in the protein.
Statistical analysis.
Enzyme kinetics were analyzed by nonlinear regression using Data Analysis Toolbox (MDL Information Systems Inc., Nashville, TN, USA) assuming first-order rate kinetics. Data represent mean ± SD of at least four independent experiments.
Results
Inhibition of 11β-HSD2 but not 11β-HSD1 by organotins.
To investigate whether organotins disrupt the control of the ratio of active to inactive glucocorticoids by inhibition of 11β-HSD enzymes, we incubated lysates of HEK-293 cells expressing recombinant 11β-HSD constructs with various concentrations of TBT and TPT and determined enzymatic activities. TBT and TPT did not inhibit the 11β-HSD1–dependent conversion of cortisone to cortisol at concentrations up to 200 μM (Table 1). In contrast the 11β-HSD2–dependent conversion of cortisol to cortisone was efficiently inhibited by both compounds with IC50 (median inhibitory concentration) values in the low micromolar range, indicating that organotins selectively abolish the 11β-HSD2–dependent inactivation of glucocorticoids.
Despite that 11β-HSD1 and 11β-HSD2 interconvert the same substrate, they are phylogenetically relatively distant enzymes, sharing only 18% identical amino acid sequence (Baker 2004). 11β-HSD2 is more closely related to 17β-HSD2, with an amino acid sequence about 45% identical. Thus, we also determined the conversion of estrone to the more potent estrogen estradiol by 17β-HSD1 and the reverse reaction by 17β-HSD2 in the presence of various concentrations of organotins. As shown in Table 1, neither TBT nor TPT inhibited 17β-HSD1. 17β-HSD2 activity was inhibited by TPT at a 6-fold higher IC50 value compared with that of 11β -HSD2, indicating that organotins preferentially inhibit 11β-HSD2.
Because trialkyltins are progressively dealkylated by microorganisms in the environment and in mammalian organs including brain, liver, and kidneys (Gadd 2000), and because the dialkyltins DBT and DPT are major metabolites with significant toxicity (Penninks et al. 1985; Seinen and Willems 1976; Snoeij et al. 1988), we included these two compounds in the study. Analysis of all four organotin compounds revealed comparable inhibitory properties both in lysates and intact cells expressing 11β-HSD2 with IC50 values in the low micromolar range (Table 1). In intact cells the trialkyltins were approximately 3-fold more potent than the dialkyltins, and the phenyltins were slightly more potent than the butyltins, well correlating with the hydrophobicity of these compounds. It is important to note that the inhibitory potency of organotins is comparable with that of the well-known 11β-HSD inhibitor GA in intact cells.
Dithiothreitol but not glutathione protects 11β-HSD2 from inhibition by organotins.
To investigate the molecular mechanism of organotin-induced inhibition of 11β-HSD2, we measured the effect of organotins in the presence or absence of the reducing agent dithiothreitol (DTT) (Figure 1). DTT alone did not significantly alter enzymatic activity. A concentration of 2 mM DTT, added simultaneously with 5 μM of the corresponding organotin compound, restored 70–80% of 11β-HSD2 activity. Upon preincubation of 11β-HSD2 with organotins for 15 min, 50–60% of the enzymatic activity could be restored (not shown), indicating that most but not all of the inhibitory effect was reversible. In contrast to the dithiol DTT, the monothiol glutathione was not able to prevent organotin-dependent inhibition of 11β-HSD2 (not shown).
We recently demonstrated that dithiocarbamates, another environmentally relevant class of compounds, inhibit 11β-HSD2. Cofactor NAD+ partially protected the enzyme, suggesting that covalent modification of the thiol group at Cys90 in the cofactor-binding region may be responsible for dithiocarbamate-induced inactivation of 11β-HSD2 (Atanasov et al. 2003). In contrast we found no protective effect of NAD+ on organotin-induced enzyme inhibition (not shown), indicating an inhibitory mechanism distinct from that of dithiocarbamates.
Functional analysis of the cysteine residues of 11β-HSD2.
We have previously shown that substitution of Cys90 by serine leads to abolished protein expression and enzymatic activity (Atanasov et al. 2003). Because NAD+ did not protect 11β-HSD2 from organotin-induced inhibition, we analyzed the functional relevance of the remaining eight cysteine residues by mutating them to serines. All nine cysteine-to-serine mutants contained a FLAG epitope at the C-terminus, allowing the quantification of the relative protein expression (Figure 2). Five of the mutant enzymes showed significantly reduced expression. No band could be detected for mutant Cys228Ser and only a very weak band for Cys90Ser was detected. As expected, no activity could be measured using lysates from cells transfected with cDNA for either of these two mutant enzymes. Significantly reduced expression was also observed for mutants Cys128Ser and Cys188Ser. When the catalytic activity of these mutant enzymes was adjusted for their reduced expression, an apparent Vmax comparable to that of wild-type 11β-HSD2 was obtained (Table 2), indicating reduced enzyme stability but intact catalytic activity. For mutant Cys264Ser, which also showed reduced expression, a slightly higher Vmax was obtained. Comparison of the kinetic parameters revealed altered kinetic parameters for mutant Cys264Ser, with a 3-fold higher apparent Km. Mutation of Cys264 to serine also resulted in a 2-fold increase in the IC50 for TBT, suggesting a role for Cys264 in the binding of cortisol and in the interaction with organotins with 11β-HSD2.
Analysis of the mode of organotin-dependent inhibition of 11β-HSD2.
We next investigated the effect of preincubation of 11β-HSD2 with either 1.5 μM TPT or 2 μM TBT for 5 or 10 min. Although there was a slight tendency toward increased inhibitory effect upon preincubation with TBT, the changes did not reach significance (Figure 3), in line with reversible inhibition. To further assess the mode of inhibition, we determined the change of apparent Km and Vmax in the absence or presence of TBT. The apparent Km increased 2- and 3-fold upon incubation with 2 μM and 3 μM TBT (62 ± 17 nM for the control compared with 137 ± 24 nM and 197 ± 33 nM for treated samples, respectively), whereas Vmax decreased slightly (2.14 ± 0.32 nmol × h−1 × mg−1 for the control compared with 1.66 ± 0.40 nmol × h−1 × mg−1 and 1.48 ± 0.48 nmol × h−1 × mg−1 for treated samples, respectively). These findings suggest a mixed-competitive mode of inhibition, with most of the inhibitory effect being reversible.
To further test this assumption, we measured the effect of diluting the enzyme–inhibitor complex (EI) (Figure 4). In case of a competitive mode of inhibition, dilution of the EI complex would lead to a decreased inhibitor concentration and reduced competition with the substrate, hence reduced enzyme inhibition. If covalent enzyme modification takes place, the dilution of the EI complex would not change the proportion of modified to unmodified molecules, meaning that after the dilution of EI there would be no change of the relative inhibition compared with the control. In case of TBT-induced inhibition of 11β-HSD2, a 2-fold and 4-fold dilution of the EI complex led to a proportional decrease of the relative inhibitory effect, indicating a transient interaction of the organotin compound with 11β-HSD2.
Inhibition of 11β-HSD2 in endogenous cell lines.
Because organisms are exposed to various sources of organotins and these chemicals undergo dealkylation in vivo, cells in tissues are generally exposed to a mixture of organotins. Therefore, we compared the activity of 11β-HSD2 in intact cells either upon incubation with DBT, TBT, DPT, or TPT alone, at concentrations 50% below their IC50 value, or after incubation with a mixture of the four chemicals (Figure 5). Whereas each compound alone reduced 11β-HSD2 activity only slightly, the mixture showed additive inhibitory effects and significantly inhibited enzymatic activity, indicating that the distinct organotins act by the same mechanism. We observed this phenomenon in transfected HEK-293 cells as well as in endogenous cell lines.
We next determined the potential of TBT to inhibit 11β-HSD2 activity in cell lines derived from tissues with endogenous expression of this enzyme, for example, placenta, renal cortical collecting duct, and colon (Figure 6). In placenta-derived JEG-3 cells and in colon-derived SW-620 cells, the inhibition of 11β-HSD2 by TBT was highly similar to that observed in HEK-293 cells. We observed approximately 2-fold stronger inhibition in renal cortical-collecting duct–derived RCCD-2 cells, with an IC50 of 0.83 ± 0.23 μM for TBT.
Discussion
Relatively little is known about the molecular targets of organotins despite their wide range of toxic effects and that they can be readily detected in the blood of humans. In this article we describe the organotin-dependent inhibition of 11β-HSD2. Our data suggest that organotins inhibit 11β-HSD2 by a mostly reversible, mixed-competitive mode of inhibition. Comparison of the kinetic parameters obtained from measurements with lysates and intact cells expressing 11β-HSD2 indicates that the trialkyltins enter the cell more easily, which explains their more potent effects in intact cells. However, dialkyltins display equal or even enhanced inhibitory potency in lysates. Organotin-induced inhibition of 11β-HSD2 was prevented by the dithiol DTT but not by the endogenous monothiol glutathione, which sugggests that two cysteine residues in close proximity might be involved in the mechanism of inhibition. This is in contrast to the inhibition of 11β-HSD2 by dithiocarbamates, which irreversibly inhibit the enzyme, probably through covalent modification of a cysteine residue by attachment of a carbamoyl group (Atanasov et al. 2003).
The inhibition of 11β-HSD2 by dithiocarbamates (Atanasov et al. 2003) and organ-otins seems to involve distinct cysteine residues, as addition of high concentrations of cofactor NAD+ partially protected from dithiocarbamates but not from organotins. Site-directed mutagenesis and functional analysis revealed an essential role of Cys128, Cys188, Cys228, and Cys264 for enzyme stability and/or catalytic activity in addition to the previously described Cys90. To begin to understand the basis for the different activities of these mutant enzymes, we applied a 3-D structural model of 11β-HSD2 (Arnold et al. 2003; Atanasov et al. 2003) to investigate the structure surrounding each of these cysteines (Figure 7). Cys48 and Cys371 are located in hydrophobic segments thought to associate with cellular membranes in the endoplasmic reticulum. These residues are outside the 260 residue core segment comprising the catalytically active domain in homologs of 11β-HSD2 and were not analyzed further. The 3-D model shows that Cys127, Cys128, and Cys248 do not interact with sites on 11β-HSD2 critical for binding of either NAD+, substrate, or the catalytically active Tyr232. Figure 7A shows that these cysteines have few stabilizing interactions with other amino acids and are oriented to the solvent, away from the catalytic pocket.
As described previously (Atanasov et al. 2003), the thiol group of Cys90 has several interactions with amino acids that stabilize Glu115 and Asp91 (Figure 7B). Glu115 has critical hydrogen bonds with the ribose hydroxyl on NAD+ that are important in stabilizing binding of the cofactor and in maintaining its orientation to the substrate. Thus, the loss of the stabilizing effects of Cys90 on the orientation of Glu115 leads to an inactive 11β-HSD2.
Similarly, the thiol group on Cys188 stabilizes several amino acids that interact with the pyrophosphate segment of NAD+ (Figure 7C). However, like Cys90, Cys188 is not directly involved in interactions with the cofactor, and it appears that if the thiol group is replaced with a serine hydroxyl group, there remains sufficient stabilization of the structure near the pyrophosphate group in mutant Cys188Ser to retain some catalytic activity.
Cys228 is in the loop that precedes the catalytically active Tyr232 (Figure 7D). Evidence from solved 3-D structures of 11β-HSD2 homologs, such as 17β-HSD1, clearly shows that this loop stabilizes the position of the nicotinamide ring and the catalytic Tyr232 with the steroid substrate to promote catalysis (Breton et al. 1996; Ghosh et al. 1994; Tanaka et al. 1996; Yamashita et al. 1999). The thiol on Cys228 stabilizes the backbone oxygen in Pro227, which is an important structural amino acid that is also in the loop that positions the catalytic tyrosine, the nicotinamide ring, and the steroid. Cys228 also stabilizes Glu277, which belongs to a helix in the substrate binding site. Thus, the thiol group on Cys228 interacts with amino acids on 11β-HSD2 that are important in substrate binding, which explains the loss of activity of mutant Cys228Ser.
Figure 7D also shows that the thiol group of Cys264 has important interactions with Leu284, Ala285, and Pro288, which are part of the helix in the C-terminal region of 11β-HSD2 that likely is important in the substrate binding site, based on analyses of 17β-HSD2 and other homologs (Tanaka et al. 1996) of 11β-HSD2. Our experimental data indicate that the serine hydroxyl group can partially take over the function of the thiol on Cys264. Together, these findings suggest that Cys228 and Cys264 may be involved in the interactions with organotins, and interference with the function of their thiol groups may be responsible for inactivation of 11β-HSD2 by organotins.
Although the chemical nature of organotin–protein interactions is not completely understood at present, it is believed that most of the properties of organotins are a result of the nature of C–Sn bonds that can be attacked by both nucleophilic and electrophilic reagents (Hoch 2001). Buck et al. recently described the interaction of organotins with the membrane protein stannin (Buck et al. 2003, 2004). A nonapeptide derived from stannin containing the sequence Cys–Trp–Cys was able to dealkylate TMT, followed by binding of the resulting DMT. Buck et al. (2003, 2004) further showed that DTT bound TMT without inducing dealkylation. Stannin efficiently dealkylated organotins with short alkyl side chains with, weak binding observed for TBT and TOT.
11β-HSD2 does not possess a Cys–Xaa–Cys motif, and Cys127 and Cys128 are unlikely to be the target residues for inhibition by organotins, as replacement of either of these residues did not affect the organotin-dependent inhibition of the mutant enzymes. Replacing Cys228 by serine completely abolished enzyme activity, indicating an important functional role of this residue. Mutating Cys264 to serine led to a well-expressed enzyme with only slightly reduced catalytic efficiency. This mutant was less sensitive to organotin-dependent inhibition.
The potency of organotins in inhibiting 11β-HSD2 is equal to or greater than that reported for other enzymes involved in steroid hormone metabolism, including cytochrome P450 aromatase (Lo et al. 2003), 5α-reductases (Doering et al. 2002), and rat testis microsomal 3β-HSD (McVey and Cooke 2003). As mentioned earlier, organotins represent ubiquitous contaminants of the water ecosystem. Although most often present at lower nanomolar concentrations, organotins accumulate in aquatic organisms, with up to 70,000 times higher organotin concentrations in plankton and other organisms than in sea water (Takahashi et al. 1999). As little as 200 g of contaminated shell-fish could be sufficient to reach the toxic daily intake of TBT (0.25 μg/kg body weight) (McVey and Cooke 2003). Another major source of organotins for humans is the drinking water in locations where PVC water pipes are used. A study of organotin levels in Canadian drinking water distributed through PVC pipes in 1996 revealed total concentrations of different organotins < 50 ng Sn/L, although in some occasions values > 250 ng Sn/L were detected (Sadiki and Williams 1999). Although the IC50 values of organotins for 11β-HSD2 in the present study were in the high nanomolar and low micromolar range, the accumulation of organotins in specific organs and tissues including brain, liver, and kidneys may lead to high local concentrations, as has been found in aquatic organisms. In addition Chicano et al. (2001) provided evidence that organotins are located in the upper part of the phospholipid palisade near the lipid–water interface and affect the degree of hydration of the phospholipid carbonyl moiety. Thus, intracellular organotin concentrations might be highest at the surface of membranes, and 11β-HSD2 might be exposed to concentrations much higher than those in the solution.
The increased glucocorticoid-mediated effects due to the inhibition of 11β-HSD2 are expected to disturb several essential physiologic processes. The effect of TBT and its major metabolite DBT to suppress T-cell–dependent immune functions by causing thymus atrophy has been extensively studied (Krajnc et al. 1984; Seinen et al. 1977a, 1977b; Snoeij et al. 1988; Tryphonas et al. 2004; Vos et al. 1984a, 1990). A single dose of TBT, DBT, or DOT induced dose-dependent reductions in the weight of the thymus, spleen, and lymph node (Seinen et al. 1977b). The effect of TBT was less pronounced and slightly delayed compared with DBT, indicating that in vivo TBT is metabolized to the more toxic DBT (Snoeij et al. 1988). The thymotoxic effects of organotins are completely reversible (Seinen et al. 1977b). A selective inhibition of the proliferation of immature CD4−/CD8+ thymocytes by organotins seems to be responsible for the observed depletion of CD4+/CD8+ thymocytes, which show a rapid turnover.
11β-HSD enzymes play a pivotal role in regulating proliferation and differentiation in various tissues. 11β-HSD1 generates active glucocorticoids and promotes differentiation, and 11β-HSD2 inactivates glucocorticoids, thereby promoting proliferation. 11β-HSD1 and 11β-HSD2 were both expressed in whole mouse thymus (Moore et al. 2000; Speirs et al. 2004), although the exact subtype-specific expression pattern of 11β-HSD enzymes remains to be determined. In the acute stress response, the high level of glucocorticoids induces thymus involution (Schuurman et al. 1992). Organotin-dependent inhibition of 11β-HSD2 may cause antiproliferative effects on immature thymocytes by increasing locally the ratio of active to inactive glucocorticoids or, alternatively, by increasing systemic glucocorticoid levels. Both organotin-induced inhibition of 11β-HSD2 and thymotoxicity are reversible.
Experiments in mice showed maximal thymocyte apoptosis 8 hr after glucocorticoid administration, followed by full recovery after 18 hr (Ishii et al. 1997). There was a significant depletion of CD4+/CD8+ thymocytes. A delayed, dose-dependent apoptosis of thymocytes, reaching maximal effect after 24 hr, was observed when mice were treated with a single dose of the 11β-HSD inhibitor GA (Horigome et al. 1999). Thymocyte apoptosis was induced dose-dependently by corticosterone in vitro. GA alone did not induce apoptosis in vitro, suggesting that elevated corticosterone levels due to inhibition of 11β-HSD2 may cause the apoptosis. Similarly, organotin-dependent inhibition of 11β-HSD2 and subsequent locally enhanced glucocorticoid levels may contribute to the immunotoxicity of these compounds. However, thymus atrophy was also observed in DBT- and TBT-treated adrenalectomized rats (Seinen and Willems 1976; Snoeij et al. 1985), and no extensive cell destruction was observed in DBT- and TBT-treated rats compared with rats treated with very high glucocorticoid concentrations (La Pushin and de Harven 1971). These findings suggest that the immunotoxic effects of organotins are caused by a glucocorticoid-dependent and a glucocorticoid-independent mechanism.
The glucocorticoid-dependent effects caused by organotins may be most critical during pregnancy, where fetal development is highly sensitive to glucocorticoids (Seckl et al. 2000). The lower birth weight and decreased weight gain in the offspring after exposure of organotins during pregnancy is a phenotype also observed as a result of exposure to excessive levels of glucocorticoids. Enhanced glucocorticoid action due to inhibition of 11β-HSD2, which in the placenta protects the fetus from high maternal levels, or due to treatment with synthetic glucocorticoids that cannot be inactivated by 11β-HSD2, such as dexamethasone (Rebuffat et al. 2004), have been associated with reduced birth weight and irreversible changes in the cardiovascular system with complications in later life (Seckl et al. 2000). Thus, inhibition of 11β-HSD2 may be one reason that offspring from pregnant rats treated with organotins show significantly reduced birth weight (Adeeko et al. 2003; Cooke et al. 2004; Crofton et al. 1989).
Conclusions
This work demonstrates the disruption of the 11β-HSD2–dependent inactivation of glucocorticoids by organotins. Various organotin compounds inhibit 11β-HSD2, mainly by a reversible mode of inhibition, and show additive effects. Endogenous glutathione cannot prevent the organotin-induced inhibition of 11β-HSD2, which explains the comparable inhibitory kinetics obtained in experiments with cell lysates and in intact cells. The results suggest that enhanced glucocorticoid concentrations due to disruption of 11β-HSD2 function may contribute to the observed organotin-dependent toxicity in glucocorticoid sensitive tissues such as thymus and placenta. Clearly, additional experiments in vivo must be performed to elucidate the relevance of organotin-dependent interference with glucocorticoid action and its pathophysiologic consequences.
We thank N. Farman (INSERM, Faculté X-Bichat, Paris, France) for providing the RCCD-2 cells, S. Andersson (University of Texas Southwestern Medical Center, Dallas, Texas) for the gift of 17β-HSD1 and 17β-HSD2 plasmids, and J. Vos (National Institute of Public Health and the Environment, Biltoven, the Netherlands) for helpful discussion. We also thank H. Jamin for excellent technical support.
A.O. is a Cloëtta Research Fellow supported by grants from the Swiss National Science Foundation (3100A0-100060 and NRP50 “Endocrine Disruptors” 4050-066575) and the Swiss Cancer League (OCS-01402-08-2003). M.E.B. and S.T. were supported by National Institute of Health grants DK41841 and HLOO4791.
Figure 1 DTT prevents organotin-dependent inhibition of 11β-HSD2. The oxidation of cortisol by 11β-HSD2 was determined using cell lysates, as described in “Materials and Methods.” Addition of DTT at a final concentration of 2 mM restored 70–80% of the activity measured in absence of organotins.
Figure 2 Expression of wild-type 11β-HSD2 and cysteine to serine mutants. C-terminally FLAG-epitope tagged wild-type and mutant 11β-HSD2 enzymes were expressed in HEK-293 cells, and protein expression was analyzed by Western blotting as described in “Materials and Methods.” After detection of the FLAG-tagged 11β-HSD2 enzymes, nitrocellulose membranes were stripped, and actin expression was detected as a control for the amount of protein loaded on the SDS gel. A representative blot from three comparable experiments is shown.
Figure 3 Effect of preincubation on 11β-HSD2 activity. The oxidation of cortisol to cortisone was determined after preincubation for 5 or 10 min with vehicle (control), 1.5 μM TPT, or 2 μM TBT in lysates of HEK-293 cells expressing 11β-HSD2. Data are mean ± SD from at least three independent experiments.
Figure 4 Effects of dilution of the enzyme-inhibitor (EI) complex on TBT-dependent inhibition of 11β-HSD2. Lysates from HEK-293 cells expressing 11β-HSD2 were split into two equal aliquots. TBT was added to the first aliquot, and the same volume of vehicle, serving as a control, was added to the second. Both aliquots were incubated for 5 min at 37°C, followed by determination of the activity of the control and EI mixture either undiluted or after a 2- or 4-fold dilution. Data are mean ± SD from at least three independent experiments measured in triplicate.
*p < 0.05.
Figure 5 Additive inhibitory effect of a mixture of organotins on 11β-HSD2 activity. Conversion of cortisol to cortisone by 11β-HSD2 stably expressed in intact HEK-293 cells was measured in a volume of 50 μL cell culture medium containing 40 nM cortisol and the corresponding amount of the organotin, as indicated (see “Materials and Methods”). Data were normalized to the control and are mean ± SD from at least three independent experiments measured in triplicate.
*Statistical significance of p < 0.01 compared with all other values.
Figure 6 Dose–response curves for TBT-induced inhibition of 11β-HSD2 in intact cells expressing endogenous 11β-HSD2. RCCD-2, rat renal cortical collecting duct cell line; JEG-3, human choriocarcinoma cell line; SW-620, human colon adenocarcinoma cell line. Details on culture conditions and activity assay in intact cells are given in “Materials and Methods.”
Figure 7 Predicted interactions of cysteine residues in the conserved core domain of 11β-HSD2. ( A) Cys127, Cys128, and Cys248 are oriented into the solvent, away from the catalytic pocket, and have few stabilizing interactions with other residues. ( B) Cys90 interacts with amino acids that stabilize Glu115 and Asp91, which have a critical role by forming hydrogen bonds to the ribose hydroxyl on NAD+ that are important to stabilize binding of the cofactor and maintain its orientation to the substrate. ( C) Cys188 is not directly involved in interactions with NAD+ but stabilizes several amino acids that interact with the pyrophosphate segment of the cofactor. ( D) The thiol on Cys228 stabilizes the position of Pro227 and Glu277, which are important for positioning of the catalytic tyrosine and the nicotinamide ring and for binding of the steroid substrate. The thiol group of Cys264 has important interactions with Leu284, Ala285, and Pro288 in the helix in the C-terminal region of 11β-HSD2, which is important for substrate binding. Predicted interatomic distances in angstroms are depicted by number. Blue: nitrogen; green: carbon; purple: phosphorus; red: oxygen; and yellow: sulfur.
Table 1 Selective inhibition [IC50 (μM)] of 11β -HSD2 by organotins.a
Cell lysate
Organotin 17β -HSD1 17β -HSD2 11β -HSD1 11β -HSD2 Intact cells 11β -HSD2
TBT > 200 > 200 > 200 1.90 ± 0.66 1.52 ± 0.43
DBT ND ND ND 1.95 ± 0.27 5.03 ± 0.70
TPT > 200 19 ± 3 > 200 3.19 ± 0.73 0.99 ± 0.24
DPT ND ND ND 1.42 ± 0.17 2.89 ± 0.59
GA ND ND ND 0.40 ± 0.08 1.01 ± 0.29
ND, not determined.
a Data are mean ± SD from at least four independent experiments measured in duplicate.
Table 2 Analysis of the kinetic parameters of wild-type 11β-HSD2 and cysteine to serine mutants.a
Km (nM) Vmax (nmol × h−1 × mg−1) TBT [IC50 ± SD (μM)]
Wild-type 62 ± 17 2.14 ± 0.32 1.90 ± 0.46
Cys48Ser 71 ± 23 1.93 ± 0.17 2.11 ± 0.33
Cys90Ser ND ND ND
Cys127Ser 67 ± 19 2.48 ± 0.36 2.23 ± 0.33
Cys128Ser 97 ± 21 3.02 ± 0.28 1.87 ± 0.53
Cys188Ser 49 ± 16 2.78 ± 0.37 1.95 ± 0.38
Cys228Ser ND ND ND
Cys248Ser 51 ± 13 2.49 ± 0.43 2.07 ± 0.47
Cys264Ser 173 ± 32 3.29 ± 0.35 4.20 ± 0.44
Cys371Ser 74 ± 20 2.08 ± 0.28 2.17 ± 0.50
ND, not determined.
a Data are mean ± SD from at least three independent experiments measured in triplicate.
==== Refs
References
Adeeko A Li D Forsyth DS Casey V Cooke GM Barthelemy J 2003 Effects of in utero tributyltin chloride exposure in the rat on pregnancy outcome Toxicol Sci 74 407 415 12773765
Arnold P Tam S Yan L Baker ME Frey FJ Odermatt A 2003 Glutamate-115 renders specificity of human 11β-hydroxysteroid dehydrogenase type 2 for the cofactor NAD(+) Mol Cell Endocrinol 201 177 187 12706305
Atanasov AG Nashev LG Schweizer RA Frick C Odermatt A 2004 Hexose-6-phosphate dehydrogenase determines the reaction direction of 11β-hydroxysteroid dehydrogenase type 1 as an oxoreductase FEBS Lett 571 129 133 15280030
Atanasov AG Tam S Rocken JM Baker ME Odermatt A 2003 Inhibition of 11β-hydroxysteroid dehydrogenase type 2 by dithiocarbamates Biochem Biophys Res Commun 308 257 262 12901862
Baker ME 2004 Evolutionary analysis of 11β-hydroxysteroid dehydrogenase-type 1, -type 2, -type 3 and 17β-hydroxysteroid dehydrogenase-type 2 in fish FEBS Lett 574 167 170 15358559
Benediktsson R Lindsay RS Noble J Seckl JR Edwards CR 1993 Glucocorticoid exposure in utero : new model for adult hypertension Lancet 341 339 341 8094115
Bhosle NB Garg A Jadhav S Harjee R Sawant SS Venkat K 2004 Butyltins in water, biofilm, animals and sediments of the west coast of India Chemosphere 57 897 907 15488580
Breton R Housset D Mazza C Fontecilla-Camps JC 1996 The structure of a complex of human 17β-hydroxysteroid dehydrogenase with estradiol and NADP+ identifies two principal targets for the design of inhibitors Structure 4 905 915 8805577
Buck BA Mascioni A Cramer CJ Veglia G 2004 Interactions of alkyltin salts with biological dithiols: dealkylation and induction of a regular β-turn structure in peptides J Am Chem Soc 126 14400 14410 15521759
Buck B Mascioni A Que L Jr Veglia G 2003 Dealkylation of organotin compounds by biological dithiols: toward the chemistry of organotin toxicity J Am Chem Soc 125 13316 13317 14583001
Chicano JJ Ortiz A Teruel JA Aranda FJ 2001 Organotin compounds alter the physical organization of phosphatidylcholine membranes Biochim Biophys Acta 1510 330 341 11342170
Coelho MR Bebianno MJ Langston WJ 2002 Routes of TBT uptake in the clam Ruditapes decussatus. II: Food as vector of TBT uptake Mar Environ Res 54 193 207 12206411
Cooke GM Tryphonas H Pulido O Caldwell D Bondy GS Forsyth D 2004 Oral (gavage), in utero and postnatal exposure of Sprague-Dawley rats to low doses of tributyltin chloride. Part 1: Toxicology, histopathology and clinical chemistry Food Chem Toxicol 42 211 220 14667468
Crofton KM Dean KF Boncek VM Rosen MB Sheets LP Chernoff N 1989 Prenatal or postnatal exposure to bis(tri-n -butyltin)oxide in the rat: postnatal evaluation of teratology and behavior Toxicol Appl Pharmacol 97 113 123 2916230
Djelidi S Beggah A Courtois-Coutry N Fay M Cluzeaud F Viengchareun S 2001 Basolateral translocation by vasopressin of the aldosterone-induced pool of latent Na-K-ATPases is accompanied by α1 subunit dephosphorylation: study in a new aldosterone-sensitive rat cortical collecting duct cell line J Am Soc Nephrol 12 1805 1818 11518773
Doering DD Steckelbroeck S Doering T Klingmuller D 2002 Effects of butyltins on human 5α-reductase type 1 and type 2 activity Steroids 67 859 887 12231121
Drake AJ Walker BR Seckl JR 2005 Intergenerational consequences of fetal programming by in utero exposure to glucocorticoids in rats Am J Physiol Regul Integr Comp Physiol 288 R34 R38 15178540
Evans S Nicholson GJ 2000 The use of imposex to assess tributyltin contamination in coastal waters and open seas Sci Total Environ 258 73 80 11007278
Fait A Ferioli A Barbieri F 1994 Organotin compounds Toxicology 91 77 82 8052990
Fent K 1996 Ecotoxicology of organotin compounds Crit Rev Toxicol 26 1 117 8833456
Fent K 2003 Ecotoxicological problems associated with contaminated sites Toxicol Lett 140–141 353 365
Gadd GM 2000 Microbial interactions with tributyltin compounds: detoxification, accumulation, and environmental fate Sci Total Environ 258 119 127 11007284
Ghosh D Wawrzak Z Weeks CM Duax WL Erman M 1994 The refined three-dimensional structure of 3α,20β-hydroxysteroid dehydrogenase and possible roles of the residues conserved in short-chain dehydrogenases Structure 2 629 640 7922040
Hasan Z Zimmer L Woolley D 1984 Time course of the effects of trimethyltin on limbic evoked potentials and distribution of tin in blood and brain in the rat Neurotoxicology 5 217 244 6542188
Hoch M 2001 Organotin compounds in the environment—an overview Appl Geochem 16 719 743
Horigome H Horigome A Homma M Hirano T Oka K 1999 Glycyrrhetinic acid-induced apoptosis in thymocytes: impact of 11β-hydroxysteroid dehydrogenase inhibition Am J Physiol 277 E624 E630 10516121
Ishii T Nakamura M Yagi H Soga H Kayaba S Gotoh T 1997 Glucocorticoid-induced thymocyte death in the murine thymus: the effect at later stages Arch Histol Cytol 60 65 78 9161690
Kozak M 1989 Context effects and inefficient initiation at non-AUG codons in eucaryotic cell-free translation systems Mol Cell Biol 9 5073 5080 2601709
Krajnc EI Wester PW Loeber JG van Leeuwen FX Vos JG Vaessen HA 1984 Toxicity of bis(tri-n -butyltin)oxide in the rat. I: Short-term effects on general parameters and on the endocrine and lymphoid systems Toxicol Appl Pharmacol 75 363 386 6474469
La Pushin RW de Harven E 1971 A study of glucocorticosteroid-induced pyknosis in the thymus and lymph node of the adrenalectomized rat J Cell Biol 50 583 597 5098862
Lindsay RS Lindsay RM Edwards CR Seckl JR 1996a Inhibition of 11-β-hydroxysteroid dehydrogenase in pregnant rats and the programming of blood pressure in the offspring Hypertension 27 1200 1204 8641724
Lindsay RS Lindsay RM Waddell BJ Seckl JR 1996b Prenatal glucocorticoid exposure leads to offspring hyperglycaemia in the rat: studies with the 11β-hydroxysteroid dehydrogenase inhibitor carbenoxolone Diabetologia 39 1299 1305 8932995
Liu JY Jiang GB 2002 Survey on the presence of butyltin compounds in chinese alcoholic beverages, determined by using headspace solid-phase microextraction coupled with gas chromatography-flame photometric detection J Agric Food Chem 50 6683 6687 12405761
Lo S Allera A Albers P Heimbrecht J Jantzen E Klingmuller D 2003 Dithioerythritol (DTE) prevents inhibitory effects of triphenyltin (TPT) on the key enzymes of the human sex steroid hormone metabolism J Steroid Biochem Mol Biol 84 569 576 12767282
McVey MJ Cooke GM 2003 Inhibition of rat testis microsomal 3β-hydroxysteroid dehydrogenase activity by tributyltin J Steroid Biochem Mol Biol 86 99 105 12943749
Moore XL Hoong I Cole TJ 2000 Expression of the 11β-hydroxysteroid dehydrogenase 2 gene in the mouse Kidney Int 57 1307 1312 10760059
Odermatt A 2004 Corticosteroid-dependent hypertension: environmental influences Swiss Med Wkly 134 4 13 14745661
Odermatt A Arnold P Stauffer A Frey BM Frey FJ 1999 The N-terminal anchor sequences of 11β-hydroxysteroid dehydrogenases determine their orientation in the endoplasmic reticulum membrane J Biol Chem 274 28762 28770 10497248
Penninks A Kuper F Spit BJ Seinen W 1985 On the mechanism of dialkyltin-induced thymus involution Immunopharmacology 10 1 10 3877030
Rebuffat AG Tam S Nawrocki AR Baker ME Frey BM Frey FJ 2004 The 11-ketosteroid 11-ketodexamethasone is a glucocorticoid receptor agonist Mol Cell Endocrinol 214 27 37 15062542
Sadiki AI Williams DT 1999 A study on organotin levels in Canadian drinking water distributed through PVC pipes Chemosphere 38 1541 1548 10070734
Schuurman HJ Van Loveren H Rozing J Vos JG 1992 Chemicals trophic for the thymus: risk for immunodeficiency and auto-immunity Int J Immunopharmacol 14 369 375 1618591
Schweizer RA Atanasov AG Frey BM Odermatt A 2003 A rapid screening assay for inhibitors of 11β-hydroxysteroid dehydrogenases (11β-HSD): flavanone selectively inhibits 11β-HSD1 reductase activity Mol Cell Endocrinol 212 41 49 14654249
Seckl JR Cleasby M Nyirenda MJ 2000 Glucocorticoids, 11β-hydroxysteroid dehydrogenase, and fetal programming Kidney Int 57 1412 1417 10760076
Seinen W Vos JG van Krieken R Penninks A Brands R Hooykaas H 1977a Toxicity of organotin compounds. III: Suppression of thymus-dependent immunity in rats by di-n -butyltindichloride and di-n -octyltindichloride Toxicol Appl Pharmacol 42 213 224 929604
Seinen W Vos JG van Spanje I Snoek M Brands R Hooykaas H 1977b Toxicity of organotin compounds. II: Comparative in vivo and in vitro studies with various organotin and organolead compounds in different animal species with special emphasis on lymphocyte cytotoxicity Toxicol Appl Pharmacol 42 197 212 929603
Seinen W Willems MI 1976 Toxicity of organotin compounds. I: Atrophy of thymus and thymus-dependent lymphoid tissue in rats fed di-n -octyltindichloride Toxicol Appl Pharmacol 35 63 75 1258060
Snoeij NJ Penninks AH Seinen W 1988 Dibutyltin and tributyltin compounds induce thymus atrophy in rats due to a selective action on thymic lymphoblasts Int J Immunopharmacol 10 891 899 2466803
Snoeij NJ van Iersel AA Penninks AH Seinen W 1985 Toxicity of triorganotin compounds: comparative in vivo studies with a series of trialkyltin compounds and triphenyltin chloride in male rats Toxicol Appl Pharmacol 81 274 286 4060154
Speirs HJ Seckl JR Brown RW 2004 Ontogeny of glucocorticoid receptor and 11β-hydroxysteroid dehydrogenase type-1 gene expression identifies potential critical periods of glucocorticoid susceptibility during development J Endocrinol 181 105 116 15072571
Stewart PM Krozowski ZS 1999 11β-Hydroxysteroid dehydrogenase Vitam Horm 57 249 324 10232052
Stewart PM Rogerson FM Mason JI 1995 Type 2 11β-hydroxysteroid dehydrogenase messenger ribonucleic acid and activity in human placenta and fetal membranes: its relationship to birth weight and putative role in fetal adrenal steroidogenesis J Clin Endocrinol Metab 80 885 890 7883847
Takahashi S Tanabe S Takeuchi II Miyazaki N 1999 Distribution and specific bioaccumulation of butyltin compounds in a marine ecosystem Arch Environ Contam Toxicol 37 50 61 10341042
Tanaka N Nonaka T Tanabe T Yoshimoto T Tsuru D Mitsui Y 1996 Crystal structures of the binary and ternary complexes of 7α-hydroxysteroid dehydrogenase from Escherichia coli Biochemistry 35 7715 7730 8672472
Tryphonas H Cooke G Caldwell D Bondy G Parenteau M Hayward S 2004 Oral (gavage), in utero and post-natal exposure of Sprague-Dawley rats to low doses of tributyltin chloride. Part II: Effects on the immune system Food Chem Toxicol 42 221 235 14667469
Vos JG de Klerk A Krajnc EI Kruizinga W van Ommen B Rozing J 1984a Toxicity of bis(tri-n -butyltin)oxide in the rat. II: Suppression of thymus-dependent immune responses and of parameters of nonspecific resistance after short-term exposure Toxicol Appl Pharmacol 75 387 408 6474470
Vos JG De Klerk A Krajnc EI Van Loveren H Rozing J 1990 Immunotoxicity of bis(tri-n -butyltin)oxide in the rat: effects on thymus-dependent immunity and on nonspecific resistance following long-term exposure in young versus aged rats Toxicol Appl Pharmacol 105 144 155 2118283
Vos JG van Logten MJ Kreeftenberg JG Kruizinga W 1984b Effect of triphenyltin hydroxide on the immune system of the rat Toxicology 29 325 336 6367147
Yamashita A Kato H Wakatsuki S Tomizaki T Nakatsu T Nakajima K 1999 Structure of tropinone reductase-II complexed with NADP+ and pseudotropine at 1.9 A resolution: implication for stereospecific substrate binding and catalysis Biochemistry 38 7630 7637 10387002
|
16263518
|
PMC1310925
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 14; 113(11):1600-1606
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8209
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8043ehp0113-00160716263519ResearchWorkgroup Report: Drinking-Water Nitrate and Health—Recent Findings and Research Needs Ward Mary H. 1deKok Theo M. 2Levallois Patrick 3Brender Jean 4Gulis Gabriel 5Nolan Bernard T. 6VanDerslice James 71 Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, Maryland, USA2 Department of Health Risk Analysis and Toxicology, University of Maastricht, the Netherlands3 Institut National de Santé Publique du Québec and Unité de recherche en santé publique, Centre Hospitalier Universitaire de Québec, Québec, Canada4 Department of Health Services Research, Texas State University, San Marcos, Texas, USA5 Department of Health Promotion Research, Southern Denmark University and Department of Public Health, University of Trnava, Slovakia6 U.S. Geological Survey, Reston, Virginia, USA7 Washington State Department of Health, Olympia, Washington, USAAddress correspondence to M.H. Ward, Occupational and Environmental Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Blvd., EPS 8104, Bethesda, MD 20892 USA. Telephone: (301) 435-4713. Fax: (301) 402-1819. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 23 6 2005 113 11 1607 1614 25 2 2005 23 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Human alteration of the nitrogen cycle has resulted in steadily accumulating nitrate in our water resources. The U.S. maximum contaminant level and World Health Organization guidelines for nitrate in drinking water were promulgated to protect infants from developing methemoglobinemia, an acute condition. Some scientists have recently suggested that the regulatory limit for nitrate is overly conservative; however, they have not thoroughly considered chronic health outcomes. In August 2004, a symposium on drinking-water nitrate and health was held at the International Society for Environmental Epidemiology meeting to evaluate nitrate exposures and associated health effects in relation to the current regulatory limit. The contribution of drinking-water nitrate toward endogenous formation of N-nitroso compounds was evaluated with a focus toward identifying subpopulations with increased rates of nitrosation. Adverse health effects may be the result of a complex interaction of the amount of nitrate ingested, the concomitant ingestion of nitrosation cofactors and precursors, and specific medical conditions that increase nitrosation. Workshop participants concluded that more experimental studies are needed and that a particularly fruitful approach may be to conduct epidemiologic studies among susceptible subgroups with increased endogenous nitrosation. The few epidemiologic studies that have evaluated intake of nitrosation precursors and/or nitrosation inhibitors have observed elevated risks for colon cancer and neural tube defects associated with drinking-water nitrate concentrations below the regulatory limit. The role of drinking-water nitrate exposure as a risk factor for specific cancers, reproductive outcomes, and other chronic health effects must be studied more thoroughly before changes to the regulatory level for nitrate in drinking water can be considered.
adverse reproductive outcomesmethemoglobinemianeoplasmsnitratenitriteN-nitroso compoundswater pollution
==== Body
Humans have altered the nitrogen cycle dramatically over the last half-century, and as a result, nitrate is steadily accumulating in our water resources. Globally, human nitrogen production has increased rapidly since 1950 and currently exceeds nitrogen fixed by natural sources by about 30% (Fields 2004). This figure compares with pre-1950 human inputs, which were a small fraction of the input from natural sources (Lambert and Driscoll 2003). Fertilizer is the largest contributor to anthropogenic nitrogen worldwide; other major sources include animal and human waste, nitrogen oxides from utilities and automobiles, and leguminous crops that fix atmospheric nitrogen (Fields 2004). These organic and inorganic sources of nitrogen are transformed to nitrate by mineralization, hydrolysis, and bacterial nitrification. Under reducing conditions, nitrate can be biologically transformed to nitrogen gas through denitrification. Nitrate not taken up by plants or denitrified migrates to streams and groundwater.
The U.S. Environmental Protection Agency (EPA) maximum contaminant level (MCL) for nitrate in drinking water of 10 mg/L nitrate-nitrogen (nitrate-N) (equivalent to 45 mg/L as nitrate) and the World Health Organization (WHO) guideline (WHO 2004b) of 50 mg/L as nitrate (equivalent to 11 mg/L as nitrate-N) were promulgated to protect against methemoglobinemia, or “blue baby syndrome,” to which infants are especially susceptible. The regulatory level is usually met for public water supplies, which are routinely monitored. Much less is known about private wells, which in the United States are usually required to be tested only when the well is constructed or when the property is sold. Some have suggested recently that the regulatory level for nitrate in drinking water is overly conservative (Avery 1999; L’hirondel and L’hirondel 2002). However, this discussion of the regulatory level has not thoroughly considered studies of other chronic health effects including cancer, adverse reproductive outcomes, and diabetes. Although a causal role for nitrate in these other health outcomes is not conclusive, recent studies that indicate possible adverse effects at nitrate levels below the MCL are of concern (Brender et al. 2004b; DeRoos et al. 2003; Ward et al. 1996; Weyer et al. 2001).
In recognition of the widespread contamination of drinking-water sources by nitrate and the potential for health effects in addition to methemoglobinemia, a symposium titled “Drinking Water Nitrate and Health: Recent Findings and Research Needs” took place at the annual meeting of the International Society for Environmental Epidemiology (1–4 August 2004, New York, New York, USA). Invited experts presented results from recent unpublished studies and summarized the state of knowledge on exposure and health effects of drinking-water nitrate, with a focus on cancer and adverse reproductive outcomes. This article summarizes the symposium discussions and recommends promising areas for future research. Specifically, we discuss the epidemiologic evidence for drinking-water nitrate and risk of specific cancers, adverse reproductive outcomes, and other health outcomes in the context of the current regulatory limit for nitrate in drinking water.
Nitrate Levels in Groundwater and Water Supplies
Nitrate is the most common chemical contaminant in the world’s groundwater aquifers (Spalding and Exner 1993). An estimated 42% of the U.S. population uses groundwater as their drinking-water supply (Hutson et al. 2004). In the United States, total nitrogen in streams and nitrate in groundwater are highest in agricultural areas, followed by urban areas and areas with mixed land use (Figure 1). The most recent data indicate that about 22% of domestic wells in agricultural areas of the United States exceeded the MCL (U.S. Geological Survey, unpublished data). In contrast, 3% of public supply wells in major aquifers (typical sources for public water supplies) exceed the MCL (U.S. Geological Survey, unpublished data).
The exposure picture is similar in the European Union. Public water supplies are largely below the WHO guideline; however, in some countries, private wells in rural areas have elevated nitrate concentrations reaching 10–15 times the recommended level (European Environment Agency 2003). Overall, nitrate levels exceeded the guideline in about one-third of the groundwater bodies for which data were available (European Environment Agency 2003). Several eastern European countries report high levels of nitrate contamination in a large proportion of private wells; for example, in Romania, 20% of 2,000 wells had nitrate levels > 23 mg/L as nitrate-N (Jedrychowski et al. 1997). Studies from other countries, including China, Botswana, Turkey, Senegal, and Mexico, report private well water levels that exceed the WHO guideline, in some instances at levels > 68 mg/L nitrate-N (WHO 2004a). Fertilizer is the main contributing factor in agricultural areas; however, nitrogen from human waste appears to be the most important source in urban areas lacking centralized water and sanitation systems. Although systematic information on nitrate levels in groundwater in other parts of the world is more limited, empirical modeling approaches have indicated that users of shallow wells in areas with high nitrogen inputs, well-drained soils, and unconsolidated rocks are most at risk of consuming high-nitrate groundwater (Nolan et al. 2002).
Methemoglobinemia
Ingested nitrate is reduced to nitrite, which binds to hemoglobin to form methemoglobin (MetHb). Methemoglobinemia occurs when elevated levels of MetHb (exceeding about 10%) interfere with the oxygen-carrying capacity of the blood. Infants are particularly susceptible to developing methemoglobinemia for several reasons, including their increased capacity to convert nitrate to nitrite and their lower levels of the enzyme cytochrome b5 reductase, which converts MetHb back to hemoglobin. Methemoglobinemia in infants fed formula made with well water with high nitrate levels was first reported in 1945 by Comly (1945). The regulatory level for nitrate in drinking-water supplies was determined after a survey of infant methemoglobinemia case reports in the United States indicated that no cases were observed at drinking-water nitrate levels < 10 mg/L nitrate-N (Walton 1951). Because an estimated 22% of domestic wells in agricultural regions of the United States exceed the nitrate MCL (U.S. Geological Survey, unpublished data), it is likely that significant numbers of infants are given water containing > 10 mg/L nitrate-N. Nevertheless, few cases of methemoglobinemia have been reported since the MCL was promulgated.
The risk of methemoglobinemia among infants depends on many factors other than the ingestion of nitrate in drinking water. Some foods and medications contain high levels of nitrate (Sanchez-Echaniz et al. 2001). Enteric infections, potentially caused by fecal bacteria contamination in wells, may lead to the endogenous production of nitrite, as evidenced by numerous published reports of infants with diarrhea and methemoglobinemia but no apparent exposure to exogenous MetHb-forming agents (Charmandari et al. 2001; Hanukoglu and Danon 1996; Levine et al. 1998; Wennmalm et al. 1993). The consumption of antioxidants such as vitamin C appears to be a protective factor. Finally, polymorphisms in the activity of cytochrome b5 reductase may mediate the effect of ingested nitrate or endogenously produced nitrite (Gupta et al. 1999).
Studies that have examined the relationship between nitrate levels in drinking water and MetHb levels in infants have produced mixed results (U.S. EPA 1991). The few experimental studies are largely negative; however, most of these studies evaluated low levels of drinking-water nitrate and included few infants. Cofactors such as diarrhea and respiratory diseases reportedly increase MetHb levels (Shearer et al. 1972; Shuval and Gruener 1972). An epidemiologic study in South Africa (Super et al. 1981) found an increase in MetHb levels in infants fed water with nitrate > 20 mg/L nitrate-N; however, clinical methemoglobinemia was rarely found. A protective effect of vitamin C intake on MetHb was noted (Super et al. 1981). More recently, a retrospective, nested case–control study in Romania found an association between nitrate exposure from drinking water and clinical methemoglobinemia, but also some evidence of an association with diarrheal disease (Zeman et al. 2002). Gupta and colleagues (1999) found cytochrome b5 reductase activities to be higher among those consuming water with high nitrate levels, indicating a level of adaptation to the consumption of high nitrate waters.
Recently, the role of nitrate exposure alone in causing methemoglobinemia has been questioned (Avery 1999; Fewtrell 2004; Hanukoglu and Danon 1996). Clearly, we need to better understand the interaction of factors that lead to methemoglobinemia to assess the relative importance of each factor and to identify the conditions under which exposure to nitrate in drinking water poses a risk of methemoglobinemia.
Nitrate Intake and Endogenous Formation of N-Nitroso Compounds
Nitrate is a precursor in the formation of N-nitroso compounds (NOC), a class of genotoxic compounds, most of which are animal carcinogens. In the human body, nitrate is a stable, inert compound that cannot be metabolized by human enzymes. However, the nitrate-reducing activity of commensal bacteria may convert nitrate into nitrite and other bioactive nitrogen compounds that affect physiological processes and human health. After ingestion, nitrate is readily absorbed from the upper gastrointestinal tract. Up to 25% is actively excreted in saliva, where about 20% is converted to nitrite by bacteria in the mouth (Spiegelhalter et al. 1976). This conversion can occur at other sites including the distal small intestine and the colon.
Under acidic conditions in the stomach, nitrite is protonated to nitrous acid (HNO2), which in turn spontaneously yields dinitrogen trioxide (N2O3), nitric oxide (NO), and nitrogen dioxide (NO2). NO is a bioactive compound known to play a role in vasodilatation and in defense against periodontal bacteria and other pathogens. N2O3, on the other hand, is a powerful nitrosating agent capable of donating NO+ to secondary and tertiary amines to form potentially carcinogenic N-nitrosamines (Leaf et al. 1989). Alternatively, HNO2 can be protonated to H2NO2, which reacts with amides to form N-nitrosamides. At neutral pH, nitrite can be reduced by bacterial activity to form NO, which can react with molecular oxygen to form the nitrosating compounds N2O3 and nitrogen tetroxide (N2O4). In addition to the acid-catalyzed and bacterial-catalyzed formation of nitrosating agents, inducible NO synthase activity of inflammatory cells can also produce NO (Ohshima and Bartsch 1994). Together, these three mechanisms of endogenous nitrosation account for an estimated 40–75% of the total human exposure to NOC (Tricker 1997). Other sources of human exposure include pre-formed NOC found in preserved meats and fish, beer, certain occupational exposures, and tobacco products (Tricker 1997).
Several studies support a direct relationship between nitrate intake and endogenous formation of NOC. High intake of drinking-water nitrate (above the MCL) is associated with an increased endogenous capacity to nitrosate proline (Mirvish et al. 1992; Moller et al. 1989). In addition, populations with high rates of esophageal and gastric cancer excrete high levels of N-nitrosoproline (Kamiyama et al. 1987; Lu et al. 1986). Nitrate intake at the acceptable daily intake level (3.67 mg/kg body weight, 0.84 mg/kg as nitrate-N) results in increased urinary excretion of NOC, particularly in combination with increased intake of dietary nitrosatable precursors (Vermeer et al. 1998). However, a Canadian population exposed to nitrate below the acceptable daily intake level showed no relationship between nitrate levels in drinking water and urinary nitrosamines (Levallois et al. 2000).
Factors that modulate endogenous nitrosation.
Although intake of high drinking-water nitrate is consistently associated with endogenous nitrosation capacity, intake of dietary nitrate is less likely to increase nitrosation, because of the presence of nitrosation inhibitors in vegetables, the major contributors to dietary nitrate intake (Bartsch et al. 1988; National Academy of Sciences 1981). Dietary compounds that inhibit endogenous nitrosation include vitamin C, which has the capacity to reduce HNO2 to NO, and alphatocopherol, which can reduce nitrite to NO. Several epidemiologic studies reported no association or inverse associations between dietary nitrate intake and human cancers (Boeing 1991; Forman 1987; Ward et al. 1996), which may be because of the antioxidants and nitrosation inhibitors in nitrate-containing foods (Bartsch et al. 1988). Inhibitory effects on nitrosation have also been described with betel nut extracts, ferulic and caffeic acid, garlic, coffee, and green tea polyphenols (Stich et al. 1984). In addition, nondietary factors such as the use of mouthwashes containing chlorhexidine can influence the endogenous nitrosating capacity (van Maanen et al. 1998).
Apart from the level of nitrosating agents, the level of nitrosatable precursors in the diet, which come predominantly from meat and fish, is a crucial factor in endogenous nitrosation. Dietary intakes of red and processed meat are of particular importance in the formation of fecal NOC (Bingham 1999; Bingham et al. 1996, 2002; Cross et al. 2003; Haorah et al. 2001). Higher consumption of red meat (600 vs. 60 g/day), but not white meat, resulted in a 3-fold increase in fecal NOC levels (Bingham et al. 1996). Colon cancer incidence is most consistently associated with consumption of red meat (beef, lamb, and pork), but not with poultry and fish (Bingham 1999). Dietary supplementation of a diet low in red meat with either heme iron or inorganic iron demonstrated that heme in particular was able to stimulate endogenous nitrosation (Cross et al. 2003), thereby providing a possible explanation for the differences in colon cancer risk between red and white meat consumption. Additionally, this linkage may be stronger for processed meat than for fresh meat because of the higher NOC and NOC precursor levels in processed meat.
Endogenous nitrosation can also be stimulated by inflammatory and other medical conditions. For instance, patients with bilharzia have an increased bladder cancer risk associated with increased urinary levels of nitrite and volatile nitrosamines, most likely generated by the reaction of inflammation-derived NO with amines present in the urine (Tricker et al. 1989). Inflammatory bowel disease is also related to both increased nitro-sation and cancer risk (Lashner et al. 1988). During inflammatory bowel disease, increased inducible NO synthase activity can produce excess NO, which is oxidized to nitrogen oxides and nitrite, which in turn react with nitrosatable precursors in colonic contents to produce NOC. Indeed, ulcerative colitis patients showed increased levels of inducible NO synthase in the colonic mucosa (Kimura et al. 1998) and of NO and nitrite in the colonic lumen (Lundberg et al. 1997; Roediger et al. 1990). Increased levels of fecal NOC have been found in patients with inflammatory bowel disease and in mice with chemically induced colitis (de Kok et al. 2005; Mirvish et al. 2003).
Health Effects Associated with Drinking-Water Nitrate
Cancer.
NOC are potent animal carcinogens, inducing tumors at multiple organ sites including the esophagus, stomach, colon, bladder, lympatics, and hematopoietic system (Bogovski and Bogovski 1981). NOC cause tumors in every animal species tested, and it is unlikely that humans are unaffected (Lijinsky 1986). The number of well-designed epidemiologic studies with individual exposure data and information on nitrosation inhibitors and precursors are few for any single cancer site, limiting the ability to draw conclusions about cancer risk.
Most studies have been ecologic in design, linking incidence or mortality rates to drinking-water nitrate levels at the town or county level. The early studies focused on stomach cancer mortality, and most used drinking-water nitrate measurements concurrent with the period of cancer mortality. Results were mixed, with some studies showing positive associations, many showing no association, and a few showing inverse associations (Cantor 1997). Recent ecologic studies of stomach cancer in Slovakia, Spain, and Hungary with historical measurements and exposure levels near or above the MCL have found positive correlations with stomach cancer incidence or mortality (Gulis et al. 2002; Morales-Suarez-Varela et al. 1995; Sandor et al. 2001). Two studies included other cancer sites. In Slovakia, incidence of non-Hodgkin lymphoma (NHL) and colon cancer was significantly elevated among men and women exposed to public supply nitrate levels of 4.5–11.3 mg/L nitrate-N (Gulis et al. 2002); there was no association with bladder and kidney cancer incidence. In Spain, there was a positive correlation between nitrate levels in public supplies and prostate cancer mortality, but no relation with bladder and colon cancer (Morales-Suarez-Varela et al. 1995).
In the past decade, several case–control and cohort studies have evaluated historical nitrate levels in public water supplies (largely < 10 mg/L nitrate-N) and risk of several cancers (Table 1). Some studies evaluated factors affecting nitrosation, such as vitamin C intake. A cohort study of older women in Iowa (USA) (Weyer et al. 2001) found a 2.8-fold and 1.8-fold risk of bladder and ovarian cancers, respectively, associated with the highest quartile (> 2.46 mg/L nitrate-N) of the long-term average nitrate levels at the current residence. They observed significant inverse associations for uterine and rectal cancer and no significant associations for NHL, leukemia, colon, rectum, pancreas, kidney, lung, and melanoma. Case–control studies of bladder (Ward et al. 2003), brain (Ward et al. 2004), colon and rectum (De Roos et al. 2003), and pancreas cancer (Coss et al. 2004) in Iowa found no association between cancer risk and average nitrate levels over almost 30 years. Each study evaluated the interaction between nitrosation inhibitors or NOC precursors and nitrate intake from drinking water. For colon cancer, there was a significant positive interaction between 10 or more years of exposure above 5 mg/L nitrate-N and both low vitamin C and high meat intake, factors likely to increase endogenous NOC formation (De Roos et al. 2003).
A case–control study of NHL in Nebraska (USA) (Ward et al. 1996) found a significant positive association between the average nitrate level in public water supplies over about 40 years and risk among men and women. In the highest quartile of nitrate (4.0 mg/L nitrate-N), risk was elevated 2-fold. However, a recent study of NHL in Iowa with similar exposure levels found no association (Ward et al. 2004). A case–control study of NHL in Minnesota (USA) (Freedman et al. 2000) with lower levels of nitrate found an inverse association among those with the highest level (> 1.5 mg/L nitrate-N). Case–control studies in Nebraska (Ward et al. 2004) and Germany (Steindorf et al. 1994) found no association with long-term average nitrate levels in public water supplies and adult brain cancer. The Nebraska study found no evidence of an interaction with vitamin C intake. A case–cohort analysis of stomach cancer within a cohort study in the Netherlands (van Loon et al. 1998) found no association with quintiles of water nitrate intake determined from public supply levels.
Specific NOC are transplacental neurocarcinogens in animal studies. A study of childhood brain cancer measured nitrate levels in water supplies using dipstick measurements, often many years after the pregnancy (Mueller et al. 2001). Measured levels of nitrate and nitrite were not associated with risk; however, women in western Washington State, one of the three study centers, who used private wells as their drinking-water source during the pregnancy had a significantly increased risk of brain cancer in their offspring.
Adverse reproductive outcomes.
In 1961, Schmitz described a possible relationship between high maternal MetHb levels and spontaneous abortion. Since then, at least 10 studies have examined the association between drinking-water nitrate and adverse reproductive outcomes. Table 2 summarizes these studies by location, study design, determination of water nitrate, and key findings. Few studies have been published regarding water nitrate and the outcomes of spontaneous abortions, stillbirths, premature birth, or intrauterine growth retardation. Results of these studies have been inconsistent, possibly indicating no true effect of water nitrate on reproductive outcomes at the levels evaluated in these studies. Alternatively, the inconsistencies may be due to the differing periods over which exposure was assessed, differing levels of water nitrate across studies, or differences in exposure to other cofactors.
Results of studies evaluating drinking-water nitrate and congenital malformations in offspring are also mixed (Table 2). Four studies (Arbuckle et al. 1988; Brender et al. 2004a, 2004b; Croen et al. 2001, Dorsch et al. 1984) found positive associations between drinking-water nitrate and congenital malformations, particularly malformations of the central nervous system, and specifically neural tube defects (NTDs). In each of these studies, water nitrate levels associated with increased risk of these defects were below the MCL, although the 95% confidence intervals (CIs) for some of the risk estimates were consistent with unity and varied by the source of water (groundwater, mixed, or surface). Two of these studies (Brender et al. 2004b; Croen et al. 2001) also examined dietary intake of nitrates and nitrates and NTDs and found minimal or no effect on risk. In a study of nitrosatable drug exposure and risk of NTDs (Brender et al. 2004b), drinking-water nitrates and dietary nitrites/total nitrites substantially modified the risk associated with this drug exposure during the periconceptional period; higher levels of nitrates in food or water significantly increased the risk of NTDs if women were exposed to such drugs.
Other health outcomes.
Animal studies suggest that nitrate at high doses can competitively inhibit iodine uptake and induce hypertrophic changes in the thyroid (Bloomfield et al. 1961). In a human biomonitoring study in the Netherlands, consumption of water with nitrate levels at or above the MCL was associated with thyroid hypertrophy (van Maanen et al. 1994) and genotoxic effects (van Maanen et al. 1996). Animal studies provide evidence that NOC can damage the pancreatic beta cells (Longnecker and Daniels 2001). Three epidemiologic studies (Kostraba et al. 1992; Parslow et al. 1997; van Maanen et al. 2000) that were ecologic in design found a positive correlation between drinking-water nitrate levels below the MCL and the incidence of type I childhood diabetes, although the association observed by van Maanen was not statistically significant. Other studies have found associations between water nitrate exposure and increased blood pressure (Pomeranz et al. 2000) and acute respiratory tract infections in children (Gupta et al. 2000).
Recommendations for Future Research
Experimental/human biomonitoring studies.
Endogenous nitrosation in humans has been demonstrated in relation to drinking-water nitrate ingestion at levels above the MCL. However, further studies are needed to determine the extent of endogenous nitrosation at intermediate drinking-water nitrate levels (5–10 mg/L as nitrate-N) and to clarify the role of nitrate from water versus food sources. Furthermore, the role of precursors and modulators of NOC formation should be more fully investigated. These future studies should be conducted among healthy individuals as well as individuals with medical conditions that increase endogenous nitrosation.
In view of the complex kinetics of NOC formation and the organ specificity of several of these compounds (Hodgson et al. 1980; Suzuki et al. 1999), more studies are needed to evaluate the relationship between nitrate intake and formation, metabolism, and excretion of NOC. Ideally, a physiologically based pharmacokinetic model should be developed as previously recommended (National Research Council 1995) to predict exposure to NOC from all sources of nitrate exposure (exogenous and endogenous), nitrite intake, the transformation of nitrate into nitrite, and antioxidant intake. However, this will require additional data on the formation of individual NOC as well as their respective toxicologic characteristics. The results of these investigations will reveal the value of different markers of NOC exposure in future epidemiologic studies. Future studies linking NOC exposure to early markers of effect or to the actual disease will clarify the role of endogenous nitrosation and NOC exposure as etiologic factors.
Because many NOC require α -hydroxylation by CYP2E1 for bioactivation and for formation of DNA adducts, it is important to investigate the influence of polymorphisms in the gene encoding for this enzyme. One study found that specific variants in this gene are linked to increased rectum cancer risk, particularly in subjects with high intake of red and processed meat, who are exposed to increased levels of NOC (Le Marchand et al. 2002). Moreover, gene expression levels of human CYP2E1 were related to cytotoxicity and DNA damage by nitrosamines in pancreatic beta-cell lines, suggesting that such gene environment interactions are also relevant in type 1 diabetes (Lees Murdock et al. 2004). These promising lines of research point to a possible interaction between drinking-water nitrate exposure and gene expression of and/or genetic variation in CYP2E1, which may also influence the risk of several adverse health outcomes associated with nitrate exposure.
Epidemiologic studies.
Methods must be developed and validated to improve estimates of current and historical exposure to nitrate via food and water, particularly for populations served by private wells, which are less likely to be routinely monitored. Future epidemiologic studies should integrate a) exposure assessment for nitrate intake from drinking water, nitrate and nitrite intake from the diet, and amines and amides from dietary and drug sources, b) endogenous exposure to NOC by analysis of relevant biological media (e.g., saliva, urine, feces), and c) reliable health risk markers (e.g., biomarkers of genotoxicity) or diagnosis of actual disease.
Future studies should include populations with well-characterized long-term exposures, including those who use private wells, which can have higher nitrate levels than public supplies. With the increasing availability of public water supply monitoring data (many U.S. states have almost 40 years of measurements), further detailed exposure assessment of populations using public supplies is also feasible. Drinking-water contaminants that may occur along with nitrate, such as agricultural pesticides, should also be evaluated. Geographic-based modeling efforts to predict the probability of high nitrate concentrations in groundwater, using information on nitrogen inputs from agricultural and urban sources (Nolan et al. 2002), is a promising approach for estimating drinking-water nitrate exposure for the population using private wells.
Additional studies of drinking-water nitrate and cancer are needed to follow up on the suggestive positive findings to date and to evaluate other cancer sites in which endogenously formed NOC may play a role. Studies of reproductive outcomes should address the exposure period most relevant for the specific outcome of interest. Maternal residential mobility between conception and birth may lead to misclassification of exposure if the water source at birth is used in studies of spontaneous abortions and congenital malformations. Studies must be of sufficient size to allow for examination of specific defects rather than groups of defects by system, because combining different defects might mask associations. More research is needed on the relation between water nitrate and the reproductive outcomes of spontaneous abortion, fetal death, premature birth, and intrauterine growth retardation.
In the design and analysis stage, future epidemiologic studies should consider factors that modulate endogenous nitrosation, as discussed above, to be able to evaluate potential interactions of water nitrate intake with these factors, thus providing stronger evidence for or against an association. In particular, studies of susceptible populations may be fruitful, and epidemiologic studies should be designed with sufficient power to evaluate risk among potentially susceptible subgroups. Such populations include patients with different forms of chronic inflammation (such as inflammatory bowel disease), patients infected with nitrate-reducing bacteria (such as in periodontal disease), those with low intake of vitamins and other known nitrosation inhibitors, or those with a history of high incidence of potentially NOC-related diseases. The people of Linxian County in China, for example, are known for their persistently low intake of several micronutrients and high risk of esophageal cancer (Blot et al. 1993). Such populations will likely benefit from preventive measures taken as a result of these investigations.
Conclusions
Adverse health effects from drinking-water nitrates are most likely the result of a complex interaction of the amount of nitrate ingested, the concomitant ingestion of nitrosating cofactors and precursors, and medical conditions of the host that may increase nitrosation. Furthermore, these effects may be attenuated by inhibitors of endogenous nitrosation such as vitamin C and alphatocopherol. We recommend that future studies take into account such complexities in understanding the relation between drinking-water nitrates and cancer, adverse reproductive outcomes, and other health outcomes.
Several authors (Avery 1999; L’hirondel and L’hirondel 2002) have questioned the importance of nitrate in drinking water as a risk factor for methemoglobinemia and have suggested that the current nitrate standard might be safely raised to 15–20 mg/L nitrate-N with no increase in methemoglobinemia cases. A better understanding of the conditions under which nitrate in drinking water poses a risk of methemoglobinemia is clearly needed, particularly in light of recent cases of methemoglobinemia associated with well water levels between 20 and 30 mg/L nitrate-N (Knobeloch et al. 2000). Most importantly, the role of nitrate as a risk factor for cancer and adverse reproductive outcomes must be more thoroughly explored before changes to nitrate water quality standards are considered.
We thank K. Cantor of the National Cancer Institute for his review of the manuscript.
Figure 1 Interquartile range of total nitrogen in streams and nitrate-N in groundwater in agricultural, urban, and mixed land use, and undeveloped areas of the United States. Upper bound of bar represents 90th percentile and lower bound represents 10th percentile. Along the top of the graph are the number of stream sampling stations and groundwater networks (group of wells in an aquifer).
Table 1 Analytic epidemiologic studies of drinking-water nitratea and cancer.
Reference, year, country Study design (case–control, cohort) Regional description Years of cancer ascertainment Exposure descriptiona Cancer sites included Summary of findings
Coss et al. 2004 USA Population-based case–control Average nitrate level in public supplies 1960–1987 (highest quartile > 2.8 mg/L); Years of exposure ≥ 7.5 and 10 mg/L Pancreas No significant associations with quartiles of average nitrate or number of years ≥ 7.5 or 10 mg/L
Incidence
Iowa 1986–1989
DeRoos et al. 2003 USA Population-based case–control Average nitrate level in public supplies 1960–1987 categorized into four levels (lowest: ≤ 1.0; highest: > 5mg/L); Years of exposure > 5 and >10 mg/L Colon No association with average level, years > 5 and 10 mg/L; Significantly elevated risk among subgroups with below median vitamin C intake or above median meat intake and 10 or more years > 5 mg/L
Incidence Rectum
Iowa 1986–1989
Freedman 2000 USA Population-based case–control Average nitrate level in public water supplies 1947–1980 (157 towns) categorized into three levels: ≤ 0.5, > 0.5 to ≤ 1.5, > 1.5 mg/L Non-Hodgkin lymphoma No increase risk with increasing exposure level. OR for > 1.5 mg/L (three cases, four controls) was 0.3 (95% CI, 0.1–0.9).
Incidence
Minnesota excluding four largest cities 1980–1982
Mueller et al. 2001 USA Population-based case–control 19 counties in San Francisco, California, area and western Washington State 1984–1990 Water source (private well, public supply) during pregnancy; dipstick measurements of nitrate and nitrite for those still living at residence during pregnancy Childhood brain No overall association with water source. Well use in western Washington State increased risk (OR = 2.6; 95% CI, 1.3–5.2); well use in Los Angeles inversely associated with risk (OR = 0.2; 95% CI, 0.1–0.8)
Steindorf et al. 1994 Germany Population-based case–control Nitrate levels in municipal supplies after 1970 (highest quartile: > 5.7 mg/L) Brain No association with average nitrate level
Incidence
Rhein-Neckar-Odenwald area 1987–1988
Van Loon et al. 1998 Netherlands Prospective cohort Nitrate intake from public supplies in 1986 and intake of tap water (quintiles; mean level in highest quintile: 3.7 mg/day) Stomach No association with quintiles of water nitrate intake (highest quintile: RR = 0.88)
Incidence 1986–1992
Ward et al. 1996 USA Population-based case–control Average nitrate level in public water supplies 1945–early 1980s categorized into quartiles (lowest: < 1.6; highest: ≥ 4.0 mg/L); Ever exposure ≥ 10 mg/L Non-Hodgkin lymphoma Significant positive trend with increasing quartiles: OR highest quartile = 2.0 (95% CI, 1.1–3.6)
Incidence
66 counties in eastern Nebraska 1983–1986
Ward et al. 2003 USA Population-based case–control Average nitrate level in public water supplies 1960–1987 (highest quartile men: 3.1 mg/L; women: 2.4 mg/L); Years of exposure ≥ 10 mg/L Bladder Inverse association with quartiles of average level among men; no association among women. Similar results for years ≥ 10 mg/L
Incidence
Iowa 1986–1989
Ward et al. 2004 USA Population-based case–control Average nitrate level in public water supplies Brain (gliomas) No association with quartiles of the average nitrate level
Incidence 1988–1993 1960–1986
66 counties of eastern Nebraska
Weyer et al. 2001 USA Prospective cohort Average nitrate level (1955–1988) in public water supplies for residence at enrollment (highest quartile: > 2.46 mg/L) Non-Hodgkin lymphoma, leukemia, colon, rectum, pancreas, kidney, bladder, breast, ovary, uterine corpus, lung and bronchus, melanoma Positive associations with average nitrate level for bladder (highest quartile OR = 2.83) and ovary (OR = 1.84) and inverse associations for uterus (highest quartile OR = 0.55) and rectal cancer (OR = 0.47)
Incidence 1986–1998
Iowa
OR, odds ratio.
a Nitrate levels presented in the original publications as mg/L nitrate were converted to mg/L nitrate-N.
Table 2 Studies of the relation between drinking-water nitratea and reproductive outcomes.
Reference, study population, study design Measurement of water nitrate Reproductive outcome Reported findings
Aschengrau et al. 1989
Massachusetts (USA) residents
Hospital case–control study Matched maternal residence at pregnancy outcome to results of tap water samples SBs through 27 weeks of gestation OR of 0.5 for SB with exposure to water nitrate levels of 0.1–5.5 mg/L relative to nondetectable levels
Grant et al. 1996
Indiana (USA)
Cluster investigation Wells tested for nitrates after cluster reported SBs Water nitrate above U.S. EPA MCL for women with SBs
Aschengrau et al. 1993
Massachusetts (USA) residents
Hospital case–control study Matched maternal residence during pregnancy or outcome to results of tap water samples Congenital anomalies, stillbirths, neonatal deaths Neither stillbirths nor congenital anomalies associated with detectable levels of water nitrate (0.2–4.5 mg/L); small positive association between water nitrates and neonatal deaths.
Super et al. 1981
South West Africa
Cross-sectional study Water sample taken from well used at time of home visit Spontaneous premature labor
Size of infant at birth No association between water from high nitrate regions and prematurity or size of infant
Bukowski et al. 2001
Prince Edward Island, Canada
Case–control study Residential postal code at time of delivery linked to nitrate level exposure map IUGR
Premature birth Dose–response relation between nitrate level and ORs for IUGR and prematurity
Scragg et al. 1982
Dorsch et al. 1984
South Australia
Case–control study Address at delivery linked to sources of water and data on nitrates Congenital malformations Elevated OR for any congenital malformation (2.8); malformations of the CNS (3.5); musculoskeletal system (2.9) if primarily drank groundwater.
Elevated ORs for congenital malformations associated with nitrate levels ≥ 5 mg/L relative to nitrate levels < 5 mg/L
Arbuckle et al. 1988
New Brunswick, Canada
Case–control study Collected and analyzed a water sample at maternal residence at time of index birth Congenital malformations of the CNS OR of 2.3 for CNS malformations with exposure to nitrate 26 mg/L relative to baseline of 0.1 mg/L
Ericson et al. 1988
All deliveries in Sweden
Case-control study Earliest known maternal address linked to water nitrate results NTDs Average water nitrate similar between cases and controls
Croen et al. 2001
California (USA)
Case–control study Linked periconceptional addresses to water companies and databases NTDs Exposure to water nitrates > 45 mg/L associated with anencephaly (OR 4.0) but not with spina bifida; increased risks for anencephaly at water nitrate levels below U.S. EPA MCL among groundwater drinkers only; dietary nitrate and nitrite not associated with NTDs
Cedergren et al. 2002
Ostergotland County, Sweden
Retrospective cohort study Linked periconceptional or early pregnancy address to water supplies using a geographic information system Any congenital cardiac defect Weak association (OR 1.2) between water nitrate ≥ 2 mg/L and cardiac malformations
Brender et al. 2004a
Brender et al. 2004b
Texas (USA)
Counties along Texas–Mexico border
Case–control study Usual periconceptional drinking-water source tested for nitrates NTDs OR of 1.9 if water nitrates ≥ 3.52 mg/L; increased water nitrate associated with spina bifida (OR 7.8) but not with anencephaly (OR 1.0); slightly inverse relation between dietary nitrite, total nitrite intake and NTDs
Abbreviations: CNS, central nervous system; IUGR, intrauterine growth retardation; SB, spontaneous abortion.
a Nitrate units are mg/L as nitrate.
==== Refs
References
Arbuckle TE Sherman GJ Corey PN Walters D Lo B 1988 Water nitrates and CNS birth defects: a population-based case-control study Arch Environ Health 43 162 167 3377550
Aschengrau A Zierler S Cohen A 1989 Quality of community drinking water and the occurrence of spontaneous abortion Arch Environ Health 44 283 290 2554824
Aschengrau A Zierler S Cohen A 1993 Quality of community drinking water and the occurrence of late adverse pregnancy outcomes Arch Environ Health 48 105 113 8476301
Avery AA 1999 Infantile methemoglobinemia: reexamining the role of drinking water nitrates Environ Health Perspect 107 1 8 10068291
Bartsch H Ohshima H Pignatelli B 1988 Inhibitors of endogenous nitrosation. Mechanisms and implications in human cancer prevention Mutat Res 202 307 324 3057363
Bingham SA 1999 High-meat diets and cancer risk Proc Nutr Soc 58 243 248 10466162
Bingham SA Hughes R Cross AJ 2002 Effect of white versus red meat on endogenous N -nitrosation in the human colon and further evidence of a dose response J Nutr 132 3522S 3525S 12421881
Bingham SA Pignatelli B Pollock JR Ellul A Malaveille C Gross G 1996 Does increased endogenous formation of N -nitroso compounds in the human colon explain the association between red meat and colon cancer? Carcinogenesis 17 515 523 8631138
Bloomfield RA Welsch CW Garner GB Muhrer ME 1961 Effect of dietary nitrate on thyroid function Science 134 1690 13870154
Blot WJ Li J Taylor PR Guo W Dawsey S 1993 Nutrition intervention trials in Linxian, China: supplementation with specific vitamin/mineral combinations, cancer incidence, and disease-specific mortality in the general population J Natl Cancer Inst 85 1483 1491 8360931
Boeing H 1991 Epidemiological research in stomach cancer: progress over the last ten years J Cancer Res Clin Oncol 117 133 143 2036128
Bogovski P Bogovski S 1981 Animal species in which N -nitroso compounds induce cancer Int J Cancer 27 471 474 7275353
Brender J Olive J Felkner M Suarez L Hendricks K Marckwardt W 2004a Intake of nitrates and nitrites and birth defects in offspring [Abstract] Epidemiology 15 S184
Brender JD Olive JM Felkner M Suarez L Marckwardt W Hendricks KA 2004b Dietary nitrites and nitrates, nitrosatable drugs, and neural tube defects Epidemiology 15 330 336 15097014
Bukowski J Somers G Bryanton J 2001 Agricultural contamination of groundwater as a possible risk factor for growth restriction or prematurity J Occup Environ Med 43 377 383 11322099
Cantor KP 1997 Drinking water and cancer Cancer Causes Control 8 292 308 9498894
Cedergren MI Selbing AJ Lofman O Kallen BAJ 2002 Chlorination byproducts and nitrate in drinking water and risk for congenital cardiac defects Environ Res 89 124 130 12123645
Charmandari E Meadows N Patel M Johnston A Benjamin N 2001 Plasma nitrate concentrations in children with infectious and noninfectious diarrhea J Pediatr Gastroenterol Nutr 32 423 427 11396807
Comly H 1945 Cyanosis in infants caused by well water JAMA 129 112 116
Coss A Lynch C Cantor KP Ward MH 2004 Pancreatic cancer and drinking water and dietary sources of nitrate and nitrite Am J Epidemiol 159 693 701 15033647
Croen LA Todoroff K Shaw GM 2001 Maternal exposure to nitrate from drinking water and diet and risk for neural tube defects Am J Epidemiol 153 325 331 11207149
Cross AJ Pollock JR Bingham SA 2003 Heam, not protein or inorganic iron, is responsible for endogenous intestinal N-nitrosation arising from red meat Cancer Res 63 2358 2360 12750250
De Kok TM Engels LG Moonen E Kleinjans JC 2005 Inflammatory bowel disease stimulates formation of carcinogenic N -nitroso compounds Gut 54 731 15831929
DeRoos AJ Ward MH Lynch CF Cantor KP 2003 Nitrate in public water systems and the risk of colon and rectum cancers Epidemiology 14 640 649 14569178
Dorsch MM Scragg RKR McMichael AJ Baghurst PA Dyer KF 1984 Congenital malformations and maternal drinking water supply in rural South Australia: a case-control study Am J Epidemiol 119 473 486 6711537
Ericson A Kallen B Lofkvist E 1988 Environmental factors in the etiology of neural tube defects: a negative study Environ Res 45 38 47 3338434
European Environment Agency 2003. Europe’s Environment: The Third Assessment. Environmental Assessment Report No. 10. Copenhagen:European Environment Agency.
Fewtrell L 2004 Drinking-water nitrate, methemoglobinemia, and global burden of disease: a discussion Environ Health Perspect 112 1371 1374 15471727
Fields S 2004 Global nitrogen: cycling out of control Environ Health Perspect 112 A557 A563
Forman D 1987 Dietary exposure to N -nitroso compounds and the risk of human cancer Cancer Surv 6 719 738 3330686
Freedman DF Cantor KP Ward MH Helzlsouer K 2000 Nitrates in drinking water and non-Hodgkin’s lymphoma: a population-based case-control study of men in Minnesota Arch Environ Health 55 326 329 11063407
Grant W Steele G Isiorho SA 1996 Spontaneous abortions possibly related to ingestion of nitrate-contaminated well water—LaGrange County, Indiana, 1991–1994 MMWR Morb Mortal Wkly Rep 45 569 572 9132576
Gulis G Czompolyova M Cerhan JR 2002 An ecologic study of nitrate in municipal drinking water and cancer incidence in Trnava District, Slovakia Environ Res 88 182 187 12051796
Gupta SK Gupta RC Gupta AB Seth AK Bassin JK Gupta A 2000 Recurrent acute respiratory infections in areas with high nitrate concentrations in drinking water Environ Health Perspect 108 363 366 10753096
Gupta SK Gupta RC Seth AK Gupta AB Bassin JK Gupta A 1999 Adaptation of cytochrome b5 reductase activity and methemoglobinemia in areas with a high nitrate concentration in drinking-water Bull WHO 77 749 753 10534899
Hanukoglu A Danon PN 1996 Endogenous methemoglobinemia associated with diarrheal disease in infancy J Pediatr Gastroenterol Nutr 23 1 7 8811515
Haorah J Zhou L Wang X Xu G Mirvish SS 2001 Determination of total N-nitroso compounds and their precursors in frankfurters, fresh meat, dried salted fish, sauces, tobacco, and tobacco smoke particulates J Agric Food Chem 49 6068 6078 11743810
Hodgson RM Wiessler M Kleihues P 1980 Preferential methylation of target organ DNA by the oesophageal carcinogen N -nitrosomethylbenzylamine Carcinogenesis 1 861 866 11219858
Hutson SS Barber NL Kenny JF Linsey KS Lumia DS Maupin MA 2004. Estimated Use of Water in the United States in 2000. USGS Circular 1268. Denver, CO:U.S. Geological Survey.
Jedrychowski W Maugeri U Bianchi I 1997 Environmental pollution in central and eastern European countries: a basis for cancer epidemiology Rev Environ Health 12 1 23 9128908
Kamiyama S Ohshima H Shimada A Saito N Bourgade MC Ziegler P 1987 Urinary excretion of N -nitrosoamino acids and nitrate by inhabitants in high- and low-risk areas for stomach cancer in northern Japan IARC Sci Publ 84 479 502
Kimura H Hokari R Miura S Shigematsu T Hirokawa M Akiba Y 1998 Increased expression of an inducible isoform of nitric oxide synthase and the formation of peroxynitrite in colonic mucosa of patients with active ulcerative colitis Gut 42 180 187 9536941
Knobeloch L Salna B Hogan A Postle J Anderson H 2000 Blue babies and nitrate-contaminated well water Environ Health Perspect 108 675 678 10903623
Kostraba JN Gay EC Rewers M Hamman RF 1992 Nitrate levels in community drinking waters and risk of IDDM, an ecologic analysis Diabetes Care 15 1505 1508 1468277
Lambert KF Driscoll C 2003. Nitrogen Pollution: From the Sources to the Sea. Hanover, NH:Hubbard Brook Research Foundation.
Lashner BA Hanauer SB Silverstein MD 1988 Optimal timing of colonoscopy to screen cancer in ulcerative colitis Ann Intern Med 108 274 278 3124681
Leaf CD Wishnok JS Tannenbaum SR 1989 l -Arginine is a precursor for nitrate biosynthesis in humans Biochem Biophys Res Commun 8 1032 1037 2783109
Lees Murdock DJ Barnett YA Barnett CR 2004 DNA damage and cytotoxicity in pancreatic beta-cells expressing human CYP2E1 Biochem Pharmacol 68 523 530 15242818
Le Marchand L Donlon T Seifried A Wilkens LR 2002 Red meat intake, CYP2E1 genetic polymorphisms, and colorectal cancer risk Cancer Epidemiol Biomarkers Prev 11 1019 1024 12376502
Levallois P Ayotte P Van Maanen JM Desrosiers T Gingras S Dallinga JW 2000 Excretion of volatile nitrosamines in a rural population in relation to food and drinking water consumption Food Chem Toxicol 38 1013 1019 11038239
Levine JJ Pettei MJ Valderrama E Gold DM Kessler BH Tractman H 1998 Nitric oxide and inflammatory bowel disease: evidence for local intestinal production in children with active colonic disease J Pediatr Gastroenterol Nutr 26 34 38 9443117
L’hirondel J L’hirondel J-L 2002. Nitrate and man: toxic, harmless, or beneficial? Wallingford, Oxfordshire, UK: CABI Publishing.
Lijinsky W 1986 The significance of N -nitroso compounds as environmental carcinogens J Environ Sci Health C4 1 45
Longnecker MP Daniels JL 2001 Environmental contaminants as etiologic factors for diabetes Environ Health Perspect 109 871 876 11744505
Lu SH Ohshima H Fu HM Tian Y Li FM Blettner M 1986 Urinary excretion of N -nitrosoamino acids and nitrate by inhabitants of high- and low-risk areas for esophageal cancer in Northern China: endogenous formation of nitrosoproline and its inhibition by vitamin C Cancer Res 46 1485 1491 3943105
Lundberg JO Herulf M Olesen M Bohr J Tysk C Wiklund NP 1997 Increased nitric oxide production in collagenous and lymphocytic colitis Eur J Clin Invest 27 869 871 9373768
Mirvish SS Grandjean AC Moller H Fike S Maynard T Jones L 1992 N -nitrosoproline excretion by rural Nebraskans drinking water of varied nitrate content Cancer Epidemiol Biomarkers Prev 1 455 461 1302557
Mirvish SS Haorah J Zhou L Hartman M Morris CR Clapper ML 2003 N -nitroso compounds in the gastrointestinal tract of rats and in the feces of mice with induced colitis or fed hot dogs or beef Carcinogenesis 24 595 603 12663523
Moller H Landt J Pedersen E Jensen P Autrup H Jensen O 1989 Endogenous nitrosation in relation to nitrate exposure from drinking water and diet in a Danish rural population Cancer Res 49 3117 3121 2720669
Morales-Suarez-Varela MM Llopis-Gonzalez A Tejerizo-Perez ML 1995 Impact of nitrates in drinking water on cancer mortality in Valencia, Spain Eur J Epidemiol 11 15 21 7489769
Mueller BA Newton K Holly EA Preston-Martin S 2001 Residential water source and the risk of childhood brain tumors Environ Health Perspect 109 551 556 11445506
National Academy of Sciences 1981. The Health Effects of Nitrate, Nitrite, and N-Nitroso Compounds. Washington, DC:National Academy of Sciences.
National Research Council 1995. Nitrate and nitrite in drinking water. Washington DC:National Academy Press.
Nolan BT Hitt KJ Ruddy BC 2002 Probability of nitrate contamination of recently recharged groundwaters in the conterminous United States Environ Sci Technol 36 2138 2145 12038822
Ohshima H Bartsch H 1994 Chronic infections and inflammatory processes as cancer risk factors: possible role of nitric oxide in carcinogenesis Mutat Res 305 253 264 7510036
Parslow RC McKinney PA Law GR Staines A Williams R Bodansky HJ 1997 Incidence of childhood diabetes mellitus in Yorkshire, northern England, is associated with nitrate in drinking water: an ecologic analysis Diabetologia 40 550 556 9165223
Pomeranz A Korzets Z Vanunu D Krystal H 2000 Elevated salt and nitrate levels in drinking water cause an increase of blood pressure in schoolchildren Kidney Blood Press Res 23 400 403 11070420
Roediger WE Lawson MJ Radcliffe BC 1990 Nitrite from inflammatory cells: a cancer risk factor in ulcerative colitis Dis Colon Rectum 33 1034 1036 2242698
Sanchez-Echaniz J Benito-Fernandez J Mintegui-Raso S 2001 Methemoglobinemia and the consumption of vegetables in infants Pediatrics 107 1024 1028 11331681
Sandor J Kiss I Farkas O Ember I 2001 Association between gastric cancer mortality and nitrate content of drinking water: ecological study on small area inequalities Eur J Epidemiol 17 443 447 11855578
Schmitz JT 1961 Methemoglobinemia—a cause of abortions? Obstet Gynecol 17 413 414 13748137
Scragg RK Dorsch MM McMichael AJ Baghurst PA 1982 Birth defects and household water supply. Epidemiological studies in the Mount Gambier region of South Australia Med J Aust 2 577 579 7162445
Shearer LA Goldsmith JR Young C Kearns OA Tamplin BR 1972 Methemoglobin levels in infants in an area with high nitrate water supply Am J Public Health 62 1174 1180 5068711
Shuval HI Gruener N 1972 Epidemiological and toxicological aspects of nitrates and nitrites in the environment Am J Public Health 62 1045 1052 5046442
Simon C Manzke H Kay H Mrowetz G 1964 On the incidence, pathogenesis and prevention of methemoglobinemia caused by nitrites [In German] Z Kinderheilkd 91 124 138 14331887
Spalding RF Exner ME 1993 Occurrence of nitrate in ground-water—a review J Environ Qual 22 392 402
Spiegelhalter B Elsenbrand G Preussman R 1976 Influence of dietary nitrate on nitrite content of human saliva: possible relevance to in vivo formation of N -nitroso compounds Food Cosmet Toxicol 14 545 548 1017769
Steindorf K Schlehofer B Becher H Hornig G Wahrendorf J 1994 Nitrate in drinking water. A case-control study on primary brain tumours with an embedded drinking water survey in Germany Int J Epidemiol 23 451 457 7960368
Stich HF Dunn BP Pignatelli B Ohshima H Bartsch H 1984 Dietary phenolics and betel nut extracts as modifiers of N -nitrosation in rat and man IARC Sci Publ 57 213 222 6533010
Super M Heese HDV MacKenzie D Dempster WS Du Plessis J Ferreira JJ 1981 An epidemiological study of well-water nitrates in a group of South West African/Namibian infants Water Res 15 1265 1270
Suzuki T Itoh S Nakajima M Hachiya N Hara T 1999 Target organ and time-course in the mutagenicity of five carcinogens in MutaMouse: a summary report of the second collaborative study of the transgenic mouse mutation assay by JEMS/MMS Mutat Res 444 259 268 10521667
Tricker AR 1997 N -nitroso compounds and man: sources of exposure, endogenous formation and occurrence in body fluids Eur J Cancer Prev 6 226 268 9306073
Tricker AR Mostafa MH Spiegelhaler B Pruessmann R 1989 Urinary excretion of nitrate, nitrate and N -nitroso compounds in Schistosomiasis and bilharzia bladder cancer patients Carcinogenesis 10 547 552 2924399
U.S. EPA 1991. Integrated Risk Information System (IRIS): Nitrate (CASRN 14797-55-8). Washington, DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/iris/subst/0076.htm [accessed 31 May 2005].
van Loon AJ Botterweck AA Goldbohm RA Brants HA van Klaveren JD van den Brandt PA 1998 Intake of nitrate and nitrite and the risk of gastric cancer: a prospective cohort study Br J Cancer 78 129 135 9662263
van Maanen JM Pachen DM Dallinga JW Kleinjans JC 1998 Formation of nitrosamines during consumption of nitrate and amine-rich foods, and the influence of mouthwashes Cancer Detect Prev 22 204 212 9618041
van Maanen JM van Dijk A Mulder K de Baets MH Menheere PC van der Heide D 1994 Consumption of drinking water with high nitrate levels causes hypertrophy of the thyroid Toxicol Lett 72 365 374 8202954
van Maanen JM Welle IJ Hageman G Dallinga GW Mertens PL Kleinjans JC 1996 Nitrate contamination of drinking water: relationship with HPRT variant frequency in lymphocyte DNA and urinary excretion of N -nitrosamines Environ Health Perspect 104 522 528 8743440
van Maanen JMS Albering HJ de Kok TMCM van Breda SGJ Curfs DMJ Vermeer ITM 2000 Does the risk of childhood diabetes mellitus require revision of the guideline values for nitrate in drinking water? Environ Health Perspect 108 457 461 10811574
Vermeer I Pachen DM Dallinga JW Kleinjans JC van Maanen JM 1998 Volatile N -nitrosamine formation after intake of nitrate at the ADI level in combination with an amine-rich diet Environ Health Perspect 106 459 463 9681972
Walton G 1951 Survey of literature relating to infant methemoglobinemia due to nitrate-contaminated water Am J Public Health 41 986 996 14847023
Ward MH Cantor KP Cerhan J Lynch CF Hartge P 2004 Drinking water nitrate and cancer: results from recent studies in the Midwestern United States [Abstract] Epidemiology 15 S214
Ward MH Cantor KP Riley D Merkle S Lynch CF 2003 Nitrate in public water supplies and risk of bladder cancer Epidemiology 14 183 190 12606884
Ward MH Mark SD Cantor KP Weisenburger DD Correa-Villasenor A Zahm SH 1996 Drinking water nitrate and the risk of non-Hodgkin’s lymphoma Epidemiology 7 465 471 8862975
Wennmalm A Benthin G Edlund A Jungersten L Kieler-Jensen N Lundlen S 1993 Metabolism and excretion of nitric oxide in humans: an experimental and clinical study Circ Res 73 1121 1127 8222083
Weyer PJ Cerhan JR Kross BC Hallberg GR Kantamneni J Breuer G 2001 Municipal drinking water nitrate level and cancer risk in older women: the Iowa Women’s Health Study Epidemiology 12 327 338 11338313
WHO 2004a. Nitrates and Nitrites in Drinking-Water. WHO/SDE/WSH/04.08/56. Rolling revision of the WHO guidelines for drinking-water quality. Draft for review and comments. Geneva:World Health Organization. Available: http://www.who.int/water_sanitation_health/dwq/chemicals/en/nitratesfull.pdf [accessed 31 May 2005].
WHO 2004b. Guidelines for Drinking Water Quality. 3rd Ed, Vol. 1, Recommendations. Geneva:World Health Organization.
Zeman CL Kross B Vlad M 2002 A nested case-control study of methemoglobinemia risk factors in children of Transylvania, Romania Environ Health Perspect 110 817 822 12153765
|
16263519
|
PMC1310926
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 23; 113(11):1607-1614
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8043
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8197ehp0113-00161516263520ResearchWorkgroup Report: Biomonitoring Study Design, Interpretation, and Communication—Lessons Learned and Path Forward Bates Michael N. 1Hamilton Joshua W. 2LaKind Judy S. 345Langenberg Patricia 5O’Malley Michael 6Snodgrass Wayne 71 School of Public Health, Division of Environmental Health Sciences, University of California, Berkeley, California, USA2 Center for Environmental Health Sciences and Department of Pharmacology and Toxicology, Dartmouth Medical School, Hanover, New Hampshire, USA3 LaKind Associates, LLC, Catonsville, Maryland, USA4 Milton S. Hershey Medical Center, Penn State College of Medicine, Hershey, Pennsylvania, USA5 Department of Epidemiology and Preventive Medicine, School of Medicine, University of Maryland, Baltimore, Maryland, USA6 Employee Health Services, University of California, Davis, California, USA7 University of Texas Medical Branch, Galveston, Texas, USAAddress correspondence to J.S. LaKind, LaKind Associates, LLC, 106 Oakdale Ave., Catonsville, MD 21228 USA. Telephone: (410) 788-8639. Fax: (410) 788-1971. E-mail:
[email protected]. LaKind is currently conducting biomonitoring-related research and related consulting services supported by RFHEE, the U.S. Environmental Protection Agency, and the Chlorine Chemistry Council. The other authors declare they have no competing financial interests.
11 2005 6 7 2005 113 11 1615 1621 12 4 2005 6 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Human biomonitoring investigations have provided data on a wide array of chemicals in blood and urine and in other tissues and fluids such as hair and human milk. These data have prompted questions such as a) What is the relationship between levels of environmental chemicals in humans and external exposures? b) What is the baseline or “background” level against which individual levels should be compared? and c) How can internal levels be used to draw conclusions about individual and/or population health? An interdisciplinary panel was convened for a 1-day workshop in November 2004 with the charge of focusing on three specific aspects of biomonitoring: characteristics of scientifically robust biomonitoring studies, interpretation of human biomonitoring data for potential risks to human health, and communication of results, uncertainties, and limitations of biomonitoring studies. In this report we describe the recommendations of the panel.
biomonitoringcommunicationdesignhuman healthinterpretationspecimen archiving
==== Body
Environmental health sciences focus on the relationship between exposures to environmental chemicals of concern and their relationship to health outcomes. The traditional method for assessing human exposures to environmental chemicals is to estimate, by empirical or modeling methods, the concentrations of chemicals of potential concern in environmental media (air, water, soil, food), and then combine this information with estimates of human exposure (e.g., estimates of daily consumption of tap water) to determine a dose. However, as analytic techniques have evolved, there has been an increasing focus on development and use of human biomonitoring (i.e., measurements of levels of environmental chemicals in human fluids such as blood, urine, or milk, and in tissues such as hair, nails, and fat) for evaluating exposure. These data have often supplemented or even supplanted estimates of exposure based on environmental measures.
Biomonitoring has been used for several decades for certain chemicals, such as for lead in blood and cotinine in urine. More recently, biomonitoring has provided data on levels of a much wider array of chemicals in various human fluids and tissues. In the United States, a systematic program of biomonitoring by the Centers for Disease Control and Prevention (CDC) has resulted in its National Report on Human Exposure to Environmental Chemicals (CDC 2003). This and other biomonitoring-based research have produced a substantial database on levels of environmental chemicals in humans. However, important questions remain: What is the relationship between these internal levels and external exposures? What is the baseline or “background” level against which individual levels should be compared? And how can internal levels be used to draw conclusions about individual and/or population health?
Uncertainties related to the relationships of exposure with internal dose and of internal dose with potential for adverse health effects have been described by the CDC (CDC 2003) and others (LaKind et al. 2005; Sexton et al. 2003; Stokstad 2004). These uncertainties were also highlighted at a recent workshop on environmental chemicals in human milk (LaKind 2005), during which a multidisciplinary group addressed questions regarding interpretation of human milk biomonitoring data for both the health of the mother and the breastfeeding infant. Recognizing that individuals from an array of disciplines have been grappling with various aspects of biomonitoring and that these disparate disciplines bring different perspectives to the table, the Research Foundation for Health and Environmental Effects (RFHEE; a co-sponsor of the human milk biomonitoring workshop) convened an interdisciplinary panel for a 1-day workshop on 13 November 2004 with the charge of focusing on three specific aspects of biomonitoring: characteristics of scientifically robust biomonitoring studies, interpretation of human biomonitoring data for potential risks to human health, and communication of results, uncertainties, and limitations of biomonitoring studies. The RFHEE sought panel members with expertise in medicine, toxicology, epidemiology, biostatistics, and risk assessment (the authors of this paper formed the workshop panel). During the workshop, the panel drew from the fields of medicine and occupational health, which have a long history of research on biomonitoring [National Institute for Occupational Safety and Health (NIOSH) 2004], as well as on the interpretation of the implications of biomonitoring results for individuals. In this report, we describe the recommendations reached during the workshop regarding biomonitoring study design, interpretation, and communication.
Recommendations of the Workshop Panel
Biomonitoring Study Design
The design of any scientific study depends on its goals and hypotheses. However, fundamental features that make for scientifically robust and credible studies exist. The panel focused on key research needs and recommendations for ensuring that the goals, hypotheses, and study design parameters are realistic, transparent, and scientifically robust.
Investigators need to gain not only approval but also acceptance for human studies.
Biomonitoring studies involving human subjects “must be conducted in a way that protects subjects’ rights and well-being,” and must have the oversight of an institutional review board (IRB), as described by the Common Rule, the International Conference on Harmonisation Good Clinical Practice guidelines, and the Declaration of Helsinki (LaKind et al. 2005). Nevertheless, biomonitoring studies may prove controversial even when approved by one or more IRBs. A recent example is a U.S. Environmental Protection Agency (EPA) study to monitor children’s exposure to pesticides [the Children’s Environmental Exposure Research Study (CHEERS); CHEERS 2005]. Although the U.S. EPA study involved monitoring rather than experimental administration of pesticides and had been approved by four separate IRBs, criticism of the CHEERS study raised by nongovernmental organizations focused on payments of up to $1,000 to subject families. This was criticized as undue inducement; criticism was also directed at partial industry funding of the study (Environmental Working Group 2004; Pesticide Action Network Updates Service 2004). In response, the U.S. EPA has cancelled the study (U.S. EPA 2005). This example highlights the need for thorough institutional review of human biomonitoring studies and the need for stakeholder acceptance, even after IRB approval is complete. Recommendations regarding ethical restrictions for biomonitoring studies have been developed (Oleskey et al. 2004). Although these were developed for pesticides, many of the recommendations have broader applicability. Researchers should also anticipate controversy from organizations and individuals who may not accept the authority of the IRBs to judge the ethical questions raised by proposed research plans.
Because human tissue banking is a critical component of biomonitoring studies, protocols for ensuring sample integrity and access are needed.
Banking of human tissue/fluid samples is important for future hypothesis testing, but it is complicated by cost, space, ethical consent (especially for genomic studies), and Health Insurance Portability and Accountability Act regulations (HIPAA 1996). Proper long-term storage conditions for specimens must be determined. The most useful matrices (e.g., blood, urine, hair) should be targeted for storage and should be linked to individual-level data (e.g., questionnaires, health information). In addition, to maximize the usefulness of such a resource, a mechanism would be needed to provide for future access to the specimens by investigators not necessarily involved in the original study.
Prioritization of chemicals selected for biomonitoring studies should be health-based but should also take into account factors such as the ability to bioaccumulate, known exposures in susceptible populations, analytic capabilities, and ability to collect and analyze the matrices that are most appropriate for each chemical of concern.
Put simply, chemicals should not be measured in humans merely because the analytic capability exists. Prioritization should be based on criteria that would yield the most useful data for maximizing public health gain.
An example of a model for prioritization of chemicals for biomonitoring is the recently awarded state biomonitoring program planning grant by the CDC to the state of New Hampshire. New Hampshire was awarded this grant for their proposal to develop biomonitoring programs to evaluate exposures to mercury in fish, MTBE (methyl tert-butyl ether) and arsenic in drinking water, and phthalates from plastics. The rationale for selecting these particular chemicals included several factors. First, these chemicals were among the top 10 environmental chemicals of greatest public health concern to that state’s population (New Hampshire has high levels of arsenic in drinking water and high levels of mercury in fish, and recognizes the growing incidences of MTBE groundwater contamination and expanding database on phthalate exposures in the general population). Also, there were readily available and validated methods for analysis of these chemicals in the appropriate biologic matrices in the population (arsenic and mercury in nails, MTBE and phthalates in urine or blood), which could be readily obtained, stored, and analyzed. Finally, there were known links to human health effects at exposures of concern (arsenic, mercury), and/or there was growing concern about potential health effects due to widespread or increasing exposure (arsenic, mercury, MTBE, phthalates) and potential for biomagnification in the food chain or accumulation in humans (mercury). Other chemicals of concern lacked one or more of these key attributes: They were not of unique or particular concern to New Hampshire, were not known to bioaccumulate, lacked an appropriate, practical or cost-effective methodology to analyze accurately, required specialized matrices (e.g., fat biopsy) that were impractical to obtain for a population study, or there was little or no toxicologic information to directly implicate them in specific health effects in humans. This selection process serves as an excellent paradigm for chemical selection for future biomonitoring studies.
Collaborations among academia, industry, and government should be encouraged to facilitate biomonitoring studies in occupational settings.
A critical component of any occupational epidemiology study is the assessment of exposure. Often industry has carried out its own monitoring, including biomonitoring of the workforce, as part of the routine assessment of employee exposure, with results residing in files or archives. However, investigators from outside the industry typically must rely on more limited data for exposure estimation, such as dates of employment, possibly obtained from union records. Collaborations with industry, built on mutual trust and respect, can often mean the availability of more specific exposure data. These can include biomonitoring results as well as environmental monitoring results and job-related data. For example, in a recent cohort study of chlorpyrifos manufacturing workers at the Dow Chemical Company, serial measurements were made over the study period of the chlorpyrifos urinary metabolite 3,5,6-trichloro-2-pyridinol and the intermediate end point, cholinesterase, allowing the investigators to carefully characterize occupational exposures and examine potential associations with health outcomes (Albers et al. 2004). A potential drawback for extrapolating these types of results (linking exposure to levels in humans) to the general population is that these occupational exposures may not be representative of exposures that would be estimated based on monitoring of air and other media, as workers often use protective clothing or work with largely enclosed processes.
The panel recommends that investigators consider the following biomonitoring study design issues:
Sampling frame.
The extent to which results can be generalized to a wider population will depend almost entirely on the sampling frame of the study. The best sampling frame is a random selection from a clearly defined population (with a high level of participation agreement), which will make results almost entirely generalizable to that population. The extent to which generalization from the study population to the general population will be restricted depends on whether a) the population is self-selected (e.g., volunteers), b) there is a high degree of nonparticipation, and c) participants are selected based on specific characteristics or features (e.g., pregnant women).
Laboratory techniques.
The laboratory performing the analyses should conduct rigorous quality assurance/quality control procedures in accordance with accepted guidelines, including calibration of instruments, running appropriate standards and blanks, performing spiking, blinded repeat and other quality assurance measures, and reporting limits of detection, variation, and other statistical parameters along with the experimental results (Needham and Wang 2002; Needham et al. 2002).
Selection of human specimen type.
In selecting the matrix for a biomonitoring study, the researcher must fully understand the utility and quality of that matrix, as well as the biologic significance and complexity. This selection is often a balance between purely scientific issues and the practicality of obtaining the matrix for a population study in a cost-effective way that is least invasive and risky to the study participant. However, matrices such as urine or blood, while easy to obtain or available from previous studies, are often not appropriate for measurement of certain chemicals.
Integrity of samples.
Improper collection, transportation, and/or storage of specimens can significantly affect the biomonitoring results (Aitio and Jarvisalo 1985; Griffin 1986; Wax et al. 2000). Biomonitoring studies should follow validated protocols that are appropriate to the matrix and analyte(s) of interest. Many samples require collection into the appropriate storage container, clean techniques for handling and storage, chemical or physical stabilization of the sample, storage under appropriate conditions, and control of the number of times a sample is removed from storage and assayed. Failure to follow these steps can produce either degradation or speciation change of the analyte or contamination of the sample from external sources (Versieck 1985). For example, until the past decade or so, it was not widely appreciated how easily a sample can become externally contaminated with chromium, principally from stainless steel and other metal components, but also from other sources. Simply dissecting and then homogenizing a rat liver in a typical laboratory blender can result in chromium contamination many times the actual biologic level in the sample. As a result, most published studies on chromium levels in human and animal tissues that did not recognize this problem contained largely inaccurate data. Analysis of trace levels of metals and other environmental chemicals often requires full clean techniques including metal-free, acid-washed sample collection and storage procedures and special clean-room analysis protocols. Likewise, collection of blood into a typical vacutainer can highly influence analysis of cadmium, plasticizers, and other compounds that may leach from rubber stoppers and walls of containers. Investigators should standardize and report their sample protocols because this can be critical in interpreting and comparing results across studies.
Reporting of nondetects.
The method used to assign a value to analytic results below the limit of detection (LOD; e.g., LOD/2) should be described because the method chosen can have an appreciable effect on the results and their interpretation (Helsel 1990). This is especially important in assessing populations in which biomonitored chemical concentrations are frequently below the detection limit. Ideally, a sensitivity analysis should be carried out to determine the effect that the method of assigning values below the LOD has on results and their interpretation.
Interpretation of Human Biomonitoring Data
Carefully conducted human biomonitoring studies serve several important functions, including a) evaluating time trends for levels of environmental chemicals in humans; b) evaluating the efficacy of regulatory action; c) assessing regional differences in levels of chemicals in populations; and d) establishing baselines and distributions of body burdens for populations. Biomonitoring information lends itself well for such interpretation, particularly for studies that are representative of populations of interest. However, besides using biomonitoring data as a marker of exposure, there is intense interest in using these data as markers of health effect(s) and establishing health-related reference levels for the measured chemicals. Alcohol is an example of a chemical for which exposure can be linked to internal dose and internal dose can be linked to effect. Lead is an example of an environmental chemical for which exposures are involuntary and for which there are sufficient data to draw associations between biomonitoring data and health, both to the individual and to a population. Decades of epidemiologic and toxicologic research on lead effects provide the underpinnings for the interpretation of blood lead level information. Not only has interpretive information been made available to the public for lead in children’s blood, but advice on medical interventions has also been developed. Both reference levels and recommended clinical interventions for other environmental chemicals are likely to evolve over time as new information is obtained; it is important to develop and make available such interpretive documents for the medical profession and the public.
Despite public pressure to provide more immediate interpretations of biomonitoring data in terms of potential for impacts on health, the development of analytic methods has provided the ability to measure extraordinarily low concentrations of a wide array of chemicals for which there are insufficient data on which to base those interpretations. In addition, it is often forgotten that the “measurement of an environmental chemical in a person’s blood or urine does not by itself mean that the chemical causes disease” (CDC 2003, p. 2). The panel discussed several issues related to interpretation of biomonitoring data, described below.
To interpret biomonitoring data in terms of health, studies are needed on the relationships between exposure and levels in the body and between levels in the body and health effect.
The first critical step is to develop an understanding of the relationship between exposure (applied dose) and body burden. This relationship can be complex, as in the case of arsenic. Although many toxic metals, such as lead, mercury, and cadmium, accumulate in human tissues such that biomonitoring can reveal the extent of long-term cumulative exposure, arsenic and other metals do not accumulate in this way. Each metal has unique pharmacokinetic and pharmacodynamic properties. For arsenic, the exposure of concern is consumption of inorganic arsenic in contaminated drinking water. Arsenic does not bioaccumulate and is readily excreted by the body with a half-life of days. As a result, total arsenic levels in blood and urine are highly variable. Moreover, because the inorganic forms are of primary toxicologic concern, if blood or urine analyses do not speciate arsenic into its various forms, the resulting data often have only limited toxicologic interpretative value. In contrast, arsenic in nails can be an excellent biomarker of exposure because it principally measures inorganic arsenic exposure, integrates exposure over several weeks, is a highly stable matrix, and is relatively impervious to external contamination or other confounders (Karagas et al. 1996, 2000, 2001a, 2001b; Nichols et al. 1998). Toenail arsenic levels were shown to closely correlate with drinking water arsenic at water levels > 1 ppb (Karagas et al. 2000). More important, toxicologic end points of concern were actually better correlated with toenail arsenic than with drinking water arsenic levels, which indicates that, in this particular situation, this bio-marker provided a more precise internal measure of individual exposure than combining measures of external levels (drinking-water arsenic) with estimates of exposure (how much water consumed per day). Such internal measures will also better integrate potential individual differences in uptake, distribution, metabolism, and excretion. This example highlights the many potential advantages of a good biomarker of exposure but also illustrates the importance of determining the appropriate matrix and methodology and the validation of this marker by comparison to external measures of exposure.
The second important step for a validated biomarker is to relate internal body burden, as assessed by the biomarker, to one or more biologic end points, such as changes in gene or protein expression, alterations in enzyme function, and specific polymorphisms or other genotype information. For example, individual arsenic toenail levels have been shown to correlate with decreases in lymphocyte expression of several DNA repair enzymes (Andrew et al. 2003a, 2003b). Such information not only provides potential biomarkers of exposure and effect but also provides direct support for the hypothesis that arsenic influences cancer risk, at least in part, by suppressing DNA repair and thereby increasing the risk from exposure to other environmental agents of concern such as sunlight (for skin cancer) and cigarette smoke chemicals (for lung and bladder cancer) (Karagas et al. 2001a, 2001b, 2004).
To be able to interpret biomonitoring data, laboratory reporting of clinical reference levels must be harmonized.
Even for some long-used markers of exposure, such as acetyl-cholinesterase and butyryl cholinesterase, reference levels have not been determined across laboratories using a consistent method (Wilson et al. 2002). There is widespread clinical confusion regarding the interpretation of reference levels, occurring most commonly when appropriate reference values are not provided by a reporting laboratory.
A recent clinical case involving a 46-year-old gardener with chronic malaise illustrates this point. An attending physician ordered a hair analysis for multiple mineral elements, including mercury and 16 additional “nutrient” elements. The gardener’s only reported exposure was having previously worked in cleaning laboratories at a nearby university where elemental mercury was used in manometers and other instruments. His hair mercury concentration was reported as 4.57 ppm, compared to the diagnostic laboratory’s reference range of 0–0.6 ppm. However, the reliability of such hair mineral analysis has been questioned (Barrett 1985; Seidel et al. 2001; Steindel and Howanitz 2001). Because of the reported abnormality of the hair mercury analysis, a repeat 24-hr urine was collected after a DMPS (sodium salt of 2,3-dimercapto-1-propane sulfonic acid) chelation challenge. The urine mercury level was 3.1 μg over the 24-hr period, or 0.83 μg/L (laboratory reference range was reported as 0–4 μg/24 hr). The measured level was similar to the mean level of urine mercury (0.77 μg/L) reported from the most recent National Health and Examination Nutrition Survey (NHANES) study and well below the 90th percentile level (3.15 μg/L) (CDC 2003). Urine mercury concentrations in humans > 100 μg/L have been associated with minor neurologic signs (Goldman et al. 2001), and urine mercury levels > 300 μg/L usually are associated with overt symptoms (Bates 1998). None of these indications was present in the cited case, but the confusion regarding the appropriate reference range may have provoked the decision to treat with DMPS. This case highlights the need for clear and consistent reference values for biomonitoring data on environmental chemicals.
Even if data on both toxicity and levels of environmental chemicals in laboratory animals are available for a given chemical, investigators should proceed with caution when attempting to link such information directly with human biomonitoring data and human health effects.
Because of species-specific variation in absorption, distribution, metabolism, and elimination, it is difficult to interpret measured levels of environmental chemicals in humans on the basis of levels of chemicals in laboratory animals. However, the goal is to eventually link animal or other laboratory data to human biomonitoring data in a way that harmonizes these data sets for more accurate and robust risk assessments. Currently, the primary approach for linking laboratory animal data to human health effects is via the risk assessment process, which relies on estimates of dose. The no observed adverse effect level (NOAEL) or the lowest observed adverse effect level (LOAEL) is used to establish a “safe” dose for humans. For some chemicals, human biomonitoring information can be used in combination with physiologically-based pharmacokinetic (PBPK) models to estimate dose (the success of this method depends on the properties of the environmental chemical and the availability of parameter information for building the PBPK model). In this way, biomonitoring data can serve as a marker of exposure (Figure 1). However, this is not equivalent to using biomonitoring data as a biomarker of effect. As noted by Bernard and Hermans (1997), biomarkers of early health effects should be stable in the biologic specimen, specific for the target tissue or cell, and sensitive to level of exposure to an insult. In an ideal world, it is desirable to have biomonitoring data serve as both biomarkers of exposure (Figure 1, left arrow) and effect (Figure 1, right arrow). The science is currently insufficiently developed for both of these purposes for most chemicals.
Environmental chemicals are a part of the public health spectrum. Because of the complex nature of disease etiology, scientists will need to obtain and analyze extensive amounts of data to fully understand which environmental chemicals, and at what levels in the body, are linked to adverse health outcomes. A key factor that is often overlooked is the inherent genetic variability of the population, which can profoundly influence disease risk even given the same environmental exposures. The study of gene–environment interactions will require an integration of exposure assessment (such as with biomonitoring) with genetic susceptibility assessment (such as with genetic polymorphism biomarkers). According to Waters and Fostel (2004, p. 943),
Predicting potential human health risks from chemical stressors raises three general challenges: the diverse properties of the thousands of chemicals and other stressors that are present in the environment; the time and dose parameters that define the relationship between exposure to a chemical and disease; and the genetic and experiential diversity of human populations and of organisms used as surrogates to determine the adverse effects of a toxicant.
Biomonitoring environmental chemicals is principally a public health tool that is part of the risk assessment framework and, with certain exceptions, is not yet informative as a marker for clinical risk.
Therefore, biomonitoring—especially the types of biomonitoring studies focused on emerging chemicals of concern for which there are limited epidemiologic and toxicologic data—is not generally useful for predicting adverse health effects to the individual. There may be cases, however, where an individual has very high, clinically relevant levels of a given chemical that could be used to assess their individual risk. For example, detection of high arsenic levels in individuals in New Hampshire as part of ongoing epidemiology and biomonitoring studies has been used as the basis for intervention in the form of suggestions for further testing and analysis for those individuals, recommendations to remediate their water, and providing additional resources for information about health effects and remediation options (M. Karagas, personal communication). However, it is important to note and to convey to such individuals that estimates of risk are based on population studies, whereas individual risks will be highly influenced by genetic background, other environmental or occupational exposures, lifestyle factors, and other individually variable factors.
Cardiovascular epidemiology provides a clear example of the complexity of documenting the usefulness and limitations of newly identified markers for clinical risk. The observation of advanced and early-onset atherosclerosis in patients with homocysteinuria prompted evaluation of plasma homocysteine as a risk factor for atherosclerosis in the broader population. An initial case–control study evaluated the level of homocysteine in patients with coronary artery disease compared to normal controls and observed significant differences in mean levels between the two groups (Kang et al. 1986). This finding by Kang et al. has been subsequently confirmed in other retrospective and cross-sectional studies (Braunwald et al. 2001). However, the results from prospective studies have been less consistent (Essebag et al. 2003). Some debate whether modestly elevated plasma homocysteine is a consequence rather than a cause of atherosclerosis (Christen 2000). Randomized trials to test whether lowering homocysteine levels will decrease risks of cardiovascular disease have shown some benefit in some subgroups of patients with preexisting coronary disease. Ongoing research is evaluating the association between homocysteine and stroke and other neurologic outcomes (Huang 2004; Wright et al. 2004).
For adequate interpretation of population-based biomonitoring data, improved population-based health data collection is essential.
Most countries have some form of death registration which is usually available to researchers. Some countries or local regions have disease-specific registries, most commonly cancer and birth defect registries. However, to make possible the interpretation of biomonitoring data, other registries are needed for health end points such as neurode-generative and respiratory effects. Newly developed registries should have a rigorous quality assurance program to ensure completeness and accuracy of records.
Communicating Results of Biomonitoring Studies
Researchers typically use the peer-reviewed scientific literature as their primary method of communication of their study design, results, and interpretation. This method is excellent for communicating with other researchers, especially those in similar disciplines. However, individuals outside of those fields will not necessarily seek out the same journals for information. On occasion, the media will become aware of an emerging scientific issue or publication, and a condensed and simplified version of the information will appear in a form available to the general public.
The scientific community must recognize the public interest in human biomonitoring studies and recognize that this interest has an effect on regulators, federal and state legislators, advocacy organizations, industry, and clinicians (who frequently find themselves on the front lines attempting to address concerns about the meaning of those studies for their patients’ health). Thus, researchers should consider venues in addition to the peer-reviewed literature so that their studies are properly communicated to relevant audiences. One risk associated with failing to do so is the potential misinterpretation of study conclusions by the media and the public. Misinterpretation is even more likely when communicating to a public that needs to be better informed about disease etiology and its multifactorial nature. General information on best practices for risk communication is available in the published literature and from government agencies [Agency for Toxic Substances and Disease Registry (ATSDR) 2001; Fischhoff 1995; National Research Council 1989; Sandman 1990]. Some key points from the workshop follow.
Investigators should be able to communicate epidemiologic concepts such as the difference between absolute and relative risk.
For example, in the pediatric condition known as Reye’s syndrome, there is an approximately 4-fold increased relative risk associated with the use of salicylates. But even following salicylate use, Reye’s syndrome remains a rare condition in absolute risk terms, with only 25 cases occurring in the United Sates in 1989 (a population incidence of 1.3 per million children under 5 years of age; Forsyth et al. 1989; MMWR 1991). Clinicians and scientists often grapple with the best way to understand and communicate risk, as has been demonstrated with basic and well-accessed information such as cancer risks (Woloshin et al. 2002).
Although the scientific community’s ability to communicate statistics and other mathematical concepts needs improvement, a larger issue is the striking lack of numeracy in the general population (Steen 1990). This is the background reality against which findings are communicated. Moreover, even when scientists attempt to communicate statistics accurately, these data can be misconstrued or miscommunicated by popular media and the web, or are in a form that is difficult for the general public to understand and interpret (Schwartz and Woloshin 2004; Woloshin et al. 2003). Methods for communicating scientific information to the public, especially pertaining to human health risk, have been addressed previously, and researchers in the field of biomonitoring should be familiar with this literature (ATSDR 2001).
For biomonitoring studies addressing health effects, investigators should distinguish between statistically significant effects and clinically significant effects.
Although the regulator or epidemiologist may be interested in the former, for clinicians and the public in general the latter is of primary interest. For example, studies reporting on the effects of polychlorinated dibenzodioxins, polychlorinated dibenzofurans, and coplanar polychlorinated biphenyls on breastfed infants provided data on levels of these environmental chemicals in the mothers’ milk samples and on infant serum levels of triiodothyronine, thyroxine, thyroid-stimulating hormone, thyroxine-binding globulin, and lymphocyte subsets (Nagayama et al. 1998a, 1998b). Effects on infant serum levels were noted, but no interpretive information was given. While this method of presenting study results is sufficient for many scientists, for the lay audience and others, additional information is desirable. For example, it is not clear whether observable adverse health effects in the infants would be anticipated or whether the serum levels were within clinically normal ranges (LaKind et al. 2004). A small biologic change in a quantitative measure can be statistically significant without necessarily indicating a change that is physiologically or toxicologically significant in terms of adverse health outcomes.
The development of an accessible Internet-based site for human biomonitoring data should be a high priority.
This type of site would allow scientists and others to share and compare data obtained from biomonitoring research (Waters and Fostel 2004). Progress with this type of public database is being made in the field of toxicogenomics, which has witnessed the development of data-exchange standards and guidelines for harmonization in data collection (Waters and Fostel 2004). A possible model for such a database is the Chemical Effects in Biological Systems knowledgebase (http://cebs.niehs.nih.gov). Models for standard-setting include the Clinical Data Interchange Standards Consortium (CDISC 2004) and the Standards for Exchange of Nonclinical Data (SEND 2004). The database for human biomonitoring studies should include the research data and reference ranges (both for population concentration data and clinical reference ranges). This type of site will require continual updating, which necessitates a long-term commitment of resources.
An established architecture is needed to communicate the meaning of individual and population-based biomonitoring results.
A concerted effort is needed to educate clinicians regarding the availability of expertise in interpreting human biomonitoring data. For most environmental chemicals, it is inappropriate to suggest that individuals consult their doctors with questions about biomonitoring data. Clinicians, especially general practitioners, are often ill prepared to answer specific questions regarding chemical exposures and health risks, nor are there obvious or readily available resources for them to obtain this information at a level and in a form that they and their patients can understand. One current option is to query a poison control center, which would likely refer them to a specialist from an established network of clinical toxicologists. Another available option is to contact an ATSDR office or a Pediatric Environmental Health Specialty Unit (PEHSU).
A nationwide effort is needed to inform physicians of the availability of medical toxicologists, such as members of American College of Medical Toxicology (American College of Medical Toxicology 2005). Efforts have been made to create electronic networks for rural physicians, such as the Rural Physicians Health Network, which links them to larger medical centers and other sources of specialized information that they can use both for specific patient inquiries and for continuing education. Such a network could be established to link physicians more generally to clinical toxicologists and public health officials that would allow them to tap into the growing knowledge base from biomonitoring, epidemiology, and similar studies. This could also provide physicians with appropriate information resources for deciding whether and how to analyze specific environmental chemicals in an individual, how to select the appropriate sample matrix and laboratory, and how to interpret the results.
Resources are the limiting factor in creating a robust and continuously updated database of human levels of environmental chemicals, linked to information that would allow these levels to be interpreted in terms of potential health effects.
The panel recommends that resources for this purpose be given a high priority. The panel recognizes that the resources required to build such a database will be significant. However, for protection of public health, it is insufficient to develop a biomonitoring database without the ability to interpret that database. Therefore, the panel recommends that funding agencies and organizations devote the resources necessary for this endeavor.
Because current exposures to certain environmental chemicals may be related to future adverse health effects, the panel recommends that an architecture be developed to support long-term storage of human specimens and that a process be established to provide for sharing of specimens as part of future investigations.
In the shorter term, physicians and others involved in health care require current information on interpretation of human biomonitoring data. Resources such as the American College of Medical Toxicology (2005) and, for pediatric issues, the PEHSUs (ATSDR 2005) are available but are not yet widely recognized. A concerted effort is required to increase the visibility of these resources and to develop additional resources for the effective communication of the interpretation of biomonitoring data.
In addition to resources for health care providers, there is a serious need for well-written, multilingual articles for the lay audience on an array of topics that would assist in improving the public’s ability to understand human biomonitoring information, including associated uncertainties and limitations. In addition to written material (i.e., manuscript-style documents), formats such as pamphlets, posters, graphic narratives, and videos are useful for reaching a wider audience.
The Research Foundation for Health and Environmental Effects (RFHEE), a nonprofit organization of the Chlorine Chemistry Council and its contributors, provided workshop support, including travel support and honoraria for the panel. The views expressed here do not necessarily reflect the views of RFHEE, its parent organization, or individual supporters contributing resources to RFHEE.
Figure 1 The continuum from exposure to adverse health effect. BBDR, biologically based dose response. With existing toxicologic and epidemiologic databases, we can more readily begin at the internal dose [identified by the star (e.g., biomonitoring data for environmental chemicals in blood)] and move along the arrow to the left, by using models such as PBPK models to obtain information on dose (exposure). At present, for most environmental chemicals, the greater challenge is to begin at the internal dose starting point and move to the right to obtain information about target tissue dose, biologic effects, and disease. (Adapted from Waters and Fostel [2004], with permission from the authors and from the Nature Publishing Group.)
==== Refs
References
Aitio A Jarvisalo J 1985 Biological monitoring of occupational exposure to toxic chemicals. Collection, processing, and storage of specimens Ann Clin Lab Sci 15 121 139 3888039
Albers JW Berent S Garabrant DH Giordani B Schweitzer SJ Garrison RP 2004 The effects of occupational exposure to chlorpyrifos on the neurologic examination of central nervous system function: a prospective cohort study J Occupational Environ Med 46 367 378
American College of Medical Toxicology 2005. Homepage. Fairfax, VA:American College of Medical Toxicology. Available: http://www.acmt.net/main/ [accessed 1 June 2005].
Andrew AS Karagas MR Hamilton JW 2003a Decreased DNA repair gene expression among individuals exposed to arsenic in United States drinking water Intl J Cancer 104 263 268
Andrew AS Karagas MR Klei L Hamilton JW 2003b Drinking water arsenic exposure is associated with decreased DNA repair gene expression Toxicol Sci 72 184
ATSDR 2001. A Primer on Health Risk Communication Principles and Practices. Atlanta, GA:Agency for Toxic Substances and Disease Registry. Available: http://www.atsdr.cdc.gov/HEC/primer.html [accessed 29 March 2005].
ATSDR 2005. Pediatric Environmental Health Specialty Units (PEHSU). Atlanta, GA:Agency for Toxic Substances and Disease Registry. Available: http://www.atsdr.cdc.gov/HEC/natorg/pehsu.html [accessed 23 September 2005].
Barrett S 1985 Commercial hair analysis. Science or scam? JAMA 254 1041 1045 4021042
Bates BA 1998. Heavy metals and inorganic metals: mercury. In: Clinical Management of Poisoning and Drug Overdose (Haddad LM, Shannon MW, Winchester JF, eds). 3rd ed. Philadelphia:W.B. Saunders, 750–756.
Bernard A Hermans C 1997 Biomonitoring of early effects on the kidney or the lung Sci Total Environ 199 205 211 9200865
Braunwald E Zipes DP Libby P eds. 2001. Heart Disease: A Textbook of Cardiovascular Medicine. 6th ed. Philadelphia: W.B. Saunders.
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:Centers for Disease Control and Prevention. Available: http://www.cdc.gov/exposurereport/ [accessed 29 March 2005].
CDISC (Clinical Data Interchange Standards Consortium) 2004. Homepage. Available: http://www.cdisc.org/ [accessed 23 September 2005].
CHEERS 2005. Children’s Environmental Exposure Research Study. Washington, DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/cheers/basic.htm [accessed 1 June 2005].
Christen WG 2000 Blood levels of homocysteine and increased risks of cardiovascular disease: causal or casual? Arch Intern Med 160 422 434 10695683
Essebag V Genest J Jr Suissa S Pilote L 2003 The nested case-control study in cardiology Am Heart J 146 581 590 14564310
Environmental Working Group 2004. Group Demands Halt to Unethical, Scientifically Questionable Study in Which EPA and Chemical Industry Pay Families to Expose Children to Pesticides. Environmental Working Group Press Release. 29 October 2004. Available: http://www.ewg.org/issues/humantesting/20041029/index.php [accessed 29 March 2005].
Fischhoff B 1995 Risk perception and communication unplugged: twenty years of progress Risk Anal 15 137 145 7597253
Forsyth BW Horwitz RI Acampora D Shapiro ED Viscoli CM Feinstein AR 1989 New epidemiologic evidence confirming that bias does not explain the aspirin/Reye’s syndrome association J Am Med Assoc 261 2517 2524
Goldman LR Shannon MW the Committee on Environmental Health 2001 Mercury in the environment: implications for pediatricians Pediatrics 108 197 205 11433078
Griffin RM 1986 Biological monitoring for heavy metals: practical concerns J Occup Med 28 615 618 3746481
Helsel DR 1990 Less than obvious—statistical treatment of data below the detection limit Environ Sci Technol 24 1766 1774
HIPAA 1996. Health Insurance Portability and Accountability Act of 1996. Public Law 104-191. Available: http://aspe.hhs.gov/admnsimp/pl104191.htm [accessed 23 September 2005].
Huang Y-C 2004 Lowering homocysteine levels does not reduce rates of stroke, coronary heart disease or death in people with ischaemic stroke Evid Based Healthcare 8 210 212
Kang SS Wong PW Cook HY Norusis M Messer JV 1986 Protein-bound homocyst(e)ine. A possible risk factor for coronary artery disease J Clin Invest 77 1482 1486 3700650
Karagas MR Le XC Morris S Blum JD Lu X Spate VL 2001a Markers of low level arsenic exposure for evaluating human cancer risks in a US population Int J Occup Med Environ Health 14 171 175 11548067
Karagas MR Morris JS Weiss JE Spate V Baskett C Greenberg ER 1996 Toenail samples as an indicator of drinking water arsenic exposure Cancer Epidemiol Biomarkers Prev 5 849 852 8896897
Karagas MR Stukel TA Morris JS Tosteson TD Spencer SK Greenberg ER 2001b Skin cancer risk in relation to toenail arsenic concentrations in a US population-based case-control study Am J Epidemiol 153 559 565 11257063
Karagas MR Tosteson TD Morris JS Demidenko E Mott LA Heaney J 2004 Incidence of transitional cell carcinoma of the bladder and arsenic exposure in New Hampshire Cancer Causes Control 15 465 472 15286466
Karagas MR Tosteson TD Morris JS Weiss JE Stannard V Spate VL 2000 Measurement of low levels of arsenic exposure: A comparison of water and toenail concentrations Am J Epidemiol 152 84 90 10901333
LaKind JS ed. 2005. Technical Workshop on Human Milk Surveillance and Biomonitoring for Environmental Chemicals in the United States. J Toxicol Environ Health special issue Vol. 68, no. 20.
LaKind JS Brent RL Dourson ML Kacew S Koren G Sonawane B 2005 Human milk biomonitoring data: interpretation and risk assessment issues J Toxicol Environ Health 68 20 1713 1770
LaKind JS Wilkins AA Berlin CM 2004 Environmental chemicals in human milk: a review of concentrations, determinants, infant exposures and health, and guidance for future research Toxicol Applied Pharmacol 198 184 208
MMWR 1991 Reye syndrome surveillance—United States, 1989 Morb Mortal Weekly Rep 40 88 90 Available: http://www.cdc.gov/mmwr/preview/mmwrhtml/00001906.htm [accessed 23 September 2005].
Nagayama J Okamura K Iida T Hirakawa H Matsueda T Tsuji H 1998a Postnatal exposure to chlorinated dioxins and related chemicals on thyroid hormone status in Japanese breast-fed infants Chemosphere 37 1789 1793 9828307
Nagayama J Tsuji H Iida T Hirakawa H Matsueda T Okamura K 1998b Postnatal exposure to chlorinated dioxins and related chemicals on lymphocyte subsets in Japanese breast-fed infants Chemosphere 37 1781 1787 9828306
Needham LL Ryan JJ Fürst P 2002 Guidelines for analysis of human milk for environmental chemicals J Toxicol Environ Health A 65 1893 1908 12470493
Needham LL Wang RY 2002 Analytic considerations for measuring environmental chemicals in breast milk Environ Health Perspect 110 A317 324 12055062
Nichols TA Morris JS Mason MM Spate VL Beskett CK Cheng TP 1998 The study of human nails as an intake monitor for arsenic using neutron activation analysis J Radioanal Nucl Chem 236 51 56
NIOSH 2004. Specific Medical Tests or Examinations Published in the Literature for OSHA-Regulated Substances. NIOSH Pub. 2005-110. Huntington, WV:National Institute for Occupational Safety and Health. Available: http://www.cdc.gov/niosh/nmed/pdfs/2005-110.pdf [accessed 29 March 2005].
National Research Council 1989. Improving Risk Communication. Washington, DC:National Academy Press.
Oleskey C Fleischman A Goldman L Hirschhorn K Landrigan PJ Lappé M 2004 Pesticide testing in humans: ethics and public policy Environ Health Perspect 112 914 919 15175182
Pesticide Action Network Updates Service 2004. Stop EPA From Testing Pesticides on Children. Available: http://www.panna.org/resources/panups/panup_20041201.dv.html [accessed 29 March 2005].
Sandman PM 1990. Getting to maybe: some communications aspects of siting hazardous waste facilities. In: Readings in Risk (Glickman TS, Gough M, eds). Washington, DC: Resources for the Future, 233–246.
Schwartz LM Woloshin S 2004 The media matter: a call for straightforward medical reporting Ann Intern Med 140 226 228 14757622
Seidel S Kreutzer R Smith D McNeel S Gilliss D 2001 Assessment of commercial laboratories performing hair mineral analysis J Am Med Assoc 285 67 72
SEND 2004. Standard for Exchange of Nonclinical Data. Review Version 1.4 March 2, 2004. Available: http://www.cdisc.org/models/send/v1.4/ [accessed 23 September 2005].
Sexton K Needham LL Pirkle JL 2003 Human biomonitoring of environmental chemicals Am Sci 92 38 45
Steen LA 1990 Numeracy Daedalus 119 211 231 Available: http://www.stolaf.edu/people/steen/Papers/numeracy.html [accessed 29 March 2005].
Steindel SJ Howanitz PJ 2001 The uncertainty of hair analysis for trace metals J Am Med Assoc 285 83 85
Stokstad E 2004 Pollution gets personal Science 304 1892 1894 15218119
U.S. EPA 2005. Children’s Environmental Exposure Research Study: this study was cancelled April 8, 2005. Washington, DC:U.S. Environmental Protection Agency. Available: http://www.epa.gov/cheers/ [accessed 12 April 2005].
Versieck J 1985 Trace elements in human body fluids and tissues Crit Rev Clin Lab Sci 22 97 184 3891229
Waters MD Fostel JM 2004 Toxicogenomics and systems toxicology: aims and prospects Nat Rev Genet 5 936 948 15573125
Wax PM Goldfarb A Cernichiari E 2000 Mercury contamination of heavy metal collection containers Vet Hum Toxicol 42 22 25 10670082
Wilson BW Henderson JD Ramirez A O’Malley MA 2002 Standardization of clinical cholinesterase measurements Intl J Toxicol 21 385 388
Woloshin S Schwartz LM Ellner A 2003 Making sense of risk information on the web Br Med J 327 695 696 14512452
Woloshin S Schwartz LM Welsh HG 2002 Risk charts: putting cancer risk in context J Natl Cancer Inst 94 799 804 12048267
Wright CB Lee HS Paik MC Stabler SP Allen RH Sacco RL 2004 Total homocysteine and cognition in a tri-ethnic cohort: the Northern Manhattan Study Neurology 63 254 260 15277617
|
16263520
|
PMC1310927
|
CC0
|
2021-01-04 23:41:31
|
no
|
Environ Health Perspect. 2005 Nov 6; 113(11):1615-1621
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8197
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8129ehp0113-00162216263521ResearchMeeting Report: Structural Determination of Environmentally Responsive Proteins Reinlib Leslie Division of Extramural Research and Training, National Institute of Environmental Health Sciences, National Institutes of Health, U.S. Department of Health and Human Services, Research Triangle Park, North Carolina, USAAddress correspondence to L. Reinlib, Susceptibility and Population Health Branch, Division of Extramural Research and Training, National Institute of Environmental Health Sciences, Room 3453, 79 TW Alexander Dr., Research Triangle Park, NC 27709-2233 USA. Telephone: (919) 541-4998. Fax: (919) 316-4606. E-mail
[email protected] author declares he has no competing financial interests.
11 2005 13 7 2005 113 11 1622 1626 18 3 2005 13 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The three-dimensional structure of gene products continues to be a missing lynchpin between linear genome sequences and our understanding of the normal and abnormal function of proteins and pathways. Enhanced activity in this area is likely to lead to better understanding of how discrete changes in molecular patterns and conformation underlie functional changes in protein complexes and, with it, sensitivity of an individual to an exposure. The National Institute of Environmental Health Sciences convened a workshop of experts in structural determination and environmental health to solicit advice for future research in structural resolution relative to environmentally responsive proteins and pathways. The highest priorities recommended by the workshop were to support studies of structure, analysis, control, and design of conformational and functional states at molecular resolution for environmentally responsive molecules and complexes; promote understanding of dynamics, kinetics, and ligand responses; investigate the mechanisms and steps in posttranslational modifications, protein partnering, impact of genetic polymorphisms on structure/function, and ligand interactions; and encourage integrated experimental and computational approaches. The workshop participants also saw value in improving the throughput and purity of protein samples and macromolecular assemblies; developing optimal processes for design, production, and assembly of macromolecular complexes; encouraging studies on protein–protein and macromolecular interactions; and examining assemblies of individual proteins and their functions in pathways of interest for environmental health.
Environmental Genome Projectgene–environment interactionsprotein structurestructural biology
==== Body
If the only tool you have is a hammer, you tend to see every problem as a nail.
—Abraham Maslow
Scientists are often no different from others in following the tendency of which Maslow spoke. Even in structural biology, a field that relies on complex mathematics, immense computing power, and cutting-edge technologies, investigations can tend to apply the specific skills at hand. However, with the emergence of genomic data and new technologies comes great challenges requiring multiple skills and scientific viewpoints. Participants at a recent workshop on structural determination of environmentally responsive proteins convened by the National Institute of Environmental Health Sciences highlighted some of these technologies but cautioned that the greatest strides will come only from seeing the problems as more than just nails.
Knowledge of the structures of individual proteins and how they fit together in macro-molecular complexes is critical to understanding function and to accelerating design of new molecular tools and more effective medicines. Although the Human Genome Project ignited an explosion of efforts to understand the blueprint of human health and disease, optimally applying the information flowing from linear gene sequences requires visualization of complete proteins in physiologically correct forms. Genomes have also been sequenced of varied pathogens and animals featured in laboratory experimentation, creating a high demand for the associated protein structures. To fulfill this demand, more advanced studies will be needed, taking advantage of the biophysics and understanding of posttranslational modifications that mandate protein folding and the ultimate form of the three-dimensional structures.
A long-term goal of these studies is to gain intimate insights into the function of protein complexes, enabling construction of ligand agonists and antagonists to assist efforts that will lead to clear mechanistic understanding and effective drugs. Among the recent successes in this arena are the structural resolutions of regions shared among the members of protein families involved in, for example, metabolism and detoxification. Key structural elements have been resolved of the components of cellular pathways—such as tyrosine kinases, G-protein subunits, and select apoptotic regulators—that mediate normal function and responses to environmental exposure and disease pathogens. Although many unique structures have been demonstrated in the population of 20,000–25,000 individual proteins that are predicted from the human genome, there is a widening gap between the expected numbers of genes and the consequent protein structures (International Human Genome Sequencing Consortium 2004). Including combinations and posttranslational modifications, about 100,000 gene products are predicted. In fact the number of sequences entered into public databases over the last several years is increasing at a much faster pace than the number of determined structures entered into the Protein Data Bank. Surprisingly, current estimates suggest that only 1,000–5,000 distinct, stable polypeptide chain folds exist in nature to accommodate the rich variety of domain structures. However, only about 700 of these distinct protein folds have been determined experimentally (Burley and Bonanno 2002). Further resolutions will be necessary to clarify the structural basis for function of the multiple components aligned in complexes and cell pathways of interest.
Accumulating evidence indicates that the structure of functional protein units is more complex than previously thought. It is becoming clear that the functions of many proteins occur as components of macromolecular complexes. Complexing may be required to fulfill a basic function (e.g., proper binding of tumor necrosis factor) or to synergize activity through, for example, altered binding affinities. BRCA-1, for example, interacts with a partner protein known as BARD1 (BRCA-1–associated ring domain). Although both BRCA-1 and BARD1 possess ubiquitin ligase activity, the combined complex is dramatically more active than either of the solitary proteins (Baer 2001). Synergies such as this may result from effects on binding affinities, efficiencies, and/or dynamics. Notably, these changes can occur in regions distal to protein active sites, suggesting that the impact of complexing may not be obvious from the examination of an individual side chain. To address the mechanisms of such effects, future studies will be needed on full-length proteins and macromolecular complexes, requiring an even greater set of skills and disciplines than in current practice.
Over the last decade, the National Institute of Environmental Health Sciences (NIEHS) has invested heavily in resequencing genes of interest in understanding the role of sequence variation in susceptibility to environmental perturbation. These environmentally responsive genes were chosen for their known or likely involvement in cellular pathways and diseases that involve environmental exposures, such as cancer, xeroderma pigmentosum, and Werner’s syndrome. As part of the NIEHS Environmental Genome Project (EGP), the resequencing and verification of about 550 genes was initiated on a set of 96 human samples obtained from the Coriell Resource Center (Coriell Institute for Medical Resources 2005; Wilson and Olden 2004). The EGP has recently decided to expand the data sets to an ethically defined panel and to explore more genes of interest to the research community. To date, approximately 280 genes have been completed, revealing more than 25,000 previously unknown polymorphisms (data available to scan and download at GeneSNPs http://www.genome.utah.edu/genesnps/). The data are useful for assessment of sensitivities based on single nucleotide polymorphisms (SNPs) in both population and basic science gene–environment projects. Individual SNPs may have significant effects on structure. Very elegant studies that indicate that intimate interactions of multiple regions within a protein contribute to overall efficiency have recently been reviewed (Tsigelny et al. 2004).
To better understand the relation of protein structural variation in environmentally responsive proteins to disease risk and resistance, NIEHS decided to appraise the state of the science and to explore optimal avenues for further research. Thus, on 26–27 April 2004, the Workshop on Structural Determination of Environmentally Responsive Proteins was convened at Snowbird Conference Center in Snowbird, Utah. The panel was composed of leading experts in the areas of crystallography, nuclear magnetic resonance, molecular biology, genomics, and environmental health sciences.
The workshop participants considered a variety of cutting-edge concepts and applications in structural biology. These included protein dynamics, protein–protein influences in macromolecular complexes, ligand responses, the impact of gene polymorphisms on predicted structures, and posttranslational modifications. Discussions also focused on the special requirements of studies of membrane proteins and the advantages of functionally based ligand design. A unique aspect of the workshop was the emphasis placed on environmentally responsive proteins and issues in environmental health sciences. For example, the structures discussed at the meeting included plasma membrane mercury transporters and P450 proteins. However, many of the topics and questions raised will no doubt be of general interest, and advances in these areas will likely be useful to a variety of research endeavors.
The workshop participants produced a set of ambitious, but practical, goals and prioritized recommendations that are discussed below.
Recommendations
In considering how to optimize research resources in reaching specific goals in structural biology, the workshop participants strongly encouraged integrated, multidisciplinary programs that would maximally integrate basic science, computing, mathematics, and engineering. Although outstanding workers are found in all these fields, they approach their subject matter from disparate viewpoints and appear to speak different languages. It appears that communication and thus efficiency of operation are lacking among investigators in multiple, complementary areas. In accordance, the workshop participants recognized the need for cross-training among molecular biologists, geneticists, computer scientists, and mathematicians, especially among young investigators. Trainees with backgrounds in biochemistry, molecular biology, and physiology, as well as emerging areas, should be allowed to gain valuable skills in mathematics and computer science that could be applied to structural biology questions, particularly as they pertain to environmental health sciences.
The workshop participants recommended that interdisciplinary teams bring their talents to bear on gene products and pathways of interest to the environmental health research community. The workshop participants expressed the sentiment that such a focus was unlikely to come from the research community at large without leadership from the NIEHS and National Institutes of Health (NIH). Currently neglected areas highlighted for further investigation include—but are not limited to—complexes involving bioinorganic substances, for example, vanadate, aluminum fluoride, and borate; proteins in pathways influenced by environmental contaminants or dietary factors, such as endocrine disruptors; and DNA repair proteins.
Also, a variety of membrane proteins and membrane receptor complexes are considered to be understudied. Although proteins associated with biologic membranes comprise approximately 30% of the genome encoded peptides, only about 2% of the three-dimensional structures deposited in the Protein Data Bank—92 membrane proteins—are membrane proteins (White 2005; Zhou et al. 2004). The lower number of high-resolution three-dimensional structures makes homology modeling, in which existing structures are used as templates, difficult to apply to membrane proteins (Zhou et al. 2004). For environmental health studies, membrane proteins of interest include the components of stress signaling pathways and ion channels involved in the transport of xenobiotics. These include aryl hydrocarbon receptors, multidrug-resistance proteins, and transporters that facilitate uptake, metabolism, and clearance of environmental toxicants such as transporters of methylmercury and inorganic mercury. Membrane macromolecular complexes, in particular, present a number of laborious and complex tasks to resolve and are unlikely to be targets of interest for pharmaceutical companies attempting to bring drugs to market. These subjects could be timely for investigation by academic researchers.
After considering a variety of exciting new findings and technologies, the workshop participants recommended a set of specific scientific goals to enhance future research and contribute to the understanding of the structural and functional relationships of proteins and macromolecular complexes.
The workshop participants prioritized their suggestions into two groups. The recommendations with the highest priorities are discussed in further detail below. The secondary priorities were seen as later steps for investigation.
Highest Priorities
Support studies of structure, analysis, control, and design of conformational and functional states at molecular resolution for environmentally responsive molecules and complexes
Promote understanding of dynamics, kinetics, and ligand responses
Investigate the mechanisms and steps in posttranslational modifications, protein partnering, impact of genetic polymorphisms on structure/function, and ligand interactions
Encourage integrated experimental and computational approaches.
Secondary Priorities
Improve the throughput and purity of protein samples and macromolecular assemblies (e.g., environmentally responsive membrane proteins)
Develop optimal processes for design, production, and assembly of macromolecular complexes
Encourage studies on protein–protein (macromolecule) interactions
Examine assemblies of individual proteins and their functions in pathways of interest.
Support studies of structure, analysis, control, and design leading to understanding of conformational and functional states at molecular resolution for environmentally responsive molecules and complexes.
There is a need to interface structural findings with biochemical outcomes, at both the in vitro and in vivo levels. The workshop participants pointed out that many talented investigators exploit a particular local skill, such as crystallography, but assistance from other scientific fields is needed to integrate these findings into a physiologic or even clinical model. Important questions permeate the field that require multiple viewpoints: How does structure lead to catalytic activity in protein kinases and other complexes of interest? How do protein dynamics play into the conformational changes that modulate function? How does misfolding lead to defective physiologic conditions?
Efforts to identify and determine the mechanisms and results of posttranslational modifications are inadequate. Gene sequencing provides a first step, but the ultimate amino acid chain can be modified radically in cellular processing. Efforts need to be redoubled to determine the final peptides. This point highlights the lack of studies performed under conditions that replicate the intracellular milieu. The final structure of a protein or complex of interest may differ significantly from that determined in ultrapure preparations. This is not a call to reintroduce impurities into samples, but rather to appreciate the influence of intracellular conditions on structure and function.
Continued focus is needed in determining molecular resolution that provides insights into interactions among proteins and in complexes. The complex need not be limited to proteins. The role of RNA, for example, appears to be underappreciated (Chien et al. 2004).
Strategies are needed to predict function from structure. Although investigators have learned much from biochemical considerations, structural resolution is frequently seen as an end point instead of a beginning for in-depth functional studies. To be most useful, structural determinations must be paired with models of how individual subunits interact with other molecules (Aloy et al. 2005). For example, DNA-associated proteins—DNA polymerases, glycosylases, and alkylases—are structurally diverse, but the relation of the known variations to function is not well understood. There are also a variety of proteins that fold into their “native state” on binding (De Lorenzi et al. 2004). Little is being done to discern how docking works in these cases. Docking methods attempt to maximally exploit all the available structural and chemical information possible from proteins, ligands, and protein–ligand complexes (van Dijk et al. 2005). “Guided docking’“ incorporates some degree of chemical information to actively guide the orientation of the ligand into the binding site (Fradera and Mestres 2004). Further work is needed to perfect such predictive models.
Promote understanding of structural dynamics, kinetics, and ligand responses.
The workshop participants agreed that the fourth dimension must be considered to clarify the kinetics, specificities, affinities, and function of proteins. One limitation of current investigations is the fixed point in time in which structures are generally solved. The dynamics of a protein of interest or its interactions with neighboring proteins in its functional pathways are likely to be key to understanding the ultimate physiologic roles. Time- and temperature-dependent dynamics of domain fluctuations have been demonstrated in protein kinases and human estrogen receptors and are likely to be integral to the structure–function relationship of many other proteins. Thus, although “snapshots” of regional protein structures are accumulating, much less is known of their place in macromolecular complexes or with regard to time. Time dependence could be a critical factor influencing conformation and behavior of side chains and flexible regions. Importantly, an action in one domain could affect other, distally located sites, an event termed allostery (Kern and Zuiderweg 2004). Consideration of the time domain is often overlooked in structural studies but could be an essential part of the overall mechanism of biologic reactions. Quantitative time-dependent kinetic analysis would be expected to lead to new avenues that will produce working models of protein complexes and interactions in pathways. The results will elucidate the mechanisms of normal physiology, susceptibility, and disease.
The structures of crystallized proteins must be examined under varied conditions, not just in cell-free systems, to better understand the constraints, limits, and flexibilities of macromolecular complexes. For example, perturbing the system would likely reveal more information about binding specificity. This approach may not be applicable to SNP studies but would apply to design of ligands based on known DNA sequences.
Structures of membrane proteins pose unique problems, but the reward would seem to be worth the efforts. These proteins are often first responders to exogenous stimuli and mediate second messengers and other signaling processes, highlighting their importance in environmental health. Membrane proteins are often in relatively low abundance, and their study will require development of more robust expression systems to increase both yield and purity. Studies will also need region-specific labels that do not impede protein function of interaction with the membrane environment.
Investigate the mechanisms and steps in posttranslational modifications, protein partnering, impact of genetic polymorphisms on structure/function, and ligand interactions.
The workshop participants indicated that three-dimensional structure is only a piece of the puzzle. Detailed atomic resolutions are also needed, as are insights into biochemical functions. To construct clear models relating structure with function, projects need to determine what the cellular function is for a protein complex and how that function relates to phenotype and susceptibility. Studies, then, may need to be performed on full-length proteins and under conditions replicating the natural milieu.
Proteins may need to be chosen that participate in complexes and interact with other proteins or nucleic acids. One example is the cold-shock proteins in bacteria. A major challenge is to make proteins amenable to study, especially for solving complex structures and assemblies. Better prediction is needed to determine the most promising protein fragments to study to optimize efficiency of time and cost. Much needed are new probes targeted to specific conformational states and individual steps of posttranslational modification. Binding agents are of particular interest for membrane proteins.
A compelling case was made for studies of the impact of SNPs on structure and function. Population studies are contributing a large amount of data linking SNPs with disease susceptibility. Combining these data with structural insights would increase the potential for improved mechanistic understanding and drug design. Tsigelny et al. (2004) provides a comprehensive overview of how multiple SNPs may affect P450 protein structures, such as aromatase. The authors suggest that visualizing the proteins allows focusing on likely sites controlling function, specificity for substrates, and the associated kinetics. Following this course, they suggest that the most significant impact would result if a particular SNP occurred in an area affecting “substrate recognition sites” or “substrate and product passage sites.” These types of value-added studies are encouraged because they indicate how structural insights, in providing new views of proteins, can lead to the design of novel ligands.
Encourage integrated experimental and computational approaches.
Future investigations will require even more integration of information from diverse sources, especially in consideration of macromolecular assemblies. Fortunately, the technologies of crystallography and comparative modeling become very powerful when combined (for review, see Davis and Sali 2005). For example, it is impractical to measure the functional impact of every possible SNP at all positions in each protein of interest. Thus, prediction based on general principles of protein structure will save time and energy. There are, in fact, publicly accessible web servers to do just this, such asLarge Scale-SNP (Rachel and Sali 2005). The server accepts input specifications for a structure and a single amino acid mutation and outputs a prediction of whether or not the mutant protein is impaired, as well as associated justifications and altered features (Karchin et al. 2005; Pieper et al. 2004). The system has worked well for several known proteins and SNPs, such as human BRCA-1 domains. A key issue is to relate the results of the model with physiologic impact. It is not hard to imagine that physiologists and biochemists will be in greater demand to collaborate on structural biology projects, just as microarrays and genomics applications have become common in population studies.
Computational protein design also lends itself well to producing novel proteins and systems. Such designs can provide mechanistic insights into the workings of complexes. The vast number of possible protein structures based on the 20 common amino acids presents a dilemma for experimentalists. Function-based computer design allows for a multitude of parameters that can be tested in silico. A simple example is to examine open versus closed conformations in the absence or presence of ligands that are expected to bind based on conserved sequences. Proteins such as enzymes could also be designed that interact with environmental pollutants. For example, a theoretical protozyme that mediates ester hydrolysis by thioredoxin could be configured that would likely have measurable activity in reaction mechanisms (Bolon and Mayo 2001). The computer model would allow testing of limited mutagenesis that would indicate the degree of effectiveness of potential ligands.
Summary
The last decade has seen an enormous expansion of insights into macromolecular structures indicating that proteins have a dynamic and complicated existence. Research in structural biology has exploded over the last decade, following a plethora of data on the genomes of humans and model organisms, and the advent of affordable computer power.
The NIEHS has an impressive history of supporting gene–environment studies. The NIEHS’s investments include extensive resequencing of > 250 human environmentally responsive genes, molecular epidemiology planning grants to form the basis for future projects, and a future program to resequence genes of interest in the laboratory rat. Building on these past initiatives, better understanding of the structure and, with it, molecular function of proteins of interest are of high priority. Along with improvements in the ability to predict and visualize protein structures come new challenges. One would like to understand not just how individual proteins operate, but also how they complex with other proteins, nucleic acids, and substrates. It follows that better ligands—and thus more promising drugs—could be constructed based on the three-dimensional images of the complex.
In a recent review of the complexities of DNA replication, Bruce Alberts (2003) worries that “a generation of biologists may have become lulled into believing that the essence of a biological mechanism has been captured, and the entire problem therefore solved” by construction and examination of two-dimensional cartoons of cell pathways. Alberts suggests that studies of biologic processes, such as DNA replication, will require collaborations of physicists, chemists, and structural and molecular biologists, the goal being to define the atomic structures of all the relevant proteins and the associated kinetics for the enzyme reactions. To this mix, we might add physiologists, biochemists, and epidemiologists to bring the scientific endeavor full circle to public health.
Ultimately, these steps, as fostered by NIH programs, such as the NIEHS EGP and the National Institute of General Medical Sciences Protein Structure Initiative, are expected to lead to improved research tools and more effective therapeutic drugs. It is important to look beyond “the hammer and the nail” for the best combination of techniques and strategies for the challenges ahead.
I thank P. Jennings for chairing the meeting and P. Mastin and D. Balshaw for their contributions and energy in organizing the workshop. I am also deeply indebted to the workshop participants for their thoughts and enthusiasm in producing the recommendations discussed in this article and for critical readings of the manuscript.
Participants in the NIEHS Workshop on Structural Determination of Environmentally Responsive Proteins
Patricia Jennings (chair) Division of Physical Sciences, Chemistry, and Biochemistry
University of California at San Diego
La Jolla, California
Leslie Reinlib (co-organizer) Division of Extramural Research and Training
NIEHS
Research Triangle Park, North Carolina
David Balshaw (co-organizer) Division of Extramural Research and Training
NIEHS
Research Triangle Park, North Carolina
Thomas Ellenberger Department of Biological Chemistry and Molecular Pharmacology
Harvard Medical School
Boston, Massachusetts
Traci Hall Division of Intramural Research
NIEHS
Research Triangle Park, North Carolina
Dorothee Kern Department of Biochemistry
Brandeis University
Boston, Massachusetts
Shohei Koide Department of Biochemistry and Molecular Biology
University of Chicago
Chicago, Illinois
James Halpert Departments of Pharmacology and Toxicology
University of Texas Medical Branch
Galveston, Texas
Stephen L. Mayo Division of Biology
California Institute of Technology
Pasadena, California
Gaetano T. Montelione Northeast Structural Genomics Research Consortium and
Department of Molecular Biology and Biochemistry
Rutgers, The State University of New Jersey
Piscataway, New Jersey
John Norvell National Institute for General Medical Sciences
NIH
Bethesda, Maryland
Stanley Opella Center for Nuclear Magnetic Resonance Spectroscopy and Imaging of Proteins
University of California at San Diego
La Jolla, California
Robert London Division of Intramural Research
NIEHS
Research Triangle Park, North Carolina
Andrej Sali Departments of Biopharmaceutical Sciences and Pharmaceutical Chemistry
California Institute for Quantitative Biomedical Research
University of California
San Francisco, California
==== Refs
References
Alberts B 2003 DNA Replication and recombination Nature 421 431 435 12540917
Aloy P Pichaud M Russell RB 2005 Protein complexes: structure prediction challenges for the 21st century Curr Opin Struct Biol 15 15 22 15718128
Baer R 2001 With the ends in sight: images from the BRCA1 tumor suppressor Nature Struct Biol 8 822 824 11573079
Bolon DN Mayo SL 2001 Enzyme-like proteins by computational design Proc Natl Acad Sci USA 98 14274 14279 11724958
Burley SK Bonanno JB 2002 Structuring the universe of proteins Annu Rev Genomics Hum Genet 3 243 262 12194989
Chien CY Xu Y Xiao R Aramini JM Sahasrabudhe PV Krug RM 2004 Biophysical characterization of the complex between double-stranded RNA and the N-terminal domain of the NS1 protein from influenza A virus: evidence for a novel RNA-binding mode Biochemistry 43 1950 1962 14967035
Coriell Institute for Medical Resources 2005. Coriell Cell Repositories. Camden, NJ:Coriell Institute for Medical Resources. Available: http://locus.umdnj.edu/ [accessed 1 July 2005].
Davis FP Sali A 2005 PIBASE: a comprehensive database of structurally defined protein domain interfaces Bioinformatics 21 1901 1907 15657096
De Lorenzi E Giorgetti S Grossi S Merlini G Caccialanza G Bellotti V 2004 Pharmaceutical strategies against amyloidosis: old and new drugs in targeting a “protein misfolding disease” Curr Med Chem 11 1065 1084 15078166
Fradera X Mestres J 2004 Guiding docking approaches to structure-based designs and screening Curr Top Med Chem 4 687 700 15032682
International Human Genome Sequencing Consortium 2004 Finishing the euchromatic sequence of the human genome Nature 431 931 945 15496913
Karchin R Diekhans M Kelly L Thomas DJ Pieper U Eswar U 2005 LS-SNP: large-scale annotation of coding non-synonymous SNPs based on multiple information sources Bioinformatics 21 2814 2820 15827081
Kern D Zuiderweg ERP 2004 The role of dynamics in allosteric regulation Curr Opin Struct Biol 13 748 757 14675554
Rachel K Sali A 2004. LS-SNP: Large Scale Human SNP Annotation. San Francisco:University of California, San Francisco. Available: http://salilab.org/LS-SNP [accessed 27 June 2005].
Pieper U Eswar N Braberg H Madhusudhan MS Davis FP Stuart AC 2004 MODBASE, a database of annotated comparative protein structure models, and associated resources Nucleic Acids Res 32 D217 D222 14681398
Tsigelny IF Kotlovyji V Wasserman L 2004 SNP analysis combined with protein structure prediction defines structure-functional relationships in cancer related cytochrome P450 estrogen metabolism Curr Med Chem 11 525 538 15032601
Van Dijk AD Boelens R Bonvin AM 2005 Data-driven docking for the study of biomolecular complexes FEBS J 272 293 312 15654870
White SH 2005. Membrane proteins of known structure. Irvine:University of California, Irvine. Available: http://blanco.biomol.uci.edu/Membrane_Proteins_xtal.html [accessed 20 June 2005].
Wilson S Olden K 2004 The environmental genome project: phase I and beyond Mol Intervent 4 147 156
Zhou C Zhang Y Zhou Y 2004 Structure prediction of membrane proteins Genomics Proteomics Bioinformatics 2 1 5 15629037
|
16263521
|
PMC1310928
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 13; 113(11):1622-1626
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8129
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8033ehp0113-00162716263522ResearchEnvironmental MedicineTubular and Glomerular Kidney Effects in Swedish Women with Low Environmental Cadmium Exposure Åkesson Agneta 1Lundh Thomas 2Vahter Marie 1Bjellerup Per 3Lidfeldt Jonas 4Nerbrand Christina 5Samsioe Göran 6Strömberg Ulf 2Skerfving Staffan 21 Institute of Environmental Medicine, Division of Metals and Health, Karolinska Institutet, Stockholm, Sweden2 Department of Occupational and Environmental Medicine, University Hospital, Lund, Sweden3 Department of Clinical Chemistry, Karolinska University Hospital, Huddinge, Sweden4 Department of Community Health, Malmö University Hospital, Malmö, Sweden5 Department of Medicine, and6 Department of Gynecology and Obstetrics, University Hospital, Lund, SwedenAddress correspondence to A. Åkesson, Institute of Environmental Medicine, Division of Metals and Health, Karolinska Institutet, Box 210, 171 77 Stockholm, Sweden. Telephone: 46-8-524-875-42. Fax: 46-8-33-70-39. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 11 7 2005 113 11 1627 1631 21 2 2005 11 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Cadmium is a well-known nephrotoxic agent in food and tobacco, but the exposure level that is critical for kidney effects in the general population is not defined. Within a population-based women’s health survey in southern Sweden (Women’s Health in the Lund Area, WHILA), we investigated cadmium exposure in relation to tubular and glomerular function, from 1999 through early 2000 in 820 women (71% participation rate) 53–64 years of age. Multiple linear regression showed cadmium in blood (median, 0.38 μg/L) and urine (0.52 μg/L; density adjusted = 0.67 μg/g creatinine) to be significantly associated with effects on renal tubules (as indicated by increased levels of human complex-forming protein and N-acetyl-β-d-glucosaminidase in urine), after adjusting for age, body mass index, blood lead, diabetes, hypertension, and regular use of nephrotoxic drugs. The associations remained significant even at the low exposure in women who had never smoked. We also found associations with markers of glomerular effects: glomerular filtration rate and creatinine clearance. Significant effects were seen already at a mean urinary cadmium level of 0.6 μg/L (0.8 μg/g creatinine). Cadmium potentiated diabetes-induced effects on kidney. In conclusion, tubular renal effects occurred at lower cadmium levels than previously demonstrated, and more important, glomerular effects were also observed. Although the effects were small, they may represent early signs of adverse effects, affecting large segments of the population. Subjects with diabetes seem to be at increased risk.
cadmiumdiabetesenvironmental exposureglomerular effectshypertensionkidneyleadpopulation-basedtubular effectswomen
==== Body
Identification of risk factors for chronic renal failure is essential in order to prevent reduction of life quality and life expectancy and to minimize the high costs of treatment. Cadmium is a widespread environmental pollutant known to cause renal damage (Järup et al. 1998). Apart from smoking, the major sources of cadmium exposure in the general population are cereals, vegetables, and shellfish. There is increasing evidence that toxic effects may occur at much lower exposure levels (Alfven et al. 2000; Buchet et al. 1990; Järup et al. 2000; Noonan et al. 2002) than those observed in occupational settings or in severely polluted environments. Still, the attempts to estimate the level of critical exposure for kidney effects have so far displayed large variations. Furthermore, possible effects in populations residing in areas with no particular industrial cadmium emission are undetermined.
Cadmium accumulates in the renal cortex and induces tubular toxicity (Barbier et al. 2005), which is first detected as increased urinary excretion of low-molecular-weight proteins and tubular enzymes. Glomerular dysfunction may also emerge, as demonstrated in heavily exposed subjects (Järup et al. 1995; Kido et al. 1990; Roels et al. 1989). It is not known, however, whether the glomerulus is affected by long-term low-level environmental exposure. Diabetes, an increasing health problem in many areas (King et al. 1998) and one of the leading causes of incident end-stage renal disease (Hostetter 2001), has been suggested to augment the risk of cadmium-induced kidney damage (Buchet et al. 1990). Also, hypertension and intake of nephrotoxic nonsteroid anti-inflammatory drugs (NSAIDs) (Fored et al. 2001) might interact with cadmium. However, these possible interactions need to be confirmed.
The aim of the present investigation was to assess the association between cadmium concentrations in blood and urine and a series of markers of tubular and glomerular function. To minimize dilution of the effects, we focused on women at the age when the accumulation of cadmium in the kidney is at its maximum. Women have increased cadmium accumulation compared with men (Järup et al. 1998; Nishijo et al. 2004b), due to a higher dietary cadmium absorption at low body iron stores (Åkesson et al. 2002; Berglund et al. 1994). In addition, we assessed whether diabetes, hypertension, and the use of NSAIDs increased the risk. The study was conducted in an area without particular industrial cadmium emission.
Materials and Methods
Study population.
The Women’s Health in the Lund Area (WHILA) study, a population-based study of all women 50–59 years of age in the community of Lund, southern Sweden (n = 10,766), started in December 1995 (Lidfeldt et al. 2001) and was extended in June 1999 to include health aspects of cadmium. This cohort was considered optimal for elucidation of remaining questions about dose–response relationships at low-dose cadmium exposure. The participation rate was 71% (n = 820). The exclusion criteria were renal cancer (n = 1) and lithium treatment (n = 3). Data were collected on various comorbidities, including diabetes and hypertension. Women were classed as having diabetes if they had a positive history, or if they had a non-fasting glucose > 8 mmol/L followed by a positive result in the oral glucose-tolerance test. Women were classed as hypertensive if they had received antihypertensive treatment or had a measured systolic and/or diastolic blood pressure ≥160 and ≥95, respectively (mean of two measurements after 15 min rest in seated position). Lists of medications and data on smoking were obtained, and weight and height were measured. Participants were asked to submit morning first-voided urine and blood samples. We obtained morning spot urine from 813 women and blood samples from 742. All samples were collected during 8 months from June 1999 through January 2000. The ethics committee at Lund University approved the WHILA study, and oral informed consent was obtained.
Analyses of exposure and kidney function.
Measurements included cadmium in blood as a measure mainly of ongoing exposure (expected to be fairly constant over time) and urine as a measure of body burden (cadmium in urine correlates well with cadmium in the kidney cortex; Järup et al. 1998; Orlowski et al. 1998). To control for possible confounding/effect modification, we also determined lead in blood (Lin et al. 2003). We used the following effect markers: cystatin C in serum (Dharnidharka et al. 2002) for calculation of glomerular filtration rate (GFR), creatinine clearance as markers of glomerular function, and human complex-forming protein (protein HC, α1-microglobulin), N-acetyl-β-d-glucosaminidase (NAG), and calcium in urine as markers of tubular damage.
We measured cadmium, lead, and calcium using inductively coupled plasma mass spectrometry (Barany et al. 1997). Cystatin C was determined by immunonephelometry (Dade Behring, Marburg, Germany). GFR = 77.24 × (cystatin C)−1.2623 (Larsson et al. 2004) and creatinine clearance = [(140 – age) × body weight (kg)]/[0.85 × serum creatinine (μM)] (Harmoinen et al. 2003). Creatinine was measured using a modified kinetic Jaffé method (Roche Diagnostics, Mannheim, Germany). We determined urinary protein HC by Mancini technique and polyclonal antibodies (DAKO A/S, Glostrup, Denmark) (Järup et al. 2000), and urinary NAG with a colorimetric method (Roche, Shionogi & Co. Ltd., Osaka, Japan).
Urinary spot samples need to be adjusted for dilution. Creatinine adjustment is most common, but a comparison of density and creatinine adjusted urinary cadmium indicated that creatinine did not adjust for all dilution-related variation of cadmium in urine. Because creatinine excretion is dependent upon meat intake and muscle mass (Davies et al. 2002; Suwazono et al. 2005), we chose to correct all urinary markers by the mean urinary density (1.015 g/mL) according to [urinary cadmium × (1.015 × 1,000) –1,000]/[(urinary density × 1,000) – 1,000]. However, creatinine-adjusted values are given for comparison.
Analytical performance.
All the equipment was tested, and possible contamination was below the limit of detection. For cadmium and lead in blood and cadmium and calcium in urine, the limits of detection were 0.12 μg/L, 0.26 μg/L, 0.31 μg/L, and 1.6 mg/L, respectively. For results below the limit of detection (mainly urinary cadmium), the concentration was set as the value factually obtained in the analysis. The imprecision of the method, calculated as the coefficient of variation for duplicate measurements, was 7.4 and 3.1% for cadmium and lead in blood and 8.5 and 6.4% for cadmium and calcium in urine. The analytical accuracy for blood (Seronorm, batch 404107; Nycomed, Oslo, Norway) was as follows (mean ± SD): 0.67 ± 0.08 for cadmium and 29 ± 1.1 for lead (n = 21; recommended, 0.67–0.70 and 31–39 μg/L, respectively). The data for certified blood samples from the U.K. National External Quality Assessment Service (n = 11) deviated on average by ± 7.9% for target values of 1.8–8.9 μg cadmium/L and ± 6.1% for 52–352 μg lead/L. The results for urine (Seronorm, batch 102021) were 140 ± 8.1 mg/L (n = 20; recommended 130 mg/L) for calcium and 0.45 ± 0.07 μg/L (n = 20; recommended, 0.35 μg/L) for cadmium. The result for the certified urine samples from Centre de Toxicologie du Quebec Interlaboratory Comparison Program for cadmium was 0.76 ± 0.09 and 3.6 ± 0.22 μg/L (n = 11; certified 0.79 and 3.6 μg/L), respectively.
The imprecision was 2.7% for cystatin C (n = 6), 16% for protein HC (n = 10; limit of detection = 0.7 mg/L), and < 10% for urinary NAG (n = 68).
Statistical analyses.
We used Spearman’s rank correlation analysis to assess univariate associations. The cadmium-associated kidney effect markers were further evaluated in multiple linear regression models, where each kidney effect marker was evaluated in relation to cadmium and confounders/covariates. The dependent effect markers (continuous) were not dichotomized (Ragland 1992). No log-transformation was needed as indicated by residual and goodness-of-fit analyses. We evaluated possible effect modification (interactions) for cadmium and lead. Lowest observed effect levels were assessed for each effect marker for categorized urinary cadmium using Dunnett’s post hoc test, including significant confounders and covariates in the models. All tests were two sided, and statistics were performed using SPSS (version 12.01; SPSS Inc., Chicago, IL, USA).
Results
The study population characteristics, exposure variables, and kidney effect markers are presented in Table 1. The proportion of subjects with diabetes was slightly higher in the present study compared with those participating in the whole WHILA cohort (6.4%) (Lidfeldt 2003). The proportion of hypertensive subjects was, however, similar (Lidfeldt 2003). Those who had ever smoked had 90% higher cadmium concentrations in blood and 40% higher in urine compared with never-smokers, who had 0.30 μg/L and 0.45 μg/L cadmium in blood and urine, respectively.
The univariate associations between cadmium and kidney effect markers as well as those with possible confounders and effect modifiers are shown in Table 2. Cadmium in both blood and urine was associated with all kidney effect markers except serum creatinine and urinary calcium, which were not included in further analysis. Using cystatin C instead of estimated GFR, and creatinine-adjusted markers in urine instead of density adjusted ones had no major impact on the results.
Multivariate analyses.
In the multiple linear regression analysis, each of the cadmium-associated kidney effect markers was tested separately, as were cadmium levels in urine and blood (Table 3). We included in the models the covariates age, body mass index (BMI), and blood lead, as well as the possible kidney-effect modifiers diabetes, hypertension, and use of NSAIDs. Never-smokers were analyzed separately. Cadmium in urine was significantly associated with GFR, creatinine clearance, protein HC, and NAG, after controlling for confounders and adjusting for other covariables (Table 3). Similar results were obtained for blood cadmium. In never-smokers, cadmium remained associated with protein HC and NAG and became significantly associated with creatinine clearance (blood cadmium). Lead was significantly associated with GFR and creatinine clearance.
Because there were no associations between cadmium and GFR or creatinine clearance (for urinary cadmium) in never-smokers, we tested whether there was confounding through a non-cadmium-dependent effect of smoking by including pack-years in the multiple regression models. Smoking (pack-years) was not significantly associated with GFR or creatinine clearance (p ≥0.1).
We assessed possible interactions between cadmium and blood lead, diabetes (insulin-treated vs. the rest), hypertension, or use of NSAIDs, and between blood lead and diabetes, hypertension, or use of NSAIDs. For NAG, there was an interaction between urinary cadmium and diabetes (insulin-treated vs. other diabetics and nondiabetics; regression coefficients (β): diabetics = 2.3, nondiabetics = 0.8; R2 = 0.10; p = 0.042) (Table 3). For protein HC, there was a close to significant interaction between urinary cadmium and diabetes (β: diabetics = 4.0, nondiabetics = 1.3; R2 = 0.09; p = 0.07), which became significant in never-smokers (β: diabetics = 27, nondiabetics = 1.9; R2 = 0.18; p < 0.001). Similar interactions were observed between blood cadmium and diabetes. Hypertension, NSAID use, and blood lead showed no significant interactions with cadmium exposure. However, there was an interaction between blood lead and diabetes for GFR in never-smokers (p = 0.005).
Lowest observed effect level.
Protein HC, NAG (diabetics excluded) (Figure 1A), and creatinine clearance (Figure 1B), after adjustment for blood lead, differed significantly between the group with lowest exposure level (urinary cadmium < 0.5 μg/L; mean, 0.36 μg cadmium/L = 0.48 μg cadmium/g creatinine) and that with the next lowest exposure level (0.50–0.75 μg/L; mean, 0.61 μg cadmium/L = 0.79 μg cadmium/g creatinine). For GFR, the group with the next highest exposure level [urinary cadmium, 0.75–1 μg cadmium/L; mean, 0.86 μg cadmium/L = 1.0 μg cadmium/g creatinine; adjusted for age, BMI, and blood lead (each categorized into four groups) and for NSAID use (into 0 or 1); Figure 1C] differed from the lowest level. For blood cadmium, associations were present in the exposure category 0.5–1 μg/L (mean, 0.69 μg/L) for protein HC (p = 0.036) and NAG (p = 0.024). For GFR, an association was seen only at blood cadmium > 1 μg/L (mean, 1.8 μg/L; p < 0.001) after adjustment for significant covariates.
Discussion
This population-based study of upper-middle-age women, representative of the general population of southern Sweden, showed clear associations between cadmium and the renal tubular-effect markers protein HC and NAG, even at the low levels of cadmium found in never-smokers. Cadmium potentiated the diabetes-induced effects on the kidney. There was also a clear association between cadmium and GFR or creatinine clearance.
This study has several methodologic advantages, including the large sample size and high participation rate, individual exposure assessment with high analytical accuracy, and inclusion of several different outcomes of renal effects. Despite the low cadmium concentrations, we had a high analytical accuracy. Any imprecision would have caused a bias toward the null.
The study population differed somewhat from the total WHILA population and Sweden (4–7%) (Lidfeldt 2003). Hence, there was a slight overrepresentation of diabetics, which may cause an overestimate of cadmium effects. However, because we controlled for diabetes in the statistical models, this is not a problem. Overcontrol and collinearity may occur in a statistical analysis such as that performed in this study. Smoking is then an obvious problem, which we handled by separate analysis in never-smokers. Lead and BMI were included, which means a risk of some overcontrol.
Another problem, common in the interpretation of data from cross-sectional studies, is that the exposure is measured at the same time as the effects, which may not be the etiologically relevant period. This may be problematic for blood cadmium, because it largely reflects recent exposure, but not for urinary cadmium, which is a good estimate of the integrated low-level exposure over decades (Järup et al. 1998). It is known that kidney deterioration, due to both aging and high cadmium exposure, increases the excretion of cadmium in urine, resulting in lower kidney cadmium and eventually lower urinary cadmium. However, the present participants were below the age when the kidney cadmium starts to decrease, and the exposure was relatively low.
The present cadmium concentrations are comparable with, or slightly higher than, those in other recent studies from Sweden (Åkesson et al. 2002; Järup et al. 2000; Olsson et al. 2002) and the United States [Centers for Disease Control and Prevention (CDC) 2003; Noonan et al. 2002; Paschal et al. 2000] but lower than those in more contaminated areas of Europe (Buchet et al. 1990; Hotz et al. 1999) and much lower than in certain areas in Japan (Suwazono et al. 2000; Yamanaka et al. 1998). Despite the present low cadmium levels, there were clear effects on the kidney. The associations between cadmium and biomarkers of several different renal effects support causality. It is unlikely that they are merely a result of parallel phenomena, impaired tubular reabsorption (protein HC), or increased general turnover of tubular cells (NAG). The associations with blood cadmium also preclude such an interpretation. Because smoking is a major source of cadmium exposure (Järup et al. 1998), the possibility of confounding through a non-cadmium-dependent effect of smoking must be considered. However, because we found cadmium-associated effects on NAG and protein HC even in never-smokers, and there was no effect of smoking on creatinine clearance or GFR, this is unlikely.
The lead levels were low. The association between blood lead and GFR and creatinine clearance may indicate either an effect on GFR at low lead exposure (Lin et al. 2003) or reverse causality. In the case of cadmium, reverse causality seems highly unlikely. Even though data may imply that a decrease in GFR causes increased blood cadmium concentrations, the inverse associations between the glomerular effect markers and cadmium in urine rather indicate that reduced GFR does not reduce the clearance of cadmium. In addition, lead is bound to high-molecular-weight plasma albumin, and cadmium to metallothionein, a small polypeptide that is easily filtered through the glomerulus.
The lowest observed effect level, defined as the mean urinary cadmium in the exposure category that displayed significantly different levels of effect markers compared with the lowest urinary cadmium category, was 0.6 μg cadmium/L (0.8 μg/g creatinine), corresponding to approximately 20 μg cadmium/g kidney cortex. The lowest observed effect level is lower than in previous studies that observed effects at low-level cadmium exposure (Buchet et al. 1990; Järup et al. 2000; Noonan et al. 2002). This is probably due to homogeneity of the population, absence of healthy worker effects (Järup et al. 2000), and good precision in analyses of the exposure and effect markers.
More important, we found a similar lowest observed effect level for creatinine clearance as for the tubular markers. Although usually considered an index of glomerular function, the creatinine clearance may partly reflect a proximal tubular dysfunction, because creatinine is not only filtrated but also secreted in the tubuli (Wuyts et al. 2003). On the other hand, the cadmium-associated increase in GFR, occurring in the next highest cadmium stratum (0.86 μg/L urine = 1.0 μg/g creatinine), clearly indicates an effect on the glomerular function. An observation in this contex that supports an effect of cadmium on GFR is the reported ecologic association between end-stage renal disease and distance to cadmium-emitting industrial plants (Hellström et al. 2001).
Cadmium has been suggested to cause hypertension, but no such effect was seen here, in agreement with other studies (Staessen et al. 2000). Also, we did not observe any synergism between cadmium and hypertension on the kidney effects. However, there might be a dilution of the group by cases with mild hypertension. Further, it has been reported, both from experimental and epidemiologic studies, that cadmium increases the risk of type II diabetes (Han et al. 2003; Schwartz et al. 2003), which was not supported by the present study. As expected, diabetes affected the kidney function, although only in insulin-dependent women, of whom about half had type II diabetes. More important, we found an interaction between cadmium and diabetes, as suggested in previous studies (Buchet et al. 1990). Hence, the lowest observed effect level is expected to be lower in diabetics but could not be evaluated because of too few cases. The incidence of diabetes is increasing (King et al. 1998), and because diabetes is the leading cause of end-stage renal disease (Hostetter 2001), this has important public health implications. The incidence of renal replacement therapy in Sweden is 125 per million, with an estimated increased prevalence of 5% per year (Swedish National Board of Health and Welfare 2003).
The nephrotoxic effects in the present study appear small in a clinical context, and only a few percent of the variances were explained by cadmium. However, the increase in the effect markers indicates renal toxicity, which should be considered an early sign of severe health effects (Nishijo et al. 2004a). Because it concerns a large segment of the population worldwide, the results are of public health concern. Although the cadmium-induced kidney effects in several studies have been associated with decreasing GFR (Järup et al. 1995; Kido et al. 1990; Roels et al. 1989), a positive aspect is that progression of the very early effect may not always occur when the exposure is substantially decreased (Hotz et al. 1999). It should, however, be emphasized that in areas with exposure to cadmium mainly through diet, the long half-time of cadmium in the soil will hamper a decrease of the exposure. Thus, far-reaching mitigation will be needed in addition to actions against smoking.
The late A. Schütz made invaluable contributions to the study. We thank H. Ottosson, A. Akantis, A.-M. Åberg, and B. Erdling for skillful technical assistance.
Funding was provided by the Swedish Research Council/Medicine; Medical Faculty of Lund University; Karolinska Institutet; the National Swedish Environmental Protection Agency; the Swedish Foundation for Strategic and Environmental Research; the Swedish Society of Medicine, Primary Care, R&D, County Council of Skåne; the Swedish Research Council for Environment, Agricultural Sciences, and Spatial Planning; and the Swedish Council for Working Life and Social Research.
Figure 1 Associations (crude) between urinary NAG (A), creatinine clearance (B), and GFR (C) and urinary cadmium (categorized) in a population-based study from 1999 through early 2000 on 816 women in southern Sweden. Boxes indicate 25th, 50th (solid line), and 75th percentiles, and whiskers indicate minimum and maximum, excluding outliers (circles; a few, not shown in the figure but included in all the calculations). Numbers inside boxes indicate the number of samples. The dashed line indicates the median in the lowest urinary cadmium exposure category. p-Values for differences between the lowest exposure group and the following groups are indicated (Dunnett’s test including the significant confounders and covariates according to Table 3).
Table 1 Participant characteristics and data on exposure and kidney effect markers in a population-based study from 1999 through early 2000 on 816 women in southern Sweden.
Variable (unit) Median (5–95% percentiles) No. of samples
Population characteristic
Age (years) 58 (54–63) 816
BMI (kg/m2) 26.2 (20.3–33.9)
Smokers: never/former/current (%) 54/25/21a
Diabetes: all/insulin dependent (%) 10/1.7a
Hypertension: all/drug treated (%) 31/18a
Regular use of NSAIDs (%) 6a
Exposure variables
Blood cadmium (μg/L) 0.38 (0.16–1.8) 725
Urinary cadmium (μg/L)b 0.52 (0.24–1.3) 807
Urinary cadmium (μg/g creatinine) 0.67 (0.31–1.6)
Blood lead (μg/L) 22 (11–46) 726
Kidney effect markers
Serum cystatin C (mg/L) 0.81 (0.65–1.0) 721
GFR (mL/min)c 101 (74–133)
Serum creatinine (μmol/L) 92 (73–116) 713
Creatinine clearance (mL/min)d 72 (51–105)
Urinary protein HC (μg/L)b 2.4 (0.98–7.9) 806
Urinary protein HC (mg/g creatinine) 3.1 (0.13–1.2)
Urinary NAG (U/L)b 1.2 (0.22–3.6) 806
Urinary NAG (U/g creatinine) 1.4 (1.1–11)
Urinary calcium (mg/L)b 135 (56–267) 809
Urinary calcium (mg/g creatinine) 170 (62–366)
BMI, body mass index.
a Data are presented as percent.
b Adjusted to mean density 1.015 g/mL.
c Calculated: 77.24 × (serum cystatin C)−1.2623.
d Calculated: [(140 – age) × body weight (kg)]/[0.85 × serum creatinine (μM)]. Mean urinary creatinine = 0.85 g/L; conversion factors: cadmium: 1 μg = 8.89 nmol; 1.0 μg/g creatinine ≈ 1.0 nmol/mmol creatinine; lead: 1 μg = 4.83 nmol.
Table 2 Associations between exposure and effect markers (Spearman’s rank correlation coefficients).
Age BMI Pack-years Blood cadmium Urinary cadmium Blood lead GFR Serum creatinine Creatinine clearance Urinary protein HC
Blood cadmium −0.01 −0.14# 0.56#
Urinary cadmium −0.02 −0.15# 0.42# 0.57#
Blood lead −0.03 −0.08* 0.18# 0.20# 0.15#
GFR −0.28# −0.27# −0.05 −0.13# −0.12# −0.11*
Serum creatinine 0.12* −0.08* 0.00 0.02 0.05 0.13# −0.38#
Creatinine clearance NR NR −0.02 −0.08* −0.13# −0.13# 0.11# −0.62#
Urinary protein HC 0.05 −0.18# 0.08* 0.15# 0.18# −0.01 −0.05 −0.02 −0.11*
Urinary NAG 0.06 −0.03 0.12# 0.13# 0.23# 0.02 −0.13# 0.09* −0.09* 0.21#
Urinary calcium −0.03 −0.04 −0.03 0.01 −0.02 0.12# 0.16# −0.15# 0.06 −0.04
Abbreviations: BMI, body mass index; NR, not relevant, as included in the calculation of creatinine clearance.
* p ≤0.05.
# p ≤0.001.
Table 3 Associations between markers of cadmium exposure and effects in a population-based study on 816 Swedish women, allowing for other risk factors, performed in all subjects and never-smokers separately.
All
Never-smokers
Dependent variable Independent variable β 95% CI R2 β 95% CI R2
GFR (mL/min) Urinary cadmium (μg/L) −7.9 −11 to −4.3 0.15 −5.0 −11 to 0.9 0.16
Age (year) −1.5 −1.9 to −1.0 −1.3 −1.9 to −0.7
BMI (kg/m2) −1.0 −1.3 to −0.7 −1.1 −1.5 to −0.7
Blood lead (μg/L) −0.20 −0.32 to −0.09 −0.26 −0.43 to −0.09
Diabetesa NS −25 −46 to −5.0
Hypertensionb NS NS
NSAIDsc −6.8 −12 to −1.2 NS
Blood cadmium −4.2 −6.6 to −1.9 0.15 −6.0 −15 to 3.0 0.16
Age (year) −1.5 −1.9 to −1.0 −1.2 −1.8 to −0.7
BMI (kg/m2) −1.0 −1.3 to −0.7 −1.1 −1.6 to −0.7
Blood lead (μg/L) −0.2 −0.3 to −0.07 −0.2* −0.4 to −0.07
Diabetes NS −25 −45 to −4.8
Hypertension NS NS
NSAIDs −6.1 −12 to −0.5 NS
Creatinine clearance (mL/min) Urinary cadmium (μg/L) −4.3 −8.0 to −0.7 0.03 −3.5 −9.9 to 2.8 0.05
Blood lead (μg/L) −0.18 −0.30 to −0.06 −0.3 −0.5 to −0.1
Diabetes NS −29 −51 to −8.0
Hypertension NS NS
NSAIDs NS NS
Blood cadmium −1.6 −4.0 to 0.7 0.03 −9.8 −19 to −0.5 0.06
Blood lead (μg/L) −0.18 −0.30 to −0.07 −0.3 −0.5 to −0.1
Diabetes NS −29 −50 to −8.0
Hypertension 3.0 0.3 to 5.9 NS
NSAIDs NS NS
Urinary protein HC (μg/L) Urinary cadmium (μg/L) 1.4 0.9 to 1.8 0.09 2.1* 1.3 to 2.8 0.13
Age (year) NS NS
BMI (kg/m2) −0.06 −0.10 to −0.02 −0.08 −0.13 to −0.02
Blood lead (μg/L) NS NS
Diabetes 3.5 2.2 to 4.9 5.1 3.1 to 7.1
Hypertension NS 0.57 0.05 to 1.1
NSAIDs NS NS
Blood cadmium 0.5 0.2 to 0.8 0.06 1.7* 0.5 to 3.0 0.09
Age (year) NS NS
BMI (kg/m2) −0.07 −0.11 to −0.02 −0.08 −0.14 to −0.02
Blood lead (μg/L) NS NS
Diabetes 3.5 2.2 to 4.8 5.8 3.4 to 8.1
Hypertension NS NS
NSAIDs NS NS
Urinary NAG (U/L) Urinary cadmium (μg/L) 0.9* 0.6 to 1.1 0.09 0.8 0.4 to 1.2 0.10
Age (year) NS NS
BMI (kg/m2) NS NS
Blood lead (μg/L) NS NS
Diabetes 1.5 0.9 to 2.2 3.0 1.9 to 4.1
Hypertension NS NS
NSAIDs NS NS
Blood cadmium 0.4 0.2 to 0.5 0.05 0.5 −0.05 to 1.1 0.06
Age (year) NS NS
BMI (kg/m2) NS NS
Blood lead (μg/L) NS NS
Diabetes 1.5 0.8 to 2.1 2.9 1.7 to 4.0
Hypertension NS NS
NSAIDs NS NS
Abbreviations: β, regression coefficient; 95% CI, 95% confidence interval; adjusted R2, explained variance; NS, not significant.
a Insulin treated vs. all others, yes = 1.
b Hypertension, yes = 1.
c NSAIDs, yes = 1.
* Significant interaction with diabetes (described in text).
==== Refs
References
Åkesson A Berglund M Schutz A Bjellerup P Bremme K Vahter M 2002 Cadmium exposure in pregnancy and lactation in relation to iron status Am J Public Health 92 2 284 287 11818307
Alfven T Elinder CG Carlsson MD Grubb A Hellström L Persson B 2000 Low-level cadmium exposure and osteoporosis J Bone Miner Res 15 8 1579 1586 10934657
Barany E Bergdahl IA Schutz A Skerfving S Oskarsson A 1997 Inductively coupled plasma mass spectrometry for direct multi-element analysis of diluted human blood and serum J Anal Atomic Spectrom 12 1005 1009
Barbier O Dauby A Jacquillet G Tauc M Poujeol P Cougnon M 2005 Zinc and cadmium interactions in a renal cell line derived from rabbit proximal tubule Nephron Physiol 99 3 74 84
Berglund M Åkesson A Nermell B Vahter M 1994 Intestinal absorption of dietary cadmium in women depends on body iron stores and fiber intake Environ Health Perspect 102 1058 1066 7713018
Buchet JP Lauwerys R Roels H Bernard A Bruaux P Claeys F 1990 Renal effects of cadmium body burden of the general population Lancet 336 8717 699 702 1975890
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:Center for Disease Control and Prevention.
Davies KM Heaney RP Rafferty K 2002 Decline in muscle mass with age in women: a longitudinal study using an indirect measure Metabolism 51 7 935 939 12077745
Dharnidharka VR Kwon C Stevens G 2002 Serum cystatin C is superior to serum creatinine as a marker of kidney function: a meta-analysis Am J Kidney Dis 40 2 221 226 12148093
Fored CM Ejerblad E Lindblad P Fryzek JP Dickman PW Signorello LB 2001 Acetaminophen, aspirin, and chronic renal failure N Engl J Med 345 25 1801 1808 11752356
Han JC Park SY Hah BG Choi GH Kim YK Kwon TH 2003 Cadmium induces impaired glucose tolerance in rat by down-regulating GLUT4 expression in adipocytes Arch Biochem Biophys 413 2 213 220 12729619
Harmoinen A Lehtimaki T Korpela M Turjanmaa V Saha H 2003 Diagnostic accuracies of plasma creatinine, cystatin C, and glomerular filtration rate calculated by the Cockcroft-Gault and Levey (MDRD) formulas Clin Chem 49 7 1223 1225 12816933
Hellström L Elinder CG Dahlberg B Lundberg M Järup L Persson B 2001 Cadmium exposure and end-stage renal disease Am J Kidney Dis 38 5 1001 1008 11684553
Hostetter TH 2001 Prevention of end-stage renal disease due to type 2 diabetes N Engl J Med 345 12 910 912 11565525
Hotz P Buchet JP Bernard A Lison D Lauwerys R 1999 Renal effects of low-level environmental cadmium exposure: 5-year follow-up of a subcohort from the Cadmibel study Lancet 354 9189 1508 1513 10551497
Järup L Berglund M Elinder CG Nordberg G Vahter M 1998 Health effects of cadmium exposure—a review of the literature and a risk estimate Scand J Work Environ Health 24 suppl 1 1 51 9569444
Järup L Hellström L Alfven T Carlsson MD Grubb A Persson B 2000 Low level exposure to cadmium and early kidney damage: the OSCAR study Occup Environ Med 57 10 668 672 10984338
Järup L Persson B Elinder CG 1995 Decreased glomerular filtration rate in solderers exposed to cadmium Occup Environ Med 52 12 818 822 8563845
Kido T Nogawa K Ishizaki M Honda R Tsuritani I Yamada Y 1990 Long-term observation of serum creatinine and arterial blood pH in persons with cadmium-induced renal dysfunction Arch Environ Health 45 1 35 41 2317087
King H Aubert RE Herman WH 1998 Global burden of diabetes, 1995–2025: prevalence, numerical estimates, and projections Diabetes Care 21 9 1414 1431 9727886
Larsson A Malm J Grubb A Hansson LO 2004 Calculation of glomerular filtration rate expressed in mL/min from plasma cystatin C values in mg/L Scand J Clin Lab Invest 64 1 25 30 15025426
Lidfeldt J 2003. Health in Middle-Aged Women, with Emphasis on Features of the Metabolic Syndrome and Related Markers for Diabetes [PhD Thesis]. Lund, Sweden:Lund University.
Lidfeldt J Nerbrand C Samsioe G Schersten B Agardh CD 2001 A screening procedure detecting high-yield candidates for OGTT. The Women’s Health in the Lund Area (WHILA) study: a population based study of middle-aged Swedish women Eur J Epidemiol 17 10 943 951 12188015
Lin JL Lin-Tan DT Hsu KH Yu CC 2003 Environmental lead exposure and progression of chronic renal diseases in patients without diabetes N Engl J Med 348 4 277 286 12540640
Nishijo M Nakagawa H Morikawa Y Kuriwaki J Katsuyuki M Kido T 2004a Mortality in a cadmium polluted area in Japan Biometals 17 5 535 538 15688859
Nishijo M Satarug S Honda R Tsuritani I Aoshima K 2004b The gender differences in health effects of environmental cadmium exposure and potential mechanisms Mol Cell Biochem 255 1–2 87 92 14971649
Noonan CW Sarasua SM Campagna D Kathman SJ Lybarger JA Mueller PW 2002 Effects of exposure to low levels of environmental cadmium on renal biomarkers Environ Health Perspect 110 151 155 11836143
Olsson IM Bensryd I Lundh T Ottosson H Skerfving S Oskarsson A 2002 Cadmium in blood and urine—impact of sex, age, dietary intake, iron status, and former smoking—association of renal effects Environ Health Perspect 110 1185 1190 12460796
Orlowski C Piotrowski JK Subdys JK Gross A 1998 Urinary cadmium as indicator of renal cadmium in humans: an autopsy study Hum Exp Toxicol 17 6 302 306 9688352
Paschal DC Burt V Caudill SP Gunter EW Pirkle JL Sampson EJ 2000 Exposure of the U.S. population aged 6 years and older to cadmium: 1988–1994 Arch Environ Contam Toxicol 38 3 377 383 10667937
Ragland DR 1992 Dichotomizing continuous outcome variables: dependence of the magnitude of association and statistical power on the cutpoint Epidemiology 3 5 434 440 1391136
Roels HA Lauwerys RR Buchet JP Bernard AM Vos A Oversteyns M 1989 Health significance of cadmium induced renal dysfunction: a five year follow up Br J Ind Med 46 11 755 764 2686749
Schwartz GG Il’yasova D Ivanova A 2003 Urinary cadmium, impaired fasting glucose, and diabetes in the NHANES III Diabetes Care 26 468 470 12547882
Swedish National Board of Health and Welfare 2003. Renal Replacement Therapy in Sweden 1991–2002 [in Swedish]. Stockholm, Sweden:National Board of Health and Welfare.
Staessen JA Kuznetsova T Roels HA Emelianov D Fagard R 2000 Exposure to cadmium and conventional and ambulatory blood pressures in a prospective population study. Public Health and Environmental Exposure to Cadmium Study Group Am J Hypertens 13 2 146 156 10701814
Suwazono Y Akesson A Alfven T Jarup L Vahter M 2005 Creatinine versus specific gravity adjusted urinary cadmium concentrations Biomarkers 10 2–3 117 126 16076727
Suwazono Y Kobayashi E Okubo Y Nogawa K Kido T Nakagawa H 2000 Renal effects of cadmium exposure in cadmium non-polluted areas in Japan Environ Res 84 1 44 55 10991781
Wuyts B Bernard D Van den Noortgate N Van de Walle J Van Vlem B De Smet R 2003 Reevaluation of formulas for predicting creatinine clearance in adults and children, using compensated creatinine methods Clin Chem 49 6 pt 1 1011 1014 12766016
Yamanaka O Kobayashi E Nogawa K Suwazono Y Sakurada I Kido T 1998 Association between renal effects and cadmium exposure in cadmium-nonpolluted area in Japan Environ Res 77 1 1 8 9593622
|
16263522
|
PMC1310929
|
CC0
|
2021-01-04 23:41:32
|
no
|
Environ Health Perspect. 2005 Nov 11; 113(11):1627-1631
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8033
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8159ehp0113-00163216263523ResearchChildren's HealthDecline of Ambient Air Pollution Levels and Improved Respiratory Health in Swiss Children Bayer-Oglesby Lucy 1Grize Leticia 1Gassner Markus 2Takken-Sahli Kathy 3Sennhauser Felix H. 4Neu Urs 5Schindler Christian 1Braun-Fahrländer Charlotte 11 Institute of Social and Preventive Medicine of the University of Basel, Basel, Switzerland2 School Health Service, Grabs, Switzerland3 School Health Service, Zürich, Switzerland4 University Children’s Hospital, Zürich, Switzerland5 Institute of Geography, University of Bern, Bern, SwitzerlandAddress correspondence to L. Bayer-Oglesby, Institute of Social and Preventive Medicine, University of Basel, Steinengraben 49, CH-4051 Basel, Switzerland. Telephone: 41-61-267-60-66. Fax: 41-61-267-61-90. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 21 6 2005 113 11 1632 1637 1 4 2005 21 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The causality of observed associations between air pollution and respiratory health in children is still subject to debate. If reduced air pollution exposure resulted in improved respiratory health of children, this would argue in favor of a causal relation. We investigated whether a rather moderate decline of air pollution levels in the 1990s in Switzerland was associated with a reduction in respiratory symptoms and diseases in school children. In nine Swiss communities, 9,591 children participated in cross-sectional health assessments between 1992 and 2001. Their parents completed identical questionnaires on health status and covariates. We assigned to each child an estimate of regional particles with an aerodynamic diameter < 10 μg/m3 (PM10) and determined change in PM10 since the first survey. Adjusted for socioeconomic, health-related, and indoor factors, declining PM10 was associated in logistic regression models with declining prevalence of chronic cough [odds ratio (OR) per 10-μg/m3 decline = 0.65, 95% confidence interval (CI), 0.54–0.79], bronchitis (OR = 0.66; 95% CI, 0.55–0.80), common cold (OR = 0.78; 95% CI, 0.68–0.89), nocturnal dry cough (OR = 0.70; 95% CI, 0.60–0.83), and conjunctivitis symptoms (OR = 0.81; 95% CI, 0.70–0.95). Changes in prevalence of sneezing during pollen season, asthma, and hay fever were not associated with the PM10 reduction. Our findings show that the reduction of air pollution exposures contributes to improved respiratory health in children. No threshold of adverse effects of PM10 was apparent because we observed the beneficial effects for relatively small changes of rather moderate air pollution levels. Current air pollution levels in Switzerland still exceed limit values of the Swiss Clean Air Act; thus, children’s health can be improved further.
air pollutionchildrencross-sectional surveysdeclinerespiratory healthsymptoms
==== Body
The causality of observed associations between air pollution and respiratory health in children is still subject to debate, although numerous studies have reported adverse effects of air pollution on the respiratory health of children, using indicators of general air pollution (Braun-Fahrländer et al. 1997; Chen et al. 1998; Gauderman et al. 2002; Horak et al. 2002; Hruba et al. 2001; McConnell et al. 1999) and of traffic-related air pollution (Brauer et al. 2002; Gehring et al. 2002; Hirsch et al. 1999; Janssen et al. 2003; Nicolai et al. 2003; van Vliet et al. 1997; Venn et al. 2001; Wjst et al. 1993). If it could be shown that reduced air pollution exposures improve the respiratory health of children, this would argue in favor of a causal relation. So far, only a few studies have investigated the expected beneficial effects of air pollution reduction on respiratory health in children. In cross-sectional analysis, the tremendous decline of coal combustion–related air pollution in East Germany after reunification was associated with a decline of respiratory symptoms (Heinrich 2003) and improved lung function (Frye et al. 2003) in children. In a cohort of children, those who moved within California to areas with lower PM10 (particles with an aerodynamic diameter < 10 μg/m3) levels showed increased lung function growth, whereas those moving to more polluted areas had a decreased growth (Avol et al. 2001). McConnell et al. (2003) observed that bronchitis symptoms, assessed yearly for 4 years in a cohort of children with asthma, varied with the yearly variability of PM2.5 (particles with an aerodynamic diameter < 2.5 μg/m3), nitrogen dioxide, and organic carbon.
In the first cross-sectional assessment of the Swiss Surveillance Program of Childhood Allergy and Respiratory Symptoms with Respect to Air Pollution and Climate (SCARPOL) in 1992–1993, Braun-Fahrländer et al. (1997) reported that rates of respiratory symptoms and diseases, adjusted for individual risk factors, were positively associated with PM10, NO2, and sulfur dioxide in children living in 10 urban, suburban, rural, and alpine areas of Switzerland. Since then, air pollution abatement measures (emission limits for industries, introduction of low-sulfur heating oil and catalytic converters) implemented after the Swiss Clean Air Act (1985) have led to declining air pollution levels in Switzerland [Swiss Agency for the Environment, Forest and Landscape (SAEFL) 2003; Kuebler et al. 2001]. In contrast to East Germany, where the tremendous air pollution decline in the 1990s went hand in hand with dramatic political and social changes, the political and social system in Switzerland has been very stable for many decades, which is an asset in our study. We hypothesize that if the health effects observed in SCARPOL in 1993 (Braun-Fahrländer et al. 1997) were causal, a) the observed reduction of PM10 in Switzerland since the first cross-section of SCARPOL would be associated with a reduction of prevalence rates of respiratory symptoms and diseases in the second health assessment phase, and b) the average reduction of symptom prevalence would be more pronounced in areas with stronger reduction of air pollution levels.
Methods
Study population and design.
In 10 Swiss communities covering a broad range of urbanization, air pollution levels, and climatic conditions, 10,397 school children (76.1%) ages 6–15 years have participated in cross-sectional, questionnaire-based health assessments between 1992 and 2001. For urban areas, we chose Lugano, Zürich, Bern, and Geneva; for suburban areas, Anières and Biel; for rural areas, Langnau, Payerne, and Rheintal; and for an alpine area, Montana. Because of the absence of PM10 data, we had to exclude children of Rheintal for this analysis, resulting in a sample of 9,591 children. The detailed recruiting procedure for the first cross-sectional health assessment in 1992–1993, which has also been applied for subsequent assessments, has been described previously (Braun-Fahrländer et al. 1997). Children of three school grades (first, fourth, and eighth) were recruited in the first phase in 1992–1993; in the second phase, one grade was enrolled each school year (first grade in 1998–1999, eighth grade in 1999–2000, and fourth grade in 2000–2001) (Table 1). This resulted in two repeated cross-sectional surveys for each age-group that are 6, 7, and 8 years apart for the first, eighth and fourth grade, respectively. The ethics committees of the Universities of Geneva and Bern approved the study protocol.
Health assessment.
For all participating children, we collected identical parent-completed questionnaires on health status, family history of disease, spare-time activities, indoor exposures, and residential situation. The questionnaire included the core questions on asthma and allergy of the International Study of Asthma and Allergy in Childhood (ISAAC) (Asher et al. 1995). Definitions of symptoms and diseases examined in this analysis are given in Table 2.
Assessment of air pollution exposures.
We assigned to each child an estimate of regional PM10 for the year preceding the questionnaire date, obtained from one fixed monitoring station in each community. Children were living within a few (3–5) kilometers of the monitors. Monitors were located in the centers of the communities, with the exception of the rural monitors in Payerne and Montana. Röösli et al. (2000, 2001) have demonstrated that in Switzerland, PM10 levels are homogeneously distributed within regions and are not significantly affected by local traffic, justifying the single-monitor approach for the assignment of PM10 exposures. Because PM10 measurements started in four communities not before 1993, we assigned annual means of 1993 to all children participating in the first cross-section (1992–1993 school year). Annual means of PM10 have been estimated for 1993 and for 1997–2000. We converted Harvard Impactor data of 1993 to DIGITEL HiVol values based on collocated measurements of the two monitors for 24 months (Krütli and Monn 1999). Between 1997 and 2000, PM10 was measured in the nine regions with DIGITEL HiVol samplers (DIGITEL 1999). In addition, we obtained temperature measurements from the fixed monitoring stations for calculating the number of cold days (days with the maximum temperature below zero degrees Celsius) for each region and year.
Statistical analysis.
To analyze the association between change of air pollution levels and change of respiratory health, we used multivariate logistic regression models. For the children participating in the second health assessment phase (school years 1998–1999, and 2000–2001), change in PM10 was calculated as the difference between the assigned PM10 estimate and the 1993 baseline values corresponding to their area. For the children participating in the first health assessment phase (school year 1992–1993), change in PM10 was set to zero. In addition to change in PM10, a dummy variable for each region was included in the regression models. To test for community correlation possibly introduced by clustering of uncontrolled covariates, we also evaluated random-effect models.
For the nine health end points, we computed adjusted odds ratios (ORs) associated with a decline of 10 μg/m3 in PM10. A priori, our regression models also included those covariates that had an impact on the effect estimates or were identified as confounders of air pollution effects in the first cross-sectional analysis of 1992–1993 (Braun-Fahrländer et al. 1997). Covariates included
Socioeconomic factors (age, sex, nationality, parental education, number of siblings, farming status)
Health-related factors (low birth weight, breastfeeding, child who smokes, family history of asthma, bronchitis, and/or atopy)
Indoor factors (mother who smokes, humidity, mode of heating and cooking, carpeting, pets allowed in bedroom)
Avoidance behavior with respect to allergies (carpet or pets removed for health reasons)
Questionnaire-related factors (person who completed questionnaire).
These covariates proved relevant in the multivariate model also for analyzing the impact of change of PM10 on respiratory symptoms. Age was included as a categoric variable (three groups according to school grades) because preliminary analysis suggested a nonlinear association between age and the evaluated health outcomes. In the first cross-section, all questionnaires were completed during wintertime to avoid confounding by season. The cross-sectional assessments of the second phase had to be spread over the whole school year for logistic reasons. A dummy variable for the month when the questionnaire was completed was included in the multivariate logistic regression models to adjust for possible reporting bias by season of the interview.
We evaluated whether secular trends had occurred between 1992–1993 and 1998–2001 that could be related to changing prevalence of the investigated symptoms and diseases—namely, climatic factors (milder or colder winters), participation rates, and mother’s concern about an association between environmental exposure and children’s respiratory health.
We further tested the final models for interactions between change of PM10 on the one hand and covariates such as age group, sex, family history of allergic diseases (asthma and/or atopy), asthma ever of child, smoker (child and/or mother), and indoor exposures (heating and/or cooking) on the other. The fit of the final models was evaluated.
To evaluate whether the average reduction of symptom prevalence is more pronounced in areas with stronger reduction of air pollution, we computed covariate-adjusted prevalence by community for the first (1992–1993) and second health assessment phase (1998–2001). To visualize the associations, we plotted the mean region-specific change in adjusted prevalence between the first and second phase against the respective mean change in PM10 levels. Corresponding Pearson correlation coefficients for the associations between these aggregate data were computed.
All analyses were conducted with Stata Statistical Software, Release 8.0.SE (StataCorp, College Station, TX, USA).
Results
PM10 levels, adjusted prevalence, and covariates 1992–2001.
Figure 1 shows PM10 levels at fixed monitoring sites in nine study regions of SCARPOL in 1993 and between 1997 and 2000. Across the nine study regions, the average decline of PM10 between 1993 and 2000 was 9.8 μg/m3 (29%). The average absolute decline in the urban and suburban areas Anières, Bern, Biel, Geneva, Lugano, and Zürich (12.7 μg/m3) was about three times as strong compared with the rural and alpine areas Langnau, Payerne, and Montana (4.0 μg/m3).
The adjusted prevalence of all investigated health end points declined between 1992–1993 and 1998–2001 (Table 3). Both the absolute and relative declines were stronger for the nonallergic outcomes chronic cough, bronchitis, common cold, nocturnal dry cough, and conjunctivitis symptoms (4.5–8.9% absolute decline of prevalence, on average, across the nine regions) compared with the allergy-associated end points sneezing during pollen season, asthma, and hay fever (0.4–1.7%). A tendency of a stronger absolute decline in suburban areas compared with rural/alpine areas was observed for the nonallergic, but not for the allergy-associated, outcomes.
Table 4 shows the distribution in the first (1992–1993) and second (1998–2001) health assessment phase of the covariates included in the multivariate models for analyzing the association between change of air pollution and change of prevalence. Excluded are children with missing data for one or more covariates. The most striking time trend is the increase in self-reported smoking of eighth graders from 6.4 to 16.3% (p < 0.0001). Mothers’ environmental concerns had declined on average from 78.9 to 75.6% (p = 0.001). The average annual number of cold days (days with the maximum temperature below zero degrees Celsius) had declined from 15 in 1992–1993 to 12 in 1998–2002 (p < 0.0001) across all study regions, with the strongest decline in Anières (from 10 to 3). An increase in the number of cold days was recorded in the alpine area Montana (from 21 to 38). Because the generally milder winters (with the exception of Montana) and the attenuated environmental concerns would be expected to move in the same direction as declining air pollution levels, that is, toward lower prevalence of reported symptoms and diseases, the logistic regression models were adjusted for the two secular trends. Participation rates in the four cross-sections (69.9, 82.4, 75.3, and 75.0%, respectively) indicated no secular trend.
Change in PM10 exposure versus change in prevalence.
Figure 2 shows that declining levels of PM10 were associated with declining prevalence of chronic cough, bronchitis, common cold, nocturnal dry cough, and conjunctivitis symptoms. For wheezing, sneezing, asthma, and hay fever, no significant association could be seen with declining PM10 levels. We found no effect modification by age group, sex, family history of allergic diseases, asthma of child, smoking, or indoor exposures. Random effect models did not change the effect estimates.
Mothers’ concerns regarding air pollution and children’s respiratory health were significant predictors for reported bronchitis, common cold, nocturnal dry cough, conjunctivitis symptoms, wheeze, and asthma, whereas the number of cold days was not significantly associated with reported symptoms and diseases (data not shown). Without adjustment for the temporal trends of mothers’ beliefs (on individual level) and number of cold days (on area level), the effect estimates were slightly stronger for chronic cough, common cold, nocturnal dry cough, and conjunctivitis symptoms and reached significance for wheeze (data not shown). Besides change in PM10, the covariates age, family history of bronchitis, child’s smoking, indoor humidity, and removal of carpets were the strongest significant predictors for chronic cough and bronchitis, while for asthma and hay fever, this applied to sex, age, family history of asthma and atopy, and removal of carpets and pets (data not shown). Crude estimates were quite similar to adjusted ORs (data not shown). The fit of the models was generally satisfactory according to Hosmer-Lemeshow chi-square (8 d.f.).
Figure 3 illustrates that, on an aggregate level, across regions the mean change in adjusted prevalence of nocturnal dry cough is associated with the mean change in PM10 levels (r
Pearson = 0.81, p = 0.008). The strongest decline of adjusted prevalence of nocturnal dry cough was observed in Geneva, Lugano, and Anières, where the strongest reduction of PM10 had also been achieved. Similar associations were observed for chronic cough (r = 0.78; p = 0.02) and conjunctivitis symptoms (r = 0.69; p = 0.04) (Figure 3), whereas for common cold (r = 0.48; p = 0.19) and bronchitis (r = 0.10; p = 0.80), the associations across regions were weaker and not significant.
Discussion
We showed that decreasing levels of PM10 were associated with declining prevalence rates of those respiratory symptoms and diseases associated with air pollution in the first cross-sectional analysis of SCARPOL (Braun-Fahrländer et al. 1997). The reduction in prevalence rates was larger in areas with a stronger decrease in PM10 levels. Decreasing environmental concerns of mothers (Swiss Society for Applied Social Research 2003) over time contributed to the observed decrease in respiratory symptoms and diseases but did not explain the association with air pollution. Adverse effects of PM10 have no apparent threshold, as we observed the beneficial effects for relatively small changes in rather moderate air pollution levels. We therefore conclude that even relatively small reductions in air pollution levels may improve children’s respiratory health.
Our findings are consistent with the improvement of nonallergic respiratory morbidity in children along with declining air pollution levels reported for East Germany (Heinrich et al. 2002; Kramer et al. 1999), although baseline levels and decline in Switzerland (SAEFL 2003) were much smaller. They are also in line with the few intervention studies that have investigated the impact of changing air pollution levels on children’s lung function growth (Avol et al. 2001; Frye et al. 2003; Neuberger et al. 2002) and bronchial responsiveness (Wong et al. 1998) and on mortality in adults (Clancy et al. 2002; Hedley et al. 2002). All these studies have found improved respiratory health or reduced respiratory and cardiovascular mortality after mitigation of ambient air pollution exposures. The consistency of these findings suggests that the observed associations between air pollution and respiratory health outcomes may be causal.
In our study, declining PM10 levels were not associated with changes in prevalence of asthma, hay fever, and sneezing during pollen season. No adverse effects of PM10 were observed for these allergy-associated health outcomes in cross-sectional analyses of SCARPOL (Braun-Fahrländer et al. 1997), and they have shown only a very small average decline in our study population and stable prevalence over the last decade in Swiss adolescents (Braun-Fahrländer et al. 2004). A similar contrast between nonallergic and allergy-associated health outcomes in children and declining air pollution levels has been reported by Kramer et al. (1999). Hirsch et al. (1999) reported significant associations of NO2, carbon monoxide, and benzene with bronchitis and morning cough but not with allergy-associated end points. A few studies using traffic counts or proximity to street as exposure proxy found positive associations with sensitization and allergy-related symptoms (Nicolai et al. 2003; van Vliet et al. 1997; Venn et al. 2001; Wjst et al. 1993). We cannot exclude such effects, but for our analysis we had no such data available.
Adjustment for the observed time trends of declining environmental concerns of mothers and reduced number of cold days over the study period did not markedly change the effect estimates. The monitoring of influenza epidemics by the Swiss Federal Office of Public Health (SFOPH) does not suggest a decrease in influenza between 1992 and 2001, which might have confounded our findings, but indicates random fluctuations between years (SFOPH 2001). The same is true for the number of hourly ozone concentrations exceeding 120 μg/m3 [Federal Commission for Air Hygiene (EKL) 2004]. For evaluation of the impact of other possible secular trends such as changes in health habits or medication use, we had no data available. Confounding of our cross-sectional findings by political or social time trends is very unlikely. In Switzerland, the system has been very stable throughout the study period (and was for many decades before), in contrast to the social changes that went hand in hand with air pollution reduction in East Germany (Frye et al. 2003; Heinrich 2003). Thus, uncontrolled confounding or secular trends are unlikely to explain our finings.
Our study is limited in that the comparison for each school grade is based on two points in time only, which are 6, 7, and 8 years apart for the first, eighth, and fourth graders, respectively. The difference in absolute change between the three age groups has been taken into account by design in the multivariate logistic regression models. However, we cannot evaluate whether the relevant time frame for the observed associations between air pollution reduction and improved respiratory health is long term (several years) or rather the year-to-year variability of air pollution levels, as recent Californian findings suggest (McConnell et al. 2003). For lifetime prevalence of asthma and hay fever, the relevance of the investigated change of exposure over a few years could be questioned, particularly for teenage children (eighth graders) who were exposed to higher air pollution levels in their early years of life, compared with first graders. Zmirou et al. (2004) report that exposure to traffic exhausts before the age of 3 years is associated with asthma in school children, but not lifelong exposures. In our data, no effect modification by age could be observed for asthma and hay fever, and their lifetime prevalence has been stable over the last decade in Swiss adolescents (Braun-Fahrländer et al. 2004).
We conclude that air pollution abatement measures implemented in Switzerland in the 1990s that resulted in moderately reduced air pollution exposures (SAEFL 2003; Kuebler et al. 2001) have successfully contributed to improved respiratory health in Swiss schoolchildren. Thus, not only dramatic changes (Heinrich 2003), but also modest improvements of ambient air pollution seem to be beneficial for children’s respiratory health. The larger reduction in symptom rates in areas with a stronger decrease in PM10 levels supports the causality of observed associations between air pollution and respiratory health in children. Our findings do not suggest a threshold for adverse effects of PM10, because we observed beneficial effects of rather small PM10 reductions in a moderately polluted environment. In urban regions and in the proximity of streets with high traffic volume, current PM10 levels still exceed limit values of the Swiss Clean Air Act (SAEFL 2003). Therefore, it can be assumed that there is still a potential for further improvement of both ambient air pollution and children’s health in Switzerland.
We thank the School Health Services for organizing the survey; the children, parents, and teachers for their cooperation; and J. Schwartz and N. Künzli for valuable comments.
This study received financial support from the Swiss National Science Foundation (grant 4026-033109), the Swiss Federal Office of Public Health, the Swiss Agency for the Environment, Forest and Landscape (FE/SAEFL 2000.I.08, FE/SAEFL 810.98.50), the Lung Associations Zürich and St. Gallen, and the cantonal health services of Zürich, St. Gallen, Valais, Vaud, Geneva, and Bern.
Figure 1 Annual means of PM10 levelsa assigned to children of the first (1993) and second (1997–2000) health assessment phase in nine SCARPOL regions.
aMeasured with DIGITEL HiVol Samplers. 1993 data converted from Harvard Impactor data.
Figure 2 Adjusted ORsa and 95% CIs of symptoms and respiratory diseases in SCARPOL associated with a decline of 10 μg/m3 PM10 levels.
aAdjusted for age, sex, nationality, parental education, number of siblings, farming status, low birth weight, breast-feeding, child who smokes, family history of asthma, bronchitis, and/or atopy, mother who smokes, indoor humidity, mode of heating and cooking, carpeting, pets allowed in bedroom, removal of carpet and/or pets for health reasons, person who completed questionnaire, month when questionnaire was completed, number of days with the maximum temperature < 0°C, belief of mother that there is an association between environmental exposures and children’s respiratory health, and region.
Figure 3 Mean change in adjusted prevalencea (1998–2001 to 1992–1993) versus mean change in regional annual averages of PM10 (1997–2000 to 1993) for nocturnal dry cough, chronic cough, and conjunctivitis symptoms across nine SCARPOL regions. Abbreviations: An, Anières; Be, Bern; Bi, Biel; Ge, Geneva; La, Langnau; Lu, Lugano; Mo, Montana; Pa, Payerne; Zh, Zürich.
aAdjusted for age, sex, nationality, parental education, number of siblings, farming status, low birth weight, breastfeeding, child who smokes, family history of asthma, bronchitis, and/or atopy, mother who smokes, indoor humidity, mode of heating and cooking, carpeting, pets allowed in bedroom, removal of carpet and/or pets for health reasons, person who completed questionnaire, month when questionnaire was completed, number of days with the maximum temperature < 0°C, and belief of mother that there is an association between environmental exposures and children’s respiratory health.
Table 1 Number of participating children by health assessment phase and school grade.
First phase Second phase
School grade [age (years)] 1992–1993a 1998–1999 1999–2000 2000–2001 Total 1992–2001
1st (6–7) 1,405 2,077 0 0 3,482
4th (9–11) 1,143 0 0 1,377 2,520
8th (13–14) 1,478 0 2,106 0 3,584
Total 4,026 2,077 2,106 1,377 9,591
a Surveys were conducted during a school year, which includes 2 calendar years.
Table 2 Definition of symptoms and diseases.
Symptom or disease Positive answer to the following question(s):
Chronic cough In the last 12 months, has your child had a cough associated with a respiratory infection lasting for more than 4 weeks?
Bronchitis In the last 12 months, has your child had bronchitis?
Common cold In the last 12 months, has your child had a common colda?
Nocturnal dry cough In the last 12 months, has your child had a dry cough at night, apart from a cough associated with a cold or a chest infection?
Conjunctivitis symptoms In the last 12 months, has your child had itchy or irritated eyes when he/she did not have a problem with the nose? (not caused by chlorinated water)
Wheeze In the last 12 months, has your child had wheezing or whistling in the chest?
Sneezing In the last 12 months, has your child had a problem with sneezing, or a runny or blocked nose when he/she did not have a cold or the flu and this occurred during pollen season (March–September)?
Asthma Has your child ever had asthma?
Hay fever Has your child ever had hay fever?
a In the German translation (grippe), this includes the flu.
Table 3 Adjusted prevalence of health outcomes and their change across all, urban/suburban,a and rural/alpineb regions.
Average of adjusted prevalence (%)c
Symptom or disease 1992–1993 (95% CI) 1998–2001 (95% CI) Absolute change Percent change
Chronic cough
All regions 12.0 (8.9–16.2) 7.9 (5.8–10.7) −4.1 34.4
Urban/suburban 13.9 (10.6–18.2) 9.2 (6.9–12.2) −4.7 34.1
Rural/alpine 8.2 (5.4–12.2) 5.3 (3.5–7.9) −2.9 35.5
Bronchitis
All regions 14.9 (11.2–19.6) 9.0 (6.7–12.0) −5.9 39.9
Urban/suburban 15.3 (11.9–19.5) 9.2 (7.1–12.0) −6.1 39.7
Rural/alpine 14.1 (9.9–19.9) 8.4 (5.8–12.1) −5.7 40.3
Common cold
All regions 35.0 (29.8–40.6) 26.1 (21.9–30.8) −8.9 25.4
Urban/suburban 35.7 (30.9–40.7) 26.7 (22.8–30.9) −9.0 25.2
Rural/alpine 33.7 (27.6–40.4) 25.0 (20.1–30.6) −8.7 25.9
Nocturnal dry cough
All regions 18.7 (14.6–23.7) 13.3 (10.3–17.0) −5.4 29.0
Urban/suburban 20.7 (16.6–25.4) 14.8 (11.7–18.4) −5.9 28.6
Rural/alpine 14.8 (10.7–20.2) 10.3 (7.5–14.3) −4.5 30.2
Conjunctivitis symptoms
All regions 19.7 (15.6–24.7) 15.2 (11.9–19.2) −4.5 23.0
Urban/suburban 21.1 (17.1–25.7) 16.3 (13.1–20.1) −4.8 22.7
Rural/alpine 17.1 (12.6–22.8) 13.0 (9.6–17.5) −4.0 23.7
Wheeze
All regions 8.2 (5.6–11.8) 6.1 (4.2–8.9) −2.0 25.0
Urban/suburban 8.5 (6.1–11.9) 6.4 (4.6–9.0) −2.1 24.9
Rural/alpine 7.4 (4.7–11.7) 5.5 (3.5–8.8) −1.9 25.1
Sneeze
All regions 8.9 (6.3–12.5) 7.2 (5.2–10.1) −1.7 18.8
Urban/suburban 8.7 (6.4–11.7) 7.1 (5.2–9.5) −1.6 18.8
Rural/alpine 9.3 (6.1–14.0) 7.6 (5.0–11.3) −1.7 18.8
Asthma
All regions 8.2 (5.7–11.8) 7.5 (5.2–10.6) −0.7 8.7
Urban/suburban 7.5 (5.4–10.4) 6.8 (4.9–9.5) −0.7 8.7
Rural/alpine 9.5 (6.2–14.4) 8.7 (5.8–13.0) −0.8 8.5
Hay fever
All regions 9.8 (7.1–13.5) 9.4 (6.9–12.7) −0.4 4.6
Urban/suburban 9.4 (7.1–12.5) 9.0 (6.8–11.8) −0.4 4.6
Rural/alpine 10.6 (7.2–15.5) 10.1 (7.0–14.6) −0.5 4.5
CI, confidence interval.
a Urban/suburban regions: Anières, Bern, Biel, Geneva, Lugano, Zürich.
b Rural/alpine regions: Langnau, Payerne, Montana.
c Adjusted for age, sex, nationality, parental education, number of siblings, farming status, low birth weight, breast-feeding, child who smokes, family history of asthma, bronchitis, and/or atopy, mother who smokes, humidity, mode of heating and cooking, carpeting, pets allowed in bedroom, removal of carpet and/or pets for health reasons, person who completed questionnaire, month when the questionnaire was completed, number of days with the maximum temperature < 0°C, and belief of mother that there is an association between environmental exposures and children’s respiratory health.
Table 4 Distribution of covariates in the first and second health assessment phase (all regions combined).
Characteristic 1992–1993 (n = 3,024) n (%) 1998–2001 (n = 4,428) n (%) p-Valuea
Sex (male) 1,550 (51.3) 2,191 (49.5) 0.139
Nationality
Swiss 2,288 (75.7) 3,214 (72.6) 0.003
Parental educationb
Low 446 (14.8) 500 (11.3) < 0.0001
Low-middle 436 (14.4) 458 (10.3)
Middle 949 (31.4) 1,294 (29.2)
Middle-high 516 (17.1) 852 (19.2)
High 677 (22.4) 1,324 (29.9)
No. of siblings
0 449 (14.9) 600 (13.6) < 0.0001
1 1,729 (57.2) 2,341 (52.9)
2 624 (20.6) 1,091 (24.6)
≥3 222 (7.3) 396 (8.9)
Farmingc 117 (3.9) 183 (4.1) 0.57
Low birth weight (< 2,500 g) 340 (11.2) 547 (12.4) 0.146
Family history of diseased 1,490 (49.3) 2,418 (54.6) < 0.0001
Breast-feeding (any) 2,436 (80.6) 3,829 (86.5) < 0.0001
Mother smokes 800 (26.5) 1,102 (24.9) 0.127
Child smokes (8th graders; n = 2,661) 67 (6.4) 263 (16.3) < 0.0001
Indoor humiditye 809 (26.8) 1,116 (25.2) 0.133
Central heating 243 (8.0) 520 (11.7) < 0.0001
Cooking mode
Electric 2,335 (77.2) 3,611 (81.6) < 0.0001
Wood 71 (2.4) 85 (1.9)
Gas 618 (20.4) 732 (16.5)
Floor type
Wood 545 (18.0) 1,798 (40.4) < 0.0001
Single carpet 460 (15.2) 772 (17.4)
Wall-to-wall carpet 2,019 (66.8) 1,867 (42.2)
Pets
No pets 1,451 (48.0) 2,031 (45.9) < 0.0001
Pets in house 731 (24.2) 1,163 (26.3)
Pets in bedroom 842 (27.8) 1,234 (27.9)
Removal of carpetf 85 (2.8) 251 (5.7) < 0.0001
Removal of petsf 68 (2.3) 96 (2.2) 0.816
Mother completed questionnaire 2,702 (89.4) 3,918 (88.5) 0.242
Environmental concerng 2,385 (78.9) 3,346 (75.6) 0.001
No. of cold daysh
gions 15 12 < 0.0001
Anières 10 3 < 0.0001
Bern 18 15 < 0.0001
Biel 13 8 < 0.0001
Geneva 10 5 < 0.0001
Langnau 20 19 0.02
Lugano 0 1 < 0.0001
Montana 21 38 < 0.0001
Payerne 22 16 < 0.0001
Zürich 21 17 < 0.0001
a Comparison of 1992–1993 and 1998–2001 using chi-square or t-tests as appropriate.
b Low: father and mother have no professional training; low-middle: father or mother has professional training of < 2 years; middle: father or mother has professional training of 2–4 years; middle-high: father or mother has academic training; high: father and mother have academic training.
c Family of child is full-time or part-time farming.
d Father and/or mother and/or siblings have asthma and/or atopy and/or chronic bronchitis.
e Mildew or water damage in the flat.
f Because of allergy or asthma of child.
g Mother believes that there is an association between environmental exposures and children’s respiratory health.
h Number of days with the maximum temperature < 0°C, assessed at the local fixed monitoring station.
==== Refs
References
Asher MI Keil U Anderson HR Beasley R Crane J Martinez F 1995 International Study of Asthma and Allergies in Childhood (ISAAC): rationale and methods Eur Respir J 8 3 483 491 7789502
Avol EL Gauderman WJ Tan SM London SJ Peters JM 2001 Respiratory effects of relocating to areas of differing air pollution levels Am J Respir Crit Care Med 164 11 2067 2072 11739136
Brauer M Hoek G Van Vliet P Meliefste K Fischer PH Wijga A 2002 Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children Am J Respir Crit Care Med 166 8 1092 1098 12379553
Braun-Fahrländer C Gassner M Grize L Takken-Sahli K Neu U Stricker T 2004 No further increase in asthma, hay fever and atopic sensitisation in adolescents living in Switzerland Eur Respir J 23 3 407 413 15065830
Braun-Fahrländer C Vuille JC Sennhauser FH Neu U Künzle T Grize L 1997 Respiratory health and long-term exposure to air pollutants in Swiss schoolchildren. SCARPOL Team. Swiss Study on Childhood Allergy and Respiratory Symptoms with Respect to Air Pollution, Climate and Pollen Am J Respir Crit Care Med 155 3 1042 1049 9116984
Chen PC Lai YM Wang JD Yang CY Hwang JS Kuo HW 1998 Adverse effect of air pollution on respiratory health of primary school children in Taiwan Environ Health Perspect 106 331 335 9618349
Clancy L Goodman P Sinclair H Dockery DW 2002 Effect of air-pollution control on death rates in Dublin, Ireland: an intervention study Lancet 360 9341 1210 1214 12401247
DIGITEL 1999. Operation Manual and User’s Handbook for Automated Particle Sampler, Type DIGITEL DHA-80 and DIGITEL HVS-HDI2 Operation Module [in German]. Hegnau, Switzerland:DIGITEL AG.
EKL (Federal Commission for Air Hygiene) 2004. Summer smog [in German]. Available: http://www.umwelt-schweiz.ch/buwal/de/fachgebiete/fg_luft/themen/sommersmog/ [accessed 28 September 2005].
Frye C Hoelscher B Cyrys J Wjst M Wichmann HE Heinrich J 2003 Association of lung function with declining ambient air pollution Environ Health Perspect 111 383 387 12611668
Gauderman WJ Gilliland GF Vora H Avol E Stram D McConnell R 2002 Association between air pollution and lung function growth in southern California children: results from a second cohort Am J Respir Crit Care Med 166 1 76 84 12091175
Gehring U Cyrys J Sedlmeir G Brunekreef B Bellander T Fischer P 2002 Traffic-related air pollution and respiratory health during the first 2 yrs of life Eur Respir J 19 4 690 698 11998999
Hedley AJ Wong CM Thach TQ Ma S Lam TH Anderson HR 2002 Cardiorespiratory and all-cause mortality after restrictions on sulphur content of fuel in Hong Kong: an intervention study Lancet 360 9346 1646 1652 12457788
Heinrich J 2003 Nonallergic respiratory morbidity improved along with a decline of traditional air pollution levels: a review Eur Respir J Suppl 40 64s 69s
Heinrich J Hoelscher B Frye C Meyer I Pitz M Cyrys J 2002 Improved air quality in reunified Germany and decreases in respiratory symptoms Epidemiology 13 4 394 401 12094093
Hirsch T Weiland SK von Mutius E Safeca AF Grafe H Csaplovics E 1999 Inner city air pollution and respiratory health and atopy in children Eur Respir J 14 3 669 677 10543291
Horak F Jr Studnicka M Gartner C Spengler JD Tauber E Urbanek R 2002 Particulate matter and lung function growth in children: a 3-yr follow-up study in Austrian schoolchildren Eur Respir J 19 5 838 845 12030722
Hruba F Fabianova E Koppova K Vandenberg JJ 2001 Childhood respiratory symptoms, hospital admissions, and long-term exposure to airborne particulate matter J Expo Anal Environ Epidemiol 11 1 33 40 11246799
Janssen NA Brunekreef B van Vliet P Aarts F Meliefste K Harssema H 2003 The relationship between air pollution from heavy traffic and allergic sensitization, bronchial hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren Environ Health Perspect 111 1512 1518 12948892
Kramer U Behrendt H Dolgner R Ranft U Ring J Willer H 1999 Airway diseases and allergies in East and West German children during the first 5 years after reunification: time trends and the impact of sulphur dioxide and total suspended particles Int J Epidemiol 28 5 865 873 10597984
Krütli P Monn C 1999. PM10-Comparison Measurements between Harvard Low-Flow-Rate-Sharp-Cut-Impactor and DIGITEL DHA-80. Institute for Hygiene and Applied Physiology, ETHZ. Report No UH-99-10. Zurich:Swiss Federal Institute of Technology Zurich.
Kuebler J van den Bergh H Russell AG 2001 Long-term trends of primary and secondary pollutant concentrations in Switzerland and their response to emission controls and economic changes Atmos Environ 35 1351 1363
McConnell R Berhane K Gilliland F London SJ Vora H Avol E 1999 Air pollution and bronchitic symptoms in Southern California children with asthma Environ Health Perspect 107 757 760 10464077
McConnell R Berhane K Gilliland F Molitor J Thomas D Lurmann F 2003 Prospective study of air pollution and bronchitic symptoms in children with asthma Am J Respir Crit Care Med 168 7 790 797 12893648
Neuberger M Moshammer H Kundi M 2002 Declining ambient air pollution and lung function improvement in Austrian children Atmos Environ 36 1733 1736
Nicolai T Carr D Weiland SK Duhme H von Ehrenstein O Wagner C 2003 Urban traffic and pollutant exposure related to respiratory outcomes and atopy in a large sample of children Eur Respir J 21 6 956 963 12797488
Röösli M Braun-Fahrländer C Künzli N Oglesby L Theis G Camenzind M 2000 Spatial variability of different fractions of particulate matter within an urban environment and between urban and rural sites J Air Waste Manag Assoc 50 7 1115 1124 10939205
Röösli M Theis G Künzli N Staehelin J Mathys P Oglesby L 2001 Temporal and spatial variation of the chemical composition of PM10 at urban and rural sites in the Basel area, Switzerland Atmos Environ 35 21 3701 3713
SAEFL 2003. NABEL—Air Pollution Levels 2002 [in German]. Schriftenreihe Umwelt. No. 360. Bern:Swiss Agency for the Environment, Forest and Landscape.
SFOPH (Swiss Federal Office of Public Health) 2001. Infectious Disease, Switzerland, 2001: Virus Influenza. Available: http://www.bag.admin.ch/infreporting/gs01/p21.pdf [accessed 31 May 2005].
Swiss Clean Air Act 1985. Luftreinhalte-Verordnung vom 16. Dezember 1985 (LRV) [in German]. SR-Number 814.318.142.1.
Swiss Society for Applied Social Research 2003. The worry barometer 2003 of Credit Suisse [in German]. Bern:GfS-Research Institute. Available: http://emagazine.credit-suisse.com/_product_documents/_articles/41997/Sorgenbarometer-2003-final.pdf [accessed 31 May 2005].
van Vliet P Knape M de Hartog J Janssen N Harssema H Brunekreef B 1997 Motor vehicle exhaust and chronic respiratory symptoms in children living near freeways Environ Res 74 2 122 132 9339225
Venn AJ Lewis SA Cooper M Hubbard R Britton J 2001 Living near a main road and the risk of wheezing illness in children Am J Respir Crit Care Med 164 12 2177 2180 11751183
Wjst M Reitmeir P Dold S Wulff A Nicolai T von Loeffelholz-Colberg EF 1993 Road traffic and adverse effects on respiratory health in children BMJ 307 6904 596 600 7691304
Wong CM Lam TH Peters J Hedley AJ Ong SG Tam AY 1998 Comparison between two districts of the effects of an air pollution intervention on bronchial responsiveness in primary school children in Hong Kong J Epidemiol Community Health 52 9 571 578 10320858
Zmirou D Gauvin S Pin I Momas I Sahraoui F Just J 2004 Traffic related air pollution and incidence of childhood asthma: results of the Vesta case-control study J Epidemiol Community Health 58 1 18 23 14684722
|
16263523
|
PMC1310930
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 21; 113(11):1632-1637
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8159
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8111ehp0113-00163816263524ResearchChildren's HealthBirth Outcomes and Prenatal Exposure to Ozone, Carbon Monoxide, and Particulate Matter: Results from the Children’s Health Study Salam Muhammad T. 1Millstein Joshua 1Li Yu-Fen 1Lurmann Frederick W. 2Margolis Helene G. 3Gilliland Frank D. 11 Department of Preventive Medicine, University of Southern California, Keck School of Medicine, Los Angeles, California, USA2 Sonoma Technology Inc., Petaluma, California, USA3 Air Resources Board, State of California, Sacramento, California, USAAddress correspondence to F.D. Gilliland, Department of Preventive Medicine, USC Keck School of Medicine, 1540 Alcazar St., CHP 236, Los Angeles, CA 90033 USA. Telephone: (323) 442-1096. Fax: (323) 442-3272. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 18 7 2005 113 11 1638 1644 9 3 2005 18 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. Exposures to ambient air pollutants have been associated with adverse birth outcomes. We investigated the effects of air pollutants on birth weight mediated by reduced fetal growth among term infants who were born in California during 1975–1987 and who participated in the Children’s Health Study. Birth certificates provided maternal reproductive history and residence location at birth. Sociodemographic factors and maternal smoking during pregnancy were collected by questionnaire. Monthly average air pollutant levels were interpolated from monitors to the ZIP code of maternal residence at childbirth. Results from linear mixed-effects regression models showed that a 12-ppb increase in 24-hr ozone averaged over the entire pregnancy was associated with 47.2 g lower birth weight [95% confidence interval (CI), 27.4–67.0 g], and this association was most robust for exposures during the second and third trimesters. A 1.4-ppm difference in first-trimester carbon monoxide exposure was associated with 21.7 g lower birth weight (95% CI, 1.1–42.3 g) and 20% increased risk of intrauterine growth retardation (95% CI, 1.0–1.4). First-trimester CO and third-trimester O3 exposures were associated with 20% increased risk of intrauterine growth retardation. A 20-μg/m3 difference in levels of particulate matter ≤ 10 μm in aerodynamic diameter (PM10) during the third trimester was associated with a 21.7-g lower birth weight (95% CI, 1.1–42.2 g), but this association was reduced and not significant after adjusting for O3. In summary, O3 exposure during the second and third trimesters and CO exposure during the first trimester were associated with reduced birth weight.
air pollutionbirth weightcarbon monoxideintrauterine growth retardationmaternal exposurenitrogen dioxideozoneparticulate matter
==== Body
A growing body of evidence indicates that maternal exposures to air pollutants, including ozone, carbon monoxide, particulate matter with an aerodynamic diameter of ≤ 10 μm or ≤ 2.5 μm (PM10 and PM2.5, respectively), nitrogen dioxide, and sulfur dioxide, are associated with adverse pregnancy outcomes. A major research focus has been to investigate the effects of air pollution on birth weight, low birth weight (LBW; < 2,500 g), and intrauterine growth retardation (IUGR).
Williams et al. (1977) first reported an inverse association between air pollution (based on the ambient levels of CO, NO2, and O3) and birth weight in Los Angeles, California, in the 1970s. Subsequently, several studies examining the effects of air pollution on fetal growth have been conducted in the United States (Chen et al. 2002; Maisonet et al. 2001; Parker et al. 2005; Ritz and Yu 1999), Canada (Liu et al. 2003), the United Kingdom (Bobak et al. 2001), the Czech Republic (Bobak 2000; Bobak and Leon 1999), China (Wang et al. 1997), South Korea (Ha et al. 2001; Lee et al. 2003), Taiwan (Lin et al. 2004; Yang et al. 2003), and Brazil (Gouveia et al. 2004).
Data from studies that were conducted in term infants showed significant associations between CO, PM10, and PM2.5 and reduced birth weight (Chen et al. 2002; Gouveia et al. 2004; Ha et al. 2001; Lee et al. 2003; Maisonet et al. 2001; Parker et al. 2005; Ritz and Yu 1999; Yang et al. 2003) and IUGR (Dejmek et al. 1999; Ha et al. 2001; Maisonet et al. 2001; Parker et al. 2005; Ritz and Yu 1999). However, the trimester-specific findings from these studies are inconsistent because they showed different windows of increased risk of reduced birth weight and IUGR for each pollutant. The explanations for the variations in results remain unclear. Although studies suggest that O3 and NO2 may be associated with birth weight, these pollutants have received less attention in epidemiologic studies.
To further investigate the effects of O3, CO, PM10, and NO2 on reduced birth weight mediated by IUGR, we conducted a study among children who were born in southern California, a region with widely varying air pollution levels. We examined the associations between birth weight, LBW, and IUGR in term newborns and air pollution data in participants of the Children’s Health Study (CHS), a population-based study of children residing in 12 southern California communities.
Materials and Methods
Study population.
The elements of the CHS have been described previously (Peters et al. 1999a, 1999b). The population for this study was a subset of 6,259 participants in the CHS who were born in California between 1975 and 1987 and were recruited in four cohorts from public school classrooms from grades 4, 7, and 10 in 12 southern California communities in 1993–1996. At CHS entry, parents or guardians of each participating student provided written informed consent and completed a self-administered questionnaire, which included detailed demographic information as well as information regarding the mother’s and child’s respiratory health and exposure to different environmental factors that might contribute to the risk of reduced birth weight or could influence air pollution exposures. Race/ethnicity of the child was grouped as non-Hispanic white, Hispanic white, African American or black, Asian and Pacific Islander, and “other.” Maternal smoking during pregnancy was assigned using self-report of any smoking during pregnancy. Socioeconomic status (SES) was determined from questionnaire items related to parent’s educational level and annual family income at study entry (Table 1).
Birth information.
Birth weight, gestational age, and other reproductive data were obtained from California birth certificates. Birth certificate information for California-born children who participated in the CHS was obtained by computerized linkage of participants with the California Department of Health Services Birth Statistical Master Files and Birth Cohort Files. Of the 6,259 cohort participants, 5,013 were born in California according to parental report, and 4,842 were matched to a birth certificate record. Because preterm birth is associated with adverse birth outcomes and our analyses were focused on the effects of air pollutants on birth weight mediated by reduced fetal growth as opposed to reduced gestational age, we restricted our analyses in children who were born at term with gestational ages between 37 and 44 weeks, consistent with the methods used by earlier researchers. This criterion resulted in the exclusion of 941 children who were born before the 37th week of gestation, and our final sample consisted of 3,901 children. Data obtained from California birth certificates included birth weight, gestational age, maternal age at birth, maternal residence ZIP code at birth, parity, months since last live birth, gestational diabetes, and marital status. The estimated date of conception was assigned using the birth date and gestational age, corrected for the average 2-week difference between the last menstrual period and conception. First trimester was defined as gestational age < 13 weeks, second trimester was 13–27 weeks, and third trimester was 28 weeks to birth. We defined LBW as birth weight < 2,500 g in term infants and defined IUGR by less than the 15th percentile of predicted birth weight based on gestational age and sex in term infants.
Air pollution exposure assessment.
Air pollution estimates were assigned using the ZIP code of the maternal residence at birth and monthly average air pollution estimates for each ZIP code interpolated from monitoring data obtained from the U.S. Environmental Protection Agency’s (EPA) Aerometric Information Retrieval System through written request in 1999. Currently the 1999 data are available under the Air Quality System database of the U.S. EPA (U.S. EPA 2005). Temperature data were acquired from National Weather Service surface observations, the South Coast Air Quality Management District, the Bay Area Air Quality Management District, and CHS air pollution monitoring stations. Elevations above sea level were derived from 1-km horizontal resolution geophysical data from the U.S. Geological Survey’s Earth Resources Observation Systems Data Center (U.S. Geological Survey 2005). Elevations were assigned to the ZIP code centroid of the maternal birth residence. Exposure estimates were calculated for monthly average of 24-hr O3 (O3[24hr]), 1000 hr to 1800 hr O3 (O3[10–6]), NO2, CO, PM10, and temperature by spatially interpolated monthly average levels to the geometric centroid coordinates of the ZIP code boundaries for the ZIP code region of the maternal birth residences.
The interpolation used inverse-distance-squared weighting based on data from up to the three nearest monitoring stations located within 50 km of the ZIP code centroid (100 km for O3). Because the spatial gradients in monthly average data are smaller than hourly or daily data, the method was relative insensitive to the number of stations included in the interpolation and maximum interpolation radius. Three stations and a 50-km radius were selected because they preserved local gradients slightly better than did larger numbers of stations and greater distances. In cases where air monitoring data were available from a station located within 5 km of a ZIP code centroid, the assignments were based on the nearest station’s data rather than interpolation. For this cohort, 29% of the NO2 assignments and approximately 40% of the O3, PM10, and CO assignments were based on the nearest station’s data. Each mother’s exposures to O3, NO2, PM10, and CO during each trimester were estimated using weighted averages of the calendar month averages. Although other studies included SO2 exposures, it was not included in this study because ambient SO2 concentrations are generally low in California and are measured at too few monitoring stations to make reliable exposure assignments.
Statistical analyses.
The effect of each air pollutant on birth weight and 95% confidence intervals (CIs) were computed by fitting linear mixed-effects models using the SAS procedure GENMOD (SAS version 8.2; SAS Institute Inc., Cary, NC). Because of the nonlinearity in the relationship of birth weight with gestational age, a cubic polynomial in gestational age was necessary to adequately capture the nonlinearity. We therefore included linear, quadratic, and cubic terms for gestational age in all final models. The models also included maternal age, months since last live birth, parity, maternal smoking during pregnancy, SES, marital status at childbirth, gestational diabetes, and child’s sex, race/ethnicity, and school grades (4th, 7th, and 10th) of the CHS cohort as fixed effects, and CHS study community as a random effect. To account for confounding by seasonal determinants of birth weight, the adjusted analyses included six terms for the basis matrix of a b-spline on Julian day of birth, with knots at days 91, 183, and 274 (De Boor 2001), because this has been suggested to be a more rigorous method of adjusting for temporal variations than adjusting for month or season of birth. We used missing indicators for missing data on maternal smoking during pregnancy, parity, SES, and race/ethnicity.
We considered the effects of pollutant exposures during the entire pregnancy and during each trimester. To illustrate the magnitude of the birth weight differences by air pollutant level, we presented the estimates scaled to the approximate interquartile ranges (IQRs) of average exposure for each pollutant. We also computed odds ratios (ORs) and 95% CIs for the association between pollutant levels and LBW and IUGR using appropriate logistic regression models that included the same adjustment variables as in the mixed-effects models. For pollutants that showed significant associations with birth weight in trimester-specific analysis, we fitted models that included average levels of those specific pollutants during specific trimesters. We fitted an additional model for O3 that included three terms for average levels during each of the three trimesters.
For pollutants that were correlated with temperature and altitude, we conducted sensitivity analyses adding linear terms for elevation above sea level, temperature, and season of birth to the multivariate models. We also fitted a model with indicator variables for the deciles of O3[24hr] exposure averaged over the entire pregnancy and presented the difference in birth weight associated with each decile of O3[24hr] exposure relative to the lowest decile. We assessed departure from linear relationships between birth weight and pollutants by examining appropriate residual plots and by comparing nested models with linear terms and categorical terms for each pollutant using partial F-tests. All significance tests were two sided at the 0.05 level.
Results
The sample included almost equal proportions of males and females (Table 1). About 60% of subjects were non-Hispanic, and 28% were Hispanic whites. Average birth weight was 3,487 g, with 72 infants (1.8%) with LBW and 585 infants (15%) with IUGR. Approximately 18% of mothers smoked during pregnancy. Few mothers had gestational diabetes (n = 18) or gave twin births (n = 25).
Table 2 shows the means ± SDs of the air pollutants, temperature, elevation, and their correlations with one another. For the trimester-specific averages, we observed strong positive correlation between O3[24hr] and O3[10–6] with correlation coefficient (r) > 0.9, and moderate positive correlation between O3 and temperature with r approximately 0.6. Daily O3[24hr] was also positively correlated with PM10 (r ~ 0.5) but negatively correlated with NO2 (r ~ −0.1) and CO (r ~ −0.3). Elevation was positively correlated with O3 (r ~ 0.3 −0.4) but negatively with CO (r ~ −0.3) and NO2 (r ~ −0.15). O3 levels in the first and second trimesters and the second and third trimesters were positively correlated (r = 0.31 in both instances), but levels in the first and third trimesters were negatively correlated (r = −0.25; data not shown).
Maternal exposure to O3 averaged over the entire pregnancy was associated with reduced birth weight for both the O3[24hr] and the O3[10–6] average metrics (Table 3). For O3[24hr] levels during the entire pregnancy, mean birth weight was lower by 47.2 g (95% CI, 27.4–67.0 g) across the IQR of 12 ppb. Exposures to PM10, NO2, and CO when averaged over the entire pregnancy were not significantly associated with birth weight.
The inverse association between ambient O3 levels and birth weight was stronger for exposure occurring during the second and third trimester for both the O3[24hr] and O3[10–6] average metrics (Table 3). With each IQR increase (i.e., 16 ppb and 17 ppb in the second and the third trimesters, respectively) in average O3[24hr] during the second and the third trimesters, birth weights were lower by 32.3 g (95% CI, 13.7–50.9 g) and 35.3 g (95% CI, 15.9–54.7 g), respectively. Trimester-specific analyses showed a statistically significant inverse association between CO concentrations in the first trimester and birth weight. Over an IQR in CO of 1.4 ppm, mean birth weight was lower by 21.7 g (95% CI, 1.1–42.3 g). The effects of PM10 were largest for the third trimester, and with each IQR increase in PM10 (i.e., 20 μg/m3), birth weight was lower by 21.7 g (95% CI, 1.1–42.2 g).
In sensitivity analyses where further adjustments were made for temperature, elevation, and season of birth, the effect estimates were somewhat reduced for the average O3 exposure over the entire pregnancy [Table S1 of the supplemental material (http://ehp.niehs.nih.gov/members/2005/8111/supplemental.pdf)]. However, greater birth weight deficits were observed for the second-trimester O3 exposures after such adjustment compared with the estimates presented in Table 3. Greater influence of these adjustment variables was observed on the CO effects on birth weight. The direction of association between CO exposures over the entire pregnancy and during the second and third trimesters changed from a positive to a negative one, and the negative association between first-trimester CO exposure and birth weight was attenuated. For PM10, the effect estimate for the third trimester was reduced and was no longer statistically significant. However, the second-trimester PM10 results became stronger and achieved borderline significance.
In two-pollutant models that examined the joint effects of air pollutants that showed significant associations in trimester-specific analyses, CO effects were stronger in a two-pollutant model that included O3[24hr] and CO, and birth weight was lower by 28.6 g (95% CI, 6.9–50.4 g) per 1.4-ppm increase in ambient CO levels (Table 4). In contrast, in a two-pollutant model that included O3[24hr] and PM10, the inverse association between PM10 and birth weight during the third trimester was attenuated and no longer was statistically significant. Including the three trimester-specific averages for O3 in one model attenuated the effect estimates for the second trimester but did not change the estimates for the third trimester.
Maternal O3 exposures averaged over the entire pregnancy and during the third trimester and CO exposure during the first trimester were significantly associated with increased risk of IUGR (Table 5). A 17-ppb difference in maternal O3[24hr] exposure during the third trimester increased the risk of IUGR by 20% (95% CI, 1.0–1.3). Over an IQR in CO of 1.4 ppm during the first trimester, the risk of IUGR increased by 20% (95% CI, 1.0–1.4). A 17-ppb difference in the third-trimester O3 exposures was associated with 40% increased risk of LBW (95% CI, 1.0–1.9); however, the estimates had substantial uncertainty due to the small number of LBW newborns.
We also observed a dose–response relationship of birth weight with average O3[24hr] that was clearest above 30-ppb exposure levels (Figure 1). Relative to the lowest decile of average O3[24hr] estimates for the next five lowest deciles were approximately −40 g to −50 g, with no clear trend and with 95% confidence bounds that included zero. The highest four deciles of O3 exposure showed an approximately linear decrease in birth weight, and all four 95% CIs excluded zero [estimates of birth weight deficits (grams) for the four uppermost deciles of exposure, in ranked order by decile median of exposure: −73.7 (95% CI, −139.0 to −7.4); −91.6 (95% CI, −157.9 to −25.3); −103.5 (95% CI, −170.3 to −36.7); −148.3 (95% CI, −214.7 to −81.9)]. There were no consistent dose–response relationships between birth weight and CO and PM10 from exposure during pregnancy or by trimester (data not shown).
The crude effect estimates without any adjustment for maternal (i.e., maternal age, race, smoking status) or fetal (i.e., gestational age, sex) factors were similar to the adjusted estimates. The directions and magnitude of the effect estimates did not change in sensitivity analyses after we excluded the plural births, infants born to mothers with gestational diabetes, and subjects with missing covariate information on maternal race/ethnicity, smoking during pregnancy, parity, and SES.
Discussion
We observed that ambient CO levels in the first trimester and O3 levels in the second and the third trimesters were independently associated with lower birth weight and IUGR in term infants through the mechanism of reduced fetal growth. Although PM10 exposure in the third trimester was associated with reduced birth weight, the PM10 effect was not statistically significant in a two-pollutant model that included PM10 and O3. A clear pattern of increasing deficits in birth weight with increasing levels of O3 was observed for 24-hr average O3 levels above 30 ppb, but no such trend was apparent for CO and PM10. NO2 levels were not associated with birth weight in this study. Although the differences in birth weight were small on average, those in the highest O3 exposure group had deficits of a magnitude equivalent (~ 150 g) to those observed after exposure to cigarette smoke.
Our finding of an inverse association between maternal CO exposure during the first trimester and birth weight and IUGR is consistent with earlier reports (Gouveia et al. 2004; Ha et al. 2001; Lee et al. 2003). Although others have observed a significant negative effect of first-trimester CO exposure on birth weight (Maisonet et al. 2001; Ritz and Yu 1999), in a study of term infants born in California in 2000, Parker et al. (2005) did not observe a significant association between CO and birth weight. This lack of a strong association between CO and birth weight in a recent California birth cohort could be due to > 3-fold reduction in average third-trimester CO exposure levels from 1989–1993 to 2000 (2.5 and 0.8 ppm, respectively) (Parker et al. 2005; Ritz and Yu 1999).
Several lines of evidence support the plausibility of a negative effect of CO exposure on birth weight. CO reduces oxygen-carrying capacity of maternal hemoglobin, which could adversely affect O2 delivery to fetal circulation. Because CO crosses the placental barrier (Sangalli et al. 2003) and fetal hemoglobin has greater affinity for binding CO than does adult hemoglobin, O2 delivery to fetal tissues is further compromised (Di Cera et al. 1989). The resultant tissue hypoxia has the potential to reduce fetal growth.
The robust inverse association of O3 concentrations with reduced birth weight is consistent with the early study of Williams et al. (1977) conducted in Los Angeles, California, in the 1970s. The present study found smaller deficits in birth weight across the range of O3 concentrations than those observed in the study conducted by Williams et al., likely because they included children who were born preterm, and the levels of O3 and other pollutants in the 1970s were much higher than levels during our study period. The 150-g deficit we observed in the highest decile of exposure in our study is of the same order of magnitude as the 314-g weight reduction in babies from areas with high O3 levels compared with those living in the least-polluted areas, as was observed by Williams et al. (1977).
Of the more recent published reports that assessed the role of different ambient air pollutants on birth weight, a limited number assessed the effects of O3 levels. Data from these studies suggested an inverse association between current ambient O3 levels and birth weight (Chen et al. 2002; Gouveia et al. 2004; Ha et al. 2001). Chen et al. (2002) observed a borderline significant inverse association between ambient O3 levels and birth weight in term infants born in Washoe County, Nevada, between 1991 and 1999. Mean 8-hr O3 levels in Washoe County were substantially lower compared with the levels observed in this study (27.8 vs. 50.6 ppb). In Seoul, South Korea, with a median O3 level of 22.4 ppb, Ha et al. (2001) also observed a significant association between levels of O3 in the third trimester and LBW. Our results are consistent with the findings from these studies. In the study of term infants in Los Angeles, conducted by Ritz and Yu (1999), CO effects were the focus of the research, and associations of O3 with birth weight were not reported. In addition, in this study, we observed significant reductions in birth weight with O3[24hr] concentrations ≥ 30 ppb. This may provide an explanation for not observing any association between O3 and birth weight in areas with lower ambient O3 levels. However, interpreting this observation as evidence for a threshold is not justified without further investigation.
In addition to epidemiologic reports, experimental studies in animal models also support a role of O3 in reduced birth weight and suggest that the effect could be mediated through modulation of maternal inflammatory processes. In rats, neutrophilic inflammation was proportional to O3 dose, and pregnant rats were more susceptible to acute pulmonary inflammation from O3 than were virgin rats (Gunnison and Hatch 1999). The increased susceptibility during pregnancy arises from higher O3 doses to the respiratory epithelial lining fluid (RELF) due to higher alveolar ventilation produced by increased tidal volumes and decreased ascorbic acid levels in the RELF. Kavlock et al. (1979) reported that mid and late gestational exposure to O3 increased embryo toxicity in rats with evidence for a reduction in weight gain 6 days after birth. Bignami et al. (1994) observed reduced postnatal weight gain in mice whose mothers were exposed to 1.2 ppm O3 over days 7–17 of gestation. Because pregnancy duration in mice is about 3 weeks (Thibodeaux et al. 2003), 7–17 days of gestation would correspond to the second and the third trimesters in mice. In a second study by this group, mice exposed prenatally to 0.6 ppm O3 showed a long-lasting reduction in body weight (Dell’Omo et al. 1995).
Similar experimental studies of exposures cannot be conducted in pregnant women; however, in healthy human volunteers, effects of O3 exposures on biomarkers have been observed, including increased peripheral neutrophil and lower ascorbic acid levels 6 hr after 0.2 ppm O3 exposure for 2 hr (Mudway et al. 1999). Because pregnant women have higher alveolar ventilation than do nonpregnant women (Barclay 1997), the level of exposures during pregnancy is likely to be greater, and similar inflammatory responses could be more pronounced in pregnant women. Furthermore, inflammation due to O3 results in the release of lipid peroxidation products and inflammatory cytokines into circulation (Hemmingsen et al. 1999; Larini and Bocci 2005). This could adversely affect placental circulation and function and can affect fetal growth. Further research is needed to define the mechanisms of O3 on the maternal–fetal unit.
Our estimates of about a 22-g decrease in birth weight per 20-μg/m3 increase in PM10 during the third trimester was comparable with the 11-g decrease in birth weight for a 10-μg/m3 increase in PM10 levels, as observed by Chen et al. (2002). However, after adjustment for O3, the estimate was reduced by about 50% and was not statistically significant. Further investigation of the relationship of PM10 and birth outcomes is needed, especially in the context of exposures to ambient levels of O3 and PM10.
Although a South Korean research group observed a significant negative effect of first-trimester NO2 exposure on birth weight in two studies (Ha et al. 2001; Lee et al. 2003) where the study population of the first study was part of the second study, other researchers did not observe any significant association between NO2 and birth weight (Gouveia et al. 2004; Lin et al. 2004; Liu et al. 2003). Moreover, there was a high correlation between NO2, CO, and total suspended particles or PM10 in the studies conducted in South Korea, and it is not clear whether NO2 had an independent negative effect in these studies.
Although third-trimester O3 exposure was positively associated with LBW, we did not observe any significant association between any of the measured pollutants and LBW in our sample for exposures averaged over the entire pregnancy or for exposures during the first and second trimesters. One reason for observing significant associations between air pollutants and birth weight on a continuous scale but not on a categorical scale could be due to the low prevalence (i.e., < 2%) of LBW. Because a set cut-point of 2,500 g is used to define LBW, and the prevalence of LBW in term infants in developed countries is low, the use of LBW may not be ideal for assessing the impact of air pollution on birth weight in developed countries.
In contrast with earlier studies, the present study had information on maternal smoking and SES to adjust for potential confounding by these factors. Because maternal smoking during pregnancy may affect the associations between air pollutants and birth weight, and we did not have enough power to detect any significant effect modification, we restricted our analysis to mothers who did not smoke during their pregnancy with the index child and observed similar results (data not shown).
The findings from this study must be interpreted in light of the methods used to assign exposure during the period of pregnancy. Air pollution exposures were assigned based on the centroid of the ZIP code of maternal residence at childbirth and may not reflect the actual exposure levels for the duration of the entire pregnancy. Studies have shown that about 20–25% of pregnant women change their residences during pregnancy (Khoury et al. 1988; Shaw and Malcoe 1992). However, pregnant women most often move to a different house in the same locality (Fell et al. 2004), probably because of the proximity to their workplaces, health care providers, and schools for those with older children. Residential address from birth certificates most often reflects maternal location during the last trimester (Schulman et al. 1993). This may have minimized the measurement error in exposure assessment during the third trimester, and as such, the third-trimester–specific results are likely to have less error than do the other trimester-specific results. We also lacked information on maternal employment location and commuting patterns. Because the resultant misclassification in exposure assignment is not likely to be differential, it may have attenuated the risk estimates. Exposure misclassification may explain the null findings for NO2, but this would not account for the deficits in birth weight that we observed from exposures to the other pollutants. Because we did not have direct measures of maternal exposures to indoor and occupational pollutants during pregnancy, we could not address the effects of these exposures on the birth outcomes.
Another source for error in the air pollution estimates may be the distance between maternal residence and the air monitoring stations used to assign exposure. The proportion of data on O3, PM10, NO2, and CO interpolated from monitoring stations < 5 km from the maternal birth residence ranged from 29.2 to 41.1%, whereas 51.2–61.3% were between 5 and 25 km, and the remaining 2.8–9.6% were interpolated from stations between 25 and 50 km for PM10, NO2, and CO and between 25 and 100 km for O3 [Table S2 of the supplemental material (http://ehp.niehs.nih.gov/members/2005/8111/supplemental.pdf)]. We conducted sensitivity analyses to assess the influence of distance of the monitoring stations on the associations between air pollutants and birth outcomes excluding women living farther from these monitors, and the results showed little change (data not shown). Furthermore, a recent study in California showed high correlation (r ~ 0.9) of air pollutant levels between monitors located within 5 miles (i.e., ~ 8 km) from the maternal residence and those located at the county level (Basu et al. 2004), suggesting that the ZIP-code–based air pollution data could be a good proxy for exposure levels at the residences and in the region surrounding residences where subjects spend most of their time.
Although measurement errors of pollutants in single-pollutant models are likely to attenuate the magnitude of birth weight deficits associated with air pollutants, the findings from two-pollutant models as well as those adjusted for temperature and elevation should be evaluated with some caution given the correlation structure between these environmental factors. In two-pollutant models where levels of pollutants are correlated and the measurement error in one pollutant is larger than the other, Zeger et al. (2000) showed that the effects of the relatively poorly measured pollutant are decreased and could introduce positive bias in the effect of the better measured pollutant in the presence of strong negative correlation between the measurement errors of the two pollutants. We would expect attenuation in the CO effect estimate in the first trimester in a two-pollutant model that included O3, because the measurement error in CO was possibly larger than O3 and the pollutants were negatively correlated (r ~ 0.3). However, the CO effect estimates remained robust in a two-pollutant model, suggesting an independent CO effect on birth weight in the first trimester. PM10 and O3, on the other hand, were positively correlated in the third trimester (r ~ 0.5), and the effect of PM10 was reduced about 2-fold in the two-pollutant model. Because a strong negative correlation between the measurement errors in PM10 and O3 seems unlikely, the independent PM10 effects may be smaller.
Measures of maternal nutrition (i.e., prepregnancy weight-for-height, gestational weight gain, and intake of nutrients) were not available to assess the potential effects of these factors on the associations observed in this study. Although maternal nutrition may be a determinant of birth weight in the developing world, Kramer (1987) did not observe any significant role of maternal nutrition in LBW in developed countries. Also, earlier literature has shown that maternal race/ethnicity (Cohen et al. 2001), maternal education (Kramer et al. 2000), and maternal age (Haiek and Lederman 1989) are associated with maternal nutrition. Because we adjusted for race, maternal education, and maternal age and found little evidence for confounding in our analyses, we may have indirectly adjusted for the nutritional differences during pregnancy. Therefore, maternal nutrition is unlikely to confound the associations of air pollutants with birth weight in this study.
We also considered the possibility of bias from the potential effects of air pollutants on preterm births. In California, birth certificate data for birth weight have been found to be valid and reliable (Von Behren and Reynolds 2003). However, the data on gestational age based on maternal recall of last menstrual period may be less accurate. We may have included some preterm births in our analysis, but because our outcome of interest was reduction in mean birth weight, our results are not likely to be biased by the effects of air pollutants on gestational age at birth.
In conclusion, we observed an association between lower birth weight and IUGR with O3 concentrations. The second- and third-trimester O3 levels were most strongly associated with deficits in birth weight. In addition, CO levels in the first trimester were associated with about a 22-g reduction in birth weight over an IQR of 1.4 ppm. These findings suggest that ambient CO in the first trimester and O3 in the second and third trimesters are determinants of birth weight and IUGR. Because exposures to the levels of ambient air pollutants observed in this study are common, and fetal growth is an important determinant for childhood and adult morbidity and mortality, our findings are likely to have important public health and regulatory implications.
Supplementary Material
supplemental material Supplemental material is available online at http://ehp.niehs.nih.gov/members/2005/8111/supplemental.pdf
We thank S.H. Alcorn for technical assistance in estimating air pollution levels.
This research was funded by the National Institute of Environmental Health Sciences (grants 5P01 ES009581, 5P01 ES011627, and 5P30 ES07048); the U.S. Environmental Protection Agency (grants R826708-01 and RD831861-01); the National Heart, Lung, and Blood Institute (grant 5R01HL61768); the California Air Resources Board (contract 94-331); and the Hastings Foundation.
Figure 1 Birth weight deficit by decile of O3[24hr] exposure averaged over the entire pregnancy compared with the decile group with the lowest O3 exposure. Deficits are plotted against the decile-group–specific median O3 exposure. Error bars represent 95% CIs. Indicator variables for each decile of O3[24hr] exposure (except the least-exposed group) were included in a mixed model. For detailed modeling information, see “Materials and Methods.”
Table 1 Characteristics of study subjects (n = 3,901)a born at term in California between 1975 and 1987.
Characteristic No. (%)
Gestational age (days)b 281.7 ± 10.3
Birth weight (g)b 3,486.7 ± 489.0
Sex (male) 1,888 (48.4)
LBW (< 2,500 g) 72 (1.8)
Term infants with IUGR 585 (15.0)
Maternal smoking during pregnancy 717 (18.8)
Unmarried at childbirth 632 (16.2)
Maternal gestational diabetes 18 (0.5)
Race/ethnicity
Non-Hispanic white 2,315 (59.7)
Hispanic white 1,096 (28.3)
African American/black 183 (4.7)
Asian/Pacific Islander 100 (2.6)
Other 184 (4.7)
Maternal age at childbirth (years)
< 20 356 (9.1)
20–22 607 (15.6)
23–29 1,917 (49.1)
30–32 550 (14.1)
33–35 298 (7.6)
≥36 173 (4.4)
Parity
1 1,630 (41.9)
2 1,299 (33.4)
3 603 (15.5)
4 225 (5.8)
5 68 (1.8)
6–9 64 (1.6)
Months since last live birth
0 (plural births) 25 (1.3)
10–25 537 (27.4)
26–36 464 (23.7)
36–57 446 (22.8)
≥58 485 (24.8)
SES based on parent’s/guardian’s education and annual household income
Less than high school education or annual income < $15,000 838 (22.0)
Completed high school with annual income ≥$15,000 655 (17.2)
Some college education with annual income ≥$15,000 1,421 (37.2)
Completed 4-year college with annual income ≥$15,000 310 (8.1)
Graduate-level education or annual family income ≥$100,000 593 (15.5)
a Data are presented as n (%) unless otherwise specified. Subjects with missing data were not used to calculate the percentages.
b Data presented as mean ± SD.
Table 2 Correlations for average exposures within each trimester of pregnancy.
Correlations
Exposure Mean ± SD O3[10–6] O3[24hr] PM10 NO2 CO Temperature Elevation
Entire pregnancy
O3[10–6] (ppb) 50.6 ± 17.5 1 0.85 0.54 0.26 0.00 0.41 0.43
O3[24hr] (ppb) 27.3 ± 8.7 1 0.20 −0.10 −0.27 0.17 0.60
PM10 (μg/m3) 45.8 ± 12.9 1 0.55 0.41 0.49 −0.02
NO2 (ppb) 36.1 ± 15.4 1 0.69 0.54 −0.16
CO (ppm) 1.8 ± 0.9 1 0.32 −0.35
Temperature (°F) 61.2 ± 4.2 1 −0.19
Elevation (m) 219.1 ± 263.5 1
First trimester
O3[10–6] (ppb) 51.0 ± 28.2 1 0.92 0.54 0.12 −0.17 0.69 0.28
O3[24hr] (ppb) 27.5 ± 14.1 1 0.34 −0.11 −0.32 0.60 0.38
PM10 (μg/m3) 46.6 ± 15.9 1 0.48 0.29 0.47 0.00
NO2 (ppb) 36.6 ± 16.9 1 0.68 0.20 −0.13
CO (ppm) 1.8 ± 1.1 1 −0.07 −0.29
Temperature (°F) 61.9 ± 7.3 1 −0.12
Second trimester
O3[10–6] (ppb) 49.9 ± 25.5 1 0.91 0.50 0.10 −0.20 0.67 0.27
O3[24hr] (ppb) 27.0 ± 12.8 1 0.27 −0.15 −0.37 0.57 0.38
PM10 (μg/m3) 45.4 ± 14.8 1 0.53 0.35 0.48 −0.04
NO2 (ppb) 36.2 ± 16.9 1 0.72 0.21 −0.15
CO (ppm) 1.8 ± 1.1 1 −0.04 −0.30
Temperature (°F) 61.5 ± 6.8 1 −0.14
Third trimester
O3[10–6] (ppb) 51.1 ± 27.1 1 0.91 0.52 0.11 −0.22 0.72 0.29
O3[24hr] (ppb) 27.5 ± 13.3 1 0.31 −0.14 −0.40 0.63 0.40
PM10 (μg/m3) 45.4 ± 15.5 1 0.52 0.37 0.48 −0.01
NO2 (ppb) 35.5 ± 16.6 1 0.71 0.18 −0.15
CO (ppm) 1.8 ± 1.1 1 −0.05 −0.30
Temperature (°F) 61.7 ± 7.1 1 −0.08
Table 3 Effects of air pollutants on birth weight from single-pollutant models.a
IQR Birth weight [g (95% CI)]b p-Value
Entire pregnancy
O3[10–6] 26 ppb −49.9 (−72.0 to −27.8) < 0.001
O3[24hr] 12 ppb −47.2 (−67.0 to −27.4) < 0.001
PM10 18 μg/m3 −19.9 (−43.6 to 3.8) 0.10
NO2 25 ppb −7.2 (−34.7 to 20.4) 0.61
CO 1.2 ppm 2.2 (−20.1 to 24.4) 0.85
First trimester
O3[10–6] 33 ppb −5.7 (−23.2 to 11.8) 0.52
O3[24hr] 17 ppb −10.4 (−28.6 to 7.7) 0.26
PM10 20 μg/m3 −3.0 (−22.7 to 16.7) 0.76
NO2 25 ppb −15.3 (−39.7 to 9.2) 0.22
CO 1.4 ppm −21.7 (−42.3 to −1.1) 0.04
Second trimester
O3[10–6] 29 ppb −25.7 (−42.7 to −8.7) 0.003
O3[24hr] 16 ppb −32.1 (−50.7 to −13.4) < 0.001
PM10 19 μg/m3 −15.7 (−36.1 to 4.7) 0.13
NO2 25 ppb 1.9 (−23.1 to 26.9) 0.88
CO 1.4 ppm 11.3 (−9.7 to 32.3) 0.29
Third trimester
O3[10–6] 33 ppb −36.7 (−54.9 to −18.5) < 0.001
O3[24hr] 17 ppb −35.2 (−54.6 to −15.8) < 0.001
PM10 20 μg/m3 −21.7 (−42.2 to −1.1) 0.04
NO2 25 ppb −6.1 (−31.0 to 18.9) 0.63
CO 1.3 ppm 11.8 (−8.4 to 32.1) 0.25
a For detailed modeling information, see “Materials and Methods.”
b Minus sign denotes reduction in mean birth weight.
Table 4 Effects of air pollutants on birth weight estimated from multipollutant models.a
IQR Birth weight [g (95% CI)]b p-Value
Model 1: first trimester
O3[24hr] 17 ppb −17.8 (−37.3 to 1.6) 0.07
CO 1.4 ppm −28.6 (−50.4 to −6.9) 0.01
Model 2: third trimester
O3[24hr] 17 ppb −31.6 (−52.2 to −11.0) 0.003
PM10 20 μg/m3 −10.8 (−31.8 to 10.2) 0.31
Model 3: all trimesters
O3[24hr] (first trimester) 17 ppb −18.6 (−39.0 to 1.7) 0.07
O3[24hr] (second trimester) 16 ppb −17.5 (−38.6 to 3.7) 0.11
O3[24hr] (third trimester) 17 ppb −36.9 (−58.9 to −15.0) < 0.001
a For detailed modeling information, see “Materials and Methods.”
b Minus sign denotes reduction in mean birth weight.
Table 5 Effects of air pollutants on LBW and IUGR from single-pollutant models.
IUGR
LBW
IQR OR (95% CI)a p-Value OR (95% CI)a p-Value
Entire pregnancy
O3[10–6] 26 ppb 1.2 (1.0 to 1.5) 0.02 1.5 (0.9 to 2.3) 0.10
O3[24hr] 12 ppb 1.2 (1.0 to 1.4) 0.01 1.3 (0.9 to 1.8) 0.18
PM10 18 μg/m3 1.1 (0.9 to 1.3) 0.49 1.3 (0.8 to 2.2) 0.22
NO2 25 ppb 1.1 (0.9 to 1.3) 0.51 0.8 (0.4 to 1.4) 0.44
CO 1.2 ppm 1.0 (0.9 to 1.2) 0.62 0.8 (0.6 to 1.3) 0.41
First trimester
O3[10–6] 33 ppb 1.0 (0.9 to 1.1) 0.92 1.0 (0.7 to 1.3) 0.92
O3[24hr] 17 ppb 1.0 (0.9 to 1.2) 0.72 1.0 (0.7 to 1.3) 0.87
PM10 20 μg/m3 1.0 (0.9 to 1.2) 0.94 1.0 (0.7 to 1.5) 0.85
NO2 25 ppb 1.2 (1.0 to 1.4) 0.07 0.9 (0.5 to 1.5) 0.67
CO 1.4 ppm 1.2 (1.0 to 1.4) 0.01 1.0 (0.7 to 1.5) 0.90
Second trimester
O3[10–6] 29 ppb 1.1 (1.0 to 1.2) 0.18 1.1 (0.8 to 1.5) 0.47
O3[24hr] 16 ppb 1.1 (1.0 to 1.2) 0.10 1.1 (0.8 to 1.5) 0.65
PM10 19 μg/m3 1.0 (0.9 to 1.2) 0.74 1.2 (0.8 to 1.7) 0.34
NO2 25 ppb 1.0 (0.8 to 1.2) 0.94 1.0 (0.6 to 1.6) 0.90
CO 1.4 ppm 1.0 (0.9 to 1.1) 0.72 0.9 (0.6 to 1.3) 0.66
Third trimester
O3[10–6] 33 ppb 1.2 (1.0 to 1.3) 0.01 1.4 (1.1 to 1.9) 0.02
O3[24hr] 17 ppb 1.2 (1.0 to 1.3) 0.02 1.4 (1.0 to 1.9) 0.03
PM10 20 μg/m3 1.1 (0.9 to 1.3) 0.29 1.3 (0.9 to 1.9) 0.13
NO2 25 ppb 1.0 (0.8 to 1.2) 0.93 0.6 (0.4 to 1.1) 0.11
CO 1.3 ppm 1.0 (0.8 to 1.1) 0.51 0.7 (0.5 to 1.1) 0.09
a For detailed modeling information, see “Materials and Methods.”
==== Refs
References
Barclay ML 1997. Critical physiologic alterations in pregnancy. In: Emergency Care of the Woman (Pearlman MD, Tintinalli JE, eds). New York:McGraw Hill, 303–312.
Basu R Woodruff TJ Parker JD Saulnier L Schoendorf KC 2004 Comparing exposure metrics in the relationship between PM2.5 and birth weight in California J Expo Anal Environ Epidemiol 14 5 391 396 15361898
Bignami G Musi B Dell’Omo G Laviola G Alleva E 1994 Limited effects of ozone exposure during pregnancy on physical and neurobehavioral development of CD-1 mice Toxicol Appl Pharmacol 129 2 264 271 7992316
Bobak M 2000 Outdoor air pollution, low birth weight, and prematurity Environ Health Perspect 108 173 176 10656859
Bobak M Leon DA 1999 Pregnancy outcomes and outdoor air pollution: an ecological study in districts of the Czech Republic 1986–8 Occup Environ Med 56 8 539 543 10492651
Bobak M Richards M Wadsworth M 2001 Air pollution and birth weight in Britain in 1946 Epidemiology 12 3 358 359 11338317
Chen L Yang W Jennison BL Goodrich A Omaye ST 2002 Air pollution and birth weight in northern Nevada, 1991–1999 Inhal Toxicol 14 2 141 157 12122577
Cohen GR Curet LB Levine RJ Ewell MG Morris CD Catalano PM 2001 Ethnicity, nutrition, and birth outcomes in nulliparous women Am J Obstet Gynecol 185 3 660 667 11568795
De Boor C 2001. A Practical Guide to Splines. New York:Springer-Verlag.
Dejmek J Selevan SG Benes I Solansky I Šrám RJ 1999 Fetal growth and maternal exposure to particulate matter during pregnancy Environ Health Perspect 107 475 480 10339448
Dell’Omo G Fiore M Petruzzi S Alleva E Bignami G 1995 Neurobehavioral development of CD-1 mice after combined gestational and postnatal exposure to ozone Arch Toxicol 69 9 608 616 8660138
Di Cera E Doyle ML Morgan MS De Cristofaro R Landolfi R Bizzi B 1989 Carbon monoxide and oxygen binding to human hemoglobin F0 Biochemistry 28 6 2631 2638 2471551
Fell DB Dodds L King WD 2004 Residential mobility during pregnancy Paediatr Perinat Epidemiol 18 6 408 414 15535816
Gouveia N Bremner SA Novaes HM 2004 Association between ambient air pollution and birth weight in São Paulo, Brazil J Epidemiol Community Health 58 1 11 17 14684720
Gunnison AF Hatch GE 1999 O3-induced inflammation in prepregnant, pregnant, and lactating rats correlates with O3 dose estimated by 18O Am J Physiol 276 2 pt 1 L332 L340 9950896
Ha EH Hong YC Lee BE Woo BH Schwartz J Christiani DC 2001 Is air pollution a risk factor for low birth weight in Seoul? Epidemiology 12 6 643 648 11679791
Haiek L Lederman SA 1989 The relationship between maternal weight for height and term birth weight in teens and adult women J Adolesc Health Care 10 1 16 22 2921184
Hemmingsen A Allen JT Zhang S Mortensen J Spiteri MA 1999 Early detection of ozone-induced hydroperoxides in epithelial cells by a novel infrared spectroscopic method Free Radic Res 31 5 437 448 10547188
Kavlock R Daston G Grabowski CT 1979 Studies on the developmental toxicity of ozone. I. Prenatal effects Toxicol Appl Pharmacol 48 1 pt 1 19 28 452041
Khoury MJ Stewart W Weinstein A Panny S Lindsay P Eisenberg M 1988 Residential mobility during pregnancy: implications for environmental teratogenesis J Clin Epidemiol 41 1 15 20 3335868
Kramer MS 1987 Determinants of low birth weight: methodological assessment and meta-analysis Bull WHO 65 5 663 737 3322602
Kramer MS Seguin L Lydon J Goulet L 2000 Socio-economic disparities in pregnancy outcome: why do the poor fare so poorly? Paediatr Perinat Epidemiol 14 3 194 210 10949211
Larini A Bocci V 2005 Effects of ozone on isolated peripheral blood mononuclear cells Toxicol In Vitro 19 1 55 61 15582356
Lee BE Ha EH Park HS Kim YJ Hong YC Kim H 2003 Exposure to air pollution during different gestational phases contributes to risks of low birth weight Hum Reprod 18 3 638 643 12615838
Lin CM Li CY Yang GY Mao IF 2004 Association between maternal exposure to elevated ambient sulfur dioxide during pregnancy and term low birth weight Environ Res 96 1 41 50 15261783
Liu S Krewski D Shi Y Chen Y Burnett RT 2003 Association between gaseous ambient air pollutants and adverse pregnancy outcomes in Vancouver, Canada Environ Health Perspect 111 1773 1778 14594630
Maisonet M Bush TJ Correa A Jaakkola JJ 2001 Relation between ambient air pollution and low birth weight in the northeastern United States Environ Health Perspect 109 suppl 3 351 356 11427384
Mudway IS Krishna MT Frew AJ MacLeod D Sandstrom T Holgate ST 1999 Compromised concentrations of ascorbate in fluid lining the respiratory tract in human subjects after exposure to ozone Occup Environ Med 56 7 473 481 10472319
Parker JD Woodruff TJ Basu R Schoendorf KC 2005 Air pollution and birth weight among term infants in California Pediatrics 115 1 121 128 15629991
Peters JM Avol E Gauderman WJ Linn WS Navidi W London SJ 1999a A study of twelve southern California communities with differing levels and types of air pollution. II. Effects on pulmonary function Am J Respir Crit Care Med 159 3 768 775 10051249
Peters JM Avol E Navidi W London SJ Gauderman WJ Lurmann F 1999b A study of twelve southern California communities with differing levels and types of air pollution. I. Prevalence of respiratory morbidity Am J Respir Crit Care Med 159 3 760 767 10051248
Ritz B Yu F 1999 The effect of ambient carbon monoxide on low birth weight among children born in southern California between 1989 and 1993 Environ Health Perspect 107 17 25 9872713
Sangalli MR McLean AJ Peek MJ Rivory LP Le Couteur DG 2003 Carbon monoxide disposition and permeability-surface area product in the foetal circulation of the perfused term human placenta Placenta 24 1 8 11 12495653
Schulman J Selvin S Shaw GM Malcoe LH 1993 Exposure misclassification due to residential mobility during pregnancy in epidemiologic investigations of congenital malformations Arch Environ Health 48 2 114 119 8476302
Shaw GM Malcoe LH 1992 Residential mobility during pregnancy for mothers of infants with or without congenital cardiac anomalies: a reprint Arch Environ Health 47 3 236 238 1596108
Thibodeaux JR Hanson RG Rogers JM Grey BE Barbee BD Richards JH 2003 Exposure to perfluorooctane sulfonate during pregnancy in rat and mouse. I: maternal and prenatal evaluations Toxicol Sci 74 2 369 381 12773773
U.S. EPA (U.S. Environmental Protection Agency) 2005. Technology Transfer Network Air Quality System (AQS). Available: http://www.epa.gov/ttn/airs/airsaqs/detaildata/downloadaqsdata.htm [accessed 5 October 2005].
U.S. Geological Survey 2005. Earth Resources Observation Systems Data Center. Available: http://edc.usgs.gov/products/elevation/gtopo30.html [accessed 5 October 2005].
Von Behren J Reynolds P 2003 Birth characteristics and brain cancers in young children Int J Epidemiol 32 2 248 256 12714545
Wang X Ding H Ryan L Xu X 1997 Association between air pollution and low birth weight: a community-based study Environ Health Perspect 105 514 520 9222137
Williams L Spence A Tideman SC 1977 Implications of the observed effects of air pollution on birth weight Soc Biol 24 1 1 9 877600
Yang CY Tseng YT Chang CC 2003 Effects of air pollution on birth weight among children born between 1995 and 1997 in Kaohsiung, Taiwan J Toxicol Environ Health A 66 9 807 816 12746128
Zeger SL Thomas D Dominici F Samet JM Schwartz J Dockery D 2000 Exposure measurement error in time-series studies of air pollution: concepts and consequences Environ Health Perspect 108 419 426 10811568
|
16263524
|
PMC1310931
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 18; 113(11):1638-1644
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8111
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.7994ehp0113-00164516263525ResearchChildren's HealthIn Utero Exposure to Dioxins and Polychlorinated Biphenyls and Its Relations to Thyroid Function and Growth Hormone in Newborns Wang Shu-Li 1Su Pen-Hua 2Jong Shiang-Bin 3Guo Yueliang L. 4Chou Wei-Ling 1Päpke Olaf 51 Division of Environmental Health and Occupational Medicine, National Health Research Institutes, and Graduate Institute of Occupational Safety and Health, Kaohsiung Medical University, Kaohsiung, Taiwan2 Institute of Medicine, Chung Shan Medical University, and Department of Pediatrics, Chung Shan Medical University Hospital, Taichung, Taiwan3 Department of Nuclear Medicine, Kaohsiung Medical University Hospital, Kaohsiung, Taiwan4 Department of Occupational and Environmental Health, College of Medicine, National Cheng Kung University, Tainan, Taiwan5 ERGO Research Laboratory, Hamburg, GermanyAddress correspondence to S.-L. Wang, No. 35, Keyan Rd., Zhunan Town, Miaoli County 350, Taiwan. Telephone: 886-935-285848. Fax: 886-7-316 2725. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 27 6 2005 113 11 1645 1650 4 2 2005 27 6 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. The aim of this study is to examine the association between transplacental exposure to dioxins/polychlorinated biphenyls (PCBs) and thyroid and growth hormones in newborns. We recruited 118 pregnant women, between 25 and 34 years of age, at the obstetric clinic. Personal data collected included reproductive and medical histories and physical factors. Clinicians gathered placental and umbilical cord serum upon delivery and carefully scored the 118 newborns, making both structural and functional assessments. We analyzed placentas for 17 polychlorinated dibenzo-p-dioxins and dibenzofurans and 12 dioxin-like PCB congeners with the World Health Organization–defined toxic equivalent factors, and six indicator PCBs by high-resolution gas chromatography and high-resolution mass spectrometry. We analyzed thyroid and growth hormones from cord serum using radioimmunoassay. Insulin-like growth factor (IGF)-1, IGF-binding globulin-3, and thyroxine × yroid-stimulating hormone (T4 × TSH) were significantly associated with increased placental weight and Quetelet index (in kilograms per square meter; correlation coefficient r = 0.2–0.3; p < 0.05). Multivariate analyses showed independently and significantly decreased free T4 (FT4) × TSH with increasing non-ortho PCBs (r = −0.2; p < 0.05). We suggest that significant FT4 feedback alterations to the hypothalamus result from in utero exposure to non-ortho PCBs. Considering the vast existence of bioaccumulated dioxins and PCBs and the resultant body burden in modern society, we suggest routine screening of both thyroid hormone levels and thyroid function in newborns.
dioxinsinfantplacentaprenatal exposure delayed effectsthyroid hormones
==== Body
Polychlorinated dibenzo-p-dioxins and dibenzo-furans (PCDD/PCDFs), known together as dioxins, and polychlorinated biphenyls (PCBs) are ubiquitous, persistent organic pollutants that act as endocrine disruptors in eco- and bio-systems. Various industries, the combustion of solid wastes, and forest fires generate these xenobiotic pollutants. Human milk surveys have established a high body burden for these substances in long-industrialized countries, such as those in Europe (Schecter and Piskac 2001). These contaminants are highly lipophilic and are potentially bioamplified through the food chain. Humans accumulate such compounds mainly from foods of animal origin, including fish (Smith and Gangolli 2002), meat, dairy products, and processed foods such as fish oil (Chen et al. 2003; Rideout and Teschke 2004). PCDD/PCDFs and PCBs have long half-lives of 7–10 years in humans. The body burden and related health effects are particularly concerning in countries with high population density, such as Taiwan, due potentially to the increased numbers of factories, solid waste incinerators (Tango et al. 2004), and the volume of imported foods from animal sources.
Our previous study established the transfer of PCDD/PCDFs and PCBs from mother to infant via the placenta (Wang et al. 2004) in the general population. These endocrine-disrupting effects upon thyroid function might significantly affect growth and development during fast growth stages of the central nervous system in the human fetus (Koopman-Esseboom et al. 1994; Pluim et al. 1992). These effects on fetal growth and development, particularly as related to the ectodermal layer (Guo et al. 2000; Jacobson and Jacobson 1996), are irreversible and can last for several years of life (Chen et al. 1992; Jacobson and Jacobson 1996; Patandin et al. 1999). Dental health might also be affected (Alaluusua et al. 1999; Wang et al. 2003); however, this mechanism awaits further investigation. Early hypo- or hyperthyroidism or thyroid receptor alterations can cause permanent behavioral, intellectual, and neurologic dysfunction, particularly during the second and third months of gestation (Utiger 1999). Human thyroid function is similar to that of other vertebrates, but the thyroid system is well developed and functioning at birth (Fisher and Brown 2000). Animal studies have shown that dioxins and PCBs might affect thyroid hormone secretion, transport, and/or action (Leatherland 1999). More recent studies have shown that low-dose 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) might cause rat thyroid morphologic and functional abnormalities through gestational and lactational exposure (Nishimura et al. 2003).
Different dioxin and PCB congeners might have different toxicities. The approach that uses toxic equivalent factors to summarize toxicity is subject to debate particularly for dioxin-like PCBs (Van den Berg et al. 1998), because the toxicity of dioxin-like PCBs in the presence of non-dioxin-like PCBs cannot be summarized simply by additive effects. The present study provides an opportunity to examine both single and summarized levels.
There have been inconsistent outcomes regarding thyroid function alterations resulting from dioxins and PCBs in humans with background exposure in both the Netherlands (Koopman-Esseboom et al. 1994; Pluim et al. 1993) and Japan (Matsuura et al. 2001; Nagayama et al. 1998). Such diverse thyroid hormone findings might result from using different matrices (Kimbrough and Krouskas 2001) for measuring exposure (i.e., breast milk or venous blood), various specimen collection time points, and/or studied isomers (dioxins, indicator PCBs) for different populations at unique sample sizes. The aim of this study is to examine the association between transplacental exposure to various dioxin and PCB congeners and thyroid hormone status in 118 mother–newborn pairs from the general Taiwanese population, taking into account growth hormones and birth outcomes.
Materials and Methods
Subjects.
This is part of a prospective study of dioxins/PCBs for the general population in central Taiwan. Subject recruitment was described in a previous study (Wang et al. 2004) with a detailed birth cohort (Chao et al. 2004). In summary, we invited all pregnant women visiting the local medical center between December 2000 and November 2001 to participate in this study. We interviewed 430 subjects at the obstetric clinic, collecting personal data that included reproductive and medical histories, and physical factors. The subjects were all between 25 and 34 years of age with a single pregnancy and without known complications, cigarette smoking, or alcohol consumption during the pregnancy. Clinicians gathered placental and umbilical cord blood upon neonate delivery. One well-trained nurse carefully scored all newborns, including both structural and functional assessments, which a physician verified.
The Human Ethical Committee of the National Health Research Institutes in Taiwan reviewed and approved the study protocol. This study followed the ethical standards formulated from the Helsinki Declarations of 1964 and revised in 2000 (World Medical Association 2000). Each of the participants provided informed consent after receiving a detailed explanation of the study and its potential consequences.
Specimen and data collections.
Immediately after collection on the ward, technicians divided each placenta into two equally symmetrical parts, taking half for the study sample. Technicians then sliced each sample into pieces, storing 100 g of each sample in Teflon-lined cap glass bottles for future dioxin and PCB measurements, and then shipped the frozen placental sections to O. Päpke in ERGO Research Laboratory (Hamburg, Germany), which is certified by the World Health Organization (WHO), for analysis. This laboratory regularly and successfully participates in international interlaboratory studies, including PCDD/PCDFs in human milk, beef, and fish liver (Päpke 1998).
Analyses of PCDD/PCDFs and PCBs were performed according to a previously published method (Päpke et al. 1994). Briefly, we extracted 100-g placenta samples with n-pentane after adding an internal standard, such as 13C12-PCDD/PCDF or 13C12-PCB. We determined the lipid content of the samples using a gravimetric method before cleanup in a multicolumn system. The congeners, with the WHO-defined toxic equivalent factors (Van den Berg et al. 1998) of seventeen 2,3,7,8-substituted PCDD/PCDFs, 12 dioxin-like PCBs including non-ortho and mono-ortho PCBs, and six indicator PCBs [International Union for Pure and Applied Chemistry (IUPAC) congeners 28, 52, 101, 138, 153, and 180] were analyzed by gas chromatography (Hewlett Packard GC 5890 Series II; GMI Inc., Ramsey, MN, USA) with high-resolution mass spectrometry (VG-AutoSpec, Manchester, UK). We purchased authentic standards of native dioxin-like PCBs and PCDD/PCDFs from AkkuStandard Inc. (New Haven, CT, USA). We obtained indicator PCB standards from LGC Promochem (Wesel, Germany). We then measured two isotope masses for each component. We used internal/external standard mixtures via the isotope dilution method for quantification.
In our study we examined 118 newborns with complete data including dioxin/PCB levels in the placenta and thyroid hormone status via the cord serum. We noted increased thyroid hormone levels during the first few hours of life due to the expected postparturition surge of thyroid-stimulating hormone (TSH). Thyroid hormone concentration measurements in the cord blood were similar to those measured in the venous blood at 24–72 hr or 2 weeks of age (Mashang and Thornton 1999). We collected cord blood upon umbilical cord ligation, which we completed within 1 min of childbirth; thus, we used cord levels as indices of thyroid function without phlebotomizing the infants. The decreased number of subjects reported is a result of the economic dioxin analyses necessary to achieve a statistical power sufficient to generate a valid conclusion. Thyroid and related growth hormones were measured using radio-immunoassay methods, including triiodothyronine (T3), thyroxine (T4), thyroid-stimulating hormone (TSH; thyrotropin), free T4 (FT4), T3 uptake, thyroid-binding globulin (TBG), insulin-like growth factor (IGF)-1, and IGF-binding globulin-3 (BP3). We carried out blind duplicates for every 10 samples. We purchased T3, T4, TSH, IGF-1, and BP3 commercialized kits from Daiichi Radioisotope Laboratory (Tokyo, Japan). FT4 kits were from CIS Bio International US Inc. (Bedford, MA, USA); T3 uptake kits were from Diagnostic Products Corporation (Los Angeles, CA, USA); TBG kits were from CIS BIO International (Marcoule, France).
Statistical methods.
We verified the data distribution of each continuous variable for normality and generally noted data skewing slightly to the right. When data distribution was significantly beyond the normal distribution range, we used the Mann-Whitney U-test for determining differences (i.e., hormone levels) between high- and low-exposure groups; otherwise, we performed the Student t-test. We used the chi-square test for categorical variables. We calculated the multiplied concentrations of FT4 × TSH based on the normal hypothalamic–pituitary axis in which decreased T4 feeds back to the hypothalamus and thereby stimulates the anterior pituitary to secret TSH (Scanlon and Hall 1998). We also calculated the concentration ratio of T4:TBG because TBG is the main T4 carrier protein, and generally there is positive association between the concentrations of T4 and TBG (Porterfield 2001). We used Spearman correlation analyses to evaluate the association between PCDD/PCDF and PCB levels and hormone levels. For the present report, we confirmed the results using Pearson correlation analyses after log transformation. We used a general linear model to examine the relationship of the hormone log values to the dioxin/PCB exposure values. To increase statistical power, we combined the results of male and female infants when their patterns were the same. We carried out multivariate linear regression analyses to adjust for age and other appropriate variables. We used the Statistical Package for Social Science (version 10.0; SPSS, Chicago, IL, USA) to execute all statistical analyses. For data presentation, we used international units (IU), except for Figures 1 and 2 in which IU was not used to illustrate different hormone levels in single figures.
Results
The newborns were generally healthy with a mean (± SD) body weight of 3,229 ± 371 and Apgar scores at 1 and 5 min of 8.4 and 9.7, respectively. The mothers had a mean (± SD) age of 29 ± 4 years (Table 1). Male infants tended to have greater birth weights and head girths (p < 0.1) than did female infants. Approximately half of the infants were first parity, with an even distribution of males and females. Higher dioxin/PCB exposure groups correlated significantly with increased maternal age and longer infant length. Higher parity might confound the length relationship, although this relationship did not persist with multivariate analysis. One- and 5-min Apgar scores were both slightly lower in the high-exposure group, but this was without statistical significance (p = 0.19, 0.35, respectively). Table 2 demonstrates significantly increased concentrations of T3, TBG, and BP3 and decreased TSH in female infants, whereas in males, decreased TSH values were not statistically significant. The 95% confidence intervals (CIs) for T3, T4, TSH, and FT4 were 54.3–60.6 (ng/dL), 8.4–9.1 (μg/dL), 6.8–8.9 (μU/mL), and 0.77–0.85 (ng/dL), respectively. The concentrations in the present study were within the normal range for cord blood (Mashang and Thornton 1999). We excluded one subject who was seen in follow-up for thyroid dysfunction and dropped out of the study from the present report.
Maternal age was significantly associated with decreased gestational age and Apgar scores (Table 3). Further analyses showed that maternal age correlated with increased parity and was associated with increased infant weight and length. Therefore, the greater the parity, the shorter the gestational age for an infant with a normal-range weight. IGF-1 and BP3 concentrations in cord serum were significantly and positively associated with greater placental weight, birth weight, head girth, and Quetelet index (QI) (correlation coefficient, 0.2–0.3). FT4 × TSH and T4 × TSH were both significantly and positively associated with placental weight and QI (r = 0.2–0.3). Maternal thyroid hormones at the third trimester of pregnancy did not correlate with birth outcomes; therefore, these are not included in the present report.
We examined whether the key growth- and development-related hormones in cord blood were associated with dioxin/PCB body burden after adjusting for maternal age (Table 4). We found significant correlations mainly for female infants. First, increased T3 levels were associated with increased PCDFs. Second, decreased TSH levels were associated with increased PCDDs and dioxin-like PCBs. Third, increased TBG levels were associated with increased PCDD/PCDFs. Fourth, decreased FT4 × TSH levels were associated with increased non-ortho PCBs. Fifth, decreased T4:TBG and T3 uptake:T3 levels were associated with increased PCDD/PCDFs. The concentrations of PCB-138, -153, and -180 were not significantly associated with key hormone levels and related only somewhat to T4 and FT4 levels in females.
Table 5 shows the results of multivariate analyses for significant relationships between the hormone and dioxin/PCB levels after adjusting for maternal age and other dioxin/PCB isomers. We combined PCDDs and PCDFs with the same relationship pattern to increase statistical power. We noted the continuation of a significant and positive association between T4 concentrations with levels of PCDD/PCDFs. We also noted a significant relationship between TBG levels and increased PCDF levels. FT4 × TSH levels were negatively associated with non-ortho PCB toxic equivalents (TEq; r = −0.25) and positively related to concentrations of PCB-138, -153, and -180 (r = 0.02; data not shown) independent of age and other congeners. We noted a negative relationship between FT4 levels and non-ortho PCB levels. In addition, maternal age was negatively associated with T4, TBG, and FT4 × TSH. Mainly because of low levels and values below the detection limit, mono-ortho PCBs have not demonstrated a significant association with hormone levels.
We also noted increased T3 and T4 concentrations, as well as decreased TSH and FT4 × TSH, with increasing total dioxins and PCB TEq levels, the latter two statistically significant for female infants (Figure 1A). For male infants, similar patterns were not statistically significant (Figure 1B). Non-ortho PCBs were associated with decreased FT4 × TSH for both sexes after examining various congener groups (Figure 2). Each subject’s level was plotted for non-ortho PCBs against FT4 × TSH (Figure 3). Both linear and nonlinear models demonstrated a significant negative association. The 95% CIs for values of FT4 × TSH and T4 × TSH were 5.5–7.4 ng/dL × μU/mL and 58.3–78.5 μg/dL × μU/mL, respectively.
Discussion
This is the first study demonstrating the in utero effects, as measured in the placenta, of dioxins and PCBs on thyroid function, considering various growth factors in newborn infants from the general population. Thyroid function is crucial for infant growth and development (Polk and Fisher 1998), and previous studies of 38 healthy, term infants have shown borderline significance for increased T4 and TBG concentrations in upper-median dioxin-exposed groups as measured in breast milk (Pluim et al. 1992, 1993). Here, we demonstrate consistently increased T3, T4, and TBG levels in cord blood correlating to upper-median exposure groups for female but not male infants. The increases in T4 and TBG levels seen with increasing dioxins persisted even after adjusting for maternal age and PCB exposure.
We further demonstrated decreased FT4 × TSH levels with increasing non-ortho PCBs, which remained significant even after adjusting for other dioxin and PCB congeners. This phenomenon was most noteworthy in female infants. FT4 × TSH levels could be predicted with significance by using a simple linear regression model with non-ortho PCB TEq levels as predictors. FT4 is the major hormone affecting hypothalamus-stimulating TSH (Scanlon and Hall 1998). FT4 feedback might not be effective enough to stimulate the hypothalamus to secrete TSH, resulting in decreasing TSH levels. We also demonstrated that TSH and FT4 × TSH are significantly and positively associated with placental weight and QI, signifying the general growth impact from the two hormones. We recommend routine thyroid function screening in newborns as a protective measure in the face of ubiquitous environmental endocrine disruptors, particularly for bioaccumulated and transplacental dioxins and the PCB body burden. Such screening should include both individual thyroid hormone levels (i.e., TSH and/or T4) and relationship values (i.e., FT4 × TSH) for functional evaluations. We did not emphasize the significant reduction of T3 and T4 concentrations with increasing non-ortho PCBs because of the positive association between the hormone levels and dioxins, particularly for T4 and PCDDs, possibly due to the increased TBG concentration with increasing PCDDs. Various conger considerations are important because of the multiple exposures that humans receive.
Regarding sex differences, we noted significant increases in BP3 in high-exposure female infants in the present study. Female infants exposed to slightly higher dioxin and PCB levels were without statistical significance, however, and these female infants generally had similar hormone levels to the male infants. In addition, female infants were slightly smaller and had higher bilirubin levels than did male infants. It was easier to differentiate hormone levels by upper- and lower-median dioxin/PCB levels in females, which indicated greater sensitively to endocrine disruptors in female compared with male neonates. Sex differences in thyroid hormones were also found in Michigan Lake fish eaters (Persky et al. 2001). Most experts consider dioxins antiestrogenic with context-dependent effects, which might explain the greater hormonal outcomes observed in females. Estrogen receptors in thyroid cells (Scanlon and Hall 1998) and greater responses of TSH to thyrotropin-releasing hormones in females might also contribute. Further investigation is necessary to elucidate the reasons behind these sex differences.
Prior research had different results than those of the present study. In a study of residents living in a highly industrialized region of the Netherlands (Koopman-Esseboom et al. 1994), dioxins and PCBs in breast milk collected at 14 days postpartum showed no association with T4, T3, FT4, and TSH concentrations in cord blood but showed significantly decreased T4 and increased TSH levels at 2 weeks and 3 months postpartum. Among these 78 mother–infant pairs collected from surrounding industrial areas of Rotterdam, TSH varied from 10.0 μIU/mL at birth to 2.3 μIU/mL at 2 weeks. Dioxins and PCBs in breast milk ranked high in Europe at 30.8 pg TEq/g fat compared with 15.1 in the present study; however, the method detection limit has been largely lowered because of technical improvements in the last 10 years. In addition, dioxin and PCB levels in breast milk might decrease with parity and duration of the feeding; therefore, it is difficult to make direct comparisons between the Dutch study and the present study. Nonetheless, both studies drew the same conclusions regarding altered thyroid hormone status as related to dioxins and PCB exposure, and both conclude that the data warrant further follow-up.
More recent studies of the general population in Japan showed no association between TSH and FT4 concentrations at 1 year with dioxins and PCBs in breast milk collected at 30 postpartum days (Matsuura et al. 2001). This indicated that breast-feeding might not change thyroid hormone levels significantly. As recommended by Rogan and Miller (1989), the ectodermal effects of in utero longer term dioxin and PCB exposure needs to be examined. Our study of dioxins and PCBs in 118 placentas demonstrated associated hormonal changes, while accounting for various growth factors and covariates. For instance, we noted higher dioxin and PCB body loads with increased maternal age, which results in increased transplacental and lactational exposure to fetus and infant, respectively. We suggest that researchers consider age when evaluating the effects of persistent thyroid hormone disruptors. In the present study, we found that maternal age was associated with increased dioxin levels, decreased gestational age, and decreased T4 and TSH concentrations, within 25–34 years of age. Maternal age adjustments in our study might explain why our outcomes differed from those of previous studies.
Animal studies have indicated that thyroid function changes according to gestational and lactational exposure to TCDD and PCBs (Fisher and Brown 2000; Nishimura et al. 2003; Rosiak et al. 1997). In the offspring of Holtzman rats sensitive to TCDD, serum T4 levels decreased on postnatal day 21 and increased by postnatal day 49 when exposed to a low-dose (either 200 or 800 ng/kg) at gestational day 15 (Nishimura et al. 2003). T3 increased at both postnatal day 21 and postnatal day 49, but this increase was less significant. Nishimura et al. (2003) concluded the induction of the uridine diphosphate-glucuronosyltransferase-1 gene in the liver increases T4 metabolism. In this animal study, hyperplasia of the thyroid gland might result from decreased T4 and sustained high TSH levels with a corresponding feedback system disorder. We observed greater T3 levels but did not observe increased TSH levels in female neonates in relation to increased PCDFs. Background exposure might affect humans differently than animal models. In addition, levels of TBG, the major T4 carrier protein in the peripheral blood (LaFranchi 2000), were significantly increased with dioxin exposure. TBG might be increased during pregnancy secondary to maternal estrogen effects (Mashang and Thornton 1999); therefore, researchers recommended further analyses of estrogens in cord blood. We found FT4 and T3 uptake were slightly and negatively associated with dioxins, indicating there might be fewer unoccupied binding sites on TBG. Additional studies to examine carrier protein changes using animal models would help further clarify regulation processes and help define the reference levels for pathologic diagnosis.
Although indicator PCBs have been associated with decreased T4 levels and sustained TSH stimulation in Sprague-Dawley pups (Ness et al. 1993), the present study found no major correlation between the various thyroid hormones and PCB-138, -153, and -180. The only correlation noted in this study was that of increased T4 levels with increasing concentrations of non-dioxin-like PCBs in female infants as determined by univariate analysis. Differences in exposure levels of endocrine disruptors might reveal different effects (Brouwer et al. 1999).
There are several methodologic concerns in studies of this nature. First, it is difficult to draw venous blood from infants who are < 3 months of age. Nonetheless, the pattern of hormonal changes remained consistent at 1 and 11 postnatal weeks according to observation of infants in the general population of the Netherlands (Pluim et al. 1993). Second, we did not include routine TSH regulatory monitoring results in the present report because venous concentrations varied with time points of the blood draws; thus, this was not suitable data for comparison. The pregnant women in our study did not have iodine insufficiency concerns to our understanding, probably because of the proximity to the sea and dietary habits, which include the intake of processed foods using seaweed containing iodine. For exposure measurements, human milk dioxin/PCB levels decreased substantially with the duration of breast-feeding. The percentage of all congener-specific analyses for PCDD/PCDF and PCB levels greater than the detection limit is 90% for placenta, compared with 75% for human milk and 43% for cord serum. Placental analyses might prove to be good indicators of in utero exposure for newborns.
Conclusions
We found significantly decreased FT4 feedback to the hypothalamus resulting from in utero exposure to non-ortho PCBs in neonates from the general population. The mechanism of action of the compound treatment remains to be determined. We also recommend routine screening of both individual thyroid hormonal levels and thyroid function in newborns, considering the pervasive existence of bioaccumulated dioxins and PCBs and the resulting body burden in modern society. We recommend further studies that include antiestrogenic effects to examine the positive correlation between TBG and increased dioxins. It is worthwhile to follow the growth and development for those with altered thyroid status.
We are grateful to L.-Y. Lin for placenta gathering, C.-N. Huang for providing endocrinology documents, H.-Y. Yu for specimen collections, and K.-H. Chang for final statistical assistance.
The study was performed under grants from the National Health Research Institutes (EO-091-PP-01 and EO-092-PP-05), Taiwan.
The scientific content of the manuscript has been reviewed and approved for publication by the Division of Environmental Health and Occupational Medicine (DEHOM) of the National Health Research Institutes (NHRI). Approval for publication does not necessarily signify that the content reflects the view and policies of the DEHOM/NHRI, or condemnation or endorsement and recommendation for use on this issue presented.
Figure 1 Concentrations of T3 (ng/dL), T4 (μg/dL), TSH (μU/mL), and FT4 × TSH (ng/dL × μU/mL) in cord serum according to quartiles of total dioxins and PCBs (pg-TEq/g lipid) in placenta: (A) females; (B) males.
*p < 0.05; #p < 0.1.
Figure 2 Concentrations of T3 (ng/dL), T4 (μg/dL), TSH (μU/mL), and FT4 × TSH (ng/dL × μU/mL) in cord serum according to quartiles of non-ortho PCBs (pg-TEq/g lipid) in placenta: (A) females; (B) males.
*p < 0.05; #p < 0.1.
Figure 3 Decreased FT4 × TSH levels according to concentrations of non-ortho PCBs. By general linear model: FT4 × TSH = 8.32–1.05 × (non-ortho PCB); R = 0.245, p = 0.009. By quadratic model: FT4 × TSH = 9.06–1.77 × (non-ortho PCB) + 0.116 × (non-ortho PCB)2; R = 0.261, p = 0.022.
Table 1 Birth outcomes by sex and median of dioxin and PCB TEq level in the newborn babies (mean ± SD).
Sex (n)
Dioxin/PCB TEqa (pg/g lipid)
Factor (unit) Female (n = 62) Male (n = 57) p-Valueb Low (< 15.1; n = 59) High (≥ 15.1; n = 60) p-Valueb Total factor (unit)
Dioxin/PCB TEq 16.38 ± 6.76 16.01 ± 5.44 0.75 11.59 ± 2.43 20.74 ± 5.24 < 0.001 16.20 (6.14)
Mother’s age 29.81 ± 4.60 28.65 ± 4.11 0.15 27.54 ± 4.03 30.93 ± 4.10 < 0.001 29.35 (4.04)
Placenta weight (g) 574.4 ± 150.4 571.7 ± 138.6 0.92 582.5 ± 165.5 563.7 ± 119.7 0.50 568.2 (122.5)
Placenta fat content (%) 0.75 ± 0.10 0.75 ± 0.13 0.98 0.75 ± 0.11 0.75 ± 0.11 0.94 0.77 (0.12)
Gestational age (weeks) 38.64 ± 1.33 39.15 ± 1.42 0.05 39.00 ± 1.31 38.76 ± 1.47 0.35 38.80 (1.37)
Baby birth weight (g) 3,051 ± 411.7 3,181 ± 416.3 0.09 3,062 ± 416.6 3,163 ± 415.3 0.19 3,229 (371.1)
Baby birth length (cm) 51.09 ± 2.27 51.66 ± 2.46 0.19 50.90 ± 2.34 51.82 ± 2.33 0.03 51.86 (2.21)
QI (kg/m2) 11.66 ± 1.27 11.90 ± 1.18 0.29 11.78 ± 1.22 11.77 ± 1.25 0.95 12.01 (1.20)
Baby head girth (cm) 33.23 ± 1.37 33.69 ± 1.21 0.05 33.39 ± 1.38 33.51 ± 1.25 0.62 33.69 (1.25)
Baby chest girth (cm) 32.75 ± 1.70 32.94 ± 1.73 0.55 32.82 ± 1.65 32.86 ± 1.79 0.91 33.07 (1.64)
1st min Apgar score 8.59 ± 1.28 8.32 ± 0.76 0.16 8.55 ± 1.34 8.37 ± 0.71 0.35 8.35 (0.75)
5th min Apgar score 9.92 ± 1.13 9.68 ± 0.47 0.15 9.91 ± 1.16 9.70 ± 0.46 0.19 9.73 (0.45)
Baby bilirubin (mg/dL) 8.80 ± 2.61 7.93 ± 2.31 0.07 8.62 ± 2.64 8.09 ± 2.33 0.27 7.96 (2.30)
Parity (%)
1st 53 54 0.95c 58 50 0.63c 54 (64)
2nd 32 30 27 35 31 (37)
≥ 3rd 15 16 15 15 15 (18)
a The 17 PCDD/F and 12 dioxin-like PCB congeners with the WHO toxic equivalent factors were measured in the placenta.
b Mann-Whitney U-test adjusted for mother’s age for all the characteristics examined.
c Chi-square test p-value.
Table 2 Hormone concentrations by median of dioxins/PCB TEq (pg/g lipid, mean ± SD) level in the female and male babies.
Female
Male
Hormone Low (< 15.1; n = 31) High (≥ 15.1; n = 31) p-Valuea Total Low (< 15.2; n = 28) High (≥ 15.2; n = 29) p-Valuea Total
T3 (nmol/L) 0.79 ± 0.18 0.91 ± 0.28 0.05# 0.85 ± 0.24 0.96 ± 0.32 0.88 ± 0.24 0.34 0.92 ± 0.28
T4 (nmol/L) 105.8 ± 22.7 115.7 ± 19.6 0.07# 110.8 ± 21.7 112.9 ± 28.8 116.5 ± 21.3 0.59 114.8 ± 25.0
T3 uptake (%) 30.5 ± 4.36 28.9 ± 4.08 0.15 29.7 ± 4.26 28.1 ± 5.22 28.0 ± 3.44 0.94 28.1 ± 4.37
FT4 (pmol/L) 10.7 ± 3.2 10.8 ± 2.3 0.85 10.7 ± 2.8 10.4 ± 2.4 9.93 ± 2.7 0.44 10.2 ± 2.6
TSH (mU/L) 9.23 ± 5.7 6.86 ± 5.3 0.10# 8.03 ± 5.6 8.46 ± 6.5 6.99 ± 5.5 0.38 7.67 ± 6.0
TBG (mg/L) 21.9 ± 3.95 24.6 ± 4.82 0.03* 23.2 ± 4.57 25.0 ± 7.21 25.1 ± 4.47 0.94 25.0 ± 5.86
IGF-1 (ng/dL) 85.2 ± 38.6 99.1 ± 36.0 0.15 95.2 ± 37.7 82.3 ± 35.3 86.0 ± 36.6 0.70 84.3 ± 34.7
BP3 (ng/dL) 1.23 ± 0.44 1.64 ± 0.84 0.02* 1.43 ± 0.70 1.54 ± 0.86 1.38 ± 0.87 0.51 1.46 ± 0.86
a Student t-test or Mann-Whitney U-test when data distribution was significantly beyond the normal distribution range:
* p < 0.05,
# p < 0.1.
Table 3 Correlationa between thyroid hormone concentrations in cord serum and birth-related indices.
Hormones Maternal age (years) Gestational age (weeks) Placenta weight (g) Birth weight (g) Birth length (cm) Baby head girth (cm) QI (kg/m2) 1st min Apgar score 5th min Apgar score Baby bilirubin (mg/dL)
Maternal age 1 −0.271** −0.004 −0.018 0.040 0.023 −0.103 −0.191* −0.025* −0.054
Gestational age −0.271** 1 0.099 0.308*** 0.136 0.169# 0.355*** 0.210* 0.208* −0.015
T3 (nmol/L) −0.069 0.218 −0.157 0.046 −0.006 0.044 0.105 −0.036 −0.042 0.004
T4 (nmol/L) −0.105 0.137 0.020 −0.029 −0.120 −0.075 0.129 0.086 0.044 −0.070
TSH (μU/L) −0.216 0.148 0.288** 0.155# −0.039 0.055 0.205* 0.054 0.066 0.018
T3 uptake (%) 0.025 −0.244** 0.178# 0.022 0.074 0.106 −0.084 −0.049 −0.032 −0.030
FT4 (pmol/L) −0.042 −0.005 0.080 0.009 −0.140 −0.055 0.171# 0.042 0.115 −0.085
TBG (mg/L) −0.063 0.213* 0.051 0.009 −0.107 0.001 0.154 0.120 0.085 −0.062
IGF-1 (nmol/L) 0.069 −0.164# 0.279** 0.309*** 0.112 0.321*** 0.267** −0.139 −0.062 −0.169#
BP3 (nmol/L) −0.086 0.023 0.237* 0.218* 0.050 0.300** 0.234* −0.078 −0.015 −0.119
FT4 × TSH −0.224* 0.148 0.237* 0.117 −0.095 0.016 0.226* 0.081 0.154 −0.039
T4 × TSH −0.255** 0.168# 0.281** 0.118 −0.089 0.002 0.229** 0.085 0.097 0.006
T4:TBG −0.031 −0.007 0.004 −0.064 −0.088 −0.125 −0.009 0.008 −0.044 0.015
T3 uptake:T3 0.059 −0.248** 0.177# −0.011 0.066 0.027 −0.116 0.011 0.040 −0.008
FT4:T3 0.023 −0.213* 0.174# −0.054 −0.091 −0.044 0.001 −0.012 0.031 −0.079
a Spearman correlation:
*** p < 0.001,
** p < 0.01,
* p < 0.05,
# p < 0.1.
Table 4 Correlationa between PCDD/Fs and PCB levels in toxic equivalence and thyroid hormone concentrations in cord serum by infant’s sex.
Hormone Infant sex Mother’s age PCDDs PCDFs Non-ortho PCBs Mono-ortho PCBs PCB-138, -153, -180
T3 (nmol/L) F 0.097 0.214 0.300* 0.216 0.227# 0.074
M −0.220 −0.090 −0.039 −0.008 0.070 0.035
T4 (nmol/L) F 0.066 0.191 0.253# 0.189 0.191 0.282#
M −0.236# 0.116 0.231# −0.137 0.172 0.183
TSH (μU/L) F −0.223# −0.265* −0.233# −0.320* −0.264* −0.154
M −0.225 −0.214 0.015 −0.130 −0.239# −0.078
T3 uptake (%) F −0.059 −0.177 −0.172 0.074 0.024 0.019
M 0.057 0.007 −0.048 −0.003 −0.237# −0.179
FT4 (pmol/L) F 0.062 −0.004 −0.017 0.058 0.022 0.247#
M −0.172 −0.182 0.058 −0.302* −0.178 −0.033
TBG (mg/L) F 0.121 0.319* 0.368** 0.118 0.189 0.136
M −0.176 −0.039 0.118 −0.141 0.107 0.054
IGF-1 (nmol/L) F 0.175 0.073 0.053 −0.035 −0.110 −0.196
M −0.084 −0.045 −0.006 −0.179 −0.061 −0.018
BP3 (nmol/L) F 0.161 0.185 0.254# −0.097 −0.061 −0.215
M −0.329* −0.190 −0.141 −0.204 −0.253 −0.208
FT4 × TSH F −0.239# −0.218 −0.187 −0.294* −0.243# −0.210
M −0.249 −0.210 0.047 −0.291 −0.276 −0.035
T4:TBG F −0.067 −0.305* −0.270# 0.046 −0.068 0.082
M −0.002 0.082 0.049 0.057 −0.002 0.110
T3 uptake:T3 F −0.073 −0.224# −0.282* −0.144 −0.152 < 0.001
M 0.146 0.021 −0.023 0.014 −0.156 −0.151
FT4:T3 F −0.063 −0.188 −0.242# −0.134 −0.162 0.131
M 0.088 −0.059 0.023 −0.188 −0.176 −0.040
a Spearman correlation:
** p < 0.01,
* p < 0.05,
# p < 0.1.
Table 5 Correlation coefficientsa between PCDD/F and PCB levels and thyroid hormone concentrations in cord serum by stepwise multivariate linear regression.
Hormone Mother’s age PCDD/Fs Non-ortho PCBs
T4 −0.274** 0.202# —
TBG −0.201# 0.286* —
FT4 — — −0.277**
FT4 × TSH −0.218* — −0.246*
a The hormones levels were log-transformed for normality to use the parametric method.
** p < 0.01,
* p < 0.05,
# p < 0.1.
==== Refs
References
Alaluusua S Lukinmaa P-L Torppa J Tuomisto J Vartiainen T 1999 Developing teeth as biomarker of dioxin exposure [Letter] Lancet 353 206 9923879
Brouwer A Longnecker MP Birnbaum LS Cogliano J Kostyniak P Moore J 1999 Characterization of potential endocrine-related health effects at low-dose levels of exposure to PCBs Environ Health Perspect 107 suppl 4 639 649 10421775
Chao HR Wang SL Lee CC Yu HY Lu YK Päpke O 2004 Level of polychlorinated dibenzo-p -dioxins, dibenzofurans and biphenyls (PCDD/Fs, PCBs) in human milk and the input to infant body burden Food Chem Toxicol 42 1299 1308 15207381
Chen HL Lee CC Liao PC Guo YL Chen CH Su HJ 2003 Associations between dietary intake and serum polychlorinated dibenzo-p -dioxin and dibenzofuran (PCDD/F) levels in Taiwanese Environ Res 91 172 178 12648480
Chen YC Guo YL Hsu CC Rogan WJ 1992 Cognitive development of Yu-Cheng (“oil disease”) children prenatally exposed to heat-degraded PCBs JAMA 268 3213 3218 1433761
Fisher DA Brown RS 2000. Thyroid physiology in the perinatal period and during childhood. In: Werner & Ingbar’s The Thyroid: A Fundamental and Clinical Text (Utiger R-D, ed). Vol 81. Philadelphia:Lippincott Williams & Wilkins, 959–972.
Guo YL Hsu PC Hsu CC Lambert GH 2000 Semen quality after prenatal exposure to polychlorinated biphenyls and dibenzofurans Lancet 356 1240 1241 11072947
Jacobson JL Jacobson SW 1996 Intellectual impairment in children exposed to polychlorinated biphenyls in utero N Engl J Med 335 783 789 8703183
Kimbrough RD Krouskas CA 2001 Polychlorinated biphenyls, dibenzo-p -dioxins, and dibenzofurans and birth weight and immune and thyroid function in children Regul Toxicol Pharmacol 34 42 52 11502155
Koopman-Esseboom C Morse D-C Weisglas-Kuperus N Lutkeschipholt I-J Van der Paauw C-G Tuinstra L-G 1994 Effects of dioxins and polychlorinated biphenyls on thyroid hormone status of pregnant women and their infants Pediatr Res 36 468 473 7816522
LaFranchi S 2004. Disorders of the thyroid gland. Nelson Textbook of Pediatrics (Behrman RE, Kliegman RM, Jenson HB, eds) 17th Ed. Philadelphia:W.B. Saunders, 1870–1890.
Leatherland J-F 1999. The effects on the immune system, as well as the effects on the liver and thyroid. In: Environmental Endocrine Disruptors: An Evolutionary Perspective (Guillette LJ, Crain DA. eds). London:Taylor & Francis, 248–289.
Mashang T Thornton PS 1999. Endocrine disorders of the newborn. In: Neonatology: Pathophysiology and Management of the Newborn (Avery GB, Fletcher MA, MacDonald MG, eds). Philadelphia:Lippincott Williams & Wilkins, 859–886.
Matsuura N Uchiyama T Tada H Nakamura Y Kondo N Morita M 2001 Effects of dioxins and polychlorinated biphenyls (PCBs) on thyroid function in infants born in Japan—the second report from research on environmental health Chemosphere 45 1167 1171 11695630
Nagayama J Okamura K Iida T Hirakawa H Matsueda T Tsuji H 1998 Postnatal exposure to chlorinated dioxins and related chemicals on thyroid hormone status in Japanese breast-fed infants Chemosphere 37 1789 1793 9828307
Ness DK Schantz SL Moshtaghian J Hansen LG 1993 Effects of perinatal exposure to specific PCB congeners on thyroid hormone concentrations and thyroid histology in the rat Toxicol Lett 68 311 323 8516785
Nishimura N Yonemoto J Miyabara Y Sato M Tohyama C 2003 Rat thyroid hyperplasia induced by gestational and lactational exposure to 2,3,7,8-tetrachlorodibenzo-p -dioxin Endocrinology 144 2075 2083 12697716
Päpke O 1998 PCDD/PCDF: human background data for Germany, a 10-year experience Environ Health Perspect 106 suppl 2 723 731 9599723
Päpke O Ball M Lis A 1994 PCDD/PCDF in humans, a 1993-update of background data Chemosphere 29 2355 2360 7850384
Patandin S Lanting CI Mulder PG Boersma ER Sauer PJ Weisglas-Kuperus N 1999 Effects of environmental exposure to polychlorinated biphenyls and dioxins on cognitive abilities in Dutch children at 42 months of age J Pediatr 134 33 41 9880446
Persky V Turyk M Anderson H-A Hanrahan L-P Falk C Steenport D-N 2001 The effects of PCB exposure and fish consumption on endogenous hormones Environ Health Perspect 109 1275 1283 11748036
Pluim H-J de Vijlder J-J Olie K Kok J-H Vulsma T van Tijn D-A 1993 Effects of pre- and postnatal exposure to chlorinated dioxins and furans on human neonatal thyroid hormone concentrations Environ Health Perspect 101 504 508 8137779
Pluim H-J Koppe J-G Olie K Vd Slikke J-W Kok J-H Vulsma T 1992 Effects of dioxins on thyroid function in newborn babies [Letter] Lancet 339 1303 1349705
Polk DH Fisher DA 1998. Fetal and neonatal endocrinology. In: Endocrinology (Hadley ME, ed). Englewood Cliffs, NJ:Prentice Hall, 2239–2257.
Porterfield SP 2001. Thyroid gland. In: Endocrine Physiology (Porterfield SP, ed). London:Mosby, 59–84.
Rideout K Teschke K 2004 Potential for increased human food-borne exposure to PCDD/F when recycling sewage sludge on agricultural land Environ Health Perspect 112 959 969 15198915
Rogan WJ Miller RW 1989 Prenatal exposure to polychlorinated biphenyls [Letter] Lancet 2 1216 2572923
Rosiak KL Seo BW Chu I Francis BM 1997 Effects of maternal exposure to chlorinated diphenyl ethers on thyroid hormone concentrations in maternal and juvenile rats J Environ Sci Health B 32 377 393 9177011
Scanlon MF Hall R 1998. Thyrotropin-releasing hormone: basic and clinical aspects. In: Williams Textbook of Endocrinology (Larsen P-R, ed). Philadelphia:W.B. Saunders Company, 192.
Schecter AJ Piskac AL 2001. PCBs, dioxins, and dibenzofurans: measured levels and toxic equivalents in blood, milk and food from various countries. In: PCBs: Recent Advances in Environmental Toxicology and Health Effects (Hansen LG, ed). Lexington, KY:University Press of Kentucky, 161–168.
Smith AG Gangolli SD 2002 Organochlorine chemicals in seafood: occurrence and health concerns Food Chem Toxicol 40 767 779 11983271
Tango T Fujita T Tanihata T Minowa M Doi Y Kato N 2004 Risk of adverse reproductive outcomes associated with proximity to municipal solid waste incinerators with high dioxin emission levels in Japan J Epidemiol 14 83 93 15242064
Utiger RD 1999 Maternal hypothyroidism and fetal development N Engl J Med 341 601 602 10451467
Van den Berg M Birnbaum L Bosveld AT Brunstrom B Cook P Feeley M 1998 Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife Environ Health Perspect 106 775 792 9831538
Wang SL Chen TT Hsu JF Hsu CC Chang LW Ryan JJ 2003 Neonatal and childhood teeth in relation to perinatal exposure to polychlorinated biphenyls and dibenzofurans: observations of the Yucheng children in Taiwan Environ Res 93 131 137 12963397
Wang SL Lin CY Guo YL Lin LY Chou WL Chang LW 2004 Infant exposure to polychlorinated dibenzo-p -dioxins, dibenzofurans and biphenyls (PCDD/Fs, PCBs)—correlation between prenatal and postnatal exposure Chemosphere 54 1459 1473 14659948
World Medical Association 2000 Declaration of Helsinki: Ethical principles for medical research involving human subjects Bull Med Eth 162 8 11
|
16263525
|
PMC1310932
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 27; 113(11):1645-1650
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.7994
|
oa_comm
|
==== Front
Environ Health PerspectEnviron. Health PerspectEnvironmental Health Perspectives0091-67651552-9924National Institute of Environmental Health Sciences 10.1289/ehp.8022ehp0113-00165116263526ResearchChildren's HealthBiologic Monitoring to Characterize Organophosphorus Pesticide Exposure among Children and Workers: An Analysis of Recent Studies in Washington State Fenske Richard A. 1Lu Chensheng 2Curl Cynthia L. 3Shirai Jeffry H. 1Kissel John C. 11 Department of Environmental and Occupational Health Sciences, School of Public Health and Community Medicine, University of Washington, Seattle, Washington, USA2 Department of Environmental and Occupational Health, Rollins School of Public Health, Emory University, Atlanta, Georgia, USA3 Integral Consulting Inc., Boulder, Colorado, USAAddress correspondence to R.A. Fenske, University of Washington, Box 357234, Health Sciences Building, F-233, 1959 N.E. Pacific St., Seattle, WA 98195-7234 USA. Telephone: (206) 543-0916. Fax: (206) 616-2687. E-mail:
[email protected] authors declare they have no competing financial interests.
11 2005 6 7 2005 113 11 1651 1657 16 2 2005 6 7 2005 2005Publication of EHP lies in the public domain and is therefore without copyright. All text from EHP may be reprinted freely. Use of materials published in EHP should be acknowledged (for example, ?Reproduced with permission from Environmental Health Perspectives?); pertinent reference information should be provided for the article from which the material was reproduced. Articles from EHP, especially the News section, may contain photographs or illustrations copyrighted by other commercial organizations or individuals that may not be used without obtaining prior approval from the holder of the copyright. We examined findings from five organophosphorus pesticide biomonitoring studies conducted in Washington State between 1994 and 1999. We compared urinary dimethylthiophosphate (DMTP) concentrations for all study groups and composite dimethyl alkylphosphate (DMAP) concentrations for selected groups. Children of pesticide applicators had substantially higher metabolite levels than did Seattle children and farmworker children (median DMTP, 25 μg/L; p < 0.0001). Metabolite levels of children living in agricultural communities were elevated during periods of crop spraying. Median DMTP concentrations for Seattle children and farmworker children did not differ significantly (6.1 and 5.8 μg/L DMTP, respectively; p = 0.73); however, the DMAP concentrations were higher for Seattle children than for farmworker children (117 and 87 nmol/L DMAP, respectively; p = 0.007). DMTP concentrations of U.S. children 6–11 years of age (1999–2000 National Health and Nutrition Examination Survey population) were higher than those of Seattle children and farmworker children at the 75th, 90th, and 95th percentiles. DMTP concentrations for workers actively engaged in apple thinning were 50 times higher than DMTP concentrations for farmworkers sampled outside of peak exposure periods. We conclude that workers who have direct contact with pesticides should continue to be the focus of public health interventions and that elevated child exposures in agricultural communities may occur during active crop-spraying periods and from living with a pesticide applicator. Timing of sample collection is critical for the proper interpretation of pesticide biomarkers excreted relatively soon after exposure. We surmise that differences in dietary exposure can explain the similar exposures observed among farmworker children, children living in the Seattle metropolitan area, and children sampled nationally.
agricultural communitiesagricultural workersbiologic monitoringchildrendialkylphosphatesorganophosphorus pesticidespesticide exposure
==== Body
Children may experience greater risks from pesticide exposures than adults because of behavioral, dietary, and physiologic characteristics associated with development (National Research Council 1993). University of Washington researchers began an investigation of children’s exposure to pesticides in 1991, with particular emphasis on organophosphorus (OP) pesticide exposures of presumed high-risk populations, such as children of pesticide applicators, children of farmworkers, and children living in agricultural regions with substantial agricultural pesticide use (Curl et al. 2002; Fenske et al. 2002; Koch et al. 2002; Loewenherz et al. 1997; Lu et al. 2000, 2001; Simcox et al. 1999). These studies have suggested that children with parents who apply pesticides in agriculture and who live in agricultural areas during active crop spraying receive higher OP pesticide exposure than do other children.
Each of these studies has employed biologic monitoring of urinary dialkylphosphate (DAP) metabolites to yield information on OP pesticide exposure. Biologic monitoring is a valuable tool in exposure assessment, allowing for integrated measurement of exposure from all pathways and routes. Biologic monitoring has been used effectively to evaluate exposures to populations across studies and across time (Barr et al. 2004a; National Research Council 1991). This approach has been used in many regional studies to determine OP pesticide exposures among young children (Adgate et al. 2001; Aprea et al. 2000; Heudorf and Angerer 2001; Heudorf et al. 2004; Mills and Zahm 2001; Morgan et al. 2005; Royster et al. 2002; Shalat et al. 2003).
More than 30 OP pesticides are registered for use in Washington State, and many of these pesticides do not have urinary metabolites that can be considered selective. When we began the Washington State biologic monitoring studies in 1994, selective metabolites were not available for azinphosmethyl and phosmet, the primary OP pesticides used in the region. We therefore developed an assay for the DAP metabolites (Moate et al. 1999). The DAP metabolites are the common products of OP pesticide metabolism and integrate exposure from most registered OP pesticides (Barr et al. 2004a). The National Center for Environmental Health now includes the DAPs among the metabolites assayed as part of the National Health and Nutrition Examination Survey (NHANES). Their most recent reports present DAP levels for participants 6–59 years of age and thus provides information on current levels in the general U.S. population [Centers for Disease Control and Prevention (CDC) 2003, 2005]. One concern in the use of exposure biomarkers is the possibility that the compounds being assayed could appear in food or the environment as degradation products (Lu et al. 2005; Morgan et al. 2005). As with any breakdown products that are urinary biomarkers of toxicant exposure, if exposure to the breakdown product occurs, and if this compound is absorbed efficiently into the body and excreted in the urine unchanged, then its appearance in urine samples could confound interpretation of such measurements. To date, however, no published studies have demonstrated that DAPs behave in this fashion.
The purpose of the present analysis was to examine OP pesticide metabolite concentrations in five Washington State studies, conducting both qualitative and quantitative (statistical) comparisons, with special attention to issues of study design and sampling that make such comparisons problematic. The five studies include apple thinners exposed to OP pesticides when reentering fields after applications (Simcox et al. 1999); children of agricultural pesticide applicators, many of whom lived near pesticide-treated farmland (Loewenherz et al. 1997; Lu et al. 2000); children living in the Seattle metropolitan area whose parents were not occupationally exposed to pesticides (Lu et al. 2001); children living in an agricultural community whose parents were not involved significantly in agricultural production (Koch et al. 2002); and children living in households with adults employed as farmworkers in a variety of agricultural activities (Curl et al. 2002; Thompson et al. 2003). The aforementioned biologic monitoring data from the CDC provide an opportunity to draw comparisons between the study populations and the general U.S. population (Barr et al. 2004a; CDC 2003). It is our hope that this analysis will prove useful in the development of future study designs and sampling plans for population-based pesticide exposure studies.
Materials and Methods
We used seven populations enrolled in five independent studies, all of which took place in Washington State between 1994 and 1999, in this analysis. Our previous reports included detailed descriptions of population recruitment, sample collection, and sample analysis (Curl et al. 2002; Koch et al. 2002; Loewenherz et al. 1997; Lu et al. 2000, 2001; Simcox et al. 1999; Thompson et al. 2003). Tables 1 and 2 provide geographic location, number and age of participants, and number of samples collected in these studies and describe relevant occupational or para-occupational factors, as well as relationships between sampling time and active crop spraying with OP pesticides.
Study designs.
Three studies included in this analysis were cross-sectional in design. The first was conducted in Douglas and Chelan Counties in the Wenatchee Valley region of central Washington State. The primary industry in this region is tree fruit production, and the area includes many small family orchards. Forty-nine families of pesticide applicators participated in this study, including 72 children between the ages of 2 and 6 years (applicator children). Each child provided two individual urinary voids separated by 3–7 days. Sample collection occurred between May and July of 1995, during the time that dimethyl OP pesticides were applied to control the codling moth (Figure 1). A complete discussion of the methods for this study is provided by Loewenherz et al. (1997) and Lu et al. (2000).
The second cross-sectional study occurred in the Seattle metropolitan area (King and Snohomish counties). Participants included 110 children 2–5 years of age from 96 families. Parents of children enrolled in this study were not occupationally exposed to pesticides. Participating families were originally separated into one of two groups based on socioeconomic status, but no difference was observed in OP pesticide metabolite levels based on community or family income (Lu et al. 2001). These children were considered a single group (Seattle children) in the present analysis. Each child provided one sample in the spring of 1998 (May–June) and another in the fall of 1998 (September–November), as indicated in Figure 1. Methods for this study are described by Lu et al. (2001).
The third cross-sectional study was designed as a baseline survey of communities enrolled in a multiyear community intervention project. It included 24 communities in the Yakima Valley region of Washington State. This area, like the Wenatchee Valley, is a very productive agricultural region that includes tree fruit production. Dimethyl OP pesticides are applied in the late spring to protect against the codling moth, and other OP pesticides are used as a part of agricultural production. Two hundred eighteen households including both an adult farmworker and a child 2–6 years of age were enrolled in this study. The farmworkers were engaged in a variety of work tasks, such as harvesting, pruning, planting, weeding, and irrigating. About 20% reported some involvement in pesticide mixing, loading, or application. Urine samples were provided by 213 adults (farmworkers) involved in agricultural tasks such as harvesting, weeding, and pesticide applications, as well as by 211 farmworker children. Urine samples for farmworkers and farmworker children consisted of a composite of either two or three voids, each separated by a minimum of 3 days, and all collected within a 2-week period. These samples were collected between July and October of 1999 (Figure 1). Methods for this study are described by Curl et al. (2002) and Thompson et al. (2003).
The two remaining studies included in this analysis focused on relatively small populations but sampled participants repeatedly over time. Both were conducted in the Wenatchee region of Washington State. The first study investigated dimethyl OP pesticide exposures in a group of 20 adult apple thinners; sampling took place between May and July of 1994. These workers entered treated fields soon after OP pesticide applications (Figure 1). Each worker provided between 7 and 21 individual voids; a total of 293 voids from these workers were included in this analysis. Methods for this study are described by Simcox et al. (1999).
The last study included in the analysis was a longitudinal study of OP pesticide exposures among preschool children living in an agricultural community. A group of 44 children 2–5 years of age provided biweekly urine samples for up to 1 year. Sampling for this study was conducted between December 1997 and August 1999. This encompassed two periods of active crop spraying (Figure 1). Koch et al. (2002) reported that dimethyl OP pesticide metabolite levels were significantly elevated in these children during the periods of active crop spraying. Therefore, samples collected during periods of active crop spraying (farm community children, spray season) were considered separately from those collected during other times of the year (farm community children, nonspray season). This analysis included 274 samples collected from the 44 children during the spray season and 694 samples collected during the nonspray season. Methods for this study are described by Koch et al. (2002).
Sample analysis.
Urine samples collected in all studies were analyzed for DAP metabolites common to most OP pesticides. We selected dimethyl DAP metabolites as the focus for comparisons across studies because they were substantially and consistently higher than diethyl metabolites in all studies. Dimethylthiophosphate (DMTP) was selected as the best comparative indicator of exposure across all studies. DMTP has been found to be the dominant of the three dimethyl DAP metabolites in samples analyzed in all of our studies (Curl et al. 2002, 2003; Koch et al. 2002; Lu et al. 2000, 2001; Simcox et al. 1999), as well as in the 1999–2000 NHANES population (NHANES III; CDC 2003).
Sample analysis for all of the studies was conducted by the University of Washington Environmental Health Laboratory following the procedure described by Moate et al. (1999). This procedure involved solid-phase extraction, azeotropic distillation, and derivatization with pentafluoro(methyl)benzylbromide, followed by gas chromatographic detection. That all samples were analyzed within the same laboratory provides reassurance that urinary metabolite levels can be compared directly, because results of such assays have been shown to vary across laboratories (James et al. 2003).
Samples with metabolite concentrations below the limit of detection (LOD) were assigned one-half the value of the LOD for this analysis. The LOD in the later studies was lower than for the 1994 study of apple thinners and the 1995 study of applicator families (e.g., 40 μg/L LOD for DMTP in 1994, 20 μg/L in 1995, 1.1 μg/L in 1998 and 1999). To reduce the relative impact of this higher detection limit, we assumed the lower detection limits for all studies. This assumption lowered the estimates of exposure for the apple thinner and applicator children populations relative to the other populations. Comparison of the data below the 25th percentile is problematic because of these differences in detection limits.
Data analysis.
In the 1995 study of applicator children, DMTP levels measured in the two samples (collected 3–7 days apart) were averaged to yield one value per child. This is essentially equivalent to the procedure employed in the 1999 farmworker and farmworker children study, where equal volumes of two or three samples (collected within 2 weeks) were pooled and the resultant composite sample was analyzed to yield one value per child. Two samples were also collected per child in the 1998 Seattle children study; the first in the spring and the second several months later (Lu et al. 2001) showed no difference in DAP metabolite levels related to season of sample collection, so these samples were averaged to yield one value per child. In a few cases for each of these studies, only one sample was provided or analyzed per child. In these instances, that sample was assumed to provide the best available estimate of the child’s exposure. It seems unlikely that these sampling differences could produce substantial differences for the analyses conducted.
Distributions of urinary DMTP concentrations (micrograms of metabolite per liter of urine) were created for the populations sampled in the three cross-sectional studies. These distributions were compared at various percentiles, and cumulative frequency distributions were created to describe the data. The metabolite levels for two of the populations were not normally distributed; therefore, nonparametric tests including the Wilcoxon matched-pairs signed-rank test for paired samples and the Mann-Whitney U-test for independent samples were used to determine significant differences between groups. All analyses were performed using the statistical package Stata 6.0 (StataCorp, College Station, TX) or SPSS 10.0.5 (SPSS Inc., Chicago, IL).
The three dimethyl DAP concentrations (dimethylphosphate, DMTP, and dimethyl-dithiophosphate) for each sample were converted to their molar equivalents and summed to produce a composite dimethyl alkylphosphate (DMAP) value (nanomoles of metabolite per liter of urine) for the Seattle children and the farmworker children to allow a more thorough comparison of these two populations. The Mann-Whitney U-test for independent samples was used to compare these groups.
Differences in sampling strategies between these three studies and the two repeated-measures studies precluded direct statistical comparisons across all groups. To assess the exposure of the apple thinner and farm community children study populations, we calculated the arithmetic mean DMTP level for each participant. Distributions of these mean concentrations were created and compared at various percentiles.
Creatinine was measured in most of these studies, but we chose not to adjust the values because of concerns regarding the validity of such adjustments, particularly for children (Barr et al. 2005).
Results
The five studies examined in this analysis included 437 children and 233 adults, who provided > 2,000 urine samples. Maximum values and selected percentiles for DMTP levels in the urine of the adult farmworkers, applicator children, farmworker children, and Seattle children in the cross-sectional studies are presented in Table 3. At the median (50th percentile), DMTP concentrations were highest in children of applicators (25 μg/L), followed by adult farmworkers (10 μg/L), and then children living in the Seattle area and children of farmworkers (6.1 and 5.8 μg/L, respectively). This pattern continued at the 75th percentile, with applicator children higher than adult farmworkers (44 vs. 32 μg/L), and Seattle children higher than farmworker children (17 vs. 13 μg/L). Even at the 90th percentile, the DMTP concentrations for applicator children were greater than for adult farmworkers (110 vs. 99 μg/L) and the other two groups. At the 95th percentile, however, this trend changed, with adult farmworkers having the highest concentrations (180 μg/L), followed by applicator children (130 μg/L), farmworker children (50 μg/L), and finally Seattle children (39 μg/L). The cumulative distributions for the top 50th percentile of these four populations are presented in Figure 2.
Metabolite concentrations in applicator children were higher than those in adult farmworkers (p = 0.02), although this difference was not statistically significant when correction for multiple comparisons was included in the analysis. DMTP concentrations in the urine of the applicator children were higher than those of either the farmworker children (Mann-Whitney U-test, p < 0.0001) or the Seattle children (Mann-Whitney U-test, p < 0.0001). Adult farmworker concentrations were higher than those of either the farmworker children (Wilcoxon matched pairs, p < 0.0001) or the Seattle children (Mann-Whitney U-test, p = 0.002). DMTP concentrations in farmworker children were not significantly different from those in Seattle children (p = 0.73). This information is summarized in Table 4.
We compared DMTP levels in these four populations qualitatively with DMTP levels in the NHANES III population survey (CDC 2003) for children 6–11 years of age (the youngest population sampled) and Mexican Americans (most of the adult farmworkers in the 1999 cross-sectional study were Hispanic). Table 3 presents DMTP levels for the 25th, 50th, 75th, 90th, and 95th percentiles. Applicator children had higher levels than did the NHANES III children at all percentiles. Farmworker children and Seattle children had higher DMTP concentrations at the 50th percentile than did the NHANES III children, but not at the higher percentiles. DMTP concentrations for the adult farmworkers were consistently higher than the NHANES III Mexican-American subgroup.
Table 5 presents the distributions for composite DMAP concentrations from the Seattle children and the farmworker children. DMAP concentrations were significantly higher for the Seattle children compared with the farmworker children (median values of 117 nmol/L and 87 nmol/L, respectively; p = 0.007). At the 50th percentile, DMAP concentrations were similar for the NHANES III children (6–11 years of age) and the farmworker children. At the 75th and 90th percentiles, the NHANES III values were similar to the Seattle children values and substantially higher than the farmworker children values. At the 95th percentile, the NHANES III values were substantially higher than either the Seattle children or farmworker children values.
Distributions of DMTP levels in the repeated-measures studies are presented in Table 6, which provides the 25th, 50th, 75th and 95th percentiles of the arithmetic mean urinary DMTP concentrations for the apple thinners, the farm community children (spray season), and the farm community children (nonspray season). DMTP concentrations among the apple thinners, who worked in fields soon after crop spraying, were one to two orders of magnitude greater than those of the farm community children. The 50th percentile DMTP concentration for the apple thinners (530 μg/L; Table 5) was > 50 times higher than that of the adult farmworkers (10 μg/L; Table 2). As reported by Koch et al. (2002), metabolite concentrations were significantly higher for the farm community children during the spray season than during the nonspray season.
Figure 2 includes data points describing the 75th and 95th percentiles of the arithmetic mean urinary DMTP concentrations for the farm community children during both the spray and nonspray seasons. Metabolite levels for the apple thinners were beyond the scale of this figure. Figure 2 demonstrates that, at the 75th percentile, all children except for the applicator children had roughly similar metabolite levels. At the 95th percentile, the farm community children (spray season) demonstrated higher levels than did the Seattle children, the farmworker children, and the farm community children (nonspray season). The maximum value for the farm community children (spray season) also exceeded the maximum value for the farmworker children [180 μg/L (Table 6) and 140 μg/L (Table 3), respectively].
Discussion
The two most striking findings from this analysis were the relatively high levels of metabolites among applicator children compared with the other study groups and the similarity in metabolite levels between farmworker children sampled in 1999 and Seattle metropolitan area children sampled in 1998.
DMTP levels in children of pesticide applicators sampled in 1995 were significantly higher than those in the farmworker children in 1999, and were also higher than those in farmworkers in 1999 up to the 90th percentile. Direct comparison of these populations is problematic because pest control strategies may have changed over the 4 years that separated these studies. Pesticide use practices in this region are geared to many factors, including crop type, weather, pest infestation levels, and adoption of less chemical-intensive integrated pest management techniques. Additionally, the 1995 study sampled children during the active crop spraying season, whereas the 1999 sampling occurred for the most part after the peak spraying period.
One way to gauge changes in pesticide use that is less affected by the time of sampling is to examine pesticide concentrations in household dust in these studies. Such an approach is based on the observation that pesticide concentrations in house dust are less susceptible to short-term fluctuations than are urinary metabolite measurements, and on the assumption that changes in pesticide levels in dust reflect changes in pesticide use practices and therefore differences in exposure opportunity. In a previous study, we noted that ethyl parathion concentrations in house dust decreased dramatically after this compound was withdrawn from agricultural use (Fenske et al. 2002). Figure 3 provides the median house dust concentrations of azinphos-methyl and phosmet—the primary dimethyl OP pesticides used in regional tree fruit production—from studies conducted in 1992 (Simcox et al. 1995), 1995 (Lu et al. 2000), and 1999 (Curl et al. 2002). These concentrations decreased over time: Azinphosmethyl concentrations in 1999 were about half those measured in 1995, and phosmet concentrations decreased even more dramatically. Other factors that might explain this difference include relatively high parental exposures during pesticide handling with consequent para-occupational exposure for the children, and the close proximity of the homes of many applicators to pesticide-treated farmland.
A second striking finding from this analysis was the lack of a significant difference between DMTP levels in farmworker children and Seattle children (Table 3) and the significantly higher DMAP concentrations among Seattle children compared with those in the farmworker children (Table 5). We had anticipated that the farmworker children would exhibit higher metabolite concentrations, given the widespread use of dimethyl OP pesticides in the Yakima Valley where they resided. The Seattle population consisted of 2- to 5-year-old healthy children living in the Seattle metropolitan area who had no known risk factors for pesticide exposure; less than half of the parents of these children reported use of any pesticides on lawns, gardens, indoors, or pets (Lu et al. 2001). The farmworker children were of a similar age, resided in the lower Yakima Valley about 150 miles east of Seattle, and were also considered a healthy population; all farmworker children lived with at least one adult actively engaged in farm labor (Thompson et al. 2003). It appears that most of these farmworker children had dimethyl OP pesticide metabolite concentrations lower than or indistinguishable from those of children living in urban and suburban environments whose parents did not work with pesticides. We have also examined the DMTP and DMAP metabolite concentrations of child subgroups in the Yakima Valley study based on the agricultural task of the adult farmworker (e.g., pesticide application, crop thinning) and did not find differences across these subgroups (Fenske et al. 2004). That the farmworker children did not have unusually high metabolite levels is further confirmed by comparison of these data with the 1999–2000 NHANES data for children 6–11 years of age (Tables 3, 5). Data comparisons across laboratories can be problematic, but in this case the CDC laboratory that analyzed the NHANES samples and the University of Washington laboratory that analyzed the present results had participated in a round-robin test for the DAP compounds, and these labs were found to produce comparable results (James et al. 2003).
DAP metabolite measurements probably reflect exposures that occurred in the previous 3–5 days. Most of the farmworker children’s samples were collected well after dimethyl OP pesticide crop spraying in the region, so the measured urinary metabolite concentrations did not necessarily capture peak exposures for this population that may have occurred in the late spring and early summer. The longitudinal study of children in an agricultural community reviewed here (Koch et al. 2002) provides persuasive evidence that DAP metabolite levels can increase during such spraying periods, and the 1995 study of children of pesticide applicators sampled during the active crop spraying period had higher metabolite levels than did other children in the same community (Loewenherz et al. 1997; Lu et al. 2000).
Our comparison of metabolite levels in farmworker children and Seattle children led us to conduct a dietary exposure study in the Seattle metropolitan area. We examined OP pesticide exposures in preschool children who consumed primarily organic juice and produce and those of children who consumed conventional (nonorganic) foods (Curl et al. 2003). We found that diet appeared to be the primary contributor to OP pesticide exposure among these children: The median DMAP concentration for children consuming organic juice and produce was about five times lower than for children with conventional diets (30 vs. 170 nmol/L, respectively). The median DMAP concentration for the farmworker children in the Yakima Valley study was 87 nmol/L. We speculate that a conventional diet rich with juices and fresh produce may have been more common among the Seattle children compared with the farmworker children. Supporting evidence for the importance of dietary differences was reported recently for a family in Germany (Heudorf et al. 2004). DAP concentrations were relatively low in a father and son compared with the mother’s levels. It was then learned that the mother had a special diet with a very high intake of fresh fruit. Substitution of supermarket fruit with organic (pesticide-free) fruit reduced the mother’s DAP concentration to those of her family members.
A third finding of this analysis was the 50-fold difference in DMTP concentrations between apple thinners and adult farmworkers. Although the apple thinner study was conducted in 1994, the metabolite levels measured at the time are considered representative of 1999 exposures, based on dislodgeable foliar residues measured at the time, allowable application rates, and U.S. Environmental Protection Agency estimates (Fenske et al. 2003). Biologic measurements of farmwork-ers’ pesticide exposure need to be collected during active work periods to capture peak exposures. It is interesting to note, however, that the adult farmworkers in the 1999 study had higher DMTP levels than did the NHANES III Mexican-American population, despite the fact that most farmworker samples were collected after peak exposures. This comparison supports the view that farmwork-ers represent an important subpopulation of Mexican Americans with respect to pesticide exposure.
Several other studies have employed DAP metabolites to estimate OP pesticide exposures among children. DAP concentration data from most of these studies have been compared in a recent publication (Barr et al. 2004a). A 1995 study in central Italy included collection of spot urine samples from 195 school children 6–7 years of age (Aprea et al. 2000). The primary findings of the study were that DAP concentrations were higher if pest control operations had been performed inside or outside the house in the preceding month, and that higher DAP concentrations were found for children than for a comparable adult population. Results were expressed as nanomole per gram of creatinine and so could not be compared directly with the present data. A 1998 study in Frankfurt am Main, Germany, involved collection of spot urine samples from residents in former U.S. Forces housing, including 309 children ≤5 years of age (Heudorf and Angerer 2001; Heudorf et al. 2004). These children had higher DAP concentrations than did adults (Heudorf and Angerer 2001). The median DMTP concentration among these children was 18.8 μg/L (Heudorf et al. 2004), exceeding the median concentrations of all of our study groups except applicator children. The authors concluded that the primary source of OP pesticide exposure in this population was from diet, and that these exposures in children may reach and even exceed the World Health Organization’s acceptable daily intake values (Heudorf et al. 2004).
A 1997 study of farmworker families near Fresno, California, included collection of spot urine samples from 9 children and 18 adults (Mills and Zahm 2001). Most samples did not have detectable levels of the six DAP metabolites. Frequency of detection was higher among children than among adults. The mean DMTP concentration (13 μg/L) for children was similar to the mean value (14 μg/L) found in the study of farmworker children in Washington (Curl et al. 2002).
A 2000 study in an agricultural community near the U.S.–Mexico border in southern Texas included collection of spot urine samples from 41 children (Shalat et al. 2003). Only eight of these samples (19.6%) had detectable levels of DMTP. The authors concluded that wipe samples of children’s hands may serve as a better exposure metric for epidemiologic studies than do house dust samples. Comparison with the Washington State studies was not possible because DMTP detection frequency was low and DMAP concentrations were not reported.
A 2000 study in Imperial County, California, focused on 20 children 11–17 months of age. The study examined the relationship between proximity of homes to treated farmland and DAP concentrations in the urine of the young children living in these homes (Royster et al. 2002). The median DMTP concentration reported was 3.2 μg/L, or about one-half that observed in our studies of farmworker and Seattle children. No significant difference was found between DAP concentrations of children living within 400 m (one-quarter mile) and those living more than 400 m from an agricultural field. It is not clear that the statistical analysis (nonparametric Mann-Whitney U-test) had sufficient power to detect a difference, given the sample sizes of five and nine, respectively. The authors did not compare DAP concentrations in the children living nearest to (< 400 m) and farthest from (> 800 m) farmland. It is interesting to note that the mean DAP concentration in the first group was 4.4 times higher than the mean concentration for the second group (123 vs. 28 μg/g creatinine), indicating that the most highly exposed children lived nearer to farmland. The 1995 study of applicator children in Washington State found that DAP concentrations were higher in children who lived closer to agricultural fields (Loewenherz et al. 1997; Lu et al. 2000), using several distance categories that were < 400 m, and a larger sample size. This study also focused on children more likely to spend time outdoors (2- to 5-year-olds) than did the Imperial County study (1- to 2-year-olds).
Conclusions
This analysis makes evident that measurements of short-lived metabolites in urine show marked variability both within and across different studies. Children of pesticide applicators sampled during the active spraying season exhibited high DAP concentrations relative to other child populations. Children in an agricultural community without pesticide applicators in their households exhibited higher DAP concentrations during the active crop spraying season than during the rest of the year. Children of farmworkers sampled largely outside of the peak spraying season had DAP concentrations similar to or lower than those of children in the Seattle metropolitan area. This analysis highlights the importance of sample timing in biomarker studies of pesticide exposure and suggests that identification of high-exposure subpopulations in urban and rural communities can be challenging. Future studies should be designed as longitudinal investigations with frequent repeated measurements to capture peak exposures and characterize intrapersonal and interpersonal variability. This analysis also highlights the difficulty of designing epidemiologic studies to evaluate potential health effects of pesticide exposure in children. It cannot be assumed that children in agricultural communities have higher exposures than children in other environments, without taking into account crop spraying patterns and parental contact with pesticides. Studies that seek to categorize children’s exposure in these communities will need to sample both peak and nonpeak exposure periods and will need to evaluate multiple exposure pathways.
Contributors to the studies discussed include J. Camp, G. Coronado, I. Islas, C. Loewenherz, G. Kedan, D. Knutson, D. Koch, T. Moate, N. Simcox, and B. Thompson.
This work was supported by the U.S. Environmental Protection Agency (EPA) STAR program (R819186), the National Institute for Occupational Safety and Health Agricultural Centers program (U07/CCU012926), and the EPA/National Institute of Environmental Health Sciences Child Health Center program (R826886/PO1ES09601). Contents are the responsibility of the authors and do not necessarily represent the official view of these agencies.
Figure 1 Timeline for urine sampling. The solid box indicates the agricultural dimethyl OP pesticide spray season for 1994, 1995, and 1998. The dashed box indicates the agricultural dimethyl OP pesticide spray season for 1999.
Figure 2 Cumulative frequency distribution of the top 50th percentile of urinary DMTP concentrations (μg/L) of children of pesticide applicators, children living in the Seattle metropolitan area, children of farmworkers, and adult farmworkers. Diamonds indicate the 75th and 95th percentiles for children living in farming communities during the nonspray season (blue) and the spray season (orange).
Figure 3 Median pesticide concentrations in house dust from households with agricultural workers in Washington State, 1992–1999. Data for 1992 from Simcox et al. (1995); for 1995 from Lu et al. (2000); for 1999 from Curl et al. (2002).
Table 1 Characteristics of populations in five Washington State studies with cross-sectional sampling design.
Location Collection period Sample size (n) Samples collected Age (years) Relation to agricultural production Sampling time frame
Applicator children Wenatchee Valley May–Jul 1995 72 137 2–6 Parent works as a pesticide applicator During active spray season
Seattle metro children Seattle metro area May–Jun 1998
Oct 1998 110 207 2–5 None NA
Farmworker children Yakima Valley Jul–Oct 1999 211 211a 2–6 Household member works as fieldworker of applicator Most samples collected during nonspray season
Adult farmworkers Yakima Valley Jul–Oct 1999 213 213 ≥19 Employed as a field worker or pesticide applicator Most samples collected during nonspray season
NA, not applicable.
a Each of these samples represents a composite of equal volumes of two or three individual voids, each separated by a minimum of 3 days and all collected within a 2-week period.
Table 2 Characteristics of populations in five Washington State studies with repeated-measures sampling design.
Location Collection period Sample size (n) Samples collected Age (years) Relation to agricultural production Sampling time frame
Apple thinners Wenatchee Valley May–Jul 1994 20 293 ≥19 Employed as an apple thinner During active spray season
Farm community children (spray season) Wenatchee Valley May–Jul 1998
Jun–Aug 1999 44 274 2–5 Reside in an agricultural community During active spray season
Farm community children (nonspray season) Wenatchee Valley Dec 1997–Apr 1998
Aug 1998–May 1999 44 694 2–5 Reside in an agricultural community During active spray season
Table 3 Urinary DMTP concentrations (μg/L) for participants from three cross-sectional Washington State studies and NHANES III data for children 6–11 years of age in the general U.S. population.
Percentile
Population 25th 50th 75th 90th 95th 97th Maximum
Adult farmworkers 3.3 10 32 99 180 250 2,000
Applicator children 8.2 25 44 110 130 180 220
Farmworker children 2.4 5.8 13 33 50 57 140
Seattle children 2.4 6.1 17 29 39 42 60
NHANESa children 6–11 years old < LOD 4.1 20 40 62 — —
NHANESa Mexican Americans < LOD 2.0 10 38 130 — —
—, data not reported.
a Reported by CDC (2003).
Table 4 Statistical analysis of differences in urinary DMTP concentrations between populations.
Population with higher exposure
Comparison population Adult farmworkers Applicator children Farmworker children
Applicator children p = 0.02a,b — —
Farmworker children p = 0.0001c p < 0.0001a —
Seattle children p = 0.002a p < 0.0001a No significant differencea,d
a Mann-Whitney U-test for independent samples.
b Applicator children had higher DMTP levels than did adult farmworkers (p = 0.02), but this difference was not considered significant because of multiple comparisons (Bonferroni adjustment; p < 0.008 necessary for significance).
c Wilcoxon signed-rank test for paired samples.
d DMTP levels for Seattle children and farmworker children were not different (p = 0.73).
Table 5 Composite DMAP concentrations (nmol/L) for Seattle metropolitan area children, Yakima Valley farmworker children, and NHANES III children 6–11 years of age.
Percentile
Population No. 25th 50th 75th 90th 95th
Seattle children 110 63 117b 250 453 545
Farmworker children 211 50 87b 174 378 522
NHANESa 471 23 91 270 460 679
a Data from Barr et al. (2004, Table 4).
b Significantly different, Mann-Whitney U-test, p = 0.007.
Table 6 Arithemetic mean urinary DMTP concentrations (μg/L) for participants in repeated-measures studies.
Percentilea
Population 25th 50th 75th 95th Maximum
Apple thinners 310 530 610 1,100 1,100
Farm community children (spray season) 5.5 7.8 14 84 180
Farm community children (nonspray season) 3.8 5.5 9.5 18 45
a Percentiles of arithmetic means of repeated measures for individuals in study population.
==== Refs
References
Adgate JL Barr DB Clayton CA Eberly LE Freeman NC Lioy PJ 2001 Measurement of children’s exposure to pesticides: analysis of urinary metabolite levels in a probability-based sample Environ Health Perspect 109 583 590 11445512
Aprea C Strambi M Novelli MT Lunghini L Bozzi N 2000 Biologic monitoring of exposure to organophosphorus pesticides in 195 Italian children Environ Health Perspect 108 521 525 10856025
Barr DB Bravo R Weerasekera G Caltabiano LM Whitehead RD Jr Olsson AO 2004 Concentrations of dialkyl phosphate metabolites of organophosphorus pesticides in the U.S. population Environ Health Perspect 112 186 200 14754573
Barr DB Wilder LC Caudill SP Gonzalez AJ Needham LL Pirkle JL 2005 Urinary creatinine concentrations in the U.S. population: implications for urinary biological monitoring measurements Environ Health Perspect 113 192 200 10.1289/ehp.7337.15687057
CDC 2003. Second National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:Centers for Disease Control and Prevention.
CDC 2005. Third National Report on Human Exposure to Environmental Chemicals. Atlanta, GA:Centers for Disease Control and Prevention.
Curl CL Fenske RA Elgethun K 2003 Organophosphorus pesticide exposure to urban and suburban preschool children with organic and conventional diets Environ Health Perspect 111 377 382 12611667
Curl CL Fenske RA Kissel JC Shirai JH Moate TF Griffith W 2002 Evaluation of take-home organophosphorus pesticide exposure among agricultural workers and their children Environ Health Perspect 110 A787 A792 12460819
Fenske RA Curl CL Kissel JC 2003 The effect of the 14-day agricultural restricted entry interval on azinphosmethyl exposures in a group of apple thinners in Washington State Regul Toxicol Pharmacol 38 91 97 12878058
Fenske RA Kissel JC Curl CL Shirai JH Galvin K 2004 Agricultural task not predictive of children’s exposure to organophosphorus pesticides Environ Health Perspect 112 A865 A866 15531415
Fenske RA Lu C Barr D Needham L 2002 Children’s exposure to chlorpyrifos and parathion in an agricultural community in central Washington State Environ Health Perspect 110 549 553 12003762
Heudorf U Angerer J 2001 Metabolites of organophosphorous insecticides in urine specimens from inhabitants of a residential area Environ Res 86 80 87 11386745
Heudorf U Angerer J Drexler H 2004 Current internal exposure to pesticides in children and adolescents in Germany: urinary levels of metabolites of pyrethroid and organophosphorus insecticides Int Arch Occup Environ Health 77 67 72 14551781
James RR Hern SC Robertson GL Schumacher BA 2003. Comparison of Methods for the Determination of Alkyl Phosphates in Urine. Report No. 600/R-03/075. Washington, DC:U.S. Environmental Protection Agency Office of Research and Development. Available: http://www.epa.gov/heasdweb/children/pdf/alkyl-phosphates-hern.pdf [accessed 10 January 2005].
Koch D Lu CA Fisker-Andersen J Jolley L Fenske RA 2002 Temporal association of children’s pesticide exposure and agricultural spraying: report of a longitudinal biological monitoring study Environ Health Perspect 110 829 833 12153767
Loewenherz C Fenske RA Simcox NJ Bellamy G Kalman D 1997 Biological monitoring of organophosphorus pesticide exposure among children of agricultural workers in central Washington State Environ Health Perspect 105 1344 1353 9405329
Lu C Bravo R Caltabiano LM Irish RM Weerasekera G Barr DB 2005 The presence of dialkylphosphates in fresh fruit juices: implication for organophosphorus pesticide exposure and risk assessments J Toxicol Environ Health A 68 209 227 15762180
Lu C Fenske RA Simcox NJ Kalman D 2000 Pesticide exposure of children in an agricultural community: evidence of household proximity to farmland and take-home exposure pathways Environ Res 84 290 302 11097803
Lu C Knutson DE Fisker-Andersen J Fenske RA 2001 Biological monitoring survey of organophosphorus pesticide exposure among preschool children in the Seattle metropolitan area Environ Health Perspect 109 299 303 11333193
Mills PK Zahm SH 2001 Organophosphate pesticide residues in urine of farmworkers and their children in Fresno County, California Am J Ind Med 40 571 577 11675626
Moate T Lu C Fenske RA Hahne R Kalman DA 1999 Improved cleanup and determination of dialkyl phosphates in the urine of children exposed to organophosphorous insecticides J Anal Toxicol 23 230 236 10445484
Morgan MK Sheldon LS Croghan CW Jones PA Robertson GL Chuang JC 2005 Exposures of preschool children to chlorpyrifos and its degradation product 3,5,6-trichloro-2-pyridinol in their everyday environments J Expo Anal Environ Epidemiol 15 297 309 15367928
National Research Council 1991. Monitoring Human Tissues for Toxic Substances. Washington, DC:National Academy Press.
National Research Council 1993. Pesticides in the Diets of Infants and Children. Washington, DC:National Academy Press.
Royster MO Hilborn ED Barr D Carty CL Rhoney S Walsh D 2002 A pilot study of global positioning system/geographical information system measurement of residential proximity to agricultural fields and urinary organophosphate metabolite concentrations in toddlers J Expo Anal Environ Epidemiol 12 433 440 12415492
Shalat SL Donnelly KC Freeman NC Calvin JA Ramesh S Jimenez M 2003 Nondietary ingestion of pesticides by children in an agricultural community on the US/Mexico border: preliminary results J Expo Anal Environ Epidemiol 13 42 50 12595883
Simcox NJ Camp J Kalman D Stebbins A Bellamy G Lee I-C 1999 Farmworker exposure to organophosphorus pesticide residues during apple thinning in central Washington State Am Ind Hyg Assoc J 60 752 761 10635541
Simcox NJ Fenske RA Wolz S Lee I-C Kalman D 1995 Pesticides in housedust and soil: exposure pathways for children of agricultural families Environ Health Perspect 103 1126 1134 8747019
Thompson B Coronado GD Grossman JE Puschel K Solomon CC Islas I 2003 Pesticide take-home pathway among children of agricultural workers: study design, methods, and baseline findings J Occup Environ Med 45 1 42 53 12553178
|
16263526
|
PMC1310933
|
CC0
|
2021-01-04 23:41:33
|
no
|
Environ Health Perspect. 2005 Nov 6; 113(11):1651-1657
|
utf-8
|
Environ Health Perspect
| 2,005 |
10.1289/ehp.8022
|
oa_comm
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.